Thom Davis, B. J. Tomlinson, Jim Ledbetter (Auth.), R. G. Ross Jr. (Eds.) - Cryocoolers 11-Springer US (2002)

CRYOCOOLERS 11
A publication of the International Cryocooler Conference
CRYOCOOLERS 11
Edited by
R. G. Ross, Jr.
Jet Propulsion Laboratory
California Institute of Technology
Pasadena, California
KLUWER ACADEMIC PUBLISHERS

NEW YORK, BOSTON, DORDRECHT, LONDON, MOSCOW
H%RRN ,6%1 0-306-47112-4
3ULQW,6%1 0-306-46567-1
.OXZHU$FDGHPLF3XEOLVKHUV
1HZ<RUN%RVWRQ'RUGUHFKW/RQGRQ0RVFRZ
3ULQW1.OXZHU$FDGHPLF3OHQXP3XEOLVKHUV
1HZ<RUN
$OOULJKWVUHVHUYHG
1RSDUWRIWKLVH%RRNPD\EHUHSURGXFHGRUWUDQVPLWWHGLQDQ\IRUPRUE\DQ\PHDQVHOHFWURQLF
PHFKDQLFDOUHFRUGLQJRURWKHUZLVHZLWKRXWZULWWHQFRQVHQWIURPWKH3XEOLVKHU
&UHDWHGLQWKH8QLWHG6WDWHVRI$PHULFD
9LVLW.OXZHU2QOLQHDW KWWSNOXZHURQOLQHFRP
DQG.OXZHU
VH%RRNVWRUHDW KWWSHERRNVNOXZHURQOLQHFRP
Preface
Over the last two years we have witnessed a continuation in the breakthrough shift toward
pulse tube cryocoolers for long-life, high-reliability cryocooler applications. One class of pulse
tubes that has reached maturity is referred to as "Stirling type" because they are based on the
linear Oxford Stirling-cooler type compressor; they generally provide cooling in the 30 to 100 K
temperature range and operate at frequencies from 30 to 60 Hz. The other type of pulse tube
cooler making great advances is the so-called "Gifford-McMahon type." Pulse tube coolers of this
type use a G-M type compressor and lower frequency operation to achieve temperatures in the 2 to
10 K temperature range. Nearly a third of this proceedings covers these new developments in the
pulse tube arena.
Complementing the work on low-temperature pulse tubes is substantial continued progress on
rare earth regenerator materials and Gifford-McMahon coolers. These technologies continue to
make great progress in opening up the 2 - 4 K market. Also in the commercial sector, continued
interest is being shown in the development of long-life, low-cost cryocoolers for the emerging high
temperature superconductor electronics market, particularly the cellular telephone base-station
market. At higher temperature levels, closed-cycle J-T or throttle-cycle refrigerators are taking
advantage of mixed refrigerant gases, spearheaded in the former USSR, to achieve low-cost cryo-
cooler systems in the 65 - 80 K temperature range. Tactical Stirling cryocoolers, the mainstay of
the defense industry, continue to find application in cost-constrained commercial applications and
space missions; the significant development here is the cost-effective incorporation of Oxford-like
flexure spring piston supports so as to achieve an extended-life, low-cost product.
The objective of Cryocoolers 11 is to archive these latest developments and performance
measurements by drawing upon the work of the leading international experts in the field of cryo-
coolers. In particular, this book is based on their contributions at the 11th International Cryocool-
er Conference, which was held in Keystone, Colorado, in June 2000. The program of this confer-
ence consisted of 127 papers; of these, 98 are published here. Although this is the eleventh
meeting of the conference, which has met every two years since 1980, the authors’ works have
only been made available to the public in hardcover book form since 1994. This book is thus the
fourth volume in this new series of hardcover texts for users and developers of cryocoolers.
Because this book is designed to be an archival reference for users of cryocoolers as much as
for developers of cryocoolers, extra effort has been made to provide a thorough Subject Index that
covers the referenced cryocoolers by type and manufacturer’s name, as well as by the scientific or
engineering subject matter. Extensive referencing of test and measurement data, and application
and integration experience, is included under specific index entries. Contributing organizations
are also listed in the Subject Index to assist in finding the work of a known institution, laboratory,
or manufacturer. To aide those attempting to locate a particular contributor’s work, a separate
Author Index is provided, listing all authors and coauthors.
Prior to 1994, proceedings of the International Cryocooler Conference were published as
informal reports by the particular government organization sponsoring the conference – typically
a different organization for each conference. A listing of previous conference proceedings is
presented in the Proceedings Index, at the rear of this book. Most of the previous proceedings
were printed in limited quantity and are out of print at this time.
v
vi PREFACE
The content of Cryocoolers 11 is organized into 19 chapters, starting first with an introduc-
tory chapter providing summaries of major government cryocooler development and test pro-
grams. The next several chapters address cryocooler technologies organized by type of cooler,
starting with regenerative coolers; these include Stirling cryocoolers, pulse tube cryocoolers,
Gifford-McMahon cryocoolers, and associated regenerator research. Next, Turbo-Brayton, Joule-
Thomson, and sorption cryocoolers, as well as sub-Kelvin refrigerators are covered in a progres-
sion of lowering temperatures. The technology-specific chapters end with a chapter on Optical
Refrigeration; this provides a glimpse into the future with miniature solid-state refrigerators using
advanced optical-based refrigeration cycles. The last four chapters deal with cryocooler reliability
investigations, integration technologies, and experience to date in a number of representative
space and commercial applications. The articles in these last four chapters contain a wealth of
information for the potential user of cryocoolers, as well as for the developer.
The expanding availability of low-cost, reliable cryocoolers is making major advances in a
number of fields. It is hoped that this book will serve as a valuable reference to all those faced
with the challenges of developing and using cryocoolers.
Ronald G. Ross, Jr.

Acknowledgments
The International Cryocooler Conference Board wishes to thank Ball Aerospace & Technolo-
gies Corp., which hosted the 11th ICC, and to express its deepest appreciation to the Conference
Organizing Committee, whose members dedicated many hours to organizing and managing the
conduct of the Conference. Members of the Organizing Committee and Board for the 11th ICC
include:
CONFERENCE CO-CHAIRS PROGRAM COMMITTEE

Rodney Oonk, Ball Aerospace John Brisson, MIT
Richard Reinker, Ball Aerospace William Burt, TRW
Stephen castles, NASA/GSFC
CONFERENCE ADMINISTRATORS Peter Kerney, Leybold
Margueritte Sommers, Ball Aerospace Eric Marquardt, NIST
Dianne Fisher, Ball Aerospace Lawrence wade, JPL
PROGRAM CHAIRMAN ADVISORY BOARD

Klaus Timmerhaus, Univ. of Colorado Guobang Chen, zhejiang Univ., china
Thom Davis, AFRL
Dave Glaister, Ball Aerospace
CONFERENCE SECRETARY
Geoff Green, MAPC
Jill Bruning, Nichols Research Corp.
Tom Kawecki, NRL
Peter Kittel, NASA/ARC
PUBLICATIONS Ralph Longsworth, APD Cryogenics
Ron Ross, Jet Propulsion Laboratory Yoichi Matsubara, Nihon Univ., Japan
Ted Nast, Lockheed Martin ATC
TREASURER Martin Nisenoff, NRL
Ray Radebaugh, NIST Walter Swift, Creare
In addition to the Committee and Board, key staff personnel made invaluable contributions to
the preparations and conduct of the conference. Special recognition is due C. Hall, P. Irwin, J.M.
Lee, R. Mestas, B. Oonk, A. Ravex, B. Reinker, C. Stoyanof, M. Stoyanof, and J. Timmerhaus.
vii
Contents
Government Cryocooler Development and Test Programs 1

Military Space Cryogenic Cooling Requirements for the 21st Century ...... 1
T.M. Davis and B.J. Tomlinson, Kirtland AFB, NM; and J.D. Ledbetter, Mission
Research Corp., Albuquerque, NM
Status of Programs for the DoD Family of Linear Drive Cryogenic

Coolers for Weapon Systems .................... .................. 11
W.E. Salazar, US Army Night Vision, Fort Belvoir, VA
Air Force Research Laboratory Cryocooler Characterization and

Endurance Update .............................................. 17
B.J. Tomlinson, C.H. Yoneshige, AFRL, Kirtland AFB, NM; and N.S. Abhyankar, Dynacs
Engin., Albuquerque, NM
Air Force Research Laboratory Cryocooler Reliability Initiatives ........... 27

S. Blankenship and T.L. Fountain, Georgia Inst. of Tech., Atlanta, GA; and T.M. Davis
and B.J. Tomlinson, AFRL, Kirtland AFB, NM
Space Stirling Cryocooler Developments 35

Protoflight Spacecraft Cryocooler Performance Results ................... 35
K. Price, Raytheon, El Segundo, CA; and J. Reilly, N. Abhyankar, and B.J. Tomlinson, AFRL,
Kirtland AFB, NM
Characterization of Raytheon’s 60 K 2 W Protoflight Spacecraft

Cryocooler .................................................... 45
N.S. Abhyankar, Dynacs Engin., Albuquerque, NM; and C.H. Yoneshige, B.J. Tomlinson, and J.
Reilly, AFRL, Kirtland AFB, NM
The Development of a 10 K Closed Cycle Stirling Cooler for Space Use ...... 55
G. Baker, D. Féger, and A. Little, Astrium, Stevenage, UK; A.H. Orlowska,
T. Bradshaw, and M. Crook, RAL, Chilian, UK; B.J. Tomlinson, AFRL, Kirtland
AFB, NM; and A. Sargeant, Cubic Appl. Inc., Lacey, WA
Development of a 12 K Stirling Cycle Precooler for a 6 K Hybrid

Cooler System ................................................. 63
W.J. Gully, D.S. Glaister, and D.W. Simmons, Ball Aerospace, Boulder, CO
Thermodynamic Optimization of Multi-Stage Cryocoolers ................ 69

C.S. Kirkconnell and K.D. Price, Raytheon, El Segundo, CA
ix
x CONTENTS
Long-Life Tactical and Commercial Stirling Coolers 79

The Advent of Low-Cost Cryocoolers ................................. 79
R.Z. Unger, R.B. Wiseman, and M.R. Hummon, Sunpower, Inc., Athens, OH
Performance and Reliability Improvements in a Low-Cost Stirling

Cycle Cryocooler ............................................... 87
M. Hanes, Superconductor Tech. Inc., Santa Barbara, CA
Development of a Long-Life Stirling Cryocooler ........................ 97

Y. Ikuta, Y. Suzuki, K. Kanao, and N. Watanabe, Sumitomo Heavy Indus.,
Hiratsuka, Kanagawa, Japan
Flexure Springs Applied to Low-Cost Linear Drive Cryocoolers ............ 103

R.M. Rawlings and S. Miskimins, DRS Infrared Tech., Dallas, TX
High Reliability Coolers under Development at Signaal-USFA ............. 111

M. Meijers, A.A.J. Benschop, and J.C. Mullié, Signaal-USFA, Eindhoven,
The Netherlands
Long-Life Commercial Pulse Tube Coolers 119

Development of a Long-Life Stirling Pulse Tube Cryocooler for
Superconducting Filter Subsystems ................................. 119
Y. Hiratsuka, Daikin Indus., Osaka, Japan; K. Murayama, Y. Maeda, F. Imai, and K.Y. Kang,
Daikin Envir. Lab, Tsukuba, Japan; and Y. Matsubara, Nihon Univ., Funabashi, Japan
Development of a 5 W at 65 K Air-Cooled Pulse Tube Cryocooler ........... 125

S-Y Kim, J-J Park, S-T Kim, W-S Chung, and H-K Lee, LG Electronics Inc., Seoul, Korea
Space Pulse Tube Cryocooler Developments 131

TES FPC Flight Pulse Tube Cooler System ............................ 131
J. Raab, S. Abedzadeh, R. Colbert, J. Godden, D. Harvey, and C. Jaco, TRW,
Redondo Beach, CA
The AIM Space Cryocooler Program ................................. 139

I. Rühlich, H. Korf, and Th. Wiedmann, AEG Infrarot-Module, Heilbronn, Germany
Miniature Pulse Tube Cryocooler for Space Applications ................. 145

T.C. Nast, P.J. Champagne, V. Kotsubo, J. Olson, A. Collaco, and B. Evtimov,
Lockheed Martin ATC, Palo Alto, CA; T. Renna, Lockheed Martin Communications,
Newtown, PA; and R. Clappier, Clappier Consulting, Discovery Bay, CA
Gamma-Ray Pulse Tube Cooler Development and Testing ................ 155

R.G. Ross, Jr., D.L. Johnson, and A. Metzger, JPL, Pasadena, CA; V. Kotsubo,
B. Evtimov, J. Olson, and T. Nast, Lockheed-Martin ATC, Palo Alto, CA; and
R.M. Rawlings, DRS Infrared Tech., Dallas, TX
High Efficiency Pulse Tube Cooler ................................... 163

E. Tward, C.K. Chan, J. Raab, T. Nguyen, and R. Colbert, TRW, Redondo Beach,
CA; and T. Davis, AFRL, Kirtland AFB, NM
CONTENTS xi
High Performance Flight Cryocooler Compressor ....................... 169

P.B. Bailey and M.W. Dadd, Oxford Univ., Oxford, UK; N. Hill and C.F. Cheuk,
Hymatic Engin. Co., Redditch, UK; and J. Raab and E. Tward, TRW, Redondo Beach, CA
Vibration Reduction in Balanced Linear Compressors .................... 175

M.W. Dadd., P.B. Bailey, and G. Davey, Oxford Univ., Oxford, UK; and T. Davis
and B.J. Tomlinson, AFRL, Kirtland AFB, NM
95K High Efficiency Cryocooler Program .............................183

K. Price, Raytheon, El Segundo, CA; and V. Urbancek, AFRL, Kirtland AFB, NM
Design and Test of the NIST/Lockheed Martin Miniature Pulse Tube

Flight Cryocooler .............................................. 189
P.E. Bradley and R. Radebaugh, NIST, Boulder, CO; J.H. Xiao, Johnson and Johnson,
Somerville, NJ; and D.R. Ladner, Lockheed Martin Astronautics, Denver, CO
Low-Cost Pulse Tube Liquefier for In-Situ Resource Utilization ............199

C.M. Martin and J.L. Martin, Mesoscopic Devices, Broomfield, CO
GM-Type Pulse Tube Coolers for Low Temperatures 205

Performance Characteristics of a 4 K Pulse Tube in Current Applications ..... 205
C. Wang and P.E. Gifford, Cryomech, Inc., Syracuse, NY
Experimental Study of a 4K Pulse Tube Cryocooler ..................... 213

S. Fujimoto, T. Kurihara, T. Oodo, Y.M. Kang, Daikin Ltd., Tsukuba, Japan;
T. Numazawa, Nat. Res. Inst. for Metals, Tsukuba, Japan; and Y. Matsubara,
Nihon Univ., Chiba, Japan
GM-Type Two-Stage Pulse Tube Cooler with High Efficiency ............. 221
A. Hofmann, Karlsruhe Inst. for Tech. Physics, Karlsruhe, Germany; H. Pan and L.
Oellrich, Univ. of Karlsruhe, Karlsruhe, Germany
Developments on Single and Double Stage GM Type Pulse Tube

Cryorefrigerators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
J.M Poncet, A. Ravex, and I. Charles, CEA/DRFMC/Service des Basses Temperatures,
Grenoble, France
30 - 50 K Single Stage Pulse Tube Refrigerator for HTS Applications ........ 235
J. Yuan, J. Maguire, A. Sidi-Yekhlef, and P. Winn, American Superconductor Co.,
Westborough, MA
Two-Stage 4K Pulse Tube Refrigerator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243

S. Zhu, M. Ichikawa, M. Nogawa, and T. Inoue, Aisin Seiki Co., Ltd., Kariya, Aichi, Japan
Compressor-Specific Design of a Single Stage Pulse Tube Refrigerator ....... 249

J.M. Pfotenhauer and J.H. Baik, Univ. of Wisconsin, Madison, WI
Hybrid Cryocoolers Using Pulse Tubes 259

A Novel Multi-Stage Expander Concept ............................... 259
C.S. Kirkconnell, K.D. Price, M.C. Barr, and J.T. Russo, Raytheon, El Segundo, CA
xii CONTENTS
Numerical Study of a New Type of 4K GM/PT Hybrid Refrigerator . . . . . . 265

L. Liu, L. Gong, J. Liang, and L. Zhang, Cryogenic Lab, Chinese Acad. of Sci.,
Beijing, China
Thermally Actuated 3He Pulse Tube Cooler .......................... 273

Y. Matsubara, H. Kobayashi, and S.L. Zhou, Atomic Energy Res. Inst., Nihon Univ.,
Chiba, Japan
Investigation of Helium and Nitrogen Mixtures in a Pulse Tube

Refrigerator ................................................... 281
Z.H. Gan and G.B. Chen, Zhejiang Univ., Hangzhou, China; and G. Thummes
and C. Heiden, Univ. of Giessen, Giessen, Germany
Pulse Tube Refrigeration with a Combined Cooling and Freezing

Cycle for HTSC Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
G. Chen, Z. Gan, L. Qiu and J. Yu, Zhejiang Univ., Hangzhou, China
Experimental Investigation of a Pulse Tube Refrigerator Driven

by a Thermoacoustic Prime Mover ................................. 301
L.M. Qiu, G.B. Chen, N. Jiang, Y.L. Jiang, and J.P. Yu, Zhejiang Univ.,
Hangzhou, China
Design, Development, and Operation of a Thermo-Acoustic

Refrigerator Cooling to below -60°C .............................. 309
M.E.H. Tijani, J. Zeegers, and A.T.A.M. de Waele, Eindhoven Univ. of Tech.,
Eindhoven, The Netherlands
Pulse Tube Analysis and Experimental Measurements 317

Design of a Miniature Pulse Tube Refrigerator . . . . . . . . . . . . . . . . . . . . . . . . . 317
A. Halouane, French Inst. of Petroleum, Rueil-Malmaison, France; and
J-C. Marechal and Y. Simon, Ecole Normale Supérieure, Paris, France
Investigation of a Single Stage Four-Valve Pulse Tube Refrigerator

for High Cooling Power ......................................... 327
T. Schmauder A. Waldauf, M. Thürk, R. Wagner, and P. Seidel, Univ. of Jena,
Jena, Germany
Analysis and Experimental Research of a Multi-Bypass Version

Pulse Tube Refrigerator ........................................ 337
L.W. Yang, J.T. Liang and Y. Zhou, Chinese Academy of Sciences, Beijing, China
Experimental Study of the Heat Transfer in Pulse Tubes ................. 345

S. Jeong, K. Nam and M.G. Kim, Korea Adv. Inst. of Sci. and Tech., Taejon,
Korea; and H.-M. Chang and E.S. Jeong, Hong Ik Univ., Seoul, Korea
Shuttle Loss in Pulse Tubes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353

L. W. Yang, Chinese Academy of Sciences, Beijing, China
Numerical Study of Gas Dynamics Inside of a Pulse Tube Refrigerator . . . . . . 363

Y. Hozumi, Chiyoda Corp., Yokohama, Japan; M. Murakami, Univ. of Tsukuba,
Tsukuba, Japan; and M. Shiraishi, ME Lab, MITI, Tsukuba, Japan
CONTENTS xiii
Visualization of DC Gas Flows in a Double-Inlet Pulse Tube

Refrigerator with a Second Orifice Valve ............................ 371
M. Shiraishi and A. Nakano, ME Lab, MITI, Tsukuba, Japan; K. Takamatsu and
M. Murakami, Univ. of Tsukuba, Tsukuba, Japan; T. Iida, NASDA, Tsukuba, Japan;
and Y. Hozumi, Chiyoda Corp., Yokohama, Japan
GM Refrigerator Developments 381

A Gifford-McMahon Cycle Cryocooler below 2K ..................... 381
T. Satoh, Sumitomo Heavy Ind., Kanagawa, Japan; A. Onishi, Sumitomo Heavy Ind.,
Tokyo, Japan; I. Umehara, Y. Adachi, and K. Sato, Yokohama Nat ’l Univ., Yokohama, Japan;
and E.J. Minehara, FEL Lab, Japan Atomic Energy Res. Inst., Naka, Japan
High Efficiency, Single-Stage GM Cryorefrigerators Optimized for

20 to 40K . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 387
C. Wang and P.E. Gifford, Cryomech, Inc., Syracuse, NY
Remote Cooling with a G-M Cryocooler by Use of Cold Electro-

magnetic Valves Driving an External Flow Loop ..................... 393
K.M. Ceridon and J.L. Smith, Jr., MIT, Cambridge, MA
Optimum Intermediate Temperatures of Two-Stage Gifford-McMahon

Type Coolers .................................................. 401
T.C. Chuang, Raytheon-RCSI, Philadelphia, PA; S. Yoshida, Taiyo Toyo Sanso, Co., Kawasaki,
Japan; and T.H.K. Frederking, UCLA, Los Angeles, CA
Regenerator Analysis and Materials Developments 409

Regenerator Behavior with Heat Input or Removal at Intermediate
Temperatures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 409
R. Radebaugh, E.D. Marquardt, J. Gary, and A. O’Gallagher, NIST, Boulder, CO
Measurement of Heat Conduction through Metal Spheres ................ 419

M.A. Lewis and R. Radebaugh, NIST, Boulder, CO
Innovative Technology for Low Temperature Regenerators ............... 427

L. Tuchinskiy and R. Loutfy, MER Corp., Tucson, AZ; and B.J. Tomlinson, AFRL,
Kirtland AFB, NM
Ductile, High Heat Capacity, Magnetic Regenerator Alloys for the

10 to 80 K Temperature Range .................................... 433
K.A. Gschneidner, Jr., A.O. Pecharsky, and V.K. Pecharsky, Ames Lab, Iowa State Univ., Ames, IA
Low Temperature Properties of HoSb, DySb, and GdSb .................. 443

H. Nakane and S. Yamazaki, Kogakuin Univ., Tokyo, Japan; H. Fujishiro, Iwate Univ., Morioka,
Japan; T. Yamaguchi and S. Yoshizawa, Meisei Univ., Tokyo, Japan; T. Numazawa, Nat. Res. Inst.
for Metal, Tsukuba, Japan; and M. Okamura, Toshiba Corp., Yokohama, Japan
Manufacturing Considerations for Rare Earth Powders Used in

Cryocooler and Magnetic Refrigerator Applications .................... 449
S.A. Miller, J.D. Nicholson, Starmet Corp., Concord, MA; and K.A. Gschneidner, Jr.,
A.O. Pecharsky, and V.K. Pecharsky, Ames Laboratory, Iowa State Univ., Ames, IA
Magnetothermal Properties of Polycrystalline Gd2In . . . . . . . . . . . . . . . . . . 457

M.I. Ilyn and A.M. Tishin, Moscow State Univ., Moscow Russia; K.A. Gschneidner, Jr., V.K.
Pecharsky, and A.O. Pecharsky, Ames Labs, Iowa State Univ., Ames, IA
xiv CONTENTS
New Regenerator Material for Sub-4 K Cryocoolers ...................... 465

T. Numazawa, O. Arai and A. Sato, Tsukuba Magnet Lab, Nat. Res. Inst. for Metals, Tsukuba,
Japan; S. Fujimoto, T. Oodo, and Y.M. Rang, Daikin, Ltd., Tsukuba, Japan;
and T. Yanagitani, Konoshima Chemical Co., Kagawa, Japan
New Regenerator Materials for Use in Pulse Tube Coolers ................ 475
A. Kashani and B.P.M. Helvensteijn, Atlas Scientific, Sunnyvale, CA, P. Kittel,
NASA/ARC, Moffett Field, CA; and K.A. Gschneidner, Jr., V.K. Pecharsky, and
A.O. Pecharsky, Ames Labs, Iowa State Univ., Ames, IA
Turbo-Brayton Cryocooler Developments 481

Advanced Developments for Low Temperature Turbo-Brayton
Cryocoolers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 481
J.A. McCormick, G.F. Nellis, H. Sixsmith, M.V. Zagarola, M.G. Izenson and
W.L. Swift, Creare, Hanover, NH; and JA. Gibbon, NASA/GSFC, Greenbelt, MD
Life and Reliability Characteristics of Turbo-Brayton Coolers .............. 489

J.J. Breedlove, M. V. Zagarola, G.F. Nellis, F.X. Dolan, and W.L. Swift, Creare,
Hanover, NH; and J.A. Gibbon, NASA/GSFC, Greenbelt, MD
A Flexible Turbo-Brayton Cryocooler Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 499

P.L. Whitehouse, NASA/GSFC, Greenbelt, MD; and G.F. Nellis and M.V. Zagarola,
Creare, Inc., Hanover, NH
J-T and Throttle-Cycle Cryocooler Developments 505

A 10 K Cryocooler for Space Applications ............................. 505
D.S. Glaister, W.J. Gully, G.P. Wright and D.W. Simmons, Ball Aerospace,
Boulder, CO; and B.J. Tomlinson, AFRL, Kirtland AFB, NM
Modern Trends in Designing Small-ScaleThrottle-Cycle Coolers

Operating with Mixed Refrigerants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 513
M. Boiarski and A. Khatri, IGC-APD Cryogenics, Allentown, PA; O. Podcherniaev,
IGC-Polycold Sys., San Rafael, CA; and V. Kovalenko, Moscow Power Engin. Inst.,
Moscow, Russia
Thermodynamic Analysis of an Mixed-Refrigerant Auto-Cascade

J-T Cryocooler with Distributed Heat Load .......................... 523
M.Q. Gong, E.C. Luo, J.T. Liang, Y. Zhou, and J.F. Wu, Chinese Academy of Sciences, Beijing,
China
Sorption Cryocooler Developments 531

PLANCK Sorption Cooler Initial Compressor Element
Performance Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 531
C.G. Paine, R.C. Bowman Jr., D. Pearson, M.E. Schmelzel, P. Bhandari, and
L.A. Wade, JPL, Pasadena, CA
Sizing and Dynamic Performance Prediction Tools for 20 K

Hydrogen Sorption Cryocoolers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 541
P.Bhandari, M. Prina, R.C. Bowman and L.A. Wade, JPL, Pasadena, CA; and
M. Ahart, Princeton Univ., Princeton, NJ
CONTENTS xv
165 K Microcooler Operating with a Sorption Compressor and a

Micromachined Cold Stage ....................................... 551
J.F. Burger, H.J. Holland, J.H. Seppenwoolde, J.W. Berenschot, H.J.M. ter Brake, J.G.E.
Gardeniers, M. Elwenspoek and H. Rogalla, Univ. of Twente, Enschede, The Netherlands
Sub-Kelvin Refrigerator Developments 561

Double Stage Helium Sorption Coolers ............................... 561
L. Duband, CEA/DRFMC/Service des Basses Températures, Grenoble, France
Sub-Kelvin Sorption Coolers for Space Application ...................... 567

L. Duband, CEA/DRFMC, Grenoble, France; B. Collaudin, ESTEC, Noordwijk, The Nether-
lands; and P. Jamotton, Centre Spatial de Liège, Belgium
Closed-Cycle Cooling of Infrared Detectors to 0.25 K for the Polatron ....... 577
R.S. Bhatia, J.J. Bock, V.V. Hristov, W.C. Jones, A.E. Lange, J. Leong, P.V. Mason, B.J. Philhour
and G. Sirbi, Caltech, Pasadena, CA; S.E. Church and B.G. Keating, Stanford Univ.,
Standford, CA; J.G. Glenn, Univ. of Colorado, Boulder, CO; S.T. Chase, Chase Research Ltd.,
Sheffield, UK; and P.A.R. Ade and C.V. Haynes, QMW College, London, UK
A Continuous Adiabatic Demagnetization Refrigerator for Use

with Mechanical Coolers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 587
P. Shirron, E. Canavan, M. DiPirro, M. Jackson, J. Panek and J. Tuttle, NASA/GSFC,
Greenbelt, MD; and N. Abbondante, M. Grabowski and M. Hirsch, Worcester Polytechnic
Institute, Worcester, MA
Reaching 96 mK by a Pulse-Tube Precooled Adiabatic Demagnetization

Refrigerator ................................................... 597
G. Thummes and M. Theiß, Inst. of Applied Physics, Giessen, Germany; and
M. Bühler and J. Höhne, CSP GmbH, Ismaning, Germany
Dissipation in Metal Welded Bellows and Its Consequences for

Sub-Kelvin Refrigerators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 605
C.L. Phillips and J.G. Brisson, MIT, Cambridge, MA
Optical Refrigeration Developments 613

Design and Predicted Performance of an Optical Cryocooler for a
Focal Plane Application .......................................... 613
G.L. Mills, A.J. Mord and P.A. Slaymaker, Ball Aerospace, Boulder, CO
Optical Refrigeration Using Anti-Stokes Fluorescence from

Molecular Dyes ................................................ 621
G. Rumbles, B. Heeg, and J.L. Lloyd (née Clark), Imperial College, London, UK; P.A.
DeBarber, MetroLaser, Inc., Irvine, CA; and B.J. Tomlinson, AFRL, Kirtland AFB, NM
Solid-State Optical Cooler Developments .............................. 631

B.C. Edwards, J.E. Anderson, and R.I. Epstein, Los Alamos Nat’l Lab, Los Alamos, NM; and
C. W. Hoyt and M. Sheik-Bahae, Univ. of New Mexico, Albuquerque, NM
Cryocooler Reliability Investigations and Analyses 637

Cryocooler Reliability and Redundancy Considerations for Long-Life
Space Missions ................................................. 637
R.G. Ross, Jr., JPL, Pasadena, CA
xvi CONTENTS
Space Cryocooler Contamination Lessons Learned and Recommended

Control Procedures ............................................. 649
S. Castles, NASA/GSFC, Greenbelt, MD; K.D. Price, Raytheon, El Segundo, CA; D.S. Glaister
and W.J. Gully, Ball Aerospace, Boulder, CO; J. Reilly, AFRL, Kirtland AFB, NM; and T. Nast
and V. Kotsubo, Lockheed-Martin, Palo Alto, CA
Cryocooler Contamination Study: Temperature Dependence

of Outgassing ................................................. 659
S.W.K. Yuan and D.T. Kuo, BAE Systems, Sylmar, CA
BAE’s Life Test Results on Various Linear Coolers and Their Correlation
with a First Order Life Estimation Method .......................... 665
D.T. Kuo, T.D. Lody and S.W.K. Yuan, BAE Systems, Sylmar, CA
Initial Observations from the Disassembly and Inspection of the

TRW 3503 and Creare SSRB ...................................... 673
B.J. Tomlinson and C.H. Yoneshige, AFRL, Kirtland AFB, NM; and M.L. Martin,
Dynacs Engin., Albuquerque, NM
Cryocooler Integration Technologies and Materials 681

Cryogenic Material Properties Database ............................... 681
E.D. Marquardt, J.P. Le, and R. Radebaugh, NIST, Boulder, CO
Experimental Results on the Thermal Contact Conductance of

Ag-Filled Epoxied Junctions at Cryogenic Temperatures ................ 689
Z. Wang, A. Devpura, and P.E. Phelan, Arizona State Univ., Tempe, AZ
A Fail-Safe Experiment Stand for Cryocooler Characterization ............. 699

C.H. Yoneshige, J.P. Kallman, G. Lybarger, AFRL, Kirtland AFB, NM; and
N.S. Abhyankar and M.L. Martin, Dynacs Engin., Albuquerque, NM
Development and Testing of a Gimbal Thermal Transport System .......... 707

D. Bugby, B. Marland, and C. Stouffer, Swales Aerospace, Beltsville, MD; and
B. Tomlinson and T. Davis, AFRL, Kirtland AFB, NM
Cryocooler Interface System ........................................ 719

G.S. Willen, Tech. Applications, Inc., Boulder, CO; and B.J. Thomlinson, AFRL,
Kirtland AFB, NM
Development and Testing of a High Performance Cryogenic

Thermal Switch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 729
B. Marland, D. Bugby, and C. Stouffer, Swales Aerospace, Beltsville, MD; and
B. Tomlinson and T. Davis, AFRL, Kirtland AFB, NM
Thermally Conductive Vibration Isolation System for Cryocoolers .......... 739

G.S. Willen, Tech. Appl., Inc., Boulder, CO; and E.M. Flint, CSA Engin.,
Mountain View, CA
Advanced Cryogenic Integration and Cooling Technology for

Space-Based Long Term Cryogen Storage . . . . . . . . . . . . . . . . . . . . . . . . . . . . 749
B.J. Tomlinson and T.M. Davis, AFRL, Kirtland AFB, NM; and J.D. Ledbetter, Mission
Research Corp., Albuquerque, NM
CONTENTS xvii
Space Cryocooler Applications 759

MOPITT On-Orbit Stirling Cycle Cooler Performance . . . . . . . . . . . . . . . . . . . 759
G.S. Mand and J.R. Drummond, Univ. of Toronto, Toronto, Canada; and D. Henry
and J. Hackett, COM DEV Inter., Cambridge, Ontario, Canada
HIRDLS Instrument Flight Cryocooler Subsystem Integration and

Acceptance Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 769
W.C. Kiehl, D.J. Berry, D.S. Glaister, J. Richards, and R.G. Stack, Ball Aerospace, Boulder, CO
Low-Temperature, Low-Vibration Cryocooler for Next Generation

Space Telescope Instruments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 775
R.L. Oonk, D.S. Glaister, W.J. Gully and M.D. Lieber, Ball Aerospace, Boulder, CO
Commercial Cryocooler Applications 783

Considerations in Applying Open Cycle J-T Cryostats to Cryosurgery ....... 783
R.C. Longsworth, 1GC-APD Cryogenics, Allentown, PA
Interference Characterization of Cryocoolers for a High-Tc SQUID-

Based Fetal Heart Monitor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 793
A.P. Rijpma, M.R. Bangma, H.A. Reincke, E. de Vries, H.J. Holland, H.J.M. ter Brake
and H. Rogalla, Univ. of Twente, Enschede, The Netherlands
Vapor Precooling in a Pulse Tube Liquefier . . . . . . . . . . . . . . . . . . . . . . . . . . . . 803

E.D. Marquardt, R. Radebaugh, and A.P. Peskin, NIST, Boulder, CO
Terrestrial Applications of Zero-Boil-Off Cryogen Storage . . . . . . . . . . . . . . . . 809

L.J. Salerno and P. Kittel, NASA/ARC, Moffett Field, CA; J. Gaby, NASA/GRC,
Cleveland, OH; R. Johnson, NASA/KSC, FL; and E.D. Marquardt, NIST, Boulder, CO
Indexes 817
Proceedings Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 817
Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 819
Subject Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 821

Military Space Cryogenic Cooling Requirements
for the 21st Century
Thom Davis1, B. J. Tomlinson1, and Jim Ledbetter2

1
Space Vehicles Directorate, Air Force Research Laboratory
Kirtland AFB, NM 87117-5776
2
Mission Research Corporation
Albuquerque, NM, USA 87106-4266
ABSTRACT
Current space cryocooler developments have achieved performance and capability that have
made the use of active refrigeration in space missions feasible. Space flight demonstrations such as
the Sandia National Laboratory Cobra Brass and Multispectral Thermal Imager missions, the Na-
tional Aeronautics and Space Administration SABER, Hyperion, and AIRS missions baselined and
implemented active refrigeration to achieve mission goals. The NASA retrofit of the NICMOS
cooling system on the Hubble Space Telescope, due to be installed during a 2001 servicing mis-
sion, will use a reverse Brayton cycle cryocooler to provide cooling for the NICMOS sensor due to
a prematurely depleted cryogen dewar. These applications of cryocooler technology validate the
improved mission capabilities and reliability and lifetime confidence in active refrigeration in space.
Past development efforts have focused primarily on reliability and the achievement of long
life. However, looking ahead at 21st century military space applications, there are improvements
needed in several aspects of current cooling technology including higher capacity cooling loads,
mass reduction, and improvement in efficiency, low temperature performance, and lifetimes greater
than 10 years. In addition, cryogenic integration technology must be developed to allow efficient
cryocooler to cooled component integration. Significant improvements in cryocooler technology can
easily be overshadowed by gross parasitic heat loads and unacceptable cryogenic system penalties.
This paper focuses on mid-term and out-year cooling requirements for the Air Force Space
Based Infrared System Low, Space Based Laser, Advanced Space Based Infrared System, and
other Department of Defense space missions.
INTRODUCTION
With the advent of the Strategic Defense Initiative in the mid 1980s, the Department of
Defense recognized the improved mission capabilities of cryogenic cooling for detectors in
space applications. Cooled detectors allow the collection of photons at longer wavelengths,
allowing vast improvements in identification and discrimination capability with a minimum of
sensor aperture growth. Smaller aperture produces cheaper, lighter sensors, much easier to
host in a space-based environment. Other space missions such as communications, remote sensing,
and weather monitoring can benefit from subsystems using cryogenic technology including super
Cryocoolers 11, edited by R.G. Ross, Jr.

Kluwer Academic/Plenum Publishers, 2001 1
2 GOVERNMENT CRYOCOOLER DEVELOPMENT AND TEST PROGRAMS
conducting electronics, high data rate signal processors, and high speed/low power analog to
digital converters.
Priorities for the Air Force Research Laboratory cryocooler effort are to develop and
demonstrate space qualifiable cryogenic technologies required to meet future requirements for
Air Force and Department of Defense (DoD) missions. Other objectives are to develop state-of-
the-art cryocooler technology, characterize and evaluate the performance of development
hardware, pursue advanced concepts for future spacecraft missions, and work to enhance
cryocooler to spacecraft integration. Cryocooler development is tracked under the Defense
Technology Objectives initiative with performance improvement objectives having been
established for life, power, mass, and vibration. In addition to cryocooler development,
utilization of improved integration technologies ensures an optimum cryogenic thermal
management system is developed that limit or eliminates operational constraints imposed on the
spacecraft platform. Progress is reviewed annually at DoD level. Collaboration with other
government development activities and private industry has been a major strength of the AFRL
program. This has resulted in leveraging of scarce development funding and more rapid
transition of cryocooler technology to the space community. Components that significantly
improve the efficiency, extend life, reduce mass, or limit induced vibration are developed and
transitioned into next generation cryocooler designs.
The Air Force Research Laboratory (AFRL) and its predecessors, Phillips Laboratory and
the Air Force Space Technology Center, has been the primary agent of the Ballistic Missile
Defense Organization (BMDO) and the Space Based Infrared Low program office for the
development of low capacity cryogenic refrigerators and integration technologies for space
applications since the mid 1980s. These cryocooler development programs concentrated on
addressing the negative impacts of mechanical refrigerators on optical space systems: induced
line of sight vibration, longevity, power consumption, and mass. Additionally, initiatives funded
through the Air Force Science and Technology budget have addressed critical issues for other
Department of Defense users of cryogenic technology.
Early development efforts were on comparatively large capacity machines to support
cooling requirements for the Space Surveillance and Tracking System (SSTS). The protoflight
Cryocooler program produced two three-stage 10K cryocoolers (Contractors: Air Research and
Arthur D. Little) for cooling of the long wave silicon focal plane arrays. An additional program
aimed at 10K primarily developed by NASA’s Jet Propulsion Laboratory and Aerojet using
sorption, culminated with the BESTCE Shuttle flight experiment in 1995. The Standard
Spacecraft Cryocooler program (SSC) initiated in 1990 marked a change in emphasis from
relatively large machines to more compact and efficient cryocoolers aimed at meeting cooling
needs in the range from 60K to 150K for MWIR applications. Using Oxford Stirling cycle
technology developed primarily in the United Kingdom, these machines utilized linear drive
motors and tight clearance seal non-contacting piston shafts. The pulse tube cryocooler, a
variation of this technology, replaces the actively moving expander piston with a non-moving
regenerator and pulse tube. AFRL has also pursued alternate cryocooler concepts including
reverse Brayton cycle designs; and for extremely low temperature cooling (~10K), variant using
Joule-Thomson combinations and improved Stirling and pulse tube designs are being considered.
As user confidence in cryocooler reliability has improved, focus is also being placed on reduced
mass and improved efficiency.
REQUIREMENTS IDENTIFICATION
The primary purpose of AFRL cryogenic technology development program is to support
military unique mission requirements. An essential element of long-term Air Force planning is
to ensure users have superior military capabilities by quickly developing and sustaining the right
weapon systems. It is important to fully understand users' operational requirements, to develop a
range of alternative solutions and identify the critical, enabling, and enhancing technologies. At
the heart of this effort are the Air Force Technical Planning Integrated Product Teams (TPIPTs)
who facilitate the planning and development of technically superior and affordable solutions to
MILITARY SPACE COOLER REQUIREMENTS FOR 21ST CENTURY 3
operational needs. TPIPTs who facilitate the planning and development of technically superior
and affordable solutions to operational needs. A typical TPIPT is facilitated by Product Center
development planners or program office planners and consists of members from operational
commands, laboratories, air logistic centers, test centers, program offices, system engineering,
and intelligence agencies.
TPIPTs do development planning for the operational mission or mission support areas to
include near-term planning through long-term planning -- "now until forever." TPIPTs support
the user's Mission Area Assessments, Mission Needs Analyses and development of Mission Area
Plans and road maps to project future operational capabilities. The TPIPT process includes the
user and Air Force team members to gather, analyze, coordinate, and disseminate information in
each area. TPIPTs are also a critical component of the Air Force technology master process and
serve, as the primary source of weapon system technology needs. TPIPT products document
candidate systems solutions to user needs, development roadmaps, and technology investment
recommendations. The primary mission area for cryogenic cooling is the Space Control TPIPT
that addresses space surveillance, counter space, missile warning, and space based ballistic
missile defense command, control, and missile defense tasks.
Other requirements are obtained from direct coordination with various Air Force and DoD
program offices assessing their technical needs against current technology and desired mission
improvements. One other important source for identifying technical voids is the systems
contractor who eventually develops the next generation space surveillance systems.
Coordination with the contractor payload developers is essential to accurately forecast needed
advancements in cryogenic cooling technology. The end product of the planning process is
technology roadmaps and technology investment plans, which define specific technical
objectives and expected funding. Finally, AFRL works closely with other government
cryocooler developers to leverage scarce technology funding and assure research efforts are not
duplicated.
CURRENT DEVELOPMENT PROGRAMS

The primary emphasis of the current cryocooler development efforts is to support the
Engineering Manufacturing Development (EMD) requirements for the Space Based Infrared
System Low (SBIRS Low) satellite program. Several machines developed by AFRL were based
lined for the SBIRS Low Flight Demonstration System (recently terminated) and Cobra Brass
flight experiment. The Cobra Brass experiment and then Brilliant Eyes program office
developed requirements for the TRW 150K Protoflight Spacecraft Cryocooler (PSC). This “mini
pulse tube” machine has a large capacity (>2W) and high efficiency for 150K cooling, and more
limited capability at colder temperatures (down to about 65K). A TRW mini pulse tube was
flown on the unsuccessful NASA SSTI satellite launched in 1997 and two other units are
currently flying on the Cobra Brass payload. A larger capacity TRW pulse cryocooler originally
developed with BMDO funded was launched in early 2000 on the Department of Energy’s Multi
Spectral Thermal Imager Satellite. AFRL managed the development of the space qualified
cryocooler and flight electronics for Sandia National Laboratories. The NASA/Langley SABER
flight experiment scheduled for early 2001 launch is also using a mini pulse tube supplied by
AFRL.
Figure 1. TRW 150K Miniature Pulse Tube Cryocooler.

Figure 2. TRW 60K Pulse Tube Cryocooler Integrated into the

Multispectral Thermal Imager Payload.
The SBIRS Low operational system (EMD) is currently scheduled for deployment in 2006.
The major drivers for EMD cryocooler requirements are increased duty cycle, higher cooling
loads for the tracking sensors, increased cooling loads for the fore optics, 10 year design life with
complete mechanical and electronic redundancy, and higher tolerance to radiation environments.
The cryocoolers must also have a low system mass penalty and improved efficiency. While the
final EMD design is still evolving, current concepts require two focal planes in the track sensor
imager to stare at a target simultaneously in both the MWIR and LWIR bands. Dual temperature
cryocoolers offer attractive system benefits over single stage cryocoolers if redundant
cryocoolers are mandated.
System mass penalty is defined as the sum of the cryocooler mass, the mass of the electrical
power system necessary to drive it, control electronics mass, and the radiator area mass needed
to reject the waste heat. It is used as a measure of cryocooler impact on spacecraft design. The
Air Force Research Laboratory is addressing the issue of system mass penalty in several ways.
Thermal Storage Units (TSU) have been developed to absorb the wide thermal load variations
during peak duty cycle and allow system designers to size the cryocooler for the average load
instead of the peak heat load. An Air Force funded 60K TSU was successfully flight
demonstrated in October 1998 aboard the STS-95 Shuttle mission. Under the Swales Aerospace
CRYOBUS program, several cryogenic integration technologies are being designed and
fabricated for potential ground demonstrations. This program is developing cryogenic Thermal
Switches (CTSW) with high thermal resistance in the “OFF” state and low resistance in the
“ON” state to make redundant cryocooler designs feasible. Additionally, the program is
addressing various integration approaches for the optics cooling and heat rejection problems. A
cryogenic capillary pumped loop, flexible cryogenic link, ambient loop heat pipe, and cryogenic
looped heat pipe are being developed to support EMD needs.
Figure 3. Conceptual TRW 95K High Figure 4. Conceptual Raytheon 95K High
Efficiency Pulse Tube Cryocooler. Efficiency Cryocooler.
An additional requirement for SBIRS Low EMD is a cryocooler to support cooling of the
increased heat loads on the fore optics. Two AFRL programs, jointly funded by BMDO and
SBIRS Low, have the objective to design and develop an advanced high efficiency cryocooler to
meet on-gimbal optics cooling requirements of 10 Watts @ 95K. The 24-month programs will
emphasize ease of integration, low EMI signatures, low EMC susceptibility, and the ability to
survive the launch loads of any existing Air Force launch vehicle.
A contract was awarded to TRW in September 1998 for a potential SBIRS Low optics
cooler producing 10 Watts of cooling at 95K with the added objectives of minimal cryocooler
mass and input power. TRW’s design objectives are for a pulse tube cryocooler with mechanical
mass of 4 kilograms and a specific power of less than 10 W/W. The AF, BMDO, and NASA are
jointly funding the program. NASA is considering the cooler as a candidate for the space
transportation system on the Mars Exploration Mission. While still under development, the
program has already made significant improvements of the previous state-of-the-art in reduced
mass, weight, and input power. A second contract was awarded to Raytheon in June 1999. This
parallel effort uses a unique two-stage hybrid Stirling expander and pulse tube cold head design
to achieve the desired cooling loads. An important element of both programs is the inclusion of
producibility issues as an important program objective and the development of radiation
hardened flight electronics.
FUTURE SPACE SURVEILLANCE NEEDS

Improved discrimination utilizing high performance multicolor and multi-spectral focal
planes will provide a significant improvements in operational capability for surveillance and
missile tracking and detection and is being considered as a block change to the SBIRS Low
system following deployment of the initial constellation. Multiple spacecraft applications could
require near 10K operation with the use of Si:As infrared sensors for missions such as mid-
course missile detection and spectroscopy surveillance where silicon is preferred for wavelength
and/or uniformity. Traditionally stored cryogens have been used where low temperature
operation is required, but large system penalties with Dewars and prohibitive mass penalties for
most missions. Dewars are mostly applicable for short duration (< 1 year) experiments or very
small cooling loads (<10 mW). The end result is that efficient, low mass; active cryocoolers are
needed to support the low temperature cooling requirements for Very Long Wave Infrared
(VLWIR) focal planes.
Near term cooling requirements for available 128 x 128 Si:As focal plane arrays are
estimated at between 50 to 150 milliWatts depending on the manufacturer and AC power
dissipation. In all likelihood, future systems will require larger arrays with the resulting increase
in heat loads. In addition to the cooling of the tracking sensor, VLWIR surveillance systems will
cool of the aft optics at temperatures projected from 40 to 60K. While current cryocooler
requirements have addressed missile defense and tracking missions, cryocoolers are also being
used for hyperspectral applications such as proliferation detection and treaty verification. Near
term payloads are adapting previously coolers. Future systems will require cryocoolers with
much larger cooling capability to meet the expected increase in focal size with projected cooling
loads as high as five Watts at 35K. Large capacity, multi-load cryocoolers capable of cooling
sensors, optics, and optical benches will greatly simplify the cooling approach for next
generation payloads.
Two current development programs are aimed at developing a low temperature capability.
A BMDO funded near term effort with Astrium (formerly MMS) and Rutherford Appleton
Laboratory developed a two stage Stirling cryocooler with a capacity of between 0.045 Watts at
10.3K. Astrium delivered the cryocooler to AFRL in early May 2000 for performance
characterization and endurance evaluation. An Air Force funded program with design goals of
250 milliwatts @ 10K and with a specific power of <1000W/W was initiated in May 1998 with
Ball Aerospace Corporation in Boulder, Colorado. This cryocooler uses an existing Stirling pre-
cooler combined with a Joule-Tompson based Redstone Interface to achieve cooling at 10K.
Figure 5. Ball Aerospace 10K Cryocooler. Figure 6. Ball Aerospace 10K Cryocooler.
The current program schedule calls for delivery of an Engineering Development Model to AFRL
in May 2001.
Another candidate technology that could support both the 10K and high capacity 35K
cooling needs is Creare reverse Brayton cryocooler based on 65K Single Stage Reverse Brayton
(SSRB) cryocooler developed with BMDO, Air Force, NASA funding. A modified version of
this design was successfully flight demonstrated on STS-95 and will be installed on the Hubble
Telescope during an early 2001 flight servicing mission. Under a recently awarded Air Force
Small Business Innovative Research contract, Creare is evaluating designs for a range of cooling
loads and temperatures at both 35K and 10K. Minimal total input power for the cooling and
operating life in excess of 10 years are program design objectives. Low frequency vibration
(<150Hz) in this unit is negligible due to the operation of the turbine at such high rotational
speeds that results only in low Q high frequency vibration that have negligible effects on the
‘jitter’ of the sensor. A radial flow heat exchanger is also being incorporated into the design will
reduce volume by 75% and be 70% lighter than the existing SSRB heat exchanger. If approved
for Phase II, the program will produce an engineering development model at the temperature of
most interest to Air Force systems developers.
SPACE APPLICATIONS OF CRYOGENIC INTEGRATION TECHNOLOGY

Cryogenic system integration is becoming more and more important to the overall use of
cryocoolers in space. Depending on the application, various cryogenic and ambient thermal
management technologies are needed to augment and improve the cryocooler capabilities. The
realization that to ensure mission success the cryogenic application must be viewed as a system
and not as a component level mix and match is pushing mission planners to consider “end-to-
end” issues within the cryogenic system. Issues such as cryocooler redundancy, remote and
flexible cryogenic and ambient heat transport, thermal storage, and more efficient cryogenic
system integration schemes are evident in near and far term system designs and mission
applications.
Thermal straps are currently the most common technology for cryogenic integration, but
disadvantages associated with these devices include limitations on length, temperature stability
through the strap, and imposed parasitic heat loads. Although adequate for many current
applications, future systems will require additional capabilities that include long transport
distances (over 2 meters), thermal storage for duty cycle heat load applications, stringent
temperature stability, and thermal switching.
Cryocooler redundancy is a primary consideration within the cryogenic system for space
applications. In order to meet the long life goals (10+ years) for mission performance, system
developers are considering redundant cryocoolers to assure mission success. For single cooling
stage coolers (and no additional coolers cooling higher temperature radiation shields) the
question of redundancy is reduced to selecting one of two options. The first is to select a larger
capacity cooler to meet the useful cooling load of the sensor and the parasitic heat load imposed
by the off cooler. The second is to utilize a cryogenic thermal switch to reduce the parasitic load
from the off cooler.
There are several methods of achieving thermal switching and include technologies such as
“gas-gap” and mechanical thermal switches, diode heat pipes, loop heat pipes (LHP), and
capillary pumped loops (CPL). All of these technologies provide a high conductance when “on”
and have the capability to provide a low conductance when “off” to isolate the warm cryocooler.
The reduction in parasitic loads of all these devices versus thermal straps is on the order of 3 to 6
fold, but system designers must consider the overall increase in system complexity and decrease
in system reliability versus a simple strap.
The need for thermal switching is diminished by the use of multistage / multiload
cryocoolers. In a conceptual redundant system with these coolers, the higher temperature, more
efficient upper stages are cross-strapped with the off cooler. This allows the parasitic heat load
to be intercepted at a higher temperature and minimize the power increase to the operating cooler
and the parasitic heat load on the cooled sensor.
Heat pipes, LHPs, and CPLs are passively pumped systems. This means that the source of
pumping for the working fluid is provided by capillary action in a wick structure. These devices
have the capability for long distance, efficient cryogenic heat transport. In the case of the LHP
and the CPL, the cooling loop has an evaporator to absorb the cryogenic heat load and a
condenser to reject that heat at the cryocooler cold tip. These components are connected via
long thin walled tubes and provide excellent thermal and vibration isolation between the cooler
cold tip and the sensor. Additional capabilities that are being developed are multiple evaporator
and condenser interfaces for these loops.
Actively pumped systems utilize an ambient temperature compressor, an interface for the
cryogenic load, and a cryocooler as a precooler for the loop. This concept is very flexible for
many different system concepts, however it has an added concern for system reliability due to
the compressor. Technology is under development to address concerns for flexible cryogenic
heat transport across a two-axis gimbal and development of a hybrid system to achieve 10 Kelvin
cooling.
Figure 7. Cullimore & Ring – Swales Figure 8. Swales Aerospace 65 K Thermal

Aerospace – NASA GSCF Cryogenic Capillary Storage Unit (CRYOTSU).
Pumped Loop.
Figure 9. Technology Applications Inc. Figure 10. Swales Aerospace Cryogenic

Cryocooler Interface System. Thermal Switch.
Thermal storage and temperature stability is also a consideration for future systems.
Thermal storage is a consideration for systems that have a significant heat load duty cycle. The
thermal storage unit (TSU) is a phase change or sensible heat device that allows the cryocooler to
be sized for the average heat load of the system and not the peak. When the heat load is low, the
cryocooler reverses (for solid-liquid systems), recondenses (for liquid-gaseous systems), or
recharges (for variable temperature devices that are part of a cooling loop) the TSU. When the
load is high the cooler remains cooling at the average level and the TSU absorbs the additional
heat at constant or near constant temperature. TSUs do provide temperature stability, but future
systems may require much more stringent control of the sensor temperature.
Future cryogenic applications will be extremely system design dependent. However,
cryogenic integration technology is currently under development to increase performance over
the state of the art, significantly reduce cryogenic system penalties, and enable new system
concepts.
POTENTIAL SPACE-BASED LASER CRYOGENIC REQUIREMENTS

Cryocoolers potentially are enabling technology for future SBL space systems, which will
have significant cooling requirements for the cryogenic fuel. A range of issues are currently
being addressed including the use of cryogenic gas or liquid storage, the large volume
requirements for mission life, and the cost for on-orbit replenishment. Cryocooler integration for
large tanks will be a significant issue. High capacity cryocoolers and long term (>20 years) on
orbit propellant storage are potentially enabling technology for future High Energy Laser (HEL)
space systems, orbital transfer vehicles, and on orbit propellant depots.
A number of critical issues are currently being evaluated by the SBL program office and
contractor teams regarding the use of elevated temperatures for cryo-gas, use of multi-load
cryocoolers and reduction of storage pressure. Other significant concerns for cryogenic
applications in space based systems requiring long term cryogen storage includes substantial
cooling requirements for subcritical cryogens, cryocooler redundancy, on orbit cryogen transfer
from vehicle to vehicle, large shield cooling, long term gas and liquid cryogen storage, large
distributed cooling surfaces, cryogenic system integration, and the significant spacecraft system
penalties due to mass and input power.
AFRL has pursued low capacity cryocooler concepts including reverse Brayton cycle, single
and multiple stage Stirling cycle, advanced Joule-Thomson cycle, and Pulse Tube (Stirling cycle
variant) designs and the technology development spans a wide range of cooling temperatures
(from ~10 Kelvin to 150 Kelvin) and heat loads (up to 10 Watts at 95 Kelvin). Additionally,
AFRL has pursued advanced cryogenic integration technology including cryogenic thermal
switches, cryogenic heat transport, thermal storage, and cryogenic integration schemes to reduce
system mass and input power penalties.
Current cryogenic integration and cryocooler development programs address the negative
impacts of the cryogenic system on optical space systems: including induced line of sight
vibration, longevity, power consumption, mass, thermal transport, thermal storage, and thermal
switching. However, the cryogenic cooling requirements for future Air Force systems may
require large capacity cryogenic cooling, extremely mass and power efficient mechanical
refrigerators, and significant improvements in long term on orbit cryogen storage. The technical
efforts at AFRL concentrate on exploratory and advanced development programs that focus on
the development of technology from concept and breadboard engineering models to protoflight
models that are geared to experimental characterization and technology transition for flight
demonstrations and, potentially, operational programs.
CONCLUSIONS
The AFRL leveraged approach to cryocooler development is providing technology to
support near term requirements for SBIRS Low and other DoD programs. Cryocooler mass has
been steadily reduced, cryocooler induced line of. sight vibration has fallen below the value
allocated for other sources, and user confidence in cryocooler reliability has improved. The
cryogenic integration effort at AFRL is developing hardware to meet out-year cryogenic inte-
gration requirements. Integration developments lead to more efficient cryocooler / cryogenic
systems, which further enhance the mission capabilities of this already enabling technology.
Future systems will benefit from increased investments in multi-temperature, larger capacity
cryocoolers, and more compact, radiation hardened flight electronics. Pursuit of advanced con-
cepts such as optical cooling could produce systems with no moving parts, and use of MEMS
for improved fabrication technology can reduce weight and improve performance with the
resulting a much enhanced integrated sensor/cooler/processing packaging, improved reliabil-
ity, and enabling surveillance at very long wavelengths.
REFERENCES
1. Swift, W.L., “Single Stage Reverse Brayton Cryocooler: Performance of the Engineering Model,”
Cryocoolers 8, Plenum Press, New York (1995), pp. 499-506.
2. Burt, W.W., and Chan, C.K., “Demonstration of a High Performance 35 K Pulse Tube Cryocooler,”
3. Davis, T.M., Reilly, J., and Tomlinson, B.J., “Air Force Research Laboratory Cryocooler Technology
Development,” Cryocoolers 10, R.G. Ross, Jr., Ed., Plenum Press, New York (1999), pp. 21-32.
4. Curran, D.G., “Use of Two Stages of Cooling to Reduce Space Based Laser (SBL) Cooling Require-
ments for Both IFX and EMD Cryocooler Procurement,” Aerospace Corporation Thermal Control
Department briefing, Mar 99.
5. Orlowska, A.H., Bradshaw, T.W., Scull, S., Tomlinson, B.J., “Progress Towards the Development of
a 10K Closed Cycle Cooler for Space Use,” Cryocoolers 10, R.G. Ross, Jr., Ed., Plenum Press, New
York (1999), pp. 67-76.
6. Bugby, D., Stouffer, C., Davis, T., Tomlinson, B. J., Rich, M., Ku, J., Swanson, T., and Glaister, D.,
“Development of Advanced Cryogenic Integration Solutions,” Cryocoolers 10, R.G. Ross, Jr., Ed.,
Plenum Press, New York (1999), pp. 671-687.
Status of Programs for the DoD Family of
Linear Drive Cryogenic Coolers for
Weapon Systems
W.E. Salazar
U. S. Army Communications and Electronics Command
Research, Development, and Engineering Center
Night Vision and Electronic Sensors Directorate
Fort Belvoir, VA 22060-5806
ABSTRACT
The Standard Advanced Dewar Assembly (SADA) is the critical module in the Department of
Defense (DOD) standardization effort of second-generation thermal imaging systems. DOD has
established a family of SADAs to address high performance (SADA I), mid-to-high performance
(SADA II), and compact class (SADA III) systems. SADAs consist of the Infrared Focal Plane
Array (IRFPA), Dewar, Command & Control Electronics (C&CE), and the cryogenic coolers.
SADAs are used in weapons systems such as Comanche, the M1 Abrams tank, the M2 Bradley
fighting vehicle, and the Javelin CLU.
The linear drive cryocoolers maintain the Infrared Focal Plane Arrays (IRFPAs) at the desired
operating temperature. Stirling linear drive cryocoolers are being used in place of Stirling rotary
coolers. DOD has defined a family of tactical linear drive coolers in support of the family of
SADAs. These coolers are required to have low input power, a quick cool-down tune, low vibra-
tion output, low audible noise, and higher reliability. This paper (1) outlines the characteristics of
each cooler, (2) presents the status and results of qualification tests, and (3) presents the status and
test results of efforts to increase cryocooler reliability. Flexure-spring designs of the 0.15 watt and
1.0 watt coolers are currently in reliability growth testing.
INTRODUCTION
The US Department of Defense (DoD) has chartered a strategy to standardize second genera-
tion infrared (IR) components throughout the services. A family of second-generation (2nd Gen.)
infrared-imaging critical components called the Standard Advanced Dewar Assemblies (SADA’s)
has been developed to support this strategy. SADA I is designed to address requirements for high
performance systems, SADA II for mid-to-high performance, and SADA III for compact class
systems. SADAs consist of the Infrared Focal Plane Array (IRFPA), dewar, Command & Control
Electronics (C&CE), and the cryogenic cooler.
The US Army CECOM Night Vision and Electronics Sensors Directorate (NVESD) has de-
veloped a family of Stirling cycle linear drive coolers, shown in Fig. 1, in support of this standard-

Figure 1. Family of linear drive coolers.
ization effort.1 These coolers address the shortcomings of rotary coolers such as low reliability,
poor shelf life, multi-axes vibration & torque, excessive acoustic noise, and poor temperature sta-
bility of the detector array. This paper highlights the latest developments and results involving US
Army programs for linear drive coolers. Table 1 highlights the key parameters of the family of
coolers.
PROGRAMS FOR DoD FAMILY OF LINEAR DRIVE COOLERS 13
QUALIFICATION REQUIREMENTS FOR CRYOCOOLERS

SADAs and linear drive coolers are products that require qualification prior to first delivery.
These components are qualified once they pass through a series of tests approved by NVESD and
the procurement activity. The government or the cryocooler manufacturer may perform these tests.
The government approves all test procedures, equipment, and test facilities prior to testing. Some
of the weapon systems supported by this qualification effort include the Army’s Second Generation
Forward Looking Infrared (FLIR) Horizontal Technology Integration (HTI), Comanche, Apache,
Javelin, Improved TOW Acquisition Sensor, and Long Range Advanced Scout Sensor Suite
(LRAS3). The tests in Table 2 are part of the qualification effort for linear drive coolers.
0.15-WATT LINEAR DRIVE COOLERS

The 0.15-watt linear drive cooler was developed for second-generation FLIR man-portable
applications. The 0.15-watt cooler from DRS Infrared Technologies (formerly Texas Instruments)
was originally qualified in 1997 for use in the Javelin Command Launch Unit (CLU). Javelin is an
anti-tank missile system.
This 0.15-watt cooler was re-qualified in 1999 following an Army funded Manufacturing Tech-
nology (Mantech) program with DRS.2,3 This Mantech program was performed on both the 0.15-
watt and 1.0-watt coolers with funding from the US Army Mantech program, the Program Manager
for Night Vision Reconnaissance, Surveillance and Target Acquisition (PM-NV/RSTA), and the
Program Manager for Javelin. It developed new production techniques and processes that reduced
the cost to manufacture the 0.15-watt cooler by 30%. It focused on process improvements to the
compressor clearance seals, gas decontamination process, motor manufacturing, and cooler final
assembly. It also replaced the compressor helical spring suspension system with a flat plate, flexure
spring, suspension system. The goal of changing to a flexure springs system was to simplify motor
assembly and double the life of the coolers from 4,000 to 8000 hours MTTF. A significant cost
avoidance will be realized due to lower manufacturing costs and higher reliability coolers, and the
potential production requirement for more than 7,000 of these coolers in the next 15 years.
The 0.15-watt Javelin cooler with flexure springs is currently in full-scale production. It passed
all required Javelin detector dewar cooler assembly performance and qualification tests prior to
production.
Three Javelin coolers began life testing on February 1, 1999 at DRS. These life test coolers
have accumulated an average of 4777 relevant hours of operation through January 28, 2000. Two
units, MSN 18 and MSN 13, have 5289 hours each and the third unit has 3754 hours since being
returned to the life test. Javelin cooler numbers 18 and 13 have been failure free. The only anomaly
during the period occurred when the chamber failed to hold the prescribed temperatures during the cycle.

The 1.0-watt cooler is the focus of significant efforts and investments to qualify multiple sources,
reduce manufacturing costs, and increase their reliability. These coolers are used with SADA II,
and are critical components of many DoD programs to include the Army’s 2nd Generation FLIR
Horizontal Technology Integration (2nd Gen. FLIR HTI) program and the improved TOW acquisi-
tion system.
Status of Qualification Efforts
The DRS Infrared Technologies 1.0-watt coolers were first qualified in 1997. DRS is one of
the main suppliers of 1.0-watt coolers to the Army. AEG Infrared Modules (AIM) of Germany is
also a qualified supplier. AIM was qualified in 1998 through a Foreign Comparative Testing (FCT)
program with NVESD and the Army’s Program Manager for FLIRs (PM FLIR). The FCT program
provided funds to purchase and test several AIM coolers. Both cooler manufacturers demonstrated
reliability over the 4,000-hour MTTF requirement. DRS coolers accumulated an average of 6,486
hours and the AIM coolers accumulated 4,753 hours.
Qualification testing of Litton Life Support 1.0-watt coolers is near completion. These coolers
have passed most formal qualification tests. Only the Electromagnetic Radiation and the reliability
tests remain to be completed. The coolers have accumulated an average of 2,500 hours in reliability
testing. NVESD and PM FLIR support Litton’s formal qualification testing.
Manufacturing Technology (Mantech) Efforts
The 1.0-watt cooler was also the beneficiary of an Army Mantech effort that resulted in a 32%
decrease in cooler manufacturing costs at DRS. As mentioned before, this Mantech program was
performed on both the 0.15-watt and 1.0-watt coolers with funding from the US Army Mantech
program, the Program Manager for Night Vision Reconnaissance, Surveillance and Target Acqui-
sition (PM-NV/RSTA), and the Program Manager for Javelin. This Mantech program focused on
manufacturing process improvements to the compressor clearance seals, gas decontamination pro-
cess, regenerator/expander design, motor manufacturing, and cooler final assembly. The Mantech
effort was completed in 1998 with the completion of environmental and reliability tests. This pro-
gram established a lower cooler price threshold that is impacting the competitive procurement of
current and future procurements (12,000 coolers projected in the next 20 years). In order to maxi-
mize competition, the Mantech program included a technology transfer effort that provided DRS
reports and briefings to approved cooler manufacturers.
Reliability Improvements
The DRS 1.0-watt cooler with flexure springs is currently in qualification and life testing. This
effort is funded by the Army’s Operation and Support Cost Reduction (OSCR) Program and man-
aged by NVESD. The goal is to increase the life of the 1.0-watt cooler from 4,000 hours to 8,000
PROGRAMS FOR DoD FAMILY OF LINEAR DRIVE COOLERS 15
Figure 2. Reliability Test Profile, 1.0-watt linear cooler with external electronics.
hours Mean Time to Failure (MTTF) in order to reduce operation and support costs. The lessons
learned in the development of the DRS 0.15-watt Javelin cooler with flexure springs were applied
to the 1.0-watt cooler effort.
Figure 2 depicts the reliability test profile for 1.0-watt coolers. Six 1.0-watt coolers with flexure
springs have been built, and test results show that they match the performance of the 1.0-watt
coolers (helical springs) currently qualified and in production. Several modifications have been
necessary since testing started in 1999. One modification deals with the addition of more rugged
clamping mechanisms after the flexure spring stack shifted during the mechanical shock and me-
chanical vibration tests.
Reliability testing on three 1.0-watt coolers with flexure springs began on April 16, 1999. One
unit was removed from testing due to unacceptable vibration output levels. A defective magnet
assembly caused the unacceptable vibration output levels. This unit was replaced with two addi-
tional coolers to total four coolers in life test. These reliability test coolers have surpassed the
required 4,000 relevant hours of operation with an accumulated average count of 4,600 hours.
Testing will continue until 8,000 hours MTTF are demonstrated or until failure of the coolers.
Additional tests are required to demonstrate full conformance to qualification requirements.

The 1.75-watt cooler is designed to address the needs of the high performance second genera-
tion infrared imaging systems that will use a SADA I or other equivalent performing system whose
cooling capacity requirements or faster cool-down times cannot be met with a 1.0-watt cooler.
These coolers are currently used in many DoD programs to include the Army’s Comanche and
Apache helicopters.
16 GOVERNMENT CRYOCOOLER DEVELOPMENT AND TEST PROG RAMS
Several AIM 1.75-watt coolers were purchased and are undergoing evaluation and formal quali-
fication testing as part of a Foreign Comparative Test (FCT) program with NVESD and the Pro-
gram Manager for Comanche (PM Comanche). Several 1.75-watt coolers have been successfully
integrated into Comanche, Apache, and several other high performance FLIR systems to include
Quantum Well FLIRs. The production requirement for these coolers is estimated at over 8,000 in
the next 20 years.
FUTURE COOLER EFFORTS

Army efforts will continue to focus on reducing manufacturing costs, improving reliability,
and improving the performance of the family of linear drive Stirling cryocoolers.
NVESD, PM NV/RSTA, and PM Comanche are supporting efforts to increase the reliability of
1.0-watt coolers to 12,000 hours MTTF and the 1.75-watt cooler to 8,000 hours MTTF. Fiscal year
2001 funds are possible for these efforts. A related effort will aim to reduce the manufacturing costs
of the 1.75-watt coolers through the transfer of manufacturing processes developed under the 1.0-
watt and 0.15-watt cooler Mantech program.
With the advent of longer life coolers (8000-12,000 hours) there is a need to develop a method
to shorten the reliability testing while demonstrating cooler life. NVESD is exploring alternatives
to the current reliability test method. The current test method demonstrates cooler life by actually
running the coolers until failure. With the current method the coolers accumulate 450 hours of run-
time per month. It will take 30 months to demonstrate a 12,0000-hour cooler introduction of a valid
accelerated reliability test. The cooler is still the least reliable component in FLIR systems and
there is a need to accelerate the introduction of longer life cryocoolers into DoD systems.
SUMMARY
A family of second-generation (2nd Gen.) infrared-imaging critical components called the Stan-
dard Advanced Dewar Assemblies (SADA’s) has been developed to support a DoD standardization
strategy. A family of linear drive coolers has also been established in support of this standardization
strategy.
The US Army CECOM Night Vision Directorate in conjunction with several US Army Pro-
gram Managers has embarked in linear drive cooler efforts aimed at qualifying coolers, reducing
cooler prices, and increasing cooler reliability.
Qualification test efforts for the DRS 0.15-watt flexure spring and for the DRS & AIM 1.0-
watt coolers were successfully completed in the last five years. Qualification testing for the 1.75-
watt AIM cooler and two additional 1.0-watt cooler designs (DRS flexure springs and Litton) is
ongoing. Some of the weapon systems supported by these qualification efforts include the Army’s
Second Generation Forward Looking Infrared (FLIR) Horizontal Technology Integration (HTI),
Comanche, Apache, Javelin, Improved TOW Acquisition Sensor, and Long Range Advanced Scout
Sensor Suite (LRAS3).
The US Army Manufacturing Technology program and the Operation and Support Cost Re-
duction program were successful in reducing manufacturing costs and in implementing flexure
spring designs aimed at increasing the reliability of the 0.15-watt and 1.0-watt coolers. Additional
work is planned to transfer these improvements to the 1.75-watt coolers and to further increase the
reliability of 1.0-watt coolers.
REFERENCES
1. J. Shaffer and H. Dunmire, “The DOD Family of Linear Drive Coolers for Weapons Systems,”
2. Raytheon Texas Instruments Systems, “Linear Drive Cooler Mantech Program,” Contract DAAB07-
95-C-J513 Industry Review, December 1997.
3. DRS Infrared Technologies, “Linear Drive Cooler Mantech Program,” Contract DAAB07-95-C-
J513 Industry Review, February 1999.
Air Force Research Laboratory Cryocooler
Characterization and Endurance Update
B. J. Tomlinson1, C. H. Yoneshige1, and N. S. Abhyankar2

1
Air Force Research Laboratory
Kirtland AFB, NM, USA 87117-5776
2
Dynacs Engineering
ABSTRACT
The Air Force Research Laboratory (AFRL) has been instrumental in advancing space
cryocooler technology through cryocooler development and characterization of the. long-term
performance of different types of cryocoolers. These coolers were developed to support the
long life space mission requirements of the United States Air Force SBIRS-Low Program Of-
fice, the Ballistic Missile Defense Organization (BMDO), the National Aeronautics and Space
Administration and the Department of Defense. Long life cryocooler applications include cool-
ing infrared sensors, focal planes, optics and electronic circuits for various space missions of
national interest.
The main objective of this paper is to present the status of the cryocoolers currently under-
going characterization and endurance evaluation at AFRL. The information gained through
these processes is shared with industry partners, cryocooler developers, technology sponsors,
and technology users. This feedback is essential for cryocooler design enhancements and fu-
ture cryogenic technology development efforts. There are two cryocoolers undergoing charac-
terization at AFRL. These include the Astrium (formerly Matra Marconi Space) 10 K Cryo-
cooler, and the TRW 150 K miniature pulse tube (150K MPT). In addition, there are five
cryocoolers undergoing endurance evaluation at AFRL These include the Raytheon Protoflight
Spacecraft Cryocooler, Raytheon Standard Spacecraft Cryocooler (SSC) II, TRW 3585, TRW
6020, and the Defense Evaluation and Research Agency (DERA) cryocooler. This paper in-
cludes each cryocooler’s status, and update or initial report on its performance, elapsed run-
time hours, performance anomalies and updated characterization and endurance evaluation
data.
INTRODUCTION
The most critical characteristics of cryocoolers for strategic space applications are life-
time and reliability. This is what distinguishes them from their short-term tactical cousins. The
primary purpose of the Air Force Research Laboratory Cryogenic Cooling Research Facility
(CCRF) is to explore the thermodynamic performance characteristics of one-of-a-kind or first-
of-a-kind engineering design model or protoflight space cryocoolers and assess their lifetime and

reliability to help meet technology needs for the Air Force, the Ballistic Missile Defense
Organization (BMDO), and Department of Defense (DoD) strategic space cryogenic cooling
applications. Cryocoolers are enabling technology for cryogenic optical systems and infrared
sensors.
During characterization, each cryocooler undergoes a rigorous series of experiments, which
determine its thermodynamic performance capabilities. The effects that sensor duty cycling,
changes in heat rejection temperature, heat load, and input power have on the cryocooler’s
performance are investigated. During the endurance evaluation, the cryocooler is constantly run
at its nominal operating parameters until it meets pre-determined failure criteria. One of the
major issues with establishing confidence in the reliability of cryocooler technology is the
cryocoolers’ 10+ year design life and the absence of accepted accelerated testing methods.
The CCRF is equipped with one 6 foot, two 36 inch, eight 24 inch and two 28 inch thermal
vacuum chambers. Each of these chambers is capable of high vacuum (~10-7 torr), which helps
closely simulate a space environment, and has independently controlled conductive heat rejection
surfaces to simulate a potential spacecraft thermal interface. Each chamber also has multiple
feed-throughs for chiller fluid lines, cryocooler control electronics, data acquisition
instrumentation, and computerized environmental controls. Cryocoolers can also be
characterized in a table-top configuration, especially cryocoolers that are not qualifiable for
space flight, but have significant heritage to current or in development space flight cryocoolers.
CHARACTERIZATION
Characterization consists of two major portions. The first part of the characterization
process is the initial baseline / acceptance evaluation. This evaluation is done to ensure that the
cryocooler meets the manufacturer’s performance specifications and verify the “as delivered”
performance prior to continuing the experiment. Before the cooler even reaches the laboratory,
AFRL engineers have spent time with the contractor to gather vital data necessary to design the
experimental test stand and prepare an adequate plan for characterization. The second part of the
process is the actual characterization of the cryocooler. AFRL engineers and technicians
perform experiments with the cooler to fully understand and map its thermodynamic
performance and examine the complex interrelationships between the different operating
parameters. During this process, it is possible to concentrate on various areas of interest where
the technology under examination could potentially be applied. This is part of the development
life cycle for cryocoolers and requires inputs from technology users, government, and system
integrators.
Initial Baseline Evaluation
When a cryocooler arrives at AFRL, it normally undergoes an initial baseline/acceptance
evaluation. This evaluation consists of refrigeration performance baselines, nominal load lines,
low temperature stability trials, cool-down to lowest temperature trials (no heat load applied) and
stiction tests. This phase of characterization also serves as a screening method to identify any
manufacturing defects or shipping and handling damage. There are two cryocoolers currently in
this phase of characterization. They are the TRW 150K miniature pulse tube (MPT) and the
Matra Marconi Space (MMS) 10K Cryocooler.
TRW 150K MPT. The TRW MPT is designed to lift 1W @ 150K at various heat rejection
temperatures. The AFRL MPT #006 was provided to NASA in exchange for MPT#002. The
#006 has been qualified and integrated into the NASA Sounding of the Atmosphere using
Broadband Emission Radiometry (SABER) experiment. MPT #002 is currently at AFRL and is
being integrated with a 6-axis dynamometer for a functional test and an induced vibration trial.
The cooler will then be moved to a 24” thermal vacuum chamber for an initial baseline
evaluation and characterization.
Astrium 10K. The Astrium 10K cryocooler (formerly Matra Marconi Space) is designed to
lift 45 mW at 10.4 K. It was delivered to AFRL in early May 00 and was set up on a table top
for a functional test. This cryocooler will be integrated with a 36” thermal vacuum chamber for
AFRL COOLER CHARACTERIZATION AND ENDURANCE UPDATE 19
Figure 1. Astrium 10K cryocooler.
its initial baseline evaluation and characterization. Figure 1 shows the Astrium 10K cryocooler
in its table-top configuration.
Performance Characterization
In order to fully understand the cryocooler technology developed for the Air Force and DoD,
the CCRF conducts experiments, which are meant to explore the full thermodynamic range of
performance for each cryocooler. The AFRL objectives for characterization are as follows:
1. Characterize the full thermodynamic performance envelope of emerging cryocooler

technologies to establish parametric performance models.
2. Feedback performance data, models, and lessons learned to parties involved in the
cryocooler development process.
During the characterization phase, the cryocooler goes through a series of experiments
designed to characterize its thermodynamic idiosyncrasies versus predicted performance as well
as its potential to be integrated with a spacecraft cryogenic system. Load lines, refrigeration
performance baselines, temperature stability experiments, transient heat rejection thermal
response, cool-down curves, and off-state conduction (determination of the parasitic heat load)
are normally done during this phase of characterization. This data is fed into parametric
performance models to provide spacecraft systems integrators tools to model the cryocooler
performance versus specific mission requirements.
AFRL is expecting the re-delivery of the Ball Aerospace 35/60K cryocooler. This
cryocooler was designed to lift 0.4W @ 35K and 0.6W @ 60K. It is a split-Stirling cryocooler
with 3 stages and has the ability to lift 2 heat loads at 2 different temperatures. The initial
baseline/acceptance evaluation was performed to evaluate its nominal performance. The

cryocooler was returned to Ball for retrofit of the expander cold head due to a design flaw
discovered on the NASA Ball 30K cryocooler. The cooler is currently in the final stages of re-
assembly and should be delivered to AFRL in July 00.
ENDURANCE EVALUATION
AFRL objectives for endurance evaluation have evolved to meet the following requirements
for demonstration of the lifetime and reliability of cryocooler technology developed for the Air
Force, BMDO and the DoD:
1. The focus of the long life endurance evaluation of cryocooler technologies is on the
characterization of the long term performance degradation, system reliability, reliability
contributors and detractors, an accurate assessment of the lifetime of the technology, and
the development of accepted methods for the accelerated testing of long life space
cryocoolers.
2. AFRL provides feedback to cryogenic technology developers, users, and spacecraft
developers in the form of technical reports and conference presentations of data and
lessons learned in order to aid follow-on development efforts and add to the body of
reliability data on cryocooler long term performance.
The endurance evaluation is normally run until the cryocooler meets predetermined failure
criteria. If the cryocooler continues to perform nominally past its design lifetime requirement, it
will be allowed to run until it does meet the failure requirements. Even though endurance data is
valuable for understanding the life and reliability characteristics of a cryocooler, an endurance
evaluation that lasts for more than 5 years will not meet technology insertion freeze dates for
critical Air Force and DoD programs or provide the necessary data to impact follow-on
technology development programs of similar heritage. Thus, AFRL is working on developing
accepted accelerated testing methods to help increase confidence in the reliability of emerging
technologies, while still meeting technology insertion freeze dates.
There are five cryocoolers currently undergoing endurance evaluation at AFRL. Table 1
shows these cryocoolers and their nominal operating conditions. These operating conditions are
maintained throughout the endurance evaluation, except for periodic load line checks, which are
done to track any performance drift of the cryocooler. Each of the cryocoolers’ heat rejection
temperatures is cycled above and below the nominal heat rejection temperature listed in Table 1.
This allows engineers to monitor steady state performance over the design thermal rejection
temperature range for each cryocooler. It also allows AFRL engineers to perform baseline
evaluations at different heat rejection temperatures.
For comparison purposes, 300K is defined as the nominal heat rejection temperature for all
cryocoolers. The rejection temperature range is defined by the cryocooler’s design margins
based on an intended orbital transient temperature profile, it’s sensitivity to coefficient of thermal
expansion effects and thermodynamic performance limits.
The endurance evaluation can be run in a thermal vacuum chamber or on a table-top.
Coolers are run on table top if they are not space flight qualifiable, but have a significant heritage
to existing or in development space cryocoolers. AFRL currently has a total of 5 cryocoolers
undergoing an endurance evaluation. Table 2 shows a summary of the coolers in endurance.

There are two cryocoolers running in a table-top experiment set-up. They are the Defense
Evaluation and Research Agency (DERA) cryocooler and the TRW 3585 pulse tube cryocooler.
DERA Cryocooler. The DERA cryocooler is a miniature split-Stirling cooler designed to
lift 0.25W @ 65K. It was developed under sponsorship from BMDO to examine the
development of cryocoolers for Infrared Focal Plane Array (FPA) applications at wavelengths
from ~3µm to 15µm with cooling to 65K and below. The lifetime of this cryocooler was
estimated to be about 20,000 hours. The cryocooler design incorporates the use of flat section
springs to support the compressor pistons. The compressor unit contains two pistons set back-to-
back, which work on a common compression space. The free displacer is balanced using a
balance mass opposed to the displacer piston. Sliding plastic seals are used inside the
cryocooler. The plastic seals ensure no metal-to-metal rubbing contact along all running
clearances.
TRW 3585. The TRW 3585 pulse tube cryocooler has accumulated over 19,000 hours. It is
designed to lift 0.85W @ 35K. This is an engineering design model single stage orifice pulse
tube with a 20cc compressor. This cooler was the first of three units developed under the AFRL
TRW 35K Pulse Tube Cryocooler program. This cooler has limited follow-on applications due
Figure 2. TRW 6020 load line comparison between JPL and AFRL (Nov 1998 data, rejection
temperature 293 Kelvin).
to evolutionary changes in the compressor and cold head design, but is being evaluated because
it has significantly contributed to the family heritage of the TRW/Oxford Stirling compressor.
Due to an error discovered in the calibration of the capacitance sensor electronics, this
cryocooler is currently being re-baselined for comparison to data taken at the Jet Propulsion
Laboratory in 1993.
AFRL has three cryocoolers undergoing an endurance evaluation in a thermal vacuum
chamber. They are the Raytheon (formerly Hughes) Standard Spacecraft Cryocooler (SSC) II,
the Raytheon Protoflight Spacecraft Cryocooler (PSC) and the TRW 6020 pulse tube cryocooler.
Raytheon SSC II. The SSC II is a split-Stirling cryocooler designed to lift 2W @ 65K with
a specific power of 30 W/W. During its initial baseline/acceptance evaluation, the SSC II
specific power exceeded its design specifications and the cooler developed hardware problems.
The cooler was returned to Raytheon for refurbishing and sent back to AFRL. A short mini-
characterization was completed for design and “as delivered performance” verification. During
the mini-characterization, the cooler was not able to maintain 2W @ 65K without occasional
tripping possibly due to compressor warm-up at 300K rejection temperature. The nominal
operation point was then reset to 1.5W @ 65K in safer mode. The cooler has been running in
endurance at this new operation point since January 2000.
Raytheon PSC. The Raytheon PSC is a split-Stirling cryocooler designed to lift 2W @
60K. This cryocooler is the latest design from Raytheon and is directly related to the older
Raytheon SSC II currently under endurance at AFRL. During its initial baseline/acceptance
evaluation, the cooler demonstrated the ability to efficiently lift 1.2W @ 35K.
After characterization, the PSC entered its endurance evaluation, but was shut down
temporarily due to an electronics problem in the control rack. With the problem was solved, the
cooler was moved to a 24” thermal vacuum chamber and is continuing with endurance
evaluation.
TRW 6020. The TRW 6020 pulse tube cryocooler is a single stage orifice pulse tube with a
10cc dual opposed compressor designed to lift 2W @ 60K. It was integrated into a 24” thermal
vacuum chamber for characterization and long term endurance evaluation. Its heat rejection
temperature is set by a copper block interface to a computer controlled chilled fluid loop.
Apparent discrepancies in the characterization data led AFRL to conduct a detailed review
of all the data on this cryocooler and also examine similar data discrepancies with the TRW 3585
and the TRW 3503 pulse tubes. The first indication of discrepancies in the data was in the form
of a distinct difference in performance compared to data gathered during a characterization
performed in 1994-5 at the NASA Jet Propulsion Laboratory. Additionally, data taken at AFRL
indicates a possible shift in parasitic load on the cold block over time. Figure 2 shows the
apparent performance shift based on data taken in November 1998. The assumptions were that
the parasitic load on the cold end was similar to the JPL set-up and that the stroke conversion
values as provided by TRW were correct.
An investigation into the apparent differences in performance revealed that the conversion
factors used to determine the stroke length on compressors A and B from the capacitance sensor
box readings were incorrect. The conversion factors used during the November 1998 data
comparison were the factors that TRW provided in the operation manual for the cryocooler.
However, when the 6020 was undergoing its initial characterization at JPL, the 6020 had to be
returned to TRW to stake the capacitance sensor cap internal to the cooler. During this time,
TRW recalibrated the capacitance sensor electronics and provided these values to JPL. When
the cooler arrived at AFRL it was assumed that the conversion values were the ones listed in the
operations manual. These conversion factors for the capacitance sensor readings were put into
the data acquisition software and contributed to the fact that the cryocooler appeared as if it was
experiencing degradation in performance.
Stiction tests completed recently at AFRL showed that the conversion factors being used
were indeed not correct and different from the conversion factors used by JPL. Figures 3 and 4
show traces from stiction tests completed in May and June 2000 at AFRL. New conversion
factors were calculated using this “soft” calibration and the LabVIEW™ data acquisition
software was updated.
Figure 3. Stiction Trace from Compressor 1. Figure 4. Stiction Trace from Compressor 2.
The JPL load lines were reaccomplished at AFRL, and the similarity in the data shows no
significant degradation in performance. Figure 5 shows the load line comparison for data taken
in May 2000 and Figure 6 is a multivariable plot that includes AFRL and JPL data. Figure 7 is
a plot of AFRL data that attempted to match the temperatures and heat load of a JPL 293K,
11mm load line. The plot shows that the input power is nearly identical, but the stroke needed
to achieve the JPL performance was 0.23 mm longer on average. However, this brings up
another potential source of error in that the JPL stroke data on average differed 0.15 mm from
compressors A to B. The AFRL data only varied 0.06 mm from side to side. Apparently, the
Figure 5. May 2000 TRW 6020 load line comparison between JPL and AFRL data (rejection
temperature 293 Kelvin).
Figure 6. Multivariable Plot for May 2000 TRW 6020 JPL and AFRL data (rejection temperature
293 Kelvin).
Figure 7. June 2000 TRW 6020 load line comparison between JPL and AFRL data.
output of each channel of the amplifier has a distinct impact on the actual stroke for each
compressor. Differences seen in the plots are most likely accounted for by instrumentation and
measurement errors, but it is apparent that the 6020 performance is very close to the data that
was taken at JPL almost five years ago. A complete error analysis is still being constructed.
An additional source of error that contributed to the apparent difference in cryocooler
performance from JPL to AFRL in the fact that the parasitic load on the cold end was signifi-
cantly larger at AFRL in November 1998. The JPL quoted measurement was acquired utilizing
a cryocooler intercept to the ”off’ 6020. The method at AFRL is a parasitic warm-up test that
allows the measurement of an effective warm up parasitic on the cold block. JPL had values of
-0.5 W for a parasitic load compared to AFRL which had a high of 1.4 W in Jan 00 (Table 3).
It was determined that the multilayer insulation (MLI) wrap, the heater, and the sensor leads
were inadequate and were the source of the large discrepancy in parasitic load.
To alleviate the parasitic load on the 6020, AFRL engineers and technicians designed a
new MLI blanket and wrapping scheme, the instrument leads were replaced with very thin
wires to minimize the parasitic heat conduction, and the oversized cryogenic heater was re-
moved and replaced with a 5W power resistor (JPL used 0.25 W wire-wound ceramic resis-
tors). This effort culminated in the reduction of the effective parasitic to 0.6 W.
Based on these findings it is apparent that a stiction-calibration of the capacitance sensor
electronics and a parasitic warm-up test are needed components of baseline performance checks
during endurance evaluation. The TRW 6020 endurance evaluation plan has been updated to
include these procedures and similar calibrations and parasitic load mitigation efforts are un-
derway for the TRW 3585 cryocooler.
SUMMARY
The AFRL Cryogenic Cooling Research Facility continues to characterize and evaluate
the long life performance of emerging technology for strategic space cryocoolers for the Air
Force, BMDO, and the DoD. Design and execution of these experiments are critical to ensure
accurate examination of the technology. Long life endurance evaluation is a crucial compo-
nent for system integrators and technology users to determine life and reliability heritage.
ACKNOWLKDGMENTS
This work was a team effort and included contributions from Mr. John Kallman (AFRL),
Mr. George Lybarger (AFRL), and Mr. Michael Martin (Dynacs Engineering).
Air Force Research Laboratory Cryocooler
Reliability Initiatives
S. Blankenship and T. Lynn Fountain

Georgia Tech Research Institute
Atlanta, GA
T. M. Davis and B. J. Tomlinson
Kirtland AFB, NM 87117
ABSTRACT
The primary concern of spacecraft developers when considering active refrigeration for space
missions is lifetime and reliability. Lack of confidence in current cryocooler technology to achieve
mission performance goals and achieve the necessary lifetime is a deterrent that often precludes the
consideration of cryocooler technology for many space applications.
The Air Force Research Laboratory (AFRL), through the Georgia Tech Institute of Technol-
ogy, has conducted two workshops with government and industry developers and technology users
to address cryocooler reliability. These workshops highlighted critical issues associated with me-
chanical, electronic, and software reliability for cryocooler systems. Also addressed were issues on
contamination control, performance testing, acceptance testing, manufacturing in-process testing,
environmental qualification screening, endurance evaluation and demonstration, and producibility
issues.
In addition, AFRL has pursued research and development efforts to augment cryocooler tech-
nology development with the aim of increasing reliability. Efforts such as vibration mitigation
through improved compressor design, improved flexure bearings, reduced gas contamination, and
potential accelerated testing approaches have been explored.
AFRL in-house laboratory characterization and endurance evaluation of cryocooler technol-
ogy has also contributed to the lifetime and reliability database for emerging technologies. This
information is invaluable as a tool for spacecraft developers to understand the lifetime and reliabil-
ity potential of a candidate cryocooler technology and establish a heritage database for cryocooler
families. Additionally, AFRL has made direct contributions to improvements in follow-on designs
through the feedback of performance and endurance data to industry developers.
INTRODUCTION
Lifetime and reliability are driving concerns for the use of active cryogenic cooling technology
in space. Military, commercial, and scientific applications have driven the requirements for the
development of long life (10+ years), high reliability cryocoolers for three decades. The scope of
development issues for active refrigeration includes the mechanical unit itself, the power condi-

tioning and control electronics, and the software utilized for cryocooler operation. Recent develop-
ments in the state of the art have vastly improved the current generation of cryocooler technology,
but significant issues remain and chiefly center around the reliability of the devices utilized for long
life mission applications.
The Air Force Research Laboratory (AFRL), Space Vehicles Directorate, Cryogenic Technol-
ogy Group has pursued research and development efforts to develop cryocooler technology to meet
lifetime, reliability, and thermodynamic performance to meet Air Force and Department of De-
fence mission requirements for over 15 years. Recently completed AFRL advanced technology
development programs such as the Raytheon Protoflight Spacecraft Cryocooler, the Ball Aero-
space 35/60K Cryocooler, and the Astrium (formerly Matra Marconi Space) 10K Cryocooler have
made an impact on the state of the art for mechanical refrigeration. Additionally, current ongoing
development programs such as the High Efficiency 95K Cryocooler programs with TRW and Ray-
theon, and the Small Business Innovative Research programs to develop cryocooler and cryogenic
integration technology are making contributions to pushing the state of the art. Through such
research efforts as vibration mitigation through improved compressor design with Oxford Univer-
sity, improved flexure bearing design with the Aerospace Corporation, and potential accelerated
testing approaches with the Ukrainian Institute for Low Temperature Physics many aspects for
basic improvement of emerging cryocooler technology have been explored.
Quantifying the lifetime and reliability of long life cryocooler technology is elusive. Many of
the mechanical refrigerators that have been developed or are under development are usually unique
or have very low production numbers. Additionally, designs mature and evolve from cooler to
cooler to accommodate new improvements or to meet customer specifications. These changes
effect the design heritage and any prediction of cryocooler reliability. A large unknown in the
useful lifetime prediction for cryocooler performance is the long-term degradation components that
are observed only over thousands of hours of operation.
As a necessary component to understanding the problems associated with developmental cryo-
cooler technology, AFRL in-house laboratory characterization and endurance evaluation of engi-
neering design model and first-of-a-kind protoflight cryocoolers has a significant role in the life-
time and reliability database for emerging technologies. This information is invaluable as a tool for
spacecraft developers to understand the lifetime and reliability potential of a candidate cryocooler
technology and establish a heritage database for cryocooler families. AFRL has also made direct
contributions to improvements in follow-on designs through the feedback of performance and en-
durance data to industry developers.
THE CRYOCOOLER RELIABILITY WORKSHOPS

The Air Force Research Laboratory Space Vehicles Directorate (AFRL/VS) at Kirtland Air
Force Base and the Space Technology Advanced Research Center (STAR) at Georgia Institute of
Technology in Atlanta sponsored a series of workshops on the reliability of cryocoolers for space
applications. The First Cryocooler Reliability Workshop was held on 17 – 18 September 1998 in
Albuquerque, New Mexico, and the Second Cryocooler Reliability Workshop was held in Manhat-
tan Beach, California on 23 - 24 August 1999.
These workshops are a mechanism for involving the cryocooler community in developing
general approaches to reliability problems. In the workshops, a subset of experts met in private for
two days to identify and prioritize cryocooler reliability issues and to recommend approaches to
resolve the issues. By concentrating on the technical questions in a cooperative atmosphere with a
breadth of viewpoints, new insights and solutions were encouraged.
Participation in the workshops was determined by invitation. The two workshops drew par-
ticipants with knowledge and interest in cryocooler reliability from government, industry and
academia. Civilian as well as military interests were well represented. The first workshop drew 25
attendees and the second 35. (A list of the organizations represented at the two workshops is
contained in Table 1.) The diversity in participants was very important to the goals of the work-
shops.
AFRL CRYOCOOLER RELIABILITY INITIATIVES 29
The workshop process began with introductory, plenary sessions to define the scope of the
workshop and to establish participant interest areas. The workshops then broke into two or three
smaller parallel sessions. Each of these was charged to review a particular subset of the problem, to
prioritize issues, and to propose suggested approaches to the issues. The last afternoon of the
workshops was devoted debriefs by the leaders of the parallel sessions. Abbreviated workshop
agendas are shown below in Tables 2 and 3.
WORKSHOP RESULTS
The First Workshop concentrated solely on mechanical reliability of active refrigeration. As
one of the recommendations from that meeting, the Second Workshop expanded the scope to in-
clude electronic and software reliability as well. A summary of each workshop including the ses-
sion debriefs was produced on CD-ROM along with the available presentations. Information is
also available at http://www.star.gatech.edu under Workshops.
Cryocooler Mechanical Reliability Issues

The most significant cryocooler mechanical reliability issue from both workshops was con-
cerned with the perceptions of the acquisition and user communities regarding cryocooler maturity
and suitability for space flight. Several presentations in the plenary session of both meetings ad-
dressed this issue, and the parallel sessions revisited the question in several forms. The majority of
system designers and users have not recognized significant improvements in cryocooler mechani-
cal reliability.
Extensive experimentation and performance evaluation is the proposed mechanism for dispel-
ling ignorance and doubt. The workshops spent significant energy in considering how to character-
ize cryocoolers, how to capture lessons learned from experiments and qualification testing, and
how to distribute the results widely.
Much of current experimentation and testing will characterize known mechanical failure modes
including failures from contamination, wear, leakage, mechanical failure, and other sources. To
continue improvements in mechanical reliability, the unknown failure modes are more important to
identify. While it is possible to design for and screen for known failure mechanisms, an adequate
screening for the unknowns must also be devised. Both cryocooler system and component level
screening are necessary.
In addition to the “screening” type of testing, such as environmental qualification testing, long-
term endurance evaluation plays a major role in identifying and tracking long time constant perfor-
mance degradation. Endurance evaluation is difficult due to the unavailability of accepted acceler-
ated testing methods that allow the cooler to be operated in a fashion that provides reliability data
for long term operation in a greatly reduced time period. However, the nature of current cryocooler
technology apparently prohibits this type of test. Cryocoolers must be run over thousands of
hours, usually in thermal vacuum chambers with sophisticated instrumentation, to track per-
formance and provide user confidence in the design. Continuing to provide for this type of
performance evaluation is essential to continue to refine the reliability database for space cryo-
cooler technology.
To verify units and designs, there is a need for full space qualification units. For the low
number of builds involved in most cryocooler efforts, the cost of independent qualification units is
usually prohibitive. With the expected move to manufacturing lines for cryocoolers, the need to
develop qualification unit test standards and procedures will become vital.
Two summary recommendations concerning mechanical reliability resulted from the work-
shops:
(1) Develop good mechanical design principles and practices
Good design principles and practices should be documented, controllable, and repeatable. Some
examples are the TRW High Efficiency design for manufacturing compressors and a paper on
contamination lessons learned presented at the Second Workshop. Properly implemented, these
standards would result in such features as larger, more predictable design margins, less unit-to-unit
performance variation, and streamlined process improvement.
(2) Continue testing and experimentation programs
Identifying new failure modes and characterizing reliability requires continued, well constructed
experimental and qualification testing programs. A continuation of long-term endurance evalua-
tion on coolers is essential to these goals. Cryocooler system and component level screening test-
ing need to be expanded.
Cryocooler Electronic Reliability Issues

Most electronic reliability issues for spacecraft are not specific to cryocoolers. The major
cryocooler-specific electronic reliability issue is the reliability of the controller. The specific issues
that were identified include the lack of radiation-hardened parts for some functions, the reliability
of sensors, the relatively large part count, the effects of electromagnetic interference, and the com-
plexity of the control objectives. Approaches to resolving these issues were identified from several
alternatives. Since the community is moving to second generation designs, the timing is right for
the implementation of some of these recommendations:
(1) Develop standardization across design teams
The community needs to move toward standardization and integration. The standards must
maintain design flexibility to meet unique system requirements, so the question of the acceptable
level of standardization must be faced. Once defined, the effectiveness of standard building blocks
for circuit and package integration needs to be confirmed. Feedback with system and mechanical
designers must be included in the process. The presentation by Jim Lyke at the Second Workshop
included a program with many of the desired properties.
(2) Define a standard digital data interface for cryocoolers
By identifying potential common core standards defined by digital inputs/outputs, there should
be reliability benefits for all subsystems, not just the electronic controller.
(3) Develop a cryocooler-unique testing standard for controllers
There are no data specifically concerned with the reliability of cryocooler controllers. A test-
ing standard and a testing program are needed to characterize cryocooler electronics reliability.
Cryocooler Software Reliability Issues
The important cryocooler software Issues sorted into two areas: organizational and technical
issues. The organizational issues are concerned with how computer scientists/software engineers
are integrated into the cryocooler team. Generally, the software community is not directly a part of
the mechanical or electronics communities. Consequently, the software developers usually lack
enough operating experience to know the level of functionality required, they have very little expe-
rience with space cryocoolers, and builds are so infrequent that there is little continuity of experi-
enced personnel and a loss of lessons learned.
The technical issues that arise are partly the result of the lack of experience by the software
developers. The requirements are often overly complex, there is not enough margin in the require-
ments or the design, and the differences between an initial design with commercial parts and a final
design with radiation-hardened parts are often overlooked.
The inclusion of software reliability in the Second Workshop was important to initiate a dis-
cussion of this issue. The discussion was informed by personal experiences of cryocooler develop-
ers, but did not include the perspective of software developers, particularly ones who specialize in
software reliability. An understanding of software failure was implicit in the comments, but a
definition of software reliability did not emerge. The most detailed recommendations concerned
methods to avoid complexity in software as a means of improving software reliability, without a
consideration of the potential effects on overall cooler effectiveness and reliability.
The discussions led to suggestions that are collected below under three major headings:
(1) Integrate software developers into cryocooler development programs
Integrating software workers more deeply into cryocooler development programs could re-
solve some of the issues. Keeping a record of lessons learned in cryocooler software development
would aid in getting new workers up to speed. These would explicitly include reference to margins
in computing capability.
(2) Minimize the need for software
Reducing the dependence on software could contribute to resolving all software problems,
including reliability issues. This might not always be possible, but some activities that are implica-
tions of this approach were explored in the Second Workshop.
(3) Establish software testing benchmarks and standards

A set of cryocooler software testing methods and standards would enable the community to
track progress and predict reliability. This approach will require software reliability expertise gen-
erally lacking in the cryocooler community.
General Workshop Recommendations

The perspective of the cryocooler community, as represented in the workshops, is very posi-
tive. Advances in the last decade have raised cryocooler mechanical reliability to the levels re-
quired for spacecraft application, to the point of making electronic and software reliability as much
the issue as mechanical reliability. The cryocooler community has resolved a difficult mechanical
research and development issue and is eager to see the results applied on spacecraft. Several flight
programs have baselined cryocooler technology and over the next several years reliability predic-
tions, system integration issues, on orbit problems (or lack thereof), and improvements in the state
of the art of technology will affect research and development and quantification of cryocooler
reliability.
Consequently, two major needs were identified by the workshops. The first is to demonstrate
and to advertise the improved mechanical reliability by developing and distributing supporting test
data from ongoing performance characterization and endurance evaluation activities. The second
is to attack the remaining reliability issues, not only in the mechanical systems, but also in electron-
ics and software. The general recommendations of the workshops, summarized in Table 4 as action
items for specific areas of the community, were intended to address these two major needs.
SUMMARY
The Cryocooler Reliability Workshops have proved to be very useful for determining criti-
cal issues in space cryocooler reliability. Expanded awareness of these issues from the technology
developer and user communities will allow more efficient development and technology investment
strategies across industry and government. Comprehensive use of active cryogenic refrigeration in
space through a growing number of operational flight programs is pushing the need for defined and
quantifiable cryocooler reliability. It is recognized that reliability must be pursued as an integral
part of cryocooler development.
ACKNOWLEDGMENTS
The authors would like to acknowledge the significant and essential contributions of the
workshop participants, the briefing presenters, and the facilitators toward making these workshops
a success.
Protoflight Spacecraft Cryocooler
Performance Results
Kenneth Price1, John Reilly2, Nandu Abhyankar2, Ben Tomlinson2

1
Raytheon Systems Company
El Segundo, CA, 90245, USA
2
Albuquerque, NM, 87117, USA
ABSTRACT
The Protoflight Spacecraft Cryocooler (PSC) is a flight-qualified Stirling cryocooler that deliv-
ers 1.2 W refrigeration at 35 K and 3 W refrigeration at 60 K. This Oxford-class unit employs three
finger tangential flexures in the compressor module. These flexures have been shown to provide
smoother, lower vibration piston motion than previously obtained with conventional three finger
spiral flexures. Acceptance Tests validated the required performance capabilities and Qualification
Tests validated the cooler for flight. Acceptance Tests included performance mapping at rejection
temperatures from 275 K to 325 K, residual vibration measurements for each module, temperature
stability under various conditions, cold tip motion, cold tip side load capacity. Qualification Tests
included three-axis random vibration, thermal cycling, hermeticity and EMI/EMC tests. The 12.5
kg cooler was delivered with a brassboard command and control module that provides simulta-
neous temperature and Adaptive Feed-Forward vibration control. The PSC with brassboard elec-
tronics have been delivered to the Air Force Research Lab in Albuquerque, where additional testing
has validated the results obtained at Raytheon.
INTRODUCTION
Air Force Research Laboratory awarded the PSC Program1 to Raytheon Company to build,
test, and deliver a flight qualified Thermo-Mechanical Unit (TMU) and a brassboard Electronic
Control /Power Conditioner (EC/PC). Objectives included improving thermodynamic performance,
reducing weight, and upgrading hardware to flight quality in comparison to the previously built
65K Standard Spacecraft Cryocooler Engineering Model. The PSC system has been successfully
assembled, tested and delivered. Acceptance Tests performed at Raytheon and subsequently re-
peated and expanded at the AFRL and JPL demonstrate that the PSC TMU performance require-
ments have been met or exceeded in all categories.
In particular, outstanding performance has been achieved over the 25K to 120K operating range.
Up to 3W refrigeration can be delivered at 60K and up to 1.2W at 35K. The efficiency of the cooler
at 35K is a remarkable 85.9W/W, measured at the motor inputs.
The large data base now established by AFRL, Raytheon, and JPL will provide a definitive
benchmark against which to measure the performance of the PSC during life testing planned to start

36 SPACE STIRLING CRYOCOOLER DEVELOPMENTS
Figure 1. Protoflight Spacecraft Cryocooler.
at AFRL in late 1999. Additional cooler characterization will be performed by AFRL prior to life
test. Emphasis will be placed on the PSC’s ability to continue operating efficiently over the full
rejection temperature range.
CRYOCOOLER SYSTEM DESCRIPTION

The PSC TMU is a single stage cryocooler housed in two modules connected by a compliant
steel transfer line. See Figure 1. Refrigeration is produced at the tip of the cold cylinder, where the
user’s thermal interface is a cylindrical copper projection. Waste heat is removed from the ex-
pander module at a flange located at the base of the cold cylinder. Compressor waste heat is re-
moved from a flange at the center of the housing.
The cryocooler weight requirement is specified as the combination of physical weight plus a
weight penalty† proportional to power consumption. Penalty weight is defined as 0.3 kg/W times
the motor power drawn by the TMU when supporting 2W at 60K and rejecting heat to 300K. This
combination is required to be less than 33.0 kg.
The PSC compressor module weighs 7.0 kg and the expander, 5.5 kg for a TMU physical
weight of 12.5 kg. This is a 30% reduction compared to Raytheon’s predecessor programs such as
the 65K SSC.2 Since the PSC consumes 57W at the specified operating point, the weight penalty is
17.1 kg. Therefore, the combined physical and penalty weight is only 29.6 kg, which are 3.4 kg
(10.3%) below the allowed maximum.
Each module is internally balanced for optimal control of residual vibrations. The compressor
employs two pistons working in opposition against a common compression chamber located at the
center of the module. The expander employs a single displacer piston to produce refrigeration. A
matching mass driven in opposition dynamically balances the displacer piston. Residual vibrations
produced by each module are corrected by the electronics via an Adaptive Feed Forward (AFF)
algorithm operating from feedback obtained by monitoring the three load cells located at each
module’s three symmetric mounting points.
The PSC TMU design retains significant design heritage to the previously built 65K Standard
Spacecraft Cryocooler (SSC) and its successor, the Improved Standard Spacecraft Cryocooler (ISSC).
This decision was implemented to retain legacy to the extensive life test data obtained on ISSC unit
1 and 2 (over 23,000 hours each) and flight test data from ISSC unit 3 (NASA Cryogenic System
Experiment, STS-51). Design modifications from these earlier machines were made to improve
†
Weight penalty is a measure of the weight impact a cryocooler’s power consumption imposes on a spacecraft.
It is an estimated weight of hardware required to supply electrical power and to dissipate waste heat.
PSC CRYOCOOLER PERFORMANCE RESULTS 37
Figure 2. Cross-sections of the PSC compressor (top) and expander (bottom).

thermodynamic performance, reduce residual vibration, reduce weight, and increase ruggedness
and reliability.
Figure 2 shows cross sections of the two modules. The compressor is nominally 418mm long
and 103mm in diameter. The expander is nominally 384mm long and 124mm in diameter. The cold
cylinder extends 102mm beyond the expander’s warm-end waste heat rejection surface. This ar-
rangement typically enables the cold cylinder to project into a sensor cold volume and places it
beneficially adjacent to the user’s load.
QUALIFICATION AND ACCEPTANCE TEST PROGRAMS
Qualification and Acceptance Testing was performed as specified in the Technical Require-
ments Document and the cooler has been shown to meet or exceed all requirements. Qualification
Tests included thermal cycling between 233K and 327K and 12.9 Grms random vibration on each
of three orthogonal axes. To verify alignment stability of the pistons, sticktion hysteresis loops
were recorded before and after each vibration test and before, during, and after each thermal cycle.
Housing gas hermeticity was verified by helium leak testing after final assembly and again prior to
delivery.
Acceptance Tests performed at Raytheon included a variety of thermodynamic performance
characteristics, measurement of cold tip side load capacity, hermeticity, cold tip motion, and tem-
perature stability and repeatability. Self-generated vibrations and cooler EMI were measured at the
JPL. Most of these tests have been repeated at the AFRL with virtually identical results.
Figure 3. PSC thermodynamic performance Figure 4. PSC load lines at three rejection
map at 300K rejection temperature. temperature and at intermediate power levels are
very consistent.
CRYOCOOLER THERMODYNAMIC PERFORMANCE

Figure 3 is a performance map covering the cold tip temperatures from 35K to 120K, all with
rejection temperature at 300K. At the design point, 2W at 60K, the specific power is 28.5 W/W,
referenced to motor power. The efficiency relative to Carnot is 14.0%. (Carnot efficiency is defined
as measured COP over Carnot COP between rejection and load temperatures expressed as per
cent.) In addition, the PSC can support up to 3.0W at 60K, which is a 50% performance margin over
design requirement.
Figure 3 also shows test data taken at the AFRL overlaid on Raytheon test data. The AFRL data
is shown as large circles, Raytheon data as solid lines. The AFRL also took additional data at 90K,
shown as a dotted line. The close correlation between the two data sets indicates both accuracy and
repeatability of the data despite the use of different cold tip temperature sensors, load heaters, and
instrumentation. Raytheon took its data in late 1997 and the AFRL data was taken in 1999. Housing
hermeticity has been continuously verified throughout the 1 1/2 years since delivery.
Although the design specification did not require performance below 60K, the single stage PSC
delivers up to 1.2W at 35K for only 101 W of compressor motor power. The specific power at that
point is 84.2 W/W, corresponding to a robust efficiency relative to Carnot of 9%. This high level of
performance is proving to be the most significant accomplishment of the cooler.
Finally, Figure 4 shows three load lines for the cooler at the piston strokes corresponding to
rated power (i.e., 2W at 60K.) The load lines correspond to rejection temperatures of 275K, 300K,
and 325K. Significantly, the PSC delivers almost identical refrigeration at each rejection tempera-
ture and achieves no-load temperature just above 20K.
CRYOCOOLER SELF-GENERATED VIBRATION

The compressor’s self generated vibration is required to be less than 0.2 Nrms at any frequency,
and the expander’s, less than 0.1 Nrms. Vibrations from each module were measured indepen-
dently by the JPL to gauge the vibration contribution by each. Data are shown in Figures 5 and 6
demonstrate that the cryocooler’s internally generated vibration was well under the government’s
maximum limit. The test data shown below include vibrations with and without Adaptive Feed
Forward (AFF) vibration control. The AFF control was only active on the drive axis in each module
(the Z-axis.) Cross axis vibrations (X- and Y-axes) were not controlled. Due to processor speed
constraints (the breadboard uses a Motorola 68000 operating at 8 MHz) only the first six harmonics
in each module could be controlled within a reasonable computation cycle time. Although a higher
speed processor would enable more harmonics to be controlled, this was found to be unnecessary to
achieve performance requirement.
The performance of the compressor’s three finger tangential flexures, Figure 7a, demonstrated
the predicted low vibration behavior of the Aerospace Corporation patented design. The tangential
Figure 5. PSC compressor self-generated vibration at 2W heat lift at 60K. Cross-Axis vibrations on
left, drive-axis vibrations on right. White columns control OFF, dark columns, control ON.
Figure 6. PSC expander self-generated vibration at 2 W heat lift at 60 K. Cross-axis vibrations on

left, drive-axis vibrations on right. White columns control OFF, dark columns, control ON.
Figure 7a. Compressor suspension flexures. Figure 7b. Expander suspension flexures.
flexures were expected to displace more smoothly than conventional spiral flexure designs used in
most Oxford class coolers, resulting in a significant reduction in cross-axis self-generated vibra-
tion.
The expander module employed previously designed twelve finger spiral flexures, Figure 7b.
This is because data collected from earlier Raytheon coolers indicated cross axis vibrations were
low and did not require further reduction. This was validated, as shown in the test results.
COLD TIP SIDE LOAD CAPACITY

In order to achieve reliable Stirling cryocooler operation, the oscillating displacer and cold
cylinder must not contact during operation. Therefore, after assembly with proper alignment, the
cold cylinder must be stiff enough to carry incidental side loads without deflecting into the moving
displacer. The PSC was required to support a side load greater than 6N in any direction. To validate
this requirement, the PSC was designed with displacer electrically isolated from the cylinder so that
contact resulting from side loads can be detected by a simple conductance measurement. This is a
very sensitive measurement and can be performed during static test at ambient and during operation
at any temperature. Measurements show this requirement was exceeded by more than a factor of
three with a minimum side load capacity of 19.6N.
CORRELATION OF TEST PROCEDURES

The close correlation between Raytheon and AFRL thermodynamic performance data was ob-
tained because both teams agreed to the same test methodologies, clear definitions of test param-
eters and used virtually identical instrumentation. Three key examples include:
1. The heat rejection temperature was defined as the temperature of the liquid cooled heat sink
at the cooler’s waste heat thermal interface, as measured on the user’s side of the interface.
Significantly, temperature sensors were positioned along the waste heat flow path. This
includes the effect of contact thermal resistance at the interface and rejection temperature is
measured at a location that is of most significance to the user.
2. Raytheon and AFRL used identical cold end instrumentation and radiation shielding. Cold
end instrumentation was mounted to a copper ring clamped to the cryocooler cold tip to
represent the user’s side of the cold tip thermal interface. Two calibrated silicon diode tem-
perature sensors and a resistive heater were mounted to the ring so that measurements of
cooler performance included the adverse effect of the interface thermal resistance. Again,
the temperature sensors were located at the point of greatest significance to the user: on the
user’s side of the thermal interface contact resistance.
3. Power consumed by the motor cables was deducted from measured power. This was done
because a user’s cable resistance is likely to differ from the test cables. To enable a user to
account for cable power, maps of current vs. thermal load at constant load and rejection
temperature were developed. Figure 8 is the current map corresponding to the performance
map shown in Figure 3.
CONTAMINATION CONTROL
Long-term cryocooler reliability relies on execution of quantifiable charge gas contamination
control procedures. For the PSC and other Raytheon coolers, a charging procedure developed with
the AFRL has been implemented that consistently achieves the required low levels of contamina-
tion. To confirm adequate control, gas samples are taken periodically throughout the charging pro-
cedure and sent to Pernicka Corporation in Fort Collins, CO for mass spectroscopy analysis. By
agreement with AFRL, Raytheon follows Pernicka instructions regarding sampling hardware and
procedures and reported analysis results are accepted as definitive. Contamination levels of the
various gases, both condensable and non-condensable, are reported in per cent by volume (which is
directly convertible to parts per million.) From this data and knowledge regarding total gas mass
within the cooler, the contaminant levels are calculated as a solid volume. Experience indicates that
Figure 8. The motor current map corresponding to the performance map shown in Figure 3
is developed to aid the system integrator in selecting cable design parameters.
the PSC and similar size coolers can tolerate at least 5 cubic millimeters. At this level of contami-
nation, no adverse effects are observed. Table 1 shows final gas charge analysis. The table com-
bines gas specie and volume, provided by Pernicka, with calculations of specie mass and volume
performed by Raytheon. The contamination level is only 2.23 cubic millimeters. No contamination
effects have been observed during testing at Raytheon, JPL, or AFRL.
PERFORMANCE VERSATILITY
The PSC’s wide performance versatility is one of the program’s major accomplishments. In
1992 the AF was focused on the need for a cryocooler system that could efficiently supply 2 watts
of cooling at 60K. During the course of the program, the AF interest in longer wavelength focal
plane arrays extended refrigerator performance requirements to the 35K range. To address this
need, the performance and acceptance test structure was expanded to map the PSC’s maximum
performance envelope. It was determined that the PSC, without design modification, outperformed
both in efficiency and cooling loads, existing cryocoolers specifically designed to perform in the
35K region. At 35K, the PSC was able to cool loads over one watt with thermo-mechanical effi-
ciencies under 90w/w from a reject temperature of 300K. In addition, the system was able to deliver
up to 3W at 60K, a 50% increase over the 2W cooling load requirement.
COMMENTS ON IPT PROGRAM MANAGEMENT

The PSC Integrated Product Team (IPT), including the government Project Officer and key
Raytheon personnel, encountered and solved a number of problems during the course of the pro-
gram. This successful IPT provides an excellent model that can be used by the Air Force to improve
program management in the future. Program success can be directly linked to the restructuring of
the Statement of Work (SOW) and Technical Requirements Document (TRD) in early 1995. This
was driven by both technical problems and limited funds following the completion of the Critical
Design Review (CDR) in late 1994. The result of these efforts is a system that has exceeded all
initial government requirements. The system was completed with additional funds that was about
50% of the funds spent up to and including the CDR. The initial funding had been almost ex-
hausted due to two separate developmental problems.
The first problem was excessive vibrations transverse to the drive axis caused by non-uniform
displacement of the compressor’s three finger spiral flexure bearings. The solution to this problem
was to replace the spiral flexures with the newly developed Aerospace patented three finger tangen-
tial flexures. The tangential flexure design was found to displace smoothly, resulting in far greater
transverse motion stability. The changeover required significant redesign of the flexure support
system and the drive motor, which was completed by late 1994. The new flexures reduced trans-
verse axis vibrations by a factor of about 40 and reduced axial vibrations by a factor of at least 10.
The second problem was the high cost and projected unsatisfactory performance expected to be
obtained from axial vibration control feedback electronics. Specifically, the high-speed processor
and related circuitry to be provided by SCI was intended to perform the numerically demanding
narrow bandwidth vibration control algorithm in near real time, but the project suffered from cost
and technical problems. The technical limitation was the inability to correct vibrations above the
third harmonic. The cost and performance issues compromised the program. The technical prob-
lem was solved in early 1995 when the government and Raytheon agreed to replace the narrow-
bandwidth vibration control algorithm with a feed-forward vibration control algorithm that could
be implemented in simpler, lower cost electronics. This system could separately correct excessive
vibration levels in the first six harmonics, which was sufficient to achieve program requirements.
The feed-forward algorithm was also found to more effectively reject electrically generated feed-
back noise because it sampled vibrations over several complete machines cycles before calculating
a solution. The cost problem was addressed by rescoping electronics maturity from prototype to an
engineering design model (EDM) implemented within a commercial laboratory grade rack.
DISCUSSION OF SELECTED TECHNICAL PROBLEMS

Several problems arose during the final development of the PSC system that were resolved to
meet or exceed program requirements. Each is discussed below:
a. One compressor motor coil shell buckled during pressurization testing in Dec of 95. The
motor coil assembly is comprised of a thin wall titanium bobbin on which the coil is wound
and a thin wall titanium cover shell welded over the coil to contain potential gas contami-
nants. Both bobbin and shell are thin wall elements to minimize eddy current losses and to
minimize magnet gap distance. Coil lead wires pass through the bobbin via two glass sealed
connector pins. In order to resist buckling in the pressurized cryocooler, the titanium coil
structure must be well supported on the inside. Encapsulating the coils in sufficient filler
material to make a line-to line assembly between titanium and filler provides the required
support. This coil had been built up with insufficient filler, resulting in a gap between coil
and titanium cover large enough to result in buckling when pressurized. A new coil with the
proper assembly tolerances was obtained and the program continued.
b. A piece of the tip of the expander’s titanium piston broke off during early testing in Jan 1996.
The location of the fracture was at a hole near the cold end of the piston that had been cut by
electrical discharge machining (EDM.) A series of these holes is used to secure a clip that
retains the regenerator packing inside the piston. The primary cause of the fracture was that
the EDM process melts and recasts titanium in the zone immediately adjacent to the hole,
resulting in embrittlement at the edges of the hole. The recast material had not been removed
after EDM. To correct the problem, a stainless steel displacer piston was substituted for the
titanium piston. Stainless steel pistons had previously been successfully used in all previous
Raytheon Oxford class coolers. The subsequent thermo-mechanical performance to date has
been outstanding.
c. The most critical problem encountered from a time and cost standpoint was loss of compres-
sor and expander housing hermeticity, also detected in Jan of 96. Hermeticity was lost when
the electrical connectors were electron beam welded to the housings. These connectors con-
tain multiple glass-to-metal 16-guage pins that are used to transfer motor power and position
sensor leads to/from the compressor and expander modules. Localized heating by the electron
beam generated thermal stresses that caused the glass to separate from the pins and/or shell,
resulting in leakage paths. Fortunately, the problem had already being solved for a set of
coolers being produced for the SBIRS Low Program. For these coolers, an inconel connector
shell was developed to replace the previously used steel shells. Inconel welded successfully
because it has a lower coefficient of expansion, which reduces stress during nonuniform heat-
ing, and distributes heat more uniformly, which reduces the thermal strain potential during
welding. Reworking the PSC to incorporate these connectors required complete disassembly
of both modules and remanufacture of several housing components, which extended develop-
ment schedule and increased costs substantially.
d. The high cost to build flight-qualified electronics to drive and control the cooler led to rescoping
of the contract to build laboratory grade rack electronics. The electronics were adapted from
a previously developed design and used a combination of new hardware, spare hardware from
a previous build, and some hand wired circuitry. Unfortunately, the remaining funds for elec-
tronics were barely sufficient for the task, leading to workmanship and quality control prob-
lems. Correcting the problems caused schedule slips and testing delays.
e. Access to key personnel proved to be a problem during periods of SBIRS Low program
activity that required experienced cryogenics engineers and technicians. SBIRS Low’s higher
priority and tight schedule caused the PSC program schedule to slip 6 months while selected
personnel were redirected to build and integrate cryocoolers for the SBIRS Low Flight Dem-
onstration System.
SUMMARY
Despite a significant number of problems and setbacks, the PSC delivered in 1997 has proven
to exceed all cryocooler performance requirements. As described, all difficulties were resolved
through the close working relationship of the entire IPT, particularly between the Raytheon Pro-
gram Manager and the AF Project Officer. The excessive cross axis vibration in the compressor
module caused by the original three-finger spiral flexure design was corrected by using tangential
flexure bearings. Two control electronics problems were solved the first time was with the SCI
processor development, which significantly impacted both cost and schedule. The second incident
was the repair and corrections to the engineering model electronics at the time Raytheon was pre-
paring for cryocooler acceptance testing. Other problems such as the motor coil buckling and the
tip of the expander piston breaking off could be expected on any program; however the loss of
personnel to the SBIRS Low program for the good part of a year was never anticipated. What
seemed like an equally crushing occurrence was the leakage problem with the electrical connectors.
This problem took over four of five months to resolve. What cannot be minimized is the outstand-
ing product that has resulted from this endeavor. Not only does the cryocooler have the capability
to cool greater than 50% of the original contracted cooling load at 60K with outstanding efficiency
(under 30 w/w), but the cooler has demonstrated a great cooling load versatility over a wide cooling
spectrum. This spectrum goes far beyond what was called-out in the original contract. The most
noteworthy of all the cryocoolers capability is the outstanding performance at 35K. It is able to
provide over a watt of cooling for less than 90w/w.
ACKNOWLEDGMENTS
This work was sponsored by the Ballistic Missile Defense Organization. The Air Force Re-
search Laboratory, Albuquerque, NM, managed the project. D. Johnson, S. Collins and P. Narvaez
of the JPL and T. Pollack and M. Kieffer of Raytheon made significant contributions in acquiring
the data reported herein.
REFERENCES
1. Price, K.D., Barr, M.C. and Kramer, G., “Prototype Spacecraft Cryocooler Progress,” Cryocool-
ers 9, Plenum Press, New York (1997), pp. 29-34.
2. 65 K Standard Spacecraft Cryocooler Program Final Report, Contract #F29601-89-C-0082, Hughes
Aircraft Company, Electro-Optical Systems, El Segundo, CA, November 1995.
Characterization of Raytheon´s 60 K 2W
Protoflight Spacecraft Cryocooler
N. S. Abhyankar2, C. H. Yoneshige1, B. J. Tomlinson1 and J. Reilly1

1
2
Dynacs Engineering Co.
ABSTRACT
The Air Force Research Laboratory Cryogenic Cooling Research Facility is supported by
the Ballistic Missile Defense Organization (BMDO) and the U.S. Air Force SBIRS Low
Program Office. It was created to characterize the thermodynamic performance and long life
potential of cryogenic cooling technologies developed by various defense industry contractors.
The objectives of the characterization process are to explore the cooler’s ability to perform at its
design point and to map its range of thermodynamic operation. This provides a detailed
performance envelope for alternative space applications and aids in providing valuable feedback
to cryocooler developers. This paper provides an overview of the AFRL characterization of the
Protoflight Spacecraft Cryocooler (PSC) built by Raytheon Systems Co.
The PSC is a split Stirling cryocooler with dual, opposing motion compressors and a
displacer, which uses a mass balancer to reduce vibration. It is designed to lift a heat load of 2W
at 60K, with a nominal rejection temperature of 300K. The characterization of the PSC involved
a series of experiments, including cool-down to lowest temperature, design point verification,
parameter optimization, stiction tests, long term-stability, transient thermal response (due to
orbital temperature variation) and a proto-qualification thermal vacuum test. The performance
map is charted providing a graphical display of important parameters including specific power,
calculated as input power per watt of cooling. The off-nominal performance evaluation included
lifting 1.2W at 35K and 6W at 120K, while operating within recommended total input power and
stroke boundaries.
INTRODUCTION
The Air Force Research Laboratory Cryogenic Cooling Research Facility (CCRF) is
engaged in the long-term functional and thermal evaluation of various types of cryocoolers. The
thrust of operations is to characterize these coolers and verify their design requirements, to
provide feedback to cryocooler developers for further improvements in future designs and to
allow potential users to evaluate and compare different cryocoolers for specific space flight
missions. The CCRF, with assistance from the AFRL and the Ballistic Missile Defense
Organization (BMDO) is equipped with vacuum chambers, instrumentation, material resources

and personnel to carry out short and long-term evaluations of the cryocoolers. In the facility, a
cryocooler can undergo numerous years of continuous operation to identify and evaluate its
efficiency, reliability and performance changes over time.
Raytheon’s Prototype Spacecraft Cryocooler (60K PSC) is a split Stirling, long life (7+
years) space cryocooler designed to lift 2 W of heat at 60K. This cryocooler is a protoflight unit
intended to demonstrate the advancement of Stirling cryocooler technology for use in long life
strategic space applications. It is designed to operate for 7 to 10 years on orbit (continuous duty
cycle) with very high reliability (>0.95). The motive for developing the 60K PSC was provided
by the Space Based Infrared System Low (SBIRS Low), BMDO and the AFRL management
team for cooling infrared (IR) sensors aboard surveillance spacecraft. The 60K PSC incorporated
lessons learned from Raytheon’s internal research and development coolers (the Improved
Standard Spacecraft Cryocoolers) and the Air Force sponsored Standard Spacecraft Cryocoolers.
The 60K PSC has other potential cooling applications including chilled electronics and
superconducting devices.
DESCRIPTION OF THE CRYOCOOLER
The 60K PSC incorporates the use of both spiral and tangential flexure bearings. The
compressor unit is a dual dynamically balanced gas piston design, which works on a common
compression space. The expander is balanced using a balance mass operating in an opposed
motion to the expander piston, thereby cancelling vibration. Clearance seals with liners ensure no
metal-to-metal rubbing or wearing contact along all running clearances. The motor coils are
canned to mitigate the potential for wires or potting material to out-gas and contaminate the
helium working fluid.
The cryocooler is operated using a single control electronics rack that is convectively cooled
internally. The cryocooler has been validated for operation with a rejection temperature (TR)
between 2°C (275K) and 52°C (325K). The control electronics are not qualified for space flight
in their present form.
Specialized Mounting Fixtures

As shown in Figure 1, the mechanical fixtures were specially designed for this experiment to
allow the cryocooler to be mounted inside a 36” thermal vacuum chamber. The heat rejection
surface is provided by standard cold plates, which have chiller fluid transfer lines for cooling and
electrical coils for heating. The rejection temperature at the interface is automatically maintained
by temperature controllers switching between cooling and heating.
Figure 1. 60K PSC integrated with a 36” thermal vacuum chamber.

CHARACTERIZATION OF RAYTHEON’s 60 K 2W PSC COOLER 47
Cold End
There are two Lakeshore silicon diodes attached to the cold end of the PSC. One of them is
directly connected to the PSC controller electronics, which allows the use of an automatic
temperature controller. The other diode is used for independent monitoring and performance
verification through the PSC data acquisition software. The calibration curves, which range from
4K to 330K, were inserted into the temperature controller instrument to ensure accurate
measurements. The diodes and the heater are wired with cryogenic grade, low conduction
twisted wires. The entire cold finger assembly is shielded from the vacuum environment and
from the expander base by multi-layer insulation (MLI). The MLI is isolated from the cold finger
by baffles. This set-up helps reduce conductive and radiative parasitic losses and give a more
accurate estimation of the cryocooler’s performance.
Instrumentation for the Controller, Data Acquisition and Environmental Control
The instrumentation for the PSC is separated into three stacks: one set for controlling the
cryocooler itself, another for data acquisition (including power and temperature measurements)
and a third for monitoring and controlling environmental parameters (such as heat rejection
temperature and vacuum pressure). The instruments are controlled by LabVIEW™ via IEEE-
GPIB or RS232 communication interfaces. An automated data acquisition system records the
experiment data and environmental conditions. The data acquisition is also provided by
LabVIEW™ software.
CRYOCOOLER CHARACTERIZATION
The PSC was integrated in a 36” thermal vacuum chamber and connected to its peripheral
equipment according to procedures established by AFRL and Raytheon personnel. A test
readiness review was held, where laboratory personnel presented their understanding of the
performance and safety requirements and showed that the experiment stand was adequate to
carry out the characterization activities. A series of experiments were performed as an
acceptance evaluation before the more elaborate characterization experiments were undertaken.
The characterization plan was created, which established the sequence of operations according to
an experiment matrix. The characterization plan included the following evaluation activities:
1. Optimization Evaluation
2. Characterization Load Lines/Performance Mapping at Different Heat Rejection
Temperatures
3. Temperature Stability at Different Heat Rejections (With and Without Heat Load)
4. Transient Thermal Response
5. Cool-down at Different Heat Rejection Temperature (With and Without Heat Load)
6. Parasitic Heat Load Determination (Off State Conduction)
7.Thermal Cycle/Thermal Vacuum
Optimization Evaluation
The performance of the cooler depends on various parameters such as compressor stroke
length, expander stroke length, the phase angle between the compressors and expander, operation
frequency, piston offsets, and heat rejection temperature. To ensure the most efficient cryocooler
performance, these parameters must be optimized. In the laboratory, optimization is carried out
by varying one parameter at a time while all other operation parameters and conditions are kept
the same. The cold end temperature and input power are used as performance evaluation
parameters. These are recorded for a set of heat loads while obtaining the sensitivity of the
performance to the chosen parameter. The dimensional effect of the temperature and power
values is avoided by defining the efficiency factor in terms of a percent Carnot efficiency, which
is given by the ratio of thermodynamic coefficient of performance (COP) to the Carnot
Figure 2. Percent Carnot COP for expander stroke length variation.
efficiency. (The thermodynamic COP is defined as the total input power minus the I2R losses
divided by the cooling power expressed as a heat load.) As an example, the variation of the
expander stroke length is presented in Figure 2. The fixed parameters were as follows:
Compressor A stroke length = 7.5mm, Phase Angle = 72.4°, and Operation Frequency = 35 Hz.
These values were carefully monitored by changing the command signals. Vibration control was
turned off and the heat rejection temperature was 300K. Data points were obtained for discrete
heat loads between 1W and 5W.
As seen in the figure, the percent Carnot COP for an expander stroke length of 2.7mm is
higher than that obtained for 2.5mm. The maximum expander stroke length at a rejection
temperature of 300K is about 3.2mm and decreases to 3.0mm at a rejection temperature of 325K.
In order to maintain a large enough safety margin and to lower the input power, for a fixed stroke
length on compressor A, 2.5mm was chosen as the nominal expander stroke length. If a lower
heat rejection temperature were used, the expander stroke length could be slightly increased
without any significant increase in input power.
Characterization Load Lines/Performance Mapping
A baseline depicting input power and cold end temperature changes over a range of heat
loads provides a picture of the cryocooler’s nominal performance, which can be tracked
periodically during a long-term endurance evaluation. The off-nominal performance map shows
the cryocooler’s entire performance with different heat loads, stroke lengths, and lowest cold end
temperature. The limits of the off-nominal performance evaluation are based on the cryocooler’s
upper limit of input power (~100W for PSC) and the highest recommended operation
temperature (~150K for PSC), which were established by payload considerations and the
properties of the motor coils. The data is obtained as a series of points at steady state conditions.
The performance map is shown as a carpet plot of input power versus heat load for constant
Figure 3. Performance map for heat rejection temperature of 300K.
compressor stroke lengths and constant cold end temperature. It can also include specific power
lines, which are used as a tool to compare cryocooler performance. The specific power is the
inverse of the thermodynamic COP. In comparing two different cryocoolers, at comparable
conditions, the cooler with the lower specific power is considered as a better performer. The
performance map also allows the estimation of additional information for constant temperature
or constant stroke lines without actually performing the experiments along those lines.
Performance maps are dependent on heat rejection temperature.
The performance map for a heat rejection temperature of 300K is shown in Figure 3. The
data points are obtained at steady state for fixed stroke lengths ranging from 2 mm to 8.5 mm
and fixed heat loads from 0W to 6W. The input power and resultant cold end temperature values
were analyzed, processed using curve fits, and reduced to get constant temperature lines. The
dark lines show the characteristics of performance at constant cold head temperature and
constant stroke length. The constant temperature lines are verified by running a set of constant
temperature load lines. These data points are shown as filled circles in Figure 3. The curve fit
trend lines provide a comparable accuracy for the constant temperature lines. The specific power
lines are shown as dotted lines ranging from 5W/W to 150W/W. The data point for a 2W heat
load with the cold end temperature at 60K shows that the specific power is below 30W/W, as
specified in the design requirements. Similar performance maps were generated for heat rejection
temperature of 275K and 325K.
Cool-down Curves
Cool-down curves provide important information regarding the cooling effectiveness of the
cryocooler. The time to cool down to the design temperature and the lowest possible temperature
can be used as one of the performance criteria for comparing two different cryocoolers or for the
selection of a cryocooler for a rapid cooling mission requirement.
Figure 4. Cool-down curves for heat rejection temperatures of 325K, 300K and 275K with no heat
load and a 2W heat load.
A total of six cases were considered. For the first three cases, heat load was not applied, and
the heat rejection temperature was set at 300K, 275K and 325K. For the next three cases, a 2W
heat load was applied when the cold end temperature approached 225K at the same three heat
rejection temperatures. The reason the 2W load was not applied until the cold end reached 225K
was to protect the resistor heater, which is rated for 0.1W at room temperature, from burning up.
The six curves are shown together in Figure 4. The time is scaled so that there is a common zero
where the cooling starts.
During the cooldown, the input command is held constant by the LabVIEW™ controller VI,
branding it as a constant input command rather than a constant stroke cool-down since the stroke
length changes as the cooler cools down.
Design Point Performance

The operational parameters for the 60K at 2W design point are given in Table 1 for all three
heat rejection temperatures. They are provided as a reference for future evaluation.
Temperature Stability
A space flight cryocooler must show temperature stability for a constant heat load and other
fixed conditions. It should maintain a given cold end temperature for a certain length of time.
From a laboratory-experiment standpoint, the temperature should stay within a fraction of a
Kelvin for a period of one hour. A larger window is given for a 24-hour period. Although this is
not a pass/fail criterion, it is a desirable feature. In Figure 5, a long term (>48 hours) stability
evaluation for a heat rejection temperature of 300K and a 1W heat load is shown. The stability
Figure 5. Temperature stability at heat rejection temperature of 300K with 1W heat load.
criterion of ± 0.1K for 1 hour and ± 0.5K for 24 hours was selected. The banded profiles
superimposed on the actual cold end temperature are shifted by an hour with a value change of ±
0.1K and by 24 hours with a value change off 0.5K. For strict stability, the cold end temperature
should stay within these bands. This was achieved for a heat rejection temperature of 300K and a
heat load of 1W, as shown in Figure 5. The bands take into account the fact that rack electronics
are influenced by the ambient temperature in the laboratory, which fluctuates between daytime
and nighttime outdoor temperatures. This made it difficult to maintain a given temperature for a
fixed heat load. During the stability evaluation, the input commands for the cryocooler
components were fixed through the LabVIEW™ software controller VI.
Transient Thermal Response.
For a cryocooler to be considered for cooling space electronics, its sensitivity to the orbital
fluctuation in heat rejection temperature must be determined. The transient thermal response
(TTR) evaluation done on the PSC assumed a sinsusoidal variation in heat rejection temperature
between 295K and 305K with a 90-minute period. The heat load was fixed at 2W. Figure 6
shows the variation in rejection temperature and the corresponding variation in cold end
temperature. The rack electronics temperature is also included to show its variation in the same
time frame and extrapolate its effect on the cold end temperature. The cold end temperature is
shown at the bottom of the plot on the primary axis and shows a cyclic variation due to the
variation of the heat rejection temperature, as well as a slight creep due to the drift in rack
electronics temperature. The overall temperature variation during this time period is about ±
0.5K.
Thermal Cycle/Thermal Vacuum Effects (233/330K Proto-Qualification Level Cycles)
The thermal cycle evaluation is performed by putting the entire cryocooler assembly through
extreme changes in temperature, as are normally encountered in typical space flight conditions.
The idea behind this experiment is to pick out any failure mechanisms that arise as a result of a
Figure 6. Transient thermal response with heat rejection temperature of 300±5K, 90min period.
Figure 7. Thermal cycle/thermal vacuum heat rejection temperature profile.
mismatch in coefficients of thermal expansion. The PSC was cycled through a predefined profile
of heat rejection temperatures ranging from 230 K to 328 K as shown in Figure 7. The
cryocooler was operational while the heat rejection temperature was between 275K and 325K.
The dwell time on the first and last cycle lasted for 6 hours at each extreme temperature. For
shorter cycles, the dwell time was limited to 1 hour. As the experiment progressed, it became
apparent that the limited chiller capability did not allow a defined dwell time at the lowest
temperature. The four cycle test was then repeated with a 3 hour dwell time to account for
temperature settling. Figure 7 shows the entire process, which lasted for about 5 days. The
controlled profile was generated automatically through control software. The dwell time, rise/fall
rate and start/end temperature settings were adjustable. The desired temperature change rate was
1K/min. Stiction tests were performed during the 325K dwell times of the first and last cycles to
verify that no contact was being made between the piston and the cylinder.
CONCLUSION
The Raytheon 60K PSC performed well at the design point of 2W @ 60K with a heat
rejection temperature of 300K. The design point settings required less than 30W/W specific
power showing an improvement in performance over split-Stirling space cryocoolers of the past.
The optimization of parameters was carried out to obtain the most efficient combination of
operation parameters. A performance map was generated for heat rejection temperatures of
275K, 300K and 325K. A long-term stability evaluation showed that the dependence of the
laboratory grade rack electronics on the ambient laboratory temperature caused the cold end
temperature to fluctuate. The proto-qualification thermal vacuum/thermal cycle evaluation did
not bring out any failure mechanisms with the heat rejection temperature between 230K and
328K. The PSC was moved to a 24” chamber and is undergoing an endurance evaluation. So far,
the PSC has accumulated over 3600 hours of operation.
The Development of a 10 K Closed Cycle
Stirling Cooler for Space Use
G. Baker†, D. Féger†, A. Little†, A.H. Orlowska#, T.W. Bradshaw#,

M. Crook#, B.J. Tomlinson*, and A. Sargeant+
†
Astrium (UK) Ltd, Stevenage, United Kingdom
#
Rutherford Appleton Laboratory, Chilton, Didcot, United Kingdom
*US Air Force Research Laboratory, Kirtland AFB, New Mexico, USA
+
Cubic Applications Inc, Lacey, Washington, USA
ABSTRACT
A two-stage, split, Stirling cycle 10 K cooler is being developed for space applications to
achieve the requirement for cooling silicon-based IR detectors. This program has been sponsored
by the US Ballistic Missile Development Organization. It is a further extension of a Rutherford
Appleton Laboratory (RAL) 20 K cooler being space qualified at Astrium (formerly Matra Marconi
Space (MMS)). New features include optimized geometry and enhanced regenerator materials as
well as a larger compressor system.
This paper describes the experimentation performed to develop the optimized geometry for
the cooler displacer. The experimentation was achieved using existing two-stage Stirling cycle
20 K cooler hardware. Operation at 10 K is a challenging concept for a Stirling cooler and
necessitated the development of new heat exchanger materials and configurations optimized for
this new temperature range.
The proof of concept program included the detail design of the optimized displacer and
compressor, and the manufacture and assembly leading to proof of concept laboratory testing.
During testing, the cooler reached a base temperature of 9.4 K.
INTRODUCTION
This 10 K cooler has been derived from an initial 500 mW at 35 K US Air Force Research
Laboratory development effort, which was modified in 1997 to focus on a 45 mW at 10.3 K
requirement to meet the operational temperature and heat lift needs of Si:As IR sensors.
The effort was conducted in collaboration with RAL, and proposed to meet the new 10.3 K
requirement using a two-stage Stirling cooler based on the experience gamed with the design of
the MMS 20-50 K cooler, but using rare earth type regenerator materials and an increased swept
volume; the MMS 20-50 K cooler has a 12 K base temperature and was developed for the Euro-
pean Space Agency.
DEVELOPMENT STRATEGY
The major problem that prevents a regenerative cycle from reaching very low temperatures is
the regenerator heat capacity. As the temperature falls, the density of the gas passing through the
regenerator increases, and hence the heat flow into the regenerator also rises. Unfortunately, the
heat capacity of most conventional regenerator materials falls as the temperature decreases.
Improvements in regenerators have therefore focused on those materials which undergo mag-
netic transitions, at an appropriate temperature, with a large magnetic contribution to the specific
heat in order to get a large volumetric specific heat in the 9 to 20 K temperature range. The
optimization of regenerator materials has to also fulfil manufacturing and cooler operational re-
quirements. The correct operational parameters are extremely important and any gains made with
a better choice of regenerator material may be cancelled by a poor design. In particular, the
regenerator must be porous—with a high surface area to volume ratio—allowing the working fluid
to pass though it, and facilitating good heat exchange. At the same time, the regenerator-material
trade-off must take into account the thermal and fluidic performances, as well as manufacturability
and compatibility with the cooler environment, such as launch vibration or extreme temperature
cycling during operation. A systematic review of regenerator candidate materials was performed
to identify materials having the highest volumetric heat capacity, and, at the same time, mechani-
cal and chemical properties compatible with the manufacturing process and launch environment.
The selected materials went through a qualification process with regard to their mechanical,
thermal and chemical compatibility.
The second problem of low temperature in any refrigeration systems is one of efficiency. The
maximum Carnot coefficient of performance at 10 K is 3.5%, and a small Stirling cycle cooler
would typically only operate at 1 % of Carnot at 10 K. This implies that the input power required
to obtain 45 mW at 10 K would be well above the capability of a pair of existing MMS cooler
compressors such as those used in the MMS 20-50 K cooler, which are limited to 50 W input
power.
As the current cooler is a proof-of-concept model, and the key issue is the design of a regen-
erator to provide enough efficiency at temperatures below 20 K, MMS proposed to focus the
development effort on this part of the cooler. For the compressor, it was decided to use two pairs
of MMS 20-50 K cooler compressors; they would be used as a development tool to test the most
critical component of the cooler, the regenerator.
For the same reasons, as well as to constrain costs, the displacer mechanism and its momen-
tum balancer were based on the existing MMS 20-50 K design.
In order to give confidence in the ability of this type of cooler to achieve the specified
performance, both laboratory test work and mathematical modelling have been utilized. The com-
putational modelling has focused on predicting the effect on cooler performance of geometrical
and material changes to the regenerator design. The test effort has been used to correlate the
mathematical model and to demonstrate actual measured performances of various displacer con-
figurations.
This approach resulted in a first prototype running at RAL in 1998 on which different candi-
date regenerator materials were tested for performance while MMS was qualifying the corre-
sponding manufacturing processes to insure that they would meet mechanical and thermal envi-
ronments. This first prototype reached a 9.9 K base temperature in November 1998.
Based on these results and a pressure drop analysis (see Orlowska1), a new displacer design
was defined and manufactured to further improve the performance. This displacer was, in 1999,
connected to the existing two pairs of compressors to make a proof of concept model of the 10 K
cooler whose design and testing are described hereafter.
DESCRIPTION OF THE 10 K PROOF OF CONCEPT COOLER

Figure 1 gives an overview of the overall configuration of the cooler. Although this is a
laboratory proof of concept cooler, it is made of flight-type hardware and follows a flight like
architecture:
- Two pairs of back-to-back MMS 20-50 K cooler compressors
- One MMS 20-50 K cooler displacer mechanism
- One MMS 20-50 K cooler momentum balancer
10 K CLOSED CYCLE STIRLING COOLER FOR SPACE USE 57
Figure 1. 10 K cooler overview.
Two testing configurations have been considered for this prototype: one at ambient atmo-
sphere, with air cooling, at MMS premises, and the other in a thermal vacuum chamber, with
conductive cooling, at the US Air Force Research Laboratory at Kirtland Air Force Base, To be
able to operate in vacuum the whole cooler is mounted on a single base plate, which mimics the
spacecraft mechanical and thermal interfaces. The four compressors are mounted in two co-axial
pairs to minimize the level of force exported on the assembly. Each pair of compressors has a
transfer line and a manifold leading from the compressed volume inside the cylinder head to the
displacer. The compressor pairs are run in anti-phase with respect to each other.
The compressor pairs are mounted on a simple supporting structure. This structure is made
from aluminium alloy, and transfers heat from the unit heads to the base plate. The compressors
are attached to the structure via adapter plates with thermal gaskets fitted to enhance the thermal
conduction across the joint. An aluminium thermal housing is fitted to each compressor linking
the copper cylinder head to the support structure to increase the heat rejection capability. A
thermocouple is attached to each of the compressor heads for temperature monitoring during
operation.
The compressor support structure is bolted to a base plate using thermal filler at the interface
to reduce the compressor head temperature in a conductively cooled configuration.
A number of options were considered for the manifold and transfer line assembly. These
included:
- Four compressors feeding into one manifold connected to the displacer;
- Two manifolds joining the two pipes from each compressor pair and leading to a third
manifold joining these two pipes, with a single pipe leading to the displacer.
The design selected features two transfer lines leading to the two compressor pairs to separate
inlet ports in the displacer head. The benefits of this arrangement include:
- Use of existing compressor assembly without modification;
- Low overall transfer line length and manifold complexity;
- More even pressure distribution in the displacer.
A cylindrical support tube which surrounds the momentum balancer and the displacer mecha-
nisms provides at the same tune mechanical support and conductive heat sinkage. This is repre-
sentative of a flight bracket, with the exception that provision of force transducer is not included.
The exported vibration of this assembly is low level, but not actively controlled. The concept can
be extended to include force transducers between the displacer mounting plate and the tube assem-
bly. In that case, flexible thermal straps would be used to link the displacer mounting plate to the
support tube.
The cold finger itself was initially equipped with a launch support tube to provide structural
support to the cold finger during launch vibration. Based on the MMS 20-50 K design, this
support tube features resilient snubbers giving a small clearance at the cold end of the cold finger
to limit deflection and loads during vibration with limited heat leak. Thermal blankets are mounted
inside the launch support tube to insulate the cold finger.
This launch support tube was not required for this proof of concept for which no launch
vibration requirements were specified by the customer. For the case of the MMS 20 -50 K cooler,
the European Space Agency put stringent requirements on the cooler side loads capabilities, so, as
explained below, this support tube was later-on removed to improve the cooler with respect to heat
leak and heat shield performance.
TEST SET UP
For test purposes at ambient pressure, the cold finger was surrounded by a vacuum can,
which was linked to a vacuum pump. This vacuum can was equipped with an electrical feedthrough
for the thermal sensors and the heater wires.
Figure 2 gives an overview of the cooler test configuration. The cooler was tested on a fill and
purge test bench with which it was possible to evacuate the cooler and adjust as needed its fill
pressure. Several fens blowing on the displacer and the four compressors provided forced-air
cooling. Thermocouples were located on the compressor heads and on the displacer flange to
monitor heatsink temperatures. The cold tip was equipped with two high precision temperature
sensors. An additional sensor provided the mid-stage temperature. The cold tip was equipped as
well with a heater to simulate the user’s heat lift requirement.
The compressors, displacer, and momentum balancer were driven by two sets of laboratory
drive electronics, each driving one of the pair of compressors, and either the displacer or the
momentum balancer. One of the drive electronics was used as the “master”, to which the second
one was “slaved” to provide adequate phasing and operating frequencies between all the different
components.
Figure 2. 10 K test set up overview; note launch support tube around cold finger.
TESTS RESULTS
Performance testing of the optimized cooler started in September 1999, but the cooler only
achieved 40 K. MMS and RAL ran a joint investigation of the performance. Investigation focused
quickly on the displacer and it was decided to disassemble it to check: 1) the regenerator material
packing, 2) the regenerator geometry, and 3) the cold finger bore geometry.
The regenerator geometry was found to be faulty, generating an oversized clearance between
it and the cold finger bore, leading to bypass flow of gas toward the cold tip expansion volume.
Based on these findings, and in agreement with the customer, a recovery plan was set up at the
beginning of October, to re manufacture a new regenerator. The displacer was rebuilt by the end
of November.
After rebuilding the displacer, the cooler achieved 12 K during initial testing, confirming the
investigation findings. Testing was re started in January 2000 at MMS after vacuum baking, to
remove any moisture in the system.
Optimization of the cooler was carried out with different fill pressures. The lowest tempera-
ture achieved on the cold tip was then 11.2 K with a mid stage temperature slightly under 200 K.
Review with the customer and RAL of these second performance tests results concluded that the
cold stage was performing correctly, but that the temperature of the mid stage was much higher
than expected. One of the main explanations identified was that there could be some extra heat
leaks into the system, especially as RAL pointed out that they did not consider, either in their
model or in their previous 10 K prototype, the implementation of the above-mentioned launch
support tube; the tube was inherited from the MMS 20 K cooler design. As the customer had, on
this proof of concept model, no requirement with regard to launch vibration, this launch support
tube was removed. At the same tune, the Multi Layer Insulation (MLI) was reviewed and consid-
ered non-optimized as well, so it was decided to: 1) remove the launch support tube, 2) redesign
and re manufacture the MLI around the cold finger, and 3) design and manufacture an aluminium
heat shield attached to the mid stage. Routing of the wires of the cold tip temperature sensors was
reviewed too, to reduce the corresponding heat leaks.
As thermalization of the displacer was identified also as being critical, an additional copper
heat sink was designed and manufactured, using the bolted interface of the former launch support
tube to provide additional heat sink around the base of the cold finger. This extra heat sink was
implemented later in final testing so the impact of the heat leak reduction actions could be seen
initially.
Testing was done on the cooler to verify the improvement on performance seen on the cold
tip, due to these modifications. The heat load to be applied to the cold tip to reach the temperature
achieved before these heat leak reduction actions was 67 mW. The mid stage temperature was
reduced by 30 K. Further optimization, especially with regard to fill pressure, led to a base
temperature of 10.4 K, and to 11.6 K with a 45 mW heat load.
Following these results, the cooler was warmed up to ambient in order to implement the
additional displacer heat sink, replace the cold finger open bracket with a flight-like cylindrical
bracket, and to crimp the fill ports; this brought the cooler to its deliverable configuration.
Further optimization tests with regard to phase, frequency, and displacer and compressor
strokes were then performed in March 2000. These led to an ultimate record breaking base
temperature for a Stirling space cooler of 9.4 K, and 10.4 K with 45 mW heat load as shown in
Fig. 3.
CONCLUSIONS
This test campaign, summarized in Fig. 4, successfully demonstrated the feasibility of reach-
ing the 10 K temperature range with a space-rated, two-stage Stirling cooler.
This achievement is mainly due to the successful use of advanced rare earth regenerator
materials. With 10.4 K, the target specification of 45 mW at 10.3 K was nearly met during test
with ambient-air cooling; this cooling was not able to keep the compressors heads and displacer
flange below 30 C. During the foreseen thermal vacuum tests at the US Air Force Research
Figure 3. 10 K proof of concept cooler heat load line.
Laboratory, tests at colder heat rejection temperatures should allow the cooler to overcome this
present limitation and obtain the ultimate performance of the cold regenerator.
With regard to design, the regenerator was successful, especially the cold stage. The mid
stage temperature being 30 K higher than foreseen, we therefore plan to review its sizing to further
lower its temperature. This mid stage redesign should be performed taking into account the heat
shielding and sensor interface strategies, as they are key parameters for the overall instrument
cryogenic performance, and should be investigated in detail to review the cold finger design
accordingly.
The displacer mechanism should be reviewed to improve its heat sinking capability; the
displacer waste heat is estimated at 60 W, leading to a displacer flange temperature as high as
30°C when using ambient (20°C) forced-air convection cooling.
Figure 4. 10 K proof of concept cooler temperature history.

The four 20-50 K cooler compressors used to feed the 10 K displacer in this proof of concept
cooler fulfilled their task with limited heat sink capability and input power margins. A new com-
pressor, with a greater swept volume, should be developed to meet the 10 K cooler requirements.
This should reduce the mass and cost of the cooler and increase its efficiency.
This future work should lead to a qualified and optimized Stirling cooler able to provide up to
100 mW in the 10 K temperature range, either to meet the requirements of future Silicon based
sensors or to act as a pre cooler for Joule Thomson coolers operating in the 3 to 4 K temperature
range.
ACKNOWLEDGMENT
This development work was funded by the Ballistic Missile Defence Office.
REFERENCES
1. A.H. Orlowska, T.W. Bradshaw and S. Scull, “Progress Towards the Development of a 10K Closed
Cycle Cooler for Space Use,” Cryocoolers 10, Plenum Publishers, New York (1999), pp. 67-76.
Development of a 12 K Stirling Cycle Precooler
for a 6 K Hybrid Cooler System
W.J. Gully, D.S. Glaister, and D.W. Simmons

Ball Aerospace & Technologies Corp.
Boulder, CO, USA 80306
ABSTRACT
There is a need for reliable, space-qualified mechanical coolers for temperatures of 10
Kelvin and below for use in both X-ray and infrared systems. Our analysis shows that hybrids
consisting of a Joule-Thomson (J-T) cooler coupled with a mechanical cooler are the most
efficient when used in a low Earth orbit. We are developing a hybrid cryocooler consisting of a
helium J-T system coupled to a Stirling cycle mechanical cooler for these applications. We plan
to use the relatively mature Stirling cycle mechanical cooler to provide all of the precooling, and
use the J-T recuperative system for refrigeration below 20 K, a region that has historically been
difficult for regenerative coolers. In this paper we discuss our work in developing this precooler
on our NASA Explorer 6 K program. A discussion of the J-T portion of this system will occur
elsewhere in this conference.1
Our precooler work is a continuance of our previous work on linear Stirling coolers. We
have developed one-, two-, and three-stage Stirling coolers for various NASA and Department of
Defense programs. For the Explorer 6 K, we have focused on adapting our three-stage Stirling
cooler for operation at lower temperatures. In this program we are using breadboard tests to
acquire both an empirical knowledge and an analytical understanding of the factors that affect the
low-temperature performance of our cooler. This information will be used to develop a space-
qualified precooler during a later phase of the system development.
In our initial tests we have demonstrated improved performance when we substituted a lead
shot regenerator for the previous regenerator constructed from phosphor bronze screens.
Although the data suggests future directions for more cooling, at present the performance falls
short of our analytical expectations.
MOTIVATION FOR A HYBRID COOLER

Our use of a hybrid system reflects the difficulties in achieving the low temperatures with
just a regenerative mechanical cryocooler.2 With a hybrid, the effort below 20 K is shared
between the mechanical cooler and the J-T cooler (Figure 1). Our analysis suggests that the
hybrid system can be designed to run with approximately the same power efficiency with the
precooler operating anywhere from 12 K to 16 K. Our task is to learn what we can actually
achieve with our precooler. The applications we are looking at, a sub-Kelvin ADR for the X-ray
work and a medium/far infrared wavelength detector for the infrared, require about 100 mW of
cooling at this temperature.

Figure 1. A schematic of a hybrid cooler, consisting of a J-T system

coupled to a multistage mechanical cooler.
There are a number of intrinsic problems that regenerative coolers face at these tempera-
tures. The density of helium gas has become significant and there is a lack of sufficient matrix
heat capacity to store the internal energy of the working fluid during the cycle. In addition,
because of the high densities, miniscule dead volumes in the cold end hold substantial amounts
of nonproductive helium and degrade the performance. The pressure dependence of the heat
capacity in the helium working fluid leads to a phenomenon in the regenerator similar to the
imbalance in a recuperative heat exchanger and causes unavoidable heat leaks.
There are additional problems associated with the use of the spring suspended linear cool-
ers. These coolers were developed to eliminate the need for a wearing seal in both the compres-
sor and in the cold head. With a stiff radial suspension, the gap clearance can be made small
enough to dynamically support a pressure wave at sufficiently high frequencies. These coolers
typically operate above 30 Hz to sustain the pressure and to run on resonance. Early work has
been done on the extra losses in the cold head due to this effect.3
In contrast, commercial mechanical coolers with wearing seals easily provide watts of cool-
ing at these temperatures. Because the regenerator heat current scales with operating frequency,
the ability to run at low frequencies (typically a few Hz in these coolers) is a thermal advantage
and results in higher effectiveness. Due to our reliability requirements, we must use our
noncontacting Stirling precooler and operate at the higher frequencies.
6 K CRYOCOOLER TEST PROGRAM

We are using a three-stage Stirling cycle breadboard cooler originally developed for the Air
Force 35/60 K program. At the time our goal was to provide two separate refrigeration stages on
the same cold finger. In this effort we use the multiple stages to provide efficient operation at
low temperatures. With three refrigeration stages, we can divide the temperature drop across
more regenerators and reduce the heat load on the last stage. This compensates for the increased
ineffectiveness and allows for the lowest possible temperatures at the cold tip.
The multistage expander offers several advantages as a precooler for a J-T system. The
intermediate stages are used to absorb the heat due to the inefficiencies of the various recupera-
tive heat exchangers that precool the fluid for the J-T system.
12 K STIRLING PRECOOLER FOR A 6 K HYBRID CRYOCOOLER 65
Figure 2. The Stirling cycle displacer breadboard used in the development of the low-temperature
cooler.
Figure 3. The baseline thermodynamic performance of the 35/60 K cooler at its lowest temperatures.
The baseline configuration of our breadboard cooler is shown in Figure 2. Its three
expansion stages are indicated by steps in the outer tube. Both middle and cold stages have
interfaces for attaching loads. In this particular case the warmest stage, which typically runs
around 180 K, is used primarily to improve the thermodynamic efficiency of the unit and has no
external load.
This breadboard is particularly suited to regenerator studies. It has demountable interfaces at
the base, midstage, and cold tip, which allow the external regenerators to be removed and
replaced without much difficulty. In particular, this can be done without disturbing the displacer
drive mechanism.
Our goal was to understand the specific influence of a number of factors at low
temperatures, so we began by making a careful measurement of the performance of the existing
35/60 K cooler. In this configuration we used typical wire screen regenerators. The results are
shown in Figure 3. As an example, when the cooler was lifting 0.1 watt at the cold tip, and had
no load on the midstage, the cold tip was at 20.5 K and the midstage was at 42 K. The input
power in these circumstances was approximately 48 electrical watts to the compressor.
Figure 4. The breadboard cooler cold finger, showing the larger diameter outer regenerator on our
displacer drive for the lead regenerator. The original outer tube for the screen regenerator is to the left.
We performed trend tests to explore the cooler’s limitations near its lowest temperatures.
The strongest indication of regenerator limitations was the lack of response to increased pressure
and frequency. These trends generally agreed with the results of our analytical models,4 and
could be explained by the vanishing capacity and subsequent loss in the regenerator.
These trends and the analytic models suggested that the cryocooler was regenerator limited.
Consequently, an improvement in the matrix capacity should produce a substantial improvement
in the low-temperature performance because of improved efficiency. Improving the efficiency
removes a “heat leak” between the midstage and the cold tip, and we predicted a quarter watt
increase in lift at 20 K, along with a similar loss of heat lift at the midstage.
TEST WITH LEAD SPHERES

Our first effort for converting to work at low temperatures was to introduce a lead (95/5 Pb-
Sb “babbit”) regenerator. This is a traditional choice for a low-temperature regenerator because
of its relatively high capacity, because it is well characterized, and because it is available as fine
shot, a form that can be used in a regenerator. We procured the lead from Clad Metal Industries.5
Our goal was to understand its real impact on our low-temperature performance.
The regenerator influences the cryocooler performance in a number of ways, and it can be
difficult to separate them out. In an ideal situation we could have obtained woven lead screen
with the same porosities and flow factors as before, so the substitution would be limited to the
material properties only. However, we have tried to size the lead regenerator to mimic as many of
these flow properties as we could. The remaining determinant factor in the cooler performance
should be the change in matrix material capacity. The cold finger, with the larger cold
regenerator, is shown in Figure 4.
12 K STIRLING PRECOOLER FOR A 6 K HYBRID CRYOCOOLER 67
Figure 5. Thermal performance of the breadboard cooler with lead

ball regenerator and midstage (mid) and cold tip (ct) heat loads as
indicated.
The key results of our tests are shown in Figure 5. The cooler had the same pressure ratio
and input power as before. But as a result of the substitution of the lead ball regenerator, there
was a substantial shift up in no-load temperature at the midstage of approximately 5 degrees
Kelvin, and a drop in temperature of approximately 2 K at the cold tip. Both shifts were qualita-
tively expected, but quantitatively the improvement at the cold tip seems to be too small. The
midstage shift suggests the loss of 300 mW of indirect refrigeration, presumably by the regen-
erator loss, but the cold tip seems to have improved only by roughly 100 mW. We initially
suspected that flow restrictions at the midstage may have reduced the gross capacity at both
stages and explained this change, but subsequent testing with larger channels showed that not to
be the case.
Our auxiliary trend studies now indicate a direction for improvement, but are also some-
what contradictory, which means more work is needed before the low-temperature performance
will be understood. As before, the trends indicate that the cooler is regenerator limited, in that it
does not respond to increased pressure and input power with more refrigeration. But the cold tip
responds strongly to compressor stroke to the end of its range, suggesting that the next step is to
increase the compressor’s displacement relative to the existing hardware. Conversely, the cooler
does not respond to displacer stroke, suggesting ways to eliminate dead volume.
One trend study with the lead regenerator was of special interest. As discussed in the intro-
duction, there are solid arguments for the need to operate a cooler at low frequencies for effi-
cient operation at low temperatures. This motivated a series of tests in which we maintained the
same strokes and pressure, but dropped the frequency. The result for the cold tip temperature (at
a load of 100 mW) is shown in Figure 6. There is a general improvement as we go to lower
frequencies, but it is not particularly strong.
UPCOMING WORK
Because of the previous results, we are currently rebuilding our compressor with increased
displacement. We also are making a number of small changes to the displacer hardware to wring
out some amounts of cold dead volume. The goal is to find the right proportions for best perfor-
mance. Regenerator work up to this point has focused on the cold regenerator. We expect to also
modify the upper regenerators to both improve the heat lift in the vicinity of 40 K, and if pos-
Figure 6. Thermal performance as a function of drive frequency for

/the breadboard cooler with a lead ball regenerator.
sible, reduce dead volume to enhance the performance of all stages. We will explore whether
this be best done with modified screen mesh regenerators, or with powders of more exotic mate-
rials. The goal is to develop the best cooler possible by modifying the regenerators on our bread-
board hardware.
ACKNOWLEDGMENTS
This work is supported by a grant from the Explorer Technology Initiative program at the
Goddard Space Flight Center, NAS5-99239. We also would like to thank Steve Castles at GSFC
for his permission to use hardware developed on the 30 K multistage cryocooler program in
support of this work.
REFERENCES
1. Gully, W.J., Lester, J., Levenduski, R.C., Simmons, D.W., Wright, G.P., Tomlinson, B.J., Davis,
T.M., and Reilly, J., “Rotary Vane Compressor Development for a 10 K Cryogenic Cooler,” Cryo-
coolers 11, Plenum Press, New York (2001).
2. Bradshaw, T.W., Orlowska, A.H., Jewell, C., Jones, B.G., and Scull, S., “Improvements to the
Cooling Power of a Space Qualified Two-Stage Stirling Cycle Cooler,” Cryocoolers 9, Plenum Press,
New York (1997), pp. 79-88.
3. Keung, C., and Lindale, E., “Effect of Leakage Through Clearance Seals on the Performance ofa 10
K Stirling-Cycle Refrigerator,” Proc. of the Third Cryocooler Conference, N.B.S. Pub. 698 (1985),
pp. 127-134.
4. Gary, John, Daney, David E., and Radebaugh, Ray, “A Computational Model for a Regenerator,”
Proc. of the Third Cryocooler Conference, N.B.S. Pub. 698 (1985), pp. 199-211.
5. Clad Metal Industries, Inc., 40-T Edison Ave., Oakland, NJ 07436 USA.
Thermodynamic Optimization of Multi-Stage
Cryocoolers
C. S. Kirkconnell and K. D. Price

Raytheon Electronic Systems
El Segundo, California, USA 90245
ABSTRACT
Active Stirling class cryocoolers, including pulse tube coolers, are complex, difficult to
optimize machines. The large number of characteristics and properties associated with
geometry, materials, gas properties, heat transfer devices, flow manifolds, mechanical
mechanisms, and electro-mechanical devices that determine a particular machine’s performance
make quick optimization difficult. Single-stage coolers are now sufficiently well understood that
design optimization is reasonably straightforward. However, multi-stage coolers compound the
design problem by virtue of the dramatically enlarged number of variables, and optimization is
still a challenge. Often, multi-stage machines are “optimized” by a brute force search of the
design space or design decisions are made based on overly generalized or inaccurate assumptions
about relationships between variables. The schedule-constrained time typically available to
perform optimization procedures combined with the large number of variables and their complex
interaction results in sub-optimal products. This paper presents a concept for optimization that
more rapidly converges on an optimal design.
INTRODUCTION
This paper presents a cryocooler design optimization method based on analysis of exergy
flow. Exergy analysis follows the destruction of energy availability from the machine input to
the low temperature stage(s) where refrigeration is produced.
First, the terms and approach of the exergy method are reviewed and clarified with reference
to thermodynamic analysis of one-stage coolers. Then the analysis is extended to two-stage
coolers. This is followed by a discussion of the optimization technique in one- and two-stage
cryocoolers and, finally, the generalization of the method to multi-stage cryocoolers.
THERMODYNAMIC ANALYSIS
Single-Stage Cryocooler Thermodynamics
A careful examination of the thermodynamics of the comparatively simple single-stage
cryocooler is helpful in describing the operation of the more complicated multi-stage devices.
For that reason, a brief review of fundamental single-stage cryocooler thermodynamics is
presented.
The conventional approach to cryocooler thermodynamic design involves the variation of
design parameters, such as component volumes, regenerator matrix type and dimensions, etc., so

Figure 1. Energy flow schematic for Figure 2. Temperature entropy diagram for a single-stage
single-stage refrigerator; arrow Stirling refrigerator.
denotes energy flow direction.
that the capacity objective is achieved with the maximum thermodynamic efficiency.
ARCOPTR is an example of a widely-used model for this type of optimization.1 Though
challenging in practice because of the complicated and interdependent physical phenomena
involved, this approach is conceptually simple. Consider the energy flow diagram for a single-
stage cooler shown in Figure 1. The conservation of energy dictates that
and the coefficient of performance (COP) is given by
The design parameters are therefore varied such that is maximized for the desired at
The effectiveness of this approach is hampered by an imperfect knowledge of the detailed
internal thermodynamics and the vast trade space afforded by the multitudinous design options.
The primary shortcoming of the above approach is that it provides no figure of merit by
which to evaluate the relative goodness of the achieved. The standard figure of merit, the
Carnot COP is obtained by considering a reversible refrigerator operating between the
temperature extremes of interest, and is shown in innumerable thermodynamics texts to be
defined as follows:
A Second Law COP can now be defined in terms of Eqns. (2) and (3):
In words, Eq. (3) states that is the maximum amount of refrigeration that can be
obtained from a refrigerator operating between and given an input power of This can
be clearly shown on a temperature-entropy (T-s) diagram for a refrigerator operating on a
Stirling cycle (Figure 2). The area bounded by the curves is the net refrigeration produced. The
actual refrigeration is less than the maximum possible (reversible) refrigeration for a refrigerator
operating between the given temperature extremes because of practical inefficiencies
(irreversibilities) which can be roughly grouped into two categories, pneumatic losses and
parasitic losses. Pressure drops, rounding of the corners of the T-s area due to real-world limits
THERMODYNAMIC OPTIMIZATION OF MULTI-STAGE COOLERS 71
over phase control, seal leakage losses, and other such losses that decrease the amount of gross
refrigeration produced are termed here “pneumatic losses.” Conduction losses, radiative loads,
regenerator inefficiency, and other losses that consume a portion of the gross refrigeration
produced are differentiated into another category called “parasitic losses.” Multiple expressions
for the net refrigeration rate arise from these definitions:
Optimum refrigerator performance can also be sought through the method of entropy
generation minimization. Referring back to Fig. 1, it is clear that the rate of entropy generation
is given by
Efficiency is maximized when is minimized. Though intuitively evident from these simple
expressions, it is interesting to note that the equivalence of the First Law and Second Law
methods can be demonstrated by considering the implications of irreversibilities on a
refrigeration cycle from two distinct perspectives. Using the interpretation provided above that
fixes the input power and attributes the difference between the reversible refrigeration rate and
the actual refrigeration rate to the irreversibilities in the system, an expression can be obtained
for the lost refrigeration in terms of the other system parameters:
Using an alternative interpretation in which the refrigeration rate is fixed and the irreversibilities
manifest as an increase in the actual required input power above that required for a reversible
system, i.e.,
a similar expression to Eq. (7) can be obtained for the lost input power:
The algebraic combination of Eqns. (7) and (9) reduces to the simple expression
Using the definition for lost work (power) provided in Bejan and elsewhere, this analysis finally
yields to an expression for the lost refrigeration capacity in terms of the entropy generation:
This expression demonstrates that the First Law optimization approach of minimizing lost
refrigeration capacity is functionally equivalent to minimizing the total rate of entropy
generation, whatever the source. (The reference temperature can be taken as for a
refrigerator rejecting waste heat to the ambient environment.)
All of these concepts of energy flow and irreversibility come together in the exergy flow map
provided in Figure 3, which is modeled after the techniques demonstrated by Bejan.2 The input
exergy, also called availability, provided by the compressor flows down from the warm end
reservoir to the refrigeration temperature where refrigeration is produced. Exergy is
destroyed through the pneumatic and parasitic loss mechanisms. The exergy content of a power
interaction is equal to the power, while the exergy associated with a heat transfer interaction is a
function of the temperature at which the heat transfer occurs:
Figure 3. Exergy flow diagram for a single- Figure 4. Energy flow diagram for a particular
stage refrigerator. Exergy is destroyed through type of two-stage refrigerator. Other two-stage
parasitic and pneumatic losses. configurations may dictate alternative energy
flow diagrams.
Thus the rejection of heat to ambient carries with it zero exergy. A refrigerator’s
minimum achievable “no load” temperature occurs at the temperature where all of the input
exergy has been destroyed through irreversible losses. Of note is the fact that the distribution of
losses between pneumatic and parasitic is immaterial from the standpoint of optimization
through entropy generation minimization for single-stage refrigerators. As shown in the next
section, this is not the case for multi-stage devices.
Two-Stage Cryocooler Thermodynamics

The thermodynamic analysis and optimization of a two-stage refrigerator follows directly
from the analytical approach used for a single-stage device. The stages of multi-stage
refrigerators can be arranged thermodynamically in many ways; an energy flow schematic for
the type of staging of present interest, a single ambient compressor with the coldest stage
rejecting heat to the intermediate stage, is provided in Figure 4. The conservation of energy
equation is again defined at the boundaries of the refrigerator and is given by
The input power that drives both stages originates at from the ambient compressor.
Therefore, the definitions for both the first stage and second stage Carnot COPs are based upon
the ambient temperature
The standard energy flow diagram used in Figure 4 is somewhat misleading in that it appears to
show an engine residing between and and this leads to the temptation to use in Eq.
(14a). This is incorrect for the single compressor problem. If one were to use it can be
shown that the perfectly reversible system yields negative entropy generation, not zero, which
violates the Second Law of Thermodynamics. In contrast, if a multi-stage refrigerator were to be
considered in which individual compressors operating between each temperature level were used
to pump heat between those adjacent temperature reservoirs, then the Carnot COP definitions
would be based upon the temperatures bounding each stage.
The input power can be conceptually partitioned into the individual portions required to drive
each stage. The reversible input power can thus be defined as
where and are the actual net refrigeration rates. The reversible input power for each
stage represents the minimum theoretical power required to deliver the prescribed refrigeration
rate between and the refrigeration temperature. The irreversible, or lost, power can be
similarly partitioned, and the approach demonstrated for the single-stage cooler in developing
Eq. (10) leads to similar looking expressions for the two-stage device:
The input power lost due to irreversibilities in the first stage represents an overall system loss,
but some of the second stage loss, in particular that portion due to parasitic losses between
and is partially recoverable. The energy transfer from to through conduction losses
and regenerator enthalpy flux due to heat exchanger inefficiency decreases the capacity at the
second stage, but it increases the capacity at the first stage by the same amount. Therefore, the
parasitic loss portion of is partially recovered:
Note that the recoverable fraction of the second-stage power lost due to parasitics is given by the
ratio of the Carnot efficiencies,
The thermodynamic intricacies of the two-stage cooler analysis are captured in the exergy
flow map provided in Figure 5. The flow of input exergy from ambient and its partitioning
between the refrigeration stages illustrates the proper selection of in Eq. (14a) as the warm
end reference temperature. The map also shows that the second stage parasitic loss carries with
it positive exergy, that a portion is destroyed in the transferring of capacity from the colder stage
to the intermediate stage, and that the remainder represents the recoverable exergy. The
recoverable exergy can be shown to be equal to the recoverable power defined in Eq. (17) by
considering the net exergy change due to the second stage parasitic loss component:
Figure 5. Exergy flow map for two-stage cryocooler. Parasitic heat transfer from middle stage to cold
stage increases the capacity at and decreases the capacity at by the same amount. This represents a
net exergy loss to the system.
The right hand side can be expressed in terms of the Carnot efficiencies:
This expression reduces to
where the second term on the right hand side represents the exergy destruction and the first term
is the recovered exergy. Note that the right hand side is always negative; although there is partial
recovery, some exergy is destroyed through the second stage parasitic loss.
CHARACTERIZATION AND OPTIMIZATION TECHNIQUES

Single-Stage Cryocoolers
Thermodynamic efficiency is maximized when the rate of entropy generation is minimized.
The optimization of single-stage cryocoolers through numerical analysis typically involves the
variation of design parameters such as component size, operating frequency, and phase angles in
search of a maximum refrigeration capacity-to-input power ratio. This is simply an indirect
method of minimizing entropy generation. For a typical design study, the refrigeration load,
refrigeration temperature, and nominal heat rejection temperature are known, and the input
power required to deliver the needed capacity is being minimized. Referring to the expression
for entropy generation in Eq. (6), the only unknown on the right hand side is so entropy
generation is minimized when the heat rejection rate is minimized. The reversible input power is
fixed by the refrigeration load and the temperature extremes, so is known as well.
Combining Eqns. (1) and (8) yields
in which only is unknown. Therefore, the task reduces to the minimization of

The intent of introducing these Second Law considerations is not to replace the parametric
characterization of design variables in cryocooler optimization, rather the purpose is to improve
the parametric studies by guiding the efforts through an improved understanding of the physics
of the system. The Second Law techniques are particularly useful for characterizing the
efficiency of the cryocooler using as a figure of merit. For example, if a cryocooler is
sufficiently well understood to express the losses in terms of known system parameters, then
determined from the entropy generation calculations, can be used to describe the efficiency of
the cryocooler over a range of refrigeration temperatures. Assume the refrigeration losses can be
approximated by an expression of the form
where
and and are known. (Ideally, the cryocooler is designed such that is minimized
in the vicinity of the design point corresponding to From Eqns. (3) and (10), the total input
power is given by
The Second Law COP follows immediately from the above using Eqns. (2) and (4). Examples of
where this type of characterization might prove useful include design efforts to rescale an
existing cryocooler to increase capacity and thermal system trade studies in which various
cryocoolers are being considered for applications outside their originally intended and better
characterized range of operation.
Two-Stage Cryocoolers
The increased thermodynamic complexity of a multi-stage cryocooler gives rise to a
substantially more challenging task of parametric characterization than for a single-stage cooler,
thus making the practical application of Second Law techniques all the more valuable. Consider
a two-stage cryocooler design in which the refrigeration loads and temperatures at both stages
are prescribed together with the heat rejection temperature. As described above, entropy
generation minimization is equivalent to reducing the total lost input power, which is given by
the algebraic combination of Eqns. (16a), (16b), and (17):
Substituting and combining terms yields
Expressions are needed for the lost refrigeration in the form of the general equation
and for the second-stage parasitic loss
The above expressions assume the parasitic loss between the intermediate and coldest stage is
driven by the temperature difference between those stages, as would be expected for a
conductive or convective loss, but allows for the possibility of more complicated interactions
between the stages in determining the total loss, which includes the pneumatic loss component.
However, since the adjacent boundary temperatures likely dominate the total loss between
stages, Eq. (27) can be simplified:
As an example, consider an application in which a two-stage cryocooler design is being

considered to meet a single prescribed refrigeration load and temperature. No intermediate
refrigeration capacity is required, so the temperature at the intermediate stage is immaterial to the
user. Given this flexibility, the cryocooler designer seeks the optimum intermediate temperature
such that the design capacity at the coldest stage is achieved with minimum entropy generation.
Using a simplified form of the loss equation from Eq. (22) in which it is assumed any non-
linearities with respect to temperature are captured in the first coefficient, the following
expressions for total lost refrigeration are obtained:
For the purposes of this example, assume the parasitic loss is a linear function of the temperature
difference (conduction and convection dominated):
The optimum intermediate temperature occurs where the total lost input power (Eq. 26) is
minimized, i.e., where
The partial derivative of interest is calculated through several sequential applications of the chain
rule, only a few of which are shown below for the sake of conciseness.
The chain rule is applied again for each loss component. For example,
The functions and and the second-stage parasitic loss constant are assumed to be known,
which is consistent with the premise that the performance characteristics of each stage is
reasonably well understood, and it is the optimum combination of those stages into a single
device which is sought. Eventually, through substitution and further application of the chain
rule, the lost power derivative in Eq. (33) is reduced to a singular expression in terms of the
system’s one unknown, The derivative can then be calculated over the temperature range of
interest and the optimum value for thus identified.
DISCUSSION
Multi-Stage Cryocooler Figure of Merit
The application of Second Law principles to cryocooler thermodynamic design helps guide
the design by providing a figure of merit that relates performance to that of an ideal cryocooler.
Eq. (4) defines that figure of merit, for a single-stage cryocooler. For two-stage and other
multi-stage cryocoolers, the concept of exergy flow can be used to define similar figures of
merit. As shown in Eq. (12b), the exergy associated with a heat transfer interaction between a
system and its surroundings is proportional to the inverse of the Carnot COP corresponding to
the temperature at which the heat transfer process occurs. For a cryocooler, the refrigeration
capacities occurring at various temperature levels can be normalized to a single refrigeration
capacity at a single arbitrary temperature, and the resulting normalized efficiency can then be
compared to that of a single-stage Carnot refrigerator. Typically, the temperature at the coldest
stage is used to normalize the capacities because it is the achievement of positive refrigeration at
that temperature level which stresses the design. It follows that the normalized “pseudo single
stage” refrigeration for an n-stage refrigerator is given by the expression
The COP and Second Law efficiency definitions follow directly from Eqns. (2), (4), and (35):
These expressions are useful in comparing multi-stage cryocoolers to each other and to single-
stage cryocoolers, and they can also be used to evaluate the relative efficiency between varied
temperature distributions and heat loads for a multi-stage cryocooler operating. The latter is
essentially a generalized extension of the two-stage example used above where the net
intermediate refrigeration is allowed to vary from zero.
Extension of Exergy Flow Concepts to n-Stage Cryocoolers

The thermodynamic analysis of cryocoolers with more than two cold stages follows directly
from the two-stage cryocooler analysis. Assuming a single ambient compressor is driving the
multi-stage expander, the Carnot efficiency for each stage is based upon the ambient temperature
and the refrigeration temperature for that stage, i.e.,
The concepts of lost power at each stage, the partial recovery of exergy flow due to parasitic heat
transfer losses, and the minimization of total lost power for maximum efficiency all translate
directly from the two-stage cryocooler analysis. The number of terms in the total lost power
expression (Eq. 25) grows with n, so the calculation of the partial derivatives required to
characterize the efficiency of the cryocooler over the range of interest becomes more
complicated, but the fundamental approach is the same. The value of employing these
techniques actually increases as the number of refrigeration stages increases because the
voluminous trade space associated with multi-stage cryocoolers becomes nearly impossible to
adequately characterize with brute force parametric design trades.
CONCLUSION
The introduction of exergy flow concepts into the thermodynamic analysis of cryogenic
refrigerators provides a clear illustration of how internal irreversibilities, which destroy exergy,
are manifested at the system boundary as decreased performance. The decreased performance
can be interpreted as either a loss in refrigeration capacity for a given input power or an increase
in the required input power for a desired refrigeration capacity. The Second Law optimization
approach of minimizing entropy generation (i.e., exergy destruction) is shown to yield particular
advantage for multi-stage refrigerators because the trade space is broad and the thermodynamic
interactions complex. Interestingly, Streich reached a similar conclusion in his exergy analysis
of a quite different problem, the thermodynamic characterization of a mixing process involving
two natural gas streams.3 He concluded, in part, that exergy analysis is particularly useful for
“pioneering” and “systematic studies,” terms which aptly describe the exercise of developing and
optimizing a multi-stage cryogenic refrigerator.
Future work is planned in which quantitative loss correlations for a multi-stage cryocooler
will be substituted into the analytical model described herein to demonstrate the proposed
Second Law optimization approach.
REFERENCES
1. Roach, Pat R. and Kashani, AH, “A simple modeling program for orifice pulse tube cryocoolers,”
2. Bejan, Adrian, Advanced Engineering Thermodynamics, John Wiley and Sons, Inc., New York
(1988), pp. 111-123.
3. Streich, Martin, “Opportunities and limits for exergy analysis in cryogenics,” Chem. Eng. Technol.,
vol. 19 (1996), pp. 498-502.
The Advent of Low Cost Cryocoolers
R.Z. Unger, R.B. Wiseman and M. R. Hummon

Sunpower, Inc.
Athens, OH 45701 USA
ABSTRACT
A new cryocooler, the M87, has been designed. It combines the well-tested thermodynamics
of the M77 cryocooler with features developed to permit the manufacture of large-volume free-
piston (linear) compressors. The price of the cryocooler is projected to be $2,000 per unit in lots of
10,000, with availability in the first quarter of 2001. In comparison to the M77 cryocooler, the
M87 shows enhanced rated efficiency (20%), improved cooling power, and reduced length and
mass. The M87 is rated at 7.5 W lift at 77 K with 150 W(e) input. The enabling market for
development of the M87 is the growing need for portable oxygen therapy for home use. The
development of significant new markets and new products based on the availability of a low cost
cryocooler is anticipated. For instance, in the telecommunications field alone, estimates of the
world-wide market by 2005 range from 120,000 to 400,000 units/year.
INTRODUCTION
Many cryocooler designs have been developed and tested over the past dozen years. In the
absence of a large market for applications of high temperature super-conductivity (HTS), there has
been little impetus to undertake the development work and investment required for mass manufac-
ture of any of these designs. In turn, the lack of low-cost cryocoolers has been an impediment to the
development of products that require such technology.
Production of a low cost cryocooler requires a design appropriate for low cost manufacturing.
Sunpower, Inc. has developed an appropriate design and is now undertaking production in a pilot
manufacturing facility. The new cryocooler, the M87, was designed for high volume manufacture
and is based on a proven thermodynamic design, Sunpower’s M77 cryocooler. The M77, also a
free-piston Stirling cryocooler, is a low volume, relatively high cost unit widely used for aerospace
testing. Extensive experience with other free-piston designs contributed to this effort, especially
work on linear compressors intended for very high volume, low cost white goods.
The first application of the M87 will be in a novel medical device under development by In-X,
Inc. of Denver, Colorado. Their system will supply liquid oxygen to meet the needs of patients
requiring home-based portable oxygen therapy. The well-established, high volume market for
home oxygen therapy provides the enabling market for the initial manufacture of the M87 cryo-
cooler.
This paper describes the features of free-piston technology that support the M87 cryocooler
design, giving examples based on the M77 cryocooler and on linear compressors. The description
of the M87 design includes new techniques that support mass manufacturing (patents pending) and

80 LONG-LIFE TACTICAL AND COMMERCIAL STIRLING COOLERS
are broadly applicable to free-piston technology. Both the equipment and the processes used in
manufactured are outlined. Finally, we speculate on additional markets that are likely to develop
with the availability of reliable, low-cost cryocoolers.
FREE-PISTON TECHNOLOGY
Free-piston designs at Sunpower have a number of features in common. There is no crank
mechanism. The piston moves back and forth within a closely fitting cylinder, typically at a reso-
nant frequency determined by the overall design. The piston and, where present, the displacer,
require no oil or lubrication, but instead are supported by gas bearings using the working gas of the
machine. With non-contact gas bearings and clearance seals, there are no life-limiting wear com-
ponents and the mechanical efficiency is high. Controlling the amplitude of the piston stroke via
input voltage can easily and continuously modulate the output of these machines.1 ,2 Frequency is
constant. A compliant member, together with planar springs and gas bearings, permits the use of
tight but standard manufacturing tolerances.3 An integral linear motor / linear alternator is sealed
within the machine,4 with the permanent magnets attached to a cost effective drawn or rolled sheet
stainless steel part.5 In a Stirling engine, piston motion is converted into AC electricity. In a
Stirling cooler or in a linear (free-piston) compressor, electricity drives the piston motion to imple-
ment the Stirling cycle in the case of a cooler, or the vapor compression cycle or pumping action in
the case of a compressor. The free-piston design is applicable to a wide range of designs and
applications.
M77 Cryocooler
The M77 cryocooler (Figure 1) is an example of a Stirling cooler using free-piston technology.
This cooler was designed for low volume assembly in our research and development lab; neverthe-
less, its cost, $35,000, is well below competing options. The first units of this design were deliv-
ered in 1992 and used for HTSC filter applications. To date, a total of 71 units have been delivered
to university laboratories, private industry, and U.S. government agencies. Test uses include nuclear
detection, earth-orbital weather balloons, oxygen liquefaction, and space applications. The M77
will cool germanium detectors in NASA’s High Energy Solar Spectroscopic Imager (HESSI), a
satellite-based observatory of X-rays and gamma rays from the sun.6 Tests of the M77 by NASA-
Goddard resulted in the following statement: “In extensive studies over the past two years at GSFC,
M77 coolers have been vibrated to the GEVS mandated 14.1 Grms, run under thermal vacuum
conditions from -25C to +30C and life-tested already (continuing) for hrs. Monitoring
during these tests showed no internal contamination of the working gas. Units already tested at
GSFC are fully qualified for flight, and will be used for HESSI.7 ” A related higher temperature
cooler produced under license by Global Cooling Manufacturing, Athens, Ohio, has been incorpo-
rated into commercial equipment to test jet fuel by ISL, Verson, France.8 An older high tempera-
ture cooler, in nearly continuous operation at Sunpower since 1995, has accumulated to date (June
Figure 1. M77 Cryocooler with Passive Balancer.

THE ADVENT OF LOW-COST CRYOCOOLERS 81
Figure 2. Linear Compressor.
2000) just over 40,000 hours of operation without repair, adjustment, or degradation of perfor-
mance. In all, this combined experience proves the general design of free-piston machines and the
material used, and validates the thermodynamics of the coolers.
Linear Compressor
Sunpower has developed a free-piston, fractional horsepower hermetic compressor for use in
domestic refrigeration (Figure 2). The design is intended for high volume mass manufacturing and
low price (approximately $40 each). The free-piston design is appropriate for the long life expected
for refrigeration compressors, 10 to 20 years of service. This reliability will meet the industry need
for negligible warranty rates for the industry standard long warranty period (2 to 5 years). These
compressors have higher efficiency than conventional compressors9 ,10 for several reasons: the de-
sign is mechanically efficient; losses associated with conversion from rotary to linear motion are
eliminated; modulated operation through stroke control eliminates frequent on-off surges, losses,
and wear. Recent analysis using pre-production units demonstrates that the linear compressor is
30% more efficient than conventional units.11 The very low piston side loads permit either oil-less
operation with gas bearings, or the use of low viscosity oil. For manufacturing purposes, the com-
pliance concept2 supports ease of assembly, with no alignment required. This compressor design is
entering the final stages in preparation for mass production, with production units expected in
2001.11 While improved performance is demonstrated, note that cost targets must still be met in
order for the linear compressor to enter widespread use.
M87 Cryocooler Design

To produce the M87 cryocooler design appropriate for manufacture, Sunpower combined its
experience with the thermodynamically robust M77 cryocooler and techniques from the linear com-
pressor design that support manufacturing. Additional intellectual property was also developed for
this project, and will serve other free-piston applications.
Table 1 compares aspects of the M77 and the M87. Figure 3 shows the thermal performance of
the M77. Thermal performance for the M87 is shown in Figure 4. Overall, the M87 reduces mass
by 23%, increases lift at 77K, and improves rated efficiency by 20%.
The M77 is a general-use cryocooler rated at 4 W cooling power (heat lift) at 77 K to 40°C
reject with 100 W(e) input. Heat can be rejected through the use of a cooling loop (liquid) or fins
(air). Cooler orientation during operation is unrestricted. The controller is operated from a DC
Figure 3. Thermal performance of M77 cryocooler; reject temperature 40°C (313K).
Figure 4. Thermal performance of M87 cryocooler; reject temperature 55°C (328K).

Figure 5. A. M87 cryocooler, 3-D cutaway view (left). B. Photograph (right).
source and incorporates a closed loop temperature control. Vibration is controlled through either a
passive unit (a mass-spring combination) or an active unit (a driven linear motor). The M87 is
designed specifically for liquefying oxygen (Figure 5). Cooling power is 7.5 W at 77 K and 10 W
at 87 K with 150 W(e) input. Reject temperature is 55°C into ambient air through fins. The M87 is
orientation dependent (vertical, cold-end down) and has no temperature control. Vibration attenu-
ation is through a passive mass-spring combination. The driver-controller is triac based,12 using 60
Hz 120 V AC for power. The piston is centered for start-up by a DC circuit that lifts the piston to
position.13
To achieve a design suitable for manufacture, the M87 part count, the number of weld/brazed
joints, and the use of polymers and glue joints were each reduced by 50% or more in comparison to
the M77. The M87 uses a single gas bearing system for both the piston and displacer, implemented
by means of a greatly simplified design.14 A new design for the heat exchangers and heat path
improves heat transfer and simplifies manufacture.15 The regenerator is constructed of a random
fiber material.
M87 Manufacturing Facility
To support the development of the manufacturing facility, Sunpower recognized that its exper-
tise and practices were unlikely to support a manufacturing start-up effectively. A separate manu-
facturing facility (6,000 sq ft) was developed with the capacity to manufacture over 30,000 cryo-
coolers per year with three shifts. An experienced manufacturing engineer was hired to oversee
both start-up and manufacturing, and a newly hired, entirely separate team is assigned to M87
manufacturing activities.
The M87 is composed nearly entirely of vendor-supplied parts. These are assembled at our
facility into five major sub-assemblies (one example in Figure 6). Sub-assemblies are then baked
out at controlled temperature under vacuum to remove all contaminating gases, and assembled and
then sealed by plasma welding in dry nitrogen (Figure 7). The coolers are then transferred to an
evacuation / charging station (Figure 8). Table 2 lists the equipment in place to support these
activities. An investment of over $3 million has been required to set up, equip, and debug the
manufacturing facility and processes.
Figure 6. M87 cryocooler sub-assembly.
Figure 7. M87 cryocoolers ready for final weld in glove box.
Figure 8. M87 cryocooler pre-production batch, evacuation and charging station.

Manufacturing Strategy
The general manufacturing plan is to focus on assembly processes and some specialized test-
ing and finishing processes. Purchased parts (over 90% of machined parts inventory) will be sched-
uled and received on a just-in-time basis. We will use a batch process. Batch size, initially, will
equal one month’s production. When the monthly batch quantity exceeds the vacuum oven batch
quantity, the job batch quantities will equal the oven batch capacity. The initial production cycle
for a batch will span two-and-a-half to three calendar months. As a result, there will be multiple
active jobs in the system at any given time in order to maintain monthly shipping schedules. Strict
document control systems and incoming inspection will be used to maximize efficiency, yield,
productivity, and end product quality.
Additional on-line inspection of work-in-process quality control sampling of work in progress
will be used to minimize scrap and rework. Statistical process control will be implemented over
time. Employees will be trained in statistical process control procedures and processes will be
prioritized on the basis of potential benefit.
OTHER APPLICATIONS FOR LOW COST CRYOCOOLERS

The telecommunications industry is seeking ways to enhance cell phone service while reduc-
ing cost by using cooled HTSC components. The M87 can meet some of this need without modi-
fication. The design is also well understood, and can be fairly easily modified to enhance perfor-
mance at other lifts and under differing conditions.16 The manufacturing techniques and intellec-
tual property already developed will allow further manufacturing to occur even more easily. For
the telecommunications market, reasonable projections suggest a market by 2005 for at least 120,000
units per year, and perhaps over 400,000 per year. With the availability of mass-produced, low cost
cryocoolers, other markets are likely to develop.
CONCLUSIONS
A new cryocooler, the M87, has been developed and is being placed into production at a spe-
cially built manufacturing facility. The cooler combines excellent thermodynamic performance
with a volume manufacturing design. The initial and enabling market is a unique oxygen therapy
device. Production units will be available in the first quarter of 2001. Price in lots of 10,000 is
expected to about $2,000 per unit. With the availability of mass-produced cryocoolers, it can be
predicted that other applications will develop for the M87 itself, and for related designs optimized
for other specifications, including the telecommunications market.
ACKNOWLEDGMENT
The authors and Sunpower thank Charles Henry of In-X, Inc., for his early and continuing
support for manufacture of the M87 cryocooler.
REFERENCES
1. U.S. Patent 5,342,176; Method and Apparatus for Measuring Piston Position in a Free Piston Com-
pressor, 8-30-1994.
2. U.S. Patent 5,496,176; Method and Apparatus for Measuring Piston Position in a Free Piston Com-
pressor, 3-5-1996.
3. U.S. Patent 5,525,845; Fluid Bearing with Compliant Linkage for Centering Reciprocating Bodies,
6-11-1996
4. U.S. Patent 4,602,174; Electromechanical Transducer Particularly Suitable for a Linear Alternator
Driven by a Free-Piston Stirling Engine, 7-22-1986.
5. U.S. Patent 5,642,088; Magnet Ring Support, 6-24-97.
6. http://hesperia.gsfc.nasa.gov/hessi/; http://ssl.berkeley.edu/hessi/
7. http://hessi.ssl.berkeley.edu/instrument/cryocooler.html
8. http://www.isl-france.com/fpp_more.htm
9. Van der Walt, N.R., and R.Z. Unger, “Linear Compressors—A Maturing Technology,” Proceedings
45th International Appliance Technical Conference, University of Wisconsin, Madison, Wisconsin
(May 1994).
10. Unger, Reuven, “Development and Testing of a Linear Compressor Sized for the European Market,”
Presented at International Appliance Technology Conference, Purdue University, West Lafayette, In-
diana, May 10-12, 1999. Available at: http://www.sunpower.com/tech papers/pub74/iatc99.html
11. Lee, H.K., G. Y. Song, E.P. Hong, K.B. Park, J. Y. Yoo, and W.H. Jung, “Development of the Linear
Compressor for a Household Refrigerator,” to be presented at the 2000 International Compressor
Engineering Conference, Purdue University, West Lafayette, Indiana. (July 25-28, 2000).
12. U.S. Patent 5,592,073; Triac Control Circuit, 1-7-97.
13. DC Centering of Free Piston Machine, U.S. Patent application pending.
14. Gas Bearing and Method of Making a Gas Bearing for a Free Piston Machine, U.S. Patent applica-
tion pending.
15. Heat Exchanger and Method of Constructing Same, U.S. Patent application pending.
16. SAUCE, Sunpower’s proprietary Stirling design software.
Performance and Reliability Improvements
in a Low-Cost Stirling Cycle Cryocooler
M. Hanes
Superconductor Technologies Incorporated
Santa Barbara, CA 93111
ABSTRACT
The use of a free piston, gas-bearing Stirling cycle cryocooler for commercial high
temperature superconductor (HTS) applications dictates the cooler must not only be low cost, but
also have long life and high reliability. Over the past two years Superconductor Technologies
Inc. (STI) has integrated a new cryocooler design into our commercial HTS systems. This cooler
is an improvement over the previous cooler we were using in this application however, further
improvements were still attained with this new design, which presently has achieved a 1/COP of
<17 w/w at 77K cold end temperature and 23°C ambient temperature. This paper will describe
the performance, life, reliability and manufacturing cost improvements made to the cooler design
over this time period. Data will be presented detailing the performance of the cooler over an
ambient temperature range of 20°C to 70°C at various operating conditions. In addition to
laboratory results, a Weibull analysis of life and reliability data from over 140 units in the field
with a combined runtime of over 300,000 hours will be presented.
INTRODUCTION
Superconductor Technologies Inc. (STI) has as its primary product high temperature
superconductors (HTS) which are utilized as filters in cellular applications to provide increased
performance over conventional filters. One of the ramifications of utilizing HTS in the filters is
the necessity of having to maintain the HTS at cryogenic temperatures. There are a multitude of
cryogenic refrigerators which could be considered for this application; STI chose to use a Stirling
cycle, free piston, linear motor design. This design provides for a compact, efficient, long life
system without the need for helium lines connecting the cold end to the compressor, as with a
Gifford McMahon type cooler. If the application involves mounting the HTS products on a
tower, or other remote locations, the elimination of the helium lines becomes a more significant
advantage. The Stirling cycle cooler designed for this application must have a long life to allow
the cryogenically cooled filters to compete effectively with the conventional, ambient
temperature filters.
The performance requirements for this application were determined to be 4 watts of lift at
room ambient with less than 100 watts input power, an operating range of -40°C to 60°C, a mass
of less than 5 kg. The target for the cooler life is 60,000 hours with no required maintenance.

DESIGN OVERVIEW
The cooler which STI is currently manufacturing and incorporating into its HTS filter
systems consists of the following design features:
Integral design – the compression and expansion volumes are contained within a common
pressure vessel, as opposed to a split design, where the two volumes are connected by a transfer
line.
Motor – the cooler uses an internal coil linear motor with radially magnetized Neodymium –
iron – boron magnets.
Displacer – is of the “free piston” design where a pressure difference between two volumes
acts on the cross sectional area of a drive piston to provide the correct displacer motion.
Regenerator – a non metallic material is used for the regenerator in this design. This
provides an inexpensive regenerator which performs well down to below 50 Kelvin.
Gas bearing – a gas bearing is utilized to prevent the compressor piston from coming into
contact with the compressor cylinder. This eliminates both the wear and debris generation which
would result if there is contact between these surfaces.
Variations from previous design. The most significant change was to the linear motor.
The previous design utilized an external coil with Samarium cobalt magnets and the current
design uses an internal coil with magnetized Neodimium – iron – boron magnets. In addition to
these fundamental differences in design, the new motor has superior performance. The
efficiency of the new motor is approximately 40% higher than the old one, resulting in the same
cooler performance with only 100 watts input power as opposed to 140. This is achieved in part
from the internal coil, which allows smaller air gaps in the motor’s magnetic circuit. This also
allows for the use of smaller magnets, which reduces cooler cost. Another factor for the
increased efficiency is the use of individual laminations for the internal motor iron, as opposed to
a solid iron with machined slots. The individual laminations greatly reduce the eddy current
losses within the internal iron, which not only reduces input power, but decreases the internal
temperature of the cooler. Both designs utilized individual laminations for the outer iron.
Another advantage of the latest motor is the lower amount of magnetic side force for a given
amount of misalignment between the motor components. This is especially critical on a design
which utilizes gas bearings as the means for preventing contact between the piston and the
cylinder. Figure 1 shows the relative magnitudes of the side force vs. the offset. This lower side
force not only increases the effectiveness of the gas bearings, but it allows use of less precision
parts and for simpler and less time consuming assembly of the cooler, which results in a lower
cost cooler.
Lastly, the force constant of the new motor is more constant over the displacement of the
motor along its axis. This results in a cooler which is easier to design driver electronics for and
to use the motor’s back emf to approximate the location of the piston. The difference in the force
constants are shown in Figure 2.
GAS BEARING DESIGN PRINCIPLES

In order to overcome the magnetic side force and weight of the motor this cooler utilizes a
gas bearing scheme to ensure long cooler life. The gas bearing eliminates virtually all contact
between the compressor piston and the compressor cylinder, hence eliminating friction and wear.
The piston essentially floats on a thin layer of helium gas, which is the same gas used as the
working fluid for the thermodynamic processes within the cooler. A cross section of the piston -
gas bearing - cylinder assembly is shown below in Figure 3. During cooler operation, the high
pressure reservoir is kept at a relatively constant and high pressure by the action of the check
valve. During the portion of the cycle where the working pressure in the warm end of the cooler
IMPROVEMENTS IN A LOW-COST STIRLING CRYOCOOLER 89
Figure 1. Motor magnetic side force vs. magnet offset.
Figure 2. Motor force constant profiles.

Figure 3. Typical piston gas bearing layout.
is higher than the pressure of the high pressure reservoir, helium flows from the warm end into
the reservoir and “recharges” it. During the time when the warm end pressure is lower than the
reservoir pressure, the check valve is closed, preventing helium from escaping from the reservoir.
During the entire cycle, helium is flowing from the reservoir through the piston flow restrictors
and into the bounce volume. The three pressures within the system are shown in Figure 4.
As shown on the graph, all the pressures initially start at the same level. As the cooler
begins to run, the pressure in the reservoir begins to pump up to an almost constant level. The
magnitude of the fluctuation in the reservoir pressure is a function of the reservoir volume and
the piston flow restrictor flow rates. Therefore, if these parameters are designed correctly, the gas
bearing will operate over an almost constant pressure difference, in spite of the oscillatory nature
of the pressure in the warm end, or working, volume of the cooler.
Figure 5 shows an expanded piston - cylinder gap to illustrate the principles of a gas bearing
supported piston. A piston which is supported by a gas bearing will have the flow resistance of
the piston flow restrictor approximately equal to the flow resistance of the annular gap between
the piston and the cylinder, when the piston is centered in the cylinder. This results in the
Figure 4. Pressures in various cooler volumes.

Figure 5. Gas bearing with piston off axis.
pressure in the gas bearing pad being approximately halfway between the reservoir and the
bounce volume pressures. Also, when the piston is centered, the pressures in the pads are equal
on all sides of the piston, and there are no net bearing forces acting on the piston. However,
when the piston is forced off center, as depicted in the above drawing, the resistance of “gap 2”
becomes lower than that of “gap 1” and the pressure in the gas bearing pad associated with gap 1
increases (becomes more closely coupled to the higher pressure reservoir), while at the same time
the pressure in the pad on the opposite side decreases (becomes more closely coupled to the
lower pressure bounce volume). This results in a pressure difference between the two sides of
the piston, which act upon the projected area of the piston to provide a centering force. Since the
flow resistance of the gap is proportional to the inverse of the gap width cubed, large pressure
differences will exist for very small piston offsets. This self centering gas bearing will have a
spring constant in the range of 10,000 lb/in per set of gas bearings, which is adequate to prevent
piston to cylinder contact, and ensure the longevity of the cooler.
IMPROVEMENTS TO CURRENT DESIGN

Since the incorporation of the new cryocooler design there have been several improvements
to the design which are intended to improve the efficiency and/or the reliability of the cooler.
Two of the most significant changes are discussed below.
Piston centering port modification. During operation there is a continuous flow of helium
out of the warm end volume of the cooler, into the gas bearing reservoir, out the gas bearing
restrictors and into the bounce volume (see Figure 3). If this flow was not compensated for, the
average pressure in the bounce volume would become higher than the average pressure in the
warm end volume. The effect of this would be to bias the piston toward the warm end volume
and cause it to hit the compressor cylinder at an unacceptably low piston displacement (or input
power). To offset this flow of gas from the working volume to the bounce volume and
“centering port” circuit is used. This circuit allows gas to flow back from the bounce volume to
the working volume. In the initial design, under certain operating conditions, this circuit was not
allowing enough gas to return to the working volume. This resulted in the cooler knocking
(piston hitting the compressor bore) at a low input power. The centering port circuit was
modified to keep the average pressures on the two sides of the piston more closely matched. The
results of this modification are shown in Figure 6. The vertical axis shows the average (or
midstroke) location of the piston as a function of input power. As can be seen, with the “no
mod” centering ports, the piston drifts toward the bore as soon as power is applied to the cooler.
compressor bore at the extremity of its displacement. With the modified centering port circuit,
the piston initially drifts away from the compressor bore, until at 80 watts, it has returned to the
zero power location. Under the conditions which the cooler normally operates, 80 watts is
typically the max power required and the redesigned centering ports have the piston centered in
the motor at this point, which provides the highest operating efficiency. When higher input
power is required, such as at high ambient temperatures, dewar degrade or cooler efficiency
degrade, the piston will not hit the compressor bore before the motor reaches its maximum input
power. The details of this design are not shown here as a patent has been applied for and not yet
issued.
Cold end efficiency improvement. A key component to the overall efficiency of a cryocooler is
how well heat is transferred from the heat load to the helium in the cold end of the cooler. The
more efficiently this heat transfer occurs, the lower the temperature difference between the heat
load and the helium in the cold end will be. This results in the helium not having to be as cold,
and, hence, a more efficient cooler. An effort was undertaken to increase this cold end heat
transfer efficiency. There are essentially three resistances to this flow of heat; 1) the interface
resistance between the heat load and the heat acceptor portion of the cold end, 2) the conduction
resistance across the heat acceptor, and 3) the convection resistance from the heat acceptor to the
helium gas inside the cold end. It is this last resistance which is the dominant term in the overall
resistance, therefor, this was the area which was to be improved upon. This resistance is defined
as 1/(hA), where h = convective heat transfer coefficient, and A = heat transfer area. Increasing
either of these parameters will reduce the heat transfer resistance. The modifications which were
made to the cold end increased both of these parameters. Again, the details of this design are not
shown here as a patent has been applied for and not yet issued.
Regenerator material change. This cooler used a non metallic material for the regenerator.
Although this is an inexpensive method to achieve an effective regenerator, the variation in the
characteristics, particularly the diameter, of the purchased material was causing variations in the
Figure 6. Piston drift vs. input power.

performance from cooler to cooler. This occurred as a result of the variation of the regenerator
effectiveness and from the variation in pressure drop across the regenerator, which, in a “free
piston” displacer cryocooler, will have an effect on the compressor piston to displacer phasing,
which affects cooler efficiency. An effort was made to find a similar material which would not
only be more consistent, but also increase the regenerator effectiveness. In order to increase the
regenerator effectiveness a material with a smaller diameter was desired, as this would increase
the surface area for a given regenerator mass. A new material was found, and it did meet the
requirements which were desired. The results of the new regenerator material, cold end
efficiency improvement and the modified centering port design are shown in Figure 7. The
modification to the centering ports allow the cooler “with mods” to run at the higher power and
the new regenerator material and higher efficiency cold end account for the higher lift. At 100
watts input power the new regenerator material accounts for ~0.5 watt increase and the high
efficiency cold end adds the other 1.5 watt increase.
COOLER RELIABILITY
The goal for life is over 60,000 hours (5 years) of continuous operation with no maintenance
requirements. We currently have about 140 units in the field with a combined run time of over
490,000 thousand hours with no failures. Figure 8 shows a Weibull analysis of the coolers in the
field as of 5/9/00. This analysis predicts a 56,000 hour characteristic life based on the current
data. The slope of the curve on the Weibull graph (beta) is based on the types of failures which
have occurred. Since there are no failures to determine this value, it is based on the failures of
the previous design. Although this analysis is not complete, it does show that the goal of a
60,000 hour cooler life is attainable.
SUMMARY
Cooler performance. Table 1 is a summary of the cooler characteristics for a range of
parameters and Figure 9 shows the performance of the cooler at 23 °C and 60°C ambient
temperature over a range of input power. The COP of the cooler at 100 watt input and 23° C is
0.063 watts lift / watt input power. The Carnot efficiency at these conditions is 0.35, which
yields an efficiency of 18% of Carnot.
Figure 7. Cooler lift vs. input power, current and original design.
Figure 8. Weibull analysis of current coolers in the field.

Figure 9. Cooler performance summary graph.
Conclusion. The cooler as originally designed provided adequate performance for the
application it was intended to be used in, however the performance improvements not only
reduce the power required by the HTS system, it also increases the life of the cooler in the event
of failure mode which results in a slow degrade in performance. Future improvements to the
cooler include incorporating a gas bearing system on the displacer. Although very little wear has
been observed on the displacer in the current design, it is the only part in the cooler which has
moving contact during operation.
Development of a Long-Life Stirling Cryocooler
Y. Ikuta, Y. Suzuki, K. Kanao and N. Watanabe

Research and Development Center
Sumitomo Heavy Industries, Ltd.
63-30, Yuhigaoka, Hiratsuka, Kanagawa, 254-0806, Japan
ABSTRACT
A Stirling cryocooler for high temperature superconducting devices has been developed. High
efficiency and long lifetime are emphasized in the development. Its performance is 6 W at 70 K
with 150 W input under 23°C ambient conditions; at 50°C ambient, it produces 5 W at 70 K with
150 W consumption. The mean time before failure for the design is over 40,000 hours. Flexure
springs are used to support the compressor cylinders and the displacer in the cold head to realize
this lifetime. The flexure springs are designed using finite element modeling (FEM) to reduce the
stress to below the fatigue limit.
INTRODUCTION
High Temperature Superconducting (HTS) filters (as used in ground base stations for cellular
phone systems) are passive devices and must be cooled to 60-80 K. The needed cooling power is
expected to be 2-5 W at 70 K. Over 5 years lifetime is required.
Sumitomo Heavy Industries is a leading Stirling cryocooler manufacturer in Japan. The cool-
ers have been developed and manufactured since the 1980s, and are mainly used for cooling sen-
sors, especially infrared sensors. The sensor systems require a cooling power of 0.3-1.5 W at 80 K.
In 1998, development of a new cryocooler with more cooling power and high reliability was initi-
ated for the new HTS markets.
SPECIFICATION AND DESIGN

The main cooler specifications are listed in the Table 1.

To achieve these specifications, the development process was divided into two phases. In the
first phase, the proto1 cooler was designed using a combination of a conventional linear motor and
a conventional free-displacer cold head, and without using long-life technologies; it was made to
examine the cooling power of the design. In the second phase, the proto2 cooler was made, adding
long-life technologies and a high efficiency linear motor to the protol design.
Cooling Power Design

The target of the cooling power was set at 5 W at 70 K under 23°C ambient conditions. This is
the expected requirement for HTS devices. Design calculations were based on Walker’s equations,
Schdmit’s equations1 and our experience. The results are considered as the average cooling power
at the beginning of life. The guaranteed cooling power will take into consideration the results of
lifetime qualification and the variation of cooling power in mass production.
The target for the input power was set at less than 150 W. The reason is the difficulty of heat
rejection from the cooler surface to the atmosphere. In the test of the proto1 with 250 W input, the
surface temperature of the cooler was 50°C higher than the ambient. Generally speaking, if the
cooler is designed smaller, the input power becomes larger. That makes the heat rejection more
difficult and the heat sinks larger. In other words, a low input power cooler makes the heat rejection
of the whole system easier.
Mechanical Design for Longer Lifetime

Figure 1 illustrates the overall mechanical design of the cooler. The main factors affecting
cooler lifetime include: piston and displacer seal wear, gas contamination, and gas leakage.
Prevention of Seal Wear
In the proto2 design, a linear motor was incorporated to reduce the load on the piston seal. The
moving parts in the proto2 compressor consist of the cylinders and coils supported by flexure
springs at both ends of the moving elements.2 This structure has the advantage of keeping the gap
between the pistons and the cylinders constant, but also has the disadvantage of making the system
heavier. From the viewpoint of reliability, the moving cylinder structure was adopted. The dis-
placer in the cold head is also supported by flexure springs, which are located at two points along
the rod that extends to the room temperature side.
Figure 1. Schematic of the proto2 cooler.

DEVELOPMENT OF A LONG-LIFE STIRLING CRYOCOOLER 99
The flexure springs were designed using FEM to reduce the stress to below the fatigue limit.
Ordinary fatigue characteristics of materials are generally tested to less than cycles. On the
other hand, our lifetime target is 40,000 hours; that means over cycles. In the region over 107
cycles, what happens is unknown. Conventional fatigue design theory has not yet been evaluated in
this region. Therefore, fatigue tests were carried out with various flexures. A variety of shapes of
springs with various stresses were tested. All the spring were designed using the same method, and
manufactured using photo-etching.
In the flexure life test, linear motors are used to force the springs to displace at the same
amplitude and speed as in the cooler. The tests are being performed in atmosphere, at room
temperature, and the experimental apparatus is cooled by a small fan to avoid overheating. There is
no vibration except the basic mode. To date, all ten samples of the flexure springs have survived
one year of running, and the test is still ongoing. Other flexure springs with higher stresses broke
earlier as expected.
Prevention of Gas Contamination

Gas contamination can result from outgassing from internal cooler components, especially
from nonmetallic materials used as insulators. In our design, two directions have been tried to
reduce outgassing. One is minimization of the use of nonmetallic materials; the other is avoiding
overheating. The latter direction is based on the thought that the hotter the inside of the cooler, the
more gases come out of the materials. Therefore, reducing the input power to the linear motors was
tried.
Prevention of Gas Leakage
The problem of gas leakage has been addressed by using a double seal structure. If this is not
enough, a welded structure will be adopted as an alternative.
CHARACTERISTICS OF THE COOLERS

In this section, characteristics of the developed coolers are described. Measurements of the
dependency of thermal performance on ambient temperature were performed after first making
sure that the surface temperature of the compressor was equal to the ambient temperature. Figure 2
is a photograph of the proto2 cooler in the thermal chamber.
Figure 2. Photograph of the proto2 cooler in the thermal chamber.

Figure 3. Cooling power of the proto2 cooler.
Figure 4. Ambient temperature dependency of the cooling power at 60 K.
Ambient Temperature Dependency

The experimental results, measured according to the test conditions in Table 2, are summa-
rized in Fig. 3. The cooling power is the heat dissapated by an electric resistance heater on the cold
stage; measurements were made using the 4-wire method. The design cooling power is 5 W at
70 K, but the achieved performance is 6.0 W at 70 K, with a COP of 3.9% (23°C ambient).
In Figs. 4 to 6, the relationship between the cooling power and the ambient operating tempera-
ture of the proto1 and proto2 coolers is shown. From these results, the proto2 cooler is seen to have
DEVELOPMENT OF A LONG-LIFE STIRLING CRYOCOOLER 101
1.5 times more cooling power than the proto1 cooler at 23°C. The reasons responsible for this
improvement are the reduction of the dead volume in the cold head and improvements made to the
linear motor. The effect of the dead volume reduction is confirmed by our design calculations.
Also, the cooling power in cold ambient environments was improved; this relates to the thick-
ness of the compressor piston clearance seal. The clearance seal of the proto2 cooler is thinner than
that of the proto1 cooler.
CONCLUSIONS AND FUTURE WORK

The development target for cooling power has been satisfied. The COP data for the cooler are
listed in Table 3, where the COP is defined at each temperature as the ratio of the heater input
power to the input power to the cooler (not to the drive electronics).
From Table 3, it is seen that world class performance was achieved. The mechanical design
has also been verified via the flexure spring fatigue test.
Measuring the cooler's resistance to contamination remains to be done. After this, the design
lifetime of 40,000 hours will have been established.
REFERENCES
1. Walker, G., Cryocoolers, Plenum Press, New York (1983).
2. Aubon, C.R. and Peters, N.R., “Miniature Long-life Tactical Stirling Cryocoolers,” Cryocoolers 9,
Plenum Press, New York (1997), pp. 109-11
Flexure Springs Applied to Low-Cost
Linear Drive Cryocoolers
R.M. Rawlings and S. Miskimins

DRS Infrared Technologies, L.P.
Dallas, Texas USA 75243
ABSTRACT
Flexure spring suspensions have demonstrated the ability to provide long operating lifetimes
for cryocoolers intended for space-based applications. Insertion of this technology into coolers
intended for tactical or commercial application has been slow due to cost considerations. This
paper describes the development and testing of a flexure spring system for small tactical cryo-
coolers that provides a doubling of operating life while costing approximately the same as the tra-
ditional helical coil spring suspension system.
The flexure spring system described in this paper successfully achieves the high radial stiff-
ness characteristic of the flexure spring design in a low-cost package. In addition, the concept has
been implemented in cryocoolers weighing less than a pound and smaller than a soft drink can.
This design has been qualified for use in U.S. Army applications. Qualification and life test data
is presented to demonstrate the robustness of the design in tactical environments. The producibil-
ity of the design is evidenced by the on-going production of these coolers for various applications.
INTRODUCTION
For a number of years the development of long-life cryocoolers for spacecraft applications
has been reported in the literature.1,2 These coolers have all shared a design heritage from the
Oxford Stirling cooler; namely flexure springs or bearings as they have been called. The specific
implementation or configuration varied from investigator to investigator, but the use of these
springs to suspend the moving masses within the compressor was universal. Also consistent from
unit to unit was the high cost of producing these “space” coolers, some perhaps approaching the
$1-million figure.
At the same time, tactical cooler developers were designing and producing linear drive cryo-
coolers for sensor systems designed for use in day-to-day battle operations. These units have
shared design philosophies that emphasize low weight, low input power, and low cost. Lifetime
requirements have been an order of magnitude less than the requirements imposed on the space
coolers. These tactical cryocoolers have provided improvements in performance, induced vibra-
tion, audible noise, and operating life when compared to the earlier rotary cooler designs they
replaced. These improvements, however, came with an increase in acquisition costs.
In 1995 the U.S. Army, through the Night Vision and Electronics Sensors Directorate
(NVESD) began a Linear Cooler Manufacturing Technology program with DRS Infrared Tech-

nologies (a part of Texas Instruments at that time) to reduce the cost of linear drive coolers for
U.S Department of Defense programs. The goal of the program was to reduce the acquisition cost
of linear coolers by 30 percent.
As a part of on-going development at DRS Infrared Technologies, initial prototypes of flex-
ure spring compressors with the same form factors as standard tactical coolers had been designed
and built. It was proposed that a flexure spring compressor for tactical cryocoolers would offer a
life cycle cost reduction due to the increase in life that it would provide. The potential for main-
taining or even reducing the acquisition cost was also recognized, if the design were approached
accordingly. The flexure spring compressor initiatives were added to the Manufacturing Technol-
ogy program with the specific objectives of developing flexure spring compressors for the Javelin
system’s Command and Launch Unit (CLU) and the U.S. Army’s Standard Advanced Dewar
Assembly (SADA)-II
The balance of this paper describes the design considerations, the resulting designs, and test
results for those designs. These results, coupled with the on-going production for these programs,
illustrate the success of the effort.
DESIGN CONSIDERATIONS
The cooler development objectives required that a number of factors be considered during the
design phase. As is the case in any design, some of these factors tend to be mutually exclusive.
Some carry more weight than others, and the importance of each factor may vary as the design
progresses and additional issues arise and are addressed. The primary considerations in undertak-
ing the designs for these two coolers included:
• Ease of Assembly – During the design phase an overriding consideration was the ease of
assembly. These cooler designs have been developed for production programs, not prototype
or single unit installations. As an example, the Javelin program will build and deliver nearly
5,000 nightsight systems in a six-year period. Cooler deliveries to support this program will
require 70 coolers per month- nearly 20 per week. Such rates cannot be attained or sustained
if the subassemblies and final assembly require significant effort to assemble, align, or adjust.
• Compatibility with existing designs - DRS Infrared Technology has been supplying linear
drive cryocoolers since the mid-1980s to a variety programs. There is now a significant
installed base of these coolers. The using programs have developed systems around these
designs; allowing a specific space envelope and weight budget, and providing well-defined
electrical and thermal interfaces to the cooler. Design modifications to these coolers must
maintain a backwards compatibility with these fielded systems. New cryocooler designs can-
not require more space, different electrical characteristics, or more thermal management. This
forces the flexure spring cooler design to be form, fit, and function interchangeable with the
previously fielded helical spring designs.
• Less than benign installations – Tactical cryocoolers are not destined to spend their lifetimes
quietly producing refrigeration in the corner of some laboratory. They are developed to pro-
vide cooling in systems that will be subject to wind buffet at Mach 1, tracked vehicle vibra-
tion, roof mounted on an off-road personnel carrier, or slung over the shoulder of infantry
personnel as they work their way through rugged terrain. Above all else, the flexure spring
design for tactical cryocoolers had to consider the handling and operational scenarios to which
the cooler would eventually be subject. The design had to take into account mechanical
shocks, operation in extreme ambient temperatures, exposure to long-term external vibration
sources, and handling in battlefield conditions. The design must be mechanically robust.
• Cost Impact – As second-generation night vision systems have matured, investments have
been made by the using community, most notably the U.S. Army Night Vision Laboratory and
Defense Advanced Research Projects Agency (DARPA), along with the component manufac-
turers, to drive down life cycle costs, including cost of acquisition. Design changes to the
cryocooler are improvements only if the net result is a reduction in total life cost of ownership
for the user. The implementation of flexure springs into the tactical cooler designs had to
FLEXURE SPRINGS APPLIED TO LOW-COST CRYOCOOLERS 105
result in a reduction in the total cost of ownership. Some increase in the cost of acquisition
could be offset by the increase in operating life; however, DRS’ goal was to implement flex-
ure springs into the tactical designs with no increase in cost of acquisition; resulting in a
reduction in the total cost of ownership for the using programs.
• Operating Lifetime Improvement – DRS’ standard helical coil spring cooler designs have
repeatedly demonstrated lifetimes in excess of the 4,000-hour operational requirement.3 As a
goal, the incorporation of flexure supports into the compressor designs was designed to pro-
duce a doubling of this lifetime. Throughout the design process, this goal was employed to
drive design decisions.
FLEXURE SPRING DESIGN AND COMPRESSOR CONSTRUCTION
Compressor Design
Given the constraints identified in the previous section, the task remained to develop a flex-
ure design that would provide the required axial spring rates, provide the radial stiffness charac-
teristics that are the primary benefit of the flexure spring, and fit within the space envelope of the
existing compressors. The design process that was implemented was iterative in nature.
The first step was to identify a suspension configuration that would fit within the length con-
straints. Since the requirements for the design included working with the installed base of cooler
drive electronics, the electrical characteristics of the compressor could not be significantly
altered. The factors such as induction, coil resistance, back EMF, etc. had to be unchanged. Tra-
ditional flexure suspensions are fairly long in nature, to provide maximum radial stiffness. The
overall length of the tactical cooler compressor (less than 5 inches for a dual-opposed piston
design) necessitated a short spring system.
After a basic design philosophy for the suspension system (hereafter referred to as the spring
stack) was determined, the actual design of the flexure was undertaken. This again was an itera-
tive process, attempting to balance out the conflicting design parameters of spring rate, radial
stiffness, maximum allowed piston stroke, and induced stress in the material. The balancing act
had to arrive at a stack design that would produce the same axial spring force as the helical coil
spring suspension system (to maintain the system’s resonant frequency) while not overstressing
the spring material. In addition, for cost considerations the number of flexures in each stack was
desired to be kept to a minimum. Each additional spring would add cost in both material cost and
increase the time required for assembly and complexity of the alignment process.
The detail design of the flexure was also driven by cost considerations:
• Standard design rules for photo-etch processing were followed to ensure that the flexures
could be produced by a variety of photo-etch vendors using standard industry processes.
• The material chosen for the flexures was readily available blued and tempered spring steel
in stock thicknesses and widths, again to insure availability and lower cost.
When building a limited number of coolers, such cost considerations might be secondary.
With the production rates required for these coolers, piece part cost and availability becomes a
major factor. As a reference point, to support the Javelin cooler build requirements, over 15,000
individual flexure springs need to be procured each year; they must be mass-producible.
The final design of the flexure for the 1/5-watt compressor is shown in Figure 1. Comparison
with the adjacent coin illustrates the small geometries that were required for this design. The
compressor housing is only 1.5 inches in diameter.
The final stack design then used the flexure design and addressed the remaining issues of
retaining the same spring-mass-damper characteristics of the helical coil design. Overall mass
was adjusted and various design modifications were made to adjust the center of gravity of the
moving components (to minimized the potential for side-loading of the clearance seals and/or ori-
entation sensitivity of the compressor). The completed stack assembly is shown in Figure 2.
Figure 1. Javelin compressor flexure spring.
Assembly Processes
The compressor assembly process begins with the spring stacks. These are assembled in fix-
tures that locate the various components in relationship to one another as each component is
added to the assembly. Once all the comonents have been assembled they are clamped in place.
Completed stacks are then bonded to the magnet assemblies to complete the mechanism assem-
bly. Mechanisms are subjected to testing for stiction, to verify proper alignment of the stack, and
basic compressor output functions, to verify proper operation when mated to the expander assem-
bly. Following this testing, the compressor are welded into their housings, mated with an
expander assembly, and eventually delivered to the using programs.
Figure 2. Javelin compressor flexure spring stack assembly.

ENVIRONMENTAL AND LIFE TESTING

The flexure spring compressors discussed in the previous sections were designed as replace-
ments for cryocoolers already in production and fielded in various systems. The currently fielded
designs have been proven through numerous qualification and life test series, in addition to suc-
cessful operation in the field. Prior to switching to the flexure design, some level of testing, vali-
dation, and evaluation is required to verify that the new designs are suitable for the intended
applications.
Once the designs were complete, prototype units were assembled for engineering character-
ization and evaluation. After it was determined that the designs were performing as intended,
additional compressors were assembled for environmental qualification and life testing. The first
design to be tested was the 1/5-watt compressor, specifically in the Javelin configuration. The
series of environmental tests performed on the three test units is shown in Table 1.
At the same time, three additional units were started into a life test to verify reliability
improvements. These units were mated to the Javelin detector-Dewar assemblies to provide the
actual refrigeration demand seen by the cooler in its application. The test cycle for the life test is
illustrated in Figure 3. As can be seen, the cycle includes thermal soaks and operation at temper-
atures ranging from –32C to +52C. These coolers are still running in test and have, at this time,
accumulated over 21,000 hours of operation for a current mean-time-to-failure (MTTF) in excess
of 7,000 hours. As the testing continues the MTTF grows at the rate of about 600 hours per
month. Figure 4 shows the performance of one of the life test coolers since the inception of the
test. The data clearly shows that the cooler performance has been extremely stable and no degra-
dation has occurred.
The one-watt design was undertaken after the 1/5-watt design and has not progressed as far
through the qualification and life test process to date. The qualification testing process is very
similar to that of the 1/5-watt design. The test sequence is listed in Table 2.
One of the primary concerns during the design of the 1-watt flexure compressor was the
extreme magnitude (250 g’s over 10 milliseconds) of the mechanical shock that the unit must sur-
vive as a part of the qualification procedure. To mitigate the risk, several early prototypes were
assembled and subjected to the shock requirement. The prototype units met the requirement and
the design moved forward. During the qualification testing, however, externally applied mechan-
ical shock and vibration proved to be a stumbling block for the flexure design. Analysis of the
problem led to a modification of the flexure stack assembly, which is now being prototyped for
re-testing.
Figure 3. Reliability Test Cycle.
Figure 4. Javelin flexure cooler reliability test data.

Figure 5.1-watt SADA-II cooler reliability test data.
As is the case with the 1/5-watt flexure design, 1-watt flexure coolers are also in a life test. In
this instance, four units are running in test. The test profile is the same (Figure 3). These coolers
have not been in test as long as the 1/5-watt units but have accumulated over 19,000 total opera-
tional hours for a current MTTF approaching 5,000 hours. These coolers are also accumulating
additional MTTF at the rate of approximately 600 hours per month. Figure 5 shows the perfor-
mance of one of the test units since the start of the test. The anomalies in performance that are
seen at various points in the graph are caused by test equipment problems, demonstrating that the
coolers are more reliable than the environmental test chamber and computer-controlled test equip-
ment.
SUMMARY AND CONCLUSIONS

This paper has detailed the development, test, and performance of flexure spring designs
adapted to low-cost tactical cryocoolers. The adaptation has successfully combined the flexure
spring technology utilized to provide long-life for space coolers with the small-sized, low-cost
compressors used in tactical thermal sensor systems. The compressors incorporate flexure
springs for long-life but have been designed such that the cost of implementation is on the same
order as that for the standard helical coil spring designs. These new designs are being proven
through qualification testing. The true measure of the success of the program has been the intro-
duction of the 1/5-watt flexure design into the current Javelin production program. These coolers
are being built at a rate of nearly 70 units per month, demonstrating that the goals for ease of
assembly have been achieved. The continuing life tests have demonstrated reliability improve-
ments under demanding operational scenarios and environments.
ACKNOWLEDGMENT
Much of the work described in this paper was performed by DRS Infrared Technologies
under contract to the U.S. Army Night Vision and Electronic Sensors Directorate , Contract
DAAB07-95-C-J513.
REFERENCES
1 Jones, B.G., Development for Space Use of BAe’s Improved Single-Stage Stirling Cycle Cooler for
Applications in the Range 50-80K,” Cryocoolers 8, Plenum Publishing Corp., New York, 1995, pp.
1-11.
2 Nast, T.C., et.al., “Design, Performance, and Testing of the Lockheed-Developed Mechanical Cryo-
cooler,” Cryocoolers 8, Plenum Publishing Corp., New York, 1995, pp. 55-67.
3 Rawlings, R.M., Granger, C.E., and Hinrichs, G.W., “Linear Drive Stirling Cryocoolers: Qualifica-
tion and Life Testing Results”, Cryocoolers 8, Plenum Publishing Corp., New York, 1995, pp. 121-
127.
High Reliability Coolers under Development
at Signaal-USFA
M. Meijers, A. A. J. Benschop and J.C. Mullié

Signaal-USFA
ABSTRACT
Since 1997 Signaal Usfa has been working on the development of high reliability cryocool-
ers. These coolers have been developed with the objective of eliminating the lifetime determin-
ing factors of conventional cryogenic tactical coolers. The intention of this study was the devel-
opment of a family of cryocoolers that could be used to cover a large range of cooling powers.
Today, these developments have resulted in a new range of flexure-bearing cryocoolers currently
available at Signaal Usfa, with cooling performances between 0.5 and 3 W at 80K and estimated
lifetimes of more than 20,000 hours MTTF.
The basis for the extended lifetime of these coolers is our unique flexure-bearing compressor
with moving-magnet technology. Inside the linear dual-opposed-piston compressor both moving
pistons are fully supported at the back and front of the piston by optimized flexure bearings.
With this flexure-bearing suspension, side loads on the piston seals are avoided. A proper align-
ment procedure ensures no contact between piston and cylinder during operation, resulting in the
absence of wear of the piston coating, which normally determines the lifetime of a cooler.
The moving-magnet technology, as applied in our flexure-bearing compressor, has several
major advantages over moving-coil linear motors as applied in most conventional linear com-
pressors. First of all, the coils, known to be a possible source for gas contamination, can be
placed outside the hermetically-sealed compressor housing containing the working gas. Avoiding
any synthetic materials inside the cooler reduces the risk of gas contamination during the life of
the cooler. The fact that the coils can be placed outside the hermetically-sealed compressor also
means that no glass feedthroughs are required. In this way, risks of glass feedthrough leakage due
to extreme temperature shocks or mechanical shocks are no longer present. Finally, the absence
of moving coils in the compressor design also means that flying leads to supply power to the
coils are no longer present.
Several qualification tests have been performed on flexure-bearing cryocoolers with differ-
ent sizes of coldfingers, resulting in technical specifications currently available at Signaal Usfa
on all presented cooler types in the above-mentioned range. A flexure-bearing cooler for appli-
cations requiring over 6 watts cooling power at 80K is currently under development as well.
Tests performed on a first prototype of this cooler have been successful, with a measured cooling
performance of more than 8W at 80K for a 23°C ambient temperature, with 150 Wac of input
power.

LIFETIME LIMITATIONS OF A LINEAR STIRLING CRYOCOOLER

In conventional linear Stirling cryocoolers the lifetime of the cooler is limited by the com-
pressor. Inside a linear compressor usually two pistons, driven by moving coil linear motors, are
translating in opposite phase, generating a pressure wave in the compression space between the
pistons, which is connected to the warm end of the coldfinger. The magnitude of the generated
pressure wave directly determines the cooling performance obtained with a certain Stirling cold-
finger or pulse tube.
To prevent gas leakage along the pistons inside the compressor, which reduces the generated
pressure wave and thus the cooling performance, close tolerance seals are often applied. The
principle of close tolerance seals is a very small annular gap between the piston and the cylinder,
typically with a length of a few centimeters, which prevents gas from flowing from the compres-
sion space to the larger (buffer) space behind the pistons.
To limit the gas flow along the pistons to an acceptable level, the gap between the piston and
the cylinder should be as small as possible, but still allow piston movement at different ambient
temperatures between –52°C and +71°C. In practice, the initial gap between the piston and cyl-
inder will be around 10 microns. Increase of this gap due to wear of the coating applied on the
piston increases the flow along the piston. In fact, the gas flow through the annular gap between
the piston and cylinder is dependent on the gap height to the third power.
To determine the impact of an increase of the gap height between piston and cylinder on the
efficiency of the cooler, several measurements were performed on a Stirling cooler at Signaal
Usfa. In Fig. 1, the results of these measurements are depicted for a Stirling cooler with a 5 mm
cold finger. In this figure a graph is shown in which the measured cooling performance is plotted
against the input power to the cooler. Data are presented for the standard initial gap of 10 mi-
crons between piston and cylinder, as well as for larger gaps of 17 microns and 20 microns. One
can clearly see the large impact of the flow losses along the piston on the efficiency of the cooler.
From these measurements it was concluded that the specified cooling performance is no longer
met for piston diameter reductions of more than about 10 microns.
Figure 1. Results of measurements to investigate the impact of the gap between piston and
cylinder on the efficiency of the cooler.
HIGH RELIABILITY COOLER DEVELOPMENT AT SIGNAAL-USFA 113
HIGH RELIABILITY COOLERS AT SIGNAAL-USFA: A HISTORICAL OVERVIEW

Lifetime tests on various cooler types at Signaal Usfa have shown that, with conventional
compressor wear of a PTFE-based coating due to contact between piston and cylinder, the life is
limited to about 4,000 to 10,000 hours MTTF, depending of the cooler type. In order to ensure
higher MTTF values of more than 20,000 hours, wear of the pistons should be completely elimi-
nated. The only way to ensure this is by avoiding contact between piston and cylinder.
Study of possibilities to avoid contact between piston and cylinder (1997)
In 1997 Signaal Usfa initiated an internal development program investigating different pos-
sibilities to increase the lifetime of linear Stirling cryocoolers. The key goal in this development
was to avoid contact between piston and cylinder, as this was considered the only possibility to
obtain lifetimes of more than 20,000 hours. Two different principles with which this could be
achieved were studied and worked out in different compressor designs. In the end, these com-
pressor designs were compared on a number of criteria like: compressor efficiency, design com-
plexity, compressor size and weight, and cost/price of the cooler.
The first principle studied was the application of gas bearings. With gas bearings, the piston
and cylinder are separated by a thin layer of gas under pressure. Two different ways in which the
build-up of gas pressure can be achieved, often referred to as dynamic gas bearings and static gas
bearings, were studied and worked out in detail.
Although gas bearings are applied by several manufacturers of linear compressors, we con-
cluded in a final comparison that they scored worst on almost all the criteria mentioned above.
The main disadvantages with dynamic gas bearings were the presence of an extra motor to rotate
the piston, and the connection of the rotating piston to the compressor housing. With the static
gas principle, the reduction of efficiency due to the extra flow losses and the unavoidable pump-
up effect in the compressor led to the decision not to continue with this design principle either.
Contact between piston and cylinder can also be avoided when the moving piston is sus-
pended by mechanical springs that offer a high radial stiffness and allow easy movement in the
axial direction. This can be achieved with several types of springs, but we have studied a suspen-
sion in three folded leaf-springs at both piston ends and a suspension with flexure bearings at
both piston ends. Again, both possibilities were worked out in detail and, after several tests were
performed on both the folded leaf-springs and the flexure bearings, the conclusion was drawn
that, on all criteria mentioned above, flexure bearings are the most interesting solution.
First prototype flexure-bearing cryocooler (1998)
Based on studies performed in 1997 a first prototype flexure-bearing cooler was build and
tested in 1998. Figure 2 shows a picture of this cooler.
Figure 2. First prototype flexure-bearing cryocooler (1998).

With this first prototype flexure-bearing compressor, which was connected to a 7mm Stir-
ling coldfinger, several tests have been performed that prove that the flexure-bearing principle
and the specially developed alignment procedure are working correctly. As can be seen in Fig. 2,
the compressor itself is rather bulky due to the presence of flanges and bolt connections incorpo-
rated to allow the compressor to be sealed with O-rings. However, this enabled us to dismount
the compressor, to inspect components after several tests, and to verify that the flexure-bearing
suspension was working as expected.
Although several imperfections in the compressor design were found, the fact that the com-
pressor worked correctly after the first assembly, and the working principle of the flexure-bearing
suspension was verified, was very promising. At the end of 1998 the cooler was put in life test;
up to now (May 2000) it has accumulated over 12,000 hours of operation without degradation of
performance. The cooler has always been running at 80% of its maximum power generating a
cooling power of approximately 750 mW at 80K, with a 23°C ambient temperature.
Redesign of the first prototype flexure-bearing compressor (1999)
In the beginning of 1999 the flexure-bearing compressor was completely redesigned. Based
on the experience obtained from tests on the first prototype several small design changes were
introduced to simplify the compressor assembly. Also the design was made more compact with
the objective of hermetically sealing the compressor by laserwelding to allow a complete qualifi-
cation test program to be performed. Beside this, the flexures were also redesigned by increasing
the axial spring stiffness to limit piston drift under all circumstances.
To supply current to the moving coils inside a linear compressor, flying leads are commonly
applied; these are electrical wires connected to the glass feedthroughs in the compressor housing
at one end, and the coils of the linear motor at the other end. However, the flying-lead connec-
tions are critical with respect to reliability and require high precision in mounting during produc-
tion. In the redesign of the flexure-bearing compressor, an innovative solution was introduced to
replace the flying leads.
Based on the experience obtained with flexure-bearing and Finite Element Modeling (FEM)
calculations, special spiral arms were designed to supply current to the moving coils in the re-
designed flexure-bearing compressor. These “flexible leads” are very easy to mount during the
compressor assembly, and can be designed with FEM tools to ensure the absence of failures. To
illustrate the principle, Figs. 3 and 4 show the FEM model of a flexible lead and a photograph of
a flexible lead mounted on a coil assembly.
Figure 3. Photograph of a flexible lead Figure 4. FEM model of a flexible lead (right).
mounted on a coil assembly (left).
Several flexure-bearing coolers have been built with 10mm Stirling coldfingers and submit-
ted to various qualification tests. Cooling performance levels measured at ambient temperatures
ranging from -52°C to +71°C were excellent and showed that the efficiency of the cooler was
according to expectations. The length of the compressor is 180 mm with a diameter of 75 mm;
the resulting total cooler weight is 3.5 kg. Two coolers were put in a lifetime test at the beginning
of 2000 and are still running without degradation.
Development of a compressor with moving magnet motor technology (1999)
Parallel with the redesign of the first prototype flexure-bearing compressor in 1999, the de-
velopment of a compressor with moving-magnet technology was also initiated at Signaal Usfa.
With a moving-magnet linear motor, the magnets are connected directly to the pistons and the
coilholders are part of the compressor housing. This offers a number of advantages that increase
reliability of the cooler and allow the compressor design to be more compact.
First of all, the fact that the coils are no longer moving means that flying leads or flexible
leads are no longer necessary to supply current to the coils. The absence of these components
simplifies the compressor design and assembly, and reduces the compressor length.
In the final flexure-bearing compressor design, the coil-holder on which the coils are wound
is pan of the housing of the compressor. This means that the coils are outside this housing and
outside of the working gas. As the coil insulation consists of a synthetic material that absorbs
moisture, bake out of components and curing of the cooler under high vacuum at elevated tem-
peratures is probably the most critical process in the cooler production. With the removal of the
coils outside the working gas, no more outgassing-components are present inside the cooler. This
reduces the risk of gas contamination during the life of the cooler.
Finally, the fact that the coils are located outside the hermetically-sealed compressor means
that also glass feedthroughs are no longer required. Glass feedthroughs are known to be critical
components that can crack under extreme temperature shocks and severe mechanical stresses,
resulting in gas leakage.
All these advantages have lead to the conclusion that, for a high reliability cryocooler, mov-
ing magnet technology in combination with a flexure-bearing suspension is the best solution to
guarantee long lifetimes.
Final design of the flexure-bearing cryocooler with moving magnets (1999/2000)
In the final design of the flexure-bearing compressor, we have been able to reduce the com-
pressor diameter to 60 mm and the length to 165 mm, resulting in a cooler mass of 2.4 kg. FEM
calculations performed on the flexures have lead to an optimized design with a high radial stiff-
ness and low peak stress levels. FEM calculations performed on the moving magnet linear motor
have lead to an optimized motor with an effective motor efficiency of more than 70%. Figures 5
and 6 show photographs of a cooler with a 10 mm coldfinger and a flexure-bearing subassembly.
Figure 5. Photograph of the final Figure 6. Photograph of a flexure-bearing

flexure-bearing cooler (left). assembly (right).
Close to the end of 1999 several coolers were built and tested. In the beginning of 2000 two
of these coolers have successfully undergone a complete qualification test program including
shock and vibration tests. Furthermore, we have put four coolers in lifetime test. One of these
four coolers has also been tested in a centrifuge in which we have applied radial accelerations on
the compressor for 500 hours. We have increased the radial accelerations starting with 2g up to a
maximum of 10g. These tests have not resulted in degradation of the cooler. In March 2000,
three coolers with 5mm Stirling coldfingers were put in lifetime test.
NEW SIGNAAL USFA FLEXURE-BEARING CRYOCOOLER FAMILY
Based on the final design of the flexure-bearing compressor with moving-magnet linear
motor, a complete new range of Stirling cryocoolers referred to as the LSF cryocooler family has
been developed. This family consists of four main types of slip-on cryocoolers with 5mm, 7mm,
10mm and 13mm Stirling coldfingers. Qualification tests performed on these four coolers have
resulted in technical specifications that are currently available at Signaal Usfa. In Fig. 7, four
graphs with specified performances of these coolers are depicted.
Figure 7. Specified cooling performance at different ambient temperature levels

for all four coolers of the LSF cryocooler family.
FUTURE DEVELOPMENTS
Besides the development of the new flexure-bearing cryocooler family, Signaal Usfa has
several other development programs running. Two of these new developments are the develop-
ment of a miniature pulse tube cryocooler driven by a flexure-bearing compressor, and develop-
ment of a high capacity Stirling cryocooler with a flexure-bearing compressor.
Flexure-bearing compressor with pulse tube coldfinger
During the past years, Signaal Usfa has been working on an analytic model of pulse tube re-
frigerators based on harmonic approximations. This tool has made it possible to perform optimi-
zation of different design parameters determining pulse tube efficiency and helped us in under-
standing pulse tube behavior.
During the past months we have been performing tests on a prototype pulse tube coldfinger
which has a tube diameter of 5mm. The pulse tube design is a U-shape configuration with iner-
tance tube and single bypass. Up to this moment we have measured a cooling performance of
500 mW at 80K with this pulse tube at room temperature conditions. To drive the pulse tube, a
flexure-bearing compressor was adapted to match the volume and drive frequency of the pulse
tube. Our goal is to increase the cooling performance up to 1 W at 80K and further industrialize
the combination of pulse tube and flexure-bearing compressor.
High capacity Stirling cryocooler with flexure bearings
During the past year we have also been working on the design of a high capacity Stirling
cooler. The initial goal with this cooler was to obtain a cooling power of more than 6 W at 80K
with a lifetime of more than 20,000 hours MTTF. To achieve this we have designed a cooler
based on a Stirling coldfinger with a diameter of 20 mm.
During the beginning of this year we have build a first prototype of this cooler and we are
currently running an extensive test program on this cooler. With this first prototype we have
measured a cooling performances of more than 8 W at 80K in room temperature conditions with
an electrical input power of only 150 Wac to the cooler.
We are currently optimizing different parameters of the coldfinger that determine the effi-
ciency. In the meantime we are designing a flexure-bearing compressor with moving magnet
technology to drive the coldfinger in a final design that will be tested before the end of this year.
The diameter of the flexure-bearing compressor will be less than 90 mm and the length of the
compressor will be less than 200 mm, resulting in a relatively compact cooler considering the
high cooling performance.
CONCLUSIONS
From the presented work, it may be concluded that:
• After three years of work we have been able to develop a range of affordable and compact
flexure-bearing cryocoolers. These coolers are currently available for prices that are of the
same order of magnitude as conventional tactical coolers currently available in the market.
• The design of the flexure-bearing compressor, which has a diameter of only 60 mm and a
length of 165 mm, is such that it can be matched to different Stirling coldfingers and pulse
tubes with comparable void volumes.
• Inside the compressor design, full support of the pistons via flexure bearings ensures the ab-
sence of piston wear. The application of moving-magnet technology has resulted in the coils
being outside of the working gas, thus reducing the risk of gas contamination and eliminating
critical components such as flying leads and glass feedthroughs.
• Many qualification tests performed on these coolers have shown that the compressor design
can withstand the severe environmental conditions required for tactical applications.
• Lifetime tests that are running at Signaal Usfa indicate that the expected lifetime of more
than 20,000 hours will be achieved. It’s just a matter of time to verify lifetime results in
practice.
• New developments currently running at Signaal Usfa offer opportunities for future markets
requiring ultra-low vibration levels or cooling powers of more than 6 W at 80K combined
with long lifetimes exceeding 20,000 hours MTTF. With our first prototype high capacity
cooler, a cooling performance of more than 8 W at 80K with only 150 Wac input power has
been measured.
Development of a Long-Life Stirling Pulse Tube
Cryocooler for a Superconducting
Filter Subsystem
Y. Hiratsuka1, K. Murayama2, Y. Maeda2, F. Imai2,

K. Y. Kang2 and Y. Matsubara3
1
DAIKIN Industries, Ltd.
Semiconductor Equipment Department
Osaka 592-8331, Japan
2
DAIKIN Environmental Laboratory, Ltd.
Tsukuba, Ibaraki 305-0841, Japan
3
Atomic Energy Research Institute, Ninon University
Funabashi, Chiba 274-0063, Japan
ABSTRACT
We have developed pulse tube cryocoolers for high temperature superconducting (HTS)
filter subsystems used in the base stations of mobile telecommunication systems. In July 1999,
we reported on our development of a 5 W Stirling pulse tube cryocooler with a contact-type
compressor,1 with a cooling capacity of 5.5 W at 80 K for 200 W of input power. However,
demands for a smaller-sized cryocooler with higher efficiency and with 5-year reliability
prompted us to develop such a cryocooler with a U-type expander and a flexure-bearing-sup-
ported linear compressor with opposed pistons.
We have developed an HTS filter and a long-life Stirling pulse tube cryocooler to cool the
filter whose cooling capacity is around 1W at 80 K, as previously discussed in a progress
report.2 For a compressor input power of 60 W at an operating frequency of 52 Hz and a
pressure-volume (P-V) work of 26 W, and for a compressor efficiency of 45%, this cryocooler
achieved a cooling capacity of 1.05 W at 80 K (0.63 W at 70 K), a specific power of 92 W/W,
5.5% Carnot (3.9% Carnot at 70 K), and a specific P-V work of 40 W/W, with a minimum
temperature of 57 K in an ambient of 23°C.
The key devices of this filter subsystem are an HTS filter and a low noise amplifier (LNA).
The HTS filter is made from a YBCO HTS thin film and has a fractional bandwidth below
1.2% at 2 GHz and has a minimum insertion loss at 0.3dB. The HTS filter and the LNA are
operated at a constant temperature of 70 K and the cooling capacity needed by them is 0.6 W.
We integrated them with the cryocooler into a subsystem, and the external dimensions of this
system are 194 mm high, 180 mm wide, 250 mm deep, with a total volume of 8.7 L.

120 LONG-LIFE COMMERCIAL PULSE TUBE COOLERS
INTRODUCTION
The development of high temperature superconducting (HTS) devices has increased the
demand for small-sized cryocoolers. In 1992, E. Tward et al.3 developed a Stirling-type pulse
tube cryocooler with a cooling capacity of 1 W at 80 K, similar to the power required for a
conventional Stirling cryocooler. In 1998, J.L. Martin et al.4 designed a low-cost prototype
Stirling-type pulse tube cryocooler for civilian electronics applications, especially for cooling
superconducting filter subsystems for base stations of mobile telecommunication systems. Fur-
thermore, in 1999, S-Y. Kim et al.5 addressed HTS cryocoolers with a target cost of $1,000 for
quantities of 10,000 cryocooler systems per year. They designed a cryocooler to demonstrate
the feasibility of making a long-life, low-noise cryocooler for cooling HTS devices.
In 1999, we reported on the development of a 5 W at 80 K class pulse tube cryocooler for
HTS filter subsystems. Miniaturization of subsystems along with the demand for reliable cryo-
coolers prompted us to further develop a non-contact type compressor whose cooling capacity
is 1 W at 80 K. We next developed an HTS filter and then combined it with the pulse tube
cryocooler into an integrated subsystem.
This paper describes the development of our pulse tube cryocooler with a target cooling
capacity of 1 W at 80 K (0.6 W at 70 K), and its associated HTS filter subsystem.
GENERAL DESIGN
The specifications for the cooler are shown in Table 1, while Figs. 1 and 2 show a sche-
matic and a photograph of the unit. The cryocooler is a split Stirling-type pulse tube with
opposed pistons that are driven by a linear motor. The compressor is connected to the expander
by tubes that are between 20 mm and 100 mm long. The cooling method is forced air-cooling.
The system is filled with helium gas up to 3.1 MPa. During thermal testing the unit is mounted
in a vacuum chamber evacuated to about 0.13 MPa ( torr).
Figure 1. Schematic of the pulse tube cryocooler. Figure 2. Photograph of cryocooler.

A LONG-LIFE PT CRYOCOOLER FOR HTS FILTER SUBSYSTEMS 121
Expander
The cooling capacity of the U-type expander is 10% inferior to that of an in-line type.
However, we adopted the U-type expander in which the regenerator and pulse tube are ar-
ranged in parallel because it decreases the size. The phase control system uses an inertance
tube to prevent instability of the DC flow.
Compressor
To develop a cryocooler that has a long-life and requires no maintenance for at least five
years, we developed a non-contact type compressor in which the piston is arranged with bear-
ings to decrease the size of the cryocooler, and with 'tangential flexures' to significantly im-
prove the radial stiffness. The clearance of the piston and cylinder can thus be kept to less than
15 µm and the piston has no seal. This non-contact type compressor has an outer diameter of
76 mm and a length of 169 mm. We compared this non-contact type compressor with a con-
tact-type compressor developed for Stirling cryocoolers. In the comparison tests, the piston
position was monitored by using a laser vibrometer, and the mass flow rate through the phase
shifter was measured using pressure transducers mounted near the compressor discharge head,
near the hot side of the regenerator, and near the pulse tube. These measurements were used to
calculate both the pressure-volume (P-V) work of the compressor and the equivalent P-V work
of the expander.
RESULTS AND DISCUSSION

Cryocooler Performance
The diameter and length of the regenerator, pulse tube, and inertance tubes were opti-
mized to achieve the maximum cooling capacity. A comparison of the cooling capacity of the
non-contact and contact-type compressors is shown in Figure 3. For the contact-type compres-
sor, for 60 W input power, the no-load temperature was 56 K, the cooling capacity was 0.65 W
at 70 K (1.1 W at 80K), and the P-V work was about 27 W. In contrast, for the non-contact
type compressor, the no-load temperature was 57 K, the cooling capacity was 0.63 W at 70 K
(1.05 W at 80 K), which is slightly lower than that for the contact type compressor, and the P-V
work was about 26 W with a 23°C ambient. The efficiency of the non-contact type compressor
was 45%; however, if a higher efficiency compressor is used, the cooling capacity at a given
temperature should increase.
Figure 4 shows the efficiency and cooling capacity at 70 K as a function of compressor
input power. The efficiency is 3.9% Carnot at 70 K for 60 W of input power and the cooling
capacity is relatively independent of input power. Figure 5 shows the cooling capacity and
Figure 3. Measured cryocooler cooling Figure 4. Measured cryocooler cooling

capacity vs. cold head temperature. capacity vs. compressor power.

capacity vs. cold head temperature. capacity vs. inclination angle.
%Carnot efficiency as a function of cold head temperature. When the cold head temperature is
higher than the pulse tube cryocooler shows superior and more linear temperature charac-
teristics compared with those of the Stirling cryocooler.
Influence of Pulse Tube Inclination

The performance of pulse tube cryocoolers can be affected by natural convection, which is
affected by the inclination of the pulse tube. In July 1999, we reported that cryocooler perfor-
mance was unaffected by the pulse tube inclination. In general, high operating frequency and
small diameter pulse tubes reduce the convective heat loss. For confirmation, we measured the
effect of pulse tube inclination on the performance of the cryocooler.
Figure 6 shows the cooling capacity at 70 K and 80 K for cold head orientations of 0° to
180° downward from vertical. The data show that the cooling capacity is significantly reduced
when the inclination is 135°, which differs from previous results in which the cooling capacity
was independent of inclination.2 This difference might be due to the difference in shape be-
tween the in-line type and U-type expanders.
Influence of Environmental Temperature

It was necessary to confirm the effect of environment temperature on cryocooler perfor-
mance because the device is used outside. To operate the HTS filter for this system, the cool-
ing capacity necessary is 0.6 W at 70 K. We tested the system by first setting the environment
temperature to 60°C, 40°C, 25°C and 23°C, respectively, via a thermostat, and then changing
the operating frequency and compressor input power to optimize the cooling capacity. Our
results show that the cooling capacity at an environment temperature of 60°C was significantly
lower than that at an ambient of 23°C (Figure 7). However, this system did not reach the target
value, and at about 40°C the system becomes stroke limited.
The effect of operating frequency on the cooling capacity at environment temperatures of
25°C and 60°C is shown in Figure 8. The optimal operating frequency is seen to depend on the
environment temperature. This is because the temperature and internal pressure of the cryo-
cooler gas increase with increases in the environment temperature, and the increased pressure
causes the optimum operating conditions for the compressor and inertance tube to change.
Vibration
Pulse tube cryocoolers, which have no moving parts in the cold section, are more attrac-
tive than other small-sized cryocoolers because of their high reliability, simpler construction,
and lower vibration levels. Vibration is not a significant problem in the filter subsystem, but,
A LONG-LIFE PT CRYOCOOLER FOR HTS FILTER SUBSYSTEMS 123

capacity vs. environmental temperature. capacity vs. operating frequency.
Figure 9. Vibration measurement method. Figure 10. Measured cryocooler vibration.
for example, there are applications like an electron microscope that require almost no vibra-
tion. We therefore used a laser vibrometer to measure the vibration of a 70 K cold head (Fig.
9). To extract only the vibration at the cold head, we simultaneously measured the vibration at
the flange of the expander, and then subtracted it from the cold head vibration measurement.
Figure 10 shows the pressure of the pulse tube hot end (solid line) and vibration at the cold
head (dotted line). The 0-peak vibration amplitude is around +0.6 µm , with a total peak-peak
amplitude of 1.2 µm, and has a period similar to that of the pressure amplitude. We calculated
the axial extension of the expander cylinder from this pressure amplitude, and found that the
value calculated was approximately +0.94 mm, which is similar to the measured value. For
comparison, we also measured the vibration of a Stirling cryocooler, and found that it was
similar to that of a pulse tube cryocooler.
Filter subsystem
The key devices of this filter subsystem are an HTS filter and an LNA. Figure 11 shows
the frequency response of an HTS filter. This filter is made by using a YBCO HTS thin film,
and has a fractional bandwidth below 1.2% at 2 GHz and has a minimum insertion loss at
0.3 dB. Figure 12 shows a photograph of the subsystem. The external dimensions of this
system are 194 mm high, 180 mm wide, 250 mm deep, with a total volume of 8.7 L. The
cooling capacity needed by the HTS filter and LNA is 0.6 W at 70 K. The caloric output for the
amplifier is 0.4 W, the quantity of radiation parasitics is 0.1W, and the parasitic heat conduc-
tion down the cable is 0.1W. The vacuum container theoretically lasts for 10 years.
Figure 11. Frequency response of HTS filter. Figure 12. Photograph of subsystem.
CONCLUSIONS
We developed an HTS filter and a long-life Stirling pulse tube cryocooler to cool the filter.
We then integrated them into a subsystem. The significant results of our research are as fol-
lows:
1. For an input power of 60 W, the no-load temperature of the cryocooler was 57 K and the
cooling capacity was 0.63 W at 70 K, and 1.05 W at 80 K.
2. Vibration in the cold head of this pulse tube cryocooler is primarily caused by the cold-head
elongation due to the cycling pressure. The vibration amplitude is about 1.2 mm peak to peak.
3. Environmental temperature dependence of this cryocooler is 15 mW/°C.
4. The cooling capacity decreases by 35% when the pulse tube is inclined 135° from the vertical.
5. The HTS filter has a fractional bandwidth below 1.2% at 2 GHz and has a minimum insertion
loss of 0.3 dB.
6. The integrated subsystem is small and light weight, having a total volume of only 8.7 L.
REFERENCES
1. Y. Hiratsuka et al., “Development of a 1 to 5 W at 80 K Stirling Pulse Tube Cryocooler,” Cryocoolers
10, Plenum Press, New York (1999), pp. 149-155.
2. Y. Hiratsuka et al., “Development of a 5 W at 80 K Stirling Pulse Tube Cryocooler,” Advances in
Cryogenic Engin., Plenum Press, New York (2000).
3. E. Tward et al.,“Miniature Pulse Tube Cooler,” 7th International Cryocooler Conference Proceed-
ings, Air Force Phillips Laboratory Report PL-CP--93-1001, Kirtland Air Force Base, NM, April 1993,
p. 113.
4. J.L. Martin et al.,“Design Consideration for Industrial Cryocoolers,” Cryocoolers 10, Plenum Press,
New York (1999), pp. 181-189.
5. S-Y Kim et al.,“Development of Low-Cost Pulse Tube Cryocooler for HTS Applications,” Advances
in Cryogenic Engin., Plenum Press, New York (2000).
Development of a 5W at 65 K Air-Cooled
Pulse Tube Cryocooler
S-Y Kim, J-J Park, S-T Kim, W-S Chung, H-K Lee
LG Electronics Inc. Digital Appliance Lab. COMP Team
Seoul 153-023, Korea
ABSTRACT
LG Electronics (LGE) has developed an air-cooled Pulse Tube Cryocooler (PTC) for HTS
applications. The air-cooled PTC provides 5.5 W of cooling at 65 K and 25°C ambient with 280 W
of input power with a single-acting linear compressor. Its performance is greater than that of an
LGE water-cooled PTC of last year, 4.9 W at 65 K and 20°C, with 270 W of input power. The air-
cooled PTC, compared to the water-cooled one, has a lower-efficiency linear motor, a 3% decrease
due to cutting down the cost of the linear motor, and a higher temperature at the surface of the
aftercooler, 30°C higher than the water-cooled one. Thus, the performance has improved signifi-
cantly.
We have optimized the geometry of the regenerator, the pulse tube, and the inertance tube
using both simulation and test; a significant increase of the performance is noted. Also, to maintain
the low cost target, we have decreased the cost of the linear motor and used a common heat ex-
changer for air-cooling.
INTRODUCTION
A pulse tube Cryocooler has a lot of advantages over a Stirling Cryocooler because it has no
moving parts in the low-temperature region. This means much more reliable operation and much
lower vibration in the cold region. Ideally, Stirling cryocoolers have better thermal efficiency than
PTC. However, some recent studies2,3 have suggested that PTC using an ‘inertance’ tube can
generate the phase shift needed to make the PTC operate as well as a Stirling cryocooler. The
inertance tube is a long, thin tube that allows the phase between the pressure and mass flow in the
pulse tube to be adjusted to an extent that was not previously possible.
The air-cooled pulse tube cryocooler developed by LGE aims at HTS applications such as
cooling HTS RF filters in wireless communication systems. The compressor that drives the in-line
pulse tube is based on a moving-magnet design combined with flexure supports. This PTC design
approach strives for a better price-to-performance ratio, while maintaining a long life. Our target
cost is $1,000 for quantities of 10,000 units per year. The PTC has a high potential for very low cost
and demonstrates high cooling power.
The design of the PTC was driven by the challenging low cost and reliability targets associated
with the requirements of wireless applications. The M-CALC II report3 suggested a goal of life and
cost for low-cost commercial cryocoolers. The 40,000-hour continuous duty life limits the design
options available, and the cost goal of $1,000 for quantities of 10,000 a year for the complete

cryocooler means that this cooler must achieve goals that no other cryocooler has yet met. We
continue to cut down the cost of the PTC such as by reducing the size of the linear motor magnet
and decreasing the high-precision machining. This is because the linear motor is the key compo-
nent and the compressor cost reported5,6 is over 50% of the total system cost. In the air-cooled PTC,
we are also using a radial copper fin type of heat exchanger and a common fan at the aftercooler to
achieve a lower level of cost.
We have developed a performance prediction program based on the ‘thermodynamic non-
symmetry’ effect1 and tried to find the optimum design point for the main components such as the
regenerator, pulse tube and inertance tube. The performance must be improved significantly be-
cause the efficiency of the linear motor was decreased, 3% lower due to the low cost design, and the
heat rejection temperature was increased, 30°C higher than the water-cooled one.7 Currently, the
performance tests show very good results, 60% higher than what the water-cooled one achieved last
year.
COMPRESSOR
The low-cost compressor consists of three main parts: a clearance-sealed piston/cylinder, spi-
ral flexure linear bearings, and a moving magnet type of linear motor. These are enclosed in a
common pressure vessel with an aftercooler interface to the pulse tube. The layout of the compres-
sor in combination with the in-line pulse tube is shown in Figure 1. This unit has a motor diameter
of 130 mm and a total prototype vessel dimension of 170 mm diameter, including vessel flange,
and 245 mm long, including aftercooler. The compressor mass is 10 kg, and the total mass is 12 kg.
Production vessels will be welded shut, not bolted. This will save some mass and remove the
flanges.
The moving magnet motor consists of an outer stator with coil, an inner stator, and a magnet
assembly on which the magnets are attached. The magnet assembly is connected to a reciprocating
shaft supported on each end by flexures and connected to the piston. Moving magnet motors re-
move the need for problem-prone flexing leads, so they simplify the structure greatly. The motor is
designed for a maximum of 400 W of mechanical output, and a maximum of 20 mm of stroke. It is
tuned by adjusting the moving mass and springs so that it works at 60 Hz from 220 VAC. The linear
motor, which includes radial-laminations stacked automatically, can be produced as easily as a
traditional motor in the mass production line, and uses a very small Nd-Fe-B magnet.
The piston moves without rubbing contact in the cylinder to produce the pressure wave that
drives the pulse tube. The clearance seal is maintained in a centered radial position by flexures, and
the radial clearance is 25~35 mm. The clearance seal requires very tight tolerances and high-preci-
sion alignment. Because the clearance seals may be too expensive for mass production, one of the
Figure 1. Layout of the air-cooled pulse tube cryocooler.

DEVELOPMENT OF A 5W AT 65 K AIR-COOLED PT COOLER 127
primary concerns is to considerably reduce the cost of their alignment. This is accomplished by
reducing the high-precision machining, devising a simplified mechanical architecture with fewer
parts, and simplifying the alignment of the piston supports. The front flexures, combined with the
piston and the shaft, are bolted to the cylinder assembly maintaining the piston in the center of the
cylinder; the backside flexures, bolted to a backside frame, are combined with the shaft and the
magnet assembly in a way that maintains the pre-alignment. This robust and inexpensive design
shows only a small amount of loss of delivered PV power.
OPTIMIZATION
The LGE pulse tube has an in-line shape with an inertance tube to provide the proper phase
shift. Stacked copper screens are used in the cold end, the warm end, and in the aftercooler; stain-
less steel 400 mesh screens are used in the regenerator. The PTC uses a 1 inch inner diameter thin
stainless steel tube for the regenerator, and a 1/2 inch tube for the pulse tube. The charge pressure is
2.7 MPa and the amplitude is 0.5 MPa. The pulse tube consists of many components that have
complex thermal interactions when a part is changed, so it is very hard to optimize each component.
Therefore, we decided to first increase the inner aftercooler volume and the outer surface of the air-
cooled PTC to provide a higher overall heat transfer coefficient. Then, the regenerator, pulse tube
and inertance tube, which were expected to be the main parameters with the largest effect on the
performance, were optimized using simulations and tests.
Figure 2 shows the predicted performance from the simulations as a function of regenerator
volume and pulse tube volume. The swept volume and the aftercooler volume are fixed and the
inertance tube is tuned to achieve maximum COP in every case. From the figure, it is clear that
there is a strong trade-off between the regenerator loss and the pressure drop inside the regenerator.
The optimum design point of the regenerator is easily found, although the optimum pulse tube
volume varies slightly when the regenerator volume is changed. In the small regenerator volume
region, in which the regenerator loss is dominant, we can see the performance rises steeply when
the regenerator volume is increased. However, in the pressure drop dominant region, the curves
show a gentle slope.
Figure 3 shows a comparison between the simulation and test results. The test results have
lower performance than the simulations due to some assumptions in the simulation—adiabatic
compressor, perfect heat exchanger and so on—and some heat losses. However, the trends are very
similar.
The optimum point of the regenerator and the pulse tube volume is not changed with the vari-
ous inertance tubes, but the simulation and test results show the performance of the PTC is changed
Figure 2. COP Predictions as a function of regenerator and pulse tube volume: Vre regenerator
volume, Vpt pulse tube volume, Vs swept volume; the inertance tube is tuned in every each case.
Figure 3. COP as a function of regenerator and pulse tube volume (25°C ambient temperature,
vacuum torr): Vre regenerator volume, Vpt pulse tube volume, Vs swept volume; the inertance tube is
tuned in each case.
significantly when the length to diameter ratio of the inertance tube, L/D, is increased (Figure 4). In
each case, the length and diameter of the inertance tube are tuned to maximize the COP of the PTC.
Thus, the length to diameter ratio is increased when the diameter is increased. The test results show
a different tendency beyond the particular L/D of the inertance tube, in which the performance goes
down steeply. Therefore, we can conclude that the length to diameter parameter of the inertance
tube has a significant effect on the performance and there is a critical point, which means the
phenomena inside the inertance tube seems to be changed suddenly beyond that point.
Figure 5 shows the performance test results of the optimized PTC in the torr vacuum
chamber at 25 °C ambient temperature. The curve shows the 5.5 W at 65 K, and the 48 K no-heat-
load temperature. The cooling capacity of the optimized cooler is slightly improved over what the
water-cooled one achieved last year; for the same conditions, the optimized one shows 60% higher
cooling capacity. The total input power is 280 W, and the PV work is 240 W, 86% of indicated
efficiency and 87% of estimated motor efficiency.

The air-cooled optimized PTC developed by LGE has a better price-to-performance ratio than
that of our water-cooled PTC of last year, even with the lower efficiency linear motor and the
Figure 4. COP as a function of inertance tube L/D ratio (25°C ambient temp., vacuum torr).
DEVELOPMENT OF A 5W AT 65 K AIR-COOLED PT COOLER 129
Figure 5. Cooling capacity at the cold end (25°C ambient temp., vacuum torr).
higher heat rejection temperature of the aftercooler. These changes were the result of cutting down
the cost of the linear motor and air-cooling. In addition, these results indicate that the PTC will be
able to meet the challenging cost goals and the performance suggested by the HTS application
companies.
REFERENCES
1. J. Liang, A. Ravex and P. Rolland, “Study on pulse tube refrigeration”, Cryogenics, Vol. 36 (1996),
pp. 87-106.
2. D.L. Gardner, and G.W. Swift, “Use of Inertance in Orifice Pulse Tube Refrigerator,” Cryogenics,
Vol. 37 (1997), pp. 117-121.
3. Pat R. Roach and Ali Kashani, “Pulse Tube Coolers with an Inertance Tube: Theory, Modeling and
Practice,” Advances in Cryogenic Engineering, Vol. 43, Plenum Press, New York (1998), pp. 1895-
1902.
4. M. Nisenoff, Cryocoolers for Electronic Technologies, M-CALC II Workshop Report, San Diego,
1998.
5. T. Nast, P. Champagne, and V. Kotsubo, “Development of a Low-Cost Unlimited-Life Pulse-Tube
Cryocooler for Commercial Applications,” Advances in Cryogenic Engineering, Vol. 43, Plenum
Press, New York (1998), pp. 2047-2053.
6. J.L. Martin, J.A. Corey, and C.M. Martin, “A Pulse Tube Cryocooler for Telecommunications
Applications,” Cryocoolers 10, Plenum Press, New York (1999), pp. 181-189.
7. S-Y Kim et al., “Development of low-cost Pulse Tube Cryocooler for HTS Applications,” Advances
in Cryogenic Engineering, Vol. 45, Plenum Press, New York (2000), pp. 19-24.
TES FPC Flight Pulse Tube Cooler System
J. Raab, S. Abedzadeh, R. Colbert, J. Godden, D. Harvey, C. Jaco

TRW
One Space Park
Redondo Beach, CA 90278 USA
ABSTRACT
The TRW Tropospheric Emission Spectrometer (TES) Focal Plane Cooler (FPC) features two
integral pulse tube cryocoolers that independently control the temperature of the two instrument
focal planes. The TES mission acquires high-resolution ozone concentration data in the earth’s
troposphere in order to better understand the ozone: where it comes from and its interaction with
other chemicals in the atmosphere. TES is scheduled to fly on the EOS-Aura platform in 2002.
The TES FPC program delivered two flight coolers and electronics, and one flight spare cooler
and electronics in November 1999. This paper presents data collected on the flight coolers during
acceptance testing. Tests included thermal performance mapping at various reject temperatures and
power levels, launch vibration testing, EMC/EMI testing, and self-induced vibration testing.
Designed conservatively for a six-year life, the coolers are required to provide 1W cooling at
57K while rejecting to 35°C with less than 63W input power to the electronics. The system (cooler
and electronics) required mass is less than 17.1 kg. The system also includes radiation-hardened
control electronics and provides cooler control functions with a software-controlled microproces-
sor.
INTRODUCTION
The TES FPC program delivered two flight cooler systems and one flight spare cooler system,
plus ground support electronics (GSE) to interface with the cooler system. The program was per-
formed for the Jet Propulsion Laboratory (JPL) over a 31-month period. The TES mechanical
cooler is shown in Figures 1 and 2.
The cooler provides focal plane array (FPA) cooling via a thermal strap and rejects heat to a
loop heat pipe attached to a radiator. The TES FPC system, which is the latest version of the TRW
100 series coolers, consists of the mechanical pulse tube (MPT) cooler with attached accelerometer
electronics, and separately, the cooler control electronics (CCE).
The mechanical cooler is derived from the AIRS cooler which was a split pulse tube cooler. The
TES cooler has been reconfigured into an integral configuration with the same cold head and the
same compressor as the AIRS cooler. The electronics is the same basic design as the AIRS and
MTI (currently in orbit) electronics except that the producibility was upgraded and the software
was made more user friendly.
Before installation and operation of the cooler on the instrument, both the mechanical and the
electronics assemblies together with the operating software underwent flight level acceptance test-
ing, including environmental tests of launch vibration, thermal vacuum cycling, EMI/EMC testing,
132 SPACE PULSE TUBE CRYOCOOLER DEVELOPMENTS
Figure 1. Integral vibrationally-balanced pulse tube cooler.
Figure 2. TES FPC envelop for the mechanical cooler.

TES FPC FLIGHT PULSE TUBE COOLER SYSTEM 133
and burn-in. These tests, which are typical for space instruments, are performed to ensure reliabil-
ity. The cooler performance, including load lines, temperature stability, self-induced vibrational
force, and EMI/EMC properties, was measured. This paper reports the test data for one of the new
flight coolers. There was less than 6% input power difference from unit to unit at the nominal
operating condition of 1W at 57 K.
COOLER SYSTEM
The mechanical cooler (Figure 1) refrigerates via the cold block, rejecting heat at the centerplate
of the compressor. Inside the compressor, flexure springs support the moving-coil linear motor,
which drives the pistons. The springs maintain alignment for the attached non-contacting piston
that oscillates and compresses gas into the pulse tube cold head. A small clearance between the
cylinder and the piston seals the compression space. Two opposed compressor halves vibrationally
balance the compressor. The compressor is operated at the resonant frequency of 44.6 Hz.
Capacitive sensors are used to measure the position of both pistons. The output is used to mea-
sure and control dc offset and to provide overstroke protection. The pulse tube cold head is bolted
to the compressor centerplate, and is sealed with a metal seal. The centerplate conducts heat to the
radiator and incorporates the reservoir tank. The cold head components are arranged linearly:
mounting flange, regenerator, cold block, pulse tube, and warm-end heat exchanger body (or ori-
fice block). The cold head is surrounded by an H-bar that supports and provides a thermal path to
remove heat from the orifice block. The stainless steel orifice line and bypass line connect the gas
from the orifice block to the reservoir tank and to the aftercooler, respectively.
The internal wiring in the compressor is stranded, PTFE-insulated (cross-linked Teflon) wiring,
or Kapton flexible cable. All wiring exits the centerplate through ceramic-insulated pins in
feedthroughs attached to D-shell connectors for the cooler drive power and to the capacitive sen-
sors and thermistor. A separate connector is used for the redundant platinum resistance thermom-
eters (PRTs) on the cold block. Redundant accelerometers are mounted on the compressor centerplate.
Together with the signal conditioning electronics, the accelerometer provides a feedback signal to
the vibration control algorithm in the cooler control electronics (CCE).
The CCE (Figure 3) is based on our high-reliability AIRS flight design1 modified forproducibility.
New features include the horizontal slice design as shown in Figure 3 and additional internal con-
nectors to allow for slice-level testing. There are three slice subassemblies, one for control (control
slice), one for power amplifiers (power slice), and one for power conversion (converter slice). The
Figure 3. Cooler Control electronics (CCE).

slices are housed in a standard subassembly that is 225 mm (L) x 216 mm (W) x 175 mm (H). The
bottom of the housing serves as a mounting surface for direct thermal contact. The electronics in the
CCE: (1) converts the 28 Vdc primary power to the secondary power, (2) drives the cooler, and (3)
provides communication with the host and control of the cooler with a processor using software
resident in PROM. The software performs the following functions:
• Transmits spacecraft command and cooler telemetry via the RS422 data bus
• Collects the cooler state of health data
• Controls the cold block temperature
• Actively balances vibration force by controlling the waveform of the pistons
• Provides safety protection to the cooler
COOLER OPERATION AND CAPABILITIES

Table 1 summarizes system weight and capabilities. The cooler electronics provide AC drive
power at 44.6 Hz to the motors in the compressor. The compressor moving coil and piston assem-
blies are designed to resonate on their gas and mechanical springs at this drive frequency, and thus
generate a 44.6 Hz pressure wave and mass flow to the cold head. The software adjusts the stroke to
maintain the desired cold block temperature. The vibration control algorithm samples the acceler-
ometer signal and determines, by Fourier analysis, transfer gains and error signals for up to 16
harmonic frequencies. The error signal modifies the motor drive waveform to reduce vibration.
Figures 4 and 5 show the cooling load as a function of cooling temperature for different reject
temperatures and input powers. For the TES FPC nominal cooling load of 1.0 W at 57 K, the cooler
system requires 58.7 W of input power and the compressor operates at 45.5% stroke. For a TES
FPC cooling load of 0.5 W at 57 K, the cooler system requires 34.5 W of input power and the
compressor operates at 35.8% stroke.
The CCE (Figure 3) plays a critical role in the overall cooler performance. When the input bus
power was measured as a function of the output power to the compressor it fit the straight-
line correlation:
where the efficiency is 0.825, and the extrapolated tare power at zero compressor power is
Figure 4. Cryocooler performance for variable reject temperatures: 1W at 57K.
Figure 5. Cryocooler performance for variable reject temperatures: 0.5 W at 62K.

Figure 6. Temperature stability maintained by control loop during simulated orbital temperature change.
The temperature control algorithm adjusts the stroke level based on the difference between the
cold block PRT temperature and the set point temperature value. Figure 6 shows that the tempera-
ture stability of the cooler is within a 22 mK band when operating with a 1-W load at 57 K and a
baseplate temperature change of 0.21°C/min. The resolution of one bit in the temperature measure-
ment electronics is 12 mK.
The vibration control algorithm continuously updates the compressor waveform to minimize
cooler vibration. TRW’s special purpose dynamometer measures the three axes of the self-induced
vibration of the cooler. Figure 7 shows the force in the direction of piston motion (cooler axis) as
well as the two cross axes when the cooler is mounted on a rigid structure.
ENVIRONMENTAL TESTS AND COOLER ACCEPTANCE

The TES FPC acceptance testing included launch random vibration, a thermal vacuum test with
operating and non-operating temperature cycles, and burn-in. Levels and ranges for these tests are
summarized in Table 1. Repeatable cooler performance after each environmental test is used as an
acceptance criterion. The cooler was accepted because no performance change of the load line was
detected within experimental uncertainty. The measured helium leak rate was two orders of magni-
tude less than the 5-year-life criterion and satisfied a 10-year-life requirement.
EMI/EMC TEST
The cooler system must meet stringent requirements for radiated electric and magnetic fields,
conducted emissions on the input bus power lines, and electromagnetic susceptibility. Excessive
magnetic fields are a generic issue with linear-motor cryocoolers, as are excessive levels of input
ripple current. The TES FPC is an integral version of the TRW AIRS design, which required mag-
netic shielding to pass the radiated magnetic emission requirement.2 The TES FPC will be fitted
with this shield design. TRW’s newer cooler designs meet the radiated magnetic emission require-
ments without the need for shielding.3
The in-rush currents and the ripple currents for the nominal 29 V bus voltage requirement were
Figure 7. Measured self-induced force of compressor in three axes at 46 compressor stroke level and
42.6W into compressor.
recorded as 4.2 amps and 123 dB micro amps, respectively (Table 1). TRW’s latest electronics
have modified the TES FPC design to considerably lower the ripple current.4
The EMI and compatibility EMC qualification tests were performed at the TRW EMI test fa-
cilities to determine the degree of compliance to Mil-Std 461C requirements, as modified in TRW
BDA-14A-001, EMC Test Procedure for the TES FPC program. Table 2 summarizes the test ma-
trix. During the test series, two tests failed to meet the requirement. For the conducted emissions
(CE03), an overlimit condition was observed at 200 and 100 kHz in the narrowband mode (Table
2). An external filter at the input of the CCE will enable the cooler system to meet CE03.
CONCLUSIONS
The TES FPC performance met the program goals. The coolers were delivered in October 1999
and are awaiting integration with the payload.
ACKNOWLEDGMENT
The work described in this paper was carried out by TRW and sponsored by the JPL TES
Project. This report was prepared for the Jet Propulsion Laboratory, California Institute of Tech-
nology, sponsored by the National Aeronautics and Space Administration.
REFERENCES
1. Chan, C.K., J. Raab, A. Eskovitz, A.R. Carden III, R. Orsini, “AIRS Flight Qualified 55K
Pulse Tube Cooler,” Cryocooler 9, Plenum Press, NY (1997), pp. 895-904.
2. Johnson, D.L., S.A. Collins, and R.G. Ross, Jr., “EMI Performance of the AIRS Cooler and
Electronics,” Cryocooler 10, Kluwer Academic/Plenum Publishers (1999), pp. 771-775.
3. Chan, C.K., T. Nguyen, R. Colbert, J. Raab, R.G. Ross, Jr., and D.L. Johnson, “IMAS Pulse
Tube Cooler Development and Testing,” Cryocooler 10, Kluwer Academic/Plenum Publish-
ers (1999), pp. 139-147.
4. Chan, C.K., Pamela Clancy, and John Godden, “Pulse Tube Cooler for Flight Hyperspectral
Imaging,” Cryogenics, 39, Elsevier Ltd. (1999), pp. 1007-1014.
The AIM Space Cryocooler Program
I. Rühlich, H. Korf and Th. Wiedmann

AEG Infrarot-Module GmbH
Theresienstr. 2
74072 Heilbronn, Germany
ABSTRACT
AIM is developing space cryocoolers for superconducting telecommunication components.
The equipment should be available in 2001, and a mission to the International Space Station is
scheduled for 2003. The basis for the cooler development is the AIM model SL200 cooler; it has a
nominal cooling capacity of 3.5 W at 77 K, and is used for cooling high performance IR detectors.
The space mission will have a duration of three years with two years in operation. The development
was structured into the following phases: spin-off of classic design, implementation of flexure
bearings, introduction of pulse tube cold head. Results are improved COP and lifetime. The current
cooling capacity is about 4.3 W at 77 K with 96 W of input power. The expected lifetime exceeds
30,000 h. The qualification testing will start in July 2000.
INTRODUCTION
The cooler AIM SL200, which is currently mainly being used for military IR equipment and
for HTSC components, has to be improved to serve as a cooler for space missions in the 5 W-class.
The design and performance of the SL200 are an ideal basis to be improved for enhancement of
performance and reliability for mid term space missions (SL400).
The cooler development at AIM is embedded in a government program “Superconductor and
novel ceramics for communication technology of the future.” 1 A major milestone in the program
is a mission to the International Space Station (ISS) in 2003 for testing HTSC filters in a cryogenic
platform. The program is coordinated by BOSCH Telecom. To provide redundancy, two coolers
are planned for the platform.
DESIGN DESCRIPTION
The AIM SL400 has a classic linear-cooler design with a double-acting compressor and a
proven 12 mm cold head. The design has demonstrated sufficient life performance to serve for
medium term space missions, even without flexure bearings. The SL200 is a design equivalent to
the SL100, which consistently exceeds 4500 h MTTF in accelerated life testing. Rather than its
reduced cooling performance, the limiting factor preventing higher MTTF during life qualification
is the increased vibration output at end of life. Design features of all SL coolers at AIM are equiva-
lent, providing a profound basis for a thorough redesign for the space demands.

In order to improve the performance, the following components of the cooler have been rede-
signed:
• Replacement of Samarium Cobalt magnet material with high density Neodymium Iron Bo-
ron magnets
• Installation of magnetic field liner with higher magnetic field saturation and also smaller
hysteresis losses
• Optimization of the resonance conditions and of the regenerator in the cold head
The increase of life time was accomplished by:
• Special coating of piston in compressor and of displacer providing practically friction free
and wear tree operation of the clearance seal
• Current leads to moving coil like professional loudspeaker design
• Compressor spring
The current design has the potential to exceed 10,000 h MTTF. Verification is in progress
with three Engineering Models (EM).
According to step 2 of the program plan, and in parallel with the qualification of the current
configuration, a flexure bearing compressor is also in preparation together with a pulse tube cold
head (in collaboration with Gießen University). These improvements will push the MTTF to ex-
ceed 30,000 h.
Outline dimensions of the cooler are given in Fig. 1 (the length of transfer line for the space
mission is about 180 mm); photographs of the cooler and electronics are shown in Figs. 2 and 3.
The envelope dimensions of the electronics are 180 x 183 x 63 mm. Its weight is about 1.9 kg.
The electronics provides an adjustable temperature control algorithm and the ability to switch
between the regulated mode and an override fixed-output mode. The latter is used as an emergency
mode and to provide a base load in cases where the other cooler is operating in the regulated mode.
Furthermore, the input power into the electronics and the current operational mode can be moni-
tored by telemetry signals. The internal PID algorithm works with a real power control algorithm
for the motor current.
Figure 4 illustrates the configuration of the cooler in the breadboard platform. Besides the
AIM cooler (left hand side), another cooler in the sub-10W class, fabricated by LEYBOLD, is
foreseen for the experiment. Each of the coolers is equipped with a single separate electronics. Both
coolers can operate either separately or jointly, one providing the base load while the other operates
in the temperature control mode.
Figure 1. Outline dimension of cooler.

THE AIM SPACE CRYOCOOLER PROGRAM 141
Figure 2. SL400 cryocooler.
Figure 3. Space Electronics, developed by Astrium.
Figure 4. Breadboard platform with cryocoolers (courtesy BOSCH Telecom).

Figure 5. Cooling capacity of the SL400.
Figure 6. Comparison of SL400 Carnot fraction with other coolers.2,3
PERFORMANCE DATA
Typical heat load requirements for HTSC applications are in the range of about 4 W at 77 K.
As the total heat load of the cryogenic platform being developed by BOSCH Telecom is also ap-
proximately 4 W, the performance of the coolers have to exceed 4 W to meet the cooling require-
ments. Beside that, high efficiency was a major requirement for the development.
Cooler
So far 4.3 W at 77 K with 96 W of input power has been achieved. The cooling capacity vs.
input power for different ambient conditions is shown in Fig. 5. The Carnot fraction of the SL400
and other cryocoolers in the range of cooling capacities between 1 W and 10 W is shown in Fig. 6
for comparison.2,3
The specific cooling capacity is another important criterion, especially for space applications.
The cooling capacity vs. cooler mass is shown in Fig. 7. Three curves for specific cooling capacity
are given.
THE AIM SPACE CRYOCOOLER PROGRAM 143
Figure 7. Comparison of cooling capacity vs. mass for SL400 and others.2,3
Electronics
The characteristics of the electronics are as follows:
Control temperature setting 47 K - ambient
Temperature stability
Temperature sensor PT 100
Control algorithm PID
Operation voltage
Efficiency
STATUS OF QUALIFICATION
The assembly of Engineering Models for the space qualification is completed. The general
qualification at the system level will be performed at BOSCH Telecom. After the final test at AIM
the delivery is scheduled for the end of June 2000. Vibration tests of the bare cooler against the
levels given in Tables 1 and 2 have been performed successfully. The limitation for the elastic
deflection of the cold head during launch vibration has required some special solutions. The key
issue is the high mass of the thermal interface from the cold tip to the flexible thermal connection to
the cryogenic platform; this mass leads to low frequency resonances, and thus to high deflections.
BOSCH therefore has developed a special Kevlar support with high stiffness and low axial conduc-
tion. The assembly of the cold finger and support has successfully been tested at qualification
levels.
For verification against shock loads, 10 shocks per axis will be tested. The levels are given in
Table 3.
CONCLUSIONS
AIM is performing a program for the development of space cryocoolers in the 5W class.
Engineering models for the space qualification at BOSCH telecom are about to be delivered. The
cooler SL400 has a cooling capacity of 4.3 W at 77 K with 96 W of input power, which is a 12.7%
Carnot fraction. The qualification will be finished in fall 2000. A mission to ISS is scheduled for
2003.
ACKNOWLEDGMENT
Financial support by a BMBF grant (13 N 7391) is gratefully acknowledged. The authors
would like to acknowledge the good cooperation with BOSCH Telecom, LEYBOLD VAKUUM
and Astrium.
NOMENCLATURE
EM Engineering Model
g Constant of gravitation
MTTF Mean time to failure
RMS Root mean square
REFERENCES
1. Schrempp, Ch., Klauda, M., Neumann, Ch., “Design of a Cryogenic Platform for New Communi-
cation Payload Technologies,” SAE Paper 1999-01-2086, 29th International Conference on Environ-
mental Systems, Denver, 1999.
2. “Cryocooler Survey 1998,” CD-ROM by Nichols Research Corp., Albuquerque, NM.
3. Glaister, D.S. et al., “An Overview of Performance and Maturity of Long Life Cryocoolers for Space
Applications,” Cryocoolers 10, Plenum Press, NY (1999), p. 1.
Miniature Pulse Tube Cryocooler for
Space Applications
T. C. Nast, P. J. Champagne, V. Kotsubo, J. Olson,

A. Collaco and B. Evtimov
Lockheed Martin Advanced Technology Center
PaloAlto, CA 94304-1191
T. Renna
Lockheed Martin Communications and Power Center
Newton, PA
R. Clappier
Clappier Consulting
Discovery Bay, CA 94514
ABSTRACT
Lockheed Martin’s Advanced Technology Center (LM ATC) has developed a miniature,
lightweight pulse tube cryocooler system for space operation under funding from NASA/GSFC.
The cold end is a U-tube configuration, and is driven by a dual opposed piston flexure bearing
compressor. The compressor utilizes a moving magnet linear motor and incorporates a number
of features that simplify assembly and enhance reliability. This cooler is designed for 0.3 W of
cooling at 65 K with a 310 K rejection temperature, 15 W of compressor power, and a mass of
less than 1.25 kg. Three engineering model cryocoolers are to be delivered to NASA/GSFC.
Two have been completed and test data is presented here on the EM performance.
In a parallel effort, we have developed a lightweight, power efficient electronic controller
funded by LM ATC cost sharing funds for a DARPA contract. The first Engineering Model of
this controller has been completed and functional testing has verified successful operation. All
of the design goals have been met or exceeded.
INTRODUCTION
Numerous future spaceflight missions require a very light weight, compact cryocooler
system with lifetimes of 10 years or more. Cryocooler customers also want lower costs and
shorter delivery times. NASA-GSFC awarded a development contract to Lockheed Martin’s
Advanced Technology Center (LM ATC) for the thermo-mechanical system in Sept. 1997.
The cooling requirements are 0.3 W at 65 K with 15 W of compressor power with a rejection
temperature of 310 K. Qualification and cooling performance is performed from 250 to 310 K.
With the requirement for lifetimes in excess of 10 years, LM ATC selected the no-mov-
ing-parts pulse tube coldhead driven by a flexure-bearing clearance-seal compressor. At LM

ATC we have achieved pulse tube efficiencies comparable to Stirlings,1-4 such that the cooling
requirements of the NASA-GSFC contract could be met with a pulse tube. The flexure-bearing
compressor has now been demonstrated to be a reliable, robust technology. For this program,
LM ATC is using a moving magnet compressor that has simplified assembly, reduced cost and
enhanced reliability over the standard Oxford-heritage compressor with moving coil.
The NASA/GSFC contract calls for the delivery of 3 EMs. The first has completed testing,
the second is in early stages of testing, and the third is undergoing final assembly. The load
lines from EM #1 and early test results from EM #2 are presented here.
LM ATC also recognized the need for a smaller, lighter, lower cost electronic controller
with substantially improved reliability over previous versions. LM ATC initiated a program
early in 1998 to develop a second-generation electronic controller.
The circuits and control logic were designed by a consultant and LM ATC, with the devel-
opment and the manufacture of the EMs and FMs by Lockheed Martin’s Communication and
Power Center (LM CPC).
An Engineering Model of this controller has recently completed functional testing and all
goals of the design have been achieved.
SYSTEM DESCRIPTION
Cryocooler Thermo-Mechanical Unit (TMU)
The thermo-mechanical unit is an integral configuration with a U-tube coldhead directly
mounted to the compressor. This arrangement requires only a single spacecraft-mounting in-
terface for both structural support and heat rejection. Mounting options are available as re-
quired by the customer. The compressor utilizes a linear motor, with a dual-opposed configu-
ration for momentum compensation. Figure 1 shows the general configuration of the system,
which includes the electronic controller. All components of the cyrocooler are packaged within
the configuration shown. No additional hardware is required.
The U-tube coldhead typically simplifies integration to the instrument, reduces overall
weight, and eliminates the need for a second heat rejection point at the warm end of the pulse
Figure 1. Cryocooler and electronic controller.

MINIATURE PT CRYOCOOLER FOR SPACE APPLICATIONS 147
Figure 2. Static load test of cold head.
tube as required by in-line designs. The U-tube is also structurally more rigid than the in-line
coldhead, making it less susceptible to damage from launch vibration and instrument side load
forces. The side load capability of our cold tip in the weakest of two lateral axes is predicted to be
6.5 kg. Figure 2 shows the cold end under a static load test supporting 5 kg.
The compressor is designed for low manufacturing costs while still maintaining the reliability
of the flexure-bearing compressor with non-contacting piston-cylinder seals. The architecture is
based on a larger compressor originally developed by LM ATC under NASA funding for a low cost
commercial cryocooler.5 This approach utilizes a moving magnet design, with the drive coil exter-
nal to the working gas space. By placing the coil outside of the pressure vessel, we eliminate the
single largest source of contamination of the working gas, the organic coil potting. Likewise, the
position sensor’s electrically active element is also outside of the pressure vessel, and thus electri-
cal penetrations through the pressure wall are completely eliminated, removing gas leakage through
the electrical feedthrough as a potential failure mode. The stationary coil also eliminates breakage
of flexing leads of moving coil motors as a failure mode.
The compressor incorporates several self-aligning features for the piston/motor/flexure assem-
bly, and a low piecepart count which simplifies the assembly and shortens the assembly time. The
critical piston-cylinder seal utilizes a simple alignment adjustment mechanism that rapidly and
repeatably performs this task. This mechanism can be computer automated. These compressor fea-
tures will reduce costs in large volume manufacturing. In small quantities, they enhance reliability
by reducing the risk of workmanship defects. Table 1 summarizes these features.
Cooler Drive Electronics (CDE)

The electronic controller being developed has full capability for operating the cryocooler
in space environments. It has a PWM amplifier for driving the compressor, feedback control of
Figure 3. Test setup for thermodynamic testing.
piston amplitudes for temperature control, and feedback control of one of the pistons for reduction
of exported axial vibration. An RS-422 port provides interfacing to the spacecraft for command and
telemetry.
This controller is substantially lighter, more power efficient, more reliable, and consumes less
overhead power than LM ATC’s existing Stirling cooler flight electronic controller. The produc-
tion cost will also be substantially lower than the existing controller. These advances were achieved
by simplifying all aspects of the controller, including the control algorithms and the circuitry. Elimi-
nation of the displacer drive and control circuitry associated with the Stirling cycle also results in
further simplification.
SYSTEM PERFORMANCE
Thermo-mechanical Unit
Figure 3 shows the test setup for thermodynamic testing.
Thermodynamic Performance. Performance tests have been conducted on the EM unit at
various power inputs and heat rejection temperatures. Figure 4 presents the load lines at several
power inputs for heat rejection temperatures of 280 K for EM #1. Figure 5 presents the effect of
heat rejection temperature on cooling loads. The requirement of 0.3 W of cooling at 65 K with a
Figure 4. Load curves for EM #1.

Figure 5. Effect of heat rejection temperature on

cooling capability with 15 W compressor power (EM#1).
310 K reject temperature was met with 18 W of compressor power. At 80 K, the 0.5 W cooling
power requirement with a reject temperature of 310 K was met with 16.4 W of compressor
power for EM #2, and reflects improved cooling over EM #1 resulting from improvements.
Launch Vibration. Preliminary launch vibration tests have been conducted to verify in-
tegrity of the hardware. Figure 6 shows the cooler mounted on the vibration fixture and sum-
marizes the launch load vibration environment.
Life Testing. Life tests have been conducted on a version of the mini cryocooler which
was utilized for cooling of a high temperature superconducting filter package. In this test, the
compressor and cold head were identical to the mini described in this paper except it was a
“split” version in which the compressor and cold head were separated by 6 cm to facilitate
packaging.
In this test the unit ran continuously for 6,000 hours, except for infrequent power outages.
Figure 6. Launch load environment and test setup.

Figure 7. Mini pulse tube cryocooler mounted on 6 axis dynamometer.
Figure 8. Induced vibration along drive axis; 60 Hz drive frequency.
Induced Vibration
Induced vibration tests were conducted on the cryocooler mounted on the LM ATC six axis
dynamometer.
The test setup is shown in Fig. 7 and the measured forces are presented in Fig. 8 in the drive
axis. In these tests the units were run at full stroke, with no closed-loop feedback control. Forces in
the two lateral axes were below the peak value of 0.22 N. Induced vibration forces with the closed
loop of the electronic controller are in evaluation, but are expected to be lower.
Table 2 summarizes the EM parameters against the NASA/ GSFC contract specification.
Figure 9. View of top-assembly and two boards.
Electronic Controller
The cooler drive electronics (CDE) is comprised of two major subassemblies: (1) the Con-
trol Board Assembly, and (2) the Power Board Assembly. Major features of the design are
described below for each subassembly. Photographs of the control board and power board are
included in Figure 9.
Control Board Assembly. The control board contains the RS422 command and telem-
etry to the spacecraft (user interface), the temperature control function, and piston position
limiting, and signal conditioning circuits required to drive the two PWM motor drive amplifi-
ers. In addition, optional vibration cancellation control circuits (AFFECS – Analog Feedforward
Error Correction System) are included in the circuit board design and can be configured in
manufacturing depending on the level of vibration cancellation required. Most of the control
features are embedded in an FPGA. A UART is included as part of the RS422 interface. Other
options include a variable compressor drive frequency for optimization of cooler performance,
and to meet specific customer requirements.
Power Board Assembly.
Motor Drive Amplifiers. The compressor motors are driven using a high-frequency PWM am-
plifier. There are two identical amplifiers per assembly. The amplifiers include control circuit inter-
faces to the control board that enable accurate current control of the compressors in order to elimi-
nate influence of the +28 V bus on the compressor power and also to allow for accurate harmonic
cancellation. At the compressor interface, active clamp circuits are included for use during launch
to minimize compressor piston excursion by providing magnetic damping.
Power Supply. The power supply section contains a PWM DC-DC converter to provide
the necessary bias voltages to the entire CDE. The converter design is optimized for reduced
power consumption to help minimize the overall bias power required for the CDE. As part of
the power supply section, the +28 V input bus is filtered using an LC-filter to reduce both
differential and common-mode noise within the specified EMC limits. There is an option to
include an inrush current limited (may not be necessary due to external bus inrush limiters).
Mechanical Packaging. The CDE is housed in an aluminum chassis in an H-Frame con-
figuration. A view of the top-assembly was previously shown in Fig. 9. The control board is
bolt-mounted into one cavity of the chassis, while the power board is bolted into the opposing
cavity. This separation isolates the control circuits from any EMI from the PWMs and power
supply. The two cavities are enclosed using an aluminum cover design specifically to reduce
radiated emissions. Each double-sided printed circuit board utilizes both through-hole and sur-
face mount components. The entire assembly was designed to optimize use of board area lead-
ing to a minimum overall weight.
Qualification Status
CDE Functional Testing – An EM of the CDE has been assembled and tested. A brief
description of the testing follows for each level of assembly.
Control Board Assembly – The control board EM has been functionally tested at room
ambient temperature. Simulated acceleration feedback, position feedback and temperature feed-
back loops were used to verify acceptable performance. Interface through the RS-422 port was
exercised to verify proper commanding and telemetry read-back. All functions operated as
expected during this level of test.
Power Board Assembly – The power board was functionally tested at room ambient tem-
peratures to verify performance prior to integration with the control board. Each PWM ampli-
fier was exercised and performance data gathered for efficiency, output harmonic distortion
and proper drive amplitude control levels. The efficiency exceeded the 90% goals and typi-
cally measured 93% at maximum output power. The inrush limiter and power supply were also
tested to verify their expected operation.
CDE Assembly – The control board and power board were integrated and functionally
tested at room ambient temperature to ensure proper operation prior to integration with the
compressors. The current control loop was tested in this configuration.
Integrated CDE and Cryocooler Testing – Once the CDE was functionally tested, it was
integrated with the cryocooler. Testing included verification of the temperature control loop,
AFFECS vibration cancellation, and position loop.
A summary of the system characteristics and the original design goals is shown in Table 3.
Figure 10. Load line for high capacity mini version.
ADAPTABILITY AND ENHANCED COOLING CAPABILITY

A unique advantage of pulse tube coldheads is that they have very short development time
when compared with compressors and Stirling displacers. Thus, with a given compressor, coldheads
can be developed for specific customer needs to provide the optimum performance at specific tem-
peratures and cooling loads, or to provide specialized cooling configurations, such as multi-staging
, multiple coldheads6 or split systems.
We have performed numerous design studies exploring the range of cooling capacities and
configurations consistent with our compressor capability in response to specific customer require-
ments. At the high capacity end, we have achieved 3 W of cooling at 80 K with a redesigned
coldhead and 69 W of compressor power. The load lines for this configuration are presented in
Figure 10. We have also operated a modified version of the prototype cooler with over 40 W of
compressor power and produced 1.25 W at 77 K. This cooler, in a split configuration with a U-tube
coldhead, was used to cool an engineering model of a High-Temperature-Superconducting 4 GHz
Input Multiplexer for satellite communications systems,7 and accumulated 6,000 hours of semi-
continuous running without problems.
SUMMARY
Lockheed Martin’s Advanced Technology Center, under support from NASA-GSFC has de-
veloped a miniature, lightweight pulse tube cryocooler for space applications. The contract calls for
delivery of three EM units. EM #1 and #2 have been completed and are under test, and #3 is in final
assembly. They produce 0.3 W at 65 K and 0.5 W at 80 K at a reject temperature of 310 K with
18 W and 16.4 W of compressor power, respectively, and weigh only 1.32 kg.
The design approach of these units leads to reduced cost and improved reliability. Modified
versions of the same size and weight have shown cooling capabilities of 3 W at 80 K and 2 W at 65 K.
A simpler, lighter weight controller has been developed to drive the cryocooler. The controller
has a mass of 1.6 kg and greater than 90% conversion efficiency with 3 W of overhead power. It has
full capability, including temperature control and vibration control and full telemetry to the space-
craft. It has a reduced number of parts contained in two boards which result in lower cost and higher
reliability than prior versions.
ACKNOWLEDGMENT
This work was supported by NASA/GSFC (the thermal mechanical unit) and by DARPA
and Lockheed Martin internal funding (electronic controller).
REFERENCES
1. D.L. Glaister, M. Donabedian, and D. Curran, An Overview of the Performance and Maturity of Long
Life Cryocoolersfor Space Applications, Aerospace Report No. TOR-98 (1057)-3, (1998).
2. V. Kotsubo, J. R. Olson, and T. C. Nast, “Development of a 2 W at 60 K Pulse Tube Cryocooler for
Spaceborne Operation,” Cryocoolers 10, Plenum Press, New York (1999), pp. 157-161.
3. T.C. Nast, P. Champagne, J. R. Olson, and V. Kotsubo, “Development of Pulse Tube Cryocoolers for
HTS Satellite Communications,” Cryocoolers 10, Plenum Press, New York (1999), pp. 171-179.
4. W.W. Burt and C.K. Chan, “New Mid-Size High Efficiency Pulse Tube Coolers,” Cryocoolers 9,
5. T. Nast, P. Champagne, and V. Kotsubo, “Development of a Low-Cost Unlimited Life Cryocooler for
Commercial Applications,” Adv. Cryo. Eng., 43a, Plenum Press, New York (1998) p. 2047.
6. J.R. Olson, V. Kotsubo, and T.C. Nast, “Multiple Pulse Tube Coldheads Driven by a Single Compres-
sor,” Adv. Cryo. Eng., 45, Plenum Press, New York (2000).
7. T.C. Nast, B.G. Williams, V.Y. Kotsubo, J. R. Olson, and D. J. Frank, “Development of a Cryogenic
60 Channel, HTS Multiplexer,” Adv. Cryo. Eng. 45, Plenum Press, New York (2000).
Gamma-Ray Pulse Tube Cooler
Development and Testing
R.G. Ross, Jr., D.L. Johnson, A. Metzger

Jet Propulsion Laboratory, California Institute of Technology
Pasadena, California 91109
V. Kotsubo, B. Evtimov, J. Olson and T. Nast
Lockheed Martin ATC, Palo Alto. CA 94304
R.M. Rawlings
DRS Infrared Technologies, Dallas, TX 75243
ABSTRACT
For a variety of space-science applications, such as gamma-ray spectroscopy, the introduction
of cryogenic cooling via a cryocooler can greatly increase the potential science return by allowing
the use of more sensitive and lower noise detectors. At the same tune, the performance benefits
must be carefully weighed against the implementation cost, any possibility of degraded detector
performance associated with the operation of the cryocooler, and the requirement to achieve long
life. This paper describes the development, test, and performance of a novel new low-cost, low-
noise, high-reliability pulse tube cooler, designed specifically for highly cost-constrained, long-
life space missions.
The developed cooler marries two technologies: a low-cost high-reliability linear compressor
and drive electronics from the 1.75 W tactical Stirling cryocooler of DRS Infrared Technologies
(formerly Texas Instruments), and an 80 K pulse tube developed specifically for the compressor
by Lockheed Martin ATC. The successful new cooler achieves over 1.6 watts of cooling at 80 K
at 23 W/W, and has the advantages of greatly reduced vibration at the coldtip and no life-limiting
moving cold elements.
To achieve maximum life and low vibration, the compressor incorporates flat flexure springs
for piston support and uses two opposing pistons in a head-to-head configuration with linear drive
motors. The pulse tube is a compact U-tube configuration for unproved integration and is mounted
to the compressor in a split configuration with a transfer line.
INTRODUCTION
The object of this cooler development program was to make it possible to utilize high-resolu-
tion germanium (Ge) detectors for planetary gamma-ray spectroscopy on relatively low-cost space
missions involving one- to two-year operational lifetimes. Use of a germanium detector cooled to
around 80 K provides measurement sensitivities that are on average seven tunes greater than
commonly used uncooled scintillation gamma-ray detectors.
Until now, radiative cooling to space has been the best method for weight-limited, long-
duration planetary missions. Now, with reductions in size and power consumption, and improve-

ments in reliability, miniature mechanical coolers offer an increasingly attractive alternative. Their
use would eliminate the interface requirement for a three-axis stabilized spacecraft and a broad
unobstructed view to space, and would allow experiments in environments where radiative coolers
cannot function satisfactorily, notably in orbit around warm planets and on the surface of planets
and moons. Comparisons of size, mass, duty cycle and operating temperature make mechanical
cooling more attractive than radiative cooling for many planetary missions.
Developing a mechanically-cooled gamma ray spectrometer (GRS) for small, low-cost, plan-
etary missions requires a small, long-life, low-cost cooler from which vibration, capable of induc-
ing microphonics effects at the detector, has been eliminated. The cooler solution described here
is to combine the compressor from a low-cost, miniature, high-reliability, commercially-available
Stirling cycle tactical cooler, with a matched pulse tube expander made specifically for the com-
pressor. The intended result is to produce a cooler that has minimal mechanical motion at the
detector, retains the small mass and volume characteristics of the tactical cooler, and, thanks to
recent improvements in pulse tube efficiency, requires a relatively low level of spacecraft power.
The elimination of the tactical cooler’s Stirling displacer is expected to also add to the reliability
and lifetime of the cooler, and substantially reduce the vibration environment at the cold-load
interface. The use of a commercially available compressor and the simplicity of the pulse tube
design is expected to preserve most of the cost advantage of the tactical cooler relative to the
sophisticated long-life space coolers. The same relative simplicity is also expected to translate into
additional cost savings by allowing inexpensive tactical-cooler drive electronics to be used.
COMPRESSOR SELECTION AND DESIGN FEATURES

Central to achieving the cooler development objectives was the need to acquire a tactical
cryocooler compressor with proven long-life potential, low vibration, and compatibility with the
needed pulse tube expander in terms of swept volume and operating pressure. The specific design
goal was to achieve > 1.1 watt of cooling at 80 K with a compressor specific power of less than 25
watts/watt.
An analysis of available tactical cooler compressors led to the selection of an advanced 1.75 W
tactical Stirling cryocooler manufactured by DRS Infrared Technologies (formerly Texas Instru-
ments). The particular model, shown in Fig. 1, is based on an advanced linear compressor with
its two pistons operated head-to-head and supported on flexure springs to achieve long life and
good vibration suppression. This new flexure-supported compressor is one of a family of ad-
vanced flexure-equipped compressors being developed to achieve extended-life tactical coolers.1
Figure 1. DRS 1.75 W tactical cooler with drive electronics.

GAMMA-RAY PT COOLER DEVELOPMENT AND TESTING 157
Figure 2. Refrigeration performance of the DRS 1.75 W Stirling cooler as a function of input drive
voltage, coldtip temperature, and coldtip load.
The thermal performance of this cooler, using the Stirling expander shipped with the cooler, is
shown in Fig. 2. The more conventional DRS 1.75 W cooler has similar performance, but uses
helical coil springs to support the pistons and has a predicted life greater than 5000 hours. JPL
has had good success using the conventional (non-flexure-spring) DRS 0.2-watt, 1-watt, and 1.75-
watt Stirling coolers for a variety of low-cost, intermediate-life space missions.2,3,4
PULSE TUBE DESIGN AND CONSTRUCTION

The second task critical to achieving the required cooler performance was the development of
a high efficiency pulse tube expander carefully matched to the compression attributes of the DRS
compressor and the interface requirements of the JPL gamma-ray detector mounting system,
shown schematically in Fig. 3. This task, carried out by Lockheed Martin Advanced Technology
Center, involved first thoroughly characterizing the DRS compressor, then designing and fabricat-
ing a pulse tube consistent with the compressor and the gamma-ray detector cooling load and
mounting interfaces. The chosen concept was the U-shaped pulse tube shown in Fig. 3.
Compressor Characterization
To achieve an efficient design for the proposed pulse tube it was necessary to include an
accurate model of the compressor's performance in the overall pulse tube design analysis. To
acquire the needed data, the DRS flexure-bearing compressor was tested at Lockheed Martin ATC
with dead volumes to determine its characteristics. Tests were performed using three different
dead volumes (12. 1cc, 22.8cc, and 28.9cc), and five different charge pressures (300psia, 200psia,
100psia, 50psia, and 5psia).
The compressor was driven with a current controlled amplifier with an input signal given by
an HP signal generator. The power to the compressor was monitored with a Valhalla power meter,
and a calibrated pressure transducer, mounted in the compression space, monitored the pressure.
The resonant frequency at each charge pressure was determined by a frequency sweep search-
ing for the maximum voltage for a fixed drive current, for drive currents ranging from 0.1 A up to
1.3 A. At the resonant frequency, the current, voltage, power, and pressure amplitudes were
recorded. Table 1 presents a summary of the compressor parameters (for each compressor half);
most were determined from the measurements, while some were provided by DRS.
Compressor internal losses were also characterized to allow estimation of the expected effi-
ciency of the overall pulse tube cryocooler. Because the compressor exit-passage parameters were
designed for the standard DRS split-Stirling expander that has a small-diameter transfer line,
somewhat higher losses were predicted when used with the pulse tube, which requires a larger
transfer line. In the future, if more optimum performance from the compressor is desired, one
should consider enlarging the internal flow passages to tailor the compressor for improved opera-
tion with a pulse tube.
Pulse Tube Design

To achieve an efficient design for the pulse tube, detailed thermodynamic simulations were
conducted by Lockheed Martin of the entire cooler system. Key parameters included pulse tube
geometries, transfer line diameter, fill pressure, operating frequency, piston stroke, and pulse tube
reservoir-line tuning.
The resulting design was predicted to provide 1.2 W of cooling at 80 K with 30 W of total
compressor power and a piston amplitude of 2.9 mm. Note that the piston amplitude is well below
the maximum of 5 mm. The predicted cooling capacity is slightly higher than the required 1.1 W,
and the predicted specific power of 25 W/W matches the design goal.
The largest uncertainty in the prediction was the internal losses within the compressor, which,
in the dead volume tests, were particularly significant at high piston amplitudes. A conservative
empirical model was used to represent the compressor flow losses in the analysis, which tended to
reduce the piston amplitude in order to reduce the losses.
A series of parametric studies was performed to predict the sensitivity of the coldhead to
operating conditions. The efficiency of the coldhead was found to be relatively insensitive to mass
flow rates and frequencies, typical of other L-M pulse tubes. This indicates that the coldhead
design was not significantly influenced by the particular model used for the compressor losses.
Figure 4 shows the (pressure-volume) PV specific power as a function of input power. As
shown, the coldhead itself is predicted to have a PV specific power of 14 WAV at 80 K, compa-
rable to other coldheads developed at Lockheed Martin ATC.5,6 Lockheed's best in-line, high-
Figure 4. Predicted PV specific power as a Figure 5. Predicted cooling power dependence

function of input power. on charge pressure.
capacity designs are typically 11-12 W/W. The slightly lower efficiency of this pulse tube is due
in part to the U-tube configuration, the small diameter of the transferline internal to the compres-
sor, and to the smaller cooling capacity. At the lower power levels, the efficiencies are still good,
although slightly decreased from the higher power levels. At 10 W of PV power, the PV specific
power is predicted to be around 17.5 W/W.
Figure 5 shows the predicted cooling power as a function of charge pressure at 30 W of
compressor power. The frequency was varied for optimum performance, along with retiming of
the impedances. The regenerator and pulse tube remained fixed in the analysis. This plot sug-
gests that 400 psia would be a good working charge pressure, and shows that the cooler can
tolerate a slight reduction in charge pressure down to 300 psia or so without serious reduction in
performance. However, if the charge pressure is decreased down to 200 psia, the performance
begins to seriously degrade. If the regenerator and pulse tube were to be redesigned for lower
charge pressures, then the performance degradation would not be as severe as shown in Fig. 5.
The results of the overall simulation analyses indicated that the proposed coldhead should
perform well over a range of conditions. This is significant in that the compressor loss mecha-
nisms were not known in detail. Once the compressor and coldhead were integrated, it was
expected that minor tuning of the overall system would be able to achieve a good match between
coldhead and compressor.
Based on the modeling it was considered likely that the pulse tube would exceed the predicted
efficiency, since a conservative model was used for the compressor losses, a conservative model
was used for the motor force constant, and many Lockheed coldheads outperform their predic-
tions. Thus, it was expected that the coldhead would provide in excess of 1.1 W of cooling at
better than 25 W/W. In addition, the low design stroke would allow the cooler to be driven to
substantially higher strokes and power levels, although at a somewhat higher specific power.
Pulse Tube Fabrication

Once the analyses and component designs were complete, the pulse tube cooler components
were fabricated and assembled into a completed pulse tube expander. Figure 6 shows the piece
parts ready for assembly, together with a completed pulse tube. Figure 7 shows the complete
cooler setup during verification testing at Lockheed.
PULSE TUBE SYSTEM-LEVEL TESTING

After initial checkout and performance verification of the completed cooler at Lockheed
Martin ATC, extensive performance characterization testing was carried out at JPL in preparation
for planned tests to validate the vibration and EMI compatibility with an actual gamma-ray detec-
tor using the setup illustrated in Fig. 3.
Figure 8 presents the overall thermal performance measured at JPL as a function of coldend
Figure 6. Pulse tube expander piece parts and final assembly.
Figure 7. Completed
pulse tube cooler with DRS
compressor on the left, reser-
voir volume on the right, and
pulse tube with vacuum bon-
net assembly in the center.
Figure 8. Refrigeration performance of the completed gamma-ray pulse tube cooler as a function of input
drive voltage, coldtip temperature, and coldtip load.
Figure 9. Refrigeration performance of the completed gamma-ray pulse tube cooler as a function of
helium fill pressure.
temperature, coldend load, and input voltage (which is roughly proportional to stroke). Note that
the specific-power performance at 80 K is around 22 W/W, which is better than the design goal of
25 W/W, and that the overall cooler capacity is also better than the design requirement, reaching
over 1.6 watts at 80 K near full stroke (9 volts), in contrast to a requirement of 1.1 watt.
To confirm the cooler's predicted sensitivity to fill pressure and drive frequency, additional
parametric testing was conducted with these parameters as variables. The measured performance,
displayed in Figs. 9 and 10, confirm that fill pressure increases cooling capacity with minimal
effect on efficiency, while drive frequency, once the pulse tube volumes are fixed, is a relatively
sensitive parameter. For the as-fabricated pulse tube cooler, the best specific power is seen to
occur at a frequency of around 42 Hz.
Figure 10. Refrigeration performance of the completed gamma-ray pulse tube cooler as a function of drive
frequency.

This paper has described the development, test, and performance of a novel new low-cost,
low-noise, high-reliability pulse tube cooler, designed specifically for highly cost-constrained
long-life space missions such as planetary gamma-ray spectroscopy.
The developed cooler marries two technologies: a low-cost, high-reliability linear compres-
sor and drive electronics from the 1.75 W tactical Stirling cryocooler of DRS Infrared Technolo-
gies, and an 80 K pulse tube developed specifically for the compressor by Lockheed Martin ATC.
To achieve maximum life and low vibration, the compressor incorporates flat flexure springs for
piston support and uses two opposing pistons in a head-to-head configuration with linear drive
motors. The pulse tube is a compact U-tube configuration for improved integration and is mounted
to the compressor in a split configuration with a transfer line.
The successful new cooler achieves over 1.6 watts of cooling at 80 K at 23 W/W, and has the
advantage of greatly reduced vibration at the coldtip and no life-limiting moving cold elements.
ACKNOWLEDGMENT
The work described in this paper was carried out by the Jet Propulsion Laboratory, California
Institute of Technology, and by Lockheed Martin ATC under contract with JPL; it was sponsored
via the Planetary Instrument Definition and Development Program (PIDDP) through an agree-
ment with the National Aeronautics and Space Administration.
REFERENCES
1. Rawlings, R.M. and Miskimins, S.M., “Flexure Springs Applied to Low Cost Linear Drive Cryo-
coolers,” Cryocoolers 11, Kluwer Academic/Plenum Publishers, NY, 2001.
2. Glaser, R.J., Ross, R.G., Jr. and Johnson, D.L., “STRV Cryocooler Tip Motion Suppression”,
Cryocoolers 8, Plenum Publishing Corp., New York, 1995, pp. 455-463.
3. Ross, R.G., Jr., “JPL Cryocooler Development and Test Program: A 10-year Overview,” Proceed-
ings of the 1999 IEEE Aerospace Conference, Snowmass, Colorado, Cat. No. 99TH8403C, ISBN 0-
7803-5427-3, 1999, p. 5.
4. Johnson D.L., “Thermal Performance of the Texas Instruments 1-W Linear Drive Cryocooler,”
Cryocoolers 10, Kluwer Academic/Plenum Publishers, NY, 1999, pp. 95-104.
5. Kotsubo, V.,Olson, J.R., andNast.T.C., “Development of a 2W at 60K Pulse Tube Cryocooler for
Spaceborne Operation,” Cryocoolers 10, Kluwer Academic/Plenum Publishers, NY, 1999, pp. 157-
170.
6. Kotsubo, V., Olson,J.R., Champagne, P., Williams, B., Clappier, B. and Nast, T.C., “Development
of Pulse Tube Cryocoolers for HTS Satellite Communications,” Cryocoolers 10, Kluwer Academic/
Plenum Publishing Corp., NY, 1999, pp. 171-179.
High Efficiency Pulse Tube Cooler
E. Tward, C. K. Chan, J. Raab, T. Nguyen,

R. Colbert and T. Davis†
TRW, Redondo Beach, CA 90278
†
Albuquerque, NM 87117
ABSTRACT
The High Efficiency Cooler (HEC) is being developed in order to provide a long life, low
mass, high efficiency space cryocooler suitable for use on lightweight gimbaled optics on surveil-
lance missions such as SBIRS Low. This paper reports on the development and testing of this next
generation family of space pulse tube cryocoolers which feature high cooling capacity, lower mass,
lower EMI and lower self induced vibration than the current state of the art. The HEC achieves low
input power and large cooling power because of the efficiency of its pulse tube cold head and
highly efficient compressor. The low mass (<4.3 kg) results chiefly from its next generation Oxford
flexure compressor technology reported in a companion paper. The projected long lifetime and
high reliability results from use of the proven low complexity flexure compressor and pulse tube
cold head. Its low EMI is due to its self-shielding motor. The low self induced vibration results
from its internal dynamic balancing. It features the ease of integration into an instrument of a small
pulse tube cooler. The cooler achieves its 10 W at 95K cooling requirement with substantial margin
while rejecting heat to 300K.
INTRODUCTION
The High Efficiency Cryocooler achieves low input power and large cooling power because of
the efficiency of its pulse tube cold head and efficient compressor. Its low mass results from the use
of second-generation flexure compressor technology developed with Oxford University and
productionized by Hymatic Engineering. It achieves long lifetime and high reliability by using a
proven low complexity flexure compressor and pulse tube cold head. Its low EMI is due to its self-
shielding motor. It achieves low vibration through its internal dynamic balancing, and it features
the ease of integration of a small pulse tube cooler. The cooler is being developed in order to
provide a low mass, high efficiency cryocooler suitable for use on lightweight gimbaled optics on
surveillance missions such as SBIRS Low.
Surveillance systems incorporating LWIR focal planes require cooling of focal planes and
optics. This capability has long been sought as the solution to the midcourse missile flight detection
problem. During midcourse missile flight, the trajectory phase between burnout and re-entry, track-
ing and discrimination of ballistic missiles, reentry vehicles, and deployed penaids (decoys) are
difficult without the use of sensitive LWIR focal planes. In addition, space surveillance of resident
space objects for tracking and identification is needed. By collecting photons at the longer wave-

Kluwer Academic/Plenum Publishers. 2001 163
lengths emitted by these cold, dim objects, a vast improvement in identification and discrimination
capability with a minimum of sensor aperture growth can be realized. Multiple spectral bands can
be utilized to greatly improve the sensor’s ability to determine the temperature and cooldown rates,
which will aid in the discrimination of lethal objects versus decoys. Smaller apertures produce
cheaper, lighter more agile sensors, much easier to host in a space-based system.
To achieve the sensitivity at the long wavelength requires cooling of focal planes to tempera-
tures less than 40K. Optics cooling is required to minimize background noise and enhance the
detection sensitivity of these long wavelength, low temperature focal planes. Such missions cannot
contemplate use of this detection capability unless efficient, lightweight cryocoolers are available
for cooling of optics. Below the Horizon (BTH) imaging, requires the development of efficient
cryocoolers as mission-enabling technology. To achieve the rapid scanning of an agile sensor either
for missile tracking Above the Horizon (ATH) or for BTH tracking requires lightweight on-gimbal
components. The high heat loads due to BTH imaging, along with the requirement for rapid scan-
ning, drive telescope designs to use cooled on-gimbal foreoptics. For on-gimbal cooling the mis-
sion and careful thermal design determine the heat loads that the cooler must lift to be rejected by a
radiator an-gimbal. This radiator capacity is limited by the physical area that can be accommodated
on-gimbal, and is the mission limiting component. Given this radiator limit, all possible missions
can be accommodated only by improving the cooler efficiencies to the point that radiator heat
rejection is no longer the limit. If the cooler is inefficient, heat rejection of on gimbal radiators
becomes the factor limiting the system capability. For this reason, cryocooler efficiency is critically
important to the system design.
In the required 95 K temperature range the current state of the art in specific power (ratio of
input power to cooling power at 95 K) for flight qualified cryocoolers is in the range of 12 WAV.
Previous flight qualified coolers did not possess the capability to produce the required >10 watt
cooling capacity. The best previous specific mass (ratio of mechanical cooler mass to cooling power
at 95 K) was in the range of 1.5 kg/W. This project seeks to make a major improvement to the
specific power while at the same time reducing the cooler specific mass by 350% to <0.43 kg/W.
The specific power improvement goal will be achieved by developing an exceptionally efficient
pulse tube cold head whose performance is optimized at 95K. The specific mass improvement goal
will be achieved by using our next generation very low mass flexure bearing compressor technol-
ogy in the mechanical compressor. Engineering models of these compressors were demonstrated
on the IMAS project and to date have reached 3480 hours in life test with no detectable perfor-
mance change.
At the time of this writing the two protoflight compressors are complete and are reported in a
companion paper1 at this conference. Development has succeeded in producing exceptionally effi-
cient development cold heads with a Carnot efficiency of 25% at the 95K, 10W operating point
with 300K reject temperature. Flight cold heads are now in fabrication and will be integrated with
the compressors in June 2000. Delivery is scheduled for October 2000.
HIGH EFFICIENCY CRYOCOOLER
The High Efficiency Cryocooler (HEC) key requirements are given in Table 1. The input power
goal is quite ambitious requiring major strides in the cold head design. Despite the very low spe-
cific mass of this machine achieving the mass goal is low risk because of the existence of the
demonstrated compressor technology.
HIGH EFFICIENCY PULSE TUBE COOLER 165
Figure 1. High Efficiency Cryocooler. Figure 2. IMAS cooler.
The reliability and lifetime goal is representative of pulse tube cryoeoolers of this class which
has been partially verified by life testing and in-orbit operation of similar earlier-generation ma-
chines.
The HEC conceptual mechanical design is shown in Figure 1. The cooler is an integral con-
figuration pulse tube cryocooler. It incorporates a back to back flexure bearing compressor for
vibration balance and the passive in-line pulse tube cold head. The pulse tube cold head is a passive
Stirling cycle cooler in which the moving Stirling displacer is replaced with a passive expander.
This greatly increases the reliability and producibility of the cryocooler by eliminating cold moving
parts with close tolerances. Otherwise, it follows the same thermodynamic principles and uses the
same Oxford-style compressor. Because it is a much newer technology than Stirling cryoeoolers,
movement up the learning curve in recent years has been very rapid. It now has the same tempera-
ture production capability and efficiency as Stirling coolers. Ten-year lifetime is achieved by elimi-
nating all wear mechanisms. In the pulse tube cold head this is automatic since it is an all-metal
plumbing system with no moving parts. As a result, it is simple to build and has no significant
disadvantages. In the Oxford style compressors, long life is achieved via the flexures which elimi-
nate piston wear.
The flexure springs are very stiff in the direction perpendicular to the driven motion (much
stiffer than gas or magnetic bearings) so that close-tolerance gas-gap seals can be maintained and
wearing seals can be eliminated. The flexures themselves are designed for maximum stress levels
well below the material endurance limits. Non-fatiguing performance is readily validated in any
machine since over 10 cycles are accumulated in 4 to 5 days with these compressors. The working
fluid is inert dry helium with no lubricants. The drive is a direct voice coil motor similar to a
loudspeaker driver, thereby eliminating linkages. Oven baking prior to closure reduces volatile
condensables and water in the machines to negligible levels. All of these types of coolers are her-
metically metal sealed to have effectively zero detectable leakage rates of helium fluid. The pro-
cesses have been verified by life tests of similar pulse tube coolers including TRW units currently
in orbit.
The HEC cryocooler is derived from its predecessor IMAS engineering model coolers (Fig. 2)
developed for and delivered to the Jet Propulsion Laboratory.2
Figure 3. HEC Compressor.
The HEC compressor shown in Fig. 3, incorporates simple and effective mechanical and ther-
mal interfaces with the payload. The single mechanical mounting interface at the compressor
centerplate also serves as a conductive interface for removing all the heat from the cooler. To
enhance the efficiency and minimize its system impact on the payload to which it is connected, the
centerplate is designed to provide heat spreading, allowing the cooler to be mounted directly to a
radiator. A secondary alternative warm mechanical/thermal interface is provided at the centerplate
where the cold head attaches to the compressor. This provides the flexibility of changing cooler
orientation as well as the option of a vacuum interface for ground testing. The larger diameter end
cap incorporates the pulse tube cold head reservoir tank.
The 10W at 95K requirement has been met in development testing with an engineering model
compressor and development cold heads. At its present development stage the cold head is already
more efficient at this temperature than other flight units. Figure 4 gives the leadline through the
10W at 95K operating point for the most efficient of the development cold heads.
Figure 5 gives the efficiency of this cold head as a function of temperature while rejecting to
300K. The low radiated magnetic field (RE01) of the IMAS mechanical cooler at 75 watts input
power shown in Figure 6 was measured at JPL. This excellent performance at the cooler operating
frequency and harmonics results from the patented self-shielding voice coil motors. This basic
design has been carried over to the HEC and therefore identical performance should be obtained at
the same power level.
A second major benefit of this compressor design is the very low self-induced vibration over a
wide frequency range that is achievable with the cooler. This results from the very rigid motor
design and the degree of balance of the two compressor halves. Other work is ongoing to better
match the compressor halves to further reduce the need for active vibration cancellation for many
applications. The output force of the IMAS cooler shown in Figure 7 was measured on a Kistler
Figure 4. Cooler load line through 10W at 95K. Figure 5. Cold Head Efficiency.
HIGH EFFICIENCY PULSE TUBE COOLER 167
Figure 6. Radiated Magnetic Field (RE01) for the IMAS Cooler.
Figure 7. Self-Induced vibration on drive axis with 80 W input

power to cooler and tailored waveform.
dynamometer using a fixed predetermined tailored drive waveform to minimize the harmonics.
The force shown for 80 watts of input power is along the axis of motion of the compressor pistons.
In orbit the tailored waveform can typically be determined autonomously and updated in real time,
if necessary, by the existing flight electronics using feedback from an accelerometer mounted on
the cooler.
ACKNOWLEDGMENT
This project was supported by the Air Force Research Laboratory under contract F29601-98-
C-0179.
REFERENCES
1. High Performance Flight Cryocooler Compressor, P.B. Bailey, M.W. Dadd,, N. Hill, C.F. Cheuk, J.
Raab and E. Tward, Cryocoolers 11, Plenum Publishing Corp., New York (2001).
2. Chan, C.K., Ross, R.G., Jr., et al., "IMAS Pulse Tube Cooler Development and Testing," Cryocoolers
10, Plenum Publishing Corp., New York (1999), pp. 139-147.
High Performance Flight Cryocooler
Compressor
P.B. Bailey and M.W. Dadd

Oxford University, Oxford, UK
N. Hill and C. F. Cheuk

The Hymatic Engineering Company, Ltd.
Redditch, UK
J. Raab and E. Tward
TRW, Redondo Beach, CA, USA
ABSTRACT
In this paper we report on the development of a next generation flexure bearing compressor
which features high efficiency, high capacity per unit mass, enhanced producibility and ease of
integration into payloads. The compressor was developed for the 95K High Efficiency
Cryocooler programme.
The compressor achieves low mass by using small diameter flexure springs and having a
new compact design of magnetic circuit which also has the advantage of being self shielding,
thus reducing the radiated magnetic field.
A pair of compressors mounted back to back and driven in anti-phase provides low levels of
self-induced vibration, which is further improved by the rigidity of the motor and the charac-
teristics of the new motor and spring designs.
Its ease of integration results from its compact size and the incorporation of a single thermal
and mechanical mounting interface in its centreplate. The centreplate incorporates heat spread-
ing both internally for removing compressor heat as well as for spreading the heat to the radiator
to which it can be attached.
Producibility has been achieved by transferring the processes developed for manufacturing a
similar Oxford designed long life tactical cryocooler.
The compressors are being manufactured by Hymatic to a design which has evolved from
earlier machines made by Oxford University. TRW will integrate the compressors into the flight
qualified 95K High Efficiency Cryocooler which will be delivered to AFRL in October 2000.
INTRODUCTION
A new type of compact linear motor and a new flexure spring design have been developed
by Oxford University for linear compressors with the aim of meeting stringent requirements for
high efficiency and low mass. The compressors are a key part of the High Efficiency Cryocooler
(HEC) which is being developed by TRW.
Cryocoolers 11, edited by R.G. Ross. Jr.
Prior to this programme 3 single compressors (all with active balancers) and 4 balanced pair
compressors were made and assembled at Oxford for TRW, together with a fifth balanced
machine assembled by TRW. Two of the balanced pair compressors were delivered by TRW to
NASA/JPL for the New Millenium IMAS project1. These compressors, which have an identical
motor design to the High Efficiency Compressor discussed here, have been subject to extensive
thermal, vibration and EMI testing.
The original design has now been taken a step further in a three way collaboration between
Oxford, Hymatic and TRW, with the aim of making the compressor more rugged, and also
introducing a fully controlled assembly and test process more suitable for repeated and consistent
quantity production. The assembly and test processes have drawn on the manufacturing and
process technology from the Hymatic tactical Stirling cooler2. In the first half of 2000 two HEC
compressors have been built at Hymatic, using the new assembly and test processes and have
been delivered to TRW.
DESIGN PHILOSOPHY
The key feature of the Oxford design philosophy is the holistic approach to design taken -
too many cryocoolers have been designed that are almost impossible to assemble. From the
outset, details are incorporated into the design to aid the assembly and testing of the machine.
Another important feature is the elimination of many delicate and precise components, which are
replaced by simpler parts, combined with extensive use of jigs and fixtures. This approach lends
itself to larger production quantities, rather than the 'one-off' approach to earlier builds.
COMPRESSOR DESIGN
The compressor is based on the well-proven 'Oxford' principles of spiral flexure springs and
non-contacting clearance seals. The machine is a compact moving piston design, with the piston
and cylinder located within the core of the magnetic circuit of a moving coil motor.
The springs are the only component subjected to significant fatigue loading, and these are
routinely batch tested at a minimum 25% overstroke to in excess of 108 cycles. The springs have
been qualification tested, and the results from this predict a single spring arm reliability of
0.999998 and a reliability for the 96 spring arms used in each compressor of 0.9998.
The linear motor powering the compressor is a new moving-coil design that features a very
compact magnet circuit with low flux leakage and consequent high motor efficiency for the size
and power. The coil is fully supported on a former, and special attention was paid to maximize
the fill factor and increase motor efficiency. The structural integrity of the coil former facilitates
transmission of driving forces without relying on the variable strength of the coil potting
adhesive, thereby eliminating a common source of compressor failure.
The motor design is self-shielding and features extremely low levels of radiated magnetic
field. Test on the similar IMAS compressor showed that the compressor essentially met the
requirements of the MIL-STD-461C RE01 test specification measured at 7cm distance3.
The design of the compressor is such that it is inherently well balanced with the two
'compressor halves' mounted in line and operated in anti-phase. Tests on the MAS cooler
indicated very low levels of self-induced vibration with 30 W of sine wave input power. The
only harmonic above 40mN rms was the second harmonic, and this .probably arises from a
mismatch in the 'mechanical zero' position between the two compressor halves3.
The two identical compressor halves are mounted on an aluminium alloy 'Centre Plate' that
contains all of the cryocooler interfaces. Electrical power is supplied by means of a hermetic
feedthrough which is Electron Beam welded to the centreplate. One face of the centre plate
provides 45cm2 of thermal interface for heat rejection. A large flange is provided around the
connection to the pulse tube for a vacuum tank to be fitted during testing. Gas containment is by
means of metal 'O' rings between the end caps and centre plate and an aluminium gasket to seal
the fill port. A leakage rate ofbetter than 10-7 mbar litre/sec has been achieved consistently.
HIGH PERFORMANCE FLIGHT CRYOCOOLER COMPRESSOR 171
Figure 1. Completed compressor.
Particular attention is given to the locking of fasteners and component stacks to preserve the
alignment against vibration and redundant locking devices are used to enhance reliability.
The compressor has been designed to operate with a nominal 100 Watts of input power with
an additional 50% input power margin. The compressor has an overall length of 226 mm, with
end caps 57 nun in diameter and a mass of 2.45 kg.
PRODUCABILITY
Many of the assembly processes involve bonding, and these processes are usually
irreversible. Hence the key to quantity production of cryocoolers is the verification (where
possible) of each and every stage of the production process, from component manufacture to
final assembly.
Material Selection
Materials used in the assembly such as plastics, adhesives and primers are selected from an
existing knowledge base of materials with a low out-gassing rate. New materials are extensively
tested for their margin of being rendered clean by vacuum bake-out processes. A quadrupole
Mass Spectrometer is used extensively for this purpose. Only traceable materials are used in the
manufacture of components.
Component Stage
Geometric tolerances commensurate with the functional requirements of the components are
specified. During component manufacture, the geometric tolerances form the basis of the
method of work holding, while the surface finish requirement defines the manufacturing process.
The success of the component manufacturing processes stems from the realisation of the
differences between metal and plastic components when they are machined to tight tolerances.
Non-contact measuring methods including laser, optical and air gauging are used extensively to
verify the success of the manufacturing processes. Plans for transit and storage protection for
each component are designed into the manufacturing processes from the raw materials to the
finished products.
Verification and Development of Assembly Processes and Tooling
Assembly processes are verified by testing as defined in process development plans.
Sufficient quantities of test pieces are manufactured to render these tests statistically significant.
The objective of these tests is to determine the capability of the processes. Data derived from
these tests is then used as the standard to which subsequent assembly operations are controlled,
rather than the less stringent functional requirement of the assembly itself. If a process is
capable of performing to a standard, it is considered to be under control only if it consistently
attains the same standard within the natural variation of the process.
The design of the Compressor relies heavily on the use of tooling. The dimensions and
geometric tolerances of tooling is inspected and verified before being released for production.
Component Inspection
For the initial builds, 100% inspection of all dimensions is being implemented, but as the
component manufacturing processes become fully defined and robust, this will be gradually
replaced by 100% inspection of 'critical dimensions only' in future builds.
'Goods Inward' inspection of components in itself is not sufficient. It is a truism that
components are at their best immediately after manufacture - from then on it is downhill - every
operation, from finishing (deburring/frazing), inspection, cleaning and transport has the potential
to damage components. For this reason it is vital with critical components to have a functional
inspection of the part immediately prior to assembly. A detailed inspection plan has been
compiled specifying the functional inspection requirement of critical parts and the method of
inspection, which mimics the assembled conditions of the components. During the initial builds,
this inspection procedure will detect errors on parts indicating that the processes need refining.
This gateway enables improvements on the quality of components to be made before they are
assembled beyond the 'point of no return'.
In-Process Testing
Where practical every stage of the assembly process is verified by some form of in-process
testing, both to test the validity of the actual process itself and also to ensure that the process has
not had any secondary deleterious effects on the assembly.
Experience has shown that one of the main problem areas is the clearance seal – it is
difficult to achieve the correct clearance and easy to lose it. Hence many of the most important
tests are those which check the alignment of the assembly and show that there is no friction
between piston and cylinder. Placing the friction and alignment tests strategically in the
assembly process, the consistency of the free frictionless movement of the finished compressor
can be assured.
Test Facility. Many of the tests are carried out on a computer-controlled test rig which also
functions as a data logger. Using digital-to-analogue converters the computer controls both DC
and AC amplifiers for powering the compressor, together with signal conditioning and data
logging functions. Among the functions available is the facility to continuously monitor the
drive coil temperature and to shut down any test should this temperature become unacceptably
high. This facility is essential in some of the DC tests, which are slow and could easily lead to a
coil burn-out if the tests were carried out manually.
Alignment Test. This test verifies the capability of the spring suspension system to effect a
linear motion of the piston within the cylinder such that the clearance between them is main-
tained. The test is carried out immediately after the spring stacks are assembled and aligned. The
assembly is then locked to prevent movement, and the test is repeated before the piston is fitted.
HIGH PERFORMANCE FLIGHT CRYOCOOLER COMPRESSOR 173
Figure 2. Alignment Test – typical result.
To perform the test, power is applied to the motor, which is taken through three complete
stroke cycles and the run-out error measured. A least-squares polynomial is calculated through
the data points; figure 2 shows a 'screen dump' of a typical test result. Note that the sensor
used exhibits some backlash, but this is annulled when the polynomial is calculated (visible in
the centre of the trace). The repeatability of the measurements is excellent – apart from a
certain amount of 'bedding-in' during the first cycle, the three curves follow each other within
0.1 µm. The repeatability shown here is an excellent demonstration of the flexure spring
suspension system.
The results of this particular test must be treated with a certain amount of caution, as the
test is recording not only the linearity of the motion of the cylinder, but also measuring the
straightness of the cylinder itself. Thus to make any sense of the test, the 'cylindricity' of the
cylinder must be good, and is typically less than l.5µm. Much of the 'noise' apparent on the
trace is repeated on successive test cycles and is due to the surface finish of the cylinder.
' From such readings a complete picture of the combined linearity and cylindricity can be
built up, and this is then displayed as a three-dimensional plot (figure 3). From this plot the
effect of form error and alignment error can easily be separated and the true alignment error
evaluated.
Friction Tests. Tests are used to evaluate the friction between piston and cylinder - one
for dynamic friction, and one for static friction. The tests are computer-controlled and are
performed at several stages throughout the build, both before and after the piston is fitted.
The dynamic test (low frequency sweep) involves driving the compressor through one
complete cycle using a 0.01 Hz triangle wave. A curve of current against displacement is
plotted and studied for discontinuities and to observe the size of the hysteresis loop.
Process Capabilities. Using the in-process tests outlined above the capability of the
alignment and piston assembly processes can be quantified.
The alignment process is expected to produce a mean error of 2.6µm with a standard
deviation of 1.67. 99% of the spring suspensions systems that have been aligned by these
processes are expected to have an alignment error of less than 6.5µm. The smallest diametrical
clearance between the piston and cylinder should then be larger than 13µm for a true frictionless
clearance seal design to be realistic.
Figure 3. Alignment test - typical result in 3D plot.
CONCLUSION
A compact, high power and producible compressor design has been achieved. A build
procedure has been formulated and line qualified to ensure the consistency of the quality of the
compressor. Alignment accuracy of less than 6.5µm (peak-to peak) has been achieved with no
evidence of friction in the clearance seal. Innovative in-process test procedures have been
designed and are expected to benefit the future manufacture of cryocooler compressors.
ACKNOWLEDGEMENT
We acknowledge the strong support of Thom Davis of AFRL for this project.
REFERENCES
1. Chan, C.K., Nguyen, T., Colbert, R., Raab, J., Ross, R.G. Jr., Johnson, D.L., "IMAS Pulse Tube
Cooler Development and Testing", Cryocoolers 10, Plenum Press, New York (1999), pp 139-147.
2. Aubon, C.R., Peters, N. R., "Miniature Long Life Tactical Stirling Cryocoolers", Cryocoolers 9,
Plenum Press, New York (1997), pp 109-118.
3. Tward, E., Davis, T., "High Efficiency Cryocooler", Proc. AIAA Paper No. 99-4564,1999.
Vibration Reduction In Balanced Linear
Compressors
M.W.Dadd, P.B.Bailey and G.Davey

Oxford University, Oxford, UK
T.Davis and B.J. Thomlinson
Air Force Research Laboratory, Albuquerque, NM, USA
ABSTRACT
Coolers for Space applications are often powered by reciprocating compressors that use a
linear compressor technology. These can deliver the requirements for long life and high
reliability but have not yet produced acceptable uncompensated vibration levels at a reasonable
cost. If two nominally identical compressors are mounted back to back the vibration level is
reduced, but may still be too high for many applications.
Further reduction of vibration is achieved through the use of Adaptive Control systems,
which are expensive and reduce the reliability of the system. If the residual vibration can be
reduced by better matching of the two compressors, then cheaper, more reliable electronics can
be used to achieve the desired vibration level.
Under a research and development effort with the Air Force Research Laboratory, all
sources of vibration were considered but effort was concentrated on improving the matching of a
compressor pair to limit the main causes of vibration. The dynamics of a compressor were
modelled. The force generated by the coil was calculated from flux densities determined by a
finite element analysis of the magnetic circuit. The rest of the system was modelled as a damped
harmonic oscillator.
An attempt was made to reduce the residual vibration in an opposed pair of compressors by
duplicating the model to simulate a compressor pair and investigating the effect of small
variations of a number of parameters. These results were used to estimate the accuracy with
which various parameters must be matched to achieve a certain residual vibration and this
information was then used to improve the assembly of a compressor pair.
Most of the components for this compressor were already available so it was not possible to
make major changes. Detailed measurements were made on all components and assemblies so
that the vibration spectrum could be related to a compressor with documented manufacturing and
building standards.
INTRODUCTION
The Oxford University cryogenics group has been involved in the development of space
cryocoolers for many years and has produced a number of prototype coolers/compressors. A
prototype balanced compressor pair (designated the Capital Compressor) that had been designed

and manufactured at Oxford for a recent TRW space project, was found to have an inherently
low vibration level The compressor halves had been assembled carefully to achieve true
clearance seals but there had been no conscious effort to produce a low vibration machine.
The very low vibration levels required by space programmes are currently achieved by
complex drive electronics that use adaptive control algorithms. Whilst effective, this approach
has significant drawbacks:
• The complexity of the electronics hinders the achievement of high reliability
• The additional electronics increases cost and payload.
• A basic adaptive control approach is only able to improve on-axis vibration levels. For the
reduction of off-axis levels, additional force transducers and control electronics would be
required with a further impact on reliability etc.
An approach that suggested itself is the further reduction of the uncompensated vibration by
improved matching of the two compressors through better matching of components and closer
control of build processes. To achieve this goal it was recognised that a better understanding of
the forces operating would be desirable and that methods should be found to allow build quality
and component matching to be evaluated and documented. Some work has already been done on
the development of a compressor model using a dynamic simulation package called Vissim. This
was used to model an existing unbalanced compressor and was found to give good agreement
with measured operating parameters that included out of balance forces. This work was reported
in ref(l).
The project that evolved had the aim of
• Looking more closely at the forces acting in the compressors to see whether the existing
model is adequate.
• Looking at what dimensional and alignment limits are required for good balance, and what
the practical limitations might be.
• Documenting the assembly of a balanced compressor pair. It was hoped that some
improvement might be made over previous assemblies but this might not be possible given
the restriction of using largely existing components.
• Comparing the measured vibration of the assembled compressor with the values generated
by the model
At the time of writing, the compressor build had not been completed so this paper will
concentrate on describing the approach to modelling and build evaluation.
FORCES ACTING IN A LINEAR COMPRESSOR

The types of linear compressors used in space cryocoolers have the virtue that the systems
offerees operating in them are amenable to relatively simple descriptions. Figure 1. shows the
main components of a typical linear compressor. These are:
• A piston/cylinder assembly utilising a no-contact clearance seal
• A linear motor – in this case a moving coil “loudspeaker” type motor
• A suspension system comprising of two aligned sets of flexure bearings – these are usually
springs with a flat spiral geometry and are known as spiral springs or flexures.
The characteristics of the suspension system are extremely important to the operation of the
compressor. The geometry and alignment of the springs accurately define an axis along which
the spring stiffness is low and a large movement (e.g. > 10 mm) is possible without exceeding
the fatigue limit of the spring material. Perpendicular to this axis the stiffness is extremely high.
The obvious axial symmetry of this type of compressor and the distinct characteristics of the
suspension spring assembly suggest the division offorces acting into three types:
• Forces acting along the compressor axis where the spring stiffness is low– these will be
termed “On-axis” forces.
• Forces acting perpendicular to the compressor axis where the spring stiffness is very high –
these will be termed “Off-axis forces.
VIBRATION REDUCTION IN BALANCED LINEAR COMPRESSORS 177
Figure 1. Schematic diagram of typical compressor.
• Forces associated with the rotation of the spiral suspension springs. The operation of these
springs generally gives rise to a small rotation of the moving assembly. The forces exerted
on the moving assembly, are also off-axis forces but they resolve to a zero net force and a
well-defined moment acting about the compressor axis.
ON-AXIS FORCES
The On-axis forces are the dominant forces acting in the compressor. The principal forces
with large magnitudes are those intrinsic to the compressors operation:
• The force generated by the coil
• The force produced by the suspension springs as they are deflected.
• The force produced by gas pressure differences acting on the piston area.
These are balanced by the inertia force acting through the centre of gravity as the moving
assembly accelerates. Ideally the axis defined by the movement of centre of gravity and axes
along which these forces act should coincide. The result would then be a single force acting
along a single well defined axis and there would be no resulting moments. In practice there will
be offsets and angular misalignments between the axes and these will result in off axis forces and
moments perpendicular to the compressor axis. The radial stiffness and separation of the two sets
of suspension springs determine the actual movements resulting from such misalignments. If the
magnitudes of the axial forces are known then the level of misalignment that can be tolerated is
readily calculated.
There are other effects that will give rise to axial forces such as:
• Windage
• Eddy current effects due to conductive components moving in the motors magnetic field
• Shear forces acting on the piston due to gas flow in the clearance seals
• Forces generated by sections of the coil leads moving in areas of stray flux.
These are much smaller in magnitude and would only have any significance if:
• they were so uncontrollable that they resulted in big differences between the two compressor
assemblies.
• They were large enough and acted far enough off the compressor axis that they produced
sufficient moments to cause radial deflection of the springs.
It is clear that windage and eddy currents could only have any net effect if there was fairly
gross mismatching of the compressors.
The forces generated by coil leads will be very closely matched but will not be acting along
the compressor axis. The shear forces acting on the piston are sensitive to the dimensions and
geometry of the clearance seals, which may not be closely matched.
Frictional forces should not be present because in principal there are no contacting surfaces.
The detection of vibration caused by friction would indicate a fault condition. The forces
generated by friction will be very variable depending on small details of surface form etc so that
it is not possible to be too specific about its form.
OFF-AXIS FORCES
Off axis forces can only register a net effect because of some deviation from symmetry
either required in the design or as a consequence of imperfections in the compressor build.
Because of the high radial stiffness only large forces will produce significant displacements.
There is also the possibility that smaller forces may have enough leverage to cause a significant
angular deflection. Some possible forces that were considered are:
• Unbalanced gas forces acting on the sides of the piston
• Forces generated by current carrying leads interacting with the motor flux
• Forces generated by deviations from linear movement of the suspension system
The forces acting on the sides of the piston are large and are determined by the detailed
geometry of the clearance seal. Defects in the clearance seal geometry can lead to significant
imbalance and misalignment of these forces. The radial gap in a clearance seal is typically
around 10 microns for reasonable seal efficiency. With machining tolerances typically around +/-
1 micron and a similar tolerance on alignment it is clear that this effect merits careful
consideration.
The forces generated by the current leads are an inevitable part of a practical design. Their
magnitudes are readily estimated, and are likely to be insignificant.
Off–axis Forces generated by inaccuracies in the linear movement of the suspension system
need to be considered but it is likely that the linearity required for the clearance seal will
automatically ensure that they are not significant.
MOMENTS ABOUT COMPRESSOR AXIS

The magnitude of the rotation produced by the springs will be determined by certain aspects
of the spring geometry, principally the spiral arm length and curvature. The manufacturing
processes used allow these parameters to be controlled to tight limits, typically better than +/-
0.1%. It is clear that provided that the axes of the compressor halves are well aligned and that
the compressor strokes are well matched, any residual moment is likely to be insignificant.
RESONANCES
In addition to the forces already described there remains the possibility that mechanical
assemblies have resonant modes that may be excited at particular frequencies. Small out of
balance forces, whether on-axis or off-axis, could be amplified to the point that they become a
problem. The general approach to this problem is to keep the resonant frequencies of
components as high as possible by designing the moving assembly for high stiffness. The use of
materials that have some intrinsic damping can also be considered although the opportunities to
do this are limited by the need for other mechanical properties. The moving assembly/suspension
spring system does present a particular instance where resonances may be a problem. The stroke
required of the springs limits their stiffness to relatively low values.
There are a number of vibration modes that should be considered:

• Resonance of the spring arms – the inner and outer ends being nodes and the centre being an
anti-node.
• Torsional resonance of the moving assembly about the compressor axis
• Torsional resonance of the moving assembly perpendicular to the compressor axis
• Radial resonance of the moving assembly with respect to the compressor axis.
These modes of resonance were investigated in the Capital compressor being built and the
resonant frequencies were determined to be at least 400 Hz.
A STRATEGY FOR GOOD BALANCE

The aim for a well-balanced compressor is that each compressor half should only produce
an On-axis force and that these forces should equal so that when they are aligned back to back
they cancel out To approach this goal the design has to be effective, the compressor components
need to be well matched and the build quality has to be adequately controlled.
The overall strategy that has been adopted is:
• Close matching of the amplitude of On-axis forces.
• Alignment of On-axis forces to avoid generating unnecessary couples.
• Minimising of Off-axis forces
• Design of suspension system with high radial stiffness to minimise deflections.
• Mechanical design that avoids assembly resonances in the operating frequency range
• Design which includes damping (where possible) to minimise amplitudes of resonances
IMPROVING THE BALANCE OF THE CAPITAL COMPRESSOR

The above has described in general terms the effect of different forces and their possible
sources without detailed reference to their magnitudes and real significance. For the Capital
compressor the Vissim based model described below was used to calculate the principal On-axis
forces for typical operating conditions. The effects of mismatching certain parameters were
investigated. Also measurements were made of the torsional and radial stiffness of the
compressor’s suspension system. This information was used as the basis for deciding which of
the effects described were likely to have any real impact on the overall balance.
For the Capital compressor, areas open to some improvement were identified as:
• Better matching of principal On axis forces – reducing differences in axial offsets of coils,
better matching of moving masses, motor parameters i.e. no of turns in coil, field in air gap
• Reducing radial offsets between principal On axis forces
• Improving geometry of clearance seals by closer control of component manufacture and
alignment
• Improving angular alignment of compressor halves with respect to each other
ON-AXIS FORCES AND SPRING STIFFNESS FOR CAPITAL COMPRESSOR

Maximum values of principal On-axis forces 100 N
Minimum radial stiffness of suspension system 760,000 N/m
Torsional stiffness perpendicular to compressor axis (about C.O.G.) 780 Km/radian
If a maximum of 1 micron radial movement of any of point on the piston is set as a criteria
then maximum allowable values can be set for Off-axis forces and moments perpendicular to
compressor axis. With the peak value of the principal forces, maximum angular misalignments
and offsets can also be defined:
Maximum Off-axis force 0.76 N
Maximum moment perpendicular to compressor axis 0.023 N.m
Maximum angular misalignment for On axis forces 0.0076 radians
Maximum offset for On axis forces. 0.23 mm
MODEL USED TO INVESTIGATE PRINCIPAL ON-AXIS FORCES

The model used to evaluate the effects of build mismatches is briefly described below.
The differential equation defining the motion for the On-axis forces described above is:
x is the displacement of the piston assembly from its rest position

m is the moving mass, is the spring rate of the suspension springs
is the pressure difference acting across the piston over an area A.
where P is cycle pressure, is pressure in compressor body
F(x,t) is the force generated by the driving coil.
The variation in is primarily generated by the cycle pressure. This is determined by the
thermodynamic and pressure drop processes occurring in the refrigeration cycle and cannot be
simply described. An approximate model, that is simple and useful, can be developed by treating
the gas force as a spring/ damper combination The resulting differential equation is that for a
damped harmonic oscillator:
c is effective damping constant for the gas, k is total spring rate, is the
effective spring rate of the gas.
The force generated by the coil is given by
is the conductor length per axial length of coil, i(t) is the current through the coil
The differential equation defining the behaviour of the moving coil motor as an electrical
system is:
V(t) is the applied voltage, E(t) is the back emf generated by the coil, R is the coil
resistance, L is the coil inductance.
E(t) is given by:
The values of integral that are required for Eqs. (2) and (4) are accessed in
Vissim as a “Look up” table.
The look up table values were calculated in a spread sheet using flux distribution defined
by:
These equations were derived empirically using values generated by a finite element
analysis of the magnetic circuit.
SOME RESULTS OBTAINED FROM THE MODEL

The work described in Dadd et al.1 showed that model values and values measured on a
particular unbalanced compressor were close for all the main parameters i.e. instantaneous
values of forces, currents and voltage inputs. The only value that had to be adjusted to obtain
good agreement was the coil inductance. A similar comparison will be made with the Capital
compressor when its build has been completed.
The model was used to investigate how the mismatch of particular build parameters would
Figure 2. Variation of net force with axial offset of coil on one side.
Figure 3. Vibration Spectrum for coil offset of 0.2 mm.
effect the balance such parameters could be moving mass, inductance, magnetic flux distribution
etc. As an example, one finding is the dependence on the mean position of the coil in the
magnetic circuit. This can vary because of the build up of tolerances in the compressor build.
Figure 2. shows how the residual force varies with the changes in offset for one of the
compressors. Figure 3. shows the frequency spectrum for the case where one coil is centred and
the other is offset by 0.2mm. It will be seen that the residual force is becoming quite significant
i.e. 1% of value for each half and that it is dominated by the second harmonic.
MEASUREMENTS AND DOCUMENTATION OF COMPRESSOR BUILD

An important and time-consuming part of the work described in this paper was the
measurement and documentation of various parameters during the build. These measurements
included:
• Component masses
• Coil characteristics – resistance, No of turns and Inductance
• Magnetic flux distribution in gap
• Geometrical and dimensional features of individual components e.g. concentricity
• The geometrical features e.g. alignments of the compressor assemblies as they are built up.
The purpose of these measurements was twofold. Firstly to establish a build specification
on which to base any modelling or interpretation of measurements. Secondly to actually identify
areas where mismatches were significant or where components were not manufactured to
adequate standards.
Many of these measurements e.g. mass, coil resistance etc required only ordinary laboratory
equipment and will not be discussed further. However the measurement of geometrical features
and alignment required specialised metrology equipment. Many of these measurements also
needed to be taken during the compressor build when cleanliness is very important.
To enable inspection measurements to be made in a clean environment a TalyRond 100 was
installed in the clean room. The essential features of this type of instrument are:
• An accurate turntable incorporating centering and levelling adjustments
• Vertical and horizontal columns allowing accurate positioning of probes
• A number of sensitive displacement probes capable of determining surface forms e.g.
roundness, concentricity, surface roughness etc.
Whilst taking these alignment measurements it became clear that the process could be
helped by having specific measurement features e.g. easily accessible reference surfaces. This
was not possible with the capital compressor but may be considered in future designs.
CONCLUSIONS
Although it is not possible to comment with the advantage of actual vibration
measurements, some conclusions can be made:
• The existing model appears to be adequate for describing the principal On-axis forces
• The magnetic properties of the magnets and magnetic circuit components are variable. Good
matching of these components is helped if they are chosen from within a single batch.
• The matching of the on axis forces requires matching of coil offset as well as other
parameters
• The offsets between the axes of the principal forces can cause significant moments if their
alignment is not adequately controlled
• The clearance seal geometry needs to be closely controlled to avoid significant unbalanced
forces acting perpendicular to the compressor's axis.
• Future compressor designs could usefully incorporate specific features to facilitate
measurements on build alignment
In many ways good balance is synonymous with high build quality – i.e. both are concerned
with good alignments, absence of friction and the minimising of resonance. Low vibration
compressors require a high build standard: a high build standard may be demonstrated by low
vibration measurements. Assessing build quality is an important issue because of its effect on
reliability. The pursuit of low vibration may therefore have added benefits.
ACKNOWLEDGEMENTS
The Capital compressor was designed and built under a contract with TRW, who also made
compressor components available for this work.
REFERENCES
1. Dadd, M. W., Davey, G., Lion Stoppato, P.F., Bailey, P.B., “Vibration Reduction in Balanced
Linear Compressors in the 17th International Cryogenic Engineering Conference”, ICEC 17,
Institute of Physics Publishing, Bristol (1998), pp.127-131.
95 K High Efficiency Cryocooler Program
Kenneth Price1 and Capt. Vladimir Urbancek2

1
El Segundo, CA, 90245, USA
2
Kirtland AFB, NM, 87117, USA
ABSTRACT
The Air Force / Raytheon 95K High Efficiency Cryocooler (95K HEC) Program is developing
a new two-stage hybrid Stirling-pulse tube space qualified refrigerator with high heat lift
capacity, high efficiency, low weight and size, and low production costs relative to the current
state-of-the-art. The basic program will deliver a protoflight Stirling-class Thermo Mechanical
Unit (TMU) with protoflight radiation hard electronics. The cooler is designed to support 10W
heat lift from a 95K source to a 300K sink. Motor power consumption is to be less than 100W
and system power (including electronics) is to be less than 137W. The cooler is to weigh no
more than 6Kg. The TMU cold head and compressor designs are highly versatile to enable low
cost tailoring to meet the needs of a wide variety of applications. The first demonstration of this
versatility is a program option to deliver a companion high-capacity 35K cryocooler. This cooler
will also have an aggressive efficiency requirement. The 95K and 35K TMU will share over 95%
of components, resulting in significant production efficiencies. Another result of this high degree
of commonality is that each cooler can be powered and controlled by standardized Command
and Control Electronics (CCE). The only adjustments needed to match the CCE to a TMU design
are in selected logic parameters stored in ROM and in minor changes to winding ratios in two
transformers. The CCE is designed with radiation hard components, but the initial protoflight
units will be delivered with lower cost commercial substitutes, where available with the same
form, fit, and function. The 25 month program will deliver a fully flight qualified 95K system
including both TMU and CCE and, if the option is exercised, a similar flight qualified 35K
system.
INTRODUCTION
The Space Based Infrared System (LEO), or SBIRS Low, is part of the our nation’s Ballistic
Missile Defense program and it’s proposed function is to detect and track ballistic missiles
during mid course flight, which is the trajectory phase between burnout and reentry.
Discrimination of ballistic missiles, reentry vehicles, and deployed decoys during this phase is
difficult, if not impossible, without the use of long wave infrared focal planes. Achieving the
sensitivity necessary to identify and discriminate among these cold dim objects requires cooling
of the focal plane to temperatures less that 40K. Optics cooling is also required to reduce

background thermal noise and improve detection sensitivity at these long wavelengths and low
temperatures.
An advanced study of a potential SBIRS Low architecture called for gimbaled fore optics at
100-110 Kelvin that required cryogenic cooling at 95K from the fore optics cooler. The study
also called for tracking sensor/shielding cooling in two stages at 35K/60K. It was quickly
determined that state-of-the-art cryocoolers and/or radiators would not be able to achieve the
desired temperatures and heat loads in a reasonably sized or efficient package. Thus, the
government embarked on the High Efficiency Cryocooler program.
The United States Air Force is sponsoring a program to design, develop, fabricate and deliver
space qualifiable cryocoolers to meet SBIRS Low Technology needs. These coolers will support
the SBIRS Low cooling performance objectives for the potential on-gimbal optics design at an
initial design point of 95K with 10 Watts of cooling. The contract includes an option to modify
the 95K design and then deliver a cooler that produces two-stage cooling at 35K and 60K for
tracking sensor cooling. An added benefit of exercising the option task is that the two versions
of the cooler (95K and 35K/60K) may be close enough to allow endurance testing on one unit
and qualification testing on the other thereby producing a fully space qualified design.
Raytheon’s High Efficiency Cooler program kicked off in August 1999. The contract is an
innovative cost and risk sharing agreement between the government and Raytheon. Under the
arrangement Raytheon has assumed the risk and cost associated with the design of the expander
module and electronics in exchange for full data rights to these designs. The Air Force, in the
spirit of acquisition reform, is funding the lower risk design of the compressor and the complete
system fabrication costs. This approach allows the Air Force to develop a new cooler for a third
to a half of the cost of going it alone. Though the Air Force has limited data rights, it will get
full benefit of the design because the contract assures that the cooler will be made available to
any DoD contractor at a reasonable cost. On the technical side, the program was selected
because it proposed developing a unique and innovative two-stage hybrid Stirling-pulse tube
design which promises to be easily modified for split heat loads and potentially used in tracking
sensor cooling at 35K and 60K. This is a true two-stage design, not a one-stage cooler with
regenerator heat intercept that has in the past been mistakenly called two-stage. The design also
allows angling of the second stage from 0 to 90 degrees relative to the first for unique ease of
integration of the cold head into confined cryogenic spaces.
The Raytheon 95K High Efficiency Cooler (95K HEC) program will deliver a complete
protoflight quality cryocooler including compressor, expander, and flight like electronics in
September 2001. The system will be fully flight qualifiable except that contract allows the use
of non-radiation hardened parts as direct substitutes, if available, in the cryocooler control
electronics when radiation hardened parts are cost prohibitive.
THERMO MECHANICAL UNIT DESCRIPTION

The 95K HEC system is comprised of a Thermo-Mechanical Unit (TMU) and Command and
Control Electronics (CCE). Key performance requirements are listed in Table 1. The TMU is a
novel two-stage hybrid Stirling-pulse tube cryocooler comprised of separate compressor and
expander modules connected by a transfer line. See Figure 1. The TMU retains significant legacy
to Raytheon’s previously developed “Oxford” class machines. Compressor swept volume is 6cc
produced by a pair of pistons working in opposition against a common compression volume.
95 K HIGH EFFICIENCY CRYOCOOLER PROGRAM 185
Figure 1. Protoflight 95K High Efficiency Cryocooler. The hybrid Stirling-pulse tube two-stage
expander is more efficient than single stage coolers and the unusual angled cold head offers
unique integration advantages.
The two-stage expander employs a Stirling first stage and a pulse tube second stage. The
pulse tube stage is configured in a “U-tube” for compactness and structural rigidity. Warm ends
of the pulse tube and regenerator tube and the pulse tube orifice and surge volume are thermally
anchored to the first stage.
The hybrid expander has several useful characteristics compared to two-stage pulse tubes and
two-stage Stirlings. Compared to a two-stage pulse tube expander with similar heat lift, the
working gas volumes of the hybrid expander are significantly smaller and require lower gas flow
rate. This reduces the compressor’s swept volume and increases its pressure ratio.
Compared to a two-stage Stirling expander with similar heat lift, the structural elements are
much easier to construct. For example, the tight-tolerance second stage clearance seal is
eliminated. Manufacturing and alignment are no more difficult than for ordinary single stage
Stirling expanders.
The hybrid expander also offers noteworthy versatility. For example, the expander can be
configured to angle the second stage from 0 to 90 degrees relative to the first. The unit shown in
Figure 1 has a 45-degree angle between stages, which enables a redundant pair of expander cold
heads to be compactly configured as shown in Figure 2.
The aluminum compressor housing is nominally 338mm long and 97mm in diameter. The
center section of the housing is a box-like structure that has a large area heat rejection surface on
one side and an easily accessed three-point mount on the other. See Figure 1. The heat rejection
area is 61.8 sq. cm (9.6 sq. inch.) This large area, combined with aluminum’s high thermal
conductivity, can efficiently transport over 150W of waste heat from the module. The three-point
mount on the opposite surface simplifies integration into most systems. Each of the two pistons
Figure 2. Two hybrid two-stage cold heads angled at 45 degrees can be compactly integrated into
a small sensor volume. Note that the first and second stage thermal interfaces are immediately
adjacent to each other a feature that can be used to advantage in shortening thermal straps.
has a Linear Variable Differential Transducer (LVDT) position sensor. The LVDT output is fed
into the electronics to prevent motion beyond a programmed “soft stop” stored in the electronics.
The expander shown in Figure 1 is nominally 309mm long and 94mm in diameter. The two-
stage cold cylinder extends 128mm in the axial direction beyond the expander’s warm-end waste
heat rejection surface. Thermal interfaces at both cold stages and at the warm end are large flat
surfaces. Combined side load capacity of the two stages is a minimum of 100N in any direction.
As in the compressor module, the housing is aluminum, position sensors are used to limit
motion, and waste heat is efficiently removed from 21.4 sq. cm (1.66 sq. inch) of heat transfer
interface area.
The protoflight compressor module is projected to weigh 4Kg and the protoflight expander,
2.5Kg for a TMU physical weight of 6.5Kg.
Each module is internally balanced for optimal control of residual vibrations. The
compressor employs two pistons working in opposition against a common compression chamber
located at the center of the module. The expander’s first stage piston is balanced by a matching
mass driven in opposition. The pulse tube second stage does not require dynamic balancing
because the moving gas mass generates insignificant force. Residual vibrations produced by each
module are corrected by the electronics via a novel algorithm executed by the electronics
described below. The algorithm is driven by feedback from piezoelectric load washers at each
module’s mounting points.
ELECTRONICS DESCRIPTION
The Command and Control Electronics (CCE) is a two-board system packaged in an
aluminum “slice” housing. See Figure 3. The two boards are designated as the Power Board and
the Logic Board. The Power Board includes:
1. A Low Voltage Power Supply (LVPS) with multiple DC voltage output forms
2. Spacecraft power bus voltage step-up conversion
3. Two pulse width modulated (PWM) switched compressor motor amplifiers
4. EMI filters at the board inputs and outputs
95K HIGH EFFICIENCY CRYOCOOLER PROGRAM 187
Figure 3. Two board Command and Control Electronics. The board on the left is contains the
digital elements and is called the Logic Board. The other board contains Low Voltage Power
Supplies and compressor PWM amplifiers and is called the Power Board.
The LVPS generates a set of low power DC voltage forms that energize circuits within the
CCE and the expander motor amplifiers.
The Logic Board includes:
1. An RS422 serial port using RS232 protocols to manage spacecraft telemetry
2. A parallel RS232 serial port to enable use of a PC user interface in ground test
3. A 10KHz drive circuit to excite the LVDT piston position sensors
4. Analog signal converter circuits (e.g., a multiplexer and an ADC)
5. A set of gate arrays containing all logic needed for communication and control
6. Various digital system elements such as a ROM chip and a RAM chip.
7. Two expander motor amplifiers mounted to the Logic Board for packaging
convenience
8. EMI filters on the board inputs and outputs where required
The CCE is a digitally based control system that eliminates most of the conventional analog
circuitry previously used in cooler electronics. This resulted in a significant reduction in parts
count that increased reliability and reduced the number of required radiation hard Integrated
Circuits. Total parts count is less than 550 including all resistors, capacitors, diodes, and
magnetics as well as integrated circuits and FETs. Reliability at ten years is calculated to be over
97% and radiation hardness at the component level is a minimum of 200Krad.
The CCE uses gate logic to implement key functions, including:
1. Telemetry uplinks (commands) and downlinks (data and status reports)
2. Cold tip temperature control on the second stage
3. Vibration control in both compressor and expander modules
4. Piston position control over the displacer and two compressor pistons
5. Launch lock enable/disable
Electronics also include two separate smaller modules: an Input Ripple Filter (ERF) and a
preamplifier module. The ERF attenuates reflected AC current drawn from the spacecraft power
bus. The preamplifier module resides near the cryocooler to amplify low level signals from the
cold tip temperature sensor and the compressor and expander vibration sensors before
transmission to the CCE through potentially lengthy cables. Signal amplification reduces noise
susceptibility of the signals.
Two significant accomplishments demonstrated by the electronics are high power throughput
efficiency to the compressor motors and a novel and effective vibration control algorithm. Power
throughput efficiency is defined as the ratio of the power delivered to the cryocooler motors
divided by power supplied to the electronics minus power to the LVPS. Throughput power
includes power consumed by the IRF, bus voltage stepup conversion, two compressor PWM
motor amplifiers, and EMI filters. This factor is closely correlated to power drawn by the
compressor module, which in turn scales with amount of refrigeration produced by the cooler.
Power throughput efficiency is over 90% when delivering at least 25W to the cooler motors.
Power to the LVPS is approximately constant under all conditions and is designed to be about
15W. Power drawn by the LVPS includes all power required to drive circuitry within the CCE.
Figure 4. Performance of the new vibration control algorithm. The time domain based algorithm
suceeds in reducing vibrations to the noise floor.
A new vibration control algorithm was developed to work within the limited capacity of
Field Programmable Gate Arrays (FPGA). This algorithm operates entirely within the time
domain, which eliminates computationally intensive Fourier Transforms used by adaptive feed
forward algorithms. The effectiveness of the new algorithm was tested by artificially inducing a
large imbalance in a compressor module to show the algorithm converges on a low vibration
solution. The imbalance was implemented by dropping out every fifth compressor PWM power
pulse in one piston assembly. This induced the large 5.8Nrms imbalance shown at 44 Hz drive
frequency in the vibration spectrum shown in Figure 4a. Vibration levels at the next several
harmonics were on the order of 0.3Nrms.
Despite the induced fault, the algorithm successfully brought the compressor into balance
and drove the force levels down to .030Nrms or lower over the 0-500 Hz band. See the spectrum
in Figure 4b. All disturbances produced by the compressor at its drive frequency and next ten
harmonics were lowered to the noise level present in the vibration feedback signal. The
broadband noise is due to background environmental vibrations plus inherent circuit noise.
Correction below the background level is is believed to be impossible.
Temperature control has also been demonstrated and integration of the temperature control
algorithm with vibration control is currently in progress.
PROGRESS SUMMARY
A breadboard TMU is in assembly as of June 2000 for test by July. Data will be used to
refine the calibration of the hybrid cryocooler math model enabling optimization of the design.
The breadboard is close to a flight qualifiable unit. Once the performance of the breadboard is
demonstrated the design can be readily upgraded to flight quality. A new protoflight TMU will
then be built and tested for performance and flight qualification.
Most elements of the CCE circuits and logic have been demonstrated. A second set of circuit
boards is currently in assembly incorporating refinements. The logic is being integrated as a
complete package and will be burned into a set of FPGAs. These chips will be mounted to the
new Logic Board and integrated with the protoflight TMU for system test.
ACKNOWLEDGEMENTS
The Air Force Space and Missile Command and BMDO sponsored this work. The Air Force
Research Laboratory, Albuquerque, NM, managed the project.
Design and Test of the NIST/Lockheed Martin
Miniature Pulse Tube Flight Cryocooler
P. E. Bradley, R. Radebaugh
National Institute of Standards and Technology
Boulder, Colorado, USA 80303
J. H. Xiao
Johnson & Johnson Co.
Somerville, New Jersey, USA
D. R. Ladner
Lockheed Martin Astronautics Co.
Denver, Colorado, USA
ABSTRACT
A two-stage miniature pulse tube (PT) cryocooler, designed for a Space Shuttle flight demon-
stration, was built and tested at Lockheed Martin Astronautics (LMA) at Denver, CO and the NIST
Boulder Laboratory. The Miniature PT Flight Cryocooler (MPTFC) was designed to provide 0.15
W of cooling at 80 K with heat rejection at 275 K. It was developed as the smallest cryocooler of
its kind for the purpose of demonstrating launch survivability and thermal performance in a zero-g
environment. A prototype laboratory version was first built and tested to provide information on
component sizing and flow rates for comparison to numerical models. The flight version was then
fabricated as a Getaway Special (GAS) Payload. Cost containment and manned flight safety con-
straints limited the extent of the MPTFC development to achieve performance optimization. None-
theless, it reached 87 K driven by a commercially available tactical compressor with a swept vol-
ume of 0.75 cc. The on-orbit cooling performance was not demonstrated because of low battery
voltage resulting from failed primary batteries. The first off-state PT thermal conductance mea-
surements were successful, however, and the MPTFC also demonstrated the robustness of PT cryo-
coolers by surviving pro-launch vibration testing, shipping, and the launch and landing of STS-90
with no measurable performance degradation.
The design and performance optimization approach for miniature two-stage PT coolers are
discussed. Some factors that may limit performance in small-scale PT coolers are identified also.
Laboratory pre-launch and post flight performance data of the MPTFC are presented, including
cooling performance as a function of heat load and rejection temperature. Off-state conductance
results are discussed in a related but separate presentation.

The Miniature Pulse Tube Flight Cryocooler (MPTFC) flew as a NASA shuttle payload
(GAS-197) in April 1998 aboard STS-90 (Shuttle Transportation System-90) as a technology dem-
onstration experiment. The primary objectives of this experiment were to demonstrate pulse tube
(PT) cryocooler performance and “off-state” thermal conductance in a micro-gravity environment
and to verify launch survivability of miniature coolers having limited vibration mitigation features.
This project was a collaboration between Lockheed Martin Astronautics (LMA) and the National
Institute of Standards and Technology (NIST). Because of the STS schedule and the safety issues
associated with manned space missions, the experiment was subject to a number of design con-
straints. The tight development schedule was based on limited flight opportunities preceding the
construction of the International Space Station (ISS) and on a low project budget. These two con-
straints dictated the extensive use of commercially available components, including a tactical com-
pressor and drive electronics (to obviate a long-life flight compressor development effort), an inex-
pensive electromagnetic latching valve, a commercial data acquisition system, and numerous com-
mercial electronics components. Attention to flight safety issues directly impacted the MPTFC
design in terms of operating pressure, sizing for limited battery-powered operation in a cold envi-
ronment, and limited design opportunity for performance optimization. The overall flight experi-
ment design also had to address various flight hazard issues, such as mechanical and electrical
integrity, EMI, redundant fusing, diode isolation, mitigation for high temperatures, etc. The ex-
periment design had to accommodate operation over a range of STS bay temperatures from -50 to
+40°C. In addition, the experiment timeline had to conform to limited STS crew operations.
The approach for completing the project on schedule was to design and test a prototype cryo-
cooler in parallel with the overall flight hardware system definition and parts procurement. Subse-
quently, the flight hardware and flight cryocooler development and testing were also accomplished
as parallel efforts. Lockheed Martin had primary responsibility for the flight and GSE hardware
and electronics, systems engineering, and for payload management, while NIST had primary re-
sponsibility for the cryocooler development, assembly, and testing. In practice each organization
contributed to all of these tasks.
DESIGN
Coldheads
The cryocooler coldhead design selected was a two-stage U-tube geometry orifice pulse tube
(PT) system based upon the double inlet concept first introduced by Zhu, Wu, and Chen.1 The
system is schematically represented in Figure 1. This two-stage approach was arrived at based on
the design goal of reaching 80K and the miniaturization requirement in which the compressor and
coldhead are separate components. The compressor and coldhead were separated to reduce vibra-
tion at the coldhead and to balance the thermal operating loads at the compressor. This approach,
commonly referred to as multi-inlet when two or more stages are present, reduces the regenerator
loss by using a secondary orifice which diverts a small percentage (approximately 10%) of the gas
to travel directly from the compressor to the warm end of the pulse tube. This small flow bypasses
the regenerator and then compresses and expands the gas that remains at the warm end of the pulse
tube. This reduces the flow through the regenerator thus reducing the regenerator loss accordingly.
For optimal performance this configuration relies on optimized and stable flow division (provided
by the three orifice impedances), minimum void volume, maximum pressure ratio, and minimiza-
tion of any DC flows or turbulence. Analytical and numerical models such as REGEN3.1 devel-
oped by NIST2,3,4 and a thermoacoustic model developed by Xiao5,6,7,8 were employed to design both
the prototype and flight coldheads.
The prototype cooler is shown in Figure 2. The test fixture for the prototype coldhead allowed
the flow division between the primary and secondary orifices to be adjusted during operation using
external metering valves, which were modified to minimize void volume. The approximate flow
rates were easily determined for the primary and secondary orifices based on the metering valve
settings. However, the intermediate flow path distribution between the first and second stages
THE NIST/LOCKHEED MARTIN MINIATURE PT CRYOCOOLER 191
Figure 1. Schematic of double inlet configuration. Figure 2. NIST/LMA pulse tube prototype.
required manual adjustment. Therefore partial disassembly of the coldhead was necessary during
several iterations to optimize the intermediate orifice. Calculations provided the approximate final
flow impedance for this orifice. The fabrication, characterization, and best thermal performance
were accomplished within only 3 months of the initial design phase. The prototype unit used a
larger 2.5 cc laboratory compressor operated at reduced power to simulate the expected power of
the smaller 0.75 cc flight unit. Instrumentation of the prototype coldhead provided temperatures,
heat loading, and pressure data for evaluating pressure ratios and phase angles. The relatively long
transfer line and small but unavoidable void volumes in the valves limited the efficiency of the
prototype coldhead, but its low temperature performance proved the design feasibility of such a
small system. A low temperature of 84 K with a pressure ratio of about 1.23 was achieved. For a
pressure ratio of 1.26 however, 76 K was achieved. When the 0.75 cc flight compressor was
attached to the optimized prototype coldhead with its attendant void volume a temperature of only
127 K was achieved. This resulted from the much lower pressure ratio of 1.13, indicating that the
PV work was lower with the flight compressor. A pressure ratio of 1.2 to 1.25 was the design value
for the MPTFC. Steps were taken in the fabrication of the flight coldhead to minimize any void
volume in the system in order to deliver the PV work associated with the design pressure ratio.
The flight cooler (MPTFC) coldhead shown in Figure 3 has a PT volume of nearly 0.54 cc (see
Table 1 for other important coldhead dimensions). The figure provides an exploded view of the
two stages but several components are not shown. The cold end and aftercooler are made using
OFHC copper; the regenerator and pulse tubes are thin wall 304 stainless steel. The reservoir,
which is also made from 304 stainless steel, has a bracket for attaching the flight pressure trans-
ducer. Two smaller diagnostic pressure transducers communicate with the compressor and pri-
mary orifice spaces of the coldhead. The transfer line is shown prior to final flight modification,
Figure 3. NIST/LMA MPTFC Coldhead.

and the pressure relief capillary is also visible. The primary and secondary orifices (not shown) are
integrated into the aftercooler. Clamps and seal rings to connect the two stages and the fixed
radiation shield / MLI are not shown. Also not shown are two thin plates that provide impedances
to form the secondary and intermediate orifices. Many other flight-related elements including other
instrumentation such as temperature sensors and a film heater on the cold end are also omitted in
this figure. The MPTFC schematic is shown in Figure 4. Some important features of the MPTFC
design are 1) a very compact physical arrangement; 2) the aftercooler and the compressor are
conductively cooled through contact to the experiment mounting plate (EMP); 3) nylon displace-
ment stops are located at both the coldhead inter-stage and cold end. The development and test
phases for the flight coldhead required about 11 months to complete concluding with final prepara-
tions for vibration testing just prior to integration.
Coldhead Instrumentation
The MPTFC was extensively instrumented for temperature, pressure, and vibration mea-
surements. In fact there were double and triple redundancies built into the system. All of the
sensors employed were commercial off the shelf (COTS) items. Specifically, thin film platinum
RTDs, piezoresistive and piezoelectric pressure transducers, thin film heaters, and tri-axial acceler-
ometers were used. Schedule constraints required two diagnostic piezoresistive-pressure transduc-
ers to be epoxied in place for flight to mitigate the risk of GHe leakage. The electronics system was
designed and built in-house and made extensive use of COTS hardware. The DAS was configured
at a 5 minute scan rate in a “fill and hold” mode.
G-197 Design Features
Figure 5 shows the G-197 Payload minus the GAS canister enclosure. This assembly can be
referred to as a cantilevered frame support. The upper third of G-197 consists of the MPTFC ex-
periment itself, a commercial latching vacuum valve to expose the experiment to space vacuum,
Figure 4. Exploded schematic of the Figure 5. TheG-197 payload minus GAS canister.
flight cooler arrangement.
and other electrical and battery venting connecting hardware. The lower two thirds (the “battery
box”) contains the battery modules and the flight electronics module — the DAS, computer, signal
conditioners, and drive electronics for the experiment. This entire assembly, i.e., G-197, integrates
with the NASA GAS canister, which is a 5 cubic ft. cylindrical containment vessel that is capped at
both ends and evacuated during integration. There are four supports at the cantilevered end, known
as bumpers, that lock the bottom of G-197 radially into the canister by preloading against the
canister walls. This must be done in order to restrict the movement of the lower end of the battery
box during launch and landing, thereby reducing the bending loads applied at the opposite end
where the experiment attaches to the EMP with only 12 Ti alloy screws. The completed GAS
canister then encloses the experiment with all electronic connections accomplished via feedthroughs
at the canister bottom plate which holds the NASA GAS electronics. The bottom plate is the
electronic interface between G-197, the GAS relay system, and the shuttle GAS computer which is
operated by astronauts.
Special features of the G-197 GAS payload are: 1) the EMP is uninsulated and coated with
silverized Teflon tape for maximum heat rejection to a space environment, 2) the GAS canister is
evacuated to simulate a flight instrument environment, 3) The primary battery system consisted of
thirty-three 2V batteries in a 3 string redundant arrangement to power the MPTFC and its compres-
sor drive electronics at a nominal 22V. The secondary battery system consisted of four 4V batteries
in series to maintain a nominal 16V supply to power the EM valve, DAS, and computer electronics.
All batteries are polyurethane foamed into the BB and are vented to EMP relief valves, 4) “low
voltage” and “high temperature” cutoff circuits are employed (no longer required by NASA), and
5) battery voltages and GAS canister temperatures are measured for the uninsulated EMP for com-
parison to NASA numerical models developed in 1987.
Flight Safety Features
The MPTFC design employed several voluntary safety features as well as those required by
NASA to safeguard the STS and astronauts during flight. Although not all of the required safety
features were practical for mission success, they had to be accommodated in order to fly MPTFC on
the shuttle. NASA vacillated on certain requirements, but the following were final. Although the
sealed lead acid primary batteries were vented external to the GAS canister to safeguard against the
buildup of explosive gases, both low voltage and high temperature cutoff circuitry were also re-
quired. Proof pressure testing was required for the MPTFC vacuum housing to more than ten times
the maximum pressure that would exist if the cooler developed a leak or the compressor pressure
exceeded the limits of its housing. Thermal cutoff switches were employed in the drive electronics
to prevent an overheat condition within the compressor. Furthermore, a pressure relief mode, con-
sisting of a capillary that was sealed using an indium soldered cap connected to the warmest loca-
tion on the MPTFC, was required in case the MPTFC overheated. The compressor drive power was
set conservatively to eliminate excessive initial vibration. Polyswitch fusing and redundant wiring
were also required to complete the electronics package. The 3.6 V Li cell used to retain memory
was double-diode isolated and fused.
PREFLIGHT TESTS
All optimizations of both the prototype and the flight cooler (MPTFC) systems were con-
ducted in the laboratory in a bench test environment. All instrumentation and drive electronics
were installed in the best configuration for optimization and therefore were not configured (i.e.,
wired or attached) for flight. This meant that upon completion of optimization the sensors were
removed and reworked for the flight configuration, including rewiring, reattachment, and func-
tional tests.
Experiment Functional Testing
After reworking the MPTFC for the flight environment and integration with the flight electron-
ics and the flight support structure, a complete functional verification of all sensors and a perfor-
mance verification of the G-197 system were made. This of course included a final pump-out and
Figure 6. Cool down for flight system rejecting to 258 K.
gas helium pressure charge of the cooler system to 2.6 MPa. These tests included verifying that
minimum temperatures and maximum refrigeration loads were maintained, as well as the cool-
down performance. Figure 6 shows a representative cool-down for the flight system as measured
during thermal vacuum testing conducted at the LMA facility. Figure 7 shows the MPTFC perfor-
mance for a nominal rejection temperature of 273 K. This data was consistent with data measured
before the flight configuration and integration with the DAS and flight electronics rework.
EMI Testing
Although G-197 was a GAS payload, which is considered to be a very low risk to the shuttle
operations / communications when fully sealed, NASA required that the radiated EMI of the pay-
load be measured. This requirement for G-197 was due to the vacuum line, which runs from the
cooler housing via the EM valve to an EMP port for the on-orbit evacuation of the housing. The
Figure 7. Nominal cooler performance for a 273 K rejection temperature.

MPTFC and flight electronics subsystem were tested at the LMA facilities. Testing included low
frequency to GHz radiated emissions for both narrow and broad band. The G-197 radiated EMI
easily passed within the acceptable limits imposed by NASA.
Vibration Testing
Complete vibration testing was required for the G-197 payload because it slightly exceeded the
nominal limit of 91 kg (200 lb.). Both a low-level sine sweep and random vibration at qualification
and protoflight acceptance levels were performed in lateral and axial modes. These tests were
accomplished at the NASA/ARC. It is of note to the reader that these vibration levels are rather
severe and often are conservative compared to the actual launch environment. They represent a
safety concern by NASA for the structural integrity of the system. However, NASA requires this
conservative level of testing because of the uncertainty of both location of the payload and flight
vibration loads within the STS bay. G-197 was calculated to have a lowest lateral mode resonance
of 58 Hz when installed in the GAS flight canister. A measured resonant frequency of 40 Hz was
obtained using the shipping canister, still above the required 35 Hz minimum. An unexpected
result of this test was that some of the Pb-acid batteries failed in the axial mode. This was recog-
nized by a decrease in the primary string voltage from a nominal 22 V to ~20 V. However, due to
schedule constraints, coupled with the knowledge that the MPTFC experiment can be operated
effectively at 16 V, no changes or modification to the batteries were undertaken. It was believed
that there was sufficient margin to continue with the scheduled integration and launch, anticipating
that there would be no further degradation.
INTEGRATION AT NASA/KSC
Upon completion of the vibration testing and subsequent pre-integration functional check-
outs, the G-197 payload was packaged and shipped to NASA Kennedy Space Center (KSC) for
integration with the GAS flight electronics and flight canister. The integration is typically a three-
day process that takes place about 3 months prior to launch. Upon completion and sign-off of the
integration, there is no opportunity for further contact with the payload by the experiment investi-
gators. Therefore the MPTFC experiment had to be able to retain its 2.6 MPa gas helium charge
and an adequate battery charge over a three-month period. This requirement and STS safety con-
siderations effectively eliminate many types of batteries used for unmanned missions.
The integration process is a very challenging exercise for a complicated powered experiment
such as MPTFC. First the payload is unloaded from the shipping canister, which is quite similar to
the GAS flight canister except that it has no NASA electronics and acts only as a protection vessel
for shipping. However, because the support bumpers must be positioned to safeguard the experi-
ment during transit, disassembly is required at NASA/KSC. Next, a very thorough visual safety
inspection of the battery box (BB), electronics / fuse box, and experiment itself is conducted by
NASA/KSC personnel to ensure that the payload is in full compliance with all safety paperwork.
After the safety inspection is signed off, the real work to prepare the experiment for flight begins.
The battery systems must be top-charged and the BB and vent plumbing subjected to a pressure
proof test prior to a GN2 purge. The experiment housing must be evacuated and valved off before
a preliminary system functional check is made. After a final visual inspection, all fasteners are
secured for flight which involves epoxying and/or lock-wiring external screws and securing wiring
/ cables. A weigh-in of the completed experiment verifies that it is within approved limits before
NASA will proceed. Finally, the payload is installed into the flight canister, the bumpers are preloaded
for flight, and the end plate with NASA electronics / interface cable is integrated to the payload. A
final functional test is then made to ensure that the NASA electronics and payload are compatible.
This test also serves to verify EM valve operation, MPTFC cooling, and adequate battery margin.
The EM valve is reset to a closed position and a removable manual valve (which maintains a guard
vacuum in the line connecting the EM valve to the EMP port) is closed and tagged for removal prior
to flight. The GAS canister is evacuated and the payload is officially handed off to NASA/KSC for
STS integration.
FLIGHT DATA
Minimal Cooling
The cooling data collected during the STS-90 mission was unfortunately of minimal content.
After initial confirmation of primary power, the first data set consisted of 20 hrs, powered mostly
by the secondary system, which operates the coldhead heater and the DAS in the absence of pri-
mary power. Attempts to operate the cooler on-orbit were apparently foiled by a fluctuating pri-
mary voltage, which dipped below 16 V causing a designed malfunction indication at the GAS
control computer. Specifically, this led to periodic cutouts of relay ‘A’, eliminating power to the
MPTFC compressor drive electronics. This relay supposedly resets automatically at 40 minute
intervals by a NASA timer, but there was no indication of this operation in the data set. The relay
was switched off and reset manually later in the mission, again with positive initial indication but
little evidence of cooling in the data. This was unfortunate since the MPTFC can operate safely
below 16 V, thereby making the NASA-mandated cutout level for battery protection unnecessary.
It should be noted that NASA has recently eliminated malfunction circuitry requirements for GAS
payloads. Measures to perform a re-flight of this experiment (as G-785) to demonstrate the on-orbit
cooling have already been performed. The robustness of the MPTFC and its survivability without
performance degradation have already been proven.
“Off-state” Thermal Conductance Test
The on-orbit data collected from the second data set of 15 hrs included continuous cold stage
heater operation. This data comprises the on-orbit conductanc test for a pulse tube cooler in an off-
state passive mode. To date there has been little if any published data of this type. In fact, less than
a handful of PT coolers have been operated in space. This conductance data and its analysis are
presented in a related paper at this conference.9
POST-FLIGHT TESTS
Battery testing
One month after STS-90 touched down at KSC the G-197 payload was de-integrated and de-
livered to LMA personnel. A post-flight functional test determined that the primary battery system
exhibited random voltage fluctuations that varied from ~15 to 21V. Later it was determined that at
least 7 of the 33 cells were either intermittent or completely defective. At least two cells on each
redundant string were affected. This condition explained the relay cutouts during flight. The low
voltage condition was attributable to the vibration tests, the launch, and the -6°C STS bay tempera-
ture during initial MPTFC activation.
Accelerometer data and model results
Analysis of post-flight data from a miniature tri-axial accelerometer mounted at the reservoir
of the cooler during operation was evaluated using a finite element model. It predicts a lateral
displacement of 1.14 at 164 Hz for the as-built u-tube configuration. However, if a third thin
wall support member were used to stiffen each stage, the model predicts the displacement would
decrease to 0.21 at a frequency of 391 Hz. This result indicates improvement may be made to
further reduce induced vibration at the cold tip for vibration sensitive sensor packages.
Cold environment testing
Subsequent to post-flight battery refurbishment (see below), cold environment testing was
performed at rejection temperatures below the nominal 273 K. The performance as a function of
the heat rejection temperature was conducted and the results are shown in Figure 8. The heat
rejection temperature affects both stages consistently. The first stage temperature ranges from
161 K to 204 K for both a no load and a 45 mW load at the second stage based upon a rejection
temperature ranging from 244 K to 300 K. For the same rejection temperature range the second
Figure 8. MPTFC cooler performance as a function of heat rejection temperature.

stage temperature ranges from 82 K to 105 K for no load, while it ranges from 90 K to 121 K with
a 45 mW load. The cold testing also served to confirm the nominal operation of heaters located on
the flight electronics and compressor.
PERFORMANCE DATA
This system as currently configured is certainly not very efficient, having a Carnot efficiency
of only 2%. However, because of its very small size it represents an important step toward minia-
turizing PT systems. As most readers of this paper are aware, extensive experimental optimization
is required to achieve the best performance for PT cryocoolers after thorough modeling has been
completed. Often the first iteration in the system design based on the numerical model falls short of
the intended goal. This is especially true when pushing the predicting capabilities of the model into
a new scaling arena, as for the MPTFC. Although the schedule allowed for only one design itera-
tion, much has been learned about scaling miniature PT coolers, e.g., the complexities associated
with the flow distribution and optimization for a two-stage multi-inlet design. For best efficiency
the importance of minimizing parasitic heat leaks can not be overlooked.
Immediately after flight the MPTFC demonstrated no reduction in performance and there was
no detectable change in the system pressure. In fact, it has held pressure for 2.5 yrs with only a 5%
loss, but even this loss has degraded thermal performance from a no-load temperature of 111 K to
123 K at ambient rejection. A loss of only 5 % in the pressure nonetheless represents significant
seal performance for a system designed for modification flexibility.
RE-FLIGHT
At present a re-flight of the MPTFC experiment is in progress as G-785. A re-flight provides
a potential opportunity for MPTFC improvements, depending on the manifest date set by NASA.
A new battery system is always required for a GAS re-flight. Failure of the NASA-recommended
primary system batteries used in G-197 resulted in the selection of new batteries of the successful
secondary system type, but even these batteries revealed a sourcing issue. Specifically, while the
original 4V batteries and their new replacements were of the same model from the same vendor,
they were not from the same factory. A different internal design in the newer version resulted in
failure during sample vibration tests; G-785 uses older version cells.
Other improvements include a high resolution voltmeter, improved frequency diagnostics,
and an increased DAS scan rate. All paperwork and refurbishment are complete for a re-flight and
we are anticipating an opportunity in early 2001. If it becomes necessary to re-pressurize the MPTFC
to 2.6 MPa, more diagnostic sensors and some radiation baffles will be installed.
FUTURE MPTFC COLDHEAD DESIGN AND TEST

Future efforts in MPTFC research will concentrate on improving the current 2-stage coldhead,
including evaluation of DC flow, further reduction of the parasitic heat loads9, modification of
stage geometry, and performance evaluation at higher operating pressures. Furthermore, a new
one-stage design effort will be undertaken for comparison with the two-stage approach. Extending
existing numerical models with innovations will streamline the design and test of miniature sys-
tems, thereby advancing their predicting capabilities and accuracy.
ACKNOWLEDGMENTS
We especially wish to thank Dr. Joe Martin for financial and program support, Mr. Don
Hirschfield for assistance during integration at NASA/KSC, Mr. Ken Sell for intricate machining,
and Mr. Aleks Bakman for vibration analysis, all of LMA; Dr. Peter Kittel of NASA/ARC for
vibration testing; Mr. Michael Lewis of NIST for technical assistance during MPTFC testing; and
Mr. Mark Wallace of Campbell Scientific for software assistance.
REFERENCES
1. S. Zhu, P. Wu, and Z. Chen, “Double inlet pulse tube refrigerators: An important improvement,”
Cryogenics 30, 514 (1990).
2. P.J. Storch, R. Radebaugh, and J.E. Zimmerman, “Analytical Model for the Refrigeration Power of
the Orifice Pulse Tube Refrigerator,” NIST Technical Note 1343 (1990).
3. Gary, J., Daney, D.E., and Radebaugh, R., “A computational model for a regenerator,” Proc. Third
Cryocooler Conference, NIST Special Publication 698, (1985), p. 199.
4. Gary, J. and Radebaugh, R., “An improved numerical model for calculation of regenerator perfor-
mance (REGEN3.1),” Proc. Fourth Interagency Meeting on Cryocoolers, David Taylor Research
Center, Report DTRC-91/003, (1991), p. 165.
5. J. H. Xiao, “Thermoacoustic theory for regenerative cryocoolers: A case study for a pulse tube
refrigerator,” Proc. 7th International Cryocooler Conference, Air Force Report PL-CP-93-1001, Kirtland
Air Force Base. NM (1993), p. 305.
6. J. H. Xiao, “Thermoacoustic heat transportation and energy transformation, Part 1: Formulation of
the problem,” Cryogenics 35, 15 (1995).
7. J. H. Xiao, “Thermoacoustic heat transportation and energy transformation. Part 2: Isothermal wall
thermoacoustic effects,” Cryogenics 35, 21 (1995).
8. J. H. Xiao, “Thermoacoustic heat transportation and energy transformation, Part 3: Adiabatic wall
thermoacoustic effects,” Cryogenics 35, 27 (1995).
9. D. R. Ladner, P. Bradley, and R. Radebaugh, “Offstate Conductance Measurements of the NIST/
Lockheed Martin Miniature Pulse Tube Flight Cryocooler: Laboratory vs. Space,” Cryocoolers 11,
Plenum Publishers, NY (2001).
Low-Cost Pulse Tube Liquefier for In-Situ
Resource Utilization
C.M. Martin and J.L. Martin

Mesoscopic Devices, LLC
Broomfield, Colorado, USA 80020
ABSTRACT
NASA’s strategy for continued exploration of Mars is based on the concept of using
Martian resources to supplement materials brought from earth. This in-situ resource utilization
(ISRU) program allows dramatic reduction on the mass of materials that must be transported
from the Earth, and is an enabling technology for a manned mission to Mars. A key part of the
ISRU strategy is to use the Martian atmosphere along with hydrogen feedstock and chemical
reactors brought from earth to produce oxygen and a hydrocarbon as rocket fuel for the return trip
to Earth.
Any oxygen produced on Mars will need to be stored for months to years as sufficient
reserve is built up. The overall weight of the storage system is lower for liquid oxygen than
pressurized gas, so a liquefier is required. We are developing a low-cost oxygen liquefier for in-
situ resource utilization. The design point for this cooler is 20 W at 89 K rejecting to 245 K.
The liquefier uses a pulse tube cryocooler, with a linear, opposed pressure wave generator.
The pulse tube is being designed for compactness and ease of integration with the balance of the
system. The first generation cryocooler has been built and is currently being tested. Performance
predictions for this cryocooler are presented.
INTRODUCTION
Exploration of Mars will require utilization of indigenous resources to support human life
and operations.1 One component of ISRU is the manufacture and storage of rocket propellants.
Storage of these propellants can be accomplished for the lowest mass using cryogenic
liquefaction and storage.
Under support from NASA/Johnson Space Center through the Small Business Innovative
Research program, Mesoscopic Devices is developing a pulse tube cryocooler to liquefy oxygen
and propellant in support of ISRU. This cryocooler must be low-weight, capable of operating
over a large range of ambient temperatures, and scalable to large powers to support human
missions. We are one year into a two year effort and have completed the fabrication of the first
generation pulse tube liquefier. The design of this cryocooler is described below.
IN-SITU RESOURCE UTILIZATION

In-situ resource utilization involves using the indigenous resources on other solar bodies to
support their exploration. In the case of Mars, in-situ consumable production (ISCP) has the
potential to significantly reduce the cost and risk of robotic and human exploration.1 This would
involve sending feedstock, such as liquid hydrogen, to Mars along with a chemical processing
plant. The chemical processing plant would use the carbon dioxide from the Martian atmosphere
and the hydrogen brought from earth to generate rocket propellants, such as oxygen, methane,
methanol or other hydrocarbons. In addition, oxygen can be produced for crew life support. The
initial mission mass required in low-Earth orbit can be reduced by 20% by using ISCP.
ISCP requires three components: resource collection, chemical processing, and liquefaction
and storage. NASA is pursuing all three areas concurrently. The work presented in this paper
addresses the liquefaction issue. A liquefier is required to cryogenically liquefy the oxygen, and
probably the hydrocarbon, for storage.
The requirements for the liquefier are still evolving, but there are several key features that
need to be addressed in all potential designs. First, minimum weight is essential. This weight
must account for the cryocooler, as well as the storage dewar, power source and heat rejection
system.2 Unlike terrestrial applications, heat rejection will have a significant impact on the
overall weight and must be considered at the earliest stages of the design. Improved cryocooler
efficiency may increase the weight of the cooler slightly, but it may be more than compensated
for by reducing the weight of the heat rejection system.
Other key factors include the operating duty cycle and the cooling loads and temperatures
required. These have been discussed previously and are an area that continues to evolve.3 The
one thing that is certain is that initial missions, such as rock sample return missions, will have
much lower cooling requirements than later, manned missions. For this reason, a cryocooler
design that can be effectively scaled in size will provide a reduction in the risk for human
missions by demonstrating key technology in smaller, earlier missions.
CRYOCOOLER DESIGN
We selected a coaxial pulse tube cryocooler driven by a linear, opposed pressure wave
generator. The use of a coaxial cold head simplifies integration with the storage dewar, as the
cold head can be inserted directly into the dewar neck. An annular regenerator surrounds the
pulse tube. A condenser, to provide additional condensation surface area, can be bolted to the
cold tip before insertion into the dewar.
Figure 1 shows the cryocooler and Figure 2 shows the coaxial pulse tube. Table 1 shows the
key design parameters and overall dimensions.
The liquefier uses an inertance tube to generate the required phase shift. Due to the
relatively high operating frequency (60Hz), phase differences of up to 52 degrees between the
volumetric flow rate and the pressure can be generated at the entrance to the inertance tube. This
allows for near-ideal phase relationships at the cold end and regenerator. Another key advantage
of the inertance tube design is the elimination of DC flows through the pulse tube. The baseline
inertance tube is 4 mm × 2.4 m long. For testing, this tube is routed out the side of the inlet
cap to an external reservoir. This arrangement facilitates optimization of the inertance tube
parameters. In the final configuration, the inertance tube is coiled in an annular volume inside
the compressor surrounding the pistons. A 0.5 liter reservoir volume is provided inside the
pressure vessel, separate from the compressor back side.
The pulse tube is tapered to suppress acoustic streaming. We used the methodology of Swift
to calculate the optimal taper angle.4 The required angle is very small, only 1.6 degrees (full
cone angle) for this design.
The aftercooler is liquid-cooled, with 100 mesh copper screens on the helium side and
circumferential fins and grooves on the coolant side. For laboratory testing, we circulate a
mixture of ethylene glycol and water through the 6 mm high by 2 mm wide passages to reject the
heat from the cycle. For a Mars application, the aftercooler would be coupled to a radiator using
either a pumped single-phase loop or a two-phase thermosyphon or heat pipe. Design of the
Mars heat rejection system will be a significant task in the second year of our program.
Mesoscopic Devices worked closely with CFIC of Troy, NY to develop the compressor for
the liquefier. The compressor uses patented STAR motors and bent strap flexure bearings.5
LOW-COST PT LIQUEFIER FOR IN-SITU RESOURCE UTILIZATION 201
Figure 1. Oxygen liquefier.
Figure 2. Cold head, shown 180° from normal operating orientation

(cold tip is up here but would be down in operation).
STAR motors are a type of moving magnet design that uses an interdigitated stator-armature
combination and separate bobbin-wound coils. The 100 W STAR motors used for this
compressor represent the smallest STAR motors built to date. Motors of this type have already
been built at 100 W, 300 W, 2 kW and 10 kW. Linear compressors using this motor and flexure
design can be built from 200 W to 20 kW of input power, spanning two orders of magnitude in
cooling power. This provides a clear growth path from cryocoolers sized for robotic sample
return missions, which will require a few kg/day of propellant, to manned missions that might
require several hundred kg/day for breathing, exploration and propellant uses. The compressor
design parameters are also included in Table 1.
Performance Predictions
Figure 3 shows the predicted load curves for this machine. We used, DELTAE, a
thermoacoustic code, to generate the predicted load curves. The two curves are for reject
temperatures of 245 and 293 K, representing the design operating conditions in a Martian
environment, and typical laboratory conditions. The curves indicate the maximum load, limited
by either the compressor input power (200 W maximum) or the compressor stroke limit (12 mm).
LOW-COST PT LIQUEFIER FOR IN-SITU RESOURCE UTILIZATION 203
Figure 3. Predicted load curves for reject temperatures of 245 and 293 K.
SUMMARY
The Mars ISRU application requires a cryocooler that is efficient, low weight and can be
scaled to larger cooling loads to support manned missions. The pulse tube cryocooler described
here will meet these requirements. The same basic design can be scaled up to 1 kW of input
power, and an in-line version can be scaled up to 20 kW (we are currently building 4 and 20 kW
liquefiers using this same design for terrestrial applications).
ACKNOWLEDGEMENTS
This work was supported by NASA/JSC through the Small Business Innovative Research
program under contract NAS9 99081.
REFERENCES
1. NASA Technology Plan, http://actuva-www.larc.nasa.gov/techplan (1999).
2. J.L. Martin, et. al., “Low-Cost, High-Performance Cryocoolers for In-Situ Propellant Production”,
In-Situ Resource Utilization (ISRU-III) Technical Interchange Meeting, Denver, CO, February 1999.
3. P. Kittel, L.J. Salerno, and D.W. Plachta, “Cryocoolers for Human and Robotic Missions to Mars”,
Cryocoolers 10, Kluwer Academic/Plenum Publishers, New York (1999), pp. 815-821.
4. G.W. Swift, M.S. Alien, and J.J. Wollan, “Performance of a Tapered Pulse Tube”, Cryocoolers 10,
Kluwer Academic/Plenum Publishers, New York (1999), pp. 315-320.
5. US Patent 5389844, Linear Electrodynamic Machine, February 14, 1995.
Performance Characteristics of a 4 K Pulse Tube
in Current Applications
C. Wang and P. E. Gifford

Cryomech, Inc.
Syracuse, NY 13211, USA
ABSTRACT
The Cryomech Model PT405 has cryo-cooled several applications during the past year where
both its 4 K temperature and pulse tube characteristics were required. The specifications for the
PT405 have improved since it was introduced. Presently the typical cooling capacity is 0.6 W at
4.2 K on the second stage with simultaneously 30 W at 65 K on the first stage with an input
power of 4.9 kW. With the new CP900 Series compressor packages, the PT405 maintains the
same performance for both 60 Hz and 50 Hz power supplies. The force transmitted into a
cryostat by the PT405 base plate is less than 3.6 N. The same test registers a force of 178 N for a
4 K GM Cryorefrigerator. The maximum displacement (elongation of the base tube assembly
due to the cycling of the high and low pressures) at the 2nd stage heat exchanger is Some
applications of the PT405 are presented in the paper; such as, operating a low vibration cryostat,
conductively cooling a superconducting magnet, pre-cooling an ADR X-ray detector and
condensing helium.
INTRODUCTION
Since the discovery of the pulse tube cryorefrigerator, the cryogenic community has been
looking forward to benefiting from its no-displacer design. The pulse tube cryorefrigerator
promises to improve reliability, to increase the meantime between maintenance, to extend system
lifetime, and to decrease the cryorefrigerator cost in comparison with Gifford-McMahon (GM)
and Stirling Cycles. Yet, to many, the most exciting benefit of the absence of the displacers in
the PTR was the reduction of “g” forces exerted on the cryostat and lack of magnet gradient
disturbance caused as the displacer moved up and down in the GMs.
Over the past few years, significant progress in the design of pulse tube cryorefrigerators has
led to an increase in its efficiency in comparison with conventional cryorefrigerators. There are
several types of pulse tubes commercially available, such as the TRW “Stirling type” space pulse
tube1 and IWATANI “GM type” pulse tube, etc.
In the summer of 1999, Cryomech introduced a 4 K Pulse Tube, the Model PT405. Although
others had introduced single stage PTRs, the PT405’s 4 K performance provided the first
opportunity to test the PTR in several of the most challenging cryo-cooling applications. The
performance of PT405 has been improved by recent modifications and was characterized for
applications. Some of the current applications will be discussed in this paper.

206 GM-TYPE PULSE TUBE COOLERS FOR LOW TEMPERATURES
PT405 COLD HEAD

PT405 is a two-stage, 4 K pulse tube cryorefrigerator that includes a pulse tube cold head, con-
necting flexible lines and a helium compressor package. The pulse tube cold head employs a double-
inlet configuration. Figure 1 shows a photo of the PT405 cold head. A new compressor, Cryomech
Model CP950, was specially designed for the pulse tube with regard to a long meantime between
maintenance (MTBM). The compressor package supplies the cold head with pressurized helium
through flexible metal hoses. A rotary valve in the cold head, similar to that in a GM cryorefrigerator,
directs the helium gas in and out of the pulse tube system.
The 1st cooling stage of the refrigerator can provide cooling power between 35 K to 80 K, and the
nd
2 cooling stage below 4 K. Owing to some advanced designs in the whole system, the MTBM of the
PT405 is expected to be > 20,000 hours.
PERFORMANCE CHARACTERISTICS OF PT405

Cool down performance
To maintain the standard specifications for both the 50 Hz and 60 Hz electrical power supplies,
the PT405 is supplied with different compressor packages: the CP950 for 60 Hz, and the CP970 for
50 Hz. The input power in both cases is 4.9 kW with the 1st stage operating at 65 K and the 2nd stage
at 4.2 K. A typical cooling-load map for the PT405 is given in Figure 2; for example, it shows that the
cooler provides 0.6 W at 4.2 K on the second stage, simultaneously with 30 W at 65 K on the first
stage. The performance of the PT405 has been improved upward from the initial announced capaci-
ties of 0.57 W at 4.2 K and 18 W at 65 K.2
Figure 2. Typical cooling-load map of PT405.
Figure 1. Photo of PT405 cold head.

PERFORMANCE OF A 4 K PT IN CURRENT APPLICATIONS 207
Figure 3. Cool down curves for PT405 with and without mass attached.
The cool down time with a thermal mass attached for the PT405 was tested with 6.4 kg of
OFHC copper on the 1st stage and 0.9 kg of OFHC copper on the 2nd stage. Figure 3 shows the
cool down times of the PT405 with and without the added mass. It takes 60 minutes for a
unloaded PT405 to reach 4 K and 110 minutes for both stages to reach their minimum
temperatures of 2.6 K and 35 K. With the above thermal loads attached to both stages, the 2nd
stage reaches 4 K in 100 minutes, and the two stages reach their minimum temperatures in 170
minutes.
Vibration of PT405
Cryomech, Inc. has collaborated with several different groups to analyze the vibrations
generated by the PT405. Figure 4 shows the test rig used by GE R&D Center to analyze the
mounting forces exerted on the cryostat at the PT405 base plate. A load cell was mounted under
the base plate of cold head. The displacement of the 2nd stage cold heat exchanger was measured
with both an optical comparator and an accelerometer. The accelerometer was mounted on the
bottom of the 2nd stage cold heat exchanger as shown in Figure 4. The optical measurement was
made with a ST405 Cryostat (Figure 5). A laser beam penetrated through the optical windows in
the cryostat and focused on the second stage heat exchanger, measuring the movement.
The mounting force of PT405 compared with that of a mechanical driven GM
cryorefrigerator is shown in Figure 6. The mounting force of PT405 is less than 3.6 N (0.8 lb)
and for the GM 178 N (40 lb). The mounting force of pulse tube is only generated by the
pressurization and depressurization of helium in the pulse tube assembly. The curve is similar in
its wave shape to the dynamic gas pressures.
Table 1 gives the displacements of the 2nd stage cold head measured by both the optical and
the accelerometer methods. The vectors of the three-axis are given in Figure 4. The two tests
measured similar displacements. The maximum displacement around is in the vertical
direction.
Figure 4. Vibration test rig .
Figure 5. Photo of ST405 cryostat.

Figure 6. Mounting forces of pulse tube and GM cryorefrigerators.
Effects of a Magnetic Field or Orientation on the Cooling Performance

The effects of a magnetic field on the operation of the PT405 have been tested. Due to the
limitation of resources, we have not tested the PT405 in a large enough magnetic field that has
degraded its performance. We have designed the second stage regenerator with anti-
ferromagnetic materials to minimize the loss of heat capacity in field. In one test we moved a
PT405 Cold Head while operating in a cryostat at minimum temperature toward a
superconducting magnet. In this test, a field strength of 900 gauss had no affect on the rotary
valve motor or bottom temperature. Higher magnetic field will stop the motor. According to the
published information of the rare earth materials, it is expected that the low temperature
performance of the 2nd stage will be slightly decreased in the magnetic field higher than 1 Tesla.
It is suggested for maximum performance that the pulse tube cold head operate as close to
vertical as possible. “Gas mixing losses” in the pulse tubes caused by gravity decrease the
performance of both stages when the pulse tube cold head is tilted off the vertical position2.
However, mounting angels up to 30° off vertical will not degrade the performance greatly.
SOME APPLICATIONS OF PT405

Low vibration cryostat
A 3 K cryogen-free cryostat with optical access was built at Cryomech, Figure 5. The Model
ST405 has very low vibration, as can be seen in Table 1. A radiation shield is fixed on the 1st
cooling stage to reduce the radiation loss on the 2nd cooling stage and sample holder. It is
designed for two and three axis optical measurements. There are 5 holes with diameter of 12.7
mm through the radiation shield and 5 optical windows of diameter of 25.4 mm on the optical
cube. The radiation losses are crucial for the sample temperatures below 4 K. The bottom
temperature on the sample holder in the cryostat is 2.6 K, when all holes through the radiation
shield are covered with aluminum tape; 3.0 K with one hole uncovered; and 3.5 K with two holes
uncovered.
Small helium-4 liquefier or re-condenser

A small helium liquefier was developed at Cryomech using the PT405. In the system room
temperature helium gas is pre-cooled by the 1st stage and 2nd stage regenerator. Figure 7 shows a
schematic of the liquefier. The pre-cooling of the helium from the first stage temperature down
to approximately 6.5 K using the inefficiency of the 2nd stage regenerator is critical for increasing
the liquefaction rate. Numerical analysis made by the present authors calculates that the pre-
cooling heat load on the 2nd stage regenerator, decreases the PT405 2nd stage cooling capacity
by only 10% of the heat actually absorbed into the regenerator. Figure 8 is a curve of the rate of
liquefaction based on the liquid helium level in our condensation chamber. The accuracy of the
level instrument was mm, and the liquid level sensor indicated that liquid helium level
increased steadily. The pulse tube liquefier condensed 1.0 liters of liquid helium at 4.2 K in five
hours, for a corresponding liquefaction rate of 4.8 liters per day.
Cryogen-free operation of superconducting magnet

The first commercial cryogen-free superconducting magnet system cooled by the PT405 was
developed by Cryomagnetics, Inc. Figure 9 is a photo of the magnet system. The system has a
horizontal room temperature bore diameter of 32 mm. The first stage of PT405 is used for
cooling intermediate thermal shield and Bi-2223 based HTS current leads. The 2nd stage
conductively cooled NbTi-based superconducting magnet below 4 K. The complete system
takes approximately 14 hours to cool down the system to the operating temperatures: 50 K for
the thermal shield and <4 K for the magnet. The magnet generates a magnetic field up to 9 T
with 0.1% homogeneity over a 10mm diameter volume. After an intentional quench at 9.2 T, the
system recovered in temperature in 3 hours.
Figure 7. Schematic of pulse tube helium liquefier. Figure 8. Liquefaction rate of the PT Liquefier.
Figure 9. Photo of conductively cooling a Figure 10. Photo of precooling an ADR for SJT
NbTi-base SC magnet using PT405 Detector
Precooling an ADR for STJ detector

There has been considerable interest in attaching an adiabatic demagnetization refrigerator
(ADR) to the PT405 to cool superconducting tunnel junction (STJ) detectors to 100 mK. The
STJ is a powerful tool for materials microanalysis because they can accurately measure X-ray
energy. The PT405 will maintain the warm end of the two-stage ADR at 4 K, as well as cool the
shield and current leads with the first stage. The first integration of PT405 into an ADR was
made at Lawrence Livermore National Lab, Figure 10. A rigid cold finger joined the second
stage of PT405 to the 4.2 K stage of the ADR. The purpose of this rigid connection in the
preliminary test is to investigate the effects of pulse tube vibration on the detector measurement.
The energy resolution of the measurements taken of the X-rays using STJ degraded from 26 eV,
using liquid helium to 30 eV when attached to the PT405. Since no effort was made to
mechanically or electrically isolate the ADR from the pulse tube, this result is very encouraging.
CONCLUSION
During the past year, the performance characteristics of Cryomech’s 2-stage, 4 K pulse tube
cryorefrigerator have been measured in several applications. In each application, the pulse tube
improved the system performance over what was previously possible with a GM. The benefits
of the Pulse Tubes “no displacer” design are real. The PT405 has been improved to offer the
typical cooling power of 0.6 W at 4.2 K and simultaneously 30 W at 65 K for both 60 Hz and 50
Hz power supplies. The PT405 is opening up attractive applications for cryogenic cooling
around liquid helium temperatures.
ACKNOWLEDGMENT
We would like to thank R. Ackermann at GE R&D Center for proving mounting force data;
Cryomagnetics, Inc. for the data of conductively cooling a SC magnet; and J. Ullom of Lawrence
Livermore National Lab for the data of precooling ADR for SJT detectors.
REFERENCES
1. Tward, E., Chan, C.K., Jaco, C. et al, “Miniature Space Pulse Tube Cryocoolers”, Cryogenics, vol.31,
no.8(1999), pp.717-720
2. Wang, C. and Gifford, P.E., “Performance study on a two-stage 4 K pulse tube cooler”, to be published
in: Advances in Cryogenic Engineering, Vol. 45B, Plenum Press, New York (2000).
Experimental Study of a 4 K Pulse Tube
Cryocooler
S. Fujimoto1, T. Kurihara1, T. Oodo1, Y.M. Kang2,

T. Numazawa3, and Y. Matsubara4
1
DAIKIN Air-Conditioning R&D Laboratory, LTD.
3 Miyukigaoka, Tsukuba 305-0841, Japan
2
DAIKIN Environmental Laboratory, LTD.
3 Miyukigaoka, Tsukuba 305-0841, Japan
3
Tsukuba Magnet Laboratory, National Research Institute for Metals
3-13 Sakura, Tsukuba 305-0003, Japan
4
Atomic Energy Research Institute, Nihon University
7-24-1, Narashinodai, Funabashi, Chiba 274-0063, Japan
ABSTRACT
A prototype two-stage pulse tube cryocooler was previously developed that uses a double-
inlet configuration. When the second-stage regenerator was filled with conventional magnetic
regenerator materials, and the cryocooler achieved a minimum temperature of
2.89 K on the second stage and a maximum cooling capacity of 170 mW at 4.2 K. The rated input
power of the compressor unit is 3.3 kW at 50 Hz.
In this study, to improve the cooling capacity at 4.2 K for this cryocooler, we used a new
oxide regenerator material, in the second-stage regenerator. This material has a mag-
netic transition temperature of about 3.8 K, and has a considerably larger heat capacity compared
with that of and below 4 K. When was placed in the lowest-temperature
part of the second-stage regenerator, the cryocooler achieved a no-load temperature of 2.51 K at
the second stage, and a cooling capacity at 4.2 K of 250 mW. By adjusting the pulse tube size for
further optimization, a cooling capacity at 4.2 K of 288 mW and a no-load temperature of 2.42 K
was achieved.
To further evaluate the effect of we placed in the second-stage regenerator
of a 4 K Gifford-McMahon (GM) cryocooler. The cooling capacity below 4 K was improved, but
that at 4.2 K was degraded.
By numerical simulation, we determined the effect of on cooling performance both
for the 4 K pulse tube cryocooler and the 4 K GM cryocooler. For the 4 K GM cooler, the
numerical results roughly agreed with the experimental results. For the 4 K pulse tube cooler, the
numerical results showed that, again, the cooling performance at 4.2 K was degraded by using
By numerical simulation, we also examined the temperature distribution in the second-
stage regenerator and the temperature oscillation in the second-stage expansion space. The mecha-
nism that makes effective for improving the cooling performance at 4.2 K in the 4 K
pulse tube cooler remains unclear. More detailed analysis is needed to clarify this mechanism.
INTRODUCTION
We previously developed a single-stage Gifford-McMahon (GM) pulse tube cryocooler that
uses a double-inlet configuration.1 For a 1.5 kW compressor unit, the cryocooler achieves a
minimum temperature of 25.7 K and a maximum cooling capacity of 20.0 W at 80 K. We also
evaluated the dependency of the operating frequency on the inclination of the pulse tube. We
measured the cooling performance by turning the cooler upside down, and found that the effect of
the pulse tube inclination on the cooling performance at 80 K is negligible when the operating
frequency is higher than 6 Hz.
We also previously developed a two-stage pulse tube cryocooler that also uses the double-inlet
configuration.2 The rated input power of the compressor unit is 3.3 kW at 50 Hz. When the
second-stage regenerator is filled with lead spheres, the cryocooler achieves a minimum tempera-
tures of 33 K at the first stage and 5 K at the second stage, and a maximum cooling capacity of
22.9 W at 80 K at the first stage and 5.9 W at 20 K at the second stage. Preliminary measurements
using the same compressor unit showed that the two-stage cooler reached a minimum temperature
of 2.89 K and provides 170 mW at 4.2 K when conventional magnetic regenerator materials,
and are used in the second-stage regenerator.
In this study, to increase the efficiency of this two-stage cooler, we used a new oxide magnetic
regenerator material, in the lowest temperature part of the second-stage regenerator. The
significantly improved the cooling capacity both at 4.2 K and below 4.2 K. We also
placed in a second-stage regenerator of a 4 K GM cooler and determined its effect on
cooling performance. Furthermore, by using numerical simulation, we compared the effect of
in the 4 K pulse tube cooler with that in the 4 K GM cooler.
EXPERIMENTAL APPARATUS AND PROCEDURE
Two-stage 4K Pulse Tube Cryocooler

Figure 1 shows a schematic of our two-stage pulse tube cryocooler. The phase shifters for first
and second stages are double-inlet configurations. Two reservoir tanks, each with a volume of 500
are connected to the hot ends of the first and the second stage pulse tubes through needle
valves, and connected to the suction line of the compressor unit through other needle valves. The
DC fluid current, flowing from the cold end of the pulse tube to the hot end, is controlled by using
these needle valves. They also permitted control of the cooling performance. The rated input
power of the compressor unit is 3.3 kW at 50 Hz.
The first-stage regenerator was filled with about 1100 disks of 200-mesh phosphor bronze
screen. The volume of the second-stage regenerator was about The temperatures of the
first and second stage cold heads were measured by using a calibrated Pt-Co resistance thermom-
eter and a calibrated Germanium resistance thermometer, respectively. The cycle frequency was
2.0 Hz and the initial pressure was 2.1 MPa.
The regenerator cylinder can be separated into first-stage and second-stage cylinders, which
are connected by their flanges and sealed by a metallic O-ring. Thus, the regenerator materials in
the second-stage regenerator can be easily changed. First, we used a combination of lead spheres
and conventional magnetic regenerator material, and in the second-stage regenera-
tor. The materials were separated from each other by some phosphor bronze screens in the regen-
erator cylinder; lead spheres were placed in the upper half, and and were placed
each in one-fourth of the lower half. To improve the cooling capacity at 4.2 K and the terminal
temperature, we replaced pan of with a new oxide magnetic regenerator material,
in the lowest temperature part (Figure 2a).
The was developed at the National Research Institute for Metals to be used in cryo-
coolers that are designed for 4.2 K or less.3 can hold the high entropy below 4 K, has a
magnetic transition temperature of about 3.8 K, and has a heat capacity considerably higher than
that of and below 4 K (Figure 3).
EXPERIMENTAL STUDY OF A 4 K PULSE TUBE CRYOCOOLER 215
Two-stage 4K GM Cooler
To evaluate the effect of in a 4K GM cryocooler, we used a two-stage 4K GM
cryocooler (DAIKIN prototype cooler). is used in the lower half part of the second-stage
regenerator (Figure 2b). Using the same procedure as in the 4K pulse tube cooler, we varied the
volumetric ratio of by replacing part of with The rated input power of
the compressor unit is 6.5kW.
Figure 1. Schematic of two-stage cooler.
Figure 2. Schematic of second-stage regenerator. Figure 3. Specific heat of

NUMERICAL ANALYSIS METHOD

To evaluate our experimental results and to improve the cryocooler performance, based on a
recent numerical simulation of a pneumatic-driven GM cryocooler4,5 we used numerical
simulation of a two-stage 4K pulse tube cooler. Complete one-dimensional governing equations
are solved by using finite difference method. Conservation of mass, momentum and energy of
helium gas are written as follows;
Conservation of energy of helium gas and the regenerator matrix are written as follows;
Temperature of helium gas can be determined from the internal energy and density as follows;
The equation of state of real gas can be written as;
In this simulation, variable physical properties of helium and regenerator materials were
considered.
The mass flow rate through the orifice valves was calculated by using the nozzle equation.
The numerical model also includes the compressor unit to estimate the influence of its pressure
oscillation as well as its performance (Figure 4).
Figure 4. Numerical simulation model.

EXPERIMENTAL STUDY OF A 4K PULSE TUBE CRYOCOOLER 217
EXPERIMENTAL RESULTS
Figure 5(a) shows the measured cooling performance of the 4K pulse tube cooler as a
function of volumetric ratio of When was not used in the second-stage
regenerator, the terminal temperature was 2.89K and the cooling capacity at 4.2K was 170mW.
The cooling performance was unproved by using When the volumetric ratio of
to that of the total and was 0.25, the terminal temperature was 2.51K
and the cooling capacity at 4.2K was 250mW. The experimental results show that the new oxide
regenerator material, significantly increased the cooling performance both at 4.2K and
below 4.2K in the 4K pulse tube cryocooler. By adjusting the pulse tube size for further
optimization, a terminal temperature of 2.42K and a cooling capacity at 4.2K of 288mW was
achieved.
Figure 5(b) shows the cooling performance of the 4K GM cooler as a function of the
volumetric ratio of The figure shows terminal temperatures both with heat load of 0.4
W and 0.8W on the second stage. The experimental results show that was not effective
in increasing the cooling performance at 4.2K, but was considerably effective at about 3.5K.
COMPARISON BETWEEN EXPERIMENTAL AND NUMERICAL RESULTS

The experimental results show that the effectiveness of the new oxide regenerator material,
in the 4K pulse tube differs from that in the 4K GM cryocooler. Using numerical
simulation, we determined the cooling capacity both for the 4K pulse tube cooler and the 4K GM
Figure 5. Measured Cooling Performance.
Figure 6. Calculated Cooling Performance.

cryocooler. Figure 6 show that is not effective for improving the cooling performance
of either cryocooler at 4.2K but is effective for improving the cooling performance of both
cryocoolers below 4K. For the 4K GM cryocooler, the numerical results agree with the
experimental results.
If the temperature oscillation in the expansion space is considerably large, the large specific
heat of can improve the cooling performance at 4.2K. Figure 7 shows that the
temperature oscillation is too small to improve the cooling performance by using the high
specific heat below 4K. Figure 8 shows the temperature distribution of the second-stage
regenerator both for the pulse tube cooler and the GM cryocooler. No remarkable difference is
evident in the temperature distribution between the 4K pulse tube cooler and the 4K GM cooler.
CONCLUSIONS
We placed a new oxide regenerator material, in the second-stage regenerator in a
two-stage 4K pulse tube cryocooler and in a 4K GM cryocooler. For the pulse tube cryocooler,
significantly improved the cooling performance at 4.2K. The cooling capacity was
increased from 170mW to 250mW and the terminal temperature was degraded from 2.89K to 2.5
1K. By adjusting the pulse tube size for further optimization, the cryocooler achieved a cooling
capacity at 4.2K of 288mW and a terminal temperature of 2.42K. For the 4K GM cryocooler,
was not effective in improving the cooling performance at 4.2K, but was considerably
effective at about 3.5K.
By using numerical simulation, we determined the temperature distribution in the second-
Figure 7. Calculated Temperature Oscillation of helium gas in 2nd-stage expansion space.
Figure 8. Calculated Temperature Distribution of 2nd-stage regenerator.

EXPERIMENTAL STUDY OF A 4 K PULSE TUBE CRYOCOOLER 219
stage regenerator and the temperature oscillation in the second-stage expansion space. The mecha-
nism that makes effective for improving the cooling performance at 4.2 K in the 4K pulse
tube cryocooler remains unclear. More detailed analysis is needed to clarify this mechanism.
REFERENCES
1. S. Fujimoto, Y.M. Kang, and Y. Matsubara, "Development of a 5 to 20W at 80K GM Pulse Tube
Cryocooler," Cryocoolers 10, R.G. Ross, Jr., ed., Kluwer Academic/Plenum Publishers, New York
(1999), p. 213.
2. S. Fujimoto, Y.M. Kang, T. Kanyama, and Y. Matsubara, "Experimental Investigation of Some
Phase Shifting Types on Two-stage GM Pulse Tube Cryocooler," CEC/ICMC' 99, to be published
3. T. Numazawa, O. Arai, A.Sato, S. Fujimoto, T. Oodo, and Y.M. Kang, T. Yanagitani, "New
Regenerator Material for Sub-4K Cryocoolers," Cryocoolers 11, R.G. Ross, Jr., ed., Kluwer Aca-
demic/Plenum Publishers, New York (2001).
4. T. Kurihara and S. Fujimoto, "Numerical Analysis of the Performance of Pneumatic Driven 4K-GM
Refrigerator," Cryogenic Engineering (in Japanese), Vol.31 (1996), p. 197.
5. T. Kurihara, M. Okamoto, K. Sakitani, H. Torii, and H. Morishita, "Numerical and Experimental
Study of a 4K Modified-Solvay Cycle Cryocooler," Adv. Cryog. Eng. 43 (1998), p. 1791.
Nomenclature
specific heat of matrix
T temperature
equivalent diameter
u internal energy
e total energy
v velocity
f coefficient of friction
w external PV work
acceleration of gravity
density
P pressure
equivalent thermal conductivity of regenerator
q heat exchange between regenerator and cylinder wall
t time
Subscripts
g gas
m matrix
GM-Type Two-Stage Pulse Tube Cooler
with High Efficiency
A. Hofmann*, H. Pan**, L. Oellrich**

*Forschungszentrum Karlsruhe Institut für Technische Physik
D-76021 Karlsruhe, Germany
**Univ. Karlsruhe, Institut für Technische Thermodynamik und Kältetechnik
D-76128 Karlsruhe, Germany
ABSTRACT
A two-stage pulse tube refrigerator has been designed for maximum refrigeration powers at 20 K
and 50 K, when powered by a 6.5 kW of electric compressor. The modular setup of the cold head
enables easy access to all components to be modified for the optimization of the system. All gas
flows at the regenerator and at both pulse tubes are controlled by solenoid valves. Additional ad-
justment of the flow is done by throttling valves at the pulse tubes.
Two arrangements with different sizes of the second stage have been tested. With the small size
second stage, the typical cooling power achieved was 55 W at 50 K for the first stage together with
3.5 W at 20 K. This corresponds to about 5.0 % of summarized Carnot efficiency. Even higher
efficiency of 5.8 % Carnot was obtained for the system with enlarged second-stage components
operated with 40 W at 46 K plus 10.5 W at 20 K. No-load temperatures down to 8.5 K were
achieved with lead spheres for the second-stage regenerator. Some details on the design of the test
rig and on operational parameters are given.
In addition, the results are compared with numeric predictions based on a small amplitude
thermoacoustic model.
INTRODUCTION
In the early days, pulse tube coolers were considered most attractive because of their simple
design, low noise, and projected high reliability; it was believed that lower efficiency than that of a
comparable Stirling or GM cooler had to be accepted. However, now the pulse tube process is
understood as a Stirling process. With this understanding, there are no physical reasons for having
lower efficiency with a properly designed pulse tube cooler. In the past few years this has been
verified for compact single-stage Stirling-type coolers1 and also for GM-type pulse tube systems.2
With multi-stage GM-type PTRs even temperatures in the range of 2 K are being achieved.3,4 Such
temperatures have not yet been realised with conventional GM coolers. These examples show that
pulse tube coolers may be substituted for many types of conventional Stirling and GM coolers. In
the present paper, research is focused on a two-stage GM-type pulse tube cooler for operation with
a 6.5 kW input power compressor.

EXPERIMENTAL SETUP
The experimental set-up, shown schematically in Fig. 1, was designed to enable manifold
modifications. A commercial product would have required a lot of simplifications. The cooler is
operated with a conventional compressor (Coolpak 6000®) equipped with additional buffer volumes
in the suction and exhaust lines. The time-averaged flow rate is measured with a mass flow meter
(FM). The cold head has been designed for having refrigeration power at two levels, around 50 K
and around 20 K. All components are flanged. Pressures are measured at the warm end of both the
regenerator and the pulse tubes. Temperature sensors are attached on the outer surfaces of regenerators
and pulse tubes, and in addition, at both cooling stages (copper blocks with ohmic heaters).
The first-stage regenerator is filled with stainless steel mesh. Easy variation of the active length
is made possible by use of a solid plug at the upper end. The second-stage regenerator is filled with
Pb shot. Different tubes with diameters up to about 25 mm and lengths up to 200 mm can be
connected. The uncommon configuration with an inline arrangement of the first-stage pulse tube
and the second-stage regenerator has no physical reason. It was chosen to allow better adjustment
of the components. Water-cooled heat exchangers are located on the warm ends of both pulse
tubes. All gas flow in the regenerator and both pulse tubes is controlled by solenoid valves. Additional
adjustment of the flow is done by two pairs of throttling valves at the pulse tubes. The solenoid
valves can be actuated individually. The typical timing for 2 Hz operational frequency is shown in
the insert in Fig. 1. However, each configuration requires some re-adjustment.
RESULTS
Continuous improvement of the system has been achieved through manifold modification of the
hardware components together with tuning of the valve timing and needle valve settings, accompanied
by recalculations. The first step was aimed at getting the most power at the 50 K level, together
Figure 1. Schematic of the 2-stage PTR.

GM-TYPE 2-STAGE PT COOLER WITH HIGH EFFICIENCY 223
Figure 2. Refrigeration power and temperatures achieved with Coolpak® compressor (6.5kW) and
different second stage configurations.
with rather modest power at 20 K. This result is shown in Fig. 2a), where the temperatures and
are plotted together with the heater powers applied to those stages. The valves where adjusted
for getting optimum conditions for a heat load of 3 W at the second stage together with 50 W at the
first stage (circled points). The respective temperatures are 18 K and 48 K. No further adjustment
was done during the variation of the heater powers.
In a second step, the size of the second-stage components was increased to achieve more cooling
power at the second stage. Some degradation of the first-stage refrigeration power was expected.
The result is shown in Fig. 2b. Here, the system was adjusted with heat loads of 40 W and 10 W at
the first and second stages, respectively. The no-load temperatures went down to 26 K and 8.7 K.
Somewhat lower temperatures could have been achieved by readjustment at no-load.
This refrigeration chart is very close to that of commercial GM coolers designed for the same
operational range. For having a better defined characterisation of the efficiency, we calculate the
overall Carnot efficiency according to
Our respective results together with some literature data of other systems are compiled in Table 1.
In the best case, the reference point in Fig. 2b, we get an efficiency of 5.8 % Carnot. This is close to
data evaluated for commercial GM refrigerators7 operated with the same type of compressor; also
Figure 3. a) Cool-down with nominal load and b) temperature stability.
the more advanced laboratory-type 4K GM coolers5,6 are in the same range. Appreciably higher
efficiency values have been obtained for single-stage pulse tube coolers of the Stirling-type1 and of
a GM-type with active buffer phase shifter.9
Our pulse tube cooler operated with six solenoid valves and four additional needle valves has a
great number of parameters to be adjusted for tuning. One important criterion is the compensation
of the dc flow which can exist in different loops. Direct observation of the dc flow is not possible in
our system. Instead, the change in temperature distribution detected at the different components
gives valuable advice for handling. If, for instance after modification of one needle valve, the
temperature in the regenerator is increased and decreased in one pulse tube, it may be argued that a
dc flow from the regenerator to the pulse tube has been increased. Much experience is required for
the tuning. But after documentation of all adjustment parameters, excellent reproducibility of the
system is obtained even after dismantling and re-assembling of components. A typical cool-down
curve with heat load applied from the beginning and the temperature stability during a 8-h run is
shown in Fig. 3. Steady state operation is established after some 90 min, thereafter temperature
fluctuations are less than 1 K at both stages.
It has also been shown, that such pulse tube refrigerators can rather easily be adapted to specific
requirements with different distributions of the thermal loads at both stages. Moreover, a powerful
single-stage refrigerator is obtained just by disconnection of the second stage. The respective result
is shown in Fig. 4. Typically, a refrigeration power of 120 W is obtained at 70 K. But it should be
mentioned, that the same first-stage components as they have been optimised for the two-stage
operation have been used. Higher efficiency might be obtained by additional modification of those
components.
Figure 4. Refrigeration power of the first stage (second stage disconnected).

NUMERIC MODELING
Although the performance of pulse tube systems is such that they can compete with conventional
GM refrigerators, their absolute efficiency is still rather small. There is certainly some margin for
further improvement, but the question is where to concentrate the needed effort. With this objective,
we have tried to analyse pulse tube systems using numeric models. Our code8 is based on a small
amplitude thermoacoustic approximation, on the assumption of laminar flow in all components,
and on the approximation of the regenerators by equivalent parallel channel structures. This rather
crude model yields quite a good prediction of the refrigeration powers and of the temperature
distribution in both regenerators and in the pulse tubes. On that basis, the model is used for calculating
the work flow and the losses at different positions of the system.
For evaluating the different contributions in the efficiency term, it is more appropriate to consider
the term
where the efficiency is not related to Carnot, with as it is for Eq.(1), but to
which is the ideal pulse tube efficiency for all single-stage configurations where no expansion work
rejected from the pulse tube is recovered.10 With the work flow scaling and with the
flow rate fraction
this can be transformed into
Where are the effective refrigeration powers of the first and second stage, respectively, and
are the work flows (expansion powers) at the cold end of the pulse tubes. The mass flow
fraction in PT 1 is 0.6 for the present simulation, and is the work flow at the hot end of the
first regenerator.
Those results are plotted in Fig. 5. The work flow (“pV-power”) at the inlet of the first regenerator
is about 1 kW. This is supplied by the compressor with 6.5 kW of electric input power measured
for typical operational conditions. Hence, the efficiency of such a valved pressure wave generator
is in the range of This low compressor efficiency, a drawback ofthe GM operation, is
certainly the greatest loss term, but it is about the same for conventional GM coolers.
Next, we consider the first-stage regenerator. The expansion power (pV-work flow) in the first
pulse tube is 91 W. It is balanced by 44 W of enthalpy flow in the first-stage regenerator (regenerator
loss) and by 12 W of heat flow coming down through the first pulse tube (“shuttle loss”). Hence the
residual effective calculated refrigeration power is 35 W at 45 K, a value close to the measurement.
This corresponds to a first-stage efficiency with
In the second stage, the expansion power is 25 W, the regenerator loss is 5 W, and the pulse tube
loss amounts to 8 W. This gives a residual refrigeration power of 12 W at 20 K, and the respective
efficiency becomes The calculated mass flow is 60 % in the first and 40 % in the second
pulse tube. Hence, the overall scaled efficiency is a value
close to the measured Carnot efficiency. The term within the brackets may be considered as the
overall efficiency of the cold head. It amounts to 0.42. An overall efficiency in this range can only
be approached by use of a very efficient linear drive compressor, which, however, may be too bulky
and expensive for low-frequency operation, which is 2 Hz in the present case.
For a more detailed evaluation, we consider the ratio of losses to the work flow. This results in
0.29 for the first regenerator, 0.13 for the first pulse tube, 0.2 for the second regenerator, and the
greatest value of 0.32 for the second pulse tube. Hence, apart from the required improvement of the
compressor, the most effort should be spent to improve the first regenerator and the second pulse
Figure 5. Calculated work and heat flows.
tube. Those calculations, however, are based on some arbitrary assumptions such as longitudinal
thermal conductivity being the same as for the bulk material, and laminar flow in the pulse tubes.
Hence, the regenerator loss might be overestimated. But the pulse tube loss is more likely
underestimated. The reduction of those losses is no trivial problem.
Another interesting term is the expansion power removed from the pulse tubes. This is 87 W at
pulse tube 1 and 22 W for the second tube. Both terms are much smaller than the power fed into the
regenerator. Recovering the expansion power by use of piston expanders would only yield a small
improvement (in the 10 percent range) of the overall efficiency. No expansion work is recovered in
the present valved system.
CONCLUSIONS
Stable and reproducible operation is obtained with a GM-type two-stage pulse tube cooler operated
with 3 pairs of solenoid valves for controlling the flow at the warm ends of the regenerator and of
both pulse tubes. The refrigeration powers at the 50 K and at the 20 K stages can be redistributed
just by modification of the second-stage components. The overall efficiency of two-stage pulse
tube coolers is shown to be very close to that of conventional GM-coolers. But the absolute value is
still much smaller than the ideal Carnot efficiency. Most of this is caused by the valved pressure
wave generator, which is estimated to have only 15 % efficiency. The cold head itself is analysed to
have a rather good performance with more than 40% of Carnot efficiency. The computer model,
based on small amplitude approximation and on the assumption of having laminar harmonic flow
in all components, gives some hints for localising the predominant loss terms. But this model should
not yet be stressed for quantitative evaluations.
REFERENCES
1. W.W. Burt, and C.K. Chan, “New Mid-Size High Efficiency Pulse Tube Coolers,”
Cryocoolers 9, Plenum Publishing Corp., New York, 1997, pp. 173-182.
2. A. Ravex, I. Charles, L. Duband and J.M. Poncet, “Pulse Tube Development at CEA/SBT,” Proc. of
the IIR Conf., Sidney, Sept. 1999.
3. C. Wang, G. Thummes, and C. Heiden, “Experimental Study of Staging Method for Two-Stage
Pulse Tube Refrigerators for Liquid Helium Temperatures,” Cryogenics, 37 (1997), pp. 159-164.
4. M.Y. Xu, A.T.A.M. De Waele, Y.L. Ju, “A pulse tube refrigerator below 2 K,” Cryogenics, 39
(1999), pp. 865-869.
5. A. Onishi, “4K-GM Cryocoolers having little orientation dependency and small influence from mag-
netic field,” Cryogenic Engineering (J. Cryog. Eng. Soc. Japan, Tokyo), Vol. 34 (1999), pp. 233-235.
6. J.N. Chafe and G,F. Green:, “Performance of a Low Temperature Giffbrd McMahon Refrigerator
Utilizing Neodymium Disk Regenerator,” Advances in Cryogenic Engineering, Vol. 43 (1998),
pp. 1783-1790.
7. Leybold-Katalog Vakuumtechnik (1998).
8. A. Hofmann and S. Wild, “Analysis of a Two-Stage Pulse Tube Cooler by Modeling with
Thermoacoustic Theory,” Cryocoolers 10, Plenum Publishing Corp., New York, 1999, pp. 369-377.
9. S. Zhu, Y. Kakimi and Y. Matsubara, “Investigation of active-buffer pulse tube refrigerator,” Cryo-
genics, 37(1997), p. 461.
10. P. Kittel, A. Kashani, J.M. Lee and P.R. Roach, “General pulse tube theory,” Cryogenics, 36 (1996)
pp. 849-857.
Developments on Single and Double Stage
GM Type Pulse Tube Cryorefrigerators
J.M. Poncet, A. Ravex* and I. Charles

Service Basses Températures, CEA-DRFMC, 17, rue des Martyrs
38054 Grenoble Cédex 9, France
*present address : Air Liquide, DTA, BP 15
38360 Sassenage, France
ABSTRACT
Pulse tube (PT) cryocoolers with no moving parts in their cold ends are potentially able to
replace conventional Gifford MacMahon (GM) cryocoolers for most applications. Over the past
several years, CEA/SBT has undertaken basic research and prototype development of GM driven
PT cryocoolers. Systematic studies have been carried out to characterize flow distribution and
d.c. flow effects. Based on a good comprehension of these phenomena, several prototypes have
been built and optimized for various customised applications. Their performance is presented in
this paper.
1. INTRODUCTION
Several years ago, CEA/SBT initiated basic research on PT cryocoolers to understand and
analyse their operation. A model has been proposed1 and a numerical tool has been developed for
the design and optimisation of prototypes.
As suggested by Zhou2, a double inlet configuration has been systematically used for
performance improvement. It is generally stated that this configuration with a gas supply to the
tube at both the cold and warm ends reduces the mass mass flow rate through the regenerator, thus
increasing its efficiency. In fact, the major interest of the double inlet configuration is the achievement
of a better phase shift between the pressure versus mass flow oscillations at the cold end of the tube.
Gedeon3, through theoretical calculations, has predicted that the double inlet configuration may
allow for d.c. flow through the tube and regenerator that is able to produce a parasitic heat load of
the order of the PT gross cooling power. This effect has been observed4 at CEA/SBT during Stirling
type PT developments. Experimental evidence was also observed by Chen5 during the development
of a double stage 4 K GM type PT. He observed a large improvement in the performance of their
prototype by fine tuning of a so-called minor orifice (needle valve) set between the buffer and the
compressor suction line. In fact, a controlled d.c. flow is created at room temperature through this
connection which cancels the parasitic d.c. flow in the cold region. This artifact is now well understood
and commonly used.6,7
Based on a good comprehension and control of this d.c. flow, and on extensive studies on
distribution valve design and phasing control, several prototypes have been designed, optimised
and characterized for various applications.

Figure 1. SingleStage Pulse Tube (100 W at 80 K).
2. SINGLE STAGE PROTOTYPES

At the 1997 CEC in Portland, we have reported 8 results on a single stage GM type prototype
with the following characteristics :
Compressor : CTI8500 / 5 kW
Operation frequency : 2 Hz
Ultimate temperature : 26 K
Cooling power : 80 K /100 W
Configuration : U shaped
A picture of this prototype is given on Figure 1. Based on this first development, several
prototypes have been extrapolated for various applications. Performances and main features of
these coolers are reported in Table 1.
The main evolutionary developements from the initial cooler have been required by specific
applications. For example, to reduce the vibration exported to the devices to be cooled, the
distribution valve has been systematically disconnected from the cold head with a flexible
connection line up to 1 meter long.
DEVELOPMENTS ON 1 - AND 2-STAGE GM TYPE PT COOLERS 231
A well known drawback of the low frequency operated GM type PT is its performance
sensitivity to the orientation of the cold finger.8 To avoid this inconvenience, operation at 5 Hz has
been demonstrated with a large reduction in the orientation effect.
For HTS applications or single thermal shield helium cryostats, large cooling power in the
30K/50K temperature range is required. To achieve this goal, the initial prototype has been modified
by introducing lead shot in the regenerator, with the following characteristics :
Compressor : Leybold Coolpak / 6 kW
Operation frequency : 2 Hz
Ultimate temperature : 17 K
Cooling power : 30 W at 30 K or 70 W at 50 K
Configuration : U shaped
Further developments are in progress on single stage PTs to develop a full range of coaxial
type cold heads – which may allow for an easier integration – to be associated with 1.5 kW, 3 kW
and 6 kW standard GM compressor units.
3. DOUBLE STAGE PROTOTYPES

The common applications for double stage GM coolers are cryopumping and MRI cryostat
thermal shield cooling. For these applications, large cooling power at bom 80 K and 15 K temperature
levels is required. With the goal to demonstrate that PT coolers are able to achieve these requirements,
SBT/CEA has developed two GM type prototypes with overall sizes (length from 300 K flange to
second stage heat station : 325 mm, overall diameter : 95 mm for the largest prototype) equivalent
to existing GM coolers. A picture of such a double stage GM type PT prototype on a test bench is
presented in Figure 2. The performance of both prototypes, established with two types of compressors,
is reported in Table 2 and in a traditional map presentation in Figures 3 and 4. It can be observed
that the obtained performance is comparable to that of the corresponding GM coolers.
Figure 2. Double stage PT prototype on a test bench.

Figure 3
Figure 4
DEVELOPMENTS ON 1 - AND 2-STAGE GM TYPE PT COOLERS 233
The prototypes were optimized for operation at respectively 80 K and 15 K on the two stages,
thus using stainless-steel gauze discs (210 mesh) in the first stage regenerator and lead shot (200
in the second stage regenerator.
Without any geometrical modification for optimisation, these prototypes have been also operated
in the liquid helium temperature range, introducing magnetic material in the second stage regenerator.
The results obtained are summarised in Table 3. These results are encouraging for operation at LHe
temperatures and may be further improved by optimisation of the second stage regenerator and tube
geometries.
4. CONCLUSION
The performance obtained with the various single and double stage GM-type PT coolers
developed at CEA/SBT demonstrates that the new PT technology, with its potential advantages of
reliability and ease of integration, is mature for most applications.
A collaboration with industrial partners is now necessary to introduce commercial PT coolers
into the traditionnal markets for GM coolers.
REFERENCES
1. Liang, J., Ravex, A. and Rolland, P., “Study on pulse tube refrigerator Part 1 : Thermodynamic non
symmetry effect Part 2 : Theoretical modelling Part 3 : Experimental verification”, Cryogenics, vol
36 (1996), pp. 87-106.
2. S. Zhu, P. Wu and Z. Chen, “Double inlet pulse tube refrigerators: an important improvement,”
Cryogenics, vol. 30 (1990), pp. 514.
3. D. Gedeon, “DC gas flows in Stirling and pulse tube cryocoolers,” Cryocoolers 9, R.G. Ross Jr., ed.
Plenum Press, New York (1997), pp. 385.
4. L. Duband, I. Charles, A. Ravex, L. Miquet and C. Jewell, “Experimental results on inertance and
permanent flow in pulse tube coolers,” Cryocoolers 10, R.G. Rosw Jr., ed., Plenum Press, New York
(1999), pp. 281-290.
5. G. Chen, J. Zheng, L. Qiu, X. Bai, Z. Gan, P. Yan, J. Yu, T. Jin and Z. Hang, “Modification test of
staged pulse tube refrigerator for temperatures below 4 K,” Cryogenics, vol. 37 (1997), pp. 529.
6. C. Wang, G. Thummus and C. Heiden, “Control of DC gas flow in a single stage double inlet pulse
tube cooler,” Cryogenics, vol. 38 (1998), pp. 843.
7. I. Charles, L. Duband and A. Ravex, “Permanent flow in low and high frequency pulse tube coolers
– experimental results,” Cryogenics, vol. 39 (1999), pp. 777.
8. A. Ravex, J.M. Poncet, I. Charles and P. Bleuzé, “Development of low frequency pulse tube refrig-
erators,” Advances in Cryogenic Engineering, Plenum Press, New York, vol. 43A (1998), pp. 1957.
30 - 50 K Single Stage Pulse Tube Refrigerator
for HTS Applications
J. Yuan, J. Maguire, A. Sidi-Yekhlef, and P. Winn

American Superconductor Co.
Westborough, MA 01581 U.S.A.
ABSTRACT
The need for reliable cryocoolers for High Temperature Superconducting (HTS)
applications ranging from 30 to 50 K has become apparent in the past several years. Many
cryocooler designs are under development, including GM, Stirling, and pulse tube. Pulse tube
refrigerators have attracted extensive interest in recent years due to their high potential for
reliability and simplicity. This paper describes the development program of a 30 – 50 K single
stage pulse tube refrigerator for use in a HTS system at American Superconductor Co. Initial
tests indicate that the cooler can reach a minimum temperature below 20 K and has a cooling
capacity of 20 watts at 30 K, and 60 watts at 50 K, with an input power of 6 kW. Systematic
investigation of the effects of valve timing, operating frequency and compressor input power on
the cooler performance is presented in the paper. Additionally, the stability issue of the pulse
tube refrigerator is addressed. The progress towards development of a successful 30 – 50 K pulse
tube and its potential and limitations are also discussed.
INTRODUCTION
In the last decade, the significant advance of the performance levels of high temperature
superconducting (HTS) wire has made it suitable for commercially viable applications such as
synchronous motors, generators, transformers and electric power cables. Currently, to provide
the requisite current density, the HTS wire formed from BSCCO and YBCO must be operated at
temperatures around 30 K and 50 K, respectively.1 Cryogenic refrigeration systems are one of
the most crucial components for the successful commercialization of HTS applications. In most
small scale HTS applications, such as motors and generators, cryocoolers will be the most
attractive candidates due to their size , cost and reliability. Over the past two decades, since the
discovery of HTS materials, the need for reliable cryocoolers for HTS applications ranging from
30 to 50 K has become apparent. Many cryocooler designs are under development, including
GM, Stirling and pulse tube. Pulse tube refrigerators have attracted extensive interest in recent
years due to their potential for reliability and simplicity. However, few attempts have been made
to develop a single stage pulse tube refrigerator with relatively large cooling capacity at
temperatures ranging from 30 K to 50 K.
The present paper describes efforts that have been made to develop a high power 30 to 50 K
pulse tube refrigerator for potential HTS applications. Initial tests indicate that the cooler can
reach a minimum temperature below 20 K and has a cooling capacity of 20 watts at 30 K, and 60

watts at 50 K, with an input power of 6 kW. Systematic investigation of the effects of valve
timing, operating frequency, and compressor input power on the system performance are
presented in the paper. Additionally, the stability issue of the pulse tube refrigerator is addressed.
The progress towards development of a successful 30 – 50 K pulse tube, its potential and
limitations, are discussed.
DESCRIPTION OF COOLER AND TEST APPARATUS

The pulse tube cooler described in this paper is designed for temperatures ranging from 30 K
to 50 K applications. A thermodynamic model2 was used to size the pulse tube, buffer and
regenerator. The U-shape configuration is utilized for the design. The cooler flow diagram and
complete package are illustrated in Figure 1. As shown in Figure 1, the cooler utilizes a
modified active valve configuration3 that allows control of the phase shift at the warm end of the
pulse tube to optimize the cooler performance. Furthermore, this configuration allows control of
the DC flow by adjusting the opening and timing of valves 3 and 4. Finally, from a practical
point of view, this configuration makes it possible to design using a rotary valve.
The pulse tube is fabricated from a thin-wall stainless steel tube with a 45 mm outside
diameter and 0.5 mm wall thickness. The tube is 270 mm long which yields a total pulse tube
volume about 0.4 liter. To optimize the performance of the cooler at temperature ranging from
30 K to 50 K, a three layer regenerator containing different bronze mesh sizes at the warm end
and fine lead shots at cold end has been designed and constructed.
In the design process, particular attention has been paid to the cold end heat exchanger.
Optimizing the design of the cold end heat exchanger requires balance between the dead volume,
total heat transfer area, and pressure drop. Detailed calculations indicate that the heat exchanger
efficiency can be dramatically increased by carefully arranging the gas flow channels. The cold
end heat exchanger has been optimized to have a high heat transfer coefficient, large heat
transfer area and a small pressure drop and dead volume. Although a water-cooled warm end
heat exchanger is used for the current testing, the design of a helium-cooled warm end heat
exchanger is underway; this will be a part of the complete package.
For test purposes, a valve system comprising solenoid and throttle valves was used to
control the gas flow in and out of the refrigeration system. The flow coefficients of the two
throttle valves connected in series with valves 3 and 4 can be adjusted from zero to 0.5 while the
flow coefficient of throttle valve 5 can vary from zero to 0.15. Timing of the solenoid valves has
Figure 1. Pulse tube cryocooler: (a) complete packge (b) flow diagram.
30 - 50 K SINGLE STAGE PT COOLER FOR HTS APPLICATIONS 237
been accomplished through the use of a controller. Different valve timings are achieved easily by
programming. The design of the rotary valve similar to the one used in a GM cryocooler is
underway which will replace the solenoid valve system. The valve timing of the rotary valve is
fixed by the hole configuration machined into the valve plates. Nevertheless, the following test
results were obtained utilizing the solenoid valve system.
Five separate piezoresistive pressure transducers, which allow measurement of pressure
oscillations, are mounted on the warm ends of the pulse tube and regenerator, on the discharge
and suction lines of the compressor, and on the buffer, respectively. The cold head temperature
is measured by two silicon diodes. Four cartridge heaters which can supply more than 150 watts
are placed at the cold end heat block to measure the cooling capacity. In order to reduce the
thermal radiation loss at low temperatures, more than 15 layers of aluminized Mylar are wrapped
around the cold parts. A Leybold RW6000 compressor with an input power of 6 kW was used in
most of the experiments. For comparison purpose, a Leybold CoolPak 6000 with an input power
of 7 kW was also used in some of the tests. The working medium used in all tests was helium
gas.
WORKING PROCESS
Theoretical understanding of the pulse tube refrigerator is based on the analysis of the
thermodynamic cycle of the gas column within the pulse tube2. The refrigeration mechanism of
a pulse tube refrigerator basically results from gas expansion at the cold end of the pulse tube,
which is similar to that in a GM refrigerator. However, in the case of the pulse tube refrigerator,
the expansion work is transferred by a “gas piston” from the cold end to the warm end of the
pulse tube instead of a solid displacer as is in the case of GM cryocooler. The system cooling
capacity is determined by the expansion work at the cold end of the pulse tube. To maximize the
expansion work, careful control of the valve timing is necessary.
A typical valve timing for the cooler is illustrated in Figure 2. The solid lines represent the
time that valve is open. To better describe the thermodynamic cycle of the cooler, the time
dependent pressure waves within the pulse tube and the buffer obtained by the theoretical
simulation are presented in Figure 3 (a). The corresponding start points are also shown in the
same graph.
Both the compression and expansion processes can be divided into three steps as can be seen
in Figure 2 and Figure 3 (a). The entire working process includes a) (1-2 )gas is admitted to the
pulse tube from buffer, b) (2-3) gas flows into the pulse tube from compressor, c) (3-4) gas
flows from compressor into buffer through the regenerator and the pulse tube, d) (4-5) gas is
exhausted from the pulse tube to the buffer, e) (5-6) gas flows out from the pulse tube to the
compressor and f) (6-1) gas flows from the buffer to the compressor through the pulse tube and
regenerator.
Figure 2. A typical valve timing.

Figure 3. Pressure oscillation obtained by (a) theoretical simulation (b) experiments.
PERFORMANCES AND DISCUSSIONS

Cooldown
System cooldown characteristics have been first studied based on the optimized valve
timing. Figure 4 displays a typical cooldown curve with operating frequency of 1.5 Hz. As can
be seen, the cooldown process takes about 40 min. The minimum no load temperature achieved
to date is 19.8 K with an operating frequency of 1.5 Hz and a pressure ratio of 1.9.
The pressure oscillations within the pulse tube, regenerator, buffer and compressor gathered
during the experiments are displayed in Figure 3 (b). The experimental results show that the
wave patterns are very similar to those obtained by the theoretical simulation as seen in Figure 3
(a). The pressure difference between the pulse tube and regenerator is due to the combined
effects of the packed bed and throttle valves. The effect of valve timing can be easily observed
from the slope changes of the pressure waves. One should notice that, in the theoretical
simulation, the pressure drop across the regenerator has been ignored.
Figure 4. Cooldown characteristics of the refrigeration system.

Figure 5. Influence of valve timing on cooling capacity

Influence of Valve Timing
To study the influence of valve timing on the system performance, different valve timings
have been tested with same compressor discharge and suction pressures. An operating frequency
of 1.67 Hz and fixed openings of throttle valves were employed for all the tests. As an example,
the cooling capacities as a function of the timing ratio of compression to expansion are displayed
in Figure 5. For all tests, the time ratios of each individual valve and remained the
same while the time ratio of compression to expansion varies. The time ratios of compression to
expansion utilized in the experiments were 270: 330, 300:300 and 330: 270.
Figure 5 illustrates the test results with three different compression to expansion time ratios.
As can be seen, the cooler performance strongly depends on the valve timing. The minimum
temperature achieved with different time ratio 270:330, 300:300 and 330:270 are 22.5 K, 24.4 K,
and 20.5K, respectively. The best cooling capacity at 30 K was obtained with the time ratio
270:300 while the minimum temperature was obtained with a time ratio of 330:270.
Figure 6. The influence of the operating frequency on cooling capacity.

Figure 7. Cooler stability testing data over 5 days
Influence of Operating Frequency

The cooling capacity of the cooler as a function of frequency is plotted in Figure 6. All the
data in this figure results from tests run with the same valve timing and throttle valve opening.
The time ratio of compression to expansion used in experiments was 300:270. As can be seen in
Figure 6, the operating frequency of the valves displays a noticeable influence on the
refrigeration capacity. In general, there is best operating frequency for a given pulse tube system
at a specific operating temperature. As shown in Figure 6, the best performance at 50 K was
obtained with an operating frequency of 1.5 Hz while the best performance at 77 K was obtained
with an operating frequency of 1.67 Hz.
Stability
A stability test was carried out over a period of about five days. During this period, a variety
of heat loads have been applied to the cold head. As shown in Figure 7, the temperature is quite
stable and remains constant for a given load. In general, the temperature variation is less than +/-
0.5 K. The experimental results revealed that the higher the operating frequency, the larger the
temperature variations. The reason for this may be partly related to the performance of the
solenoid valves since the higher the operating frequency, the shorter the valve operating time.
Figure 8. Effects of compressor input power on cooling capacity

Influence of Compressor Input Power

To study the influence of the compressor input power on system performance, a Leybold
CoolPak 6000 compressor with input power of 7 kW was used to drive the pulse tube cooler.
Figure 8 shows the test results. In both tests, operating parameters were same except for the input
power. As can be seen, even though the minimum temperature for both cases are almost same,
the cooling capacity with CoolPak 6000 is about 16% larger than that with RW6000 for a given
temperature.
CONCLUSIONS
A single stage 30 K – 50 K pulse tube refrigerator has been designed, built and tested based
on a thermodynamic model. It has been found that the single stage pulse tube has the potential to
supply high cooling power at temperatures ranging from 30K to 50 K. Initial test results are
encouraging for future development. In the initial tests, minimum temperatures below 20 K and
60 watts at 50 K were reached with an input power of 6 kW. Typical cooler performance
demonstrated a percentage Carnot efficiency of about 5% at 50 K. Systematic investigation
indicates that valve timing has significant effect on the system performance. Tests also show
that the cooler is relatively stable during the test period. Further work is needed to design the
matching compressor and improve the overall system efficiency.
ACKNOWLEDGEMENT
The work described here was supported by American Superconductor Co.. The authors
would like to thank Don Hannus, Jay Taber and other coworkers at American Superconductor
for their assistance and support with the many test setups.
REFERENCES
1. Malozemoff, A. P., Carter, W., Fleshler, S., Fritzemeier, L., Li, Q., Masur, L., Miles, P., Parker, D.,
Parrella, R., Podtburg, E., Riley Jr. G.N., Rupich, M., Scudiere, J., Zhang, W., “HTS Wire at
Commercial Performance Levels” IEEE Transactions on Applied Superconductivity 9, 1999, pp.2469
– 2473.
2. Yuan, J and Pfotenhauer, J. M. “Thermodynamic Analysis of Active Valve Pulse Tube Refrigerator”.
Cryogenics, Vol. 39. No. 4, pp. 283 – 292.
3. Yuan, J and Pfotenhauer, J. M. “A Single Stage Five Valve Pulse Tube Refrigerator Reaching 32 K”.
Advances in Cryogenic Engineering Vol. 43, pp. 1983 - 1989
Two-Stage 4 K Pulse Tube Refrigerator
Shaowei Zhu, Masahiro Ichikawa, Masafumi Nogawa,

and Tatsuo Inoue
Second Development Department
Aisin Seiki Co., Ltd.,
Kariya, Aichi, 448-8650 JAPAN
ABSTRACT
This paper describes the manufacture and test of a thermally connected two-stage 4 K pulse
tube refrigerator. The second stage of the refrigerator involved the use of either a “two middle-
buffer with double inlet” type phase shifter or a “two middle-buffer with displacer.” The phase
shifter of the first stage was a two middle-buffer type. The connecting tubes between the cold head
and the pressure switching valves were 1.5 meter long in order to reduce the vibration and the
electromagnetic noise near the cold head. The best cooling capacity, 0.58 W at 4.2 K, was achieved
with 5.5 kW of input power and the two middle-buffer with double inlet type phase shifter.
INTRODUCTION
A 4 K pulse tube refrigerator is expected to be one of the most important refrigerators for
either laboratory use or for medical instruments such as MRIs and SQUID systems; this is because
of its low noise, low vibration; and no need to remove the cold head during maintenance. In most
papers,1,2 a gas-connected structure is adopted for the double staging of the pulse tube refrigerator.
In this paper, a thermally connected two-stage pulse tube refrigerator that consists of two indepen-
dent refrigerators for each stage is manufactured and tested. An advantage of the thermally con-
nected structure is that each stage can be driven at different conditions to achieve optimization.
In order to decrease the noise level, 1.5-meter-long connecting tubes between the pressure
switching valves and the cold head were used. One problem caused by such long connecting tubes
was a rather large amount of work loss from the compressor due to the pressure drop and void
volume in the tubes. To achieve a higher efficiency, a “two middle-buffer” type phase shifter was
used for the first stage refrigerator. For the second stage, two types of the phase snifters were
tested: 1) a “two middle-buffer combined with double inlet” type, and 2) a “two middle-buffer
combined with displacer” type.
STRUCTURE OF THE TEST REFRIGERATOR

Figure 1 shows a schematic of the test machine. R11 and PT1 are considered a single refrig-
erator, referred to as the first stage refrigerator. R21, R22, and PT2 are also considered as another
single refrigerator, called the second stage refrigerator. The cold ends of the regenerators R11 and
R21 are thermally connected to each other.
In the regenerators R11 and R21, screen meshes of copper and stainless steel are stacked. Rare
earth materials of or shots are packed in the half volume of R22 at the cold side. The

Figure 1. Schematic of 4 K pulse tube refrigerator.
other half side of R22 is Pb shots. The connecting tube between the cold head and valves is 1.5 m.
Several separate electromagnetic valves are used for V11-V26 to allow easy changing of the on/off
timing for experiments. The compressor is a conventional G-M refrigerator compressor with an
input power capacity of 6 kW.
The phase shifter of the first stage is of the “two middle-buffer” type. A set of two buffers B11
and B12, of different intermediate pressure levels, are connected at the hot end of the first pulse
tube through on/off valves V13 and V14. V11 and V12 are the high and low pressure valves,
respectively, connected to the compressor.
The phase shifter of the second stage is of the “middle-buffer with double inlet” type. Another
set of two buffers of different intermediate pressure levels B21 and B22 are connected to the bypass
line through on/off valves V23 and V24. Valves V21 and V22 are the high and low pressure
valves, respectively, connecting the second stage to the compressor. The bypass line consists of
two needle valves N1 and N2 that adjust the direction and rate of DC gas flow in the bypass line.
Needle valve N3 with buffer B23 is mainly used to adjust the displacement of the gas at the cold
end of the pulse tube PT2. The tube between N1 and V21 or V22 has an orifice-like function.
The temperature of the first stage is measured by a PtCo temperature sensor that is mounted at
the cold end of the pulse tube PT1. The second stage temperature is measured by a silicon diode
that is mounted at the cold end of the second pulse tube PT2. Several pressure gauges are installed
on each buffer and the valve side of each connecting tube.
Compared to an ordinary two-stage pulse tube refrigerator, there are some advantages for the
thermally connected type pulse tube refrigerator. These include: 1) the influence between the first
and second stage is small, 2) each stage can be operated at a different pressure or frequency with a
different compressor, and 3) the mass flow rate to each of the two refrigerator stages can be consid-
ered as half that of an ordinary two-stage pulse tube refrigerator; thus the required valve opening
area is smaller. In order to get good performance, a compact and good heat exchanger between the
first stage and second stage refrigerator is needed; this is one of the disadvantages.
Another concept is to use a gas controlled displacer, as shown in Fig. 2, to replace the double
inlet. The movement of the displacer is controlled by valve V25, which is connected to the high-
pressure line of the compressor, and valve V26, which is connected to the low-pressure line of the
compressor. Because the displacer stops the DC gas flow, a needle valve N4 is connected between
buffer B22 and the low pressure line of the compressor to introduce a suitable rate of DC gas flow.
TWO-STAGE 4 K PULSE TUBE REFRIGERATOR 245
Figure 2. Schematic of 4 K pulse tube refrigerator with displacer.
The function of the displacer is the same as that of the bypass. It supplies gas to the hot end of the
pulse tube during the pressure-increasing process, and removes gas from there during the pressure-
decreasing process. The displacer can make it easier to control the shape of the PV diagram than
with the double inlet. This method is similar to that the displacer is on bypass.3
Figure 3 shows typical valve timing for the first stage refrigerator in Fig. 1. The valve timing of
the second stage in Fig. 1 is the same as that for the first stage. The dark bold lines represent the
period of time when the valve is open; the thin lines represent when the valve is closed.
Figure 4 shows typical valve timing for the second stage refrigerator in Fig. 2. The meaning of
the line weights is the same as for Fig. 3.
Figure 3. Valve timing of two middle-buffer pulse Figure 4. Valve timing of two middle-buffer
tube refrigerator. with displacer pulse tube refrigerator.
MIDDLE-BUFFER TYPE PULSE TUBE REFRIGERATOR

One or more buffers are connected to the hot end of the pulse tube through on/off valves. The
pressure in the buffers is at an intermediate or “middle” level, and it theoretically can not be the
high pressure or the low pressure of the compressor. Because the pressure in the buffers is at a
middle level, we call this type of pulse tube refrigerator a “middle-buffer pulse tube refrigerator.”
The middle-buffer pulse tube refrigerator can be considered as a modification of the active-buffer
pulse tube refrigerator,4 if the high pressure buffer and the low pressure buffer in the active-buffer
pulse tube refrigerator are eliminated.
The advantage of the middle-buffer type is that the theoretical mass flow rate is small com-
pared to the active-buffer type, if the same numbers of buffers are used. In the case of two buffers,
the theoretical mass flow rate of the middle-buffer type is about one third of that of the active-buffer
type. This is very important for high efficiency at low temperature.
In order to explain the working process of the middle-buffer pulse tube refrigerator, a numeri-
cal simulation was conducted of the first stage refrigerator of Fig. 1. Here the assumed pressure
ratio of the compressor was two. This is the minimum pressure ratio of our conventional GM
compressor. Figure 5 is a typical calculated pressure waveform, whereas Fig. 6 is a typical PV
diagram. Figure 5 shows that the pressures in buffers B11 and B12 are at a middle level. For
comparison, the PV diagrams of the active buffer with two buffers are shown in Fig. 6. It is clearly
visible that the displacement of the gas in the middle-buffer type is reduced to nearly one third of
the active buffer case. This means the mass flow rate required from the compressor is also de-
creased to one third. Another important point is that the PV diagrams of the middle buffer type at
the cold end and at the hot end of the pulse tube are completely separated. But in the case of the
active buffer, they are not separated. This means the shuttle loss in the middle-buffer pulse tube is
significantly reduced. Though the mass flow rate is significantly reduced, the calculated efficiency
is lower than that of the active-buffer type; this is because the upper left corner and right lower
corner of the PV diagram at the cold end of the pulse tube are cut off. Therefore the area of the PV
work is smaller.
In the middle buffer pulse tube refrigerator, the swept volume of the gas from the regenerator
to the cold end of the pulse tube depends on the void volume of the regenerator, the volume of the
pulse tube, the refrigeration temperature, and the pressure-ratio of the compressor. In the tempera-
ture range of lower than 10 K, the mass flow rate, which is proportional to the density and swept
volume of the gas through the regenerator, becomes too large due to the high density of the helium
gas. Therefore, a bypass or a displacer is necessary to decrease the mass flow rate. In this experi-
ment, a two-middle-buffer configuration, combined with either a double inlet or a displacer, was
adopted as the phase shifter for the second stage.
Figure 5. Pressure wave. Figure 6. PV diagram.

TWO-STAGE 4 K PULSE TUBE REFRIGERATOR 247
Figure 7. Pressure of the second stage Figure 8. Cooling capacity of the second
with two middle-buffer and 1. Double inlet with HoCu2
double inlet. 2. Double inlet with
3. Double inlet with HoCu2
EXPERIMENT RESULTS
In the experiment, needle valves N1, N2, N3, and valve timing of V11-V26 were adjusted to
get the highest cooling power.
Figure 7 is a typical pressure waveform of the second stage when the double inlet was used. It
is clearly shown that the pressures in the buffers are near a middle level. The pressure waveform of
the first stage is the same as in Fig. 7.
Figure 8 shows the cooling power versus the temperature of the second stage for the case of
or for regenerator R22, respectively. The operating frequency is 2 Hz. When
was used, the cooling power was 0.58 W at 4.2 K with an input power of about 5.5 kW. When
was used, only 0.4 W at 4.2 K was achieved with 6 kW of input power. When the displacer
was used, 0.45 W at 4.2 K was achieved with 5.5 kW of input power. In the case where was
used, the achieved temperature decreased to 2.35 K with a smaller opening of needle valve N3.
CONCLUSIONS
A thermally connected 4 K pulse tube refrigerator has been manufactured and tested. A “two
middle-buffer” type phase shifter was used for the first stage, and a “two middle-buffer with double
inlet” or displacer was used for the second stage. A cooling capacity of 0.58 W at 4.2 K was
achieved using a 1.5-meter-long connecting tube, 5.5 kW of input power, and the two middle-
buffer with double inlet type phase shifter.
REFERENCES
1. Wang, C., Thummes, G., and Heiden, C., “Performance Study on a Two-Stage 4 K Pulse Tube Refrig-
erator,” Advances in Cryogenic Engineering, Vol. 43 (1998), pp. 2055-2062.
2. Chen, G., Qiu, L., Zheng, J., Yan, P., Gan, Z., Bai, X., and Huang, Z., “Experimental Study on a
Double-Orifice Two-stage Pulse Tube Refrigerator,” Cryogenics, Vol. 37, No.5 (1997), pp. 271-273.
3. Zhou, S.L., Thummes, G., and Matsubara, Y., “Experimental Investigation of Loss Mechanism in a
4 K Pulse Tube,” to be published in Advances in Cryogenic Engineering.
4. Zhu, S.W., Kakimi, Y., and Matsubara, Y., “Investigation of Active-buffer Pulse Tube Refrigerator,”
Cryogenics, Vol. 37, No. 8 (1997), pp. 461-471.
Compressor-Specific Design of a Single Stage
Pulse Tube Refrigerator
J. M. Pfotenhauer and J.H. Baik
University of Wisconsin - Madison

Madison, WI USA 53706
ABSTRACT
A single stage active valve pulse tube refrigerator has been designed to operate at 30 K
and provide a nominal cooling power in excess of 30 watts. This report details the various
considerations comprising the system design, focusing on the limitations imposed by the
reciprocating-type compressor commonly used for GM-style pulse tube refrigerators. We
describe a design method for GM-style pulse tubes to maximize the pulse tube cooling power
that can be produced from a compressor of fixed capacity. The method provides a physical
understanding of the various influencing factors, and is illustrated using the specifications for the
Cryomech CP640 compressor, which draws a maximum electrical power of 5.5 kW. The design
process begins by defining mass flow and compressor work as a function of the discharge and
suction pressures, thereby producing a compressor performance map. The compressor map in
turn provides a framework from which the pulse tube system geometry can be optimized for
maximum cooling power. Various real constraints, such as pressure drop through the valves and
regenerator, laminar boundary layer along the pulse tube walls, and conduction losses are
included in the design process and are shown to significantly impact the optimized result.
INTRODUCTION
In the process of designing a GM-type pulse tube refrigerator, one must select a
compressor from a finite set of fixed compressor capacities - deliverable work is not available as
a continuum. In view of this constraint it is relevant for a designer to ask, "what pulse tube
geometry will best utilize the available power from a specific compressor?" - that is, which
geometry will produce the highest cooling power for the fixed limitations of the compressor? To
date, this question has been answered largely by imperical means. For example, Fujimato et.al.1
varied the length of their pulse tube, finding that the cooling power increased with the length, but
that "the compressor power was too large for the pulse tubes used in [their] tests." Wang,
Thummes & Heiden2 found that the cooling capacity of their pulse tubes decreased as the
diameters (and volumes) were increased. An exception to the imperical approach is presented by
Ravex et.al.3 who confirm the ability of their numerical model to optimize their pulse tube
geometry for a specific compressor. The present report follows in this non-imperical direction,

but provides a general procedure that enhances the physical understanding of various constraints
along the way.
The following paragraphs outline a procedure for designing a GM-type pulse tube
cryocooler that will provide the maximum possible cooling capacity for a fixed compressor. The
procedure identifies the constraints imposed by the compressor, as well as those associated with
the pulse tube and regenerator geometries. The process begins by defining a compressor map in
terms of mass flow, compressor work, and the inlet and outlet pressures. Practical considerations
regarding the pulse tube volume and valve system determine the attainable range on the
compressor map, and an iterative method is presented for optimizing the geometry of the pulse
tube refrigerator. The specific characteristics of a Cryomech model CP640 compressor are used
to illustrate the compressor constraints. In the iterative process, we use the thermodynamic
model of Yuan & Pfotenhauer4 to calculate the geometry dependent cooling capacity of a 5-valve
pulse tube. However, in the case of other pulse tube configurations, alternate methods for
calculating the cooling power as a function of the pulse tube volume could be used instead.
COMPRESSOR CHARACTERISTICS
The mass flow delivered by a reciprocating compressor can be expressed (see for
example references 5-7) in terms of the displaced volume, compressor speed S, clearance
volume ratio the inlet and discharge pressures and the polytropic
compression exponent n according to the Eq.:
Here is the clearance volume, and the polytropic exponent n is associated with the expression
that describes the relationship between pressure and volume during the compression and
expansion processes. The possible values of n are bounded by (minimum) for an
isothermal compression process, and (maximum) for an adiabatic compression
process. Here and are the specific heat at constant pressure and constant volume
respectively. The expression given in Eq. (1) results from considerations regarding the
volumetric efficiency of a reciprocating compressor, and has been extensively verified in a recent
investigation6 of commercial compressors used in the refrigeration industry. An expression for
the electric power consumed by the compressors is also provided in the same study, and is given
as:
Here is the combined efficiency of converting electrical to mechanical and mechanical to

PV work, which may be roughly given as for a wide variety of compressors.
Although the values for the clearance volume and the polytropic exponent that are required in
Eq.s (1) and (3) are not readily available from the commercial vendors, these can be obtained
from a few performance characterization measurements for a compressor of interest. Thus it is
possible to accurately define the mass flow that will be delivered by a compressor for a given set
of inlet and outlet conditions.
COMPRESSOR-SPECIFIC DESIGN OF A 1-STAGE PT COOLER 251
The compressor map defined by Eq.s (1) and (3) for the Cryomech CP640 is shown in
Fig. 1. Using our measurements of mass flow for various inlet and outlet pressures, and an inlet
temperature of 315 K, we determined values of and for the clearance volume
ratio and polytropic exponent respectively. The sensitivity of the mass flow rate to and its
insensitivity to can be understood in terms of the expression
Except through the relatively weak dependence of the volumetric efficiency on the outlet
pressure, the mass flow is entirely determined by the inlet conditions. The compressor work, on
the other hand is dependent on both and and is maximized at the higher values of
Constraints imposed by the compressor define three limitations to mass flow that can be
identified on the compressor map. A minimum inlet (suction) pressure and a maximum outlet
(discharge) pressure define the first two of these. For the Cryomech CP640 these are respectively
defined by and The third constraint is unlikely to be encountered
for a pulse tube system, but is defined by the condition
OPTIMIZATION DESIGN PROCESS
Within the constraints imposed by the compressor, Fig. 1 displays a wide range of mass
flow rates available through many combinations of inlet and outlet pressures. The cooling power
of pulse tube refrigerators is known to depend on all three of the parameters and mass
flow rate. It is therefore of interest to determine both the region of the compressor map that
affords the largest pulse tube cooling capacities, and which regions are accessible in real systems.
The first of these questions is answered for the case of a 5-valve pulse tube using an iterative
process described by the flow diagram in Fig. 2.
Figure 1. Compressor map for Cryomech CP640 reciprocating compressor.

Figure 2. Flow diagram of iterative design process to optimize a pulse tube system geometry for
maximum cooling power given specific values of and mass flow rate.
The iterative design process begins by considering pressure losses across the inlet and
outlet valves connecting the pulse tube system to the compressor. In view of the energy losses
directly associated with such a pressure drop, it is important to minimize the flow resistance
through these valves. For all valved (GM-type) pulse tube systems, the pressure swings
experienced in the pulse tube system will be less than those produced by the compressor. In
Fig. 2 above, and represent the high and low pressures realized in the pulse tube system
respectively.
The next step requires a selection of the pulse tube operating, or cold end, temperature.
This selection is of course not arbitrary and must be based on reasonable limits for a single stage
pulse tube cooler appropriate selection of regenerator matrix material, and admittedly
some previous experience. In the examples to follow, we have used as a cold end
temperature. With the values of and mass flow rate through the pulse tube fixed, one
might expect that the pulse tube volume would be fixed. In fact, an iteration process is required
to determine the combination of pulse tube and regenerator gas volumes associated with the fixed
conditions that maximize the resulting cooling power. The ratio of the pulse tube volume to
the gas (or non-solid) volume in the regenerator characterizes the competing effects of expansion
COMPRESSOR-SPECIFIC DESIGN OF A 1-STAGE PT COOLER 253
volume and pressure loss: Large values of maximize the expansion volume in the pulse tube,
but produce large pressure drops through the regenerator. Small values of minimize the
pressure loss through the regenerator but sacrifice expansion cooling in the pulse tube.
In addition to optimizing the cooling power as a function of the length-to-diameter
aspect ratio of both the pulse tube and regenerator can be optimized for each value of and mass
flow rate. For the case of the pulse tube, the aspect ratio is chosen as large as possible to
minimize conduction losses through the walls, but not so large that the boundary layer velocity
becomes turbulent. The laminar to turbulent transition is defined8 by the condition that the
Reynolds number be less than 280. The characteristic dimension in the Reynolds number is
defined by the Stokes boundary layer thickness
where is the kinematic viscosity and is times the frequency. The laminar condition
constrains the cross sectional area of the pulse tube to be larger than a minimum defined by
For the regenerator, REGEN29 is used to balance the conduction and regenerator
ineffectiveness against the pressure drop in through the regenerator. In our designs, we have
selected a maximum allowable pressure drop of 50 kPa. Fig. 3 displays the cooling power, pulse
tube volume, and regenerator gas volume as a function of for one set of and mass flow
rate values. In this case, a decreasingly significant benefit is realized by increasing beyond 8.
In the interest of a compact design, this value of is chosen for the optimized design.
Figure 3. Geometry optimization of pulse tube and regenerator volumes for single set of
and mass flow rate values.
Figure 4. Pulse tube cooling capacities as a function of compressor characteristics.
The same process described in the previous paragraph, carried out for a variety of inlet
and outlet compressor pressures, results in the cooling powers shown in Fig. 4. At low values of
the outlet pressure, a weak pressure wave results in the pulse tube and increased values of inlet
pressure reduce the pressure oscillations and therefore the cooling power. At high values of
outlet pressure, the dependence of the cooling capacity Q on the inlet pressure is reversed. In this
case, increased inlet pressures produce an increased density at the compressor inlet, a larger mass
flow rate, and a larger cooling power. It is of obvious interest to know which parts of this map
are accessible for a real pulse tube system, and what steps can be taken to reach the region of
both high outlet and high inlet pressures.
Pulse Tube System Options
From the results displayed in Fig. 4 it is clear that the maximum outlet pressure permitted
by the compressor corresponds closely with its rated power capacity. Operating at is a
necessary, but not sufficient condition for maximizing the cooling power of a pulse tube system
attached to the compressor. It is also desirable to maximize as well. How can this be
achieved?
Beginning with the case of an ideal pulse tube system - that is, one with no losses - one
can consider the influence of pulse tube size on the pressures in the pulse tube system. Fig. 5
depicts the high and low pressures that will result in a pulse tube system as a function of the
charging pressure, for a small volume and a large volume pulse tube system. The maximum and
minimum charging pressures in the pulse tube system are constrained by the compressor's
maximum discharge and minimum suction pressures respectively. A small volume pulse tube
system will realize large pressure oscillations, while the large volume will realize small pressure
oscillations. The arrows, representing the pressure swing realized in each pulse tube system
when it is charged to its maximum allowable pressure, reveal that although the pressure
oscillations will decrease as the pulse tube volume is increased, for a constant the inlet
pressure will increase with pulse tube volume. This intuitive consideration is confirmed by the
COMPRESSOR-SPECIFIC DESIGN OF A 1 -STAGE PT COOLER 255
Figure 5. Depiction of pressures realized in pulse tubes as a function of the charging (average) pressure,
and as limited by the minimum and maximum pressures provided by the compressor.
Figure 6. Ideal cooling power vs. the optimized pulse tube volume for the various combinations
of high and low pressures produced by the Cryomech CP640 compressor.
calculations displayed in Fig. 6. Here the same cooling powers as calculated in Fig. 4 are shown
with their corresponding optimized pulse tube volumes. For the highest values of Q clearly
increases as the pulse tube volume is increased.
Real System Constraints
Based on the calculations displayed in Fig. 6, one may wonder whether an upper limit to
the pulse tube system volume exists. In a real pulse tube system, losses associated with the
pressure drop through the valves and regenerator, and conduction losses (as the pulse tube
volume increases, the optimized aspect ratio decreases) impose practical limits and define a
different optimum pulse tube volume for maximum cooling power. Repeating the same
procedure that resulted in Figs. 4 and 6, but including the losses realized in our valve system, the
50 kPa pressure drop through the regenerator, and conduction losses through the pulse tube and
regenerator structure, provides the results shown in Fig. 7. The optimum pulse tube volume falls
between 200 and 300 with a associated values of and pulse tube aspect ratio of 8.
For this design, a cooling power of 60 watts is expected at 30 K. Two additional losses have yet
to be included in our model; those associated with shuttle heat loss, and DC flows. We expect
therefore that the actual cooling power will be less than 60 watts. The significance of valve
losses are clearly evident through this design process. We are presently pursuing an improved
valve arrangement for our own pulse tube system.
The results provided in Figs. 4, 6, and 7 also permit an estimate of efficiency for GM
style pulse tube refrigerators. For the ideal case (ignoring losses) depicted in Fig. 4, one finds
that the efficiency associated with the maximum possible cooling power of a pulse tube operating
at 30 K and driven by the CP640 compressor is 0.03, or 28% of Carnot. A more realistic value,
including conduction losses and pressure drop through the regenerator is 0.015, or 14% of
Carnot.
Figure 7. Cooling power vs. optimized pulse tube volume. The optimization process accounts for
pressure losses through valves and regenerator, and conduction losses.
COMPRESSOR-SPECIFIC DESIGN OF A 1 -STAGE PT COOLER 257
Applying the above procedure for the CTI 8500 compressor used by Ravex et.al.3 we find
that an optimum pulse tube volume for operation at 30 K would be approximately three times
larger than what they used but that the aspect ratios of the pulse tube and
regenerator would be similar. Furthermore an optimum
value of and a cooling power of 82 watts is found for that case.
CONCLUSION
A design method has been described that allows one to optimize the cooling power of a
GM-style pulse tube refrigerator for a given fixed compressor. The method produces a
compressor performance map and defines an iterative procedure for optimizing the cooling
power of the pulse tube refrigerator as a function of its volume and geometry. Considerations
regarding the volume of an ideal pulse tube suggest that the volume should be made as large a
possible. However, real losses such as pressure drop through the pulse tube and conduction
losses significantly influence the process and allow one to define an optimum volume and
geometry. The efficiency of a GM-style pulse tube operating at 30 K is limited to less than 15%
of the Carnot efficiency.
ACKNOWLEDGEMENT
This work is supported by DOE through Lockheed Martin subcontract DE-AC05-960

OR22464.
REFERENCES
1. S. Fujimoto, Y.M. Kang, Y. Matsubara, "Development of a 5 to 20 W at 80 K GM Pulse Tube

Cooler," Cryocoolers 10, Plenum Press, New York (1999), pp. 213-220.
2. C. Wang, G. Thummes, C. Heiden, "Performance Study on a Two-Stage 4 K Pulse Tube Cooler,"
Advances in Cryogenic Engineering vol. 43, Plenum Press, New York (1998), pp. 2055-2062.
3. A. Ravex, J.M. Poncet, I. Charles, and P. Bleuze, "Development of Low Frequency Pulse Tube
Refrigerators," Advances in Cryogenic Engineering vol.43, Plenum Press, New York (1990), pp.
1957-1964.
4. J. Yuan and J.M. Pfotenhauer, “Thermodynamic Analysis of Active Valve Pulse Tube
Refrigerators,” Cryogenics vol. 39, (1999), pp. 283-292.
5. P. Popovic and H.N. Shapiro, "A Semi-empirical Method for Modeling a Reciprocating Compressor
in Refrigeration Systems," ASHRAE Transactions vol. 101 (2), (1995), pp. 367-382.
6. D. Jaehnig, "A Semi-Empirical Method for Modeling Reciprocating Compressors in Residential
Refrigerators and Freezers," MS Thesis, University of Wisconsin - Madison, (1999).
7. D. Jaehnig, S.A. Klein, and D.T. Reindl, "A Semi-Empirical Method for Representing Domestic
Refrigerator/Freezer Compressor Calorimeter Test Data.", ASHRAE Transactions, (June, 2000),
Minneapolis, MN
8. R. Akhavan, R.D. Kamm, and A.H. Shapiro, "An Investigation of Transition to Turbulence in
Bounded Oscillatory Stokes Flows - Part 1. Experiments," J. Fluid Mech. vol. 225, (1991), pp. 395-
422.
9. V. Arp and R. Radebaugh, "Interactive Program for Microcomputers to Calculate the Optimum
Regenerator Geometry for Cryocoolers (REGEN and REGEN2)," AFWAL-TR-87-3040 Wright-
Patterson Air Force Base, (1987).
A Novel Multi-Stage Expander Concept
C. S. Kirkconnell, K. D. Price, M. C. Barr, and J. T. Russo
Raytheon Electronic Systems
El Segundo, California, USA 90245
ABSTRACT
Raytheon has developed a novel two-stage expander for use in long life, high reliability cryo-
coolers for space and commercial applications. The expander is classified as a Stirling machine and
requires a conventional reciprocating piston compressor to drive it. The key feature is a new method
for obtaining and controlling expansion at the two stages. Thermodynamic efficiency is higher than
existing one and two stage coolers and the mechanical implementation is as simple or simpler. The
expander device is described in both thermodynamic and mechanical terms and performance pre-
dictions given.
INTRODUCTION
Raytheon has developed a hybrid Stirling-pulse tube two-stage expander module for use in long
life, high reliability space applications. In this design, the first stage is a traditional “Oxford” Stirling
flexure suspended expander and the second stage is a U-tube pulse tube. Refrigerant gas flows
through the Stirling first stage to the pulse tube second stage. The warm ends of the pulse tube, its
regenerator, and the orifice/surge volume are thermally anchored to the Stirling stage. The same
compressor powers both stages. The hybrid design offers significant advantages in thermodynamic
performance, power efficiency, ease of construction, reliability, cost, size, and weight compared to
existing cryocooler technologies. In the following sections, the thermodynamic benefits of the new
expander are discussed in comparison to existing technologies, the mechanical design is described,
and performance predictions are given.
THERMODYNAMIC BENEFITS OF NEW APPROACH

The hybrid expander achieves high efficiency by combining the best features of Stirling and
pulse tube coolers while eliminating or attenuating their inefficient features. The following briefly
describes the key advantages and disadvantages of Stirling and pulse tube one-stage expanders.
This will be followed by an explanation of how the hybrid two-stage design advantageously com-
bines these features to produce a high capacity, high efficiency cryocooler.
Stirling expanders achieve high efficiency in part because of the relatively low gas volume that
must be cyclically pressurized and depressurized. This minimizes the mass flow rate of gas into and
out of the expander relative to a comparable pulse tube cooler. The low flow rate enables better
optimization of regenerator parameters, reduces pressure drop losses through the machine, and
reduces compressor swept volume compared to a pulse tube. A limiting factor in Stirling perfor-
mance is the complexity required to obtain high heat transfer efficiency at the cold end. A Stirling

260 HYBRID CRYOCOOLERS USING PULSE TUBES
Figure 1. Stirling-Pulse Tube Hybrid Expander Concept. The two-stage hybrid expander simply
attaches a pulse tube expander stage to the cold end of a Stirling expander. The schematic shows the basic
configurations of one-stage coolers and the two-stage hybrid cooler. In the two-stage cooler, gas flows
through the Stirling stage to the pulse tube stage as in conventional Stirling class multi-stage cryocoolers.
The schematic shows a linear pulse tube, but the Raytheon hybrid configures the pulse tube stage as a
U-tube for compactness and ease of integration.
with regenerator integrated within the piston has relatively poor heat transfer at the cold end be-
cause there is no flow-through heat exchanger from the regenerator to the cold thermal interface.
The gas simply impinges at relatively low speed on the internal surfaces. To achieve high heat
transfer, the regenerator must be external to the piston (e.g., fixed to non-moving structure) so that
gas flowing into the cold expansion volume can pass through an efficient heat exchanger connected
to the thermal load. Mechanical arrangements to implement this are relatively complex and the
structures tend to have higher conductive parasitic loads that offset the benefit of improved heat
transfer.
Pulse tubes achieve good performance partially because of the high-efficiency flow-through
heat exchangers present throughout the cooler, but suffer because the relatively large working vol-
ume draws high gas mass flow rates. The high flow rates result in suboptimal regenerator param-
eters that trade heat transfer efficiency against pressure drop losses, and the larger working volume
increases the size of the compressor required to drive the expander.
When a Stirling first stage is combined with a pulse tube second stage, the gas flow rate through
the expander is reduced relative to an all pulse tube design because the Stirling stage has a smaller
gas volume compared to a pulse tube. This reduces the size of the pressure drop losses in the
expander. Furthermore, the gas volume is greatly reduced compared to an all pulse tube machine,
reducing the size of the requisite compressor.
The Stirling stage receives a high efficiency, flow-through heat exchanger because one can be
inserted in the gas flow path from the Stirling expansion volume into the pulse tube regenerator.
Furthermore, the Stirling stage can retain the regenerator integrated within the piston, which is
beneficial for manufacturing and reduces parasitic loads relative to an external regenerator design.
Finally, the complexity of the Stirling stage is no greater than the complexity of well-developed
single-stage Stirling coolers. For example, a tight tolerance, cryogenic clearance seal is not re-
quired to separate the two stages.
Since the pulse tube stage is entirely at cryogenic temperature, pressure drop losses within the
stage are naturally low. Reduced pressure drop improves regenerator design parameters resulting in
more efficient regenerator heat transfer, maximum pulse tube work-of-expansion, and low thermal
conduction losses.
A NOVEL MULTI-STAGE EXPANDER CONCEPT 261
The hybridized expander module is no more difficult to manufacture than a one-stage Stirling
yet delivers true two-stage cryocooler performance. The second stage is entirely passive, yielding a
mechanically simple design with very few tight tolerance features. The angular orientation of the
second stage with respect to the first stage is arbitrary. Like a two-stage pulse tube but distinct from
a two-stage Stirling, the absence of a moving piston in the second stage allows the user to orient the
stages for optimum integration at the system level. In-line (0°), a 90° bend, or any intermediate
angle between 0° and 90° are all possible. The required compressor swept volume is significantly
lower than would be required for either a one or two-stage pulse tube of comparable heat lift and is
not significantly larger than for a one or two-stage Stirling expander.
The expansion process in each stage is controlled by a separate mechanism that gives the hybrid
an unusually useful feature: refrigeration power can be allocated and reallocated between the two
stages on command and continuously in real time. The Stirling first stage piston is motor con-
trolled, so its motion can be altered on command. A fixed orifice and surge volume controls the
pulse tube expansion. Analysis shows that modifying the first stage piston motion will alter gas
flow rates, phase angles, and pressure ratios in both stages resulting in the reallocation of refriger-
ating power between the stages on command. Thus, refrigerating power can be biased to increase
heat lift at the second stage at the expense of first stage heat lift and vice versa. This functionality is
not available from typical single-piston, two-stage Stirling designs because the expansion phase
angles in both cryogenic stages are directly, mechanically linked. Two-stage pulse tube cryocoolers
have no active phase control, so dynamic load shifting is not possible. The partial coupling of the
first and second stage expansion phase angles and the capability to shift the refrigeration allocation
on the fly are by-products of the unique thermodynamic performance characteristics provided by
the hybrid expander design. This capability can be used to extend the performance range of a single
cryocooler to different combinations of first and second stage temperature and heat load require-
ments or the heat lift distribution between stages can be modified during operation to accommodate
fluctuating heat loads.
STIRLING-PULSE TUBE EXPANDER DESIGN

The Hybrid Expander, shown in Figure 2, is composed of a small U-tube pulse tube second
stage with surge tank attached onto a first stage consisting of a moving Stirling piston assembly,
which is essentially a resized version of the Raytheon PSC piston assembly.1 The first-stage Stirling
piston is supported on a suspension of 12-finger spiral flexures to provide long-life, non-contacting
operation. The piston is driven by a compact linear motor. A like suspension / motor arrangement
Figure 2. Raytheon Protoflight Hybrid Expander. Second stage oriented at 45° with respect to the
first stage for protoflight. Design utilizes standard Raytheon moving displacer in first stage and U-tube
pulse tube for second stage.
is packaged to drive a countermass to null the vibration output from the expander. As in all
Raytheon Stirling cryocoolers, two close-fit, non-contacting clearance seals are incorporated at
the warm end of the moving first stage piston: the “plenum” seal and the “regenerator” seal. The
plenum seal isolates the working gas volume from the non-working plenum volume where the
flexure suspension system and control motor and balance mass reside. The regenerator seal mini-
mizes leakage of gas around the regenerator to port working gas through the regenerator.
From the Stirling expansion space, the first-stage gas is ported into a copper mesh heat ex-
changer housed within the first-stage heat interface block, and this same block houses the pulse
tube surge volume. The gas ports through a 90° turn as it enters the second-stage regenerator, the
purpose being to demonstrate the hybrid’s insensitivity to relative angle between the stages. (A 45 °
orientation has been selected for protoflight, which is the configuration depicted in Figure 2.) The
gas continues through the pulse tube regenerator out to the second stage heat interface block where
the second-stage expansion occurs. At this block the gas is turned 180° and sent through another
copper mesh heat exchanger, which also serves as a flow straightener for the cold end of the pulse
tube. The gas enters the pulse tube from the heat exchanger on its way back to the first stage heat
interface block where it passes through yet another copper mesh heat exchanger / flow straightener.
Here it passes through a flow restrictor and is ported into the surge volume.
The containment housing at the warm end is made from aluminum. The first-stage thin-walled
cold cylinder is stainless steel to minimize parasitic heat loss. The first- and second-stage heat
interface blocks are made from copper and the second stage regenerator and pulse tubes are made
from inconel, also selected to minimize conduction parasitics. A variety of welds and braze joints
are used to join together these disparate materials. The complete cryocooler (compressor + ex-
pander) can be made with all welded or brazed joints except for one mechanical joint at the transfer
line. Warm end heat rejection, structural mounting and vacuum sealing are interfaced to the alumi-
num warm end housing. Electrical feedthrough headers are welded onto steel bosses on the alumi-
num housing. In total, the expander weighs about 3.2 kilograms.
DISCUSSION
Performance Predictions
Thermodynamic models have been well correlated at Raytheon for both Stirling cryocoolers
(SSC2, ISSC, PSC1, SBIRS Low) and Pulse Tube cryocoolers (35K IR&D3, Mini IR&D4, 4-Tube5,
LCC6). A numerical model of the Hybrid Expander was developed based upon these building
blocks, a fundamental purpose of the breadboard expander presently in assembly being to correlate
the new math model. Of particular interest are those portions of the model that represent features
unique to the Hybrid Expander, not relevant to either single-stage model. Examples are the pres-
ence of a flow-through heat exchanger at the cold end of the Stirling and the use of a cryogenic
surge tank for the pulse tube stage.
The maturity of the models from which the hybrid model was constructed provides confidence
that reasonably accurate performance predictions can be obtained for the breadboard unit. In the
areas of modeling uncertainty such as those noted above, conservative assumptions were used. The
nominal design point for the breadboard expander based upon the existing model is 5.6 watts at
95 K and 5.2 watts at 145 K for 75 watts input P-V power, or about 100 W input motor power for a
typical Stirling-class reciprocating compressor, at a heat rejection temperature of 305 K. Extensive
characterization of the breadboard expander is planned over a wide trade space of temperatures,
loads, frequencies, charge pressures, and phase angles. Experimental data will be provided at such
time that it becomes available.
Thermodynamic Efficiency Benefits of Hybrid Expander
The concept of “specific power,” the ratio of input power to net refrigeration capacity, has
become well established in the characterization of single-stage cryocoolers as providing an infor-
mative representation of a cooler’s efficiency. In a paper presently in work by the authors7, a
methodology for normalizing the performance of multi-stage refrigerators in terms of a common
A NOVEL MULTI-STAGE EXPANDER CONCEPT 263
“pseudo single stage” efficiency is presented. The total capacity at the various stages is normalized
to a single capacity at the coldest stage by use of the following equation:
In the above expression, is the Carnot COP based upon the refrigeration temperature at each
stage and the cryocooler’s warm rejection temperature, typically at or near ambient. From Eq. (1),
the normalized refrigeration capacity for the breadboard hybrid expander, assuming a rejection
temperature of 305 K, is 8.2 watts at 95 K. Continuing with the single-stage analogy, the specific
power for the breadboard hybrid cooler, based on 100 watts motor power, is 12.2 W/W at 95 K.
Preliminary predictions indicate that hybrid designs tuned for operation in this temperature range
should be readily capable of specific power efficiencies of 10 W/W at 95 K, with the refrigeration
capacities split between the first and second stages as desired by the user.
CONCLUSION
A novel two-stage Stirling-Pulse Tube hybrid expander has been developed by Raytheon and is
in final assembly as of the writing of this paper. The hybrid expander takes advantage of the
strengths of the individual expander designs upon which it is based. Using a Stirling expander first
stage reduces the large pressure drop loss characteristic of both single-stage and multi-stage pulse
tube expanders. The mechanical complexity of a cryogenic clearance seal is avoided entirely by
using a pulse tube second stage, and this enhances the reliability of the expander by eliminating a
potential single point failure mode. The actively controlled Stirling displacer in the hybrid pro-
vides the unique capability to shift refrigeration capacity between the stages, a feature not available
to multi-stage pulse tubes or multi-stage Stirlings with single displacer pistons. The projected
performance for the breadboard hybrid expander unit presently being assembled is 5.6 W at 95 K
and 5.2 W at 145 K for 100 W input power.
REFERENCES
1. Price, K.D., Barr, M.C. and Kramer, G., “Prototype Spacecraft Cryocooler Progress,” Cryocool-
ers 9, Plenum Press, New York (1997), pp. 29-34.
2. 65 K Standard Spacecraft Cryocooler Program Final Report, Contract #F29601-89-C-0082, Hughes
Aircraft Company, Electro-Optical Systems, El Segundo, CA; November 1995.
3. Soloski, S.C. and Mastrup, F.N., “Experimental Investigation of a Linear Orifice Pulse Tube Ex-
pander,” Cryocoolers 8, Plenum Press, New York (1995), pp. 321-328.
4. Kirkconnell, C.S., Soloski, S.C. and Price, K.D., “Experiments on the Effects of Pulse Tube Geom-
etry on PTR Performance,” Cryocoolers 9, Plenum Press, New York (1997), pp. 285-293.
5. Kirkconnell, C.S., “Experimental Investigation of a Unique Pulse Tube Expander Design,” Cryo-
coolers 10, Plenum Press, New York (1999), pp. 239-247.
6. Kirkconnell, C.S., “Experiments on the Thermodynamic Performance of a ‘U-Tube’ Pulse Tube
Expander,” Advances in Cryogenic Engineering, vol. 43 (1998), pp. 1973-1980.
7. Kirkconnell, C.S. and Price, K.D., “Thermodynamic Optimization of Multi-Stage Cryocoolers,”
Cryocoolers 11, Plenum Press, New York (2001).
Numerical Study of a New Type of 4 K GM/PT
Hybrid Refrigerator
Liqiang Liu, Linghui Gong, Jingtao Liang and Liang Zhang

Cryogenic Laboratory
Chinese Academy of Sciences
Beijing 100080, China
ABSTRACT
A new type of hybrid refrigerator operating in the liquid helium temperature region has been
proposed. The warm stage of this refrigerator is a typical G-M refrigeration cycle, on which the
cold stage of a pulse tube cooler is coupled thermodynamically. The phase shift structures for the
pulse tube are supported at the temperature of the G-M refrigeration stage. There are several meth-
ods that can be applied as the phase shifter for the pulse tube; for example, a cold piston that is
connected to the displacer of the G-M refrigerator. Physical and mathematical models have been
established to describe the unique thermophysical aspects of this type of refrigerator, and numeri-
cal methods have been employed to solve the theoretical models. Through the calculations, some
helpful results have been obtained, and the structure of the refrigerator has been optimized. Experi-
mental evaluation is now underway.
INTRODUCTION
In recent years, owing to the introduction of magnetic regenerator materials, the low tempera-
ture performance of G-M refrigerators has been improved, particularly at 4.2K, and other signifi-
cant performance enhancements have been achieved.1 However, due to the seal rings having to
operate in the low temperature environment in a G-M refrigerator, the life and the stability of the
refrigerator are seriously compromised; furthermore, the structure of its cold stage is complex. On
the other hand, substantial performance improvements have been recently achieved in pulse tube
coolers through the introduction of an orifice and double inlet as well as other phase shifters.2
Whereas the cooling power of a pulse tube cooler may be less than that of a comparable G-M
refrigerator, its cooling temperature with two stages can reach below 4.2 K. Also, the structure of
its cold stage is very simple, as the seal rings are eliminated.
Based on theoretical considerations, a new type of 4K refrigerator has been proposed in which
the warm stage is a typical G-M refrigeration cycle, onto which the cold stage of the pulse tube
cooler is coupled. This new type of refrigerator can overcome the disadvantages of G-M refrigera-
tors, and has promise for future applications that require somewhat lower cooling power (e.g.,
W at 4.2 K). At present, a prototype of this new type of refrigerator has been manufactured, and
experimental measurements are under way. In this paper, a numerical analysis and optimization
calculations are presented.
Cryocoolers 11, edited by R.G. Ross. Jr.

Figure 1. Brief structure of the new type of 4K refrigerator.
STRUCTURE OF THE NEW TYPE OF REFRIGERATOR

As shown in Fig. 1, the new type of refrigerator has two stages. The warm stage of the
refrigerator is a typical G-M refrigeration cycle; it is coupled by a pulse tube to the cold stage.
With the goal of enhancing the ability to control the pulse tube phase, and building upon the
phase shifting technique of the G-M refrigerator, a cold piston connected to the G-M displacer
has been introduced as the phase shifter for the pulse tube. Besides this mode of shifting the
pulse tube phase by a cold piston, other styles of phase shifters, such as an orifice or double
inlet, can be conveniently introduced to improve the pulse tube's performance. Through skill-
ful design, these other phase shifting concepts can be used either independently, or in combi-
nation with the cold piston. As a result, some new methods of shifting phase are introduced,
and opportunities for improving the performance of the refrigerator are increased.
From Fig. 1 one can see that the new refrigerator is not just a pulse tube cooler pre-cooled
by a G-M refrigerator. In this new concept, all gas passages are linked, and the pressure,
temperature, and mass flow oscillations share the same frequency. Therefore, analytical mod-
els of either G-M refrigerators or pulse tube coolers by themselves are not capable of explain-
ing the working mechanisms of the new refrigerator; a new theoretical model must be built. In
the present study, a numerical model for simulating the dynamic performance and characteris-
tics of the oscillating flow in the new refrigerator was developed to analyze its performance
and optimize its structure.
PHYSICAL MODEL AND GOVERNING EQUATIONS

Physical Model
Figure 2 shows the physical model of the new refrigerator that is the subject of the study.
The focus is on the low temperature stage regenerator and pulse tube. In the model, the junc-
tion between the cold chamber of the upper stage and the low temperature regenerator is taken
as the left boundary, and the inner surface of the pulse tube phase-shifter piston is taken as the
right boundary. Each method of shifting phase can be embodied by dealing with the right
boundary properly. For example, when the cold piston and the orifice phase shifters are both
used, the right boundary is movable, and the width of the grid to the left of it is varied accord-
ingly; in addition, the gas velocity through this boundary u, is determined by the orifice and
NUMERICAL STUDY OF A NEW 4 K GM/PT HYBRID COOLER 267
Figure 2. Schematic of the physical model for the new type of refrigerator.
reservoir. Otherwise, the length of the grid left to the right boundary is constant if the cold piston
is not employed, and is zero if the orifice phase shifter is not applied.
The basic assumptions in the model are as follows:
1. one-dimensional laminar compressible flow of
2. constant wall temperature of heat exchangers at cold and warm end of the pulse tube;
3. axial heat conduction neglected;
4. pressure drop in the regenerator also neglected
Governing Equations
The governing equations for the numerical study are given as follows, where the gas
velocity, u, is defined as positive if the gas flow is from left to right and as negative for the
opposite flow. Continuity equation for the gas:
where is the density of the gas. Energy equation for the gas:
where is the heat transfer area per unit volume of the matrix, h is the specific enthalpy of the
gas, is the temperature of the gas, is the temperature of the matrix, a is the coefficient of
heat transfer, which is taken from the reference 3. Energy equation for the regenerative materials:
Equation of state for the real gas of helium:
Boundary Conditions
The gas temperature at the right boundary is as follows:
where is the gas temperature in reservoir, which is taken to be the constant, is also taken
as the wall temperature of the heat exchanger at warm end. Meanwhile the wall temperature of
the heat exchanger at cold end is taken to be another constant,
The process in the reservoir is regarded as isothermal and isobaric. The gas velocity through
the right boundary (this velocity is zero if the orifice phase shifter is not employed) is determined
by a formula for a nozzle with a correction factor as follows:
where denotes the pressure in the reservoir and is the correction factor for the geometry of
the nozzle.
The motion of the piston for phase shift is given by
where is the crank angle, is the phase difference between the pressure wave and the
movement of the piston, is the stroke, is the dead gap, which generally is 0.5 mm.
The gas temperature at the left boundary is as follows:
The gas inlet temperature is maintained constant and the outlet temperature depends on the
efficiency of the regenerator.
For the left boundary, the pressure oscillation is taken as input data. The pressure varies with
as
where and denote the maximum and the minimum of the pressure oscillation respectively.
NUMERICAL METHODS AND PROCEDURE

To carry out the numerical simulation, discretization procedures were employed. The govern-
ing equations were discretized using a control-volume approach with the internal node method (the
length of the last spatial grid, from left to right, was variable if the cold piston was used as the phase
shifter). To reduce the number of spatial grids, a nonuniform grid was used. An implicit approach
for time items in the governing equations and the upwind one-order scheme were applied to achieve
calculational stability. The staggered grid approach, which uses control volumes for the velocities
that are staggered with respect to those for temperature and pressure, was also adopted to eliminate
erroneous pressure profiles. The governing equations were solved with the under-relaxation itera-
tion method. In particular, it should be noted that during the solution, close attention was paid to a
special case of zero gas velocity. This case sometimes occurred along the spatial grid when the
direction of gas velocity reverses at a certain place; when this special case occurs, the iteration
procedure is lightly divergent. An efficient method to resolve the problem is to employ the upwind
scheme strictly at every spatial grid, especially at the grid with zero gas velocity.
COMPUTATIONAL RESULTS AND DISCUSSION

The numerical analysis was first applied to the new refrigerator with the phase shifting accom-
plished merely by the cold piston (abbreviated ‘Pis.’); next it was applied assuming both the cold
piston and the orifice (abbr. Pis.-Hol.). For comparison, the performance of the orifice-only ver-
sion of the pulse tube cooler (abbr. Hol.) is also presented using the above numerical analysis.
Finally, the optimized configuration of the new refrigerator is presented. If no special explanation
is given, the parameters used are as shown in Table 1. The major operating conditions are:
MPa, (in the case of the Pis.-Hol. shifter), and the operating
frequency was (that is, the rotary speed of the refrigerator was 60 rpm).
Cooling Power
The temperature dependence of cooling power obtained using the three versions of the phase
shifter is shown in Fig. 3. It is obvious that the performance of the refrigerator with only a cold
piston as its phase shifter is not as good as expected. Nevertheless, the performance can be im-
proved (the cooling temperature can be decreased more than 4.5 K) in the case where the cold
piston is combined with another version of the phase shifter, such as an orifice. A promising future
candidate may be to combine the cold piston with a double-inlet phase shifter.
Mass Flow Rate
Figures 4 and 5 show the transient mass flow rates at the warm and cold ends of the pulse tube
with the three versions of the phase shifter, respectively. The pressure wave is also presented in
these figures to help discriminate the difference in the phases. It is found that the phase difference
between the pressure wave and the mass flow rate of the Pis.-Hol. shifter is superior to that of the
Hol. or Pis. shifters at the cold end of pulse tube. In contrast, the former is inferior to that of the
Hol. shifter at the warm end. This phenomenon predicts that the performance of the refrigerator
with the combined phase shifter will be better than that with either of the two individual phase
shifters.
Figure 3. Cooling power of three versions of shifting phase.

Figure 4. Mass flow rate at warm end. Figure 5. Mass flow rate at cold end.
Temperature Oscillations
The temperature oscillations at the warm and cold ends of the pulse tube, respectively, are
shown in Figs. 6 and 7 for the three versions of the phase shifter. To make the results comparable,
the cooling temperature of each curve is the no-load temperature where the cooling power ap-
proaches zero. It can be seen that the difference between the average temperature of the gas at the
cold end of the pulse tube and the cooling temperature with the Pis.-Hol. or Hol. shifters is much
higher than that with the Pis. shifter. The fact that the average temperature of the gas at the cold end
of the pulse tube is much lower than the cooling temperature with the Pis.-Hol. or Hol. shifters at
low cooling power suggests a serious loss in the regenerator with these two phase shifters. As for
the warm end of the pulse tube, the difference between the average temperature and for the Pis.-
Hol. or Hol. shifters is also much higher than that with the Pis. shifter; this implies that the Pis.-Hol.
and Hol. shifters have a higher capacity for pumping heat.
Optimum Structure
From the above numerical analysis we can see that to obtain better performance, the cold
piston should be combined with an orifice-type phase shifter. In the following, interest is focused
on the new refrigerator with the Pis.-Hol. type phase shifter to determine its optimum structure. For
the Pis.-Hol. type phase shifter, Figures 8-11 show the influence on the cooling power at 8.0 K of
the original phase difference, the diameter of the orifice, the volume ratio of the piston chamber to
pulse tube (keeping the stroke constant and altering the diameter), and the volume ratio of the
regenerator to the pulse tube (keeping the diameter constant and altering the length). In Fig. 11,
represents the cooling power per unit gas mass at the warm end of the regenerator. From these
figures, we can determine the optimum shifter for the new refrigerator with the Pis.-Hol. phase
shifter, that is: 1) the relation between the phases of the pressure wave and piston movement is
Figure 6. Temperature wave at warm end. Figure 7. Temperature wave at cold end.
Figure 8. Cooling power as a function of Figure 9. Cooling power as a function of

the original difference of phase. the diameter of orifice.
Figure 10. Cooling power as a function of the Figure 11. Cooling power as a function of the
volume ratio of piston to pulse tube. volume ratio of regenerator to pulse tube.
reversed, 2) the diameter of the orifice is 0.25 mm, and 3) the volume ratio of the piston chamber
to pulse tube is 0.3. It is important to note that increasing the volume of the regenerator always
improves so there is no optimum volume ratio for the regenerator to pulse tube for but the
optimum volume ratio for exists, that is 5.3.
CONCLUSIONS
We have proposed a new type of hybrid refrigerator and developed a numerical simulation to
model its performance. Some useful results have been achieved as follows:
1. The behavior of the new type of hybrid refrigerator with only a cold piston as its phase shifter
is not as good as expected; the reason is that its phase shifting capacity is very limited. How-
ever, the performance is improved for the case where a cold piston is combined with another
version of a phase shifter, such as an orifice. A future possibility is to combine the cold piston
with a double-inlet type phase shifter.
2. The loss that most affects the performance of the hybrid refrigerator is the regenerator loss.
3. The optimum configuration for the new hybrid refrigerator with a cold piston and orifice as its
phase shifter is: 1) the relation between the phases of the pressure wave and piston movement
is reversed, 2) the diameter of the orifice is 0.25 mm, 3) the volume ratio of the piston chamber
to pulse tube is 0.3, and 4) the volume ratio of the regenerator to pulse tube is 5.3 (for cooling
power efficiency).
ACKNOWLEDGMENT
This activity was supported by the National Natural Science Foundation of China and Hui Guo
Ren Yuan Foundation of CAS.
REFERENCES
1. Li, R., “Great Process in Magnetic Regenerator Material and 4K G-M Cryocooler,” Proceedings of
ICCR’98, International Academic Publishers, Beijing (1998), pp. 67-73.
2. Wang, C., Thummes, G., and Heiden, C., “A Two Stage Pulse Tube Cooler Operating below 4K,”
Cryogenics, vol. 37, no. 3 (1997), pp.159.
3. Kays, W.M. and London, A.L., Compact Heat Exchanger, 2nd ed., McGraw-Hill, New York (1964).
Thermally Actuated Pulse Tube Cooler
Y. Matsubara, H. Kobayashi and S. L. Zhou

Atomic Energy Research Institute, Nihon University
Funabasbi, Chiba 274-8501 Japan
ABSTRACT
An isotope of Helium gas, is an attractive working gas for cryocoolers that strive to
efficiently provide cooling temperatures below 4 K. The drawback of this gas is that it is
extremely expensive. To minimize the total amount of gaseous a hybrid cycle has been
proposed. A single-stage pulse tube cooler using as the working gas and either a Stirling-
type or GM-type compressor system can provide a cooling temperature around 40 K, starting
from room temperature. A secondary cycle using can then be thermally attached to this
cold head to produce a secondary cooling temperature below 4 K. The operating frequency of
the secondary cycle should be lower than 2 Hz to prevent degradation of the performance of
the lower-stage regenerator.
In this study, a thermally actuated pressure wave generator driven between the tempera-
tures of 40 K and 300 K was selected; it has no difficulty in generating a pressure wave below
a frequency of 2 Hz. Workflow analysis calculations indicate that the total amount of
working gas required for this secondary cycle may be minimized by the use of a warm ex-
pander type of phase control.
INTRODUCTION
Additional performance improvement is still required of cryocoolers designed to cool LTC
devices such as low noise signal detectors. The long-term goal is a 4 K cryocooler that is more
reliable, more compact, generates minmal disturbance of the cooled devices, and has low cost and
low required input power. To satisfy these requirements, a pulse tube cooler operating at a low
cycle frequency by a non-mechanical compressor would be a good choice. There are many papers
on 4 K pulse tube coolers. However, all of these studies have been based on mechanical compressor
systems.1-22
Since there are no moving components in the low-temperature region of a pulse tube cooler,
the most significant feature to classify it is the gas compressor system that is located at room
temperature and generates the pressure oscillation. Thus, pulse tube coolers can be classified into
two different styles: GM-type, utilizing the valved compressor, and Stirling-type, utilizing the
valveless compressor. A multi-staged GM-type pulse tube cooler is able to achieve lower
temperatures, down to 4 K, because of its lower operating frequency. However the thermodynamic
efficiency is reduced due to the work losses within the rotary valve.

Figure 1. Basic configuration of thermally actuated pulse tube cooler.
In the case of a Stirling-type pulse tube cooler, the required pressure wave is provided by
a direct-coupled compressor piston without any valves. Therefore, the driving frequency can
be increased up to 50 Hz. For required temperatures above about 40 K, the thermodynamic
efficiency is also good because it has no irreversible components such as the valves. However,
it is difficult to get lower temperatures below about 10 K because of its higher frequency
operation.
As an alternative to the Stirling-type pulse tube cooler, the Vuilleumier (VM-type) pulse
tube cooler has been reported.23 The feasibility of applying the VM cycle to a 4 K pulse tube
cooler has also been studied.24 Further modifications of the VM-type 4 K pulse tube cooler are
discussed in this paper.
BASIC MODEL OF A THERMALLY-ACTUATED PULSE TUBE COOLER

A hybrid pulse tube cooler for cooling temperatures below 4 K is schematically illustrated
in Fig. 1. It consists of two cycles thermally coupled at the cold heat exchanger (f); this allows
for a minimum usage of gas in the secondary cycle, or a different operating pressure or
frequency to be used. The operating frequency of the secondary cycle should be lower than 2
Hz to produce efficient cooling performance below 4 K. A single-stage pulse tube cooler (a ~
d) provides a cooling temperature around 40 K at the cold heat exchanger (f). This portion of
the cooler is the upper stage or primary cycle, and can be driven by either a Stirling-type or a
GM-type compressor system. The secondary cycle is pre-cooled by the primary cycle and is
driven by the thermal compressor (e ~ h) similar to that of the VM cycle. Here the thermal
compressor also rejects its heat to the primary cycle at the cold heat exchanger. Since pulse
tube coolers for the primary cycle are now commercially available, this paper focuses on the
secondary cycle.
Figure 2 highlights the basic function of the thermal compressor. A warm expander was
selected as the phase shifter. When the warm heat exchanger is heated to a temperature, and
the cold heat exchanger is cooled to a temperature, by the primary cycle, the movement of
the displacer generates a pressure wave. When the displacer moves from the warm end to the
cold end, the cold gas will be displaced to the warm end through the warm regenerator, so it is
heated to the temperature as a result, the pressure within the closed volume increases. The
reverse movement of the displacer decreases the pressure in the same way. Therefore, the back
and force movement of the displacer generates an oscillating pressure. However, if there is no
pulse tube section, no work is generated. If there is a work receiver such as a warm expander
or an orifice, then work is done and flows in the direction of the arrow indicated in the figure.
THERMALLY ACTUATED PULSE TUBE COOLER 275
Figure 2. Energy flow map for thermally driven pulse tube cooler.
Figure 3. Basic model for workflow analysis of DPT model.
This situation can be explained by an energy flow map. The work is generated within the
warm regenerator by consuming the heat flow of the opposite direction. The generated work is
transferred back to the cold end of the warm regenerator through the displacer. Most of the
work flows to the pulse tube cooler section; however a small fraction of workflow to the warm
regenerator is required, because the work amplification within the warm regenerator is limited
due to the limited temperature ratio of and according to the 2nd law of thermodynamics.
In general, this warm regenerator has a limited heat transfer surface and flow friction. The
limited heat transfer surface generates the enthalpy flow, which increases the required cooling
capacity of the primary cycle at the cold heat exchanger. The flow friction in the warm regen-
erator generates a pressure drop and decreases the work amplification, which results in addi-
tional mechanical work being required to move the displacer. Therefore, the design of the
warm regenerator is very important to make the system efficient.
A simplified workflow analytical method for the double inlet pulse tube (DPT) model is given
in Fig. 3. Sinusoidal movement of the displacer gives the variable volume, and The mass
Sow at the orifice, bypass valve, warm regenerator, and cold regenerator are given as
and respectively. It is assumed the flow friction within the regenerator is generated at the
middle of the regenerator, exclusively. With this assumption, four different pressure equations are
obtained as and The equations used in this analysis are as follows:
and are the swept volume at the hot end and the cold end of the displacer.
and are the void volume at the hot end and the cold end heat exchangers. is the angular
velocity of the cycle. are the flow coefficients at the warm regenerator, cold
regenerator, bypass valve and the orifice valve respectively. The equivalent volume at the both
end of the pulse tube, are obtained by assuming a gas piston having the variable volume
within the pulse tube. The PV work and workflow at each interesting point is solved by,
where R is the gas constant as an ideal gas (2078 for He4 and 2757 for He3). jj is the
number of time division in a cycle used in this numerical analysis. The enthalpy flow through the
warm regenerator is given by,
where is the inefficiency of the regenerator and was given as a constant value.
CALCULATED RESULT OF DPT AND WEPT MODEL

The calculated result of the case of double inlet method is given in Fig. 4. Pulse tube inner
diameter is 10 mm and the length is 350 mm. Cv value of orifice and bypass for double inlet are
fixed near their optimum opening rates and were preliminary calculated as 0.008 for orifice and
0.048 for bypass. The reservoir volume is 1 liter. The cold regenerator located between 4 K and
40 K is 17 mm in diameter and 135 mm in length. The porosity and the hydraulic diameter are
fixed to 0.4 and 0.04 mm respectively. The thermal compressor is operated between 40 K and the
room temperature 300 K. The warm regenerator inner diameter is 25 mm and the length is 100
mm. The mean pressure is 1.3 MPa and driving frequency of the displacer is 1 Hz. The hydraulic
diameter of the packing material of the warm regenerator is changed to find out the optimum
condition under the constant porosity of 0.678. At each calculation, is changed until the
cold end workflow at 4 K becomes 1 watt.
The result indicates the workflow at the Va (Wa in Fig. 4 (a)) is almost constant but it
quickly increases at dense mesh region. The enthalpy flow through the warm regenerator <H>r
decreases with increasing mesh number. The sum of these two becomes the heat flow (Q) and is
the same as the minimum required cooling capacity at 40 K to keep this temperature constant.
Here the enthalpy flow through the cold regenerator is neglected. We found the minimum heat
Figure 4. Calculated result of DPT model.
flow is 22 watts when the hydraulic diameter of the warm regenerator is around 0.07 mm. The
required displacer stroke volume is about
The workflow Wa separates into two directions, to the cold and warm regenerator. The work
flowing to the warm regenerator (Wrl in Fig. 4(b)) is amplified and flows out of the warm end of
the warm regenerator, and then separates again into two directions, to the bypass and to the warm
end of the displacer (Wh). Therefore the additional mechanical work, which is the difference of
Wa and Wh, is required to move the displacer.
In this example, the pressure ratio up to 1.9 was generated, and it decreases with the
increase of the stroke volume. However the use of higher pressure ratio range should be avoided
because the required cooling capacity at 40 K also increases for same cooling capacity at 4 K.
Similar calculation has been done for the case of the warm expander pulse tube cooler
(WEPT). A warm expander with of the swept volume and with a phase lead of 70
degrees over the displacer volume (Va) is used instead of the double inlet phase shifter system.
All other sizes and driving conditions are the same with the previous double inlet method. The
result is shown in Fig. 5 and it indicates that the most of the important parameters are not a
strong function of the warm regenerator mesh size. The minimum required cooling capacity at 40
K reduced to 14 watts. And the stroke volume of the displacer also reduced to Additional
mechanical work is negligibly small, because the work used for the double inlet line is no longer
required in this case.
Figure 5. Calculated result of WEPT model.

Figure 6. Variation of thermal compressor for 4 K pulse tube cooler.
Figure 7. Gas movement within the pulse tube and work transfer tube.
VARIATION OF THERMAL COMPRESSOR FOR 4 K PULSE TUBE COOLER

If the displacer is removed from the cold section to the room temperature, it becomes the
configuration as shown in Fig. 6 (a). Here the tube without the solid displacer is called as the
work transfer tube, because it mainly transfers the work to the lower temperature with minimum
transfer of the heat flow. The calculated result indicates the required cooling capacity at 40 K is
18.7 watts, which is somewhat between the DPT and WEPT of the previous case. The volume of
the work transfer tube is and the stroke volume of the displacer is which is
slightly larger than the previous result of DPT. However, this method has no moving parts in its
cold section, and we can separate all of the moving parts from the cold section by means of
flexible tube. Similar calculation for the double inlet model has been done. However it requires
the cooling capacity of 31 watts at 40 K, and is not comparable with the other methods.
Fig. 7 indicates the phase difference of the pressure and the gas pistons of the warm
expander method. Both of the gas piston length of the pulse tube and the work transfer tube are
almost constant within a cycle similar to the solid piston. This result is caused by the lower
pressure ratio. The PV diagrams plotted from this gas trajectory at Va and Ve indicate that this
part of the cycle represents the Stirling cycle operating between 40 K and 4 K, although the
required cooling capacity at 40 K is increased to 14 watts from 10 watts for the ideal Stirling
cycle.
Another variation of the thermal compressor is schematically given in Fig. 6 (b). It consists
of a pressure wave generator operating between 900 K and 300 K, a phase shifter and the work
transfer tube. In this particular case, the work transfer tube can be replaced with the regenerator
and it becomes an ordinary thermally precooled single stage pulse tube cooler.
Most of the example calculations in this study are based on 1 watt workflow at 4 K for He4,
and the cooling capacity is reduced to about 20 % due to the non-ideal gas properties of He4. In
the case of He3, this reduction is not clear so far, because the thermal transport properties of He3
are not available for us.
From the viewpoint of minimizing the total amount of the He3, WEPT with cold displacer
will be the best, however, the WEPT with warm displacer seems to be the best for the application
adaptability.
CONCLUSION
1. Detail of simplified workflow analysis of the thermally actuated pulse tube cooler was
given. It indicates the warm expander method is the best choice from the both viewpoints of
minimizing the amount of enclosed gas and the required cooling capacity at the precooling
temperature. Therefore this method could be applied for He3 pulse tube cooler providing the
cooling temperature below 4 K efficiently.
2. The work transfer tube has been introduced. The function of this tube is very similar to
the pulse tube, which transfers the work from end to end with minimum heat pumping effect.
Only the difference is the direction of the workflow. This tube transfers the work from high
temperature to low temperature, which is the reverse of pulse tube. Minimizing the heat flow
through the work transfer tube is the most important requirement.
REFERENCES
1. Y.Matsubara, J.L.Gao, K.Tanida, Y.Hiresaki and M.Kaneko, “An Experimental and Analytical
Investigation of 4 K Pulse Tube Refrigerator”, Proceedings of 7th International Cryocooler
Conference, (1993), pp. 166-186.
2. J.L.Gao and Y.Matsubara, "4 K Pulse Tube Refrigeration", Proceedings of 4th JSJS on Cryocoolers
and Concerned Topics, (1993), pp. 69-73.
3. J.L.Gao and Y.Matsubara, "Experimental Investigation of 4 K Pulse Tube Refrigerator", Cryogenics,
vol. 34, (1994), pp. 25-30.
4. Y. Matsubara and J.L. Gao, “Novel Configuration of Three-stage Pulse Tube Refrigerator for
Temperatures below 4 K”, Cryogenics, Vol.34 (1994) 259.
5. Y. Matsubara and J.L. Gao, “Multi-staged Pulse Tube Refrigerator for Superconducting Magnet
Applications”, Proceedings of the Fifteenth International Cryogenic Engineering Conference, (1994)
pp. 155.
6. Y.Matsubara and J.Gao, "Multi-Stage Pulse Tube Refrigerator for Temperatures below 4 K",
Cryocoolers 8, Edited by R.G.Ross.Jr, Plenum Press. (1995), pp. 345-352.
7. K.Tanaida, J.L. Gao, N.Yoshimura and Y. Matsubara, “Three-Staged Pulse Tube Refrigerator
controlled by Four-Valve Method”, Advances in Cryogenic Engineering, Vol. 41, Plenum Press,
(1996), pp. 1503-1509.
8. J.L. Gao, Y. Hiresaki and Y. Matsubara”, “A Hybrid Two-Stage Refrigerator operated at
Temperatures below 4 K”, Advances in Cryogenic Engineering, Vol. 41, Plenum Press, (1996), pp.
1495-1502.
9. J. L. Gao and Y. Matsubara, “An Inter-Phasing Pulse Tube Refrigerator for High Refrigeration
Efficiency”, Proc. of Sixteenth ICEC (1997) pp. 295-298.
10. K. Tanida, J. L. Guo Y. Hiresaki and Y. Matsubara, “Performance of the Hybrid Two-Stage Pulse
Tube Refrigerator”, Proc. of Sixteenth ICEC (1997) pp. 303-306.
11. Y. Ohtani, G. R. Chandratilleke, H. Nakagome, N.Yoshimura, Y. Matsubara, M. Narita and H. Okuda,
“Development of a Three Stage Pulse Tube Refrigerator”, Proc. of the Fifth Japanese-Sino Joint
Seminar (1997) pp. 118-122.
12. Yoshimura, N., Zhou, S.L., Matsubara, Y., Chandratilleke, Y, Ohtani, Y, Nakagome.H., Okuda, H.,
and Shinohara, S., “Conceptual Design of Space Qualified 4 K Pulse Tube Cryocooler”, Cryocoolers
10, KA/Plenum Press (1999) pp. 221-226.
13. Yoshimura, N., Zhou, S.L., Matsubara, Y, Chandratilleke, Y, Ohtani, Y, Nakagome.H., Okuda, H.,
and Shinohara, S., “Performance Dependence of a 4 K Pulse Tube Cryocooler on Working Pressure”,
Cryocoolers 10, KA/Plenum Press (1999) pp. 227-232.
14. C. Wang, G. Thummes, C. Heiden, “Effects of DC gas flow on performance of two-stage 4 K pulse
tube coolers", Cryogenics, vol. 38, no. 6 (1998), pp. 689-695.
15. G. Thummes, C. Wang, C. Heiden, “Small scale liquefaction using a two-stage 4 K pulse tube
cooler”, Cryogenics, vol. 38, no. 3 (1998), pp. 337-342.
16. C. Wang, G. Thummes, C. Heiden, “Experimental study of staging method for two-stage pulse tube
refrigerators for liquid temperatures”, Cryogenics, vol. 37, no. 12 (1997), pp. 857-863.
17. G. Chen, L. Qiu, J. Zheng, P. Yan, Z. Gan, X. Bai, Z. Huang, “Experimental study on a
double-orifice two-stage pulse tube refrigerator”, Cryogenics, vol. 37, no. 5 (1997), pp. 271-273.
18. M.Y. Xu, A.T.A.M. De Waele, Y.L. Ju, A pulse tube refrigerator below 2 K, Cryogenics, vol. 39, no.
10 (1999), pp. 865-869.
19. A. von Schneidemesser, G. Thummes, C. Heiden, “Generation of liquid helium temperatures using a
lead regenerator in a GM precooled pulse tube stage”, Cryogenics, vol. 40, no. 1 (2000), pp. 67-70.
20. A. von Schneidemesser, G. Thummes, C. Heiden, “Performance of a single-stage 4 K pulse tube
cooler with neodymium regenerator precooled with a single-stage GM refrigerator, Cryogenics, vol.
39, no. 9 (1999), pp. 783-789.
21. G. Thummes, S. Bender, C. Heiden, “Approaching the lambda line with a liquid nitrogen
precooled two-stage pulse tube refrigerator”, Cryogenics, vol.36, no. 9 (1996), pp. 709-711.
22. S.L. Zhou, G. Thummes, and Y. Matsubara, “Experimental Investigation of Loss Mechanisms in a
4 K Pulse Tube”, Advances in Cryogenic Engineering, Vol. 45, Plenum Press, (to be published).
23. Kaneko, M. and Matsubara, Y, “Thermally Actuated Pulse Tube Refrigerator”, Cryocoolers-5,
Naval Postgraduate School, USA (1988), pp. 103-112.
24. Y. Matsubara and S.L. Zhou, “Feasibility Study of Applying Thermal Compressor to 4 K Pulse Tube
Cooler”, Proceedings of the eighteenth International Cryogenic Engineering Conference, (2000), (to
be published).
Investigation of Helium and Nitrogen Mixtures
in a Pulse Tube Refrigerator
Z.H. Can, G.B. Chen, G.Thummes† and C.Heiden†

Cryogenics Lab. Zhejiang University, Hangzhou, 310027, P.R.China
†
Institute of Applied Physics, University of Giessen, 35392, Germany
ABSTRACT
Normally, the working fluid used in regenerative cryocoolers is not condensed at any point in
the working cycle. Helium is the popular working medium for such systems, rather than a gas
mixture because of its excellent thermodynamic and transport properties. However, our theoretical
analyses indicate that pure helium is not the best working fluid for regenerative refrigerators near
80 K, either for the best cooling power, or for the best coefficient of performance (COP).
Motivated by the advantages of pulse tube refrigerators, which have no moving parts in the
cold end, this paper describes an experimental investigation on a two-component, multi-phase he-
lium and nitrogen gas mixture in a single-stage pulse tube refrigerator. The experimental results
show that the COP and the cooling power can be increased to some extent near 80 K with the use of
two-component gas mixtures with less than 25% nitrogen. A relatively stable temperature platform
at the triple point of nitrogen (63.15K), which is independent of the nitrogen fraction, is obtained
when the cooling power is below 7W.
INTRODUCTION
Investigations of pulse tube refrigerators have made great progress over the past few years
since Mikulin proposed the orifice type configuration in 1985. With the application of various
novel phase shifters and construction improvements, the refrigeration temperature has continu-
ously dropped and the thermodynamic efficiency has continuously improved. So far, refrigeration
temperatures as low as 20 K have been reached with a single-stage pulse tube refrigerator.1 In
addition, cooling temperatures below 4.2 K have been obtained by some two-stage pulse tube
refrigerators,2,3 and a three-stage pulse tube refrigerator with has reached a temperature lower
than the point of 1.78K.4 The results confirm that refrigeration performance can be in-
creased if the appropriate refrigerants or mixed fluids are used in the corresponding temperature
regions. Due to their advantages of simple construction and no moving parts in the cold end, pulse
tube refrigerators provide feasible conditions for using a multi-phase fluid as the refrigerant.
One may imagine that, if a mixture of helium and nitrogen were used as the refrigerant in a
pulse tube refrigeration system, the nitrogen will liquefy when the refrigeration temperature
reaches the liquefaction point of nitrogen. In such a system, refrigeration performance could be
higher than that using pure helium, since the latent heat of the phase change of liquid nitrogen
could be used. A number of researchers are interested in utilizing mixtures as the refrigerant for

Figure 1. Schematic of a GM-type single-stage pulse tube refrigerator.
regenerative refrigerators.5,6 The possibility of using mixtures of helium and nitrogen to improve
the refrigeration performance of pulse tube refrigerators was also analyzed in our previous paper.7
However, few reports of theoretical studies or successful experimental results have been published
so far. An experiment with helium and neon mixtures as the refrigerant in a two-stage GM cooler
was described by P C Mcdonald.8 Although no benefit in cooling power was obtained from the
mixture of helium and neon, compared with pure helium in the refrigerator, a relatively stable
temperature was obtained for cooling superconducting devices at 26K, close to the triple point of
neon.
In order to increase the refrigeration performance of pulse tube refrigerators near 80K, the
authors have been exploring the possibility of using mixture gases, both theoretically9 and experi-
mentally. In this paper, we present some experimental results using helium and nitrogen mixtures
in a pulse tube refrigerator.
EXPERIMENTAL ARRANGEMENT
The experimental setup used for measuring refrigeration performance of mixture fluids con-
sists of the refrigeration system, vacuum system, measuring system, and mixture preparation sys-
tem. Figure 1 shows a sketch of the refrigeration system, which consists of a helium compressor
(Leybold, RW2), a GM-type rotary valve, and a single stage pulse tube refrigerator. Dimensions of
the regenerator and pulse tube are and respectively. A copper
tube for water cooling is wrapped and soldered around the hot end of the pulse tube. The pulse tube
refrigerator is equipped with a double-inlet configuration via two needle valves (Nupro, type M,
25div/turn) and a 0.5 liter reservoir. The temperature profile along the pulse tube wall is measured
by Pt100 resistance thermometers in Figure 1). Piezoelectric pressure sensors (Siemens,
type KPY 46R) are used for monitoring the dynamic pressures at the hot ends of the regenerator
and pulse tube as well as in the reservoir The cooling power is measured by a cali-
brated resistive heater, PBH-100 attached to the cold-end heat exchanger. The input power of
the compressor is measured by EKM 265. Pressure and temperature recordings are accomplished
by means of a data acquisition system controlled by a personal computer.
EXPERIMENT PROCEDURE
The test pulse tube refrigerator was operated with a charge pressure of 1.7 MPa at room tem-
perature (pressures quoted throughout this paper are absolute values), and at a frequency of 2 Hz.
Mixture Preparation. Dolton’s law of partial pressures may be used for preparing a mixture
of helium and nitrogen. For example, a mixture of 80% helium and 20% nitrogen can be prepared
by firstly filling the system with helium gas to 1.36 MPa, then filling nitrogen gas up to a total
pressure of 1.7 MPa. With this method: 97/3, 94/6, 92/8, 88/12, 83/17, 80/2 and 75/25 % helium/
% nitrogen gas mixtures were prepared for the experiment.
INVESTIGATION OF He AND MIXTURES IN A PT COOLER 283
Comparison Criterion. A proper comparison standard is particularly important when com-

paring the refrigeration performance of a mixed fluid with that of pure helium. Both the orifice
valve and double-inlet valve settings for the refrigeration at 80K, and for the lowest cooling tem-
perature (about 30K), are different when pure helium is used as the working fluid. In addition, it is
difficult to adjust the valve setting to a temperature below the condensation point of nitrogen due to
the multi-phase fluid that exists when a mixed fluid is used. Therefore, the optimized settings for
each refrigerant in this paper were fixed for a condition of 9W of cooling at about 80K. Further-
more, since the optimized double-inlet valve setting was found to be relatively insensitive to the
helium/nitrogen mixture ratio, the same double-inlet valve setting was adopted for all cases includ-
ing pure helium; this simplified the experimental procedure. In short, the experimental procedure
was divided into two steps. The first was to optimize the orifice valve opening for a 9W cooling
load at 80K. The second was to measure the refrigeration load-line (heat load vs. cooling tempera-
ture) with the optimized orifice setting.

Cooling Power and Coefficient of Performance
Figure 2 shows the experimental results of cooling power (a) and COP (b) for the pulse tube
refrigerator with mixtures of helium and nitrogen. The figure also gives the curves for pure helium
as a comparison. We can see that both the cooling power and COP with a mixed fluid are higher
than that of pure helium for various nitrogen fractions up to 25%. When the cooling temperature is
higher than 70K, the degree of performance improvement depends on the composition of the mix-
ture. It can be seen from Figure 2 that when the cooling power is lower than 7W, all curves in this
figure are concentrated at approximately the triple point of nitrogen (63.14K). An approximate
isothermal line near 64K is obtained under 7W of cooling power, which is independent of the
fraction of nitrogen Because of the advantage of no moving parts in the cold space of the
pulse tube, the isothermal working platform creates a very constant temperature for devices to be
cooled, and the system is resistant to degradation caused by particles of liquid or solid nitrogen that
are carried into the pulse tube by the working gas flow.
Figure 3 shows the relationship between specific cooling power and COP versus composition
of the mixture at 80K. When the fraction of helium is greater than about 85%, the refrigeration
performance using a mixture is better than that with pure helium. The maximum gains obtained in
the experiments include a 3.85% increase in cooling power with a helium fraction of 88%, and a
4.05% increase in COP with a helium fraction of 97%.
Figure 2. Cooling capacity (a) and COP (b) versus temperature.

Figure 3. Specific cooling power and COP versus fraction of mixture.
Figure 4. Phase change phenomena of 3% nitrogen gas mixture.
Two transition temperatures for helium/nitrogen mixtures can be observed in Figure 2. This
can be explained as follows. During the cooling-down process, nitrogen, as the secondary refriger-
ant in the mixture, will be condensed at its liquid point, which depends on the partial pressure of
nitrogen and the fraction nitrogen; this is the vapor-liquid transition temperature. Furthermore, the
liquid-solid transition temperature will occur at the triple point of nitrogen, 63.14K, when the liquid
nitrogen freezes. After that, the nitrogen is in a vapor-liquid-solid multi-phase state. If the nitrogen
fraction is much less, all of the nitrogen will be solidified.
Figure 4 shows the phase change phenomena of a 3% nitrogen mixture during the cooling
process. It indicates that the liquid-solid phase change occurs around 63K. Then another tempera-
ture platform appears on the cooling curve near 58K. Finally, after the nitrogen fraction in the
mixture is completely solidified and the refrigerant is almost pure helium, the refrigeration process
gradually approaches a minimum point of 33K or so.
Figure 5 and Table 1 give experimental results of the vapor-liquid phase change temperatures
which are clearly in accordance with the partial pressure of nitrogen in the mixture. However, the
liquid-solid change temperature for all tests with various mixture fractions (3-25%) is almost the
same as 63-65K; this is because the liquid-solid phase change is caused by the triple point phenom-
enon of nitrogen.
Figure 5. Vapor-liquid phase change temperature in helium-nitrogen mixture.
Temperature Profile
The locations of the eight temperature measuring points in the experiment were presented in
Figure 1. Accordingly, the temperature at the hot end of the pulse tube is 290K, due to the water
cooling heat exchanger, while the hot end of the regenerator is at room temperature (305K).
It is useful to define a dimensionless relative temperature as
where T is the variable value to be measured, and is the cold end temperature of the pulse tube;
is 290K for the pulse tube and 305K for the regenerator.
Figure 6 shows the relative temperature profile during the refrigeration process with pure he-
lium versus relative position along the pulse tube or regenerator. We can see from Figure 6(a) that
the temperature distribution profile along the regenerator shifts to the outside with increasing heat
Figure 6. Temperature profile with position along regenerator (a) and pulse tube (b).
Figure 7. Temperature profile versus heat load at solidification process.
Figure 8. Temperature profile versus heat load at complete solidification state of nitrogen.
load. Figure 6(b) gives the same curve shape in the opposite direction along the pulse tube. But the
latter seems closer together. This is in accordance with the theoretical result given by de Waele for
explaining the non-ideal gas effect.11 In conclusion, the temperature curves along both the pulse
tube and the regenerator are expanded in the direction of increasing heat load.
Figure 7 and Figure 8 show different temperature profiles during the refrigeration process with
a mixture fluid. It can be divided into three steps. The first step, when the cooling temperature drops
to above the liquid-solid phase change point, the curve shape appears the same as that with pure
helium. That is, both the temperature profile along the pulse tube and regenerator expand in the
direction of increasing heat load, as shown in Figure 7(a). The second, when the refrigeration tem-
perature is at the solidification point of nitrogen, the curves along both the pulse tube and regenera-
tor appear opposite to their state with pure helium, and shift to the inside with increasing heat load,
as shown in Figure 7(b). The third, when the refrigeration temperature is lower than the phase
change point (63.15K), the curves again show a tendency to the same shape as with pure helium, as
shown in Figure 8. This is because the working fluid becomes pure helium again, as most nitrogen
in the mixture has been completely solidified.
Effect of Orifice and Double-inlet Valves

Figure 9 shows the relationship between optimized orifice valve setting and helium frac-
tion. The optimized valve opening becomes larger with an increase in nitrogen fraction in the mix-
Figure 9. Optimized orifice valve setting versus fraction of gas mixture.
Figure 10. Refrigeration temperature versus orifice valve setting.
ture. For example, the refrigeration temperature in Figure 10 gradually drops down with an increase
of the orifice opening when 12% nitrogen is used in the mixture. The minimum refrigeration tem-
perature, 79.7K, was obtained with a 8.94W heat load and an orifice opening of 70 div. In contrast,
with a pure helium refrigerant, the optimized orifice opening is 57 div (in Figure 9) with almost the
same heat load.
Figure 11 shows coldend temperature versus double-inlet valve opening for a pure helium
refrigerant; the best opening of the double–inlet valve is 95 div. Similar data for sensitivity to
double-inlet valve opening for a 25% nitrogen mixture are shown in Figure 12. These data indicate
that the optimized opening is also 95 div with almost the same heat load. It seems that the same
double-inlet valve setting can be adopted for both cases. This fact simplifies the experimental pro-
cedure.
CONCLUSION
Experiments using a number of helium and nitrogen mixtures in a GM-type single-stage pulse
tube refrigerator have been carried out. The experimental results show that both the coefficient of
performance (COP) and the cooling power can be increased to some extent when a mixed helium
refrigerant is used with a nitrogen fraction of up to 25%. A cooling temperature platform near
63.14K has been observed for the tested fractions of nitrogen. The vapor-liquid and liquid-solid
transition temperatures of nitrogen are also observed during the pulse tube cooling process. The
Figure 11. Refrigeration temperature versus double-inlet valve setting for pure helium.
Figure 12. Double-inlet valve setting for 25% nitrogen mixture.
optimized double-inlet valve settings are the same for both pure helium and the helium/nitrogen
mixed fluid, while the optimized orifice valve setting depends on the composition of the mixture.
ACKNOWLEDGMENT
The project is financially supported by the National Natural Sciences Foundation of China and
the Deutscher Akademischer Austauschdienst (DAAD).
REFERENCES
1. Ishizaki, Y., et al., “Experimental Performance of Modified Pulse Tube Refrigerator Below 80K.
Down 23K,” 7th Intl. Cryocooler Conf. Proceeding, Part 1 (1993), p. 140.
2. Wang, C., Thummes, G., and Heiden, C., “A Two-stage Pulse Tube Cooler Operating Below 4K,”
Cryogenics, vol.37 (1997), p. 159.
3. Chen, G.B., Qiu, L.M., Zheng, J.Y., Yan, P.D., Gan, Z.H., Bai, X., and Huang, Z.X., “Experi-
mental Study on a Double-orifice Two-stage Pulse Tube Refrigerator,” Cryogenics, vol. 37 (1997), p.
271.
4. Xu, M.Y., De Waele, ATAM., and Ju, Y.L., “A Pulse Tube Refrigerator below 2K,” Cryogenics,
vol. 39, (1999), p. 865.
5. Walker, G., “Stirling-Cycle Cooling Engine with Two-Phases, Two-Components Working Fluid,”
Cryogenics, vol. 8 (1974), p. 459.
6. Patwardhan K.P. and Bapat S.L., “Cyclic Simulation of Stirling Cycle Cryogenerator Using Two
Component Two Phase Working Fluid Combinations,” Proceedings of ICCR’98, Hangzhou, Interna-
tional Academic Publishers (1998), p. 397.
7. Zhao L., Chen, G.B., and Yu, J.P., “The possibility use with Gas Mixtures in Pulse Tube Refrigera-
tion”, Cryogenics Engineering, 5 (1996), p. 15.
8. McDonald, P.C., “Self-regulating Temperature Control of a Gifford-McMahon Refrigerator for Po-
tential Use with Neon in High-Tc Power Applications,” Superconductor Science & Technology, 11
(1998), p. 817.
9. Chen, G.B., Gan, Z.H., Thummes, G., and Heiden, C., “Thermodynamic Performance Prediction
of Pulse Tube Refrigeration with Mixture Fluids,” Cryogenics, to be published.
10. De Waele ATAM., Xu, M.Y., and Ju, Y.L., “Nonideal-gas Effect in Regenerators,” Cryogenics,
vol.39 (1999), p. 847.
Pulse Tube Refrigeration with a Combined
Cooling and Freezing Cycle
for HTSC Devices
Guobang Chen, Zhihua Gan, Limin Qiu,

and Jianping Yu
Cryogenics Laboratory
Zhejiang University
Hangzhou 310027, China
ABSTRACT
Based on the analysis of a modified Brayton cycle, a combined cooling and freezing cycle
is introduced for pulse tube refrigeration with a binary refrigerant. A novel configuration of a
pulse tube refrigerator with a frozen cryogen accumulator is proposed as an example for pos-
sible applications. Some experimental phenomena explaining the working principles are pre-
sented.
INTRODUCTION
In recent years, substantial progress has been made in the study of high- SQUID-based
heart scanners cooled by small Stirling cryocoolers that are cryogen-free.1,2 When the refrig-
erators are running, a magnetic noise level as high as from 10 Hz to 100 Hz has
been measured in SQUID magnetometer tests. When the refrigerators are switched off, the
noise level drops to about (26 times lower), which is very close to the white-noise
level of a SQUID cooled by liquid nitrogen.
Additionally, recent investigations3-5 have been conducted on the use of mixed refriger-
ants to improve the refrigeration performance of pulse tube refrigerators in the 80 K region.
Since pulse tube refrigerators bear the promise of high reliability and economic operation, the
authors propose a novel configuration of a pulse tube refrigerator with a freezing cryogen
accumulator for some particular applications. One may imagine that the freezing cryogen ac-
cumulator may provide the necessary refrigeration for a certain period of time after the pulse
tube refrigerator with two-phase binary mixture is switched off.
To explore the possibility of providing such a pulse tube refrigerator, a modified Brayton
cycle with a binary mixture of nitrogen and helium was conceived and is described in detail.
Then, a combined cooling and freezing two-phase system was added to the pulse tube refrig-
erator via a freezing cryogen accumulator. Finally, some experimental phenomena explaining
the working principles of the novel pulse tube refrigerator are presented.

Figure 1. T-S diagram for the pulse tube refrigeration

cycle with two-isentropic and two-isobaric processes.
MODIFIED BRAYTON REFRIGERATION CYCLE

In order to predict the refrigeration performance of pulse tube refrigerators with a binary
mixture, the authors recommend a practical model of the refrigeration cycle which can meet
the requirements of the thermodynamic analysis. In the case of pulse tube refrigeration, the
following assumptions must be made:
1. The processes of compression and expansion of the working refrigerant that result in refrigera-
tion are adiabatic processes rather than isothermal ones
2. The rejection of the heat of compression into the coolant in the after-cooler is an isobaric
process
3. The heat transfer process occurring at the cold end of the pulse tube is also isobaric; thus, the
refrigeration effect of the pulse tube refrigerator is absorbed over a temperature range rather
than at a constant temperature
4. Heat transfer processes in the regenerator are carried out under constant pressure conditions.
The enthalpy imbalance between the compressed warm stream and the expanded cold stream in
the regenerator must also be taken into consideration.
Figure 1 shows the T-S diagram of this assumed pulse tube refrigeration cycle with two
isentropic and two isobaric processes. The working refrigerant leaving from the cold head heat
exchanger in state 1 enters the regenerator, and then is warmed up at a constant
pressure to state 2 ambient temperature). It is compressed adiabatically in the
compressor from state 2 to 3 and then passes through the after-cooler to reach state 4
The working fluid then enters the regenerator and is cooled at a constant
pressure to state 5 It is then expanded adiabatically at the cold end of the
pulse tube to its lowest refrigerating temperature in state 6 which is lower than
Tc. Finally, the refrigerant absorbs the cooling load at constant pressure P1 in the cold head
heat exchanger (temperature and returns to state 1, finishing a complete cycle.
The cycle shown in Fig. 1 is essentially a Brayton cycle. However, a conventional Brayton
cycle utilizes a recuperative heat exchanger, and turbine machines are generally used as the
compressor and expander in the cycle. In contrast, in the described pulse tube cycle, the recu-
perative heat exchanger of slotted plates is replaced by a regenerator in the pulse tube refrig-
erator, and the turbine expander is replaced by an expansion space within the pulse tube.
Thus, the thermodynamic cycle for a pulse tube refrigerator can be considered as a varia-
tion of the Brayton cycle, and will be referred to as a modified Brayton cycle in this paper.
PT WITH A COMBINED COOLING AND FREEZING CYCLE 293
Coefficient of Performance
Coefficient of performance (COP) of the modified Brayton cycle, which consists of two
isobaric and two adiabatic processes, can be expressed as follows.
The heat rejected to ambient is
The heat removed from the cold end heat exchanger at the constant pressure namely the
refrigeration effect of the system, is
The coefficient of performance of the system then can be expressed as follows
The above analysis is based on enthalpy-balanced condition in the regenerator. It must be

stressed that the phenomenon of unbalance enthalpy flow in the regenerator cannot be neglected
in the computation process. The heat rejected by compressed working refrigerant in the
regenerator is
The heat absorbed by the expanded working refrigerant in the regenerator is
In an ideal case, For example, it is true when pure helium is used as the working
refrigerant at around 80K. However, for mixture fluids, enthalpy difference of the warmer
refrigerant flux could be greater or less than that of the cooler one in the regenerator. In this case,
the enthalpy deficit occurs. The enthalpy deficit may be positive or negative depending on the
fluids and operating parameters, and can be expressed as
The enthalpy difference is an additional heat load (if it is a positive value) or refrigeration
power (if it is a negative value) of the cycle, then the Eq. (2) is rewritten as
and Eq.(3) is turned into
Calculations have shown that the enthalpy deficit cannot be eliminated if mixture fluids are
used as the working refrigerants4.
Computed Results
In the following calculations, the ambient temperature the filling pressure of
working refrigerant 1.7 MPa and the pressure ratio of the refrigeration
system are assumed, so as to make a comparison with experimental results. The refrigeration
effect and COP with varied fraction of nitrogen in the mixture are computed according to
Eqs.(7) and (8). The results are shown in Figs.2 and 3. We can find that by using a mixture of
Figure 2. Theoretical cooling power versus helium fraction in the mixed fluid.
Figure 3. Calculated COP versus helium fraction in the mixed fluid.
10% nitrogen and 90% helium, the refrigeration effect and COP can be improved by 6.7%
and 9.5%, respectively, compared with that using pure helium. Obviously, the nitrogen-helium
mixture is a promising fluid for pulse tube refrigeration at around 80 K.
VAPOR-LIQUID TWO-PHASE REFRIGERATION CYCLE

The modified Brayton refrigeration cycle (Fig. 1) cannot provide two isothermal processes,
thus the heat transfer efficiency will be lower. In addition, the cooling power of a cold gas
refrigeration system is only contributed by the sensible heat of the refrigerant, which is much
smaller than its latent heat. These drawbacks may be eliminated by utilizing a vapor-liquid
two-phase refrigeration cycle. In such a cycle, both condensation and vaporization processes
occur at a constant pressure and temperature, respectively. Thus more refrigeration effect can
be expected from the phase change process. The working fluid gives its latent heat of evapora-
tion to the cold heat exchanger, while the refrigeration temperature does not change.
The thermodynamic process describing the vapor-liquid two-phase refrigeration cycle is
shown in Fig. 4 The working refrigerant leaving from the warm end of the regenerator at
state is compressed isentropically to state 3 rejecting heat isobarically to a
coolant in process 3-4. Then the compressed refrigerant enters the regenerator and is cooled to
the saturated vapor line at state 5 Process 5-6 is a condensation process in the two-
phase region. The saturated liquid then expands isentropically from 6 to 7. In the process 7-9,
Figure 4. Vapor-liquid two-phase refrigeration cycle
the liquefied refrigerant absorbs heat from the cold heat exchanger and evaporates. Gaseous
refrigerant further absorbs heat from state 9 to state 1 then enters the regenera-
tor, being heated to state 2 completing one cycle.
The vapor-liquid two-phase refrigeration cycle is widely used in general refrigeration pro-
cesses in which the critical temperature of the refrigerant is required to be higher than room
temperature so as to reject the condensation heat to the ambient. Otherwise, the heat of con-
densation will be an interior heat source in the system. In the present case associated with
obtaining refrigeration in the 80 K temperature region, the only refrigerant that may be used is
nitrogen. It is well known that the critical temperature of nitrogen (126.2 K) is much lower
than room temperature, thus the heat transferred in process 5-6 in Fig. 4 becomes a heat load
within the refrigerator itself. It offsets the effective cooling. In addition, the enthalpy deficit in
the regenerator is possibly so high that the efficiency and benefits of the two-phase cycle
become unfavorable. A combined cooling and freezing cycle with a binary refrigerant might
be more preferable for cryogenic refrigeration, as explained in the following.
COOLING AND FREEZING COMBINED CYCLE

The T-S diagram of the proposed cooling and freezing combined cycle is shown in Fig. 5. The
diagram consists of a sensible heat refrigeration cycle called the cooling cycle (1-2-3-4-5-6-1)
and a complementary latent heat refrigeration cycle called the freezing cycle (1-2-3-4-5-7-8-10-1).
Figure 5. Cooling and freezing combined cycle.

The working refrigerant is assumed to be a mixture of helium and another component, such as
nitrogen, which has a suitable boiling-point temperature for the required application. Helium
in the mixture enables sensible heat refrigeration to temperatures as low as 30 K using a single-
stage pulse tube refrigerator; it is referred to as the cooling medium. Nitrogen has a higher
boiling point and can be condensed into its liquid phase at the cold head temperature; it is
referred to as the freezing medium, and contributes latent heat refrigeration.
The working process of the cooling and freezing combined cycle includes the following
three steps:
1. The system firstly works in accordance with the modified Brayton refrigeration cycle when the
refrigeration temperature is higher than the boiling point of nitrogen. In this case, both helium
and nitrogen play the role of a cooling media. In fact, this is a pre-cooling process of the
system, called the sensible heat refrigeration process.
2. When the cooling temperature reaches the condensation temperature of nitrogen at a corre-
sponding partial pressure, the cycle turns into a vapor-liquid two-phase refrigeration cycle.
The nitrogen then plays the role of a freezing medium, making latent heat refrigeration. In
contrast, the helium continues as a cooling medium in the sensible heat refrigeration state.
When the cooling temperature is dropped down to the liquefaction point of nitrogen, the two-
phase state appears, and nitrogen is gradually condensed at the cold head. With continued
decrease of the refrigeration temperature, more and more liquid nitrogen will be obtained. This
is the cooling and freezing combined process.
3. The total pressure of the system may slightly decrease during step 2, since some fraction of the
nitrogen has been liquefied. As a result, the gaseous component of the system gradually be-
comes nearly pure helium gas. At this point, the system transitions to work again as a sensible
heat refrigeration cycle based on pure helium as the refrigerant. Hence, we can forget the influ-
ence of the transport properties of the freezing medium nitrogen in the binary mixture, which
are normally poorer than those of pure helium.
It is found that the cooling cycle can cool the system down to a temperature considerably
below the triple point of nitrogen, where the freezing medium nitrogen is in a state of super-
cooled liquid or even solid.
PULSE TUBE REFRIGERATOR WITH A FREEZING CRYOGEN ACCUMULATOR

Based on the above analysis, the authors propose a novel pulse tube refrigerator with a
liquid or solid cryogen accumulator as shown in Fig. 6. In the pre-cooling operation, the
system is firstly in a sensible heat refrigeration process with a mixed refrigerant of helium and
nitrogen. Then, the nitrogen in the mixture begins condensing as the cooling temperature reaches
its liquefaction point. The accumulator may accept the condensed fraction of nitrogen by means
Figure 6. Pulse tube refrigerator with a freezing cryogen accumulator.

Figure 7. Cooling power versus cooling temperature of pulse tube refrigerator
of the phase separator under the cold end of the pulse tube while the system operates in a
cooling and freezing combined process. The liquefied nitrogen in the accumulator can further
be solidified when the cooling temperature drops below its triple point. The liquid or solid
freezing medium contained in the accumulator provides additional cooling capacity associated
with the latent heat of evaporation or sublimation to meet the requirements of the application,
even if the machine is switched off. The time duration of the provided refrigeration depends
on the volume of the accumulator and the cooling load of the device to be cooled. Of course,
the volume of the accumulator must reflect the possible fraction of nitrogen in the mixture of
the system.
EXPERIMENTAL OBSERVATIONS
In experiments with a G-M type pulse tube refrigerator, cooling power versus composition
of the mixture was measured for mixtures of helium and nitrogen in which the mole fraction of
nitrogen varied from The test machine was a single-stage U-type pulse tube refrig-
erator with 2.0 kW input power. Figure 7 shows an example curve of cooling power versus
temperature for a mixture of 97% He and 3% compared with that for pure helium. The
cooling power of the mixture is higher than that of pure helium when the cooling temperature
is higher than the condensing temperature (71.8 K) at the corresponding partial pressure of
nitrogen in the mixture.
It is interesting that the liquid-solid phase change at 63.14 K is visible. The experimental
results indicate that the machine can provide a cooling power at an approximate temperature
platform around 63.14 K with nitrogen fractions up to 25%. We can also see in Fig. 7 that
solid nitrogen begins to appear when the cooling temperature drops below 63.14 K. The solidi-
fication process reaches completion when the cooling temperature drops to around 58 K. Fi-
nally, the refrigeration process, when the refrigerant is reduced to almost pure helium, gradu-
ally approaches a minimum point of 35 K or so. Obviously, the possibility of producing solid
nitrogen depends on the refrigeration ability of the cryocooler.
Figure 8 shows a comparison of temperature distributions along the pulse tube and regen-
erator for pure helium and a mixture of 3% nitrogen and 97% helium, respectively. For the
mixed fluid case, we can see from Fig. 8 that temperatures along about 30% of the length of the
pulse tube are almost the same as those at the cold end of the tube. This means that some
nitrogen in the mixture has been liquefied.
Figure 9 shows temperature variations at the cold end of the pulse tube during re-heating
after the machine was switched off. The pulse tube refrigerator was operated with a mixture of
12% nitrogen and 88% helium. In this example, the machine can keep the cooling temperature
close to 64 K for about 4.3 minutes with a cooling power of 200 mW. The capacity of the
Figure 8. Temperature distribution along pulse tube and regenerator.
Figure 9. Temperature variation of pulse tube refrigerator without accumulator in re-heating process.
accumulator, which is installed under the cold end of the pulse tube and can contain of
liquid or solid nitrogen, can be calculated. The results indicate that 2.0 minutes of dwell at
63.16 K can be achieved for a 200 mW cooling power input for each of solid nitrogen.
This means that 20 minutes of dwell can be achieved via the solid-liquid phase change at 200
mW cooling power for a accumulator. In conclusion, the pulse tube with the additional
accumulator may provide a rather steady temperature near 64 K for about 24 minutes with an
applied cooling load of 200 mW.
CONCLUSIONS
Based on the above analysis, we can draw conclusions as follows:
1. A G-M type single-stage pulse tube refrigerator of 2.0 kW input power with a mixture of 3% N,
and 97% He can provide a cooling capacity of 7 to 10.5 W at 63.5 K to 90 K. This is about 7%
greater than would be obtained with pure helium. The liquid-solid phase change at 63.16 K and
below has been observed.
2. In our experiment, a pulse tube refrigerator without a frozen cryogen accumulator was able to
keep a temperature of about 64 K for about 4 minutes with a 200 mW load when the machine
was switched off. Therefore, if a frozen cryogen accumulator is added, it may be able to meet
the measurement requirements of SQUIDs by eliminating the magnetic noise caused by the
running machine.
3. Calculation results indicate that a frozen nitrogen accumulator, if installed in the cold
head of the pulse tube, can provide approximately 200 mW of cooling power at 64 K for about
20 minutes for a SQUID application.
ACKNOWLEDGMENT
The project was financially supported by the National Natural Sciences Foundation of
China. The authors would appreciate the opportunity to work with Professor C. Heiden and G.
Thummes in their laboratories at University of Giessen, Germany.
REFERENCES
1. ter Brake, H.J., et al., “Construction and Test of a High-T, SQUID-based Heart Scanner Cooled by
Small Stirling Cryocoolers,” ICEC-17, Institute of Physics Publishing (1998), pp. 341-344.
2. van den Bosch, P.J., et al., “Cryogenic Design of a High-T, SQUID-based Heart Scanner Cooled by
Small Stirling Cryocoolers,” Cryogenics, vol.37, no.3 (1997), pp.139-151.
3. Chen, G.B., et al., “Study on Two-component Gas Mixture in Regenerative Refrigerators,” ICEC-17,
Institute of Physics Publishing (1998), pp. 197-200.
4. Chen, G.B., Gan, Z.H., Thummes, G., and Heiden, C., “Thermodynamic Performance Prediction of
Pulse Tube Refrigerators with Mixture Fluids,” to be published in Cryogenics (2000).
5. Gan, Z.H., Chen, G.B., Thummes, G. and Heiden, C., “Experimental Study on Pulse Tube Refrigerator
with Helium-nitrogen Gas Mixture,” Cryocoolers 11, Kluwer Academic/Plenum Publishers, New York
(2001).
6. J. A. McCormick, et al., “Design and Test of Low Capacity Reverse Brayton Cryocooler for Refrigera-
tion at 35K and 60K,” Cryocoolers 10, Kluwer Academic/Plenum Publishers, New York (1999), pp. 421-
429.
Experimental Investigation of a Pulse Tube
Refrigerator Driven by a Thermoacoustic
Prime Mover
L.M. Qiu, G.B. Chen, N. Jiang, Y.L. Jiang and J.P. Yu

Cryogenics Lab, Zhejiang University
Hangzhou 310027, P.R.China
ABSTRACT
A breadboard pulse tube refrigerator driven by a standing-wave thermoacoustic prime
mover has been set up to study the relationship among stack, regenerator, and working fluids.
The stack of the thermoacoustic prime mover is packed with dense-mesh wire screens because
of their low cost and ease of construction. The effect of packing factor in the stack on onset
temperature, refrigeration temperature, and input power has been explored. An optimum pack-
ing factor of 1.15 pieces per millimeter has been found; this is an empirical value that provides
a compromise between enhancing the thermoacoustic effect, and decreasing the heat conduc-
tion and fluid-friction losses along the stack. The pulse tube cooler driven by the thermoacous-
tic prime mover is able to obtain refrigeration temperatures as low as 138 K and 196 K with
helium and nitrogen, respectively.
INTRODUCTION
Thermoacoustic engines can be an attractive alternative for specialized applications be-
cause of their simplicity, and their absence of lubrication, seals, and environmentally harmful
working fluids.1-3 On the other hand, research and development of pulse tube refrigerators has
reached such a stage that commercial products have been already developed for cooling tem-
peratures around 77 K and 4.2 K, respectively.4,5 One may imagine that the reliability of a
pulse tube refrigerator will be greatly improved if it is driven by a thermoacoustic prime mover.
In the past decades, researchers have developed several kinds of stacks to enhance the
thermoacoustic effect. Among these, stacks made of wire screen meshes have the merit of low
heat conduction and reasonable efficiency. S.L. Zhou, Y. Matsubara and G.B. Chen have
reported on measurements of thermoacoustic prime movers with stacks made of copper wire
mesh.6-8 They found that the overall performance of a thermoacoustic prime mover is mainly
dependent on where is the thermal penetration depth, and is the hydraulic radius.
Also, an optimized value of has been experimentally obtained.6
Additionally, the authors have recognized that the packing factor of the stack plays an
important role in influencing the thermoacoustic effect of the prime mover and the refrigera-
tion temperature of the pulse tube. Attempts to find an optimum packing factor will be of
benefit to the design of stacks made of wire mesh. Therefore, experiments have been con-

Figure 1. Schematic diagram of the experimental apparatus.
ducted to determine the optimum packing factor. Meanwhile, the pressure characteristics of
the acoustic prime mover have also been extensively studied. Finally, a refrigeration tempera-
ture as low as 138 K has been obtained using a thermoacoustic-driven pulse tube.
EXPERIMENTAL APPARATUS
The experimental apparatus consists of a 1/2-wavelength, standing-wave thermoacoustic
prime mover and a pulse tube unit as shown in Fig. 1. The thermoacoustic prime mover in-
cludes a resonator tube, hot buffers, hot heat exchangers, stacks, and cold heat exchangers.
They are symmetrically arranged on both sides of the resonator tube as shown. The hot heat
exchanger consists of a copper block with two internal heaters (about 200 W each), and an
outside heater of about 400 W. A transformer adjusts the heating power delivered to the hot
heat exchanger. The structure of the stack is similar to the regenerator used in pulse tube cool-
ers. No. 6 and No. 10 copper mesh disks are packed in turn with a proportion of Various
packing factors were tested to determine the optimum packing factor. The same copper mesh
disks were also packed in the cold heat exchanger with a proportion of 2:1. The hot section of
the thermoacoustic prime mover is insulated with ceramic fiber. The resonant tube is a stainless
steel tube, 4 m in length. Dimensions of the thermoacoustic prime mover are presented in Table 1.
The pulse tube unit, which is connected to the cold heat exchanger via a copper pipe as
shown in Figure 1, has a coaxial configuration and works with a double-inlet. Some param-
eters and data of the pulse tube refrigerator are shown in Table 2.
In the control and measuring system, a temperature controller and a voltage adjuster were
used to control the heating temperature. The heating power is calculated from the resistance of
PT COOLER DRIVEN BY A THERMOACOUSTIC PRIME MOVER 303
the heaters and the applied voltage. The maximum heating temperature is limited to around
500°C because of the structure and silver-alloy brazing of the hot heat exchanger. The pressure
oscillations were measured by a strain pressure sensor that is installed at the inlet of the pulse
tube unit. The heating temperature of the prime mover and cooling temperature at the cold end
of the pulse tube are measured by thermocouples and a Rh-Fe resistance sensor, respectively.
The accuracy of the resistance sensor is approximately 0.1 K.
Pressure Characteristics
The pressure characteristics of the prime mover connected to the pulse tube cooler were mea-
sured in the present work. Figure 2 shows the dependence of refrigeration temperature and heating
temperature on helium charge pressure. We can see that, at a certain charge pressure, the heating
temperature of the mover increases and the refrigeration temperature of the pulse tube decreases
with increasing heating power. For a fixed input power, the higher the charge pressure of the
system, the lower the refrigeration temperature that can be obtained. This can be explained from the
characteristics of pressure ratio and pressure amplitude at the inlet of the pulse tube as shown in
Figure 3. Note that the pressure amplitude increases with increased heating power, which leads to
more mass flow taking part in the refrigeration process. Meanwhile, an increase in pressure ratio
Figure 2. Dependence of charge pressure on minimum temperature and heating temperature.
Figure 3. Pressure amplitude and pressure ratio vs. heating power.

Figure 4. Dependence of charge pressure on onset temperature.
also improves the refrigeration performance of the pulse tube. Therefore, the prime mover has the
advantage of being able to adjust both the pressure amplitude and the pressure ratio by adjusting the
heating power. By comparison, it can be difficult to increase the pressure ratio or amplitude of a
system driven by some mechanical compressors.
Dependence of onset temperature on charge pressure is presented in Fig. 4. We can see that the
onset temperature at a heating power of 294.5 W increases with the charge pressure. In summary,
our inhouse-fabricated thermoacoustic driver can generate a pressure ratio of about 1.1 for helium
or nitrogen, which is adequate for a Stirling type pulse tube refrigerator.
Packing Factor of Stack

Zhou and Matsubara found that the overall performance of the thermoacoustic prime mover is
mainly dependent on They reported that a thermoacoustic prime mover with a wire mesh
stack works well with near This result was also verified in our experiments. According
to our measurements, the resonant frequency is about 70 Hz and 25 Hz for helium and nitrogen,
respectively. Dimensions of the meshes of the stack used in our experiment are listed in Table 3;
when the average temperature of the stack is 573 K, equals 0.37. Thermal penetration depths
of helium and nitrogen are calculated and shown in Fig. 5.
The packing factor of the stack is expressed as pieces per unit length for convenience. Experi-
ments for finding the optimum packing factor were done using a constant charge pressure of 1.8
MPa and the same open ratios for the orifice and second inlet. The heating temperature was kept at
by adjusting the input power; the working fluid was nitrogen. The temperature of the
cooling water was about 300 K.
Figure 6 shows the dependence of the minimum temperature and input power on packing
factor. An optimum packing factor of 1.15 was obtained, corresponding to the minimum refrigera-
tion temperature. If the packing factor is greater than 1.15, the minimum temperature will increase
sharply because of the significant heat conduction and fluid friction losses along the stack. The
temperature at the cold end of the stack was also measured to estimate the heat conduction along the
stack (see Fig. 1). This temperature could be as high as 80-110°C for a heating temperature of
500°C. If the packing factor is smaller than 1.15, the minimum temperature changes smoothly.
Figure 5. Calculated thermal penetration depth of helium and nitrogen.
Figure 6. Dependence of packing factor on minimum temperature and input power.
Figure 6 also shows that heating power increases with a decrease of the packing factor. It means
that the smaller the packing factor, the larger the heating power that can be absorbed. The depen-
dence on packing factor of onset temperature and total thermal energy consumed for onset was also
measured for a heating power of 1347.2 W (see Fig. 7). We find that onset temperature decreases
Figure 7. Dependence of packing factor on onset temperature and minimum input power.
Figure 8. Cool-down behavior of the pulse tube cooler.
with increase of packing factor. The trend of the total thermal energy consumed for onset is in
accordance with the onset temperature.
Refrigeration Characteristics of the Pulse Tube

Typical cool-down tests were carried out to study the refrigeration characteristics of the
pulse tube driven by both the thermoacoustic prime mover and a mechanical compressor. Based
on the above experiments, the packing factor of the stack was set at 1.15 pieces/mm. The
operational parameters of the pulse tube cooler are listed in Table 4. Figure 8 shows the experi-
mental results of the pulse tube unit driven by both the thermoacoustic prime mover and the
compressor. We can see in Fig. 8 that the temperature at the cold end increases slightly in the
first 19 minutes until the onset temperature of 390°C is reached. Then, the temperature de-
creases sharply and reaches the minimum temperature of 138 K within 150 minutes. In com-
parison, with the same pulse tube driven by the mechanical compressor with a swept volume of
a lowest temperature of 74 K (64 K lower than the former) was obtained in 140 min-
utes. Clearly, the efficiency of the pulse tube refrigerator driven by the acoustic driver is still
lower and must be improved.
CONCLUSIONS
1. The experimental results show that our thermoacoustic prime mover can generate a pressure
ratio of about 1.1 for nitrogen and helium, which is adequate for a Stirling type pulse tube
cryocooler.
2. An optimum packing factor of about 1.15 pieces/millimeter of the stack has been found experi-
mentally; this corresponds to the No.6 and No. 10 copper mesh disks being packed in turn with
a proportion of 1:2.
3. The pulse tube cooler driven by our thermoacoustic prime mover is able to obtain refrigeration
temperatures as low as 138 K.
ACKNOWLEDGMENTS
This work was financially supported by the National Natural Science Foundation of China.
Z.H. Gan, K. Tang, W. Zhang and J. Yan are appreciated for their contributions to the experi-
mental work.
REFERENCES
1. Rott, N., “Thermoacoustics,” Advances in Applied Mechanics, vol.20 (1980), pp. 135-175.
2. Swift, G.W., “Thermoacoustic Engine,” Acoust. Soc. Am., vol. 84 (1988), p. 1145.
3. Swift, G.W., Radebaugh, R. and Matin, R.A., “Acoustic Cryocooler,” U. S. Patent, No. 4 953 366
(1990).
4. Radebaugh, R., “Recent Development in Cryocoolers,” Proceedings of 19th International Congress
on Refrigeration, vol.3b (1995), pp. 973-988.
5. Chen, G.B., Qiu, L.M., Zheng, J.Y., Yan, P.D., Gan, Z.H., Bai, X., and Huang, Z.X., “Experimental
Study on a Double-orifice Two-stage Pulse Tube Refrigerator,” Cryogenics, vol.37 (1997), pp. 271-
273.
6. Zhou, S.L., Matsubara, Y., “Experimental Research of Thermoacoustic Prime Mover,” Cryogenics,
vol.38, no.8 (1998), pp. 813-822.
7. Chen, G.B., Jin, T., et al., “Experimental Study on a Thermoacoustic Engine with Brass Screen Stack
Matrix,” A.C.E., vol.43b (1998),pp. 713-718.
8. Bai, X., Jin, T. and Chen, G.B., “Experimental Study on a Thermoacoustic Prime Mover,” Proceed-
ings of ICCR ’98, (1998), pp. 522-525.
Design, Development, and Operation of a
Thermo-Acoustic Refrigerator Cooling
to below –60 °C
M.E.H.Tijani, J. Zeegers, A.T.A.M. de Waele

Eindhoven University of Technology
Low Temperature Group
ABSTRACT
A thermoacoustic cooler, using a resonant standing acoustic wave, has been built. It employs a
loudspeaker to sustain a standing wave in a resonance tube. In the helium-filled tube, a layered
parallel-plate structure, called stack, and two heat exchangers are installed. The interaction of the
compressed and expanded gas in the channels of the stack with the surface generates heat transport.
A description of the cooler is presented, together with the first performance measurements using
two different stacks, two different average pressures, and different dynamic pressures.
INTRODUCTION
Over the past two decades, thermoacoustic cooling has been investigated as a new cooling
technology.1-8 Thermoacoustic coolers can reach temperatures of –70°C and can have a coefficient
of performance of 20 % of Carnot. Instead of CFKs, inert gases are used.
Typically, a thermoacoustic cooler consists of an acoustic resonator (e.g. tube) filled with an
inert gas at some average pressure in which a structure with channels, called a stack, is placed.
The stack is the heart of the cooler; it is where the heat transfer takes place. At both ends of the
stack, heat exchangers are installed. The temperature of the hot heat exchanger is fixed at room
temperature; at the cold heat exchanger, cooling power is generated.
A modified loudspeaker generates sound in the form of a standing resonant wave. This wave
causes the gas particles to oscillate while compressing and expanding.
The thermoacoustic cooling cycle can be illustrated by considering a parcel of gas oscillating
along the stack surface as a response to the standing wave, as illustrated in Fig. 1. During one
period of the acoustic cycle, the parcel of gas undergoes two adiabatic steps (1 and 3), and two
constant pressure heat transfer steps (2 and 4). In step 1, the parcel of gas moves forward, in the
direction of lower pressure, expands, and cools. At this time, the parcel of gas is colder than the
local stack surface, and heat transfer from the stack to the parcel takes place (step 2). In step 3, the
parcel of gas moves back to its initial position, is compressed, and warms up. Now, in step 4, the
parcel of gas is warmer than the local stack surface and heat flows from the parcel to the stack.

Figure 1. (a) A typical gas parcel in a thermoacoustic cooler experiences a four-step cycle with two
adiabatic steps (1 and 3) and two constant-pressure heat transfers steps (2 and 4); (b) An amount of heat is
shuttled along the plate from one parcel of gas to the next; as a result, heat Q is transported from one end
of the plate to the other using acoustic energy W to compress the gas.
At this stage the parcel of gas is returned to its initial position and the cycle starts again.
Although the parcel excursion is much smaller than the length of the stack, the net effect of many
parcels along the stack is that heat is transported from one end of the stack to the other (c.f. Fig.1b).
If the hot end is fixed at room temperature, the other end cools down. Only the gas layer
approximately within a distance of one thermal penetration depth from the stack's surface
contributes to the thermoacoustic effect. The quantity is the distance across which heat can
diffuse through the gas in a time where f is the acoustic frequency, is defined in terms of the
thermal conductivity of the gas k, the gas density and its isobaric specific heat
In standing wave coolers, an important geometrical requirement is the transverse channel

dimension in the stack, which amounts 1 to 4 times and depends on the used geometry.3,4 A
detailed discussion of the theory of the thermoacoustic effect can be found in the literature.1-4
DESCRIPTION OF THE THERMOACOUSTIC COOLER

In Fig. 2, a schematic diagram of our thermoacoustic cooler is shown. It is a Hofler type
cooler.8 Thermoacoustic theory3 was applied to the design this cooler. The cooler consists mainly
of five parts: a loudspeaker, a helium-filled resonator, a stack, and two heat exchangers. From the
loudspeaker, the fabric dome was cut off near the voice coil and replaced by a thin-walled light
aluminium cone glued onto the voice coil. A rolling diaphragm is used to seal the resonator from
the loudspeaker housing.
Commercially available loudspeakers have an efficiency of 3-5%. The performance can be
improved when a loudspeaker is coupled to an acoustic resonator, provided an appropriate selection
is made of the parameters of the system. From a model simulating the coupling between the resonator
THERMO-ACOUSTIC REFRIGERATOR COOLING BELOW -60°C 311
Figure 2. Cross-sectional illustration of the thermoacoustic cooler, showing the different parts.
and the loudspeaker, it follows that the electroacoustic efficiency of the loudspeaker is maximum
when the mechanical resonance of the loudspeaker is equal to the fundamental acoustic resonance
of the resonator.6 The electroacoustic efficiency of the loudspeaker is defined as the ratio of the
output acoustic power to the input electric power delivered to the voice coil. Because, in general,
the two resonant frequencies are different, a new concept is used successfully in our cooler to
control the mechanical resonant frequency of the loudspeaker. The volume of gas behind the cone
is used as an adjustable extra spring that adds to the spring of the loudspeaker. As a consequence,
the resonant frequency of the loudspeaker can be tuned to the frequency of the resonator to obtain
high performance. We have used a system consisting of a cylindrical volume and a piston that can
be positioned at different heights, and thereby changes the volume behind the cone. This is indicated
in Fig. 2 as the cylinder and piston system. A detailed description of this tuning concept will be
published elsewhere.
The resonator was optimised for minimum viscous losses leading to a fundamental operating
frequency of nearly 430 Hz. It consists of many parts. First is a copper flange, which contains the
hot heat exchanger. This flange is used to connect the resonator to the driver housing, and ensures
a good thermal contact with the bottom of the loudspeaker housing through which cooling water
circulates. The part of the resonator which contains the stack has a low thermal conductivity to
minimize the heat flow from the hot end to the cold end of the stack. Next, a copper contraction is
used to connect the stack holder to the smaller copper resonator part and to reduce turbulence as the
cross section changes. The cold heat exchanger is soldered to the neck. Finally, the small resonator
part terminates in the stainless-steel buffer volume to complete the acoustically resonant system.
The whole system is designed to maintain a static pressure of 12 bar. Copper tubes, in which
cooling water circulates, are soldered in the bottom of the loudspeaker housing to remove the
thermoacoustic heat of the stack and of the loudspeaker.
As discussed above, the transverse dimension of the channels in the stack is determined by the
thermal penetration depth This is a function of the parameters of the type of gas and the acoustic
frequency. The cooler is designed to meet the requirements of a relatively large temperature span
of 90°C over the stack and a cooling power of 4 watts. The parameters concerning the stack position,
stack length, and layer spacing, were all taken into account for the design. The hot and cold heat
exchangers are made of copper fins, 0.1 mm thick. They consist of sine channels, and have different
sizes in the longitudinal direction. Two different stacks have been made. One stack is made by
winding spirally a long sheet of 0.06 mm thick plastic. Fishing lines, 0.3 mm thick, are used as
spacers at 5-mm intervals, glued laterally onto the sheet. The second stack consists of Mylar parallel
plates, 0.1 mm thick. The plates are also separated by 0.3-mm thick monofilament fishing line
spacers, at an interval of 7 mm. The two stacks have a diameter of 38 mm and a length of 84 mm.
The second stack was more difficult to construct, but it has a more uniform channel structure. An
electrical heater (Fig. 2) was used to apply heat to the cooler in order to determine the coefficient of
performance of the cooler at different temperatures.
Thermometers are placed at the cold heat exchanger, at the hot heat exchanger, on the
loudspeaker, and on the buffer volume. A more detailed description of the cooler will be published
later.
MEASUREMENTS AND RESULTS

The input acoustic power into the resonator is measured using a dynamic pressure transducer
placed near the pusher cone and an accelerometer on the pusher cone7 (c.f. Fig. 2). By using two
lock-in amplifiers we determined the dynamic pressure the velocity of the cone (u), and phase
difference between them. The input acoustic power is given by
The performance of the cooler is described by the coefficient of performance,3 which is given
by the ratio of the cooling load (including heat leak), Q, at the cold heat exchanger and acoustic
power, W, delivered by the loudspeaker to the resonator
It is convenient to characterise the performance of the cooler by the coefficient of perfor-

mance relative to Carnot’s coefficient of performance COPR, defined as 3
where the Carnot coefficient of performance COPC is given by 3
and are the temperatures of the hot and cold heat exchangers, respectively.
The measurement procedure is as follows: the cooler is first evacuated and filled with helium
up to the desired average pressure. Then, the vacuum vessel is evacuated until a good vacuum
Pa) is reached. Finally a power amplifier, controlling the loudspeaker, is set at the desired amplitude
of the dynamic pressure by amplifying the signal from a function generator.
A series of measurements of the electroacoustic efficiency for different piston heights are
shown in Fig. 3. Two peaks can be noticed in all plots: one peak at 430 Hz is due to the fundamental
acoustic resonance and the other is due to the mechanical resonance of the loudspeaker. The peak
due to the loudspeaker shifts as the height of the piston changes. The height can be varied from
31 mm to 123 mm. As the height becomes lower the included volume becomes smaller and the
corresponding spring constant becomes larger. This results in a shift of the mechanical resonance
of the loudspeaker to higher frequencies. A maximum efficiency is reached when the two peaks
nearly match, at 72.5 mm. When the two frequencies are equal, the efficiency is constant over a
wider frequency range, which is desirable, as the resonance frequency of the resonator decreases
when the temperature decreases during the cooldown.
Figure 3. Measured electroacoustic efficiency h as function of the frequency for nine different piston
positions. A maximum efficiency is reached when the resonance frequency of the loudspeaker and that of
the resonator are nearly equal at a height of 72.5 mm.
The loudspeaker used for these measurements had a large electro-mechanical damping. A higher
efficiency of 30-40% is possible if an appropriate loudspeaker is used in combination with our
tuning concept.
All performance measurements were made using helium as the working gas. The temperature of
the cooler decreases until steady state conditions are established (c.f. Fig. 5). Then, stepwise, a heat
load is applied to the cold end. Each time the steady state temperature is reached, a set of parameters
is recorded, consisting of f, Q, P1, u, etc. After this set of performance measurements,
parameters like static pressure and dynamic pressure can be changed for a new series.
Figure 4. Measured performance of the thermoacoustic cooler with the first (coiled) stack.
Figure 4 shows the performance measurements and the temperature difference, over the
stack for the first stack (coiled sheet) for two static pressures, 5 and 10 bar, using three different
dynamic pressures in each case. We have chosen to use the ratio of the dynamic pressure to the
static pressure instead of the dynamic pressure as the parameter. In all cases, the temperature
difference over the stack is a linear function of the heat load Q, and it increases as the dynamic
pressure increases. The coefficient of performance relative to Carnot shows a maximum, which
shifts to higher heat loads when the dynamic pressure increases.
For the second stack, the behaviour of the temperatures of the hot heat exchanger cold heat
exchanger and the buffer as function of time, are also plotted in Fig. 5a. The temperature of
the hot heat exchanger is held constant by the cooling water. The whole part of the cooler below the
cold heat exchanger cools down. As the temperature of the buffer lags that of the cold heat exchanger,
we believe that the kink in at the time of nearly 200 minutes, can be attributed to convection in
the resonator, which is set up as a consequence of the temperature gradient over the cold side. The
long cooling time constant is due to the large mass of the cold side (1300 g).
Only one measurement of the temperature difference and coefficient of performance relative to
Carnot is available for the second stack at this time. The same behaviour can be concluded as for
the first stack. The maximum coefficient of performance of the second stack for a mean pressure of
10 bar helium and a pressure ratio of 2.1% is 11.5% at a heat load of watts.
CONCLUSIONS
A thermoacoustic cooler has been designed and built. The first measurements were successful.
It cools down to below – 67°C and has a COPR of 11.5% for a static pressure of 10 bar helium and
2.1% pressure ratio. From the measurements, one can conclude that the parallel-plate stack has a
higher performance than the coiled one.
A number of improvements to the cooler are underway; a lighter aluminium resonator has
recently been constructed. It weighs only 350 g instead of the present 1300g, which will decrease
the thermal time constant and decrease the temperature gradient over the cold end to a low level.
Figure 5. Some measurements of the thermoacoustic cooler with the second stack. a) Temperatures
of the hot heat exchanger, cold heat exchanger and the buffer volume as function of time. b) Measured
temperatures and performance as function of heat load.
This will also accelerate the measurements that last normally a day per heat load for low dynamic
pressures. The use of other parallel plate stacks with different spacing and other gases is also planned
for the near future.
ACKNOWLEDGEMENT
We like to acknowledge the following persons: Lock Penders (electronics), Leo van Hout
(engineering), the assistance of the department and central workshops of our university. We like to
thank Greg Swift and Chris Espinoza of Los Alamos National Laboratories for their advise in the
engineering of the parallel plates stack. We are much indebted to Guido d’Hoogh of Philips speaker
systems for the development of the loudspeakers.
REFERENCES
1. J. Wheatley, T. Hofler, G.W. Swift, and A. Migliori, “Understanding some simple phenomena in
thermoacoustics with applications to acoustical heat engines,” Am. J. Phys, Vol. 53, no.2 (1985),
pp. 147-162.
2. J. Wheatley, T. Hofler, G.W. Swift, and A. Migliori, “An intrinsically irreversible thermoacoustic
heat engine,” J.Acoust.Soc.Am., Vol. 74, no.1 (1983), pp. 153-170.
3. G.W. Swift, “Thermoacoustic engines,” J. Acoust. Soc. Am. Vol. 84, no.4 (1988), pp.1146-1180.
4. G.W. Swift, “Thermoacoustic engines and refrigerators,” Encyclopaedia of Applied physics, Vol. 21,
(1997), pp. 245-264.
5. G.W. Swift, “Thermoacoustic engines and refrigerators,” Physics Today, (1995), pp. 22-28.
6. S.L. Garrett, “ThermoAcoustic Life Science Refrigerator,” NASA Report, no. LS-10114 (1991).
7. T.J. Hofler, “accurate acoustic power measurements with a high-intensity driver,” J. Acoust. Soc. Am.
Vol. 83, no.2 (1988), pp. 777-786.
8. T.J. Hofler, “Thermoacoustic Refrigeration Design and Performance,” Ph.D. Thesis, Physics
Department, University of California, San Diego (1986).
Design of a Miniature Pulse Tube Refrigerator
A. Halouane*, J-C Marechal and Y. Simon

Ecole Normale Superieure, 24 rue Lhomond
75005 Paris, France
* French Institute of Petroleum
92852 Rueil-Malmaison cedex, France
ABSTRACT
Using a miniature pulse tube refrigerator and a RICOR compressor of 1cc swept volume, we
have achieved a lowest temperature of 80 K at which the cooling power is 1W.
The main purpose of this paper is to explain the rules used to design each component of a
miniature PTR: regenerator, tube, reservoir, heat exchangers, and valves. The regenerator will be
investigated in the following two ways: Firstly, a complete hydrodynamic study (based on mass
and momentum conservation). Secondly, a thermal study of the system based on energy conser-
vation in the gas and the matrix. These allowed the computation of the gas temperature profile in
the regenerator and in the tube. The losses of the system were deduced.
INTRODUCTION
In an OPTR / DIPTR Pulse Tube Refrigerator1 helium gas is periodically compressed and
expanded in a closed system; the main part is a simple tube ended by two heat exchangers: cold
and hot. During compression, the gas, before entering the tube, has passed through a regenerator
filled with wire screens, whose role is to bring the gas from room temperature to the cold
temperature During expansion, the same regenerator brings the gas from to The gas is
pulsed at the entrance of the regenerator by a oscillating piston compressor.2 The system (tube,
regenerator, and heat exchangers) is in a vacuum chamber.
CHARACTERISTICS OF THE MINIATURE OPTR

The piston of the compressor, driven in a cyclic manner, sweeps (under a process somewhere
between isothermal and adiabatic) a volume at a frequency The swept
volume amplitude is around An aftercooler (heat exchanger with a dead volume of 0.4
is inserted between the compressor and the regenerator to cool the gas down to room tem-
perature
The bronze wire screens are packed along the regenerator with a cross-section of
The open gas section is A = n S, where n = 0.61 represents the porosity.
Another parameter called hydraulic radius, , defined as the ratio (open volume / exchange
section) gives an idea about the dimensions of the pores. In the present case

318 PULSE TUBE ANALYSES AND EXPERIMENTAL MEASUREMENTS
Figure 1. Description of the PTR.
The thickness of the stainless steel tube is about 0.2 mm, its length is 72 mm, and its diameter
is 4 mm Both heat exchangers are made with the same copper wire screens
soldered at the periphery of the tube over a length of 6 mm, (n=05, each has a dead
volume of
The orifice is a needle valve, and the reservoir volume is bigger than the whole system volume
as shown in Fig. 1.
In Fig. 2 we represent the temperature profile of the system obtained using the theoretical
model and we will compare it to the experimental results.
The monatomic gas used is helium; given its unity mass one can write:
and the specific heats:
Figure 2. Temperature profile of the PTR.

DESIGN OF A MINIATURE PULSE TUBE REFRIGERATOR 319
Figure 3. Electrical analogy of the PTR.
The amplitude of the mass flow rate in a cycle is an important parameter to enhance the
refrigeration power. Working at high pressure, this will increase automatically the mass flow
rate especially in miniature PTR. In our case the main pressure is :
The uniform pressure in the tube is the reference, where At
the entrance of the regenerator the pressure amplitude is bigger and the phase
angle has a lead over all others To all these measured parameters we can also add
the measure of the reservoir pressure amplitude which allows us to
compute the mass flow rate value through the orifice:
is the equivalent capacity of the reservoir in an electrical analogy pressure/voltage and

mass flow rate/ current( 3) , (6 ) (see Fig. 3).
Through the orifice and after linearizing the relation one can write :
From the measure of the pressure amplitude of the reservoir , we get using the
relation :
Typically and
The mass flow rate at the entrance of the hot heat exchanger is a bit different from
because of the dead volume. Since the transformation in the compressor could be
adiabatic, isothermal or just between them, the mass flow rate flowing out of the
compressor will be expressed by:
takes into account the mean value of the compressor and all dead volumes located
between the compressor and the regenerator entrance.
Hysteresis effect in the tube

At the hot end of the tube (respectively the cold end): at the instant t’ the gas enters in
the tube with a temperature and goes up to a maximum distance The mass
conservation of gas in the tube allows the computation of this ratio of maximum penetration
depth
where : is the capacity of the tube.

For any edge x(t), we can write the relation between the position and the pressure :
In this approximation «sinusoidal waves», the mass flow rate at the cold end is :
We find:
with:
At the cold end the « ratio » of maximum penetration depth is given by
At the moment t the gas will leave the hot end with a temperature T(t) linked to the
temperature T(t’) by the isentropic relation :
According to the first law of thermodynamics the heat evacuated at the hot heat
exchanger is expressed as :
In the case of small sinusoidal amplitudes :
When the system reaches the coldest temperature, and since the system is in a
vacuum chamber, the « energy flux » is constant along the regenerator and the tube
is the term representing the regenerator losses takes into

account the cryocooler losses and is the conduction loss of the tube
Since is small compared to then the regenerator losses are
equivalent to the energy flux along the system
Figure 4. Flow through a porous media.
STUDY OF THE REGENERATOR
Applying a small perturbation around the equilibrium point with for a cyclic
steady operation, the thermodynamic parameters P(x,t) and T(x,t) should be decomposed as
with
Starting with the hydrodynamic analysis we can write the following equations of conservation
in the regenerator:
Mass conservation
In the porous media. Figure 4, the mass flow rate is related to the mean velocity u and
the gas density ρ by the relation The mass conservation equation is then :
Momentum conservation
We can write, after simplification, the momentum conservation :
where R is the resistance per unit length of the regenerator expressed by(4), (5) :
a and b depends on the geometry of the wire matrix and it porosity (in our case a = 44 and
b = 0.4) . µ is the dynamic viscosity of the gas and
the Reynolds number.
In order to study the thermal behaviour of the regenerator ; we should know firstly the heat
exchange between the gas and the matrix (matrix = regenerator wire screens). This work was
done by Tanaka and all:
and is the thermal conductivity of the gas.

HEAT EQUATION
For an ideal gas the heat equation is expressed as :
is the matrix temperature, the term sources can be written :
where ( div q ) is the term and the viscous dissipation is represented by
is a numerical parameter
1-Analysis of steady terms in the Heat equation
Steady terms obtained from the heat equation can be written as follows:
let us compare different terms of this equation at the hot end of the regenerator; the same work
was done at the cold part and gave same conclusions. At the hot end the term
and represent exactly the same expression:
and A simple
calculation shows that is:
from this result we can assume that the mean temperature of the gas and the matrix are the
same in each point x along the regenerator.
2-Analysis of unsteady terms of the Heat equation
Unsteady terms obtained from heat equation allow us to write the following equation
let us compare different terms of this equation at the hot end of the regenerator (respectively at
the cold side). and
All these terms have the same order so we will not neglect any
one of them.
ENERGY EQUATION OF REGENERATOR MATRIX

Between sections x and x + dx of the regenerator, the matrix occupies a volume
A dx (1-n)/n. Its heat capacity per unit volume is matrix density and mass
heat capacity of the matrix); The longitudinal conduction of the
matrix and the tube is represented by:
taking into account the stainless steal tube conduction and the matrix conduction
is a coefficient depending on diameter and
thickness of the regenerator tube.
As in the heat equation, we deduce the energy conservation equation of the regenerator
matrix.
Global Equation (gas - matrix)

Combining the equation of heat conservation (24) (unsteady terms) with equation (25)
(of matrix conservation of energy) one obtains:
To this equation we can add the conservation of energy along the regenerator :
Let us introduce the complex expression of the following parameters at the first order
development :
and
and the time constant with an effective specific heat of the matrix :
Combining equations (27) and (28) we obtain the relation between the local temperature
gradient and the energy flux :
where,
Figure 5. Temperature profile in the regenerator.
COMPUTER MODELLING8
The pressure in the tube and the pressure in the reservoir

with a common phase angle will be our reference zero. The couple is completely
defined at the cold end of the regenerator. Applying the electrical analogy we start our
program with an obtained for example when we suppose that the temperature
profile is constant along the regenerator. From a cell (i=1) to the next one (i+1), starting at the
cold end, we arrive at the entrance of the regenerator where we compare with the experimental
values of (pressure amplitude), (angle phase), (compressor swept volume) and
If these parameters correspond to the experimental data we stop the program and we
print at the same time the temperature profile and the regenerator losses ; (see Figure 5).
The ratio between experimental and theoretical regenerator heat loses is :
STUDY OF THE TUBE7
The heat exchanged over a cycle between the gas and the tube-wall has a very large affect on
the refrigeration power of the system. In order to study the tube losses we can follow the same
work done with the regenerator. Considering the motion of the gas in the tube, Equations (8) and
(12) allow us to compute point by point according to x, for different time instants (where i= 1 to
10) regularly spaced in a period the instantaneous temperature profile of the gas
in the tube. Figure 6 shows the instantaneous temperature profile at the hot and cold ends of the
tube. Since the buffer gas never leaves the tube, the mean temperature profile can be estimated to
Figure 6. Instantaneous temperature profile at the hot and cold end.

Figure 7. Mean temperature profile of the gas in the tube.
Figure 8. Mean temperature profile of the tube and the gas.
Figure 9. Mean temperature profile of the tube (gas + tube).
be linear. The mean temperature profile of the gas at a moment in the tube is shown in Fig. 7.
The thermal conduction along the stainless steel tube is:
Figure 8 shows both the mean temperature profile of the tube alone, due to thermal conduc-
tion, and the gas temperature profile. In Fig. 9, the temperature profile of the tube (gas + tube)
is compared to the experimental results and shows that the theoretically predicted curves fit well
with the experimental results.
DISCUSSION
In this work we have shown that knowing the real losses of the regenerater requires knowledge
of the real temperature profile. The optimization is, then, easier. We changed the geometry and
nature of the matrix and selected the best one. The regenerator tube volume was twice the com-
pressor swept volume. For the tube, the most important thing was that its volume is near the
volume of the compressor; thus the penetration depth of the gas into the tube at both ends never
exceeded 20 to 30%. If the penetrartion was much greater, the gas coming from the cold side
could travel all the way to the hot heat exchanger, which would seriously decrease the efficiency.
The mixed gas chamber in the valve system must be as small as possible so as to reduce the dead
volume of the needle valves. The volume of the reservoir was about 100 times greater than the
compressor volume. For miniature pulse tube refrigerators, we have to focus our work on reducing
every dead volume in the system.
REFERENCES
1. Radebaugh, R., Zimerman, J., Smith, D. R. and Louie, B., “A Comparison of Three Types of Pulse
Tube Refrigerators: New Methods for Reaching 60 K,” Advances in Cryogenic Engineering, vol. 31,
Plenum Press, New York (1986), p. 779.
2. Roach, P., Kashani, A. and Lee J. M., “Theoretical Analysis of a Pulse Tube Refrigerator,” Advances
in Cryogenic Engineering, 41 B (1995), p. 1357.
3. Huang, B.J and Chuang, M.D., “System Design of Orifice Pulse Tube Refrigerator Using Linear
Network Analysis,” Cryogenics, vol. 36, no. 11 (1996), pp.1357.
4. Tanaka, M., Yamashita, I. and Chisaka, F., “Flow and Heat Transfer Characteristics of Stirling Engine
Regenerator in an Oscillating Flow,” JSME Int Series II (1990).
5. Kays, W., M. and London, A. L., “Compact Heat Exchangers,” Mc Graw-Hill Book Company.
6. Halouane, A., Marechal, J.C. and David, M., “Study by Electrical Analogy of Different Pulse Tubes
Refrigerators: Model and Experiments,” The European Physical Journal Applied Physics, vol. 4,
(1998), pp. 31-35.
7. Halouane, A., “Hydrodynamic and Thermal Study of the Pulse Tube Refrigerator: Model and Experi-
ments,” PhD Paris Vl University, April 1999, (in French ).
8. Halouane, A., Marechal, J.C., David, M. and Simon, Y., “Study of Regenerator Temperature Gradient
in a Pulse Tube Refrigerator,” Proceedings ICEC 17, IOP (1998), pp. 227-235.
Investigation of a Single Stage Four-Valve Pulse
Tube Refrigerator for High Cooling Power
T. Schmauder, A. Waldauf, M. Thürk, R. Wagner and P. Seidel

Institut für Festkörperphysik, Friedrich-Schiller-Universität Jena
D-07743 Jena, Germany
ABSTRACT
We discuss the optimization of a pulse tube refrigerator for high cooling power. Our
approach is to increase the system efficiency by analyzing and reducing the various loss
mechanisms. Because stationary losses (such as radiation and thermal conduction in the sys-
tem) as well as design principles for the regenerator are well understood, our main effort is
focused on controlling the flow behaviour of the working gas at the various tube connections
between the components. For time resolved measurements of the gas velocity and gas tem-
perature we use hot wire anemometry and thermocouples respectively. The results of this
analysis are used to improve the design especially of the cold head heat exchanger and the hot
end setup of the pulse tube. Despite the consequent separation of the in- and outlet gas at the
hot end of the pulse tube we find a strong hot end loss caused even by very simple flow
parallelizing devices at the hot end of the pulse.
A cooling power of 67 W at 70 K has been achieved. The aim of this project is a cooling
power of 100 W at 80 K for thermal shielding for magnets and for cryopumps.
INTRODUCTION
Today there are various applications for cryorefrigerators such as sensors, electronics
and superconductors cooling and cryopumping, thermal shields cooling or even cryogenic
fluids liquification for the cooling of large high-field magnets. Low cost and high reliability
as well as low interference make the pulse tube a suitable cooler for many applications.
Amongst the various types of pulse tube refrigerators, the four-valve pulse tube refrigerator
(FVPTR) has great potential for being a highly efficient pulse tube cooler.1 Compared with
the well known explanation of the basic phenomena inside of an ideal pulse tube refrigera-
tor,2-6 a clear explanation of the real effects has not been given even though a few experimen-
tal studies have been reported,7-10 in which the intrinsic behaviour of the orifice pulse tube
refrigerators was investigated. Nevertheless, open questions about the special intrinsic ef-
fects inside the FVPTR remain.
The idealized working principle of the FVPTR is illustrated in Figure 1. The working
fluid (usually Helium) is compressed by an external compressor and enters the pulse tube at
the cold end after being cooled to the cold head temperature Tcold while passing the regenera-

Figure 1. Working cycle of the four-valve pulse tube refrigerator (FVPTR): A) a) to e):
temperature distribution along regenerator and pulse tube for different phases of the cycle,
B) pressure wave during the cycle.
tor (Figure 1a). The gas in the pulse tube is adiabatically compressed from low pressure
to high pressure and heats up by a factor . In the next phase (see
Figure 1b) the part of the working gas with a higher than ambient temperature is re-
moved from the system at the hot end of the pulse tube. Next, the gas in the pulse tube is
expanded towards the cold end through the regenerator, adiabatically cooling below
before passing the cold end heat exchanger (HX). Finally, the amount of gas, which was
removed in the second phase of the working cycle is replaced. However, the replaced gas has
only temperature while the removed gas had a higher temperature, thus removing an
amount of heat from the system and generating a cooling power.
THEORETICAL CONSIDERATIONS
Generated Cooling Power
The heat removed during one cycle from a machine running a gas cycle process as de-
scribed above for the FVPTR is:
Eq. (1) can be analysed in different ways:

Separation in phases of the working cycle. In this case the cycle is broken up into
easily described steps. For the FVPTR cycle as described above, Eq. (1) reduces to:
INVESTIGATION OF 1 -STAGE 4-VALVE PT REFRIGERATOR 329
specific heat of working gas, amount of heat exchanged) with exchanged mass
(M: molar mass of working gas, gas constant, pulse tube volume,
exchange ratio) and temperatures:
Eq. (2) to (4) yield
Obviously, large values of pressure ratio, pulse tube volume and exchange ratio are
desirable.
Thermoaccoustic model. A different approach to analyse eq. (1) is to treat volume flow
and pressure wave in a harmonic approximation. The work flow at the cold end of the pulse
tube is then11:
phase angle between pressure and volume wave, amplitudes of pressure and volume
waves, respectively). The phase angle is crucial for the cooling power. It is not only
influenced by the construction details of regenerator and heat exchanger, but also by the
timing of the gas exchange at the hot end. Detailed investigations based on phasor diagrams
are given in Ref. 4.
Loss Mechanisms
There are a number of obvious loss mechanisms in the pulse tube cryocooler. Some of
them are very well understood and routinely considered when designing a cryocooler setup.5
These include losses due to heat conduction and radiation. Losses due to the finite thermal
capacity of the regenerator are also well understood and we considered them when designing
our regenerators6. There remain additional losses which are not quantitatively understood yet
and on which our experimental efforts will be focussed:
Turbulence Generation at Tube Connections between control valves, regenerator, heat
exchanger and pulse tube. As described above, the working principle of the cooler relies on a
stable temperature separation of hot and cold gas in the pulse tube. Turbulences in the gas
streams entering or leaving the tube disturb this temperature separation and thus reduce the
net cooling power. On the other hand a certain degree of turbulence helps to improve heat
transfer in regenerator and the heat exchanger.
Dead Volume and Hot End Exchange Volume. The mass of working gas exchanged at
the hot end of the pulse tube requires work from the compressor for its displacement. In
contrast to orifice/buffer pulse tube refrigerators, this work is lost and can not be recovered
when the displacement is reversed later during the cycle. For a large exchange volume
in Eq. (3)) the additional load for the compressor may overweight the gain in cooling power
as expected from Equation (5). A similar problem arises for the regenerator construction:
Low pressure drop over the regenerator requires a large effective diameter, but this increases
the dead volume in which gas is compressed but does not contribute to cooling power genera-
tion.
Hot End Loss. Gas leaving the pulse tube at the hot end will heat up the flow channel it
passes. When the replaced gas enters later through the same flow channels, it will collect this
heat and bring it back into the pulse tube. This is not desired since it reduces the amount of
heat in Eqs. (1) to (6)) removed from the system during the work cycle. Thus, the design
of the hot end of the pulse tube is crucial for the operation of the FVPTR.
EXPERIMENTAL SETUP
Refrigerator Setup
A FVPTR testbed has been set up which allows one to quickly exchange the main compo-
nents including the pulse tube, cold heat exchanger, and regenerator, to vary the timing of the
regenerator and hot end inlet and outlet valves, and to monitor the temperatures and pressures
at various points of the cryocooler. A commercial Helium compressor (CTI, model 9600)
with 5.5kW input power was used. Figure 2 A) shows a photograph of the testbed.
A schematic overview of the setup is given in Figure 2 B). We note that the inlet and
outlet channels at the hot end of the pulse tube are strictly separated to avoid hot end losses
due to regenerative effects (hot gas leaving the pulse tube heats the outlet channel walls which
would return the excess heat to the entering replacement gas later during the work cycle).
This construction enables us to avoid the use of a hot end heat exchanger. Thus, the refrigera-
tor design remains very simple since only the helium supply lines and the control valve power
line, but no active cooling, are necessary on the refrigerator unit.
Hot Wire Anemometry
To understand and minimize losses due to turbulence at the junctions of the various
components, we conducted hot wire anemometry (HWA) studies. The setup for these mea-
surements is shown schematically in Figure 3a). A continuous helium flow at 8 bar was used
as the working gas. Measurements of the gas velocity have been carried out within the pulse
tube at 6 points along the longitudinal axis using hot wire anemometers which were able to
scan over the tube diameter. The response time of the HWA-sensor is about 500µs (2kHz).
As indicated in Figure 3b) various designs of flow straighteners for the hot end, the
Figure 2. A) photograph of our FVPTR test bed, and B) schematic of the setup indicating probe
positions for temperature measurements (Si diodes and thermocouples) and pressure measurements
(piezoelectric sensors).
Figure 3. Experimental setup for hot wire anemometry measurements: a) schematic of the
entire apparatus, b) the inlet head allows for various configurations by exchanging insets
representing regenerator and pulse tube junctions.
cold end, and the regenerator intake manifold were tested by introducing the constructions
in the changeable inlet head on top of the apparatus.

Pneumatic Flow Optimization
Typical velocity profiles of the working gas right behind the intake design of the hot
end of the pulse tube, the cold end of the pulse tube and the regenerator are documented
in Figure 4a, b and c, respectively.
An intake design without any flow straightener is leading to a jet stream in the centre
of the tube ( ´s in Figure 4a). The gas penetrates more deeply into the tube increasing the
turbulence. This results in a high convective heat transfer between the gas elements and
high axial heat loss, respectively. A noticeable compensation of the mismatched flow is
reached by using a stack made from a couple of screens (+ and curves in Figure 4a). On
the other hand we have to reduce thermal mass at the hot end of the pulse tube to prevent
hot end loss9. As pointed out above, however, the losses due to flow instabilities increase for
an unfavourable designed intake device. A compromise is to be sought. An efficient flow
straightener design with low thermal mass could be a thin plate perforated with fine holes
( in Figure 4a).
At the cold end of the pulse tube the working gas stream is perpendicular to the tube
axis. This results in asymmetrical distribution of the flow as it is displayed in Figure 4b. As
the cold heat exchanger made of a stack of more than 20 screens also acts as flow
straightener, the velocity profile of the gas entering the tube from the regenerator is flat
and the incident flow is laminar.
In order to minimize regenerator losses a uniform fluid distribution over the cross
section of the regenerator is necessary. The -curve in Figure 4c indicates an asymmetrical
pattern of the flow distribution over the cross section of the regenerator due to the lack of
any plenum chamber at the end of the regenerator. On the other hand a too large volume
of the plenum chamber means a large compression and expansion of the remaining gas.
This does not contribute to the cooling power, but the necessary compression power
reduces the COP; again a compromise is to be sought. A sufficient balance of the flow was
Figure 4. Anemometry measurements of the flow velocity a) at the hot end,and b) at the cold
end of the pulse tube, c) at the regenerator inlet. Solid lines are to guide the eye.
Figure 5. Pressure and temperature waves during the working cycle at different positions along
the pulse tube for optimized (maximum cooling power) exchange volume and dc-flow.
achieved by using a plenum spacer with a height of 3 mm. The high velocities near the wall
indicate reduced mesh density due to a small gap between the stack and the regenerator
housing which can be prevented by using slightly overdimensioned discs.
Refrigerator Operation
As discussed above, the exchanged volume has significant influence on
the cooling power: according to Eq.(5), a large exchange volume is desired, while due to
dead volume loss this volume needs to be limited. Similarly, the balance between the
removed and replaced amount of gas (referred to as dc-flow) will influence the regenerator
efficiency and the coolers performance. Thus we studied the performance of the FVPTR in
dependence of these two parameters and dc-flow which we control using needle valves
at the hot end in- and outlet. For studies of the heat transfer within the pulse tube we use
thermocouple probes for time resolved measurements of the gas temperature at various
positions along the pulse tube. Figure 5 shows these temperature curves together
synchronised to the respective pressure waves. It is obvious that the steepest temperature
gradient is in the lower half of the tube and a huge temperature amplitude of more than
100K (peak-peak) is observed. Surprisingly, no indication of the temperature drop due to
the replacement gas intake is seen even very close to the hot end flow straightener.
For cryocoolers of high cooling power, the cold heat exchanger becomes a critical
component. In the pulse tube refrigerator this heat exchanger also serves as flow
conditioner at the cold end. We performed measurements of the temperature drop as well
between the cold head housing and the wires of the copper gauze serving as exchange
surface as of the drop between those wires and the working gas. The results of these
measurements are given in Figure 6A. To reduce the temperature drop per transferred heat
power we improved the contact between the cold head housing and the wire gauze disks
Figure 6. Transferrable power vs. temperature drop over the cold heat exchanger: a) for the
original cold head heat exchanger, b) for the improved heat exchanger with soldered gauze discs.
and further increased the exchange surface. The new heat exchanger (Figure 6B) has a
significant reduced temperature drop per transferred power of 0.11 K/W.
Hot End Design

We were not satisfied with the available cooling power of the test FVPTR even with
pneumatically optimized component junctions, improved heat exchanger and optimal
exchange volume and dc-flow. From the available data we assumed a problem in the hot
end flow conditioner: No sign of temperature change is recorded even very close to the flow
conditioner disk inside the pulse tube during the phase of refilling the (according to our
view of the working cycle relatively cold) exchange gas from the hot end. In addition we see
a strong overheating of the hot end of the pulse tube during cooldown time (Figure 7). The
heat is not efficiently removed from the hot end.
To overcome this problem, a variety of flow conditioning insets were tested at the hot
end of the pulse tube. Figure 8 summarizes the results. The originally used punched disc
(Figure 8 a and c) causes a phase shift between the temperature wave above and below the
disc. The result is also a shift in the volume wave due to
Figure 7. Temperature vs. time development during the cooldown phase of the refrigerator at
various points of regenerator and pulse tube. Note the strong overheating of the hot end during
initial cooldown and during operation under heat load at the cold head!
Figure 8. Temperature waves at the hot end inside the pulse tube and behind the respective
flow parallelizer: a) and b): experimental setup, c):punched disk as flow straightener, d): 4, and e):12
wire gauze disks. Note the strong and very setup dependent phase shift between pressure and
temperature wave caused by the flow conditioners!
(ideal gas, assuming p=const.), and finally a change in the phase shift in Eq. (6).
Alternative flow conditioners at the hot end are stacks of wire gauze dics (Figure 8b, d, e,
compare also Figure 4a). The phase shift of the temperature wave over this stack depends
strongly on the number of discs used. Almost no shift is seen for 4 discs (Figure 8D) and
almost 180° shift for 12 disks (Figure 8E). The regenerator performance behaves similar:
both, the ultimate cold head temperature (35 K and 29 K respectively) and the cooling
power at 80K (30 W and 70 W respectively) improve strongly when changing from 4 to 12
discs. Further increase of the number of discs in the hot end flow straightener does not
improve refrigerator performance.
Figure 9. Power characteristic of the FVPTR with different flow straighteners at the hot end
of the pulse tube.
Figure 9 contains the cooling power characteristics of the cooler with the punched disc
(dashed line) and the stacked gauze hot end flow straightener (solid line). While the
available cooling power is significantly increased, we are still concerned about the
regenerative effect of the new flow conditioner: the temperature waves of the gas inside the
disc stack in Figure 8d) and e) show a very small amplitude, indicating strong hot end loss.
CONCLUSIONS
A single stage FVPTR was designed and constructed. As the preliminary experiment the
performance of the pulse tube was investigated. The essential results can be summarized as
follows:
z The minimum no-load temperature achieved is 28 K. The observed net cooling power at
70 K is 67 W.
z The influence of several parameters on the FVPTR performance has been studied to test
their sensitivity for potential modifications in future improved designs.
z The hot-end loss due to induced regenerative effects is incompatible with the adiabaticity
required for the maximum enthalpy flow, and thus is a main loss source.
A successful design of the hot end of the pulse tube has to match the requirements of low
mixing and low turbulence of the gas by using as little thermal mass as possible for the insets
in order to prevent hot end losses. The current work has mainly focused on the reduction of
these two types of losses.
ACKNOWLEDGMENT
The authors wish to acknowledge the financial support provided for this project by the
German BMBF under grant number FKZ 13 N 7395.
REFERENCES
1. Blaurock.J., Hackenberger, R. Seidel, P. and Thürk, M., “Compact Four-Valve Pulse Tube Refrig-
erator in Coaxial Configuration,” Cryocoolers 8, Plenum Press, New York (1995), pp. 395-401.
2. Storch, P.J. and Radebaugh, R., “development and experimental test of an analytical model of the
orifice pulse tube,” Adv. Cryog. Eng., vol. 33 (1988), p 851.
3. Radebaugh, R., “A review of pulse tube refrigeration,” Adv. Cryog. Eng., vol. 35 (1990), p. 1191.
4. Hoffmann, A., Wild, S., “A Model for Analyzing Ideal Double Inlet Pulse Tube Refrigerators,”
5. Walker, G., Cryocoolers, Part I:Fundamentals, Plenum Press, New York (1983).
6. Ackermann, R.A., Cryogenic regenerative heat exchangers, Plenum Press, New York (1997).
7. Lee, J.M., Kittel, P., Timmerhaus, K.D., Radebaugh, R. “Flow patterns intrinsic to the pulse tube
refrigerator,” Proceedings of the 7th Int. Cryocooler Conf., PL-CP-93-1001, Part 1, Kirtland AFB
(1993), pp. 125-139.
8. David, M., Merechal, J.-C. and Encrenaz, P., “Measurements of instantaneous gas velocity and
temperature in a pulse tube refrigerator,” Adv. Cryog. Eng., vol. 37 (1992), pp. 939-943.
9. Gerster, J., Thürk, M., Reißig, L., Seidel, P., “Hot end loss at pulse tube refrigerators,” Cryogenics,
vol. 38 (1998), pp. 679-682.
10. Thürk. M., Brehm, H., Wagner, R., Gerster, J., Seidel, P., “Intrinsic behaviour of a four valve pulse
tube refrigerator,” Proc. ICEC 16/ ICMC, Elsevier Science, Amsterdam (1996), pp. 259-262.
11. Xiao, J.H., “Thermoacoustic heat transportation and energy transformation; part 1: formulation of
the problem,” Cryogenics, vol. 35, no. 1 (1995), pp. 15-19.
Analysis and Experimental Research of a
Multi-Bypass Version Pulse Tube Refrigerator
L.W. Yang, J.T. Liang and Y. Zhou

Cryogenic Laboratory
Chinese Academy of Sciences
Beijing 100080, China
ABSTRACT
One important configuration of a pulse tube refrigerator is the multi-bypass pulse tube refrig-
erator (MPTR). This paper analyzes the working processes of this PT configuration and presents
some experimental results taken using operating frequencies below 5 Hz. The analysis shows that
a MPTR is different from a general multi-stage cooler, but works somewhat like a multi-stage
cooler to a certain extent. Experimental results are presented and compared to former results.
The experiments demonstrate that the multi-bypass channel leads to an improvement in perfor-
mance, and that the function of the multi-bypass more resembles a double-inlet than the first-stage
of a multi-stage cooler.
INTRODUCTION
The multi-bypass pulse tube refrigerator (MPTR) was invented by Y. Zhou in 1993.1 As
shown in Fig. 1, the MPTR contains not only an orifice valve and double-inlet valve at the pulse
tube hot end, but also a connection between the middle of the regenerator and the middle of the
pulse tube; this forms a new structure. In the figure, both the pulse tube and the regenerator have
two parts, just like a two-stage pulse tube cooler.
In the past, experimental investigations of MPTRs were mainly carried out using a valveless
compressor and an operating frequency higher than These tests showed that the MPTR
structure is effective at lowering the refrigeration temperature. However, experiments at lower
Figure 1. Multi-bypass version pulse tube refrigerator.

Figure 2. Difference between gas piston and actual piston.
frequencies are not available. The objective of this paper is to analyze and do experiments using
operating frequencies generally lower than 5 Hz to lay the foundation for further research.
ANALYSIS
Actual Piston and Gas Piston
There are different viewpoints as to the working mechanism of a multi-bypass pulse tube
refrigerator. Some investigators treat it as a special version of a multi-stage cooler,4 and some treat
it just like another double-inlet PT, but at a lower temperature.3 The function of the double-inlet
is easy to understand from Figure 1, where the multi-bypass and double-inlet are similar with just
different positions. When a MPTR is compared with a cold head of a two-stage G-M refrigerator,
its functional similarity to a multi-stage cooler is also easy to understand, as shown in Figure 2.
In Figure 2 (A), with an actual moving piston, three moving volumes are formed: at the hot
end, at the middle part, and at the cold end. By controlling the phase between the pressure wave
and the piston motion, refrigeration is developed in the two lower-temperature volumes.
A similar process take place in the MPTR, as shown in Figure 2 (B). With the assumption of
a gas piston existing in the pulse tube, three moving volumes are formed, just like with a mechani-
cal piston. However, considering the compressibility and fluidity of the gas piston, the middle
chamber in the pulse tube may exhibit other effects that make the actual process more complicated
than a simple mechanical piston. Figure 2 (C) shows this feature. The gas piston in the pulse tube
is actually divided into two parts by a multi-bypass flow chamber. This assumption makes the
multi-bypass resemble a double-inlet.
For a G-M refrigerator, as shown in Figure 2 (A), the middle chamber volume change is
predetermined by the piston area and its stroke. And generally, the flow channel from the regen-
erator to the middle chamber is made large enough to minimize flow resistance; thus the gas flow
rate into the middle chamber is determined by the piston position and the pressure wave.
However, a MPTR is different. In Figure 2 (B), the cold-chamber movement is determined
by the hot-end gas movement, the gas piston change, and the middle chamber change. In contrast,
the middle chamber volume change is generally not determined by the hot-end gas movement.
This makes the phase relationship required for refrigeration a problem. In fact, the flow channel
connecting the pulse tube and the regenerator must be specially designed to limit or control the
middle chamber amplitude. This makes the phase relation of the middle chamber in a MPTR
quite different from that of a G-M refrigerator. Flow resistance becomes very important in a
MPTR.
Flow Resistance and Mass Flow

For a MPTR, a small opening of the multi-bypass valve will form a small middle chamber,
and a large opening will result in a large one. The middle chamber volume is controlled by the
RESEARCH OF A MULTI-BYPASS VERSION PT REFRIGERATOR 339
multi-bypass tube flow area, and in this process, the pressure wave before and after this flow area
becomes very important. As a comparison, at the cold end of pulse tube, the connection tube is
generally large enough to minimize flow resistance.
In the pulse tube, flow resistance is small, and pressure wave is of little relation to
position. Assuming the pressure wave at the multi-bypass position of the regenerator is the
mass flow into and out of the pulse tube multi-bypass is determined by.
where c is the flow coefficient, A is the flow area, and
Consider at time gas begins to flow into the pulse tube, then the partial
volume change will be:
When me pressure in the pulse tube changes, will change correspondingly in the pulse
tube:
where n is component to consider gas change, and for an adiabatic process

The bypass position volume change will thus be:
changes from zero to a maximum and then to zero, and generally has a special form.
Refrigeration Effect
Eq. (4) only gives the potential middle chamber volume change. The function of this flow
and the extent to which it contributes to refrigeration needs further consideration.
Assuming the pulse tube pressure is when some gas with temperature flows into
pulse tube through the multi-bypass, and when this gas with temperature Tmo flows out of the
regenerator, the pressure in pulse tube is Similar to Eq. (3), the relationship is as follows:
If and there will be a temperature drop effect when gas stays in the pulse
tube. Considering the whole process, gas flowing into the pulse tube should have an average
higher inlet pressure. In general, such a condition is obtainable in experiments with a particular
orifice opening, and then refrigeration will happen. But the degree of phase between pressures is
mainly determined by flow resistance, and the refrigeration effect is generally small. This has
been explained by others.5
REFRIGERATION SYSTEM
Design of the system is based on the viewpoint of multi-stage cooler.
As shown in Figure 1, the regenerator includes two parts: a high temperature part and a low
temperature part. The high temperature part was 120 mm long, with 28 mm outside diameter, and
0.45 mm thick stainless steel tube; it was filled with 250-mesh stainless steel screen. The low
temperature part was 110 mm long, with 20 mm outside diameter and 0.45 mm thick stainless
steel tube; it was filled with 250-mesh stainless steel screens for the first 30 mm length, and 0.2-
0.3 mm lead spheres for the rest. The pulse tube also includes two parts. The high-temperature
part is 150 mm long, with 18.4 mm outside diameter, and 0.2 mm thick stainless steel tube. The
low-temperature part is 100 mm long, with 14.4 mm outside diameter, and 0.2 mm thick stainless
steel tube. The joint between the two parts is filled with 20 pieces of 60-mesh copper screen. The
connection of the multi-bypass is made with a copper tube with an inner diameter of 1.86 mm. By
changing the minimum flow area of the multi-bypass tube, different experiments were carried out.
The pulse tube refrigerator was driven by a 750-W air-conditioning compressor. The fre-
quency of the rotary valve was easily changeable from 1 Hz to 10 Hz. All tests were done under
a pressure of about 10 bar and 700 W input power to the compressor.
In the experiments, the main parameters measured were:
a. Temperatures. There are three as shown in Figure 1: at cold tip, at the multi-bypass
position, and near the pulse tube hot end.
b. Pressure waves. One is before the regenerator, the other in the pulse tube.
c. Cooling capacity at the cold tip.
d. Multi-bypass minimum flow area.
EXPERIMENT RESULT AND ANALYSIS
Effect of Multi-bypass Area

The most important aspect of an MPTR is the multi-bypass effect. In experiments, for a fixed
multi-bypass flow area, the orifice and double-inlet were optimized to reach the lowest tempera-
ture. Six groups of experiments were done and the main parameters measured are listed in Table 1.
In the table, different forms of flow channels for multi-bypass have been converted to flow area.
The most important effect of multi-bypass area is temperature. There is an optimized flow
area to achieve the minimum cold-tip temperature this is the same as former experiments.2,3
Temperature at the multi-bypass position drops continuously with the opening of the multi-
bypass. This is easy to understand. Larger opening of the multi-bypass represents larger flow rate
and larger refrigeration at the bypass position, and thus the temperature will drop. This can also
explain the temperature of As to the relation between and it is possibly due to the mass
flow distribution. With pulse tubes, increase of the bypass flow will decrease the cold-tip flow.
An optimized lowest temperature is thus gained with a decrease of gas flow at the cold tip in
combination with a relatively high regenerator efficiency, and a relatively low middle temperature.
The pressure wave is evidently affected by the multi-bypass. Through bypass flow, flow resis-
tance is much smaller than that without multi-bypass. From no multi-bypass to largest multi-
bypass, the pressure drop changes from 0.742 to 0.923, and this makes the pressure
amplitude in the pulse tube larger.
Figure 3. Cool-down curves for three different openings of the multi-bypass valve including: no
multi-bypass maximum multi-bypass and optimum multi-bypass
Orifice and double-inlet opening are important parameters for a MPTR. Both use standard
needle valves. From Table 1, optimization of the opening of the orifice and double-inlet valve is
seen to have little relation to the flow area of the multi-bypass. This is different from that for two-
stage pulse tube refrigerators, where there are two pulse tubes, and each one has a best value. In
contrast, this feature is more like the function of a double-inlet, whose effect on orifice opening is
small. And this is the rationale for a high- and low-temperature double-inlet version.5
Cooling Down Curve and Cooling Capacity
In past experiments with a 16-Hz MPTR, there is a slow cool-down rate in comparison to an
OPTR or DPTR.2 However, this feature was not evident in this experiment, as shown in Figure 3.
The three curves in Fig. 3 are for, respectively: 1) largest multi-bypass, 2) no multi-bypass, and 3)
optimized multi-bypass. Although the lowest temperatures are different, the cooling rates are
similar.
Refrigeration capacity is very important for a cryocooler. Some experiments for MPTRs have
shown a temperature rise per unit cooling capacity that is higher than for a DPTR.2 It seems that
in this experiments, such a tendency is not very evident, as shown in Figure 4. The results for all
five groups of multi-bypass designs (groups 1 to 5 as defined in Table 1) are almost parallel and
slightly lower in slope than the DPTR results (group 6).
Figure 4. Cooling capacity of each of the six experiment groups.

Figure 5. Effect of double-inlet opening on refrigeration temperature at different orifice opening.
Relation of Orifice and Double-inlet

With the optimized multi-bypass flow area of orifice and double-inlet are adjusted
to reveal their effect on refrigeration temperature. The typical result is shown in Figure 5. There
are four groups of orifice openings. For every orifice opening, there is a best corresponding
double-inlet opening. It seems that larger opening of the orifice will need a larger opening of
double-inlet in order to reach the lowest temperature. Also, the best orifice opening for the OPTR
to gain the lowest temperature doesn' t correspond to the best orifice opening for the DPTR to gain
its lowest temperature. However, the temperature difference is only 2-4 K.
Frequency Effect
In the experiment, frequency is easy to adjust. Tests were conducted to optimize the fre-
quency. As shown in Table 2, the optimized frequency is about 2.2 Hz. And throughout the test,
this frequency led to the best results in comparison to other frequencies. Also, as with previous
experiments, frequency strongly affects the optimum opening of the orifice, while its effect on the
opening of the double-inlet is quite small.
DISCUSSION
Other results on the parameters influencing the MPTR are available. However, they are based
on a pulse tube cooler with a relatively high refrigeration temperature. In fact, the high tempera-
ture part of the regenerator and the high temperature part of the pulse tube once formed one pulse
tube refrigerator. That pulse tube refrigerator could easily reach 40 K. And with the second part
of the regenerator added, the refrigeration temperature is higher.
The whole regenerator in this paper came from one two-stage pulse tube refrigerator, and this
cooler reached a temperature of 11 K at the second stage coldtip when the high-temperature stage
was cooled to 50 K. Thus, a multi-bypass pulse tube refrigerator should not be treated as general
multi-stage cooler.
In this experiment, one thing that needs to be improved is to further lower the temperature of
high temperature part. However, the design of this part is not easily isolated, as it is also related to
the low-temperature part. Another future consideration is how to make the MPTR act like a two-
stage cooler.
CONCLUSION
The analysis shows that a MPTR is different from a general multi-stage pulse tube refrigera-
tor. This makes it difficult to design. In this investigation, several groups of experiments were
conducted to reveal the features of the MPTR at low frequency. Throughout the work, it could be
concluded that the design of the MPTR should not follow the idea of a general multi-stage unit.
Also, the experiments have shown that the MPTR resembles a double-inlet pulse tube more than
a multi-stage unit.
REFERENCES
1. Zhou, Y., Han, Y.J., “Pulse tube refrigerator research,” 7th International Cryocooler Conference
Proceedings, Air Force Phillips Laboratory Report PL-CP--93-1001, Kirtland Air Force Base, NM,
April 1993, p. 147.
2. Wang, C., Wang, S.Q., Cai, J.H., Zhou, Y., “Experimental study of multi-bypass pulse tube refrig-
erator,” Cryogenics, vol. 36 (1996), pp. 605-609.
3. Cai, J.H., Wang, J.J., Zhu, W. X., and Zhou, Y., “Experimental analysis of the multi-bypass prin-
ciple in pulse tube refrigerator,” Cryogenics, vol. 34 (1994), pp. 713-715.
4. Olson, J.R., Kotsubo, V., Champagnes, P.J., Nast, T.C., “Performance of a two-stage pulse tube
cryocooler for space application,” Cryocoolers 10, Kluwer Academic/Plenum Publishers (1999), pp.
163-170.
5. Yang, L.W., Zhou, Y., Liang, J.T., “Research of pulse tube refrigerator with high and low tempera-
ture double-inlet,” Cryogenics, vol.39 (1999), pp. 417-423.
Experimental Study of the Heat Transfer
in Pulse Tubes
S. Jeong, K. Nam, M. G. Kim, H.-M. Chang* and E. S. Jeong*

Korea Advanced Institute of Science and Technology
Department of Mechanical Engineering
Taejon, 305-701, Korea
*Hong Ik University, Dept. of Mech. Engineering
Seoul, 121-791, Korea
ABSTRACT
The present study has been conducted to observe the details of heat transfer under pulsat-
ing pressure and oscillating flow in a pulse tube. An experimental apparatus was fabricated to
measure the gas temperature, wall temperature, pressure, and the instantaneous heat flux in-
side a pulse tube. The measured gas temperature and heat flux must be corrected to compensate for
their finite time constant under oscillating flow conditions. In experiments performed from 1 Hz to
3 Hz, the phase difference between the instantaneous heat flux and the gas-wall temperature differ-
ence was clearly observed. The experimental heat fluxes were compared to theoretical correlations
such as the Complex Nusselt Number Model (CNNM) and the Variable Coefficient Model (VCM).
In general, the absolute value of the heat flux predicted by the CNNM was greater than that of the
VCM. The experiment confirmed the validity of the VCM for the instantaneous heat flux under the
pulsating pressure and oscillating flow in the warm end of the basic pulse tube.
INTRODUCTION
Oscillating flow under pulsating pressure is a common phenomenon in an engineering
system such as a pulse tube cryocooler, Stirling cryocooler, or G-M cryocooler. Due to the
complex physics and lack of experimental data, the heat exchangers in these systems are usu-
ally designed by conventional steady-state heat transfer relations that can not predict the oscil-
lating heat transfer phenomena properly. It is known that a phase shift exists between the
instantaneous heat flux and the gas-wall temperature difference under oscillating flow and
pulsating pressure conditions. The conventional Newton’s law of cooling does not contain a
term that explains this phase shift phenomenon. Kurzweg1 attempted to apply the previous
oscillating heat transfer data to a Stirling cycle heat exchanger. Gedeon2 introduced a complex
Nusselt number using the results of Kurzweg’s work. He obtained a Nusselt number for in-
compressible oscillating flow and showed the existence of the phase shift between the heat
flux at the wall and the gas-wall temperature difference when the oscillatory frequency was
high. Kornhauser3 showed that heat transfer analyses using the complex Nusselt number could
predict the experimental data well for the Stirling engine. Jeong et al.4 obtained two-dimensional

velocity and temperature profiles for the oscillating flow caused by pulsating pressure that was
induced by piston motion. He suggested a new heat transfer relation, so called the ‘Variable
Coefficient Model’. Lee et al.5 studied analytically the heat transfer of the Stirling cycle heat
exchanger. He obtained the gas and wall temperature profiles when the axial gradient of the wall
temperature was not constant and the oscillating flow only existed. He also examined the effect
of the frequency and the maximum displacement. Jeong et al.6 installed the heat flux sensor on
the outer surface of the heat exchanger for the basic pulse tube and calculated the instantaneous
heat flux at the inner wall using the measured data. In their experiment, however, the calculated
heat flux at the interior wall had an uncertainty due to the capacitance effect and the complex
geometry of the heat exchanger.
This paper describes the experiment of the instantaneous measurement of the heat flux and
the temperature at the heat exchangers of the pulse tube under oscillating flow and pulsating
pressure. The measured heat flux data were compared with the theoretical predictions that had
been previously developed.
EXPERIMENTAL CONFIGURATIONS
An experimental apparatus was fabricated as shown in Fig. 1. At the inlet of the pulse tube,
the stainless steel mesh (#200) was stacked to make one-dimensional flow between 1/4 inch tube
and 1 inch tube. Its thickness was 4 mm. The cold-end heat exchanger was made of 1 mm thick
copper tube and nicely fitted to the pulse tube. To install the heat flux gauge on the inside wall of
the heat exchanger, the flange with 7 mm thickness was brazed to the heat exchanger and
assembled with bolts. The warm-end heat exchanger had the same shape as that of the cold-end
heat exchanger, but it had the water jacket to cool the warm-end by the cooling water as shown in
Fig. 2.
Figure 1. Schematic diagram of the experimental apparatus. (CHX: cold-end heat exchanger,
WHX: warm-end heat exchanger).
EXPERIMENTAL STUDY OF HEAT TRANSFER IN PULSE TUBES 347
Figure 2. Warm-end heat exchanger. Figure 3. Sensor Installation (T/C: thermocouple).
Fig. 3 illustrates the sensor installation at the heat exchanger. A fast response thermocouple
(OMEGA Model EMQSS-010E, type-E) was inserted into the heat exchanger by reducer fitting.
The wire diameter of this thermocouple was 0.038 mm and its junction was exposed to the flow
field for the fast response characteristics. The heat flux gauge was attached to the inner wall of
the heat exchanger. It was a thick micro-foil type sensor and its time constant for the step
input was 20 ms according to the manufacturer. Thermal resistance of the heat flux gauge was
very small compared to that of the convection with the helium gas. The difference of the heat
flux between the case with the sensor and without the sensor was about 5 % of the heat flux
value. The heat flux gauge had also the T-type thermocouple that could measure the wall
temperature of the heat exchanger.
The pulse tube was made of 1 mm thick stainless steel tube with outer diameter of 25.4 mm
and the length of 200 mm. Helium compressor (CTI-cryogenics Model 8200) was used to supply
helium gas as working fluid. The rotary valve system provided the pulse tube with pulsating gas
pressure and flow. The frequency of the pulsating pressure was adjusted by the rotational speed
of the rotary valve, which was controlled by the stepper motor and the function generator.
The strain gauge type pressure transducer (Sensym Model ST2000) was installed at the inlet
of the pulse tube to monitor the pulsating pressure. The signal of the heat flux gauge was so
small (the order of and so sensitive to the environmental noise that this signal was pre-
amplified by isolation pre-amplifier (YOKOGAWA Model 313100-61E). All the experimental
data were acquired by data acquisition board (Keithley Model DAS 1600) and stored in the
personal computer when the cyclic steady-state was reached.
EXPERIMENTAL RESULTS AND DISCUSSION

As mentioned earlier, utilizing complex Nusselt number concept, Kornhauser3 proposed a
heat transfer correlation as follows:
where
k = thermal conductivity of the gas
= hydraulic diameter of the pulse tube
f = oscillating frequency of pressure wave
= real part of the complex Nusselt number
= imaginary part of the complex Nusselt number
The complex Nusselt number in Eq. (1) was determined by using the least square method from
the experimental data.3 He also presented the complex Nusselt number for various Peclet number
On the other hand, Jeong et al.4 has derived the following relation from the
two-dimensional energy equation for the case of the fluid flow between two flat plates. The heat
flux at the wall was expressed as follows:
where
2H = distance between two flat plates

R = ideal gas constant
P = pressure of the gas
The thermal conductivity of the gas and the wall temperature were assumed to be constant when
Eq. (2) was derived. When Eq. (2) was applied to the pulse tube, 2H was assumed to be the
diameter of the pulse tube.7 The fundamental difference of Eq. (2) from Eq. (1) is that the heat
flux term associated with the oscillation is derived from the direct instantaneous pressure change.
Although the temperature gradient term in Eq. (1) is replaced by the pressure gradient term in Eq.
(2), two equations describe virtually the same physical phenomenon because the temperature is
varied by the pressure change. Kornhauser’s relation contained the complex Nusselt number that
must be determined from the extensive experiment, but Jeong’s relation included no empirical
constant. Kornhauser’s relation was also known as the Complex Nusselt Number Model
(CNNM) and Jeong’s relation as the Variable Coefficient Model (VCM).
Under the pulsating pressure and oscillating flow, the heat flux and the gas-wall temperature
difference have been measured as shown in Fig. 4. In this paper, the experimental data are shown
only for the case of the closed orifice (basic pulse tube type) and no load condition at the cold-
end heat exchanger. There was an apparent phase shift between the heat flux and the temperature
difference at 2 Hz, but the phase shift was not clear at 1 Hz. The heat flux was also calculated by
Eq. (1) and Eq. (2) using the measured temperature and the pressure. The predicted flux was
compared to the measured heat flux at the cold-end and the warm-end heat exchanger as shown
in Fig. 5 and Fig. 6. The negative heat flux means the heat flow from the gas to the wall. At the
warm-end heat exchanger, both the CNNM and the VCM can predict the heat flux reasonably
well. On the other hand, Fig. 6 shows that both the results from the two theoretical relations
failed to predict the experimental data at the cold-end heat exchanger. The CNNM predicted the
heat flux variation pretty well, but with large amount of discrepancy. This result indicated that
the complex Nusselt number suggested by Kornhauser could not be applicable to the cold-end
heat exchanger. Since Kornhauser correlated the complex Nusselt number only by Peclet number
his empirical correlation could not represent the general case. Therefore, it is suggested
that another non-dimensional parameter, such as maximum Reynolds number
Figure 4. Heat flux and gas-wall temperature difference at the cold-end heat exchanger: (a) 1
Hz (b) 2 Hz
Figure 5. Heat flux at the warm-end heat exchanger: (a) 1 Hz (b) 2 Hz
Figure 6. Heat flux at the cold-end heat exchanger: (a) 1 Hz (b) 2 Hz.
should be included in the complex Nusselt number. Here, is the maximum mass
flow rate of the oscillating flow, is the viscosity of the gas and A is the cross-sectional area.
The VCM did not match the variation pattern of the heat flux in one cycle, but the cyclic
integrated value was more close to the measured value than that of the CNNM. Table 1 shows
the comparison of these values.
The CNNM had another difficulty in its application at higher frequency. Since it contains a
time derivative term of the measured temperature as shown in Eq. (1), the prediction of the heat
flux by the CNNM may require one more calculation step. As shown in Fig. 7(a), a small
variation in gas temperature was amplified by the time derivative operation and the heat flux
from this gas temperature had a large oscillation as shown in Fig. 7(b). This inevitable error
amplification could be virtually reduced by approximating the measured temperature to a smooth
function like sinusoidal wave (continuous line in Fig. 7). Fig. 8 shows, thus, the predicted heat
flux of the CNNM by this error reducing method.
CONCLUDING REMARKS
The experiment was performed to the heat exchangers of the basic pulse tube for the
frequency of 1 ~ 3 Hz. From this experiment, we recognized the following results.
(1) The VCM (Variable Coefficient Model) could predict the heat flux in the warm-end heat
exchanger of the pulse tube better than the CNNM (Complex Nusselt Number Model).
Figure 7. Heat flux from the CNNM using the curve fitted gas temperature (3 Hz, warm-end).
Figure 8. Comparison of the heat flux using the calculated complex Nusselt number. (3 Hz,
warm-end).
(2) Neither the VCM nor the CNNM predicted the heat flux at the cold-end heat exchanger
where the oscillating flow effect was noticeable. The better model is necessary for the
oscillating flow with large amplitude.
ACKNOWLEDGMENT
This research was supported by the Korea Research Foundation under the contract no. 1998-
018-E00164 and partially supported by the Brain Korea 21 project. The authors would also like
to acknowledge the support of the Ministry of Science and Technology through the Dual Use
Technology project (grant no. 99-DU-04-A-02).
REFERENCES
1. Kurzweg, U. H., "Enhanced heat conduction in fluids subjected to sinusoidal oscillations", J. Heat
Transfer, Vol. 107 (1985), pp. 459-462.
2. Gedeon, D., "Mean-parameter modeling of oscillating flow", J. Heat Transfer, Vol. 108 (1986), pp.
513-518.
3. Kornhauser, A. A., Gas-wall heat transfer during compression and expansion, Sc.D. Thesis, Dept. of
Mech. Eng., Massachusetts Institute of Technology, Cambridge, MA (1989).
4. Jeong, E. S. and Smith, Jr., J. L., "An analytic model of heat transfer with oscillating pressure",
General papers in Heat Transfer, ASME, Vol. 204 (1992), pp. 97~104.
5. Lee, D. Y., Park, S. J. and Ro, S. T., "Heat transfer by oscillating flow in a circular pipe with a
sinusoidal wall temperature distribution", Int. J. Heat Mass transfer, Vol. 38, No. 14 (1995), pp.
2529-2537.
6. Jeong, S. and Nam, K., "An experimental study on the heat transfer characteristics of the heat
exchangers in basic pulse tube refrigerator", Crycoolers 10, Plenum Publishers, New York (1999), pp.
249-256.
7. Jeong, E. S., "Secondary flow in basic pulse rube refrigerators", Cryogenics, Vol. 36, No. 5 (1996),
pp. 317-323.
Shuttle Loss in Pulse Tubes
L.W. Yang
Institute of Applied Physics
Justus-Liebig-University Giessen
D-35392 Giessen, Germany
ABSTRACT
Pulse tube refrigerators have the unique feature of a hollow tube replacing the cold piston. This
paper analyzes some losses resulting from this hollow tube due to the oscillating flow; the losses
are quite similar to the shuttle loss and pumping loss in a Stirling or G-M refrigerator. The analysis
starts with surface heat pumping in a basic-type pulse tube cooler. Then, flow features and the
origin of losses are further explained. As an important parameter, boundary layer thickness is pre-
dicted. The analysis shows that, due to the large temperature difference, the distribution of bound-
ary layer thickness along the pulse tube changes greatly. For different boundary layer thicknesses,
shuttle losses and pumping losses differ significantly. With the predicted boundary layer thickness,
the amplitude of the shuttle loss and pumping loss are just the opposite of what they are in a
traditional Stirling or G-M cooler. Pumping gas loss becomes an important loss in the pulse tube
refrigerator.
INTRODUCTION
In a cryocooler, evaluation of the losses is as important as the refrigeration mechanism.1 Gen-
erally, the various kinds of losses occupy a major part of the gross refrigeration. For example, in a
Stirling refrigerator, there are regenerator heat transfer losses and flow resistance losses, displacer
pumping gas losses and shuttle losses, conductance losses, cold tip radiation losses, and heat-ex-
changer losses. These losses make up about 80-90% of the gross refrigeration at low temperatures.
Pulse tube refrigerators have developed very fast since 1984, with the orifice and reservoir
being added at the hot end of the pulse tube.2 These improvements gave rise to a revolutionary
improvement in the refrigeration mechanism. Many theoretical methods have been used to explain
pulse tube operation based on different physical models. However, as to loss evaluation, especially
as to the loss in the pulse tube, no detailed introduction could be found in the literature.
The pulse tube working process has been treated in many references now. Some researches
consider the flow or working process in the pulse tube itself, including flow pattern research and
numerical simulation.3, 4 However, these results are not directly related to the evaluation of losses
and the design of pulse tubes. Recently, the unique gas movement and potential heat transfer in the
pulse tube was explained with a change of temperature ratio.5 Other researchers have also pointed
out the problem of shuttle losses in a pulse tube.6
Here shuttle loss and pumping loss are analyzed and calculated in detail for the first time.

Figure 1. Structure of pulse tube refrigerator and gas movement.
Figure 2. Gas movement characteristics of BPTR.
ORIGIN OF LOSSES
Limit of Surface Heat Pumping
The pulse tube refrigerator was invented in 1963 by Gifford and Longthworth.7 This first pulse
tube is referred to as a basic-type pulse tube refrigerator (BPTR), and its working mechanism is
attributed to surface heat pumping as shown in Fig. 1 (BPTR) and Fig. 2.
In Figure 1, accompanying the movement of the compressor piston, there is another gas move-
ment at the jointing position between the pulse tube and the regenerator. At this jointing position
the temperature is lowest and a temperature gradient is formed along the pulse tube wall. For such
a process, the working mechanism and its limitations are shown in Figure 2.
Assuming an adiabatic process in the pulse tube, with a pressure increase, every part of the gas
in the pulse tube will move further into the tube, with the change in position and temperature
described by Eqs. (1) and (2):
where is the adiabatic component and y is the normalized pulse tube position from the hot end.
Figure 2 illustrates the gas moving process and corresponding temperature at four initial pulse
tube positions, y = 0.2, 0.5, 0.8 and 1.0, where 1.0 represents cold end. First look at the curve with
the hot end temperature and cold end temperature Evidently, at any position
the temperature of moving gas is higher than the temperature of that based on a linear distribution.
Then, during the high pressure process, moved gas may transfer heat to the corresponding pulse
tube wall. Thus, the whole process results in heat being pumped to a higher-temperature position
from an initial lower-temperature position. Such a process is called surface heat pumping. This
shows why there is temperature drop in the pulse tube refrigerator.
SHUTTLE LOSS IN PULSE TUBES 355
However such a process has its limit or shortcoming with low temperature decreasing.
Let’s look at in Figure 2. At the position of 1.0, gas temperature change almost
coincides with assumed linear temperature distribution. Evidently, surface heat pumping will
have problem. With even lower temperature, the inverse process happens compared to
y = 0.8 and 1.0 of situation. Though the gas temperature increase upon compression,
this temperature is much lower than the temperature based on a linear distribution. The
mechanism of surface heat pumping at certain temperature ratio will terminate, and the process
turns to be a loss.
Through controlling gas temperature distribution along pulse tube, the lowest temperature
reached in theory may be low. For example, assuming that the temperature distribution has
the following form:
n= 1 means linear distribution of temperature along the wall, n>1 means there is a slow
temperature change at the hot end, n <1 means a slow temperature change at the cold end. Based
on, such a hypothesis, a group of ideal lowest temperature can be calculated as shown in Table 1.
Table 1 shows that with smaller value of n, a low temperature of about 70K is attainable.
But if temperature distribution at n= 0.2 is plotted, we will find that this is a non-realistic curve,
because the low temperature part is long and flat.
Gas Movement with Reservoir and Double-inlet

As shown in Figure 1 (OPTR DPTR), after the invention of orifice version pulse tube
refrigerator (OPTR)2 and further double-inlet version pulse tube refrigerator (DPTR),8 the above
mentioned low temperature limit has been overcome.
Different to the BPTR, the gas in the pulse tube consists of three parts for the OPTR.
Besides the gas section (here called gas piston) staying in the pulse tube and the gas section
flowing into pulse tube through the cold end, now a gas section flowing through the hot end of
pulse tube is added, as shown in Figure 1.
Neglecting mechanism, complex moving process of different gas section is considered. For
basic version, cold end flow position change has the following form according to
adiabatic process:
where is the pulse tube length, pressure wave in the pulse tube, lowest pressure
in the pulse tube, are the gas piston length at average pressure
For orifice version, assuming that hot end gas section movement is and the reservoir
has a large volume, generally there is very small pressure change in the reservoir, and position
change at hot end has the following relation:
where represents pressure in reservoir.

With double-inlet, another gas section will be added at the hot end according to
pressure difference of the following relation:
where is pressure wave at inlet of regenerator.

And then hot end flow with orifice and double inlet change to be:
where are coefficients to determine their amplitude.

With hot end flow predetermined, cold end gas flow is easily to attain. Assuming that the
gas piston change is adiabatic, a simple formula to reveal their relation is that:
All three gas groups move in the pulse tube continuously. Because of the hot end flow, the
refrigeration process is different to Figure 2, and at the same time, loss may occur.
Loss due to Gas Piston Flow
A typical result of the actual flow feature of OPTR is as shown in Figure 3.
It’s easy to notice the refrigeration mechanism. Different to Figure 2, cold tip gas flow
position forms one circle in clockwise direction. The corresponding inlet pressure at
is larger than the lowest pressure The circled area represents the theoretical
refrigeration capacity or work done to gas piston near it. Comparing to Figure 2, the circle in the
cold end is due to the circle in the hot end.
Let’s look at the gas piston movement. In the Figure 3, six groups of length and position are
depicted by double-arrow line. The length of gas piston changes only with pressure. But we
should notice that the biggest length is not at cold tip, nor the shortest length at the hot tip. Its
change with position is complex, while the process is from cold tip to hot tip and then back.
If we depict the flow curve of any point of gas piston, a circle will form and contact with
tube wall and heat transfer is inevitable. What we are really interested in is that whether this heat
transfer contributes to refrigeration or a kind of loss.
First look at cold tip. Different to basic version pulse tube, at the highest pressure, which
corresponding to the highest temperature of the gas piston, the gas piston is not corresponding to
the largest position of but at the middle of the pulse tube, and at the largest position of
the pressure drops by a certain extent. Compared to Figure 2, gas moving distance is much
larger with hot end flow adding. Also important is temperature ratio. If the lowest temperature is
lower than that determined by Table. 1, surface heat pumping will not work for cold tip, and loss
will be inevitable.
For basic version pulse tube, hot end of pulse tube always contributes to surface heat
pumping. From Fig. 3, with pressure increasing, increases first while basic version
decreases: the highest pressure is not at the shortest position, the lowest pressure is not at the
lowest position. Heat transfer is inevitable, but not transferring heat to higher temperature
position. The loss is inevitable.
In fact, the loss resulting from the movement of gas piston is just like the actual cold piston
or displacer in Stirling refrigerator or G-M refrigerator.
The most evident one is shuttle loss. When considering this loss, we could assume that gas
piston length itself could keep unchanged but the position is constantly changing. Then heat is
transferred from low temperature to high temperature gradually according to wall and gas piston
contact. This loss in Stirling refrigerator is large and important.
Another is pumping gas loss. This loss is to consider the effect of gas gap or gas clearance
between gas piston and wall. Due to viscosity, this gas gap adheres to tube wall and can not
move like gas piston. Then some more gas will go into this gas gap from cold end or gas piston
during compression and be released during expansion, and the corresponding heat transfer forms
the loss. This loss is generally small in Stirling or G-M refrigerator.
Figure 3. Typical gas piston movement in pulse tube.
METHOD TO CONSIDER SHUTTLE LOSS

Shuttle loss is considered similarly to traditional G-M type or Stirling refrigerator.1
Total heat transfer amount could be expressed as:
where Q represents heat transfer amount in one cycle, is average moving distance of the gas
piston and f is frequency.
According to the mechanism of shuttle loss, one cycle heat transfer amount:
where is gas conduction coefficient, A is heat transfer area, is temperature difference

between wall and gas piston, is heat transfer time, is distance between tube wall and gas
piston.
In theory, A, are different to traditional ones, and they are changing with
pressure wave and flow of orifice and double-inlet. These parameters should be considered in
priority.
Length of Moving Distance
in Eq.(10) represents moving distance. In ST or GM cooler, it is just related to cold

chamber length minus potential dead volume length. But for pulse tube, its value is not so easy to
determine because hot end part and cold end part generally have different moving length and the
two are combined together. Here based on Eq. (4-9), the result is deduced.
For basic version pulse tube, the largest moving distance is determined by:
For orifice version, the largest value at the hot end should be determined first, which
depends on orifice valve. Then we could roughly predict the largest value at the cold end:
For double-inlet version, we should have the largest movement at the hot end resulted
by double-inlet valve, and then hot end largest movement is:
while the cold end movement becomes:
With the upper largest values, average moving distance will be:
Length of Gas Piston

Though in Eq. (9) gas piston length at average pressure is given, its actual value is
complex to determine. The problem is that have phase relation.
From Eq. (9), at average pressure, we could get:
For basic version, the value is easy to get according to Eq. (4):
For orifice version, we could assume pressure ratio is not large and reservoir is large, then:
For double-inlet version, if generally, we have:
Besides average length of gas piston, we should notice that the actual gas piston length
changes with pressure:
For high pressure ratio such as 1.8, for low pressure ratio such as 1.2,
The gas piston of low pressure ratio more resembles actual piston.
Heat Transfer Temperature Difference

Along gas piston, a simple and reasonable assumption is linear distribution of temperature.
And in actual, wall temperature of pulse tube is in linear distribution for orifice version, but this
may be destroyed by double-inlet. Here the nonlinear effect is neglected.
When the gas piston is located at certain position the heat transfer temperature
difference is:
means certain middle position, where gas piston temperature distribution is the same to wall
temperature distribution. It should be noticed that is joint position, which separates heat
conducting to tube wall and from tube wall, or where heat transfer temperature difference
changes its direction.
Gas Gap Between Wall and Gas Piston
The most important and difficult consideration is about gas gap . If neglecting viscous of
gas, there should be no so called gap. But this will result very large loss according to Eq. (11).
The simplest consideration is to use steady state boundary layer method. The following
formula is the simplest method to calculate the boundary layer thickness (BLT) of laminar flow
through a large plate:
where is gas viscosity, x distance form plate inlet, gas density, U main flow velocity.
Accurately, flow characteristics in pulse tube does not fit upper equation. There are two
questions. One is distance x. In theory, every part of the gas forming gas piston performs cyclic
flow around certain point, length itself will also expand and shrink with pressure changing, and
actual velocity U also has such a feature. Thus in theory, every time, every position has a
different BLT based on definition of distance x, because any position will be start point of some
gas and at the same time middle position of some gas. Also, viscosity will effect BLT because it
changes with temperature.
For the convenience of calculation, here based on some main parameters in pulse tube, a
simple boundary layer thickness is determined.
Considering actual movement, assume the length parameter in Eq. (22) to be:
Considering frequency f , the average velocity will be:
Neglecting the complex of actual distance and velocity, upper two characteristic parameters
could determine the boundary layer thickness. Then Eq. (22) changes to be:
Eq. (25) means that the boundary layer is determined by gas itself, temperature, pressure and
frequency.
Viscosity and density should also be used. Viscosity is changing evidently with temperature,
while changing little with pressure at 0.1 MPa—5MPa. A best fit line is:
with temperature from 14K to 400K and unit (Accuracy will be 98%).
Density could be calculated use ideal gas state equation with accuracy of about 99%:
for helium4, R =2118.4 J/(kg K).

Calculation has been done to reveal the effects of temperature and pressure on It
shows that with the increasing of temperature and decreasing of pressure, will increase
evidently. The difference will be more than 100 times with same pressure at 15K and at 300K.
This means, according to equation (25), the thickness of boundary layer at high temperature will
be 10 times thicker than that at low temperature. Using helium gas, a group of calculated results
is shown on Table 2.
In Table 2, temperature has a very large effect on BLT. Then along the pulse tube, velocity
distribution form will change greatly from high temperature to low temperature. Also, pressure
and frequency also have evident effect on boundary layer thickness. Higher pressure and higher
frequency corresponds to smaller BLT.
Heat Conduction Coefficient

Gas heat conduction coefficient changes greatly with temperature and slowly with pressure.
Neglecting pressure effect, a simple formula is given for helium gas:
Heat Transfer Area

Considering the boundary layer thickness, the average heat transfer area will be:
where is pulse tube diameter.
Calculation of Shuttle Loss

With upper parameters, Eq. (11) could be integrated to get the result. Because along the gas
piston, parameters are changing, twice integration is needed.
First, according to energy conservation, integration of whole area and whole time will be:
From upper integration, the middle point that separates Q from positive to negative
could be gained.
Then for half cycle amount from we have:
A reasonable and simple method will be:
Upper integration process is possible. For many parameters of actual condition are coupled
together, a computer is needed to get the result easily.
In calculation, for different working conditions, shuttle loss should be calculated differently.
For example, heat transfer time is not easy to determine. There are two simple methods. One is to
consider linear distribution of time with temperature distribution. This is similar to Stirling type
pulse tube refrigerator. At very part, gas move is similar to sinusoidal wave. The other is to
divide a period into three parts: one is for gas piston staying at hot end, one is for gas piston
staying at cold end, and the third is for gas piston staying at middle part where no heat transfer
takes place. This method is suitable for G-M type pulse tube refrigerator.
METHOD TO EVALUATE PUMPING GAS LOSS

Pumping gas loss is different from shuttle loss. Generally, when shuttle loss is large,
pumping gas loss will be small. And it also relates to boundary layer thickness. Using upper
parameters, pumping loss similar to traditional method is deduced:1
The average boundary thickness is defined to be:
Mass flow rate in and out of this average thickness due to pressure wave is:
Adopting following heat transfer equation:1

where is heat transfer coefficient, is equivalent diameter, is specific heat of gas

which is nearly 21 KJ / Kmol · K for helium gas at 50-400K.
Then non-efficiency of heat transfer in the boundary layer is determined according to heat
transfer unit (NTU) method:
where is heat transfer area.

Pumping gas loss is:
LOSS ANALYSIS---COMPARISON WITH THEORETICAL REFRIGERATION

Based on thermodynamic analysis of the hot end flow of pulse tube and considering
sinusoidal pressure wave, following formula is reduced to calculate theoretical refrigeration:9
where is pressure wave amplitude and a special parameter determined by:
Combining refrigeration capacity with upper two losses, one typical calculated result
according to a typical group of parameters is given, as shown in Figure 4, where relative
proportion of shuttle loss, pumping loss and sum of two to refrigeration are given. For different
average boundary layer thickness, there will be different results of shuttle loss and pumping loss.
With the increasing of shuttle loss decreases evidently while pumping loss increases quickly.
There is a smallest value of the sum of two, which shows the best boundary layer thickness for
pulse tube. Such a thickness is easy to execute for actual piston and not easy for gas piston.
In Figure 4, the predicted point of actual boundary layer thickness is 0.57mm, which
corresponds to the smallest shuttle loss and largest pumping loss. This is not strange. Shuttle loss
becomes small with thick film preventing heat transfer, but pumping loss becomes large. In
Stirling or G-M refrigerator, pumping gas loss is much small comparing to shuttle loss, because
BLT or gas gap could be as thin as 0.1mm.
Figure 4. A group of loss evaluation result. Condition: pressure 2.5MPa, pressure ratio 1.1, frequency
50Hz, pulse tube diameter 9mm, length 70mm.
CONCLUSION
In this paper, losses due to oscillating flow in a pulse tube have been analyzed in detail. Through
analysis, some unique features have been found in the pulse tube flow. One is that surface heat
pumping has a limit. The other is that boundary layer thickness in the pulse tube may be very thick
compared to the actual piston. The above features will result in a small shuttle loss and a large
pumping loss. It is important that, independent of a thick or thin boundary layer thickness, the sum
of the shuttle loss and pumping loss will be an important loss in a pulse tube refrigerator. This may
be helpful for future pulse tube refrigerator design.
ACKNOWLEDGEMENTS
The author thanks Dr. G. Thummes (Giessen) for his helpful discussion and support.
REFERENCE
1. Bian, S.X., Small-type Cryocooler, China Mechanical Industry Press, Beijing (1985).
2. Mikulin, E.I., Tarasov, A.A., Shkrebyonock, M.P., “Low-temperature expansion pulse tubes,” Adv.
in Cryo. Eng., vol.29, Plenum Press, New York (1984), p. 629.
3. Shiraishi, M., Nakamura, N., Seo, K., Murakami, M., “Visualization study of velocity profiles and
displacements of working gas inside a pulse tube refrigerator,” Cryocoolers 9, Plenum Press, New
York (1997), pp. 355-364.
4. Lee, J.M., Kittel, P., Timmerhause, K.D., Radebaugh, R., “Higher Order Pulse Tube Modeling,”
5. Yang, L.W., Zhou Y., Liang, J.T., “Analysis of theory refrigeration and losses in pulse tube,” Adv.
in Cryo. Eng., vol. 45, Plenum Press, New York (2000).
6. Hiratsuka, Y., Kang, Y.M., Matsubara, Y., “Development of a 1 to 5W at 80K Stirling pulse tube
cryocooler”, Cryocoolers 10, Kluwer Academic/Plenum Publishers (1999), pp.149-155.
7. Gifford, W. E., Longsworth, R.C., “Pulse tube refrigeration”, Trans. ASME, J. Eng. Ind., vol. 63
(1964), p.264.
8. Zhu, S.W., Wu, P.Y., Chen, Z.Q., “Double-inlet pulse tube refrigerators: an important improve-
ment,” Cryogenics, vol. 30 (1990), p.514.
9. Yang, L.W., Research of pulse tube refrigeration mechanism and practical development, Post-doctor
Report, Institute of Mechanics, Beijing (1998), China.
Numerical Study of Gas Dynamics Inside of a
Pulse Tube Refrigerator
Yoshikazu Hozumi, Masahide Murakami1, Masao Shiraishi2

Chiyoda Corporation, 2-12-1, Tsurumichuo, Tsurumi-ku
Yokohama, 230-8601, Japan
1
Institute of Engineering Mechanics,
University of Tsukuba, Tennodai 1-1-1, Tsukuba
305-8573, Japan
2
Mechanical Engineering Laboratory, MITI
Namiki 1-2, Tsukuba, Ibaraki 305-8564, Japan
ABSTRACT
A simulation program for viscous compressible flow has been developed to study the details of
fluid motion and gas dynamics inside a pulse tube refrigerator. Axisymmetric two-dimensional
Navier-Stokes equations are solved numerically. Simulation results inside of a basic-type pulse
tube refrigerator have been reported in past research, and showed that secondary mass flow and
enthalpy flux, going from the cold end to the hot end along the tube wall, contribute to the heat
transfer. They also suggested that the boundary layer on the tube wall might play an important role
in transferring enthalpy from the cold end to the hot end of the pulse tube.
In order to investigate the heat transfer mechanisms within the entire pulse tube refrigerator,
the present research also includes the simulation of the flow and gas dynamics inside of the regen-
erator and after-cooler. The simulation results are compared with experimental data. The simula-
tion results of the pulse tube temperature profile, when compared to the experiment, are seen to be
in good agreement. The simulation results appear to well describe the gas dynamics and refrigera-
tion mechanisms of the pulse tube refrigerator.
INTRODUCTION
Pulse tube refrigerators are categorized as thermoacoustic refrigerators. One approach to ex-
plaining the refrigeration phenomenon in a pulse tube refrigerator is to note that a pulse tube refrig-
erator works the same as a Stirling cycle refrigerator; the Stirling displacer is just replaced by the
pulse tube in the pulse tube refrigerator. Thus, the pulse tube works as a gas piston similar to the
Stirling displacer.
A key feature of thermoacoustic refrigerators is their regenerator, which strongly determines
the refrigerator's performance. Heat conduction in the regenerator and phase shift between the
pressure change and the volumetric change are important phenomena in all Stirling-type refrigera-
tors. Phase shift devices are installed at the hot end of the pulse tube to improve refrigeration
performance. Examples include an orifice, an inertance tube, and a double inlet bypass connection.

A relevant question is: what happens in a heat-transfer-only pulse tube without a regenerator.
Inside of the tube, when driven with an adequate pressure pulse, there is a differential temperature
established between the closed end and the open end; the closed end gets hotter and the open end
gets colder. In order to explain this heat transfer, a surface heat pumping mechanism was suggested
by early investigators Gifford and Longsworth.1 Although the gas elements do not move directly
from the cold end to the hot end in the pulse tube, the gas transfers heat from the open end to the
closed end by successive heat-pumping cycles. Although this concept of heat pumping was sug-
gested in the 1960s, it has not been verified by experimental research. This is because the mecha-
nism is very local with a short cycle time, and experimental probes that measure temperature,
velocity, and pressure can not respond to the working gas status. However, the phenomenon had
been investigated by previous analytical researchers.2,3 The heat transfer mechanism has been
described by the Lagrange frame approach (surface heat pumping mechanism), and also by using
the pressure versus volume diagram.
In the pulse tube, a convective driving mechanism occurs in the oscillatory boundary layer at
the solid wall of the pulse tube. During the cyclic expansion and compression process, the gas
elements close to the wall experience different viscous drag associated with the alternating gas
temperature, and experience a net drift from the cold end to the hot end. This means that gas
elements around the boundary layer move from the cold end to the hot end; this is referred to as the
secondary mass flux. Since the net mass flux along the tube must be zero, so-called streaming is
introduced in the center part of the pulse tube as the secondary-flux gas returns to the cold end.
In order to investigate the heat transfer, one must evaluate the particle path of the working gas
element and the temperature difference between the gas elements and the tube side-wall. The
secondary mass flux and the streaming are observed by tracing the particle path of gas elements as
shown in Figure 1. The phenomenon can also be explained by the approach using the pressure
versus volume diagram.
Figure 2 shows pressure versus specific volume (1/ Density) diagrams for three small gas
parcels placed at different radial positions in the pulse tube. The PV diagram for the parcel closest
to the tube side wall shows large work. The diagram does not show only the refrigeration work, but
also the viscous dissipation of the gas parcel. The swept-out area of the diagram suggests the
magnitude of the conversion of pressure to work. In contrast, the gas at the center of the pulse tube
does not contribute refrigeration work. Thus, the gas on the boundary layer is what contributes the
heat transfer from the cold end to the hot end of the pulse tube.
The purpose of the present research is to evaluate the whole pulse tube refrigerator perfor-
mance. The gas dynamics in the regenerator must be investigated because the heat conduction in
the regenerator is very important to the refrigeration performance. A simulation program incorpo-
rating a regenerator has been developed, and simulation results are presented in this paper.
Figure 1. Particle path showing secondary mass flux and streaming.

NUMERICAL STUDY OF GAS DYNAMICS INSIDE PULSE TUBE 365
Figure 2. Pressure versus volume diagram on the moving parcel inside pulse tube.
SIMULATION MODEL
The state of compressible viscous flow is expressed by the Navier-Stokes equations, including
the equations of mass conservation, momentum conservation, and energy conservation. Since the
solution of discretized full three-dimensional Navier-Stokes equations places a heavy computa-
tional burden on computers, several simplified models of the governing equations have been inves-
tigated.4 Fortunately, the phenomena inside of a pulse tube can be treated as an axisymmetric flow.
This flow feature is used and axisymmetric two-dimensional Navier-Stokes equations are directly
solved for the pulse tube flow. The simulated refrigerator model matches the experimental appara-
tus at MITI's Mechanical Engineering Laboratory.5,6,7 The pulse tube dimensions are 250 mm x
15.6 mm (length x radius), and the regenerator length is 250 mm. Figure 3 shows the simulation
model of the pulse tube. Although metal screens are installed in the actual regenerator, a pipe-type
regenerator is used for the simulation modeling. The pipe-type regenerator has the advantage that
Figure 3. Schematic diagram of the pulse tube refrigerator used for simulating the axisynmetric two-
dimensional flow region: (1) Pulse tube, (2) Regenerator.
the flow inside of the pipes can be treated using the same simulation model that is applied for the
pulse tube simulation. The modeling concept is also shown in Figure 3. In reality, parallel plates
and pipes are not well suited for the real regenerator because of their longitudinal configuration. In
order to obtain the equivalent surface for heat conduction, the regenerator length in the simulation
had to be made approximately three times the length of the actual experimental apparatus. Practical
evaluation of the refrigeration performance of the entire pulse tube is available with the incorpo-
rated regenerator.
SIMULATION METHOD
The phenomena inside of the pulse tube can be treated as axisymmetric flow. Thus,
axisymmetric two dimensional viscous compressible Navier-Stokes equations were used as the
governing equations. The finite volume method was applied in the simulation. An implicit flux
split difference approximation was applied to discretize the Navier-Stokes equations. The flux split
difference approximation has an advantage in describing the compressible flow. Gauss-Seidel line
relaxation was used to solve the discretized Navier-Stokes equations. This numerical approach is a
popular method in the field of treating highly compressible flow.8 In order to close the equations,
the assumption of a perfect gas was made; helium is used as the working gas in this study.
BOUNDARY CONDITIONS
A numerical simulation requires boundary conditions for solving the governing equations.
Because of its large influence on the simulation results, a practical velocity profile must be speci-
fied for the tube inlet boundary condition. The analytical solution of the Navier-Stokes equations
for pulsating flow in the inlet of the regenerator is obtained in terms of Bessel functions.9
At the pulse tube wall, a heat conduction wall model is applied, and the wall temperature is
determined by the heat conduction from the gas temperature. The gas status inside of the boundary
layer near the tube wall is important for heat transfer, and heat conduction is also important for the
heat pumping mechanism. Mesh points are concentrated toward the wall surface and approxi-
mately eight points are imposed within the viscous boundary layer for better resolution of the
viscous shear stress on the wall.
Boundary conditions at the pulse tube hot end assume a heat exchanger is installed here, and
the amount of heat transferred is computed based on the temperature difference between the gas
and the wall. Orifice type pulse tube refrigerators can also be analyzed by the program by changing
the pulse tube hot-end boundary condition to allow flow in and out of the pulse tube though the
orifice hole, driven by the pressure difference between the tube and the reservoir.
SIMULATION RESULTS AND DISCUSSION

The first step in the numerical simulation was verification that the simulation method and
model appropriately describes the gas dynamics phenomena inside of the pulse tube.
Figure 4 shows the time variation of velocity profiles throughout the entire pulse tube simula-
tion in three locations: the cold end, the middle of tube, and the hot end. The phase difference in
oscillating flow between the tube center and near the tube wall is clearly visible. The time variation
of the temperature profile is also shown in Figure 4.
To verify the simulation method and model, the simulation results were compared with experi-
mental data obtained on the pulse tube at the Mechanical Engineering Laboratory MITI.5,6,7 The
operational frequency was 2 Hz, which is the operational frequency for best performance of this
pulse tube, and the pressure ratio was set to 1.8 to match the experiment. Figures 5 to 7 present a
comparison of the simulated and experimental gas temperature profiles versus time.
Note that the simulated gas temperature oscillation has a much larger amplitude than the ex-
perimental data; this is because the experimental temperature probes have significant heat capacity
and cannot respond to the rapidly alternating working gas temperature. The experiments tend to
show the average value. A similar level of gas temperature profile is observed.
Figure 4. The time variation of velocity

profiles and temperature profiles (velocity profiles
are observed at the cold end, middle, and hot end
of the pulse tube).
A comparison of the pulse tube wall temperature obtained from the experiment and simulation
is shown in Figure 7. The temperature profile from the simulation agrees relatively well with the
experiment and validates the simulation. Thus, the simulation appears to appropriately describe the
gas dynamics inside the pulse tube.
The refrigerator performance dependence on operational frequency was also investigated. The
optimum operational frequency for the actual pulse tube refrigerator to achieve its minimum cold-
end temperature is approximately 2 Hz.
The simulation results, presented in Figure 8, agree in that they show the tube temperature
profile decreases (improves) with lower operational frequency.
Figure 5. Pulse Tube Pressure (Basic Type, Frequency : 10Hz).

Figure 6 Pulse tube gas temperature versus time.
Figure 7. Comparison of experimental data and simulation results; pulse tube gas temperature
(basic type pulse tube, frequency = 10 Hz, pressure ratio = 1.8).
Figure 8. Pulse tube gas temperature profiles at various operational frequencies (basic type pulse
tube, pressure ratio = 1.8)
CONCLUDING REMARKS
A simulation program for viscous compressible flow has been developed to study the details of
fluid motion and gas dynamics in a pulse tube refrigerator. The gas dynamics and the heat conduc-
tion of the working gas in the regenerator are appropriately simulated by the pipe type regenerator
modeling. The simulated temperature profile of the pulse tube agrees with the experimental data.
The program has the capability to study the entire pulse tube refrigerator including the detailed gas
dynamics inside the boundary layer near the tube wall. Future work will address the optimization
of orifice and double-inlet type pulse tube refrigerators.
REFERENCES
1. W.E. Gifford and R.C. Longsworth, “Surface Heat Pumping,” Advances in Cryogenic Engineering,
Vol.11, Plenum Press, NY (1965), p. 171.
2. Y. Hozumi and M. Murakami et al., “Numerical Study of Pulse Tube Flow,” Cryocoolers 9, Plenum
Press, NY (1998), p. 321.
3. Y. Hozumi and M. Murakami, “Numerical Study of Gas dynamics inside of a Pulse Tube Refrigera-
tor,” Advances in Cryogenic Engineering, Vol.45, Plenum Press, NY (1999), p. 167.
4. J.M. Lee and P. Kittel “Higher Order Pulse Tube Modeling,” Cryocoolers 9, Plenum Press, NY
(1997), p. 345.
5. M. Shiraishi and N. Nakamura and M. Murakami, “Visualization Study of Velocity Profiles and
Displacements of Working Gas Inside a Pulse Tube Refrigerator,” Cryocoolers 9, Plenum Press, NY
(1997), p. 355.
6. M. Shiraishi et al., “Visualization Study of Flow Fields in a Pulse Tube Refrigerator,” Advances in
Cryogenic Engineering, Vol.43, Plenum Press, NY (1998).
7. M. Shiraishi et al., “Start-up Behavior of Pulse Tube Refrigerator,” Advances in Cryogenic Engineer-
ing, Vol.41, Plenum Press, NY (1996), p. 1455.
8. R.W. MacCormack and G.V.Candler, “The Solution of The Navier-Stokes Equations using Gauss-
Seidel Line Relaxation,” Computer & Fluid, Vol. 17, No.1 (1989), p. 135.
9. A.F. D’Souza and Oldenburger, Tran. ASME, D, 86-4 (1964-9), p. 589.
Visualization of DC Gas Flows in a Double-Inlet
Pulse Tube Refrigerator with a Second
Orifice Valve
M.Shiraishi, K.Takamatsu1, M.Murakami1, A.Nakano,

T.Iida2, and Y.Hozumi3
Mechanical Engineering Laboratory, MITI
Tsukuba, Ibaraki 305-8564 Japan
1
University of Tsukuba
2
National Space Development Agency of Japan
3
Chiyoda Corporation
Yokohama 230-9601 Japan
ABSTRACT
We have observed the secondary flow induced in a double-inlet pulse tube refrigerator by
using a smoke-wire flow visualization method and investigated the effects of the opening of the
bypass valve on the flow behavior of the secondary flow, especially of the DC flow. Also, the effect
of a second orifice valve on the secondary flow has been visually investigated. Based on the
observations, the relationship between the cooling performance and the dynamic behavior of the
secondary flow has been determined. It has been found that for the double-inlet pulse tube refrig-
erator, a DC flow is induced by opening the bypass valve, and that the DC flow is strengthened with
additional opening of the valve. Further, the behavior of secondary flow in the pulse tube is well
modeled as a superposition of the DC flow and the convection of acoustic streaming driven by the
oscillating main flow. The valve opening for optimum cooling performance is found to be that that
balances the DC flow and the acoustic streaming to reduce the net velocity of the secondary flow to
zero in the core region of the pulse tube. Similarly, it is found for a double-inlet pulse tube refrig-
erator with a second orifice valve, that the optimum cooling performance again corresponds to near
zero secondary flow in the core region.
INTRODUCTION
In a pulse tube refrigerator, several kinds of secondary flow are induced; these include acoustic
streaming associated with the oscillating main flow, natural convection caused by the difference in
gas temperature between the hot and cold ends, and DC flows resulting from closed-loop flow
paths as in a double-inlet configuration.1,2,3,4 Here, we define a secondary flow as any kind of
induced flow except the main oscillating flow in the pulse tube. Acoustic streaming and natural

convection lead to a deterioration in cooling performance, whereas DC flow does not always de-
grade the cooling performance. Recently, it has been empirically determined that DC flows can
have a serious effect on the cooling performance of multistage pulse tube refrigerators with several
closed-loop flow paths, especially in the case of refrigerators reaching temperatures below 4 K.5,6
How to adjust the opening of valves, i.e. tune the DC flow, is an important aspect of optimizing the
performance of such refrigerators.7,8 There is evidence that a suitable DC flow improves the cool-
ing performance, but a DC flow can also lead to unstable oscillation of the cold end temperature of
a pulse tube refrigerator.8,9
Recently, it was reported that a second (or double) orifice valve that is installed between the
reservoir and a supply or return line of the compressor can significantly improve the cooling perfor-
mance when optimally installed and adjusted.10,11 However, some valve installations bring no
advantage to the cooling performance, and the reason is not well understood. The second orifice
valve is thought to work as a control means for the secondary flow. It is of importance to under-
stand the role of the second orifice valve and its effect on the secondary flow.
The behavior of DC flows has been investigated only by measuring the change of the wall
temperature distribution. The relation between the flow behavior and cooling performance, the
temperature distribution, and the optimum condition for bypass and second orifice valves has not
yet been clarified. Visual observation is surely one of the best ways to understand the flow behav-
ior, but there are very few reports concerning visualization experiments. The objective of this study
was to observe the secondary flow in a double-inlet pulse tube refrigerator and to clarify the rela-
tionship between the flow behavior and the cooling performance. We also investigated the effects
of DC flow and second orifice valve opening on the secondary flow.
EXPERIMENT
The experimental apparatus for the visualization study is schematically shown in Fig. 1. The
pulse tube, which is 16 mm in diameter and 320 mm long, is made of a transparent plastic tube. The
regenerator is 18 mm in diameter, 170 mm long, and composed of a pile of #100 stainless-steel
screens with a wire diameter of 0.1 mm. The bypass line connects the inlet of the regenerator and
the hot end of pulse tube through the bypass valve installed near the hot end. The reservoir, which
is composed of a plastic vessel with a volume of about 10 times as large as the pulse tube, is
connected to the high or low pressure lines through the second orifice valve. Needle valves are used
as the bypass valve (Nupro, type BM), the orifice valve (type BM), and the second orifice valve
(type M). Generally, needle valves have an asymmetrical flow resistance with respect to the flow
direction; that is to say, gas flow in the direction of the arrow indication on the valve is not the same
as that in the opposite direction. As a preliminary experiment, the flow resistance was investigated
by introducing a steady flow of air. In this experiment, air was used as the working gas for the pulse
Figure 1. A schematic drawing of experimental apparatus.

VISUALIZATION OF DC GAS FLOWS IN A DOUBLE -INLET PT 373
tube refrigerator. The relationship between pressure drop and mass flow rate was obtained, and the
data were used to estimate the mass flow rate through the bypass valve from the measured pressure
difference. The resistance to air flow in the direction of the arrow is larger than that for the opposite
flow direction, but the difference is less than 12% under experimental conditions. The bypass valve
was set so that the arrow direction is toward the hot end as shown in the figure.
The pressure oscillation was generated via a rotary valve that periodically introduced pressur-
ized air of about 0.2 MPa, and then released it to the atmosphere. The smoke-wire is a thin tungsten
wire with 0.1 mm diameter. Both ends of the wire were soldered to copper supports that act as
electrodes and structural supports to keep the wire taut. The wire was stretched across a pulse tube
cross section through a pair of tube fittings installed at the middle point of the pulse tube. Pressure
transducers of the strain gauge type were installed near the inlet of the regenerator, at the hot end of
the pulse tube, and in the reservoir as shown in the figure. Mineral insulated fine thermocouples
(CA) of 0.15 mm in diameter were inserted into the gas space at the cold and hot end to evaluate the
cooling performance.
The experimental conditions of frequency and amplitude of pressure oscillation were deter-
mined by balancing competing needs. Higher frequency is desirable because it yields a larger
number of cycles in a limited time of smoke-line life. On the other hand, at higher frequency the gas
moves rapidly, and the smoke-line fades out more quickly. In the end, 6 Hz was selected as the best
frequency. As the gas velocity increases with an increase of the amplitude of the pressure oscilla-
tion, the observations also become more difficult. We found an amplitude of about 1.2, which is
defined as the ratio between the high and low pressures, as a good compromise. The opening of the
orifice valve, not the second orifice valve, was adjusted so as to maximize the difference in the gas
temperatures between the cold and hot ends under the condition of a closed bypass valve. After
that, the opening of the orifice valve was fixed throughout the experiment.
During the conduct of the experiment, the smoke-wire surface is coated with paraffin, it is
stretched in the tube, and the lead wires are connected to the high voltage pulse generator. The pulse
tube is then operated under the prescribed condition. After confirming a steady state is reached by
monitoring the gas temperature, a smoke-line is emitted. The movement of the smoke-line is viewed
and recorded using a high speed video camera with a frame rate of 400 frames/sec over 10 cycles.
The pressure and the gas temperatures are also measured.

Observation of Oscillating Main Flow and Secondary Flow
Figure 2 shows a typical oscillatory movement of the smoke-line. The smoke-line is emitted at
the moment the gas starts to move toward the hot end, so in the figure the initial motion is upward.
The shape of the smoke-line in the core region is gradually deformed from its original shape of a
Figure 2. A typical oscillatory movement of a smoke-line in the pulse tube. Vb = 0. Each picture is
selected at intervals of 20 msec from the original sequential pictures. A reversal black-and-white gradient is
used to emphasize the smoke-line.
Figure 3. A comparison of the smoke-line deformations for three different valve openings; is
near the optimum opening for maximum performance from the double-inlet configuration.
straight line into a concave parabolic line over time as the result of the secondary flow. With
increased time its dent becomes steeper, and the length of the smoke-line between the tips near both
walls and the vertex in the core region increases.
To distinguish the secondary flow component from the oscillating main flow, we examine
smoke-lines at the moment of flow reversal in each cycle. Typical results are shown in Fig. 3 for
every other smoke-line in the first 5 cycles, for three different openings of the bypass valve
Here, (and which appears in a later section) is an arbitrary unit of opening measured by a
scale on the valve. The case corresponds to a closed bypass valve, and thus to an orifice
pulse tube configuration. For this case, the length between the tips near the walls and the vertex in
the core region is elongated toward the cold (left in the figure) and the hot end (right) with the
passage of cycles. This result indicates the existence of an axisymmetric convection where the fluid
goes toward the cold end in the core region, and the associated return flow goes toward the hot end
along the peripheral wall. This convection is one of the typical secondary flows regarded as acous-
tic streaming driven by the oscillating main flow.1
In the case of which is in a region of the optimum opening, only the smoke-line near
the wall is elongated toward the hot end. The smoke-line in the core region remains around the
initial position of the first cycle. This indicates that the secondary flow is diminished to almost zero
in the core region and it only exists near the wall. For further increases in the valve opening,
50, the smoke-line drifts toward the hot end as a whole. The smoke-line deforms into an entirely
different shape from those in the former cases. The tips near the wall disappear and the concave
parabolic line in the core region changes to a convex one. This result indicates that the secondary
flow goes toward the hot end at every radial position and the velocity is much faster in the core
region than near the wall.
Velocity Estimation of Secondary Flow

The observed smoke-lines were converted into the computer-identified smoke-lines by using
an image-processing system. The results are shown in Fig. 4 as the smoke-lines at every other cycle
in the first 5 cycles, that is, 1st, 3rd and 5th cycles. This figure shows that the opening of the bypass
valve has serious effects on the flow behavior of the secondary flow in the double-inlet pulse
tube configuration. In the case of smaller than 20, the smoke-line in the core region shifts toward
the cold end with an increase of oscillation cycles, but the overall displacement across the cross
section decreases with increasing valve opening. In the case of larger than 20, the direction of
drift of the smoke-line in the core region changes toward the hot end, but the smoke-line near the
wall still keeps shifting toward the hot end. This result is well explained by considering the genera-
tion of DC flow in the direction from the cold to the hot end, which is superposed on the convection
of acoustic streaming driven by the oscillating main flow. The magnitude of the DC flow depends
on the valve opening and becomes larger with an increase in the valve opening; in contrast, the
valve opening has little affect on the convection, as long as the pressure wave attributes such as the
frequency and compression ratio are kept the same. The net secondary flow is determined as the
superposition of both flows. Consequently the direction of drift in the core region changes. It is
VISUALIZATION OF DC GAS FLOWS IN A DOUBLE-INLET PT 375
Figure 4. A change of the displacement of smoke-lines after 1, 3 and 5 cycles in the double-inlet pulse
tube configuration with the increase of bypass valve opening corresponds to the orifice pulse
tube configuration and to the optimum valve opening for the cooling performance.
worth while noting that the acoustic streaming component is not affected by the valve opening and
always exists irrespective of the valve opening.
The radial velocity profile of the secondary flow is calculated based on the drift distance of the
smoke-line during two cycles, from the 2nd to the 4th cycles. Here, a one-dimensional flow along
the axis of the pulse tube is assumed. Figure 5 shows the measured variation of the profiles with
valve opening. Negative velocity corresponds to flow toward the cold end, and positive toward the
hot end. The solid curves are the fittings of fourth-order polynomials through each profile. In the
case of the net mass flow rate passing across a cross section was calculated from the velocity
Figure 5. Change of the radial velocity profiles of secondary flow in the double-inlet pulse tube
configuration with an increase of the bypass valve opening. Solid curves are the fitting of 4th-order
polynomials through each profile. Negative velocity corresponds to the flow toward the cold end.
Figure 6. A change of the radial velocity profiles of DC flow in the double-inlet pulse tube configura-
tion with increasing the bypass valve. Solid curves are the fitting curves of quadratic polynomial. A positive
velocity corresponds to the flow toward the hot end.
profile in the figure to confirm that the net mass flow rate was balanced as it should be for acoustic
streaming convection in a closed tube of the orifice pulse tube configuration. It was found that the
mass flow rate toward the cold end in the core region agrees with that toward the hot end near the
wall within 10%. By considering the accuracy of the experiment, this result leads to the conclusion
that the net mass flow rate is well balanced and that the secondary flow is dominated by the convec-
tion of acoustic streaming.3 With increasing the core region of negative velocity disappears,
and the velocity becomes positive over the cross section, as explained in Fig. 4.
By using the data presented in Fig. 5, we extracted the component of the DC flow by subtract-
ing the component of the convection of acoustic streaming for from the measured convec-
tion with other openings. Here, we assumed that the secondary flow for the double-inlet configura-
tion is composed of a superposition of the DC flow on the convection for
The velocity profiles of the DC flow are shown in Fig. 6. Solid curves are again the fitting
curves of quadratic polynomials. The data are well correlated with the quadratic polynomial ap-
proximations, considering the accuracy of the visualization, except for small openings of 10 and
20. These parabolic velocity profiles are similar to those of the Poiseuille flow in a tube. The
velocity increases with an increase in valve opening as expected from the results in Fig. 4.
The mass flow rate of the DC flow was calculated using two methods: one based on the veloc-
ity profiles expressed by the solid curves in Fig. 6, and the other based on the measured pressure
difference through the bypass valve combined with the pressure drop characteristics that were ob-
tained from our preliminary experiment that used a steady flow (as discussed in the earlier 'Experi-
ment' section).
In the case of the latter method, the mass flow passing through the bypass valve changes the
direction of flow with respect to the pressure difference in a cycle. However, the calculation shows
that, irrespective of the valve opening, the mass flow rate of the outflow at the hot end through the
bypass valve is larger than that of the inflow. This result indicates that the DC flow in the pulse tube
is induced toward the hot end, and this agrees with the observations of Fig. 6 relative to flow
direction.
The mass flow rates, calculated using both methods, are compared in Fig. 7. The mass flow
rate calculated from the pressure drop is twice as large as that converted from the velocity profile.
This discrepancy is probably due to the measured pressure drop characteristics of the valve, and
most likely because the applied pressure drop characteristics were obtained for steady flow, not for
unsteady flow. The actual pressure difference essentially originates from the unsteady oscillating
flow. Therefore, we consider the mass flow rates converted from the velocity profiles to be closer
to reality than those calculated from the pressure drop.
Figure 7. A comparison of estimated mass flow rates calculated from the velocity profiles of DC
flow and from the measured pressure difference through the bypass valve.
Figure 8. A change of the velocity of secondary flow and the cooling performance for the double-inlet
pulse tube configuration as a function of Core corresponds to the velocity of the secondary flow in the
core region and near wall to near the wall of pulse tube.
Figure 8 shows the relation between the cooling performance and the velocity of the secondary
flow in the core region. The larger temperature difference means better performance. The results in
Fig. 8 indicate that the region of optimum performance corresponds to the region where the abso-
lute value of the velocity decreases to almost zero, i.e., where the convection is cancelled by the DC
flow in the core region. Therefore, it is concluded that the bypass valve can work as the means of
adjusting the DC flow to a suitable level in the double-inlet pulse tube configuration.
Secondary Flow in Double-inlet Configuration with a Second Orifice Valve

In the same manner as in the above experiment, the secondary flow in a double-inlet configu-
ration with a second orifice valve was visually observed, and the role of the second orifice valve
was investigated. Figure 9 shows a comparison of the smoke-line movement for two cases of con-
nections of the second orifice valve. The 'high pressure line' case is for a gas line connection be-
tween the reservoir and the high pressure supply line through a second orifice valve as shown in
Fig. 1; the 'low pressure line' data is for a connection to the low pressure return line. The double-
inlet valve was kept at a constant opening of throughout the experiment; that is the same
condition of in Fig. 4. As shown in Fig. 4, the secondary flow already exists in the pulse
tube even if the opening of the second orifice valve is zero. In the case of the secondary
flow in the core region flows toward the hot end due to a larger DC flow than convection.
For the high pressure line connection, an extra flow is induced in the direction from the hot to
Figure 9. Variation in the displacement of the smoke-lines for the double-inlet pulse tube configura-
tion for different openings of the secondary orifice valve Vso. The 'high pressure line' corresponds to a
connection between the reservoir and the high pressure supply line through the secondary orifice valve,
while the 'low pressure line' is for connection to the low pressure return line. The top figure corresponds to
the double-inlet pulse tube configuration with (see for in Fig. 4).
the cold end with increasing opening of the second orifice valve. This extra flow goes counter to the
existing DC flow and reduces the DC flow. If the opening is adjusted so that the DC flow is reduced
to a suitable level, the DC flow and the convection cancel each other in the core region as in the case
of in Fig 9. Further increase in the valve opening makes the DC flow decrease to a small
level so that the secondary flow in the core region turns toward the cold end.
With a low pressure line connection, the extra flow is induced in the same direction as the
existing DC flow, from the cold to the hot end. Both flow components flow in the same direction so
that the displacement of the smoke-lines only increases with increasing valve opening.
The velocity profiles were determined in the same manner as in the previous discussion, and
the effect of the velocity in the core region on the cooling performance was investigated. Figure 10
shows a comparison of the flow velocity and the cooling performance between both connections.
Figure 10. Measured dependence of the velocity of secondary flow and cooling performance on
secondary orifice value opening for a double-inlet pulse tube configuration. The velocity corresponds
to that in the core.
In the case of the high pressure line connection, the optimum operating region, where the
temperature difference is maximized, coincides with the region where the absolute value of the
velocity is reduced to roughly zero. On the other hand, in the case of the low pressure line connec-
tion, the velocity increases with increasing opening and the improvement in the cooling perfor-
mance is not clearly distinguishable compared to the former case. This fact indicates that the cool-
ing performance should be improved if the secondary flow in the core region is reduced to roughly
zero. This conclusion is found to hold for the double-inlet configuration without the second orifice
valve. We may, therefore, reasonably conclude that the cooling performance of the double-inlet
pulse tube refrigerator should be improved by reducing the secondary flow in the core region.
SUMMARY
We have observed the secondary flow induced in the double-inlet pulse tube refrigerator by
using a smoke-wire flow visualization method. Some results are as follows:
1. The secondary flow in a double-inlet pulse tube refrigerator is well explained by considering
the superposition of the DC flow on the convection of acoustic streaming driven by the oscillat-
ing main flow.
2. The magnitude of the DC flow increases with increasing bypass valve opening. The cooling
performance is improved by adjusting the valve opening to reduce the velocity of the secondary
flow to almost zero in the core region.
3. For a double-inlet pulse tube refrigerator with a second orifice valve, an extra flow is intro-
duced by opening the second orifice valve and the cooling performance can be improved by
adjusting the extra flow to reduce the velocity of the secondary flow in the core region to
roughly zero.
REFERENCES
1. Olson, J.R. and Swift, G.W., “Suppression of Acoustic Streaming in Tapered Pulse Tube,” Cryocool-
ers 10, Kluwer Academic/Plenum Publishers, New York (1999), pp. 307-313.
2. Thummes, G. et al., “Convective Heat Losses in Pulse Tube Coolers:Effect of Pulse Tube Inclination,”
3. Shiraishi, M. et al., “Visualization Study of Secondary Flow in an Inclined Pulse Tube Refrigerator,”
Advances in Cryogenic Engineering, Vol. 45 (2000), pp. 119-126.
4. Gedeon, D., “DC Gas Flows in Stirling and Pulse Tube Cryocoolers,” Cryocoolers 9, Plenum Press,
New York (1997), pp. 385-392.
5. Wang,C., Thummes, G. and Heiden.C., “Control of DC Gas Flow in a Single-Stage Double-inlet Pulse
Tube Cooler,” Cryogenics, Vol. 38 (1998), pp. 843-847.
6. Wang,C., Thummes, G. and Heiden.C., “Effects of DC Gas Flow on Performance of Two-stage 4 K
Pulse Tube Coolers,” Cryogenics, Vol. 38 (1998), pp. 689-695.
7. Kotsubo, V., Huang, P. and Nast, T. C., “Observation of DC Flows in a Double Inlet Pulse Tube,”
8. Charles, I., Duband, L. and Ravex, A., “Permanent Flow in Low and High Frequency Pulse Tube
Coolers-Experimental Results,” Cryogenics, Vol. 39 (1999), pp. 777-782.
9. Duband, L., et al., “Experimental Results on Inertance and Permanent Flow in Pulse Tube Coolers,”
10. Chen, G. et al., “Experimental Study on a Double-orifice Two-stage Pulse Tube Refrigerator,” Cryo-
genics, Vol. 37 (1997), pp. 271-273.
11. Yang, L., Zhou, Y. and Liang, J.,”DC Flow Analysis and Second Orifice Version Pulse Tube Refrig-
erator,” Cryogenics, Vol. 39 (1999), pp. 187-192.
A Gifford-McMahon Cycle Cryocooler
below 2 K
T. Satoh, A. Onishi*, I. Umehara**, Y. Adachi**, K. Sato**

and E.J. Minehara***
R&D Center, Sumitomo Heavy Industries, Ltd.
Hiratsuka, Kanagawa 254-0806, Japan
*Precision Products Division, Sumitomo Heavy Industries, Ltd.
Tanashi, Tokyo 188-0001, Japan
** Yokohama National University
Yokohama, Kanagawa 240-0067, Japan
***FEL Lab., Japan Atomic Energy Research Institute
Naka, Ibaraki 319-1195, Japan
ABSTRACT
According to theory, a Gifford-McMahon (GM) cycle cryocooler with 4He cannot cool
below 2 K because of the 4He superfluid transition near this temperature. However replacing
4
He by 3He removes this temperature limitation. The cooling performance of a GM cryocooler
with a magnetic regenerator material is investigated using 3He. The minimum tempera-
ture of 2.3 K with 4He goes down to 1.65 K when the 4He working fluid is replaced by 3He.
The maximum cooling capacity at 2 K is 53.9 mW with a compressor power of about 2.5 kW,
and the cooling capacity at 4.2 K is enhanced by more than 20%.
The effect of a new regenerator material on the cooling performance was also
investigated. The minimum temperature decreased to 1.64 K and the cooling capacity at 2 K
improved to 57.1 mW with the use of this material in the bottom 40% of the regenerator.
INTRODUCTION
We can now reach the 2 K temperature region very easily using a regenerative cryocooler
such as a GM cryocooler or a pulse tube cryocooler with magnetic regenerator materials. Nagao
et al.1 reached 2.09 K with a three-stage GM cryocooler, while the Giessen group2 obtained
2.07 K with a liquid nitrogen precooled pulse tube cryocooler; this is the lowest temperature
reached to date with a regenerative cryocooler.
These cryocoolers used 4He as the working fluid, and their lowest temperature is limited
by the superfluid transition of 4He near 2 K. The only possibility to reach below 2 K is using
3
He instead of 4He as the working fluid in a regenerative cryocooler. According to the theoreti-
cal analysis of Xu et al.,3 temperature does not change by adiabatic expansion or compression
if the thermal expansion coefficient is zero. Since the thermal expansion coefficient is zero for
4
He at the superfluid transition temperature, and even negative at lower temperatures, we can-
Cryocoolers 11, edited by R.G. ROM, Jr.

382 GM REFRIGERATOR DEVELOPMENTS
not expect to cool below 2 K using 4He. For 3He, on the other hand, the expansion coefficient is
still positive below 2 K, and we can expect to cool below 2 K.
Xu et al.3 investigated the performance of a three-stage pulse tube cryocooler. It was found
that the minimum temperature could be reduced from 2.19 K to 1.87 K by replacing the 4He work-
ing fluid by 3He. This indicates the possibility that a GM cryocooler with 3He could also achieve
temperatures below 2K. Following this lead, we have investigated the cooling performance of a
two-stage GM cryocooler with 3He working fluid and a magnetic regenerator material.
EXPERIMENTAL SETUP
Figure 1 shows a schematic diagram of our two-stage GM cryocooler. The compressor is a
model CKW-21 from Sumitomo Heavy Industries, Ltd. The rated input power of the standard unit
is 2.6 kW at its driving frequency of 60 Hz. However, in this experiment, to save 3He, shorter flex
hoses were used and the adsorber of the compressor was changed to a smaller one. The 3He vol-
ume was limited to about 230 liters at standard conditions and the charge pressure was about
15 bar; the purity of the 3He was 99.95%.
The cylinder of the cold head is made of thin stainless steel tube, the dimensions of which
are as follows: the first-stage cylinder diameter is 52 mm with an inner length of 191.5 mm,
and the second-stage cylinder diameter is 25 mm with an inner length of 165 mm.4 A copper
block is silver brazed to the stainless steel tube at the end of the 2nd-stage cylinder to improve
heat exchange.
Each cylinder contains a displacer, in which regenerator material is stuffed. The dis-
placer stroke is 20 mm and the displacers are driven by an AC synchronous motor that can vary
the cycle speed by changing the supply frequency. The first-stage regenerator is composed of
Figure 1. Schematic of our two-stage GM cooler.

A GIFFORD-McMAHON CYCLE CRYOCOOLER BELOW 2 K 383
Figure 2. Specific heat of magnetic regenerator materials.
#180 copper screens in the higher-temperature region, and lead spheres in the lower-tempera-
ture region. The second-stage regenerator is composed of lead spheres (150g) in the higher-
temperature side, and magnetic regenerator material in the lower-temperature side,
spheres of diameter 0.2 mm to 0.5 mm and a new material crushed powder) of size
0.2 mm to 0.5 mm were used.
Figure 2 shows the temperature dependence of specific heat for and be-
low 20 K. has a specific heat peak due to its magnetic phase transition to its anti-
ferromagnetic state at 7 K and 10 also has a similar type peak of specific heat at
2 K due to its magnetic phase transition to its anti-ferromagnetic state.6 Since it has a larger
heat capacity at 2 K than as shown in Fig. 2, is a good candidate to improve
the cooling performance in the 2K region. As shown in Table 1, four types of second-stage
regenerators were prepared to investigate the effect of the material on the refrigerator's
cooling performance below 4 K. The intake/exhaust valve timing was optimized for each case.

Test data were acquired via a germanium resistance thermometer that was mounted to
measure the second-stage temperature, and a platinum-cobalt alloy resistance thermometer
that was used to measure the first-stage temperature (see Fig. 1). An electric heater of manganin
wire was installed on each stage to allow known heat loads to be applied independently. To
reduce the radiation heat load from room temperature to the second stage, a thermal shield was
attached to the first stage.
Figure 3. Dependence of no-load Figure 4. Relationship between heat

temperature on cycle speed. load and the second-stage temperature.
No-load Temperature
Figure 3 shows the cycle speed dependence of the first and second-stage temperature
with no heat load. The No. 1 regenerator was used for this experiment. The compressor power
was about 2.5 kW for both cases. As shown in this figure, both the first and the second-stage
temperatures are lower with 3He than with 4He for all the investigated cycle speeds. The sec-
ond-stage temperature goes down when the cycle speed decreases. The lowest temperature is
2.27 K at 54 rpm with 4He, and 1.65 K at 54 rpm with 3He. Temperatures below 2 K are ob-
tained at cycle speeds lower than 90 rpm. The first-stage temperature, on the contrary, goes up
when the cycle speed decreases.
The relationship between the second-stage temperature and the heat load is shown in
Fig. 4; the cycle speed is 60 rpm. The cooling capacity of the cooler with 3He is improved
compared to that of the cooler with 4He not only in the 2 K region, but also in the higher
temperature region. The cooling capacity at 4.2 K is 0.63 W with 4He and 0.82 W with 3He; this
is improved more than 20%. The cooling capacity at 2 K with 3He is 53.9 mW.
Effect of on Cooling Performance

The specific heat of the regenerator material is a very important factor effecting the per-
formance of regenerative cryocoolers. has a larger specific heat below 2 K than
as shown in Fig. 2. Thus, is expected to be an effective regenerator material to im-
prove the cooling capacity below 2 K. Four types of the second-stage regenerators were pre-
pared to investigate the effect of on the cooling performance. Fifty percent of the
second-stage regenerator volume, the higher temperature side, was filled with lead spheres and
remained unchanged for all of the investigated regenerator configurations. In the various re-
generators, either or a combination of the two was loaded into the lower
temperature region.
The lowest temperature 1.64 K was obtained at 54 rpm and 48 rpm with the No.3 regen-
erator configuration installed and no heat load on the first stage. This is the lowest temperature
achieved by any regenerative cryocooler. The cooling capacity at 2 K is plotted against regen-
erator makeup in Fig. 5 for cycle speeds of 48 rpm and 60 rpm. The cooling capacity at 60 rpm
was improved 8% when 40% of the was replaced by Figure 6 shows the cycle
speed dependence of the cooling capacity at 2 K. The largest cooling capacity, 57.1 mW, was
obtained at 54 rpm with the No.3 regenerator.
A GIFFORD-McMAHON CYCLE CRYOCOOLER BELOW 2 K 385
Figure 5. Cooling capacity at 2K versus Figure 6. Cycle speed dependence of cool-

regenerator makeup. ing capacity with regenerator No. 3.
Figure 7. The cooling capacity below 4.2 K. The effect of

on the cooling capacity is shown in this figure.
Figure 7 shows the relationship between the heat load and the second-stage temperature
with regenerators No.1 and No.3. The heat load versus the second-stage temperature curves
cross at about 2.3 K. This figure shows that the cooling capacity is enhanced below 2.3 K by
replacing a part of the with However, the performance is poorer above 2.3 K;
this is because the specific heat of is much lower in this higher temperature region above
2.3 K. The cooling capacity at 2 K does not fall at the first stage temperature up to about 50 K.
Effect of Orientation on Cooling Performance
The cooling performance around the 4K region of a GM cooler is much influenced by the
orientation. Figure 8 shows the orientation dependence of the first and the second-stage tempera-
ture with no heat load. An orientation of 0° means that the cold head is set vertical with the room
temperature end up. The cycle speed is 60 rpm and the No.4 regenerator is used in this experiment.
The second-stage no-load temperature is nearly the same at 0° and 180°, and even lower at
180° compared to 0°. The maximum temperature is obtained at 135°, but it is still much lower than
2 K. On the other hand, the first-stage temperature goes up when the orientation angle increases.
Figure 8. Orientation dependence of Figure 9. Orientation dependence of

no-load temperature. cooling capacity at 2K
The cooling capacity at 2 K has the orientation dependence shown in Fig. 9. As shown in
this figure, the cooling capacity at 90° is the lowest, and about 17% lower than that at 0°. The
largest cooling capacity is achieved at 180°. This orientation dependence is similar to that of
the cooling capacity at 4.2 K with 4He working fluid.7
CONCLUSIONS
The cooling performance of a two-stage GM cryocooler with has been investi-
gated using 3He as the working fluid. The lowest temperature goes down from 2.27 K to 1.65 K
when the 4He is replaced by 3He, and the cooling capacity at 2 K is 53.9 mW at 60 rpm. The
cooling capacity at 4.2 K is improved more than 20% with 3He. The effect of the new magnetic
regenerator material on the cooling performance has also been investigated. The low-
est temperature goes down to 1.64 K and the cooling capacity at 2 K is improved to 57.1 mW
when 40% of the in the lower temperature end of the regenerator is replaced by NdInCu2.
A GM cooler with magnetic regenerator can be a reliable and practical 2 K cooler when the
4
He working fluid is replaced by 3He.
REFERENCES
1. M. Nagao, T. Inaguchi, H. Yoshimura, S. Nakamura, T. Yamada and M. Iwamoto, “Generation of Super-
fluid Helium by a Gifford–McMahon Cycle Cryocooler,” Proceedings of the 6th International Cryocooler
Conference, Plymouth, MA, DTRC-91/002, David Taylor Research Center (1991) pp. 37-47.
2. G. Thummes, C. Wang, S. Bender and C. Heiden, “Pulsrohrenkuhler zur erzeugung von temperaturen
im bereich des flussingen heliums”, DKV-Tagungsbericht, 23 (1996) (Jahrgang Band), pp. 147-159.
3. M.Y. Xu, A.T.A.M. de Waele and Y.L. Ju, “A Pulse Tube Refrigerator below 2K,” Cryogenics, vol. 39
(1999), pp. 865-869.
4. T. Satoh, R. Li, H. Asami, Y. Kanazawa and A. Onishi, “Development of High Efficiency 0.5W Class
4K GM Cryocooler,” Cryocoolers 10, Plenum Press, New York (1999), pp. 575-580.
5. J. Bischof, M. Divis, P. Svoboda and Z. Smetana, “Specific Heat of in Magnetic Fields,” Phys.
Stat. Sol. (a) 114 (1989), p. 229.
6. K. Sato, Y. Ishikawa and K. Mori, “Magnetic specific heat of light rare-earth Heusler compounds
( Ce, Pr, Nd and Sm),” J. Magn. Mater., 104-107 (1991), pp. 1435-1436.
7. Sumitomo Heavy Industries, Ltd. SRDK-408D catalog.
High Efficiency, Single-Stage GM Cryorefrigerators
Optimized for 20 to 40 K
C. Wang and P. E. Gifford

Cryomech, Inc.
Syracuse, NY 13211, USA
ABSTRACT
Cryomech, Inc. has developed (2) single-stage GM cryorefrigerators optimized for the 20-
40K temperature range. The Models AL230 and AL330 were designed for high efficiency and
high power cooling of HTc Superconducting Devices. The structural parameters of the AL230
and the AL330 were optimized using a numerical simulation program. The regenerative
materials used are the industry standard lead spheres and phosphor bronze screens. Both the
AL230 and the AL330 obtain the minimum temperature of <11 K. The AL230 provides 30 W at
20 K, or 65 W at 30 K when operating with a CP950 Helium Compressor Package, with an input
power of 5.4 kW (60 Hz). The AL330 provides 45 W at 20 K, or 101 W at 30 K with CP970
Compressor Package, with an input power of 7.0 kW (60 Hz).
INTRODUCTION
For the next generation of HTc Superconduting Devices; such as, Fault Current Limiters,
SMES, SC Motors and Magnets etc., to become viable products, higher capacity and lower cost
cryorefrigerators are necessary. The HTc applications require that the prototype devices increase
in size and current densities and to be cooled to < 30K. The present options of single and 2-stage
GMs, which supply 10-30 watts, are not large enough.
A single stage GM cryorefrigerator has a bottom temperature between 20 and 25 K. Because
they cannot reach temperatures below 20 K, they cannot provide the high cooling capacities at 30
K. The small temperature difference between the bottom temperature of 25 K and 30 K will not
transfer and carry away the required heat. The conventional 2-stage GM has a bottom
temperature around 7 K with the second stage, but is not as efficient in the temperature range of
20 to 40 K, as the single stage GM. Also, the cost of manufacture and maintenance for a 2-stage
GM is higher. Therefore, the conventional single-stage and 2-stage GMs have not been able to
satisfy the requirements of the new HTc superconductor applications.
Fielder et al. presented a low temperature single-stage GM1, Leybold Model RGS120-T,
which has 25 W at 20 K for 6 kW power input. This single-stage GM is simple and efficient at
the temperatures from 20 to 30 K.
Cryomech, Inc. has recently developed lower temperature single-stage GM cryorefrigerators,
the Models AL230 and AL330. A simulation program was used to optimize the structural

Figure 1. Model of single-stage GM cold head for numerical simulation
parameters of AL230 and AL330 Cold Heads. The AL230 and AL330 have bottom
temperatures of less than 11 K and are highly efficient in the temperature range of 20 K to 40 K.
DESIGN OF COLD HEADS

A conventional single-stage GM Cold Head, like Cryomech AL200, is designed for efficient
cooling in the temperature range of 50 to 80 K. The AL200 utilizes phosphor bronze screens as
the regenerative materials. The specific heat of phosphor bronze will be very low when the
temperature in the regenerator is below 50 K. Lead is a good candidate for regenerative material
in the temperature range of 20 to 50 K. For the new low temperature refrigerators, Cryomech
Model AL230 and AL330, utilizes a hybrid regenerator with phosphor bronze screens at warm
end and lead spheres at cold end.
Theoretical Optimization
A numerical simulation program was developed to optimize the structure parameters of the
refrigerator. Figure 1 shows a physical model of a single stage GM for the simulation. The
regenerator inlet is taken as the left boundary. The right boundary is the surface of the displacer.
The basic assumptions in the simulation model are as follows:
1. ideal gas;
2. one-dimensional flow of helium;
3. constant wall temperature of cold heat exchanger;
4. adiabatic expansion process in expansion chamber;
5. axial heat conduction and shuttle loss neglected.
The governing equations and method for carrying out the numerical simulations used in this
paper are the same as that in Reference 2. The left boundary conditions are: pressure wave at the
inlet of regenerator, and constant gas temperature flowing into the regenerator. The right
boundary condition is the motion of the displacer.
The normal operating parameters for the optimization program are: warm end temperature
300K, cold end temperature 30K, and operating frequency 2.4Hz.
When designing the new AL230, we used many of the components from the AL200
changing only the displacer, the displacer tube and 30 K heat exchanger. The same valve motor
assembly, valve plate, rotary valve, and displacer diameters were used. The AL330, and the 77
K AL300 were completely new products. For the AL330, the phasing of rotary valve, size of
regenerator, and heat exchanger were all optimized by the program. The procedures for the
optimization and the design are illustrated in Figure 2.
HIGH EFFICIENCY, SINGLE-STAGE GM FOR 20 TO 40 K 389
Figure 2. Diagram of optimization and design.
Figure 3. Cutaway of the AL230 and AL330.
Physical design
Figure 3 shows the cutaway of the AL230 and the AL330 Cold Heads, as well as their size. They
have rotary valves for gas distribution and the displacer is driven pneumatically. The regenerator consists
of two layers of regenerative materials: phosphor bronze screens in the upper part, lead spheres in the
lower part. A Cryomech patented cold heat exchanger is used in the refrigerators to enhance heat
transfer. Figure 4 is a photo of the AL330, AL230 and AL200 cold heads. The AL230 and the AL330
have the same length and different diameters.
Figure 4. Photo of AL330, AL230 and AL200. (a) AL330, (b) AL230, (c) AL200.
PERFORMANCE OF THE SINGLE-STAGE GM CRYOREFRIGERATORS

Performance of AL230
For the proof of principal our first attempt at a low temperature high capacity single stage
GM was the AL230. The AL230 was first tested with an in-house CP640 Helium Compressor
Package with input power around 5.5kW. Figure 5 shows cooling capacities of the AL230 with
displacer strokes from 15.2mm to 25.4mm. The shorter the displacer stroke, the lower the
bottom temperature of the cold head and the less cooling capacity in higher temperature range.
The lowest temperature, 10.5 K, was obtained with 15.2 mm stroke. Further reducing the stroke
could not lower the temperature of the cold head. The lowest temperature may be limited by the
regenerator efficiency and shuttle loss. We decided on a stroke of 17.8mm for the production of
the AL230, since it maximizes the cooling capacities in the temperature range from 20K to 50K.
Figure 5. AL230 Cooling capacities with different displacer strokes

HIGH EFFICIENCY, SINGLE-STAGE GM FOR 20 TO 40 K 391
Figure 6. Performance graph comparing AL200 and AL230.
Figure 7. Cooling capacities of AL330 and AL300

A comparison of cooling capacities of the AL200 and AL230 are displayed in Figure 6. The
curve for AL230 in Figure 6 is a typical cooling capacity, which was obtained with 17.8mm
displacer stroke and the new CP950 Compressor Package. The CP950 utilizes a helium scroll
compressor and has an efficient absorber with lifetime >20,000 hours2. The AL230 has a bottom
temperature of 11.5 K, 30 W at 20 K and 65 W at 30 K. Compared to the AL200, the AL230 has
a higher cooling power below 60 K. Our prediction for the AL230 performance is also shown in
Figure 6. We believe the discrepancy is most likely due to the fact that there was no factor for
shuttle losses included in the simulation model.
Performance of the AL330
The AL330 and AL300 operate with the CP970 Compressor Package. The CP970 also
utilizes a helium scroll compressor module with input power approximately 7kW. The AL300 is
designed for 80 K operation. Figure 7 shows the cooling capacities of the AL300 and AL330 as
well as the performance predicted by our program for the AL330. The AL300 and AL330 are
both working at the theoretically optimized stroke of 25.4 mm. The AL330 has a bottom
temperature of 10.5 K and 101 W at 30 K. It has better performance than the AL300 when the
cooling temperature is below 35 K.
Table 1 lists the cooling capacity, input power and efficiency of the AL230 and AL330. The
AL230 has 65 W at 30 K for 5.44 kW of input power, for a carnot efficiency of 10.7%. The
AL330 has 101 W at 30 K for 7.04 kW of input power, for a carnot efficiency of 12.9%. The
highest percent of carnot for the AL230 is 11.6% and for the AL330 is 14.0%. Both
cryorefrigerators have their highest efficiency near 40 K which is close to the optimization
temperature of 30 K. The AL300 and AL330 have the highest efficiency and cooling capacity to
date of any available GM cryorefrigerators.
CONCLUSION
Single-stage GM cryorefrigerators can supply the high capacity, high efficiency, 30 K

cooling requirements of many of the new HTc applications. Predictions from our simulation
program can be used to design even larger systems. The Cryomech Models AL230 and AL330
are built with a proven technology. Both cryorefrigerators have the minimum temperatures of
approximately 11 K. The AL230 provides 30 W at 20 K or 65 W at 30 K for an input power of
5.44 kW. The AL330 provides 45 W at 20 K and 101 W at 30 K for an input power of 7.04 kW.
ACKNOWLEDGMENT
We would like to thank R. Dausman and B. Zerkle for useful discussions and J. Cosco for
testing the cryorefrigerators.
REFERENCES
1. Fiedler, A., Gerban, J. and Haefner, H.U., “Efficient Single Stage Gifford-MacMahon Refrigerator
Operating at 20 K”, in: Advances in Cryogenic Engineering, Vol. 43B, Plenum Press, New York
(1997), pp.l823-1830.
2. Wang, C., Wu, P.E. and Chen, Z.Q. “Numerical Modelling of an Orifice Pulse Tube Refrigerator”,
Cryogenics, vol.32, no.9 (1992), pp.785-790.
Remote Cooling with a G-M Cryocooler by
Use of Cold Electromagnetic Valves
Driving an External Flow Loop
K. M. Ceridon and J. L. Smith Jr.

Massachusetts Institute of Technology
Cambridge, Massachusetts, USA 02139
ABSTRACT
A major limitation of Gifford-McMahon, G-M, cryocoolers is the requirement to conduction
cool the refrigeration load. A number of earlier attempts to use check valves to rectify the oscillating
pressure from the G-M expansion in an external flow loop were unsuccessful because of valve
leakage. A G-M cryocooler is being modified with the addition of Boreas style cold electromagnetic
valves to utilize the oscillating pressure to drive the unidirectional flow in an external cooling loop.
The cold valve of the Boreas cooler demonstrated minimum leakage and acceptable electrical
dissipation even at 4.5 K. Preliminary analysis indicated that the G-M blow down process could be
used to power the external flow without seriously degrading the G-M cooler capacity.
INTRODUCTION
Cold electromechanical EM valves offer significant potential for improving cryocoolers. When
fitted with two cold EM valves, a G-M cryocooler can drive an external, unidirectional flow loop
without the need for a high efficiency recuperative heat exchanger and a circulator at room
temperature. As shown in Figure 1, the EM valves allow a blow down into a cold surge volume and
a return flow through a load heat exchanger without any increase in the gas circulation in and out
of the warm end through the charge and exhaust valves. In contrast, an oscillating flow into and out
of a cold surge volume will not cool the surge volume. This is because of the regenerative heat
transfer associated with the oscillating flow which blocks of energy flow from the surge volume to
the G-M displacement volume.
Earlier attempts to rectify the oscillating pressure of a G-M cycle with cold check valves have
generally been unsuccessful. With spring loaded check valves, the gas must enter the surge volume
near the system-high-pressure and exit the surge volume near the system-low-pressure. As will be
shown, this process does not maximize the cooling capacity of the external flow loop. In addition,
the closing force is typically not sufficient enough to avoid valve leakage. Electro-mechanical
valves can overcome these problems. The cold EM valves are spring loaded in the closed position.
Clearly, commercial solenoid valves are not suitable for application as cold EM valves. In a
commercial solenoid valve, the electrical dissipation is high and the mechanical reliability is low.
On the other hand, a cold EM valve can have low dissipation. The use of high purity copper reduces
the coil resistance, which will also reduce dissipation. The high current used to open the valve is

Figure 1. Modified G-M cycle including external helium flow loop.
reduced to a low hold-open value to limit the dissipation and to prevent the impact with the full-open
stop. On closing, the energy stored in the magnetic field and the energy in the valve spring is
significantly dissipated in the control circuit at room temperature. This further reduces the cold
dissipation and reduces the impact with the valve seat. The cold EM valve is guided by elastic
flexures rather than by sliding guides. Designs of the electrical circuits that actuate the valves are
critical to low loss and long life operation.
The work on valves for the G-M cooler draws heavily on the technology developed for the cold
one-watt Boreas cryocooler [1,2]. The Boreas valve operated at 4.5 K had a cold dissipation of only
a few tens of milliwatts and proved to be quite reliable.
Cold EM valves offer significant potential for improving other cryocooler cycles, especially for
operation below 10 K where thermal regenerators have very limited heat capacity. An associated
project in cooperation with Advanced Mechanical Technology, Inc. is applying cold EM valves to a
miniature Collins cycle cryocooler for operation at 10K. The valves are to be applied to the floating
piston expanders by Jones and Smith [3].
SYSTEM DESIGN
First Order Thermodynamic Analysis of System and Processes

In order to examine how the external flow loop effects the performance of the refrigerator, a
simplified first order thermodynamic analysis of a modified G-M cycle was compared to that of an
unmodified G-M cycle. As shown in Figure 1, the principal thermodynamic components of an
unmodified G-M cycle are a regenerator, R, contained within a displacer, D, that moves between a
cold volume, C, and a warm volume, W. The compressor maintains a constant high pressure,
upstream of valve V1 and a constant low pressure, down stream of valve V2. In addition to these
components, the modified G-M cycle has two cold EM valves, V3 and V4, a large surge volume, S,
and a heat exchanger, HE. Valve V3 opens for flow from cold volume C to surge volume S. Valve
V4 opens for flow from surge volume S through heat exchanger HE to cold volume C.
For the purposes of this analysis, the unmodified G-M cycle is taken to consist of the following
four discreet processes and four equilibrium states.
State 1 W, R and C, all at D at top dead center (TDC)
Process 1-2 Close valve V2 and open V1, charge W, R and C to
State 2 W, R and C at D at TDC
Process 2-3 Move displacer from TDC to BDC
State 3 W, R and C at D at BDC
Process 3-4 Close valve V1 and open Valve V2 to blow down to
State 4 W, R and C at D at BDC
REMOTE COOLING WITH G-M COOLER BY USE OF FLOW LOOP 395
Process 4-1 Move displacer from BDC to TDC

In the modified cycle, valves V3 and V4, surge volume S, and heat exchanger HE are added to
the cycle. The mass of gas stored in the heat exchanger HE is assumed to be small. The modified
cycle consists of the following discreet processes and equilibrium states
State 1 W, R, C and S all at D at top dead center (TDC)
Process 1-2 Close valve V2 and open V1, charge W, R and C to
State 2 W, R and Cat S at D at TDC
Process 2-3 Move displacer from TDC to BDC
State 3 W, R and Cat S at D at BDC
Process 3-4 Close valve V1 and open Valve V2 until W,R and C reach
State 4 W, R and C at Sat D at BDC
Process 4-5 With V2 closed, open V3 to equalize W,R and C with S
State 5 W, R, C and S at D at BDC
Process 5-1 Close V3, open V4 and V2 to equalize W,R,C and S to then
move displacer from BDC to TDC
The following analysis presents the method used to perform a first order analysis on these two
cycles. The performance of the two cycles is then compared to determine how the external flow
loop degrades the performance of the G-M cycle.
The simplified analysis of the two cycles is based on a mass and energy balance for each
component and each process. As shown in Figure 1, the interfaces between components are: RWE
between R and W, RCE, between C and R, SI between S and C, HEI between S and HE, and HEO
between HE and C.
The G-M Cryocooler Cycle. The system starts at state 1. Process 1-2 opens valve V1 and
allowing high-pressure helium to charge the regenerator R and the warm volume W to state 2.
Process 2-3 fills the cold volume C with gas that is cooled as it flows through the regenerator to State
3. Work, flows up the displacer because of its motion at Process 3-4 expands the gas in
volume C and the gas in the regenerator to by allowing flow out through valve V2. As the
displacer moves during process 4-1 the cool gas flows through the regenerator and out valve V2.
During processes 3-4 and 4-1, heat flows from the walls of volume C to the cold gas. From an
energy and mass balance for volumes C and R, the cooling is given by
where is the displacement volume for the displacer. The first term is the net work up the
displacer from cold to warm and the second term is the net enthalpy, h, flowing from warm to cold
along the regenerator due to its ineffectiveness for the temperature span of the regenerator. The net
enthalpy flow for the commercial cooler is evaluated from the measured cooling and the displacer
work term.
Modified G-M Cycle with External Flow Loop. States 1, 2 and 3 as well as processes 1-2 and
2-3 are the same for the modified cycle. Thus the mass entering the regenerator R across section
RWE is the same for both cycles. The regenerator is assumed to have the same effectiveness and
temperature span for both cycles. In process 3-4, gas flows out through valve V2 until the pressure
in volume C and regenerator R decrease to The value for is selected as a design parameter
for the cycle. Then valve V2 closes. In process 4-5, gas flows out through valve V3 into volume S
until the pressure in volume C and regenerator R decreases to Then valve V3 closes and
valve V2 opens to start process 5-1. The displacer moves to BDC and gas exits to through valve
V2. The regenerator R, volume C and volume S go to to start the next cycle.
The operating states of the modified cycle are selected so that the regenerator has the same
warm-end net enthalpy flow and the same temperature span as in the simple cycle. From an energy
and mass balance for volumes C, R and S, the cooling QR is again given by Eq. (1) when the
electrical losses in valves V3 and V4 are neglected and there is no additional external heat leak to
volume S. The addition of the external flow loop does not influence the value of the when
Figure 2. Refrigeration temperature as a function of intermediate pressure,
processes 1 -2 and 2-3 match the simple cycle. Also, the value selected or does not influence the
refrigeration
The remaining task is to relate the cooling temperature of the heat exchanger HE to the low
temperature defined by the span of the regenerator. The required calculations are the mass and gas
temperature or energy for R, C, and S at each state 1, 2, 3, 4, 5. The mass and energy balances
together with the boundary fluxes of both energy and mass at sections RWE, RCE, SI, HEI, and HE2
allow the calculation of the energy flux across HEI during process 5-1. This is accomplished by
starting with process 1-2 and following each process in sequence. Except for process 4-5, the end
states of each process is determined by the initial states. Process 4-5 requires numerical integration
from its initial state to find the final state at pressure
By calculating the required state 1 at the end of process 5-1 for the regenerator and volume C,
the energy flux across section HEO from HE to volume C for process 5-1 is determined. If gas
crossing section HEO is assumed to be at the load temperature, then the value of the cooling load
temperature is fixed by the energy fluxes across section HEO and HEI.
Since the cooling temperature for the modified cycle is higher than for the simple cycle, the
analysis shows the penalty associated with the modification as a temperature increase at the cooling
load rather than as a reduction of the cooling The analysis also shows how the selected value for
influences the increase in cooling temperature. Figure 2 shows the cooling temperature as a
function of the intermediate pressure for the regenerator span that gave a cooling temperature of
77 K for the simple cycle of the commercial G-M system.
Figure 2 indicates that the maximum available mass should be circulated into the surge volume
and should equal Thus, process 3-4 is not necessary since it does not improve the refrigerator
performance. The mass circulation in the surge volume is also maximized when the volume S is
large compared with volume C.
This initial analysis, also indicates that, at the same cooling load, the refrigeration temperature
increases with the modifications to the cycle. When is equal to the refrigeration temperature
is 82 K. Although this higher temperature could be a result of approximations in the model, it is
likely that second-order affects cause this increase in temperature. Process 4-5 puts the cooling load,
into the mass circulating in the surge volume. As this mass decreases, a larger temperature rise
is required of the circulating mass to absorb the cooling load. The result is a larger refrigeration
temperature at
Figure 3. Cross section of cold EM valves as placed in the cold cylinder. The valve housing with
the valve assembly appears on the left and the valve housing appears on the right.
Overall, this analysis examines the cycle only on a first-order basis. Second order effects such
as losses at the valves, regenerator temperature distribution and effectiveness and external radiation
heat leak have been neglected. A more thorough analysis and experimental procedure is required to
evaluate the impact of these second order affects on the cycle performance.
Design of Cold EM Valves

The design of the EM valves, Figure 2, was selected to meet the following objectives:
• Implement an external circulation loop with minimal modification of the existing
commercial cryocooler;
• Minimize electrical and mechanical dissipation.
• Maximize operating life;
• Demonstrate an electronic valve controller that provides a programmed valve opening
force;
• Demonstrate a removable valve assembly inserted in a valve housing;
• Demonstrate a soft seat design for effective valve sealing.
The existing cooler was modified by boring two holes in the cold cylinder head for attachment
of the valve enclosures. The gas connection to the valve is through the end flange of the enclosure
which is bolted in place to compress an indium seal. Since the gas flows through the valve
assembly, all valve dissipation is effectively transferred to the helium stream. The valve assembly is
composed of the following parts:
Valve seat and valve head and stem;
A Kel-F insert in the valve head to form a soft sealing surface;
Ferromagnetic core attached to the valve stem;
Solenoidal, high purity, copper-wire winding;
Ferromagnetic back iron to close the magnetic circuit;
Two multi-leaf flexures to guide the valve and core and prevent friction;
Helical valve spring to provide valve-sealing force in addition to the pressure force.
Both the valves open into the cold volume of the cylinder so that the pressure difference across the
valve is in a direction that increases the valve sealing force.
The cold blow down valve exits the helium directly into the cold surge volume as shown in
Figure 1. The surge volume exits the gas into the load heat exchanger. For the experiment, an
electric heater in the exit line from the surge volume supplies the cooling load. The load heat
exchanger exits the helium back into the cold volume of the cylinder via the return valve.
Manufacturing and Geometric Design Considerations. The two cold EM valves V3 and V4
are attached to the cylinder head to control the flow through the external flow loop. The attachment
of the valves to the cold cylinder head is shown in Figure 3 at a 1:2 scale. The cylinder of the
cryocooler has a 3-inch internal diameter and a 0.035-inch wall thickness. The stroke of the
displacer is from 0 to 0.625-inch from the cylinder head. All components of the commercial
cryocooler are to be maintained and modified only for the valve attachment. This geometrically
limits the size and position of the valves relative to the cold cylinder.
The valve assemblies are contained within the vacuum tight housings that are welded to the
ports drilled through the cold cylinder head. The valve assemblies are assembled externally and then
placed into the housing. The valve housing is closed by end flange that is bolted to the housing and
sealed with indium for the helium-vacuum seal. The end flange has the 0.375-inch port for the tube
that connects to the external components of the flow loop.
The valves are at the maximum possible distance apart, consistent with a valve port diameter of
0.375-inch for the gas flow. The valve center lines are 1.875 inches apart and the valve housing has
an internal diameter of 1.430-inches. The valve housing is assembled by first welding an attaching
tube to the cylinder head over the valve port. Then the other parts of the housing are sequentially
welded on. The internal components are then manufactured, welded together and fully assembled as
a separate valve assembly. Finally, the valve assembly is fit into the housing. The valve assembly is
sealed against the lower flange of the housing by the force of cup-washer springs that are
compressed by the housing flange bolts.
Material and Dynamic Considerations. With the size and position of the valves set by the
geometry of the cold cylinder, the valve assembly components are designed as follows. Three
criteria are addressed in this design concept: 1) material limitations, 2) low-temperature valve
leakage and 3) dynamic constraints that arise from the operation of the G-M cycle.
Since these valves operate at low temperatures, material wear is critical. To maximize the life
of the valves, this design virtually eliminates sliding components and stress concentrations. This is
accomplished by using multi-leaf flexures to guide the motion of the single moving core. This single
moving core includes the valve stem, valve head and moving electromagnetic core. Unlike a typical
solenoid valve design, this valve is not allowed to hit the valve stops at full force. This is
accomplished by the dynamic control of the valve motion through prescribed electrical input to the
coil so that impact forces are minimized.
Electromagnetic Design and Analysis. To open the valve, enough magnetic force must be
created to overcome the pressure force on the valve head and the initial force in the valve spring.
However, once the valve begins to move, the required force to continue valve motion is substantially
reduced. Only enough magnetic force to hold the valve in the open position against the valve spring
is required. The magnetic force should then be further reduced to allow the valve spring to gently
reseat the valve in the closed position. Therefore, it is desired to first build up enough
electromagnetic energy in the coils to overcome the opening force. Once motion starts, the magnetic
energy in the coils should be reduced twice to hold the valve open to return the valve to the closed
position. Controlling the amount of electrical input to the coils will accomplish this.
In a typical solenoid valve, enough electromagnetic energy is delivered to coils to open the
valve. However, this energy is not reduced and the valve is allowed to hit the stops with full force at
a high velocity. To return the valve to its closed position, a solenoid design allows the
electromagnetic energy to be dissipated in the coils. This seats the valve at high speed using the full
force of the spring. Conventional solenoid-valve designs result in large impact forces and stresses,
valve bounce and energy dissipation that reduces cooling.
In the cold EM valve design, the electromagnetic energy in the coils is controlled by the
electrical input to the valves. Enough electrical power is supplied to the magnetic coils to open the
valves. The valve will initially open at a high velocity. The energy in the coils is then reduced
through a room temperature resistor to provide enough force to maintain the valves in the open
position. The valve spring now acts to slow down the valve motion and the valve will reach its full
open position at low velocity. Reducing the energy in the coils again to allows the valve spring to
begin closing the valve. This electromagnetic force retards the closing motion of the valve and the
valve reseats with minimum velocity. The velocity profile of the cold EM valve will be similar to
the velocity profiles produced in cam driven designs. However, unlike cam driven designs, the cold
EM valve design has no frictional components.
From initial calculations, the force required to open the valve is 53 lbf and the force required to
hold the valve open is 13 lbf. From this information, a differential analysis of the electromagnetic
design was performed using the Runge-Kutta method. It was found that the magnitude of the power
dissipation in the valves for the opening and closing motions is on the order of 10 millliwatts. These
calculations were performed using tabulated room temperature properties of pure copper. As
temperature decreases, the resistance in the windings also decreases; thus less electrical power is
required to produce a magnetic field in the ferromagnetic core. Copper wire of 99.95% purity will
be used for this design. At very low temperatures, the resistance of this copper wire will decrease to
a low value. This resistance will be experimentally determined at a later date.
Flexure Design. In order to achieve a design with no frictional components, the valve stem
must remain centered in the hole in the back iron throughout the travel of the valve. Additionally,
the valve guides should not resist the axial motion of the valve. Thin multi-leaf flexure stacks are
used to accomplish this. These multi-leaf flexures are designed to have a high radial stiffness with a
low axial stiffness.
Axial stiffness and stress concentrations are minimized by making each flexure 0.010-inch
thick. A single flexure is designed with as a 300° circular beam spanning between the moving valve
stem and the stationary housing. The circular beam is subject to torsional and bending stresses.
Each flexure is manufactured of a single sheet of spring steel and the curved beam is formed by the
milling of the required pattern into the steel sheet. Stress concentrations are minimized by
maintaining a corner radius 3/64-inch. Each three-leaf flexure consists of three single flexures
stacked with a 120° relative rotation and with 0.035-inch spacers between the single flexures.
Three-leaf flexure stacks are located above and below the moving ferromagnetic core.
The required radial stiffness of the flexure must be sufficient to overcome the instability from
the radial magnetic force exerted on the stem at its maximum off-center position. The valve stem
has a standard sliding fit clearance in the hole in the back core of 0.003-in on the radius. An
electromagnetic flux exists in this gap. The flexures stiffness was selected so that at a 0.003-inch
radial deflection, the net force will return the valve to a centered position. At a 0.003-inch radial
deflection the off-center magnetic force is 9.4 lbf. Each multi-leaf flexure stack has a stiffness that
exceeds this value.
Valve Seat Design. The head of the cold EM valve is designed with a Kel-F insert to form the
valve seat. Miller and Brisson investigated the behavior of Kel-F valve seats at room temperature
and at low temperatures. Their data shows that the required seating force necessary so that the creep
of the Kel-F at room temperature shapes the valve to the valve seat and provides a tight low
temperature seal. At room temperature with no electrical input to the electromagnetic coils, the cold
EM valves are closed with sufficient force so that the Kel-F seat deforms to the shape of the seating
surface. At low temperature, this deformation provides a tight seal for the EM valves with leak rates
well below tolerable limits [4].
CONCLUSIONS
From an initial, first-order, thermodynamic analysis, it has been determined that the addition of
an external flow loop to a G-M cycle will not significantly degrade the performance of the
cryocooler. With the cooling load the same as that for an unmodified G-M cycle, the external flow
loop increases the refrigeration temperature by 5 K. This temperature increase may be attributed to
the second-order affects that arise with the introduction of the external flow loop.
The cold EM valves are designed with good sealing properties and low frictional and impact
wear. The valve seat is made of Kel-F to reduce leakage. The valves are guided by multi-leaf
flexures that maintain the valve stem in a centered position. Since the valve is designed to hit the
stops and reseat with a negligible amount of force and momentum, the impact and valve bounce are
reduced. This is accomplished by controlling the velocity profile of the valve through controlled
electrical input to the magnetic coils. Preliminary calculations indicate that the dissipation of these
valves is on the order of a 10 milliwatts.
ACKNOWLEDGMENTS
We gratefully acknowledge the National Science Foundation and Massachusetts Institute of
Technology for the financial support of this work.
We would also like to thank Leybold Cryogenics for the donation of the cryogenic equipment
used for this analysis and experimentation.
REFERENCES
1. J.A. Crunkleton, “A new Configuration for Small-Capacity Liquid-Helium-Temperature
Cryocoolers,” 7th International Cryocooler Conference Proceedings, Vols. 1-4, Santa Fe, NM,
November 17-19, Air Force Phillips Laboratory Report PL-CP-93-1001, Kirtland Air Force
Base, NM, 1993, pp. 187-196.
2. G.R. Gallagher and J.A. Crunkleton, “Thermodynamic Analysis of the Boreas Cryocooler,”
Advances in Cryogenic Engineering, vol. 39, Plenum Press, New York (1994), pp 1543-1550.
3. R.E. Jones and J.L. Smith Jr., “Design and Testing of Experimental Free-Piston Cryogenic
Expander,” presented at the 1999 Cryogenic Engineering Conference in Montreal, Canada
(July 1999), to be published in Advances in Cryogenic Engineering, vol. 45.
4. F.K. Miller and J.G. Brisson, “Development of a low-dissipation valve for use in a cold-cycle
dilution refrigerator.” Cryogenics, vol. 39, Elsevier Science, Oxford, UK (1999), pp. 859-863.
Optimum Intermediate Temperatures of Two-Stage
Gifford-McMahon Type Coolers
T. C. Chuang*, S. Yoshida** and T. H. K. Frederking

Cryogenics Laboratory, Chemical Eng. Dept. SEAS, UCLA
Los Angeles CA 90095
* Raytheon Corp., Philadelphia, PA 19101
** Taiyo Toyo Sanso Co., Ltd., Kawasaki, Kanagawa, 210 Japan
ABSTRACT
Optimum intermediate temperatures are evaluated for two-stage Gifford-McMahon (GM) cool-
ers including shield coolers protecting cryostats with low-boiling-point liquids, e.g. liquid He-
lium-4 (He I) baths near 4 K. Recent cooler progress has raised refrigeration loads to the 1-watt
order of magnitude at 4 K. Thus, present three-stage systems with two-stage GM and single-stage
Joule-Thomson (JT) expanders are potentially replaceable as two-stage GMs reach the refrigera-
tion power of the three-stage GM/JT systems. We have extended earlier studies with the objective
of evaluating realistic optimum intermediate temperatures, e.g. of two-stage GMs, in order to re-
place previous ideal limits. Our present extended model includes previous components such as the
cascade-shunt model. Its optimum temperatures are found to be lower than ideal values, consistent
with cooler data, in the range of parameters covered. This feature relates in part to higher loss
parameters in the low temperature stage exceeding those of the elevated temperature stage.
INTRODUCTION
Three-stage GM/JT systems have been known since the 1960’s. The two-stage GM cooler,
combined with the low-temperature single JT-stage, has reached high reliability, e.g. in NASA’s
deep space network.1,2 In more recent time, modifications of the three-stage systems have become
known for the cooling of SQUIDS.3,4 In the 1990’s cooler progress has been remarkable permit-
ting two-stage GM systems (and similar coolers) for refrigeration load on the order of magnitude of
1 watt at 4 K.
In this context, the objective of the present studies is to determine the optimum intermediate
temperature between two stages of the two-stage cooler. The objective function is minimi-
zation of the total power input to the cryocooler. The calculations aim at a generic approach, to the
extent possible, using non-dimensional variables. We outline simplified early cooler models which
have led to ideal optimum temperatures for restrictive assumptions. We relax these ideal-
ized constraints using an extended model based on our preceding work of Chen et al.3,6 In lieu of
constant properties, realistic parameters are introduced; for instance, different Carnot fractions (CF)
for the two stages. Mostly, the low temperature stage has smaller CF-values than the moderate
temperature stage. Finally, we discuss model results, compare them with cooler data and present
conclusions.

Figure 1. Chen et al. cooler model, shown schematically in temperature-entropy diagram.
SIMPLIFIED SMALL CRYOCOOLER MODELS

Aside from extensive simulation codes, there have been relatively few closed-form models for
small cryocoolers. We outline a few examples of simplified cooler modeling.
Chen et al. model. W.E.W. Chen et al.5,6 have considered a “dynamic insulation” model for
multi-stage coolers. Figure 1 is a schematic diagram of the two-stage case in the temperature (T)-
entropy (S) plane. The main assumptions of the model have been introduced for the following
scenario. Cooler losses are represented by parasitic heat inputs to each stage. For a stage “j”,
the parasitic input is intercepted at the lowest temperature of the stage and “pumped” up by
power input The total power input is minimized varying the intermediate temperature be-
tween stages for a specified set of the high temperature and the low temperature (cold box
temperature For constant parameters, the model for two stages has an optimum ideal tempera-
ture equal to the geometric mean of and : The “high” temperature
usually is the environs temperature i.e. The optimum temperature ratio of a multi-stage
system with “n” stages is given by
In Fig. 1 contact with a working fluid is indicated, e.g. operation along an isobar. This case leads to
simple constraints for constant fluid properties, e.g. constant specific heat. The assumption of Carnot
cooler operation is relaxed readily assuming constant Carnot fractions for each stage. The optimum
result, equation (1) is not altered. For a stage, we have a power input of
is the Carnot power input to stage j.
Jeong-Smith cascade model.7 Figure 2 is a schematic diagram of the model’s basic unit. A
stage in the cascade system7 absorbs its refrigeration load at its low temperature “bus”. Upon power
input into the cooler stage, the Q-rate is “amplified”. The amplified rate is rejected to the next
“high-T” stage. Counting from the bottom up (instead of from the top down), the heat leaving a
stage is For constant parameters and constant Carnot fractions, the total entropy
generation rate of the multi-stage system is minimized readily. This leads to the optimum T-ratio
result, equation (1). A specific example referred to in reference 7 is the thermo-electric cooler. For
the simplified system, e.g. two-stage system, the postulate of minimum entropy generation rate is
equivalent to the postulate of total power input used in the Chen et al. model.5,6
OPTIMUM INTERMEDIATE TEMPERATURES OF 2-STAGE G-M 403
Figure 2. Jeong-Smith Figure 3. Bejan shunt model,

cascade model, schematically. schematically.
Bejan shunt model.8,9 Figure 3 shows the shunt model8,9 schematically. For small single-stage
coolers, the sum of the thermodynamic losses in the cooler is a multiple of the refrigeration load
The Bejan shunt model represents this aspect in a simplified manner. The cooler losses are lumped
together by the “shunt” carrying an entropy-generating heat flow rate. In other words, all losses are
considered to be represented by the parasitic rate of the shunt. The total refrigeration load is the
sum The resulting Carnot fraction is
When is zero, the ideal Carnot process limit is reached. In equation (2) the losses are repre-
sented by the ratio Losses are found to be high for small cryocoolers reaching ratios
of up to six and more. Bejan8,9 has used a convenient non-dimensional form of the shunt’s thermal
conductance. He normalized it using a reference entropy generation rate at the warm end
Small cooler performance results suggest that a more convenient normalization is based on the
entropy generation rate at the cold end.10,11 Using the latter, we describe the shunt-related losses by
one single loss parameter C of the cooler.19-12 We express the power input to the entire cooler in
terms of its Carnot value
is the Carnot coefficient of performance. The Carnot power input is
The “cooler map”12, CF versus provides a first order of magnitude assessment of the regimes
of various cooler types.
For illustration, a recent very small cooler is selected. Kuo et al.13 describe a miniaturized Stirling
cooler. The data set [78 K at 300 K at 0.3/14 watt/watt] leads to i.e. C significantly above
unity. This C-constraint for the mini-cooler supports the Chen et al. model5,6 (omitting ) in so far
as the equivalent parasitic of the shunt is In the present extended model, we leave the
minicooler range, relax previous conditions and incorporate the (modified) Bejan shunt model,
[Eq. (3)], into an extended Chen et al. model.
Figure 4. Extended
model with ingredients of the
Chen et al., shunt-cascade
models, schematically.
EXTENDED MODEL
Extended model characteristics. Figure 4 is a schematic drawing of the present system.

Ingredients of the three preceding models are utilized relaxing the ideal nature of the Chen et al.
model.5,6 Carnot fractions of the stages are not equal, noting that the loss parameters are different
as well. This is expected to shift the optimum temperature at the intermediate “bus” away from
the ideal system temperature
In comparison to the Chen et al. model,5,6 we have more efficient GM stages than in the small
coolers considered initially in the early modeling.5,6 Stage loss parameters are close to the order of
magnitude unity or below. Further, the cold box load is incorporated readily. The scenario
chosen for the calculations is zero externally applied refrigeration load on the intermedi-
ate “bus” at (Fig. 4).
The total power input to the two stages is
The total refrigeration load at the low-T stage (stage 1) is the sum of the cold box refrig-
eration load and the parasitic load or
The related power input into the first stage is or
The refrigeration load imposed on the upper cooler stage 2 is expressed by the cascade condi-
tion. “Amplification” of via -input provides the load imposed on stage 2.
The entire refrigeration load on stage 2 is expressed using the stage loss parameter This ad-hoc
condition is formulated as an analog of equation (6), e.g. In this context we
note that there is a lack of loss parameter information in the literature on GM - stages. Accordingly,
the power input to stage 2 is written as
Because of coupling condition (8) we arrive at two power input contributions to stage 2, and
with and
Orders of magnitude. We assess orders of magnitude for the low-T asymptotic conditions
and A simplified (truncated) set of power input contributions is
The second term on the right hand side is independent of i.e. We vary in
order to obtain a minimum of the power input. This order of magnitude estimate leads to
Thus, our qualitative result implies a lowering of below the ideal value
as the loss parameter becomes larger than
The related total power input turns out to be on the order of magnitude Therefore, in
the subsequent numerical calculations we use a non-dimensional total power input as follows:
The y-values are in the order of magnitude range from 0.1 to unity. In the model
calculations the following set is selected:
In the numerical parameter range selected for the subsequent modeling, we have increasing
roughly proportional to However, the input is a relatively weak function of This con-
dition implies that the power ratio is a distinctly decreasing function of Because of
other two-stage cooler constraints, it is necessary to have a power ratio on the order of magnitude of
ten, e.g. values of 20 or higher.22
Sample calculation. For illustration, the following cooler stage loss parameters are selected:
at a temperature ratio The low-T stage has a (normal-
ized) power input of The
power contributions to the upper stage are , in normalized form:
where The second term is
Thus, the upper stage receives The final non-
dimensional total power input is
The stage power ratio turns out to be for the present numerical set. This is
consistent with the preceding order of magnitude discussion. Values of leading to low power
ratios are physically not meaningful.
Numerical results. Results for the set [4.2 K at 300 K] are presented in Figs. 5 and 6. The non-
dimensional total power (y) is displayed as a function of
Figure 5 is based on the condition of equal loss parameters in both stages, The
curves for constant C-values show a distinct minimum. For the smallest C of 0.45, the location of
the minimum total power input is fairly close to the ideal value Trends shown for
larger C-values beyond 0.45 however are at variance with the asymptotic order of magnitude esti-
mate discussed above. Further, the related stage power ratio tends to drop toward values
below 10. Therefore, Fig. 5 suggests that the assumption is not realistic for high C-values.
Figure 6 presents curves for loss parameter sets The moderate-T stage is characterized
by a constant loss parameter kept at The parameter of the low-T stage is increased. For
a small rise in the minimum intermediate temperature is reduced but little. For a larger
mismatch in C-values however the minimum moves toward low As is raised beyond unity,
e.g. 1.2, the minimum is too low. It cannot be reached with the capability of the upper stage.
For the range of loss parameters of small coolers, there is no chance to reach 4 K with
Figure 5. Total power input Figure 6. Total power input

(normalized) versus for equal loss (normalized) versus for stage
parameters loss parameters
two stages. Instead, multi-stage cooler systems are needed. The latter are contrary to the usual
rationale applied to small cryocoolers.
From Fig. 6 we conclude that there is a relatively narrow range of loss parameter ratios
1, not too far above unity, for favorable two-stage system operation. Figure 6 suggests that running
costs require C1-values not exceeding unity. The related optimum values are lower than the
ideal value All of these constraints pertain to the set of our present assumptions.
COMPARISON WITH COOLER DATA

Table 1 presents GM-cooler parameters which reflect the two-stage system progress of the
1990’s. The data are selected for the limit of zero externally applied refrigeration load at the in-
termediate “bus” (Fig. 4). The cooler system’s low temperatures are chosen toward the low-T
end of the versus plot for various loads. The choice does not necessarily imply that the
-value coincides with the highest CF location. We select these figures in order to see whether our
model is consistent with GM cooler data. Mostly the cooler tests display rather steep characteristics
as a function of (for specific load conditions): is changed but little for the specified inter-
mediate external load.
Because of the steep characteristics the entries in Table I are considered representa-
tive of for the load conditions presumed. The cold box T L-values have decreased with time in
the 1990’s, and accordingly also and the -values.
All the -values, i. e. -values for our load conditions, are below the ideal opti-
mum temperatures of constant property models. This feature agrees with our extended model. Fur-
ther, it supports Carnot fraction behavior: the -values of the low-T stage are smaller than the
of the moderate-T stage. In this context we note further CF-values of reference 12, and recent
single-stage GM data of reference 19. For background information we refer to references 4 and 20.
Additional cooler information may be found in the manufacturer compilation of reference 21.
CONCLUSIONS
For the cold box near 4 K, both the present extended model and the cooler data support stage
loss parameters given by the inequality (for the parameter range considered). The low
temperature stage has a larger relative thermodynamic loss than the moderate temperature stage.
This implies a Carnot fraction of the low-T stage smaller than the of the moderate-T stage.
Consequently, the optimum intermediate temperature is below the ideal value
Less stringent requirements are imposed when the cold box temperature is raised to 15 K or
20 K, e.g. for high-temperature superconductor magnets. The minimum power input function is
lowered, and Carnot fractions of the stages are raised.
ACKNOWLEDGMENT
The senior author (thkf) acknowledges with appreciation the hospitality of Professor K. Andres,
Director, Walther-Meissner-Institut (WMI), Garching FRG during a brief stay (Summer 1998 / Fall
Quarter 1998). Dr. Uhlig and Mr. Hehn of WMI provided useful input in several interesting discus-
sions on coolers. Dr. Sidney Yuan’s help has been essential. It is appreciated very much.
NOMENCLATURE
C loss parameter of cooler, C of cooler stage “j”
CF Carnot fraction = for stage “j”
COP coefficient of performance
Carnot coefficient of performance
refrigeration load, stage “j”
refrigeration load imposed on cold box
parasitic heat flow rate through shunt representing stage losses
s entropy
T temperature; temperature at stage cold box temperature
intermediate temperature of two-stage cooler
W power input; for stage
Carnot value of power input
y non-dimensional power input
temperature ratio
Subscripts
1 stage 1
2 stage 2
c Carnot value
ideal ideal value, e.g.
E environs, e.g.
H high temperature value, mostly
L low temperature value
m intermediate stage
tot total
* optimum value at minimum power input
REFERENCES
1. Higa, W. H. and Wiebe, E., “One Million Hours at 4.5 K,” Proc. 1977 International Cryocooler
Conf., Boulder, CO., NBS Special Publication 508 (1978), pp. 99-107.
2. Brithcliffe, M., “A 2-Watt 5-Kelvin Closed-Cycle Refrigerator System for Micro Wave Low-Noise
Amplifiers,” Proc. 7th Intersoc. Cryog. Sympos., 1989, ASME, New York, pp. 131-134.
3. Buchanan, D. S., Paulson, D. N., Klemic, D. A. and Williamson, S. J., “Development of a Hybrid
Gifford-McMahon Joule-Thomson-Based Neuromagnetometer Cryosquid,” Proc. International Cryo-
cooler Confernece, Monterey CA, 1989, pp. 35-46.
4. Yoshida S., Umeno, T. and Kamioka, Y., “Development of a Flexibly Separated 4 K Head Refrigera-
tor For a SQUID System,” Advances in Cryogenic. Engineering, vol. 39, Plenum Press, New York
(1994), pp. 1263-1270.
5. Chen, W. E. W., Turner, J. R. and Frederking, T. H. K., “Thermodynmamic Performance Measure
for Cryogenic Vessel Insulation,” Proc. 6th Intersoc. Cryog. Sympos., AIChE Symposium Ser.. 251,
vol. 82, New York (1986), pp. 101-103.
6. Chen, W.E.W., M.S. thesis, University of California, Los Angeles, 1987.
7. Jeong S. and Smith J.L., Jr., “Optimum Temperature Staging of Cryogenic Refrigeration System,”
Cryogenics 34 (1994), pp. 929-933.
8. Bejan, A., Advanced Engineering Thermodynamics, Wiley, New York, 1985.
9. Bejan, A., Thermal Science Seminar, UCLA, Academic Year 1989-1990.
10. Ravikumar, K. V., Yoshida S. and Frederking, T.H.K., “Comparison of Performance Measures for
Cryocoolers and Refrigerators,” Intersoc. Cryog. Sympos., Houston, TX, March 1995, paper 33b,
unpublished.
11. Pinsky, C. et al., “Evaluation of the Thermodynamic Performance of Pulse Tubes,” Advances in
Cryogenic Engineering, vol. 41 (1996), pp. 1365-1372.
12. Rohlin, L. et al., “Comparison Studies of Thermodynamic Losses in Pulse Tube Cooler Compo-
nents,” Proc. ICEC 17, 17* Intern. Cryog. Eng. Conf. Bournemouth UK, IoP Institute of Physics
Publishing, Bristol, Philadelphia (1998), pp. 89-92.
13. Kuo, D. T., Loc, A. S. and Yuan, S. W. K., “Qualification of the BEI B512 Cooler, Part 1 -
Environmental Tests,” Cryocoolers 10, ed. R. G. Ross, Jr., Kluwer Academic/Plenum Publishers,
New York (1999), pp. 105-115.
14. Dr. Peter Kerney (1995), private communication on Cryodyne Model 1020.
15. Chafe, J., Green, G. and Riedy, R. C., “Neodymium Regenerator Test Results in a Standard Gifford-
McMahon Refrigerator,” 7th International Cryocooler Conference Proceedings, Air Force Phillips
Laboratory Report PL-CP—93-1001, Kirtland Air Force Base, NM, April 1993, pp. 1157-1164.
16. Uhlig, K. and Hehn, W., “3He/4He Dilution Refrigerator Precooled by Gifford-McMahon Refrigera-
tor,” Cryogenics, 37 (1997), pp. 279-282..
17. Uhlig, K. and Hehn, W., private communication (1998).
18. Yoshida, S., private communication on the Sumitomo GM [1 watt at 4 K] characteristics (2000).
19. Wang, C. and Gifford, P. E., “High Efficient Single-Stage GM Cryorefrigerator for Temperatures
around 20-40 K,” Cryocoolers 11, ed. R. G. Ross, Jr., Kluwer Academic/Plenum Publishers, New
York (2001).
20. Chuang, C., Ph.D. thesis, University of California, Los Angeles, 1981.
21. Walker, G. amd Bingham, E. R., Low-Capacity Cryogenic Refrigeration, Oxford, Clarendon (1994),
pp. 283-287.
22. Dr. T. Kuriyama, private communication at poster session of ICC 11.
Regenerator Behavior with Heat Input or Removal
at Intermediate Temperatures
Ray Radebaugh, E. D. Marquardt, J. Gary,

and A. O’Gallagher
Boulder, CO 80303
ABSTRACT
Regenerators with finite losses are capable of absorbing a limited amount of heat at inter-
mediate temperatures along their length. This paper discusses a simple analytical model and a
rigorous numerical model of regenerator behavior under the influence of heat input or heat
removal at intermediate temperatures as well as the influence of a steady mass flow superim-
posed on the oscillating mass flow within the regenerator. The finite time-averaged enthalpy
transport through the regenerator undergoes a discontinuity at the location of the heat input to
satisfy the First Law of Thermodynamics. The discontinuous enthalpy flow leads to a discon-
tinuous temperature gradient in the axial direction and to an increase in the regenerator loss
that must be absorbed at the cold end. However, the increased loss is less than the heat input
at the intermediate temperature, which allows the regenerator to provide a certain amount of
cooling without the need for a separate expansion stage. This phenomenon is particularly
useful for shield cooling and for precooling a gas continuously or at discrete regenerator loca-
tions prior to liquefaction at the cold end. For continuous precooling the total heat load can be
reduced by as much as 23%.
A comparison is made of the system performance with and without intermediate heat in-
put under various conditions. The paper presents design guidelines to determine the amount of
heat a regenerator is capable of absorbing at various temperatures. Methods for optimizing the
location of discrete heat inputs are presented. The analytical and numerical models are in very
good agreement with each other and are consistent with very limited experimental data.
INTRODUCTION
The function of a regenerator in cryocoolers is to transfer heat from an incoming, high-
pressure stream to an outgoing, low-pressure stream, just as in a recuperative heat exchanger.
The only difference is that in regenerators the heat extracted from the incoming stream is
stored temporarily in the heat capacity of the matrix before transferring it to the outgoing
stream a short time later. Even though there are discontinuities in the flow in regenerators,
there are no fundamental differences between regenerative and recuperative heat exchangers
when only time-averaged behavior is considered. The finite heat capacity of a regenerator
results in a degradation of its performance, but there is no difference between the two heat

410 REGENERATOR ANALYSIS AND MATERIALS DEVELOPMENTS
exchangers in regard to their behavior under the influence of steady external factors. For
example, the effect of a steady heat input along the length of a heat exchanger or a superim-
posed steady flow of fluid in one direction is the same for both recuperative and regenerative
heat exchangers. Therefore, the arguments being presented here will apply to both types of
heat exchangers, but the calculations performed here were carried out only with regenerators.
The second law of thermodynamics shows that it is always desirable to remove heat in a
refrigerator at the highest possible temperature since the increase in entropy flow carried by
the refrigerant to the warm end is given by
where is the change in entropy flow, is the heat input, and T is the temperature from
which the heat is being input. In the case of liquefying a gas it is desirable to have many stages
of cooling to remove the sensible heat from the gas and to cool it to the liquefaction temperature.
The last and coldest stage only removes the heat of vaporization. In practice the additional
stages may lead to a system that is too complex and costly. The simplest case is a one-stage
refrigerator used to liquefy a gas like nitrogen or oxygen. For nitrogen at atmospheric pressure
the specific enthalpy change from 300 K to the saturated vapor phase at 77 K is 234.0 J/g and the
enthalpy change during liquefaction at 77 K is 199.2 J/g. Of the total heat that must be removed
from the nitrogen, 54% should be removed at temperatures between 300 and 77 K. If this
sensible heat is removed only at the single stage operating at 77 K, then system efficiency is
decreased. But, with a single-stage refrigerator other heat sinks to remove some of the sensible
heat at a higher temperature and improve the system efficiency are not normally available. We
propose here that the heat exchanger (either recuperative or regenerative) can be used to remove
a portion of the sensible heat. This concept works only with non-ideal heat exchangers as will be
shown in the next section. A perfect heat exchanger cannot absorb heat along its length.
In the next sections we analyze the behavior of heat exchangers in three different cases that
involve heat input (or removal) to the heat exchanger along its length. In case 1 a fixed amount
of heat is input at a specific location along the length of the heat exchanger. In case 2 the heat
input is proportional to the temperature change, such as with the precooling of a gas. In this case
a portion of the total heat input is at some fixed location along the length of the heat exchanger
and the remainder is at the cold end. Case 3 is like the previous case except that the heat is
continuously removed all along the length of the heat exchanger. Case 3 also applies to the
situation of a superimposed steady flow of refrigerant through the heat exchanger. In the case of
regenerators this applies to the DC flow superimposed on the oscillating flow. The analysis
given here applies to either sign of heat flow or to either direction of steady flow. For instance
the analysis applies to the cold finger heat interceptor discussed by Johnson and Ross1 where
heat was removed at some location along the regenerator to increase the refrigeration power at
the cold end. However, the emphasis in this paper is for using the regenerator or recuperator to
absorb heat along its length in a manner to increase the system efficiency.
CASE 1, FIXED HEAT INPUT AT ONE LOCATION

Simple Analytical Model
For cryocooler operation above about 20 K the temperature profile in the regenerator is very
near linear and the regenerator loss or energy flow (time-averaged enthalpy flow plus
conduction) is approximately proportional to the temperature gradient. The contribution to the
regenerator loss due to the compression and expansion of the gas in the void space is small for
this temperature range. Therefore, for a simple model we assume the regenerator loss or energy
flow is proportional to the temperature gradient, as given by
REGENERATOR BEHAVIOR WITH HEAT INPUT 411
where is the time-averaged enthalpy flow in the regenerator (ignoring real gas effects),
is the conduction in the regenerator, is the regenerator loss or energy flow with no
heat applied to the regenerator, L is the regenerator length, is the temperature of the hot end,
and Tc is the temperature at the cold end. Heat input along the regenerator length increases the
temperature at that location so that the temperature gradient in the regenerator before and after
this location is altered as shown in Fig. 1. The increased gradient at the cold end causes an
increase in the regenerator loss to the cold end. The main question is whether this increased loss
is less than the heat input or whether the heat input simply ends up in the cold end with no
attenuation. Applying the first law of thermodynamics along the length of the regenerator shows
that the new regenerator energy flow to the cold end (region 2 in Fig. 1) is given by
where is the heat input at some intermediate location xi. To solve Eq. (3) for the new
regenerator energy flow to the cold end, we begin by introducing the following dimensionless
variables:
Figure 1 shows the use of dimensionless position and temperature. By using Eq. (2) and
these dimensionless variables, the energy balance equation Eq. (3) can be rewritten as
where the last term represents the dimensionless temperature gradient in region 1. From Fig. 1
we see that the two dimensionless temperature gradients are given by
Figure 1. Diagram showing

location of heat input and its effect
on regenerator temperature and
energy flow.
where is the dimensionless location for the heat input and is the dimensionless
temperature of the regenerator at that location. By substituting Eq. (7) into Eq. (6) and
combining that with Eq. (5) yields
Solving Eq. (8) for qreg gives the result
This equation shows that the heat input at xi is attenuated by and not all of the heat reaches
the cold end unless The ratio of the total heat load on the cold end to the heat load if the
heat were input at the cold end (heat load ratio) is given by
The dimensionless temperature at the location of the heat input qi is found by combining
Eqs. (7) and (9) to give
Figure 2 shows the locus of points for several values of qi and Fig. 3 shows the
variation of qr with for various qi from Eq. (10). We note from Fig. 2 that qi values much
above 1 give rise to significant heating of the regenerator mid section. Thus, as a general
statement we can say that the regenerator can be used to beneficially absorb heat along its length
Figure 2. Dimensionless regenerator Figure 3. Heat load ratio at cold end for
temperature at location of discrete heat input. various heat inputs at discrete locations.
only for heat inputs not much larger than the original regenerator loss. An ideal regenerator with
zero loss cannot absorb any heat along its length and the temperature at would approach
infinity since would be infinity in Eq. (11). In simple terms, the regenerator loss in an ideal
regenerator remains zero even as the temperature gradient approaches infinity. A comparison of
the calculated behavior with experiment requires that be known in order to find from a
known Typically, for 80 K cryocoolers, is comparable to the net refrigeration power,
whereas for a 60 K cryocooler may be 50% larger than the net refrigeration power.
Numerical Model
The numerical model used here for more accurate calculations is known as REGEN3.22. It
is an update of REGEN3.13, 4, which is a finite difference program using the conservation of
energy, mass, and momentum equations to describe the behavior of regenerators. One of the
additions in this new program is the ability to add or subtract heat at any location along the
regenerator and to allow for a DC flow in either direction. The baseline case used for the
5
calculations here was an optimized design similar to that for a pulse tube oxygen liquefier. The
hot and cold temperatures were 300 and 90 K. The length of the regenerator was 40 mm and it
was divided into 40 cells (41 mesh points) for these calculations. The baseline regenerator loss
(RGLOSS + HTFLUX in REGEN3.2) was 8.19 W. The RGLOSS term in REGEN3.2 is the
total enthalpy flux minus the enthalpy flux caused by pressure changes (real gas effects). The
HTFLUX term is the matrix conduction. Figure 4 compares the temperature profiles calculated
from REGEN3.2 with those from the analytical model. For zero heat input there is only a slight
deviation from linearity in the profile calculated by REGEN3.2. The midpoint dimensionless
temperature is 0.533 compared with 0.500 for a linear profile. Heat inputs or removal were at
cell midpoints and occurred at and 0.49. The results in Fig. 4 show that adding has a
greater effect on in the numerical model than it does in the analytical model. For heat removal
the two models agree very well.
Figure 5 compares the dimensionless regenerator loss calculated from the numerical model
with that from the analytical model. This figure shows that the effect of on is nearly the
same from the two models for heat input. The largest difference in the values for is 0.07.
For heat removal the largest difference is 0.27, which occurs with at the regenerator
midpoint. The temperature gradient at the cold end under those conditions is nearly zero, and the
regenerator loss would be that caused by the compression and expansion in the void space. The
temperature profile for heat input calculated by the analytical model would be in better
Figure 4. Dimensionless temperature Figure 5. Dimensionless regenerator loss

profile in regenerator for various heat inputs at as a function of heat input at two different
two different locations. locations.
agreement with the numerical model if the compression and expansion in the void space of the
regenerator were accounted for in the analytical model. That contribution is independent of the
temperature gradient, which would then indicate that a constant should be added to the
expressions for the regenerator loss in Eqs. (2) and (4). The dimensionless regenerator loss at
any location then would be expressed as
where a + b = 1. Equation 5 would then be changed to
By using the modified approximation to the regenerator loss given by Eq. (4*) we find that Eqs.
(9) and (10) do not change, but Eq. (11) becomes
The numerical results at xi = 0.49 show that when qi = 1.50. Thus, a = 0.77 and b =
0.23 give perfect agreement with the numerical model for at that value of qi and deviates only
by 0.03 at qi = 0. At qi = -2.0 for the midpoint we found that qreg = 0.27 when the temperature
gradient at the cold end went to zero. That value agrees well with b = 0.23. However, the results
for from the numerical model for heat removal agree very well with the analytical model
when b = 0. To maintain simplicity in the remaining sections, we will consider only the case of
a = 1.0 and b = 0. Such an approximation is reasonably good for a first stage regenerator, but a
second stage regenerator will have a rather large b and require the use of the modified analytical
model. For temperatures below about 20 K the regenerator temperature profile with zero heat
input to the regenerator begins to deviate considerably from linearity with a dimensionless
temperature at the center point being much less than 0.5. Thus, we do not expect that the
analytical model would be useful for temperature below about 20 K.
Figures 4 and 5 show that when heat is removed at to a heat sink at about 180 K
the regenerator loss can be reduced by 36%. For the case of regenerator loss being
comparable to the net refrigeration power, the heat intercept leads to an increase in net
refrigeration of 36% for the same power input or a reduction in input power by 36% for the same
net refrigeration power. These calculated improvements are consistent with that found
experimentally by Johnson and Ross1 with a Stirling cryocooler when a heat interceptor at 150 K
increased the net cooling power at 60 K by 30%.
CASE 2, TEMPERATURE DEPENDENT HEAT INPUT

We now consider the case where the regenerator is used to precool a gas. The amount of
heat flow at a particular location along the regenerator depends on the temperature difference
between the hot end temperature and the temperature at the specified location on the regenerator.
The gas is cooled the rest of the way to the cold temperature by the refrigeration power available
at the cold end. In this case we still consider only one heat sink located along the regenerator.
The following section deals with a continuous heat transfer along the entire length.
The total heat that must be removed by the regenerator and the cold end for this case is
given by
where for the case of fluid flow at a mass flow rate of with a constant specific heat
of The heat flow into the regenerator at the temperature is
and the heat flow into the cold end becomes

The sum of the heat flows into the cold end when the regenerator loss is considered is
which, when normalized by the baseline regenerator loss becomes
Because Ti is not known at this time, the only heat flow known at this time is from Eq.
(12). When it is normalized by it becomes qt . The other two heat flows are related to qt
in the following manner:
We now solve for in terms of the known quantities and qt . Substituting qi from Eq. (17)
into Eq. (11) allows us to write as
The solution for at any for various qt becomes
The sum of heat flows to the cold end is found by substituting Eqs. (17) and (9) into Eq. (16),
which yields
The ratio of this heat flow to the heat flow at the cold end if all the heat was delivered to the cold
end is
Figures 6 and 7 show how and vary with for different values of qt from the simple
analytical model. For all qt the optimum location for the heat intercept is at the midpoint of the
regenerator. The minimum qr is 0.889, which occurs with qt = 2.0. The value of at the
midpoint with this heat input is 0.667, which means that 1/3 of the heat is transferred to the
regenerator and 2/3 is transferred to the cold end. Thus, the use of the regenerator to precool a
gas can reduce the total heat load on the cold end by up to 11%. It should be pointed out that the
baseline heat load considered here includes the baseline regenerator loss as well as the net heat
load of cooling the gas from the warm temperature to the cold temperature. If the regenerator
loss were equal to the net refrigeration power, then precooling at the midpoint would reduce the
required net refrigeration by up to 22% at the optimum condition.
CASE 3, CONTINUOUS HEAT TRANSFER (GAS PRECOOLING)

Analytical Model
The most efficient use of the regenerator to precool a gas or a conduction member is with
continuous heat transfer all along the length of the regenerator. The first law of thermodynamics
Figure 6. Dimensionless temperature at Figure 7. Heat load ratio at cold end

location of single heat sink for various total versus heat sink location for various total heat
heat inputs. inputs.
applied to an infinitesimally small element shows that the regenerator energy flow at any
location, given by Eq. (2), undergoes a change given by
where is the heat flow caused by changing the temperature of a gas. In accordance with Eq.
(12) this heat flow is represented by By using the linear assumption in Eq. (2) for the
regenerator, Eq. (22) becomes
Equation (23) can be written in dimensionless quantities as
The solution to Eq. (24) gives the dimensionless temperature profile along the regenerator as
Figure 8 shows the dimensionless temperature profile for various qt. The dimensionless
regenerator loss or energy flow at any location with this additional heat input to the regenerator
according to Eq. (4) is given by
Figure 9 shows this regenerator energy flow along the regenerator for variousqt. The ratio of the
total heat flow to the cold end when using the regenerator for precooling to that without the
precooling is given by
Figure 8. Dimensionless temperature Figure 9. Dimensionless regenerator

profile in regenerator for continuous heat input energy flow versus position in regenerator for
or steady mass flow toward cold end. continuous heat input or steady mass flow.
A graph of qr is shown in Fig. 10 and is discussed in the following section where it is compared
with the results from the numerical model. The minimumqr of 0.770 occurs at qt = 1.80. Thus,
the continuous heat transfer can reduce the total heat load (including baseline regenerator loss)
on the cold end by a maximum of 23%.
Numerical Model (with steady mass flow)

The REGEN3.2 numerical model can be used to simulate continuous heat transfer to the
regenerator by superimposing a steady or DC mass flow on the oscillating flow. When
simulating the precooling of any gas other than helium, the steady mass flow rate of helium in
REGEN3.2 needs to be adjusted to give the same enthalpy change between the hot and cold
temperatures as for the gas to be precooled. We are assuming that the specific heats of both
gases are independent of temperature. Starting with the same baseline case discussed for Case 1,
we ran REGEN3.2 for three different steady mass flows which simulated total dimensionless
heat flows from the hot to the cold end of qt = 0.5, 2.0, and 5.0. The calculated temperature
profile for the case of qt = 2.0 is shown in Fig. 8. The numerical model shows slightly higher
temperature sensitivity than does the analytical model, Eq. (25). As discussed previously, the
difference can be explained by the neglect in the analytical model of a contribution due to
Figure 10. Heat flow ratio to cold end

with heat transfer at the optimum discrete
location and with continuous
heat transfer or steady mass flow.
compression and expansion in the void space. The regenerator energy flow calculated with the
numerical model for the case of qt = 2.0 is compared with that from the analytical model in Fig.
9. Results from the two models agree very well, particularly at the two ends of the regenerator.
Values of much greater than about 2 or 3 would tend to reduce the net refrigeration power
to zero for cryocoolers operating around 80 K. Figure 10 compares the heat flow ratio qr of
Eqs. (10) and (27) for the analytical model with that obtained from REGEN3.2. This figure
shows that the continuous heat transfer significantly reduces the cold end heat load compared
with the case of heat transfer only at the midpoint. The numerical and analytical models agree
well for both discrete and continuous heat inputs.
COMPARISON WITH EXPERIMENT

Gilman6 shows that with a heat intercept strap at a location of with
the cooling power increased from 1.65 W to 2.40 W, or a 45% gain at 60 W input power. From
Eq. (11) we find that qt = -1.56 and from Eq. (9) we find that or a 39% reduction in
the regenerator loss. The radiator load from Gilman’s results was 3.1 W, which for a qi = -1.56
gives a baseline regenerator loss of 1.99W. A 39% reduction of this loss would cause a 0.78 W
increase in refrigeration power compared with Gilman’s measured result of 0.75 W. Our nu-
merical model would predict a gain of about 0.62 W and agrees to within about 17% of the
experimental result. Our experimental results with liquefaction of nitrogen7 showed an in-
creased liquefaction rate of 17% when the incoming gas was continuously precooled by the
outer surface of the regenerator. The ratio of sensible heat of nitrogen to the heat of vaporiza-
tion is 1.18. In this example the refrigeration power is comparable to the regenerator loss, and
the calculated results from Fig. 10 show a predicted increased liquefaction rate of about 22%.
CONCLUSIONS
A simple analytical model has been developed that can be used to calculate the effect of
heat input or removal along the length of a regenerator. The model assumes a linear tempera-
ture profile in the regenerator, which is a good approximation for temperatures down to about
20 K. The model is also extended to the case of continuous heat transfer all along the length
of the regenerator and to a steady mass flow superimposed on the oscillating flow. The ana-
lytical model is in good agreement with our most recent numerical model, REGEN3.2. For
continuous precooling in gas liquefaction, the heat load can be reduced by up to 23%. The
calculated results are in reasonable agreement with experiments using heat interceptor straps
and with those obtained in the liquefaction of nitrogen with a pulse tube refrigerator.
REFERENCES
1. Johnson, D. L., and Ross, R. G., Jr., “Cryocooler Coldfinger Heat Interceptor”, Cryocoolers 8,
2. Gary, J., O’Gallagher, A., Radebaugh, R., and Marquardt, E., “REGEN3.2 Regenerator Model:
User Manual”, NIST Technical Note, to be published.
3. Gary, J., Daney, D. E., and Radebaugh, R., “A Computational Model for a Regenerator”, Proc.
Third Cryocooler Conf., NIST Special Publication 698 (1985) pp. 199-211.
4. Gary, J., and Radebaugh, R., “An Improved Model for Calculation of Regenerator Perfor-
mance (REGEN3.1)”, Proc. Fourth Interagency Meeting on Cryocoolers, David Taylor Re-
search Center Technical Report DTRC-91/003 January 1991, pp. 165-176.
5. Marquardt, E. D., and Radebaugh, R., “Pulse Tube Oxygen Liquefier”, Adv. Cryogenic Engi-
neering, vol. 45, Plenum Press, New York (2000), in press.
6. Gilman, D. C., “Cryocooler Heat Interceptor Test for the SMTS Program”, Cryocoolers 9,
Plenum Press, New York (1997), pp. 783 -793.
7. Marquardt, E. D., Radebaugh, R., and Peskin, A. P., Vapor Precooling in a Pulse Tube
Liquefier,” Cryocoolers 11, Plenum Press, New York (2001).
Measurement of Heat Conduction through
Metal Spheres
M. A. Lewis and R. Radebaugh

Boulder, Colorado, USA 80303
ABSTRACT
This paper describes the results of the measurements of heat conduction through a bed of
packed metal spheres. Spheres were packed in a fiberglass-epoxy cylinder, 24.4 mm in diameter
and 55 mm in length. The cold end of the packed bed was cooled by a Gifford-McMahon (GM)
cryocooler to cryogenic temperatures, while the hot end was maintained at room temperature. Heat
conduction through the spheres was determined from the temperature gradient in a calibrated heat
flow sensor mounted between the cold end of the packed bed and the GM cryocooler. The samples
used for these experiments consisted of stainless steel spheres, lead spheres, and copper spheres.
The spheres were screened to obtain a uniform diameter of 80 to 120 µm. Porosities of the packed
beds varied between 0.371 and 0.398. The measurements to determine the thermal conductance
were carried out with various pressures of helium gas in the void space. The results indicated, as
expected, that the helium gas between each sphere enhances the heat conduction across the contacts
between the individual spheres by several orders of magnitude compared with vacuum in the void
space. The conduction degradation factor, defined as the ratio of actual heat conduction to the heat
conduction if the metal were in the form of a solid rod of the same metal cross-sectional area, was
about 0.11 for stainless steel, 0.08 for lead, and 0.02 for copper. The conduction degradation factor
of 0.11 for stainless steel spheres agrees very well with the factor of 0.10 for stainless steel screen
measured previously in the same apparatus.
INTRODUCTION
Beds of packed spheres are commonly used as a regenerator for cryocoolers operating at
temperatures below about 80 K.1 Because of the large temperature gradient in the regenerator, heat
conduction through the packed spheres can be a significant loss. Previous studies of heat flow
through columns of packed spherical material have considered conduction in the fluid due to a
temperature gradient, conduction within solid particles, conduction from one particle to the next
through a separating film, heat transfer between particle and main body of the fluid, enthalpy car-
ried along by the moving fluid, and possible heat generation through a chemical reaction.2 How-
ever, in cryocooler regenerators, the high thermal conductivity of the helium working fluid can
transport a large fraction of the heat between the solid contacts.3 Schumann and Voss investigated
experimentally the thermal conductivity of packed beds in static gas.4 Recent work by Slavin et al.5
considers the thermal conduction in packed beds of alumina spheres in static helium gas, but the
temperature range is 100 to 500 degrees Celsius. Experimental and analytical research data for heat

Figure 1. Experimental test apparatus. Figure 2. Experimental test section.
conduction loss are not available for practical use in the design of cryocooler regenerators. The
purpose of this study is to directly measure the heat conduction loss from room temperature to
cryogenic temperatures through packed spheres in static helium gas.
EXPERIMENTAL APPARATUS AND PROCEDURE

Figure 1 shows the experimental apparatus used for the present study. The apparatus consists
mainly of a test section, a two-stage Gifford-McMahon (GM) cryocooler, a heat flow sensor, and a
vacuum vessel (not shown in Fig. 1).
The details of the test section are shown in Fig. 2 and were described previously.3 Two identi-
cal regenerators are used in this apparatus. Regenerator cylinders are made of fiberglass-epoxy,
with an inner diameter of 24.4 mm and a length of 55 mm. The wall thickness of each cylinder is 1
mm. Heat conduction along the length of the cylinder wall of a single regenerator from room
temperature to 80 K is estimated to be 0.21 W from published thermal conductivity data.6 Stainless
steel, copper, and lead spheres were used in the study. The sphere materials were carefully screened
to obtain diameters between 80 and 110 µm. The spheres were packed in the regenerator cylinder
to a height of approximately 45 mm. Helium gas lines are connected to the regenerator to change
the filling pressure in the regenerators. The pressure can be varied from vacuum to 2.0 MPa.
Multi-layer insulation was wrapped around the test section to reduce radiation heat loss.
The cold ends of both regenerators were connected to a cold plate, which was cooled by the
GM cryocooler with the heat flow sensor between them. The hot ends of both regenerators were
capped by piston-shaped water jackets. Flowing water maintained the hot end temperature at room
temperature. A bellows was attached to the lower water jacket. Changing the filling pressure of the
helium in the bellows moves the lower water-cooled piston to apply any desired load. For these
measurements, an applied load of 5.1 MPa on the packed sphere bed was used. The cold plate and
the two regenerators were free to move with respect to the water jackets, so the force exerted by the
bellows was applied equally to the two regenerator columns of packed spheres.
The heat flow sensor was mounted between the cold plate and the first stage of the GM cryo-
cooler. A flexible thermal link between the cold plate and the heat flow sensor allows for move-
ment of the cold plate when the bellows pressure is changed. The heat flow sensor consists of a
copper bar and two silicon diode thermometers. The copper bar is made of oxygen-free copper
with a cross-sectional area of 72 mm2 and a length of 135 mm. The distance between the two
thermometers is 91.3 mm. The relationship between heat flow through the copper bar and tempera-
ture difference was calibrated before these experiments by using a heater attached to the cold end.
HEAT CONDUCTION THROUGH METAL SPHERES 421
The experimental procedure was as follows. After pumping the vacuum vessel, the two-stage
GM cryocooler was turned on. Both the cold plate and the heat flow sensor were cooled by the first
stage of the GM cryocooler. The cold plate temperature was kept at a constant temperature by the
temperature controller using a silicon diode thermometer at the cold plate and an electric heater
mounted on the GM cryocooler first stage. The cold plate temperature could be varied over a wide
temperature range, but for these tests was maintained at 80 ±0.05 K. Once the cold plate tempera-
ture was set and the temperature difference at the heat flow sensor was measured, an additional heat
load was supplied to a heater mounted on the cold plate, and its effect on the temperature difference
at the copper bar was measured. The heat loads, the temperature differences, and the calibration
curve obtained previously are used to calculate the heat flow through the heat flow sensor. The
calculated heat flow here includes the heat conduction through both sphere beds, the two fiberglass-
epoxy cylinders, and other heat losses, such as radiation loss and heat conduction loss through
instrument wires. In a separate experimental run, the heat flow through the regenerators without
packed spheres was measured to provide information needed to determine the heat flow through the
columns of sphere beds only.
Characteristics of the spheres tested in this study are given in Table 1. The porosity for each
type sphere shown in this table is calculated using the density of the material and the actual mass of
spherical material used to fill each regenerator column volume to capacity. To achieve the desired
results, the regenerators along with material to be tested had to be assembled to minimize the
amount of void space that was inherent to the G-10 cylinders and sphere material. The test sections
needed to be packed efficiently to obtain low porosity values. The spherical material was poured
into the regenerators and placed on a vibration apparatus that allowed the spheres to tightly pack
together in a uniform distribution. As the material settled over the desired vibration time, the
regenerators were filled to maximum capacity. This enabled us to reduce the void volumes in the G-
10 cylinders as much as possible. Once the regenerators were packed to the necessary uniform and
solid packing, the G-10 regenerators were installed into the apparatus. Using this procedure, we
were able to obtain porosity values that were near the theoretical value expected for packed beds of
spheres for a known volume and a specific material density.

The first measurement was performed to determine the system heat leak through the two re-
generators without any spherical material packed inside and with vacuum pressure applied to the
regenerator tubes. The cold plate temperature was 80 K and the hot plate temperature was 285 K.
The total measured heat leak was 1.06 W, with 0.42 W calculated to be the heat conduction through
the cylinder walls of the two regenerators. The 1.06 W of heat flow was then subtracted from
subsequent measurements with spheres in the regenerators and various helium pressures applied to
obtain the heat conduction through the packed sphere beds.
The effect of helium gas pressure inside the regenerator on the total measured heat leak is
shown in Figure 3. These measurements were made with the cold end of the apparatus maintained
at a constant temperature of 80 K. As this figure shows, the heat leak increases rapidly with in-
creasing helium pressure until there is little significant change for pressures above 0.5 MPa. Fig-
ure 4 shows the pressure dependence of the mean free path and thermal conductivity of bulk helium
gas at 80 K, 200 K, and 300 K. Although the thermal conductivity of bulk helium at pressures to
5 MPa is almost independent of pressure, the heat leak varies with pressure conditions at pressures
Figure 3. Heat leak vs. helium gas pressure. Figure 4. Mean free path and thermal
conductivity vs. pressure dependency over
temperature ranges.
below 0.5 MPa. In order to understand the test results, we discuss the heat transfer mechanism in
the regions of molecular flow, viscous flow, and intermediate flow. Generally gas flow can be
treated as molecular flow when the mean free path, is larger than 10 times the distance, d, be-
tween the two solid plates which transfer the heat. On the other hand, when is less than 0.01 d, the
gas flow can be treated as viscous flow. In the viscous flow region, heat transfer between two solid
plates is proportional to the thermal conductivity and is independent of pressure for an ideal gas. In
the molecular flow region, the heat transfer is proportional to the gas pressure. Figure 3 shows that
at approximately 0.5 MPa the thermal conduction through the helium gas has reached the viscous
flow region where the mean free path of the helium atoms has become less than 0.01 d. Because the
temperature varies from 80 to 285 K along the length of the regenerator, the mean free path of
helium at 0.5 MPa according to Figure 4 varies from 0.010 to 0.039 µm. Therefore, the effective
distance between each spherical ball contributing most to the heat flow is about 1 to 4 µm, or 100
times the mean free path. Since the sphere is in the range of 80 to 110 µm, most of the heat is
transported by the helium gas in a region very close to the individual contacts between spheres.
According to Fig. 3, the heat leak at very low helium gas pressure approaches a value very near the
background value. Such behavior indicates that there is very little electronic heat conduction asso-
ciated with direct metallic contact between the spheres, in agreement with electrical resistance
measurements.7 The electrical resistance measurements7 indicate a heat conduction of about 4 mW
for our system.
Because of the many spheres in a packed bed, the thermal conduction through the bed is re-
duced compared to a solid bar of the same material and same cross-sectional area as the metal in the
sphere bed. Therefore, to estimate this heat leak through the packed spherical columns, a conduc-
tion degradation factor is applied. The actual conduction through a packed sphere bed is then
given as a proportional reduction to the bulk material conduction as
where
Heat flow through packed sphere bed
Conductivity degradation factor
Total cross-sectional area of regenerator
Length of packed beds
Porosity of packed beds
Temperature at cold end of regenerator
Temperature at hot end of regenerator
k: Thermal conductivity of regenerator matrix material
Figure 5. Pressure (MPa) vs. CDF for packed bed spheres.
Figure 5 shows how the conductivity degradation factor (CDF) varies with increasing pressure
within the regenerator for the stainless steel, lead, and copper sphere material. This value begins to
level out at around a pressure of 0.5 MPa, which is consistent with the previous explanation of
mean free path and thermal conductivity. The values of 0.11, 0.077, and 0.019 for the degradation
factors of stainless steel, lead, and copper material, respectively, represent our data well over the
pressure range above 0.5 MPa at a temperature of 80 K. Earlier experiments3 were performed in
this test apparatus using various screen materials as the regenerator matrix. Figure 6 shows the
conductivity degradation factor for stainless steel 400-mesh, 25.4 µm wire; 325-mesh 27.9 µm
wire; and 325-mesh, 22.9 µm wire over a wide porosity range. These porosities were obtained
using a static load of 5.1 MPa in addition to no load conditions. These data indicate that a value of
0.11 is a good representation of the conductivity degradation factor for the stainless steel screen
material and agrees with the value of 0.11 in the present investigation for stainless steel spheres.
Our previous measurements3 of phosphor bronze screen gave a degradation factor of about
0.025 compared with the value of 0.02 for the copper spheres. The slightly lower value for copper
is consistent with a lower for higher conductivity material. Table 2 gives the effective thermal
conductivity integration between 80 and 300 K of various sphere and screen materials using Equa-
tion 1. Shown for comparison is the thermal conductivity of bulk helium gas.
Figure 6. Porosity vs. CDF for stainless steel screens.

CONCLUSIONS
The heat conduction through packed spheres from room temperature to cryogenic temperature
was measured experimentally. The experimental apparatus allows for a change in the regenerator
material, the cold end temperature, and the helium gas pressure in the regenerator. The measure-
ments were performed using stainless steel, lead, and copper spheres between 80 and 120 µm in
size with a porosity of about 0.38.
The experimental results showed that the helium gas between each sphere contact point plays
an important role in transporting the heat. The heat conduction through the packed beds was en-
hanced by at least two orders of magnitude using helium gas compared to vacuum conditions inside
the regenerator. The heat conduction reached a constant maximum value for helium pressures
above 0.5 MPa. A short helium mean free path of 0.010 to 0.039 µm, indicates that most of the heat
is transported a distance of the order of 3 µm from one sphere to the next.
The conduction degradation factor, which is the ratio of actual heat conduction to the heat
conduction where the regenerator material is assumed to be bulk, was about 0.11 for the stainless
steel sphere materials, 0.077 for the lead sphere material, and 0.021 for the copper sphere material.
This factor was relatively constant for the 80 K temperature at the cold end and for pressures over
0.5 MPa. For stainless steel and lead spheres the drop off below 0.5 MPa was much more signifi-
cant than the copper spheres. This more constant value for the conductivity degradation factor for
copper spheres could be attributed to possible deformation due to pressure applied during the ex-
periments as well as a better heat transfer at the contacts
This test apparatus provided NIST with valuable information using the packed sphere col-
umns as well as the stacked screens. The conductivity degradation factors that were obtained for
the stainless steel materials of 0.11 were very consistent for both screen and spheres. These new
conductivity degradation factors for calculating the thermal conduction through packed sphere col-
umns gives valuable information for regenerator optimization. With the NIST regenerator optimi-
zation software REGEN3.1,10,11 an improved coefficient of performance for regenerators can be
achieved with proper optimization of regenerator geometry.
REFERENCES
1. Walker, G., Crycoolers, Plenum Press, New York (1983).
2. Singer, E. and Wilhelm, R.H., Chemical Engineering Progress, Vol. 46 (1950), pp. 343-357.
3. Lewis, M.A., Kuriyama, T., Kuriyama, F., Radebaugh, R., “Measurement of Heat Conduction
through Stacked Screens,” Advances in Cryogenic Engineering, Vol.43 (1998), pp. 1611-1618.
4. Schumann, T.E.W., and Voss, V., “Heat Flow Through Granulated Material,” Fuel, Vol. 13 (1934),
pp. 249-256.
5. Slavin, A.J., Londry, F.A., Harrison, J.H., “A new model for the effective thermal conductivity of
packed beds of solid spheroids: alumina in helium between 100 and 500°C,” International Journal of
Heat and Mass Transfer, Vol. 43 (1999), pp. 2059-2073.
6. Takeno, “Thermal and Mechanical Properties of Advanced Cryogenic Materials at Low Tempera-
tures,” Cryogenic Engineering, Journal of the Cryogenic Society of Japan, Vol. 21, No. 3 (1986), pp.
182-187.
7. Lee, A.C., “Contact resistance of 500 mesh regenerator screens,” Cryogenics, Vol. 34, No. 5 (1994),
pp. 451-456.
8. Simon, N.J., Drexler, E.S., Reed, R.P., Properties of Copper and Copper Alloys at Cryogenic
Temperatures, and Below, NIST Monograph 177, Material Reliability Division, Boulder, CO, Feb. 1992.
9. Childs, G.E., Bricks, L.J., Powell, R.I., Thermal Conductivity of Solids At Room Temperature and
Below, NBS Monograph 131, Cryogenics Division, Boulder, CO, Sept. 1973.
10. Gary, J., Daney,D.E., and Radebaugh, R., “A Computational Model for a Regenerator,” Proc. Third
Cryocooler Conference, NBS Special Publication 698, (1985), p. 199.
11. Gary, J. and Radebaugh, R., “An Improved Numerical Model for Calculation of Regenerator Perfor-
mance (REGEN3.1),” Proc. Fourth Interagency Meeting on Cryocoolers, Report DTRC-91/003,
David Taylor Research Center, 1991, p. 165.
Innovative Technology for Low Temperature
Regenerators
L. Tuchinskiy, R. Loutfy and B. J. Tomlinson*

MER Corp., Tucson, AZ, USA
*AFRL, Albuquerque, NM, USA
ABSTRACT
A novel approach for fabrication of cryocooler regenerators is proposed. The regenerators con-
sist of a matrix with a system of through parallel microchannels of pre-assigned shapes and sizes.
Diameters of the channels can be strictly controlled and set at any value from a few microns to a
few millimeters. The volume fraction occupied by the microchannels can be also precisely con-
trolled and set at any value from 1% to 95%. Regenerators of various customized shapes, geometri-
cal and structural characteristics to ensure desirable end properties may be produced. The patented
technique offers a possibility to fabricate cryocooler regenerators with controlled surface area and
low fluid flow resistance from any powder materials including brittle magnetic intermetallics.
Mechanical stability of the regenerators is expected to be much better than that of beds packed with
spheres. It is anticipated that the cost of manufacture using the MCS technology will be signifi-
cantly lower because of much higher yields compared to the traditional sphere production and
spherical regenerator beds packing.
INTRODUCTION
The large demand for compact, energy efficient and powerful cryocoolers is driven by recent
developments in infrared sensor detection technology, conductive cooling of conventional super-
conducting magnets for MRI systems, and wireless communications, where further increases of
tower capacity and transmission quality require superconducting-electronics-based filters and pream-
plifiers. Current applications of cryogenic regenerative materials and regenerative heat exchangers
include small cryogenic refrigerators (cryocoolers), such as Stirling, Gifford-McMahon, and pulse
tubes to reach and maintain temperatures between 4.2 and 100 K at cooling powers ranging from
100 mW to 100 W.
Regenerators are exceptionally suitable for conditions that occur in heat engines and cryogenic
coolers. The three most important requirements to regenerator matrix are: 1) a maximum thermal
storage capacity; 2) a maximum surface heat transfer area; and 3) a minimal hydraulic resistance to
the fluid flow.
Thermal storage capacity of a cryogenic regenerator matrix is proportional to the heat capacity
of a regenerator material and, therefore, successful candidate materials must have large heat capac-
ity. The heat capacities of conventional solid regenerator materials such as stainless steel or bronze
or lead rapidly decrease with decreasing temperature below ~70 K. This limits practical application
of stainless steel and bronze to temperatures above ~50 K and lead to temperatures above ~10 K.

Figure 1. Volumetric specific heat of two rare earth intermetallic compounds and lead:
Magnetic solids have received much attention because of their large heat capacities near the
magnetic ordering temperatures (Curie temperature, or Nèel temperature, The unique fea-
ture of these materials is that they exhibit sharp increase in heat capacity at temperatures near and
below 10 K. This phenomenon is shown in Fig. 1, where the volumetric specific heat of and
are compared with the volumetric specific heat of Pb.1
One of the most successful regenerator materials is that has been used in Gifford-McMahon
cryocoolers to reach 4 K with a cooling power of 1 W. A number of other rare-earth intermetallic
compounds, including ErNi, ErNi2, DyNi2, ErCo2, Er3Co, GdRh, ErRh, Nd3Ni, HoNi2, HoCu2,
have demonstrated
the greatest potential for enhancing the heat capacity at temperatures below 20K. However, effec-
tive use of these compounds in cryocoolers is limited. Because of their brittleness they are used
today only in the form of spherical particles. The brittleness of the best intermetallic regenerator
materials does not allow the use of more effective regenerators.
The critical parameter defining the effectiveness of a regenerator packing is the ratio of the heat
transfer surface area to the fluid pressure drop. In theory, the best design for regenerators is a matrix
with multiple microchannels. It has a highest ratio of heat transfer area to pressure drop. An attempt
to use this approach has been utilized in ribbon regenerators, where the channels are formed by
either dimpling or embossing the ribbon and then winding it on a mandrel. Unfortunately, the
advantages of higher thermal efficiency never have been practically achieved with the ribbon re-
generator because of channel tortuosity and associated maldistribution of the flow through the
channels.1
If the core passages are uniform and the flow is laminar, the regenerator compactness can be
expressed as2
where
p = core porosity (dimensionless)
Nu = h D / k, the appropriate Nusselt number for laminar flow (dimensionless)
the hydraulic diameter of a core passage (m)
free-flow area of the passage (m)
wetted perimeter of the passage (m)
Equation (1) shows an inverse-squared dependence of core compactness on the passage hy-
draulic diameter. This means that if everything else is held constant, the regenerator compactness
will increase dramatically as the diameter of core passages is decreased.
If the regenerator is made as a solid block with straight microchannels or regularly stacked
series of perforated plates, its effectiveness may be significantly increased. The problem is that
there are no techniques for fabricating such kind of structures from brittle magnetic intermetallics.
Should the technique be developed, it may result in the creation of higher efficiency low tempera-
ture (near and below 10 K) regenerators.
INNOVATIVE TECH FOR LOW TEMPERATURE REGENERATORS 429
The objective of this work is to demonstrate a novel approach for fabrication of high-efficiency
microchannel cryocooler regenerators from rare-earth magnetic intermetallics and other materials.
TECHNOLOGY
Our approach is based on a new, proprietary MER Corp. fabrication technique for microchannel
structures (MCS).3 MCS represent a new type of engineered powder materials consisting of a
matrix with a system of through parallel microchannels of pre-assigned shapes and sizes. Diam-
eters of the channels can be strictly controlled and set at any value from a few microns to a few
millimeters. The regenerator porosity (volume fraction occupied by the microchannels) can be also
precisely controlled and set at any value from 1% to 95%.
The MCS technology allows the fabrication of products of various customized shapes, geo-
metrical and structural characteristics to ensure desirable end properties. This technique offers a
possibility to fabricate cryocooler regenerators with controlled surface area and low fluid flow
resistance from any powder materials including brittle magnetic intermetallics. Mechanical stabil-
ity of MCS regenerators is expected to be much better than that of beds packed with spheres. It is
anticipated that the cost of manufacture using the MCS technology will be significantly lower
because of much higher yields compared to the traditional sphere production and spherical regen-
erator beds packing.
The typical technological process for the fabrication of a MCS, as illustrated in Fig. 2, consists
of the following steps. In the step 1 bi-material rods consisting of a shell and a core are produced.
The shell comprises a mixture of a matrix powder (e.g. Er3Ni) and a thermoplastic binder. The core
comprises a mixture of the binder with a channel-forming filler, which can be removed afterwards
by evaporation, melting, dissolution, etc.
The rods 1 produced in the step 1 are cut and assembled into a bundle 2 (step 2). The bundle
is re-extruded through a die of the prescribed diameter (step 3). As a result of the step 3, green
composite rods that comprise the matrix (Er3Ni powder + binder) and fibers (filler + binder) are
obtained.
Plasticity of these green rods can be controlled by temperature and mixture composition; they
can be subjected to any plastic deformation. If the rods produced by the step 3 still have filler fibers
of larger diameter than required, they can be collected into a bundle again and step 3 is repeated for
further reduction in scale. The rods of any cross-section shape can be obtained in this step. The
produced green rods are debound by heating at temperature, which provides evaporation of the
binder and filler, and sintered (step 4). The filler fibers burn out during or after debinding with the
resulting formation of the channels in the matrix. A variety of MCS with a prescribed density and
anisotropy that are tailored to desired end use can be produced by this technique.
The extrusion ratio and number of extrusion steps control the channel diameter. The porosity of the
interchannel walls is controlled by the size of the intermetallic powder and sintering conditions
(temperature and time); thus, these walls may be both solid and porous. The layers produced in the
step 3 can be stacked layer-by-layer and pressed so that the future channels in the adjacent layers
are oriented in parallel or at any pre-assigned angles. The proposed approach offers several impor-
tant advantages, as listed below:
Figure 2. Main fabrication steps for MCS.

• Multichannel regenerators can be made of any materials

• Controlled porosity may range from 1 to 90-95%
• Channel diameter in the range from a few microns to a few millimeters can be precisely
controlled
• Near-net-shape products can be produced.
The technology for fabrication of microchannel Er3Ni structures is described below.
FABRICATION OF MICROCHANNEL Er3Ni

Er3Ni is a very brittle intermetallic compound that can be relatively easily crushed and milled.
Melt Er3Ni ingots produced by Iowa State University were crushed to pieces about 3-5 mm in
average diameter and milled in a steel jar using hard-alloy ball media. After milling, the powder
was sieved through a 325-mesh sieve. The percent output of powder as a function of milling time
is shown in Fig. 3. After 5.5 to 6 hrs of milling, ~ 40 % of the source material is converted into 325-
mesh powder. The mean particle size is d50=17.5 µm, and 10% of the particles have a diameter less
than 2.5 µm, and 90% less than 47µm. Specific surface area is about 0.7 m2/cm3.
Next, a mixture of the Er3Ni powder and a binder, as well as a mixture of a filler powder with
the same binder, was prepared. These mixtures were co-extruded to produce green samples. Extru-
sion included the following steps (Fig. 4).
1. Extrusion of the bimaterial rods (1) with the (filler + binder) core and (Er3Ni + binder) shell.
2. Packing 19 rods (1) into the bundle (2) and re-extrusion of this bundle through the hexagonal
die for fabrication of the hexagonal rod (3). The rod (3) consists of an (Er3Ni + binder) matrix
and 19 (filler + binder) fibers.
3. Packing 19 rods (3) in the bundle (4) and re-extrusion for fabrication of the rod (5), which
comprises the (Er3Ni + binder) matrix and 361 (filler + binder) fibers.
Figure 3. Output of 325-mesh Er3Ni power versus milling time.
Figure 4. Extrusion steps for fabrication of Er3Ni microchannel regenerators.

INNOVATIVE TECH FOR LOW TEMPERATURE REGENERATORS 431
Figure 5. Temperature versus time for the thermal debinding and filler removal.
Figure 6. SEM micrographs of multichannel Er3Ni.
The produced extrudates were subjected to heat treatment to decompose the organic binder and
filler and remove them from the final product. Heating was performed in argon to prevent oxidation
of the Er3Ni; the heating schedule in shown in Fig. 5. According to the phase diagram, pure Er and
Ni melt at 1529°C and 1455°C, respectively. The binary compound Er3Ni, however, melts
peritectically at considerably lower temperature (the peritectic temperature is 845°C, and the liquidus
temperature is ~910°C). That enables sintering Er3Ni powders at temperatures below 845°C. In our
experiments sintering was performed at 820°C during 2 hours. After sintering, the Er3Ni samples
demonstrated uniform shrinkage without flaws and cracks, they had no visible defects, except a
micron-size black film formed on the external surface. The average diameter of the microchannels
was 60-80 µm as shown in Fig. 6.
CONCLUSIONS
The proposed approach enables a significant mass and volume reduction in parallel with an
increase in efficiency of cryocooler regenerators. The microchannel regenerators will be applicable
to pulse tube and Stirling coolers for all temperature ranges. If the regenerators are made out of
stainless steel, then they will improve the performance of coolers above 50 K due to the improved
compactness factor. Multichannel lead regenerators will improve the cooling capacity at 35 K. At
10 K, the regenerators, either of lead or a rare-earth material, will be applicable. The developed
technique may also be used for fabrication of high performance micro-heat exchangers, large and
small prime movers in Stirling engines, catalyst carriers, etc. Due to the much lighter weight, better
vibration damping and sound absorption of the microchannel structures the application possibilities
will extend to general transportation.
ACKNOWLEDGMENT
The authors acknowledge the support of the Department of Air Force on this work.
REFERENCES
1. R. Ackerman, Cryogenic regenerative heat exchangers, Plenum Press, N.Y., 1997.
2. D.S.Beck, D.G.Wilson, Gas-Turbine Regenerators, Chapman & Hall, International Thomson Pub-
lishing, 1996.
3. L. Tuchinskiy, “Multi-Channel Structures and Processes for Making Such Structures,” US Patent
5,774,779 (1998).
Ductile, High Heat Capacity, Magnetic
Regenerator Alloys for the 10 to 80 K
Temperature Range
K. A. Gschneidner, Jr.,1,2 A. O. Pecharsky1 and V. K. Pecharsky1,2

Ames Laboratory1 and Dept. of Materials Science and Engineering2
Iowa State University
Ames, IA 50011-3020, USA
ABSTRACT
New erbium-based regenerator materials have been developed as a replacement for lead in
low temperature cryocoolers. These alloys have volumetric heat capacities which are 20 to
185% larger than that of lead from 10 to 80 K. These magnetic, ductile, oxidation resistant
erbium alloys are more than 10 times stronger than lead and have a thermal conductivity ~10
times lower than lead. The alloys can easily be fabricated into spheres, foils, ribbons, and wires,
and are environmentally friendly, non-toxic materials. A layered regenerator composed of three
different erbium alloy compositions is recommended as the most efficient system to improve the
cryocoolers’ performance to “get the lead out.”
INTRODUCTION
The use of lanthanide intermetallic compounds, which exhibit low magnetic ordering
temperatures (<10 K), as cryogenic magnetic regenerator materials was pointed out by Buschow
et al.1 25 years ago. The practical use of lanthanide regenerators was not realized until 15 years
later, when Sahashi et al.2 and Kuriyama et al.3 used Er3Ni as the low temperature stage
regenerator material in a two-stage Gifford-McMahon (GM) cryocooler. In these articles the
Japanese scientists proposed that low temperature regenerator material in common use, Pb, be
partially replaced by Er3Ni. This replacement allowed one to reduce the low temperature limit of
GM crycoolers from ~10 K to ~4 K. The achieved improvement is due to the higher volumetric
heat capacity of Er3Ni relative to Pb below 25 K.
Above 10-15 K lead is the material of choice because of its high heat capacity, which is
due to its low Debye temperature of 102 K.4 Above 50 K stainless steel or bronze have much
larger heat capacities than Pb and the former are used as the upper stage regenerator materials of
a two-stage GM and other cryocoolers. Since the efficiency of cryocoolers is proportional to the
heat capacity of the regenerator material, the higher the volumetric heat capacity the greater the
amount of heat that can be transferred from the low temperature heat exchanger (or the material
to be cooled) to the hot heat exchanger (exhaust) per cycle of fluid flow through the regenerator
bed.

The utilization of the magnetic contribution to the total heat capacity offers a promising
avenue to follow to improve the efficiency of a cryocooler. The magnetic contribution to the
heat capacity arises from the magnetic ordering process itself and this can give rise to either a
very narrow heat capacity peak for a first order magnetic transformation or a modestly broader
lambda-like peak for a second order magnetic transition. There is also a third type of magnetic
heat capacity peak, which is known as a Schottky anomaly and is quite broad. It is due to the
thermal excitation of electrons from lower energy levels to higher energy levels due to the
crystalline electric field (CEF) splitting of the 4f levels of a magnetic lanthanide metal. In the
case of Er3Ni the 7 K heat capacity peak is due to a second order paramagnetic to
antiferromagnetic (on cooling) transition, but because the CEF levels are probably fairly low,5
the Schottky heat capacity contribution above 7 K prevents the total heat capacity from falling
off as rapidly as is normally observed for a second order magnetic transition.
There are many lanthanide compounds which order magnetically below 20 K, but because
most intermetallic compounds are quite brittle, this can lead to particle fragmentation when they
are used as regenerator materials in GM and other types of cryocoolers. This has been discussed
in some detail by Merida and Barclay.6 Thus we have turned our attention to ductile, magnetic
materials, primarily solid solution alloys based on pure lanthanide metals as a replacement for
lead in cryocooler regenerators. The main results obtained to date are discussed below.
HEAT CAPACITY OF MAGNETIC LANTHANIDE METALS

The magnetic ordering processes in the magnetic lanthanide metals is generally quite
complicated, and for most of them at least two ordering processes take place as the samples are
cooled below room temperature.7,8 The volumetric heat capacities for high purity,
electrotransport purified ( solid state electrolysis) heavy lanthanides (Gd, Tb, Dy, Ho and
Er) are shown in Fig. 1 for the 4 to 100 K temperature interval. Above 100 K the heat capacities
for Gd, Tb, Dy and Ho are comparable to those of bronze and stainless steel at least in the
vicinity of their respective ordering temperatures. Below about 70 K Pb has a much higher heat
capacity than either bronze or stainless steel, see Fig. 1, and is used as the low temperature
regenerator material for cooling down to ~10 K. However, Er metal has a significantly higher
volumetric heat capacity than lead from ~20 K to ~90 K, and is about the same as lead from 90
to 350 K and from 5 to 20 K. Thus Er or Er-based alloys might be an excellent substitute for Pb
as a regenerator material. However, the sharp peak at 19 K is not particularly useful because it
occurs over a narrow temperature range, 2 K. Thus if the entropy associated with this peak were
distributed over a wider temperature range, or better yet, if the peak temperature was lowered it
might have a higher heat capacity than lead below 20 K. This might be accomplished by adding
various alloying agents to Er, and this is discussed in the next section.
ERBIUM-BASED SOLID SOLUTION ALLOYS
Interstitial Alloys
The interstitial Er-based alloys containing O and C can readily be prepared by arc-melting
high purity Er with Er2O3 or graphite, respectively. The alloys containing H or N are prepared
by reacting Er metal with H2 at moderate temperatures (~500°C [~775 K]) or with N2 at elevated
temperatures (>1000°C [>1275 K]). However, most of the commercially available “pure Er”
metal already contain a substantial amount of interstitial impurities, and the interstitial
alloy (noted below) was purchased from a commercial vendor without any
additional alloying or processing.
The presence of 2.7 at.% O, plus smaller amounts of N and C, 0.3 and 0.2 at.%,
respectively, in pure Er, reduces the height of the sharp peak at 19 K and shifts it upward by
about 3 K. Furthermore, it probably combines with the smaller 25 K peak in pure Er to give a
small, but broader, heat capacity peak at 22 K (see Fig. 2). Furthermore, the 52 K peak is nearly
eliminated while the 88 K peak is shifted downward to 84 K. As a whole the volumetric heat
MAGNETIC REGENERATOR ALLOYS FOR 10 TO 80 K RANGE 435
Figure 1. The volumetric heat capacity from 0 to 100 K of high purity Gd, Tb, Dy, Ho and Er metals,
along with those of bronze, stainless steel and lead. The letters SSE stand for solid state electrolysis,
which was used to purify these metals.
capacity of pure Er between the various peaks is not changed significantly by the addition of the
interstitial elements and is still better than lead over 20 to 80 K region. Throughout the rest of
the paper this alloy will be used as the prototype Er alloy against which the heat capacities of the
other Er-based alloys will be compared.
Figure 2. The volumetric heat capacity of an Er-based interstitial alloy from 3.5 to
100 K. Also shown are the heat capacities of high purity Er and lead.
Non-lanthanide Substitutional Alloys

A five atomic percent addition of Sc or Y or Zr or Hf wiped-out the 19, 25 and 52 K peaks
and lowered the 88 K one. Below 18 K the volumetric heat capacity is slightly larger by a few
percent over that of the prototype alloys and Pb. But as a whole these alloys
are not much of an improvement, if at all, over the Er-based interstitial alloy as a regenerator
alloy.
Lanthanide Substitutional Alloys

The heavy lanthanide elements (Gd, Tb, Dy and Ho) tend to raise the magnetic transitions
temperature of Er, and thus these metals are not useful in increasing the heat capacity below
20 K. Lutetium, the last member of the lanthanide elements, has no unpaired 4f electrons, and its
behavior is similar to that observed for the non-lanthanide alloying agents Sc, Y, Zr and Hf, see
above.
The light lanthanides (La, Ce, Pr and Nd), however, have an unusual influence on the
magnetic ordering phenomena in Er. With respect to the upper (or Néel temperature) transition
of Er, all four of these elements lower it, in proportion to their size: the larger La atom lowers
88 K peak the most (to ~70 K for 5 at.%), followed by Ce (about the same as La), which is
followed by Pr (to ~75K for 5 at.%) and Nd (to ~78 K at 5 at.%). For the lower temperature
transformations, La and Ce behave quite similarly, and significantly different from that of Pr and
Nd, which have similar affects.
For La and Ce small additions (5 at.%) wipe-out the 52 K transition while rapidly raising
and merging the 19 and 25 K peaks into one. The 19/25 K peak eventually merges with the
rapidly dropping Néel temperature (88 K) at 20 at.% (La or Ce). This results in large somewhat
broad (11 K) heat capacity peak at ~40 K, but the heat capacity below 34 K and above 45 K lies
below that of the alloy. The maximum heat capacity is 1.8 J/cm3K at 40 K for
the and and the alloys would be useful if one needed a high heat capacity
material for the 35 to 45 K range.
The addition of 5 at.% Pr or Nd destroys the 25 K heat capacity anomaly shifting the
entropy toward the 19 K Curie temperature peak while reusing its temperature to about 23 K.
The 52 K peak is rapidly lowered by the initial additions of Pr reaching a minimum at
~7 at.% Pr before rising slightly and leveling-off when more than 10 at.% Pr is added. As more
Pr (or Nd) are added to the Er (~15 at.%), a double peak structure develops, which eventually
merges into one peak at ~27 at.% Pr, due to rapidly dropping of the 88 K transition temperature.
These behaviors are shown in Fig. 3a which shows the change of the transition temperatures and
Fig. 3b which shows the associated heat capacity peak values, both as a function of Pr content.
The heat capacities of a series of Er-Pr alloys are shown in Fig. 4. It is noted that for alloys
containing more than 30 at.% Pr the heat capacity is larger than that of either or
Pb below 20 K. This is shown in more detail in Fig. 5, where it is seen that has the
highest heat capacity of any of the Er-Pr alloys and the Er interstitial alloy
Pb or Er3Ni between 10 and 20 K. At 10 K the alloy’s heat capacity is 185% larger than
that of Pb and the same as that of Er3Ni.
The Nd additions follow a similar trend as shown by the Pr addition, but since the upper
transition is lowered at a slower rate, more than 30 at.% Nd is required before the two peaks
merge into one.
Based on these results we suggest the following combination of Er-based alloys as a
replacement of lead in the cryocooler regenerator: at the high temperature end the interstitial
alloy; the alloy for the intermediate temperature region; and the
alloy for the low temperature end of the regenerator. The heat capacities of these three
Er-based alloys are shown in Fig. 6 along with that of lead. It is evident from Fig. 6 that the
alloy would be the best regenerator material from 40 to 85 K having a heat
capacity 20 to 40% larger than that of Pb; while which has a heat capacity 20 to 30%
larger than that of Pb, would cover the 24 to 40 K range, and would be the most efficient
below 24 K, since its heat capacity is 20 to 185% larger than lead. Of course if cooling down
Figure 3. The variation of the magnetic ordering temperature (a) and the peak value of the heat capacity
at the respective ordering temperature (b) as a function of the Pr concentration.
much below 10 K (i.e. down to 4 K) is required, a lower temperature stage regenerator composed
of Er3Ni or HoCu2 or Nd would be needed, as is the standard practice today, when Pb is used as
the intermediate temperature stage regenerator material.
Figure 4. The volumetric heat capacity of a series of Er-Pr alloys from 3.5 to 100 K and the prototype
interstitial alloy and Pb.
Figure 5. The volumetric heat capacity of some Er-based alloys from 3 to 22 K. The heat capacities of
Er3Ni and Pb are shown for comparison.
Other Properties
In addition to the significantly higher heat capacity, these three Er alloys have other distinct
advantages over lead with a lower thermal conductivity and a higher strength; and they are not as
toxic as Pb, which is a federally regulated poison. The thermal conductivity which is shown in
Fig. 7, is nearly one order of magnitude lower than that of Pb, comparable to that of stainless
steel. The reduced thermal conductivity would lead to lower longitudinal heat losses in the
Figure 6. The heat capacities of and along with that of Pb.

Figure 7. The thermal conductivities for interstitial Er, Pb and stainless steel.
regenerator. The thermal conductivity of the Er73Pr27 and Er50Pr50 would be expected to be
slightly lower than that of the because of additional phonon scattering from the
Pr atoms in the Er matrix.
The ultimate tensile strength of interstitital Er, Pb, stainless steel and bronze are listed in
Table 1. It is evident that the tensile strength of is an order of magnitude larger
than that of Pb, and about the same as bronze. This is important if the Er-based alloy are used in
the form of spheres, since the materials are strong enough to prevent the loss of sphericity as can
happen with Pb (Sb-hardened) alloys. The strength of the Pr-Er alloys would be expected to be
about the same as that of the interstitial Er alloy. This level of strength is important when these
Er-based alloys are used as wires, screens, flat sheets and jellyrolls in cryocooler regenerators,
which one cannot do with Pb since it is so weak.
There are two other important features which need to be mentioned. One is that the Er
alloys do not oxidize like the light lanthanides, e.g. Nd, which has been used in cryocooler
regenerators. We have held and Er60Pr40 at 396±5 K for over 15
months and there has been no measurable weight gain or loss (within ±0.1 mg) for samples
weighing 3.6, 2.5 and 4.1 g, respectively. This is in contrast to Nd metal which oxidizes to form
Nd2O3 within a few hours after being exposed to ambient air at room temperature. Thus these
alloys can be easily handled and stored without any special precautions.
The second point is that since the Er-based alloys are solid solution alloys, and not
intermetallic compounds, they can be readily fabricated into spheres, wires and foils (ribbons),
see Fig. 8. Since they are ductile alloys with reasonable strength they will not decrepitate or
crumble, which can easily occur when using brittle intermetallic compounds in regenerators due
Figure 8. Fabricated forms of spherical powders – 0.30mm (12 mil) diameter; ribbon –
0.05mm (2 mil) thick; and wire – 0.30mm (12 mil) diameter.
to the high pressure pulses of the gases that are rapidly recycled through the regenerator. Details
on the fabrication of the and alloys into spherical powders are being
presented in another paper at this conference.12
Initial Test Results

An initial test using in place of Pb as the regenerator material in a single
stage pulse tube cryocooler indicated that at low frequencies the Er interstitial alloy performed
better than Pb. Additional results using are presented in another paper at this
conference.13
CONCLUSIONS AND SUMMARY

We have shown that Er-based alloys in comparison to Pb have significantly (1) higher
volumetric heat capacities from 10 to 80 K, (2) lower thermal conductivities, and (3) improved
tensile strength. The Er alloys are oxidation resistant below 396 K (123°C), and can readily be
formed into wires, ribbons (foils) and spheres. We have suggested that the most efficient
replacement for lead in a regenerator is to use a combination of three alloys at
the hot end, at the intermediate temperatures, and at the cold end. Finally we
wish to note that these Er-based alloys are not a replacement for Er3Ni, Nd or HoCu2, which are
required to reach temperatures below 10 K, but are to be used in conjunction with them if
temperatures below 8 K are required. Our theme is not only to improve cryocooler performance
but also to “get the lead out.”
ACKNOWLEDGEMENTS
This work was supported in part by Atlas Scientific, Sunnyvale, California via a SBIR, and
in part by the Materials Sciences Division, Office of Basic Energy Sciences, U.S. Department of
Energy under contract W-7405-ENG-82.
MAGNETIC REGENERATOR ALLOYS FOR10 TO 80 K RANGE 441
REFERENCES
1. Buschow, K.H.J., Olijhoek, J.F. and Miedema, A.R., “Extremely Large Heat Capacities Between 4
and 10 K”, Cryogenics, vol. 15 (1975), pp. 261-264.
2. Sahashi, M., Tokai, Y., Kuriyama, T., Nakagome, H., Li, R., Ogawa, M. and Hashimoto, T., “New
Magnetic Material R3T System with Extremely Large Heat Capacities Used as Heat Regenerators”,
Adv. Cryogen. Eng., vol. 35 (1990), pp. 1175-1182.
3. Kuriyama, T., Hakamada, R., Nakagome, H., Tokai, Y., Sahashi, M., and Li, R., Yoshida, O.,
Matsumoto, K. and Hashimoto T., “High Efficient Two-Stage GM Refrigerator with Magnetic
Materials in Liquid Helium Temperature Region”, Adv. Cryogen. Eng., vol. 35 (1990), pp. 1261-
1269.
4. Gschneidner, K.A., Jr., “Physical Properties and Interrelationships of Metallic and Semimetallic
Elements”, Solid State Phys., vol. 16 (1964), pp. 275-426.
5. Although the crystalline electric field splitting of the Er 4f electrons in the ground state multiple 4I15/2
in Er3Ni are not known, those of the isostructural Er3Co compound have been determined. The
calculated and measured Schottky heat capacity of Er3Co shows a broad peak at ~12 K which slowly
falls off with increasing temperature reaching about half the peak value at 30 K. We would expect
that the Er3Ni heat capacity would show a very similar behavior. This work was reported by:
Takahashi Saito, A., Tutai, A., Sahashi and M., Hasimoto, T., “Crystal Field Effects on Thermal and
Magnetic Properties of Er3Co”, Jpn. J. Appl. Phys., vol. 34 (1995), pp. L171-L173.
6. Merida, W.R. and Barclay, J.A., “Monolithic Regenerator Technology for Low Temperature (4 K)
Gifford-McMahon Cryocoolers”, Adv. Cryogen. Eng., vol. 43 (1998), pp. 1597-1604.
7. Sinha, S.K., “Magnetic Structures and Inelastic Neutron Scattering: Metals, Alloys and Compounds”
in Handbook on the Physics and Chemistry of Rare Earth, Gschneidner, K.A., Jr. and Eyring, L.,
eds., North-Holland Publishing Co., Amsterdam (1978), vol. 1, chap. 7, pp. 489-589.
8. McEwen, K.A., “Magnetic and Transport Properties of the Rare Earths” in Handbook on the Physics
and Chemistry of Rare Earth, Gschneidner, K.A., Jr. and Eyring, L., eds., North-Holland Publishing
Co., Amsterdam (1978), vol. 1, chap. 6, pp. 411-488.
9. Scott, T.E., “Elastic and Mechanical Properties” in Handbook on Physics and Chemistry of Rare
Earths, Gschneidner, K.A, Jr. and Eyring, L., eds., North-Holland Publishing Co., Amsterdam
(1978), vol. 1, chap. 8, pp. 591-705.
10. Anonymous in Properties and Selection: Nonferrous Alloys and Special-Purpose Materials, Metals
Handbook, 10th ed., ASM Intern., Materials Park, OH (1990), vol. 2, p. 217 (Bronze) and p. 550 (Pb).
11. Anonymous in Properties and Selection: Irons, Steels and High-Performance Alloys, Metals
Handbook, 10th ed., ASM Intern., Materials Park, OH (1990), vol. 1, p. 855.
12. Miller, S.A., Nicholson, J. D., Gschneidner, K.A., Jr., Pecharsky, A.O. and Pecharsky, V.K.,
“Manufacturing Considerations for Rare Earth Powders Used in Cryocooler and Magnetic
Refrigerator Applications”, paper presented at 11th International Cryocooler Conference, June 20-22,
2000, Keystone, Colorado.
13. Kashani, A., Helvensteijn, B.P.M., Gschneidner, K.A., Jr., Pecharsky, A.O. and Pecharsky, V.K.,
“New Regenerator Materials for Use in Pulse Tube Coolers”, paper presented at 11th International
Cryocooler Conference, June 20-22, 2000, Keystone, Colorado.
Low Temperature Properties of
HoSb, DySb, and GdSb
H. Nakane, S. Yamazaki, H. Fujishiro*, T. Yamaguchi**,

S. Yoshizawa**, T. Numazawa*** and M. Okamura****
Kogakuin University, Shinjuku-ku, Tokyo, 163-8677, Japan

*
Iwate University, Ueda, Morioka, 020-8551, Japan
**
Meisei University, Hino-shi, Tokyo, 191-8506, Japan
***
National Research Institute for Metal, Tsukuba-shi, 305-0003, Japan
****
Toshiba Corporation, Yokohama-shi, 235-8522, Japan
ABSTRACT
Materials including rare-earth and Sb compounds were developed as regenerator materials for
a 4K GM refrigerator. Sb compounds have a remarkably large peak value of specific heat in the low
temperature range. First, the heat capacities, thermal conductivity, and thermal expansion of HoSb,
DySb and GdSb, which are anti-ferromagnetic with a large spin value in the low temperature range,
were measured. Then, the results of these measurements were used to analyze the problems en-
countered when Sb compounds were used as the regenerative materials in a GM refrigerator.
As for the thermal expansion, the measured values of Sb compounds were compared with that
of fabric-impregnated phenol-formaldehyde resin, which is usually used as the wall material of a
regenerator. As regards the thermal conductivity, the thermal diffusivity was obtained by using the
measured values of thermal conductivity and specific heat. The heat penetration depth was evalu-
ated from the thermal diffusivity. Discussion of the heat penetration depth is based on the figures
obtained in a previous experiment using a GM refrigerator with HoSb packed at the cold end of the
second stage of a multi-layer regenerator.
INTRODUCTION
In a small 4 K refrigerator, the refrigeration capacity depends on the response speed of heat
exchange, i.e., the exhaustion and absorption of heat between the working gas and the materials
packed into the regenerator. In order to obtain higher effectiveness in the regenerator, the heat
capacity of the regenerator materials must be larger than that of 4He used as the working gas. Only
a magnetic material that has a large specific heat based on magnetic phase transition is effective
below 15 K. However, the specific heat of pressurized 4He gas is very large below 15 K. Since the
heat exchange region of 4He gas is wide, a single magnetic material cannot cover the specific heat
of4He gas. Multi-layer regenerators composed of several rare-earth compounds that have different
specific heat peak temperatures are usually used.
HoSb and DySb compounds were discovered to have a remarkably large specific heat peak.
HoSb compound has the peak value of 2.7 J/K cm3 at 5.3 K, and DySb, 2.0 J/Kcm 3 at 9.5 K. The
measured results of the specific heat1 of HoSb, DySb and GdSb are shown in Fig. 1. HoSb was

Figure 1. Temperature dependence of specific heat of Sb compound.
packed into a multi-layer regenerator and the refrigeration capacities were measured and are in-
cluded in the report.2
In regards to the measured values of thermal expansion, the relation between the fabric-im-
pregnated phenol-formaldehyde resin (fabric-impregnated Bakelite), which is usually used as the
wall material of regenerators, and Sb materials, which have a remarkably large peak, was analyzed
in this experiment. Then, the thermal conductivities of HoSb, DySb and GdSb were measured, and
the thermal diffusivity was obtained from the measured values of thermal conductivity and the
specific heat. The heat penetration depth of HoSb was compared with that of4He used as the work-
ing gas. Whether the driving cycle of the displacer was optimal was verified for the former experi-
ment2 in which HoSb was packed at the cold end of the second stage multi-layer regenerator of a
GM refrigerator.
EXPERIMENTS
The samples used were intermetallic compounds of HoSb, DySb and GdSb made by arc weld-
ing at the melting points of 2433 K (2160°C), 2443 K (2170°C) and 2403 K (2130°C), respectively.
Two Sb compounds were found to have remarkably large specific heat peak values in the low tempera-
ture range. One of these compounds is HoSb, with the peak value of 2.7 J/K cm3 at 5.3 K, and the
other one is DySb, with 2.0 J/K cm3 at 9.5 K. GdSb has a broad specific heat peak around 24 K.
The measuring methods of thermal conductivity and thermal expansion of such magnetic ma-
terials are explained and the measuring results are shown as follows:
Measurement and Discussion of Thermal Expansion

The thermal expansions of HoSb, DySb and GdSb were measured by a clip type dilatometer as
shown in Fig. 2. A resistance bridge was formed by four strain gauge elements (Kyowa Electronic
Instruments, KFL-1-120-C1-11) bound to a phosphor-bronze clip. The balance of the resistance
bridge was directly measured to an accuracy of 71/2 figures with an automatic thermometer bridge
(TINSLEY 5840D) by the four-wire method. At a certain temperature, the balance of the resistance
bridge changes almost linearly to the length of the specimen. To obtain the calibration values of
for the expansion, the of two copper reference specimens of different lengths was measured,
as the thermal expansion rate of copper is well known.
Figure 2. Schematic diagram of the clip type dilatometer using four strain gage elements.
LOW TEMPERATURE PROPERTIES OF HoSb, DySb AND GdSb 445
Figure 3. Thermal expansions of GdSb, DySb and HoSb.
Figure 4. Temperature dependence of Sb compounds and Bakelite.
The lengths of HoSb, DySb and GdSb samples were 10 mm, and the cross sections about 6.3
mm x 2.3 mm. The accuracy of the measurement increases with the size of the cross section. As for
the fabric-impregnated phenol-formaldehyde resin (fabric-impregnated Bakelite), which is usually
used as the wall material of regenerators, a sample of 10 mm in diameter and 10 mm in length was
cut out from a circular rod. The linear expansions of HoSb, DySb and GdSb are shown in Fig. 3. As
shown in Fig. 1, HoSb had the highest specific heat peak, DySb second, and GdSb third. However,
the sharpness of change in the linear expansion around the Néel temperature was DySb, first; HoSb,
second; and GdSb, third (not sharp). In Fig. 4, the linear expansions normalized at 300 K are ex-
pressed as percentages.
The linear expansions of the samples at 4 K were about – 0.2 % for Sb compounds, – 0.5 % for
Bakelite against the diameter and – 0.2 % against the length. The contraction of Bakelite in diam-
eter was larger than those of HoSb and DySb.
In a former experiment,2 in which HoSb was packed at the cold end of the second stage of a
multi-layer regenerator of a GM refrigerator, the effect of contraction of the wall materials was
compensated by using felt as the partition material.
Measurement and Discussion of Thermal Conductivity
The thermal conductivity was measured by the steady state heat flow method. Fig. 5 shows a
schematic view of the sample set on top of the cold head of a GM refrigerator used as a cryostat3.
Both samples were glued to the cold-finger of the refrigerator and the metal film resistance heater
(10 kW) to the sample with GE7031 varnish. AuFe(0.07 at.%)-Chromel thermocouples with a di-
ameter of 73 µm were used as thermometers. The temperature range for the measurement was from
4 to 300 K. Air from the whole chamber of the sample was removed to 1 x 10–6 Torr with an oil
diffusion pump. The samples were cut to the dimensions of 3 mm x 3 mm x 10 mm as the accuracy
Figure 5. Schematic diagram of the thermal diffusivity measurement system.
Figure 6. Temperature dependence of thermal conductivity.
of measurement inversely increases with the size of the cross section in this method. The experi-
mental results of the thermal conductivity of HoSb, DySb and GdSb are shown in Fig. 6. The
thermal conductivity of lead (Pb), stainless steel, ErNi2, ErCo2 and DyNi2 which are used as con-
ventional regenerator materials, is shown in Fig. 6 for comparison.
Heat is periodically absorbed and exhausted by the materials used in a regenerator. Thermal
diffusivity is often a more convenient parameter than thermal conductivity in discussing such a
non-steady heat transfer subject. As is well known, thermal diffusivity D is given by:
where is the thermal conductivity and Cp the volumetric specific heat. By substituting the tem-
perature properties of the specific heat and thermal conductivity of each material into Eq. (1), the
thermal diffusivities D are obtained as shown in Fig. 7. As shown in Fig. 7, the thermal diffusivities
of HoSb and DySb at the Néel temperature are very low in comparison with that of Pb. It is, there-
fore, necessary to discuss whether the materials are effective under the conditions of non-steady
heat transfer that occurs in regenerators. Obviously the amplitude of temperature change is too small at
LOW TEMPERATURE PROPERTIES OF HoSb, DySb,AND GdSb 447
Figure 7. Temperature dependence of thermal diffusivities of Sb compounds.
a position far from the surface to cause effective heat transfer between the surrounding gas and the
material. Let us consider a one-dimensional semi-infinite body model, as discussed in literature.4
In this simple model, heat is transferred in only one direction and the body is heated periodically
only on the surface. This model is often used to check the general tendency of the measured data.
We define the position where the amplitude of temperature change is 1 / e times that on the
surface as the penetration depth of heat. The penetration depth of heat Ld is described as:
We assume that the material between the surface and Ld absorbs and exhausts the heat effec-
tively, and suggest that the material dimensions used should be smaller than Ld. The heat penetra-
tion depth of three different cycles is shown as a function of thermal diffusivity in Fig. 8. The
effective range of thermal diffusivity for HoSb, DySb, Er3Ni and lead is also shown. The optimum
sizes of the regenerator materials can be estimated from Fig 8. Based on a previous experiment
with a GM refrigerator2 in which the displacer was operated at a frequency of 74 rpm and the size of
the regenerator material was 0.3 mm, we can estimate from Fig. 8 that either Sb or lead will be
effective in the heat exchange process.
Figures 9 and 10 describe the temperature dependence of the heat penetration depth of HoSb
with 4He as the working gas, as obtained from Eq. (1), and using the operating frequency as a
parameter. When the displacer was operated at a frequency of 74 rpm, a shallow heat penetration
depth was observed around the Néel temperature (5.3 K) and the penetration depth was 0.7 mm.
Figure 8. Relation between penetration depth of heat and thermal diffusivity.

Figure 9. Thermal diffusion of HoSb.
Figure 10. Thermal diffusion depth of 4He at 1.5 MPa.
The effectiveness of heat exchange should not be affected by the size of the particles as 0.3 mm
particles, smaller than 0.7 mm, were used in this experiment.
The penetration depth of the heat of 4He over 0.3 mm at 600 rpm can be usable because it is
considered larger than the hydraulic diameter.
CONCLUSION
The basic properties of HoSb, DySb and GdSb as regenerator materials were measured at a low
temperature. As for the thermal expansions, the relationship between the phenol-formaldehyde
resin (Bakelite) and the Sb materials, which have a remarkably large peak, was analyzed. The
thermal conductivity of HoSb, DySb and GdSb was also measured and the thermal diffusivity was
furthermore obtained by using the measured values of the thermal conductivity and specific heat.
The heat penetration depth was evaluated from the values of thermal diffusivity. It was compared
with the thermal expansion of 4He gas. Verification of the optimal driving cycle for the displacer
was carried out for a previous experiment in which HoSb was packed at the cold end of the second
stage multi-layer regenerator of a GM refrigerator.
REFERENCES
1. Nakane, H. et al., “Multilayer Magnetic Regenerators with an Optimum Structure around 4.2K,”
2. Nakane, H. et al., “Refrigeration Capacity of a GM Refrigerator Utilizing HoSb as its Regenerative
Material,” Advances in Cryogenic Engineering, vol. 45, (2000), pp. 397-402.
3. Ikebe, M. et al., “Simultaneous Measurement of Thermal Diffusivity and Conductivity Applied to
Bi-2223 Ceramic Superconductors”, Journal of the Physical Society of Japan, vol. 63, No. 8,
(1994), pp. 3107-3114.
4. Ogawa, M. et al., “Thermal Conductivities of Magnetic Intermetallic Compounds for Cryogenic
Regenerator,” Cryogenics, vol. 31 (1999), pp. 405-410.
Manufacturing Considerations for Rare Earth
Powders Used in Cryocooler and Magnetic
Refrigerator Applications
S. A. Miller1, J. D. Nicholson1, K. A. Gschneidner, Jr.2,3,

A. O. Pecharsky2, and V. K. Pecharsky2,3
1
Starmet Corporation
Concord, MA 01742, USA
2
Ames Laboratory and 3Department of Materials
Science and Engineering
Ames, IA 50011-3020, USA
ABSTRACT
The high chemical reactivity and oxygen affinity of the rare earth metals make them especially
difficult to produce and maintain as high purity powders. From the initial oxide reduction through
bulk melt preparation and finally powder production the sensitivity of the metals and alloys to
contamination must be attended to at all times.
Production of these powders has raised new challenges for the powder manufacturer. Process
adaptation to the unique requirements of the rare earth materials to insure a quality product will be
described. The physics of the current powder production technique will be reviewed and related to
the actual product made. Individual processing steps with their influence on final product quality
will be reviewed. Finally, the knowledge gained during the pre-production learning curve will be
used to define potential manufacturing process that could be used for the economic production of
large quantities (>100 kg) of these powders when needed.
The volumetric heat capacities of powders of Gd and Er metals, and of an Er-Pr alloy, which
were prepared by the Plasma Rotating Electrode Process (PREP) have been measured and com-
pared to the bulk starting materials. The heat capacities are essentially the same within experimen-
tal error. The three materials have been used as regenerator materials: Gd for an active magnetic
refrigerator, and Er and Er-Pr as passive regenerator materials for Gifford-McMahon and pulse
tube cryocoolers,
INTRODUCTION
The rare earth powders used in this study were made by PREP®. Powder is produced by control-
lably melting the end of a round bar of the chosen material with a plasma torch while the bar is
rapidly rotating about its longitudinal axis. The molten metal is centrifugally ejected from the
surface of the bar and forms droplets that solidify to form spherical powder particles. A conceptual
schematic of the process is shown in Figure 1.

Of the current commercially available metal atomization processes PREP has several
inherent process characteristics that make it uniquely suitable for the fabrication of powders for
high performance applications. First, because PREP is a means of contactless melting and
atomization, PREP produces powder with the highest level of cleanliness possible. This is a
critical feature for reactive, high melting temperature alloys that are aggressively corrosive in
their molten state and attack conventional ceramic crucibles and for alloys that have a high
sensitivity to very small levels of contamination. Such alloys are routinely atomized by PREP
without incurring contamination. Examples are titanium, zirconium, molybdenum and rare earth
metals and alloys. Second, PREP powder is almost perfectly spherical and essentially satellite
free, because during the atomization process droplets are dispersed and move radially away from
each other. Thus, there is little opportunity for collisions between droplets and particles and the
resulting coalescence of the two into irregularly shaped clusters. This single particle nature of
the powder spheres results in PREP powder being very free flowing and having a high packing
density, approximately 65%. Third, PREP powder combines both a tighter geometric standard
deviation and a larger median size than can be produced via other techniques.1 Finally, because
PREP atomization mechanism is produced by centrifugal forces rather than by aerodynamic drag
the powder is essentially porosity free when compared to gas atomized powder.
PREP is conducted in a vacuum/controlled atmosphere tank 2440 mm (96 in.) in diameter
by 600 mm (24 in.) long. Tank dimensions limit maximum powder size to ~1.5 mm, while
maximum spindle speeds of 2620 radians/s limit minimum powder size to ~40 µm. Usually,
powder production is conducted under inert gas; the preferred medium is helium, which offers
both improved heat transfer properties and plasma performance.
Accurately controlling the rotation speed of the bar to be melted is necessary to obtain a
desired median particle size. The diameter of the molten droplet diameter is determined by the
properties of the liquid metal, the centrifugal ejecting forces (related to rotation speed), and to a
limited extent the aerodynamics of the droplet trajectory through the inert cover gas. The
following equation predicting median droplet size is obtained from a force balance of the
centrifugal forces acting on the molten surface tending to cause atomization with the surface
tension forces resisting atomization:
CONSIDERATIONS FOR RARE EARTH POWDERS 451
where d is the median droplet diameter (microns); is rotation rate (radians/s); and are
the surface tension (dynes/cm) and density (g/cm3) of the alloy being atomized, D is the
electrode diameter (cm) and K is an empirical constant principally determined by the method of
droplet formation which is in turn controlled by the melting rate.2 For any particular alloy the
material properties are fixed and (1) can be further simplified to
where K is a constant (over a limited melting rate range) that has been determined for many alloy
systems.3 Lacking from current process understanding, however, is the ability to control the
geometric standard deviation of the powder produced. PREP like the other commercial
atomization processes is an inherently stochastic process, and the degree of spread in the particle
size distribution is outside operator control. While the amount of spread can not be controlled
directly, it is noted that the geometric standard deviation for PREP is approximately 1.2,
providing a significantly tighter particle size distribution than competitive processes.
The Pr, Gd and Er metals were purchased as chunks from Tianjiao International Trading
Company (formerly Baotou Steel and Rare Earth Company) of China. The vendor stated that the
metals were 99.9% pure. Chemical analysis indicated that the Pr metal was only 97.8a/o
(99.5w/o) pure (the major impurities were [in a/o] 0.60 O, 0.4 Li, 0.3 Al, 0.3 Fe, 0.2 Si, 0.2 C,
0.2 Nd; Gd 93a/o pure (99.7w/o) pure (the major impurities were [in a/o] 3.2 H, 1.2 O, 1.0 C, 0.8
F, 0.4 N and 0.1 Ca); and Er 96.8a/o (99.4w/o) pure (the major impurities were [in a/o] 2.7 O,
0.7 F, 0.4 H, 0.3 N, 0.2 C, 0.2 Ta, 0.1 Ca). Five to 15 kg of the metal chunks were vacuum-
melted at the Ames Laboratory in a tantalum crucible at 1600°C for Pr and Gd (1700°C for Er)
for one hour to reduce calcium and hydrogen impurities by vaporization. The molten metal was
bottom-poured to form 6.5 cm diameter rods. The rods were finish machined in preparation for
PREP processing. In the case of Gd for the first run, 26% of the initial 13.6 kg of metal yielded
spheres with a diameter < 150 µm, 32% of the spheres between 150 and 300 µm, and 5% of the
spheres with a diameter > 300 µm.
The other major losses of material (other than the unwanted size spheres were volatiles in the
melting operation (minor), slag in the casting step (intermediate), and machining (major) and the
end stubs from PREP process (intermediate). The yield of usable spheres (between 150 and 300
µm) from a second run of Gd metal was increased to 51%. Chemical analysis of the spheres
revealed that the hydrogen content was reduced to 0.08 a/o and the calcium to < 0.005 a/o, while
the other interstitial impurity contents remained essentially the same. Similar results were
obtained on the Er materials.
The Pr-73a/oEr was prepared by melting chunks of weighed amounts of Pr and Er in the
proper ratio in a Ta crucible under an Ar atmosphere at ~1565°C before drop casting the molten
alloy into a water cooled Cu crucible. The melting point of this alloy was estimated to be
1440°C. The Pr to Er ratio was determined by x-ray fluorescence analysis along the length of the
rod and a cross-section perpendicular to the rod axis. Since the Pr concentration (and thus the Er
content) varied by more than ±1 a/o, the sample was broken up and remelted. After remelting
the concentration of Pr was uniform to with ±1 a/o. After machining, this alloy was used to
make powders by the PREP process. It is important that the Pr concentration be uniform
throughout the rod, because the low temperature heat capacity varies significantly with changing
Pr concentration.4 If the Pr concentration is not reasonably uniform the produced powders will
have a range of compositions, and thus, a range of low temperature heat capacity values.
Furthermore, these powders of variable composition will be intimately mixed and impossible to
separate into various fractions with more uniform compositions and properties.
The results for the Gd powders used as an active magnetic regenerator in a magnetic
refrigerator were previously reported in 1997.6 The 1997 results showed that magnetic
refrigeration is a viable cooling technology and competitive with conventional gas compression
systems. In addition it is an environmentally friendly technology eliminating CFC, HCFC and
NH3 gases.
Before each powder type was atomized the vessel was evacuated to less than 50 µm and then
backfilled with helium to a positive pressure of 40 kPa. The bars were brought up to the desired
rotational speed and melting was commenced. Once atomization was completed the heat source
was shut off, the powder allowed to cool to ambient and then the powder was removed from the
chamber under an inert cover. A sample for sieve analysis was removed, and the remainder of
the powder sized to specification for testing. The powder was then packed and shipped under
argon. Further analysis was performed on the Er and Pr-a/o73 powders at the Ames Laboratory
and is reported elsewhere in this proceedings.4
Figure 2 is a SEM micrograph of the Pr-73a/oEr powder produced in this study. Already
classified to –250+104 µm the micrograph demonstrates the high level of sphericity and the lack
of satelliting that characterizes PREP powder. While the physics governing the PREP process is
well understood and easily mathematically reduced to equation 1 there is a lack of sufficient melt
property data, to make full use of equation 1 in a production environment. A fair amount exists
for elemental materials, but as Table 1 implies, the data is scarce for binary and more complex
alloys. As a result frequently equation 2 is employed at the expense of having to make a
prototype run in order to determine the constant K before making large quantities of powder.
Thus the rotational speed, in Table 2 was initially back calculated by using equation 1 with
the limited data and a known desired particle size d. Due to the scarcity of the required data, the
actual speed shown in Table 2 is a compromise between operator experience and the predicted
Figure 2. SEM micrograph of Pr-73a/oEr powder showing the high degree of sphericity and lack of
satelliting.
Figure 3. Erbium powder size distributions produced by the PREP process.

value. The powder from this run is then analyzed for the median size produced, that value is
entered into equation two with the speed and bar diameter employed and the process constant K
is back calculated. Equation 2 is then used with the desired median particle size to determine the
correct rotational speed for an extended production run. Figure 3 shows the particle size
distributions obtained by this procedure for Er and how rapidly the process can converge on a
desired size.
The low temperature volumetric heat capacity of the Pr-73a/oEr alloy before and after
processing the alloy into spheres is shown in Figure 4. As is evident the heat capacity of the Pr-
73a/oEr spheres are essentially the same as that of the bulk original starting alloy. The slight
difference in the heat capacities at the 36 K peak is due to the fact that the spherical
Pr-73a/oEr powders are mixed with Ag powders (~30 volume %) and compacted in order to
make the heat capacity measurements. When the known heat capacity of Ag is subtracted from
the measured value this can result in a slight subtraction error especially at temperatures above
~30 K where the heat capacity of Ag increases significantly relative to that at lower
temperatures.
CONCLUSION
Given the current level of demand for monosized, or near monosized rare earth powders for
cryocooler and magnetic refrigerator applications PREP processing offers the best current
commercial product in terms of sphericity, lack of contamination and tight size distribution.
With careful manufacturing practice powder can be produced without a loss in the properties
relevant to cryocoolers and magnetic refrigerators. However, as demand for the powder
increases the development of newer technologies to production status will be warranted to
provide improved product to the cryocooler community. From the powder producer's point of
Figure 4. The volumetric heat capacity from 4 to 79 K for both the bulk form and 100-250 µm spheres of
Pr-73a/oEr obtained by the PREP process.
view the remaining challenge is to produce a more monosized spherical powder for improved
material utilization and system performance.
ACKNOWLEDGEMENTS
Ames Laboratory is operated for the U.S. Department of Energy (DOE) by Iowa State
University under Contract No. W-7405-ENG-82. The work was partially supported by the
Advanced Energy Projects and Technology Research Division, Office of Computational and
Technology Research of DOE; by Atlas Scientific, Sunnyvale, California via a SBIR; and by the
Materials Sciences Division, Office of Basic Energy Sciences of DOE.
REFERENCES
1. B. Champagne and R. Angers, “Fabrication of Powders by the Rotating Electrode Process”, Int. J.
Powder Metall. Powder Technol., Vol 16 (No. 4), 1980, pp. 359-367.
2. S. Miller and P. Roberts, “Rotating Electrode Process”, Powder Metal technologies and Applications,
ASM handbook-Volume 7, S. Lampman et. al. eds., ASM International, Materials Park OH, 1998,
pp. 97-101.
3. B. Champagne, and R. Angers, “Size Distribution of Powders Atomized by the Rotating Electrode
Process”, Modern Development in Powder Metallurgy, Proceedings of the 1980 International Powder
Metallurgy Conference, Washington, DC, Hausner, H., Antes, H., and Smith G., eds., Metal Powder
Industries Federation, Princeton NJ, Vol 12, 1981, pp. 83-104.
4. Gschneidner, K.A., Jr., Pecharsky, A.O. and Pecharsky, V.K., “Ductile, High Heat Capacity,
Magnetic Regenerator Alloys for the 10 to 80 K Temperature Range”, paper presented at 11 th
International Cryocooler Conference, June 20-22, 2000, Keystone, Colorado.
5. Kashani, A., Helvensteijn, B.P.M., Gschneidner, K.A., Jr., Pecharsky, A.O. and Pecharsky, V.K.,
“New Regenerator Materials for Use in Pulse Tube Coolers”, paper presented at 11th International
Cryocooler Conference, June 20-22, 2000, Keystone, Colorado.
6. Zimm, C., Jastrab, A., Sternberg, A., Pecharsky, V., Gschneidner, K., Jr., Osborne, M. and Anderson,
I., “Description and Performance of a Near-room Temperature Magnetic Refrigerator”, Adv. Cryog.
Engin. vol. 43 (1998), pp. 1759-1766.
7. van Zytveld, J., “Liquid Metals and Alloys” in Handbook on the Physics and Chemistry of Rare
Earth, Gschneidner, K.A., Jr. and Eyring, L., eds., Elsevier Science Publishers B.V., Amsterdam
(1989) vol. 12, chap. 85. pp. 357-407.
Magnetothermal Properties of
Polycrystalline Gd2ln
M. I. Ilyn, A. M. Tishin
Physics Department M. V. Lomonosov Moscow State University
Moscow, Russia 119899
K. A. Gschneidner, Jr.1,2, V. K. Pecharsky,1,2 A. O. Pecharsky1

1
Ames Laboratory and Department of Materials Science and Engineering
Ames, IA, USA 50011-3020
ABSTRACT
The magnetic, thermodynamic and magnetocaloric properties of Gd2In have been
investigated as a part of a continuing study of novel lanthanide-based materials in order to
improve our knowledge about the nature of magnetic refrigerant materials and to find high-
performance alloys suitable for magnetic cooling in various temperature ranges. The dc
magnetization, ac and dc magnetic susceptibility, and heat capacity were measured from ~5 to
~350 K in magnetic fields varying from 0 to 100 kOe. Gd2In belongs to the hexagonal Ni2In-
type structure. It orders ferromagnetically at ~191 K and then antiferromagnetically at ~91 K.
The antiferromagnetic phase is easily transformed to the ferromagnetic state under the influence
of a magnetic field. The critical magnetic field varies from ~8.4 kOe at 5 K to zero at 91 K. The
magnetocaloric effect, as determined from magnetization data is in good agreement with
that calculated from the heat capacity results. The maximum adiabatic temperature rise, as
determined from the heat capacity measurements, is 7.4 K at 210 K for a magnetic field change
of 100 kOe.
INTRODUCTION
Experimental studies of the magnetocaloric properties of materials with unusual magnetic
structures can provide better understanding of the physics of the magnetocaloric effect (MCE).
To date, magnetothermal properties have been investigated in a limited number of materials with
certain types of magnetic and structural phase transformations. The physics of the MCE
occurring in the vicinity of magnetic order-order phase transitions in rare earth intermetallic
compounds is still lacking sufficient experimental and theoretical results. The unusual magnetic
properties of the Gd2In intermetallic compound were the reason for carrying out this
Cryocoolers 11, edited by R.O. Ross, Jr.

experimental study. Gamari-Seale et al.1, Jee et al.2 and McAlister3 have reported the
magnetization and electrical resistivity measurements and found that the compound Gd2In orders
ferromagnetically below 190 K and then antiferromagnetically below 99.5 K. It undergoes a
metamagnetic phase transition from an antiferromagnet to a ferromagnet in high magnetic fields
between ~4.5 and ~99 K. Taking into account what is known about the magnetic structures of
the compounds containing other lanthanide elements, which also have hexagonal crystal
structures, McAlister3 suggested that the low magnetized state (below 99.5 K) is a spiral
antiferromagnetic structure along the c axis. Furthermore, Jee et al.2 considered that there is
probably a helical ferromagnetic phase between 99.5 K and 190 K which can be transformed into
a simple ferromagnetic state by the application of an external magnetic field. Detailed
measurements of the ac magnetic susceptibility in the vicinity of the low temperature phase
transition have been carried out by Cowen et al.4 and Saran et al.5 Gamari-Seale et al.1 calculated
the paramagnetic Curie temperature and effective magnetic moment assuming Curie-Weiss
behavior of Gd2In, and obtained anomalously large value of the effective magnetic moment
per gadolinium atom. Based on their magnetic data Szade et al.6,7 suggested that Gd2In
does not obey Curie-Weiss law over a broad temperature range above the upper magnetic order-
disorder phase transition. The large value of in this compound and, correspondently, a
potentially large MCE, make Gd2In an interesting compound for an experimental study of its
magnetothermal properties.
EXPERIMENTAL
Polycrystalline Gd2In sample was prepared by arc melting of pure Gd (99.95 wt.% pure) and
In (99.99 wt.% pure) in an argon atmosphere. The gadolinium was prepared by the Materials
Preparation Center of the Ames Laboratory and the indium was purchased from Johnson
Matthey. The sample was re-melted six times with the button being turned over each time to
insure homogeneity. No heat treatment was carried out and all measurements were performed
using the as-prepared alloy. The x-ray powder diffraction study indicated no impurity phases
and confirmed the hexagonal Ni2In-type crystal structure. The space group is P63/mmc, and the
lattice parameters are: a = 5.442 Å and c = 6.767 Å, which is in excellent agreement with earlier
data. The dc magnetization, and dc and ac magnetic susceptibilities were measured from 5 to
320 K in magnetic fields up to 50 kOe using LakeShore ac/dc susceptometer. The heat capacity
was measured using an adiabatic heat pulse calorimeter8 from 3.5 to 350 K in magnetic fields
from 0 to 100 kOe.

Temperature dependencies of the dc magnetization measured in different applied magnetic
fields are shown in Fig 1. As one can see, there are two magnetic phase transitions in low
magnetic fields (2 and 5 kOe). At high temperature (above ~280 K), the low magnetized phase is
paramagnetic. The inset in Fig 1 represents a plot of inverse dc magnetic susceptibility versus
temperature. It is clearly seen that above ~280 K it behaves linearly. The effective magnetic
moment is per gadolinium atom and the paramagnetic Curie temperature is
230.4 K, assuming Curie-Weiss behavior between 280 and 320 K. The value of the effective
magnetic moment is distinctly larger than 7.94 µ B theoretically predicted for a free trivalent Gd
ion, but it is somewhat lower than that reported by Gamari-Seale et al1. The ordered magnetic
moment calculated from the saturated magnetization at 5 K after extrapolation to H = 0 is 7.23 µ B
per Gd atom, which is also slightly higher than the theoretically expected 7.0 µ B. During
cooling, the magnetization rapidly increased in the vicinity of 200 K and taking into account the
inflection point on the zero magnetic field heat capacity versus temperature (see below), we
conclude that Gd2In orders ferromagnetically at ~194 K. In the low temperature region
(T<100K), a second magnetic phase transition from a high to a weakly magnetized state is
detected. The low magnetized state completely disappears in magnetic fields exceeding ~10 kOe
MAGNETOTHERMAL PROPERTIES OF POLYCRYSTALLINE Gd2ln 459
Figure 1. The dc magnetization of Gd2In as a function of temperature. The inset displays the
behavior of the inverse dc magnetic susceptibility measured in 2 kOe Magnetic field. The line on the
inset represents a least squares fit to Curie-Weiss law.
Figure 2. The dc magnetization of Gd2In as a function of magnetic field. The inset shows the
values of critical magnetic fields, Hc, of the magnetic field induced metamagnetic transition of Gd2In
from an antiferromagnetic state to a ferromagnetic state.
Figure 3. The heat capacity of Gd2In as a function of temperature, in various magnetic fields.
Figure 4. The magnetic entropy change of Gd2In calculated from magnetization data.
(see Figs.l and 2). Isothermal magnetization, corrected for the demagnetization factor is shown
in Fig 2. Both linear initial part of the magnetization and the negative magnetocaloric effect in
low magnetic fields below ~80 K are indicative of an antiferromagnetic state in this temperature
region. We have plotted the critical magnetic field versus temperature (i.e. the magnetic field at
which the metamagnetic transition from an antiferromagnet to a ferromagnet occurs during a
field increase) in the inset in Fig. 2 and found that the temperature of this phase transition is 91 K
when taken as the point where the extrapolated Hc vs. T curve intersects with the temperature
axis. The maximum critical magnetic field of 8.4 kOe is determined as the point where
extrapolated critical fields curve intersects with the magnetic fields axis. The heat capacity of
Gd2In measured in different magnetic fields is shown in Fig 3. There are two distinct anomalies.
The first one is observed at ~95 K and is practically independent of the applied magnetic field.
The second one is observed in the vicinity of 190 K and is strongly dependent on the magnetic
field. Based on the shape of the heat capacity curves we believe that both phase transitions are
second order, which agrees with conclusions by Jee et al2 and McAlister3.
The magnetocaloric effect can be calculated as the adiabatic temperature change,
or as the isothermal magnetic entropy change Fig. 4 represents the
results obtained from numerical integration of the Maxwell relation (Tishin9) using
experimentally measured isothermal magnetization data. As expected, the magnetic entropy
change shows anomalous behavior in the vicinity of both low and high temperature magnetic
phase transitions. First, a rapid increase of occurs for all magnetic field changes near
100 K, and second, a maximum is clearly seen near 200 K. Below ~80 to ~90 K, the
sign of changes in low magnetic fields, which is consistent with the low temperature
antiferromagnetic structure of Gd2In in zero magnetic field.
The total entropy was calculated from numerical integration using experimentally measured
isofield heat capacity at several magnetic fields. We have calculated the isothermal magnetic
entropy change (Fig. 5) and the adiabatic temperature change (Fig. 6) as vertical and horizontal
segments on the S(T) diagram, respectively. Figure 7 shows the excellent agreement between the
values of the magnetic entropy change calculated from both magnetization and heat capacity
Figure 5. The magnetic entropy change of Gd2In calculated from heat capacity data.
Figure 6. The adiabatic temperature change of Gd2In calculated from heat capacity data.
Figure 7. Comparison of the magnetic entropy change calculated using heat capacity and
magnetization data.
data. The negative magnetocaloric effect is no longer visible in magnetic fields exceeding 50
kOe (Figs. 5 and 6). Although the maximum values of the MCE in Gd2In are somewhat smaller
than that in other known magnetocaloric materials in the same temperature range (e.g. in Dy it
reaches ~9 K at ~180 K for a magnetic field change from 0 to 50 kOe10), the measurable MCE
values are observed over a broad range of temperatures. This indicates that Gd2In may be
considered as a useful magnetic refrigerant material between ~150 and ~275 K, particularly when
magnetic fields greater than 50 kOe are available.
According to Bouvier et al11, the most appropriate value to describe a magnetic phase
transition point is the zero magnetic field heat capacity inflection point on the high temperature
side of the peak, which yields the transition temperature of 194 K for Gd2In (Fig. 3). However,
Tishin et al.10 suggested that there is a characteristic temperature, where the heat capacity
of the magnetic material is not affected by the magnetic field. approaches the temperature
of the maximum magnetocaloric effect, and both approach the magnetic ordering temperature in
low magnetic fields. Therefore, can be used as an indicator of the Curie (Neel)
temperature. Using this approach we have found that the ferromagnetic-paramagnetic transition
temperature of Gd2In is 190.9 K, which is much closer to the temperature reported by Jee et al.2
for this phase transitions (190 K), which was based on magnetization measurements.
CONCLUSIONS
The behavior of the inverse dc magnetic susceptibility of Gd2In above the ferromagnetic-
paramagnetic phase transition seems to follow Curie-Weiss law. However, the anomalously
large value of the effective magnetic moment 9.09 µ B per Gd atom compared to the theoretical
value of 7.94 µ B, suggests that the true paramagnetic state is not observed up to 320 K. The
ordered magnetic moment at 5 K is 7.23 µ B per Gd atom, which is also slightly larger than the
expected theoretically 7.0 µ B. Therefore, the overall magnetic moment of Gd consists of a
valence electron contribution of ~0.2 µ B and the remaining ~7.0 µ B from the localized 4f orbitals.
The strongest magnetocaloric effect in Gd2In is observed between ~150 and ~300 K, when it is in
the ferromagnetic state. The maximum value of the magnetocaloric effect in Gd2In is lower than
that obtained in the best materials which order over 200 K. However, since MCE exists over a
broad range of temperatures, Gd2In could be a useful magnetic refrigerant material for the
temperature range between ~150 and ~275 K in magnetic fields exceeding 50 kOe.
ACKNOWLEDGMENT
This work was supported by the Office of Basic Energy Sciences, Materials Sciences
Division of the U.S. Department of Energy, under Contract No. W-7405-ENG-82 (V.K.P. A.O.P.
and K.A.G.), and by a NATO Linkage Grant No. 950700 (all authors).
REFRENCES
1. Gamari-Seale H., Anagnostopoulos T., Yakinthos J. K., “Magnetic characteristics of rare-earth

indium R2In (R= Y, Nd, Sm, Gd, Tb, Dy, Ho, Er and Tm) intermetallic compounds”, J. Appl. Phys.,
vol. 50, no. 1 (1979), pp. 434-437.
2. Jee Chan-Soo, Lin C. L., Mihalisin T., Wang Xue-Qin, “Magnetization and specific heat studies of
Gd2In”, J. Appl. Phys., vol. 79, no. 8 (1996), pp. 5403-5405.
3. McAlister S. P., “Magnetic and electrical properties of Gd2In”, J. Phys. F, vol. 14 (1984), pp. 2167-
2175.
4. Cowen J. A., Williams G., “Non-linear susceptibility of Gd2In near the low temperature phase
transition”, J. Magn. Magn. Mater., vol. 171 (1997), pp. 129-134.
5. Saran M., Williams G., McAlister S. P., “Ac susceptibility of the intermetallic compound Gd2In”,
Solid State Commun., vol. 57, no.1 (1986), pp. 53-57.
6. Szade J., Neumann M., “Photoelectron spectroscopy and magnetism of some Gadolinium
intermetallic compounds”, J. Alloys Compd., vol. 236 (1996), pp. 132-136.
7. Szade J., Lachnitt J., Neumann M., “High-resolution Gd 4d photoemission from different
intermetalic compounds”, Phys. Rev. B, vol. 55, no. 3 (1997), pp. 1430-1434.
8. Pecharsky V. K., Moorman J. O., Gschneidner K. A., Jr., “A 3-350 K fast automatic small sample
calorimeter”, Rev. Sci. Instrum., vol. 68 (1997), pp. 4196-4207.
9. Tishin A.M., “Magnetocaloric effect in the vicinity of phase transitions” in Handbook of Magnetic
Materials, Buschow K. H. J. ed., North-Holland, Amsterdam, (1999), vol. 12, chap. 4, pp. 395-524.
10. Tishin A. M., Gschneidner K. A., Jr. and Pecharsky V. K., “Magnetocaloric effect and heat capacity
in the phase transition region”, Phys. Rev. B, vol. 59, no. 1 (1999), pp. 503-511.
11. Bouvier M., Lethuillier P., Schmitt D., “Specific heat in some gadolinium compaunds”, Phys. Rev. B,
vol. 43, no. 13 (1991), pp. 13137-13144.
New Regenerator Material for
Sub-4 K Cryocoolers
T. Numazawa1, O. Arai1, A. Sato1, S. Fujimoto2,

T. Oodo2, Y.M. Rang3 and T. Yanagitani4
1
Tsukuba Magnet Laboratory, National Research Institute for Metals
Tsukuba 305-0003, Japan
2
DAIKIN Air-Conditioning R&D Laboratory, Ltd.
3
DAIKIN Environmental Laboratory, Ltd.
4
Konoshima Chemical Co., Ltd.
Kagawa 769-1103, Japan
ABSTRACT
A new magnetic regenerator material for sub-4 K cryocoolers has been studied. Some rare-
earth oxide magnetic materials have the potential to provide a high heat capacity below 4 K. From
surveying the garnets and perovskites rare-earth oxides, we have found that a perovskite GdAlO3
(GAP) has a very large volumetric heat capacity below 4 K and it is useful as a regenerator
material. Fine particles of the polycrystal GAP between 100 and 500 µm sizes were fabricated by
a chemical process. The measured heat capacity of the fabricated GAP was in good agreement
with that of single crystal material, and the magnetic field dependence was negligible when apply-
ing a magnetic field of 1 T. From thermal conductivity measurements, GAP was found to have 3
~ 10 times higher thermal conductivity than that of SUS from 5 K to 20 K. A refrigeration test
was also done with a 4 K pulse tube refrigerator operating with a 3.3 kW compressor. The cooling
power was increased remarkably from 165 mW to 250 mW at 4.2 K when using GAP in the lower
temperature portion of the 2nd-stage regenerator. The minimum no-load temperature was lowered
from 2.9 K to 2.5 K. There was an optimum value for the volumetric ratio of GAP that corre-
sponded to the best cooling performance.
INTRODUCTION
Since the introduction of magnetic materials in cryogenic regenerators in place of lead (Pb),
a remarkable improvement in cooling performance has been achieved in small refrigerators below
10 K. This is because the heat capacity of the magnetic materials is large enough to cool the
helium gas in the regenerator from ~ 10 K to 4 K. Several magnetic materials such as Er3Ni,
HoCu2 and Nd have been applied in 4 K cryocoolers so for, and they have contributed greatly to
the realization of various cryogen free applications. But when refrigeration below 4 K is consid-

ered, it is observed that the regenerator materials previously developed do not provide sufficient
cooling power because the heat capacities of those materials are relatively low below 4 K.
The main purpose of the present study has been to develop a magnetic regenerator material
that has sufficient heat capacity below 4 K, and then to test it in a real refrigerator. This paper
begins with a survey of the magnetic materials for sub-4K refrigeration. We then describe the
fabrication process of a new oxide regenerator material and its thermal properties. Finally, experi-
mental results are presented on its cooling performance in a pulse tube refrigerator below 4.2 K.
EXPERIMENTAL
Criteria for Choosing Regenerator Materials below 4 K
As practical magnetic regenerator materials, rare-earth metals and their intermetallic com-
pounds are commercially available, and some reasons for choosing these materials include:
- Rare-earth elements provide a large magnetic specific heat coming from the large magnetic
moments of their 4f electrons.
- Peak temperatures in their specific heats lie between 5 K and 15 K.
- Some materials have additional specific heats associated with the Schottky anomaly above
their magnetic transition temperatures.
- The peak temperatures in the specific heats can be tailored in the intermetallic compounds.
When we look for regenerator materials for use below 4 K, it is very difficult to find suffi-
ciently high heat capacity in rare-earth metals or intermetallic compounds. Generally the mag-
netic interaction in those materials is too strong to provide a magnetic transition temperature
relating to the peak of the specific heat below 4 K. As a result, the heat capacity is very small
below 4 K, and therefore, insufficient cooling power is provided below 4 K.
As a candidate solution, the authors have proposed another approach to obtaining high heat
capacity below 4 K; this involves the use of oxide magnetic materials.2 Some rare-earth magnetic
oxides can keep their magnetic transition temperatures below 4 K because of their weak magnetic
interactions. In particular, the Gd ion has the potential to provide a large magnetic specific heat
because of no L-S coupling, i.e., holding its high degeneracy even at low temperatures.
Specific Heat Properties of Some Rare-earth Oxides

Following the above criteria, we first considered the garnet structure, R3B5O12 (R: rare-earth
ion and B: trivalent ion). Some rare-earth garnets have been used for magnetic refrigeration
because of their large magnetocaloric effects, and they also have large specific heats correspond-
ing to their magnetic orders. Typical rare-earth garnets with large specific heats such as Gd3Ga5O12
(GGG) and Dy3Al5O12 (DAG) have been investigated as regenerator materials in a previous pa-
per.2 However, not only is the garnet structure very complex, but it also has a very large unit cell.
Thus, much simpler crystal structures should be considered and we found that the perovskite
structure, RBO3 (R: rare-earth ion and B: trivalent ion) to be a promising candidate. In particular,
previous data on the molar specific heats of GdAlO3 (GAP) and DyAlO3 (DAP) single crystals
show that they have high heat capacities below 4 K.3,4
Figure 1 (a) shows the measured molar specific heats normalized by R (R=gas constant) for
GGG, DAG, GAP and DAP samples. Broad and large peaks are observed in GAP and GGG and
such broad peaks are preferred for regenerator materials. Since the Gd ion has a weak magnetic
interaction in these oxides structures, GGG and GAP are expected to keep their high degeneracy
at low temperatures. Actually, we can estimate the contribution from the Gd ion to the specific
heat when the magnetic entropy (S) is calculated as:
where C: specific heat and T: temperature.

NEW REGENERATOR MATERIAL FOR SUB-4 K CRYOCOOLERS 467
Figure 1 (a). Measured molar specific heats for GGG, DAG, GAP and DAP.
Figure 1(b) shows the calculated results on the molar entropy for GAP and DAP. The entropy
curves show clearly that the saturation value in GAP is almost equal to ln(8), which is given by
S=ln(2J+1) for the magnetic moment of J = 7/2. This indicates that the Gd ion keeps the 8-fold
degeneracy even below 4 K, while the Dy ion only has the Kramers doublet (S=ln(2) for J=1/2)
in the temperatures as seen in the curve for DAP. It is concluded that the higher level of degen-
eracy will contribute to providing a higher heat capacity.
Figure 1(c) shows the volumetric heat capacities of GGG, DAG, GAP and DAP. The large
difference occurs when we convert the specific heat data from per-unit-mole to per-unit-volume.
The perovskites, GAP and DAP, have very much higher volumetric heat capacities as compared to
those for the garnets GGG and DAG. This is because a perovskite has only 20 atoms in the unit
cell, while garnet has 160. This will clearly contribute to providing a higher heat capacity per unit
volume, which is one of the most important factors for a good regenerator material. In particular,
GAP has a broad and high volumetric heat capacity.
Figure 2 shows the volumetric heat capacity of GAP and other magnetic regenerator mate-
Figure 1 (b). Calculated molar entropy for GAP and DAP.

Figure 1 (c). Volumetric heat capacities for GOG, DAG, GAP and DAP.
rials typically used in the 4 K cryocoolers. The GAP has a considerably larger heat capacity below
4 K than those of other materials, but the heat capacity decreases sharply above 4 K. Such tempera-
ture dependence in the GAP comes from a typical behavior of antiferromagnetic. At the magnetic
transition temperature around 3.8 K, the peak value of the heat capacity in the GAP is about 4 times
larger than that of HoCu2 and the heat capacities of both materials decrease when decreasing the
temperature, but the GAP still has 2.5 times larger heat capacity than that of HoCu2 at 2 K. Thus, it
is very interesting to use the GAP as the regenerator material for sub-4 K cryocoolers.
Fabrication of the GAP regenerator material

We have chosen the polycrystal form of GAP to use as the regenerator material. The most
important reason was that the GAP was one of the single crystals which is difficult to be grown
because of the high melting point above ~2050°C. Also some advantages of the polycrystal GAP
can be shown as:
- Calcination temperature of ~1700°C lower than ~2050°C
- Low cost
Figure 2. Volumetric heat capacities for GAP and other regenerator materials used for 4 K coolers.
- Mass production by using a chemical process

- Flexibility to fabricate various forms
As the practical regenerator material, sphere form particles between 100 µm and 500 µm in
diameter are required. We developed the following fabrication method to provide such particles.
Preparation of GAP Fine Powder
- Mix the Al2O3 and the Gd2O3 in the form of ultra fine powders less than 1 µm sizes with
ethyl alcohol.
- Dry the mixed powder by heating.
- Calcine for ~3 hours at 1200°C. GAP single phase will be obtainable at this stage.
- Grind the GAP powder with a ball mill adding ethyl alcohol, and then dry by heating.
Fabrication of GAP Sphere Form Particles
- Adding the GAP fine powders with a small amount of water into a granulator; fine particles
in the sphere form can be grown.
- Calcine the granules at 1700°C.
- Filtering the proper sizes of GAP granules between 100 µm and 500 µm by using screen
meshes. Irregular sized granules can be recycled.
Figure 3 shows the photographs for the polycrystal GAP plate and the GAP granules. The plate
can be fabricated into more complex shapes. The granules were in most spherical shape, but some
irregular ones were observed such as a Rugby football shape. The theoretical density of the GAP is
about 7.4 (g/cm3). The relative density, which is defined by the ratio of the density of fabricated
polycrystal to the theoretical value, was 99 % for the GAP plate and 95 % for the GAP granules.

Thermal Properties of Polycrystal GAP
Figure 4 shows the measured volumetric heat capacity of our fabricated polycrystal GAP in
comparison with that of single crystal GAP previously reported.3 The data for the polycrystal are in
good agreement with those of the single crystal and it can be concluded that our fabricated poly-
crystal GAP is most similar to the single crystal in regards to the magnetic properties.
Magnetic field dependence of the heat capacity is another important factor as the regenerator
material because the 4 K cryocoolers are often used with superconducting magnets. Figure 5 shows
the heat capacity of the polycrystal GAP with the magnetic fields of 0 and 1 Tesla (T). Since GAP
is a typical antiferromagnetic, only 1 T of the magnetic field does not change the shape of the heat
capacity largely as seen in Fig. 5.
Figure 3. Photographs for fabricated polycrystals GAP; (a) plate with a thickness of 2 mm and
(b) granules in diameters between 100 µm and 500 µm.
Figure 4. Comparison of heat capacity between fabricated polycrystal GAP and single crystal GAP.
Figure 5. Heat capacity data of polycrystal GAP with magnetic fields of 0 T and 1 T.
Figure 6. Thermal conductivity of polycrystal GAP, Er3Ni, SUS and Er3Ni.

We also measured the thermal conductivity of the polycrystal GAP from 3 K to 20 K. Figure 6
shows the experimental result with the data of several regenerator materials previously reported.5
The increasing rate on the thermal conductivity with the temperature on the GAP is much higher
than those of other materials. The GAP has about 3 times higher thermal conductivity than that of
SUS at 5 K and the difference increases more with increasing the temperature. Such high thermal
conductivity may be possible to make the frequency of refrigeration cycle higher.
Refrigeration Test with a 4 K Pulse Tube Refrigerator
In order to test the GAP as the regenerator material, we have used a 4 K pulse tube refrigerator
manufactured by DAIKIN Industries Ltd.6 The rated input power of the compressor unit was 3.3
kW at 50 Hz. Figure 7 shows the cross-sectional view of the refrigerator. The refrigerator has two-
stage pulse tubes with regenerators. The second-stage regenerator was filled with about 48 cm3 of
lead (Pb), Er3Ni and HoCu2, where the volumetric ratio for the materials was Pb:Er3Ni:HoCu2 =
2:1:1. Since HoCu2 was set in the lower temperature portion of the regenerator, we replaced a part
of HoCu 2 with the GAP as shown in Fig.7.
Figure 8 shows the cooling power at 4.2 K and the minimum temperature without heat load as
a function of volumetric ratio (R) defined by:
Figure 7. Schematic cross-sectional view of the 4 K pulse tube refrigerator used for cooling test.
Figure 8. Volumetric ratio (R) dependence of cooling power and minimum temperature.
where V is the volume of the regenerator material. There was a strong correlation between the
cooling power and the minimum temperature. Increasing R changes the cooling power and the
minimum temperature, and we can find an optimized value around R=0.25 to provide both the
largest cooling power of 250 mW at 4.2 K and the minimum temperature of 2.5 K. This experiment
clearly shows that the GAP considerably improves the performance of the refrigerator.
Figure 9 shows the temperature dependence of the cooling power for R=0 and R=0.25. Note
that solid (R=0.25) and broken (R=0) curves show the results under the same experimental condi-
tion, so both curves must be compared first. The cooling power decreases with decreasing the
temperature, but there is a remarkable difference between R=0 and R=0.25. As expected from Fig.
8, the cooling power was increased below 4.2 K. by using the GAP, for example, the GAP provided
1.6 times larger cooling power at 4.2 K than that of HoCu2 and the value becomes much larger with
decreasing the cooling temperature. The minimum temperature was also lowered by the GAP from
2.9 K to 2.5 K. The chain curve shows the experimental data for R=0.25 under the more optimized
condition. The maximum cooling power of 288 mW at 4.2 K and the minimum temperature of 2.4
K without heat load have been obtained so far.
For GM refrigerators, a preliminary test is under going. One of test results showed that the
Figure 9. Temperature dependence of cooling power for R=0 and R=0.25.

effect of the GAP was not outstanding as seen in the above test, but an improvement in cooling
performance was observed clearly below 4 K.
More details of the experimental procedure and the analysis of the refrigeration test are given
in another paper in this same proceedings.7
CONCLUSIONS
This paper describes our approach to using oxide magnetic materials as regenerator materi-
als. In particular, GAP is found to have a very high heat capacity below 4 K, and the test results
show that it remarkably improves the cooling performance of a pulse tube refrigerator below
4.2 K. However, several problems must be overcome before it can be used as a practical regenera-
tor material. Currently known advantages and disadvantages of the oxide regenerator material
have been described, and include:
- Generally, the oxide magnetic material has the potential to provide high heat capacity at
lower temperatures, especially below 4 K.
- Thermal conductivity is higher than previous regenerator materials.
- Because it is an oxide, it is stable in air.
- Using the polycrystalline form of the material will contribute to reduced cost and ease of
fabrication into various shapes.
- Fine particles can be fabricated easily by using a chemical process.
- Although the hardness and the smoothness of the GAP granules have not been tested pre-
cisely, no broken powders were found during our cooling test with a pulse rube refrigerator;
however, some amounts of broken powders were recently reported hi the case of a GM
refrigerator.
Since there is a wide variety of cryocoolers, the expected improvement in cooling perfor-
mance from using GAP will depend on their structures. Thus, we are trying to test the material
with the help of a variety cryocooler manufacturers, and to correct identified problems. A new
method to improve the quality of the granules is under development. A Vickers hardness of ~ 900
and a relative density of 99% should be obtainable.
An interesting possibility is the use of GAP for providing lower temperatures, below 2 K,
with 3He gas. GAP appears the most likely to contribute to increased cooling capacity below 2 K.
Also some other oxides may prove to be promising at higher or lower temperatures.
ACKNOWLEDGMENTS
One of the authors (Numazawa) would like to thank the late Prof. Dr. Christoph Heiden of the
University of Giessen, who strongly recommended the study of oxide regenerator materials at low
temperatures. We would like to dedicate this paper to the memory of Prof. Dr. Heiden.
REFERENCES
1. T. Hashimoto et al., “Recent Progress on Rare Earth Magnetic Regenerator Material,” Adv. Cryog.
Eng. 37(1992), p. 859.
2. T. Numazawa, M. Okamura, O. Arai, and A. Sato, “Magnetic Regenerator Materials for Sub-2K
Refrigeration,” Adv. Cryog. Eng. 46 (1999), p. 421.
3. D. Cashion, et al., “Magnetic Properties of Antiferromagnetic GdAlO3,” J. Appl. Phys. 39(1968),
p. 1360.
4. L.H. Holmes, et al., “Magnetic Behavior of Metamagnetic DyAlO3,” Phys. Rev. B5 (1972), p. 138.
5. M. Ogawa, R. Li and T. Hashimoto, “Thermal Conductivities of Magnetic Intermetallic Compounds
for Cryogenic Regenerator,” Cryogenics, 31 (1991), p. 405.
6. S. Fujimoto, Y.M. Kang, T. Kanayama and Y. Matsubara, “Experimental Investigation of Some
Phase Shifting Types on Two-stage GM Pulse Tube Cryocooler,” Adv. Cryog. Eng., 46 (2000).
7. S. Fujimoto, T. Kurihara, T. Oodo, Y.M. Rang, T. Numazawa and Y. Matsubara, “Experimental
Study of 4K Pulse Tube Cryocooler,” Cryocoolers 11, Plenum Publishers, New York (2001).
New Regenerator Materials for Use in
Pulse Tube Coolers
A. Kashani and B.P.M. Helvensteijn

Atlas Scientific
San Jose, CA, USA 95120
P. Kittel
NASA-ARC
Moffett Field, CA, USA 94035
K.A. Gschneidner, Jr.,V.K. Pecharsky and A.O. Pecharsky

Ames Laboratoy
Ames,IA, USA 50011
ABSTRACT
A two-stage pulse tube cooler driven by a linear compressor is being developed to provide
cooling at 20 K. The first stage of the cooler will have the conventional stainless steel screen
regenerator matrix. The matrix for the second stage regenerator (<60 K) will be made from a new
class of Er based alloys which was recently developed at Ames Laboratory, in Ames, Iowa. These
alloys exhibit heat capacities that exceed that of all other materials, including lead, over a wide
range in temperature (15 K < T < 85 K). The performance of one such alloy was shown to be better
than lead when tested in a single-stage pulse tube cooler driven by a G-M compressor and operating
at 2 Hz. An effort is underway to establish their suitability at frequencies above 40 Hz. An ap-
proach to testing these alloys at low temperatures while using a low-power linear compressor is
presented.
INTRODUCTION
Future NASA missions will involve transportation and/or storage of liquid hydrogen tanks for
use as propellant. If no precautions are taken substantial amounts of liquid may be lost by evapo-
ration due to inevitable parasitic heat loads, well before cryogen utilization. Cryocoolers may
intercept parasitics, thereby, achieving zero boil off of the cryogen. This is necessary if the mission
design requires a small tank volume or if the mission duration is sufficiently long.1 Several such
missions are discussed in Ref.l, including a mission that realizes on-Mars propellant production
and made from seed and atmospheric sufficient to power a return vehicle back to
earth. A represantative cooling requirement for the seed-hydrogen tank to be transported to Mars is
2 W at 20 K plus 10 W at 60 K for cooling the heat shield.

For this and other similar applications requiring a cryocooler with high reliability and low
vibration a pulse tube cooler (PTC) driven by a linear compressor is an attractive choice. The
absence of cold moving parts in a PTC reduces wear and vibration and increases reliability. Conse-
quently, the PTC is becoming the cryogenic cooler of choice on-board spacecraft.
Single-stage PTCs, driven by linear compressors and with a stainless steel stacked-screen
regenerator matrix have achieved no-load temperatures below 40 K.2 To provide cooling at lower
temperatures, the PTC requires additional stages. In a multi-stage PTC there can be a large tem-
perature difference between the heat rejection point and the cold tip. For high performance the
regenerator materials need to have high heat capacity over the full temperature span. In general,
this requirement cannot be satisfied using a single material. For lower temperature stages, materials
other than SS are used.
Multi-stage PTCs have achieved temperatures below 4 K using low frequency G-M compres-
sors.3 However, long-term reliability, compactness and mass favors the high frequency linear com-
pressor over the G-M compressor.
To date no multi-stage PTC for use in space has yet been developed to provide cooling at
temperatures below 40 K while rejecting heat at 300 K. A two-stage PTC is being designed to
provide 10 W of cooling at 60 K and 2 W at 20 K. The matrix for the 1st stage regenerator will be
400 mesh SS screen. The matrix for the 2nd stage regenerator (<60 K) will be made from a new class
of Er alloys which has recently been developed at Ames Laboratory, in Ames, Iowa.4 The heat
capacity of these alloys exceeds that of all other materials, including lead, over a wide range in
temperature (15 K < T < 85 K). The heat capacities of these and other regenerator materials are
shown in Fig. 1. Another pertinent advantage of these alloys over lead is that they do not oxidize
readily. Also, unlike lead, they are strong enough not to lose their sphericity over time when used
as a spherical powder in a regenerator.
In-house experimentation on a GM-driven single-stage PTC validated Er as
a suitable low temperature regenerator material. In that study the regenerator consisted of two
sections. The first section had a stacked SS screen matrix; while, the second section was packed
Figure 1. Volumetric heat capacity of various materials.

NEW REGENERATOR MATERIALS FOR USE IN PT COOLERS 477
Figure 2. Rolled-foil regenerator matrix.
with either Er or lead powder. With Er in the second section, the cooler achieved a lower
no-load temperature.
Next to regenerator materials selection, the configuration given to these materials affects the
performance of the PTC. One of the more significant regenerator parameters to be maximized is the
ratio of gas-solid heat transfer to pressure drop across the regenerator. This ratio is highest for the
parallel-plate regenerator configuration.5 Implementing this concept with the Er alloys will be
explored. With proper heat treatment, a thick foil of Er has been formed. The foil can be
rolled around a low thermal conductivity mandrel, i.e., G10 (see Fig. 2). The spacing between foil
layers will be maintained by inscribing lateral ridges in the foil. Uniform ridges were formed by
Concurrent Technology on a sample Er foil. The length of each rolled-foil stack will be a fraction of
the overall regenerator length requiring several of these stacks in the regenerator. A small gap will
be maintained between adjacent stacks, thereby, enabling redistribution of the flow and minimizing
mal-distribution of the flow between neighboring flow passages.
An alternative to the parallel-plate regenerator configuration is the more traditional packed-
sphere configuration. It is appropriate that this configuration be also pursued in case the parallel-
plate configuration does not live up to its theoretical prediction. Since the proposed cooler is to be
operated at high frequencies, the particle sizes required tend to be smaller than those
used in the low frequency G-M type. It has been demonstrated that spherical powders of the Er
alloys can be manufactured even in small sizes.
Other than improving the PTC by optimizing the regenerators, performance will be enhanced
by inserting in each stage an inertance tube between the pulse tube and the respective reservoir.
Since it does not permit DC flow the inertance tube is preferred over adding a bypass. The cooler
configuration will be optimized using the PTC modeling program Sage.
MATERIAL TESTING
An effort is underway to establish the suitability of the new Er alloys at frequencies above
40 Hz using a linear compressor. To test these regenerator materials a single stage pulse tube cooler
using an inertance tube has been fabricated. The cooler is driven by a 350 W linear compressor that
can operate from 36 to 72 Hz. In order to achieve a no-load temperature of below 20 K in a single-
stage PTC, driven by this low-power compressor, liquid nitrogen is incorporated in the design to
Figure 3. Schematic of the regenerator test apparatus.
enhance the cooling power. To this end, the regenerator is made up of three sections. The matrix
material for the first two sections is 400 mesh SS screen. A heat exchanger, externally cooled by
liquid nitrogen, is inserted between these two sections. The heat exchanger is packed with 80 mesh
copper screens.
The third section which is attached directly to the second section is packed with spherical
powder of the material under test. Below this section reside the cold heat exchanger and the pulse
tube. The warm end of the pulse tube, the inertance tube and the reservoir are all maintained at
liquid nitrogen temperatures. Thus, the pulse tube warm heat exchanger is also cooled externally
with LN2.
Liquid nitrogen is supplied to the heat exchangers from a LN2 can which shares the same
vacuum enclosure with the pulse tube cold head. The inertance tube is coiled around the LN2 can.
The reservoir is made of copper and is integrated with the bottom of the LN2 can. A schemtaic and
a photograph of the test setup arc shown in Figs. 3 and 4, respectively. The applicability of the new
regenerator materials will be evaluated by measuring the performance of the test cooler. During
testing the pressure will be measured at the outlet of the compressor, at the hot end of the pulse
tubes and in the reservoirs using piezo-electric pressure transducers. From the compressor pressure
and its displacement the P-V work of the compressor is calculated. The temperature at the inlets
and the outlets of the aftercooler and the heat exchangers will be measured by thermocouples to
assure their effectiveness. The temperature of the cold heat exchanger will be measured using a
silicon diode thermometer. A heater will be placed on the cold heat exchanger to enable testing the
cooling capacity of the cooler. The experimental parameters derived from the pressure and tem-
perature data will be compared to the results of the model predictions in order to assess the applica-
bility of the regenerator material.
NEW REGENERATOR MATERIALS FOR USE IN PT COOLERS 479
Figure 4. Photograph of the regenerator test apparatus.
MODELING
The test PTC cold head described above was designed by employing the pulse tube cooler
modeling program Sage. This program allows optimization of the various cooler components. It
also accounts for the non-ideal behavior of the working fluid (helium), which is important for a
realistic design particularly at low temperatures. The modeling effort optimized the dimensions of
the pulse tube, the three regenerator sections and the inertance tube (see Tables 1 and 2).
SUMMARY
A new class of Er based alloys has been recently developed at Ames Laboratory, in Ames,
Iowa that can be used in the regenerators of multi-stage pulse tube coolers. These alloys exhibit
heat capacities that exceed that of all other materials, including lead, over a wide range in tempera-
ture (15 K < T < 85 K). A test apparatus has been constructed to validate the performance of these
alloys at frequencies above 40 Hz. It consists of a pulse tube cooler with a regenerator that has three
sections. Compensating for the low-power of the compressor the cooler design utilizes liquid nitro-
gen to achieve a no-load temperature below 20 K. The modeling program Sage has been employed
to provide the optimum paramteres for the test cooler. Initial check out of the test apparatus is
currently underway.
REFERENCES
1. Kittel, P., and Plachta, D.W., “Propellant Preservation for Mars Missions,” Adv. in CryogenicEngin.,
Plenum Press, NY, 2000.
2. Burt, W.W., Chan, C.K. “New Mid-Size High Efficiency Pulse Tube Coolers,” Cryocoolers 9,
Plenum Press, New York, 1997, p. 173.
3. Wang, C., Thummes, G., and Heiden, C., “Performance Study on a Two-Stage 4K Pulse Tube
Cooler,” Adv. in Cryogenic Engin., Plenum Press, NY, 1998.
4. Gschneidner, Jr., K.A., Pecharsky, A.O., Pecharsky, V.K., “Ductile, High Heat Capacity, Magnetic
Regenerator Alloys for the 10 to 80 K Temperature Range,” Cryocoolers 11, Plenum Press, NY,
2001.
5. Radebaugh, R. and Louie, B., “A Simple First Step to the Optimization of Regenerator Geometry,”
NBS Special Publication 698 (1985), p. 177.
Advanced Developments for Low Temperature
Turbo-Brayton Cryocoolers
J. A. McCormick1, G.F. Nellis1, H. Sixsmith1, M. V. Zagarola1,

J. A. Gibbon2, M. G. Izenson1, and W.L. Swift1
1
Creare Incorporated
Hanover, NH, 03755, USA
2
NASA Goddard Space Flight Center
Greenbelt, MD, 20771, USA
ABSTRACT
Turbo-Brayton cryocooler technology that has been space qualified and demonstrated on the
Near Infrared Camera/Multi-Object Spectrometer (NICMOS) cryocooler is being adapted for
applications with lower cooling loads at lower temperatures. The applications include sensor
cooling for space platforms and telescopes at temperatures between 4 K and 35 K, where long
life and reliable, vibration-free operation are important. This paper presents recent advances in
the miniaturization of components that are critical to these systems. Key issues addressed in
adapting the NICMOS cryocooler (NCC) technology to lower temperatures involve reducing
parasitic losses when scaling to smaller size machines. Recent advances include improvements
in the efficiency of a small, permanent magnet driven compressor that operates at up to
10,000 rev/sec in self-acting gas bearings and the successful demonstration of a 2 mm diameter
shaft operating in pressurized gas bearings.
The compressor is important for cryocoolers with input powers between 50 W and 100 W.
The miniature shaft and bearing system has applications in compressors and turbines at
temperatures from 300 K to 6 K. These two technology milestones are fundamental to achieving
exceptional thermodynamic performance in low temperature turbo-Brayton systems. This paper
discusses the development of these components and test results, and presents the implications of
their performance on cryocooler systems.
INTRODUCTION
In 1998 a single-stage turbo-Brayton cryocooler with an integral cryogenic circulating loop
was space qualified during the Hubble Orbital System Test (HOST). The test was conducted
during an 11 day mission on STS-95 in preparation for integration of the cooler with NICMOS
on the Hubble Space Telescope (HST).1 The NICMOS cryocooler is sized to provide about 8 W
of cooling at 70 K. Much of the cooling is used to absorb parasitic losses. Approximately 0.4 W
of refrigeration is needed for the detectors on the infrared (IR) instrument.
Future space missions are being planned in which cooling loads for instruments in space are
expected to be at significantly lower temperatures. As an example, NASA is presently
developing the Next Generation Space Telescope (NGST) in which the detectors for the mid-IR
instruments will require ten’s of mW at temperatures of 6 K – 8 K. Additional loads from

482 TURBO-BRAYTON CRYOCOOLER DEVELOPMENTS
detector electronics and parasitics may increase this refrigeration requirement to the 40 mW –
100 mW range. Other NASA programs such as Constellation X and the Terrestial Planet Finder,
and future Defense Department programs indicate that there is a very strong need for a reliable,
long life cryocooler that is capable of supplying low temperature refrigeration over a range of
loads. The turbo-Brayton system is an excellent candidate to meet these needs, especially in
situations where low vibration and high reliability are critical.
There are no fundamental limitations to adapting turbo-Brayton technology to these low
temperature requirements. The key components in the turbo-Brayton cooler are the miniature
turbomachines that use small high-speed rotors in gas bearings and the high performance
recuperative heat exchangers. The turbomachines and recuperative heat exchangers have an
operational history of over 20 years at temperatures as low as 4.5 K in helium refrigerators and
liquefiers that are used for cooling superconducting magnets in particle accelerators.2,3 The
continuing challenge is to reduce the size of these components so that the cycle remains efficient
at the reduced refrigeration capacities associated with low temperature space instruments.
LOW TEMPERATURE CYCLES

There are several turbo-Brayton cycle configurations that can meet low temperature
requirements. Figure 1 schematically shows a basic configuration for producing refrigeration at
temperatures in the 4 K – 20 K range. The cycle uses helium as the working fluid. At the warm
end, one or more compressors pressurize fluid in the loop and reject heat to a heat rejection
interface (a radiator or coolant loop). A warm expansion turbine is used at the cold end of the
warmest recuperator to accommodate the ineffectiveness of this recuperator. It may also provide
additional cooling to an intermediate load or to shielding for the lower stage. The cold
expansion turbine provides cooling at the lowest temperature through a cold load interface. This
turbine accommodates losses in the low temperature recuperator, parasitic heat loads at the cold
end, and the cold load.
Figure 1. Two stage low temperature turbo-Brayton cryocooler.

DEVELOPMENTS FOR TURBO-BRAYTON COOLERS 483
For space applications, centrifugal compressors with gas bearings provide long life,
reliability and low vibration. The machines are, however, inherently low pressure ratio, high
flow devices. The recuperative heat exchangers must therefore have very high thermal
effectiveness to limit the number of compressor stages that are required. Trades between
recuperator mass and number of compressors is an important consideration in arriving at an
optimum cycle.
A variation of the basic two-stage cycle is being considered for the NGST. In the NGST
version, the warm expansion turbine is removed from the cycle and parasitic heat losses from the
warm recuperator are rejected by a low temperature radiator. This is a unique variation that is
practical only if there is sufficient radiator area available at this intermediate temperature to
reject the parasitic heat load. The advantage of the approach is that it eliminates one machine.
Table 1 lists important operating and component characteristics for three versions of a low
temperature cycle. The first two columns list characteristics for a basic two-turbine system
designed to supply refrigeration at 10 K. The cooling loads are 100 mW and 200 mW,
respectively. The third column provides information on the NGST variation. The basic machine
and recuperator sizes for each of the three operating conditions are nearly identical.
The turbomachines for these cycles are smaller than the corresponding machines used in the
NICMOS cryocooler. The compressor from the NICMOS cooler is nominally a 200 W – 400 W
machine. The compressors in the low temperature cycles will operate at about 50 W – 100 W
each. The NICMOS expansion turbine is capable of about 20 W of refrigeration. The turbines
for the low temperature cycle will be designed for about 1 W output. Further reductions in the
size of the turbine would significantly improve performance figures listed in the table. The size
reductions would provide a better match to the reduced flow rates associated with the lower
temperature cycle.
The heat exchangers used in the low temperature cycle must have exceptionally high
thermal effectiveness in order for the cycle to be efficient. At the cold end of the system, the
ineffectiveness of the heat exchanger is a major portion of the net refrigeration available from the
expansion turbine. This is reflected in the figures for input power to the cycle in comparing the
100 mW version with the 200 mW version. Although the net refrigeration available to the load
differs by a factor of two, the input power only differs by about 15%. In a given heat exchanger,
mass approximately scales with the thermal ineffectiveness (1-effectiveness). The recuperator
used in the NICMOS cooler is a counterflow slotted plate design that weighed about 6 kg. It is
90 mm diameter by 560 mm long and has a thermal effectiveness of 0.9923. If the same design
is used in the low temperature cycle, the total heat exchanger mass will approach 30 kg. It is
desirable to reduce the mass of these critical elements with a different design.
The following sections discuss the present efforts directed at developing smaller machines
and lighter recuperators that will meet the unique requirements for the low temperature systems.
COMPRESSORS
The mechanical requirements for the compressor(s) for the low temperature helium cycle are
bounded by existing technology. The NICMOS compressor uses a solid rotor, three-phase
induction motor that is relatively efficient at a 300 W power level but would be much less
efficient at the power levels of the low temperature helium cycle. A compressor drive that is
derived from the NICMOS turboalternator design has been developed. The efficiency has been
increased by adopting a synchronous motor design that uses a rare-earth permanent magnet in a
hollow titanium alloy rotor. In this case, the physical dimensions of the machine are scaled up
from the smaller turboalternator. Early work on this compressor was described by McCormick,
et al.4 The development was directed toward its use in a neon single-stage turbo-Brayton cycle
designed for 1 W refrigeration at 35 K. Under those conditions, the input power to the
compressor was roughly 100 W and the design speed was about 8500 rev/s.
For the low temperature helium cycle, the design speed will be approximately 10,000 rev/s,
and the internal flow passages are optimized for helium at the appropriate pressure and flow rate.
The results from initial tests on a brassboard version of this compressor are presented in
Figure 2. These tests were performed at 10,000 rev/s in helium. The performance is presented in
the form of power train efficiency (PTE) and head coefficient as a function of flow
coefficient Design targets for the low temperature cycles are shown on the curves.
Figure 2. Performance of helium compressor at 60 W input power.

Power train efficiency is the ratio of isentropic compression power to electrical input power.
It includes compressor, motor and electrical losses in the drive electronics. Head coefficient is
the ratio of isentropic enthalpy rise to the square of the impeller tip speed. This is a non-
dimensional means of unifying pressure ratio, speed and inlet temperature conditions to a single
variable. Flow coefficient is the ratio of inlet gas velocity to the impeller tip speed. This
coefficient also unifies information from data at differing temperatures and pressures. The
power train efficiency that has been achieved is about 30% higher in this compressor than the
corresponding operating point in the larger NICMOS compressor.
EXPANSION TURBINE
Achieving the target efficiencies for the low temperature expansion turbine represents one of
the larger challenges to be met. The optimum turbine design from an aerodynamic standpoint
would include a 1 mm diameter turbine rotor operating at about 12,000 rev/s. We have
successfully built turbines using gas bearings that operate at these speeds. However, the bearings
for such machines have always operated at near-ambient temperature. Furthermore, the smallest
machine that has operated at this speed with room temperature bearings was approximately twice
the diameter desired. Reducing the size of rotors while preserving important precision flow
passage features is one of the long-term developments we continually address.
The load carrying capacity of self acting gas bearings decreases as temperatures decrease
and size is reduced. Fluid viscosity decreases with temperature, reducing the capacity of a given
bearing design to carry loads at high speed. This effectively reduces the maximum stable speed
of the machine. Furthermore, as size decreases, the clearances between the shaft and bearings
must also decrease to preserve load carrying capacity. This presents challenges in fabrication
technology. Employing room temperature bearings with a cryogenic turbine is acceptable for
higher capacity turbines, where the conducted heat leak from the bearing housing is tolerable
with respect to the desired output of the turbine. However, for a 1 W turbine at 6 K – 20 K, the
heat leak from bearings at 300 K would exceed acceptable values. A compromise is necessary.
The compromise involves choosing a slightly larger size rotor and an acceptably lower speed that
will still provide enough net refrigeration to meet the cycle requirements without imposing
excessive costs of development.
Turboalternator
Initial efforts toward the development of a suitable low temperature expansion turbine
involved assessing the ability of an existing design to operate at helium temperatures. A 2.4 mm
diameter turboexpander with a maximum operating speed of 10,000 rev/s had been built and
tested at 4.5 K in helium in the late 1980’s by Creare. This machine used an early version of the
self acting gas bearings of the type being used in the turboalternator of the NICMOS cryocooler.
The bearings operated primarily at 300 K but also behaved well down to a temperature of about
100 K. Bearing instabilities prevented reliable operation of the bearings below this temperature.
The early bearing design was modified to take into account the changes in viscosity with
temperature and to provide a better match in thermal coefficients of expansion between housing
parts, the bearing assembly and the shaft. The design was finally optimized and has been
successfully implemented in the turboalternator and circulator of the NICMOS cooler. These
bearing sets run at the same temperature as the cold end of the machine, about 70 K, and have
been used successfully in a neon cycle at 35 K. Exploratory tests have been performed with a
turboalternator assembly to assess the stability characteristics of this bearing system in helium at
lower temperatures. At 77 K, the bearings were stable to speeds above 3500 rev/s in helium. At
12 K, the lowest temperature at which tests have been performed to date, the shaft and bearings
were stable at a speed of 1000 rev/s. These initial results with the unmodified bearings are very
promising. However, higher speeds must be achieved to meet the requirements of the low
temperature cycles.
High Specific Speed Turboalternator

Because of the penalties in aerodynamic performance associated with operating at speeds of
about 1000 rev/s, an alternate configuration for the low temperature turbine is being developed.
It is identified as the High Specific Speed Turboalternator (HSSTA). It combines several
features of earlier turboexpanders and the NICMOS turboalternator. The two most important
elements of the design are the use of a high specific speed turbine rotor and the operation of the
bearing system at a higher temperature than the rotor. Specific speed is a shape characteristic for
a turbomachine that characterizes the relationships between rotational speed and diameter for a
given set of flow conditions. Higher specific speed machines are generally more efficient and
smaller, mainly because drag losses in internal flow passages are reduced. Figure 3 is a
schematic representation of the HSSTA assembly.
A high specific speed turbine rotor is used for the expansion process at the cold end of the
machine. The advantage of this configuration over the low specific speed design used in the
NICMOS turboalternator is that the rotor diameter is smaller resulting in a slight improvement in
aerodynamic efficiency and reductions in drag. In order to take advantage of this rotor
configuration, the operating speed must be in the range of 3500 – 4500 rev/s. This is achieved
by maintaining the bearing housing (the “warm” end of the assembly) at a temperature that
permits stable operation of the shaft at this speed (nominally 60 K – 70 K for these cycles). A
permanent magnet in the shaft forms the rotor of the alternator providing an electrical load for
the turbine to control speed. Most of the heat from bearing drag and alternator losses are rejected
at the intermediate stage temperature. A small portion of heat conducts back to the cold end of
the turbine through the thermal standoff and the shaft. However, because of the low conductivity
of materials in this temperature range, the penalty is acceptably low. This configuration is quite
similar to other earlier designs of turboexpanders5 except that the warm end of the machine is
now chosen to be significantly lower than 300 K and the brake circuit is replaced by an
alternator. This simplifies the system somewhat.
Figure 3. High Specific Speed Turboalternator.

Advanced 6 K Turboalternator
As a further advance, work is proceeding on significantly smaller turbines and bearing sets.
At present we are developing a turboalternator with a 2 mm diameter turbine rotor on a 1 mm
diameter shaft. Two separate gas bearing support systems are being developed that are intended
to allow the entire machine to operate at the speeds desired in the 6 K - 20 K temperature range.
One bearing system uses a small quantity of the flow from the compressor to supply gas to the
journals by means of slots in the bearing surfaces. The loss associated with the use of a portion
of the refrigerated fluid to energize the bearings must be kept low relative to the net gain in
performance resulting from the higher rotational speed and small rotor. Initial tests on a twice-
scale prototype has demonstrated the feasibility of the design. An alternate bearing system uses
self acting journal bearings in which the bearing hub is connected to an outer housing by means
of a damped structure. The inner bearing hub contains a fixed geometry consisting of a series of
convergent passages similar to the tilt pad.
RECUPERATORS
Our efforts are directed toward reducing the size and mass of the recuperators because they
account for a major portion of the overall cryocooler mass. Figure 4 gives a comparison between
the physical size and mass of two differing designs that could be used for the cold recuperator in
the low temperature cryocooler. The slotted plate heat exchanger derives from a design used in
the NICMOS cryocooler.6 Its length has been increased to raise the thermal effectiveness needed
in the cold stage of the low temperature cooler.
The Radial Flow Heat Exchanger (RFHX) is a far more compact design. It is made up of
modules consisting of plates in which alternating high pressure and low pressure flow passages
are arranged axially. Flow in the alternating passages is in the radial direction. The flow
patterns are schematically shown in Figure 5. The RFHX is a vacuum brazed assembly using
Hastelloy X to reduce mass and enhance heat transfer performance.
The basic design and fabrication methods for the RFHX have been proven. The precision
forming, blanking and brazing technology has been used to produce helium-tight stacks from
stainless steel and sub-scale helium-tight stacks from Hastelloy disks. The thermal effectiveness
and pressure loss data from these sample stacks are in good agreement with thermal performance
models. The next step in the development process is to fabricate, assemble and test a full size
heat exchanger.
Figure 4. Slotted Plate Heat Exchanger (SPHX) and Radial Flow Heat Exchanger (RFHX)
characteristics for the cold stage recuperator.
Figure 5. Flow passages in an RFHX module.
CONCLUSIONS
Significant progress has been made toward developing the components that are necessary
for an efficient low temperature turbo-Brayton cryocooler. A low-power, high efficiency
compressor has been tested at design speed and met target performance figures. Fabrication
trials for a compact light-weight recuperator have been completed on individual modules. Tests
on these modules have confirmed predictive models. The next step is to assemble and test a full
size recuperator. A new cryogenic high specific speed turboalternator design is being used to
increase the speed and efficiency for the cold stage expansion turbine. Supplemental
developments involving new cryogenic gas bearings to further reduce the size and improve the
efficiency of the cold stage offer a promising further reduction in input power to the cycle.
ACKNOWLEDGMENT
This work is supported by NASA Goddard Space Flight Center.
REFERENCES
1. Dolan, F.X., McCormick, J. A., Nellis, G.F., Sixsmith, H., Swift, W.L., Gibbon, J. A., “Flight Test
Results for the NICMOS Cryocooler”, to be published in Advances in Cryogenic Engineering, Vol.
45 (2000).
2. Swift, W.L., Schlafke, A., and Sixsmith, H., “A Small Centrifugal Pump for Circulating Cryogenic
Helium”, Advances in Cryogenic Engg., V27, Plenum Press, New York (1982), pp. 777-784.
3. Sixsmith, H., Hasenbein, R., Valenzuela, J., Theilacker, J. C., and Fuerst, J., “A Miniature Wet
Turboexpander”, Advances in Cryogenic Engg., V35, Plenum Press, New York (1990), pp. 989-995
4. McCormick, J.A., Nellis, G.F., Swift, W.L., “Design and Test of Low Capacity Reverse Brayton
Cryocooler for Refrigeration at 35 K and 60 K”, Cryocoolers 10, Kluwer Academic/Plenum
Publishers, New York (1999), pp. 421-429.
5. Swift, W.L. and Sixsmith, H., “Performance of a Long Life Reverse Brayton Cryocooler”,
Cryocoolers 7, Phillips Laboratory Air Force Materiel Command, New Mexico (1992), pp. 84-97.
6. Nellis, G.F., Dolan, F.X., McCormick, J.A., Swift, W.L., Sixsmith, H., “Reverse Brayton Cryocooler
for NICMOS”, Cryocoolers 10, Kluwer Academic/Plenum Publishers, New York (1999), pp. 431-
438.
Life and Reliability Characteristics of
Turbo-Brayton Coolers
J.J. Breedlove1, M.V. Zagarola1, G.F. Nellis1,

J.A. Gibbon2, F.X. Dolan1, and W.L. Swift1
1
Creare Incorporated
Hanover, NH 03755
2
NASA Goddard Space Flight Center
Greenbelt, MD 20771
ABSTRACT
Wear and internal contamination are two of the primary factors that influence reliable,
long-life operation of turbo-Brayton cryocoolers. This paper describes tests that have been
conducted and methods that have been developed for turbo-Brayton components and systems to
assure reliable operation.
The turbomachines used in these coolers employ self-acting gas bearings to support the
miniature high-speed shafts, thus providing vibration-free operation. Because the bearings are
self-acting, rubbing contact occurs during start-up and shutdown of the machines. Bearings and
shafts are designed to endure multiple stop/start cycles without producing particles or surface
features that would impair the proper operation of the machines. Test results are presented that
document extended operating life and start/stop cycling behavior for machines over a range of
time and temperature scales.
Contaminants such as moisture and other residual gas impurities can be a source of
degraded operation if they freeze out in sufficient quantities to block flow passages or if they
mechanically affect the operation of the machines. A post-fabrication bakeout procedure has
been successfully used to reduce residual internal volatile contamination to acceptable levels in
closed-cycle systems. The process was developed during space qualification tests on the Near
Infrared Camera/Multi-Object Spectrometer (NICMOS) Cryogenic Cooler (NCC). Moisture
levels were sampled over a seven-month time interval following the initial bakeout confirming
the effectiveness of the technique. A description of the bakeout procedure is presented.
INTRODUCTION
Turbo-Brayton cryocoolers are being developed for a variety of space applications. The
primary components in these systems include recuperative counterflow heat exchangers and
miniature high-speed turbomachines. The heat exchangers are all-metal vacuum brazed
assemblies, designed to provide effective heat transfer with minimal pressure drop. They are
totally passive components with a substantial amount of internal surface area to enhance heat
transfer. The turbomachines are small assemblies consisting of very low mass shafts operating at
high rotational speeds in gas bearings. The devices are designed to operate over a broad range of
temperatures from about 330 K down to 4 K. The materials of construction and the design
features must be compatible with these requirements. Their proper implementation provides for
vibration free operation, very long life and high reliability.
As the development of key components for turbo-Brayton coolers has progressed, parallel
efforts have been underway to identify and quantify important factors that affect the reliability
and life of these systems. The recent development of a turbo-Brayton cooler for integration with
NICMOS on the Hubble Space Telescope (HST) has provided a valuable opportunity to evaluate
the most critical issues and to implement methods that assure extended life.
Figure 1 is a schematic representation of the NICMOS Cooling System (NCS). It consists
of three distinct fluid loops: a cryogenic circulator loop, a turbo-Brayton cryocooler, and a
capillary pumped loop (CPL). The circulator loop circulates fluid between the NICMOS
detectors and the cold load interface (heat exchanger) of a single-stage reverse-Brayton
cryocooler. The circulator is a centrifugal machine driven by a permanent magnet motor. The
circulator loop is an “open” loop in the sense that a portion of it is connected to the NICMOS
dewar through bayonets in space. During ground tests it is connected to and disconnected from
several different devices that simulate the properties of the NICMOS dewar. The cryocooler is a
closed hermetic loop consisting of a centrifugal compressor, a recuperative heat exchanger, a
turboalternator, and cold and warm thermal interfaces. Heat from the system is conveyed to
radiators by a CPL. Drive electronics provide control for the system including the operating
frequency for the circulator and compressor and a resistive load to modulate turbine speed. In
order to qualify this system for the 5+ years of expected operation in space, key reliability issues
and degradation mechanisms were identified and resolved.
Figure 1. NICMOS Cooling System.

LIFE AND RELIABILITY OF TURBO-BRAYTON COOLERS 491
KEY TECHNOLOGY ELEMENTS AND RELIABILITY ISSUES

Prior experience with the development of cryogenic turbomachines has verified that the
presence of contamination is the primary source of potential problems with these systems. In
particular, the freeze-out of undesired constituents in the gas stream at the coldest portions of the
flow loop can cause blockage of flow passages, reduce heat transfer, produce mechanical
unbalance on rotating parts, and cause damage to the bearing systems. Careful cleaning
procedures during fabrication and assembly are critical to the overall integrity of the system.
However, additional steps are required to assure that the system remains contamination free after
it has been assembled and sealed. Mechanical filters are effective at trapping solids at the inlet to
machines. This can effectively eliminate wear and damage caused by debris. But filters are not
adequate protection against freeze-out in the machines.
In most previous versions of the turbo-Brayton cooler, a turboexpander has been used at the
cold end of the system. In a turboexpander, the turbine rotor operates at the required cryogenic
temperature. The bearings that support the shaft operate at room temperature. If a sufficient
quantity of moisture or other condensable contaminants are present in the system, the
contaminant may freeze out in the cold portions of the system. The accumulation of frozen
solids in the cold end of the turbine can manifest itself as a gradual decrease in the flow area at
the exit of the turbine rotor. This can produce a reduction in flow, degrading the net refrigeration
as a result. Excessive amounts of contaminants could stop the machine. However, because the
bearings operate at warmer temperatures, they are relatively unaffected by the freeze out.
In the NICMOS system, the bearings in the circulator and the turboalternator operate at
cryogenic temperatures. The journal bearings and thrust bearings rely on close clearances to
provide stable operation of the shaft at high speed. Furthermore, during some periods of system
operation, the machines are in the off state while at cryogenic temperatures. They must reliably
Start under these conditions. If frozen solids accumulate in the close clearances between the
rotors and stationary housing components, the operation of the machine will be significantly
affected. At high speeds, rubbing contact between the rotors and the stationary surrounding parts
may cause damage. In general, the damage from a high-speed rub in these areas will make the
machine inoperable. Techniques were established to determine the maximum allowable
moisture levels and to assure that the contaminants were removed after final system closure.
A second potential factor in long term reliability concerns contact between the shafts and
bearings during start up and shutdown. In the operation of the turbomachines, they are subjected
to start and stop cycles. The circulator and compressor are motor-driven centrifugal machines.
The turbine is driven by the pressure difference that is produced by the compressor. Each of the
machines uses self-acting tilt-pad bearings for radial shaft support and spiral groove thrust
bearings for axial support. Rubbing contact occurs between the rotating and the stationary
elements of this support system briefly during every start and shutdown. If the contact occurs at
sufficiently high surface velocity, the mating surfaces may become worn, reducing the
effectiveness of the bearings to support the shaft. If the wear is excessive, the machine may
become inoperable. During normal operation of the gas bearings, there is no contact, and
because of the precision employed in fabricating these parts, there is no vibration. It is only
during start/stop cycles that wear becomes an issue.
Several tests have been performed during the past 16 years at Creare to demonstrate the
robustness of the basic tilt pad bearing designs that are used in turbo-Brayton cooler components.
In one test, a 3.18 mm diameter shaft has operated for just over 15 years, running continuously at
660,000 rpm. The machine operates at room temperature using filtered air to drive the rotor.
During this time period, it has never been disassembled, cleaned or adjusted. For the first
10 years of operation, 5 stop/start cycles were conducted each week, accumulating a total of
2600 cycles during the period. During each cycle, the pressure required to lift the journal off the
pads was consistently within expected bounds, indicating qualitatively that there was no
significant wear. During each stop cycle, the surface velocity of the shaft at the point of touch
down on the pads was recorded. This figure was consistently between 1–2 m/s, well below
values that might cause wear. In a second but related test, a 3.6 mm diameter shaft operating in
room temperature dry helium was cycled through 10,000 stop/starts over a period of one week.
Comparisons between lift off pressure and touch down speed throughout the interval showed that
there was virtually no change in behavior. An inspection of the parts confirmed that there had
been no wear.
These tests provided confidence in the fundamental robustness of the tilt pad bearings at
ambient temperatures in clean, dry environments. However, two of the NICMOS machines
operate with their bearings at cryogenic temperatures. Methods for the NICMOS system were
implemented to assure that the flow loops were free of excessive moisture. And because of
concern about the potential for frozen contaminants in the circulator loop, additional tests
combining the effects of frozen contaminants with multiple stop/start cycles were conducted to
assess the tolerance of the circulator to frozen contaminants.
CONTAMINANTS
Contaminants that potentially affect the performance or reliable operation of the system
include constituents in the gas that are likely to freeze out in cold portions of the system and
solid particles of a size that may cause flow blockage or damage in the machines. Rigorous
cleaning and fabrication procedures combined with the use of filters in a circulating flow
effectively controls solid particles. More elaborate steps must be taken to assure that moisture
and other volatile contaminants are removed from the system. Cyclic purge and fill operations,
in which a system is filled with pure gas then evacuated, are effective if a sufficient number of
cycles are performed at high enough temperatures. This approach is particularly effective if the
internal surface area is low and there are no significant restrictions to flow. For systems with
large internal areas or restrictive flow passages, a circulating flow bakeout is particularly
efficient.
There is a significant amount of internal surface area in a turbo-Brayton cryocooler. In the
NICMOS cooler, the closed loop portion has approximately of internal area. Most of it is
in the recuperative heat exchanger. This component is thoroughly cleaned during the late stages
of the vacuum braze operation. However, it is highly attractive to moisture and other
constituents of air during various assembly processes that follow the vacuum braze. In order to
estimate the requirements for a thorough bakeout of the system, we assume that all internal
surfaces will adsorb about 100 molecular layers of moisture. For the NICMOS cooler, this is
roughly equivalent to 205 mg of moisture. A conservative estimate of additional moisture that is
assumed to be present in the few organics in the system accounts for about 10 mg. Other sources
of contamination, such as impurities in the charge gas and moisture trapped in virtual leaks were
significantly below the levels identified above. Table 1 lists the estimated moisture distribution
in the NICMOS cryocooler loop following its assembly. The focus of a thorough removal of
moisture was to establish a process capable of driving ~ 215 mg of moisture from the surfaces in
the system into gas and removing the moisture from the gas.
CIRCULATING BAKEOUT
A post-fabrication circulating bakeout procedure was devised to assure that condensable
contaminants were reduced to acceptable levels. Two criteria were used to assess the acceptable
residual contamination levels. Tests were performed at cryogenic temperatures in which
controlled moisture concentrations were introduced into the flow through a cryogenic
turbomachine. This qualitatively set upper limits on the tolerance of the machines to frozen
contaminants. A second criterion involved estimating the volume of frozen contaminants that
could be reasonably tolerated in the machine’s flow passages without causing anomalous
operation. This figure was then used to determine the maximum residual moisture that should be
tolerated for long-term operation.
The limiting factor in removing moisture from the system is the rate of desorption of water
molecules from internal surfaces. The exponential time constant for the water removal rate is
proportional to exp(-1/T). The total time required to achieve the desired residual moisture is
very sensitive to the temperature T at which the bakeout is conducted. It is important to maintain
all surfaces as close to uniform temperature as is practical because moisture that is desorbed
from a warm surface may readsorb onto cooler surfaces within the system.
Using the assumed mass quantity of moisture in the system listed in the table above and the
methods of Dayton2 to estimate rates of desorption based on temperature, a procedure was
developed to perform a circulating bakeout at elevated temperatures. The preliminary estimates
implied that a bakeout of less than 100 hours duration at 70°C would suffice to reduce moisture
levels to acceptable values.
The bakeout consisted of maintaining internal surfaces of the cryocooler at a temperature of
70°C while the compressor circulated neon through the loop. The cryocooler was installed in a
vacuum chamber that was filled to a pressure of 600 torr with clean nitrogen. The temperature of
the chamber and the thermal interfaces with the cooler were controlled to maintain internal
surfaces of the system at 70°C. At the nominal compressor flow rate, this means that the entire
internal volume of the cryocooler would sweep through the compressor every 4 seconds. The
bound moisture molecules were driven from the heated surfaces into the gas. A portion of the
flow was bypassed out of the cryocooler loop through a cryogenic charcoal adsorber where the
moisture was removed from the gas. Figure 2 is a schematic flow diagram of the bakeout
hardware and the cryocooler loop.
The status of the bakeout was determined by continuously monitoring the water removal
rate from the system. In order to determine the water removal rate, the mass flow rate, pressure,
and water content of the cleanup gas were measured at the inlet and the outlet. The water
content of the gas was determined by using moisture sensors to measure the water dew point
temperature of the gas. These dew point temperatures were then converted to water partial
pressures. The water removal rate was then determined using Equation (1). In this equation,
MW is molecular weight, P is pressure, and is mass flow rate through the adsorber.
The instantaneous water removal rate was calculated using Equation (1) and was
recorded at regular time intervals. A least squares method was then used to determine the values
of A and in Equation (2) that generate the best curve fit for the measured data. In Equation (2),
t is time. A similar bakeout characterization method was employed by Yuan, et al.3 Equation (3)
was then used to estimate the total amount of water that remained inside the system at time
Figure 2. Cryocooler loop arrangement with gas cleanup system.
The cryocooler system was baked out on several separate occasions following integration of
the system at Goddard Space Flight Center. In addition to complete bakeouts, moisture content
measurements were periodically made to assess the rate of outgassing that might be present in
the system. The results show that the bakeout procedure was highly effective. Over a six-month
period during which the cooler was qualified on STS-954, there was no detectable change in the
moisture content of the fluid in the loop.
Measurements of water dew point and calculated moisture removal rates for a bakeout
performed in July 1998 are shown in Figure 3. The continuous flow of gas through the adsorber
was interrupted three times during the bakeout. Each interruption lasted approximately 8 hours
and was performed to determine the non-circulating desorption rates. During these intervals, the
concentration of moisture in the neon loop increased, which caused the rate of desorption from
the surfaces into the gas to decrease. At the completion of the bakeout, the measured moisture
level in the circulating gas was approximately 0.0 ppmv and the calculated residual moisture in
the system was 0.4 mg adsorbed on the internal surfaces. Two additional measurements of
moisture level in the cryocooler were performed. The first took place 3 weeks after the bakeout,
just prior to the qualification flight on STS-95. The second measurement was made 7 months
after the bakeout, following the STS-95 flight. The measured moisture content in the gas
remained at 0.2 ppmv for both measurements.
The measurements of moisture content following the STS-95 flight confirmed that the
moisture levels had been reduced to acceptable levels in the closed cryocooler loop. Since this
portion of the NICMOS cooling system is hermetically sealed, no additional bakeout or cleaning
is required. However, the circulating loop contains bayonet connections that will not be sealed
during all operations prior to integrating the system with the NICMOS dewar in space. During
several of the qualification ground tests, the bayonets will be connected to and disconnected
from various flow loops that simulate the behavior of the tubing in the NICMOS dewar. In order
to reduce the risk of damaging the cryogenic circulator during ground tests, a bakeout procedure
similar to the one employed in the closed loop portion of the cooling system was implemented.
Temperature limitations in some of the instrumentation in the circulator loop required that the
bakeout be performed at 50°C requiring about 570 hours to reduce moisture levels to acceptable
values. Moisture measurements during the bakeout of the circulator loop concluded that the
residual moisture content in the loop was 0.8 mg at completion.
Figure 3. Dew point measurements and calculated moisture removal rates in the NICMOS.
cryocooler.
CONTAMINATION SENSITIVITY TESTS

Contamination sensitivity tests were performed on the cryogenic circulator to assess the
effects of condensable contaminants on the operation of the machine at cryogenic temperatures.
Moisture was introduced into the neon flow stream in incrementally increasing concentrations up
to a maximum of 1000 ppmv under prototypical operating conditions for the circulator. Results
show that the operation of the circulator was unaffected for moisture concentrations of less than
200 ppmv. At concentrations above 200 ppmv there was evidence of some accumulation of
frozen material in the circulator that produced erratic signals on the shaft displacement sensors.
Post-test inspection of the internal components showed that there was no permanent damage or
wear that would inhibit proper operation of the machine in non-contaminating environments.
Figure 4 is a schematic of the test facility used to assess moisture vulnerability. High purity
(99.999%) neon enters the facility at regulated pressure through a liquid nitrogen cooled
adsorber. The neon flow stream splits into two paths. A small fraction of the flow passes
through a moisture source tube and then rejoins the main flow stream immediately downstream
of the moisture source outlet. The combined stream, with moisture, flows through the circulator.
After leaving the circulator, the gas flows through a heat exchanger where it is warmed to
ambient temperature before exiting the loop through a mass flow controller and an atmospheric
vent valve.
The gas stream is cooled as it flows through tubing submerged in liquid nitrogen. The
dewar containing the liquid nitrogen can be raised and lowered to change the level of the liquid
nitrogen relative to the circulator. Moisture sensors, temperature sensors, pressure transducers,
and a mass flow controller are used to provide information about conditions in the facility.
Circulator measurements include current, voltage and shaft speed and displacement. Shaft
displacement sensors provide a sensitive indication of anomalous behavior of the rotating shaft.
The testing consisted of several operating regimes over a range of moisture levels following
an initial loop evacuation. Each test condition involved an operating period followed by a “cold
soak” interval during which the machine was not operating. At the conclusion of each cold soak,
the machine was operated through 10 successive cold start/stop cycles. Table 2 lists the key
conditions in the test matrix. Figure 5 provides typical results from these tests.
Figure 4. Schematic of the Circulator Moisture Test Facility.
Moisture levels at the inlet and exit sensing locations upstream and downstream of the
circulator are plotted in Figure 5 along with calculated estimates of the accumulated moisture
between sensing locations during a nine-hour test. The resolution of the time axis is not
sufficiently fine to illustrate details of the 10 stop/start cycles at the conclusion of each cold soak.
However, current and voltage measurements during the stop/starts showed that there was no
change in behavior during these cycles, independent of the moisture concentrations present.
These tests were useful confirmation regarding the robustness of the machines and provided
guidance in establishing final allowable moisture levels for the circulator loop. Separate
qualifying tests were used to fully demonstrate start stop capabilities at cryogenic temperatures.
A series of 2,000 cryogenic stop/start cycles were performed on a comparable circulator
assembly. Pre- and post-test inspections verified that there was no wear in the bearings and shaft
under these conditions.
Figure 5. Moisture vulnerability test profile.
CONCLUSIONS
Key factors in determining life and reliability of turbo-Brayton cryocoolers have been
discussed. Extensive life tests with gas bearings used in these systems have demonstrated that
they will operate without problems over long periods if contamination levels are kept low. This
behavior has been confirmed over a range of temperatures from ambient to 70 K. A circulating
bakeout procedure to remove contaminants from a closed system following its assembly has
been presented. The results of sampling the moisture content in the NICMOS cooler over a
7-month interval has demonstrated that the technique is effective at maintaining acceptable levels
of system purity.
ACKNOWLEDGMENT
This work was supported by the Hubble Space Telescope project at NASA Goddard Space
Flight Center.
REFERENCES
1. Nellis, G.F., Dolan, F.X., McCormick, J. A., Swift, W.L., Sixsmith, H., “Reverse Brayton Cryocooler
for NICMOS”, Cryocoolers 10, Kluwer Academic/Plenum Publishers, New York (1999), pp. 431-
438.
2. Dayton, B.B., Outgassing Rate of Contaminated Metal Surfaces, Trans. 8th National Vacuum
Symposium and 2nd International Congress on Vacuum Science and Technology, Washington, D.C.;
October, 1961.
3. Yuan, S.W.K., Kuo, D. T., and Loc, A. S., “Cryocooler Contamination Study”, to be published in
Advances in Cryogenic Engineering, Vol. 45 (2000).
4. Dolan, F.X., McCormick, J. A., Nellis, G.F., Sixsmith, H., Swift, W.L., Gibbon, J. A., “Flight Test
Results for the NICMOS Cryocooler”, to be published in Advances in Cryogenic Engineering, Vol.
45 (2000).
A Flexible Turbo-Brayton Cryocooler Model
P. L. Whitehouse, G. F. Nellis* and M. V. Zagarola*

NASA, Goddard Space Flight Center
Greenbelt, MD, USA 20771
*Creare, Inc., Hanover, NH, 03775
ABSTRACT
Cycle modeling of regenerative cryocoolers has received much attention in the literature over
the past decade or so, due in part to the complexity of the gas flow. While having received less
attention, recuperative cycle cryocoolers are somewhat easier to model in that the continuous gas
flow allows one to avoid the time discretization required by oscillatory flows and look directly at
the steady-state behavior. As the maturation of its technology nears, the tune is right to model
various configurations of a turbo-Brayton cryocooler that may be applicable to programs such as
the Next Generation Space Telescope (NGST) and Constellation-X. As the design parameters
codify for these two programs, a flexible turbo-Brayton cryocooler model is required to evaluate
the passive/active cooling trades and will certainly be needed in any follow-on design phase.
Presented here is a component-based cycle model written in Mathematica. Separate func-
tions, based on the first and second laws of thermodynamics, are used to model individual physi-
cal components. The functions are combined for a particular system configuration and solved
simultaneously for the state points and mass flow rate. This approach allows for easy mixing and
matching of components for various configurations, two of which are shown here.
DISCUSSION
Mathematica differs markedly from other programming languages due to its symbolic capa-
bilities. What other languages call procedures, functions, or subroutines are instead “rewrite
rules.” All other programming constructs are derived from this underlying term-rewriting mecha-
nism. The appearance of a definition is similar to the function declaration of other languages, but
while that is intentional, it is also superficial. The two standard parameter passing mechanisms,
pass-by-value (initialize local variables) and pass-by-reference (var in Pascal, pointers in C, and
references in C++), can only be mimicked in Mathematica.
By setting the attribute HoldAll for a definition, the arguments in a function-call are passed
unevaluated into the body of the definition, simulating the pass-by-reference mechanism men-
tioned above. The body, therefore, effectively operates on the argument itself, allowing one to
write functions that return functions. This forms the basis for a flexible turbo-Brayton cryocooler
model. In fact, this holds true of any component-based recuperative cycle coolers.
Separate procedures model individual physical components and output two equations each,
themselves functions of pressure and temperature, that are combined for a particular system con-
figuration and solved simultaneously. One of the component procedures is given in Figure 1 to

Figure 1. Component procedure for the turbo-alternator under real gas consideration, along with
SetAttribute statement for all components.
show this construction. Although some ideal-gas approximations are used in modeling the wanner
components, real thermophysical properties for helium are used throughout — fits of HEPAK data
were generated and saved as interpolated functions of pressure and temperature.
If one's model is sufficiently simple, Mathematica will be able to find symbolic derivatives of
the equations and arrive at a solution using Newton's Method. That is not the case for our model,
since it includes interpolation functions and numerical integration. In cases such as these the
FindRoot function needs to estimate the derivative numerically and solve the system using the
secant method.
Figures 2 and 3, represent two possible configurations of a 6 K turbo-Brayton cooler. Each
pair of figures includes a schematic temperature-entropy diagram and a cycle schematic. Window
Figure 2. Left) System schematic for single-stage cooler with an intermediate precooling radiator;
Right) Schematic T-S diagram of cycle showing selected points.
A FLEXIBLE TURBO-BRAYTON CRYOCOOLER MODEL 501
Figure 3. Left) System schematic for a dual-stage cooler; Right) Schematic T-S diagram of cycle
showing selected points.
Figure 4. Window showing Mathematica cycle model for single-stage cooler with an intermediate precooling
radiator as represented schematically in Figure 2.
A FLEXIBLE TURBO-BRAYTON CRYOCOOLER MODEL 503
Figure 5. Window showing Mathematica cycle model for dual-stage cooler as represented schematically
in Figure 3.
images in Figures 4 and 5 show the Mathematica code for the overall cycle definition as well as the
input parameters and execution call. The list of starting values seen inside the FindRoot function
indicates that the system is solved directly for the pressure-temperature profile of the cycle as well
as the mass flowrate. As would be expected, the rate of convergence is dependent on ones choice
of initial values being close enough to the actual solution.
REFERENCES
1. Wolfram, S., The Mathematica Book, 3rd ed., Wolfram Media/Cambridge Press, 1996.
2. Maeder, R.E., Programming in Mathematica, 3rd ed., Addison-Wesley, Reading, MA, 1996.
3. Maeder, R.E., The Mathematica Programmer II, Academic Press, San Diego, CA, 1996.
4. HEPAK, Cryodata, Inc., P.O. Box 173, Louisville, CO, 80027.
A 10 K Cryocooler for Space Applications
D.S. Glaister, W.J. Gully, G.P. Wright, and D.W. Simmons

B. J. Tomlinson
Kirtland AFB, NM, USA 87119
ABSTRACT
Ball Aerospace and the Air Force Research Laboratory are developing a high-capacity, high-
efficiency, long-life 10 K space cryocooler system to meet applications that include Very Long
Wave Infrared Instruments. To meet the stringent new requirements, which are several factors
improved over the current state of the art, the program is developing the technology for a hybrid
J-T cryocooler coupled with a Stirling precooler. This paper provides a top-level overview of the
10 K Cryocooler Development Program including a description of the cooling system, the key
technology developments, and the program status and schedule.
PROGRAM OBJECTIVES AND REQUIREMENTS

A 10 K Cryocooler is being developed at Ball Aerospace under the management of the Air
Force Research Laboratory (AFRL) with application to defense, civil, and commercial space
missions. The basic objective of the 10 K Cryocooler Development Program is to develop the
technology for a high-capacity, long-life, high-reliability 10 K cryocooler in order to enable space
missions that require cooling to 10 K and below. Of primary interest are those applications
relying on Very Long Wavelength IR (VLWIR) focal planes that could be used in support of
high-resolution space target surveillance and tracking as well as hyper-spectral imaging.
Previously, development programs for a 10 K cryocooler for space applications have been
initiated several times. However, none of these programs have produced technology with a
maturity or performance level capable of insertion into a space program.
To meet these requirements, the 10 K Cryocooler Development Program is developing a
hybrid system using a Stirling precooler (to about 15 K) in combination with a low-temperature
Joule-Thomson cooler. In Earth orbit where significant amounts of ambient temperature
precooling are required, this hybrid combination is optimum in terms of performance and
maturity for space cooling to these low temperatures. The program intends to develop the
technology to the point where only a low-risk, follow-on, flight hardware program would be
needed to produce a cryocooler for space missions. As will be discussed in more detail later in
this paper, we are concentrating on developing the J-T cooler and, in particular, the compressor,
as the Stirling precooler technology is relatively mature.
Cryocoolers 11, edited by R.C. Ross, Jr.

506 J-T AND THROTTLE-CYCLE CRYOCOOLER DEVELOPMENTS
* Average heat load.

**Motor input power divided by capacity at required cold tip temperature.
***Cryocooler mass divided by capacity at required cold tip temperature.
The 10 K Cryocooler Program was initially baselined to provide a cooling capability of 100
mW at 10 K. However, early in the program, in response to potential user input, the requirements
were increased to 250 mW at 10 K. The cooling requirement is for an average load as the hybrid
system has a thermal storage or load leveling capability, which will be discussed later. Table 1
provides a summary of the primary performance requirements and goals of the program. The
power and mass efficiency requirements of the 10 K Cryocooler represent several factors of
improvement over the current state of the art in space coolers.
10 K CRYOCOOLER SYSTEM DESCRIPTION

General System Description
In the hybrid system, most of the cooling is provided with the relatively mature Stirling
precooler, and the J-T system is used for the last stage of cooling to 10 K. By only using the J-T
in this modest role, we keep the demands on it low, which enabled us to come up with a long-life
design. The Stirling cycle is the proven leader in reliable, long-life, compact, high-efficiency
cooling from ambient to cryogenic temperatures below 35 K. However, because the Stirling is a
regenerative cycle that relies on the heat capacity of solid materials, it is limited for cooling
below 10 K where the regenerator is overwhelmed by the helium heat capacity. The J-T is a
recuperative cycle that exchanges heat directly between counter-flowing gas streams and does not
rely on an intermediate material heat capacity. Thus, the J-T is optimum for cooling at
temperatures of 10 K and below. However, without a precooler, the J-T is inherently more mass
and power inefficient in cooling from ambient temperature. This is because the high-pressure
ratios required with cooling from ambient temperature necessitate a relatively large compressor.
Thus, the hybrid combination of the J-T cold stage with the Stirling precooler exploits the
inherent advantages of both thermodynamic cycles.
As originally conceived, Ball Aerospace initially proposed a 100 mW hybrid system1
consisting of a J-T cooling stage that would be coupled to a modified Ball 35/60 K (or SB335)
split Stirling cycle precooler2,3 (Figure 1). The refrigerator consisted of a precooler, a J-T fluid
loop, and a thermal storage unit (TSU). When using the TSU to reduce the precooler
requirements during transients, the hybrid system is referred to as a “Redstone Interface.” The
J-T loop consisted of a rotary vane compressor, its associated drive electronics, and a cold head.
The cold head houses a J-T expander, a TSU, a series of heat exchangers, and a 10 K sensor
simulator. The Stirling precooler would be coupled to the Redstone Interface inside a vacuum-
insulated cold head (Figure 2).
A 10 K CRYOCOOLER FOR SPACEAPPLICATIONS 507
Figure 1. CAD rendering of original 100 mW 10 K Cryocooler.
Figure 2. Schematic of 10 K Cryocooler System.

For the original 10 K system, the modified 35/60 K’s midstage was to provide about 150
mW of cooling for the Redstone Interface’s helium flow at 40 K and the final stage was to
provide 150 mW of additional cooling at 15 K to the J-T cooler or TSU. However, when the
10 K cooling requirement grew to 250 mW, the precooling load increased beyond the capacity of
the 35/60 K Cryocooler. Thus, the plans were dropped to modify the 35/60 K precooler and a
commercial Gifford-McMahon cooler was substituted in its place. Under IR&D resources, Ball is
J-T Compressor
A key aspect of the 10 K Cryocooler and the program’s technology development is the long-
life, high-reliability J-T compressor. The program elected to use a rotary vane compressor
because they have the advantages of small size, a 2:1 or better compression ratio, low vibration,
no check valves, light weight, and extensive commercial heritage. However, the typical
commercial rotary vane compressor employs either water or oil based lubrication that is not well
suited to the conditions within a J-T cooler with a dry helium working fluid. Ball has developed a
proprietary lubrication system that works well in this environment and eliminates concerns with
condensable contaminants or particulates.
The 10 K rotary vane compressor concept is illustrated in Figure 3, which shows a cutaway
of a commercial rotary vane compressor. The rotor rotates within a fixed stator and carries
sliding vanes. These vanes form separate pockets of variable volumes of helium gas. As the rotor
rotates, the vanes are retained within their guiding slots, while also being thrown outward and
sealed against the inward facing wall of the stator.
The rotary vanes are designed to wear at predictably slow rates. Using commercial pumps
run with vanes lubricated in the same manner as the 10 K system, with a dry helium working
fluid, and vane structural loads similar to the 10 K, a vane wear database has been developed and
is illustrated in Figure 4. The figure shows how the experimental wear rate for vanes in a pump
decreases as a function of hours running time. Over 15,000 hours of wear data has been
accumulated on one pump, while two other pumps have obtained 2000 hours of data at low and
high temperature extremes. The vane wear test data has been obtained and incorporated into our
10 K compressor model. Based on this data and modeling, we predict more than a factor of 2
margin on vane wear to meet a 10-year lifetime.
Additionally, we attribute the majority of the decrease in wear rate to the slow (but steady)
polishing of the stator eccentric bore by the vanes. In the 10 K pump we have improved on this
process by having the eccentric stator bore polished to a very smooth finish that is more than two
times smoother than the commercial stators. Thus, not only are our lifetime predictions
conservative, but we predict that the 10 K pump will start off with a very low wear-rate and
require only a limited number of test hours to validate 10-year design.
Figure 3. Commercial rotary vane compressor.

A 10 K CRYOCOOLER FOR SPACE APPLICATIONS 509
Figure 4. Vane wear test data showing >10-year lifetime capability.
developing a protoflight-level, larger-capacity two-stage Stirling cryocooler which should be

completed in fall 2001. With minor cold head modifications, this unit could provide the
precooling needs of the 250 mW 10 K Cryocooler, which include about 400 mW at 15 K.
As an additional precaution, both a passive chemical getter and a depth filter are located
downstream of the compressor to prevent condensable contaminants or particulates from getting
to the J-T valve. Both the getter and filter are sized to accommodate over 10 times the measured
rates and amount of contaminants over 10 years of time from the compressor.
The overall 10 K compressor design is shown in Figure 5. It has two stages to divide the
work over more segments, and to improve the thermal efficiency of the compression. It operates
at modest pressures. Its total mass, including intercooler heat exchangers, is about 4.8 kg with
dimensions of about 12 inches in length by 4.5 inch height by 3.5 inch width.
Cold Head
The J-T portion of the 10 K Cryocooler System includes tubing for connection to the
compressor (which can be remotely located from the cold head to minimize vibration or
integration impacts), the J-T valves, heat exchangers within the J-T loop, and heat exchangers
Figure 5. 10 K Cryocooler compressor.

Figure 6. Simple tube-in-tube COOLLAR heat exchanger adequate for that unbalanced application.
Figure 7. COOLLAR J-T valve, which has sophisticated anti-plugging features.

with the precooler. These J-T cold head components for the 10 K program are highly leveraged
off the extensive heritage at Ball on programs such as COOLLAR.4 Figures 6 and 7 contain
pictures of a COOLLAR heat exchanger and J-T valve.
The heat exchangers are intimately tied to the precooler design, and the system can either
include a small precooler and high-efficiency exchangers, or more modest exchangers and a
larger precooler. In the present system, there are four exchangers of varying degrees of
complexity with efficiencies projected to be in excess of 99%.
The cold head can also contain an external or internal TSU to provide an inherent, high-
capacity load leveling capability that is unmatched by any other cryocooler technology. A modest
demonstration of the TSU load leveling capability will be included in the 10 K Cryocooler
Development Program. Candidate TSUs include a helium sorption based system internal to the
cooler which also serves as a low-temperature getter. A complimentary development program for
a high capacity (to support peak loads in excess of 1000 mW) 10 K TSU is in progress by
Redstone Engineering under a Space Based Infrared (SBIR) System contract from the AFRL.
The Redstone TSU will be integrated with the 10 K Cryocooler in 2001 following the
Development Program.
A 10 K CRYOCOOLER FOR SPACE APPLICATIONS 511
Cooler Electronics
Electronics are a significant part of these so-called “mechanical” cryocoolers. The 10 K
Cooler system is based upon Ball’s mature Stirling cryocooler drive electronics. Ball will be
delivering a flight version of these electronics in July on the HIRDLS program. The 10 K
program will be adding a new section for driving the J-T circulating pump, and for controlling
the cold tip temperature. The rotary vane compressor will use a standard brushless dc motor
driver with a sensorless commutation technology, where the timing is deduced from transients on
the motor windings themselves. In this way, the need for Hall probe position sensors on the stator
is eliminated. The motor driver also incorporates speed control, which is used to servo the 10 K
cold tip temperature. An analog section converts the temperature sensor output, with proper
compensation, into a command voltage for the rotary speed of the compressor.
PROGRAM SCHEDULE
The 10 K Cryocooler Program is on schedule to deliver an Engineering Unit 10 K J-T cooler
integrated with a laboratory quality G-M precooler to the AFRL in April 2001. The J-T cooler
will include a flight quality rotary vane compressor capable of withstanding launch vibration.
The compressor is nearly at the completion of the fabrication and assembly process. It is
scheduled to be in test in mid-July 2000. While the cold head is being built, the compressor will
be subjected to approximately 6 months of extended life testing to verify its long-life capability.
The J-T cold head is in design and scheduled to be assembled by January 2001. The cold head,
compressor, and electronics will then be integrated with the G-M precooler for testing starting in
February 2001.
SUMMARY
A high-capacity, high-efficiency, long-life 10 K space cryocooler is in development by Ball
Aerospace and the AFRL to meet applications that include Very Long Wave Infrared
Instruments. To meet the requirements that are several factors improved over the current state of
the art, the technology is being developed for a hybrid J-T cryocooler coupled with a Stirling
precooler. Culminating in the delivery in 2001 of a Engineering Unit J-T with a flight quality
rotary vane compressor, the 10 K Cryocooler Development program will have successfully
developed the technology to the point where only a low-risk flight hardware program would be
needed to provide users with a 10 K space-cooling capability.
ACKNOWLEDGMENTS
The authors wish to acknowledge the support and contributions of Thom Davis and John
Reilly of the Air Force Research Laboratory and Jim Lester of Redstone Engineering.
REFERENCES
1. Levenduski, R., Gully, W., and Lester, J., “Hybrid 10 K Cryocooler for Space Applications,”
2. Gully, W., Carrington, H., and Kiehl, W. K., “Qualification Test Results for a Dual-Temperature
Stirling Cryocooler,” Cryocoolers 10, Plenum Press, New York (1999).
3. Carrington, H., et al., “Multistage Coolers for Space Applications,” Cryocoolers 8, Plenum Press,
New York (1995), pp. 93-102.
4. Fernandez, R., and Levenduski, R., “Flight Demonstration of the Ball Joule-Thomson Cryocooler,”
Modern Trends in Designing Small-Scale
Throttle-Cycle Coolers Operating
with Mixed Refrigerants
M. Boiarski, A. Khatri, O. Podtcherniaev*, and V. Kovalenko**

IGC – APD Cryogenics, Inc., Allentown, PA 18103 USA
* IGC- Polycold Systems, San Rafael, CA 94903 USA
**Moscow Power Engineering Institute/ Moscow, Russia
ABSTRACT
The paper discusses experiences in development of coolers based on a single-stage com-
pressor operating with mixed refrigerants. Data on the refrigeration capacity and Carnot effi-
ciency were obtained for both one-stage and two-stage refrigeration cycles. It is assumed that
the two-stage cycle employs only one phase separator, which can operate at different tempera-
tures. Experimental data is presented for one-stage coolers that operate in a temperature range
of 70 K to 200 K. Data on the efficiency of counter-flow heat exchangers and experience in the
regulated throttle design are also discussed.
INTRODUCTION
Closed cycle coolers operating with Mixed Refrigerants (MR) can be built with different
configurations.1 A one-stage refrigeration cycle based on a single-stage compressor is most
attractive in development of small-scale coolers because of its simplicity. The high refrigera-
tion efficiency of the coolers based on oil-lubricated compressors has been proven in research
and many practical applications.2,3,4 A two-stage refrigeration cycle allows more freedom in
design especially to resolve freeze-out problems at temperatures below 130 K.
Reliability and lifetime of the coolers are important. They depend on compatibility of the
MR with the oil, quality of the oil separation system and design of hardware components.
Modern oil separation systems can be very reliable in long-term operation. It is proven by
experience with Gifford-McMahon and pulse tube coolers. Experience shows that the coolers
operating with the MR are quite tolerant to the oil contamination and may employ very simple
and cost-effective oil separation systems. An auto-cascade refrigeration cycle design allows
an alternate means of handling oil circulation.5 Application of a special precooling cycle to
improve on oil management has also been considered.6 These options complicate the design
and increase the cost of the cooler.
This paper discusses mainly the refrigeration performance and optimal cooler design. Both
one-stage and two-stage refrigeration cycles were selected for the comparison.

REFRIGERATION PERFORMANCE OF DIFFERENT CYCLES

Variations of the cooler schematic based on a single-stage compressor are shown in Figure 1.
Any cooler consists of a compressor unit (CM), which includes a compressor itself, condenser or
after cooler to reject heat from the cycle to the environment, and an oil separator (not shown). The
compressor unit operates at temperatures above ambient and can be integrated with the cry-
ostat assembly of a certain configuration. The cryostat includes one or more counter-flow heat
exchangers (HX), throttle devices (TH), and an evaporator interface (EV) to remove heat from
the object to be cooled at the refrigeration temperature
A multi-stage refrigeration system was proposed by A. Klimenko to liquefy natural gas. Later,
this cycle was modified to provide refrigeration capacity at with a single-stage compressor.
The number of refrigeration stages and the cryostat schematic might be different depending on the
temperature level and field of application. For cryocooler design to be studied, we selected a two -
stage cooler of the simplest configuration, which is presented in figure 1 (a). It is possible to sub-
cool the separated liquid between the phase separator and the throttle by the return flow to reduce
the losses. However, it complicates the system schematic.
An overall MR composition Z includes low-boiling and high-boiling components circulating
together through the compressor. A phase separator at temperature divides the high pressure
MR in two fractions: liquid (X), which is enriched with the high-boiling components and
vapor (Y), which has an increased content of the low-boiling components.
The vapor fraction is directed to the lower temperature part of the system. Meanwhile the
liquid fraction goes through the throttle to the return line. It provides cool down of the vapor frac-
tion to the intermediate refrigeration temperature Thus multi-stage refrigeration helps to avoid
Figure 1. MR cooler schematic: (a) two-stage cycle, (b) one-stage cycle.

THROTTLE-CYCLE COOLERS WITH MIXED REFRIGERANTS 515
freeze out by taking out high – boiling components from circulation to the low – temperature sec-
tion of the cooler.
The comparison of the cycles can be made based on the MR thermodynamic properties ob-
tained with an equation of state. The cycle model should include energy and mass-flow balances for
each component of the system. In addition, some assumptions have to be made for a minimal
temperature difference to define a pinch point in the heat exchangers. Pressure drop values
(DP) should be assumed for the supply and return lines. A computer program calculates the refrig-
eration cycle performance at any phase separator temperature selected in a range of
The phase separator efficiency which is defined as the part of the liquid fraction
supplied to the return line relative to the total liquid fraction at that point, must also be assumed. For
this set of initial data, the first stage refrigeration temperature should be selected to provide
temperature differences larger than for each HX. The two - stage cycle can be transformed to
a single-stage cycle by assuming that the efficiency of the phase separator is zero.
The Carnot Efficiency (CEF) of the cycles was determined as The
coefficient of performance (COP) related to a specified and was calculated as
where is the compressor power consumption and is the refrigeration
capacity at
Two MR compositions were selected for the analysis to provide refrigeration at
Both MRs include argon and hydrocarbons. An optimal MR provides very low extrinsic losses in
the one-stage refrigeration cycle and Carnot efficiency as high as with an ideal compres-
sor. A non-optimal MR has lower even with an ideal compressor.
Table 1 presents data for the one-stage cycle at bar and bar. High
CEF values obtained with the optimal MR is due to a very small temperature difference in the HX.
This well optimized MR provides: K - average DT, at the cold end
K, at the warm end Data also shows that replacing the isothermal compressor with an
adiabatic one causes 20 % reduction in CEF.
Table 2 gives a comparison of the one-stage and two-stage cycle at and
adiabatic compressor efficiency The other parameters are identical to those in Table 1.
Data presented in lines 1 and 4 (table 2) show the influence of on CEF in the single-stage
cycle. Increase in from 0.5 to 5 K leads to an additional 20% reduction in CEF. The refrigera-
tion performance of one-stage and two-stage cycle was calculated at different temperatures of the
phase separator and phase separator efficiencies Calculations show that recirculation of
any part of the liquid fraction causes the temperature difference in the HX to increase and reduces
CEF-lines 1,2 and 3 (Table 2).
In case of the non-optimal MR composition, the one–stage cycle is always better than the
two-stage cycle with The performance can be improved if (lines 3 and 11).
Even so the CEF values are not higher than in one-stage cycle.
From this analysis it is concluded that for a given MR composition, the one-stage refrigeration
cycle provides equal or better refrigeration performance. It can be used in the optimal cooler design
if such issues as the oil circulation and freeze out can be handled properly for the selected MR.
Two different technologies can be considered in development of the MR coolers. The differ-
ence is in the phase-state of the MR at the cryostat inlet. Based on this feature, it is possible to
define either a gas refrigerant supply (GRS) or partially liquefied refrigerant supply (LRS) technol-
ogy.2 The dew point temperature of the MR designed for the GRS technology should be less
than at the discharge pressure; but for the LRS: A major advantage of the GRS
technology is the simplicity of the system design. In this case, only single-phase flow should be
considered in the compressor-unit design. The amount of the MR charged is smaller in the system.
It allows a higher fraction of flammable components if so desired for an application. However, the
GRS technology restricts the maximal available refrigeration capacity per unit of refrigerant. The
LRS technology provides higher refrigeration capacity but requires additional efforts in the distri-
bution of two-phase flow and oil management.
For both technologies (GRS and LRS) the characteristics of idealized one-stage cycle were
analyzed. It was assumed that the compressor is ideal, Such a highly idealized
cycle gives maximal available and CEF values for a specified and a given compressor
displacement volume. In this case the intrinsic losses are caused by the entropy growth in the
throttle and counter-flow HX; they are related only to the thermodynamic properties of the MR.
Data presented in Figures 2 and 3 were obtained2 for a single-stage compressor having the displace-
ment volume of 28 1/min. The compressor discharge pressure during the calculations was always
less than bar. For any selected the optimal MR was calculated based on readily available
components. They include hydrocarbons (HC), fluorocarbons (FC) and hydro-fluorocarbons (HFC).
Some MR consists of neon (Ne) to obtain a higher Concentration of each component was
defined to avoid the MR freeze-out at TR.
These data helps to estimate maximal and minimal compressor power for a selected
by making an assumption on the compressor efficiency, and DP. An actual cooler can
be designed using the computer software that includes the HX models.3
Figure 2. Refrigeration capacity of idealized MR cooler, 28 1 / min compressor.

Figure 3. Carnot efficiency of idealized MR coolers.
EXPERIENCE IN DEVELOPMENT OF THE COOLER COMPONENTS

Evaluation of the performance of actual MR coolers can be made based on a system analysis,
which takes into account mutual interactions of the basic components.3 In the cryostat assembly,
the counter-flow heat exchanger and throttle losses have the most influence on the cooler perfor-
mance.
Although different types of HX can be integrated in the MR cooler design,7 we use a tube-in-
tube type HX. Different configurations can be used to build a low cost, compact and highly effi-
cient systems. Our experience shows that the traditional models do not predict the efficiency of the
HX operating with zeotropic MR with an acceptable accuracy. We associate it with the fact that the
hydraulic patterns of the two-phase (vapor-liquid) flows are very sensitive to design features, espe-
cially in small-scale systems. That is why new approach must be developed to correlate experimen-
tal data on the heat transfer efficiency. It was found3,8 that the assumption of the homogeneous
hydrodynamic structure of two-phase flows helps to treat the experimental data on the overall heat
transfer coefficient (HTC). The model can be tuned using limited experimental information for a
given HX configuration. A function which includes the mass flow rate and MR properties, can
be used to plot the test data, Figure 4. This graph helps to interpolate and sometimes extrapolate
data.
Initially, this model was tested with the HX having seven tubes for the return MR flow.3 New
data were obtained for a bigger size HX having 10 tubes. It was tested with different refrigerants
consisting of five components. Deviations between experimental and calculated data at different
points presented in Figure 4 were less than 10 %.
Figure 4. Experimental data on the HTC for the five component refrigerants.
The computer software calculates the HTC values based on the set of parameters that can be
measured only at the inlet and outlet of the HX. Data on the compressor performance and refrigera-
tion capacity should be available to improve the accuracy of the calculations. For small-scale cool-
ers this method is especially valuable. It allows estimation of HTC values using data obtained from
a typical refrigeration performance test. This model combined with the limited experimental data
can be used to predict the cooler performance3 based on the system analysis. It also allows evalua-
tion of the HX efficiency (HEE) defined as where and is calculated
refrigeration capacity data for real and idealized cycle, accordingly.
Calculated and experimental data on the actual cooler performance is presented in Table 3.
They were obtained for the HX having 10 tubes in 1. The adjustable coefficients have been defined
based on data presented in line 4. Measured and calculated DT in the HX are in a good agreement:
- measured, from the model; – measured,
calculated. In the presented regimes this HX is not efficient (HEE < 0.5). For example, in line 4 a
value of W is than W due to the extrinsic losses in the HX. Data from this table
shows that the model predicts with an acceptable accuracy in order to evaluate influence of the
HX on the cooler performance.
The optimal HX design should satisfy conflicting requirements in providing low values for DT
and DP. Table 4 presents characteristics of the tested HX of different configurations. The MR based
on Ar and HC were designed to operate in a temperature range of Test data on the
HX efficiency operating at is presented in Table 5. Lines 1 to 3 are related to a compo-
sition, which provides GRS technology. The line 4 presents data on the LRS regime.
The deviation of calculated and experimental data on did not exceed 15%. The best per-
formance was achieved for this test with the big HX No. 2 (line 2): Although a small
value of in the return line is an advantage of the HX- No. 3, the HEE was less than 0.5 with the
tested MR.
Figure 5. Cool-down performance: laminar throttle versus capillary.
The throttle device is another cooler component, which is important in the optimal design. It
influences the cooldown time, reliability, and noise spectrum of the cooler. The throttle device
might be as simple as a capillary or an orifice. However, in some applications it might be of a very
sophisticated design to improve the system functionality.9
Two strategies can be considered to enhance the cooldown performance. One strategy is re-
lated to process management. Increase in the mass flow rate during cool down might be achieved
either with special MR design10 or by having laminar flow in the fixed area throttle.11 Comparison
of the laminar throttle versus capillary is shown in Figure 5. The test data shows that the laminar
throttle provides higher-pressure value in the suction line. In turn, it allows the higher MR flow
rate and reduction in the cool down time Another strategy considers different approaches in
the design of a throttle device, which permits an increase in the MR flow. For a small-scale coolers
a compact bimetal valve was designed and successfully tested.9 This valve can be also used to
increase during steady state operation.
Experience in the cooler component design and the developed computer software give oppor-
tunities to meet many different customer requirements. In many cases a low cost cooler of a stan-
dard configuration can be used to operate efficiently with different MR designed for the particular
temperature range. Adjustments in the evaporator interface can be made to customize the cooler.
ENHANCED PERFORMANCE OF MODIFIED COOLERS

Refrigeration capacity and Carnot efficiency are among the prime concerns for small-scale
cooler design. Both the compressor power consumption and the cooler dimensions depend on
and CEF. In addition, increased refrigeration capacity gives more opportunities in meeting the
customer requirements. As it is shown above it might be achieved with the optimal design of the
MR and hardware for the chosen cooler configuration and refrigeration technology: either GRS or
LRS.
Application of the optimized hardware for the GRS technology can increase by 1.5 times
compared to the first-generation coolers. Data on the refrigeration map comparison is given else-
where 2,3for coolers that operate in the temperature range from 75 to 140 K.
Employment of the LRS technology requires some modification of the compressor unit and
the cryostat components. The configuration of the cooler is identical for the GRS and LRS technol-
ogy. Experimental data on the refrigeration performance is given in table 6 for the cooler based a
single-stage compressor having 28 1 /min displacement. includes 40 W consumed by the fans
removing heat from the compressor. The high Carnot efficiency achieved is not
a limit yet. It can be further improved by reducing compressor losses and throttle optimization.
It is important to note that data for MR #5 was obtained with the perlite-type heat insulation
without vacuum in the cryostat. This might be a big advantage in many applications.
A typical refrigeration map for the cooler with a capillary throttle is presented in
Figure 6. This cooler, which employs LRS technology gives more stability in with variations
in a range of W up to
Figure 6. Refrigeration map of the cooler employing the LRS technology.
Figure 7. Bimetal throttle valve influence on the refrigeration map.
The optimal throttle design improves both cool down and steady-state performance. Opening
the valve increases evaporator pressure, and the MR flow rate. It gives more freedom in design-
ing a rapid cool down cooler. A cooler with a solenoid throttle valve was designed for computer
cooling4. It has an intrinsic cool down time less that 10 min when operating with MR #4 (table
6); was less than 1 hour with 2 kg copper mass attached to the evaporator.
Application of the bimetal valve, which can smoothly regulate the MR flow rate with tempera-
ture essentially changes the shape of the refrigeration map A typical refrigeration map
measured over a broad temperature range for the cooler with a capillary throttle has a maximum of
capacity at a low temperature, as shown in Figure 7. At higher temperatures is less than
(max) due to a flow restriction caused by a capillary or an orifice.
A cooler with the bimetal valve installed in parallel to a capillary can support a heat load
increase over a wide temperature range up to Test data for the MR #5 (Table 6) is also presented
in Figure 7. The valve design is compact and suitable for a small-scale cooler. It provides powerful
small-scale cooler and gives more flexibility to meet the customer requirements on the system
functionality.
SUMMARY
New approaches and experience in developing mixed refrigerant technology allow an optimal
MR and hardware design, which improves the performance of low cost coolers based on one-stage
compressors. In the temperature range of 100 to 200 K, Carnot efficiency as high as 0.12... 0.15 has
been achieved.
MR coolers that include the regulated throttle valve allow both a rapid cool-down and a highly
efficient steady-state operation with increased heat load. It gives more flexibility to meet customer
requirements.
The single-stage refrigeration cycle provides better refrigeration performance compared to the
two-stage cycle, which includes the phase separation process at or below the ambient temperature.
The two-stage cycle might be considered for a small-scale cooler design to improve the oil manage-
ment and avoid freeze-out problem in some applications.
ACKNOWLEDGEMENT
We highly appreciate a help of S. Harold of IGC-APD Cryogenics in developing and testing
coolers. R. Logsworth is thanked for useful discussions of the paper.
REFERENCES
1. Boiarski, M., Brodianski, V., Longsworth, R., “Retrospective of Mixed-Refrigerant Technology and
Modern Status of Cryocoolers Based on One-Stage, Oil-Lubricated Compressors,” Advances in Cryo-
genic Engineering, Plenum Press, New York: Vol.43 (1998), pp.1701 – 1708.
2. Boiarski, M., Khatri, A., Podcherniaev, O. “Enhanced Refrigeration Performance of the Throttle-Cycle
Coolers Operating with Mixed Refrigerants,” presented at CEC/ICMC meeting, July 12-16, 1999;
Montreal, Canada.
3. Boiarski, M., Khatri, A., Kovalenko, V., “Design optimization of the throttle-cycle cooler with mixed
refrigerant”, Cryocoolers 10, Edited by R.G.Ross, Jr./Plenum Press, New York, 1999.
4. Ellsworth, M., et al, “Performance of a mixed-refrigerant system designed for computer cooling,” pre-
sented at CEC/ICMC meeting, July 12–16, 1999; Montreal, Canada.
5. Little, W., “Self-cleaning low-temperature refrigeration system”, US Patent, No. 5,617,739,1997.
6. Alexeev, A., Haberstroh, Ch.., Quack, H., “Mixed gas J-T cryocooler with precooling stage,” Cryocool-
ers 10, Edited by R.G.Ross, Jr./ Plenum Press, New York, 1999.
7. Luo, E., at al., “Experimental comparison of mixed-refrigerant Joule-Thomson Cryocoolers with two
types of counterflow heat exchangers,” Cryocoolers 10, Edited by R.G.Ross, Jr./ Plenum Press, New
York, 1999.
8. Landa, Yu., Gresin, A., “Design method for J-T – microcooler heat-exchanger applying multicompo-
nent refrigerant,” Proceedings the 19th Int. Cong.Refr, 1995.
9. Khatri, A., Boiarski, M., “Development of the rapid cool-down close-cycle coolers operating with mixed
refrigerants,” Presented to the ICEC 18, Feb. 2000, India.
10. Longsworth, R., Boiarski, M., Khatri, A. “Cryostat refrigeration system using mixed refrigerants in a
closed vapor compression cycle having a fixed flow restrictor,” US Patent No.5,579,654,1996.
11. Hill, D., at al, “Low vibration throttling device for throttle-cycle refrigerators,” US Patent No. 5,875,651,
1999.
Thermodynamic Analysis of a Mixed-Refrigerant
Auto-Cascade J-T Cryocooler with Distributed
Heat Loads
M.Q. Gong, E.C. Luo, J.T. Liang, Y. Zhou, and J.F. Wu

Cryogenic Laboratory, Chinese Academy of Sciences
Beijing, China, 100080
ABSTRACT
In this paper, an effort is made to go further into the thermodynamic process of the Auto-
Cascade refrigeration cycle using multicomponent zeotropic mixtures as working fluids. The
exergy method is employed to analyze the thermodynamic characteristics of components and the
whole refrigeration cycle. Specially, the condition that there are extra-distributed heat loads
along the heat exchangers is considered. Extensive comparison is made between the single stage
J-T cycle and a typical Auto-Cascade cycle at the same condition. At the calculation conditions
presented in this paper, the total exergy gained in the MARC is 6.6% better than the single stage
cycle, and 9.5% better in the situation with distributing heat loads. The results show that using
appropriate Auto-Cascade cycle can improve the performance of the refrigerator.
INTRODUCTION
In recent years, there has been a remarkable development of the mixed-refrigerant Joule-
Thomson cryocooler. The research history of the mixed-refrigerant Auto-Cascade Refrigeration
Cycle (MARC) can cast back to 1930’s.1 However, it is not until 1959 that this MARC had been
carried out successfully by A. P. Kleemenko2 in a natural gas liquefaction process. Accordingly,
the mixed-refrigerant Auto-Cascade refrigeration cycle is also called Kleemenko cycle. From
then on, many researchers carried out their study on this refrigeration system.3,4,5 The reason
why there are so many people interested in this cycle is that without any modification of the
hardware the Auto-Cascade cycle system can produce cryogenic and ultra-low cooling over a
large temperature range from below liquid nitrogen temperature (77K) to wanner than 230K, the
conventional vapor-compression cycle temperature range. In this cycle one or more intermediate
phase separators (liquid-vapor separators) are employed, which avoids the compressor lubricant
and high boiling point components from plugging in the coldest section. In addition, driven by a
modified commercial air-conditioning compressor, this type of the mixed-refrigerant Auto-
Cascade throttle cryocooler is quite reliable and flexible in many applications such as infrared
detectors, cryosurgical devices, HTS devices, material studies, biomedical storage, etc.
However, the internal mechanism of the mixed-refrigerant Auto-Cascade refrigeration is not
very clear. Compared to the original Linde-Hampson J-T refrigeration cycle, the thermodynamic
processes of the Auto-Cascade refrigeration cycle are more complicated. Generally, there are
heat losses in counter current flow heat exchangers and some other elements from the insulation.
And in most gas liquefaction industry, the gases should be pre-cooled before being liquefied. All
these cases can be considered as distributing heat load along the refrigeration process. It will be
more complicated to analyze the thermodynamic process of the refrigeration cycle with
distributing heat load. This article discusses this problem of a mixed-refrigerant Auto-Cascade
throttle refrigeration cycle. The analysis includes prediction of the mixture properties, and
calculation of the sub-thermodynamic process for each element such as the heat exchanger,
liquid-vapor separator, and throttle device.
AUTO-CASCADE REFRIGERATION CYCLE DESCRIPTION

The number of the heat exchangers and phase separators in the Auto-Cascade cycle varies
flexibly based on the required cooling temperature. There are different arrangement types of the
mixed-refrigerant Auto-Cascade refrigeration cycle for different cooling temperature or different
applications. Fig. 1 shows a basic flow pattern of the single stage throttle cycle. Fig. 2 shows a
basic flow pattern and heat exchanger arrangements of the two stage Auto-Cascade cycle. The
system, illustrated in Fig. 1, consists of a compressor, an after-cooler, a counter-current flow heat
exchanger, a throttle device that is usually a capillary or an orifice, and an evaporator. All of the
refrigerant mixture flows through the complete circuit. This arrangement is better suited to
smaller and simpler systems. Any compressor lubricant entrained in the circulating refrigerant
mixture requires proper management, e.g. the employment of an oil separator in the cycle, to
avoid plugging problem in the coldest section. The two stage Auto-Cascade system, illustrated in
Fig. 2, employs four counter-current flow heat exchangers, two vapor-liquid separators. In this
system, after each cooling step in two counter-current flow heat exchangers, a phase separator
removes the condensate from the vapor stream. A throttling device controls the exiting liquid
flow. The liquid passing through the throttling device usually becomes two phase blend. The two
phase blend mixed with the low pressure return stream cools the coming high pressure stream.
Finally, the high pressure stream passes through the end throttling device to produce cooling
capacity at the lowest refrigeration temperature.
Figure 1. Flow schematic of a single-stage mixed-refrigerant J-T refrigeration cycle.
Figure 2. Flow schematic of a two-stage mixed-refrigerant Auto-Cascade J-T refrigeration cycle.

ANALYSIS OF J-T COOLER WITH DISTRIBUTED HEAT LOAD 525
THERMODYNAMIC PROCESS ANALYSIS OF THE MARC

The thermodynamic analysis of the MARC is based on the following group of equations:
(1) The mass conservation equation, called M Equation.
(2) Vapor-Liquid equilibrium equations, called E Equations.
(3) The mole conservation equation, called S Equation.
(4) Energy conservation equation, called H Equation.
The equation group presented above is commonly called MESH equation group. However,
it is not enough using energy conservation equation to ensure that the refrigeration process
performs properly; it must be considered from the second law of the thermodynamics. Then, the
exergy method is employed in the analysis of the performance of the MARC. The exergy
equation of the refrigeration cycle is presented as the following:
where W is the input power, is the exergy of the cooling capacity, is the exergy loss of
each element of the cycle, is the exergy efficiency of the refrigeration cycle. The following
task is to find out the exergy loss of the each element of the MARC.
Compressor
In the hermetic air-conditioning compressor, it is an adiabatic compressing process,
illustrated in Fig. 3 (A). The ideal adiabatic compressing process is an isoentropic process. There
is no exergy loss. However, in the real compressing process, there is entropy generation. The
exergy loss in the compressor can be expressed as the following:
where H is enthalpy, E is exergy, S is entropy, is the ambient temperature. Subscripts 1 and 2

denote the inlet and outlet of the compressor, respectively.
After Cooler
It is an isobaric cooling process in the after cooler, illustrated in Fig. 3 (B). There are two
parts of the exergy loss in the after cooler. One is the internal loss of the after cooler, the other is
the exergy of the heat emission. These are expressed as follows:
where: is the heat emission of the after cooler.
Counter Current Flow Heat Exchanger

The heat exchanger is the most important device in the cycle, illustrated in Fig. 3 (C). In the
counter current flow heat exchanger, the return low pressure stream cools down the coming high
pressure flow. In cryogenic heat exchanger, with the exergy analysis method, the income is the
cooling exergy increase of the high pressure flow, on the other hand, the payment is the cooling
exergy reduction of the return low pressure flow. When there are extra distributing heat loads
along the heat exchanger, the income is the cooling exergy increase and the exergy of the extra
heat loads. The energy and exergy equations are expressed as follows:
where: are heat exchanged in the heat exchanger, high pressure stream
exergy changed, low pressure stream exergy changed, the extra heat load, and the exergy of the
extra heat load, respectively. If is the exergy income of the refrigeration cycle, e.g. in the
gas liquefaction process, Eq. (13) is used in the calculation of the exergy loss of the recuperative
process of the heat exchanger. When the heat load is the cooling capacity loss, the exergy loss of
the heat exchanger can be expressed as Eq. (14).
Figure 3. The thermodynamic process schematic of each device of the Auto-Cascade cycle.
Liquid -Vapor Separator

The liquid-vapor separator is another key device in the multicomponent zeotropic mixed-
refrigerant Auto-Cascade refrigeration cycle, illustrated in Fig. 3 (D). The high pressure stream
is separated into two phase flow in the device. The vapor stream goes down to the next section,
while the liquid stream passes through the throttling valve to join in the return stream cooling the
coming high pressure stream. The calculation equations are expressed as follows:
where: F is the mass flow; are mole fraction of input stream, vapor stream and liquid
stream, respectively; subscript v, l denote the vapor and liquid phase of the stream. The emphasis
is put on the calculation of the phase equilibrium of the mixture. In this article, the exergy loss of
the L-V separator is not considered.
Throttling Device
The adiabatic throttling process is a classical irreversible process, illustrated in Fig. 3 (E).
The energy and exergy equations are expressed as follows:
Blending Process
The process that the liquid stream passing the throttling device blends with the return stream
is another classical irreversible process, illustrated in Fig. 3 (F). In the refrigeration cycle, it is an
isobaric and adiabatic process. The equations are:
Evaporator
It is an isobaric process in the evaporator, illustrated in Fig. 3 (G). The calculation equations
are:
Optimization Model
Thus, with the above equations one can simulate the thermodynamic performance of the
mixed-refrigerant Auto-Cascade J-T refrigerator. An optimization model of this refrigerator can
also be developed. The objective function with the constraining conditions are expressed as
follows:
where: is the cooling temperature, is the number j stage L-V separation temperature;
are the high and low pressure, respectively; are the temperature difference
and the pressure drop in the heat exchanger.
THERMODYNAMIC PERFORMANCE SIMULATION

Based on the knowledge described above, a program has been developed for the simulation
of the thermodynamic process of the MARC. The main task of this article is focused on the
internal mechanism of the refrigeration cycle, especially on the thermodynamic parameters
coupling in the cycle. Therefore, the compressing and separating processes are considered as
ideal processes in the calculation, that is, no exergy loss occurred. The thermodynamic
performances of the ideal refrigeration cycles, illustrated in Fig. 1 and 2, are simulated at the
same given calculation condition. The calculation conditions are presented as followings:
Mixture:
Temperature: 120K, 300K; Pressure:
The ideal cycle simulation results are presented in Table 1. From Table 1, it can be easily
found that at the same condition the thermodynamic efficiency of the MARC is larger than that
of the single stage J-T cycle. The improvement of the thermodynamic performance is mostly as a
result of the exergy loss reduction in the heat exchangers. Shown in Table 1, the exergy loss of
the. heat exchanger is up to 41.2% of the total exergy input, and 57.1% of the total exergy loss of
the single stage cycle. Compared to the single stage, the exergy loss of all heat exchangers is
27.2% of the total exergy input, and 41.5% of the total exergy loss of the mixed-refrigerant
Auto-Cascade refrigeration cycle.
To know the thermodynamic performance of the MARC with distributing heat loads, a
situation of methane liquefaction process is calculated, illustrated in Fig. 4.
Figure 4. One situation of distributing heat load: liquefaction of the methane.

The simulation results of extra heat loads in the two cycles are presented in Table 2. From
Table 2 one can find that the Auto-Cascade refrigeration cycle is more suitable in the distributed
heat loads situation than single stage cycle. The total exergy gained in the MARC is about 28.9%
of total input exergy, which is larger than the 19.4% in the single stage. Compared with Table 1,
the merit of the MARC with distributed heat loads is obvious. However, more efforts should be
made in the extensive study in the future.
CONCLUSION AND DISCUSSION

This paper presents an initial effort on the study of the mixed-refrigerant Auto-Cascade
refrigeration cycle. The thermodynamic models of the multicomponent zeotropic mixed-
refrigerant Auto-Cascade cycle and its components have been established. Based on the first and
second laws of the thermodynamics, an exergy analysis method and the optimization model have
been developed. A computer program based on this method has been established. The
calculations show that the MARC can reduce the exergy loss of the heat exchangers, and
therefore can improve the thermodynamic performance of the cycle compared to a single-stage
J-T cycle. Further study shows that this cycle can reduce the inequality of the heat capacity of
the high and low pressure flows in the counter current flow heat exchanger, and finally reduce
the end throttling exergy loss. However, to verify this conclusion, more efforts should be made in
the future theoretically and experimentally.
On the other hand, one must realize that it is only the appropriate arrangement that can
improve the performance of the cycle. That is, the cycle may be too complex to be employed in
some situations. Therefore, whether to use a single stage or multi-stage refrigeration cycle
depends on the general considerations.
In this paper, an effort is made to analyze the thermodynamic performance of the mixed-
refrigerant Auto-Cascade refrigeration cycle theoretically. Based on this analysis, an
experimental set-up is being fabricated, and further experimental testing is planned.
ACKNOWLEDGMENT
This work was financially supported by the National Natural Sciences Foundation of China
under the contract No. 59706002.
REFERENCES
1. W. J., Podbielniak, U.S. patent 2041725 (1936).
2. A. P. Kleemenko, “One flow cascade cycle,” Proceedings of Xth International Conference of
Refrigeration, l-a-6(1959), pp. 34-39.
3. Dale J. Missimer, “Refrigerant conversion of Auto-refrigerating Cascade (ARC) systems,” Int. J.
Refrig. Vol.20, No.3 (1997), pp. 201-207.
4. B.C. Luo, M.Q. Gong, Y. Zhou, “Experimental Investigation of a Mixed-Refrigerant J-T Cryocooler
Operating from 30 to 60K,” Advances in Cryogenic Engineering, Vol.45 (2000), pp. 315-322.
5. W. A. Little, “Kleemenko cycle coolers: Low cost refrigeration at cryogenic temperatures,”
Proceedings of IECE17 (1998), pp. 1-9.
PLANCK Sorption Cooler Initial Compressor
Element Performance Tests
Christopher G. Paine, Robert C. Bowman Jr, David Pearson, Michael E.

Schmelzel, Pradeep Bhandari, Lawrence A. Wade
Jet Propulsion Laboratory, California Institute of Technology

Pasadena CA 91109-8099
ABSTRACT
PLANCK is an ESA-led mission to map the cosmic microwave background using
bolometric and heterodyne instruments; both instruments require cooling, one to ~20K, the other
to 0.1K. JPL is developing a sorption-based hydrogen cooler to provide 18—20 K cooling to the
two instruments. The system mass and power limitations require tradeoffs in thermal design.
To demonstrate achievement of an acceptable design, three compressor elements of a flight-
like configuration have been built and are undergoing characterization and life tests. The
compressor elements utilize a alloy for reversible hydrogen storage, resistive
heaters, and an aluminum foam matrix for thermal uniformity, all contained within a high-
pressure vessel. A gas-gap switch provides adjustable thermal isolation. Initial results indicate
that hydriding alloy bulk and surface contamination levels are insignificant, and that reversible
storage capability is near theoretical limits. We report on static and dynamic thermal
characteristics of the compressor elements, and gas supply characteristics related to operational
modes of a cooler. We then indicate what further characterization will be performed.
I. INTRODUCTION
The Jet Propulsion Laboratory (JPL) is developing continuous-duty hydrogen sorption
coolers for ESA’s PLANCK satellite mission, which will launch in 2007 to continue
investigations of the cosmic microwave background. PLANCK’s Low Frequency Instrument
employs High Electron Mobility Transistor amplifiers cooled to ~20 K; the sorption cooler will
supply this environment directly, extracting about 1 watt of heat at this temperature. The High
Frequency Instrument utilizes bolometers operating at 0.1 K; the sorption cooler provides ~150
mW at 18 K as precooling for a mechanical helium J-T refrigerator and an open-cycle dilution
refrigerator cooling chain. Information regarding the thermal architecture and many other aspects
of the PLANCK mission is available at the ESA website1 and in a recent publication2.
Continuous-duty sorption cryocoolers employing hydrogen as the working fluid and metal
hydrides as the sorbent have been discussed by Freeman3. The PLANCK mission requires
approximately 24 months cumulative operation, including ground testing. These requirements
necessitate pushing the state of the art in hydride compressor lifetime4 by a factor of ~100,
within the constraints of mass and power appropriate to a space mission. Both efficiency of

532 SORPTION CRYOCOOLER DEVELOPMENTS
operation of the J-T expander and degradation rate of the metal hydride sorbent increase with
increased working pressure of the hydrogen fluid; optimal performance is thus a tradeoff between
efficiency of operation and degradation rate of components.
The various components of the PLANCK cooler have been described at length1. In this report
we focus only on the active component of the compressor system, the compressor element, which
stores the hydrogen working fluid, releases it at high pressure to generate cooling via Joule-Thomson
expansion, and absorbs the hydrogen at low pressure to maintain the cold temperature at the liquid
hydrogen reservoir. Compressor elements have been fabricated, utilizing materials and construc-
tion techniques expected to optimize the cooler performance for the mission and to minimize deg-
radation in the hydride sorbent during cycling. The purpose of the work described herein is the
operation of these compressor elements under conditions similar to those of a functioning cooler, to
validate the design and construction techniques, gain information about the operational characteris-
tics, and quantify the degradation rate to be expected in long-term operation. This information will
be used to iterate to an optimal design for the flight cooler system.
We begin with an explanation of the operation of a single compressor element, followed by a
description of the apparatus that has been constructed to test the various characteristics of the com-
pressor element. Results of the characterizations are given, with discussion of the ramifications of
those characteristics for development of a flight cooler. Finally we indicate the direction of future
characterization and modifications planned for the test apparatus.
II. COMPRESSOR ELEMENT DESCRIPTION AND OPERATION

Figure 1 is an assembly view of a single compressor element (CE). The central portion of the
outer shell is aluminum, alloy 6061-T6. The outer 75 mm on each end are 316L stainless steel,
joined to the central portion via inertial welding prior to final machining. The inner vessel has a
316L wall 1.22 mm thick, containing the hydriding alloy distributed in an alumi-
num foam matrix [ERC Corp, Oakland CA] to enhance thermal uniformity. Three penetrations of
the inner vessel allow egress and return of the hydrogen gas, access to the single thermocouple
inside the inner vessel, and power to the resistive heater elements, also located inside the vessel.
The inner vessel is supported by the outer shell only at the ends, via axial pins. All assembly joints
are performed in an argon atmosphere using automatic orbital tube welding equipment [Weld Logic
Corp].
Figure 1. Assembly view of single compressor element, showing outer shell, inner vessel, gas-gap
switch-port, and three access penetrations. For details of assembly see Reference 5.
PLANCK SORPTION COOLER INITIAL COMPRESSOR TESTS 533
Figure 2. Van t’Hoff plot of equilibrium isotherms for the alloy chosen. The boundary of the trapezoidal
region shows the path in P:x space followed by the compressor element: Heatup begins in the lower right,
with large H:M and low pressure, moves to high temperature with small change in H:M as high-pressure
hydrogen gas fills the ullage volume, remains at constant pressure with decreasing H:M as hydrogen is
desorbed at constant pressure, returns to low pressure with small change, and returns to initial condition
as hydrogen is reabsorbed at constant pressure.
The inner vessel and outer shell are separated by a radial space of 0.75 mm which forms the
conductive link of a gas-gap thermal switch5,7; the two sections are thermally decoupled when
the annular region separating them is evacuated. The facing surfaces are gold plated [Epner
Technology Inc, Brooklyn, NY] to reduce radiative thermal transport and to chemically passivate
the surfaces.
Hydrogen is reversibly stored in the alloy, with equilibrium pressure and
temperature as shown in Figure 2. The cycle of operation of a single CE in a cooler can be
described briefly as follows: The vessel containing the hydrogen storage media is heated to drive
off hydrogen at high pressure, typically ~50 bar, which is then cooled to well below the inversion
temperature of 205 K (typically to ~50—60K), and passes through a Joule-Thomson expansion
orifice to collect as liquid hydrogen. When the source vessel is depleted of hydrogen it is cooled;
it is then able to reabsorb hydrogen gas boiling off from the liquid reservoir, typically at a
pressure ~0.3 bar. An arrangement of checkvalves permits a CE to desorb into a manifold at high
pressure, then absorb hydrogen supplied by another identical CE. Four or more such CEs are
required, operating sequentially, for continuous-duty cooling.
At 283 K, with a H:M ratio of 5.3, the equilibrium pressure of over the hydriding alloy is
~0.4 bar. The hydriding alloy could absorb more hydrogen, but only at a higher equilibrium
pressure. Since it is the reabsorption pressure of that determines the minimum temperature of
the liquid reservoir, the maximum allowable H:M ratio is determine by the required cold tip
temperature. Typically the H:M ratio is varied over the relatively flat plateau region, as indicated
in Figure 2 by the trapezoidal box.
To begin desorbing from the CE, the gas-gap region is evacuated and the inner vessel
heated. Flow of gas from the vessel will begin when the internal pressure exceeds the external
system pressure, and will continue so long as energy is added to desorb the gas and maintain
vessel temperature, until the alloy is depleted of hydrogen. As small H:M the equilibrium
temperature at fixed pressure rises steeply, thus in practice the desorption is terminated prior to
complete depletion, at a temperature set by practical considerations. In this application the
maximum temperature is derived from considerations of alloy stability6.
With the desorption phase of the cycle complete, the gas-gap region is filled to ~30 torr of
gas, thermally coupling the inner vessel to the outer shell and thence to a heat sink. When the
equilibrium pressure over the depleted alloy decreases below the external system pressure,
which is approximately the vapor pressure over the liquid reservoir, hydrogen begins to
resorb into the alloy, and continues so long as the heat of absorption is removed via the gas-gap
conductance. Eventually the alloy returns to the initial state of temperature, pressure, and H:M
ratio.
III. LIFECYCLE TESTS: APPARATUS DESIGN, TEST REGIMEN, RESULTS

As described in References 4 and 6, high temperature and pressure of is the condition
most deleterious to the hydriding alloy, for the effect on both intrinsic and extrinsic degradation.
To simulate the conditions a CE would experience in the operation of a cooler, we have
constructed an apparatus to cycle CEs through the four phases described above. The apparatus
faithfully reproduces the heatup, desorption, and cooldown phases in all respects, allowing
variation of all relevant parameters and conditions.
Three CEs were fabricated for this effort, known as the LifeCycling testing. All materials,
handling, and construction techniques were identical to those intended for the flight devices.
Each of the three CEs was bolted to an aluminum heat sink, which was maintained at constant
temperature via a circulating water bath and chiller [Neslab Instruments Inc. RT-6221]. A Type
K thermocouple inside each CE, and another Type K and a Si diode [Lakeshore Cryotronics
DT470] in close proximity to the CE on each heat sink provided thermometry. Readout wiring
for the thermometers and electrical power for the resistive heaters passed through the wall of a
vacuum chamber via hermetic connectors. To simulate the operational conditions, the chamber
was evacuated during testing.
The gas-handling portion of the LifeCycle apparatus is fabricated entirely of 316L stainless
steel tubing electropolished to the standards of the semiconductor industry. All valves and
components were cleaned to similar specifications. Prior to exposing the hydriding alloy to the
system, all tubing and valves were vacuum baked with an oil-free pumping system [Leybold
TOPS turbomolecular pump] to remove contaminants, primarily water with some hydrocarbons
and argon. The contaminant level was monitored with an RGA system [Stanford Research
RGA200], which verifies elimination of impurities before exposing the alloys to hydrogen.
Figure 3 shows schematically the layout of all plumbing, expansion tanks, valves and
pressure sensors, for the testing of the three CEs. There are three identical channels, one for each
CE; the only common connection is the gas source for the gas-gap switches. Characterization
proceeded as follows; static thermal characteristics of the CE structure prior to introduction of
hydrogen: storage capacity of the hydriding alloy: dynamic thermal characteristics of the CE
with hydrogen as per operational conditions: and behavior of the CE during desorption, with
particular emphasis on the reversible storage capacity and kinetics of gas desorption. These four
sets of tests, and the interpretation of the results, are detailed below.
A. STATIC THERMAL TESTS: PARASITICS AND GAS-GAP CONDUCTANCE

Thermal coupling, conductive and radiative, between the inner vessel and the outer shell
was measured as a function of and of gas-gap pressure. Conductance was determined by
applying constant power to the inner vessel, which was either evacuated or filled with argon at
about 1 bar, and observing the equilibrium temperature of the inner vessel as the outer shell was
Figure 3. Schematic layout of the LifeCycle apparatus.
held at constant temperature. Research-grade hydrogen was supplied to the gas-gap switch
volume via a mechanical regulator and a Brooks 5866 pressure controller, with the pressure set
via computer interface. The control and data acquisition system will be detailed in the section
on dynamic thermal characterization.
With gas-gap pressure less than torr, the parasitics are a small component of the total
system power, <2%. The parasitic losses are comfortably smaller than required for CEs in the
flight cooler. Conductance of the gas-gap during the resorption phase, when the is
important for the effective operation of the cooler cold end. Results from these tests are shown
in Figure 4. The gas-gap conductance exhibits the expected saturation as the mean free path in
the gas approaches the wall separation; the measured conductance at gas-gap pressure of 30
torr is 1.4 times that required to extract the heat of reabsorption at the required pressure of
hydrogen over the alloy.
B. HYDRIDE ALLOY STORAGE CAPACITY

The quantity of hydriding alloy included in each CE is known to within 0.1%, as is the
theoretical capacity for reversible reaction with hydrogen. The amount of hydrogen that the hy-
dride is capable of storing is then an indicator of the quality of the hydride and of the assembly
and handling techniques. Prior to introducing hydrogen, the ullage volume of the inner vessels
and of all other components of the has handling system were determined to <1%. Each of the
CEs was initially reacted three times with high purity hydrogen prior to the start of the thermal
Figure 4. Conductance of gas-gap thermal switch. CE 1 is lower points, CEs 2 and 3 are upper
points. The solid line is measurement on a test sample.
tests. These hydrogen capacities are summarized in Table 1 and demonstrate excellent
uniformity and complete activation.
C. DYNAMIC THERMAL TEST: HEATUP AND COOLDOWN CHARACTERISTICS
Of the energy dissipated by the resistive heaters to achieve and maintain desorption
temperature of the hydride vessel, the only portion that contributes to heat lift at cryogenic
temperature is the heat of desorption of hydrogen from the alloy. Of the energy remaining, the
enthalpy added to the hydride vessel and its contents is the most significant by a large margin. It
is thus beneficial that the inner vessel be designed with minimal thermal mass, consistent with
mechanical strength requirements. As the heatup phase must be completed within a time
determined by the operational requirements of the cooler, the thermal mass is the most
significant factor in determining the electrical input power required. The specific heat of the
hydriding alloy at high H:M and high temperature, where the pressure is significantly larger than
1 bar, has not previously been measured accurately. Here we report only the measurements
indicating the power necessary to heat the inner vessel to desorption pressure in the required
time; more detailed results related to the hydride properties will be reported elsewhere.
In an operating cooler of the PLANCK design, a CE will start the heatup phase from close
to the sink temperature of ~280 K, with H:M of ~5..3. The heatup test began with this condition,
with the gas-gap pressure below torr. Constant voltage was applied at the resistive heater via
a power supply [Hewlett Packard 6032A]; voltage and current were monitored each second, and
Figure 5. Heatup, desorption, and cooldown time-temperature, time-pressure, and mass flow rate
profiles for compressor elements operated in 667 second phases, with constant power during heatup of
240 W, constant power of 120 W during desorption. For a single cycle, the pressure oscillates as the
controller adjusts to hold 735 psi; the temperature continues to increase slowly as the CE desorbs,
reaching a maximum of 450 K. The mass flow (units not shown, lower curves) initially oscillates, then
undergoes variation of several percent during the desorption phase.
Joule power calculated. In the constant-voltage mode, the power varied by less than 1% during
the heatup cycle. The operational requirement, derived from the cooling requirements for this
cooler design, is that the heatup phase take no more than 667 seconds. The voltage required to
bring the inner vessel to a temperature at which the pressure of hydrogen is 50 bar, the pressure
at which hydrogen will begin to flow out of the CE and into the remainder of the system, is
determined by iteration to meet this requirement. Figure 5 shows the heatup phase time-
temperature and time-pressure profiles. The electrical power required is in all cases within the
system requirement of 240 watts for a CE in heatup phase.
Extraction of enthalpy from the inner vessel, as it cools following desorption phase, is
likewise an important parameter in the operation of the cooler. The hydride vessel must cool
sufficiently to begin reabsorbing hydrogen, else the cooler will cease to function. Although prior
analyses7 have found this requirement to be a non-critical contributor to the gas-gap design, it is
significant in that the heat transferred to the sink will influence the thermal environment
experienced by other components of the cooler. We have thus measured the time required for the
depleted CE to cool to the absorption temperature following removal of Joule power from the
heater. The measurement begins with a CE with H:M = 2.0 and T = ~450 K, as would be the
case at the completion of the desorption phase. The gas-gap pressure is quickly raised to 30 torr,
and the resulting temperature of the inner vessel and heat sink monitored. The time-temperature
profile is also shown in Figure 5. The thermal power conducted to the sink is initially large,
~1000 W, and decreases rapidly as the enthalpy is extracted. It is clear that the cooling time
requirement of 667 seconds is easily satisfied by the existing design at this gas-gap pressure, and
is expected to be adequate for gas-gap pressures as low as 20 torr.
D. HYDRIDE DESORPTION-ABSORPTION CYCLING: CAPACITY AND KINETICS
The greatest portion of the LifeCycling Apparatus is dedicated to an examination of the
behavior of the CEs during the desorption cycle, wherein the CE supplies hydrogen to the high-
pressure portion of the cooler system. The intent is to examine in detail the dynamic behavior of
the hydriding alloy, the containing vessel, and the entire compressor element; and, once the
dynamic behavior is well-understood, to operate the CE for an extended period to determine
usable lifetime. The apparatus constructed for this purpose runs each CE through the heatup
phase at constant voltage or power, desorbs the hydrogen gas at constant pressure, cools the CE
without allowing absorption, then transfers the hydrogen back into the CE. This simulates the
cooler operation in all except the final, absorption, phase, in that the absorption does not occur at
constant pressure as would be the case in a functioning cooler. At the inception of this program,
the degradation rate attainable for the hydride alloy in a temperature cycling CE was the least
well known, and thus highest risk, component of the design space. Separate tests are underway to
evaluate the intrinsic degradation rate8; the purpose of the tests reported here is evaluation of the
implementation, which includes the alloy, the materials and handling techniques used in the
fabrication of the CE, and the effects of cycling the hydride through many desorption—
absorption cycles. Operation of the flight cooler over the totality of ground testing and flight
mission will require approximately 18,000 cycles of each CE. We plan to operate the individual
CEs for at least this many cycles.
Figure 3 shows schematically the three independent channels of the LifeCycle apparatus.
The CEs connect through the wall of the vacuum chamber via a valve pair. As the temperature of
the CE inner vessel rises during heatup phase, the hydrogen gas flows to the pressure control
valve [Brooks 5866RT], which contains a PID servo loop and a mass flow sensor. A capacitive
pressure transducer [MKS 850B] upstream of the valve monitors gas pressure and supplies this
information to the control valve; when the upstream pressure exceeds the setpoint of the control
valve, the valve opens to allow hydrogen to escape into the expansion volume. Pressure in the
expansion volume is monitored by an identical transducer. The PID loop maintains constant
desorption pressure in the CE until the end of desorption phase, when power to the CE heater is
removed and the gas-gap switched to conductive mode. At the end of the cooldown phase, an
electropneumatic bypass valve [Nupro HBVCR4] is opened, allowing hydrogen to flow from the
expansion volume back into the CE, returning the system to the initial condition, whereupon the
cycle is repeated.
Hydrogen for the gas-gap switch volume is supplied from a bottle of research grade The
pressure in the gas-gap is set by another 5866RT operating in downstream control mode. The
gas-gap is evacuated via a molecular drag pump [Hovac DRI-2].
The cycling of the apparatus is controlled via a PC running Lab VIEW 5.1 under MS
Windows NT 4.0. The power supplies are controlled via GPIB, data is collected via two A—D
converter cards [National Instruments PC-MIO-16XE-50], and the setpoints for the desorption
and gas-gap pressure control valves is set from the computer via a D—A card [National
Instruments AT-AO-10]. Timing of the electropneumatic valves is controlled by the software,
and the valves actuated via a relay card driven from the RS232 bus [National Instruments CB-
50LP and SC-2062]. The software collects data from all of the pressure and mass flow sensors
and all the thermometers, to allow monitoring of all aspects of the cycle.
Considerable care was exercised in the design of the cycling apparatus and in particular in
the operational modes of the control system, to guarantee that no dangerous condition could exist
as a result of single-point component failure or computer problems. In addition to software
checks on parameter ranges, a keep-alive system was implemented such that if the software were
to cease operation, the CE heater power would be interrupted within seconds. This was
accomplished with a Watchdog relay [Macromatic TR-51822-08TV05], which requires a
periodic reset signal, generated in this case from the computer software. Tests, intentional and
otherwise, have demonstrated this system to be reliable.
Figure 5 shows the temperature, pressure, and mass flow as a function of time, for a single
compressor element over three cycles. Prior to testing, it was expected that the desorption phase
would proceed with constant power supplied to the resistive heaters, with nearly constant mass
flow proceeding from desorption of hydrogen at constant pressure as most of the heater power
goes into heat of desorption. However, it was found that under these conditions we experience
larger mass flows during the initial portion of the desorption phase, where the H:M is larger and
Figure 6. Detail of the power profile utilized during desorption tests to achieve mass flow constant to
within 1%.
the heat of desorption smaller, reduced mass flow at midcycle, and increased mass flow at the
end of desorption phase as the temperature increases and specific heat is at a minimum. Profiling
the desorption phase power has reduced the output mass flow variation to <1%, which has
important ramification for the operation of the cooler system but is beyond the scope of this
report. Figure 6 is a detail of the power profile utilized during desorption, and the mass flow
attained in this manner.
E. FUTURE WORK
Not reported here is the rate at which the alloy reabsorbs hydrogen, as the apparatus is not
instrumented for that measurement. In an operational cooler, the reabsorption takes place at
essentially fixed mass flow, with the absorption pressure determining the cold head temperature.
As discussed above, the reabsorption phase is not believed to be critical for hydride degradation.
At the time of design, it was not expected that this apparatus would be required to provide
detailed examination of the reabsorption phase; changes in the program have made it advisable to
examine this portion of the cycle in greater detail, and modifications to one channel of the
apparatus are under way to permit detailed examination of the reabsorption phase. These results
will be reported at a later time.
Upon completion of these modifications, the remaining two CE channels will be dedicated
to long-term cycling to determine the usable lifetime of compressor elements under operational
conditions; this is a critical parameter for the cooler design, as it determines by what factor BOL
performance must be above the EOL requirements, with ramifications on launch mass, power
requirements, and thermal radiator sizing. These tests are expected to begin in the summer of
2000.
IV. SUMMARY
We have constructed an apparatus to test three compressor elements of a design to be used
in the PLANCK sorption cooler. Initial results for the thermal behavior of the compressor
elements as regards power and mass requirements indicate significant margin in the design. The
hydriding alloy shows storage capacity essentially at the theoretical limit, no
evidence of intrinsic or extrinsic contamination, and excellent reversible storage capacity. The
hydride results are thus a strong indication that the starting material and handling and construction
techniques are adequate to provide for good hydride life. The pressure-temperature function relat-
ing the rate of gas released from the CEs has proven to be more complex than anticipated, but the
reasons for this are understood and operational methods to take advantage of the behavior are under
development. In all, we have gained significantly more information from this effort than was ex-
pected at inception, and anticipate considerably more as the apparatus capabilities are extended.
ACKNOWLEDGMENT
The work described was performed at the Jet Propulsion Laboratory, California Institute of
Technology, under contract with the National Aeronautics and Space Administration.
REFERENCES
1. http://astro.estec.esa.nl/SA-general/Projects/Planck/.
2. Collaudin, B., Passvogel, T., “The FIRST and PLANCK ‘Carrier’ missions. Description of the
cryogenic systems,” Cryogenics, vol 39 (1999), pp. 157-165.
3. Freeman, B.D. et al, “Progress towards the development of hydrogen sorption cryocoolers for space
applications,” Int. I. Hydrogen Energy, vol 22, no. 12 (1997), pp. 1125-1134.
4. L.A. Wide, J.J. Wu, S. Bard, T. B. Flanagan, J.D. Clewley, and S. Lou, “Performance, Reliability,
and Life of Hydride Compressor Components for 10 to 30 K Sorption Cryocoolers,” Advances in
Cryogenic Engineering, vol. 39, Plenum Press, New York (1994), pp. 1491-1498.
5. Wide, L.A. et al, “Hydrogen Sorption Cryocoolers for the PLANCK Mission,” Advances in Cryo-
genic Engineering, vol 45, Plenum Press, New York (2000).
6. Bowman Jr, R.C., et al., “The effect of tin on the degradation of metal hydrides during
thermal cycling,” J. Alloys. Compounds, vol. 217 (1995), pp. 185-192.
7. Prina, M., Bhandari, P., Bowman, R.C., Paine, C.G., Wade, L.A., “Development of Gas gap heat
switch actuator for the Planck sorption cryocooler”, Proceedings of 10th Internaltional Cryocooler
Conference (1999).
8. Bowman, R.C., Lindensmith, C.A., to be published.
Sizing and Dynamic Performance Prediction Tools
for 20 K Hydrogen Sorption Cryocoolers
P. Bhandari1, M. Prina1, M. Ahart2, R. C. Bowman1, L. A. Wade1

1
Pasadena, CA 91109
2
Princeton University,
Princeton,NJ 08544
ABSTRACT
Two continuous operation 18 K/20 K sorption coolers are being developed by the Jet
Propulsion Laboratory (JPL) as a NASA contribution to the European Space Agency (ESA)
Planck mission that is currently planned for a 2007 launch. The individual sorption coolers
will each be capable of providing a total of about 200 mW of cooling at 18 K and 1.4 Wat 20 K
given passive radiative pre-cooling at 50 K. The hydrogen sorption coolers will directly cool
the Low Frequency Instrument HEMT amplifiers to approximately 20 K and will also serve to
intercept parasitics and pre-cool a RAL 4.5 K closed-cycle helium J-T cooler to 18 K for the
separate High Frequency Instrument. To design the Planck sorption coolers a general sizing
model and a detailed performance prediction model have been developed.
In this paper we describe the underlying relationships that determine the required size of a
18/20K hydrogen sorption cooler, and the approach utilized to model and optimize designs. In
addition an initial comparison with the first data on a compressor element test will be pre-
sented while experimental data for the rest of the cooler components are not available and will be
validated in the near future.
INTRODUCTION
Planck will carry two instruments: the High Frequency Instrument (HFI) and the Low Fre-
quency Instrument (LFI) that will observe and image the full sky in nine spectral bands between 30
and 857 GHz. The hydrogen sorption coolers will directly cool the Low Frequency Instrument
HEMT amplifiers to approximately 20 K and will also serve to intercept parasitics and pre-cool a
RAL 4.5 K closed-cycle helium J-T cooler to 18 K for the separate High Frequency Instrument.1
The PLANCK individual sorption coolers will each be capable of providing a total of 200 mW of
cooling at 18 K and 1.4 W at 20 K given passive radiative pre-cooling at 50 K. The cooler design
concept is presented along with the detailed design and performance predictions.
The sorption cooler performs a simple thermodynamic cycle based on hydrogen
compression up to 5 MPa, hydrogen gas pre-cooling by the three radiators at about 170 K,
100 K and 50 K, further cooling due to the heat recovery by the cold low pressure gas stream,

expansion through a J-T expansion valve and evaporation at the cold stage. The key element of the
20 K sorption cooler is the compressor, an absorption machine that pumps hydrogen by thermally
cycling several sorbent compressor units.
The principle of operation of the sorption compressor is based on the properties of a unique
sorption material which can absorb large amounts of hydrogen at relatively low
pressures and low temperature, and which will desorb to produce high-pressure hydrogen when
heated in a limited volume. Heating of the sorbent is accomplished by electrical resistance heaters
while the cooling is achieved by thermally connecting the compressor element to a radiator. The
system is periodically cycled between heating and cooling cycles, producing high-pressure gas
intermittently. In order not to lose excessive amounts of heat during the heating cycle, a heat switch
is provided to alternately isolate the sorbent bed from the radiator during the heating cycle, and to
connect it to the radiator thermally during the cooling cycle.
The mass of hydrogen stored in a unit mass of sorbent at equilibrium is plotted on the horizon-
tal axis vs. the pressure on the vertical axis in Fig.1. Each curve is an “isotherm” at constant tem-
perature. The sorbent in contact with hydrogen gas can exist at equilibrium anywhere on one of the
isotherms. If hydrogen is added to or removed from the sorbent at constant temperature, the system
will move along an isotherm. Absorption of hydrogen at constant low temperature is represented by
the line Heating from “low T” to “high T” in a confined space is represented by
Continuously removing hydrogen from the sorbent at constant temperature is represented by
Cooling depleted sorbent from high to low temperature in a confined space is represented by
As a sorption compressor element (i.e. sorbent bed) is taken through these four steps in a cycle,
it will intake low pressure hydrogen and output high-pressure hydrogen on an intermittent basis. If
the high-pressure hydrogen is pre-cooled with radiators to below the inversion temperature and
then expanded through a Joule-Thomson expansion orifice (J-T) the high-pressure gas will par-
tially liquefy, producing liquid refrigerant at low pressure for sensor systems. Heat from the sensors
evaporates liquid hydrogen, and the low-pressure gaseous hydrogen is re-circulated back to the
sorbent for compression. This system can be depicted schematically as shown in Figure 1. Such a
system will periodically produce liquid refrigerant in a cycle involving the four steps shown in
Figure 1.
In order to produce a continuous stream of liquid refrigerant, we employ several such sorption
beds and stagger their phases so that at any given time, one is desorbing while the others are either
Figure 1. Schematic of a single compressor cooler and idealized compression cycle.

PREDICTION TOOLS FOR 20 K SORPTION CRYOCOOLERS 543
heating, cooling, or re-absorbing low pressure gas. In such a system, there is a basic clock time
period over which each step of the process is conducted.
The compressor assembly shown in Figure 2 is composed of six identical sorption
compressor elements, each filled with metal hydride and provided with independent heating and
cooling which will be described later. Each compressor element is connected to both the high
pressure and low pressure sides of the plumbing system through check valves, which allow gas
flow in a single direction only. The check valves are indicated on the schematic as single
arrows, which indicate the direction of gas flow through them. In addition to the compressors,
there are five one-liter high-pressure stabilization tanks connected to the high pressure side of
the system to damp out oscillations of the high pressure gas, and a low pressure stabilization
sorbent bed to damp out pressure fluctuations of the low pressure gas.
Refrigerant travels from the compressors through a series of heat exchangers and radiators,
which provide pre-cooling to approximately 50 K, through the J-T expander. The compressor
assembly is comprised of the six compressor elements, high-pressure stabilization tanks, the low
pressure stabilization bed, check valves, and manifolding. The compressor assembly mounts
directly onto the heat rejection radiator. This radiator is sized to reject the cooler input power at
270 K +10 K/-20 K. A single compressor element is comprised of two concentric cylinders
closed with end caps. The inner of these tubes contains 626 g of hydride
material and the outer forms a vacuum jacket around the inner cylinder. This vacuum jacket is
used as a gas-gap heat switch4.
Figure 2. Planck sorption cooler schematic. All the components are shown: three radiator and
four heat exchangers, the cold heads and the compressor. The points shown are used in the
energetic balance of the components. The dynamic model uses these points to transfer
information (pressure, temperature and mass flow of the gas) between the cooler components. The
arrows in front of each bed sorption bed are check valves, allowing flow only in the arrow
direction.
The heater passes through the hydride material and is designed to uniformly distribute heat
to ensure a high degree of temperature uniformity with a maximum temperature of 465 K. Heat
transfer to the hydride is provided by aluminum foam that fills the inner cylinder and makes
tight contact with the heater. A vent tube passes through the center of the hydride material. The
heater, thermocouple, and vent-tube leads run from the end cap of the inner cylinder to that of
the outer cylinder. The space between the inner compressor vessel and the outer tube assembly
is used as a gas-gap thermal switch.
FUNDAMENTAL RELATIONSHIP FOR COOLER SIZING

The sorption cooler is essentially a Joule-Thomson refrigeration machine, that is a
thermodynamic machine that pumps heat from the cold stage at 20 K to the lowest temperature
pre-cooling sink. The other heat exchangers and radiator are used to minimize the heat load on
the coldest radiator and therefore to reduce its temperature. Cooling is obtained by the
evaporation of liquid hydrogen expanded in a bi-phase region by a J-T restrictor. It should be
observed that thermodynamically the cooling power at the 20 K stage is dependent on the
temperature of the lowest stage pre-cooler and not on the higher stage shields temperatures. A
simple heat balance of the fluid at the high pressure inlet and the low pressure outlet of the heat
exchanger 4 and the heat inserted in the cold head, as shown in Figure 2, yields the ideal mass
flow rate required for the cooler as follows:
where is the mass flow rate of the working fluid, is the heat inserted in the cold
head by the detector – required cooling capacity, and are the enthalpies of the hydrogen
leaving (low pressure) and entering (high pressure) the counter flow cold heat exchanger 4,
respectively.
Based on the efficiency of the counter flow heat exchange 1, 2, 3, and 4, and on the
temperature of the shields, it is possible to determine the temperature and pressure in the points
shown in the schematic. The heat loads at the radiators are consequently derived from these
parameters by calculating the enthalpies from the pressure and the temperatures.
The compressor operation is based on the absorption and desorption of hydrogen in
hydrides. The stoichiometric composition of the hydride is a design variable and
can be used to optimize the power, mass and life of the cooler. The basic cycle for Planck
consists of six equally timed phases for each of the six beds: 1) heating a bed to pressurize the
stored gas to a pressure level of about 50 bars with no mass removal from the bed (line B
in Fig.l), followed by desorption (phase 2, line in Fig.1) at constant pressure (and
nearly constant temperature) via a check valve into the high pressure ballast tank. This is then
followed by phase 3, which essentially cools the bed to close to the radiator temperature via the
filled gas-gap (line in Fig.1). The final phases (4, 5 and 6) allow the cold bed to absorb
the gas exiting the cold head via the low pressure check valve and manifold (line in
Fig.l). Each successive bed is phase shifted by one phase to allow for continuous operation of
the cooler without breaks to produce constant flow rates and cooling powers as a function of
time.
The assumed cycle time of 4000 s for the Planck 18/20K cooler (phase time
667 s), is then used to calculate the total mass of hydrogen to be released by a compressor
element during the desorption phase by simply multiplying the mass flow rate by the phase
duration.
The minimum quantity of hydride needed to release is the product of and the
reversible specific hydrogen storage capacity of the hydride (approximately 1% wt.). The sizing
program also accounts for the mass of required to fill the ullage (void) volume in the
compressor element and in the lines up to the check valve during the heat-up phase. The
compressor element is not fully packed with hydride for two major reasons: there is a ~ 20 %
volume expansion of the hydride when it absorbs hydrogen up to its maximum capacity; and,
the hydride is inserted in the compressor element as powder with a practical packing density
limit. Based on those observations and on previous experience, the packing density is assumed
to be 45%. Hence, the rest of the volume (~55%) represent the ullage volume due to the
hydride powder not filling the entire available volume. The gas used for pressurization of
the compressor element can be considered as a wasted capacity of hydride because it is used to
maintain the high pressure in the element and not to flow gas into the cold head. Based on
the compressor element baseline design it can be accounted as a 5 to 10% loss in the full
reversible hydride capacity.
The EXCEL sizing model is used to go through a parametric cooler design based on simple
inputs of the radiator temperatures, the required refrigeration and the compressor radiator
temperature. Assuming the cycle described and a baseline mechanical design, the model
performs the optimization of compressor power and mass by selecting the most suitable hydride
alloy and the baseline design utilizing it. The optimization is done by sizing various coolers for
a range of high pressures and hydride alloys followed by selecting the smallest power and mass
combination. The optimization also accounts for the degradation of the hydrogen storage
capacity in the hydrides as a function of temperatures. It should be observed that the strength
and the limit of the model is its baseline parametric design. For example, the model assumes
that the compressor element cylindrical vessel is made of stainless steel vessel and based on this
assumption it calculates its dimensions. No other options are considered in terms of vessel
material and shape.
In table 1 the results of a sample program run on the Planck cooler are reported.
DYNAMIC MODELING PERFORMANCE OF THE COOLER

The thermodynamic model does not give any information on the dynamic behavior of the
cooler such as pressure variation upstream and down stream the J-T expander, the temperature
distribution in the compressor element or mass flow rate variations. A dynamic model of the
cooler has been built to predict the coolers dynamic performance and to evaluate its overall
behavior by modeling the cooler as a set of components exchanging hydrogen gas characterized
by pressure, temperature and gas flow as shown in Figure 2. The check valves are characterized
by a gas flow-delta P curve across the valves. At each time step, a numerical solution of the
component set determines the hydrogen mass flow on the J-T, the quantity of hydrogen
absorbed and desorbed by each compressor element.
The dynamic model inputs are:
a) The spacecraft interfaces,
b) The heater power sequence, and
c) The mechanical design of each components. The guidelines derived from the EXCEL
sizing models were used to design the main cooler components such as the check
valves, the filters and the connecting tube sizes as mentioned earlier.
For the sorption cooler, the “heart” of the dynamical model is the compressor modeling.
The equations describing the compressor element can be summarized in
Equation 3a represent the heat transport in the compressor element. The heat accumulated
or released by an infinitesimal volume is the net conducted (the first term) in addition to the heat
externally supplied to the volume. Conduction through the element is accomplished by the
combination of aluminum foam and hydride powder. The source term has been split in two
terms: the external heat source Q (heat supplied by the heaters) and the heat generated by the
chemical reaction:
(4)
The heat generated can be accounted by the product of the variation of the hydrogen
content (Cone) in the infinitesimal mass and the heat of reaction Equation
3b describes the quantity of hydrogen generated or absorbed per unit time in the compressor
element. The overall mass flow rate generated or absorbed by the entire compressor element is
the net sum of the rate of the concentration range over the compressor element. It should be
observed that during each phase (described in the introduction) both the direct and reverse
reaction can occur simultaneously in different spots of the compressor element, depending on
the temperature and hydrogen concentration distributions.
Equation 3c describes the compressor’s interaction with the tank and the high and low
pressure manifolds. When the pressure inside the element is below the minimal pressure to open
the low pressure check valve it is possible for the gas coming out of the liquid reservoirs to flow
through the valve and to be absorbed in the bed. The gas flow is described by the characteristic
behavior of the check valve (gas flow vs. pressure drop) described by
When the pressure inside the compressor element is above that value but not high enough to
open the high pressure check valve there is no interaction of the compressor element with the
rest of the cooler. Once the pressure exceeds the minimum pressure required to open the high
pressure check valve the high pressure hydrogen gas starts to flow from the compressor element
to the tank. The gas flow through the check valve is determined by the high pressure check
valve characteristics. Equation 3d is a mass balance inside the compressor element between the
hydride and the ullage volume around it, based on the ideal gas law.
The set of the differential equations presented are very difficult to solve with numeric
methods. The equations describe a time dependent parabolic integrative-derivative problem with
the complication of the terms introduced by the pressure and concentration variation. The
approach used to reach the numerical solution is obtained by considering the concentration
variation as a result of the temperature variation. A finite difference method (SINDA) with a
forward time step method has been used to predict the temperature distribution inside the
compressor element. Based on the temperature variation between two time steps the quantity of
gas absorbed independently by each node in the element is evaluated by assuming that within the
time steps the gas generated or absorbed did not interact with the rest of the hydride but only
with the ullage volume in the bed. For each of the node temperature values obtained after the
time step, the corresponding saturation pressure of that node is calculated using the hydride
properties. The difference between the saturation pressure so obtained and the ullage pressure at
the end of the previous time step is then used to compute the mass of hydrogen lost or gained by
each node to the ullage within that time step. Equation 4 is then used to compute the resultant
heat lost or gained due to the reaction of the hydride and the hydrogen gas. The heat generated is
then applied as an additional heat load (gain or loss) to the nodes where the reaction occurred for
the next time step.
A simplified analogy could be helpful in the explanation of the pressure modeling. Each
hydride node can be considered to be a small pore of a sponge. Between two time steps each
pore interacts with the rest of the sponge only thermally and reaches a new thermal equilibrium.
Associated with this equilibrium a bubble of gas is generated or absorbed by the pore. The
pressure of the gas liberated or absorbed by the sponge is obtained by letting the bubbles
equilibrate at the end of each time step. The time step used in this model was chosen after
reaching an asymptotic solution of the predicted performance as a function of the time steps.
Based on the symmetries in the compressor element only a section of it has been modeled
assuming a uniform heat and hydride distribution inside it. In Figure 3 the map of the compressor
nodes is presented.
The node density is increased in the vicinity of the compressor element ends to take into
account the end effects of the supports. Based on the measured properties of the hydride the
compressor performance has been predicted in terms of temperature and concentration
distribution, pressure, and hydrogen mass flow rate as a function of time. Figure 4 shows the
model’s prediction and the results obtained for a compressor element [a detailed description of
the test performed on the element used for the model validation can be found in 6].
Figure 3. Cross section and side view of the compressor element nodes locations. There are three radial
layers of hydride and six axial layer, with a higher layer density towards the end cap of the vessel.
Figure 4. Comparison between test results on a compressor element and SINDA model prediction under
the same operative conditions (power profile, radiator temperature, desorption pressure limit).
It is interesting to note that the predicted power has been correlated accurately after fine-tuning
of the measured heat losses in the ends and by radiation in the gas gap area. The end cap heat losses
were measured on an identical compressor element unit with a bar of copper inside, instead of
hydride and aluminum foam, to minimize the thermal gradients within it.
The initial performance of the sorption cooler have been predicted using the model in terms of
mass flow variations, average bed temperature and pressure fluctuations as shown in Figure 5.
Figure 5. Compressor element performance pressure, average bed temperature and average hydrogen
concentration predictions for three cycles.
An independent model of the cryostat with all the heat exchangers has also been developed.
Once the compressor model has been completely validated with the test data, the cryostat model
will be integrated with the rest of the model. The entire cooler model will then be validated by
the test data on the complete breadboard cooler that is being built at JPL.
CONCLUSIONS
The sizing and performance prediction models have proven to be powerful tools in the
design and performance evaluation of the cooler and of its components. The latest data from the
compressor element performance test has been used to fine-tune the compressor model. Use of
these models has provided great insight into the inner workings of the cooler and will prove to
be even more powerful tools once the integrated model of the entire cooler is completed.
ACKNOWLEDGMENT
The research described in this paper was carried out by the Jet Propulsion Laboratory,
California Institute of Technology, under a contract with the National Aeronautics and Space
Administration.
REFERENCES
1
L. A. Wade, P. Bhandari, R. C. Bowman, Jr., C. Paine, G. Morgante, C. A. Lindensmith, D. Crumb, M.
Prina, R. Sugimura, D. Rapp, “Hydrogen Sorption Cryocoolers for the Planck Mission”, Adv. Cryogenic
Engineering Vol. 45
2
R.C. Bowman, J.S. Cantrell, T.W. Ellis, T.D. Flanagan, J.D. Clewley, S.Luo, “Hydriding Properties of
the Pseudo-Binary Alloys Hydrogen Energy Process X, edited by D.L Block, T.N.
VeziroGlu (Int. Assoc. Hydrogen Energy, Ural Fables, FL, 1994) pp 1199-1207
3
B.D. Freeman, E.L. Ryba, R.C. Bowman Jr., J.R. Phillips, “Progress toward the Development of
Hydrogen Sorption Cryocoolers for Space Applications”, Int. J. Hydrogen Energy, 22, (1997) 1137
4
M. Prina, P. Bhandari, R. C. Bowman Jr., C. G. Paine, L. A. Wade, “Development of Gas Gap Heat
Switch Actuator for the Planck Sorption Cryocooler”, Adv. Cryogenic Engineering Vol. 45
5
S. Luo, W. Luo, J.D. Clewey. T.B. Flanagan, L.A. Wade, “Thermodynamic studies of the H
system form x=0 to 0.5”, J. Alloys Comp. 231 (1995) 467-472
6
C.P. Paine, R.C. Bowman Jr., D. Pearson, M.E. Schmelzel, P. Bhandari, L.A. Wade, “PLANCK
Sorption Cooler initial Compressor Element Performance Test”, Cryocoolers 11, Plenum Press, New
York, 2001
165 K Microcooler Operating with a Sorption
Compressor and a Micromachined Cold Stage
J.F. Burger, H.J. Holland, J.H. Seppenwoolde, E. Berenschot,

H.J.M. ter Brake, J.G.E. Gardeniers, M. Elwenspoek and H. Rogalla
University of Twente, Faculty of Applied Physics
P.O. Box 217,7500 AE Enschede, The Netherlands
ABSTRACT
This paper presents the fabrication of a microcooler consisting of small stainless steel
sorption compressor cells, micromachined silicon check valves, and a micromachined cold stage
that incorporates glass-tube heat exchangers. The design, fabrication and experiments on the
different elements are described. Two compressor cells were thermally cycled to investigate the
dynamic behaviour. The cold stage could reach a stable temperature of 169 K with a cooling
power of about 200 mW.
INTRODUCTION
During recent years a rapid development has taken place of LT-electronics and especially of
superconducting devices. However, there exists a gap between this development and the
availability of enabling technologies that are essential for the commercialization of LT-
electronics1. These enabling technologies are low-cost, highly reliable cryogenic refrigeration
systems and energy-efficient cryogenic packaging of the LT-electronic device with the cryogenic
refrigerators. Much effort is currently put in the development of such reliable and cheap coolers,
but typically small systems are still rather large in terms of size (> 1 kg) and cooling power
(> 1 W). Low-temperature applications requiring very little cooling power, such as a single chip
with a low noise amplifier or a superconducting magnetometer, would benefit from very small
closed-cycle coolers. Such coolers do not exist and in this respect it was suggested3 that
micromachining techniques can be attractive for miniature cooler components, such as heat
exchangers, check valves or compressors. As a typical example in this respect, this paper
presents realized components for a microcooling system, which consists of a sorption compressor
with a micromachined cold stage.
The sorption/Joule-Thomson (JT) cycle was identified as a potential candidate for the
development of a microminiature cooler aiming at a cooling power in the range of 10 mW at
80K3. The advantage of this cycle is the absence of wear-related moving parts, except for some
check valves. This facilitates scaling down of the system to very small sizes, it minimizes
electromagnetic and mechanical interferences (which is important for many applications), and it

Figure 1. Sorption cooler set-up. Figure 2. Schematic of compressor cycle
offers the potential of a long life time. The operating principle of sorption coolers is discussed in
several publications4,5, and is briefly summarized below.
A sorption cooler consists of a compressor unit, a counterflow heat exchanger, and a JT
expansion valve, see Fig. 1. Compressed gas corning out of the compressor unit is cooled to the
environmental temperature after which it is fed into the first counterflow heat exchanger. A
(thermoelectric) precooler such as depicted in Fig. 1 may be applied to improve the system
performance. Next, the compressed refrigerant is expanded in the JT valve to provide
refrigeration. The low pressure refrigerant then returns through the recuperative heat exchangers
to the compressor unit. The compressor unit basically contains four sorption cells and several
check valves to control the gas flows. Low and high pressures are generated by the cyclic ad- and
desorption of a working gas on a sorption material, which is accomplished by cooling and
heating of the sorption material. The gas can either be physically adsorbed onto or chemically
absorbed into various solids. Usually, heating occurs with an electrical heater and cooling is done
with a heat-switch between the sorption cell and a heat sink on the outside (typically a gas-gap
switch). A compressor cycle of one cell is schematically shown in Fig. 2. The cell is heated
during sections A and B, and cooled during C and D. During sections A and C both valves of the
cell are closed, and the cell is in a regenerating phase. During sections B and D one of the valves
is opened; the cell generates a high pressure gas flow out of the cell during B, and a low pressure
gas flow into the cell during D. It is important to notice that the valves are passive check valves;
the cycle is driven by the temperature induced pressure variations of the compressor cells. In our
case, a complete compressor cycle typically takes about 10 minutes.
A thermodynamic analysis of sorption coolers was presented elsewhere6. In that analysis the
Coefficients of Performance (COP) of the compressor and the cold stage were modelled
separately, both for quasi-static conditions. It followed that the compressor performance strongly
depends on the sorption material, the gas, the compressor material, and the temperatures and
pressures the compressor operates on. Microporous activated carbon with a high internal surface
area was chosen as the sorption material for a first demonstrator cooler, in combination with
xenon (or alternatively ethylene) as a refrigerant gas that is able to cool to about 165 K. This
sorption cooler can be used as a first stage of a two-stage sorption cooler that is able to cool to
80 K. For a low pressure of 1 bar, an optimum compressor performance was found for
compression to 10–20 bar and compressor cells operating between 300 and 600 K. However, it
165 K MICROCOOLER WITH A SORPTION COMPRESSOR 553
appeared that the cold stage required somewhat higher pressures (> 60 bar) to perform well. It
was shown that precooling of the gas by means of a miniature thermoelectric cooler connected
somewhere halfway the counterflow heat exchanger results in a dramatic improvement of the
performance at a pressure of 20 bar, resulting in an overall cooler performance of about 3%. We
aim at a cooling power of 200 mW, requiring a compressor input power of less than 10 W and a
xenon massflow of about 2.5 mg/s (or 0.5 mg/s for ethylene). These parameters define the
requirements for the individual cooler elements.
SORPTION COMPRESSOR
Requirements. (1) The ratio of the cylinder thermal mass to the sorber thermal mass should
be minimal in order to maximize the compressor COP6. The cylinder thermal mass includes the
thermal mass of the heater, thermocouple and other stainless steel components. These thermal
masses should, therefore, be minimal as well. (2) A gas-gap heat switch is required
between the inner sorption cell and the outer cylinder. Conduction and radiation losses between
the inner and outer cylinders should be minimized. Furthermore, the gas-gap vacuum space
should be perfectly sealed (all welded) and without any materials that could show outgassing so
that it can be reversibly actuated by a metal hydride sorber7. (3) The sorption cell should contain
highly porous charcoal with a grain size of minimal to maintain a low pressure drop over
the sorption bed. As a consequence, a filter is required in the gas connection line to keep the
sorber material in the cell. (4) The fabrication should be as simple as possible.
Design and realization. Fig. 3a shows a cross-sectional drawing of one sorption
compressor cell with integrated gas-gap heat switch. All components are made of stainless steel
316L, except the thermocouple sheath which is made of stainless steel 304. The different
components are assembled by laser-welding and high-temperature soldering.
The E-type thermocouple consists of a outer diameter stainless steel sheath which
contains the two thermocouple wires, isolated by a MgO ceramic powder8. A
thermocouple with a length of 0.5 meter and a total resistance of 380 ohm was coiled and fixed
as depicted in Fig. 3a. Long-duration experiments showed that these thermocouples are also
suitable to be applied as an isolated heater, so that in fact a combined heater/thermocouple
results. In this heater experiment, five thermocouples were cycled 3.105 times between 300 K and
600 K with an input power of 30 W; no significant degradation of the heater or thermocouple
fucntion was observed.
Figure 3. (a) Cross sectional drawing of one sorption compressor cell, (b) Photo’s of some elements.
The inner sorption cylinder has a wall thickness of and is laser-welded to the end-
caps. One end-cap is integrated with the sintered filter the gas supply tube, the
thermocouple and its support. The thermocouple is connected to the outside world via the gas
supply tube, which reduces the number of feedthroughs through the gas-gap vacuum space. The
inner sorption cell is suspended by two ‘wheels’ with three thin spokes, see Fig. 3a. This
construction facilitates accurate radial positioning, longitudinal flexibility to account for thermal
expansion differences between both cylinders and low thermal conduction losses between both
cylinders. The top closure of the outer cylinder contains an extension to reduce the thermal
conduction losses via the longer gas supply tube. The lower closure can be supplied with a thin-
film ZrNi hydrogen actuator to control the hydrogen pressure in the gas-gap. In the current
version of the compressor that is used for the described experiments, a gas supply in combination
with a small vacuum pump is used to control the pressure in the gas gap. Fig. 3b shows
photographs of the compressor components and an assembled compressor cell.
Experiments. To study the behaviour of the sorption compressor cells and to compare it
with our models, cycling experiments were carried out with a combination of two compressor
cells. Fig. 4a shows the experimental set-up that is designed to operate the two cells through ab-
and desorption cycles as depicted in Fig. 2, where the two cells are 180 degrees out of phase. The
metering valve was used to create the desired flow restriction. A data acquisition system was
used to measure and store the following parameters: the electrical input powers, temperatures and
pressures of the two sorption cells, the pressures in the two gas-gap heat switches, and the
massflow from cell one to two (the sensor could only measure flow in one direction). To run the
experiment automatically, the data acquisition system could control the following parameters: the
input powers of the two cells, the active valve between the two cells that controls the massflow
period, and the active valves that control the pressure in the two gas-gap heat switches. An
ethylene gas supply system and a vacuum pump were connected to the system via two valves; in
this way the system could be pumped and purged before filling with ethylene gas. Care was taken
in minimizing the dead volumes of the system because dead volumes deteriorate the compressor
performance6. However, the active valve and the massflow sensor still contributed significantly
to the dead volume, 30% and 50% of the sorption cell volume, respectively. In a real sorption
compressor these components are not present, and a much better performance can be expected.
Fig. 4b shows a typical measurement of the temperatures and pressures in the two cells. Also
depicted are the state of the active valve between the cells and the massflow from cell 1 to 2. The
measurement clearly shows the alternating periods of the two cells. The phases A-D of Fig. 2 are
indicated in Fig. 4b on the pressure-curve of cell 2. The sudden pressure drop and pressure peak
of cell 1 are caused by the dead volume of the active valve. Furthermore, it can be seen that the
Figure 4. (a) Measurement set-up for flow-tests between two sorption compressor cells, (b) Typical
measurement results (see text).
pressure was cycled between 2 and 15 bar instead of the required 1 and 20 bar. The experiments
were carried out on a slightly older version of the sorption cell with a NiCr heater wrapped
around the cell instead of the more robust thermocouple-heater. The NiCr heater was limited in
terms of temperatures and input powers that could be applied. However, the obtained pressures
and massflows matched with the modelled values6, and it is expected that the required pressures
can be obtained with the current design of the sorption cells, especially because the extra dead
volumes are not present anymore.
CHECK VALVES
Requirements. The requirements for the check valves are derived from the cooler parameter
settings as given in the introduction: (1) The valves have to stand periodically pressure
differences up to the maximum pressure difference of 20 bar; 50 bar is chosen as a safe limit. (2)
Gas leakage in the closed direction is a loss factor and should be kept below a small fraction of
the normal gas flow in forward direction (< 1% of 2.5 mg/s). (3) The valve should close
immediately as soon as the pressure difference reverses. Consequently, the valve should be
normally closed but without a significant spring force to reduce the pressure drop in forward
direction. (4) In forward direction, the massflows through the high- and low-pressure valves are
the same but the absolute gas pressures and densities are a factor of 20 different. In both cases,
the pressure drop should be a small fraction of the absolute pressure. By designing a check valve
Figure 5. (a) Design impression of the high-pressure check valve. Spring dimensions (height x width x
length): boss: 1 x 1 mm. (b) Photo of top wafer with 4 KOH-etched bosses (part of a
10-valve manifold), (c) Photo of the backside of 2 manifolds with 10 valves, prior to separation of the
manifolds and integration with glass-tubes, (d) Photo of a section of the bottom wafer.
that fulfills this requirement for the low-pressure gas, the same valve will certainly also fulfill the
requirements for the high-pressure gas. (5) From experiments with a large scale sorption cooler it
appeared that leaking check valves (of a commercial type) were the major cause of malfunction
of the system9. This was explained by contamination of the valve seat. From this we concluded
that prevention of possible contamination is a major requirement.
Design and realization. The valve consists of a thick plate (boss) suspended by four thin
springs10, see Fig. 5. The thin springs behave like single clamped beams, thus facilitating high
deflections which are required to obtain a low pressure drop in the forward direction. When the
valve is in forward direction, the gas is able to flow through the holes surrounding the springs.
The entire valve construction, including the interfacing gas lines, is made out of two silicon
wafers that are covered by two glass wafers, all bonded together. The design is such that the gas
lines on both sides of the valve can cross in the upper and lower wafers. This is required for the
construction of an integrated valve manifold that has only six connections to the outside world:
four to the compressor cells and two to the cold stage. The valve seat is made out of the polished
surface of the wafers, facilitating a perfect fit and alignment of both sides of the valve seat.
Stiction of the fragile spring beams to the bottom wafer is prevented by etching a cavity in the
bottom wafer under the beams. Stiction or bonding of the valve seat is prevented by application
of a nitride coating11. The maximum allowed deflection of the boss can be selected by choosing
an appropriate thickness of the boss relative to the wafer thickness. A sieve is constructed
in the inflow line of the valve to trap possible contamination. The sieve is made of a row of
pillars standing in the channel. Fluidic and mechanical modelling was applied to find the proper
dimensions of the valve construction; this is described in more detail elsewhere10.
Experiments. Experiments with samples bonded with Pyrex wafers showed that the
valves could withstand pressures up to 65 bar in closed direction with a leakage flow that was not
detectable At higher pressures, glass wafer 1 burst. Samples bonded with thicker
Pyrex wafers were tested up to 125 bar and did not burst at all.
With the valves in forward direction, the pressure drop was measured as a function of the
massflow. For an absolute nitrogen pressure of 2 bar and a massflow of 1 mg/s, a pressure drop
of approximately 30 mbar was measured. It was shown that this pressure drop was more or less
equally divided over the connecting tubes, the filter and the valve itself.
To test the influence of contamination on the closing behaviour of the valve, small particles
of were blown through a valve without integrated filter. At bar a leakage flow
was measured of 1.2 mg/s, which is very significant (in the same range as the forward flow). This
illustrates the importance to trap contaminant particles in a filter.
Finally, a long duration experiment was done involving many forward/closed switchings.
Fig. 6 shows that the leakage flow reduced from an initial 3% to 0.5% of the forward flow after a
few thousand cycles. This behaviour is probably caused by a small particle or irregularity in the
Figure 6. Leakage flow as a function of the number of forward-closed cycles of a valve.

Figure 7. T-S diagram for the cooling cycle depicted in figure 1.
valve seat, which is hammered into the seat during the repeated loading. Other samples showed a
similar leakage behaviour or did not show leakage at all.
COLD STAGE
Requirements. The requirements for the cold stage are determined by the cooler design of
Fig. 1. The associated T-S diagram of the cold stage is shown in Fig. 7, the numbered states in
the cycle correspond to the numbers in Fig. 1. In the diagram, isobars and isenthalps are given, as
well as the enthalpy changes that occur in the counterflow heat exchangers, condenser and
evaporator, assuming ideal operation. Under that condition, the cooling power is given by
To obtain the required gross cooling power of 200 mW, an ethylene massflow of
0.5 mg/s is needed. The majority of the enthalpy change is created during condensation
from state 3 to 4. A proper design of the condenser is, therefore, a major requirement so that the
two-phase fluid reaches full condensation at state 4.
To obtain a significant cooling power, the thermal losses on the cold stage (conduction and
radiation) should be limited, for instance below 50 mW. Because silicon exhibits a very high
thermal conductivity, a different material with a low conductivity such as glass should be used
for the construction of the counterflow heat exchangers. In constrast, it is attractive to use a high
conductivity material (silicon) for the condenser and evaporator to maintain uniform
temperatures, independent of the supplied thermal loads.
Design and realization. Fig. 8 shows the design of the cold stage, which is made of three
micromachined silicon components with two glass-tube counterflow heat exchangers in between.
All three silicon parts are constructed by fusion bonding of two thick silicon wafers in
which channels and spaces are etched by KOH etching. After processing and separating these
silicon samples, the glass tube heat exchangers are glued into the samples, after which integration
with a small vacuum flange follows – see Fig. 9. The applied glass tubes are commercially
available12 and have inner/outer diameters of 0.25/0.36 mm and 0.53/0.67 mm, respectively. Two
glass support tubes are added parallel to the two counterflow heat exchangers to add more
stability to the system.
The left silicon part is called the ‘splitter’, and makes it possible to supply separate
connection lines to the high and low pressure channels of the first counterflow heat exchanger. In
the condenser, the high pressure fluid is able to condense in the long meandering channel which
Figure 8. Top: cold stage design, bottom: details and cross sections of the condenser and restriction.
is etched in the silicon. The low pressure fluid returning from the second counterflow heat
exchanger directly connects in the condenser to the low pressure annulus of the first counterflow
heat exchanger. The high pressure fluid that enters the restriction/evaporator flows through an
etched channel to the entrance of the flow restriction, which typically consists of a 4 mm wide,
shallow channel with a length of about 3 mm. Because the restriction/evaporator is made of
a high conductivity material, the high pressure fluid that enters the silicon part easily cools to the
low temperature of the evaporator before it enters the flow restriction. This is represented by the
dotted extra heat exchanger in Fig. 1 and step 5-5a in the T-S diagram of Fig. 7. The low pressure
liquid that exits the flow restriction (in state 6a in Fig. 7) is collected in the liquid bath of the
evaporator, which connects at the top side to the low pressure annulus of the second counterflow
heat exchanger. In the present design, the orientation of the cold stage is important: the low
pressure exit of the liquid bath should be oriented vertically upward so that gravity keeps the
liquid in the bath and only vapor exits the liquid bath, except when the liquid bath is full. Both
the flow restriction and the boiler structure are supported by pillars to prevent excessive bending
stresses due to the high gas pressures of 20 bar that may be present. Furthermore, the condenser
and the restriction/evaporator contain an etched channel that can be used to insert an external
thick thermocouple to measure the temperature of these cooler parts.
The high pressure inner glass tubes are glued into the condenser and restriction/evaporator
via a so-called ‘glue hole’, whereas the outer glass tube is glued at the entrance of the sample, see
Fig. 8. This construction facilitates a robust separate connection of the high and low pressures.
The surface area of the condenser fits approximately to the surface of a two stage
thermoelectric cooler, which is a MI2012T-type fabricated by Marlow Industries13. In the same
way, a thermal load (some device) can be attached to the restriction/evaporator. In the current
design, however, for test purposes a thin film heater is deposited on the restriction/evaporator and
on the condenser. This heater can be used to study the behaviour of the cold stage. To limit the
radiation load on the cold stage, a gold layer is deposited on both sides of the cooler.
Figure 9. (a) Part of a structured silicon wafer with splitter, condenser and restriction/evaporator
samples, (b) Completely assembled cold stage connected to a vacuum flange with wiring to the thin film
heaters. The coin (Dutch 25 cents) is 18 mm in diameter.
Experiments. Different versions of the cold stage were characterized in a small vacuum
chamber. The high pressure input of a cold stage was connected to an ethylene gas bottle, with a
zeolite filter in between to trap possible contaminant gases. A data-acquisition system was used
to measure and store the following parameters: the temperatures of the condenser and evaporator,
the input powers into the TE-cooler and heaters, the input high pressure and the massflow going
into the system and coming out of the system. A typical measurement is depicted in Fig. l0a. The
important steps are numbered in the figure and are briefly discussed below.
(1) The high pressure is applied to the cooler and the evaporator cools to a temperature
slightly below ambient. The cooling power due to the enthalpy change produced at ambient
temperature, is too small to overcome the thermal losses and reach lower temperatures.
(2) The TE-cooler is started and temperature-controlled at 238 K (-35 °C), 6 K below the
condensation temperature at 20 bar. (3) The fluid starts to condense in the condenser. For a short
period, the ingoing massflow exceeds the outcoming massflow to compensate for the liquid
volume that is now being collected in the condenser. Also, the evaporator starts to cool rapidly
because of the increased cooling power of the liquid ethylene. (4) The high pressure fluid now
starts to enter the restriction as a liquid, which increases the massflow because of changing fluid
density and viscosity. (5) The low pressure boiling temperature is reached and the boiler starts to
fill with low pressure liquid. This explains why the ingoing massflow exceeds the outcoming
massflow for a while. (6) Because the cooling power exceeds the applied thermal load, two-
Figure 10. (a) Typical measurement of a start-up of the cold stage, (b) Step-by-step increase of the heater
load on the evaporator. At the boiler contents evaporates and starts to rise.
phase fluid exits the evaporator. Capillary effects explain the variations in the outgoing
massflow. These variations were dependent on the fraction of the two phases.
In the measurement of Fig. 1 Ob, the thin-film heater on the evaporator is used to determine
the net cooling power. During the initial increase of the heat load, a small temperature increase
can be observed, which may be explained by the temperature gradient that is present between the
wall of the boiler and the boiling liquid. At the liquid in the boiler starts to
evaporate, which indicates that the total thermal load (heater + losses) exceeds the cooling
power. The first two rapid temperature increases are probably caused by capillary forces, which
play an important role because of the small channel dimensions.
CONCLUSIONS
A sorption cooler operating with small stainless steel sorption compressor cells,
micromachined check valves and a micromachined cold stage was designed, and elements are
constructed and tested. Manufacturing of the sorption cells was simplified by the application of a
sheathed thermocouple that could also be applied as heater element. The cold stage could
reach a stable low temperature of 169 K with a gross cooling power of about 200 mW, provided
that no capillary pressure drops occurred in the return line.
ACKNOWLEDGEMENTS
This research is supported by the Dutch Technology Foundation (STW).
REFERENCES
1. Nisenoff, M., Cryocoolers and high temperature superconductors: advancing toward commercial
applications, Cryocoolers 8, Plenum Press, New York (1995), pp. 913-917.
2. G. Walker and R. Bingham, Micro and nano Cryocoolers: speculation on future development, Proc.
of the 6th Int. Cryocooler Conf. (1990), pp. 363-375.
3. Burger, J.F., ter Brake, H.J.M., Elwenspoek, M., Rogalla, H., Microcooling: Study on the
application of micromechanical techniques, Cryocoolers 9, Plenum Press, New York (1997),
pp. 687-696.
4. Burger, J.F., Components for a cryogenic microcooler: a vapor compression cold stage operating
on a sorption compressor, Ph.D. Thesis, Twente University, The Netherlands (2000).
5. Wade, L.A., An overview of the development of sorption refrigeration, Adv. in Cryogenic Eng. vol.
37 (1992), pp. 1095-1106.
6. Burger, J.F., Holland, H.J., Wade, L.A., ter Brake, H.J.M., Rogalla, H., Thermodynamic
considerations on a microminiature sorption cooler, Cryocoolers 10, Plenum Press, New York
(1999), pp. 553-564.
7. Gardeniers, J.G.E., Burger, J.F., van Egmond, H., Holland, H.J., ter Brake, H.J.M. and Elwenspoek,
M., ZrNi thin films for fast reversible hydrogen pressure actuation, Proc. of Aktuator 2000, Bremen,
Germany (2000).
8. Omegaclad shielded thermocouple wire, Omega, Inc., Stamford, Connecticut 06907-0047, USA.
9. S. Bard, J. Wu, P. Karlmann, C. Mirate, L. Wade, Component reliability testing of long-life sorption
coolers, Proc. of the 6th Int. Cryocooler Conf. (1990).
10. J.F. Burger, M.C. van der Wekken, E. Berenschot, H.J. Holland, H.J.M, ter Brake, H. Rogalla,
J.G.E. Gardeniers and M. Elwenspoek, High pressure check valve for application in a miniature
cryogenic sorption cooler, Proc of IEEE MEMS 99 (1999).
11. C. Gui, Direct wafer bonding with chemical mechanical polishing, Ph.D. Thesis, Twente University
(1998).
12. Supelco/Sigma-Aldrich Corp., Bellefonte, PA, USA.
13. Marlow Ind. Inc., 10451 Vista Park Road, Dallas, Texas 75238-1645, USA.
Double Stage Helium Sorption Coolers
L. Duband
CEA/DRFMC/Service des Basses Températures
Grenoble, France
ABSTRACT
We have developed two double stage helium sorption coolers. The first unit comprises a 3He
stage coupled to a 4He stage. This cooler can be operated from any cold heat sink with adequate
cooling power (typically at less than 5 K, and provides temperature down to 260 mK. It is
a self contained cooler with no moving parts, which can be simply thermally attached to a cold heat
sink. Its operation is controlled by two heaters. The second unit is a double stage 3He cooler with
the same features. It requires for its operation a cold heat sink at a temperature below 3 K, and
provides temperatures down to 204 mK. Both double stage coolers are currently being experimentally
characterized, and preliminary results are presented.
INTRODUCTION
Recent advances in detector technologies have boosted the interest in helium sorption
cryocoolers and in particular 3He coolers. These sub-Kelvin sorption coolers provide a wide
range of heat lift capability at temperatures below 400 mK. They have no moving parts, are
vibrationless, and can be designed to be self contained and compact with a high duty cycle
efficiency. Helium adsorption coolers rely on the capability of porous materials to adsorb or
release a gas when cyclically cooled or heated. Using this physical process one can design a
compressor/pump which, by managing the gas pressure in a closed system, can condense liquid
at some appropriate location, and then perform an evaporative pumping on the liquid bath to
reduce its temperature. Consequently, it requires a pre-cooling stage at a temperature lower
than the helium liquid-vapor transition, i.e. with enough cooling power. This can either
be a helium bath (typical for laboratory units, or on a payload like FIRST), or a mechanical
cooler (for instance a 3He Joule Thomson cooler or a pulse tube cooler).
CEA-SBT has acquired a large experience in sorption cooler technology.1 From the
developments intended for space applications, 3He and 4He laboratory units have been developed
and are regularly supplied to various customers. One potential drawback of the 3He sorption
cooler is indeed the requirement for their operation of a heat sink temperature below the 3He
critical temperature, i.e. 3.2 K. Thus, in general these coolers are operated from a pumped
helium bath leading to a reduced autonomy for the cryostat, induced vibration generated by the
mechanical pumps, etc.... Furthermore interests have been expressed in decreasing their
operating temperature toward 200 mK; such a cooler could then be associated with solid state
refrigeration methods (tunneling refrigerator) to provide temperatures down to 100 mK or below.
Consequently, we have developed two double stage helium sorption coolers. The first one can
provide temperatures below 300 mK from a 4.2 K bath. Unlike other systems,2 this cooler

562 SUB-KELVIN REFRIGERATOR DEVELOPMENTS
features two separated evaporators allowing for more flexibility in its operating modes. The
second double stage sorption cooler does require a heat sink below 3 K but has been designed to
extend the possible ultimate temperatures. Both cooler’s architecture can be easily upgraded to
be compatible with the space environment.
COOLERS DESCRIPTION
Fig. 1 shows the general architecture of both double stage units. Each stage closely
resembles a regular sorption cooler and comprises a sorption pump, a condenser and an
evaporator, with the exception of the 3He stage ( stage) which also includes a heat exchanger.
This heat exchanger is thermally coupled to the 4He condenser ( stage). The 4He evaporator is
3
thermally connected to the He condenser ( stage) and contains a diaphragm to prevent any
large loss due to the superfluid film during cold operation. One specific feature of this
architecture is the presence of two separate evaporators to allow for more efficient operation as
described below. The 3He / 3He unit is identical but does not require a diaphragm. The following
description focuses on the 4He / 3He unit, but operation of the second cooler is similar.
Once both stages have been charged with helium gas and permanently sealed, the cooler is
thermally coupled to the cold heat sink (we assume 4.2 K) and operation can begin. Two
operating modes are possible, in mode [A] the cycling is as follow :
- Both sorption pumps are heated (to 40 and 50 K, respectively, for 3He and 4He). At these
temperatures most of the helium gas previously adsorbed into the activated charcoal, is released
and consequently the pressure inside the cooler increases. The 4He condenser is thermally
coupled to the heat sink, and once the pressure reaches the saturated vapor pressure at that
temperature, liquid is condensed and then drops by gravity into the 4He evaporator.
- The heater on the 4He sorption pump is then turned off: its temperature decreases and the 4He
evaporator temperature drops to below 3 K; in the meantime the 3He sorption pump is
maintained at 40 K. The 3He evaporator follows closely the 4He evaporator mainly because of
convective effects, until the temperature at the 3He condenser (4He evaporator) falls below 3 K,
at which point 3He liquid is condensed and collected into the 3He evaporator. Both evaporators
continue to cool down, until the 4He stage runs out of liquid.
- At this point the 3He sorption pump heater is turned off, the pump cools down and the 3He
evaporator drops quickly to where it remains stable until the liquid 3He is in turn
exhausted. Note that it is also possible to turn off the 3He pump heater and let this one cool
before the 4He stage actually runs out of liquid. In this case the cooler is operated for a limited
amount of time with both evaporators cold (see hereafter).
Figure 1 . Schematic of the double stage cooler

DOUBLE STAGE HELIUM SORPTION COOLERS 563
While the cooler is running at or even while the evaporator is cooling down, it is
possible to recycle again the stage (mode [B]); this is made possible because the two
evaporators are decoupled. In this case the cooler is operated with both evaporators cold, leading
to a substantial gain on the performance of the stage (lower parasitic load) whose ultimate
temperature drops to When the stage runs out of liquid helium (which usually comes
first), the temperature of the stage simply rises back to Once all liquids are exhausted
the cooler can be recycled. Note that the full operation of the cooler is basically controlled by
means of two heaters.
The heat exchanger serves two functions: first, it precools the hot desorbed gas to 4.2 K,
preventing the corresponding enthalpy from being dissipated on the stage, and second, it
reduces significantly the conductive load due to the pumping tube; both features allow the design
of a more efficient stage.
The stage has been sized such that its cooling power, or rather its available cooling
capacity in terms of Joules, is enough to extract the gas enthalpy from 4.2 and about 2 K,
and then to evacuate the latent heat of condensation. Of course other constraints like the
maximum condensation efficiency, liquid loss during the cooldown process, conductive loads,
superfluid film, etc... have been taken into account in the sizing.
Operation of the unit is fairly similar, except both stages are condensed at the
same time and both sorption pumps are cooled simultaneously. The temperature of the first stage
drops to below 300 mK and acts as a thermal buffer or shunt for the second stage, which reaches
temperature significantly lower.
For practical reasons this unit has been sized on the basis of the unit, and is more a
demonstrator than an optimized cooler. Table 1 summarizes the main design specifications of the
two units.
EXPERIMENTAL PERFORMANCE
Both double stage coolers have been tested in a helium cryostat. To gain on the power
dissipated and on the kinetics of the cycle, gas gap heat switches3 have been used for the thermal
link between both sorption pumps and the cryostat cold plate. These gas gap heat switches
provide at 4.2 K ON and OFF conductance, respectively, of the order of 35 mW/K and
0.017 mW/K (ratio over 2000). The presence or absence of gas within the switch is controlled by
a miniature sorption pump, whose temperature thus determines the switch state (ON for
– typical input power of 2 mW, OFF for ). Thus, in this case, full operation of the cooler
is controlled by four heaters. The initial cooldown from room temperature down to 4.2 K takes
approximately half a day. The unit is shown in the test cryostat in Fig. 2.
double stage cooler

The curves of Fig. 3 display the time evolution of the temperatures during recycling. The
ON and OFF position of the heat switch are clearly seen. The complete recycling, i.e. to
requires about 2 hours. For this particular cycle, the input energies to the sorption pumps are
1440 and 834 Joules, respectively, for the and stages, with typical maximum input
powers of 500 mW. This corresponds to a consumption of about 0.9 liter of liquid from the
main cryostat.
Figure 2 . Picture of double stage cooler mounted in the test cryostat

(unit shown upside down).
This number can be substantially decreased by adjusting the sorption pumps temperature timing
(the heating of the pump can be delayed) and by optimizing the recycling sequence to get the
required hold time. In addition, this unit is a rather large system; its size could be reduced to gain
on the recycling time and input energy at the cost of course of a reduced autonomy.
In this mode [A], the stage reaches an ultimate temperature of 308 mK.
Once the cooler is at 308 mK, the measured hold time without any applied load is about 15
hours. As described before it is possible to recycle the stage during this phase (mode [B]). In
this case the evaporator cools to about 840 mK as the evaporator drops to 260 mK. This
recycling sequence is shown on Fig. 4. In this mode we have measured about 8 hours at 260 mK,
followed by 13 hours at 308 mK (once the stage runs out). The cooling power curves for
both operating mode are reported in Fig. 5a.
Figure 3 . cooler recycling curves – First mode (ie to 308 mK).

DOUBLE STAGE HELIUM SORPTION COOLERS 565
Figure 4 . cooler recycling curves – Second mode (ie to 260 mK).
double stage cooler

In this case a cold heat sink at a temperature below 3 K is required. For the present
experiment the helium cryostat was pumped down to about 2 K. Both sorption pumps are then
heated to 40 K and cooled simultaneously. The ultimate temperatures of both coolers were found
to be 310 mK and 204 mK respectively for the first and second stage. The cooling power is
shown on Fig. 5b. The hold time has not been measured yet, but is expected to be in excess of
one day, limited by the hold time of the first stage. Once the cooler is recycled it is possible to
bring back the test cryostat to 4.2 K. In this case the first and second stages run, respectively, at
350 mK and 225 mK. Table 2 summarizes the thermal performance of both double stage coolers.
The 3He/4He unit has been recently mounted on the cold end of a 4K pulse tube cooler
developed at CEA-SBT.4 The available cooling power at 4.2 K is compatible with the operation
of the double stage sorption cooler. This 300 K - 300 mK system is currently being tested, and if
successful will be the first liquid cryogen free cryocooler able to cover over 3 orders of
magnitude in temperature, from 300 to 0.3 K.
Figure 5a . cooler cooling power Figure 5b . cooler cooling power

for both operating modes (bath temperature : 1.7 K)
ACKNOWLEDGEMENT
We wish to thank L. Clerc, L. Miquet and V. Salvador who performed the experimental
testing of both coolers. L. Guillemet contributed to the overall design of the coolers and
produced all 3D drawings.
This development work has been supported by the European Space Agency (ESA) and
CEA-SBT internal funding. The double stage helium sorption coolers will find a direct
application in the S-Cam program, at the Space Science Department of ESA.
CONCLUSION
Two double stage helium sorption coolers have been designed and experimentally
characterized. The first unit can be operated from any cold heat sink at temperature below 5 K,
and provides ultimate temperatures down to 260 mK. The second unit requires a cold heat sink
below 3 K, but reaches temperatures down to 204 mK. Both units have no moving parts, are self
contained, do not require any mechanical or vacuum connections, and are fully controlled by
electrical heaters.
REFERENCE
1. Duband, L. and Collaudin, B., “Sorption coolers development at CEA-SBT,” Cryogenics, vol. 39
(1999), pp. 659-663.
2. Dall’Oglio, G., Fischer, W., Martinis, L., and Pizzo, L., “Improved refrigerator,”
Cryogenics, vol. 33 (1993), pp. 213-214.
3. Duband, L., “A thermal switch for use at liquid helium temperature in space borne cryogenic
systems,” Cryocoolers 8, Plenum Press, New York (1995), pp. 731-741.
4. Poncet, J.M., Ravex, A. and Charles, I., “Developments on single and double stage GM type pulse
tube cryorefrigerators,” Cryocoolers 11, Plenum Press, New York (2001).
Sub-Kelvin Sorption Coolers for
Space Application
L. Duband, B. Collaudin* and P. Jamotton**

CEA/DRFMC/Service des Basses Températures
Grenoble, France
* European Space Research and Technology Centre
Noordwijk, The Netherlands
** Centre Spatial de Liège, Belgium
ABSTRACT
The European Space Agency’s FIRST satellite will be a multi-user observatory looking at
the universe in the infrared and submillimeter wavelength range from 80 to It
accommodates three instruments, all of which have detectors or mixers operating in the range
0.3 K – 2 K. A large superfluid helium (He II) dewar provides a cold heat sink at around 1.8 K,
with the last cooling stage, down to 0.3 K, effected inside the instruments using a recyclable
sorption cooler developed at CEA-SBT.
Although these hardware developments are expected to begin in fall 2000, CEA-SBT, in
collaboration with CSL, was awarded in September 1998 an ESA Technological Research
Program (TRP) contract whose objective is to design, manufacture, test and qualify for space
an engineering model of a sorption cooler. In this framework, two prototype coolers, one
and one have been designed and fabricated. To benefit the future development of the
FIRST coolers, the prototype was sized according to the main specifications of the SPIRE
instrument. These two prototypes have been assembled, and the test plan, which shall lead to the
qualification of the coolers, is currently being performed. Preliminary results are presented.
In parallel with these projects and prior to the detailed design of the engineering models, a
laboratory sorption cooler prototype which can be run with or was developed to further
study specific aspects and answer various questions. For instance, convective effects that could
severely impact the performance during ground test operation, and superfluid film behavior were
both investigated. In addition, various measurements were performed on the titanium alloy Ta6V
and Kevlar 29. These results are presented as well.
INTRODUCTION
The FIRST/Planck mission of the ESA Horizon 2000 Science Program accommodates in
the carrier option two spacecraft for a joined launch on an Ariane 5 launcher into an L2 orbit in
2007. FIRST, the Far Infrared and Submillimeter Telescope, is a multi-user observatory type
mission performing astronomical investigations in the infrared and submillimeter wavelength
range. The FIRST payload consists of three instruments built by large scientific consortia: HIFI
(Heterodyne Instrument for First), PACS (Photo-conductor Array Camera & Spectrometer) and

SPIRE (Spectral and Photometric Imaging Receiver). The instruments are mounted in the FIRST
payload module. The spacecraft provides the environment for astronomical observations in the
infrared wavelength range from about 80 to 670 microns, requiring cryogenic temperatures for
the cold focal plane units. SPIRE will cover the spectral range using bolometric
detectors. The cooling of the detector arrays down to 300 mK will be effected by a sorption
cooler developed at CEA-SBT. Until recently, only the SPIRE instrument required a 0.3 K
sorption cooler, but recent events indicate PACS will also be using such a system.
These sub-Kelvin sorption coolers provide a wide range of heat lift capability at
temperatures below 400 mK. Helium adsorption coolers rely on the capability of porous
materials to adsorb or release a gas when cyclically cooled or heated. Using this physical process
one can design a compressor/pump which, by managing the gas pressure in a closed system, can
condense liquid at some appropriate location, and then perform an evaporative pumping on the
liquid bath to reduce its temperature.
Helium sorption refrigerators have no moving parts, are vibrationless, and can be designed
to be self contained and compact with a high duty-cycle efficiency. These features and the
expected reliability that follows, make them very attractive for space applications. In addition,
the thermal and mechanical interfaces are fairly simple. The links to ambient temperature are
limited to the heater wires used to drive the sorption pump and the heat switches. It should be
noted that this type of cooler is the last stage of a cooling system. It requires a pre-cooling stage
at a temperature lower than the helium liquid-vapor transition ( for and for ),
with enough cooling power. This can either be a helium bath (typical for laboratory units, or on a
payload like FIRST), or a mechanical cooler (for instance a Joule-Thomson cooler or pulse
tube cooler). The drawbacks are a relatively poor thermodynamic efficiency compared to Carnot,
due to the heat of adsorption, and non-continuous operation.
The above features, associated with recent advances in detector technologies, have boosted
the interest in these cooling techniques.
SORPTION COOLER – BASIC PRINCIPLE

The principle of operation of a sorption cooler has been described in numerous papers and
the reader is referred to the relevant publications.1,2 As shown in Fig. 1, the cooler is basically
made of six components designated as a sorption pump SP, an evaporator EV connected via a
pumping line to a thermal shunt TS comprising a heat exchanger, two gas gap heat switches HSP
and HSE, respectively, connected to the sorption pump and evaporator, and finally a support
structure SST. SP, EV, TS and the pumping line are assembled to form a unique component that
is the actual “heart” of the cooler. This component is held within the SST, which provides firm
mechanical support (launch environment) while minimizing any parasitic conductive load on the
cooler (low temperature environment).
Heat switches are then required for operation of the cooler. The two switches are used to
control the temperature gradient.
Figure 1. Schematic of a sorption cooler.

SUB-KELVIN SORPTION COOLERS FOR SPACE APPLICATION 569
During the condensation phase, the switches are set such that the sorption pump SP can be
heated to release the helium gas and such that liquid condensation occurs into the evaporator EV
maintained as the coldest point. The liquid is held in EV by capillary attraction inside some
porous material; both the surface tension and the vapor pressure provide forces that drive and
hold the liquid at the coldest point. Then the switches are set such that the sorption pump is
thermally grounded to the heat sink and such that the evaporator is thermally isolated. The
sorption pump provides evaporative pumping on the liquid helium bath whose temperature
quickly drops to sub-Kelvin temperatures.
Cryogenic switches based upon different physical mechanisms can be used. Gas gap heat
switches have been selected as the preferred design for the present projects. The gas gap heat
switch utilizes concentric copper cylinders separated by a small gap which is filled with or
emptied of He gas to achieve the switching action. The thermal separation between the two ends
is achieved by a thin-walled tube that also provides the mechanical support. The presence or
absence of gas is controlled by a miniature cryogenic adsorption pump that can be temperature
regulated. Thus, one of their main features is the absence of moving parts, and consequently,
operation of the cooler is fully controlled by three heaters. Note that the heat switch on the
sorption pump can be replaced by a passive thermal link, but at the cost of some additional
power on the cold heat sink. The design and operation of a system is slightly different,
mostly because of the superfluid film that creeps up the wall of the pump tube and can
dramatically degrade the performance. To reduce this effect, a diaphragm with a small hole can
be installed in the pumping line (see below).
ESA TECHNICAL RESEARCH PROGRAM

CEA-SBT, which has many years of experience in sorption cooler technology,2,3 was
awarded, in collaboration with CSL (Centre Spatial de Liege), an ESA Technological Research
Program (TRP) contract whose objective was to design, manufacture, test, and qualify for space
an engineering model sorption cooler. CSL is in charge of the project’s Product and Quality
Assurance for the overall duration of the project. In this framework, two prototype coolers, one
and one have been designed and manufactured and are currently undergoing
qualification. Table 1 summarizes some of the main specifications for the unit. The unit
was designed following the same overall specifications with the exception that it must be able to
thermally recycle the unit, i.e. the evaporator provides the cold heat sink for the
evaporator, and supplies enough cooling power to cool down and condense the gas.
Prior to the detailed design of these coolers, a sorption cooler prototype was built to assess
some specific aspects, namely the convective effect and the superfluid film behavior. This
prototype features the components previously described, associated with a simplified laboratory
support structure.
Convective Effects
Although convective effects are not relevant for zero-g operation, they could substantially
affect the performance during ground tests. In some cases they might even prevent the sorption
cooler from being recycled and then qualification tests from being performed. Indeed, convective
effects may take place during the condensation phase while the sorption pump is heated. During
this phase, the pressure inside the cooler varies from a few tenths of a torr to a few hundred torr.
Obviously, if the cooler is oriented such that the hot sorption pump is underneath the cold
evaporator, convection will certainly occur.
Measurements were performed on the cooler prototype oriented at various angles. These
experiments have demonstrated for instance that convection in horizontal tubes is even more
effective than in tubes orientated with the warm end vertically down. Figure 1 displays the
power required to maintain the sorption pump at 40 K as a function of the pumping line angle (0°
corresponds to the right-side-up position (pump directly above the evaporator), 90° to the
horizontal position, and 180° to the upside-down position). This power is a direct signature of
the convective effects. The data obtained without gas inside the cooler are also reported in Fig. 2.
During these tests the cold heat sink was a pumped helium cryostat maintained at around 1.8 K.
In these figures we have also reported the associated evaporator temperature for this particular
prototype. The evaporator is connected to the bath via a gas gap heat switch, and thus its
temperature is driven by the thermal load associated with the convective effects. The results
clearly indicate the cooler cannot be recycled for angles above 70-80°. The second curve
displays more detailed experimental results for angles around 80°. It can be seen that the
convective effects are not significant for angles up to 75° for this particular prototype (it is
interesting to note that the 15° angle roughly corresponds to the tilt angle for which the sorption
pump is in direct view of the thermal shunt from an horizontal point of view).
An interesting feature is that convection is more efficient at angles close to 135° than in the
full upside-down position. This type of behavior, although not clearly understood, has been
reported by other authors.4
In order to avoid this effect in most, but perhaps not all orientations, one could use a line
made of a succession of sections with various orientations so that ideally in any position one of
these sections is oriented with its warm side up. However, even if convection is suppressed in
some sections of the pumping line, its overall thermal conductance is expected to be larger
during ground tests than in orbit, and thus the thermal efficiency of the cooler during ground
testing may be affected. Another solution is to insert, for instance, some felt inside the line to
attenuate or suppress the convective effect. This solution has been experimentally validated;
however, one has to find a balance between convective effects and the ultimate temperature,
since more felt means larger pressure drops.
Figure 2. Recycling phase – power required to maintain the sorption pump at 40 K.

Prototype pumping line geometry 10 mm, overall length 160 mm.
Obviously it is not desirable to degrade the cooler performance for the overall mission
duration just to compensate for some ground effects. Thus, the best solution is simply to avoid,
on ground, any orientation for the cooler between 75° and 180° (this angle depends somewhat on
the geometry of the pumping line); the integration of the cooler inside the flight cryostat shall
permit us to address this constraint, allowing testing of the overall system in various orientations
(cryostat can be tilted in any position in a cone of angle 150°).
It is important to note that during the low temperature phase, the orientation does not matter
any more, and the cooler can be operated in any position.
For the SPIRE instrument the cooler includes a straight pumping line for performance and
manufacturing reasons. Consequently, the cooler will be mounted in such a way that recycling
will be possible with the full satellite tilted to a maximum angle of 15°.
Superfluid Film Behavior

An annoying source of parasitic load in the case of the cooler is the superfluid film
behavior. Indeed, when a chamber contains liquid He II (temperature below the point), a
mobile film, also known as the Rollin film, covers the walls of the container up to a height where
it is warm enough for it to evaporate. As it evaporates, more helium flows up through this film so
that the process leads to a continuous loss of helium. This direct mass loss can have a strong
impact on the performance of both the operating temperature and the hold time (the latent heat is
not extracted from this “lost” helium leading to a cooling loss). In addition, this film can carry a
substantial amount of heat by conduction between the cold and wanner part of the device.
The physical properties of this film have been extensively studied (see for instance5,6). Smith
and Boorse have published a very interesting series of papers7,8,9,10 on the subject, in which in
particular they have investigated the role of substrate, surface finish, temperature, and film
height. Results of interest to the present work include:
• The thickness of the film is practically independent of the temperature and is a few tenths of
nanometers thick (typical values are 250 to 350 A)
• The flow velocity can be as high as 250 cm/s
• The transfer from one side to the other is restricted by the narrowest part of the connecting
surface above the liquid
• The rate of transfer depends on the temperature: from 0 at the point it rises to typically
however values above have been reported for “dirty” surfaces
• The rate of transfer seems to partially depend on the quality (finish) and nature of the surface.
Clean glass is reported to be one of the best choices, as dirty metal surfaces can have transfer
rates substantially higher. The surface finish can lead to variations of about 30%.
The most common practical solution used to attenuate this effect is to implement a flow
restriction in the pumping line in the form of a diaphragm with a small hole. From the above
remarks, it can be calculated that the cooling loss at, for instance 1 K (a typical operating
temperature of a cooler), reaches a value of per cm of wet perimeter for clean glass,
and can in some cases be as high as 1.5 mW/cm for a metal surface. A 1 cm pumping line could
result, for example, in cooling losses ranging from 1 mW to 5 mW, values prohibitively too large
for a sorption cooler.
Measurements of this effect were performed for a stainless steel diaphragm with a small
hole, and the experimental data are best described by the following equation (see also Fig. 3a):
(where is the diaphragm hole diameter in mm)
Of course the size of the diaphragm hole has a direct impact on the pressure drop and thus on the
ultimate temperature of the cooler, as illustrated on Fig. 3b; for this particular set of experiments,
the ultimate temperature is reported versus diaphragm hole diameter. Thus, a typical diaphragm
hole size of results in an acceptable load of less than while in the meantime, it
does not significantly impact the ultimate temperature of the cooler. Smaller diaphragm holes
may be blocked more easily by impurities.
Figure 3a. Parasitic load due to superfluid Figure 3b. Effect of diaphragm hole size
film per mm of wet perimeter. on cooler ultimate temperature.
ESA TRP PROTOTYPES

Following these results both coolers have been sized using a software tool developed at
CEA-SBT. The limit on the average power dissipated on the helium cryostat over a full cycle
calls for a precise knowledge and control of all energies and power involved in the cycle.
Figure 4 displays a 3D drawing of the cooler, and Table 2 summarizes the main
characteristics of both coolers. Most of the cooler, i.e. the evaporator and sorption pump
envelope, the pumping line and heat-switch tubing and support structure, is made of titanium
alloy TA6V. Measurements performed in the laboratory show that even at low temperature this
alloy exhibits a lower thermal conductivity than stainless steel. Consequently, this space-
qualified material provides three benefits: 1) a lower thermal conductivity leading to lower
parasitic loads, 2) a higher mechanical strength leading to a reduction of the tube wall thickness
(reduced parasitic loads), and finally 3) a lower density leading to reduced overall mass (and
consequently to better mechanical performance and a lower support structure thermal load).
Figure 4. 3D schematic of 3He cooler.

Both heat switches are mounted off the support structure. However, because this structure is
mounted off a 4K plate, an intermediate TA6V tube mechanically connects the switches to the
structure. This tube provides strong mechanical fixation and good thermal isolation between the
copper base of the switch and the 4K heat sink. Both heat switches include snubbers to prevent
any excessive lateral motion of the copper end and the miniature sorption pump. These switches
are charged with gas at room temperature and are permanently sealed. The use of gas
substantially improves the performance in the ON position, since, in the 2K range, has a
thermal conductivity three times larger than
In Fig. 4 the Kevlar suspension structure is clearly seen. The tensioning of the Kevlar cords
is made using capstans. Hard stops are included to limit the displacement of the sorption pump
and evaporator in case of a Kevlar cord failure. In addition these hard stops are adjustable and
allow us to firmly hold the pump and evaporator while installing the Kevlar ropes and tensioning
the system. One potential problem with Kevlar could be the loss of tension over time due to
creep (the storage period of an unattended cooler could exceed a year). In order to assess this
aspect, measurements have been performed on a given length of Kevlar cord tensioned in
between two fixed points and instrumented with a force transducer. Two samples have been
characterized: not baked, and after an 80°C bake for about a week. The results shown in Fig. 5
indicate that creep of this particular cord will not impact the mechanical performance of the
suspension structure, since after a couple days (time for the cords to “settle down” - the first
decrease is attributed to slipping of the cords), a 10% change in tension requires about 22 years.
One critical aspect is the leak-before-burst requirement. Both coolers are designed so that
the pumping line (with a calculated burst pressure of 35 MPa) is the weakest point with regards
to internal pressure. In order to demonstrate the leak-before-burst behavior, a test was performed
on a dummy TA6V piece representative of the line. For practical and safety reasons, the piece
was pressure tested to failure with water. However, to check for any major difference with gas, a
first set of tests was performed on thin walled stainless steel tube – burst pressure 1.6
MPa) with water and gas. The results indicate the failure mode is about the same: the tube cracks
open in the middle along its length and the fluid is exhausted. With gas inside, because of the
remaining pressure inside the tube, the crack propagates to some length. For the titanium piece,
the recorded burst pressure is 36 MPa. In this case the visual aspect of the crack is quite different
and shows a succession of small steps. This behavior may be due to the difference in ductility
between stainless steel (up to 60% breaking elongation) and titanium (less than 15%).
Figure 5. Creep of Kevlar 29 (sample characteristics: length 220 mm, cord diameter 0.5 mm).
Performance
A qualification test plan has been established and comprises a test sequence to be performed
on the sorption coolers, together with analysis and other verification methods. The test program
covers the functional performance and environmental tests required to provide confidence in the
ability of the coolers to meet the specified requirements. Performance measurements have been
made before and after the environmental tests to detect changes in performance parameters that
may indicate a potential failure.
The main tests are listed in chronological order in Table 3 together with a summary of the
results. Most of these data have been obtained very recently; they will be further analyzed and
presented in a subsequent publication.
Both coolers have passed the vibration test sequence without apparent failure or helium
leakage. Measurements were done on both thermal switches, on the supporting structure at the
evaporator and pump side, on the evaporator, on the pump cold ends, and on the thermal shunt
on the pumping line.
As the cooler is smaller, and thus lighter than the cooler, the eigenfrequencies are
higher for the first one compared to the latter. The first measured eigenmode is a bending of the
overall structure in the axis of the pumping tube. In axes transverse to the pumping tube axis, the
first eigenfrequencies are higher and quite similar due to the relative symmetry of the coolers
with regards to the pumping tube axis.
The thermal switches are fixed by one end, and behave like cantilever beams. Moreover, as
a flexible copper strap is bolted on the free end, the eigenfrequency is lowered by the mass of it.
The first eigenfrequencies in the transverse axes are then quite low, and vary from one thermal
switch to the other. The snubbers are then very efficient in limiting the thermal-switch end
motion and are mandatory to prevent them from damage. The first eigenfrequency in their axis is
very much higher. An abbreviated summary of the eigenfrequencies is given in Table 4.
The post-vibration-test thermal performance of the cooler was checked and found to be
similar to previous measurements (see Fig. 6). The unit shall be tested shortly.
Figure 6. Ultimate temperature versus cooling power for both coolers, under various orientation.
A dedicated test cryostat has been developed at CEA-SBT. This cryostat has been designed
to be operated in any orientation between –90° and +90°. Thus, by correctly mounting the cooler,
it is possible to check for its operation with or against gravity. Note, however, that when the
cryostat is used at 4.2 K (bath not pumped), it can only be tilted to +/- 75° because of convective
effects in the main tube. Figure 6 displays the cooling power curves in various orientations for
both coolers. Figure 7 shows two pictures of the cooler, mounted in the test cryostat at
CEA-SBT and on the vibration table at CSL.
CONCLUSION
Two engineering models of sorption coolers have been developed and qualified for space
applications. They provide temperatures down to 278 mK and 912 mK, respectively, for the
and unit. Once cycled, these coolers are insensitive to orientation and can even be operated
against gravity. Both coolers have been successfully vibration tested.
The flight model cooler of the SPIRE and PACS instruments onboard the FIRST satellite
will be designed based on this heritage.
ACKNOWLEDGEMENT
We wish to thank L. Clerc and L. Miquet who successfully conducted the experimental
testing of both coolers. L. Guillemet contributed to the overall design of the coolers and
produced all 3D drawings. Finally R. Vallcorba performed a numerical modeling of the cooler to
analyze its mechanical behavior.
Figure 7. cooler mounted in the test cryostat and on the vibration table.
REFERENCES
1. Torre, J.P. and Chanin G., “Miniature Liquid Refrigerator,” Rev. Sci. Instrum., 56 (1985),
pp. 318-320.
2. Duband, L., “Space-borne helium adsorption coolers,” ESA SP-401 (1997), pp. 357-360.
3. Duband, L., Hui, L. and Lange, A.E., “A space-borne refrigerator,” Cryogenics, 30 (1990),
pp. 263-270.
4. G. Thummes, G., Schreiber, M., Landgraf, R. and Heiden, C., “Convective Heat Losses in Pulse
Tube Coolers : Effect of Pulse Tube Inclination,” Cryocoolers 9, Plenum Press, New York (1997),
pp. 393-402.
5. Keesom, W.H., Helium, Elsevier Publishing Company (1942), pp. 301-312.
6. White, G.K., Experimental Techniques in Low-Temperature Physics, Oxford University Press
(1987).
7. Smith, B. and Boorse, H.A., “Helium II film transport. The role of substrate,” Phys. Rev., Vol. 98,
No. 2(1955), pp. 328-336.
8. Smith, B. and Boorse, H.A., “Helium II film transport. The role of surface finish,” Phys. Rev.,
Vol. 99, No. 2 (1955), pp. 346-358.
9. Smith, B. and Boorse, H.A., “Helium II film transport. The role of film height,” Phys. Rev.,
Vol. 99, No. 2 (1955), pp. 358-366.
10. Smith, B. and Boorse, H.A., “Helium II film transport. The role of temperature,” Phys. Rev.,
Vol. 99, No. 2 (1955), pp. 367-370.
Closed-Cycle Cooling of Infrared Detectors to
0.25 K for the Polatron
R.S. Bhatia, J. J. Bock, V.V. Hristov, W.C. Jones, A.E. Lange,

J. Leong, P.V. Mason, B.J. Philhour & G. Sirbi
MS 59-33, California Institute of Technology, Pasadena, California 91125, USA
S.E. Church & B.G. Keating

Dept. of Physics, Stanford University, Stanford, CA 94305-4060, USA
J. Glenn
CASA CB-389, University of Colorado, Boulder, CO 80309, USA
S.T. Chase
Chase Research Ltd., 35 Wostenholm Rd., Sheffield S7 1LB, United Kingdom
P.A.R. Ade and C.V. Haynes
Astrophysics Laboratory, QMW College, London El 4NS, United Kingdom
ABSTRACT
We have integrated a 4 K mechanical cryocooler with a three stage sorption
refrigerator to achieve cooling down to 0.25 K. The cryocooler consists of two Gifford-
MacMahon stages which cool to below its inversion temperature for use in a Joule-Thomson
expansion stage. This in turn provides cooling below the critical temperature of The
sorption refrigerator is specified to achieve a heatlift of 3 microwatts at 0.25 K for 12 hours. This
cryogenic system will be used for cooling of the bolometric detectors and optics in the Polatron,
a ground-based receiver for measurement of the polarisation of the cosmic microwave
background radiation at a frequency of 90 GHz. Cooling of sensitive detectors by mechanical
cryocoolers can lead to sensitivity degradation due to a combination of microphonic pickup,
electromagnetic interference and thermal dissipative heating. We describe the design features
incorporated at system level to mitigate these effects. The Polatron also serves as a testbed for
technologies to be used on the High Frequency Instrument on the ESA/NASA Planck satellite to
measure the temperature anisotropies and polarisation of the cosmic microwave background
radiation.

INTRODUCTION
The cosmic microwave background (CMB) radiation is the remnant signature on the sky of
the Big Bang which occurred between 12 and 15 billion years ago. Measurements of the
temperature variations of the sky as a function of angular scale can help to determine the value of
the density of the Universe, the Hubble constant, the amount of dark matter and the amount of
vacuum energy in the Universe. Many different ground-based, balloon-borne and satellite-borne
instruments have been developed to measure these temperature anisotropies and the latest results
indicate that the Universe is flat1. Measurements of the polarisation of the CMB radiation can
give us additional information, for example on gravity waves and the reionisation history of the
Universe2. We have designed and built the Polatron receiver to measure the polarisation of the
CMB, which has yet to be detected.
Observations will be made using the Owens Valley Radio Observatory 5.5 metre telescope,
which has been optimised to have small sidelobes and a low level of warm loading for sensitive
CMB observations. The general assembly of the Polatron receiver is shown in Figure 1. The
incoming polarised radiation is modulated by a rotating quartz half-waveplate. A series of optical
filters3 allow radiation only at GHz to pass through into an orthomode transducer, which
splits the radiation into two orthogonal linear polarisation components. Feedhorns4 couple each
of these two polarised components to a bolometric detector cooled to 250 mK to achieve the
Figure 1 . Polatron General Arrangement.

CLOSED-CYCLE COOLING TO 0.25 K FOR THE POLATRON 579
required detector sensitivity for each Stokes parameter of Because of the low level
of the signal (expected to be one part in ), long integration times are necessary to achieve the
required signal-to-noise. We also wish to make observations remotely, and have therefore chosen
to use closed cycle cooling of the detectors so that liquid cryogens will not need to be topped up
every few days at the telescope. Cryocoolers can lead to problems with microphonic pickup and
EMI, and as discussed below we use a variety of techniques to reduce these effects and the level
of experimental systematics in general5.
BOLOMETERS
In a bolometer, incident radiation causes a temperature change of the absorber that is
connected to a heat sink through a weak thermal link. Bolometers are the preferred type of
detector at submillimetre wavelengths because of their high responsivity at these wavelengths
and also their high absorptive efficiency. The bolometers used for the Polatron are of the ‘spider-
web’ type6, as shown in Figure 2. The mesh absorber is made by etching away in the shape of a
spider’s web a thin film ( ) of low stress that is deposited onto a silicon wafer. After
the nitride has been etched, the wafer is diced and placed in a silicon etch, leaving only the mesh
of nitride connected by a small number of legs to a solid silicon frame. Chromium and gold are
then evaporated onto the centre region of the absorber, producing a network of wires. The
physical absorber area is now very small, but the capture cross section to thermal radiation is still
the full area of the web. This leads to a significant reduction in total heat capacity and hence an
increase in speed. Also, the suspended mass is now small, and so the eigenfrequency for this sub-
structure is correspondingly much higher (in the kHz range). The two bolometers (one for each
polarisation) are read out in a bridge circuit to improve rejection of noise common to both
readout channels7. JFET source followers are used for impedance conversion to reduce the length
of high-impedance signal wiring8,9. The bolometers are AC biased in order to modulate the signal
frequencies above the 1/f noise of JFET preamplifiers. We have also implemented an active
control circuit using germanium thermistors10 for control of the temperature of the bolometer
stage.
CRYOCOOLER
The HS-4 cryocooler is manufactured by APD Cryogenics11 and consists of a two-stage
Gifford-MacMahon system which precools to below its inversion temperature. A third stage
then uses Joule-Thomson expansion to achieve a temperature below the critical temperature of
for operation of the sorption cooler.
Bolometric detectors are inherently sensitive to thermal noise. In addition, variation in the
temperatures of the shields and/or optical filters modulates the thermal power incident on the
bolometer, and hence creates an additional source of experimental systematic noise. Although
two-stage G-M systems are now commercially available to achieve temperatures of 4 K, the
temperature variations of a G-M cooler can be significant12. For this reason, the three-stage
configuration was chosen. This consideration is of course somewhat offset by the increased
complexity and susceptibility to contamination of such a system. Excellent thermal stability of
the J-T stage is expected provided that the incident heat load does not boil off all the liquid
that collects at the cooler tip. Our cryocooler system is an off-the-shelf system with some minor
modifications performed by APD Cryogenics to slightly reduce the vibration at the cold head
displex rotator.
The nominal operating temperatures of the two stages of the G-M system are 80 K and 20 K.
The heatlift at 80 K is 50 W. The base temperature achieved for the two-stage configuration is 12
K and with the third stage added it achieves 3.5 K. The cooler specification requires a heatlift of
1 W at 4.2 K. Load lines for different settings of the J-T expansion orifice are shown in Figure 3.
Figure 2 . Spider Web Bolometer. Figure 3 . Cryocooler Load Curves.

Both the G-M and J-T cycles are driven using two identical oil-lubricated compressors,
linked in series so that the supply pressure to the G-M circuit is 18 bar (gauge) and to the J-T
circuit is 24 bar (gauge). The compressors are rated for a mean time before servicing of 10000
hours. They are cooled using a water chiller which circulates water at 9 °C. The compressors are
linked to the cold head using 16 metre stainless steel braided gas lines. The J-T circuit contains a
warm filter at the compressor manifold and a cold filter immediately prior to the expansion
orifice.
Flexible thermal links made from copper are used at the cryocooler 4 K and 80 K stages. A
pneumatically driven vibration isolator is used at the interface to the cryocooler vacuum can to
reduce transmitted vibration at 300 K. We plan to electrically isolate the 4 K can from the
cryocooler using a sapphire disc to reduce the susceptibility of the bolometers to conducted EMI.
3 STAGE SORPTION REFRIGERATOR DESCRIPTION

A schematic of the refrigerator is shown in Figure 4 together with an assembly drawing in
Figure 5. The detailed design and manufacture was undertaken by Chase Research Ltd. The first
and second stages of the refrigerator contain and respectively and are collectively
referred to as the intercooler. The third stage contains and is referred to as the ultracooler.
The design specifications for the refrigerator are as follows. The required performance at the
ultracooler still is 250 mK for 12 hours with load, and at the intercooler still is 500 mK
for 12 hours with load. The required duty cycle is to be greater than 83 % (i.e. total cycle
time should be less than 2 hours from start of cycling to final stable operation). The allowable
physical envelope is defined to fit within the footprint and volume seen in Figure 3.
Room temperature charge pressures for the refrigerator are 95 bar with 99.95% purity
and 99.999% purity The intercooler contains 7 S.T.P. litres of and 4 S.T.P. litres of
The ultracooler contains 3 S.T.P. litres of Type 321 austenitic chromium-nickel stainless
steel is used for fabrication of the cryopumps. This grade has added titanium to reduce the
possibility of weld failure. As an adsorbent, each pump contains 3.5 g of activated charcoal per
S.T.P. litre of gas13. The pump tubes are made from either 321 or 316 (which has superior
corrosion resistance) austenitic stainless steel to reduce the parasitic loading on the stills as the
pumps are heated. To heat each pump, a current is passed through a metal film resistor.
The pump shields are made from type C101 copper. A thermal short is soldered half way along
each pump tube to the 4 K plate to extract the majority of the enthalpy of the desorbed gas during
condensation and therefore reduce the loading on the condensation points.
The intercooler has only one still assembly with two chambers: the chamber is nested
inside the chamber to reduce the Kapitza thermal boundary resistance between the two
Figure 4 . Schematic of three-stage refrigerator.

systems. The condensation point is thermally isolated from the cold plate using G-10
supports, and is directly connected to the JT expansion stage of the cryocooler using two OFHC
copper braids. This reduces the peak condensation point temperature during heating of the
pump. The intercooler still in turn functions as the condensation point heatsink for the
intercooler A thermal strap connects this to the ultracooler pump tube so that the intercooler
liquid then serves as the condensation point heatsink for the ultracooler The stills are
made from type C103 (OFHC) copper.
A heat exchanger is located between the intercooler still and the baseplate to extract the
enthalpy of the cold gas flowing from the still to the baseplate. This suppresses most of the
parasitic load which would otherwise find its way to the stills. Still temperatures are monitored
using germanium resistance thermometers14 which are read out using an AC resistance bridge15.
Pump temperatures are monitored using silicon diode thermometers16 with a excitation
current.
Figure 5 . Three-stage sorption refrigerator.

Following heating of the pumps to desorb the helium, the pumps must be cooled to enable
pumping on the condensed liquid to commence. A passive mechanical heatswitch is used on the
ultracooler pump. An active gas-gap heat switch17 is used on each of the intercooler pumps.
This type of heatswitch has been space qualified by Duband for use on the IRTS mission18 . It is
comprised of a stainless steel case containing gas, a small quantity of charcoal and two
concentric copper cylinders. A diode thermometer16 and a metal film resistor functioning
as a heater are attached using stycast19 to the charcoal housing. The heat switch is attached to the
pump via a copper strap. When the charcoal is at 4 K, the helium is adsorbed on the charcoal.
When the charcoal is heated to above ~ 15 K, the is desorbed into the gap between the two
cylinders, thereby thermally shorting them. The heatswitch requires power dissipation to
remain conductive, and cools to the off state (10 K) in ten minutes.
SORPTION REFRIGERATOR PERFORMANCE

On initial cooldown from room temperature, the pumps cool much more rapidly than the
stills. All the helium is then adsorbed by the charcoal in the pumps so that the stills would then
take an additional day to cool to 4 K. To expedite this cooldown, the pumps are kept at ~ 20 K so
that through the desorbed gas they function as gas-gap heat switches and help cool the stills.
The following description of the cycle procedure refers to the temperature versus time plot
for the pumps and stills shown in Figure 6. The pump is heated using a current of ~ 100 mA
to a temperature of 45 K. Above ~ 18 K, the is desorbed from the charcoal. The pressure
increase and the sub-critical condensation point temperature allow condensation of the at the
coldest point of the system, which is heatsunk to the cryocooler cold tip. Although the
condensation point initially warms to 4.7 K, it then cools again to 4.1 K. The condensed
drips into the still. The pump is stabilised at 45 K for ~ 5 minutes, then the heater current is
reduced to zero and the charcoal allowed to cool down via the heat switch. The charcoal pumps
on the liquid and the reduction in vapour pressure reduces the temperature of the liquid
to 0.920 K.
The ultracooler pump is cycled in a similar manner. The liquid from the intercooler
cools the ultracooler condensation point to below the 3.3 K critical temperature of As
the pump cools down again to 4 K via the tin heat leak, the pumping action of the ultracooler
charcoal reduces the vapour pressure above the liquified in the ultracooler still, reducing its
temperature to 0.3 K. The reduced heatload on the intercooler still allows its temperature to drop
to 0.75 K.
The enthalpy of the gas desorbed from the ultracooler pump reduces the residual left in
the still until, under the parasitic heatload, the intercooler still once again starts to warm up.
The pump must once again be recycled and this is performed in the same manner as
described above. As the pump is cooling, the intercooler pump is cycled. The intercooler
pump is heated to 22 K and stabilised at this temperature for 15 minutes, then allowed to
cool back to 4 K via its gas-gap heat switch. Care must be taken to ensure that before pumping
on the intercooler liquid commences, there is no residual in the outer still which would
thermally load the colder The additional heatload on the ultracooler during the intercooler
cycle warms the ultracooler still only to 0.4 K, which subsequently cools to its base temperature
as the intercooler still cools.
We have written an automatic refrigerator cycling procedure using LabVIEW20 which uses
the GPIB interface to command a programmable power supply. This is used to drive the correct
levels of current through the pump heaters for the required times, and is currently used in open
loop mode. This will serve to standardise and considerably simplify preparations for operation at
the telescope.
We have verified that the refrigerator meets the design specifications when mounted in a
standard liquid cryostat. When operated from the cryocooler, we have reached a base
Figure 6 . Sorption refrigerator cycle procedure .
temperature of 280 mK with a hold time of 2 hours. Due to mechanical envelope restrictions, this
refrigerator was designed with short horizontal pump tubes. In practise this has meant that the
cooler must be tipped to minimise convective loading and to achieve adequate condensation
efficiency. These mechanical envelope constraints have been removed and we have therefore
redesigned the refrigerator with longer and vertical pump tubes. However, we have not yet tested
this configuration.
It has been necessary to carefully optimise the cycles to reduce the peak loading on the
cryocooler. For example, our initial cycles were performed with high pump temperatures (35
K) and with condensation point temperatures close to the critical temperature of 3.3 K. We have
subsequently reduced these significantly. Figure 7 shows the hold time of the intercooler alone as
a function of peak still temperature and it can be seen that sufficient is desorbed at a pump
temperature as low as 22 K to achieve a reasonable hold time.
Figure 8 shows the thermal conductivity of the braids from the condensation point to the
cooler cold tip (via the braids), and from the 4 K plate to the cooler tip (via the foil). These are 74
and respectively. These conductivity levels are adequate for the levels of
power dissipation at various point of the sorption cycle. We therefore do not envisage that there
will be a problem with the added Kapitza resistances of the sapphire spacer which we will add to
electrically isolate the cryocooler from the 4 K cold plate.
Figure 7 . IC hold time as a function of peak Figure 8. Thermal conductivities to JT tip.

still and pump temperature during recyling.
RF FILTERS
The reciprocating motors and the electronics systems used for control of cryocooler moving
elements generate electromagnetic interference. For example, control system power supplies
often use pulse width modulation to reduce their power consumption which is generally very
limited in a satellite application. High impedance bolometric systems are extremely susceptible
to RF noise pickup on the readout and bias wiring. They are also susceptible to RF heating
because bolometers cannot discriminate between different types of thermal input. Limitations on
cryogenic power dissipation mean that active filters cannot be used. Passive filters have been
used on the SCUBA instrument and have flown on the Far Infrared Photometer on the Infrared
Telescope in Space21.
For the Polatron, we filter all wires going into the 4 K RF clean environment using Murata
EMI filters22. In order to test the response of the filter, an RF-tight box with BNC breakouts at
each end was used in conjunction with a firewall to separate the noisy signals from the ‘clean’
signals within the box. The filter was placed through a hole in the firewall. The ground plane of
the filter and the firewall were electrically connected with epoxy23, which also served to fill up
the remaining space in the hole. The box was then stuffed with Eccosorb LS19 foam for added
high frequency suppression. A control was also made using an identical box with a single wire
going through the firewall instead of the filter. The measurements were made using a sweep
oscillator24 and spectrum analyzer25. For the cold testing, the Murata filter was thermally cycled
three times by alternating between 77 K and 300 K. The cold response of the filter was then
measured by submersing the box in liquid nitrogen. After analyzing the results of the warm test,
we decided that it would be valuable to test the effects of the Eccosorb. Thus, another run was
made (using the box with no filter) without the Eccosorb.
The results are shown in Figure 9. The warm filter performed to specification. The rejection
in the stop band from about 500 MHz to 1 GHz is greater than 60 dB. From 1 GHz to 10 GHz,
we see that the filter does not provide much attenuation (difference between filter and control
box responses in Figure 9). However, the box and Eccosorb give extra loss, resulting in a
minimum attenuation of around 50 dB at 2 GHz. For the cold testing, we see that the thermal
shocking did not adversely affect the filter for it still responds well at 77 K. There is some loss in
attenuation over most of the frequencies, but this is not appreciable.
Figure 9. RF filter performance.
SUMMARY AND STATUS

We have designed and built a bolometric receiver which uses a three-stage sorption
refrigerator precooled by a three-stage mechanical cryocooler. We have achieved a base
temperature of 280 mK which is low enough to give reasonable bolometer sensitivity, but have
only been able to maintain this for a limited hold time. We have identified the necessary
modifications to the sorption refrigerator and are implementing these changes in time to start
observations during the winter 2000 observing season.
ACKNOWLEDGEMENTS
We wish to thank the Caltech Physics Machine Shop for fabrication of parts, and Jeff
Beeman at LBNL for supply of the NTD germanium for the temperature control thermistors and
the bolometers. Amanda Mainzer wrote the intial version of the LabVIEW fridge cycling code.
The bolometers were fabricated at the JPL Center for Space Microelectronics Technology. This
work was partly supported by NASA Innovative Research Grant NAG5-3465 and NASA NAG5-
6573 for US involvement in Planck. The construction of the Polatron is funded by a Caltech/JPL
President's Fund Grant PF-414, NASA Grant NAG5-6573 and NSF Grant AST-9900868.
REFERENCES
1. De Bernardis, P. et al., ‘A Flat Universe from High-resolution Maps of the Cosmic Microwave
Background Radiation’ Nature 404 (27 April 2000) 955-959
2. Hu, W. and White, M., ‘A CMB Polarisation Primer’, New Astronomy 2 (1997) 323-344
3. Lee, C., Ade, P.A.R. and Haynes, C.V., ‘Self Supporting Filters for Compact Focal Plane Designs’,
Proceedings of the 30th ESLAB Symposium, ‘Submillimetre and Far-Infrared Space
Instrumentation’, 25-27th September 1996, Noordwijk, The Netherlands, ESA SP-388 (1996) 81-83
4. Church, S.E., Philhour, B.J., Lange, A.E., Ade, P.A.R., Maffei, B., Nartallo-Garcia, R. and
Dragovan, M., ‘A Compact High-efficiency Feed Structure for Cosmic Microwave Background
Astronomy as Millimeter Wavelengths’, Proceedings of the 30th ESLAB Symposium,
‘Submillimetre and Far-Infrared Space Instrumentation’, 25-27th September 1996, Noordwijk, The
Netherlands, ESA SP-388 (1996) 77-80
5. Bhatia, R.S., Bock, J.J., Ade, P.A.R., Benoît, A., Bradshaw, T.W., Crill, B.P., Griffin, M.J.,
Hepburn, I.D., Hristov, V.V., Lange, A. E., Mason, P.V., Murray, A.G., Orlowska, A.H. and Turner,
A.D., ‘The Susceptibility of Incoherent Detector Systems to Cryocooler Microphonics’, Cryogenics
39 8(1999)701-715
6. Bock, J. J., Mauskopf, P. D. and Lange, A. ‘Silicon Nitride Micro-mesh Bolometers’, Proceedings
of the 30th ESLAB Symposium, ‘Submillimetre and Far-Infrared Space Instrumentation’, 25-27th
September 1996, Noordwijk, The Netherlands, ESA SP-388 (1996) 119-122
7. Rieke, F.M., Lange, A.E., Beeman, J.W. and Haller, E.E. ‘An AC Bridge Readout for Bolometric
Detectors’ IEEE Transactions on Nuclear Science, 36 1 (February 1989) 946-949
8. NJ450 Process JFET, InterFET Corporation, 1000 N. Shiloh Rd., Garland, Texas 75042, USA
9. Low, F.J., ‘Application of JFETs to Low Background Focal Planes in Space Infrared Astronomy -
Scientific/Military Thrusts and Instrumentation’, Proc. SPIE 280 (1981) 56-60
10. Neutron transmutation doped germanium thermistor, J.W. Beeman, M/S 2-200, Lawrence-Berkeley
national Laboratory, Berkeley, CA 94720, USA
11. Longsworth, R.C., ‘4 K Gifford McMahon/Joule-Thomson Cycle Refrigerators’, Cryogenic
Engineering Conference, June 14-18 (1987)
12. Plambeck, R.L., ‘Improved Seal for a 4 K Gifford-McMahon Cryocooler’, Proceedings of the 8th
International Cryocooler Conference, Vail, Colorado, 28-30 June 1994, 795-801 Ed. R. G. Ross,
Jr., Plenum Press, New York (1995)
13. Duband, L., Alsop, D., Lange, A. and Kittel, P., ‘A Rocket-borne 3He Refrigerator’, Advances in
Cryogenic Engineering 35B (1989) 1447-1456
14. Model GR-200 Series GRT Sensor, Lake Shore Cryotronics, Inc., Westerville, Ohio 43081-2399,
USA
15. AVS-47 AC Resistance Bridge, RV-Elektroniikka Oy Picowatt, Vantaa, Finland. Distributed by
Oxford Instruments Ltd., Witney, Oxfordshire, UK
16. Model DT470 Series Diode Sensor, Lake Shore Cryotronics, Inc., Westerville, Ohio 43081-2399,
USA
17. Frank, D.J. and Nast, T.C., ‘Getter-activated Cryogenic Thermal Switch’, Advances in Cryogenic
Engineering 31 (1986) 933
18. Freund, M. M., Duband, L., Lange, A. E., Matsumoto, T., Murakami, H., Hirao, T. and Sato, S.,
‘Design and Flight Performance of a Spaceborne Refrigerator for the Infrared Telescope in
Space’, Cryogenics 38 4 (1998) 435-443
19. Emerson & Cuming, Inc., AN-72, Woburn, MA 01888, USA
20. National Instruments Corporation, 11500 N. Mopac Expressway Austin, TX 78759-3504, USA
21. Freund, M. M., Hirao, T., Hristov, V., Chegwidden, S., Matsumoto, T. and Lange, A. E., ‘Compact
Low-pass Electrical Filters for Cryogenic Detectors’, Review of Scientific Instruments 66 3 (March
1995) 2638-2640
22. VFM41R01C222N16-27surface mount EMI chip filter, Murata Corporation
23. H20E Silver Epoxy, Epotek, 14 Fortune Drive, Billerica, MA 01821-3972, USA
24. HP 8350B (with 0.01–20 GHz HP 83592B), Hewlett-Packard, 3000 Hanover Street, Palo Alto, CA
94304-1185, USA
25. HP 8563A Spectrum Analyser, Hewlett-Packard, 3000 Hanover Street, Palo Alto, CA 94304-1185,
USA
A Continuous Adiabatic Demagnetization
Refrigerator for Use with Mechanical Coolers
P. Shirron1, N. Abbondante2, E. Canavan1, M. DiPirro1, M. Grabowski2,

M. Hirsch2, M. Jackson1, J. Panek1 and J. Tuttle1
1
NASA/Goddard Space Flight Center
Code 552
Greenbelt, MD, USA 20771
2
Worcester Polytechnic Institute/Department of Physics
Worcester, MA, USA 01609
ABSTRACT
Sub-kelvin refrigeration is an increasingly vital technology for space missions. Future
missions are pushing the development of refrigerators with high cooling power, low operating
temperature, and, as always, low mass. In this paper we report on the development of an
adiabatic demagnetization refrigerator (ADR) which can produce continuous cooling at
temperatures of 50 mK or lower, with a cooling power goal of The design uses multiple
stages to cascade heat from a continuously-cooled stage up to a heat sink. The serial
arrangement makes it possible to add stages to extend the operating range to lower temperature,
or to raise the heat rejection temperature. Compared to conventional single-shot ADRs, this
system achieves higher cooling power per unit mass and higher efficiency. The ADR is being
designed to operate with a heat sink as warm as 10-12 K to make it compatible with a wide
variety of mechanical coolers as part of a versatile, cryogen-free low temperature cooling
system. A two-stage system has been constructed and a proof-of-principle demonstration was
conducted at 100 mK. Details of the design and test results, as well as the direction of future
work, are discussed.
INTRODUCTION
Adiabatic demagnetization of a magnetocaloric material (that is, a material that exhibits a
marked change in entropy due to an applied magnetic field) is a decades-old method for
producing sub-Kelvin temperatures in the laboratory. Recently ADRs have been developed for
space flight1,2,3 to cool infrared and x-ray detectors to below 0.1K. More low temperature
detectors are planned for future space flights, including superconducting tunnel junctions
coupled to rf-single electron transistors4, transition edge sensors5 and spiderweb bolometers6, to
measure x-ray, ultraviolet, infrared, and submillimeter radiation. Refrigeration is needed not
only to cool the detectors to sub-Kelvin temperatures, but to cool the detector surroundings, or in

the case of far infrared wavelengths, entire telescopes, to a few Kelvin. While mechanical
coolers can achieve 4 K, they do so only at the expensive of a large input power and low
thermodynamic efficiency. An ADR attached to the low temperature end of a mechanical cooler
can improve the situation significantly.
The present state of ADR technology is represented by the system built for the X-Ray
Spectrometer (XRS) instrument3 and the ADR designed for SIRTF1. The XRS ADR uses a 920
g ferric ammonium alum (FAA) salt pill, a 2 T magnet and a gas-gap heat switch linking the
ADR to a superfluid helium bath. Operating at 60 mK, it achieves a hold time of 36 hours with a
parasitic heat load of and a detector cooling power of The SIRTF ADR is a
two-stage system intended to be used with a 4.2 K (or warmer) heat sink, and uses FAA for low
temperature detector cooling and gadolinium gallium garnet (GGG) as a guard stage to absorb
the majority of parasitic heat loads. SIRTF no longer uses an ADR, but the design has been
implemented by NIST for development of their x-ray spectrometer. The cooling power for this
system, operating at 0.1 K, is also only a fraction of a microwatt.
Future missions require much higher cooling power and lower operating temperature than
either of these ADRs would be able to achieve with reasonable hold times and overall mass. It
will also be necessary, or at least highly advantageous, to extend the ADR’s operating
temperature range to ~10 K in order to mate it to a mechanical cryocooler. Having the ADR
span a larger range will improve the performance of the overall system by allowing the
cryocooler to operate where its efficiency is greatest. Extrapolating present designs, however, to
meet these requirements results in fairly massive systems, at least in the 40-50 kg range. In this
paper we describe a continuous-duty ADR that will be able to meet these requirements for less
than 10 kg mass. Its operation has been described in detail previously, so here we just
summarize its main features.
The continuous ADR uses multiple stages which are connected in series between the heat
sink and the load to be cooled. During most of the cycle, the load is maintained at constant
temperature by a slow demagnetization of the “continuous” stage. Periodically, this stage is
recycled by cooling the adjacent stage to a lower temperature and turning on a heat switch. The
net heat flow out of the stage will cause the control system to magnetize the refrigerant to
maintain constant temperature. Once the recycling is complete, the heat switch is turned off and
the adjacent stage is returned to a higher temperature so that it may transfer the heat to the next
higher stage. The process is repeated until the heat is transferred to a heat sink.
The advantages of the continuous ADR are that:
• it achieves continuous refrigeration even at very low temperature
• it achieves higher cooling power per unit mass because each stage can be recycled much
more frequently
• it can easily be expanded to operate over a wider temperature range than conventional
ADRs; this is possible because each stage spans only a fraction of the total range, and one
simply adds additional stages to span the total range desired
• it is less massive because of the smaller refrigerant volumes needed and because the
smaller operating range of each stage requires much less powerful magnets
• it is more efficient because each stage is recycled more slowly than in a conventional
system, and because parasitic heat inputs are absorbed at higher temperature
• finally, the continuous nature allows heat to be dumped from the warmest stage to the
heat sink at a slow, controlled rate, instead of in infrequent bursts as is typical for single-
shot systems, thereby reducing the peak heat load on the cryocooler
Having multiple ADR stages that are controlled independently does increase the complexity
somewhat, but the performance gains and reduced mass more than offset these costs. Based on
preliminary design studies, a continuous ADR operating at 50 mK using a 10 K sink
could be made with a mass as small as 10 kg. This represents an order of magnitude increase in
cooling power over existing systems, while the mass is actually reduced. Such a system
envelops the cooling requirements for all currently planned space missions, and the high sink
temperature will allow it to operate with a wide range of mechanical coolers under development.
CONTINUOUS ADR FOR USE WITH MECHANICAL COOLERS 589
Figure 1. Two-stage ADR developed to demonstrate continuous low temperature cooling.
DEMONSTRATION OF CONTINUOUS COOLING

Since the design of a full ADR system, including the number of stages needed to span a
given temperature range, the size of each stage, etc., depends most strongly on the characteristics
of the continuous stage, our present effort focuses on conducting a demonstration of continuous
cooling at low temperature. The primary goal was to characterize the extent to which one ADR
stage could be operated at a constant temperature while heat was periodically offloaded to a
second stage. The demonstration was performed using two stages connected in series as shown
in Fig. 1.
The continuous stage consists of a 42 g chrome potassium alum (CPA) salt pill, a 0.1 T
magnet, and a thin layer of magnetic shielding connected to a much larger stage (consisting of a
730 g FAA salt pill and a 2 T magnet) through a superconducting indium heat switch. The FAA
stage has enough cooling capacity to initially cool the CPA to its operating point then to act as its
heat sink for several days without needing to be recycled. The mass of CPA was less than the
container was designed to hold (52 g) due to problems encountered when growing the salt, but it
is enough to give the salt pill a hold time of ~1 hour with a heat load of at 0.1 K.
The configuration of the continuous stage is unusual in that the magnet is physically and
thermally anchored to the CPA salt pill. Ordinarily the magnet is attached to the heat sink and
the salt pill is suspended inside it. However, in our case the arrangement simplifies the
mechanical design and actually has no negative impact on performance because the stage
operates at constant temperature. The only potential concern relates to hysteresis losses within
the superconducting windings. We avoid this by using niobium-clad NbTi wire. As a Type I
superconductor, the niobium cladding will carry all of the current without any hysteresis. Above
niobium’s critical field of ~0.1 T, the magnet will generate hysteresis heating as magnetic flux
penetrates into the NbTi.
The design of the salt pill, or more specifically the internal thermal bus, also represents a
significant departure from convention. The standard practice when using hydrated salts is to
grow the salt onto a matrix of gold wires distributed inside a hermetic enclosure. Typically only
a few percent of the volume is used for the bus. In our case the thermal bus must be much more
conductive and have larger surface area to accommodate the higher cooling power and smaller
salt volume.
The salt pill is shown at various stages of the fabrication process in Fig. 2. The thermal bus
is cut from a single piece of OFHC copper using a wire EDM process. Two sequences of cuts,
with a 90° rotation in between, result in a uniform matrix of copper “fingers” on a square grid.
The fingers are cut as finely as possible to produce the largest possible surface area for the
amount of copper remaining. The goal was 0.25 mm fingers spaced apart by 0.50 mm, but
warping of the copper during machining made it necessary to increase the size of the fingers.
The final dimensions for the salt pill used in this work were 0.60 mm fingers with a center-to-
center spacing of 1.20 mm, yielding less than half of the desired surface area.
The heat switch design is fairly standard but has a few features of note. It is made with
copper flanges on each end that can be bolted to the thermal buses of both the CPA and FAA salt
pills. This allowed it to be fabricated and tested separately from the other components. The
flanges are structurally supported by a cylindrical shell of Vespel. This made the heat switch
strong enough to be the sole structural support for the continuous stage. The switching element
was made from high purity indium which was melted into a cylindrical cavity
defined by the Vespel and copper flanges. The switch is activated by a pair of superconducting
Helmholtz coils that produce a field perpendicular to the direction of heat flow.
The switch was designed to have an on-state conductance of 8 mW/K at 50 mK, and an off-
state conductance 3-4 orders of magnitude smaller. The high on conductance is crucial for
obtaining high cooling power. Thermal conductance measurements for the as-built switch,
including the two bolted end joints, are shown in Fig. 3. The on conductance is more than a
factor of 4 smaller than anticipated, due possibly to impurities in the indium or to excessive joint
resistance. Consequently the present ADR is limited to a smaller cooling power than the
goal.
Continuous Cooling Tests

Continuous cooling tests were conducted at 100 mK. While our eventual goal is to operate
at 50 mK or below, we are at present limited to higher temperature because of the limited cooling
power of the FAA stage. A “run” consisted of initially demagnetizing the FAA to cool both the
Figure 2. The continuous stage salt pill at various stages of construction.

Figure 3. Thermal conductivity of the indium heat switch in the normal and superconducting states.
FAA and CPA stages to 0.1 K. As the system cools, the superconducting heat switch is turned
on and the current in the CPA’s magnet is adjusted to a midrange value. Once the system
equilibrates at 0.1 K, a temperature controller is engaged to maintain constant temperature. The
control thermometer is located on a thermal strap bolted between the superconducting heat
switch and the CPA salt pill. This strap is used to thermally connect a load to the ADR, thus we
are properly controlling the temperature of the load, not simply the temperature of the CPA.
With temperature control established, a simple sequencing routine takes over to
automatically recycle the CPA when its cooling capacity is nearly exhausted. Typical results are
shown in Fig. 4a. The characteristic sawtooth pattern in CPA current results from alternately
magnetizing the CPA as it transfers heat to the FAA stage, then demagnetizing the CPA as it
actively cools its load and absorbs parasitic heat. The details of this sequence are as follows:
1. Throughout most of the cycle, the CPA is thermally isolated from the FAA (heat switch
is off) and is receiving heat from its load and parasitic sources. To maintain constant
temperature, the CPA is slowly demagnetized. Meanwhile, the FAA is stationed at a
higher temperature to absorb its parasitic heat loads more efficiently.
2. When the current approaches zero (or falls below a preset threshold), the recycling
process begins:
2a. The FAA is cycled down to 0.1 K
2b. The superconducting heat switch is turned on.
2c. The FAA is demagnetized to a slightly lower temperature, ~.095 K. This begins a
flow of heat from the CPA to the FAA. The net flow of heat out of the CPA
causes its temperature controller to begin magnetizing it
2d. When the CPA is fully magnetized, the FAA is brought back to 0.1 K and the heat
switch is turned off. As the FAA temperature rises, the CPA’s controller gradually
switches back to a slow demagnetization as heat flows into the CPA.
2e. The FAA is returned to its higher temperature.
In a system containing multiple stages, a similar process will be used to transfer heat from
the FAA to the next higher stage, and so on until the heat was transferred to the sink. The main
difference is that the wanner stages do not need to be controlled to a high level of stability since
their temperatures do not directly affect the stability of the continuous stage. As a result, the
transfers can occur more rapidly and with a much less sophisticated control algorithm.
Temperature Stability
The temperature of the continuous stage could be precisely controlled using either digital or
analog (Linear Research Model 130) temperature controllers. The analog controller, however,
provided smoother switching between magnetizing and demagnetizing as the net heat load to
CPA reversed direction. As seen in Fig. 4b, the temperature noise at 0.1 K is rms. This
Figure 4. (a) Shows the characteristic sawtooth pattern in current needed to maintain the
cold stage at a 100 mK as heat is periodically offloaded to the upper FAA stage; (b) shows
the temperature stability presently achieved during a cycle.
is almost entirely electronic noise from the resistance bridge used to readout the control
thermometer, but there also appears to be some amplification of the noise by the temperature
controller when the heat switch is turned off. This results from having tuned the temperature
controller to the fast thermal time constant of the system when the heat switch is on, so that it is
not optimized when the heat switch is off. A programmable controller should alleviate this
problem.
In Fig. 4b, one can also see spikes in temperature as the FAA is cooled to begin recycling
the CPA. These “warming” spikes last only a few seconds and are never more than about 500
in amplitude. There are usually 2-3 per recycling event, although in some cases there are
more than a dozen. The short duration and the fact that there is no net change magnetic field
after the system stabilizes suggest that the effect does not involve a cooling or heating of the
entire salt pill, but is perhaps a local effect near the thermometer. The origin is still uncertain;
possibilities include superconducting effects in the CPA magnet that cause discrete changes in
the field and a sudden pulse of eddy current heating. This is consistent with the apparent link
between the spikes and a change in direction of the magnet current, but it fails to explain why the
effect doesn’t occur when the system switches from magnetizing to demagnetizing. Whatever
the cause, we have found that operating the system using a very limited field range greatly
reduces or eliminates the spikes. Therefore we conclude that this is not an intrinsic problem, but
one that can eventually be eliminated.
Cooling Power
For the cycles shown in Fig. 4a, the CPA rejects heat at a constant rate of for 0.58
hours, and is subjected to a parasitic heat load through the heat switch of approximately
for 3.1 hours. These values were determined by measuring the relative magnetization (and
demagnetization) rates for a nominal cycle and when an additional known heat load (of 1 or 2
was applied to the salt. The overhead for the recycling operation (the time required to turn
the heat switch on and off and to ramp the FAA’s temperature up and down) is 0.23 hours. The
total time used to recycle the CPA is therefore 0.81 hours, or 20% of the total cycle time.
As we add heat from a load, the amount of time needed for the recycling operation will
increase, and the hold time will decrease. Beyond some heat load, there will be too little hold
time available to recycle the FAA. This threshold represents the maximum cooling power of the
ADR. At present a reasonable estimate is the value for which the hold and recycle times are
equal. For our current configuration where we use a 5 mK temperature drop across the heat
switch, we find a cooling power limit of at 100 mK. We can increase this value, at the
expense of efficiency, by increasing the gradient across the heat switch. Over a small range, the
cooling power will simply scale with the gradient, so that with a 10 mK gradient, the maximum
cooling power is about However, much beyond this, the overhead on the recycling
process begins to be a dominant factor; a practical limit for the cooling power is about at
100 mK.
One nice feature of the continuous ADR is that its cooling power is not strongly temperature
dependent. In fact, at 100 mK and below it is nearly independent of temperature since in this
range the entropy capacity of the salt (for a fixed magnetic field range) is inversely proportional
to temperature, and therefore the heat it can absorb is constant. The efficiency will, however,
decrease with temperature since proportionally larger gradients will be needed across the heat
switch to recycle the salt.
Thermodynamic Efficiency
In the design of the continuous ADR, there were several trade-offs that had to be made that
potentially could have degraded the ADR’s efficiency. Mostly these had to do with increasing
the amount of copper used in the heat switch and thermal bus in order to obtain large thermal
conductances. These components are close to the CPA magnet and could have been a source of
significant eddy current heating.
To address this concern, we characterized the efficiency of the continuous stage at two
levels. The first quantifies the performance of the salt pill itself in terms of the amount of heat it
can absorb during the demagnetization portion of the cycle compared to the amount of heat that
is rejected during as it is magnetized. The second, which is more appropriate for comparing the
present ADR to other refrigerators, was to determine the heat rejected to the FAA heat sink
compared to that absorbed during the cooling portion of the cycle.
The salt pill’s efficiency was determined to be 94% by taking the ratio of the heat absorbed
and rejected as it was demagnetized and magnetized, respectively, over the same magnetic field
range. This is only slightly less than the maximum that could be achieved given the fact that the
CPA’s temperature is approximately 99.4 mK when it is absorbing heat from the load, and 104
mK when it is rejecting heat. For this reason we conclude that eddy current heating in the heat
switch and salt pill is not a significant concern.
The same basic procedure was used to determine the heat flow into the FAA during
recycling, and we arrive at a figure of 88% for the efficiency of the continuous stage at 100 mK.
Again, the 6% drop is also almost entirely accounted for by the 5 mK drop across the heat switch
during recycling events.
FUTURE WORK
The two primary directions for future work are to continue developing the components
needed for a continuous cooling demonstration at 50 mK, and to begin work on materials and
components for the high temperature stages (1-10 K). For the low temperature work, we are
developing a new liquid-gap heat switch that can operate at sub-Kelvin temperatures. Along
with another CPA salt pill and magnet, this will be part of an intermediate stage between the
existing CPA and FAA stages. The heat switch has sufficiently low off-state conductance that
the FAA can be raised well above 1 K with low parasitic heat flow This three-stage
ADR will be able to produce true continuous cooling at 50 mK while the uppermost (FAA) stage
rejects its heat to a liquid helium bath at ~1.3 K.
At the upper end of the desired temperature range, magnets and magnetocaloric materials
are the most critical components. Magnets become a particular issue because standard high-field
NbTi magnets have a useful range only up to ~7 K and it is necessary to use a higher temperature
superconductor such as But although magnets are an established

technology for producing high fields at 4.2 K, very little work has been done to demonstrate their
ability to operate at 10 K or above in a vacuum environment. In addition, commercially
available magnets have relatively low field-to-current ratios, at best on the order of 0.1
T/A. Our effort here will be to identify any constraints on operating magnets in vacuum
(ramp rates, maximum field, etc.) and to investigate using a new small gauge wire which
can be reacted before being wound into a magnet. This wire is promising for significantly
increasing the field-to-current ratio and simplifying the magnet fabrication process.
We are also expanding efforts to develop magnetocaloric materials which have good
entropy capacity in the 1-10 K range while requiring the smallest magnetic fields. The largest
mass for an ADR system is usually the magnet (and certainly so if magnetic shielding is
considered), so developing materials with lower field requirements translates to lower system
mass. Some very promising results have recently been obtained with Gadolinium and
Dysprosium Gallium Garnet (GGG/DGG) mixtures9, and with iron-doping of GGG10. In
addition to producing and characterizing their magnetic properties, a significant effort will be
required to determine the requirements for packaging these materials to maximize their thermal
conductance.
SUMMARY
We have performed the first demonstration of continuous cooling at sub-Kelvin
temperatures using an ADR. The two-stage system achieved excellent temperature stability of
better than rms at 100 mK over a 24-hour period. The present system has demonstrated a 2
cooling power at 100 mK with an efficiency of 88%. It is capable of cooling powers up to 6
(with reduced efficiency), but with changes which are currently being made to the hardware
cooling powers of in excess of are expected. For the continuous ADR, the cooling power
is relatively independent of temperature, so this performance will also hold at our desired
operating temperature of 50 mK and below.
Future work will focus on developing all of the components needed for a multi-stage ADR
capable of operating continuously at temperature down to 50 mK using a 10 K heat sink.
ACKNOWLEDGMENTS
This work was supported by NASA/Goddard Space Flight Center’s Director’s Discretionary
Fund.
REFERENCES
1. Hagmann, C. and Richards, P.L., Two-stage Magnetic Refrigerator for Astronomical Applications
with Reservoir Temperatures above 4K, Cryogenics 34:221 (1994).
2. McCammon, D., Almy, R., Deiker, S., Morgenthaler, J., Kelley, R.L., Marshall, F.J., Moseley, S.H.,
Stahle, C.K. and Szymkowiak, A.E., A Sounding Rocket Payload for X-ray Astronomy Employing
High-resolution Microcalorimeters" Nucl. Instr. & Methods A 370:266 (1996).
3. Serlemitsos, A.T., SanSebastian, M. and Kunes, E., Design of a Spaceworthy Adiabatic
Demagnetization Refrigerator, Cryogenics 32:117 (1992).
4. R.J. Schoelkopf, S.H. Moseley, C.M. Stahle, P. Wahlgren and P. Delsing, A Concept for a
Submillimeter-wave Single-photon Counter, submitted for publication in IEEE Transactions on
Applied Superconductivity 9 (1999).
5. K.D. Irwin, G.C. Hilton, D.A. Wollman, and J.M. Martinis, X-ray Detection using a Superconducting
Transition-edge Sensor Microcalorimeter with Electrothermal Feedback, Appl. Phys. Lett. 69:1945
(1996).
6. J.J. Bock, et al., Silicon Nitride Micromesh Bolometer Arrays for SPIRE, Proc. of the SPIE, 3357:
297-304 (1998).
7. S. Breon, et al., The XRS Low Temperature Cryogenic System Ground Performance Tests Results,
Cryogenics 39: 677(1999).
8. P.J. Shirron, E.R. Canavan, M.J. DiPirro, J.G. Tuttle, and C.J. Yeager, A Multi-stage Continuous-
duty Adiabatic Demagnetization Refrigerator, accepted for pub. in Adv. Cryo. Eng. 45 (2000).
9. Numazawa, T., Sato, A., Large Magnetocaloric Effect in (DyxGd1-x)Ga5012 for Magnetic
Refrigeration below 1 K, Proc. of the ICEC17, Bournemouth, UK, p. 287 (1998).
10. Shull, R.D., McMichael, R.D., Ritter, J.J., Swartzendruber, L.J. and Bennett, L.H., Magnetic
Nanocomposites as Magnetic Refrigerants, Proc. of the 7th Int’l Cryocooler Conf., Santa Fe, NM, p.
1133 (1992).
Reaching 96 mK by a Pulse-Tube Precooled
Adiabatic Demagnetization Refrigerator
G. Thummes1, M. Theiß1, M. Bühler2, and J. Höhne2

1
Institute of Applied Physics, University of Giessen, D-35392 Giessen, Germany
2
CSP Cryogenic Spectrometers GmbH, D-85737 Ismaning, Germany
ABSTRACT
We report on the first operation of an adiabatic demagnetization refrigerator (ADR) that
employs a two-stage 4 K pulse tube refrigerator (PTR) for precooling. The ADR stage consists of a
ferric ammonium alum (FAA) salt pill located in the bore of a commercial superconducting
magnet. Precooling of the FAA pill is accomplished by means of a mechanical heat switch
thermally connected to the 2nd stage cold end of the PTR. The PTR, which was developed at the
University of Giessen, simultaneously provides cooling powers of 0.43 W at 4.2 K and 10 W at 57
K at the 2nd and 1st stage, respectively. The PTR is operated by use of a 6 kW GM-compressor in
combination with a rotary valve. The magnet is conductively cooled by thermally connecting the
Al-bobbin to the 2nd stage cold end of the PTR. The part of the current leads extending from the 1st
stage cold end to the magnet terminals consists of high-Tc Bi-2223 tapes with low thermal
conductance. After a cool-down period of about 13 h, the temperatures of the 1st and 2nd stage were
about 59 K and 2.8 K. Using a sweep rate of 10 A/minute the magnet was successfully charged
with a current of 55 A, corresponding to a field of 4.4 T. After demagnetization the FAA pill
attained a minimum temperature of 96 mK. Then the pill slowly warmed-up to 100 mK after l½
hours and to 140 mK after 4 hours. Lower warm-up rates and lower temperatures could be
expected after reducing the heat leak to the ADR stage.
INTRODUCTION
Since the pioneering work of Matsubara and Gao1, multistage pulse tube refrigerators (PTRs)
that operate at liquid helium temperatures have experienced considerable progress in cooling
power. While the first 4 K-PTRs1,2 consisted of three stages and reached only low cooling powers
of 25-30 mW at 4.2 K, the recently developed two-stage systems3-5 attain cooling powers of up to
0.5 W at 4.2 K. Such a cooling power is still a factor of two or three below that of commercial two-
stage Gifford-McMahon coolers.6 However, the intrinsic low level of mechanical vibrations of
PTRs makes them very attractive for applications that require a cryogen-free low-noise cooling
technique.
In particular, 4 K PTRs could be used for precooling of refrigeration systems at sub-Kelvin
temperatures without the need of a complex vibrational decoupling of the very low temperature
stage. A comparatively simple and reliable method to achieve temperatures below 100 mK is that
of adiabatic demagnetization of paramagnetic salts, e.g. Refs. 7 and 8. In an adiabatic
demagnetization refrigerator (ADR), a paramagnetic solid that is in thermal contact to a precooling

stage, is at first isothermally magnetized in a magnetic field of several Tesla, then it is thermally
isolated from the surroundings and finally demagnetized adiabatically to a field near zero, whereby
considerable cooling results7. When combined with an appropriate cryocooler, an ADR can provide
very low temperatures without the use of cryogenic liquids. Nowadays ADRs find increasing
interest for cooling of superconducting detectors (microcalorimeters or tunnel-junctions), as used
e.g. in high-resolution x-ray spectroscopy for microanalysis of materials9.
Recently, it was shown that a 4 K two-stage PTR is capable to conductively cool a small
superconducting 2.8 T magnet10. Here we present the first operation of an ADR by use of a
two-stage PTR. The field for magnetization of the paramagnetic salt is provided by a
superconducting 5 T NbTi-magnet which is directly coupled to the 2nd stage cold end of the PTR.
EXPERIMENTAL DETAILS
Two-Stage Pulse Tube Refrigerator
Description of PTR. Fig. 1 a) displays a schematic drawing of the cold head of the two-stage
PTR, and a photograph of it is shown in Fig. 2. The pressure oscillation in the refrigerator is
generated by means of a valved helium-compressor for Gifford-McMahon coolers (Leybold model
CP 6000, nominal input power 6 kW) in combination with a motor driven rotary valve which
periodically connects the high and low pressure side of the compressor to the main inlet of the PTR
cold head, as described in more detail in Refs. 2-3. In order to reduce vibrations and noise from the
mechanical rotary valve, the valve is separated from the main inlet by means of a pressure line with
length of 30 cm.
The dimensions of the two pulse tubes and regenerators in Fig. 1 a) are essentially the same as
that of the formerly described PTR4. Some modifications have been made in order to make the cold
head more reliable and user-friendly. The cold ends of the two stages are made of rigid copper
blocks that are connected to the stainless steel tubes by vacuum brazing. As the former versions2-3,
the PTR is operated in double-inlet configuration. The two reservoirs and the four flow impedances
(needle valves) for controlling the phase shift between pressure and mass flow oscillation are
located just outside the vacuum jacket. In contrast to the former PTR cold head4, the present system
operates without an additional valve for controlling the de-flow in the cold head. Previous work
have shown that proper adjustment of de-mass flow in the cold stage is essential for achieving
liquid helium temperatures4,11. In the present system dc-flow control is accomplished by adjusting
the timing of the rotary valve and the flow symmetry of the second-inlet needle valves that connect
the warm ends of regenerators and pulse tubes (see Figs. 1 a) and 2).
In view of operation in high magnetic fields the 2nd stage regenerator tube was packed with
antiferromagnetic spheres (Toshiba) in the coldest part, and with ErNi (Toshiba) and Pb-
5%Sb (Indium corporation) spheres in the middle and warm part, respectively.
Cooling Power. The PTR was filled at room temperature with the standard charging pressure
of 15.5 bar for the CP 6000 compressor. After cooling down to minimum temperatures on both
stages, this corresponded to an average pressure of 12 bar and a peak-to-peak pressure variation of
9.7 bar, as measured at the main inlet at the operating frequency of 1 Hz. A typical load map of the
two-stage PTR is depicted in Fig. 3. The data were taken with fixed needle valve settings that
before were optimized for minimum temperature with a applied load of 0.3 W on the 2nd stage. In
that case the minimum no-load temperatures are 2.56 K at the 2nd and 39.2 K at the 1st stage, and
the PTR provides cooling powers of 10 W at 57 K on the 1st stage simultaneously with 0.4 W at
4.17 K on the 2nd stage.
Adiabatic Demagnetization Refrigerator
The integration of the ADR into the two-stage PTR is illustrated in Fig. 1 b). For clarity,
only the cold end of the 2nd pulse tube and the cold platforms of 1st and 2nd stage are shown in
this Figure. The main parts of the ADR are the superconducting magnet with current leads and
the paramagnetic salt pill with thermal switch, which are described below in more detail. Not
shown in Fig. 1 are the two copper radiation shields that are attached to the cold platforms of the
REACHING 96 mK BY PULSE TUBE PRECOOLED ADR 599
Figure 1. a) Schematic drawing of the two-stage PTR cold head; b) integration of the adiabatic
demagnetization refrigerator into the PTR. and TP denote the thermometers at the lstand
2 stage, the magnet surface, and the FAA pill, respectively. Note: from the PTR only a part of the 2nd
nd
pulse tube and the cold platforms of 1st and 2nd stage are shown in b).
1st and 2nd stage.

Magnet Coil and Current Leads. The magnetic field around the paramagetic salt is
provided by a commercial superconducting solenoid (Cryogenic Ltd.). The magnet coil is made
of multifilament NbTi-wire wound on an aluminum bobbin with a clear bore of 40.6 mm and a
length of 154 mm. The coil inductance is 1.4 H and the mass is about 5 kg. The specified field of
this magnet is 5.0 T at 4.2 K, and the current to field ratio is 12.5 A/T. The magnet is cooled via
the aluminum bobbin, which is thermally anchored to the copper cold end of the 2nd PTR stage.
Each of the two current leads consists of three sections that were connected in series by soft
soldering. The first section from room temperature to the 1st stage cold platform is made of
brass tape introducing a conductive heat load of about 8 W to the 1st stage cold platform. The
second section extending to the cold platform of the 2nd stage, is made of high-Tc BiSrCaCuO
Figure 2. Photograph of PTR cold head.
Figure 3. Load map of two-stage pulse tube refrigerator. and denote the cooling
power and temperature of 1st and 2nd stage, respectively.
(Bi-2223) superconducting tapes with Ag-15 wt.% Au sheath of low thermal conductance
(supplied by BICC Cables Ltd.). These tapes with cross section of mm have a
critical current of 45 A at 77 K in self magnetic field. Two Bi-2223 tapes in parallel, each about
200 mm long, are used for each of the two current leads. The thermal load to the 2nd stage cold
platform from these four tapes is estimated to be less than 60 mW. The final section of the current
leads, which is connected to the magnet terminals, consists of multifilament NbTi-wire.
Paramagnetic Salt Pill and Thermal Switch. In our experiments the paramagnetic salt for
magnetic refrigeration was ferric ammonium alum (FAA). FAA salt has a high molar spin
entropy of In 6 in zero field at temperatures above 2 K, and has a low enough magnetic
ordering temperature of about 35 mK that makes possible to reach temperatures below 0.1 K by
adiabatic demagnetization7,8.
The FAA salt is enclosed in a hermetically sealed stainless steel cylinder with epoxy end
caps. Thermal contact between the FAA salt crystals and the outside of the ADR stage is
achieved by a copper rod extending through the lower end cap of the pill housing (see Fig. 1 b)).
Inside the housing the copper rod is thermally attached to a bundle of fine gold wires that are
distributed in the salt, see e.g. Ref. 8 for more details.
The FAA pill is suspended inside the bore of the magnet coil by Kevlar filaments with very
low thermal conductance. Thermal contact between the upper end of the pill housing and the 2nd
stage cold end is accomplished by means of a mechanical heat switch that is mounted on the 2nd
stage cold platform. A strong dc-motor serves to operate the switch via a shaft and a worm gear,
in order to achieve the high contact pressure necessary for low thermal resistance. As indicated
in Fig. 1 b) the shaft consists of two pieces that can be de-coupled after use, to avoid additional
heat load to the 2nd stage cold platform.
Thermometry. During the ADR tests, temperatures at the 2nd stage heat exchanger and at
the magnet surface were monitored using calibrated carbon glass resistance thermometers and
that of the 1st stage using a Pt-100 thermometer. The positions are indicated in Fig. 1 b)
and The temperature of the FAA pill was measured by a specially prepared Speer carbon
resistor that was fixed to the copper rod at the lower end of the pill housing ( in Fig. 1 b)). This
Speer resistor was previously calibrated down to 10 mK against the magnetic susceptibilty of a
CMN thermometer in a dilution refrigerator.
ADR TEST RESULTS

Typical cool down curves of the two pulse tube stages with built-in ADR are shown in Fig.
4 on a logarithmic temperature scale. The temperature of the 2nd stage drops below 4.2 K after
12½ hours of operation and attains a temperature of 2.8 K after 13 hours. The first stage cold
platform is then at 59 K.
The magnet was successfully charged to a current of up to 55 A, corresponding to a central
field of up to 4.4 T, for more than 10 times. A maximum sweep rate of 10 A/min was used.
The temperatures of the 2nd PTR stage, the magnet surface, and FAA pill during for two
successive adiabatic demagnetization experiments are displayed in Fig. 5. During ramping of the
magnet current the thermal switch to the salt pill was closed. In the first run in Fig. 5, the magnet
coil was charged with a field of 4.0 T using a rate of 0.8 T/min. Upon ramping the temperature
of the 2nd stage increased from 2.8 K to 2.9 K and that of the magnet surface from 2.9 to 3.6 K.
The rather high temperature rise of the FAA pill from 2.9 K to 4.6 K is due to the heat of
magnetization that has to be removed by heat conduction through the thermal switch. At constant
field of 4.0 T the system slowly cooled back to the initial temperatures within one hour. This
indicates that the heat load to the 2nd stage resulting from ohmic losses in the current lead system
are negligibly small in the present design.
After opening the motor-operated thermal switch, the magnetic field was slowly ramped
down at a rate of 0.16 T/min in order to keep eddy current heating low. As seen from Fig. 5, at
zero current (as indicated by in the Figure), the temperature at the FAA pill was 230 mK.
Figure 4. Cool down curves of the two PTR stages with the ADR attached.
Figure 5. Temperature of 2nd PTR stage magnet surface and FAA pill vs time for two
successive adiabatic demagnetization tests.
The actual field surrounding the salt pill immediately after the current was swept to zero is
not known, but is supposed to be of the order of 10 mT originating from trapped flux in the
NbTi-windings.
During the next 12 hours the salt pill warmed-up to about 2 K. Then the experiment was
repeated using a higher field of 4.4. T. After demagnetization an even lower minimum
temperature of 96 mK was achieved. This lower temperature is ascribed to the higher
magnetization field and to a lower precooling temperature of 2.1 K of the FAA pill, as compared
to the initial temperature of 2.9 K in the first run in Fig. 5.
Fig. 6 shows the warm-up behavior of the FAA pill after demagnetization to 96 mK plotted on
Figure 6. Warm-up behavior of the FAA pill after adiabatic demagnetization to 96 mK. The steps of
mK are due to the finite resolution of the digital data acquisition system.
a linear temperature scale. Within the first 1 ½ hours the pill temperature rises only slightly from
96 mK to 100 mK, and then increases further to 140 mK after 4 hours.
CONCLUSIONS
The above results demonstrate that a two-stage 4 K pulse tube refrigerator is capable to
conductively precool an adiabatic demagnetization refrigerator. The newly built PTR provides
cooling powers of 0.43 W at 4.2 K and 10 W at 57 K at 2nd and 1st stage, respectively, which are
sufficient to cool the NbTi-magnet coil and the FAA salt pill to temperatures below 3 K.
A minimum temperature of 96 mK was reached after demagnetization, starting from a field
of 4.4 T. A temperature of 100 mK is already sufficient for cooling of cryogenic detectors for
high-resolution x-ray spectroscopy.9 However, for practical applications a warm-up rate of the
ADR stage that is well below that achieved in the present tests, is highly desirable. We expect a
lower warm-up rate and also a lower minimum temperature after reducing the heat leak to the
ADR stage.
ACKNOWLEDGMENTS
We are grateful to Christoph Heiden, who died on 28 March 2000, for many helpful
comments and continuous support of the work. We thank Wolfgang Blend] (BICC Cables Ltd.,
Wrexham U.K.) for supplying us with Bi-2223 tapes.
REFERENCES
1. Matsubara, Y. and Gao, J.L., "Novel Configuration of Three-Stage Pulse Tube Refrigerator for
Temperatures below 4 K", Cryogenics, vol. 34 (1994), p. 259.
2. Thummes, G., Bender, S., and Heiden, C., "Approaching the 4He Lambda Line with a Liquid Nitrogen
Precooled Two-Stage Pulse Tube Refrigerator", Cryogenics, vol. 36 (1996), p. 709.
3. Wang, C., Thummes, G., and Heiden, C., "A Two-Stage Pulse Tube Cooler Operating below 4 K",
Cryogenics, vol. 37 (1997), p. 159.
4. Wang, C., Thummes, G., and Heiden, C., "Performance Study on a Two-Stage 4 K Pulse Tube Cooler",
Advances in Cryogenic Engineering, vol. 43, Plenum Press, New York (1998), p, 2055.
5. Wang. C. and Gifford, P.E., "0.5 W Class Two-Stage 4 K Pulse Tube Cryorefrigerator", Advances in
Cryogenic Engineering, vol. 45 (2000), in the press.
6. Satoh, T., Onishi, A., et al., "Development of 1.5 W 4 K G-M Cryocoooler with Magnetic Regenerator
Material", Advances in Cryogenic Engineering, vol. 41, Plenum Press, New York (1996), p. 1631.
7. Pobell, F., Matter and Methods at Low Temperatures, Springer-Verlag, Berlin (1992), pp. 148-157.
8. Wilson, G.W. and Timbie, P.T., "Construction Techniques for Adiabatic Demagnetization
Refrigerators Using Ferric Ammonium Alum", Cryogenics, vol. 39 (1999), p. 319.
9. Frank, M., et al., "Cryogenic High-Resolution X-Ray Spectrometers for SR-XRF and Microanalysis",
J. Synchrotron Radiation, vol. 5 (1998), p 515.
10. C. Wang,C., Thummes, G., Heiden, C., Best, K.-J., and Oswald, B., "Cryogen free operation of a
Niobium-Tin magnet using a two-stage pulse tube cooler", IEEE Transactions on Applied
Superconductivity, vol. 9 (1999), p. 402.
11. Chen G.B., Qiu, L.M., and Zhen, J.Y., "Experimental Study on a Double-Orifice Two-Stage Pulse
Tube refrigerator", Cryogenics, vol. 37 (1997), p. 271.
Dissipation in Metal Welded Bellows and Its
Consequences for Sub-Kelvin Refrigerators
Carolyn L. Phillips and J.G. Brisson

Cryogenic Engineering Laboratory
Massachusetts Institute of Technology
Cambridge, Massachusetts, USA 02139
ABSTRACT
The energy dissipation of welded bellows due to flexure is measured for several types of
Senior Flexonics welded bellows at 1.4 K. The energy dissipated by the bellows is found to vary
from 0 to 350 µJ/cycle for strokes between 0 and 1 cm. The results are non-dimensionalized using
the manufacturer’s specifications in an attempt to generate a characteristic equation to estimate
the energy dissipation rate for the entire family of Senior Flexonics bellows. The dissipation rates
are compared to the overall losses in a superfluid Stirling refrigerator that have used this type of
bellows. The losses due to bellows flexure account for less than 5% of the overall Stirling refrig-
erator losses. An ultra-low temperature isothermal expander made using these bellows is modeled
using a 3% 3He-4He mixture. The results suggest that the expander built with these bellows can
provide cooling below 100 mK.
INTRODUCTION
A new class of sub-Kelvin refrigerators that use moving parts at low temperature was intro-
duced to the low temperature community with the development of the first superfluid Stirling
refrigerator (SSR) by Kotsubo and Swift in 1990.1 Superfluid Stirling refrigerators have shown
typical cooling powers of milliwatts below 1 K and have, to date, achieved ultimate temperatures
of 170 mK.2 Other members of this class include the superfluid orifice pulse-tube refrigerator3,
the cold cycle dilution refrigerator4 (CCDR) and the superfluid Joule-Thomson refrigerator5 (SJTR).
The waste heat of these machines is typically exhausted to a pumped He bath at a temperature
between 1 and 2 K.
The working fluid for these machines is superfluid 3He-4He mixture. The concentration of the
normal 3He component of the liquid is increased by using a superfluid compressor. The compres-
sor has superleak-bypassed pistons that compress the normal 3He component of the 3He-4He
mixture while allowing the superfluid 4He component to flow freely between the piston and a
ballast volume. The compressor operates in a vacuum at temperatures below 2 K where the
working fluid is superfluid. Sliding seals of a standard piston/cylinder apparatus cannot be used in
the compressor as they are too dissipative and would allow the superfluid component of the
mixture to leak into the vacuum space. The pressure oscillations in these refrigerators are gener-
ally below an atmosphere. Bellows are ideal for use as a piston/cylinder in these applications

because of their low dissipation, leak tight structure, and ability to tolerate the low internal pres-
sures present in these refrigerators.
There has been significant work on many superfluid Stirling refrigerator designs.6 Unfortu-
nately, due to the dearth of data, most of these designs have been made with little or no informa-
tion on the flexing bellows dissipation. Since the bellows dissipation was unknown, the overall
losses could not be specifically assigned to the heat exchangers or bellows with confidence. Brisson
and Swift realized that this was an issue and reported small-displacement dissipation measure-
ments for the bellows they used in their SSR.7 Unfortunately, subsequent SSR performance mea-
surements utilized piston strokes that were outside the range of their dissipation measurements.
Here, we report on measurements of the dissipation of bellows used in the SSR’s. First we
discuss the experimental apparatus and procedure followed by the results. We then attempt to
generalize these results for an entire family of bellows using a simple model and dimensional
analysis. We then briefly compare our bellows dissipation results with the SSR performance re-
sults. Finally, the potential performance of an isothermal expander built with these bellows is
discussed. The results indicate that such an expander could be used to cool below 100 mK.
APPARATUS
The apparatus is shown in Fig. 1. The top platform consists of a 4He evaporation refrigerator
that is maintained at a temperature of 1.4 K. The platforms are suspended in a vacuum space and
are structurally supported using Kapton/epoxy composite stantions (not shown). These Kapton/
epoxy stantions also provide thermal isolation between the platforms. A weak thermal link be-
Flgure 1. Schematic diagram of the test apparatus.

DISSIPATION IN METAL WELDED BELLOWS 607
tween each of the lower platforms and the refrigeration platform is provided by a large brass rod.
The brass rod that thermally links platform 1 to the refrigeration platform is 1.9 cm in diameter
and 25 cm long. The brass rod that thermally links platform 2 to the refrigeration platform is 2.5
cm in diameter and 50 cm long. One end of each rod is thermally connected to the refrigeration
platform using a flexible stack of six OFHC copper sheets to avoid thermally induced stresses in
the apparatus as it is cooled down. Each copper sheet is approximately 1.5 cm by 4 cm by 0.13
nun thick.
The two bellows mounted on platform 1 are driven sinusoidally by a thin-walled stainless-
steel drive shaft. This shaft is driven by a room-temperature motor-cam mounted on the top of the
cryostat. The stroke is directly measured at low temperature using a linear transducer8 mounted
on the drive shaft at 4 K (not shown in Fig. 1). The bellows on platform 1 are connected by three
“cage” bars that pass through oversize holes in platform 1; two of these bars are shown in Fig. 1.
There is no contact between the cage bars and the walls of the oversize holes. The two bellows on
platform 2 are mounted in a similar way and are driven by a 2.2 cm long, 0.95 cm diameter Teflon
rod that mechanically connects the bottom bellows on platform 1 to the top bellows on platform 2.
The bellows used are of the welded bellows type made by Senior Flexonics.9 All the bellows used
in the SSR have contoured diaphragms that nest into each other to minimize the clearance volume
in the compressed bellows. The bellows are made from 347 stainless steel. The bellows on plat-
form 1 are 60050-1 edge welded bellows. The manufacturer’s specifications for these bellows
appear in Table 1. The bellows’ stainless steel flanges are soft soldered onto brass flanges that are, in
turn, bolted to the copper platform. The bellows on platform 2 are 60035-2 edge-welded bellows. Once
again, the manufacturer’s specifications for these bellows appear in Table 1.
Each platform is instrumented with a calibrated resistance thermometer and a heater, de-
picted in Fig. 1 as rectangles with symbol “T” and “H”, respectively. The thermometer resis-
tances are measured using standard four point AC measurement techniques. The heaters are wired
with four leads to allow the voltage across the heater and the current through the heater to be
directly determined.
PROCEDURE
The refrigerator platform was cooled to a steady 1.4 K and the platforms were allowed to
come to thermal equilibrium with the platform. The bellows were then driven sinusoidally with a
stroke of 0.98 cm and a period of 11.1 seconds. The system was left operating for several hours to
allow the platform temperatures to come to steady state. This long tune-constant is a consequence
of the large mass of copper in each of the platforms. (The platforms are the isothermal heat
exchangers used in our previous SSR studies.) The long time-constant dampens out any short term
temperature oscillations in the platform as the bellows are flexed. Once the temperatures of the
platforms stabilized, the temperatures were noted and the bellows drive was shut off. The plat-
forms were then maintained at their steady state “bellows flex” temperature by electrically heating
the appropriate platform heater. After the power to the heaters and the temperatures of the plat-
forms stabilized, the currents through the heaters were measured. The power dissipated in the
heaters was calculated using previously measured 4 K resistances for the manganin wire heaters.
The procedure was repeated for strokes of 0.38, 0.45, 0.66, 0.81, and 1.03 cm with periods
of 5.89, 6.01, 5.96, 6.1, and 9.5 seconds, respectively.
Figure 2. A plot of the measured bellows dissipation as a function of stroke length. Each curve is
labeled with the corresponding bellows type.

The dissipation/cycle versus stroke is shown in Fig. 2. This figure includes the bellows dissi-
pation reported by Brisson and Swift10 for type 60030-1 Senior Flexonics bellows. Please note that
these data are scaled for a single bellows despite the measurements being made with the two
opposing bellows per platform shown in Fig. 1.
There are many models for the elastic behavior of bellows11 that depend on the specific
geometry of the bellows. Here we would like to develop a very simple model for the dissipation to
allow the scaling of our measured results to this entire class of Senior Flexonics bellows. This
simple model would provide the designer of ultra-low-temperature cryocoolers with reasonable
estimates of bellows dissipation that will allow him to assess the design tradeoffs to optimize a
given cryocooler design. The bellows is a stack of annular disks (diaphragms) alternately welded
in their inner and outer radii. The diaphragms are shaped to improve the performance but the
specifics of the shape are proprietary. For the moment, we will assume that all the bellows dissi-
pation occurs in the bead of the weld. It seems reasonable then, that the energy dissipation per unit
length of the weld bead will depend only on the variation of an angle θ that the bellows diaphragm
makes with the horizontal. The variation of the angle, ∆θ, can be estimated from the number of
convolutions in the bellows N (two diaphragms constitute a convolution), the stroke S, and the
radial width of the annular diaphragm L (which is the same as the difference of the inner bellows
radius from the outer bellow radius). Assuming that the angle 6 is small, our assumptions can be
expressed as
where is the energy dissipated by the bellows per cycle, is the outer bellows diameter,
is the inner bellows diameter, and f is a function yet to be determined. Our data and that of Brisson
and Swift are plotted in Fig. 3. The agreement between the type 60030-1 and 60035-2 bellows is
very good. The type 60050-1 bellows, however, delivers much better performance on this basis
than the other two bellows. A fit to the type 60030-1 and 60035-2 data can be used as an upper
bound for the dissipation in the type 60050-1 bellows. The exponential fit shown in Fig. 4 to the
type 60030-1 and 60035-2 dissipation values is
Figure 3. Energy dissipated per unit weld bead length per cycle versus the normalized stroke as per
Eq. 1.
Figure 4. A plot of non-dimensional energy dissipation per cycle versus the non-dimensional stroke
as per Eq. 3 for the three sets of bellows data. The fit curve is Eq. 4.
where A is 0.010 µJ/cm cycle, and has the value 62. Unfortunately, this model is not easily
generalized and the data necessary to generalize this model are unavailable from the manufacturer.
Another approach is to non-dimensionalize the data using parameters that are published by the
manufacturer. Table 1 contains the manufacturer’s values for the bellows spring rate, k, and the
maximum recommended stroke, Using these values to non-dimensionalize the energy dissi-
pation and the stroke we suggest
Figure 5. Percentage of total losses due to bellows 60050-1 dissipation versus refrigeration
temperature in Patel and Brisson’s single-stage SSR.
where G is a function to be determined. In Fig. 4, the data are plotted on the basis of Eq. 3 and
fitted to an exponential function of the form:
where C has a value of and has a value of 5.8. This fit overestimates the dissipation for
the small strokes reported by Brisson and Swift. Assuming that there are no other relevant dimen-
sionless groups, Eq. 4 can be used as an estimator for the dissipation in the Senior Flexonics OTS
type bellows. (Senior Flexonics specifies 16 basic types in this product line with outside diameters
that vary from 0.95 to 40.6 cm.)
Bellows Dissipation in the SSR

There are significant loss mechanisms in the superfluid Stirling refrigerator and there has
been an open question as to how large the contribution of the bellows flexure losses are in com-
parison to the overall losses. Pate] and Brisson12 assert that the losses in their single-stage SSR are
primarily due to recuperator losses. We have taken their cooling power data and their theoretical
phonon-roton cooling power curve presented in their Fig. 17a and compared their total losses to
our measured bellows losses. This is done by using our data to generate a bellows dissipation rate
using their 0.69 cm stroke value and the fact that they use a 60050-1 bellows for their expander
stage. The total loss is calculated by taking the difference between the actual cooling power and
their phonon-roton model predictions. The ratio of these results as a function of the expander
temperature is shown in Fig. 5. Clearly, throughout the low temperature range of their measure-
ments the bellows losses account for less than five percent of the overall losses in their SSR.
Temperature Limits on using Bellows Expanders
It is interesting to consider the consequences of these bellows dissipation results on the limits
of performance of mechanical expanders for sub-Kelvin refrigeration. For this we will examine a
highly idealized model of an isothermal expander with non-dissipative valves. All the expander
losses are assumed to be linked to the dissipation due to the direct flexing of the bellows as
Figure 6. Performance plots for an isothermal expander versus temperature, a) Cooling power
versus refrigeration temperature for the dissipative expander, b) Ratio of cooling power of dissipative to
ideal expander as a function of refrigeration temperature.
described by Eq. 4 above. We ignore the effect of pressure oscillations internal to the bellows
on dissipation. The working fluid entering the expander is a liquid 3 % 3He-4He mixture. The
expander piston is bypassed by a superleak to a large 4He reservoir to allow the expander to
operate only on the 3He component of the mixture. We will assume for the purposes of this
discussion that the 3He behaves as an ideal gas throughout the expansion process and the 4He
component behaves as an inert atmosphere. The molar volume of the 3He is assumed to be of the
form where x is the molar concentration of 3He. The working fluid is
assumed to enter and exit the expander at the temperature of the expander. The clearance volume
in the expander for a given bellows is determined by assuming there is a 0.125 mm wide gap in
each convolution of the bellows when the bellows is in the compressed state. The analysis requires
finding the optimal valve timing and stroke for both the non-dissipative and dissipative cases. The
calculated optimal cooling powers of the expander are plotted in Fig. 6a as a function of tempera-
ture for the dissipative case. Figure 6b shows the ratio of the cooling power using dissipative
bellows (that follow Eq. 4) to that of using non-dissipative bellows versus refrigeration tempera-
ture. The analysis suggests that these bellows can be used to provide cooling to temperatures
below 100 mK using 3He-4He mixtures. According to Fig. 6b, more than 40% of the maximum
available cooling power is available to the non-ideal expander using bellows type 60050-1 for
temperature above 50 mK. It should be noted that the Fermi temperature for the 3He in a 3%
mixture is about 220 mK so that the use of the Boltzmann ideal gas model is questionable for
temperatures well below 100 mK.
CONCLUSIONS
We have measured the energy dissipation in bellows that have been used in superfluid Stirling
refrigerators. We have proposed a non-dimensional relation to estimate the low-temperature dissi-
pation for a complete class of bellows supplied by Senior Flexonics. Our measurements suggest
that bellows dissipation effects in the SSR has not been significant in measurements made to date.
Furthermore, theoretical analysis of an isothermal expander using the dissipative bellows suggests
that cooling can be achieved with moving parts to temperatures below 100 mK.
ACKNOWLEDGMENT
We gratefully acknowledge the support of the National Science Foundation.
REFERENCES
1. V. Kotsubo and G.W. Swift, “Superfluid Stirling-cycle refrigeration below 1 Kelvin,” J. Law Temp.
Phys., 83 (1991), p. 217.
2. A. Watanabe, G.W. Swift and J.G. Brisson, “Measurements with a recuperative superfluid Stirling
refrigerator,” Advances in Cryogenic Engineering, Edited by P. Kittel, v. 41, Plenum, NY (1996),
pp. 1527-1533.
3. A. Watanabe, G.W. Swift and J.G. Brisson, “Superfluid orifice pulse tube refrigerator below 1
Kelvin,” Advances in Cryogenic Engineering, Edited by P. Kittel, v. 41, Plenum Press, NY (1996),
pp. 1519-1526.
4. J.G. Brisson, “Cold-cycle dilution refrigeration”, J. Low Temp. Phys., 111 (1998), p. 181.
5. J.G. Brisson, “Superfluid Joule-Thomson refrigeration,” J. Low Temp. Phys., 120 (2000).
6. See for example:
A.B. Patel and J.G. Brisson, “Preliminary experimental results using a two stage superfluid Stirling
refrigerator,” Cryocoolers 10, Edited by R. Ross, Plenum Press, New York (1999), pp. 655-662.
A.B. Patel and J.G. Brisson, “Experimental performance of a single stage superfluid Stirling refrig-
erator using a small plastic recuperator,” J. Low Temp. Phys., 111 (1998), p. 217.
7. J.G. Brisson and G.W. Swift, “Superfluid Stirling refrigerator with a counterflow regenerator,”
Proceedings of the Seventh Cryocooler Conference, Phillips Laboratory, Kirtland AFB, NM (1993)
p. 460.
8. K. A. Backes and J.G. Brisson, “An inductive position sensor for the measurement of large displace-
ments at low temperature,” Cryogenics, 69 (1998), pp. 599-602 .
9. Senior Flexonics Inc., Sharon, MA
10. J.G. Brisson and G.W. Swift, “Superfluid Stirling refrigerator with a counterflow regenerator,”
Proceedings of the Seventh Cryocoolers Conference, Phillips Laboratory, Kirtland AFB, NM (1993),
p. 460.
11. See for example:
J.F. Wilson, “Mechanics of bellows: a critical survey,” Int. 3. Mech. Sci., 26, 11/12 (1984), p. 593-605.
12. A.B. Patel and J.G. Brisson, “Experimental evaluation of a single stage superfluid Stirling refrigerator
using a large plastic recuperator”, J. Low Temp. Phys., 118 (2000), pp.189-206.
Design and Predicted Performance of an Optical
Cryocooler for a Focal Plane Application
G.L. Mills, A.J. Mord, P.A. Slaymaker

ABSTRACT
Optical refrigeration by fluorescence is a new concept in refrigeration that represents a
revolutionary advance in small cryocoolers. It uses anti-Stokes fluorescence to remove heat from
a glass or crystal that is pumped with laser light.
We have completed a numerical model of the cooling process and designed an optical
crycooler for a focal plane cooling application. Such a cryocooler appears feasible and has
several advantages when compared to conventional devices, including small size, long life (no
moving parts), zero vibration, low electromagnetic interference (EMI), and potentially low cost.
Performance estimates of the device are presented based on the known performance of the
cooling material and a detailed thermal model. The performance is compared to existing
mechanical and thermoelectric coolers.
INTRODUCTION
The basic principle of cooling by anti-Stokes fluorescence was suggested as early as 1929,1
but it was not until 1995 that the actual cooling of a solid was first demonstrated by Epstein et al.
at Los Alamos National Laboratory (LANL) using Yb doped Zirconium Fluoride (Yb:ZBLAN)
glass.2,3 In 1996, Clark and Rumbles reported cooling in a dye solution of rhodamine 101 and
ethanol.4 A collaborative effort by LANL and Ball Aerospace resulted in a cylinder of
Yb:ZBLAN cooling 48 °C below the ambient temperature.5 Gosnell has reported cooling of
65 °C in a Yb:ZBLAN fiber.6
The fundamental refrigeration cycle of fluorescent cooling is simple. In the case of the Yb:
ZBLAN material, the presence of the internal electric fields of the host ZBLAN material cause
the ground and first excited states of the ion to be split into multi-level manifolds as shown
in Figure 1. A photon from a laser tuned appropriately will be absorbed only by an ion that has
been thermally excited to the highest level of the ground-state manifold, and will promote that
ion to the lowest level of the excited-state manifold. When that ion decays radiatively, it can fall
to any of the four ground-state levels. On average the outgoing fluorescent photon will therefore
carry slightly more energy than the pump photon absorbed. By selectively “picking off” the
“hottest” ions, this process depletes the population of the highest ground-state level. Thermal
equilibrium is reestablished when another ion is promoted to that level by absorbing a phonon
from the host material. The absorption of this phonon constitutes the refrigeration.

614 OPTICAL REFRIGERATION DEVELOPMENTS
Figure 1. The photon-phonon refrigeration cycle results from the energy levels of the ion in
the ZBLAN glass host material.
In summary, a dopant ion absorbs a pump photon and the photon is re-emitted slightly
bluer (higher energy). This energy difference comes from thermal vibrations (phonons) of host
material.
The simplest implementation of a cryocooler based on this principle is a simple Yb:ZBLAN
cylinder (cooling element) with high-reflectivity mirrors deposited on the ends, as shown in
Figure 2. The pump beam is introduced through a small feed hole in one mirror, and then
bounces back and forth until it is absorbed. A key feature of this arrangement is that the pump
light is confined to a nearly parallel beam, while the fluorescence is emitted randomly into
steradians. This makes it possible to allow the fluorescence to escape while trapping the pump
light inside. The fluorescent photons that are nearly parallel to the pump beam are also trapped.
They are reabsorbed and then simply try again with a small and calculable degradation to the
overall efficiency.
DESIGN STUDY
At Ball, we performed a detailed design study of a complete cryocooler to verify feasibility
and to determine the advantages and disadvantages of this technology. We chose to design an
optical cooler for cooling a generic focal plane. This application was chosen because it seemed
well suited to the capabilities and advantages of optical cooling and because of its broad use by
possible scientific, military, and commercial users.
Figure 2. High-reflectivity mirrors provide long pump path.

OPTICAL CRYOCOOLER FOR A FOCAL PLANE APPLICATION 615
Figure 3. Ball optical cryocooler design concept for an infrared detector/cryocooler dewar. This
specific design is capable of producing 400 mW heat lift at 80 K with a 21 W laser when the heat sink is
300 K.
Figure 3 shows our design concept for an optical cryocooler detector dewar. The cooling
element is bonded directly to the focal plane structure in order to absorb the heat. The
fluorescence is absorbed by a heat sink with high absorbtivity at this wavelength. The cooling
element and focal plane are supported by a folded tube of low thermal conductivity material, in a
manner similar to dewars that have been used for focal planes cooled by mechanical cryocoolers.
Note that the optical cooling element and heat sink are small compared to the structure that is
needed to support a focal plane at cryogenic temperatures.
An optical fiber is aligned and focused to transmit the light from the laser into the cooling
element. For the 400 mW cooler shown, a 20 W diode laser package would be required. This
would be a package made up of a number of individual diode laser modules, plus optical fiber
coupling, with dimensions of approximately 14 cm by 40 cm by 5 cm and a mass of 1.4 kg. It
could be located remotely from the dewar to optimize heat transfer from the laser modules, and
reduce EMI or magnetic fields, at the focal plane or for other system reasons.
PERFORMANCE MODELING
The performance of an optical cryocooler is limited by two kinds of effects: escape of pump
photons from the cooling element before they can do any good, and absorption of pump or
fluorescent photons by processes that produce heat. The heating effects are by far the more
serious.
The known pump photon loss mechanisms are: escape through the mirrors, escape past the
edges of the mirrors, escape through the feed hole, and scattering in the bulk material. The
known heating mechanisms are direct absorption by contaminants; absorption by followed
by “hopping” of the excitation to a contaminant (especially Fe); absorption by the mirrors or by
material beneath them; absorption by the cold stage of the cryocooler of photons reflected by the
heat sink; and radiative and conductive heat load from the cryocooler structure.
Three additional effects come into play: saturation of the absorption at high-power
densities, reabsorption and recycling of those fluorescent photons that cannot escape from the
sides of the cooling element, and absorption reddening of the fluorescence that escapes. These
effects are calculable and none is a significant limitation in an optimized design.
Because these many effects interact with the design parameters in different ways, a
comprehensive simulation is required to predict the performance and to understand the design
Figure 4. Performance of design concept optical cryocooler when pumped at optimum wavelength.
tradeoffs involved. Ball therefore developed a simulation we call our Photon Model that
addresses all of the photon effects (except scattering), and coupled it with a standard thermal
network model that estimates the heat transfer. At the heart of this simulation, is a realistic model
of the atomic transitions, including transition strengths, temperature-dependent level populations,
and line widths. The transition strengths and line widths were developed by LANL from their
spectral measurements. This simulation package has been our primary tool so far in evaluating
the technology and guiding our designs.
The expected cryocooler performance depends on the interplay between cooling element
dimensions and dopant concentration; dielectric mirror transmission and absorption; width of the
imperfect edge of the dielectric mirrors (unless low-index cladding is used to confine the pump
beam away from the mirror edges; size of the feed hole; pump wavelength and power; and
temperature.
Figure 4 shows the estimated performance of the conceptual cooler when pumped with 4 and
21 W of laser power with a 300 K heat sink. The ZBLAN was assumed to have been doped with
2% and the mirrors to be 99.999% reflective (“5 nines”). The cooling element was sized to
absorb 21 W without performance degradation due to saturation of the cooling ions. The
asymptotic limit is set for this particular design by the combination of saturation of the Yb
absorption at high-power density and the decreasing Boltzman population of the upper member
of the ground state manifold at low temperature.
There are materials other than Yb:ZBLAN that have the potential for cooling and may have
better performance due to wider spacing between the electron energy levels. Sheik-Bahae and
LANL have recently reported cooling in thulium doped ZBLAN when pumped at 1900 nm.7
COMPARISON TO OTHER REFRIGERATION TECHNOLOGIES

Currently, two different technologies are commonly used to actively cool focal planes,
depending on the temperature required. Thermoelectric coolers (TEC) are capable of cooling
focal planes to 180 K starting from an ambient temperature of 300 K. They are small,
lightweight, and, being solid state, have no vibration. TECs have the main disadvantage of
having efficiency that drops off rapidly with decreasing temperature becoming zero around
180K.
Mechanical coolers, such as Stirling cycle and pulse tube coolers, can produce tempera-
tures below 20 K and are more efficient than TECs. However, mechanical coolers are much
larger than TECs and can produce vibrations that must be canceled or isolated.
On spacecraft, focal planes are sometimes cooled with passive radiators, the performance
of which is highly dependent on mission parameters and spacecraft configuration. For simplic-
ity, we will not compare optical cryocoolers to passive radiators in this paper.
Vibration
Because a solid-state diode is used as the pump laser, there are no moving parts and there-
fore no vibration. This is an obvious advantage for imaging focal planes, an advantage that
cannot occur with mechanical coolers. TECs have this advantage as well.
Electromagnetic and Magnetic Noise

The pump laser can be located remotely from the cryocooler dewar allowing very low
levels of electromagnetic interference at the focal plane. This can be important if the focal
plane contains EMI or magnetic field sensitive devices such as SQUIDs. Optical coolers have
this advantage over all other active coolers. Work has been done on split-Stirling cycle coolers
with low magnetic noise,8 but such systems have reduced reliability because there is no elec-
tric displacer motor to overcome the effects of working gas contamination.
Reliability and Lifetime

An optical cryocooler has no moving parts, which enhances reliability and life. The laser
appears to be the lifetime limiting component. Currently, commercial diode modules have a
lifetime of several years in continuous operation. A laser package would consist of many diode
modules lasers feeding a single output optical fiber using optical Y junctions. This arrange-
ment would have inherent redundancy because the modules almost always fail shorted and
have a distribution of lifetimes that is gaussian. Additional redundant modules could be added
if necessary.
Some mechanical coolers and TECs have proven lifetimes in excess of five years, how-
ever, redundant coolers can be added only with the use of a heat switch because of the high
”off-state” conductance of these coolers.
Ruggedness
The cooling element is separated from the heat sink by a gap, and thus is inherently pro-
tected from physical stress. The glass, although brittle, has a compact form factor that will
allow it to withstand high accelerations. TECs and the cold tip of mechanical coolers have to
be isolated from loads by S-links or other compliant devices.
Cryocooler Mass and Volume

The overall estimated optical cryocooler mass, including the laser, will be smaller than a
mechanical cooler for typical focal plane cooling loads. An optical cryocooler, however, will
be more massive than a TEC. The lack of vibration allows the cooling element to be closely
integrated with the focal plane because S-links or other vibration and load isolation devices are
not required. Figure 5 illustrates the relative sizes and masses of the three technologies in a
typical installation that is cooling a 2.5-cm-square focal plane with a total load of 400 mW.
Figure 5. Relative sizes of the cryocooling technologies shown to scale for a 400 mW load. The
electronic and laser packages that can be located remotely are not shown, but are included in the mass
estimate.
Efficiency and System Mass

We have calculated the power efficiency of an optical cooler based on our photon model-
ing and information from laser vendors. Solid-state lasers currently have an efficiency be-
tween 25% and 60% when made up of diode modules. We assumed 50% laser efficiency for
the data shown. The calculated specific power of an optical cooler is shown with the measured
specific power of other cooling technologies in Figure 6.
The lower efficiency of optical cryocoolers and TECs reduces their advantage in power
and mass sensitive applications such as spacecraft. Mass is required not only for the power
generation, but also for rejecting the waste heat with a radiator. We used overall system mass
data for mechanical coolers using data from Glaister et al.,9 but increased the cooling load 20%
to account for the cold finger and S-link heat load. For the optical cooler and TEC we used a
system penalty of 0.28 kg/W, which is obtained adjusting data from Glaister and Curran10 to
account for the lower structure and heat transport requirements. Based on cooling temperature
and load, regions of minimum system mass were determined and plotted in Figure 7. As shown,
for spacecraft applications, optical cryocooling will likely have the lowest system mass when
the load is less than 1.0 W and the temperature is between 80 and 200 K. Optical cryocooling,
in effect, extends benefits of solid state cooling to this new, lower temperature region.
Cost
Optical cryocoolers will be producible at low cost in volume production; there are no
high- precision mechanisms involved, and all the components used are ones with proven track
records in cost-sensitive applications. Mechanical coolers have inherently high costs associ-
ated with high-tolerance moving parts. High-performance, multistage TECs are frequently as-
sembled by hand, but some manufacturers have achieved low cost in volume production of
single-stage TECs by using automated assembly. Even the hand-assembled TECs have a large
cost advantage over mechanical coolers. Optical coolers are likely to have a similar cost ad-
vantage over mechanical coolers.
Figure 6. Specific power of cryocoolers, including power supply, with 300 K sink temperature. The
pulse tube cooler data is for one with 0.5 to 3.5 W capacity and the Stirling cooler data is for one with 2.3
to 7.2 W capacity. The efficiency of mechanical coolers decreases with size.
Figure 7. Approximate regions of lowest system mass for an optimized spacecraft application.
CONCLUSIONS
Based on our design and analysis, we conclude that optical cryocooling is a feasible method
for cooling focal planes and has a distinct niche in extended solid-state cooling to those tem-
peratures that cannot be achieved efficiently (or at all) by TECs. Optical cryocoolers have a
clear advantage over mechanical coolers in vibration, ruggedness, EMI and magnetic field,
and cooler mass. Nevertheless, they are at a disadvantage to mechanical coolers based on effi-
ciency. Optical cryocoolers have an advantage over TECs in minimum operating temperature,
magnetic field, and ruggedness. They are at a disadvantage to TECs in cooler mass, however.
In a fully optimized spacecraft focal plane application, an optical cryocooler using the
Yb:ZBLAN material will likely have a lower system mass than mechanical coolers or TECs in
the 80 to 200 K temperature range for a cooling capacity of less than 1.0 W.
REFERENCES
1. Pringsheim, P., Z. Phys 57 (1929), p. 739.
2. Epstein, R.I., M.I. Buchwald, B.C. Edwards, T.R. Gosnell and C.E. Mugan, “Observations of Laser-
Induced Fluorescent Cooling of a Solid,” Nature 377, 500 (1995).
3. Epstein, R.I., et al., “Fluorescent Refrigeration,” U.S. Patent. No. 5,447,032 (University of California).
4. Clark, J.L., G. Rumbles, Phys. Rev. Letters, 76, 2037 (1996).
5. Edwards, B.C., J.E. Anderson, R.I. Epstein, G.L. Mills, and A. J. Mord, “Demonstration of a Solid-
State Optical Cooler: An Approach to Cryogenic Refrigeration,” Journal of Applied Physics, 86 (1999).
6. Gosnell, T.R. “Laser Cooling of a Solid by 65 K Starting from Room Temperature,” Optics Letters,
vol. 24, no. 15 (1999).
7. Edwards, B.C., J.E. Anderson, R.I. Epstein, “Solid State Optical Cooler Developments,” Cryocoolers
11, Plenum Publishers, New York (2001).
8. ter Brake, H. J.M., P. J. van den Bosch, and H. J. Holland, “Magnetic Noise of Small Stirling Coolers,”
Advances in Cryogenic Engineering, vol. 39(1994), pp. 1287–1295.
9. Glaister, D.S., M. Donabedian, D.G.T. Curran, and T. Davis, “An Overview of the Performance and
Maturity of Long Life Cryocoolers for Space Applications,” Aerospace Corporation Report, TOR-98
(1057)-3 (August 1998).
10. Glaister, D.S., and D.G.T. Curran, “Spacecraft Cryocooler System Integration Trades and Optimiza-
tion,” Cryocoolers 9, Plenum Press, New York (1997), pp. 873-884.
Optical Refrigeration Using Anti-Stokes
Fluorescence from Molecular Dyes
G.Rumbles, B.Heeg and J.L.Lloyd (née Clark)

Centre for Electronic Materials and Devices, Imperial College
London, SW7 2AY, UK
P.A.DeBarber
MetroLaser, Inc.
Irvine, CA 92614, USA
B.J.Tomlinson
Kirtland AFB, NM 87117, USA
ABSTRACT
Irradiating a sample of a luminescent material into the low energy tail of the first electronic
absorption band generates anti-Stokes luminescence and provides a means of removing thermal
energy from the sample and thus lowering its temperature. A recent study of the molecular dye,
rhodamine 101, dissolved in acidified ethanol has shown this novel, optical cooling effect. We
examine the merits of using dye molecules and discuss how these can be modified to provide a
suitable material that can be used as a cooling medium in an optical refrigerator. The issues of
Stokes loss, re-absorption, excited state lifetime and radiative energy transfer in the context of
optimizing the intrinsic cooling efficiency of the material are also discussed.
INTRODUCTION
The prediction that anti-Stokes luminescence could be used to effect optical cooling in the
condensed phase was made by Pringsheim,1 Vavilov2 and Landau3 as far back as 1946. The experi-
mental realization of this unusual phenomenon, however, was not made until very recently. Within
the past 5 years, there have been reports of cooling in glasses doped with rare earth elements,4,5,6,7,8,9,10
semi-conductors11,12 and molecular dyes in solution.13,14,16,17,18 The concept is counter-intuitive, as
the irradiation of most materials with intense laser light leads to heating, burning and even explo-
sions, and these effects are well known and understood. The normal, or Stokes luminescence oc-
curs to lower energy or longer wavelength than that used to excite it. Anti-Stokes luminescence,
however, occurs to higher energy, or shorter wavelength, than that used to excite it. A further and
more critical constraint on the material, is a high photoluminescence quantum yield (PLQY), such
that in excess of 98% of the absorbed photons are eventually re-emitted. To achieve high external
PLQY values in semi-conductors, samples have been first cooled,12 although the possibility of
Cryocoolers 11, edited by R.C. Ross, Jr.

ambient temperature cooling is close to reality.11 For the doped glasses and molecular dyes, PLQY
values close to 100% are attainable at ambient temperatures. The work of groups at Los Alamos
National Laboratory,19,20 Ball Aerospace20 and, more recently, the University of Queensland10 fo-
cuses on the use of ytterbium ions as the active species in ZBLANP and ZBLAN glasses.
Advances in these areas are discussed elsewhere in these proceedings.21,22 In all the cases men-
tioned, the physics of the optical cooling is via an anti-Stokes luminescence mechanism and only
the cooling materials differ.
In this article, we summarize our work on molecular dyes dissolved in fluid solution as the
anti-Stokes laser cooling (ASLC) medium. Only two distinct reports have been made in this area;
the initial work of Drexhage13 and our group at Imperial College.14,16,18 Drexhage and co-workers
saw local cooling in solutions of the laser dye rhodamine 6G, and they demonstrated the impor-
tance of sample and solvent purity in obtaining a sample with a near-unity PLQY. We observed a
drop in sample temperature of 4 K from ambient using the laser dye rhodamine 101 in acidified
ethanol. In this article, we report on the factors that limit the efficiency of ASLC and the models
that we have developed to define a system that can form the heart of an optical refrigerator. It is
worth stressing that we do not consider the molecular dye-based materials to be a competitor to the
glasses, but an alternative approach to achieve the same objective. In both cases, the
materials offer the same benefits to refrigeration by being pumped optically. Benefits such as vibra-
tion-free, non-mechanical operation; the possibility of working in a hostile environment, where
metallic components cannot be tolerated; and the ability be accessed remotely, with the pumping
power transmitted optically.
The full experimental results and conclusions of our initial work can be found elsewhere18,
but the salient details are provided for completeness. The theory of ASLC is very similar for all
ASLC-material classes. For the molecular system, it can be described by the simple energy level
diagram shown in Fig. 1. The ground and first excited singlet states of the dye molecule are denoted
by respectively, with the more closely spaced lines and representing the manifold
of vibrational states in each electronic state. At ambient temperatures, the energy separation be-
tween these levels is small with respect to the thermal energy, kT, (200 or 25 meV), and the
population of these states can therefore be determined by the simple Boltzmann distribution func-
tion. The three arrows on Fig. 1 depict a) the maximum of the absorption, b) the peak of the emis-
sion and c) a transition that results in anti-Stokes luminescence. The data in Fig. 2 are the absorp-
tion and emission spectra of a dilute solution of rhodamine 101 in acidified ethanol. The emission
spectrum, which is independent of the excitation wavelength, was recorded with excitation at 632.8
nm and the sharp peak on the spectrum at this wavelength is the scattered light from the beam
obtained from a helium-neon laser. The broad spectra and lack of discrete transitions are due to
strong interactions between the molecule and the solvent. Since the states are both good
Figure 1 Schematic energy level diagram for the dye molecule, rhodamine 101, in solution showing
the four important radiative transitions: a) maximum absorption, b) maximum emission and c) excitation
wavelength for anti-Stokes luminescence.
OPTICAL COOLING USING ANTI-STOKES FLUORESCENCE 623
Figure 2 Absorption (Abs) and Emission (Fluor) spectra of rhodamine 101 in acidified ethanol
M). The sharp peak at 632.8 nm originates from scattered excitation light.
singlet states, there is no change of spin for a radiative transition and therefore the luminescence
can be referred to more formally as fluorescence. Inspection of the data in Fig. 2 show two key
points i) the absorption at 632.8 nm is very weak and ii) the majority of the emission is to higher
energy of the excitation light. The average wavelength of the emission is at 604.5 nm, on the red
side of the emission maximum. The energy difference between 632.8 nm (15803 1.96 eV) and
604.5 nm (16543 2.05 eV) is 740 or 0.09 eV. This energy corresponds to the average
energy removed from the system for every absorbed photon. If an excited state decays non-radiatively,
then the full 15803 photon energy is transferred into thermal energy, which is a factor of 18
times higher than the average energy removed by the anti-Stokes emission. Thus no net removal of
thermal energy occurs if the PLQY is below 95%, although in reality this figure needs to be even
higher, as will be explained later.
The weak absorption, or high transmission, of the sample at 632.8 nm results in only a small
fraction of the incident photons being extinguished, with the vast majority passing through the
sample. In our original experiment, only a single pass through the sample was possible and a mere
2% of the incident light was absorbed.
The overall cooling power, available under these conditions can be described by Eq. 1:
Where
Thus, for a pump power, of 250 mW, a transmission, T, of 98% and an intrinsic effi-
ciency, of 5% gives a cooling power, of only 250 W. These values are close to those of
our original experiments and they assume an optimum PLQY of 100%. By keeping the parasitic
heat load on the sample of low heat capacity to a minimum, a temperature drop of 3 K was observed
over a period of four hours.14
To realize a cooling power of 100 mW it is necessary increase the three individual terms in
Eq. 1. This article addresses the issues that influence these three terms and examines methods of
epitomizing them. Within this context, the merits of using a molecular dye-based cooling material
to achieve the 100 mW goal for cooling power are examined.
DISCUSSION
Pump power,
Molecular dyes afford a number of attractive features as the active species in an optical cool-
ing medium and these are summarized in Table 1. The ability to use chemical synthesis and add
different substituents to the molecule to alter the position of the absorption and emission spectra of
the dye molecules, is a very powerful tool. It allows tuning of the wavelengths for pumping to
coincide with those available from a variety of laser sources, specifically those based on diode
lasers. The rhodamine 101 molecule, shown in Fig. 3, is based on the xanthene moiety, as are all the
rhodamine class of dyes. These are all good laser dyes that are commercially available in sufficient
quantities and at reasonable cost. The essential feature of the rhodamine 101 structure for ASLC
applications, is the role of the two propyl chains on each nitrogen atom that links them to the main
xanthene core. These groups prevent the nitrogen atoms from rotating and thus create a very rigid
structure that reduces internal conversion, which is a non-radiative decay mechanism that leads to
vibrational relaxation and unwanted parasitic, thermal energy23. The acid-substituted phenyl ring at
the base of the structure does not couple strongly with the transition, but it can affect the position of
the electronic transition. The acidification of the solvent, it should be noted, ensures that this acid
group is protonated to give the cationic form of the dye, which is in dynamic equilibrium with the
zwitterionic form. Table 2 lists four different substituents that are available on this position and
their effect on the wavelength of absorption maximum.
Figure 3 Central xanthene moiety that forms the core of the series of rhodamine dyes listed in Table 2.
The effect of the substituent group, R, on the absorption maximum can be seen clearly, shift-
ing from 568 nm for rhodamine 101 to 682 nm for rhodamine 800. The emission spectra are red-
shifted too and each molecule exhibits a high PLQY. Although these are all commercially available
laser dyes, it does not necessarily follow that they can operate in a laser-cooling medium. We are
currently investigating these and other types of dyes as possible candidates. It will be important to
test each dye in an ASLC apparatus, in order to verify that the PLQY values are in excess of the
required 95% limit. It is worth noting that this apparatus also provides a new and valuable tool for
measuring absolute PLQY values.
The wavelengths required for cooling with rhodamine 101 in acidified ethanol range from
610 nm to 635 nm. Our initial setup used the 3W all-lines output from an argon ion laser to pump a
jet stream dye laser and provided up to 300 mW of pump power. This system is now run in parallel
with the 3 W from a dye laser pumped by the 10W output of an intracavity-doubled laser.
The most suitable laser dye for these experiments is rhodamine 101 itself, as the lasing wavelengths
coincide exactly with those required for optical cooling. For newly-synthesized or existing dye
molecules that require wavelengths in the range 550 nm to 700 nm, the dye laser is a suitable choice
of pump laser. In the range 700 nm to 1000 nm, it can be replaced by a solid-state Ti:Sapphire laser.
Although neither the dye or Ti:Sapphire lasers are ideal for a laser-cooled device, due to their high
power consumption, bulk and difficulty of use, they are ideally suited to the evaluation in the
laboratory of potential cooling materials. The choice of diode lasers or diode laser-pumped solid
state lasers are far more suitable for commercial devices as they use less power, they are more
compact, easier to use and last longer. The available wavelengths start at 400 nm and range from
600 nm to 1600 nm in discrete wavelengths. The technology is advancing rapidly with, for ex-
ample, 635 nm available at low power and 810 run available at high powers. The list of available
laser-diode wavelengths and powers will change significantly over the next 5 years, as new applica-
tions are found. Development of new laser diode wavelengths may even be influenced by the needs
of the opto-cryocooling market.
Not only can chemistry shift the absorption maximum, it can also provide a method of facili-
tating processability. The addition of large, bulky side-groups to the R-substituent, or even to the
main xanthene core, can modify the solubility of the molecule in different solvents, without affect-
ing the electronic properties. These groups can also be used to inhibit dye aggregation, an effect
that is discussed below and that has been used with great success in the photographic industry24.
The choice of liquids in device applications is often considered detrimental, although they should
not be discounted. Encapsulation of liquid solutions into various hosts is possible. Furthermore, the
ability to append groups to the dye molecule that can provide linkages to polymer and sol-gel
precursors is an elegant and simple method of creating solid samples, although it is not without its
problems. For many years, the laser community has been attempting this approach at creating solid-
state dye lasers and with some degree of success25. We plan to ‘hitch a ride’ on the findings of this
work and prepare solid-state dye-cooling media using the same preparation procedures. Since the
laser-cooling medium is pumped off-resonance, relative to those working on the lasers, the trouble-
some issue of the degradation of the dyes will be less problematic.
It is the powerful contribution that chemical synthesis can make that is emphasized, although
there is the constraint that the molecular manipulation, or engineering, must create a molecule that
retains a near-100% PLQY in the ASLC medium. The science of how the chemistry affects the all-
important non-radiative deactivation pathways is not yet precise and often a serendipitous approach
must be adopted. The ability to affect the radiative pathway for these dyes is far harder task, with all
the dyes exhibiting large natural, radiative rate constants, with values falling in the range
These high values are large with respect to the non-radiative processes and lead to high
PLQY values. The decay of the excited state is dominated by the radiative rate constant, which
leads to an excited state lifetime in the nanosecond time domain. This is a very useful attribute,
which allows for the potential of a dye molecule to be excited many times per second. Rhodamine
101 in ethanol, for example, has an excited state, or fluorescence, lifetime of 4.1 ns. Allowing for 5
decay times to relax completely, this allows for excitations per second. The important impact
of this is to increase greatly the power of the pump laser at which the absorption saturates. Although
this issue has yet to be tested, it may prove to be beneficial in future experiments.
%Absorption, 1-T
The fraction of the pump beam absorbed by the sample is given in Eq. 1 by the term, 1 – T.
[This is correctly known as the %absorption, although to avoid confusion with absorbance the term
is not used here*.] The absorbance of the sample in the low wavelength tail of the absorption
spectrum is inherently low and consequently the transmission, T, is close to 100%. To increase
1 - T, the sample concentration or pathlength must therefore be increased. In our original experi-
ment, the sample pathlength was chosen to be < 5 mm and the dye concentration mol
The transmission of this sample was > 95%, depending on the exact wavelength, and therefore
1 – T had a maximum value of only 5%. The concentration could not be increased further as the dye
molecules have a tendency to form aggregates that absorb strongly in the pumping region. Aggre-
gates are often non-luminescent, with efficient non-radiative decay channels that lead to sample
heating and they should therefore be avoided. For the future, however, by attaching large, bulky,
ballast groups to the dye molecules, aggregation can be inhibited24.
An increase in the sample pathlength can be accomplished with a long sample or by using a
multi-passing configuration, such as that used by Los Alamos21 and Ball Aerospace22 and discussed
further in these proceedings. A transmission of 95% in 5 mm, corresponds to a sample absorbance
of 0.022. A more desirable transmission of 1% is equivalent to an absorbance of 2, which is a factor
of 100 times greater. Therefore, in order to extinguish 99% of the pump light, the sample must be
multi-passed almost 100 times. Our original sample was contained in a cylindrical sample cell of
low optical quality and the experiment could not be reconfigured to achieve this optimum configu-
ration. We are currently modifying the sample cell to optimize the number of passes through the
sample made by the pump light. The importance of the multi-passing geometry must not be under-
estimated, as the more effective this can be made, the larger the intrinsic efficiency, can be made,
as discussed below.
Intrinsic anti-Stokes luminescence cooling, ASLC, efficiency,

The intrinsic efficiency of the anti-Stokes luminescence cooling, is determined by the
energy difference between the pumping photon energy and the average energy of the photons emit-
ted by the sample. Using the data displayed in Fig. 1, the average wavelength of the emission is
approximately 604.5 nm and therefore the ASLC efficiency, given by when using
a pump wavelength of 632.8 nm, is –4.7%. The negative sign ensures the sign of the cooling power
* Absorbance = (100% – %absorption)
Figure 4 Absorption and emission spectra of rhodamine 101 in acified ethanol recorded at concentra-
tions ranging from (1) M up to (9) M. The spectra were recorded in a conventional cuvette of 10
mm cross-section, with emission detected from the centre.
is consistent with the thermodynamic convention of energy removal from the sample. To achieve
more efficient cooling the value of should be as large as possible and therefore and must
be well separated in wavelength (and energy). By shifting further to the red (lower energy) the
absorbance decreases and hence the number of multi-passes required to extinguish the pump beam
must be increased significantly, as mentioned earlier. The absorbance drops off exponentially and
in our original experiments negligible cooling was observed at 652 nm.
In contrast, must be shifted to the blue (higher energy) in order to optimize The short-
est wavelength available for any dye-based sample is defined by the average photon energy of the
molecular emission spectrum; a spectrum that can only be obtained at very low concentrations,
where re-absorption effects are negligible. The data shown in Fig. 1 were recorded under such
conditions and therefore the efficiency derived from these are optimum for the dye
in this solvent. Unfortunately, the conditions for efficient ASLC require high concentration in order
to attain significant absorption at the pump wavelength. Under these conditions, the emission spec-
trum can be severely distorted by re-absorption effects that result from the overlap of the emission
spectrum with the absorption spectrum. The overlap region can be seen from Fig. 1 to occur at ca.
590 nm, where the red absorption tail overlaps the blue emission tail. A consequence of this overlap
is the attenuation of the blue emission with respect to the red emission and is described as the inner-
filter effect. The observed spectrum is distorted and appears to be blue-shifted with respect to lower
concentration spectra; it has a longer and hence a lower efficiency, We have examined this
effect in some detail in order to reduce the impact that the inner-filter effect has on the ASLC
efficiency.18 In a previous report, we modeled the impact of sample geometry and dye concentra-
tion on the observed emission spectrum and verified the predictions experimentally.26 A similar
comparison has been reported more recently and the impact on anti-Stokes cooling addressed.27
The simple model used in these instances, however, did not take into consideration the effect of re-
emission from those molecules excited indirectly by the emission spectrum, an effect known as
radiative energy transfer. The high PLQY values required for laser cooling require that this effect
be taken into consideration. By modeling the path of initially-emitted photons by a random walk,
we can determine the cooling efficiency of a range of dye systems and sample geometries. The full
details of this work will be published elsewhere, but a series of the measured spectra used to verify
the use of these calculations can be seen in Fig. 4. The spectra are similar at low concentration but
as the concentration increases, they shift to longer wavelength (lower energy) due to the re-absorp-
tion effect. A simple solution to this problem is to use a long sample cell, which helps to increase
1-T, and with a small cross-section, which minimizes the re-absorption effect. With increasing
length, the sample heat capacity does increase, but by taking the idea to an extreme limit of using an
optical fibre, this effect can be made small and the re-absorption effect almost eliminated. For the
doped glasses, this geometry has resulted in temperature drops down to 65K below ambient.8
For device applications, however, a compromise needs to be made, as it is a high cooling
power, that is required to compete with extrinsic heat loads, not a large temperature drop in a
low heat capacity sample.
CONCLUSIONS
Molecular dyes with photoluminescence quantum yields of close to 100%, such as rhodamine
101, are ideal candidates for the active component of an optical cryocooling medium. Since the
photoluminescence emanates from a fully allowed, radiative electronic transition, the excited state
lifetimes are short and this enables high powers to be used as a pump source without saturating the
absorption transition. The absorption and emission spectra are well-separated in energy, leading to
a Stokes loss that helps reduce the inner-filter effect and enables the majority of the high energy
photons to be emitted by the sample, without being attenuated by re-absorption effects. In order to
generate high cooling powers using a dye-based sample it is necessary to design a sample that is
amenable to efficient multi-passing in order to extinguish a high percentage of the pump beam.
Using these criteria, we have designed a new sample cell that will allow us to extinguish 90% of the
pump beam by multi-passing and utilize an intrinsic ASLC efficiency of –5%. With the available
pump power of 2.5 W, this will yield a cooling power of over 110 mW, a value that is usable and far
exceeds our original efforts. By choosing a different dye system, we are also exploring the possibil-
ity of tuning the absorption wavelength to coincide with the outputs of commercial, high power
diode-lasers. In addition, different sample geometries are being examined to enable the optimum
intrinsic efficiency to be used in a multi-passing configuration.
ACKNOWLEDGMENTS
Funding for this work was partially supported by the Air Force Research Laboratory through
a Phase I SBIR, Contract Number F29601-99-C-0121, and through the physics college of the UK
Engineering and Physical Science Research Council, EPSRC.
REFERENCES
1. Pringsheim, P., “Some remarks concerning the difference between luminescence and tem-
perature radiation. Anti-Stokes fluorescence,” J. Phys., USSR vol. 10, (1946), pp. 495-498.
2. Vavilov, S., “Photoluminescence and thermodynamics,” J. Phys., USSR vol. 10 (1946), pp.
499-502.
3. Landau, L., “On the thermodynamics of photoluminescence,” J. Phys., USSR vol. 10 (1946),
pp. 503-506.
4. Epstein, R.I., Buchwald, ML, Edwards, B.C., Gosnell T.R. and Mungan, C.E., “Observation
of laser-induced fluorescent cooling of a solid,” Nature, vol. 377 (1995), pp. 500-502.
5. Mungan, C.E., Buchwald, M.I., Edwards, B.C., Epstein, R.I. and Gosnell, T.R., “Laser
cooling of a solid by 16 K starting from room temperature,” Phys. Rev. Lett., vol. 78(6)
(1997), pp. 1030-1033.
6. Mungan, C.E., Buchwald, M.I., Edwards, B.C., Epstein, R.I. and Gosnell, T.R., “Internal
laser cooling of -doped glass measured between 100 and 300 K,” Appl. Phys. Lett., vol.
71(11), (1997), pp. 1458-1460.
7. Luo, X., Eisaman, M.D. and Gosnell, T.R., “Laser cooling of a solid by 21 K starting from
room temperature,” Optics Letters, vol. 23(8), (1998), pp. 639-641.
8. Gosnell, T.R., “Laser cooling of a solid by 65 K starting from room temperature,” Optics
Letters, vol 24(15), (1999), pp. 1041-1043.
9. Murtagh, M.T., Sigel, G.H., Fajardo, J.C., Edwards, B.C. and Epstein, R.I., “Laser-induced
fluorescent cooling of rare-earth-doped fluoride glasses,” J. Non-cryst. Solids., vol. 253,
(1999), pp. 50-57.
10. Rayner, A., Friese M.E.J., Truscott, A.G.Heckenberg, N.R. and Rubinsztein-Dunlop, H,,
“ Laser cooling of a solid from ambient temperature,” J. Mod. Opt., (in press 2000).
11. Gauck, H., Gfroerer, T.H., Renn, M.J., Cornell, E.A. and Bertness, K.A., “External radiative
quantum efficiency of 96% from a GaAs/GaInP heterostructure,” Appl. Phys. A - Mat. Sci &
process., vol. 64(2), (1997), pp. 143-147.
12. Finkeissen, E., Potemski, M., Wyder, P., Vina, L. and Weimann, G., “Cooling of a semicon-
ductor by luminescence up-conversion,” Appl. Phys. Lett., vol. 75(9), (1999), pp. 1258-1260.
13. Zander, C. and Drexhage, K.H., Advances in Photochemistry, Wiley, New York, (1995),
pp.59-78.
14. Clark, J.L. and Rumbles, G., “Laser cooling in the condensed phase by frequency up-conver-
sion”, Phys. Rev. Lett., vol. 76 (1996), pp. 2037-2040.
15. Mungan, C.E. and Gosnell, T.R., “Laser cooling in the condensed phase by frequency up-
conversion - Comment,” Phys. Rev. Lett., vol. 77 (1996), pp. 2840.
16. Rumbles, G. and Clark, J.L., “Laser cooling in the condensed phase by frequency up-conver-
sion - Reply,” Phys. Rev. Lett., vol. 77 (1996), pp. 2841.
17. Sarkisov, S., Curley, M., Wilkosz, A. and Grymalsky, V., “Optical channel waveguides
formed by upconverted photobleaching of dye-doped polymer film in regime of dark spatial
soliton”, Optics Comms., vol. 161, (1999), pp. 132-140.
18. Clark, J.L., Miller, P.P. and Rumbles, G., “Red edge photophysics of ethanolic rhodamine
101 and the observation of laser cooling in the condensed phase,” J. Phys. Chem. A., vol.
102(24), (1998), pp- 4428-4437.
19. Edwards, B.C., Buchwald, M.I., and Epstein, R.I., “Development of the Los Alamos solid-
state optical refrigerator,” Rev. Sci. Inst., vol. 69(5), (1998), pp. 2050-2055.
20. Edwards, B.C., Anderson, J.E., Epstein, R.I., Mills, G.L. and Mord, A.J., “Demonstration of
a solid-state optical cooler: An approach to cryogenic refrigeration,” J. Appl. Phys., vol.
86(11), (1999), pp. 6489-6493.
21. Edwards, B.C., Anderson, J.E., Epstein, R.I., “Solid-state optical coolers developments,”
Cryocoolers 11, Plenum Press, New York (2001).
22. Mord, A.J., Mills, G.L. and Slaymaker, P. A., “Design and performance of an optical cryo-
cooler for a focal plane application”, Cryocoolers 11, Plenum Press, New York (2001).
23. Karstens, K. and Kobs, K., J. Phys. Chem., vol. 84, (1980), pp. 1871.
24. James, T.H., “Theory of the photographic process”, Ed., Macmillan, New York (1977),
Ch. 12, pp. 365.
25. Rahn, M.D. and King, T.A., “High-performance solid-state dye laser based on perylene-
orange-doped polycom glass,” J.Mod.Optics, vol. 45(6), (1998), pp. 1259-1267.
26. Dhami, S., deMello, A.J., Rumbles, G., Bishop, S.M., Phillips, D. and Beeby, A., “Phthalo-
cyanine fluorescence at high concentrations: Dimers or re-absorption effect?”, Photochem.
Photobiol., vol. 61(4), (1995), pp. 341-346.
27. Frey, R., Micheron, F. and Pocholle, J.P., “Comparison of Peltier and anti-Stokes optical
cooling,” J. Appl. Phys., vol. 87(9), (2000), pp. 4489-4498.
Solid-State Optical Cooler Developments
B. C. Edwards, J. E. Anderson and R. I. Epstein

Los Alamos National Laboratory
Los Alamos, New Mexico, USA 87544
C. W. Hoyt, and M. Sheik-Bahae

Department of Physics and Astronomy, Optical Sciences and Engineering
University of New Mexico, Albuquerque, New Mexico, USA 87131
ABSTRACT
Optical cooling of solids was first demonstrated in 1995. Since that time our efforts have
concentrated on using this phenomenon to produce a viable optical cryocooler. A bench-top,
solid-state optical cooler was demonstrated recently with 54 °C of cooling from room temperature
and a heat lift of 25 mW when it was pumped with 1.6 watts of laser light. Based on the bench-top
cooler results a compact, rugged, self-contained system with laser-diode pumping is being con-
structed as a prototype for research and commercial applications. Recent results, designs and
plans for future work are discussed. In addition to prototype work, our efforts have continued on
producing additional optical materials which demonstrate optical cooling and will improve the
efficiency and useful temperature range of optical coolers. New materials with promising results
are discussed.
INTRODUCTION
The basic principles for optical refrigeration, cooling by anti-Stokes fluorescence were first
discussed in 1929.1 Much more recently, optical refrigeration has been demonstrated experimen-
tally with ytterbium doped fluoride glass, Yb:ZBLAN,2,3,4 and in laser dye solutions.5,6 Since these
first demonstrations several studies have investigated materials for their potential in optical refrig-
eration.7,8 Two independent, theoretical studies found that Yb:ZBLAN could be used as the work-
ing material in an optical refrigerator that operates below 80K with efficiencies comparable to
those of small mechanical cryocoolers.9,10 Optical refrigerators such as these would be valuable for
many low-power cooling applications including cooling radiation detectors, high-temperature su-
perconductors and electronics. Since optical refrigerators are solid-state devices without any mov-
ing parts or thermal connections between the cold and warm stages, other than structural supports,
they would be free of vibrations, mechanically robust, reliable and represent a new approach to
cryogenic refrigeration. The basic principle of optical cooling involves pumping an optical solid
with light that is absorbed, effectively combined with thermal energy and re-radiated. This process
removes energy or heat from the material. In our system the Yb ions in our glass have a ground
state manifold and effectively only one excited-state manifold. When the Yb-doped material is

pumped at wavelengths longer than the mean fluorescence, ions are excited from the higher
energies of the ground state manifold to the lower energies of the first excited manifold. The
system returns to thermal equilibrium by absorbing phonons and then radiates a photon. On the
average this radiated photon will have a higher energy than the pump photons. The difference in
the photon energies is heat removed from the optical material.
The first stage of research on laser cooling in solids and liquids was theoretical, The second
stage has been the experimental demonstration of the principle. The third stage we believe will
be the practical application of this process. To compete with highly developed cryocooler
technologies an optical cryocooler must be demonstrated to have comparable performance,
distinct advantages in some areas and it must be inexpensive to manufacture. Our efforts at Los
Alamos National Laboratory since the initial laboratory demonstration have been directed toward
answering these practical application questions.
At the end of 1999 the results of a bench-top cooler were published11 which laid the ground
work for production of a self-contained prototype. The bench-top cooler implemented many of
the aspects of a practical cryocooler and provided valuable information on what studies still
needed to be completed before a commercial production program could be implemented.
EXPERIMENT
The research on optical cooling in solids at Los Alamos National Laboratory has focused on
two areas. The first is construction of a prototype cryocooler utilizing optical cooling. The
second is searching for new materials that exhibit optical cooling.
Recent bench-top cooler experiments have demonstrated a 54°C temperature drop from
room temperature (achieved with improvements on experiment in reference 11). These tests were
done with an Argon ion-pumped Ti:Sapphire laser producing 1.6 watts of pump power for the
cooler. These experiments were completed inside a standard vacuum chamber with external
optics, optical fiber supports and no cold finger. The next step is the construction of a self-
contained system that can be used as a design testbed for a commercial cryocooler.
The prototype cryocooler under construction is shown in Figure 1. This design was based on
the bench-top system with modifications to the structural support, optical coupling for the pump
light, cold finger and vacuum system. In the prototype the cooling material, coldfinger and
support structures are enclosed in a small vacuum that is pumped out through a port on the end
flange.
Figure 1. Schematic of the prototype cryocooler currently under construction at Los Alamos
National Laboratory. The input laser comes in through the optical fiber on the right side of the figure.
The pump light is focused by the lens coupling through a hole in the mirror on the Yb:ZBLANP cooling
element. The cooling element, transparent coupling, coldfinger and part of the support tubes cool down.
A vacuum is held between the leftmost coupling lens and the rightmost flange in the figure. The object to
be cooled will be located inside the G-10 tubes and attached to the left end of the coldfinger.
SOLID-STATE OPTICAL COOLER DEVELOPMENTS 633
The mechanical support structure is based on a dewer design that utilizes coaxial G-10 tubes
coupled on the ends. This design allows for a very stiff structure while limiting the thermal
conductive load. In the current design the tubes each have 76 microns thick walls and are
approximately 1.9 cm long with inside diameters of 1.55, 1.75 and 1.98 cm. Using a thermal
conductivity for G-10 of 0.0028 W/cm/°C we find that for a temperature difference of 200°C
there will be a thermal load from conduction of 8 mW. The internal walls of the vacuum will be
coated with layers that have low emissivity at thermal wavelengths (0.02 and 0.05). The cooling
element, transparent coupler, coldfinger and G-10 tubes can be assumed to be blackbody
absorbers in the worse case. Accounting for the geometry of the system the calculated radiative
thermal load is approximately 100 mW for a 200°C temperature difference. The total thermal
load on the cold components should be less than 108 mW for a 200°C temperature drop in the
cooling element.
Optical Coupling
The optical coupling design was tested and found to have a 86% throughput from coupling
into the fiber to light through a 600 micron hole at the location of the cooling element. This spot
size matches the hole size in the mirror of the cooling element. The lenses are made of sapphire
and thermally sunk to the outer chamber.
Coldfinger/Cooling Element Coupling
The coldfinger and its coupling to the cooling element are designed to minimize the
parasitic heating due to pump laser leakage and fluorescent light leakage as well as provide a
highly conductive thermal link between the cooling element and the coldfinger. The coldfinger
is made from aluminum, the transparent coupler is made of sapphire and the cooling element is
Yb:ZBLANP with dielectric coatings on the ends. The materials for the coldfinger and coupler
were chosen for their various thermal properties but the coupler may be made from ZBLANP in
the future because of thermal expansion issues. The coldfinger and coupler will be mounted
together with thermally conducting epoxy (the coupler will have a silver mirror coating on this
surface) and the coupler and cooling element will be joined with < 2 microns of optical epoxy.
When pumped in a vacuum with 1.5 W of laser power at Yb fluorescence wavelengths, the
coupler / cooling element interface design did not change in temperature. This interface does
however show mechanical problems when subjected to rapid temperature changes due to the
thermal expansion differences between ZBLANP and sapphire. The coupler material may be
changed to ZBLANP to overcome this thermal problem. This three element heart of the cooler
should allow for good performance of the cooling element, a good thermal connection to the
object to be cooled and eliminate any stray light escaping from the cooling element through the
dielectric mirror.
Current Status
To date all of the components of the prototype have been completed and in testing. At the
completion of the component testing the prototype will be assembled in stages with some
additional testing and the completed prototype is expected to be tested during the summer of
2000.
New cooling materials
In parallel we have been studying new materials that may exhibit cooling. Recently two
new materials have been found to cool when pumped with the proper wavelength of light. The
first is Yb:YAG and the second is Tm: ZBLANP.
The Yb:YAG was pumped at wavelengths from 990 nm to 1050 nm and found to cool
0.36°C at 1030 nm with approximately 10 mW of absorbed pump power (Fig. 2). The advantages
of a Yb:YAG over Yb:ZBLANP include: 1) possible higher efficiency at lower temperatures due
to the narrower lines in the YAG, 2) improved cooler performance due to higher thermal
conductivity of YAG, 3) better mechanical stability of YAG, and 4) better substrate for dielectric
mirror deposition.
Figure 2. Optical cooling of Yb:YAG.

White objects are cooler in this image. The
pump laser is entering from the right and
passing through the Yb:YAG sample. No
laser light is hitting the reference sample.
We have also obtained a 1 cm block of TmrZBLANP and pumped it at wavelengths

between 1.75 and 2.05 microns (Fig. 3). The Tm:ZBLANP cooled 1.2°C with approximately 90
mW of absorbed pump power. These results were obtained by viewing the Tm:ZBLANP with a
thermal infrared camera and comparing it to a reference ZBLANP block. The full results of these
experiments have been submitted for publication. For the same laser efficiencies, a cooler
constructed with thulium as the active ion instead of ytterbium would be twice as efficient. The
reason for the higher efficiency is that the input photons have one half the energy for the thulium
system as compared to the ytterbium system but each photon can remove the same amount of
Figure 3.Temperature change, normalized to incident power, verses pump wavelength for a
Tm:ZBLANP sample. The insets are two thermal images corresponding to 1.9 mm (cooling) and 2.0 mm
(heating). The solid line is a theoretical model fit to the data.
SOLID-STATE OPTICAL COOLER DEVELOPMENTS 635
energy. The current difficulty with implementing a thulium-based system is the availability of
high-efficiency 2 micron diode lasers.
In addition we have begun studies of cooling in semiconductors. Semiconductors have been
discussed as optical cooling materials12,13 and examined experimentally 14, l5 but no net cooling
has been observed. The difficulty in achieving bulk cooling in a semiconductor and with
implementing a semiconductor cooling element in a system has been the high index of refraction
in the materials that trap the fluorescence and eventually produces net heating. This is not a
trivial problem and our work is just beginning on how to get past this hurdle.
CONCLUSION
Optical cooling is solids has become accepted as demonstrated experimentally. The next
step is to implement this laboratory effect in a practical system. Our program at Los Alamos
National Laboratory is pushing toward a demonstration cryocooler which could prove the
commercial feasibility of an optical cryocooler. The prototype is under construction and
expected to be tested in the near future. In addition, one of the best ways to improve the
efficiency of an optical cryocooler is with a cooling material with better performance
characteristics. We have demonstrated cooling in two new materials, Yb:YAG and
Tm:ZBLANP, that show promise for better performance in a cyrocooler system.
ACKNOWLEDGEMENTS
This work was conducted under the auspices of DOE and supported in part by IGPP/LANL
and by a NUCOR grant from the University of California. We thank Charles Wilkerson, Melvin
Buchwald, and Alan Gibbs for their help and advice during this project.
REFERENCES
1. L. P. Pringsheim, Z. Physik, 57, 739 (1929); L. Landau, J. Phys. (Moscow) , 10, 503 (1946);
A. Kastler, J. Phys. Radium , 11, 255 (1950); S. Yatsiv, Ed., Advances in Quantum
Electronics (Columbia University Press, New York, 1961).
2. Epstein, R. I., Buchwald, M. I., Edwards, B. C., Gosnell, T. R. & Mungan, C. E., Nature,
377, 500 (1995);
3. C. E., Buchwald, M. I, Edwards, B. C., Epstein, R. I. & Gosnell, T. R., Phys. Rev. Letters,
78,1030 (1997); Mungan, C. E., Buchwald, M. I, Edwards, B. C., Epstein, R. I. & Gosnell,
T. R., App. Phy. Lett.,, 71, 1458 (1997); Gosnell, T.R., Optics Letters, 24, 1041 (1999).
4. Lou, X., Eisaman, M.D., Gosnell, T.R., Optics Letters, 23, 639 (1998)
5. Clark, J.L. and Rumbles, G., Phys. Rev. Lett. 76, 2037 (1996); Rumbles, G., and Clark, J.L.
Phys. Rev. Lett. 77, 2841 (1996); Zander, C. and Drexhage, K.H., in Advances in
Photochemistry, Vol. 20 (Wiley, 1995), p. 59.
6. Clark, J.L., Miller P.P., Rumbles, G., J. of Physical Chemistry, 102, 4428 (1998)
7. Montoya, A.E., Sanzgarcia, J.A., Bausa, L.E., Spectrochimica Acta Part A Molecular and
Biomolecular Spectroscopy, 54, 2081 (1998); Fajardo, J. C., Sigel, G. H., Jr., Edwards, B. C.,
Epstein, R., I., Gosnell, T. R. & Mungan, C. E., J. of Non-Crystalline Solids, 213, pp 95-100
(1997); M. T., Sigel, G. H., Fajardo, J. C., Jr., Edwards, B. C., & Epstein, R., I., J. of Non-
Crystalline Solids, in press (1999); Mungan, C. E., Edwards, B. C., Epstein, R. I., Gosnell, T.
R., & Buchwald, M. I. Mat. Sci. Forum , 239, 501-504 (1997).
8. Lei, G., Anderson, J. E., Buchwald, M. I, Edwards, B. C., Epstein, R. I., Murtagh, M. T., &
Sigel, G. H., Jr., IEEE Journal of Quantum Electronics, 34, 1839-1845 (1998)
9. Edwards, B. C., Buchwald, M. I., Epstein, R. I., Review of Scientific Instruments, 69, 2050-
2055 (1998).
10. Lamouche, G.,Lavallard, P., Suris, R., Grousson, R., Journal of Applied Physics, 84, 509,
(1998).
11. Edwards, B. C., Anderson, J. E., Epstein, R. I., Journal of Applied Physics, 86, 6489,
(1999).
12. Oraevsky, A. N., J. of Russian Laser Research, 17, 471-479 (1996)
13. Rivlin, L. A., Zadernovsky, A. A.,Optics Communications 139, 219-222 (1997)
14. Finkeissen E., Potemski M, Wyder P., Vina L. & Weimann G., Applied Phys. Let., 75,1258
(1999)
15. Gauck, H., Gfroerer, T.H., Renn, M.J., Cornell, E.A., Bertness, K.A., Applied Physics A
Materials Science & Processing, 64, 143 (1997)
Cryocooler Reliability and Redundancy
Considerations for Long-Life Space Missions
R.G. Ross, Jr.

Pasadena, CA 91109
ABSTRACT
One means of achieving high reliability with cryogenic payloads involving cryocoolers is to
incorporate redundancy, either in the form of redundant coolers and/or redundant drive electron-
ics. To access the redundant elements, electrical and/or heat switches must also be incorporated.
Although the redundant elements protect against a possible failure, the increased system complex-
ity and increased cryogenic load associated with the incorporation also have a negative effect on
reliability that must be taken into account.
This paper presents an analysis of the reliability advantages and disadvantages of a variety of
cryocooler redundancy options, based on their total reliability, mass, and power impact at the
cryogenic system level. The paper begins with developing an approach for quantifying the prob-
ability of failure of the key subassemblies, such as coolers, electronics, and heat switches, associ-
ated with the redundancy; the analysis considers the subassembly's state of development, the
complexity and testability of its critical failure mechanisms, and the effect of the total cryogenic
load on its reliability. Means are also presented for estimating the total cryogenic load as influ-
enced by the addition of the redundant elements.
Finally, the overall system performance (reliability, mass, and power) of the various cryo-
cooler redundancy options is computed using the failure probabilities of the individual elements,
and the system interrelationships of the elements.
INTRODUCTION
Achieving high reliability is a key design driver for cryogenic systems required to provide
continuous cooling during multi-year space missions. There are three key steps to achieving high
reliability:
1) Use highly tested, robust components, such as cryocoolers and electronics that incorporate
well understood design principles with a proven history of high reliability.
2) Thoroughly predict the cryogenic refrigeration load over the mission life-cycle and incor-
porate significant margin to cover load growth and cryocooler performance loss over time.
3) Incorporate redundant components to protect against individual component failures.
This said, the problem faced with most cryogenic systems is: "How do I assess the reliability
of a one-of-a-kind cryocooler or cooler system component?" and "How do I trade off the reliabil-

638 CRYOCOOLER RELIABILITY INVESTIGATIONS AND ANALYSES
Figure 1. Example cryocooler redundancy options.
ity gain achieved from redundancy against the losses associated with increased system complexity
and cryogenic load growth caused by adding the redundancy?"
The analysis presented in this paper was conducted to clarify these issues and provide a
means for selecting an optimal configuration for a high-reliability cryocooler system. To help
focus the discussion, Fig. 1 illustrates four example cooler system configurations.
In the sections that follow, the problem of assessing total cryogenic system reliability perfor-
mance has been broken down into three computational steps: 1) quantifying the reliability (or
probability of failure) of key cryogenic system subassemblies such as cryocoolers, cryocooler
electronics, and heatswitches; 2) determining the total system cryogenic load associated with
candidate redundancy options, and 3) quantifying the overall system performance associated with
candidate cooler-system options including the mass and power performance of the overall system,
and its overall reliability.
ESTIMATING RELIABILITY OF KEY SYSTEM SUBASSEMBLIES

One of the most difficult issues is the problem of quantifying the reliability (or probability of
failure) of an individual one-of-a-kind subassembly such as a cryocooler or heatswitch. Typically,
such units are custom built for each application, and little or no quantitative reliability or life test
data exist. For such subassemblies, one means of assessing their reliability is to first utilize expert
knowledge of the unit's detailed design to identify each important failure mechanism associated
with the unit's design features. This list will include such items as leakage of seals, fatigue of
flexed elements, contamination of gasses, structural failure during launch vibration, etc.
Next, the probability of failure of each of these important mechanisms is estimated based on
key attributes of the failure mechanism known to correlate with failure probability. These include:
• The extent to which the mechanism's underlying physics are well understood
• The level of complexity of the mechanism's parameter dependencies
• The accuracy of predictive analytical and experimental test methods that have been or will be
used to design out the possibility of failure of the mechanism
• The degree to which the actual flight hardware can be and will be tested and validated with
respect to the mechanism
The last step is then to combine the mechanism-level failure probabilities into a probability of
failure for the total assembly level. For failure probabilities that are much less than one
the assembly-level probability of failure is well approximated as just the sum of the individual
mechanism probabilities, i.e. for all mechanisms i = 1, N. The reliability of a unit
is just one minus the failure probability, i.e. R = 1 - P.
CRYOCOOLER RELIABILITY FOR LONG-LIFE SPACE MISSIONS 639
To provide a consistent framework for making the mechanism-level failure probability assess-
ments, Table 1 presents a series of benchmarks created by this author and drawn from experience
with a broad variety of space hardware. Included in each failure probability level is a qualitative
description of the attributes responsible for the assignment of that level.
Cryocooler Reliability
As a first step in understanding cooler system reliability, it is useful to quantify the reliability
of a variety of individual representative long-life space cryocooler types. The identified cryo-
cooler failure mechanisms, shown in Table 2, are drawn from 10 years of personal knowledge of
the development and testing of coolers at a number of space-cooler manufacturers, both in the
U.S. and overseas. Similarly, the individual mechanism-level assessments are based on generic
cooler sensitivities observed over the years. However, the mechanism-level failure probabilities
reflect observations integrated over a number of similar cooler designs and must be corrected for
any given design if that design has made a special effort to resolve a particular issue. As an
example, Stirling coolers as a category suffer from a high sensitivity to side loading of the dis-
placer coldfinger, which can cause internal rubbing and wear. However, some manufacturers,
such as Raytheon, have greatly increased the robustness of their space-qualified Stirling-cooler
displacers with respect to this issue and would deserve a much lower failure probability for "Ex-
pander blowby due to long term wear."
The qualitative data displayed in Table 2 not only provide a useful assessment of overall
cooler reliability, but also provide insight into where to concentrate efforts to improve reliability.
One can see that unproved robustness with respect to contamination and leaks are key priorities
drawn from this author' s experience. Similarly, it can be seen that the use of pulse tube expanders
has eliminated a number of failure mechanisms associated with Stirling expanders.
One aspect of cooler reliability not included in Table 2 is the sensitivity of cooler probability
of failure to input power level. Sensitivity of reliability to power level is well known for driven
mechanisms such as cars, planes, and motors, and one can project a similar sensitivity to power
level for some cooler failure mechanisms. It is important to assess this sensitivity because fraction
loading level is an important trade-off parameter for cooler selection and sizing.
Table 3 presents this author's estimate of the effect of fraction piston stroke or fraction of
maximum input power on the failure probability of the pulse tube cooler design introduced as the
left-most column in Table 2. The various fraction power and fraction stroke column headings
reflect the nonlinear drop in power which accompanies reduction in stroke. Note that the failure
probabilities in Table 2 match the 85%-stroke operating point in Table 3; this is because an 85%-
stroke is considered the nominal design point for a cooler. Also note that Mure mechanisms such
as leakage are unaffected by power level, whereas others such as contamination from piston
rubbing are considered to be strongly dependent on input power and stroke.
Cryocooler Electronics Reliability
In addition to the mechanical cooler itself, most cryocooler systems include a relatively com-
plex set of cooler drive electronics used to generate the AC drive current from the 28 VDC
spacecraft power bus, to control the cooler operation, and to communicate digitally with the host
spacecraft or instrument. Although there are well developed ways of estimating electronics reli-
ability, Table 4 presents an abbreviated assessment of probability of failure reflecting the four
main ingredients of the cooler electronics including its power drive electronics, analog sensor
electronics, digital control electronics, and controlling software. As with the mechanical cooler,
power level has been included as a parameter to allow it to be used in subsequent trade studies.
Heat Switch Reliability

The use of redundant cryocoolers generally involves the consideration of heat switches to
isolate the thermal load of the turned-off (backup) cooler from the primary operating cooler, and
to connect the backup cooler when it is needed. The off-cooler heat load is the heat conducted
down the cold finger of the non-operating cooler, and it can be half of the total load if heat
switches are not used. Because heat switches are generally considered to be relatively high-risk
devices that add significant parasitic loads, heat switch reliability and performance must be fully
included in any trade study of cooler redundancy options.
As shown in Fig. 2, there are three generic types of heatswitch designs that have received
widespread attention: 1) the so called CTE-based switch, 2) the gas-gap thermal switch fed by a
hydride sorption pump, and 3) the gas-gap thermal switch fed by bottled gas.
The CTE-based switch utilizes materials of widely different Coefficients of Thermal Expan-
sion (CTE) to cause an outer element of a high-CTE material such as aluminum or copper to
shrink tightly around a low-CTE material such as molybdenum or beryllium when the high-CTE
material is cooled. The classic challenge with this design is the possibility of coldwelding of the
mating surfaces over long time periods, and the fact that available CTEs result in the gap between
the two materials being quite small This small gap makes the design quite vulnerable
to shorting out from small side loads. These reliability risk areas of the CTE-based thermal
switch are highlighted in the right hand column of Table 5 and summed to achieve an estimated
total Mure probability for this design of about 7.5%.
Figure 2. Three generic heat switch options.

The second type of beat switch design, the gas-gap switch, is also based on maintaining a
very small gap between the switch halves and therefore suffers from the same high sensitivity to
side loading; however, in this case the gap is designed to be always open so that cold welding is not
an issue. To actuate a gas-gap switch, a gas such as hydrogen is introduced into the gap to cause
conduction across the gap. The required filling and evacuation of the gas has its own set of
reliability issues, and these are noted in the gas-gap columns of Table 5.
In this author's assessment, the CTE-based switch is the more reliable of the three, but all
have failure probabilities similar to the cryocooler they are designed to be integrated with. This
low reliability prediction for heat switches is driven by the lack of good predictive analysis and test
methods for failure mechanisms such as cold welding, long-term leakage, gaseous contamination,
and small movements from warpage or side loads through flexbraids.
ESTIMATING CRYOGENIC LOAD GROWTH DUE TO REDUNDANCY

In addition to the reliability of the individual cryocooler and heatswitch assemblies, another
key consideration that must to be included into a cryogenic system trade study is the increased
cryogenic load associated with the introduction of redundant elements. As noted in Fig. 3, this
load growth has four components: 1) thermal conduction down added structural supports and
plumbing, 2) the radiation load to the external surfaces of added cold components, 3) the effective
load due to the thermal drop across added 'on' heat switches and thermal flexbraids, and 4) the
thermal conduction load through 'off' heatswitches and coolers.
Figure 3. Cryogenic loads associated with incorporating redundancy.

In a mature design, these loads could be computed with good accuracy based on the design
details. However, for a generic trade study early in the design process, what is needed is a means
of estimating each of these four loads based on rough sizing and mass estimates.
Structural Support Conduction. To estimate increased structural support conduction, what
is desired is a generic relationship between overall support conductance and supported mass for
typical launch loading conditions. This was derived by examining three flight-proven designs for
cryogenic structural supports: one using a Z-fold fiberglass tube that supports the TES-instrument
focal plane assembly, one using a hermetic glass tube that supports the AIRS-instrument focal
plane assembly, and a third using taunt fiberglass bands that supports the MICAS-instrument focal
plane assembly. Table 6 summarizes the very similar conductance computed for these three
diverse designs and presents the derived rule-of-thumb conductance load of 2 µW/gram·K.
Radiation Load. Increased radiation load due to added cold surface area is estimated based
on an effective emittance value of 0.05 for gold-plated surfaces or small-area MLI blankets typical
of those used in a cryocooler cold-end assembly.
Effective 'On-state Conduction' Load. The effective 'On-state conduction' load is the
additional load that a cooler would have to carry at a given cold-end temperature to make up for
the fact that it has to run colder (at temperature because of the thermal drop through
a conducting heatswitch and/or thermal flexbraid assembly. This effective load is sized to
yield the same cooler input power with load at temperature as would be required with
load (L) at temperature This Dload per is just the slope (watts/K) of the load curve
for the cryocooler of interest. It can also be roughly approximated using Eq. 1, where the con-
stant varies from about 0.8 to 1.2 mW/W·K for typical high-efficiency space cryocoolers.
(1)
In Eq. 1, SP is the specific power of the cryocooler at the cryogenic load temperature and AT
is the predicted temperature drop (K) through the 'On-state conductance'.
Off-state Conduction Load. The off-state conduction load down through a cryocooler cold-
finger and 'off' heatswitch is computed from the combined 'off' thermal resistance of the cooler
plus the heat switch. A typical value for a cryocooler 'off' thermal resistance is around 500 K/W,
and a typical value for a heatswitch is in the range of 1000 to 2000 K/W.
COMPUTING TOTAL SYSTEM-LEVEL THERMAL PERFORMANCE

At this point the necessary data have been generated to allow various cryocooler systems to be
compared on the basis of total thermal performance. Once the thermal performance has been
computed, the next section will then build up the total reliability performance.
To allow the various cryocooler redundancy options presented in Fig. 1 to be compared
thermally, one needs to develop detailed conceptual designs for each system including realistic
mass, size and parasitic-load estimates. This has been done for the four representative mechanical
systems illustrated in Fig. 4. These systems include a single cooler with no redundancy, a redun-
dant pair of coolers with CTE-based heat switches, and two variations of redundant coolers with
no heat switches. In all cases, the cold end is assumed to be at 60 K and incorporates radiation
shields tied to an assumed 160 K passively-cooled optical bench to minimize radiation loading of
the 60 K cold-end surfaces. In the second case of redundant coolers with no heat switches, the
Figure 4. Conceptual designs of four cooler redundancy options.
160 K shield is also tied to both cryocooler coldfingers to act as a 160 K heat interceptor.1,2 To
allow its thermal performance to be assessed, each system has been drawn to scale and mated to a
representative focal plane dewar assembly taken from the JPL AIRS instrument.3
For the case with the 160 K heat interceptor, the performance is assumed to follow the data
measured at JPL for the MMS 80 K Stirling cooler as presented in Fig. 5.1 From these data the
use of a 160 K heat interceptor is seen to reduce the parasitic conduction load down through the
non-operating cryocooler by 60%, and to also reduce the required input power of the operating
cooler by around 40%.
Table 7 presents the computed thermal performance data for five cryocooler redundancy
options: four with two redundant coolers (two with heat switches and two without), and one single
cooler with no redundancy to serve as a reference.
Figure 5. Measured performance gain at 60 K as a function of intercept

temperature for the use of a heat interceptor on the MMS 80 K cryocooler.1
The left-most two columns in Table 7 are for redundant cooler systems incorporating a JPL-
designed dual CTE-based heatswitch shown in Fig. 6. The new conceptual design for a CTE-
based heatswitch was specifically generated for this study because no heatswitch design was found
in the literature that had useful performance. The concept shown in Fig. 6 was designed to yield
a light-weight low-surface-area configuration with a Z-fold fiberglass tube that provides the needed
rigidity to prevent a one-pound side load from shorting out the CTE gap. The side load itself is
maintained below a pound by incorporating integral flexbraids into the cooler-attachment end of
the switches; these flexbraids serve a double duty by also providing the needed flex coupling to the
cooler coldfinger itself.
As shown in Fig. 4, the bodies of the switches are directly supported off of the 60 K focal-
plane assembly as is done in the actual AIRS flight instrument; this minimizes conductive loads by
making the structural conduction path to the 160 K optical bench instead of to the 300 K cryo-
cooler support structure.
The difference between the left-most two columns in Table 7 is the existence or absence of a
160 K radiation shield surrounding the heatswitch assembly. In the left-most column, the 160 K
radiation shield is absent, i.e. a background radiation temperature of 300 K is assumed for the
heatswitch assembly surfaces. This 300 K background leads to a very high (~ 300 mW) parasitic
radiation load, which is why all the other options have the 160 K radiation shield included.
Figure 6. JPL conceptual design for compact dual CTE-based heat switch.
Note that the total cooler load with these highly optimized heat switches and the 160 K shield
is about 200 mW higher than the single non-redundant cooler, and about 400 mW less than the
fully-redundant, no-heatswitch case. Thus, even these heat switches are only about 60% effective
at removing the conductive load of the off-cooler. Other heat switch concepts in the literature4 are
much heavier and larger, and are therefore much worse.
The lesson learned is that heat switches must be low in external surface area, be light weight,
and be enclosed with cryogenic radiation shields if they are to significantly reduce system parasitics
associated with redundant coolers. Whereas a very good heat switch implementation may reduce
parasitic loads by 50 to 70%, a poor heat switch implementation can lead to minimal load reduc-
tion, or may even increase parasitic loads.
In contrast to the use of heat switches, the center column of Table 7 describes the performance
of dual coolers without heat switches, but with a heat interceptor at the radiation shield tempera-
ture. This concept is seen to offer parasitics comparable to the use of good heat switches, and also
has the advantage of considerably improved cooler efficiency, as noted in Fig. 5.
COMPUTING TOTAL SYSTEM-LEVEL RELIABILITY PERFORMANCE

Now that the assembly-level failure probabilities (Tables 2 - 5) and system-level thermal
performance (Table 6) of the various redundancy options have been computed, this section com-
bines these to compute the total system-level reliability performance. As a first step in this process
it is useful to review the computational rules by which assembly-level failure probabilities com-
bine to form system-level failure probabilities.
System Reliability Computation
When various assemblies are connected in series and parallel to provide redundancy, the
manner in which their reliabilities (or failure probabilities) combine is described by classic prob-
ability theory.5 Although the definition of reliability is just one minus failure probability, the
resulting equations used to combine reliabilities are more complex than the equivalent equations
used to combine failure probabilities. For this reason, we have chosen here to use failure probabil-
ity (P), and not reliability (R=1 - P).
A typical cooler system, such as that shown in Fig. 7, is said to be made up of a series/parallel
combination of elements or assemblies, each with a failure probability When is small
compared to one (say P< 10%), the combined failure probability of a series of assemblies (such as
a cooler in series with a heat switch) is just the sum of then- individual failure probabilities, i.e.
This is true as long as the failures are statistically independent, i.e. the failure of one
does not influence the failure of the others, and each must work for the system to work.
On the other hand, when multiple assemblies are placed in parallel so that the system works
if any one parallel branch is functional, then the combined failure probability of the system is just
the product of the individual failure probabilities, i.e.
Using these relationships, the combined failure probability of the system shown in Fig. 7 can
be computed in terms to the cooler failure probability and heat switch failure probability
This computational methodology will be used to examine the
system-level reliability of the various cryocooler redundancy options explored in Table 7.
Figure 7. Classical dual cooler with heat switches configuration.

However, before doing that, it is instructive to examine a somewhat more complicated case to
validate the accuracy of the derived relationship.
CTE Heatswitch System Reliability

For the case where the heatswitches in Fig. 7 are CTE-based switches governed by the failure
probabilities in Table 5, the data suggest that these switches tend to only fail shorted. Thus, a
viable system can include a failed switch with an operating cooler together with a failed cooler
with an operating switch. This two-failure case was explored using a more sophisticated statistical
analysis techniques based on the binomial distribution function.5 For this case it is found that one
component failure is never a system failure, two thirds of two-failure scenarios are system failures,
and all cases of three or four failures are a system failure.
For the case where the heat switch and cryocooler failure probability are equal, which is not
an unreasonable assumption, the resulting system-level reliability is plotted in Fig. 8. From this
figure it is seen that redundant coolers with no heat switches (e.g. AIRS), or 100%-reliable heat
switches, provides significant reliability enhancement. Also, incorporation of heat switches with
the same reliability as the cooler raises system failure probability 4× over systems with no switches.
Reliability Summary for Cryocooler Redundancy Options
With the above background, we are now in a position to compute the overall system-level
reliabilities and mass/power performance of the various cryocooler redundancy options presented
originally in Fig 1. This is done in Table 8.
From this table it is seen that most of the redundancy options involving dual coolers or dual
electronics roughly half the probability of failure of a single (no redundancy) cooler system. For
the cases involving dual coolers, this is accompanied by a relatively large increase in system mass
and power. The clear winner in this analysis is the option involving full redundancy, no heat
switches, and the use of a 160 K heat interceptor to pick up the off-cooler parasitic loads and to
improve the efficiency of the operating cooler.

In this paper, an approach to cooler redundancy trade-offs has been developed and demon-
strated. The approach includes mechanism-level assessment of the reliability of key system as-
semblies (coolers, switches, and electronics), includes reliability dependency on cooler power
Figure 8. Affect of cooler redundancy on cooler system reliability.

level, includes means of estimating increased parasitic loads associated with redundancy, and
includes overall system mass, power, and cooling-load impacts.
Top level conclusions include: 1) The highest reliability is achieved with lightly-loaded, fully
redundant coolers with heat interceptors to reduce the parasitic load, 2) the addition of heat
switches can improve the system thermal efficiency somewhat, but with a significant increase in
failure probability, 3) the use of redundant electronics only (with an electrical switch) has similar
reliability to a system with heat switches, but with lower mass and power, and 4) a single cooler
provides the lightest weight, lowest power, and least cost, but may have marginal reliability for a
high-reliability mission.
ACKNOWLEDGMENT
The work described in this paper was carried out by the Jet Propulsion Laboratory, California
Institute of Technology, and was sponsored via Task Order 15144 by the National Oceanic &
Atmospheric Administration (NOAA) National Environmental Satellite Data and Information Service
(NESDIS) through an agreement with the National Aeronautics and Space Administration.
REFERENCES
1. Johnson, D.L. and Ross, R.G., Jr., “Cryocooler Coldfinger Heat Interceptor”, Cryocoolers 8, Ple-
num Press, New York, 1995, pp. 709-717.
2. Gilman, D.C., “Cryocooler Heat Interceptor Test for the SMTS Program”, Cryocoolers 9, Plenum
Press, New York, 1997, pp. 783-793.
3. Ross, R.G., Jr., et al., “ AIRS PFM Pulse Tube Cooler System-level Performance,” Cryocoolers 10,
Kluwer Academic/Plenum Publishers, NY, 1999, pp. 119-128.
4. Bugby, D., et al., “Development of Advanced Cryogenic Integration Solutions,” Cryocoolers 10,
Kluwer Academic/Plenum Publishers, NY, 1999, p. 684.
5. ARINC Research Corp., Reliability Engineering, Prentice-Hall, Inc., Englewood Cliffe, NJ (1964).
Space Cryocooler Contamination Lessons Learned
and Recommended Control Procedures
S. Castles1, K.D. Price2, D.S. Glaister3, W. Gully3,

J. Reilly4, T. Nast5 and V. Kotsubo5
1
NASA/GSFC, Greenbelt, MD, 20771
2
El Segundo, CA 90245
3
Ball Aerospace Boulder, CO 80301
4
5
Lockheed Martin Advanced Technology Center
Palo Alto, CA 94304
ABSTRACT
One of the most important characteristics of a space cryocooler is its reliability over a lifetime,
typically in excess of 7 years. While design improvements have reduced the probability of me-
chanical failure, performance degradation directly traceable to internal contamination has been
observed across the industry in significant numbers, both in endurance test units and flight units.
Therefore, the risk posed by excessive contamination is still a major concern and should be ad-
dressed in a consistent standard throughout industry.
This paper first describes the cryocooler contamination problem in general terms and then
describes one company’s efforts and experiences in addressing and resolving the problems. The
general discussion identifies a number of sources of internal contamination, the subsequent degra-
dation and failure mechanisms, and the observed impact to a cryocooler’s operational efficiency.
The paper suggests some specific contamination prevention procedures and their resultant benefi-
cial impact on cooler performance. The most common and preventable source of contamination is
the actual working gas charge, and through a comprehensive charge, purge, and fill procedure, the
risk of even long term contamination can be significantly minimized. One major source of unex-
pected contamination is from the vendor supplied gas bottles. The addition of an in-line getter or
cold trap to the fill lines helps reduce such contamination. Internal contamination sources that
slowly release volatile gases can also be a problem.
This paper also describes the lessons learned during Raytheon’s Standard Spacecraft Cryo-
cooler (SSC) and Improved Standard Spacecraft Cryocooler (ISSC) development programs in the
area of internal contamination, and the subsequent cryocooler gas sample analysis. The exception-
ally impressive results before and after the incorporation of identified changes are presented.
Based on the authors’ observations, the cryocoolers are much more robust to contamination
than expected. Contaminant levels generally have to exceed 100 ppm before noticeable piston-to-
cylinder or cold-tip temperature degradation is observed.

INTRODUCTION
The authors and their contributions to this paper represent a relatively broad cross section of
the space cryocooler development from both government and industry. Ken Price is the lead engi-
neer at Raytheon for the development and production of several cryocoolers including the SSC,
ISSC, and Protoflight Spacecraft Cryocooler (PSC) lines. John Reilly is associated with the AFRL,
and has been the government’s program manager for several cryocooler programs including the
SSC and PSC, the Ball Stirling, Joules-Thomson 10K-cryocooler program and the Creare Minia-
ture Reverse Brayton Cryocooler (MRBC). Dave Glaister recently joined Ball as a cryogenic sys-
tems engineer, but was with Aerospace Corporation providing technical support to the AFRL for
nearly all of their cryocooler development programs over the past 3.5 years.
One of the most important characteristics of a space cryocooler is its reliability over a lifetime,
typically in excess of 7 years. This long life requirement is the single factor that most distinguishes
the space cryocoolers from terrestrial-based technology, and is a primary factor driving space cryo-
cooler design. The concerns over mechanical related failure mechanisms (such as wear or fatigue)
has decreased with the advent of the ”Oxford” style flexure, which has been utilized extensively by
both Stirling and pulse tube space cryocoolers during the last decade. There have been successful
multi-year endurance test demonstrations of a number of cryocooler units. However, while design
and assembly improvements have reduced the probability of mechanical failure, the risk of internal
contamination is still significant. Many of the endurance test and flight units have experienced
some level of performance degradation related to internal contamination. Thus, the relative impor-
tance of anomalous performance resulting from internal contamination has actually increased.
CRYOCOOLER CONTAMINATION SOURCES

Table 1 contains a list of the most common cryocooler contaminants and their saturation
temperatures as a function of partial vapor pressure and gas concentration. Both vapor-to-solid
(sublimation) and vapor-to-liquid (condensation) transitions are included. Water is the most
common and pervasive contaminant and, at the concentration levels of interest, freezes below
273 K. Nitrogen and carbon dioxide contamination can also be significant and appear between
about 40 to 50 K and 120 to 140 K, respectively.
Contamination sources can arbitrarily be subdivided into two classes: ”free” gas contami-
nants and ”bound” gas contaminants. Free gas contaminants are those mixing freely with the
helium charge gas such as argon. They generally can be easily purged from the system by
simple replacement with pure helium gas. Bound gas contaminants are more difficult to re-
move because they are confined within the cooler by some mechanism. Examples include elec-
trically polarized molecules that stick to metallic surfaces and organic vapors evolving from
adhesives as they slowly cure. Removal of bound contaminants requires procedures beyond
simple gas replacement.
SPACE CRYOCOOLER CONTAMINATION LESSONS LEARNED 651
As will be subsequently discussed in this paper, these procedures extend from thoughtful
choices made in the design phase with the selection of the materials to be used in the manufac-
ture of the cryocooler, to the highly difficult task of minimizing bound contaminant sources.
Also discussed are the final gas charging procedures and selected hardware to be utilized in the
final gas fill/purge operations.
The following are the most common sources of contamination associated with space cryo-
coolers and suggested approached to resolving these problems:
1. Source Gas Contamination. Those working in the industry have found that high purity he-
lium canisters purchased directly from the major helium suppliers are often seriously contami-
nated with water, nitrogen, oxygen, carbon dioxide and/or argon. Therefore, until this situation
is corrected, procedures to filter the delivered gas must be implemented. Although some of
these contaminants sometimes behave as bound gases (e.g., polarized molecules), they behave
as free gas when being removed from the source gas. Filtering of these contaminants can be
implemented through high quality in-line gettering or cold trapping them upstream before the
gas enters the cryocooler.
2. Charge Fitting Leaks. There can be leaks in the lines and fittings that connect the gas source
to the cryocooler that introduce free gas contaminants (air) independent of the source gas qual-
ity. This source can be minimized through use of high quality pressure and vacuum compatible
fittings. Pressure testing for leaks before filling the cooler is advised.
3. Trapped Gas or Virtual Leaks. Free contaminants can be trapped within poorly vented
volumes inside the cooler. Examples of such volumes include the small space at the end of
blind screw holes. Tightly torqued screws retard flow of gas in and out of such spaces so they
are difficult to purge during charging operations. During the long operating life of the cooler,
such contaminants will slowly migrate into the cooler, generally finding their way into the cold
end. These contamination sources are best eliminated in the design phase by identifying them
and providing efficient vent ports. Trapped spaces that cannot be well vented require more
extensive and thorough purging operations.
4. Internal Cooler Contamination Generation. Bound gas contaminants originate in and on
materials that can desorb or outgas condensable vapors or gases. Examples include polarized
molecules adhering to metallic surfaces and organic vapors evolving from adhesives with long
curing time-constants. These sources can be reduced by minimizing the use of organic/porous
(o/p) materials, or hermetically isolating, whenever practical, any area where the o/p material is
used. Finally, baking out the cooler at high temperatures will activate bound contaminants into
the free gas stream prior to and during the purge/fill procedure, enabling their removal.
5. Cryocooler Leaks. Once charged and sealed, the cryocooler can develop leaks, which intro-
duce contaminants by reverse diffusion processes. This source can be minimized through her-
metic seals (welds or metal o- or c-rings, instead of organic o-rings) and more extensive and
thorough leak checks.
6. Chemistry in a Wearing Cooler. Gaseous contamination can be produced if a plastic liner is
allowed to wear significantly. The combination of local heating and stress can break down the
plastic and cause the release one or more troublesome gaseous products. The result is a contin-
ued accumulation of light gases in proportion to the running time of the cooler. This process is
analogous to the breakdown of ball bearing lubricants in conventional rotary mechanical cool-
ers. Thankfully, the spring supported linear cryocoolers under consideration in this article vir-
tually eliminate such wear.
CONTAMINATION RELATED CRYOCOOLER DEGRADATION

AND FAILURE MECHANISMS
Contamination related degradation and failure mechanisms for space cryocoolers generally
fall into two categories resulting from the collection of frozen contaminants toward the cold end:
1. Interference between a Stirling displacer piston and cylinder that limits piston motion.
2. Blockage of the flow passages in regenerators manifolds and/or heat exchangers.
When a cryocooler is initially charged or has been in a dormant state for an extended
period of time at ambient temperatures, contaminants within the working gas will be relatively
evenly distributed. The major portion of the working gas is within the compressor volume and
the majority of that gas is usually on the motor/flexure side of the compressor clearance seal.
Thus, most of the contaminants would not be initially near the cold end of the cryocooler.
However, during cryocooler operation, through a combination of pumping through the clear-
ance seals and vapor pressure differences, the condensable contaminants can migrate to the
coldest portions of the cryocooler.
For some Stirling space cryocoolers, (pulse tube coolers do not have a displacer piston
and, thus, are not subject to displacer clearance related contamination degradation), a clear-
ance seal is maintained between the displacer piston and cylinder wall. This clearance may be
as small as (0.5 mils) near the base of the displacer. The piston assembly is usually
supported by flexure springs that have high radial, and low axial stiffness. Thus, theoretically,
there should be no contact or wear between the moving pistons and the cylinders. Fortunately,
the critical areas for sealing in the expander are at or near ambient temperature, and thus are
unlikely to have condensation of contaminants. The clearances around the cryogenic portions
of the expander piston are not as critical to the unit’s efficiency, and can be significantly larger
than the clearances at the seals. The current generation of protoflight single-stage Stirling cryo-
coolers usually have expander cold tip clearances in excess of .075mm (.003 inch) to provide
significant tolerance to contaminant accumulation in this volume.
This is not the case with multi-stage coolers where one or more clearance seals may be at
cryogenic temperatures. For multistage coolers to work most efficiently, they require clear-
ance seals between the various expansion stages. Therefore, these coolers have tight clearance
at low temperatures that introduce extra sensitivity to contaminants. In fact, Ball Aerospace
(BA) typically found that these contaminants enhanced the performance of their cryocoolers!
For example, one of BA’s 30K cryocoolers typically had gaps wider than thermally opti-
mum in order to insure non-contacting, (piston to cylinder) motion. BA found that if they
employed helium contaminated by “air”, (500 ppm) the low temperature performance at the
cold tip was roughly 50% better than the 30 K performance using the high purity helium BA
normally employed. BA presumes that the condensed air in the cold gap(s) tightened the cold
seal. In fact, BA detected displacer “stiction” during these periods of improved performance.
The performance enhancement and the detected stiction disappeared when the working charge
was replaced with high purity helium, and the cold tip was heated above approximately 60 K.
This unexpected bonus cooling had initially compromised on-going cryocooler development
efforts, because it distorted the results of the controlled changes being made on the cryocooler
to further optimize the cooler performance during that same period of time.
The freezing of a significant percentage of the contaminants on the displacer piston wall
near the cold tip can result in the contaminant changing the sensitive alignment of the piston,
which may cause piston to cylinder wall contact. At the clearance seals, the piston is typically
coated with a thin layer of a plastic such as Vespel® or Rulon® to minimize any likelihood of
contact. Frequent contact at the seals could eventually erode the plastic liner and result in
metal-to-metal contact. The plastic liner typically does not extend much farther up the piston
from the seals. (If contact occurs above the liner, it would be metal-to-metal contact and result
in creating small metal burrs and filings, which would exponentially accelerate the degrada-
tion of the piston and cylinder surfaces.) Overall, if the freezing of contaminants on the dis-
placer piston cause piston to cylinder contact, the initial impact may not cause a noticeable
degradation in the cooler performance. The contact can instead be detected by electrical conti-
nuity monitoring to detect any piston to cylinder contact. However, if the contact continues to
occur on a frequent basis, the tight clearances will quickly degrade and result in a catastrophic
or “hard” cooler failure.
Contaminants can also build up in the cold tip expansion space of both Stirling and pulse
tube cryocoolers. If a large amount of frozen material is deposited in this space, it can degrade
the critical heat transfer at the cold tip or impede the stroke of the Stirling displacer piston.
Contamination impacts both pulse tube and Stirling regenerator flow passages (usually
stacks of metal screens) typically resulting in a much more gradual “soft” failure of cryocooler
performance degradation. In order to maintain both input power and cooling load efficiencies,
Stirling and pulse tube space cryocooler regenerators require uniform distribution of gas flow
through the regenerator passages which are typically between 1 and 3 mils in diameter. If some
flow passages are blocked within a segment of the regenerator screens by frozen contaminants,
the gas flow will be redirected through other passages. This increases flow rates through the
unblocked portions of the regenerator which increases pressure drop and effectively reduces
regenerator heat transfer surface area. Both of these effects combine to decrease the net cool-
ing at the cryocooler cold tip. Migration of contaminants to the cryogenic portions of the cryo-
cooler, or the freezing of contaminants within the regenerator flow passages are usually ob-
served as a progressive degradation in the net cooling and generally result in reduced heat lift
capacity.
RAYTHEON CRYOCOOLER CONTAMINATION EXPERIENCE

The following section describes experiences (including measured and observed effects of
contamination) with early Raytheon programs and the aggressive methods Raytheon and the
Air Force Research Laboratory (AFRL) employed to minimize future contamination problems
with their cryocoolers. The Raytheon SSC #2 is a single-stage, Stirling cycle engineering de-
velopment model cryocooler with a design point of 2 W of cooling at 65 K. Raytheon manu-
factured the cryocooler during the 1989 to 1994 timeframe and delivered the unit to the AFRL
in August of 1995. The cryocooler characterization at AFRL was initiated in December of
1995. After approximately 3 months of operational characterization testing, piston-to-cylinder
contact was noted after a sustained weekend cooldown period.
The anomalous operation provided a strong indication that the cryocooler could have in-
ternal contamination or that some malfunction caused a misalignment of the expander’s piston
to cylinder clearance gap. At this point, the AFRL program manager John Reilly requested
Aerospace Corp. to perform a dynamic X-ray analysis of the expander’s piston to cylinder
interaction. When X-ray images showed proper alignment of the expander piston, Mr. Reilly
arranged for Pernicka Corporation of Ft Lewis Co to perform an analysis of the cryocooler’s
helium gas charge in a further effort to determine the causes of the earlier detected piston-to-
cylinder contact.
The Pernicka gas analysis of the SSC #2 cooler utilized mass spectrometry (Model PC-
421 based on a quadrupole mass filter). The analysis had a relative error of less than 1 ppm.
However, because the measurement is referenced to a calibration standard, the absolute accu-
racy is limited by the quality of the standard. For water, Pernicka uses a NIST standard with an
accuracy of +28 ppm. The process by which a gas sample is extracted from a cryocooler may
result in the largest error contribution. It is critical that the sampling equipment (valves, lines,
and fittings) be free of contaminants. This requires adequate and repeated purging of the sam-
pling equipment prior to measurement. During the sample extraction, it is also critical that the
valves and fittings do not leak and introduce atmospheric contaminants.
Table 2 includes a summary of the SSC #2 gas analysis performed by Pernicka. It also
includes a sample that was taken from the ISSC #1 (Improved Standard Spacecraft Cryocooler
which was the successor to the SSC line) cooler after 23,500 hours of operation without a
failure. A summary of the analyses of these samples and the associated observed performance
of the cryocooler is also included in Table 2. Also included in the table is the manufacturer’s
specification for the 99.999% helium gas used to charge the coolers.
There are several significant conclusions, which can be made from the gas analyses:
1. The level of contaminants that were found in SSC #2 exceeded the source gas specification. The
98-ppm of water alone is ten times the gas container specifications. In follow-on discussions
with contamination experts, the consensus believed that the 99.999% helium quality only re-
fers to the gas prior to the filling of the gas canister at the gas vendor. (Unfortunately, the
canisters are generally not checked, cleaned or purged prior to being refilled. Thus, the gas
canister itself can easily introduce contaminants into the previously pure gas. Helium gas vend-
ers usually have several grades of helium purity, the highest rated at 99.9999% pure, but they
only sample one canister from a number of canisters filled over a given period of time. There is
no assurance that another canister bottle from that same batch will contain an acceptable purity
level, because that canister may have been contaminated to some level before being filled.
Additionally, the attachment of a pressure regulator at the gas vendor’s facility can introduce
contaminants into the gas bottle that were in the fittings between the regulator and bottle.
2. Raytheon believes the dominant source of SSC #2 contamination was the charge gas. This does
not discount the possibility of outgassing of bound gas contaminants as the source of contami-
nation that could have built up continually over a period of months. Once detected and purged,
there may have been few bound contaminants left in the system. Since there was no initial gas
analysis available to support this position, it is only conjecture. Initially, Raytheon had not
gettered or trapped the helium working fluid prior to charging the cryocooler.
3. The cryocoolers are much less sensitive to contamination than previous analysis had pro-
jected. The SSC #2 had a small expander piston-to-cylinder wall cryogenic clearance rela-
tive to Raytheon’s more recently designed SBIRS-Low and PSC coolers. Although it had a
reasonably large level of contaminants, the SSC #2 cooler's performance degraded after
many hours of operation, prior to a sustained period of unattended cool-down. This was
over a weekend that allowed the moisture contaminants to migrate to the cold head area
and subsequently restricted the free piston motion. The ISSC #1 gas sample was taken
after the cooler had run nearly continuously for about 23,500 hours. The ISSC #1 cooler
displayed a detectable, but reversible decrease in cooling capability after about 12 months
of continuous operation due to contaminants that limited displacer piston motion. The life
test continued through 33 months of operation without failure. During occasional shut-
downs for instrumentation calibration or unexpected power outages, contaminants redis-
tributed within the cryocooler and the cooler’s performance returned to original levels.
Overall, the SSC #2 and ISSC #1 results indicate that if contaminants are present in a high
enough concentration (200 to 300 ppm) under general test and analysis conditions, cryo-
cooler performance degradation can occur within the first year.
BALL AEROSPACE CONTAMINATION EXPERIENCE

In general BA’s experience closely matches that outlined above by Raytheon. When designing
a new cooler, BA’s approach was to run an end-to-end program. They would begin with a failure
mechanism analyses. This led to design budgets and process details for keeping the contaminants
within a tolerable level. In practice the levels derived, about 10 ppm, have been found to be well
below the levels that would result in noticeable undesired effects. BA had one occasion to check the
durability of their cleaning process. They had a seal failure (traced to an improperly designed bond
joint) after approximately 10,000 hours of running time on our two stage GSFC cooler. Before
opening the cooler, they sampled the several year old gas charge with a mass spectrometer analyzer
at a local vendor (Chematox). BA used a broad mass scan of 10 ppm sensitivity through 120 mass
units, and select integration enhancements to 10 ppb for water, carbon dioxide, methane, and
perfluorinated methane. They did not detect any contaminants in three successive samples of
the gas charge.
CONTAMINATION CONTROL CONFERENCE

In October of 1996, Raytheon and the AFRL jointly sponsored a contamination review
meeting with personnel from AFRL, Raytheon, AF SMC/SBIRS Low, NASA/GSFC, Aero-
space, JPL, and Pernicka. The meeting covered the SSC #2 gas analysis results, previous Ray-
theon contamination control methods, and recommendations for improved control. The recom-
mendations from that working group are summarized in Table 3.
The table also contains a summary of the current modified Raytheon control methods in
place as of 1998. The table subjectively indicates which recommendations were strongly en-
couraged by the working group. Some significant recommendations include the following:
1. Increase vacuum bake out time during assembly to further outgas long time constant contami-
nants.
2. Add an in-line getter or cold trap to the fill line to prevent/reduce contamination originating in
the gas bottles.
3. Keep the cryocooler and the fill line hardware filled with dry helium at all times.
4. Use stainless steel fill lines wherever possible.
5. Increase temperature of the hot purge to at least 75C to 80C with the precaution of employing
materials that do not degrade at the vacuum bake and hot purge temperatures.
6. Obtain at least one reference sample of the source gas for analysis along with fill gas samples
taken during the purge/fill process. Use this information to characterize and verify the mini-
mum number of cycles needed to achieve suitably purified helium within the cryocooler.
RAYTHEON’S CORRECTIVE MEASURES AND RESULTS

Following the meeting, Raytheon implemented several improvements to their contamina-
tion control methods for all of the cryocooler programs including the PSC and the SBIRS-Low
Flight Demonstration System. The improvements included the following: vacuum bake-outs
were extended from an initial minimum of 2 to 3 days, an in line getter was added to the fill
process, the fill hardware is kept filled with dry helium, over 80% of the copper fill lines were
replaced with stainless steel, the displacer was oscillated during the dwell times of the fill/
purge process, and a reference fill/purge process was performed with periodic gas analysis.
Aerospace recommended the use of a helium gas purifier or getter instead of a liquid nitro-
gen cold trap. Some of the advantages of the gas purifier are: 1) That it does not require char-
acterization and periodic calibration, 2) That it is relatively insensitive to the amount of con-
taminants in the flow of about 2000 parts per million (ppm), 3) That it can reduce contaminants
down to the parts per billion (ppb) level, 4) That it does not require thorough cleaning and it
does not generate contaminants. Two important disadvantages are that the gas purifier does
not absorb nitrogen and that it has a finite life. The limited life of the purifier requires that
careful logging entries be maintained to ensure replacement of the purifier at the appropriate
times, (which is not during a cryocooler fill/purge procedure).
In implementing the contamination lessons learned, Raytheon upgraded their fill/purge
procedures. These procedures were implemented on the Raytheon SBIRS-Low #1 cryocooler.
Table 4 provides a summary of the gas analyses (done by Pernicka) of the samples taken dur-
ing the reference run. The table again shows very high levels (especially of nitrogen) of con-
taminants in the source gas. However, downstream of the purifier, the water contaminants had
been reduced to near zero and the nitrogen reduced by a factor of 10 due to the efficient fill,
bake-out and purge process. After 15 cycles, the total measured contaminant level had dropped
below 10 ppm as compared to over 1500 ppm for the source gas. Although it is not yet fully
understood why the nitrogen levels dropped so dramatically, this reference test confirmed that
the greatest improvement to the cryocooler contamination control is the filtering of the source
gas.
CONCLUSIONS
One of the most important concerns with space cryocoolers is their long term (> 10 years)
reliability. Contamination continues to remain a primary threat to space cryocooler’s sustained
reliability. This paper focuses on the two primary sources of contamination, free gas contami-
nants, (mixing freely with the helium gas) and bound gas contaminants (such as electrically
polarized molecules that stick to metal surfaces and organic vapors evolving over time from
adhesives). Some of the resultant cryocooler degradation scenarios resulting from contamina-
tion buildup are discussed. The paper identifies five general sources of contamination, but
specifically stresses concern with: Source gas contamination, trapped gas or virtual leaks, and
internal cooler contamination generation. Methods are identified to combat each of these sources.
It further cites the need to filter the source gas.
The second portion of the paper deals with Raytheon’s experience with contamination,
particularly focusing on their SSC #2 cryocooler. After a initial X-ray analysis by Aerospace
and a subsequent gas analysis by Pernicka determined that the cooler’s gas was contaminated,
a contamination conference was held at Raytheon which brought in some of the best contami-
nation experts from government and industry to recommend more comprehensive contamina-
tion procedures. These are listed in Table 3 of this report. The report then cites what additional
measures Raytheon implemented to minimize/prevent any future cryocooler contamination
problems with their cryocoolers. Finally, the paper identifies the excellent results Raytheon
has had over the last several years with the implementation of their new, more comprehensive
procedures.
REFERENCES
There are a limited number of reference documents available in this area. Two recom-
mended papers are:
1. Hall, J. L., and Ross, R. G., Jr., “Gas contamination Effects on Pulse Tube Performance,” Cryocool-
ers 10, Plenum Publishing Corp., New York (1999), pp. 343-350.
2. Hall, J.L., Ross, R.G., Jr. and Le, A.K., “A Contaminant Ice Visualization Experiment in a Glass Pulse
Tube,” Adv. in Cryogenic Engineering, Plenum Publishing Corp., New York ( 2000) (In Press).
Cryocooler Contamination Study:
Temperature Dependence of Outgassing
S.W.K. Yuan and D.T. Kuo

BAE Systems, Cryogenic Products
Sylmar,CA 91342
ABSTRACT
With the advance in technology, the life of the tactical cryocooler has extended way beyond
the conventional 4,000 hours milestone1. And with the life extension, the contamination control of
the manufacturing process has to be revisited, or cooler life may fall short of expectation, because
outgassing is a function of time. This means that cooler components need to be baked out at a
higher temperature or for a longer period of time. This in turn requires the knowledge of outgassing
rate as a function of temperature. Contamination study of foreign gases and liquids in a cryocooler
was discussed elsewhere2. Outgassing curves of water, alcohol, and acetone at 71°C were pre-
sented in the above study. This is a follow-up study emphasizing the temperature dependence of
outgassing.
INTRODUCTION
As the mean time to failure (MTTF) of cryocoolers approaches and surpasses 10,000 hours,
contamination becomes a crucial factor that impacts cooler’s life. Cooler manufacturers are faced
with the task of further reducing the beginning-of-life contaminant levels so that the end-of-life
levels will still lead to performance that meet the specification. This translates to longer bake-out
time at elevated temperatures. To shorten the process time, bake-out at a higher temperature is
called for. This requires knowledge of outgassing as a function of temperature.
Outgassing properties of various gaseous and liquid contaminants in cryocoolers have been
investigated in Ref. 2. Due to the lack of outgassing data as a function of temperature, a linear
function was adopted, assuming desorption of gases from a monolayer3. Further testings con-
ducted at BAE Systems indicated that the temperature dependence of outgassing is nonlinear. The
test setup and procedure are described in Ref. 2, except that the test chamber temperature was
varied between 23°C and 100°C. Figures 1 to 3 show the concentration of acetone, alcohol and
water as a function of temperature respectively. As one can see the data show an exponential
dependence of concentration as a function of temperature, with much higher outgassing rates at
elevated ambient temperatures.
The high outgassing rate at elevated temperature was first attributed to the possibility of phase
transition. As the boiling points of the liquids are exceeded, vaporization of the liquids may have
resulted in the high contents of concentrations being measured. Further testing on the outgassing
property of air shows that it exhibits similar exponential behavior. This confirms that the expo-

Figure 1. Concentration of acetone vs. temperature.
Figure 2. Concentration of alcohol vs. temperature.
Figure 3. Concentration of water vs. temperature.

TEMPERATURE DEPENDENCE OF OUTGASSING 661
Figure 4. Concentration of air vs. temperature.
nential outgassing property as a function of temperature is intrinsic to all gases and liquid tested.
Using a least squares curve approach, the above data can be fitted with the following equation.
(1)
where m is the concentration, T is the ambient temperature, and are constants.

Knowing the temperature dependence of the outgassing property, one can then combine with
the time dependence function reported in Reference 1 to arrive at the following equation.
(2)
Where is a function of the initial contaminant level in the system (i.e., the larger the value of
the higher the contamination level), B is the temperature dependence constant found in Eq. (1), and
A is the time dependent constant found in Reference 2. The values of these constants are listed in
Table 1.
Equation (2) is plotted in Figs 5, 6, and 7 for acetone, alcohol and water respectively at three
temperatures, 100°C, 71°C, and 23°C. Experimental data at 71°C are also included.
Figures 5, 6, and 7 are extremely useful in devising the bake-out process. The same level of bake-
out can be achieved with a much shorter time at an elevated temperature. For example, to reach the
same concentration level, the bake-out time for alcohol and water at 100°C is only one tenth that of
71°C. As for acetone, the bake-out time at 100°C is only a quarter that of 71°C.
As the life of cryocoolers extends beyond 10,000 hours3, the tolerance of contaminants in the
cooler becomes smaller. Detection of small level of contaminants at the beginning-of-life can often
be difficult especially at relatively low temperatures. The current procedure of BAE’s gas chro-
matograph (GC) sampling of the working gas calls for two hours of bake-out time at 71°C before
Figure 5. Acetone concentration as a function of time.
Figure 6. Alcohol concentration as a function of time.
Figure 7. Water concentration as a function of time.

TEMPERATURE DEPENDENCE OF OUTGASSING 663
taking data. By elevating the temperature to above 71C, one can increase the outgassing rate (thus
resulting in a more accurate reading of the concentration) and shorten the process time.
Although it seems feasible to bake-out coolers at the highest temperature the properties of the
cooler materials would allow, precautions must be taken not to expose the motor magnets to tem-
peratures higher than the manufacturer’s specification, which may result in irreversible demagneti-
zation of the magnets.
With Eq. (2), one can proceed to calculate the outgassing rate (dm/dt) by taking the derivative
of the equation.
(3)
The outgassing rates of acetone, alcohol and water at various temperatures are plotted in Figs.
8, 9, and 10, respectively. As mentioned before, outgassing at high ambient is far more effective,
with a steep rise between 71°C and 100°C. The outgassing rate is about three times higher in 100C
compared to 71C for acetone, and about five times higher for alcohol and water.
Figure 8. Outgassing rate of Acetone.
Figure 9. Outgassing rate of alcohol.

Figure 10. Outgassing rate of water.
CONCLUSIONS
The temperature dependence of outgassing has been studied in this paper. Outgassing curves
of acetone, alcohol and water are presented as a function of temperature. It was found that the
outgassing of the acetone, alcohol, water and air, is an exponential function of the ambient tempera-
ture. Results of this study are very useful for defining bake-out processes of cooler manufacturing.
It can also be used to cut down the process time by baking out components and subassemblies at
high ambient temperatures.
REFERENCES
1. Kuo, D.T., and Yuan, S.W.K, First Order Life Estimation and Its Correlation with Experimental
Data, parallel paper in this conference.
2. Roth.,A., Vacuum Technology, North-Holland (1989), pp. 187.
3. Yuan, S.W.K., Kuo, D.T., Cryocooler Contamination Study, to be published in Proc. of Advances in
Cryogenic Engineering, Vol. 45, 1999.
BAE's Life Test Results on Various Linear Coolers
and Their Correlation with a First Order
Life Estimation Method
D.T. Kuo, T.D. Lody and S.W.K. Yuan

BAE Systems, Cryogenic Products
Sylmar, CA 91342
ABSTRACT
Life test results of various models of BAE Stirling coolers are presented in this paper together
with a first order life approximation model.
A cryocooler life estimation method based on the Watt-Hour approach has been developed
elsewhere1. According to this method, the total energy of a cryocooler (i.e., the product of mean
input power and total operating time) is conserved. From actual life test data of input power rise as
a function of time, the energy of the cooler in Watt-Hour can be calculated by integrating the life
test curve. With this knowledge and the specification, one can proceed to estimate life. The big-
gest disadvantage of this method is that it requires prior knowledge of the life test data before life
estimation can be performed. In the present paper, a first order approach is used to estimate the rise
in input power as a function of time, which can then be used in life estimation.
INTRODUCTION
BAE Systems has conducted life test on various linear motor coolers, including B512C (Ref.
2), B602C (Ref. 3) and B1000E coolers. Conditions of the life tests are summarized in Table 1.
Life test data of three different models of coolers in input power vs. time can be found in
Figures la, 2a, and 3a. Life test data of cooldown time vs. time are plotted in Figures 1b, 2b, and
3b. And life test data of minimum refrigeration vs. time are presented in Figures 1c, 2c, and 3c,
respectively.

Figure 1a. Input Power of B512C cooler vs. time
Figure 1b. Cooldown time of the B512C cooler vs. time.
Figure 1c. Minimum refrigeration of the B512C cooler vs. time.

BAE’s LIFE TEST RESULTS ON VARIOUS LINEAR COOLERS 667
Figure 2a. Input power of the B602C cooler vs. time.
Figure 2b. Cooldown time of the B602C cooler vs. time.
Figure 2c. Minimum refrigeration of the B602C cooler vs. time.

Figure 3a. Input power of B1000E cooler vs. time.
Figure 3b. Cooldown time of B1000E cooler vs. time.
Figure 3c. Minimum refrigeration of B1000E vs. time.

The life test of the BAE B602C (0.6 W linear) cooler was terminated at 10,008 hours when it
failed the specification. The failure was caused by particulate contamination which blocked the
flow in the regenerator. After installing a new regenerator, the performance of the cooler was
restored. This proved that the compressor was not the root cause for failure. The life tests of both
the B512C (0.5 W linear) coolers and the B1000E (1 W linear) cooler are still in progress, with the
MTTF of the B512C cooler and the B1000E cooler exceeding 9,000 hours and 5,000 hours respec-
tively.
Cooler Life Estimation and Correlation with Experimental Data

Cooler life estimation using a Watt-Hour approach was evaluated in Ref. 1. The method
assumes that the total watt-hours of a cooler are conserved, i.e., running the cooler at low power
will extend its life and vice versa. Under normal conditions, cooler life is a function of compres-
sor piston wear. As the clearance gap grows (which increases the blow-by losses) due to wear, the
driver needs to drive the piston harder to make up for the lost performance. This in turn increases
the power (see Fig. 4a). When the power exceeds the user’s specification, the end-of-life of the
cooler has been reached. Given the experimental data of the input power increase as a function of
time (power curve, Fig. 4a), one can then calculate the total watt-hours of the cooler by integrating
the power curve to calculate the total area underneath it. For the example in Fig. 4a, an integra-
tion from the initial power of 8 W to the specification of 14 W, gives us the total watt-hours of the
cooler that is then entered into Fig. 4b as data point a. The procedure is repeated for various
initial powers to come up with Fig. 4b. Of course, if one starts with an initial power of 14 W, the
cooler would have essentially no life as indicated in point b of Fig. 4b.
A major shortcoming of this method is that it takes extensive life test data to predict life.
Moreover, due to the variation in performance from cooler to cooler, data taken from one cooler
may not be applicable to others. This means that a large number of coolers need to be tested before
the life of an average cooler can be determined.
In this paper, a simple first order model is proposed to estimate the rise in input power as a
function of time (power curve). The slope of the power curve (Fig. 4a) is assumed to be propor-
tional to the heat load, the ambient temperature, and the charge pressure, and inversely proportional
to the cold tip temperature and the bearing area of the piston seal.
The effect of heat load and cold tip temperature on life can be deduced from Figs. 5 and 6.
Heat load (cooling capacity) versus input power is plotted in Fig. 5, and heat load versus cold tip
temperature is plotted in Fig. 6, for the B1000E cooler. Heat load is almost a linear function of the
input power within the range of interest. Large heat loads require high input powers which reduce
the cooler life. The linear dependence of heat load on cold tip temperature (in Fig. 6) suggests that
the cold tip temperature is also a linear function of life. With the same input power, the heat load a
Figure 4. Watt-Hour life estimation.

Figure 5. Heat load vs. input power at 78K. Figure 6. Heat load vs. cold tip temperature.
Figure 7. Input power vs. ambient temperature. Figure 8. Input power vs. piston wear.
cooler can alleviate is less at low cold tip temperatures. This means that for the same heat load, one
has to apply a higher input power at a low cold tip temperature, which shortens the life of the cooler.
In Fig. 7, the influence of ambient temperature on input power is depicted. The data were
taken on a B512C cooler at various ambient temperatures at 78 K and with 300 mW heat load. The
data appear to be quite linear between 0 to 60°C, with a much sharper rise at temperatures above 60°C.
The effect of surface area on life can be obtained from Figure 8, which shows the predicted
performance of a cooler (by a second order cooler simulation model) versus compressor piston gap.
Since the piston wear is inversely proportional to the piston surface area, and the piston gap is a
measure of the piston wear, Fig. 8 implies an inverse linear function between cooler life and bear-
ing surface of the piston.
With the above information, one can come up with a simple first order model, by assuming
that the slope of the power curve (Fig. 4a) in W/hr is
(1)
where P is the charge pressure in bars, Q is the heat load applied to the cooler in watts, TH is the
ambient temperature in °C, A is the piston area and is the cold tip temperature in K. Cont
is a constant that equals to 5.475E-5.
The slope of the power curve (Fig. 4a) can be calculated by Eq. (1) with the parameters listed
in Table 1. The predictions are then compared to the life test data of the three coolers in Table 2.
The model gives a good prediction on the slope of the power curve for both the B512C and the
B602C coolers despite a wide range of differences in cooler parameters (Table 1). The correlation
of the first order model with the life test data of the B1000E cooler is not possible due to a decrease
in input power (negative power curve slope) as the cooler wears in during the life test. More run
time is needed in order to validate the model. Equation (1) can be applied to other Stirling coolers,
for the trends described in Figs. 5 to 8 are generic to most coolers. In order to apply Eq. (1), one
must have some life test data as depicted in Fig. 4a to obtain the constant (Cont) in Eq. (1), With
this information, one can proceed to estimate life of different models of coolers of the same design,
or coolers of the same model operated at different conditions.
As depicted in Figs. 1 to 3, some of the parameters to be monitored during a life test include,
input power, cooldown time and minimum refrigeration. Failure in meeting the specification in
any of these three criteria constitutes a failure of the cooler. The life estimation discussed in this
paper can be applied to all three criteria mentioned above. One simply measures the slopes of the
life test data in W/hr (rise in input power vs. time in Figs. 1a, 2a and 3a), minutes/hour (increase in
cooldown time vs. time in Figs. 1b, 2b and 3b) and mW/hr (decrease in minimum refrigeration vs.
time in Figs. 1c, 2c and 3c). To estimate the life of the same build of coolers under another set of
operating conditions, simply apply Eq. (1). The slopes of the life curves are directly proportional to
the ambient temperature, heat load, charge pressure, and inversely proportional to piston area and
coldtip temperature. For example, if the ambient temperature is doubled, one would expect the
slopes to be doubled and life of the cooler halved, and if the heat load is halved, the slopes are
halved, and cooler life doubled, etc.
Precautions must be taken in using Eq. (1), not to exceed the range that this simple approach
is intended for. For example, a cooler without any heat load applied will not have infinite life or
zero slope (for the power curve). Also, the equation is only valid for ambient temperatures above
0°C. Generally speaking, the room temperature data is a worst case estimation for life at sub-zero
(°C) temperatures.
CONCLUSIONS
Life test results of BAE’s B512C, B602C and B1000E coolers were presented. The B602C cooler
exceeded 10,000 hours of life test, with the life test of both the B512C (> 9,000 hours) and the
B1000E (> 5,400 hours) coolers still in progress. A simple first order life test estimation is sug-
gested, which gave good correlation to the life test data of the B512C and the B602C coolers. More
data are needed to validate the model against the B1000E cooler. The effect of charge pressure on
life should be further studied. The life estimation method proposed in this paper can be used to
predict cooler life as limited by the constraints of input power, cooldown time, and minimum re-
frigeration.
REFERENCES
1. Kuo. D.T., Loc, A.S., Lody, T.D., and Yuan S.W.K., “Cryocooler Life Estimation and It’s
Correlation with Experimental Data,” Advances of Cryogenic Engineering, Plenum Press, NY,
vol. 45 (2000).
2 Yuan, S.W.K., Kuo, D.T., and Loc, A.S., “Qualification of the BEI B512 Cooler, Part 1 -
Environmental Tests,” Cryocoolers 10, Plenum Press, NY, pp. 105-110.
3. Yuan, S.W.K., Kuo, D.T., Loc, A.S., and Lody, T.D., “Performance and Qualification of
BEI’S 600 mW Linear Motor Cooler,” Advances of Cryogenic Engineering, Plenum Press,
NY, vol. 45 (2000).
Initial Observations from the Disassembly and
Inspection of the TRW 3503 and Creare SSRB
B. J. Tomlinson, Jr. and C. H. Yoneshige

M. L. Martin
Dynacs Engineering
ABSTRACT
The importance of long term endurance evaluation data on cryocooler technology cannot be
overemphasized. As useful and important as operational data is failure analysis data and the
lessons learned from those failures. During fiscal year 1999, the TRW 3503 pulse tube and the
Creare Single Stage Reverse Brayton (SSRB) stopped operating during endurance evaluation at
the Air Force Research Laboratory (AFRL) Cryogenic Cooling Research Facility (CCRF).
On 8-9 November 1998, the TRW 3503 pulse tube cryocooler suffered a system overheat as
a result of an environmental control electronics failure. The cooler kept operating throughout the
overheat and continued to run after the environmental system was stabilized. The TRW 3503
continued to run with only a small performance degradation of approximately +2K over the
design point with a heat load of 0.3W until it tripped and shut down on 20 April 1999. The
cooler would not restart, so AFRL engineers and technicians worked with TRW to investigate
the cause of the tripping.
On 19 December 1998, the Creare SSRB shut down due to an enviromental system fail-safe
overheat condition on the compressor. Subsequent attempts to restart the cooler were
unsuccessful. The compressor started normally, but the free spinning turboexpander (TE) was
not able to lift off of its bearings and start spinning. AFRL personnel worked with Creare to
investigate the cause of the TE’s inability to lift up and spin. This paper will discuss the
investigation of both cryocoolers’ inability to restart, initial observations in post-endurance
disassembly, and lessons learned from these experiences.
INTRODUCTION
The importance of long term endurance evaluation data on cryocooler technology cannot be
overemphasized. As useful and important as operation data is failure analysis data and lessons
learned from those failures. Some of the potential contributors to cryocooler unreliability include
wear, performance drift, fatigue, material creep, gaseous contamination particulate/compound
contamination and clogging, material and workmanship defects, inadequate machining process

development, magnetic circuit degradation, assembly errors, material thermal expansion

mismatches, and long-term alignment instability.
In the case of the TRW 3503 Pulse Tube Cryocoler, the environmental system failure and
resultant overheat condition pushed this technology far beyond any design condition ever forseen
by TRW engineers. Even with the overheat, the cryocooler continued to operate, with only a
small degragation in thermodynamic performance, for over 5 months and shut down due to a
failure to stake a component, which was previously identified by TRW during the construction of
the cooler, and but was not fixed by the government. In the case of the Creare Single State
Reverse Brayton (SSRB) cryocooler, the environmental condition that caused the SSRB to shut
down was not nearly hot enough to damage the compressor. However, the system (specifically
the turboexpander) would not start due to a balancing problem that has been corrected in flight
hardware development at Creare. AFRL teamed up with both TRW and Creare to investigate and
document these failures.
TRW 3503 PULSE TUBE CRYOCOOLER

The TRW 3503 Pulse Tube Cryocooler is a protoflight unit with a design point of 0.3W @
35K. It went through characterization at the NASA Jet Propulsion Laboratory (JPL) and AFRL,
and then began its endurance evaluation at AFRL. It operated for more than 12,000 hours. This
cryocooler was developed under the TRW 35K Pulse Tube Program, and is similar in design to
the TRW 6020, the NASA TESS and AIRS cryocoolers, and the Multispectral Thermal Imager
(MTI) Space Cryocooler.
Environmental System Failure and Cryocooler Overheat

On 8-9 November 1998, the TRW 3503 environmental system electronics failed and the
cryocooler significantly overheated. An investigation of the environmental control electronics
revealed that one of the two IOTech 488/4 Digital to Analog (D/A) converters failed. The D/A
converters link the Lab VIEW™ dataport to the environmental system electronics by converting a
digital signal from the dataport to a current, which commands the chiller fluid flow vavles to
some increment between fully opened and fully closed. When the D/A converter failed, it
stopped sending current signals to the valves. The environmental control electronics read this as
a zero amp “signal” and commanded the vavles to fully close, depriving the cryocooler of its
heat rejection interface.
Examination of the data revealed several interesting things about the temperature excursion.
The flow valves for the TRW 3503 were commanded to close at ~2:50 AM on 8 November,
which caused the heat rejection temperature and case temperature to rise. The heat rejection
temperature continued to rise, surpassing 330K (57° C) at 4:48 AM that same morning. The heat
rejection temperature and cryocooler case temperature remained significantly above the bakeout
temperature for almost 23 hours. The cooler case temperature reached a maximum temperature
of 428.6K and the cold tip temperature reached a maximum temperature of ~65K. Figure 1
shows the heat rejection temperature, the cold end temperature, and the compressor strokes of the
TRW 3503 over the period of the temperature excursion.
During the entire time that the chiller fluid valves were closed, the cooler continued to
operate. As the case temperature rose, the stroke shortened (reaching stroke lengths as small as
9mm compared to the nominal stroke of 12 mm), which is why the cooler didn’t trip and shut
down. Once the heat rejection temperature and case temperatures reached steady state at ~9:00
AM on 9 November 1998, AFRL engineers ran a comparative load line with the heat rejection
temperature set at 300K. This load line was compared to one that was run on 6 November 1998.
As a testament to the robustness of this design, the TRW 3503 endured extremely high
environmental temperatures for a significant period of time, and was still able to return to nearly
the same performance. There was not much change in the average stroke lengths (the average
stroke length required to reach design point on 6 November was 11.93 mm and increased to
11.97 mm on 9 November), and the input power required to reach design point only increased by
OBSERVATIONS OF THE TRW 3503 AND CREARE SSRB 675
4W after the overheat condition. After the comparative load line, the cooler was returned to its
nominal conditions of 0.3W @ 35K, with an average stroke length of 11.97 mm. The TRW 3503
continued the endurance evaluation and continued to run with only a slight performance
degradation of ~2K with a 0.3W heat load.
System Shutdown and Diagnosis of the Problem

On 20 April 1999, at approximately 4:00 AM, the TRW 3503 tripped and shut down. The
Heater Interlock System (a fail-safe electronic circuit to prevent the heater from continually
running if the cryocooler stops operation) for the cryogenic heater cut power to the load resistor
on the cold block, protecting the system from another overheat. Upon arriving at the lab at ~4:20
AM, an AFRL technician tried unsuccessfully to restart the cooler.
From 20-23 April 1999, AFRL engineers and technicians worked with TRW to investigate
the situation. Tests were done to eliminate the TRW electronics rack, signal amplifiers and
capacitance sensor electronics box. Current displacement and oscilloscope traces taken of
compressor 2 and its capacitance sensor showed anomalous readings (Figure 2). The problem lay
somewhere inside the cooler, and the decision was made to perform a post-endurance evaluation
disassembly and inspection. These activities included an analysis of the working gas inside the
cooler, transport of the cooler to TRW for disassembly and failure analysis, and return of the
hardware to AFRL for post-disassembly stiction tests and determination of final disposition of
the hardware.
In Figure 2, channel 1 shows the current probe trace for compressor 2, channel 2 shows the
drive output of the TRW electronics, channel 3 shows the drive 2 output of the Techron
amplifier, and channel 4 shows the compressor 2 capacitance sensor output. Note the distortions
in the compressor 2 current and the capacitance sensor traces show an overall distortion and
periodic spikes in the current and displacement.
Figure 1. Case temperatures on the TRW 3503.

Figure 2. Traces from the TRW 3503.
Gas Sampling and Analysis

On 22 May 1999, the TRW 3503 was taken to Pernicka Corporation in Fort Collins,
Colorado for a gas analysis. There was a relatively large amount of carbon dioxide seen in the
analysis results. This was to be expected if there were rubbing parts inside the cooler. Also of
note is the relatively small amount of.organic materials seen in the gas composition. This
provides evidence that the epoxies used in the cooler did not significantly out-gas, even at the
high temperatures reached during the overheat condition.
3503 Cryocooler Disassembly at TRW

The cryocooler was taken to TRw for disassembly and a failure analysis on 27-28 May
1999. The compressor housings were removed, and there was no evidence that the flexures were
rubbing inside the cooler. The cooler was tested for shorts and all of the readings showed very
high or infinite resistance. Then each compressor was run at 5-10 at 24 – 44 Hz. The current
traces seen were smooth and clean and showed no indication of a short in the cooler.
Upon removing the 9-pin connectors, some feathering was visible on the Teflon casings
used to insulated the power leads, which caused some concern, The wires were looked at and
photographed under a CCD microscope. The feathering was caused by sharp edges in the holes
that the wires were fed through. The holes are in a position that doesn’t allow for de-burring or
smoothing on the inside surface of the structure.
Stiction tests on both compressors showed no current spikes, which would have indicated
that contact was occurring inside the cooler. Increasing the stroke length, however, did show
anomalous current traces similar to those seen at AFRL.
Figure 3. Capacitance sensor target with the locking nut detached.

The next step was to electronically disconnect the capacitance sensors and try running the
compressors seperately. Compressor 2 started making a ticking noise. Compressor 1 ran with no
problems.
The technician at TRW suspected that something might have happened to the capacitance
sensor itself. The capacitance sensor in compressor 2 was pulled out of its target, which revealed
that the locking nut on the capacitance sensor had backed off and was loose. The loose nut was
causing the cooler to trip erroneously by hitting the end stop and simulating over-stroking. The
target vibrating inside the cooler caused the rattling noise. In this prototype unit, the locking nut
had not been staked to the target. Figure 3 shows the capacitance sensor with the locking nut
detached.
The locking nut was reattached to the capacitance sensor target, and current traces were
taken again. The cooler was taken back to AFRL and current/voltage traces were taken with the
TRW 3503’s capacitance sensor box hooked up. Again, the master gain was set at ~0.1. The
traces taken after the disassembly show no anomalous spikes or readings, proving that the loose
locking nut on the capacitance sensor target was indeed what was causing the TRW 3503 to trip
erroneously.
CREARE SINGLE STAGE REVERSIBLE BRAYTON CRYOCOOLER

The Creare Single-Stage Reverse Brayton (SSRB) cryocooler was developed under
sponsorship by the Ballistic Missile Defense Organization (BMDO), monitored by AFRL and
managed by NASA Goddard Space Flight Center (GSFC). The Creare SSRB was designed to lift
a 5W heat load at 65K, and operate for 10 years on-orbit. The SSRB technology was chosen by
NASA to augment the dewar on the Hubble Space Telescope for the NICMOS system. The
SSRB logged over 29,000 hours of operation at AFRL.
Environmental System Failure and System Shutdown

On 19 December 1998, the Creare SSRB automatically shut down when its case temperature
exceeded the limit set on its temperature safety circuit. This was not an excessive overheat, it
only reached the set point of the safety limit. On 21 December 1998, AFRL personnel
investigated the cause of the shutdown and found that the circulating pump motor in the chiller
failed. This stopped the flow of chilling fluid to the compressor heat rejection interface.
Restart Attempts
An attempt to start the cooler was made on 28 December 1998. The plan was to start the
cooler, clean the gas (using the installed getter), and continue with the endurance evaluation. The
compressor started normally, but the turboexpander would not lift up and spin (the expander is
not driven, it is a free spinning device). During this event, several other attempts were made to
restart the cooler, but none were successful. During the restart attempts, a failed power supply in
the compressor speed proble signal conditioning circuit was discovered. The circuit board was
repaired and a new capacitance probe power supply was built. Upon installation of the new
components, two more unsuccessful attempts were made to start the SSRB.
Mr. Frank Dolan from Creare came to AFRL on 2-4 March 1999 to assist in troubleshooting
the turboexpander (TE) and attempt to restart the cooler without dismantling it. During his visit,
the cryocooler was heated using the internal load heaters. This adequately heated the expander
bearings closest to the load. The compressor was on during this time, but no spin was detected
from the TE. The cryocooler was then evacuated for an extended time, and then refilled with dry
neon. The compressor was engaged, but again, no spin was detected from the TE. The vacuum
chamber bell jar was then removed, and the TE assembly was heated with a heat gun to ~340K.
The compressor was on during this process, but no spin was detected from the TE. As a result,
the decision was made to take the TE apart and and inspect it to find out why it would not lift up
and spin.
Turboexpander Disassembly at Creare

Frank Dolan and John White traveled to AFRL on 7 Jun 1999 to remove the TE from the
SSRB and pack it up for return to Creare. The gas was purged from the system and the multi-
layer insulation was removed just enough to expose the TE. The TE was removed and all ports
were covered with clean room tape to keep the system relatively clean.
The TE was returned to Creare, and systematically dismantled in Creare’s class 10,000 clean
room. AFRL personnel were present to observe and assist with the analysis. Components of
interest were examined with optical microscopes and photographed.
There were several discrepancies noted during the disassembly and inspection. Debris was
found in several components, and scratches were found on the surface of the brake inlet and on
the TE shaft. Some of the scratches appear to be from the balancing process, while some one of
the scratches (seen on the area that would have been covered by the labyrinth seal) looked like
rubbing might have caused it. An inspection of the labyrinth seal revealed burrs and sharp edges
that had not been smoothed, which may have caused the scratch on the shaft.
Glass beads were also found lodged between the turbine blades at the tip of the shaft. The
glass beads are from bead blasting that was done after the shaft was Tiodized. Beads were found
in 11 of the shaft passages, and one passage had 3 beads in it. The beads were only found on the
exhaust side. No beads were found in the brake channels. Creare admitted that the shaft was not
inspected upon its return from Tiodizing.
When the TE was lifted manually, it would not drop back down under its own weight. Evem
with separate bearings, the shaft wouldn’t drop under its own weight, but would drop after being
lightly pushed manually.
The results of the disassembly and inspection of the TE suggest that the reason the TE could
not lift up and spin is that the shaft was rubbing somewhere inside the housing wall, and the
resulting friction made it impossible to lift. One hypothesis is that the shaft was balanced before
it was sent to be Tiodized, but after being Tiodized, bead blasted and having glass beads lodged
in the passages between the turbine blades, it became unbalanced. Thus, upon its return to
Creare, it had to be re-balanced, which caused the numerous scratches seen on the shaft. Re-
balancing also caused a high spot on the shaft, which decreased the clearance between the shaft
and the housing wall. If any of the glass beads fell out during operation of the cooler, the shaft
would no longer be balanced. This might cause it to lean in one direction and rub against the
housing wall.
LESSONS LEARNED
In order to prevent another cryocooler temperature excursion, two safety measures were
implemented at AFRL. First, the AFRL technicians replaced the faulty D/A converter, and
rewired the system so that in the event of a similar D/A converter failure, only one cold plate
valve per chamber would close. The other one would continue to function normally and remove
heat from the cooler.
The second safety measure implemented was the addition of case temperature inputs to the
ISACC 24-hour notification system. If the case temperature of a cryocooler exceeds a certain
value, an alarm signal will be sent, and the ISACC system will notify AFRL personnel of the
situation.
These two experiences have also emphasized the importance of considering maufacturing
processes when designing cryocoolers. TRW now stakes all capacitance sensor target locking
nuts. TRW recognized this as a potential problem prior to and during the development of the
3503, but the government decided! not to open up the cooler and stake the component during the
original build. In addition, all coolers with the same heritage as the 3503 were already staked.
Since building the SSRB, Creare procured a class 10,000 clean room as well as high powered
optical microscopes for inspecting components before they are integrated with the rest of the
cryocooler, as well as refined assembly procedures for flight cryocoolers.
The obvious robustness of these technologies is apparent even through significant overheats
and foreign particle contamination. The TRW 3503 would have continued running, at a degraded
level of performance, if the capacitance sensor locking nut had been staked. The Creare SSRB
was an engineering model non-hermetic cooler and was not designed for long life evaluation and
had the Creare SSRB had the level of cleaning now evident in current generation processes, the
SSRB would still be continuing in its endurance evaluation. However, these post-endurance
evaluations have provided cryocooler developers and users with significan insights into the
robustness and the potential for long life of cryocooler technology.
REFERENCES
1. Correspondence with William Burt, TRW, April 1999 to June 1999.
2. Dolan, Frank. “Trip Report – Visit to AFRL to Restart the Creare SSRB Engineering Model
Turboexpander”, Creare Incorporated Memorandum (1999).
3. Correspondence with Frank Dolan, Creare Inc., December 1998 to June 1999.
Cryogenic Material Properties Database
E.D. Marquardt, J.P. Le, and Ray Radebaugh

Boulder, CO 80303
ABSTRACT
NIST has published at least two references compiling cryogenic material properties. These
include the Handbook on Materials for Superconducting Machinery and the LNG Materials &
Fluids. Neither has been updated since 1977 and are currently out of print. While there is a great
deal of published data on cryogenic material properties, it is often difficult to find and not in a
form that is convenient to use. We have begun a new program to collect, compile, and correlate
property information for materials used in cryogenics. The initial phase of this program has
focused on picking simple models to use for thermal conductivity, thermal expansion, and
specific heat. We have broken down the temperature scale into four ranges: a) less than 4 K, b) 4
K to77 K, c) 77 K to 300 K, and d) 300 K to the melting point. Initial materials that we have
compiled include oxygen free copper, 6061-T6 aluminum, G-10 fiberglass epoxy, 718 Inconel,
Kevlar, niobium titanium (NbTi), beryllium copper, polyamide (nylon), polyimide, 304 stainless
steel, Teflon, and Ti-6Al-4V titanium alloy. Correlations are given for each material and
property over some of the temperature range. We will continue to add new materials and increase
the temperature range. We hope to offer these material properties as subroutines that can be
called from your own code or from within commercial software packages. We will also identify
where new measurements need to be made to give complete property prediction from 50 mK to
the melting point.
INTRODUCTION
The explosive growth of cryogenics in the early 50’s led to much interest in material
properties at low temperatures. Important fundamental theory and measurements of low
temperature material properties were performed in the 50’s, 60’s, and 70’s. The results of this
large amount of work has become fragmented and dispersed in many different publications, most
of which are out of print and difficult to find. Old time engineers often have a file filled with old
graphs; young engineers often don’t know how to find this information. Since most of the work
was performed before the desktop computer became available, when data can be found, it is
published in simple tables or graphically, making the information difficult to accurately
determine and use.
NIST has begun a program to gather cryogenic material property data and make it available
in a form that is useful to engineers. Initially we tried to use models based upon fundamental
physics but it soon became apparent that the models could not accurately predict properties over

682 CRYOCOOLER INTEGRATION TECHNOLOGIES AND MATERIALS
a large temperature range and over different materials. Our current approach is to choose a few
simple types of equations such as polynomial or logarithmic polynomials and determine the
coefficients of different materials and properties. This will allow engineers to use the equations
to predict material properties in a variety of ways including commercial software packages or
their own code. Integrated and average values can easily be determined from the equations.
These equations are not meant to provide any physical insight into the property or to provide
‘standard’ values but are for working engineers that require accurate values.
MATERIALS
Initial materials that we have compiled include oxygen free copper, 6061-T6 aluminum, G-
10CR fiberglass epoxy, 718 Inconel, Kevlar 49, niobium titanium (NbTi), beryllium copper,
polyamide (nylon), polyimide, 304 stainless steel, Teflon, and Ti-6A1-4V titanium alloy. These
were chosen as some of the most common materials used in cryogenic systems in a variety of
fields.
MATERIAL PROPERTIES
Thermal Conductivity
Widely divergent values of thermal conductivity for the same material are often reported in
the literature. For comparatively pure materials (like copper), the differences are due mainly to
slight material differences that have large effects on transport properties, such as thermal
conductivity, at cryogenic temperatures. At 10 K, the thermal conductivity of commercial
oxygen free copper for two samples can be different by more then a factor of 20 while the same
samples at room temperature would be within 4%. It is also not uncommon for some
experimental results to have uncertainties as high as 50%. Part of our program is to critically
evaluate the literature to determine the best property values. Data references used to generate
predictive equations will be reported.
The general form of the equation for thermal conductivity, k, is
where a, b, c, d, e, f, g, h, and i are the fitted coefficients, and T is the temperature. These are
common logarithms. While this may seem like an excessive number of terms to use, it was
determined that in order to fit the data over the large temperature range, we required a large
number of terms. It should also be noted that all the digits provided for the coefficients should be
used, any truncation can lead to significant errors. Tables 1A and 1B show the coefficients for a
variety of metals and non-metals. Equation 2 is the thermal conductivity for an average sample
of oxygen free copper. It should be noted that thermal conductivity for oxygen free copper can
CRYOGENIC MATERIAL PROPERTIES DATABASE 683
vary widely depending upon the residual resistivity ratio, RRR, and this equation should be used
with caution. The thermal conductivities are displayed graphically in Figure 1.
Specific Heat
The specific heat is the amount of heat energy per unit mass required to cause a unit increase
in the temperature of a material, the ratio of the change in energy to the change in temperature.
Specific heats are strong functions of temperature, especially below 200 K. Models for specific
heat began in the 1871 with Boltzmann and were further refined by Einstein and Debye in the
early part of the 20th century. While there are many variations of these first models, they
generally only provide accurate results for materials with perfect crystal lattice structures. The
Figure 1. Thermal conductivity of various materials.

specific heat of many of the engineering materials of interest here is not described well by these
simple models. The general form of the equation is the same as Equation 1. Table 2 shows the
coefficients for the specific heat. Figure 2 graphically shows the specific heats.
Thermal Expansion
From an atomic perspective, thermal expansion is caused by an increase in the average
distance between the atoms. This results from the asymmetric curvature of the potential energy
versus interatomic distance. The anisotropy results from the differences in the coulomb attraction
and the interatomic repulsive forces.
Different metals and alloys with different heat treatments, grain sizes, or rolling directions
introduce only small differences in thermal expansion. Thus, a generalization can be made that
literature values for thermal expansion are probably good for a like material to within 5%. This is
because the thermal expansion depends explicitly on the nature of the atomic bond, and only
those changes that alter a large number of the bonds can affect its value. In general, large
Figure 2. Specific heat of various materials.

changes in composition (10 to 20%) are necessary to produce significant changes in the thermal
expansion (~5%), and different heat treatments or conditions do not produce significant changes
unless phase changes are involved.8
Most of the literature reports the integrated linear thermal expansion as a percent change in
length from some original length generally measured at 293 K,
Where is the length at some temperature T and is the length at 293 K. While this is a
practical way of measuring thermal expansion, the more fundamental property is the coefficient
of linear thermal expansion,
The coefficient of linear thermal expansion is much less reported in the literature. In
principal, we can simply take the derivative of the integrated linear thermal expansion that
results in the coefficient of linear thermal expansion. While we have had success with this
method over limited temperature ranges, we have not yet determined an equation form for the
integrated expansion value that results in a good approximation of coefficient of linear thermal
expansion. For the time being, we will report the integrated linear thermal expansion as a change
in length and provide the coefficient of linear thermal expansion when it is directly reported in
the literature. The general form of the equation for integrated linear thermal expansion is
Tables 3A and 3B provide the coefficients for the various materials while Figure 3 plots the
integrated linear thermal expansions.
FUTURE PLANS
We plan to continually add new materials, properties, and to expand the useful temperature
range of the predictive equations for engineering use. We will report results in the literature but
will also update our website on a continual basis. The initial phase of the program was a learning
Figure 3. Integrated linear thermal expansion of various materials.
experience on how to handle the information in the literature as well as for the development of a
standard set of basic equation types used to fit experimental data. By using just a few types of
equations, we hope to make the information easier to use. We shall now focus on developing
large numbers of equations for a variety of materials and properties. Please check our web site at
http://www.boulder.nist.gov/div838/cryogenics.html for updated information.
REFERENCES
1. Berman, R., Foster, E.L., and Rosenberg, H.M., "The Thermal Conductivity of Some
Technical Materials at Low Temperature." Britain Journal of Applied Physics, 1955. 6: p.
181-182.
2. Child, G., Ericks, L.J., and Powell, R.L., Thermal Conductivity of Solids at Room
Temperatures and Below. 1973, National Bureau of Standards: Boulder, CO.
3. Corruccini, R.J. and Gniewek, J.J., Thermal Expansion of Technical Solids at Low
Temperatures. 1961, National Bureau of Standards: Boulder, CO.
4. Cryogenic Division, Handbook on Materials for Superconducting Machinery. Mechanical,
thermal, electrical and magnetic properties of structure materials. 1974, National Bureau of
Standards: Boulder, CO.
5. Cryogenic Division, LNG Materials and Fluids. 1977, National Bureau of Standards:
Boulder, CO.
6. Johnson, V.J., WADD Technical Report. Part II: Properties of Solids. A Compendium of
The Properties of Materials at Low Temperature (phase I). 1960, National Bureau of
Standard: Boulder, CO.
7. Powells, R.W., Schawartz, D., and Johnston, H.L., The Thermal Conductivity of Metals and
Alloys at Low Temperature. 1951, Ohio State University.
8. Reed, R.P. and Clark, A.F., Materials at Low Temperature. 1983, Boulder, CO: American
Society for Metals.
9. Rule, D.L., Smith, D.L., and Sparks, L.L., Thermal Conductivity of a Polyimide Film
Between 4.2 and 300K, With and Without Alumina Particles as Filler. 1990, National
Institute of Standards and Technology: Boulder, CO.
10. Simon, N.J., Drexter, E.S., and Reed, R.P., Properties of Copper Alloys at Cryogenic
Temperature. 1992, National Institute of Standards and Technology: Boulder, CO.
11. Touloukian, Y.S., Recommended Values of The Thermophysical Properties of Eight Alloys,
Major Constituents and Their Oxides. 1965, Purdue University.
12. Veres, H.M., Thermal Properties Database for Materials at Cryogenic Temperatures. Vol. 1.
13. Ziegler, W.T., Mullins, J.C., and Hwa, S.C.O., "Specific Heat and Thermal Conductivity of
Four Commercial Titanium Alloys from 20-300K," Advances in Cryogenic Engineering
Vol. 8, pp. 268-277.
Experimental Results on the Thermal Contact
Conductance of Ag-Filled Epoxied Junctions at
Cryogenic Temperatures
Z. Wang, A. Devpura, and P.E. Phelan

Arizona State University, Mechanical & Aerospace Engineering
Tempe, AZ 85287-6106 USA
ABSTRACT
The thermal contact conductance across an epoxied copper junction loaded with Ag (silver)
particles was investigated at cryogenic temperatures. Thermal contact conductance, or its
inverse, thermal contact resistance, consists of two components: thermal contact resistance at the
copper/Ag-particle epoxy interfaces, and the thermal conduction resistance across the Ag-
particle epoxy slab. The effects of the Ag-particle volume fraction, and the average interface
temperature of the epoxied junction are both evaluated. Increasing the Ag-particle fraction
increases the conductance above that for a plain epoxied sample, by as much as one order of
magnitude for a Ag particle fraction of 30%. A critical Ag particle volume fraction is observed
in the measurements, below which value the thermal conductance of the epoxied junction
increases only slightly with increasing particle fraction.
INTRODUCTION
When heat flows across an interface between two solids in contact, a relatively high thermal
resistance is encountered because of the imperfect junction. The imperfect junction is caused by
the roughness of the surfaces on a microscopic scale, such that the real contact surface of the
interface is only a few of the scattered contact spots. Usually, the ratio of the real contact area to
the nominal contact area is very small. Thermal contact conductance is defined as
where q is the heat flux across the interface, the interfacial temperature
drop, and the thermal contact resistance (TCR). In many cases, the energy transfer across the
interface is of concern. It is often desired to enhance or isolate the heat transfer at the junction.
Therefore, the prediction of is very important.1
The thermal conductance between contacting solids varies considerably, depending upon the
thermal and physical characteristics of the materials and the conditions of the junction. In order
to control we usually consider several parameters as controlling variables, such as the
mechanical and thermophysical properties of the materials composing the junction, the surface
characteristics of the contacting surfaces, the mechanics of the contact, and the interstitial
materials. In the real world, due to the fixed structure of the entire system, relatively speaking it

is easy to change the interstitial material. The interstitial materials’ thermal conductivity,
hardness and thickness, and the values of the corresponding properties of the base materials,
dictate the final value in thermal contact conductance
Fletcher2 defined the efficiency as the ratio of the logarithms of the conductances with and
without the interstitial material,
and the nondimensionalized conductance, which takes the thickness of the interstitial material
into account,
where t is the thickness of the interstitial layer and is the equivalent gap thickness. The
subscript cm refers to the control material and bj represents the bare junction. These two
parameters compare the effectiveness of the insert, which is used for the control of thermal
conductance. However, they cannot predict from the operating conditions and the material
properties of the interstitial layer. Since there are only limited data available for either for
bare junctions or with an interstitial material, especially at cryogenic temperatures, a large
number of experiments must be performed in order to determine and for different interstitial
materials.
Here we investigate one particular category of interstitial materials, Ag-particle-filled
Stycast epoxy (Type 2850-FT, Catalyst 9), which may be used at cryogenic temperatures where
high values are required. The junction is formed by two copper pieces in a flat-plate geometry.
EXPERIMENT
Figure 1 presents a schematic of the test column, which is utilized here to investigate an Ag-
filled epoxied junction, but was previously used to investigate for other materials in contact at
cryogenic temperatures.3,4,5 It consists of the test samples, and two calibrated heat flux meters
which are pressed together and aligned to the centerline of the apparatus. The alignment can be
adjusted by the point contacts of the ball bearings. The test specimens are made of two solid
copper cylinders, epoxied together with Ag-particle-filled Stycast cryogenic epoxy.
The thermal conductivity of the heat flux meters was previously calibrated against a
reference 304SS bar obtained from the National Institute of Standards and Technology. The
temperature drops along the heat flux meters and across the sample interface are measured by
differential Type E thermocouples. The differential thermocouples are calibrated at a single point
by immersing one junction into liquid nitrogen and the other into an ice bath, and applying the
resulting percent deviation from the standard table to all measurements. The silicon diodes
mounted on the copper heat mounts are utilized as the reference temperatures for the differential
thermocouples. The thermocouples are inserted into diameter holes drilled to the center
lines of the samples and heat flux meters, and which are partially filled with fine copper powder
in order to provide reliable thermal contact.
Indium foil is inserted at all junctions besides the sample interface in order to reduce
extraneous contact resistance. Two thermal shields are placed above and below the heat flux
meters in order to reduce the radiation heat transfer. The entire test column is surrounded by a
thermal shield to minimize heat gain from the ambient.
The outputs of the thermocouples, silicon diodes, and the load cell amplifier are monitored
and recorded by a Macintosh Power PC computer, then simultaneously converted and analyzed
by a Labview control program. If the temperature drops across the sample interface as well as
those along the upper and lower heat flux meters are all within a set criterion for the last 100 data
points, the system is considered to be in steady state. Starting with the reference silicon diodes
located on the copper heat mounts, each temperature point is calculated sequentially from the
THERMAL CONTACT CONDUCTANCE OF Ag-FILLED EPOXY 691
Figure 1. Schematic diagram of the experimental apparatus.
differential thermocouple readings. The temperature drop at the sample interface is over-
determined by the calculation downwards and upwards, which are averaged to determine the
final The heat flux at the sample interface is considered as the average of the upper and
lower heat flux meters.
All measurements are conducted under a vacuum condition of torr on average,
which is measured by a cold cathode vacuum gauge.
EXPERIMENTAL UNCERTAINTY
The total experimental uncertainty of the thermal contact conductance is from the
uncertainties in q and respectively. Since the heat flux q is the average of the measurements
of the upper and lower heat flux meters, the uncertainty of is calculated from:6
where and are the uncertainties in q from the upper and the lower heat flux
meters, which are calculated from:
where and are the uncertainties in the thermal conductivity K of the heat flux meters
and the thermocouple readings, and L and the length and the temperature drop along the
heat flux meters.
The total uncertainty in thermal contact conductance, is calculated from:
where and are the uncertainties in q and The total experimental uncertainty can be
divided into two parts: bias errors caused by instrument specifications and precision errors (95%
confidence level) caused by scatter in the data. For each different error, the final value is
calculated from the root sum square of these two parts accordingly.
Under cryogenic temperatures, the radiative heat gain from the ambient to the test samples is
negligible. The maximum radiative heat gain was estimated to be less than 5 mW, while the
normal heat flow along the axial direction of the sample is about 2.5 ~ 3.0 W.
Both junctions of the differential thermocouple positioned across the sample interface are
mounted in holes located 2.4 mm from the interface. The resulting temperature drop across the
copper is estimated as 1.4% of the across the interface and is considered to be negligible.
SAMPLE PREPARATION AND CHARACTERISTICS

The purpose of this investigation is to determine how of filled epoxied junctions varies
with the particle volume fraction at cryogenic temperatures. The temperature ranges from 40 to
230 K. At these low temperatures, conventional epoxies are prone to cracking due to thermal
expansion mismatch. Hence, our experiments are performed with an epoxy that is designed
specially for cryogenic use. By analogy with the percolation behavior of electrically conductive
composite materials,7 we know that the larger the difference between the thermal conductivities
of the filler and the matrix materials, the more liable we are to observe the percolation
phenomenon in the experiments. Pure silver has one of the highest thermal conductivities among
all the metals, and it is very stable under typical experimental conditions. Therefore, the metal
basis silver powder with 3N purity, purchased from Alfa Aesar, was chosen as the filler material.
Basically, the particles are not spherical, as they have irregular shapes, but Alfa Aesar provides a
size parameter to describe the particles. The size ranges from For the epoxy, there are
several suitable cryogenic candidates. By considering, however, that the particle distribution of
the samples will be examined by an optical microscope, we select Stycast epoxy, having a dark
color, as the matrix material. This provides good contrast, and makes it easy to distinguish the
white Ag filler particles from the black epoxy background.
The surface treatment method and roughnesses for the copper samples are tabulated in Table
1, where the surface parameters (roughness) and m (mean slope) are defined elsewhere.3 Note
that all surface measurements were performed at room temperature. The surface conditions of all
samples are similar to one other.

A certain amount of silver powder and Stycast epoxy are blended together, according to the
desired Ag volume fraction Since the densities of both materials are available, can be
controlled by controlling the mass of silver powder mixed in with the epoxy according to
where and are the masses of Ag particles and of the epoxy, and the densities
of Ag and of the epoxy, and the mass of epoxy removed before mixing, in order to arrive
at the desired
The thickness of the epoxy mixture layer is constricted to the thickness of a filler gauge (250
After the experiments, the actual thickness was examined under an SEM (Scanning
Electrical Microscope). Table 2 shows the measured thicknesses and slopes, i.e., deviations from
a perfectly parallel interface where the slope would be zero degrees, of the four samples.
In order to observe the distribution of Ag particles in the epoxy, we examine a cross section
of the epoxy under an optical microscope, with 100X magnification. The image is subsequently
digitized for analysis, and the minimum and maximum of the x, y coordinates are set as 0.0 and
1.0. Figure 2 shows a local image of the particle distribution, for the sample containing 25% Ag
particles.
It was desired to gauge the randomness of the Ag particle distribution. Accordingly, Chi-
Square tests were performed separately on the x and y position coordinates:8
Figure 2. A partial image of the cross-section for the 25% Ag fraction sample under an optical
microscope (original image with 100X magnification).
Figure 3. Comparison of among a pure epoxy junction and four Ag-filled epoxied junctions.
where and are the observed and expected values in the ith interval, the population
variance, and k the number of intervals. For the test sample having 25% volume fraction of Ag
particles, the results show that the Ag particles are randomly distributed over the cross-section.
The correlation between the x and y coordinates, r, was also found to be close to zero (less than
0.05), so that the following conclusion can be made:9 the Ag-particles are uniformly distributed
in the epoxy, with a confidence level of 95%. Furthermore, since the cross-section examined
was chosen randomly, we conclude that the distribution of particles is random throughout each
of the epoxied junctions.

Four test specimens were investigated for the of Ag-filled epoxied junctions. As was
shown in Table 2, the volume fractions of Ag particles were approximately 10%, 20%, 25% and
30%, respectively. Separate experiments were conducted for these four different Ag volume
fractions. Great effort was made to maintain the consistency of all experimental conditions for all
the samples during the measurements, so the difference among the volume fractions can be
counted as the only variable to explain the different results.
Figure 3 shows the experimental of the four samples compared with that of a pure epoxy
junction sample.10 The thickness of the pure epoxy junction is close to the average
thickness of the Ag-filled ones, and other experimental conditions are the same, so the difference
in is primarily due to the Ag particles. AH four different Ag fraction curves show clearly a
trend that increases with an increase in the average junction temperature. The rate of increase
is larger in the low-temperature range, while in the high-temperature range, the curve becomes
more flat. Around 50 K, there is a big leap in thermal contact conductance. For is
very close to that for the pure epoxy data. There is a trend, though still not obvious, that the
data are a little higher than those of the pure epoxy after a pivot temperature (150 K).
Figure 4. Effect of varying the Ag-particle fraction on (junction temperatures near 100 K, 85 K,
and 70 K).
The curve shows that the two sets of data are still close to each other, but the pivot
temperature is lower (110 K). At low temperatures (below 70 K), the and pure epoxy
curves are close, however, above 70 K there is a pronounced increase in of the
sample over that of the pure epoxy. For is greater than that for pure epoxy over the
entire temperature range. The difference between the Ag-filled and pure epoxy increases with
increasing while the pivot temperatures fall with increasing The existence of the Ag-
particle filler in the Stycast epoxy apparently enhances as expected, with the enhancement
being the most dramatic for
The error bars of the measurements, as displayed in Fig. 3, are generally small, except in the
low-temperature range (less than 50 K). As the average temperature at the interface rises, the
error decreases rapidly.
Figure 4 shows the effect of varying the Ag-particle volume fraction on at several values
of temperature (100 K, 80 K, and 70 K). The participation of the Ag-particles in the junction
changes differently according to the magnitude of For the first three samples
20%, and 25%), there is only a small increase in However, (30%) for all
temperatures is special in the figures: there is a trend to rapidly increase The of the
junctions includes two parts, the thermal conductance of the epoxy composite,
and the two thermal boundary resistances at the interfaces, which add to determine the
overall
where is the effective thermal conductivity of the epoxy composite, and is assumed to
have the same value at each epoxy/copper interface.
The value of is determined, in part, by the samples’ surface conditions. Table 1 shows
that the surface roughnesses of all samples are relatively close to one other. We therefore assume
that remains unchanged from sample to sample, leaving as the only factor that influences
Here, we estimate from the value determined from two pure Stycast epoxied samples,
giving at
The values of effective thermal conductivity can now be extracted from the measured
using Eq. (8), and the resulting values are given in Table 3.
Two approaches that provide upper and lower bounds to where is the thermal
conductivity of the epoxy matrix, are to (i) assume the matrix and filler lie in series (lower
bound), and (ii) assume the matrix and filler lie in parallel (upper bound). The corresponding
relation for the series approximation is:11
and that for the parallel approximation is:11
For the values of calculated from Eqs. (9) and (10) are shown in Table 3,
together with the experimental results. All the experimental data lie between the upper and lower
bounds. For the first three fractions (below 30%), the parallel approximation is much higher than
the experimental data, while the series approach makes a better estimation. However, for the last
fraction (near 30%), the experimental data are substantially larger than the lower bound, and
draw closer to the upper bound. The parallel approximation can be interpreted as the extreme
percolation case, in which all the Ag particles are in contact and form a path through the slab,
and there is no thermal contact resistance between the Ag particles. In reality, however, the
thermal contact resistance between the Ag particles is non-negligible, and not all the Ag particles
come together to form a high-conductivity path across the slab. Therefore, the measured value of
is lower than that given from the parallel approximation.
CONCLUSIONS
The thermal contact conductance for four different Ag-particle volume fractions of filled
epoxied junctions is investigated at cryogenic temperatures, where the average interface
temperature ranges from 40 K to 210 K. Thermal contact conductance increases as the junction
temperature increases, with the increase in the low-temperature range being more significant.
The results indicate that increasing the Ag-particle volume fraction increases in accord with
expectations, for the greater the number of Ag particles, the greater the possibility that thermal
shortcuts of Ag particles are formed. The presence of the Ag particles enhances beyond that
measured for a plain epoxied junction, with the enhancement for the curve being as
much as one order of magnitude.
ACKNOWLEDGMENTS
All the authors wish to thank Lisa De Bellis for her assistance in the experiment. P.E.P.
gratefully acknowledges the support of the National Science Foundation through a CAREER
Award (Grant No. CTS-9696003), and matching funds provided by Raytheon.
REFERENCES
1. Madhusudana, C.V., Thermal Contact Conductance, Springer, New York (1996), pp. 77-99.
2. Fletcher, L.S., “Recent Developments in Contact Conductance Heat Transfer,” J. Heat Transfer, vol.
110 (1988), pp. 1059-1069.
3. Phelan, P.E., & Mei, S., “Experimental Results on the Thermal Contact Resistance of G-10CR
Composites at Cryogenic Temperatures,” Paper No. AJTE99/6156, Proc. of the 5th ASME/JSME
Joint Thermal Engineering Conference, San Diego, CA, March 15-19 (1999).
4. Zhao, L., & Phelan, P.E., “Thermal Contact Conductance Across Filled Polyimide Films at
Cryogenic Temperatures,” Cryogenics, vol. 39 (1999), pp. 803-809.
5. De Bellis, L., Phelan, P.E., Drake, P., & Kroebig, W., “Measurement of the Thermal Properties of
Epoxied Titanium Contacts at Cryogenic Temperatures,” to appear in Advances in Cryogenic
Engineering, vol. 46 (2000).
6. Holman, J.P., Experimental Methods for Engineers, McGraw-Hill, Inc., New York (1994), p. 66.
7. Joy, T., “Percolation in a Thin Ply of Unidirectional Composite,” J. Composite Materials, vol. 13,
(1979), pp. 72-78.
8. Banks, J., & Carson, J.S., Discrete-Event System Simulation, 2nd Ed., Prentice-Hall of India, New
Delhi (1998), p. 303.
9. Swartz, C.E., Used Math, State University of NY at Stony Brook, New York (1973), p. 100.
10. De Bellis, L., & Phelan, P.E., “Measurement and Prediction of the Contact Conductance Across
Epoxied Copper Contacts at Cryogenic Temperatures,” to be presented at the ASME International
Mechanical Engineering Congress & Exposition, Orlando, Florida, November 5-10 (2000).
11. P. E. Phelan, P.E., & Niemann, R.C., “Effective Thermal Conductivity of a Thin, Randomly
Oriented Composite Material,” J. Heat Transfer, vol. 120 (1998), pp. 971-976.
A Fail-Safe Experiment Stand for Cryocooler
Characterization
C. H. Yoneshige1, N. S. Abhyankar2, J.P. Kallman1, G. W. Lybarger1

and M. L. Martin2
1
Space Vehicles Directorate, Air Force Research Laboratory
2
Dynacs Engineering Co
ABSTRACT
Understanding possible failure mechanisms is the first step in the prevention and impact
minimization of experiment stand failures in any research facility. As the facility encompasses a
wider range of multiplicity of experiments, these mechanisms can become coupled, requiring
careful planning and preventative measures. This paper will discuss the possible experiment
stand failure modes and safeguards used to minimize the impact that experiment stand failures
have on the cryocooler characterization process at the Cryogenic Cooling Research Facility at
the Air Force Research Laboratory (AFRL). The facility is equipped with one 6 foot, two 18
inch, two 36 inch and eight 24 inch thermal vacuum chambers along with all of the equipment,
specialized tools and instruments required for the short term performance characterization and
long term endurance evaluation of cryogenic coolers. The laboratory characterizes cryocoolers
based on various thermodynamic cycles, such as the Stirling cycle, pulse tube (Stirling cycle
variant) and reverse Brayton cycle, at various levels of technical maturity. These cryocoolers
were developed to satisfy space mission requirements from the United States Air Force (SBIRS-
Low Program Office), The Ballistic Missile Defense Organization (BMDO), the National
Aeronautics and Space Administration (NASA) and the Department of Defense.
The prevention of experiment stand failures related to cryocooler characterization will be
discussed from different standpoints, including the vacuum system, equipment interfaces,
electronics, software programs, communication systems, and electronic remote notification
systems.
INTRODUCTION
Failures may occur on the one component that was assumed to be “non critical” and was
therefore not addressed in the fail-safe set up of the experiment stand. Most potential component
failures are accounted for during routine maintenance. However, every so often, a component
fails and produces unexpected results. This leads to the reevaluation of the experiment stand set-
up and the revision of the routine maintenance list. The experiment stand fail-safe set up is

critical for 24-hour laboratory operations encompassing multiple experiments with extensive
instrumentation.
The Air Force Research Laboratory Cryogenic Cooling Research Facility (CCRF)
characterizes unique, space qualifiable cryocoolers for thermodynamic and long term (~10 years)
performance in a space like environment. The CCRF was built and maintained under joint
sponsorship by the Ballistic Missile Defense Organization (BMDO) and the United States Air
Force (USAF). The facility is equipped with two 18”, eight 24”, two 36”, and one 6’ thermal
vacuum chambers. Most cryocoolers are characterized in the thermal vacuum chambers.
However, coolers that are not space qualifiable, but have heritage to space cryocooler designs are
tested on table-top experiment stands with a vacuum bonnet covering the cold end only.
Cryocooler development is supported by BMDO, AFRL, the SBIRS-Low program office
and NASA to meet specific mission requirements. The lab has characterized cryocoolers from
TRW, Raytheon, Ball Aerospace, Creare, Lockheed Martin, Matra Marconi Space, and other
agencies. These cryocoolers are based on different thermodynamic cycles, such as the Stirling
cycle, pulse tube (Stirling cycle variant), and reverse-Brayton cycle. Each cryocooler’s
thermodynamic performance envelope is established. Then it is put into an endurance evaluation
and run until it meets predetermined failure criteria.
Cryocooler characterization involves a series of functional experiments, such as load line
trials, temperature stability trials, transient thermal response trials, cool-down curves, and
parasitic heat load determination, which explore the cryocooler’s thermodynamic
capability/capacity as well as its compatibility with its proposed spacecraft, its mission and its
subsystems. A complete performance envelope is determined for each cryocooler, providing a
range of operation points which can be used by space mission planners to find existing
cryocoolers suited for their specific mission requirements. The endurance evaluation monitors
the cryocooler’s long-term behavior and helps to build confidence in new cryocooler
technologies. It also provides insight into failure mechanisms. All of the data collected provides
feedback to cryocooler researchers and manufacturers regarding design improvements based on
lessons learned.
A typical experiment stand for cryocooler characterization consists of the following
subsystems:
- Vacuum System
- Temperature Rejection System and Interfaces
- Electronics
- Software
All of these subsystems interact while the experiment is running. Failure modes for each
subsystem must be addressed both independently and collectively during cryocooler integration.
This paper addresses the identification of critical system components and the implementation of
safety features to protect experimental hardware in case of a component failure.
IDENTIFICATION OF CRITICAL COMPONENTS AND FAIL-SAFE FEATURES
A typical experiment stand used for cryocooler characterization is shown in Figure 1. This
experiment stand is set up for a split-Stirling cryocooler and includes the heat rejection interface.
Cables, fluid lines, and thermocouple wires are passed through air-tight feedthroughs which are
located around the bottom of each vacuum chamber.
Remote Notification System
While a cryocooler is operational, critical operation parameters are continuously monitored

to make sure they stay within their specified limits. If the limit is crossed, an alarm condition
exists, and a signal is sent through a custom-built alarm interface box to the Intelligent System
for Automatic Control and Communication (ISACC) remote notification system. The ISACC
system can monitor equipment and environmental conditions using 16 analog or digital inputs. It
can also be programmed to switch up to 8 digital outputs and four analog outputs in any
A FAIL-SAFE STAND FOR CRYOCOOLER CHARACTERIZATION 701
Figure1. A typical cryocooler characterization experiment stand.
combination. If an alarm is detected, ISACC can dial up to 8 phone numbers and communicate in
voice or data. One technician or engineer is placed on-call each week and carries a pager, and is
the first person called in case of an alarm. If this person cannot be reached, ISACC is
programmed to call other lab personnel. Utilization of this system allows cryocooler
characterization experiments to be run unattended 24 hours a day, seven days a week. Specific
alarm conditions that have been set will be covered in the following sections.
Vacuum System
The vacuum system consists of a roughing (vane) pump, turbo pump, gate valve, and
instrumentation feedthroughs. In this section, only the mechanical aspects of the vacuum system
will be addressed. The electronics and instrumentation will be covered in a later section. Possible
component failures and their consequences are described, and their fail-safe solutions are
presented. Some of the possible component failures include the following:
Pump Failures. If the turbo pump fails, and the chamber pressure is lower than the
minimum pressure that the roughing pump can sustain, oil from the roughing pump can back
stream into the chamber and contaminate the surfaces. As a result of this risk, a system was set
up to close the gate valve in the event that the pressure in the vacuum chamber exceeds a set-
point value. A vacuum loss alarm was also set up and connected to the remote notification
system. If the roughing pump fails, the turbo pump will fail and cause an increase in the vacuum
pressure, again triggering the vacuum loss alarm.
Gate Valve Failure. The gate valve is pneumatically operated (positive pressure is required
for the gate valve to remain open). If the shop air compressor fails or AC power is lost, the gate
valve will close, protecting the vacuum inside the chamber.
Heat Rejection System and Interfaces
When a cryocooler is operating in orbit, the heat it generates is normally rejected to deep
space through a radiator. In the laboratory heat is rejected to a surface, which is kept at a constant
temperature by alternately running a refrigerant through a chiller loop and heating it with a
heater element. The optimal heat rejection temperature for each cryocooler is specified in its
design specifications. The fluid from the chiller flows through a spiral of copper pipes which lie
under the heat rejection surface. Heat is exchanged with the cryocooler by conduction at the heat
rejection surface and then by convection with the chiller fluid, which returns to a constant
temperature chiller reservoir. The fluid enters and exits the chamber through special feed-
throughs. The amount of fluid flowing through the system is determined by flow valves, which
are controlled manually or by an electronic controller.
If any part of the heat rejection system fails, the risk of a system overheat, thermal runaway,
or loss of vacuum exists. Operation at high temperatures can damage the cryocooler. Some of the
possible component failures include the following:
Chiller. The chiller is probably the most sensitive component of the heat rejection system. It
can be affected by commercial power outages and has many parts succeptible to failure. The
pump motor, refrigeration unit and plumbing lines are all possible failure mechanisms.
Precautionary measures were taken to ensure that any chiller component failures would not
cause damage to the cryocooler under evaluation. In case of a commercial power failure, the
chillers are hooked up to a back up power generator, which is activated within about 30 seconds
of the initial power loss. The chillers are programmed to restart once the generators are turned
on.
Alarms were also set up to notify lab personnel in case of any component failures. If the
pump motor fails, there is a direct interface to the ISACC system. If there is a leak in the fluid
lines, there is a low-flow alarm, which notifies lab personnel if fluid flow falls below a specified
limit. If the leak is inside the vacuum chamber, the chiller fluid will increase the pressure inside
the chamber and the vacuum loss alarm will activate.
Flow Valves. The flow valves are controlled by computer software which sets the amount
that the flow valves are open to maintain the desired flow rate. The valves are set to fail open,
which protects the test article from overheating.
Heating Units. Either serpentine heaters or cal rods heat the chiller fluid and maintain a
specified heat rejection temperature. The method used for heating the chiller fluid depends on the
requirements of each individual cryocooler or cryogenic technology experiment. The only
consequence of a heating element failure is the inability to run the experiment at a higher heat
rejection temperature. This poses no risk to a cryogenic experiment.
Electronics
There are four electronics subsystems used in the experiment stand. One subsystem monitors
and controls the environmental system, another consists of instrumentation used to collect data
from the experiment, another consists of power supplies use to run the cryocooler and its
electronics and a fourth is used to monitor the experiment stand and activate the remote
notification system. One last item – probably the most important – is the set of cables used to
connect all the electronic subsystems together.
Environmental. The environmental system electronics consists of controllers and
instrumentation used to maintain a vacuum inside each thermal vacuum chamber and control the
Figure 2. Vacuum Chamber Environmental Electronics.
heat rejection temperature. Figure 2 is a block diagram showing a typical vacuum chamber
environmental set-up.
The Tempscan monitors the environmental thermocouples. It is also used as a digital I/O
device which controls the vane pump and opens or closes the gate valve through relays, indicated
by a letter R in the figure. If the Tempscan fails, the data from the environmental thermocouples
is lost, subsequently the gate valve closes and the vane pump shuts off. There is no threat to the
test article if the data from the environmental thermocouples is lost. Closing the gate valve
protects the vacuum inside the chamber for a little while. When the vacuum pressure increases
enough, the vacuum loss alarm is triggered and lab personnel are notified.
The bus converter provides serial communication between the General Purpose Interface
Bus (GPIB) and the chiller, Varian and Omegas. The Varian reads the vacuum pressure and
updates the environmental software. The omegas control the flow valves. If the bus converter
fails, the chillers will continue to run, the Varian will lose the ability to update the vacuum
pressure reading on the software, and the omegas will keep running at the last input setting. None
of these consequences poses any threat to the experiment running in the chamber.
However, if one of the Omegas fails, the flow valves will close. A zero signal equates to
fully closed. This deprives the test article of the chiller fluid that keeps the heat rejection
temperature down. For this reason, thermocouples were placed on the warmest running parts of
the cryocooler compressor (and expander, if applicable) casing. The readings from these
thermocouples are fed into a case temperature alarm interface box. If the case temperature goes
above the alarm limit, the case temperature alarm activates and notifies lab personnel.
If the Varian ion gage fails to get emissions, an alarm is activated. If the turbo pump
controller fails, the turbo pump stops functioning, the vacuum pressure rises and the vacuum loss
alarm is activated.
Data Acquisition Instrumentation. The two major types of instrumentation used to

monitor cryocooler characterization are thermometry instruments and power meters.
Thermocouples measure the ambient temperature in the vacuum chamber, and are attached to
different parts of the cryocooler and test stand to provide a thermal map of the experiment.
Diodes and Platinum Resistance Thermometers (PRTs) measure the temperature of the cold head
(and each additional stage of a multi-stage cryocooler). Fluke Data Acquisition Units monitor the
thermocouples, and either Lakeshore or Conductus temperature controllers monitor the diodes
and PRTs.
Valhalla power meters monitor input power (in terms of current and voltage) to the
cryocooler. The drive power is wired through the Valhalla. If the Valhalla fails, the input power
continues to go through to the compressor. The only loss is the input power data. The same holds
true if a Fluke, Lakeshore or Conductus fails. The data losses pose no threat to the cryocooler.
All of the data acquisition instruments are on a calibration schedule to make sure that the
data is traceable to NIST standards. This calibration cycle also helps find problems with the
equipment and keep up on maintenance.
Power Supplies. Some cryocoolers require external power supplies. Others have their
power supplies inside their control electronics. If any of the power supplies fail, the cooler stops
running. If a cooler stops running or trips, an alarm is activated to alert lab personnel to the
situation.
Cables. Failures can also occur in cables at their connector junctions. The cables run from
the electronics boxes or instruments through the vacuum chamber feed-throughs and to their
respective components inside the chamber. The wires are individually soldered, and can come
loose if the cables are bent or moved on a regular basis. This can cause intermittent contact,
which is one of the most difficult problems to diagnose.
The best way to prevent cable failures is to be extremely careful when building the cables
and harnesses. Once the cables are built, they go through an extensive visual and mechanical
inspection and bench testing procedure.
Software
In terms of the experiment stand, the only software concerned is the environmental control
software and the data acquisition software. Both sets of software were written in-house using
Lab VIEW™ software. The environmental software monitors and controls the environmental
conditions in each thermal vacuum chamber. The environmental software is not used in every
vacuum chamber. In some cases, it is easier to control the vacuum system manually.
The data acquisition software monitors and records parameters, which include input power,
input voltage and input current and cryocooler cold end, body and heat rejection temperatures.
The external heat load is also set through this software. Some cryocoolers come with stroke
length monitoring devices, such as LVDT or capacitance sensors. Their responses in terms of
voltage are converted into stroke lengths in terms of millimeters. The conversion factors are
provided by the cryocooler manufacturers and verified by AFRL engineers. Figure 3 shows a
data acquisition front panel for Raytheon’s PSC, a protoflight spacecraft cryocooler. The graphs
are for endurance evaluation at its design point of 60 K at 2 watts of load.
The software programs, in general, monitor the system for normal operation and for outside
of its set limits. The operation limits can be set and adjusted from the software. Based on these
limits, alarm conditions are established. If an alarm condition is detected, the system is set up to
take data in smaller time intervals, giving lab personnel a better picture of what causes the alarm
to activate and the subsequent consequences.
Figure 3. Data Acquisition Front Panel.
CONCLUSIONS
In the course of developing the Cryogenic Cooling Research Facility, many lessons were
learned regarding the fail-safe set-up of experiment stands. In an effort to avoid over-alarming
the system and wasting time and resources, the consequences of each component failure were
considered and alarms were placed where they could look for those consequences. It turned out
that a few alarms could cover multiple component failures.
One of the greatest lessons learned is the importance of effective communication between
the engineers and technicians. With everyone clear on what direction each experiment was
taking, the proper fail-safe mechanisms were put into place, leading to systems that operated
longer and provided greater protection for experiments running unattended.
Development and Testing of a Gimbal
Thermal Transport System
D. Bugby, B. Marland, and C. Stouffer

Swales Aerospace
Beltsville, MD, USA 20705
B. Tomlinson, T. Davis
ABSTRACT
This paper presents development details of a two-phase thermal transport system for
carrying cryocooler waste heat across a 2-axis gimbal with low line-induced torque. Applications
for this system include space-based remote sensing spacecraft with gimbaled cryogenic optics
and/or infrared sensors. The described system will use standard loop heat pipe (LHP) technology
and small diameter stainless steel bellows tubing to transport 50-200 W of waste heat over a
distance of a few meters while still meeting strict gimbal flexibility requirements. The gimbal
will be motorized in each axis to study whether rapid gimbal movement affects thermal transport
system operation. Heaters will simulate cryocooler waste heat. Preliminary results from a
pressurized, multi-line, non-motorized gimbal mock-up indicated the line-induced torque could
be acceptably low if the routing is appropriately managed. The paper will describe the design and
development effort. Test results were not available at the time the paper was written.
INTRODUCTION
Cryocooled infrared payloads are being considered increasingly by the DoD and NASA for
space-based missions that require advanced imaging and tracking capabilities. In many instances,
these systems operate best when the optics/sensors can be independently slewed relative to the
spacecraft. Thus, a gimbaled optical/sensor system is the preferred approach, but the problem
becomes how to reject the heat generated by on-gimbal cryocoolers to space. Typically, on-
gimbal radiators cannot be used due to their poor radiative efficiency, large weight/volume, and
adverse impact on system performance. And, as cryogenic cooling loads increase in the future,
this thermal management problem will become even more severe. In addition, the problem of
transporting heat across an isolated interface is also a concern for infrared detectors and optical
benches that need to be structurally isolated. There is also a similar problem when attempting to
use a deployable/steerable radiator to reduce the size of a spacecraft launch vehicle and meet a
reduced envelope. The solution to this long-standing thermal problem is the subject of this paper.

The solution is to utilize appropriately routed stainless steel bellows tubing and an ambient
loop heat pipe (LHP) two-phase heat transport system to continuously move cryocooler waste
heat across the gimbal to a spacecraft-mounted radiator. Although typical LHP systems are
completely passive with no moving parts, the challenge is to manage the line routing so that the
gimbaled sensor/optical system can still meet its requirements for 2-axis motion, low torque, low
jitter, high reliability, and long lifetime. The principal technological uncertainty is whether
gimbal motion during LHP operation has any unknown dynamic and potentially deleterious
effect on the system's ability to meet mission requirements. Figure 1 illustrates the concept.
To meet the technical need outlined above, this paper presents the development details of an
AFRL funded program to design, fabricate, and test a two-phase thermal transport system for
carrying ambient waste heat across a 2-axis gimbal. This program, which is part of a broader
AFRL initiative known as CRYOBUS (see Bugby1), has been given the name GATTS (Gimbal
Ambient Thermal Transport System). The paper is organized as follows. First, technical
background information is provided on infrared sensor pointing methods, LHP technology
options, and flex line routing options. Next, the overall program is described including the
GATTS objectives, requirements/specifications, design calculations/trades, design approach, and
test plans. Lastly, the conclusions of the program and future plans (at the time this paper was
written in June 2000) are outlined and discussed.
BACKGROUND
The following presents background information on infrared sensor/optical system pointing
methods, LHP technology options, and flex line technology.
Pointing Methods
There are four possible approaches for pointing an infrared sensor/optical system at a target.
Arranged from least to most agile, the advantages/disadvantages of each are described below.
Spacecraft Pointing. In this option, the infrared sensor system is contained entirely on the
spacecraft and the spacecraft is slewed toward the target. The advantages of this pointing method
are simplicity (no additional optical components are required), narrow field of view (so the
optics can be well baffled and stray light will be minimized), and spacecraft compactness. The
disadvantages are poor pointing agility (large momentum wheels), high power, degraded
communication/power during pointing (re-pointing required for high gain antennas and solar
arrays), and degraded space-view during pointing (important if cryogenic, passively-cooled
detectors are used).
Figure 1. Notional Representation of the Across-Gimbal Thermal Transport System

A GIMBAL THERMAL TRANSPORT SYSTEM 709
Instrument Pointing. In this option, the entire instrument is contained within a canister on
a 2-axis gimbal. The advantages of this pointing method are pointing agility (the
mass/momentum disturbance is less than that of the spacecraft pointing option), uninterrupted
communication/power during pointing (the orientation of antennas/solar arrays/radiation coolers
is determined by the spacecraft attitude), simplicity (no additional optical components are
required), and narrow field of view (so the optics can be well baffled and stray light will be
minimized). The disadvantages are mechanical complexity (a large canister on a 2-axis gimbal is
needed), and electrical/thermal complexity (power, data, and thermal transport lines must be
routed from the telescope across the gimbal to the spacecraft).
Telescope Pointing. In this option, a portion of the optical system is included on a 2-axis
gimbal, and the rest of the system is on the spacecraft. The advantages of this approach are
pointing agility (the mass/momentum disturbance is less than that of the instrument pointing
option), a narrow field of view (so the optics can be well baffled and stray light will be
minimized), and system simplicity (power, cryogenic lines, and focal plane arrays (FPAs) are
located on the spacecraft). The disadvantages are optical complexity (the telescope's optical path
must be split between the canister and the spacecraft with at least one additional fold mirror) and
image rotation.
Scan Mirror Pointing. In this option, the instrument is on the spacecraft and a scan mirror
is on a 2-axis gimbal. The optical system has a wide field of view and the scanning mirror is used
to acquire and track targets. The advantages here are pointing agility (the mass/momentum
disturbance is less than that of the telescope pointing option), optical simplicity (all curved
telescope optics and focal planes are on the spacecraft body), and electrical/thermal simplicity
(power, cryogenic lines, and FPAs can be located on the spacecraft body). The disadvantages are
mechanical complexity (the system uses scan mirror on a two-axis gimbal), image rotation, and
stray light (due to the wider field of view).
Conclusion. For cryogenic infrared sensor applications where stray light must be
minimized, where agile target acquisition and tracking are needed, and where real-time
communications are needed during pointing, the telescope pointing and instrument pointing
options are preferred. If cryogenic temperature optics are needed, and gimbal mass/volume and
cryocooler jitter are to be minimized, then the ability to transport either ambient or cryogenic
cooling across the gimbal is an absolute necessity. The solution to this problem requires two
equally challenging developments: (1) a reliable ambient or cryogenic thermal transport system;
and (2) a reliable across-gimbal flex line system. Background information on these two topics is
provided below.
LHP Technology Options

Presented below is an overview of LHP technology followed by discussions of the
operational principles, pumping ability, and design/operational constraints of a standard LHP as
well as a brief description of two other LHP-based alternatives.
Overview. The loop heat pipe (LHP) is a revolutionary two-phase heat transport device
needed to solve the gimbal heat transport problem. In the "standard" LHP configuration (two
other options are discussed below), it is a two-phase, liquid/vapor fluid loop with five basic
components: evaporator, condenser, reservoir, transfer lines, and the working fluid. An LHP
reservoir is sometimes referred to as the "hydro-accumulator" or the "compensation chamber".
Figure 2 illustrates a flow diagram of a standard LHP two-phase heat transport system. In a
standard LHP, liquid is vaporized at the evaporator, which is mounted next to the heat source
(the on-gimbal cryocoolers). The resulting vapor flows within small diameter tubing to the
condenser, which is mounted next to the cooling source (the spacecraft-mounted radiator). The
vapor then condenses to liquid and flows back, again within small diameter tubing, to the
evaporator. An additional wick structure spans the physical distance between the reservoir and
the evaporator. This arrangement allows reliable loop start-up without preconditioning, it
provides passive management of the working fluid (typically ammonia) over a range of operating
temperatures, and it automatically regulates loop operating temperature over a wide range of
source/sink operating conditions.
Operational Principles. Based on heat pipe principles, fluid circulation in an LHP is driven
by surface tension forces (capillary action) that create very small pressure differences in the
liquid and vapor phases. Like the heat pipe, the LHP is a completely passive device with no
moving parts to wear out. When heat is added to the surface of a porous "wick" located in the
evaporator, liquid is vaporized and capillary action acts to replenish the vaporized fluid. To
accomplish this fluid replenishment, capillary action "pulls" liquid back into the evaporator. At
the same time, at the vapor outlet of the evaporator, the just-vaporized vapor "pushes" the
previously generated vapor towards the condenser. In this way, fluid is circulated in a loop and
heat is transported from the source to the sink. One significant advantage the LHP has over the
conventional heat pipe stems from the small, isolated wick in the LHP evaporator, which enables
the use of very small-diameter flexible tubing between the evaporator and condenser. In a
conventional heat pipe, a continuous evaporator-to-condenser wick (typically narrow slots or
grooves) necessitates the use of thicker-walled, not readily flexible tubing. This advantage also
translates into increased pumping ability of the LHP because smaller pore-size wicks can be
used.
Pumping Ability. The fluid pumping ability of a standard LHP (and the two other LHP
options described below) increases as the pore size of the porous wick is reduced. At 300 K, an
ammonia-filled LHP with a wick can overcome an adverse elevation of over 6 meters of
liquid ammonia. This pumping capability can also overcome accelerations due to gimbal
starting/turning while maintaining significant heat transport capability. The ability to use very
small pore size wicks in LHPs stems from the fact that LHP wicks are positioned at just one
specific location in the evaporator, which minimizes viscous pressure drop in the wick structure.
In all cases where pumping against gravity or external forces is required, the LHP is the
preferred heat transport solution. The system payoff for having against-gravity pumping
capability is that an LHP-enabled gimbaled spacecraft can most likely be tested in any
orientation. Thus, the LHP is the only reasonable option for across-gimbal heat transfer.
Design/Operational Constraints. In the standard LHP, the condenser must return
subcooled liquid to the evaporator. Often, the subcooler is a separate heat exchanger on the
space-facing radiator downstream of the condenser. Subcooling is required to compensate for
ambient heat absorbed by the liquid return line, ambient heat absorbed by the reservoir, and
gimbaled-payload waste heat conducted across the wick structure. Sufficient subcooling must be
provided to balance these heat inputs, otherwise the loop operating temperature will rise (i.e., the
loop will "autoregulate") until an energy balance is achieved. Typically, it is desirable to
minimize these heat inputs in order to achieve a minimum operating temperatures. It is also
necessary to separate the liquid return line from the vapor transport line across the gimbaled
joint. Finally, it is necessary that the condenser (i.e., the radiator) be configured to provide
optimum subcooling as required to achieve the desired system operating temperature. For the
systems considered herein, these design constraints mean the following. Sufficient isolation of
the liquid return line is needed considering the probable long line lengths, the fact that (relative
to rigid lines) flex line volume is comparatively large, the fact that mechanical constraints
required to achieve the needed flexibility across the joint may result in significant parasitic heat
input, and the fact that there may be large heat leaks into a reservoir that is not "cold-biased."
Thus, in a standard LHP, the reservoir must be either exposed to a cold environment (i.e., space)
or protected from a hot environment. Thus, with a standard LHP, the reservoir and evaporator
should not be wrapped together within a single MLI blanket.
Thermoelectric Option. As indicated, the standard LHP design approach requires two
separate lines across the gimbal per LHP loop. One novel design alternative, as illustrated in
Figure 3, is to mount a thermoelectric cooler (TEC) onto the reservoir to provide the cold-biased
environment. The heat dissipated by the TEC is added to the normal evaporator heat load with a
thermal strap. What this design option does is remove all constraints on LHP evaporator and
reservoir placement. The reservoir no longer needs to be cold-biased or protected from a hot
environment. In addition, the liquid and vapor lines no longer need to be thermally separated. A
coaxial arrangement, where the liquid return line flows inside the larger vapor outer line, is the
plumbing system of choice. Because of this coax arrangement, the liquid line can be a highly
flexible (non-hermetically sealed) material like Teflon. This plumbing approach has been
successfully demonstrated by Swales Aerospace on the HOST flight experiment. The objective
there was to reduce the diameter and improve the flexibility of the transport line bundle in an
LHP-like system that will ultimately be installed on the Hubble Space Telescope during
Servicing Mission 4 (SM-4).
Vapor Flush Option. Another novel design alternative for solving the problem addressed in
this paper is the LHP vapor flush design option, which is commonly referred to as the advanced
LHP or ALHP (see Hoang2). The ALHP flow diagram is provided in Figure 4. As indicated,
there are actually two evaporators in an ALHP. The first evaporator, located at the heat source
(the on-gimbal cryocooler), is a reservoir-less LHP evaporator. The second evaporator, located at
the cooling source (the radiator), is a typical LHP evaporator and reservoir. The ALHP uses two
superimposed flow loops. The first loop, denoted as the "steady-state" loop, carries the brunt of
the cryocooler heat load. The second loop, denoted as the "vapor flush" loop, carries as much
heat as is necessary to flush vapor from the on-gimbal evaporator. A small amount of heater
power (typically about 5% of the total dissipation) must be added to the on-radiator evaporator,
in order to provide the needed vapor flush mass flow. The ALHP has three important operational
advantages over the standard LHP configuration, and two advantages over the TEC option. First,
although there are three across-gimbal lines per loop, a tri-axial arrangement of the three lines is
possible. Thus, the three lines can be coalesced into a single across-gimbal transport line, similar
to the arrangement described above for the TEC option. Second, the LHP reservoir has been
moved off-gimbal, thus the ALHP has less on-gimbal weight (perhaps 1 kg per LHP loop) than
the TEC or standard LHP options. Third, the vapor flushing action of the on-radiator
evaporator/reservoir system significantly reduces the need for subcooling. Thus, the radiator can
be made more compact than it can in either other LHP option. The ALHP option was developed
by TTH, Inc. The ALHP has been successfully demonstrated in laboratory testing.
Figure 2. Standard Loop Heat Pipe (LHP) Flow Diagram.
Figure 3. Loop Heat Pipe with Thermoelectric Cooler (TEC).

Figure 4. Advanced Loop Heat Pipe (ALHP) Flow Diagram.
Flex Line Technology

As described above, LHP technology solves a portion of the gimbal thermal transport
problem. However, additional technology development was needed in two areas: (1) small
diameter flexible lines; and (2) 2-axis flex line routing schemes. In the former technology area,
flex lines were needed with an internal diameter of 0.32 cm (0.125 inch) or less that could
contain an internal pressure of at least 4.2 MPa (615 psi), which is ammonia's vapor pressure at
353 K, a worst-case hot temperature for STS payloads. The basic problem was that stainless steel
bellows tubing is typically not available in diameters less than 0.64 cm (0.25 inch). In the latter
technology area, bellows tubing must be flexed primarily in just one plane, so 2-axis flexibility
can be best accomplished by separating the flex line into two separate flexible sections. To
address these two needs, a preliminary search for available flex lines and a preliminary study of
possible line routing schemes were both carried out prior to the GATTS program.
Small Diameter Flex Lines. During the preliminary flex line search, just one supplier of
reasonably small diameter stainless steel bellows tubing was found. This particular bellows
tubing is a helical, seamless design with an ID of 0.39 cm (0.155 inch) and an OD of 0.66 cm
(0.26 inch). Figure 5 illustrates an 18 cm long section of this tubing with brazed-on end fittings.
Proof tests showed that that this tubing would not deform at pressures up to 10 MPa (1500 psi).
Flex Line Routing. During the preliminary study to evaluate across-gimbal line routing
schemes, the basic concept that was developed is shown in Figure 6. As indicated, the azimuth
axis flexible section consisted of 2 circumferential wraps (of a 3-line bundle) to give 360° of
flexibility and the elevation axis flexible section consisted of 1 circumferential wrap to give 180°
of flexibility. Rigid lines were used elsewhere. The 2-axis gimbal used to evaluate this line
routing scheme was constructed from drawings provided by the AFRL. This gimbal was
designated as the Gimbal Demonstration Unit (GDU). With this set-up, tests were conducted
examining the effects of line-induced torque and internal pressurization. During the GATTS
program, this line routing scheme was refined to reduce line length and improve repeatability.
Figure 5. Flexible Bellows Tubing Sample for 2-Axis Gimbal Thermal Transport.
Figure 6. Preliminary Line Management Scheme for 2-Axis Gimbal Thermal Transport.
Preliminary testing of line-induced torque with the configuration illustrated above indicated
that each flexible line added about 0.14 m-N (20 in-oz) of torque in each axis, so a 3-line bundle
added about 0.42 m-N (60 in-oz) of torque to each axis of the gimbal. Preliminary testing of
system pressurization effects with the configuration illustrate above indicated that torque was
independent of pressure to well above 0.7 MPa (100 psi), which would be the approximate
pressure of an ammonia-charged LHP operating near room temperature.
PROGRAM DESCRIPTION
This section of the paper described the GATTS program in detail including the program
objectives, requirements/specifications, design calculations/trades, design approach, test plans,
and test results to date.
Program Objectives
The goals of the GATTS program are to analyze, design, manufacture, and test a full-scale,
2-axis gimbal with an accompanying flexible ambient LHP thermal transport system. The
CRYOBUS 2-axis Gimbal Development Unit (GDU) is the starting point for the program. The
plan is to incorporate motors, position encoders, torque measurement instrumentation,
temperature sensors, and cryocooler simulators (heaters) onto the GDU. The goal is a motorized
system whose axes can be independently slewed over their entire range of motion during testing.
To manage line motion during flexing, the goal was to design a flight-like flex line routing and
guide system, an appropriately sized flexible standard LHP heat transport system, and the
required ground support equipment (GSE) for testing. The ultimate goal is to test the assembled
gimbal LHP heat transport system continuously for 2-4 weeks to identify any unknown dynamic
or potentially deleterious effect on the system's ability to meet the mission requirements outlined
below. The system will be tested in air under ambient laboratory conditions.
Requirements/Specifications
The basic requirements/specifications for the GATTS program are listed in Table 1. These
were prescribed and/or derived by AFRL and Swales as representative of a range of missions
requiring across-gimbal ambient thermal transport capabilities.
Design Calculations/Trades
A range of design calculations and trades were done to determine the test orientation
limitations (if any), bellows line diameter and length, bellows line pressure handling capability,
optimal evaporator placement, and motor torque requirements.
Test Orientation Limitations. One of the primary advantages of the LHP over
conventional heat pipes is the strong capillary pumping action in the evaporator. Because
capillary pressure is inversely proportional to the effective wick pore radius, this pumping
ability increases as the wick pore size is reduced. In fact, in many cases virtually all ground test
orientation limitations can be eliminated. To achieve this favorable result, the maximum against-
gravity pumping height (H) must exceed the physical separation distance between the condenser
and evaporator. With currently achievable evaporator pore radii of about with nickel wicks,
LHP evaporators with ammonia as the working fluid can overcome 6 meters of adverse
elevation. Equations (1-2) illustrate the relationships between surface tension pore
radius liquid density and H. Thus, GATTS and most small-to-medium sized systems
using an across-gimbal LHP can probably be tested in any orientation. One final factor to
consider is pressure drop in the lines, and this topic is discussed next.
Line Diameter and Length. To reduce line length and diameter, maximize line flexibility,
minimize torque, and minimize pressure drop, calculations were carried out for the flexible
tubing to determine pressure drop as a function of line diameter and length. The results of the
calculations are as follows. With ammonia as the working fluid, the pressure drop for a 200 W
heat load within a 1.25 m long section of the selected flex line was about 5% of the available
capillary pumping head and about 7 times higher than that of smooth-walled tubing.
Bellows Line Pressure Containment With ammonia as the working fluid, system
pressures will range from 0.54 MPa (80 psi) at 280 K to about 1.4 MPa (200 psi) at 310 K. This
working pressure range will pose no difficulty for the selected flex line given its proof test results
presented earlier.
Evaporator Placement Considerations. During gimbal movement, the fluid within the
LHP will be subject to accelerations. To minimize the impact of these accelerations on LHP
operation, the evaporator will be located as close as possible to the axis of rotation of each axis.
Test results are available which indicate that accelerations of up to 5g will not cause deleterious
effects in LHP systems. As a further precaution, the centerlines of the LHP reservoir and
evaporator (typically coincident with each other) will be parallel with the elevation axis of
rotation. When oriented nominally (see Figure 6) or upside down, this placement will prevent the
reservoir from being below the evaporator with respect to gravity (although certain sideways
orientations may be problematic). Additionally, the reservoir will be radially inboard of the
evaporator with respect to the azimuth axis, which will force fluid in the reservoir to feed into the
evaporator as the azimuth axis rotates in either direction.
Motor Torque/Placement and Axis Motion. To ensure the procured motors are located
optimally and have sufficient torque to drive each GDU axis enough to achieve the maximum
slewing rates listed in Table 1, a design study to quantify system torque and evaluate optimal
motor placement was implemented. The results of this design study are provided below.
The elevation motor will be mounted to the elevation on the opposite side of the gimbal
from the flex lines. It will be capable of producing approximate 0.56 m-N of torque,
which includes about 0.22 m-N of margin. The planned motion for the elevation axis
is to accelerate linearly at from –90° to 0° and then decelerate at from 0°
to +90°. This acceleration-deceleration profile will yield a maximum angular velocity of about
200 deg/s. The time required for a complete –90° to +90° elevation slew is about 1.8 seconds.
The azimuth motor will be mounted to the azimuth axis underneath the azimuth bearing. It
will be capable of producing approximately 2.2 m-N of torque, which includes a
margin of about 0.56 m-N The planned motion for the azimuth axis is to accelerate
linearly at from –200° to 0° and then decelerate at from 0° to +200°. This
motion will yield a maximum angular velocity of around 280 deg/s. It may be preferable to
modify this plan so that the motor maintains a constant velocity once an angular velocity of 200
deg/s is achieved. The time required for a complete –200° to +200° slew is about 2.8 seconds.
Design Approach
The primary goals in designing the GATTS test system included but were not limited to the
following: (a) maximize the repeatability/reliability of the flex line system; (b) minimize the flex
line and total line lengths; (c) minimize the line-induced torque, (d) minimize the weight of the
on-gimbal LHP, and (e) ensure sufficient instrumentation/test monitoring equipment to answer
the major technology question of whether gimbal motion affects LHP operation. An equally
important technology question, which will be part of the GATTS follow-on program assuming
this effort is successful, is whether LHP flow in conjunction with gimbal motion introduces
excessive jitter into the optical/sensor system. In the interest of brevity, the remainder of this
section just addresses the issues of line management and length minimization.
Line Management and Length Minimization. From prior testing, the line routing scheme
developed prior to GATTS (shown in Figure 6) had excessive line length, questionable launch
stability, and uncontrolled line movement in both axes. To rectify these deficiencies, a study was
carried out to look at alternative line routing methods. In this study, the schemes all centered on
the concept of "minimum dynamic bend radius". The minimum dynamic bend radius specified
by the manufacturer of the selected flex line is 7.5 cm (3 inches). The line management concepts
that evolved from the study are illustrated schematically in Figure 7.
In the elevation axis routing method, only 20 cm (8 inches) of flex line are used. One of
the nice features of this routing method is that the flex line is not flexed (or stressed) in its
neutral position at 0°. This feature means that there will be very good line controllability during
both launch and on-orbit operation. In the azimuth axis, because of the much greater range of
motion required (i.e., more than one complete revolution is needed), the flex line needs to be
about 100 cm (40 inches) in length. One unique feature of the azimuth line routing method,
which avoids having the lines rub against one another, is the helical wrap shown in Figure 7.
Although this method means that the lines will not bend strictly in plane, the slight out-of-plane
bending does not appear to be a problem. However, compared to the elevation axis line routing
method, the azimuth axis will probably require a more substantial launch lock subsystem to
prevent the lines from moving excessively during launch as well as a low-torque line
management restraint subsystem to ensure repeatable wrapping and unwrapping of the lines
during operation. The launch restraint subsystems will be developed as part of the GATTS
follow-on program, and will not be required for this effort.
Test Plans
This section of the paper addresses the plans for the test set-up, provides a preliminary test
matrix, and discusses the anticipated results. Again, at the time this paper was written, the test
hardware was still in the design stage.
Test Set-up. The test set-up for the GATTS program will be similar to the configuration
illustrated in Figure 8. Plans for modifying the existing GDU to meet GATTS program goals
include the following: (1) the GDU will be modified with the addition of motors on the azimuth
and elevation axes; (2) a single flexible LHP system will be utilized (although the GATTS
system will be designed to accommodate an additional flexible LHP system should follow-on
work for a redundant flight system be warranted); and (3) the LHP evaporator and reservoir will
be aligned along the elevation axis with the reservoir inboard of the evaporator. This evaporator-
to-reservoir orientation ensures that azimuth gimbal movement always forces reservoir fluid into
the evaporator.
Test Matrix. At the time of this writing, three primary sets of tests are planned for the
gimbal thermal transport system. The first will be an accelerated limited life test with the system
pressurized with nitrogen to identify any unanticipated lifetime issues with respect to the line
routing scheme that might cause the system to fail prematurely. Once this test has demonstrated
sufficient lifetime in each axis (e.g., greater than 50,000 cycles or 2.5 times the requirement of
20,000 cycles listed in Table 1), a new set of flex lines will be added and the system will be
charged with ammonia for the second set of tests. In this second set of tests, LHP system
performance will be generated without gimbal motion for use as a comparison baseline. Finally,
in the third set of tests, the GATTS LHP will be tested with simultaneous gimbal motion in each
axis. The types of subtests that will comprise this third set of tests has not been determined.
Ideally, all sorts of combinations of axis motions will be investigated to identify system
limitations (if any).
Anticipated Results. Based on the favorable results presented by who successfully
tested an LHP under accelerations up to 4.7g, it is anticipated that the LHP system will be able to
Figure 7. Illustration of GATTS Flex Line Management/Length Minimization Approach.

Figure 8. Test Set-Up for the GATTS Program.
operate successfully during gimbal motion. The maximum accelerations in the GATTS system
will be about 0.5 g (in each axis), hence gimbal motion should not affect LHP operation.
CONCLUSIONS
This paper has presented the rationale, technical background, and current plans for the
development and testing of an across-gimbal ambient thermal transport system. The application
for this system is for cryocooled infrared payloads for space-based DoD and NASA missions that
require advanced imaging and tracking capabilities. The status of this development effort, known
as GATTS (for gimbal ambient thermal transport system), is that the test system is still in the
design phase. However, no problems in successfully achieving the program goals are foreseen. In
about 2 months from the time this paper was written (in June 2000), manufacture and assembly
of the test hardware will have been completed. Testing will commence shortly thereafter and test
results should be available another month or so after that. When available, test results will be
published pending approval by the Air Force.
ACKNOWLEDGMENT
The authors would like to gratefully acknowledge the funding, program management, and
technical management support provided by the Air Force Research Laboratory for the work
performed in this effort, which was part of a broader research program known as CRYOBUS.
REFERENCES
1. Bugby, D., Brennan, P., et al., "Development of an Integrated Cryogenic Bus for Spacecraft
Applications," Space Technology and Applications International Forum (STAIF-97), January 1997.
2. Hoang, T., Kim, J., and Cheung, K., "Design and Test of a Proof-of-Concept Advanced Capillary
Pumped Loop," SAE 27th International Conference on Environmental Systems, July 1997.
3. Ku, J., Kaya, T., et al., "Testing of a Loop Heat Pipe Subjected to Variable Accelerating Forces,"
Spacecraft Thermal Control Technology Workshop, The Aerospace Corporation, El Segundo, CA,
March, 2000.
Cryocooler Interface System
G.S. Willen
Technology Applications, Inc.
B.J. Tomlinson
Kirtland AFB, MM, USA 87117
ABSTRACT
For actively cooled cryogenic systems it is usually necessary to locate the cryocooler in close
proximity to the cooled assembly. This places highly demanding requirements on the cryocooler,
its integration into the spacecraft, and introduces unwanted electrical, magnetic, and mechanical
disturbances. A unique Cryocooler Interface System is being developed in which the cryocooler
can be remotely located from the cooled elements, virtually eliminating cryocooler disturbances.
This system accommodates cooling across gimballed axes, provides nearly constant temperature
cooling under variable loads, can cool large area and/or distributed elements, and simplifies cool-
ing system integration.
The capability to provide cooling across a gimbal axis is important in advanced surveillance
systems such as the Spaced Based IR System-Low (SBIRS-Low) Segment, a near-term Air Force
program for developing and deploying a constellation of low-earth orbiting observation satellites
that incorporate cooled optics mounted on a two-axis gimbal. Mounting the cryocooler on the
gimbal has a number of drawbacks; the most serious are weight, vibration, and heat rejection. To
address these issues, a Cryocooler Interface System is being developed under an AFRL/VSSS
Phase II Small Business Innovative Research (SBIR) program (Contract No. F29601-99-0009) for
cooling the gimbal-mounted optics on SBIRS-Low.
INTRODUCTION
The emphasis on aerospace cryocooler development has focused on reliability, efficiency, and
vibration reduction. These goals have been generally realized; however, many practical issues
concerning cryocooler integration remain unresolved. Some of the more critical issues are: 1)
providing cooling across gimbal axes, 2) large area or distributed cooling, 3) effective cryocooler
heat rejection, and 4) overall cooling system integration.
Technology Applications, Inc. (TAI) is developing a unique Cryocooler Interface System (CIS)
in which the cryocooler can be located remotely from the cooled elements. The CIS accommodates
cooling across gimbaled axes, provides near constant temperature cooling under variable loads,
and greatly simplifies cooling system integration. It also allows cooling of large area and/or dis-
tributed elements, virtually eliminates cryocooler disturbances, is insensitive to gravity level, of-
fers ease of redundancy, and can be used with most types of cryocoolers.
Mounting a cryocooler on-gimbal adds weight, increases inertia, is a source of vibration, and
requires routing power and control wires across the gimbal axes. Since cryocooler heat absorption

is highly localized, high conductivity paths need to be incorporated into the gimbal mounted optical
system, adding weight and introducing temperature gradients. Perhaps the most critical issue is
effectively rejecting cryocooler-generated heat. For heat loads of 10 to 15 watts, the amount of heat
rejected is on the order of 100 to 150 watts; this requires a large radiator and, because the gimbal
can be pointed in arbitrary directions, the radiator performance can vary as a function of its view to
space.
Conventional approaches for cooling large area focal plane or optical assemblies using thermal
straps can induce large temperature gradients across the structure, are thermally inefficient, add
weight, and can put excessive stress on the cryocooler cold tip. Since spacecraft heat rejection
radiators are not typically located near the cooled elements, a solid conductor and/or heat pipes
must be employed to transfer cryocooler-generated heat to the radiator. This not only adds weight,
it increases the cryocooler internal temperature, thereby increasing input power and reducing cryo-
cooler efficiency, reliability, and lifetime.
Currently, the spacecraft system engineer must commit to a cooling system early in the design
phase. As the design evolves and system requirements change, the impact upon the cooler and its
integration into the spacecraft can lead to excessive cost and schedule growth, often compromising
overall system performance. The CIS will provide the spacecraft system engineers with a high
degree of cooling system design flexibility allowing them to design their systems without locking
into a specific cryocooler and a location.
SYSTEM DESCRIPTION
The CIS, shown in Figure 1, consists of a low-pressure fluid-cooling loop that absorbs heat
from the optics and rejects it at the cryocooler; the major CIS components are identified by the
bold call-outs. The fluid loop incorporates flexible segments that provide the range of motion for
routing the coolant lines across the gimbal axis. All CIS components, except for the transfer line,
can be located wherever it is most convenient; for the cryocooler, this will generally be in the
proximity of the spacecraft heat rejection radiator. Once the optical bench heat exchanger,
cryocooler, and circulation compressor assemblies are integrated into the spacecraft, the transfer
lines are installed connecting the two assemblies. Any changes in the spacecraft design will
primarily affect only the transfer line routing. The key CIS design requirements are given in
Table 1.
Figure 1. Cryocooler Interface System concept provides system flexibility.

CRYOCOOLER INTERFACE SYSTEM 721
Full cooling system redundancy is easily accommodated without the need for thermal
switches, as the coolant loop for the non-operating cryocooler adds only a few tens of milliwatts
parasitic heat load. If full redundancy is not required, several levels of redundancy can be
implemented wherein only the most failure prone components such as the cryocooler,
compressors, and electronic control are duplicated.
The Interface System operation is illustrated by following the path of the working fluid
through the block diagram in Figure 2, and corresponding state points on the Temperature-
Entropy diagram in Figure 3. For an optics temperature of 110 K, the circulating coolant is
methane, which has a normal boiling point of 111.6 K and is ideal for cooling in the 100 K to
120 K range.
The methane return gas is compressed adiabatically (1-2) to a pressure slightly higher than the
saturation pressure corresponding to the cryocooler cold tip temperature (for 120 K, the methane
pressure must be psia). The heat of compression is removed by cooling the methane to near
ambient temperature with a simple heat exchanger attached to the spacecraft heat rejection bus
(2-3). Next, the cold return gas from the optics precools the high-pressure methane stream in a
counter-flow heat exchanger (3-4). The methane is then condensed (4-5) and subcooled (5-6) at
constant pressure as it passes through the cold tip heat exchanger.
The subcooled liquid methane flows through a rigid, insulated transfer line that is connected
to the optics via flexible, low-stiffness line sections designed to route the coolant across the
gimbal axes (6-7). It is then expanded through a throttle valve to a pressure of 12 psia and a
corresponding temperature of 109 K (7-8) and routed through a heat exchanger thermally
attached to the optical bench (8-9) where the liquid phase boils at constant temperature,
absorbing the heat load. After the methane exits the optical bench heat exchanger, it is routed
through the return transfer line (9-10), through the counter-flow heat exchanger (10-1), and into
the compressor inlet. Since the optical bench heat exchanger temperature is determined by the
coolant loop pressure, the cryocooler operating temperature can be 5-to-10 degrees above the
optics temperature and vary as much as This reduces both the input power and
temperature control requirements imposed upon cryocoolers used in conventional cooling system
approaches.
Gas storage volumes are located on both the high- and low-pressure sides of the compressor.
Each is sized to maintain the absolute pressure in the system within safe limits when the system
is warmed to ambient temperature, and to buffer against optics heat exchanger temperature
variations that result from system pressure variations due to variable heat loads.
Figure 2. Cryocooler Interface System functional block diagram.
Figure 3. Cryocooler Interface System thermodynamic cycle.
It is not desirable to periodically vary the cryocooler heat lift to match the large SBIRS-Low
load variations, nor is it efficient to always operate the cryocooler at maximum load, throwing
away the excess cooling. A unique feature of the CIS is its ability to be easily integrated with an
external phase-change thermal storage unit (TSU). For SBIRS-Low, a TSU using 2-
methylpentane as the phase change storage medium is a promising option. Since 2-
methylpentane freezes at 119.6 K, yet remains liquid up to 60°C, there is no need for a large,
external expansion chamber. However, because 2-methylpentane freezes at 119.6 K, the
methane supply pressure must be psia to subcool the liquid methane.
The circulation system compressor is the only active component in the system. It has a
volumetric flow rate of three standard liters per minute and a pressure ratio of about 4:1. TAI is
developing a long-lifetime rotary compressor for the CIS program.
The cryocooler cold tip heat exchanger is a small diameter stainless steel tube wrapped
around a copper spool that is attached to the cryocooler cold tip. If a TSU is used, the cold-tip
heat exchanger will be integrated into the TSU. Since the coolant flow through cold tip heat
exchanger is mostly liquid, its effectiveness is greater than 0.98.
The throttle valve is located downstream of the cryocooler and reduces the liquid methane
pressure to approximately 12 psia, corresponding to a saturation temperature of 109 K. A
capillary tube was chosen because, for the same pressure drop, its diameter is considerably larger
than an equivalent orifice, making it less sensitive to contamination. It is also more stable, as the
flowrate changes only slightly with temperature, whereas, orifice flow is very sensitive to
temperature changes.
The optical bench heat exchanger consists of a small diameter tube attached to the optical
bench; it is designed to minimize spatial and temporal temperature gradients. In a zero-gravity
environment, the two-phase methane will continuously wet the tube walls. Bending the coolant
tubes in a serpentine shape will assure the liquid in the flow stream continuously contacts the
tube walls.
Filter assemblies are incorporated into the circulation system to minimize the potential for
throttle valve plugging. Getters cannot be used in the CIS since the methane coolant is strongly
adsorbed by the getters. Of the potential outgassing contaminates, only water and carbon dioxide
will condense; their freezing temperatures are 273.3 K and 216.5 K, respectively. Prior to
charging the system with methane, the CIS will be thoroughly cleaned, purged with hot nitrogen,
and evacuated. What little water and carbon dioxide are left will, for the most part, freeze out on
the counterflow and cold-tip heat exchangers. In the event the throttle valve becomes plugged, a
defrost heater installed on the throttle valve can be activated to thaw any potential plugging
resulting from condensibles.
The methane transfer line is composed of flexible sections that cross the gimbal axis
connected by rigid sections. One of the most demanding requirements is transferring coolant
across the gimbal axis. To accommodate the gimbal motions, the gimbal axis transfer line
assembly incorporates low-stiffness, flexible sections that cross the azimuth and elevation axes.
These flexible transfer lines are small diameter 300 series stainless steel tubes formed into a
helical configuration, winding and unwinding as the axes rotate. The azimuth axis rotates
between the elevation axis rotates from 0° to 90° and back.
The primary issues are achieving the full range of motions, minimizing torque on the gimbal,
and meeting the 400,000-cycle lifetime requirement. Since the transfer lines constitute the major
heat leak into the system, and every milliwatt of heat leak translates into a milliwatt of additional
load on the cryocooler; thus, the lines are gold coated to minimize heat leak. The gimbal axis
transfer line assembly is shown in Figure 4.
Figure 4. Gimbal axis line routing.

COMPONENT DEVELOPMENT TESTING

Component development testing of the azimuth and elevation axis flexible line segments and
the breadboard methane circulation system has been ongoing since Phase I of the SBIR program.
The circulation compressor is scheduled for performance and life testing starting in July 2000.
Full-scale models of the azimuth and elevation axis gimbal joints were constructed using
programmable stepper motors to provide the full range of motion. Representative flexible line
segments were fabricated and tested; the test set-up is shown in Figure 5. Preliminary azimuth
axis testing, conducted during the Phase I program, demonstrated over cycles of The
elevation axis flexible segment, consisting of both a supply and return line, was cycled over
times (0° to 90° and back) with no signs of fatigue or wear. One complete gimbal axis
cycle every 100-minute orbit equates to 5,260 cycles per year; the 2.5 million cycles
demonstrated is equivalent to 475 years.
The elevation axis torque was measured prior to and after completion of testing; the torque
was oz-in for both cases. This torque value is an average of six measurements; it represents
the net torque after the bearing torque was subtracted. All tests were conducted at room
temperature. Since the modulus of elasticity of 300 series stainless steel increases about 20% at
110 K, the gimbal torque will increase slightly at operating temperatures. The azimuth axis coil
has a larger diameter than the elevation axis coil; therefore, its net torque is expected to be less
than 3 oz-in.
The methane circulation system was tested to demonstrate the CIS remote cooling capability
and assess its operating characteristics. This test set-up and the accompanying circulation system
schematic are shown in Figure 6. The circulation system dewar assembly includes a liquid
nitrogen reservoir that is used in place of a cryocooler to absorb the circulation system
heat load, the counterflow and cold-tip heat exchangers, and the capillary throttle valve (refer to
the components enclosed by the dashed box). The transfer line from the throttle valve to the load
end is approximately one-meter long. The load-end heat exchanger is a one-meter long tube
coiled around a one-inch diameter copper mandrel with an imbedded cartridge heater that
supplies the variable cold-end heat load. A heater is attached to the reservoir cold tip, and a
temperature controller is used to maintain the methane temperature exiting the cold-tip heat
exchanger at
Figure 5. Azimuth and elevation flexible line segment test set-ups.

Figure 6. Methane circulation system test set-up.
Typical circulation system performance is illustrated by the temperature profiles in Figure 7.

The upper temperature data are the cold-tip outlet and throttle valve inlet temperatures (T7 and
T2). The lower temperature data are the throttle valve outlet, and the load heat exchanger inlet
and outlet temperatures (T3, T4, and T5). At the start of testing, the circulation system was
stabilized at a nominal 8 watts cold-end heat load. The cold-end heater power was then varied
between six and twelve watts. As shown by Figure 7, the load-end temperature generally stayed
within over the load variation, except at the high and low load extremes.
The conductance between the reservoir and cold-tip heat exchanger was sufficiently low
that, for a cold-end load of 12 watts, the methane temperature exiting the cold-tip heat exchanger
started rising above 119 K and was no longer subcooled. In response, the load-end temperature
increased 3 K above the 109 K nominal target temperature. When the cold-end load was
lowered to 6 watts, the cold-tip heater did not have sufficient power to maintain the methane
temperature at 118K, consequently, the methane was highly subcooled and the load-end
temperature dropped to 106 K.
The cold-end temperature variation with load observed in this test can be attributed primarily
to the test set-up, which was designed for a maximum load variation between 6 and 12 watts. A
commercial linear compressor was used; this limited the supply pressure to about 28 psia at a
flow rate of 30 mg/s. In addition, the cold-end heat exchanger has very little thermal mass. An
operational system will use a circulation compressor sized for the load and the optics subsystem,
having a much higher thermal mass, will greatly attenuate any temperature variation.
Figure 7. Circulation system test results.
PROTOTYPE INTERFACE SYSTEM TESTING

Testing of the prototype CIS under simulated operational environments will be conducted
using the test set-up illustrated in Figure 8. The test set-up will be instrumented to provide the
data necessary to fully characterize the CIS performance and operation; the instrumentation and
their locations are defined by the system schematic in Figure 9.
A representative set of gimbal axis motions and optics heat loads will be used in the prototype
test to simulate a realistic mission scenario. By cycling the gimbal once a minute, a ten-year on-
orbit simulation can be completed in approximately five weeks. The test will be performed in a
vacuum torr, and at room temperature.
Figure 8. Prototype Interface System test set-up.


A unique Cryocooler Interface System is being developed in which the cryocooler can be
remotely located from the cooled elements. Its capability to provide cooling across a gimbal
axis is important in advanced surveillance systems such as cooling the SBIRS-Low optics.
Critical component tests have demonstrated both the capability for effectively transferring
coolant across the gimbal axes and the operational stability of the circulation system. Prototype
system testing will demonstrate the overall system operation and capabilities.
ACKNOWLEDGMENTS
The authors wish to acknowledge the support of TAI personnel Steve Nieczkoski and Edward
Myers for their considerable contributions in the design, analysis, and testing of the Cryocooler
Interface System.
Development and Testing of a High Performance
Cryogenic Thermal Switch
B. Marland, D. Bugby, and C. Stouffer

Swales Aerospace
Beltsville, MD, USA 20705
B. Tomlinson and T. Davis

ABSTRACT
This paper presents development details and performance test results of a high performance
cryogenic thermal switch (CTSW) for coupling redundant cryocoolers to cryogenic components
with minimal off-cooler parasitics. Because gas-gap, hydride-pumped CTSW designs have not
reliably met performance goals of an "on" resistance less than 2 K/W and an "off" resistance
greater than 1000 K/W, a simpler, more reliable device was sought. The device that was
developed is based on the reversible and highly reliable physical process of differential thermal
contraction/expansion of stainless steel relative to beryllium. The 250 gram Swales differential
coefficient of thermal expansion (CTE) CTSW (or SDCC) has just 3 machined parts: two
cylindrical beryllium discs (same diameter, different lengths) and a thin-walled stainless steel
tube. In ground testing, the SDCC demonstrated an "off" resistance of 1400 K/W and an "on"
resistance of 1.2-2.0 K/W over a cold end temperature range of 25-50 K and a warm end
temperature range of 230-300 K. Other issues addressed in the paper include alternative gas-gap
(non hydride-pumped) designs, the option of "cross-strapping" and CTSW reliability. Finally, an
advanced SDCC design is also briefly described that can reduce CTSW mass to just 50 grams
with virtually no (surface area-induced) parasitic heat input into the cryogenic system.
INTRODUCTION
The reliability of today's cryocoolers remains a limiting design constraint for many long-
life, low risk cryogenic space applications. As a result, near term cryocooler space applications
will probably require cryocooler redundancy, which is accompanied by an additional parasitic
heat load from the non-operating cryocooler. For typical space cryocoolers operating without a
thermal switching device, the parasitic load stems primarily from conduction through the non-
operating cyrocooler expander. The thermal resistance of this conductive path is generally 400-
500 K/W for space cryocoolers. As a result, the parasitic load due to the non-operating
cryocooler is approximately 0.5 W ([285-60]/450) at 60 K, Bugby1.

In order to minimize cooling and input power requirements, a reliable cryogenic thermal
switch is desirable. A properly designed CTSW increases the thermal isolation between the
instrument and expander body of the non-operating cryocooler and reduces the parasitic load
from the non-operating cryocooler by 65-80%, Bugby1. As a result, cryocooler cooling and input
power requirements are substantially reduced.
The benefit of a CTSW becomes increasingly pronounced as the instrument operating
temperature and cryocooler efficiency are decreased. From 10-30 K, the cooling and input power
requirements needed to overcome an additional parasitic heat load of 0.5 W are prohibitive. In
this operating regime, a CTSW becomes essential. Figure 1 illustrates a redundantly cooled dual
CTSW system.
By reducing the parasitic penalty for each non-operating cooler, CTSWs invite the thermal
systems engineer to consider the use of multiple redundant coolers, lower reliability coolers and
low-cost tactical coolers. Figure 2 illustrates the use of multiple CTSWs in conjunction with an
array of low-cost tactical coolers.
CRYOGENIC THERMAL SWITCH PERFORMANCE REQUIREMENTS

The development of the cryogenic thermal switches presented in this paper was funded by
an AFRL-sponsored initiative to incorporate new and enabling cryogenic technologies into space
systems. This initiative, dubbed the Integrated Cryogenic Bus (ICB), endeavors to combine a
Figure 1. Dual CTSW System. Figure 2. CTSW/Low-Cost Cryocooler System.

DEVELOPMENT OF A CRYOGENIC THERMAL SWITCH 731
range of cryogenic integration solutions to meet the needs of future space applications. The
design of the SDCC is driven by the ICB established CTSW performance requirements, outlined
in Table 1.
SWALES DIFFERENTIAL THERMAL EXPANSION CTSW (SDCC) OVERVIEW

In an effort to meet the ICB performance requirements while providing a high reliability
CTSW, a novel and promising concept was invented. The SDCC design, which has just three
parts, is based on the reversible and highly reliable physical process of differential thermal
contraction/expansion of stainless steel relative to beryllium. The SDCC design is an evolution
of a gas-gap design developed by Swales and flown on STS-95 as part of the CRYOTSU
Hitchhiker Flight Experiment in October 1998. The Swales gas-gap design demonstrated the
ICB performance requirements in ground testing but due to the impressive potential of the SDCC
and system level reliability concerns with the gas-gap CTSW (which will be elaborated upon in
subsequent sections) emphasis for CTSW development shifted to the SDCC design. Unlike gas-
gap CTSW designs, the SDCC design does not utilize a working fluid. As a result, no hermetic
seals or getter/hydride pump to actuate the switch are needed and switch reliability is greatly
enhanced. Figure 3 illustrates the SDCC design. Swales Aerospace is in the process of patenting
this design.
The SDCC is actuated "on"/"off" by contracting/expanding the stainless steel tube relative to
the beryllium discs. When the smaller beryllium disc is cooled sufficiently, the higher CTE of
the stainless steel tube compared with the beryllium causes the narrow flat gap in the switch to
close, turning the switch "on". The time needed to actuate the SDCC "on" is minimized by the
low mass/thermal impedance of the smaller beryllium disc. For low CTE/beryllium instruments,
the beryllium construction of the SDCC allows for direct mounting, thereby, reducing the
number of flexible thermal links needed and greatly enhancing the effective "on" performance of
the system.
By warming the smaller disc or a portion of the stainless steel tube, a temperature gradient
in the tube is created, causing the tube to expand, the narrow gap to open and the SDCC to turn
"off". For missions utilizing redundant or multiple cryocoolers, once a cryocooler fails or is
turned off, it is desirable to thermally isolate that cooler from the cryogenic instrument as rapidly
as possible. By temporarily applying power to a small heater (approximately 0.3 W was needed
for the prototype SDCC) on the center of the stainless steel tube, the tube expands and the SDCC
is rapidly turned "off" ("on" to "off" actuation times are less than 5 minutes for the prototype
unit). Once in the "off" condition, the small beryllium disc becomes thermal isolated and its
temperature may be increased with heater power or allowed to increase due to parasitic heating.
Figure 3. SDCC Design.

After the smaller disc warms sufficiently (approximately 230 K with the prototype SDCC), the
SDCC will remain in the "off" condition without heater power and the smaller disc will operate
at a temperature above 230 K (275 K during SDCC prototype testing). The SDCC has been
operated at temperatures as low as 25 K but operation could be easily extended to below 4 K.
SWALES DIFFERENTIAL THERMAL EXPANSION CTSW (SDCC) DESIGN

In addition to its novel three-piece construction and unique beryllium interface
compatibility, the SDCC has several distinguishing design features, (1) the extremely narrow
(0.07 mm nominally), highly precise flat gap that separates the beryllium cylindrical parts, (2)
the thin-walled stainless steel tube which aligns the beryllium parts and provides structural
support, (3) the large size difference between the two beryllium components and (4) the
proprietary gold plating process applied to each beryllium part to maximize the contact
conductance.
As mentioned previously, a narrow flat gap separates the two beryllium parts. A major
advantage of the flat gap SDCC design over other mechanical and gas-gap CTSW designs is that
the gap width can be verified at any time, not just prior to final assembly. As a result, the flat
gap design significantly reduces the risk of a thermal short when the switch is "off". In addition,
by varying the gap width, the temperature at which the switch turns "on"/"off" may be adjusted.
In order to maintain the narrow flat gap and provide the thermal switching mechanism of
differential thermal contraction, a thin walled stainless steel tube precisely aligns the two
beryllium cylindrical parts. Aside from providing critical alignment, the stainless steel tube
represents the primary support structure in the SDCC. For a given cross sectional area, the thin-
walled design (0.13 mm as shown in Figure 1) maximizes the bending stiffness of the tube and
provides significantly more structural support than a solid rod with equal cross section. By
mounting the larger beryllium part directly to the cryogenic component, the stainless steel tube
only supports the mass of the smaller beryllium piece (which is minimized for structural and
thermal response considerations) and a portion of the flexible cryocooler interface. The tube
diameter is sized to provide enough bending stiffness to avoid mechanical contact of the gap
faces when the smaller beryllium part is subjected to a 13 g lateral load, approximately 3 N. A
robust fundamental lateral frequency of more than 170 Hz and positive stress margins with a
40 g lateral load result from this sizing. The structural design of the SDCC is more robust than
its predecessor, the hydrogen gas-gap CTSW, which was successfully qualified and flown on
STS-95 on the CRYOTSU Hitchhiker Flight Experiment.
While the switch is "off", the thin walled construction of the stainless steel tube represents
the primary thermal path from a non-operating cryocooler to the instrument, minimizing its cross
sectional area is critical. The tube's conductive resistance of approximately 2000 K/W for the
operating range described in Table 1 ensures effective thermal isolation while the switch is "off".
During "on" operation, the relative differential contraction of the stainless steel tube relative to
beryllium closes the gap and provides approximately 450 N (analytical estimate) of contact force
between the gap surfaces. The flat gap design of the SDCC loads the stainless steel tube axially
(its stiffest direction) ensuring large stress margins during "on" operation.
Operationally, the large size difference between the two beryllium components also plays an
important role. During "off" operation, the large temperature gradient along the length of the
tube and the thermal contraction of the larger beryllium part maintain a gap in the system. The
length of the larger beryllium component relative to the assembly length ensures sufficient
beryllium thermal contraction to maintain the SDCC in the "off" condition without heater power
on the tube.
Finally, a proprietary gold plating process applied to the gap surfaces of each beryllium
component reduces the contact resistance between these surfaces during "on" operation. The
SDCC has been cycled over 30 times with the gap operating force of 450 N and an additional 20
times in thermal cycling without any evidence of surface degradation or cold-welding. Life-cycle
testing awaits to fully qualify the SDCC for flight applications.
SDCC TEST METHODOLOGY AND RESULTS

In order to accurately determine "on" and "off" performance of CTSWs, simple yet effective
test methods were developed during the Swales gas-gap CTSW development effort for the STS-
95 Hitchhiker Flight Experiment. These methods, as they relate to the SDCC, will now be
discussed. In order to conduct an "on" test with the SDCC, the switch is turned on by mounting
the larger beryllium part to the cold head of a cryocooler and reducing the temperature of the
surrounding environment until the switch turns "on". Once the switch is "on", the surrounding
environment may be warmed and the switch will remain "on". When steady state temperatures in
the switch and environment are reached, an initial temperature difference across the switch
is measured where is the temperature of the warm end and is the
temperature of the cold end). This step provides the datum from which the temperature (relative)
measurement errors and the effects of external parasitics can be eliminated. Then, a known
heater power is applied to the warm end and a new is measured. The "on" conductance is
the heater power divided by the change in temperature difference. Eq. (1) provides the analytical
relationship.
The "off" test is conducted by first warming the surrounding environment to ambient
conditions and then turning the switch "off" by applying power to the heater on the center of the
tube. Once the smaller beryllium part reaches a critical temperature (approximately 230 K with
the prototype SDCC), the tube heater may be turned off and the switch will remain "off". With
the SDCC in the "off" condition, the "off" performance is easily determined. The required heater
power needed to equilibrate the warm end and surrounding environment temperatures, is
then applied to the warm end. This heater power "zeroes out" the parasitics. The "off" resistance
is the resulting divided by the heater power. Eq. (2) illustrates the analytical relationship.
Figures 4, 5 and 6 demonstrate ground testing results for the SDCC. The test results are
highlighted by an "on" resistance of 1.2 K/W at 50 K, 2.0 K/W at 25 K and an "off" resistance of
Figure 4. SDCC "ON"/"OFF" Performance.

Figure 5. SDCC "ON" Performance 25-35 K.
Figure 6. SDCC "ON" Performance 45-55 K.
1400 K/W. The reduced "on" performance of the switch at decreased temperatures is attributed
to the decreased thermal conductivity of the beryllium through the constricted heat path at the
contact interface. The rapid transitions from "off" to "on" (at 230 K) and from "on" to "off" as
well as the stable behavior in both the "off" and "on" conditions should also be noted. In
addition, it is evident from the data and the design descriptions above that the SDCC operating
range may be easily extended to below 4 K.
GAS-GAP CRYOGENIC THERMAL SWITCH OVERVIEW

As previously discussed, Swales began CTSW development work under the ICB initiative.
A hydrogen gas-gap CTSW was developed and flown as one of four advanced cryogenic
integration devices flown on the STS-95 CRYOTSU Hitchhiker flight experiment in October
1998 under this initiative. Figure 7 illustrates the layout of the CRYOTSU flight experiment. No
degradation in switch performance due to launch loads was observed upon comparison of pre-
flight and post-flight laboratory test results.
The concept for the hydrogen gas-gap cryogenic thermal switch utilizing a getter/hydride
pump is based on a JPL design, Johnson2, which used a zirconium-nickel hydride pump. A
hydrogen gas-gap CTSW and hydride pump/getter system is shown in Figure 8. In principle, the
hydrogen gas-gap CTSW is nominally "off" (evacuated) until actuated "on" by heating a metal
hydride getter, which evolves hydrogen and provides thermal conductance across the gap.
Although the Swales hydrogen gas-gap CTSW demonstrated an "off" resistance greater than
1000 K/W with both a Zr-V-Fe metal getter and a turbopump and an "on" resistance less than 1.0
K/W, the SDCC design remains the preferable solution for near term space-borne applications,
due to the higher anticipated reliability of a fluid-less/getter-less system. A potentially more
reliable solution than the gas-gap/getter system is shown in Figure 9. Latching solenoid valves
similar to those used on the HST Nicmos Cooling System, Nellis3, would be baselined for such a
system. Further details on this system, the Swales gas-gap CTSW design, and Swales gas-gap
CTSW test data may be found in Marland4.
Figure 7. Layout of the CRYOTSU Flight Experiment.
Figure 8. Hydrogen Gas-Gap CTSW and Hydride Pump System.

Figure 9. Hydrogen Gas-Gap CTSW and Solenoid Valve/Helium Tank System.
Figure 10. Multi-Stage Cooler "Cross-Strapping" Concept.
CRYOCOOLER CROSS-STRAPPING
One argument against the use of CTSWs is the potential option of "cross-strapping" in multi-
stage cooler systems. Figure 10 illustrates the concept. Cross-strapping, which is simply a
thermal linking across the lower stages of redundant cryocoolers, uses the excess cooling
capacity in the "on" cooler's lower stages to pump heat away from the "off" cooler's lower stages
with little performance degradation of the "on" cooler's highest stage. The net result is that all the
stages of die "off" cooler run considerably colder than they would if cross-strapping were not
used. However, cross-strapping is not an option for single-stage systems, and most of the long-
life coolers developed and flight qualified to date are, in fact, single-stage coolers. Thus, only if
multi-stage coolers become more prevalent will cross-strapping reduce the need for CTSWs.
Furthermore, even with cross-strapping, the thermal resistance of a properly designed CTSW is
significantly greater than the thermal resistance between the two highest stages of the cooler. As
a result, the CTSW increases the thermal isolation between the non-operating cryocooler and
cryogenic component and reduces parasitic heat leaks.
CTSW RELIABILITY
Based on the aforementioned discussions, it is evident that the system benefits of CTSWs
are substantial. The decision to implement CTSWs should therefore be based on reliability. For
space applications, the CTSW should have a reliability of nearly 100%. Although the SDCC has
not completed a rigorous flight qualification process, the anticipated SDCC reliability is
expected to meet the rigorous standards of flight applications. The operation of the SDCC is
based on the reversible and highly reliable physical process of differential thermal
contraction/expansion of stainless steel relative to beryllium. The simple three-piece design and
successful thermal and mechanical cycling ground testing of the SDCC enforce the reliability
expectations. The SDCC is now ready for flight qualification life cycle testing.
ADVANCED SDCC DESIGN

By modifying the SDCC mounting scheme and large beryllium part, the SDCC mass and
surface area-induced parasitics may be greatly reduced without effecting system operation or
performance. Figure 11 illustrates the concept. The design shown in Figure 11 has a mass of
just 50 grams with virtually no surface area-induced parasitics. As with the prototype SDCC
design, direct mounting to the cryogenic component eliminates the need for a flexible link
between the SDCC and cryogenic component, providing a significant improvement to the system
"on" performance and mass.
SUMMARY
The primary objective of this paper is to describe the design, operation, and test results
of the Swales differential CTE CTSW (or SDCC), which is designed to couple redundant
cryocoolers to cryogenic components with minimal off-cooler parasitics. With its simple
three-piece construction and highly-reliable, repeatable actuating mechanism of differential
thermal contraction, the SDCC is intended to provide a high reliability CTSW to meet the
DoD performance goals of <2 K/W "on" resistance and >1000 K/W "off" resistance. In
ground testing, the SDCC easily met these requirements, demonstrating an "off" resistance of
1400 K/W and an "on" resistance of 1.2-2.0 K/W over a cold end temperature range 25-50 K
and a warm end temperature range of 230-300 K. In addition to the SDCC design and test
results, related CTSW topics such as test methodology, gas-gap CTSWs and cross-strapping
are addressed. Finally, an advanced SDCC design with a mass of less than 50 grams and
virtually no area-induced parasitics is introduced.
Cryogenic thermal switches are critical cooling and input power saving devices, which
represent an important part of the Integrated Cryogenic Bus (ICB) initiative to incorporate
new and enabling cryogenic technologies into space systems. Offering potential high
reliability, simple construction, and unsurpassed performance, the SDCC provides intriguing
promise for the near future implementation of cryogenic thermal switches in space
applications. This paper is intended to provide the cryogenic space applications community
with a SDCC status report.
Figure 11. Advanced SDCC Design.

ACKNOWLEDGMENTS
The authors would like to acknowledge the Air Force Research Laboratory for providing the
funding for this work. We would also like to acknowledge the efforts of Thorn Davis and B.J.
Tomlinson of AFRL.
REFERENCES
1. Bugby, D., Stouffer, C., Hagood, B., et. al, "Development and Testing of the CRYOTSU Flight
Experiment," Space Technology and Applications International Forum (STAIF-99), M. El-Genk
editor, AIP Conference Proceedings No. 458, Albuquerque, NM, 1999, pp. 2-3.
2. Johnson, D. and Wu, J., "Feasibility Demonstration of a Thermal Switch for Dual Temperature IR
Focal Plane Cooling," Cryocoolers 9, Plenum Press, New York (1996).
3. Nellis, G., Dolan, F., Swift, W., and Sixsmith, H., "Reverse Brayton Cooler for NICMOS,"
Cryocoolers 10, Kluwer Academic/Plenum Publishers, New York (1998).
4. Marland, B., Bugby, D., Stouffer, C., "Development and Testing of Advanced Cryogenic Thermal
Switch Concepts," Space Technology and Applications International Forum (STAIF-00), M. El-
Genk editor, AIP Conference Proceedings No. 504, Albuquerque, NM, 2000.
Thermally Conductive Vibration Isolation
System for Cryocoolers
G. S. Willen
Technology Applications, Inc.
E. M. Flint
CSA Engineering, Inc.
Mountain View, CA, USA 94043
ABSTRACT
With the increasing demand for high temperature cryoelectronics and sensitive spaceborne
infrared imaging and detection systems, reliable mechanical cryocoolers have become an enabling
technology. These cryocoolers share two undesirable characteristics, vibration and heat genera-
tion. Vibration causes deflections in attached structures, adversely affecting sensor systems align-
ment and inducing spurious electrical signals. Heat generation increases cryocooler temperature,
thereby reducing efficiency, reliability, and lifetime.
Technology Applications, Inc. (TAI), with the support of CSA Engineering, Inc., developed a
multi-axis Thermally Conductive Vibration Isolation System (TCVIS) for cryocoolers under a SBIR
Phase II program for the Space Vehicles Directorate of AFRL (Contract No. F29601-97-C-0114).
The vibration isolation goals were a multi-axis 50:1 vibration reduction at the cryocooler drive
frequency, and a 10:1 reduction at two or more harmonics.
INTRODUCTION
Split-Stirling cryocoolers generally use opposed compressors whose vibration can be signifi-
cantly reduced by electronically adjusting their phase relationship. Back-to-back operation is not
feasible for most cryocooler expanders and active axial vibration cancellation is employed; how-
ever, reduction of axial vibration has no effect on the lateral vibration. Therefore, this development
study focused on multi-axis vibration isolation for cryocooler expanders.
When isolating the expander vibration from the structure to which it is attached, the expander
itself becomes effectively isolated from the structure. This poses a problem for in-space applica-
tions where expander heat removal depends upon conduction. Copper braid or aluminum foil
straps are thermally inefficient, heavy, and difficult to integrate into the system. In addition they
are generally too stiff to decouple the expander vibration from the structure. Therefore, a light-
weight, flexible, effective heat removal subsystem was integrated into the TCVIS design.
The TCVIS, shown in Figure 1, fits between the cryocooler and detector dewar assembly. It
incorporates six-axis vibration reduction that uses compact, efficient electromagnetic actuators,
force sensors, and passive isolators. This system has demonstrated a hundred-fold reduction in
vibration at the fundamental frequency with significant reduction at multiple higher harmonics.
Cryocoolers I I , edited by R.G. Ross, Jr.
Figure 1. Thermally conductive vibration isolation system.
The primary features incorporated into the design are ease of integration into cooling systems and
adaptability to many different cryocoolers.
The TCVIS was designed to accommodate a 1 -watt split-Stirling tactical cryocooler; a Hughes
Model HM7050C-514 used in this program. These type of cryocoolers have demonstrated in ex-
cess of 6,000 hours operating lifetime, making them a low cost alternative for many ground, air-
borne, and space-based tactical infrared surveillance systems. For applications where vibration
control and effective heat rejection are required, the TCVIS developed under this SBIR Phase II can
provide those capabilities.
DESCRIPTION
The major TCVIS assemblies and components are shown in Figure 2. The cryocooler is mounted
to an aluminum flange that conducts heat from the expander to a high conductance, flexible graph-
ite fiber thermal strap. The flange assembly consists of an aluminum upper flange that is attached
to a stainless steel lower flange by a welded bellows assembly. The lower flange bolts to the
detector dewar housing; the cryocooler expander is attached to the upper flange. The bellows
connecting the two flanges forms a vacuum tight assembly and provides a low axial stiffness path
that effectively decouples the expander vibration forces from the lower flange.
The vibration isolation assembly consists of six electromagnetic actuator struts arranged in a
Stewart configuration that provides the capability for both axial and radial vibration cancellation.
The upper ends of the actuators are attached to three posts that are attached to the upper flange; the
lower actuator ends are attached to the lower flange. Vibration forces generated by the expander
are counteracted by the actuators, attenuating the forces transmitted to the lower flange.
Heat Transport Subsystem

Heat generated by gas compression is localized in the expander head. This heat is conductively
transferred from the expander head to the upper aluminum flange via a conductive copper collar
THERMALLY CONDUCTIVE VIBRATION ISOLATION SYSTEM 741
Figure 2. TCVIS expanded view

showing the major subassemblies.
attached to the expander. The heat is then conducted to the flexible graphite fiber thermal strap
assembly that is bolted to the aluminum flange. The collar is made from oxygen free high conduc-
tivity (OFHC) copper, slips over the expander head, and is clamped with two cap screws to provide
good thermal contact. It is bolted to the upper flange near the thermal strap-mounting interface.
This assembly is shown in Figure 3.
The thermal strap assembly conducts the expander heat from the upper flange to the heat rejec-
tion interface. It is the most important heat transfer element between the cryocooler and the heat
rejection radiator, as over half of the temperature drop occurs across the strap. In addition to effec-
tive heat rejection, the thermal strap has sufficient flexibility to effectively attenuate vibration,
either from the TCVIS to the spacecraft structure, or from the spacecraft to the TCVIS.
Figure 3. Heat transport subsystem.

Figure 4. Flexible graphite fiber thermal strap assembly.
The low-stiffness, high-conductance requirement led to the development of a lightweight, flex-

ible, high-thermal conductivity graphite fiber thermal strap. The strap assembly, shown in Fig-
ure 4, consists of a single row of 20-fiber bundles and two aluminum end fittings. It is made from
high thermal conductivity K-l 100 graphite fibers attached to aluminum end fittings. These fibers
have 2.75 times the thermal conductivity of copper and 1/4 the density; on a per unit mass basis,
they are theoretically 11 times more weight efficient than copper, and over six times more weight
efficient than aluminum. The theoretical maximum conductance is 0.45 W/K; the measured ther-
mal strap conductance was 0.35 W/K, yielding a thermal efficiency of 78%. The weight of the strap
is 26.5 g, approximately half the weight of a comparable aluminum foil strap.
Vibration Isolation Subsystem

The six-axis vibration isolation system (VIS6) reduces both the axial and radial forces gener-
ated by the cryocooler. To provide multi-axis vibration isolation, the isolation system is arranged
in a Stewart (i.e., hexapod) configuration, in which the six struts are oriented to provide control
authority over all six principal degrees of rigid body motion. A stand-alone signal processor was
developed using a feed-forward control algorithm to reduce the first five disturbance tones (the
primary and four harmonics).
The VIS6 consists of two subsystems: the transducer hexapod that contains the six struts and
the electronics. The electronics consist of the load cell signal conditioning and the electronic sup-
port package. The VIS6 system schematic is shown in Figure 5. The six struts are arranged in
symmetric pattern of three repeated pairs around the cryocooler expander; this arrangement is iden-
tified as the vibration assembly in Figure 2.
Each strut, shown in Figure 6, consists of a single electromagnetic proof mass actuator (PMA),
a viscoelastic passive stage, a piezoelectric wafer load cell, and two axial end flexures; it weighs
approximately 120 g. The electromagnetic PMA provides the counter force required by the control
subsystem. A current supply power amplifier in the electronic support package (ESP) supports the
actuator in each strut. Similarly, signal conditioning is provided to the wafer load cell of each strut
by a dedicated operational amplifier. The average resonance of the actuators is 44 Hz, at which they
can generate 0.47 N/V. At 200 Hz the actuators can generate 0.133 N/V. In the low frequency
range (below resonance) the generated force was limited by the 23.3 mils end-stop limit.
The filtered-x least mean square (FXLMS) algorithm was used as the basis for the active vibra-
tion control. FXLMS is a version of the least mean square algorithm in which the reference
signal is filtered through an estimate of the secondary path dynamics before it is used in the output
and primary plant update.
SUBSYSTEM TEST RESULTS

Tests were conducted on both the heat transport and vibration isolation subsystems to demon-
strate performance and to characterize the individual subsystems. The thermal testing was limited
to measuring thermal strap conductance. Extensive testing of the vibration isolation subsystem
was conducted to fully characterized its operation and performance.
Figure 5. VIS6 system schematic. Figure 6. VIS6 strut configuration.
Thermal Strap Test Results

The thermal straps were characterized by measuring their overall thermal conductance. These
tests were performed at the National Institute of Standards and Technology (NIST) in Boulder, CO
and repeated at TAI; both sets of test data were in good agreement. The thermal conductance was
measured and compared with theoretical predictions to determine overall strap conductance and
thermal efficiency. Each strap was tested at a minimum of five different power levels to determine
if the conductance was dependent on power.
A flexible strap with 20-fiber bundles was fabricated and tested; refer to Figure 4. Its measured
conductance was 0.35 W/K; this value represents an average of five data points taken at different
heater power settings. The theoretical maximum conductance is 0.45 W/K; therefore, the thermal
efficiency of the flexible strap is 78%. Different power settings were used to verify the test mea-
surement consistency and to assess whether increasing the amount of heat transferred (heat flux)
affected the conductance. The conductance variation with power and the average temperature drop
across the strap for each series of tests is shown in Figure 7. These results were typical for all
thermal straps tested and showed the conductance to be independent of heat flux. The data also
provide a check on the “goodness” of the conductance testing. For a constant conductance, the
should approach zero as the heater power approaches zero. As shown in Figure 7, the linear ex-
trapolation of the nearly intersects the origin.
Figure 7. Thermal conductance test results for a 20-fiber bundle strap.

Figure 8. Comparison of open and closed loop behavior in the time domain for all six load cells at 4
Vrms cryocooler drive voltage.
To verify these results, two additional thermal straps were fabricated and tested. Both were the
same design, but they were fabricated with 18- and 22-fiber bundles. Their measured conductances
were 0.34 and 0.36 W/K, respectively, which is consistent with the 0.35 W/K conductance of the
20-fiber bundle thermal strap.
Vibration Isolation Test Results
The vibration isolation subsystem performance was characterized over a cryocooler drive volt-
age range of 2-to-10 Vrms (10 Vrms corresponds to full power). Figure 8 illustrates a representa-
tive example of the vibration subsystem performance. It corresponds to the load cell response of
strut 4 at a cryocooler drive voltage of 4 Vrms. Using the strut load cell locations and orientations,
the axial and radial loads generated by the cryocooler and transmitted through the load cells were
calculated. On average, the primary frequency was reduced 40.8 dB (100:1) in the axial direction
and between 31.5 to 37.1 dB (37:1 to 70:1) in the two radial directions. Peak reductions of 45.6 dB
(190:1) occurred at an 8 Vrms cryocooler drive voltage for the axial direction and 42.5 dB (130:1)
at 4 Vrms for the radial x-direction. The next four tones that were controlled were reduced on
average 21.8 dB (12:1) in the axial direction and between 15 to 19.4 dB (5:1 to 9:1) in the two radial
directions.
Overall performance results stayed relatively consistent versus the cryocooler drive level, as
illustrated by Figure 9, which shows the axial and radial vibration force reductions for three cryo-
cooler drive voltages. The radial reduction is an average of the two transverse radial components.
The maximum axial and radial reductions of –40 and –35 dB, respectively, occur at the cryocooler
Figure 9. Vibration isolation performance for 8, 9, 10 Vrms cryocooler drive voltage.

Figure 10. TCVIS test setup with thermocouple locations.
drive frequency (dark bars); the –20 to –25 dB reduction is an average of the next four harmonics
(light bars). These results confirm that the VIS6 subsystem met or exceeded all vibration isolation
goals and requirements.
INTEGRATED SYSTEM TEST

To demonstrate the TCVIS can meet operational requirements in a space environment, an
integrated system test was conducted at NIST in Boulder, CO. The TCVIS was mounted onto a
base plate with the heat sink attached to a liquid cooled cold plate, as shown in Figure 10. The test
setup was attached to a heavy ballast plate using two fiberglass channels for thermal isolation. It
was then placed inside a vacuum chamber where a diffusion pump maintained the vacuum below
torr and a recirculating chiller provided the cooling for the heat rejection interface over a –20°C
to +20°C temperature range.
The criteria for determining successful TCVIS operation during the Integrated System Test
was defined as 1) demonstrating a reduction in vibration forces comparable to those demonstrated
during the subsystem level testing and 2) maintaining the cryocooler expander temperature below
70°C under the worst-case conditions of maximum cryocooler power and heat rejection interface
temperature. Exact duplication of the extensive vibration subsystem testing was not expected be-
cause of the TCVIS mounting in the NIST vacuum chamber and the many additional vibrational
modes associated with the vacuum chamber, diffusion pump operation, chiller flow transfer lines,
and general disturbances throughout the laboratory.
This test demonstrated the TCVIS met its target of multi-axis, vibration isolation, and that the
thermal transport subsystem maintained the cryocooler expander within its operating temperature
range. Vibration force reduction in the simulated space environment was demonstrated by reading
the broadband vibration signal from the load cells, and viewing each strut load cell output on a
digital oscilloscope. Thermal performance was determined by measuring critical component tem-
peratures and the temperature distribution along the TCVIS conductive path. The broadband vibra-
tion reduction for the integrated test was not as great as that measured during the controlled sub-
system tests but was generally in good agreement with the subsystem level test.
The average steady state expander temperature, for a heat rejection interface temperature of
20°C, ranged from 60°C to 68°C as the power was varied from 80 to 100%. As expected, the
temperatures increased as the power increased. Over the full power range, the thermal transport
subsystem was able to keep the expander temperature below 70°C, within the safe operating range
of tactical cryocoolers.
Figure 11. VRS1 mounted to the cryocooler expander.
TECHNOLOGY SPIN-OFFS
Two technology spin-offs, a single-axis vibration reduction system (VRS1) and a flexible ther-
mal strap assembly, were developed during this program. The single-axis vibration reduction sys-
tem is a counter-force based system that uses an electromagnetic proof mass actuator (PMA), accel-
eration based error signal, and a feedforward control algorithm implemented on a stand-alone pro-
cessor. The VRS1, shown in Figure 11, is simple, compact, and attaches to the cryocooler or
expander to facilitate ease of integration. It consists of a transducer unit, attachment fitting, control
electronics package and associated cabling. The transducer unit (actuator and sensor) is bolted to a
circular aluminum mounting ring that slips over, and is clamped to, the cryocooler expander head.
When assembled, the actuator end cap sits on a ring of material separating it from the expander
head, limiting heat conduction into the actuator. A flat-heat-rejection interface is machined into the
aluminum mounting ring with three tapped holes for easy attachment of a thermal strap.
The VRS1 reduces 11 disturbance tones, the primary and the first 10 higher order harmonics.
Figure 12 shows representative results from the measurement of the VRS1 performance when
mounted on the VIS6 hexapod with the cryocooler driven at 6 Vrms. The primary and second
harmonic disturbances were reduced by over two orders of magnitude (42.5 and 40.5 dB). Over the
Figure 12. VRS1 performance measured by the accelerometer error sensor, 6 Vrms drive voltage.
Figure 13. Graphite fiber thermal strap assembly.
bandwidth of operation (550 Hz), the measured axial rms acceleration disturbance levels were
reduced, on average, by more than an order of magnitude (26 dB). At 10 Vrms, the primary and
second harmonic disturbances were reduced by 18 and 28 dB, respectively; the higher harmonics
demonstrated an average force reduction of 28 dB. The decrease in performance at the primary and
second harmonic is most likely due to VRS1 stroke limitations.
This system in its entirety adds an additional 1.3 kg, of which 88% is associated with the
electronics, cabling, and power conditioning. The system requires 9.6 W in standby mode, and
13.0 W when in the control mode. It is easy to retrofit onto existing systems and only requires
access to the cryocooler drive signal and 18-36 V DC power.
Flexible Graphite Thermal Strap Assembly

A lightweight, flexible, high-conductance, K1100 graphite fiber thermal strap (GFTS) assem-
bly was developed. A 28.5-cm long GFTS with a conductance of 0.20 W/K has demonstrated a
78% weight savings over a solid copper (OFHC) bar of the same conductance. An equivalent
aluminum (1100F) rod would have a diameter of 0.75 in and weigh 2.7 times more than the GFTS.
These copper and aluminum conductor weights assume solid rods; flexible braid or foil conductors
would weigh substantially more.
The GFTS configuration tested is pictured in Figure 13; it consists of two rows of K1100
carbon fiber bundles and two aluminum end fittings. Each row is made up of 20 fiber bundles; each
fiber bundle consists of 20,000 individual fibers. Three 28.5-cm straps were fabricated with 1, 2,
and 3 rows, respectively. Their theoretical and measured conductances along with their thermal
efficiencies are summarized in Table 1. The thermal efficiency of the three straps shows a signifi-
cant decrease when increasing the number of rows. The difference in thermal efficiency appears to
be linear with the number of rows, about 8.5% for each row. To see if this trend is valid for four
rows, a four-row engineering prototype strap, 24.8-cm long, with 88 total fiber bundles was tested.
Its theoretical and measured conductances were 0.61 and 0.31 W/K, respectively. This yielded a
thermal efficiency of 51%, indicating that the fourth row does not offer any significant advantage.
Therefore, the bulk of the inefficiency appears to be related to the end fitting conductance losses.

A TCVIS was developed that fits between the cryocooler and the sensor housing dewar assem-
bly. It integrates both a thermal transport and a six-axis vibration cancellation subsystem into a
compact, lightweight assembly. This system was designed to accommodate a 1-watt split-Stirling
tactical cryocooler expander; however, it is adaptable to many different cryocoolers.
At the cryocooler drive frequency, the TCVIS demonstrated more than 40 dB reduction in the
axial direction and between 31 and 37 dB in the two radial directions; the next four harmonics were
reduced an average of 22 dB. The flexible thermal strap successfully limited vibration transmis-
sion and maintained the cryocooler expander well within its recommended operating temperature.
Using the technologies developed in this program, two innovative technology spin-offs were
developed: a single-axis vibration cancellation system for tactical cryocooler expanders and a flex-
ible thermal strap assembly. The single-axis vibration reduction system reduced 11 disturbance
tones. The primary and second harmonics, the dominant axial disturbances, were reduced by over
two orders of magnitude; the higher harmonics were reduced by more than an order of magnitude.
The graphite fiber thermal strap assemblies demonstrated thermal efficiencies up to 92%, yielding
a 65 to 80 percent weight savings over aluminum or copper straps for the same conductance. These
straps have sufficient flexibility to accommodate installation and alignment tolerances, relative
structural movements due to vibration and thermal contractions, and limited bending.
ACKNOWEDGEMENTS
The authors wish to acknowledge Mr. Michael Evert, Mr. Patrick Flannery, and Dr. Eric Ander-
son of CSA Engineering, Inc., for their significant contribution in the development of the vibration
isolation subsystem; Mr. Edward Myers and Ms. Jennifer Lock of Technology Applications, Inc.,
for their contributions to the system design, analysis, and testing; and Dr. Dino Sciulli, AFRL
Technical Project Officer, for his assistance and guidance throughout the project.
Advanced Cryogenic Integration and Cooling
Technology for Space-Based Long Term
Cryogen Storage
B. J. Tomlinson and T. M. Davis

Kirtland AFB, NM 87117
J. D. Ledbetter
Mission Research Corporation

ABSTRACT
The Air Force Research Laboratory (AFRL), Space Vehicles Directorate has been the lead
Department of Defense (DoD) agency for the development of low capacity cryogenic
refrigerators and integration technologies for space applications since the late 1980s. High
capacity cryocoolers and long term (>20 years) on orbit propellant storage are potentially
enabling technology for future High Energy Laser (HEL) space systems, orbital transfer vehicles,
and on orbit propellant depots. Cryogenic applications in space based systems requiring long
term cryogen storage includes: significant cooling requirements for subcritical cryogens,
cryocooler redundancy issues, on orbit cryogen transfer from vehicle to vehicle, large shield
cooling, long term gas and liquid cryogen storage, large distributed cooling surfaces, cryogenic
system integration issues, and significant spacecraft system penalties due to mass and input
power. AFRL has pursued low capacity cryocooler concepts including reverse Brayton cycle,
single and multiple stage Stirling cycle, advanced Joule-Thomson cycle, and Pulse Tube (Stirling
cycle variant) designs. The cryocooler technology spans a wide range of cooling temperatures
(from ~10 Kelvin to 150 Kelvin) and heat loads (up to 10 Watts at 95 Kelvin). Additionally,
AFRL has pursued advanced cryogenic integration technology including cryogenic thermal
switches, cryogenic heat transport, thermal storage, and cryogenic integration schemes to reduce
system mass and input power penalties.
Current cryogenic integration and cryocooler development programs address the negative
impacts of the cryogenic system on optical space systems including: induced line of sight
vibration, longevity, power consumption, mass, thermal transport, thermal storage, and thermal
switching. However, the cryogenic cooling requirements for future Air Force systems may
require large capacity cryogenic cooling, extremely mass and power efficient mechanical
refrigerators, and significant improvements in long term on orbit cryogen storage. The technical
efforts at AFRL concentrate on exploratory and advanced development programs that focus on

the development of technology from concept and breadboard engineering models to protoflight
models that are geared to experimental characterization and technology transition for flight
demonstrations and, potentially, operational programs.
INTRODUCTION
For the last decade, developments in cryocooler technology have focused in on low capacity
heat lift in the 35 Kelvin to 150 Kelvin temperature range with recent developments in cooling at
10 Kelvin. Essential elements of the technology have evolved to establish a technology base for
long life cryocoolers designed to meet reliability, mass, vibration, and thermodynamic
performance requirements. In addition, cryogenic integration technology has also focused on
low capacity thermal issues in the cryogenic system. Evolving mission requirements for
infrared, multi, and hyperspectral focal plane and optics cooling are still driving the technology
development to lighter, more efficient and reliable cryocoolers.1
Future Air Force and Department of Defence (DoD) systems including High Energy Laser
(HEL) space systems, orbital transfer vehicles, and on orbit propellant depots all have
requirements for the long-term storage of cryogens, also known as zero-boil off systems, in a
liquid or gaseous state for use as propellant or chemical reactants. In addition to the Air Force
and DoD system concepts, the National Aeronautic and Space Administration (NASA) has been
exploring technology for human and robotic space exploration. The long-term cryogen storage
requirements cover numerous issues including increased cooling capacity refrigerators,
cryogenic system redundancy, cooling of large distributed surfaces, cryogenic system
integration, and significant system power and mass reduction.
This problem is not a new one. The idea of long-term on-orbit cryogen storage has been
considered in the literature for at least three decades. One of the largest obstacles to
implementing the past system concepts was the lack of reliable, sufficiently efficient, and low
mass cryogenic cooling technologies. Current technology development for sensor and optics
cooling has provided a potential bridge for the development of feasible long-term cryogen
storage systems.2
The requirements for the long-term storage of cryogens are still emerging and could change
many times before the completion and implementation of a flight system. However, there are
many common attributes of the possible system configurations that can be covered by an
organized technology development program. Past AFRL technology development programs
have focused on lower heat lift capacity, long life technology and significant system issues such
as reduction of induced vibration, mass, and input power. However, current AFRL programs
provide a starting point for technology development to meet the advanced requirements. The
following sections provide an overview of current AFRL programs for cryocooler and cryogenic
integration development and outlines potential plans to meet the emerging requirements.
POTENTIAL REQUIREMENTS
There are potential requirements on the horizon that will require expansion of past
developments to increase the reliable lifetime, reduce the mass, significantly increase the
capacity, and increase the power efficiency of cryocooler and cryogenic integration technology.
System concepts for Air Force HEL systems and orbital transfer vehicles are in their infancy and
still evolving. Specific cryogenic capabilities are very system design dependent, but there are
common themes for the range of technology requirements.
Table 1 provides some examples of potential cooling requirements of large tanks and shields
for on-orbit reactant storage for a conceptual space based HEL system. However, the cooling
temperatures and loads are only part of the problem. The issues of how this system is integrated,
potential redundancy, the large cooling surfaces on the shields and tanks, and the system impact
of cooler mass and input power are significant and especially system dependent. Decisions on
development of large capacity coolers, gangs of small capacity coolers, single stage coolers, or
multiple stage coolers depend on the particular cryogenic system schemes that will be pursued.
SPACE-BASED LONG TERM CRYOGEN STORAGE 751
As an example, redundant multistage cryocoolers could mitigate the need for cryogenic
thermal switches to offset the parasitic heat load of the “off” redundant cryocooler(s) by cross-
straping the stages and intercepting the heat at the more efficient warmer stages of the cooler to
prevent most of the heat load from reaching the less efficient colder stages of the “on” cooler.
The Aerospace Corporation has examined several potential scenarios for cryogenic storage of
reactants for space based HEL systems. This includes trades on cryogenic gas versus liquid
systems, large versus small coolers, small versus large tanks, and other system level attributes.
An additional consideration was the comparison of system cooling performance by utilizing
hypothetical multistage cryocoolers that were based on extensions of current technology. The
results of these initial studies have shown that there is a wide range of system impacts depending
on the specific system design. As expected, the tank heat leaks were shown to have a large
impact on required cryocooler cooling power and with lower efficiency coolers the input power
required was staggering for even a small system. However, with more effective tank thermal
isolation and multilayer insulation, and with an efficient cryocooler, the input power and mass
penalties become manageable.3 An additional finding was the comparison of systems employing
single stage cryocoolers or multistage coolers for a conceptual 100 Kelvin tank of gaseous
cryogen. The results from Aerospace show that the multistage systems have significant system
benefits over single stage cooler systems.4
Cryogen storage in space and for the exploration of extraterrestrial bodies is also a
technology need for the National Aeronautical and Space Administration (NASA) Mars
exploration programs. Human and robotic missions to Mars will require cryogen liquefaction,
storage, and transfer for propellant and breathable atmosphere during the long trip to Mars and
for storage on the surface for use at a later time. Kittel, et. al., has examined the problem of
cryogen storage and liquefaction cooling for conceptual Mars exploration and although the
mission lifetime requirements are on the order of 6 months to five years, the technology required
to meet these mission requirements will apply to future AF and DoD missions. Potential cooler
requirements for the Mars missions are shown in Table 2 and are a mixed combination of various
missions and durations up to 1.4 years.5
As system concepts for AF and DoD missions evolve, the requirements for advanced
cryogenic integration and cooling technology for space based long-term cryogen storage will
change. However, there are current development activities that are addressing many of the grass
roots issues associated with the cooling system requirements for these advanced systems. As is
the case on many development programs that span decades, the component level technology
maturity and capabilities are major drivers in the system design. Careful development and
investment into cooler and cryogenic integration technology will allow for the timely maturity of
the technology through both ground proof-of-concept demonstrations, prototype hardware

development, and flight qualification through short or long duration space missions and
Advanced Concept Technology Demonstrations (ACTD).
CURRENT DEVELOPMENT ACTIVITY
The Refrigerators
When examining the potential system concepts that are on the horizon it becomes evident
that some of the requirements will need the capability for large capacity cooling. As the system
concepts evolve, the extent of the increase in capacity over the current state of the art designs
will become clearer. There is the possibility of utilizing a single large cryocooler to produce all
of the necessary cooling for a given requirement, or sharing the load between two or more
smaller coolers. This is a system design, integration, and reliability issue and will be considered
during the evolution of the future concepts.
Although significant improvements in the state-of-the-art have been realized, there are still
development issues to be resolved, especially in cooling ranging from 10 to 50 Kelvin.
However, the developments such as improved non-contacting “Oxford” flexure bearing
compressors, miniature gas bearing turbine compressors and expanders, pre-cooled and
lubrication-less Joule-Thomson systems, and marked improvements in pulse tube technology
have greatly expanded the use of cryocooler technology in a broad range of space applications
and are baselined for many near-term and future missions as “enabling” technology. This is a
reflection of the increased mission need for cooled sensors and optics, but is also a reflection of
the increasing maturity of cryocooler and cryogenic integration technology.
AFRL is currently developing higher capacity cryocoolers to meet the Space Based Infrared
System Low (SBIRS Low) technology requirements for the Engineering Manufacturing
Development phase of the program development. It was identified at the beginning of the AFRL
technology development that there was a need for highly efficient, low mass cryocoolers for
cooling the gimbaled fore optics of a conceptual space based system. The need arose due to the
large cooling load that would be required to cool the necessary optics, the need to minimize the
waste heat on gimbal by increasing efficiency (due to limited radiator area on gimbal), and to
reduce the mass as much as possible due to the large amplification of the mass penalty on
gimbal. The requirement was determined to be 10 Watts of cooling at 95 Kelvin, with minimal
mass and a specific power of at least 10 Watt input power / Watt cooling or lower. The two
programs currently underway at AFRL are the TRW 95K High Efficiency Cryocooler (TRW
HEC) and the Raytheon 95K High Efficiency Cryocooler (Raytheon HEC). Both development
programs are leading to technologies that will provide the framework for the follow-on
generations of high capacity machines.
The TRW 95K High Efficiency Cooler is the latest development in pulse tube technology.
The cooler is a vast improvement in pulse tube cryocooler technology due to a combination and
culmination of various technology developments. The cooler is designed to lift 10 W of heat at
95 K with an overall cooler specific power of 10 W/W.TRW has transitioned the “Oxford” linear
compressor technology developed under IR&D and under the IMAS cryocooler development
program for the Jet Propulsion Laboratory. The compressor developed for the 95K program has
greatly reduced mass over the state of the art and will have 6 cubic centimeters of swept volume
and a compressor efficiency of 83%. The mass of the whole cooler will only be 4.5 kg. The first
cooler under this program will be completed and delivered to AFRL in November 2000.
NASA Ames Research Center is utilizing this AFRL / TRW development to procure an
additional unit under this contract for demonstration of the Mars liquefaction requirements
discussed in the requirements section. NASA will utilize the cooler for ground demonstrations
of the in-situ liquefaction and storage of gases simulating the Martian atmosphere in preparation
for near and mid-term follow-on flight programs to Mars.
In its present form, a number of these coolers would be required to meet the 80 K and below
cooling requirements described in Table 1. However, it is reasonable to believe that with a
development program, the technology used to create the compact and power efficient compressor
can be scaled up to much larger swept volumes. TRW has past experience in large capacity
pulse tube development and, with some development, can scale up their recent developments to
accommodate the higher capacity cooling requirements. Although this may sound like a
“simple” extension of current technology, it is not. There are significant issues in the design of a
large capacity pulse tube cryocooler. Issues such as non-linear scaling of internal design
features, heat rejection interfaces (external and internal), and regenerators may pose significant
research and development problems for cooler designers.
The Raytheon 95K High Efficiency Cryocooler (Raytheon HEC) is the latest extension of the
successful technology development under the Raytheon Protoflight Spacecraft Cryocooler
program, delivered to the Air Force and is currently in endurance evaluation at AFRL. The
Raytheon HEC cooler is poised to be a very versatile for sensor and optics cooling in addition to
many other different potential applications. The program requirements for this cooler are
identical to the TRW HEC cooler program and is being designed for 10 W at 95 K, with a
specific power of <10 WAV. The Raytheon HEC is still in development, but there is significant
potential for this technology to be scaled up to the high capacity levels. Several technical
hurdles remain in this program and a critical design review is currently scheduled for October
2000. At the time of this writing additional details on the design of the cooler are not available
for public release.
Multistage technology is developing field for the current generation of long life space
cryocoolers. Significant system benefits can be realized with multi-temperature and multi-load
capable cryocoolers. For instance, integration schemes that employ redundant multistage
cryocoolers can alleviate the need for thermal switches to lower the parasitic load due to the fact
the parasitic heat is intercepted mainly at the higher temperature, more efficient, stages of the
“on” cooler. In addition, multistage and multi-load coolers also have the benefit of reducing the
total number of coolers to meet the cooling requirements and increase the overall system
reliability. Instead of using three single stage coolers to cool multiple loads at different
temperatures, one successfully developed multistage cooler will suffice and reduce the overall
uncertainty in the system reliability.
However, there are many hurdles to the development of multistage cryocoolers. Piston –
displacer Stirling cycle coolers have the nasty problem of cryogenic clearance seals and have the
difficult machining problem of multiple diameter cylinder displacers running in a cryogenic
clearance seal. Multistage pulse tube cryocoolers are not as mature as the single stage pulse tube
coolers recently developed or are currently in development. There is significant room for
analytical modeling and research in the very complex multistage pulse tube design to improve an
often trial-and-error design and build process. The reverse Brayton multistage coolers have been
demonstrated, but not with the current technology that is being utilized in the Creare NICMOS
cryocooler or the Creare 65K Turbo Brayton, which is currently the most viable reverse Brayton
technology for long life space use. Part of the problem lies in the need for more efficient heat
exchangers and the development of the high capacity turbo alternator. Although J-T systems are
better understood, the technology has not matured for efficient use in space. Several programs,
most recently the COOLAR program6, have pursued the development of J-T systems and have
run into problems with working fluid contamination and the need for a reliable high pressure
ratio long-life compressor. The current hybrid J-T program, discussed in later paragraphs, will
utilize a different concept to precool a J-T loop with a Stirling cooler and leverage the best of the
lower temperature J-T system and the best of the higher temperature Stirling cycle pre-cooler.
However this is still in the development phase and still needs to be demonstrated.
The Ball Aerospace Stirling cycle two stage 30K and the three stage 35/60K are examples of
multistage Stirling technology. The Ball technology is an extension of the technology developed
at the Rutherford Appleton Laboratory in the United Kingdom and is one of the original
“Oxford” class cryocoolers. Ball has made numerous technology advancements in the coolers,
most notably the fixed regenerators vs. the moving regenerators of the original design. The Ball
30K cooler is currently under life test at Goddard Space Flight Center and the 35/60K cryocooler
is scheduled for delivery to the AFRL, pending the completion of a design retrofit, in July 2000.
Both of these coolers are potential candidates for future technology development for increased
multistage capacity. The most notable issues in this development are the complex machining
process for the displacers and the reduction of the induced vibration.7,8
The lower temperature cooling requirements for cryogen storage will benefit from the current
technology development for 10 Kelvin cooling. The conceptual cooling requirements at 20
Kelvin are steep in terms of total load and could impose severe penalties on future systems if the
cooler technology is inefficient and massive. At this low temperature and relatively high loads,
the technology hurdles are difficult at best. There are several current AFRL programs exploring
the low capacity applications at these low temperatures that will provide insight into the
development paths to pursue in future high capacity technology developments at 20 Kelvin.
The Astrium (formerly Matra Marconi Space) 10 Kelvin cryocooler has been delivered to
AFRL and currently is being integrated into a vacuum chamber for characterization and
endurance evaluation. This cooler has been shown to produce 0.045 Watts of cooling at 10.4
Kelvin with 198 Watts of input power (cooler mass ~32 kg).9 The system is not optimized and
has four of the standard 20 Kelvin compressors ganged to provide sufficient swept volume for a
redesigned expander cold head. The cold head is a two stage device similar to the heritage 20 K
cooler, but with additional geometry and material changes implemented to accommodate the 10
K cooling goal. Although this has been a successful proof-of-concept program, it is insufficient
to meet some of the cooling requirements that have been tentatively identified for 20 Kelvin (ref.
Table 1). This cooler can lift roughly 0.4 Watts at 20 Kelvin and to meet the potential 20 K high
capacity requirements the system would need 50 units with 10,000 Watts of input power and
1,600 kg of mass! Although the first stage of the cooler is only currently being used to cool a
shield around the second stage cold tip which is lifting a useful heat load at 10 Kelvin, there is
the possibility of utilizing this two-stage design for larger heat loads at each cooling stage.
Another promising cooler technology to reach 10 Kelvin is the Ball Aerospace 10 Kelvin
Cryocooler, which is a proof of concept cooler that will produce 0.250 Watts at 10 K. This
cooler is a J-T cycle pre-cooled by a three-stage higher temperature cooler. The J-T side of this
hybrid is also know as the “Redstone Interface” and has been reported in the literature. The pre-
cooler would only have to be interfaces and could be any technology that can provide the cooling
at the desired temperatures. The key to the successful development of this system is the
development of the long-life rotary vane compressor for the J-T loop and the development of
small, compact and highly efficient heat exchangers. The reliable and efficient operation of the
compressor and the heat exchangers are critical to the success of this program. It is possible to
see die scaled up version of this idea for the cooling of the lower temperature requirements for
cryogen storage. However, there is a corresponding increase in the size of the J-T compressor,
the heat exchangers, and the pre-cooler that must be carefully sized and designed to meet system
needs. A system demonstration of the proof of concept 0.25 Watts at 10 Kelvin is scheduled for
December 2000.
Cryogenic Systems Issues

Cryogenic integration has been developed separately from the cryocooler technology to
augment the cooling system capabilities. Past “plug-in” component level developments have
focused on potential system benefits of long cryogenic transport distances across low
temperature differences for remote location of the cryocooler without imposing heavy system
parasitic penalties, thermal storage for load leveling capability in the case of duty cycle cooling,
thermal switching for de-coupling an “off” redundant cryocooler from the system and mitigating
significant heat load parasitics, and ambient heat rejection links for the cooler (e.g. heat pipes,
capillary pumped loops, loop heat pipes, etc.). However, the recognition of designing a
cryogenic system for maximum efficiency has prompted the need for a system level view that
encompasses all of the cryogenic system components. This includes the interface to the cooled
component through the cryogenic link, the cooler, the ambient interface with the cooler, and the
ambient link to the heat rejection surface. This system level view allows for the design of more
efficient systems by developing technology that leverages and accommodates all beneficial
aspects of the various parts of the cryogenic system to meet future mission requirements.
To meet the requirements for the long-term storage of cryogens on orbit, there will probably
be large cryogenic areas that will require cooling. At present, most cryocooler technology is
capable of producing cooling at a small interface, usually However, the cooling
requirements for the large (potentially 10’s of radiation shields and large cryogen tanks will
most likely utilize distributed cooling systems to minimize the temperature differential across the
cooled area. Whether this is a cryocooler that can cool over a large surface, or is a cryocooler
that has multiple interfaces, or is a cryogenic integration bus that has multiple heat interfaces that
connects to one or more coolers is a question of system design and technology maturity.
Another consideration in the cryogenic system is the issue of cryogenic system redundancy.
It is fairly evident that future systems will require reliable system lifetimes of 15, 20, and more
years and this stringent requirement may drive system designers to employ redundant cryogenic
systems to ensure the 20+ year capability. The particular tools that system designers will employ
will vary from system concept to concept, but such technologies as thermal switches and
multiple condenser heat transport systems will serve as technology toolboxes in system design to
accommodate multiple cryocoolers integrated into the system.
There are many technologies that AFRL has developed and is currently developing for
cryogenic system integration requirements. It is logical that follow-on technology development
programs would leverage this technology development and continue the maturing of the
technology to meet the larger capacity or system design unique needs for long term stored
cryogens. The following paragraphs highlight some of the technologies that could be useful for
future long-term cryogen storage systems.
Swales Aerospace (Beltsville, MD) has been developing cryogenic system integration
technology for AFRL under the Phase II CRYOTSU Small Business Innovative Research
contract and the CRYOBUS program for almost five years. This program has produced flight
hardware that was demonstrated in the Space Shuttle STS-95 GAS canister experiment,
CRYOTSU. The AFRL technologies demonstrated were the 65K Cryogenic Thermal Storage
Unit (CRYOTSU) and the gas-gap Cryogenic Thermal Switch. As part of the joint flight
experiment, NASA GSFC also had included the Cryogenic Capillary Pumped Loop (CCPL),
designed by Cullimore and Ring and fabricated by Swales Aerospace, experiment to demonstrate
the feasibility of passive cryogenic nitrogen heat transport in micro-gravity.10
Under the Swales program, there were two different cryogenic thermal switches developed.
The first was a gas-gap thermal switch that leveraged technology, experimental results, and
proof-of-concept hardware developed by the Jet Propulsion Laboratory. The Swales switch
improved the JPL design and achieved excellent performance in ground characterization without
the hydride pump used to fill and evacuate the switch. For the flight, the original JPL hydride
pump was integrated with the flight switch and the complete system was integrated into the
payload and sent to KSC for final integration into the shuttle. Due to the short flight integration
schedule, there was inadequate time to characterize the performance of the switch with the
original JPL hydride pump and it was discovered during flight operations that the pump was
irreversibly contaminated. However, subsequent development has demonstrated that the gas gap
switch technology is viable. Swales continued the cryogenic switch development by designing a
mechnical thermal switch that is capable of producing 1.2 K/W (at 50K) resistance when on and
1400 K/W resistance when off. The device is simple and capable of repeatable cycles of
operation. Swales is currently completing the prototype development for AFRL and should
deliver by December 2000.
Also under development with Swales is the ambient Flexible Heat Transport System. This
system is designed to support the requirements for SBIRS Low to provide ambient heat rejection
across the two-axis gimbal to the optics coolers, currently under development and discussed
previously. This system will allow system designers to reject to 110+ Watts of waste heat from
the “on” cooler to the satellite radiators instead of being forced to dump the heat to an on-gimbal
radiator with its uncertain pointing attitude (it would be fixed to the gimbal) and its considerable
mass penalty. Swales has successfully completed several major milestones in this development
and will deliver a working prototype system, that will be integrated into a motorized 2-axis
gimbal mock-up, to AFRL in December 2000.
Another potential candidate technology for cryogenic heat transport is the Technology
Applications Incorporated Cryocooler Interface System. This concept utilizes a pre-cooler for a
low pressure J-T expansion loop. The loop employs a counterflow heat exchanger, expansion
valve, cryogenic load heat exchanger, an ambient compressor, and associated filters and plenum
volumes. The system design requirements are built around the technology needs for across the
gimbal cryogenic cooling for SBIRS Low. A successful development would allow system
designers to remove the relatively massive coolers and the vibration source off gimbal and allow
10-15 Watts of cooling at the fore optics to 100-120 Kelvin through the 2-axis gimbal.
Technology hurdles for this program include the development of a reliable long-life ambient
temperature compressor, a flexible transfer line system capable of 5 million plus cycles, and
adequately addressing the issue of internal gas contamination. The development is currently
under a Phase II Small Business Innovative Research contract and is scheduled to deliver a
prototype system in April 2001.
Recent New Starts

Several new programs have just recently started that have potential uses to meet the
requirements for on orbit cryogen storage. Each of these programs was started as a Phase I
Small Business Innovative Research (SBIR) proposals under the Ballistic Missile Defense
Organization topics or the Air Force topics. The programs are managed by the Air Force
Research Laboratory and seek to demonstrate proof-of-concept developments that could be
carried to Phase II and III for advanced development and demonstration.
TTH Research (Laurel, MD) has been awarded an Air Force Phase I SBIR contract to
develop a flexible cryogenic loop heat pipe (CLHP) to address cryogenic cooling requirements
across a two axis gimbal to support SBIRS Low technology needs. The passive nature of the
CLHP combined with a “self starting” pumping scheme is very attractive to the near term SBIRS
Low mission and could provide a starting point for the development of higher capacity devices to
serve as parts of a cryogenic cooling bus for advanced cryogen storage applications. The
program began in May 2000 and should complete Phase I in February 2001.
Technology Applications Incorporated (Boulder, CO) also has a new Ballistic Missile
Defense Organization (BMDO) Phase I SBIR to develop the Efficient Low Temperature
Cryocooler (ELTC) concept. This system, if successful, will provide cooling to below 45 Kelvin
with a Carnot efficiency of greater than 10%. The system employs a higher temperature, highly
efficient pre-cooler to a fluid loop that utilizes a novel expansion engine. This concept could
serve as a hybird cryocooler / cryogenic integration system to potentially provide high capacity,
distributed cryogenic cooling for large radiation shields and cryogen storage tanks for advanced
concepts. Phase I will provide demonstration hardware to prove the feasibility of the concept, a
final report, and set the stage for prototype development in Phase II by January 2001.
Another BMDO Phase I SBIR is with Sierra Lobo, Inc. (Fremont, OH) to develop a system
of producing and maintaining densified cryogens on orbit. This technology is, of course, directly
applicable to the cryogen storage application. The company will develop the methods,
procedures, and hardware to allow the storage of cryogens at higher densities. The program will
develop critical monitoring sensors for micro-gravity and develop tank structural requirements,
boil off and re-circulation techniques, and cryocooler integration techniques. This program will
complete in November 2000 and will prepare for a Phase II follow-on demonstration.
Creare Inc. (Hanover, NH) has been developing reverse Brayton technology for some time
and is currently engaged in developing the critical turboaltemator technology for high capacity
operation under an Air Force Phase I Small Business Innovative Research contract. The
program goal for Phase I is to develop the necessary component technology by February 2001 to
prepare for a Phase II closed loop cryocooler demonstration. Additional work is being pursued
by NASA Goddard Space Flight Center with Creare to develop this technology for use at 10K
and could provide the necessary leap in technology to efficiently meet much of the low
temperature cooling requirements for cryogen storage.
An alternative technology for the higher temperature, high load cooling requirements is the
development of the vapor compression cycle cryocooler for use in space. Historically, this
technology has not been exploited for space due to the complexities facing the separation of the
mixed gases (multiple component gases are needed to reach lower temperatures), the lack of an
acceptable long life compressor, and the unknown susceptibility of the system to working fluid
contamination. However, the Air Force has awarded a Phase I Small Business Innovative
Research contract to Mainstream Engineering (Rockledge, FL) to pursue a multi-refrigerant
vapor compression cycle cooler to reach as low as 77 Kelvin. Although very immature for
space, this technology may provide the technology groundwork needed to develop a long life
space cooler for the higher temperatures and capacity cooling requirements in a simple, low cost,
easily integration system for cryogen storage. The Phase I of this SBIR program should
complete in December 2000.
THE WAY FORWARD

Although a great deal of technology has been discussed, there are areas where AFRL will
investigate that are potential technology needs for future cryogen storage requirements. Issues
associated with high conductivity (ambient and cryogenic) thermal interfaces, quick disconnect
and re-connect thermal interfaces, much higher capacity cryocoolers, and long life, high pressure
ratio (DC flow) compressors are some of the technologies that need to be considered.
Technology roadmaps are currently in place to address technology development for advanced
cryogenic cooling systems to address mission concept needs out to 2010 and support such
requirements as integrated cryogenic cooling systems, advanced optics cryocooler technology,
and 35 Kelvin and 10 Kelvin cryocooling systems for Very Long Wave Infrared (VLWIR) space
based systems. Much of the technology development planned on this roadmap dovetails with the
development of many of the cryogenic system issues needed for future cryogen storage systems.
However, as general mission requirements and concepts evolve into more specific technology
needs, this roadmap can accommodate the development plan for cryogen storage technology.
As part of an overall technology development program, future Small Business Innovative
Research topics will directly address some of the needs for cryogenic systems related to cryogen
storage. Technical developments will be sought for improved cryogenic and ambient thermal
interfaces, quick disconnect cryogenic and ambient thermal systems, and advanced cryocooler
technology.
Development leveraging across the various government organizations is becoming an all-
important tool to the successful development of advanced technology. It has already been
identified that NASA has a definite requirement for long-term cryogen storage and has already
leveraged a current AFRL program for a potential candidate cooler for oxygen and methane
liquefaction and storage. AFRL will continue the long history of technology leveraging with
NASA and DoD organizations to maximize the research and development investment.
SUMMARY
AFRL is developing technology that will meet the future requirements of long term cryogen
storage in space for various mission applications. Heritage to past and current AFRL cryogenic
technology programs will allow a jump-off point for developing low mass, highly efficient
cryogenic systems that enable critical mission goals. Future programs will remain flexible to
accommodate evolving performance requirements and maintain development plans to meet
program schedules. As it has been shown in the past for cryocooler and component level
cryogenic integration technology, parallel development paths are necessary to encompass broad
system design applications. This approach will serve to maintain options for system designers
and maximize system efficiency.
ACKNOWLEDGMENTS
The authors would like to acknowledge the contributions of Dave G. Curran and Martin
Donabedian from the Aerospace Corporation, for contributions on the potential cooling
requirements and system analyses for HEL systems.
REFERENCES
1. Davis, T. M, Reilly, J., and Tomlinson, B. J., “Air Force Research Laboratory Cryocooler
Technology Development,” Cryocoolers 10, R. G. Ross, Jr., Ed., Plenum Press, New York (1999),
pp. 21-32.
2. Schuster, J. R., and Brown, N. S., “Long Term Orbital Storage of Cryogenic Propellents for
Advanced Space Transportation Missions,” conference proceedings of the 20th Annual Electronics
and Aerospace Systems Conference – Technology for Space Leadership, Washington D.C., Oct 1987.
3. Donabedian, M., “Cryogenic System Thermal Analysis for Conceptual High Energy Laser,”
Aerospace Corporation Thermal Control Department briefing, Mar 99.
4. Curran, D. G., “Use of Two Stages of Cooling to Reduce Space Based Laser (SBL) Cooling
Requirements for Both IFX and EMD Cryocooler Procurement,” Aerospace Corporation Thermal
Control Department briefing, Mar 99.
5. Kittel, P., Salerno, L. J., and Plachta, D. W., “Cryocoolers for Human and Robotic Missions to Mars,”
Cryocoolers 10, R. G. Ross, Jr., Ed., Plenum Press, New York (1999), pp. 815-821.
6. Fernandez, R., and Levenduski, R., “Flight Demonstration of the Ball Joule-Thomson Cryocooler,”
Cryocoolers 10, R. G. Ross, Jr., Ed., Plenum Press, New York (1999), pp. 449-456.
7. Berry, D., Carrington, H., Gully, W. J., Luebbert, M., and Hubbard, M., “System Test Performance
for the Ball Two-Stage Stirling Cycle Cryocooler,” Cryocoolers 9, R. G. Ross, Jr., Ed., Plenum Press,
New York (1997), pp. 69-77.
8. Gully, W. J., Carrington, H., and Kiehl, W., “Qualification Test Results for a Dual-Temperature
Stirling Cryocooler,” Cryocoolers 10, R. G . Ross, Jr., Ed., Plenum Press, New York (1999), pp. 59-
65.
9. Orlowska, A. H., Bradshaw, T. W., Scull, S., Tomlinson, B. J., “Progress Towards the Development
of a 10K Closed Cycle Cooler for Space Use,” Cryocoolers 10, R. G. Ross, Jr., Ed., Plenum Press,
New York (1999), pp. 67-76.
10. Bugby, D., Stouffer, C., Davis, T., Tomlinson, B. J., Rich, M., Ku, J., Swanson, T., and Glaister, D.,
“Development of Advanced Cryogenic Integration Solutions,” Cryocoolers 10, R. G. Ross, Jr., Ed.,
MOPITT On-Orbit Stirling Cycle Cooler
Performance
G.S. Mand1, J.R. Drummond1, D. Henry2 and J. Hackett2

1University
of Toronto
Department of Physics
Toronto, Ontario, Canada M5S 1A7
2COM DEV International
Cambridge, Ontario, Canada N1R 7H6
ABSTRACT
The Measurements of Pollution In The Troposphere (MOPITT) instrument was launched
aboard the Terra spacecraft (formerly known as EOS AM-1), from Vandenburg Air Force
Base, California on Dec 18th 1999. At present the instrument is in normal operations mode,
having undergone its outgas and activation phases.
The instrument uses a pair of Matra Marconi 50-80K Stirling Cycle Coolers interfaced to
a set of Lockheed Martin low vibration cooler drive electronics (CDE). The cooler sub-system
is mounted to the MOPITT baseplate, directly over a capillary pump heat transfer system
(CPHTS). The CPHTS allows the coolers to dump their heat via a remote radiator panel. The
displacers are in a back to back configuration and each cold tip is interfaced to a detector nest
holding sixteen indium antimonide detectors cooled to less than 100K. The coolers are also
run in a vibration cancellation mode using the CDE position digital error correction system
(PDECS), in order to minimise focal plane jitter.
This paper will describe the MOPITT cooler sub-system performance during the
instrument’s activation and normal operation sequence (March 3rd to June 1st 2000). In particu-
lar the cooler characteristics are presented. This includes cooler cold tip and detector thermal
performance, any affects of orbital temperature fluctuations on the cooler sub-system and con-
tamination (ice) build up on the cold tip and detectors. Finally, the vibration levels will be
characterised.
MOPITT is one of the first instruments to use the Matra Marconi 50-80K coolers on-orbit,
hence the data should provide an interesting insight into cooler performance and longevity.
The MOPITT instrument has been funded by the Canadian Space Agency, built by COM
DEV International and operated by the University of Toronto Instrument Operations Team
(IOT).

760 SPACE CRYOCOOLER APPLICATIONS
INTRODUCTION
The Measurements Of Pollution In The Troposphere (MOPITT) instrument will measure
some of the pollutants in the lower atmosphere, in particular the global concentrations of carbon
monoxide (CO) and methane The results will not only be used to map the global CO and
concentrations but will also be assimilated into 3-D models in order to study the chemistry
and dynamics of the lower atmosphere. CO and concentrations will be measured using
correlation spectroscopy. The CO profile measurements are made using upwelling thermal
radiance in the fundamental band. The troposphere is resolved into about four layers with
approximately 3km vertical resolution, 22km horizontal resolution and 10% accuracy. CO and
column measurements are made using reflected solar radiance in the CO and the
bands. The horizontal resolution is 22km with a 10% and 1% precision requirement
for the CO and columns respectively.
The instrument, along with a suite of four other instruments, aboard the Terra spacecraft
(formerly known as EOS AM-1) was successfully launched by an Atlas IIAS from Vandenburg
Airforce Base, California on Dec 18th 1999. After the initial spacecraft check out and outgassing
phases the instrument doors were opened on February 28th, followed by a decontamination of the
calibration targets and the cooler sub-system. The coolers were switched on on March 2nd and the
instrument brought up to nominal SCIENCE mode. Other then periodic calibration in the
activation phase the instrument has been in its nominal mode of operation since that time. In
particular the coolers have been in continuous operation.
MOPITT GENERAL INSTRUMENT DESCRIPTION

MOPITT is a scanning, nadir viewing eight channel IR radiometer. Figure 1 shows an
isometric layout. The instrument has two identical "mirror imaged" optical tables with
calibration sources, scan mirrors, choppers, modulators and cold dewar assemblies containing the
cold optics and detector packages. Each detector is a four by one array, giving four pixels in-line
along the velocity vector. This results in a 22 x 88 km instantaneous field of view (this is extended
by the use of cross track scanning). The dewar assembly is cooled by a pair of low vibration, back
to back Stirling Cycle Coolers (SCC's). The largest heat dissipating units, namely the coolers and
cooler drive electronics module, are located directly above the coldplate, other critical electronic
modules are placed close to the coldplate. The coldplate provides a stable thermal environment
and is used as the thermal sink for all modules except the main power supply module which is
thermally isolated from the baseplate and radiatively cooled to space (Figure 1).
A more complete description of the science goals, the measurement methodology and the
instrument description can be obtained elsewhere1.
Figure 1. MOPITT Isometric Layout. The baseplate dimensions are

MOPITT ON-ORBIT STIRLING COOLER PERFORMANCE 761
MOPITT COOLER AND DETECTOR SUB-SYSTEM

MOPITT makes measurements in the range using indium antimonide detectors.
For optimal performance, that is maximum signal to noise ratio, these detectors need to be
cooled to <105K. This is achieved by using a pair of 50-80K mechanical Matra Marconi Stirling
Cycle Coolers (SCC’s) interfaced to a Lockheed Martin low vibration cooler drive electronics
module (CDE).
Figure 2 shows the cooler configuration as constrained by the instrument optical layout.
MOPITT uses a pair of Matra Marconi coolers, the compressors are mounted back to back in
order to minimise the axial vibration. The compressors are also mounted such that they have a
common heat dump interface and are mounted low and close to the baseplate in order to
maximise the heat dump to the baseplate and to the coldplate system. The displacers are also
mounted in line, in a back to back configuration such that the cold tips point outwards. The in-
line configuration once again minimises vibration and the tips interface with the two detector
packages, one on each optical table. Each detector nest, shown in Figure 3, consists of four
detector packages, each of which has a four by one pixel array with cooled pre-amplifier, cold
optics and a cold filter. The detector nest is connected to the cold tip, via a shrink fit cup and
plug arrangement and vibration isolated via the use of thermal copper braids. Each nest also has
a 200 mW decontamination foil heater bonded to the inside of the main block holding the
detector capsules. Each detector nest represents a 0.8W heat load.
The cooler is driven by the Lockheed Martin CDE, incorporating the digital error correction
system (DECS) to further reduce cooler vibration to acceptable levels. The DECS system can be
run in two modes, either by using the piston position sensor (PPO) output for the feedback loop
or by using the accelerometers mounted on the body for the feedback loop. The cooler can then
be run in PDECS or ADECS mode, at present the coolers are being operated under PDECS mode
but could be switched to ADECS mode if required.
In addition to vibration cancellation, cold tip temperature control is also required so that the
coolers will cool to the desired set point. This is achieved by having cold tip 1 temperature
controlled, it’s set point being 78K. The compressor stroke is then adjusted for tip 2 to minimise
vibration.
Figure 4 shows the cooler orientation in the instrument, it should be noted that the cooler z-
axis, the piston axis, is perpendicular to the spacecraft velocity vector and the launch g-forces.
Figure 2. Cooler sub-system in its final configuration before integration

Figure 3. MOPITT detector nest. Figure 4. Cooler layout within MOPITT.
ON-ORBIT COOLER PERFORMANCE

Prior to cooler activation the detector nests were first outgassed at an elevated temperature
using the detector decontamination heaters. The decontamination cycle was as follows, both nest
heaters were switched on and after the 24 hours detector nest 1 and 2 temperatures had stablised
at 317.5K and 318.8K respectively. The detectors were maintained at this temperature for the
next 28 hours before the decontamination heaters were switched off and the coolers switched on
16 hours later.
Initial Cool Down Characteristics

The cooler sub-system was first activated on-orbit on March 2nd 2000 (TERRA operations
day 062) during a 25 minute real time spacecraft contact. All critical cooler telemetry (motor
currents, piston strokes, vibration levels etc) was carefully trended and observed at the initial turn
on and for the next 24 hours to ensure a nominal cool down case. The results of this cool down
are shown in Figures 5a-f.
Figure 5a and b show the cooler cold tip and associated detector nest cool down. Figure 5a
shows that tip 1 cools down to its set point temperature of 78K in approximately 13.5 hours and
from that point on the temperature control loop maintains that tip at a constant temperature.
Detector nest 1 follows the cool down curve and once stable there is a 12K delta between the tip
and the nest. There is a further 0.2K delta between the nest and the cold filter. Tip 2 (Figure 5b)
shows similar cool down characteristics, it cools down to 71K. Once again there is a 12K delta
between the tip and the nest temperature and the filter is a further 1K warmer then the detector
nest. Comparison with the last pre-flight cool down (April 1998) shows that the base
temperatures of the cold tips and detectors are within 2K and that the cool down time to stability
is the same to within 15 minutes. Figure 5c and 5d show the compressor piston strokes and
associated motor currents. At the start of the cool down the piston strokes are ramped up to the
maximum allowable setting, these being 5.1 mm and 4.9 mm respectively for the two
compressors (this constraint is due to the total instrument power consumption and not the
cooler). As the temperature set point is reached on tip 1 compressor 1 stroke is “backed off” to
4.6mm, resulting in compressor 2 stroke also being reduced to 4.6mm in order to maintain low
vibration.
Figure 5a. Cooler 1 cool down. Figure 5b. Cooler 2 cool down.
Figure 5c. Compressor 1 stroke and current. Figure 5d. Compressor 2 stroke and current.
Figure 5e. Compressor vibration levels. Figure 5f. Displacer vibration levels.
During the cool down phase the compressor motor currents are 0.39A and 0.42A and once
the stroke decreases the current decreases by approximately 0.02A. For this entire period the
displacer strokes are 2.1mm and 1.9mm respectively and the currents are constant. Comparison
of the strokes and current draw with pre-flight cool downs show that the strokes agree to within
0.1mm and the current agrees to within 0.01 A, once again indicating that the cooler is behaving
nominally.
Figures 5e and 5f show the compressor and displacer vibration levels respectively. For both
the compressor and displacer the PDECS controlled axial vibration levels are reduced below
0.12Nrms. The uncontrolled off axis x and y vibration levels are below 0.3Nrms for the
compressors and below 0.15Nrms for the displacers (the y-axis vibration is the larger).
Comparison of these vibration levels with pre-flight cool down data shows that the levels agree
to within 0.05Nrms. Both pre-flight and on-orbit plots show the same trends. For the
compressors the axial vibration is lowest followed by the x and y axis vibration. For the
displacers the x-axis vibration is the lowest followed by the z and y axis both of which are of
comparable magnitude.
Careful analysis of the relevant data for the first cooler turn on in orbit and comparison
with the last pre-flight cool down shows that the coolers have successfully survived launch and
are operating in a nominal fashion. All parameters were within their normal operating range and
matched closely the numbers from the last pre-flight cool down in April 1998.
Cooler Long Term Trend Characteristics

Since the initial cool down the coolers have been left in the same operational mode, that is,
they have been running uninterrupted for 100 days as of June 8th 2000. During this period all
important cooler telemetry has been monitored and trended. Trend plots of the top level
telemetry from March 3rd, following the successful cool down, to June 1st (these dates correspond
to operations day 063 to 153) are presented in Figures 6a-f.
Figure 6a shows the cooler 1 tip and detector nest thermal trend plot. The approximately
0.1K drop in the tip and nest temperatures on operations day 77 is correlated to a planned
decrease in the set point of the MOPITT coldplate system which in turn led to a 2K reduction in
the baseplate temperature. This had the effect of also reducing the CDE temperature by 2K
which in turn led to a slight change in the reference and therefore shows up as a small
temperature change. This is verified by looking at the temperature changes on operations day 79-
84 and days 129-131 when the tip temperature has gone up and returned to its nominal value. In
both cases these were calibration event time periods and the on-board calibration targets had
been heated up to 465K. This lead to a change in the baseplate and hence CDE temperatures
which once again shifted the reference values slightly and shows up as a small temperature
change. Further evidence of a CDE temperature change effect can be seen by noting the
correlation in the tip and nest temperature changes with the compressor stroke and motor current
(Fig 6c). Had this been a true temperature change there should be an anti-correlation effect
between temperature, power and stroke, that is, as temperature decreases both the stroke and
power should increase. Therefore, the perturbations in tip and nest 1 temperature can be
explained by CDE thermal changes. The overall temperature change in tip 1, from operations day
88 – 153 is 0.05K, this is as expected since this tip is controlled with a set point of 78K. The nest
temperature has increased by 0.32K and the compressor stroke and power has increased by
0.13mm and 0.01 A respectively, in order to maintain the cooler at its set point. This general
increase is because of an increasing heat load on the system due to contamination of the cold
surfaces over time.
Figures 6b and 6d show cooler 2 performance over time. Figure 6b shows that between
operations day 76 and 87 the tip temperature has decreased by 4K in about 18 hours and then
gone back up to 72K in under 6 hours (the detector nest shows a similar but slightly smaller
change). Figure 6d shows that the motor power has, as expected an anti-correlation to the
temperature, that is the current has gone up as the tip temperature went down. However, the
expected anti-correlation between the temperature and stroke is not observed. The stroke
behavior for compressor 2 is very similar to that of compressor 1 but does not follow the
compressor 2 temperature or motor current characteristics.
The change in cooler 2 performance, over this time period is being carefully investigated and as
yet no correlation between any instrument and spacecraft activity has been found. Cooler 2
performance is being carefully trended and this effect has not repeated itself (Fig 6b). The overall
temperature change in tip 2 and nest 2 since day 88 is 1K and 1.25K respectively, with a
Figure 6a. Cooler 1 thermal trend plot. Figure 6b. Cooler 2 thermal trend plot.
Figure 6c. Compressor 1 current and stroke. Figure 6d. Compressor 2 current and stroke.
Figure 6e. Compressor vibration levels. Figure 6f. Displacer vibration levels.
corresponding change in stroke and current of 0.13mm and 0.01A respectively. It should be
noted that tip 2 temperature has changed significantly more then tip 1, this is as expected since
tip 1 is controlled and tip 2 is “floating” and hence cooler 2 reacts to any changes in cooler 1
which is trying to maintain tip 1 at 78K. In both cases the nest temperatures have increased by
approximately 0.25K from the baseline tip temperatures and the strokes and motor currents have
increased by similar amounts.
The compressor and displacer vibration levels are shown in Figures 6e and 6f respectively.
The slight changes in the compressor z-axis vibration levels on days 68-71 and 79-84 can be
correlated to baseplate and hence CDE temperature changes due to the calibration targets being
hot. The more gradual, but small, increase in z-axis vibration from day 97 to 105 cannot be
correlated to compressor and displacer strokes and currents. The increase in the compressor x-
axis vibration can be correlated to the changes in tip 2 temperature and compressor 2 motor
current (Figures 6b and d). The displacer x and z-axis show the same increase in vibration level
as did the compressor z-axis through days 97-105. In fact the displacer z-axis vibration level
increases significantly more then the compressor z-axis vibration. The displacer y-axis vibration
changed gradually from days 63- 88 and since then has remained fairly constant. It should also
be noted that the compressor z-axis and the displacer z and x-axis have a larger range of value
between days 110-130. It is almost as if the vibration levels were changing between two states
before settling down to the new values.
The cooler performance over the first hundred days of operation has been trended and
presented. Some artifacts of the trend plots can be explained and correlated to instrument events
and activities. However the change in tip 2 temperature on days 67-87 and the change in some of
the cooler vibration levels on days 97-105 are as yet unexplained.
Cooler Short Term Trend Characteristics

The coolers are mounted on the instrument baseplate and reject heat via the baseplate to the
coldplate, in fact the coolers being the largest heat dissipating units in MOPITT are mounted
directly over the coldplate. The coldplate operates on a capillary pump heat transfer system using
a saturated ammonia loop and dumps the instrument heat to a remote space view radiator. This
coolant loop also allows a reservoir set point temperature to be maintained. The use of such a
coldplate system ensures that the instrument is thermally stable over the orbit time and hence
fluctuations in the cooler tip temperatures due to fluctuations in the heat sink temperature are
expected to be small over these time scales.
The orbital variation on the instrument baseplate and the compressor and displacer body
temperatures is approximately 0.1K. The resulting variation in the tip temperatures is shown in
Figure 7a. As expected there is no orbital component in tip 1 temperature (0.01K) since that is
under the temperature control loop, however, tip 2 which is floating, shows an orbital variation
of approximately 0.04K. It is interesting to note that although there is no discernable orbital
variation in tip 1, the four detectors on that cooler show an orbital temperature variation ranging
from 0.04K to 0.1K (Figure 7b). The four detectors in nest 2 show a similar variation. At present
it is assumed that this additional component may be due from the orbital variation in flux falling
on the detectors acting as an additional load. It may also be due to the orbital changes in the heat
leaks into the dewar acting as a load (eg dewar window temperature changes over the orbit).
In addition to the orbital variation, tip and detector nest 2 shows a longer term,
approximately 26 hour temperature oscillation. Figure 7c shows that the tip temperature varies
by approximately 0.08K over this time period.
Cooler Cold Tip Contamination

As has been seen above (Figures 6a-d) the tip and nest temperatures are gradually increasing
as are the strokes and motor currents. This implies that the load on the cooler is gradually
increasing due to contaminant build up on the cold surfaces. This in fact is most likely due to
water condensing and icing on these surfaces. As a result of this build up the coolers have to be
Figure 7a. Tip temperature orbital fluctuation. Figure 7b. Nest 1 orbital fluctuations.
Figure 7c. Longer term tip 2 variation. Figure 8. MOPITT gain trend plot.
periodically switched off and the tip and detector package outgassed by using the
decontamination heaters on each nest. From the cooler perspective the cooler is being run in a
“backed off” mode to achieve the desired nest temperatures and has plenty of margin to offset
the effect of contamination and hence increase the time between outgas cycles. However, from
the science data point of view the detector nest packages contain cold filters and optics and as
these occlude optical transmission decreases and as a result the system gain also decreases. The
change in channel gain is being used as a more precise indicator of the contamination than the
cooler characteristics (this was done pre-flight and the gain returned to the original levels once
an outgas cycle had been conducted).
Figure 8 shows the gain change for channels 1 and 3 (two of the most sensitive channels). It
shows that the gain is decreasing at a rate of 0.06%/day corresponding to a 1% gain change every
seventeen days. To date the gain has decreased by about 6%. Since MOPITT conducts a two
point gain measurement every eleven minutes this long term gain drift is easily compensated for
over these time periods.
The baseline MOPITT instrument team criteria for cooler decontamination is a gain decrease of
8-10%, this implies that the first decontamination cycle will occur approximately 150 days after
initial turn on.
CONCLUSION
The on-orbit cooler performance results from activation on March 2nd to June 1st 2000 have
been presented. It has been shown that the initial cool down occurred in a nominal fashion, all
data was within limits and in close agreement with the last pre-flight cool down in April 1998.
Furthermore, the long term trend plots show the cooler is behaving as expected with one or two
as yet unexplained single time events. At the orbital time scale the variations are <0.1K mainly
due to the coldplate system providing a stable thermal platform. It has also been shown that the
time between outgas cycles will be a minimum of five months and as time progresses this could
extend to a decontamination every ten months or more.
To date the coolers are operating nominally and within limits, the coolers have been running
uninterrupted for over a 100 days (2400hrs).
ACKNOWLEDGEMENTS
The MOPITT instrument has been funded by the Canadian Space Agency, built by COM
DEV International and operated by the University of Toronto Instrument Operations Team.
REFERENCES
1. Drummond, J. R., MOPITT Mission Description Document, 1996.
HIRDLS Instrument Flight Cryocooler Subsystem
Integration and Acceptance Testing
W. Kiehl, D.J. Berry, D.S. Glaister, J. Richards and R.G. Stack

ABSTRACT
Ball Aerospace has been performing integration and acceptance of its flight Stirling-cycle
mechanical cryocooler for the High Resolution Dynamic Limb Sounder (HIRDLS), an
instrument slated to fly on the Earth Observing System Chemistry Platform. The HIRDLS cooler
subsystem (CSS), developed under subcontract for the NASA Goddard Space Flight Center Earth
Observing System Program, consists of a sophisticated and highly reliable, single-stage, fixed
regenerator, Stirling cryocooler and its drive electronics, interfacing hardware, cryocooler support
bracketry, and the cryocooler radiator.
The HIRDLS CSS provides 60 K cooling to the vibration isolated detector subsystem while
the cooler is supported in a heat rejecting radiator attached firmly to the instrument structure. The
electronics with embedded software are mounted on another panel of the instrument structure.
The engineering model was delivered in July 1999, and the flight model is scheduled for
delivery in July 2000.
INTRODUCTION
Ball has built and is integrating its single-stage, Stirling-cycle mechanical cryocooler onto
the High Resolution Dynamic Limb Sounder (HIRDLS) instrument slated to fly on the Earth
Observing System Chemistry Platform. This program, funded by the Goddard Space Flight
Center (GSFC) Earth Observing System Program, is leveraged off the NASA 30 K Phase IV
cryocooler that is currently under life test at GSFC.
HIRDLS Cryocooler Description

The HIRDLS cryocooler consists of a compressor, displacer, and a cooler control unit. These
are shown in Figures 1 and 2. The compressor is a dual package of opposed, Oxford style, linear
motors. The displacer is a single-stage, fixed regenerator cold head.
The cooler control unit contains the redundant power converters, pulse-width modulation
motor drivers, a linear motor drive for the displacer, control electronics, and data collection and
telemetry electronics. Power is received from the instrument power supply via the spacecraft
noisy bus. Cryocooler operational data is relayed to the instrument processor on an RS-422
interface. It is radiation hardened to 30 krad. Redundant launch locks have been successfully
tested to a random level of 14.1-g root mean squared and a 15-g sine wave.

Figure 1. The engineering model of the HIRDLS cryocooler is shown mounted in the radiator
during launch vibration testing.
The cooler control unit is packaged in an aluminum chassis for heat rejection and
electromagnetic interference control. Worst-case analysis shows a chassis temperature of 35 °C.
Flight software provides all the control and telemetry necessary to operate the cryocooler.
Motor frequency and phase between the motors are user selectable. The amplitude of the motor
strokes are either manually set or automatically set by a temperature control feedback loop.
Performance
In tests of the flight unit at 39 Hz and rejecting to 300 K heat sink, the HIRDLS cryocooler
lifted 1 W at 60 K for 65 W total power, including flight electronics operating at 75.3% of total
stroke. Exported vibration was measured on a 6-axis dynamometer under full operating
conditions. Vibration was measured on the compressor and displacer separately. With no active
vibration cancellation, the measured levels up to 10 harmonics were all well under the
specification of the HIRDLS instrument.
Figure 2. The cooler control unit is a compact electronics package.

HIRDLS FLIGHT CRYOCOOLER SUBSYSTEM I&T 771
HIRDLS Cryogenic Subsystem Description

The HIRDLS cryocooler is mounted directly to the ambient radiator to reduce integration
complexity and to reduce the total mass of the system. The weight-relieved radiator panel
provides the necessary waste heat rejection while also acting as the structural mount for the
cryocooler. The close proximity between the radiator and the cryocooler does away with the
massive thermal bus that would otherwise be necessary. The radiator mounts to the spacecraft
through four blade-style flexures to mitigate effects of thermal expansion differences between the
bus and the radiator, and the instrument-to-radiator heat leak path. The photograph of the
HIRDLS cryocooler engineering model mounted in the radiator is shown in Figure 1. The mass
of the cryogenic subsystem (including electronics) is approximately 27 kg.
A cryovac housing surrounds the cold finger. During integration this mates with the detector
subsystem. Before integration, the cryovac housing is capped off so the subsystem can be
independently tested.
HIRDLS Cold Transport, Insulation, and Support System

HIRDLS uses a very efficient system to provide flexible heat transport to the detector
assembly, conductive and radiative insulation, and structural support for launch vibration loads.
The system is illustrated in Figures 3 and 4. The system uses multi-layer insulation (MLI)
blankets; high-strength, low-conductivity Kevlar tension straps; a low-mass, low-temperature-
drop conduction rod; and a flexible thermal strap, or S-link. The heat from the detector is
transported to the cooler with minimal temperature drop while maintaining low mass and
flexibility for integration and vibration isolation. The cold link mass is held rigidly to minimize
its movement and induced loads during launch vibration while adding minimal additional heat
load parasitics.
As shown in the figures, heat is transported from the detector assembly, through a flexible S-
link, a solid cold rod, and a cold tip, to the cryocooler. The S-link is composed of multiple layers
of aluminum foil, the cold rod is made of a beryllium alloy, and the cold tip is made of copper.
Including interfaces, the overall temperature drop over a distance of about 6 inches, from the
detector to the cryocooler, was tested at less than 2 K for a heat load near 1 W @ 60 K.
Figure 3. High-strength, low-conductivity Kevlar strap configuration as attached to HIRDLS

cryocooler cold tip without the MLI or additional cold links.
Figure 4. Schematic cutaway of HIRDLS cryocooler with straps, support tube, cold rod, S-link, and
MLI blankets.
The mass of the cold link assembly is supported off a titanium support tube by eight Kevlar
straps. The support tube base attaches to the cryocooler displacer baseplate, which is at ambient
temperature. The Kevlar straps are mounted in an opposing manner and pre-loaded to hold the
cold tip. Each strap can hold over 500 lbsf of tension with minimal cross-sectional area and
associated conduction parasitics. This support then insulates the thin-walled cryocooler displacer
tube walls from side loads induced by the movement of the cold link assembly during launch.
Figure 5. Representation of the HIRDLS cryogenic subsystem mounted on the instrument payload.
HIRDLS FLIGHT CRYOCOOLER SUBSYSTEM I&T 773
A three-part, seven-layer MLI blanket covers the cryogenic hardware including the displacer
and a portion of the detector assembly. For the engineering model, where only four Kevlar straps
were used, the overall effective emittance of the blanket was less than 0.02 based on model
correlations to test. Four more straps were added to the flight model to reduce structural loads
during launch. With the additional strap penetrations, the blanket effective emittance is about
0.025.
System Integration
The aim of this design was to facilitate system integration by developing a modular
subsystem that can be tested as a unit and then easily integrated into the spacecraft. Figure 5
shows a representation of the HIRDLS cryogenic subsystem mounted on the instrument payload.
With a system that can be easily integrated and tested as a unit, we have provided the HIRDLS
customer with a low-risk approach to a cryocooler system.
SUMMARY
Ball has developed a modular, easily integratable cryogenic system for the HIRDLS
instrument.
ACKNOWLEDGMENT
We would like to thank Brenda Costanzo of Lockheed Martin Space Systems Company for
her support, encouragement, and understanding during the completion of this project.
Low-Temperature, Low-Vibration Cryocooler for
Next Generation Space Telescope Instruments
R.L. Oonk, D.S. Glaister, W.J. Gully, and M.D. Lieber

ABSTRACT
Proposed mid-wave infrared instruments for NASA’s Next Generation Space Telescope
(NGST) all use focal planes cooled to temperatures ranging from 6-8 K. This is well below the
30 K passive cooling capabilities of NGST. For these low-temperature focal planes, low-
vibration active cryocooling will be needed. Ball is currently developing a Joule-Thomson (J-T)
cryocooler that reaches 10 K or lower for the Air Force. This cryocooler features an oil-free, very
low-vibration, rotary vane helium compressor with an operating lifetime well in excess of 10
years with no maintenance. This cryocooler development program, along with past technology
developments at Ball, serves as the basis for a 6-8 K cryocooler to be used for NGST
instruments. The NGST cryocooler is a helium/hydrogen cascaded J-T system that takes
maximum advantage of the unique thermal environment available on NGST to maximize power
efficiency, while providing vibration-free cooling at the focal plane interface.
The cryocooler’s design is highly leveraged off of the Air Force 10 K design. The hydrogen
and helium compressors are scaled versions of the 10 K rotary vane compressor. The heat
exchangers are also scaled versions of the designs used on the 10 K cryocooler system. The
electronics are simple commercial-based brushless dc motor controllers (without processor and
minimal overhead).
INTRODUCTION
The Next Generation Space Telescope (NGST), shown in Figure 1, is a space-based
astronomical observatory currently planned for launch in 2008. The basic characteristics of
NGST reflect what is required to meet its scientific objectives. The baseline wavelength range of
NGST will be from to The most important targets NGST will observe are
exceedingly faint. To accomplish this, NGST will need the greatest possible sensitivity, which
necessitates a large, 8-m collecting area. NGST also must be precisely pointed, which results in
its very low tolerance to jitter.
To obtain the very low background needed for its infrared observations, the mirrors and
focal planes of NGST must operate at low temperature. To accomplish this, NGST will be flown
in a heliocentric orbit to allow maximum use of passive radiant cooling. The entire NGST

Figure 1. The Next Generation Space Telescope (NGST) is a space-based infrared observatory
scheduled for launch in 2008.
observatory is designed to be a giant passive radiator. While the solar arrays and spacecraft bus
run at 250-300 K, the long boom and large, staged sunshades thermally isolate these warm
assemblies from the telescope and integrated science instrument module (ISIM), enabling these
parts of the observatory to operate at 30-35 K.
Although it is not currently part of the baseline mission, extension of NGST’s wavelength
coverage to or longer in the mid-infrared will probably be added in the near future.
Adding a mid-wave instrument for observations at these longer wavelengths will require focal
plane cooling to 6-8 K, well below what is possible with passive radiant cooling.
NGST INSTRUMENT COOLING REQUIREMENTS

Table 1 summarizes the cooling requirements for an NGST mid-wave infrared instrument.
The focal plane temperature is driven by the requirement to observe in the wavelength
range. This also drives the requirement for the cold shield temperature.
Perhaps the most stringent requirement for NGST is the allowable line-of-sight (LOS)
disturbance. This requirement probably will preclude the use of any vibration-inducing
mechanisms, such as linear motor driven compressors, in the vicinity of the ISIM. Any
mechanical compressor will probably be located remotely from the ISIM, either on the payload
tower or the spacecraft bus, in order to meet the LOS jitter requirement. In addition, it may need
to be mechanically isolated, in a fashion similar to the reaction wheels (also located on the
spacecraft bus) used for the observatory’s pointing control.
NGST’s passive cooling capabilities also provide potential benefits to a cryocooler. The
ISIM has the capability to reject a small amount of waste heat from a cryocooler at 30-35 K. In
addition, space is available along the support boom to add radiators capable of rejecting larger
amounts of heat at temperatures between 40 and 200 K.
LOW-TEMPERATURE CRYOCOOLER FOR NGST 777
POTENTIAL APPROACHES
Potential approaches for cryocooling an NGST mid-wave infrared instrument include:
• Stored cryogen or LHe)
• Reverse turbo-Brayton
• Sorption Joule-Thomson (J-T)
• Stirling/J-T hybrid
• Two-stage helium/hydrogen J-T (the focus of this paper)
Each approach is discussed briefly below.
Stored Cryogen. The low heat load and low temperature environment afforded by the
radiant cooling make a stored cryogen system feasible for NGST mid-wave infrared focal plane
cooling, at least for the 5-year minimum lifetime requirement. A stored cryogen system also has
the advantages of zero vibration, zero power consumption, and high reliability. However, stored
cryogens always have a finite life, making them less attractive than a closed-cycle system.
Reverse Turbo-Brayton. This approach, under development by Creare, and described by
Swift and Zagarola,1 can meet the allowable LOS disturbance. Because it is a recuperative cycle,
it can be integrated into NGST so that the turbo compressor is located near the warm end of the
observatory, whereas the cold turbo expander is located inside the ISIM. However, turbo-Brayton
cryocoolers are typically less efficient than other closed-cycle approaches due to the inherent
inefficiencies of operating turbo-machinery at very low heat lifts. Also, reliable operation of the
turbo-expander at 6-8 K remains to be proven.
Sorption Joule-Thomson. Like the reverse turbo-Brayton, the sorption J-T cryocooler,
under development at JPL and described by Wade et al.,2 can meet the allowable LOS
disturbance. Also like the turbo-Brayton, it is a recuperative cycle, so it can be readily integrated
into the observatory. The disadvantages of the sorption J-T system are the numerous valves
required, which leads to reliability concerns, plus the inherently low thermodynamic efficiency of
the thermally driven compressors.
Stirling/J-T Hybrid. In this approach, a Stirling (or pulse-tube) cryocooler is used as a pre-
cooler for a helium J-T cooler. This approach has the advantage of using a relatively mature
Stirling cooler in a temperature range where it is reasonably efficient (~20 K), coupled to a
sorption or mechanically driven helium J-T cooler in its most efficient temperature range. The
disadvantage of this approach is the difficulty of integrating the Stirling cooler into the
observatory. Isolating the telescope from the vibration produced by the Stirling compressor
forces the latter to be located remotely from the ISIM. In addition, the high-power dissipation of
the compressor also forces it to be located far away from the 30-35 K ISIM. The only alternative
is to locate the Stirling cooler near the warm end of the observatory, but this results in 10-15
meters of cold J-T loop plumbing between the Stirling cold finger and the ISIM, which
dramatically increases the parasitic heat leak into the system.
Two-Stage Helium/Hydrogen J-T. This approach uses a classic, cascaded J-T system using
a hydrogen loop pre-cooling a helium loop. Unlike a typical ground-based application that
requires a nitrogen loop to pre-cool the hydrogen below its inversion temperature, active pre-
cooling of the hydrogen loop is not needed, because it is provided by the observatory’s passive
radiant cooling. The approach uses mechanically driven compressors, described in more detail
below, for both the helium and hydrogen loops. Like the turbo-Brayton and sorption J-T systems,
the helium/hydrogen J-T approach is a recuperative cycle, so it can be readily integrated into the
observatory. It is also more thermodynamically efficient than the turbo-Brayton and sorption J-T
systems. Because it relies on mechanically driven compressors that produce vibration, meeting
the LOS disturbance requirements must be demonstrated.
CRYOCOOLER DESIGN
As described in detail below, the design of the helium/hydrogen cryocooler for NGST is a
highly leveraged approach, and is based heavily on prior and current Ball cryocooler
development programs. Figure 2 is a schematic of the helium/hydrogen J-T cryocooler. Figure 3
shows how the cryocooler is integrated into the NGST observatory. Focal plane cooling at 6-8 K
is provided by the helium loop. Cold shield cooling and pre-cooling of the helium loop is
provided by the hydrogen loop. Both loops are heat sunk to the ISIM radiator, operating at 33 K.
The helium and hydrogen compressors are located at the warm end of the observatory, and reject
heat to a radiator operating between 220 and 270 K.
The designs of the He and compressors are derivatives of the compressor Ball is
developing for the Air Force Research Laboratory (AFRL) 10 K Cryocooler Development
Program, which is described in more detail by Glaister et al.3 Figure 4 shows the design layout
of the 10 K Cryocooler rotary vane compressor. The two-stage compressor features sliding vanes
that have a demonstrated operating life in excess of 10 years. Performance testing of this
compressor is scheduled for July 2000, which will verify both thermodynamic performance and
exported vibration characteristics of the unit.
For the NGST application, which requires 10 mW of cooling at 6 K, the required helium
compressor is approximately a 65-percent scale version of the AFRL 10 K unit. It is a two-stage
unit, with inlet pressure of 40 psia, outlet pressure of 80 psia, and a mass flow rate of 8.6 mg/s.
Due to its lower mass flow rate, it is predicted to have seven times lower induced vibration than
the AFRL 10 K unit. The required hydrogen compressor required is approximately a 135-percent
Figure 2. Helium/hydrogen Joule-Thomson cryocooler takes maximum advantage of NGST

thermal environment (for 10 mW at 6 K).
Figure 3. Helium/hydrogen cryocooler readily integrates into the NGST observatory.
Figure 4. 10 K Cryocooler two-stage helium compressor.
scale version of the AFRL 10 K unit. It is also a two-stage unit with inlet pressure of 4.5 psia,
outlet pressure of 9 psia, and a mass flow rate of 9.8 mg/s. Due to its mass flow rate and lower
absolute operating pressures, it is predicted to have an order of magnitude lower induced
vibration than the AFRL 10 K unit.
The design of the heat exchangers, J-T valves, and contamination control assemblies will
build on J-T cryocooler development done at Ball since 1982. The most notable result of this
development was the Cryogenic On-Orbit Long-Life Active Refrigerator (COOLLAR) program.
This government-funded program developed a space-qualified two-stage nitrogen-based J-T
cryocooler, and successfully demonstrated on-orbit operation during space shuttle mission
Figure 5. COOLLAR space shuttle demonstration flight in August 1997 successfully demonstrated
the cold head technologies needed for the NGST cryocooler.
STS-85 in August 1997 (Figure 5). The performance of the oil-lubricated compressor, non-
plugging J-T valves, heat exchangers, and contamination control system were all successfully
demonstrated during 300 hours of continuous on-orbit operation.
The overall features and benefits of the helium/hydrogen J-T cryocooler for the NGST
application are summarized in Table 2.
PERFORMANCE
As summarized in Table 1, the cryocooler meets or exceeds the estimated
requirements for NGST with minimal power and mass penalties relative to alternative cooling
technologies. Table 3 provides a more detailed summary of the cooler system for a 10 mW load
at 6 K, including individual component masses and sizes. A similar detailed design has also been
developed for a system having 20 mW cooling capacity at 6 K.
LOS disturbance is perhaps the most stressing NGST requirement for a mechanical
cryocooler. We have estimated the induced vibration from the NGST He and compressors,
using a detailed compressor model developed for the 10 K Cryocooler Program. The exported
vibration is driven by the compressor torque, because there are minimal linear forces generated
by the rotary vane compressors. Figure 6 provides a time history plot for torque for the first and
second stages of the helium compressor, which has higher levels than the hydrogen unit. These
predicted torque disturbances were input into an NGST opto-mechanical structural model to
yield the LOS disturbances. Using conservative assumptions for the coincidence of cooler peak
frequencies to NGST harmonics, the model predicts that the jitter requirements are marginally
exceeded with the compressors hard-mounted to the payload tower. However, the model also
Figure 6. Predicted torque variations from the first and second stages of the NGST He rotary vane
compressor. These torque disturbances were input into an NGST optomechanical structural model to
yield the LOS disturbances.
predicts that the disturbances are easily reduced below the jitter requirements by using passive
isolation mounts similar to those baselined for the spacecraft reaction wheels.
SUMMARY
We have developed a conceptual design of a helium/hydrogen J-T cryocooler that can
provide 6-8 K focal plane cooling for proposed NGST mid-wave infrared instruments. The
cryocooler takes maximum advantage of the unique thermal environment available on NGST to
maximize power efficiency, while providing vibration-free cooling at the focal plane interface.
Our approach uses a classic, cascaded J-T system, with a passive radiator pre-cooling the first-
stage hydrogen loop, and the hydrogen loop pre-cooling the second-stage helium loop. This
cryocooler can be readily integrated into the observatory, and is more thermodynamically
efficient than alternative cryocooling approaches. Our modeling has also shown that LOS
disturbances caused by the compressor’s exported vibration are well below NGST limits.
The J-T cryocooler design is highly leveraged off of prior cryocooler designs, giving
high confidence that the predicted performance can be realized in actual hardware. The hydrogen
and helium compressors are scaled versions of a previously developed rotary vane compressor.
The heat exchangers are also scaled versions of the designs used on the prior J-T cryocooler
systems.
REFERENCES
1. Swift, W.L, and Zagarola, M.V., “Turbo Brayton Cryocooler for NGST,” presentation at Woods
Hole-Hyannis NGST Technology Conference (available at
http://www.ngst.nasa.gov/public_docs.html) (September 1999).
2. Wade, L.A., Lindensmith, C.A., Bowman, R.C., Paine, C., and Crumb, D., “Vibration-Free Cooling
of Infrared Astrophysics Instruments,” presentation at Woods Hole-Hyannis NGST Technology
Conference (available at http://www.ngst.nasa.gov/public_docs.html) (September 1999).
3. Glaister, D.S., Gully, W.J., Wright, G.P., Simmons, D.W., and Tomlinson, B.J., “A 10 K Cryocooler
for Space Applications,” paper #124 to be presented at the 11th International Cryocooler Conference,
Keystone, Colorado (June 2000).
Considerations in Applying Open Cycle
J-T Cryostats to Cryosurgery
R.C. Longsworth
IGC-APD Cryogenics
Allentown, PA 18103-4783
ABSTRACT
Some of the same characteristics of open cycle Joule Thompson (JT) cryostats that have
made them attractive for military applications also make them attractive for cryosurgical applica-
tions. These include their small size, ability to cool fast, long term storage, and potentially small
and flexible connecting lines. One characteristic that is different is the nature of the heat load.
Cooling tissue requires a high initial cooling rate but the load decreases as the ice ball grows
because of the low thermal conductivity of tissue. Characteristics of ice ball formation in tissue
including temperature distribution, heat flux, and fraction cooled below -20°C have been calcu-
lated for different size cylindrical and spherical cryoprobes and different probe temperatures.
Cooling characteristics of different gases are discussed including gas-cooling effects in
bottles, flow rates and cooling rates. Maximum refrigeration rates at 40 MPa, are presented for
Ar and kr in finned-tube and matrix-tube heat exchangers in the size range of interest for
cryoprobes. The relative merits of finned-tube and matrix-tube heat exchangers are discussed.
Test data is presented for ice ball formation in beef liver using a 3.45 mm diameter by 19
mm long cold tip probe cooled by 2.5 L Ar at an initial pressure of 42 MPa using a finned-tube
JT heat exchanger.
INTRODUCTION
There has been a continual exploration of cryosurgical procedures and equipment since the
mid 1960s.1 Some of the recent advances have resulted from the ability to monitor the formation
of the ice ball by ultrasound and MRI. Cryosurgical procedures have been developed for a very
wide range of applications that require different size cryoprobes having different cooling rates
and temperatures. A cryogenic catheter for treating heart arrhythmia2 requires a very small cold
tip with small flexible cryogen supply/vent tubes. One advantage of cryoprobes in treating
neurological problems is that the tissue can be cooled to the point where the electrical function is
interrupted but the tissue is not killed. If the temperature of the probe can be controlled then the
correct position of the probe can be checked before the temperature is reduced to the point where
tissue is killed.
The JT cryoprobe developed and reported in this paper was designed to freeze tumors in the

784 COMMERCIAL CRYOCOOLER APPLICATIONS
prostate gland but the requirements for tumors in other organs such as the liver are similar.
Cryoprobe requirements for each case can vary in terms of the number that might be grouped
together, their diameter, cold zone length and cooling capacity.
Research work is on going in understanding the mechanisms by which cryosurgery kills
tissue, and differences for different tissues. Following is a summary of procedures that cause cell
death1,
• Cooling to < -20 C, (-40 C preferred)
• Slow rate of warming
• Multiple freeze/thaw cycles
Tissue death due to the disruption of blood flow in the region outside the –20 C zone, and the
stimulation of an immune response due to freezing3, have also been studied.
APD Cryogenics has been making small JT heat exchangers since the early ‘60s to cool
infrared detectors in military applications. We have recently reduced the involvement in military
applications of this technology and have been investigating applications to cryosurgery. Finned-
tube heat exchangers have been the standard for these systems but matrix-tube heat exchangers4
which we developed in the mid ‘80s offer another option. JT cooling has the same advantages in
cryosurgery relative to closed cycle coolers or liquid cryogens as it does in military applications.
These are:
• Use of pressurized gas bottles that can be stored for long periods of time.
• Very small heat exchangers.
• Very fast cool down.
• High heat transfer rates.
• Can be run without electricity.
JT CRYOPROBE
Gases such as and can provide cooling at their normal boiling points of 195 K (-78
C) and 185 K (-88 C) respectively by direct expansion from a pressurized bottle. Gases such as
and require a counterflow heat exchanger to precool the gas in order to
reach their normal boiling point (NBP) temperatures. The JT cooling effect increases with
pressure for these gases up to about 40 MPa (6,000 psi) above which it decreases. The total JT
cooling that is available per L of gas, assuming no losses in the heat exchanger, as the gas bottle
empties to atmospheric pressure is higher for gases with higher boiling points. Table 1
summarizes the JT cooling/L at 40 MPa and 300 K and the NBP for each of these gases. Another
relationship is that the size of the heat exchanger needed for efficient JT cooling is greater for
low NBP gases than high NBP gases. Ar and Kr are selected for analysis in this paper because
they are inert and have favorable JT cooling characteristics.
APPLYING OPEN CYCLE J-T CRYOSTATS TO CRYOSURGERY 785
Figure 1 Construction of experimental finned-tube heat exchanger cryoprobe with thermocouples

(upper) and a matrix-tube heat exchanger (lower).
An experimental cryoprobe designed for use with Ar was made as shown in figure 1 (top).
High pressure gas flows through the coiled finned-tube then through a small diameter nozzle tube
and emerges as a gas/liquid spray at the tip of the probe. It then cools a 19 mm long zone before
reentering the heat exchanger. Cold gas flowing through the fins cools the high pressure gas in a
counter flow heat transfer relationship. The location of thermocouple temperature sensors that
were used in the tests are shown. The one at the tip would be incorporated in a standard
cryoprobe. This design is unique in that it has a second supply tube connected at the inlet end to
a low pressure (< 1 MPa) source of Ar through a valve and to the heat exchanger mandrel. While
the high pressure supply valve is open for cooling the low pressure valve is closed. After closing
the high pressure supply valve the low pressure valve is opened and there is a reverse flow of gas
through the cold zone and heat exchanger fins for fast warm up5. Also shown in figure 1
(bottom) is a cryoprobe with a matrix-tube heat exchanger. High pressure gas flows through the
tube and low pressure return gas through the matrix which serves the same heat transfer function
as the fins. The matrix-tube heat exchanger has the high pressure gas which is warm in the tube
on the outside while the finned tube heat exchanger has the colder low pressure gas on the
outside. This may be an important difference in some applications.
Tests were run with the finned-tube cryoprobe inserted > 50 mm into beef liver which was
placed in a glass beaker and maintained at 310 K (37 C) by surrounding the beaker with a
temperature controlled water bath. Four thermocouples were inserted in small diameter stainless
steel tubes so they projected out the ends. The tubes were held in a bracket that spaced them at 5
mm increments radially from the centerline of the cryoprobe. They were then inserted into the
liver so they were opposite the mid point of the cold zone.
Figure 2 shows temperatures recorded from the start of Ar flowing through the JT heat
exchanger from a 2.5 L bottle charged to 42 Mpa. Temperature of the liquid Ar inside the probe
dropped below 105 K in < 15 s. The thermocouples on the outside of the probe cooled to < 114 K
in < 22 s at the tip, < 122 K in < 60 s in the middle, and < 135 K in < 75 s at the upper end of the
cold zone. The delay in cooldown along the cold end reflects a cooling of the cryoprobe and a
heat load from the liver that exceeds the cryostat capacity initially. The heat transfer rate is
highest at the cold tip where the turbulence and liquid fraction are highest. This cold end was
designed with a small radial gap to promote good heat transfer but the test results show that some
additional internal surface is needed to reduce the temperature difference between the liquid in
the tip and the outside surface of the probe.
Figure 2 Temperatures recorded from thermocouples during test of experimental cryoprobe in

beef liver, 2.5 L, 42 MPa, Ar, 310 K bath temperature.
The temperature in the liver at 5 mm drops below 203 K (-70 C) relatively fast but never gets
below 183 K. About 7 minutes after starting the flow of gas the heat load exceeds the JT cooling
capacity and the cold end warms up starting at the end farthest from the tip. The ice ball reaches
a maximum radius of about 17 mm at approximately 14 minutes. The tip is still cold at 16
minutes when the high pressure supply valve is closed and the warm up valve is opened. Low
pressure Ar at room temperature flows to the tip first, then through the cold zone. One minute
after starting the warm up the thermocouples on the cold end are in the range of 257 K to 262 K
and warming slowly. Warming may be controlled by adjusting the flow rate or by heating the
warm up gas. Pressure in the supply bottle dropped from 42.1 MPa to 19.9 Mpa during the test.
Flow rate and JT cooling will be discussed later.
ICE BALL MODEL

Models have been developed for spherical and cylindrical ice balls using a spread sheet that
assumes a cryoprobe of radius Ri surrounded by layers having a uniform thickness, dR. A
column is used for each layer which is added incrementally. The first rows of the spread sheet
calculate the surface areas for each increment using the average thickness of the outer layer and
the heat flow rate to the outer layer using a heat transfer coefficient, h The outer shell is
assumed to freeze at 273 K as a result of heat flowing to the next inner ring by conduction, k
(W/cm K). Both of these coefficients are determined from experiment and are different for
different tissues and different patients. The model assumes a constant value of k but k is actually
temperature dependent. A probe temperature, Tc, is assumed and the heat flow rate from the
outer shell at 273 K to the probe is calculated. A maximum cooling rate for the cryostat can be
entered and the next row can list the lesser heat flow rate. Using this heat flow rate, the steady
state temperature can be calculated for each layer with successive rows representing one
additional layer. The probe temperature will be > Tc if the cryostat capacity is less than the heat
flow rate calculated for Tc.
A separate block of rows is used to calculate the heat that has to be removed from each layer
as a new layer is added. A value of 400 J/g is used for the outer layer to account for the cooling
from 310 K to 273 K and freezing. A value of 2.0 J/g K is used for the specific heat of the solid.
The heat to be removed to form each new layer is summed, then divided by the cooling rate to
determine how long it takes to form each new layer. These time increments are added to get the
elapsed time. The cooling capacity of the cryostat is calculated as a function of this time. The
model works for the last layer that forms even if the JT cooling has dropped below the heat
influx rate and the probe temperature increases. The model has not been extended past this point
of the warm up phase.
Figure 3 presents four plots from the spreadsheet for a cylindrical probe of the same diameter
as the experimental probe shown in figure 1, assuming a surface temperature of 100 K.
Calculated JT cooling during the test and temperature data are plotted along with calculated
values from the model. Similar plots were done for the spherical model. The length of the
cylindrical probe and the diameter of the spherical probe were varied along with the h and k
values of the tissue to see what values best matched the test data. It had been observed when
forming ice balls in a clear gel that the shape is a tear drop that extends above the cold zone. It
was thus not surprising to find that the cylindrical model provided a better match and the length
of the assumed cold zone had to be greater than the actual cold zone to get a reasonable match.
Figure 3A shows a high rate of heat flow to the probe initially with an assumed maximum
rate of 50 W. By 2 minutes the heat flow rate has dropped to 31 W and decreases slowly as the
ice ball grows. The JT cooling that is available exceeds the calculated heat flow rate into the
probe even though the length has been assumed to be almost 60 % longer than it actually is for
about 10 minutes. Having the heat exchanger inserted into the liver without insulation puts
additional heat into the cryostat that probably explains why the test unit started to warm up
earlier. Figure 3B shows plots of temperature vs. time for layers that are near the thermocouples
in the test. The parameters used in the calculation were selected to give a good match at 10 mm
radius and to minimize the differences at 5 mm and 15 mm. Figure 3C plots the ice ball radius
and the radius at –20 C vs. time. The fraction of the volume that is below – 20 C is plotted in
figure 3D.
The cylindrical and spherical models were used to calculate the effects of different probe
temperatures and diameters. Using the same thermal properties as previously and the same 30
mm cylindrical probe length, heat loads and fraction below – 20 C at 6 minutes have been
calculated for probe temperatures of 100 K and 140 K. Results are presented in figure 4.
Observations to be made from these plots with regard to the fraction below – 20 C are,
• The fraction below – 20 C increases as the radius of the probe is increased.
• The fraction below – 20 C is greater for a cylindrical probe than a spherical probe of the
same radius.
• A significant fraction of the ice ball is warmer than – 20 C for all probes.
• The fraction that is warmer than – 20 C is greater for a temperature of 140 K than 100 K
but the difference may not be significant e.g. .50 vs. .57 for the cylindrical test probe size.
One advantage of having a group of cylindrical probes spaced so the ice balls touch at about 5
minutes is that the ice that forms between them will cool to lower temperatures so the fraction of
the total ice ball below – 20 C will be much greater, even if the probe temperature is warmer.
Heat flow rates to multiple cylindrical probes will be less than isolated probes because there is no
heat flow across the center plane between probes.
JT CRYOSTAT FLOW EFFECTS

The JT cryostat that was tested is said to be operating in the “blow-down” mode, that is the
cryostat has a fixed nozzle and gas is supplied at the bottle pressure. JT cooling is the product of
the mass flow rate and the enthalpy difference between the gas leaving at low pressure and the
Figure 3 Plots from spreadsheet with cylindrical model of ice ball that most closely matches the
test data of figure 2.
gas entering at high pressure. This difference peaks at about 40 MPa for Ar at 300 K and drops as
the bottle pressure drops. As a first approximation it can be assumed that the gas in the bottle
stays at room temperature when the cryostat is operating, and that the flow rate is proportional to
the pressure in the bottle. Differences can be significant for fast blow-down cryostats, first
because the gas in the bottle will drop in temperature, and second because the properties of the
gas flowing through the nozzle change and are influenced by the amount of precooling of the
high pressure gas before it reaches the nozzle.
Figure 4 Fraction of ice ball below – 20 C and heat load at 6 minutes calculated from cylindrical and
spherical models at 100 K and 140 K vs. probe radius.
It is possible to see how much the expansion of the gas in the bottle differs from being
isothermal by recording the increase in pressure after flow is stopped and the gas returns to
ambient temperature. For the test reported above, Ar flowing from a 2.5 L steel bottle for about
16 m, the pressure increased < 1 %. In applications where the blow down is a lot faster the
difference may be 10 % to 20 %.
Examples of the second effect are shown in figure 5. Flow rate of the Ar during the test in
liver was calculated assuming isothermal expansion in the bottle using the pressure transducer
readings. The readings lacked the last significant digit so the values are scattered around the
second order curve fit that is drawn labeled Ar 2. In order to see the departure from flow being
proportional to pressure the flow rate is first divided by pressure then normalized by dividing by
the first reading after the cryostat starts producing liquid. Plots for another cryostat that is
different in construction and operation are shown for Ar and Kr. These flow rates were calculated
from the pressure at the inlet to a vacuum pump connected to the cryostat outlet. The test labeled
Ar 1 shows very little departure from the assumption of flow being proportional to pressure
while the test with Kr shows a large departure.
Figure 5 Experimental plots of flow rate/bottle pressure normalized by dividing by initial cold
value to show deviations of assumption that flow is proportional to pressure. Ar 2 is from the test in liver.
JT CRYOSTAT COOLING
Calculations have been done for the cooling capacity of both finned-tube and matrix-tube
heat exchangers in the size range that might be used for cryosurgical applications. Figure 6 has
plots of cooling capacity and length vs probe diameter (not radius) for Ar and Kr at 40 MPa.
Calculations are done assuming bottle volumes and nozzle sizes that will cool an ice ball for
about 12 minutes. Heat exchangers are sized for efficiencies of about 80 % and return side
pressure drop of about 700 kPa at 6 minutes. This means that the liquid Ar in the cryostat would
be at about 110 K. It is seen that the matrix-tube heat exchanger is larger in diameter but shorter
than a finned-tube cryostat with the same capacity. A cryostat that is designed for Kr can be
significantly smaller than one designed for Ar.
Figure 7 shows plots of the ratios of the JT cooling at 40 MPa as presented in figure 6
divided by the ice ball heat loads at 6 minutes as presented in figure 4. A spherical ice ball
requires about 1.5 times as much cooling initially as it does at 6 minutes and a cylindrical probe
needs 1.8 to 2 times as much, for the tissue properties used in this analysis. All of the heat
exchangers that are plotted in figure 6 can produce more refrigeration than is needed for spherical
probes that are the same diameter as the heat exchanger. Cylindrical probes that have an
effective 30 mm long cold zone need to have the following minimum diameter heat exchangers,
Kr finned-tube, 2 mm, Kr matrix-tube, 2.7 mm, Ar finned-tube, 3.4 mm, and Ar matrix-tube, 4.9
mm. The matrix-tube heat exchanger shown in figure 1 has the same diameter as the finned-tube
unit but a shorter cold zone to compensate for its reduced capacity. Other design options are to
reduce the flow rate if the heat exchanger is oversized or adding an insulating sheath if a smaller
diameter has sufficient cooling.
SUMMARY
• Simple thermal models of spherical and cylindrical ice balls as they grow around a
cryoprobe have been developed and used to analyze heat loads and temperature profiles. Heat
loads are determined by the size and shape of the cryoprobe with differences due to refrigerant
temperature and tissue properties. These models provide a useful tool for studying how different
JT cooler system designs and operating protocols match a wide range of cooling requirements.
Figure 6 Maximum JT cooling and length for finned-tube and matrix-tube heat exchangers with
Ar and Kr at 40 MPa, 300 K vs. heat exchanger diameter.
Figure 7 Ratio of JT cooling from figure 6 to heat loads from figure 4 vs. cryoprobe diameter.
Spherical probes need a ratio > 1.5 and cylindrical probes need > 1.8.
• The models also show that a significant fraction of the ice ball does not cool below –20 C.
This means that controlled warming rates, multiple probes, repeated freeze-thaw cycles etc. may
be needed to kill the tissue.
• Examples have been given of the cooling capacity and length of finned-tube and matrix-tube
heat exchangers operating with Ar and Kr as a function of their diameter. Finned-tube heat
exchangers can produce more cooling than matrix-tube heat exchangers for a given diameter but
they are longer. JT cooling rates have been compared with heat loads from spherical and
cylindrical probes and are found to be sufficient to cool a wide range of cryoprobe sizes. Tests
with a cylindrical 3.4 mm diameter finned-tube cryostat in liver demonstrated sufficient cooling
to keep a 19 mm long cold zone cold for > 8 minutes during which time the ice ball radius grew
to 18 mm.
• A cryosurgical probe is to be considered as part of a larger system that involves issues of
safety, convenience to insert and control, support requirements, cost, compatibility with ice ball
monitoring means, etc.. Open cycle JT cooling has some unique characteristics which may have
advantages over other cooling methods. These include the small size of the heat exchanger,
small size warm gas lines, and adaptability to different operating protocols.
REFERENCES
1. Dobak, J. “A Review of Cryobiology and Cryosurgery”, Advances in Cryogenic
Engineering, Vol 43, Plenum Press, NY, 1998.
2. Marquadt, E.D., Radebaugh, R., and Dobak, J. “A Cryogenic Catheter for Treating Heart
Arrhythmia”, Advances in Cryogenic Engineering, Vol 43, Plenum Press, NY, 1998.
3. Ablin,R. J. “Cryoimmunotherapy fo the Treatment of Cancer” presented at 5th Annual
Meeting of Society of Urological Cryosurgeons, Tuscon, AZ Jan 1998.
4. Longsworth, R. C. and Steyert, W. A. “Fast Cooldown J-T Refrigerators for IR Detectors”,
Proceedings of the Interagency Cryocooler Conference/ Easton, MD; September 24, 1986.
5. Longsworth, R. C. “Cryo-Probe”, US patent #5,452,582 Sept. 1995.
Interference Characterization of Cryocoolers for a
High-Tc SQUID-Based Fetal Heart Monitor
A.P. Rijpma, M.R. Bangma*, H.A. Reincke, E. de Vries, H.J. Holland,

H.J.M. ter Brake, H. Rogalla
University of Twente, Faculty of Applied Physics
P.O. Box 217, 7500 AE Enschede, The Netherlands
*present address: KPN-Research, Area Middleware
Leidschendam, The Netherlands
ABSTRACT
The FHARMON-project at the University of Twente aims at a high-Tc SQUID based fetal
heart monitor for use in standard clinical environments. Besides the suppression of
environmental magnetic noise, the cooling of this fetal heart monitor is an important issue. For
maximum flexibility, we intend to apply a closed-cycle cryocooler instead of a liquid nitrogen
cryostat. Because of the extreme sensitivity of SQUID magnetometers, the interference caused
by the cryocooler is of major importance. This concerns electromagnetic interference (EMI),
mechanical vibrations and temperature fluctuations. We have developed measuring techniques to
characterize coolers in this respect. The characterization procedures were tested on a Signaal
USFA 7058 Stirling cooler and a Leybold RGD210 GM-cooler. In the paper the measuring
techniques are described along with interference characteristics of an APD Cryotiger and a
Ricor/AirLiquide K535 Stirling cooler. Also, the impact on the design of the fetal heart monitor
is considered.
INTRODUCTION
Fetal Heart Monitor
In the past decades, several groups showed that adult magneteocardiograms (MCGs) were
not only feasible, but on occasion even displayed a clearer indication of heart disorder than
electric ECG-measurements.1 Nevertheless, in general, the additional information that is obtained
does not outweigh the higher cost and complexity of an MCG system. For fetal heart monitoring,
however, the situation is different. In this case, the measurement of the ECG often fails or
becomes distorted, whereas the MCG-recordings are mostly successful.2,3 For this reason, the
FHARMON project aims at the design of a relatively inexpensive and easy-to-handle fMCG
system for use in a clinical environment.

SQUID measuring head

The measuring head of the first FHARMON demonstrator will consist of an alumina holder
containing three SQUID sensors. Alumina is applied to have sufficient thermal contact between
SQUIDs and the cold tip of the cooler. Metal cannot be used for this purpose, because of Johnson
noise and eddy current effects.4,5 The three SQUIDs will be combined electronically into a spatial
filter, i.e. a second-order axial gradiometer.4,5 The SQUIDs are separated by 6 cm in axial
direction. This distance, the baseline, resulted as an optimum from measurements with a variable
baseline configuration.6 For proper gradiometer operation the SQUIDs need to be parallel to one
another. However, imperfections are unavoidable. Due to the imperfections the system will also
be sensitive for magnetic fields orthogonal to the axis. This is called imbalance.4,5 In order to
compensate for this imbalance, two reference SQUIDs can be used that measure these off-axis
components. Based on introductory experiments, the environmental background field is
estimated to have a white noise level of In order to reduce the noise due to imbalance
to below the intrinsic system noise of this imbalance should be below A more
detailed discussion on these imbalance effects was presented elsewhere.6
Cooling
As the aim is an easy-to-handle system, it was decided to use cryocoolers rather than
cryogens. This can be implemented in several ways, as is discussed in more detail in reference 6.
The cooler, for example, may be put on top of the sensor insert and run continuously. For this we
require ‘low-noise’ coolers (JT-coolers, GM-type pulse tube coolers). More noisy coolers
(Stirling, GM) can be applied if time or space separation is used. Space separation will require a
thermal interface of typically 2 m in length. As a result, additional cooling power is required.
More problematic, however, is the practical realization of the interface. It will require a cryogenic
fluid circulation or a thermal strap. This significantly adds to the complexity of the system and it
is likely to be a costly and life-limiting component. For this reason, low-noise coolers are the
preferred option. Therefore, the JT APD-Cryotiger is our primary test cooler.7 Additionally, the
‘noisy’ Ricor K535 was tested for application in a system with thermal interface.8
REQUIREMENTS
Two types of requirements can be distinguished; commercial and technical. The commercial
requirements cover price (USD5000), lifetime and maintenance The
technical requirements include the elementary conditions for SQUID operation, such as operating
temperature and cooling power. It also deals with noise from the cooler: thermal, mechanical and
magnetic disturbances. It is important to note that these interference requirements may be relaxed
if appropriate measures are taken. For instance, temperature control can be applied to reduce
cold-tip temperature drift.
Temperature
A SQUID sensor is intrinsically a flux sensor based on superconducting technology. It has to
be cooled to below 84 K in order to work. The sensitivity of a SQUID magnetometer depends on
the temperature, because the effective sensing area, by which field is translated to flux, is
temperature dependent. If a SQUID is operated in a background field, this change in the effective
area can be seen as a change of magnetic field, thus as noise. However, temperature variations in
the measuring band of 0.5-100 Hz are not relevant since they will be damped by the thermal mass
of the insert. Therefore, we are left with a requirement on the drift in temperature, which will
affect the balance of the gradiometer and thus its capability of rejecting the environmental noise.
The allowed temperature drift is calculated at Furthermore, as SQUID
properties (e.g. sensitivity) are more stable at somewhat lower temperatures, i.e. around 60 K, it
would be desirable to have an operating temperature around that value.9
CRYOCOOLERS FOR A HIGH-Tc SQUID HEART MONITOR 795
Electromagnetic interference
Cryocoolers generate electromagnetic noise. Supply currents, transformers and compressors
cause noise due to currents and moving magnets. The metallic parts of a cooler generate Johnson
noise, which is mostly relevant if the metal is close to the sensors. Magnetic impurities will
generate EMI due to mechanical vibrations. To meet the requirement of a noise field smaller than
with a baseline of 6 cm, the second-order gradient field due to the cooler should be
below 28 in the measuring band of 0.5-100 Hz. An exception is made for the mains
frequency of 50 Hz. For the mains frequency we state that the noise contribution of the cooler
should be lower than the background field, which results in requirements of 10 nT for the field
component, 3 nT/m field gradient and for the second order gradient.6 For DC fields we
state that the contribution of the cooler should be lower than the DC background field (earth
field), which is about 50 µT strong with a gradient of 200 nT/m.4,5
Mechanical interference
Movement of the sensors in the background field introduces noise. Translations are not
problematic in this respect. With an imbalance of translations in the millimeter-range are
still acceptable.6 Much more problematic are rotations. The DC earth field is about 50 µT. With
an imbalance of and an acceptable noise level of this gives allowed rotations of
Ideally, the cooler itself would stay below these requirements. If not, then
vibrations can be damped by adding mass. Alternatively, the alumina holder could be
mechanically decoupled from the cooler. In both cases, the forces and torques of the cooler are
required to calculate the resulting vibrations of the alumina holder.
MEASUREMENT TECHNIQUES
Temperature
In order to measure the temperature fluctuations, Lakeshore SoftCal™ Silicon diodes are
used.10 These diodes have an absolute temperature accuracy of 0.15 K in the range of 60 K to
345 K. The relative accuracy is determined by fluctuations in the diode supply current and
readout amplifier and is much better. For example, the measured voltage drift of the readout
electronics corresponded to about 10 mK/day.
The electromagnetic interference is measured with a Bartington 3-axis fluxgate
magnetometer, type MAG-03MS, with a resolution of Both remanent magnetic
fields, as well as fields from active components, such as drive coils, are measured. The remanent
magnetism is measured by slowly rotating the cooler on a turntable and measuring the field
fluctuation at a fixed position. The field from active components is measured in a similar way.
The turntable is, however, stopped at a number of angles to allow measurement of the field.
Two methods are used to measure mechanical vibrations. The first basically gives a cooler
freedom of motion and then measures accelerations at a number of positions around the cooler.
To accomplish this, the cooler is rigidly mounted on a wooden board with is suspended from the
ceiling by means of springs (Fig. 1). The acceleration sensors are also mounted on this board
(Seika type BDK3).12 Taking into account (inertial) masses, the forces and torques exerted by the
cooler are calculated.
Figure 1. The USFA 7058 cooler in the set-up Figure 2. The vibrations of the RGD210 cooler
for the measurement of forces and torques. The as measured with a construction as shown in figure 1.
cooler is mounted on a plate along with acceleration The bandwidth is 1/16 Hz.
sensors. This plate in turn is supported by springs,
allowing freedom of motion. To support different
masses and sizes of coolers several plates and
springs were available.
Since the vibrations of the APD-Cryotiger cold head are small, a vibration measuring method
with a high resolution is needed. For that purpose, we developed a technique based on a SQUID
that is placed in an applied magnetic field.13 The SQUID is mounted on the cold tip of the cooler. It
measures the field along the axis of the cold head. In this way, the translation along the axis of the
cold head and the two rotations around the other two axes can be measured. For measuring the
rotations of the cold head, a uniform field is applied orthogonal to the SQUID’s axis of sensitivity.
Due to rotation, the SQUID measures a field
where is the applied field and and the amplitude and phase of the rotation. For the
translation measurement a field gradient is created. The SQUID response is now given by
where X is the displacement. Because in this arrangement, direct attachment of the SQUID to the
metallic cold tip of the cryocooler would introduce too much Johnson noise, an alumina cylinder
with a length of 4 cm and a diameter of 4½ cm was used to separate the SQUID from the cold tip.
INTRODUCTORY MEASUREMENTS
Two coolers were available for testing the measurement procedures: the USFA 7058 Stirling
cooler and the Leybold RGD210 GM cooler. Both were available from previous experiments.14,15,16
The USFA cooler only required a power supply to operate. As the RGD210 cooler had not been
used for several years, it had to be refilled with helium-gas before it could be used. No further
servicing was performed, as the cooler was only used to test the measurement procedures.
The accelerometer approach was tested with the Leybold GM cooler. Figure 2 shows an
acceleration spectrum of the Leybold GM cold head. Clearly visible are harmonics of the 2 Hz
drive frequency and a 46 Hz peak from the compressor, which is transferred by the gas lines. The
2 Hz peak in the acceleration spectrum results in translations of roughly
Figure 3. Cooldown-curve of the Leybold Figure 4. Short-term temperature fluctuation

RGD210 GM-cooler. The end temperatures of 1st of the lst (bottom) and 2nd (top) stage of the
and 2nd stage are 69 K and 33 K respectively. RGD210 cooler. The fluctuations are relative to
the operating temperature.
Figure 5. Remanent magnetic field of the Figure 6. Magnetic field of the USFA 7058
USFA7058 cooler measured at 0.5 m distance. The cooler measured at 0.5 m distance. The cooler is
cooler is rotated with about 0.5 Hz. The top and rotated, and the tangential and radial field
bottom lines correspond to the radial and is measured. The line corresponds with the field
tangential fields, respectively. from a theoretical dipole.
Temperature
Figures 3 and 4 show the cooldown curve and temperature stability of the GM-cooler. From
this we can see that the second stage becomes active after the first stage comes close to the final
temperature. The stability curve shows fluctuations of about 4 mK and 20 mK with the drive
frequency of 2 Hz for the first and second stage, respectively. The relatively small fluctuation of the
first stage compared to that of the second stage is explained by a higher thermal mass
Figures 5 and 6 show the remanent magnetic field of a USFA7058 cooler and the field produced
by the drive currents respectively. Both were measured at a distance of 0.5 m. The results correspond
well to a theoretical response from a dipole source. The graphs correspond to moments of for
the remanent magnetisation and for the moment caused by the drive currents.
CRYOTIGER RESULTS
To investigate whether the requirements can be met, measurements are performed on an APD
Cryotiger with a non-magnetic High Performance cold head, which is connected to the compressor
by means of flexible gas lines with a length of 7.5 m.7
Figure 7. Cooldown-curve of the Cryotiger. Figure 8. Cooling power of the Cryotiger as a

The lowest temperature obtained without load is function of temperature. The dots are measured
about 67 K. data-points, whereas the line is a typical
performance curve (PT13-mixture; standard cold
head; 60 Hz mains) taken from the manual.
Figure 9. Short-term temperature fluctuations

of the Cryotiger relative to the operating
temperature ( 70 K).
Temperature
The temperature is relevant in a number of aspects; the cooldown-curve (Fig. 7), the cooling
power (Fig. 8), the short-term temperature stability (Fig. 9) and the long-term temperature
stability (Fig. 10).
The cooldown cycle, without significant load, is about 2 hours, which is acceptable.
The cooling power complies with the specifications and goes up to about 5 watts at 77 K. As
we only expect a load of 0.5-1 watt this is more than sufficient.
The short-term temperature fluctuations stay well below our requirement of 0.1 K and are in
agreement with the results of Hohmann.17
The long-term temperature stability, however, does not meet our requirements. Large
temperature fluctuations were present at the cold tip of the Cryotiger in various experiments. In
Fig. 10 both the outside temperature of the vacuum housing as well as the cold tip temperature is
shown. The ambient temperature shows some large fluctuations due to sunshine and draft.
However, the temperature fluctuations of the cold tip do not coincide and the cause should be
sought elsewhere. In our opinion the temperature instability is due to (partial) clogging of the
cold stage. Just before every increase of about 10 K there is a small temperature drop of about
1 K. This is consistent with a pressure drop of the cryogenic liquid due to congestion of the high-
pressure gas supply line. Recovery might indicate that the congestion has cleared. Another
indication for clogging, was the fact, that ‘normal’ operation could be restored by switching off
the cooler and allowing the cold head temperature to rise above 120 K. This might indicate that a
component of the PT13 gas mixture (or a contaminant) freezes out and blocks the gas flow.18
Other users have also noticed similar problems concerning the instability of the operating
temperature.19,20
Figure 10. Long-term temperature stability of the Cryotiger. Around day 23 the cooler was
switched off. The bottom line shows the cold tip temperature. The top line depicts the simultaneously
recorded temperature of the cold head housing (i.e. ambient temperature).
The noise of the compressor and cold head is examined separately. From the turntable
measurements it followed that the remanent magnetic field of the compressor at 1 m is about
50 nT with a gradient of 70 nT/m. At 40 cm from the cold tip, the remanent magnetic field of the
cold head is about 0.1 µT with a gradient of 1 µT/m. This is fairly high for a ‘non-magnetic’ cold
head.
To determine the spatial distribution of the 50 Hz EMI of the compressor the magnetic field
around the compressor was measured at distances of 1 to 3 m and heights from 0 to 1.5 m. The
results are shown in Fig. 11. At about 2 m distance from the compressor the noise level from the
compressor approaches the environmental noise.
Within the resolution of the measurement no magnetic noise (beside the remanent field)
could be measured around the cold tip. With a bandwidth of 1/16 Hz and a sensor noise level of
this means that the 50 Hz magnetic noise field of the cold head is below 2 pT, which
clearly meets the requirement.
The mechanical interference of the compressor was measured with the accelerometers.
These measurements showed forces and torques up to 1.4 N and 0.4 Nm, respectively. The
corresponding frequency was 49 Hz.
Figure 11. Magnetic field of the Cryotiger Figure 12. Field measured by the SQUID
compressor measured at several angles in the sensor, when the cooler is operated in a applied
horizontal plane. At each angle, the distance from large magnetic field.
the compressor was varied from 1 to 3 m and the
height Z from 0 up to 1.5 m.
A typical vibration spectrum measured with an applied field of 0.5 mT is depicted in

Fig. 12. In this frequency spectrum one can clearly see the 50 Hz and higher harmonics peaks,
due to background power line interference. The noise at 25 Hz is caused by a vacuum pump. A
closer look at the spectrum, as shown in Fig. 12, reveals that there are not only 50 Hz peaks and
higher harmonics but also 49 Hz and harmonics. These are caused by the cold tip vibrations.
Detailed analysis resulted in the following: the cold tip of the Cryotiger translates with an
amplitude of at a frequency of 49 Hz along the cold tip axis. Hohmann17 and Hill21
report similar results. The rotations around the x and y axes yield a maximum inclination of
K535 MEASUREMENTS
The Ricor K535 cooler is a dual opposed piston Stirling crycooler with integrated cold head.
The temperature was kept constant at a preset value by the accompanying software and hardware.
If the cooler could not reach this temperature in 2 hours it automatically shut down. According to
the specifications, the cooling power is 6 watts at 77 K, with an input power of 200 watts.
Temperature
The cooldown curve for this cooler is shown in Fig. 13. In about 10 minutes it cools to a
temperature of about 50 K. The second half of the figure shows what happens if the cooler is
operated in the so-called regenerative mode in which it basically acts as a heater.
Figure 14 shows the remanent magnetic field of the cooler as a function of time when it is
slowly rotated on the turntable with about 0.1 Hz. Figure 15 shows the 46 Hz drive frequency
magnetic field from the cooler as a function of the measurement angle. Note the difference with
figures 5 and 6, which is due to a different arrangement of the piston drive coils, causing the
K535 cooler to show quadrupole behavior. Assuming a separation of 17 cm between the two
pistons, the moment of the two dipoles is for the remanent field. The quadrupole
corresponding to 46 Hz drive frequency is built from two dipoles with a strength of
The mechanical vibrations were measured with the accelerometers. Along the cold finger
axis a force of 10 N was measured at the 46 Hz drive frequency. Taking into account only the
cooler mass, this corresponds to an acceleration of The torque was 0.34 Nm (rotation:
The drive axis showed a force of 4 N and a torque below 0.1 Nm
The force and torque on the remaining axis amounted to 6N and
Figure 13. Cooldown-curve of the K535 cooler followed

by a regenerative cycle, which heats the cooler.
Figure 14. Remanent magnetic field of the Figure 15. Magnetic field of the K535 cooler
K535 cooler measured at 0.5 m distance. The measured at 0.5 m distance. The cooler is rotated,
cooler is rotated with about 0.1 Hz. The top and and the tangential and radial field is
bottom line are the radial and tangential field, measured. The lines corresponds with the field
respectively. The offset is due to the earth field. from a theoretical quadrupole.
DISCUSSION
The measurement methods all gave reasonable and reproducible results. Especially the
measurement of vibrations with the SQUID sensor in an applied field (gradient) allows for very
accurate measurements. The results of the Cryotiger measurements are, in general, well within
our requirements. The vibration levels even allow the insert to be mounted directly on the cold
tip. However, as the sensors require a stable temperature below 80 K, the large long-term
temperature fluctuations are a source of major concern. Nonetheless, we plan to use the Cryotiger
in our first FHARMON demonstrator. As expected, the K535 has more magnetic and mechanical
noise than the Cryotiger cold head. Its power, however, allows for use of the cooler in
combination with a thermal link. In this respect, the quadrupole behavior of its EMI will ensure a
fast decrease of interference with increasing distance. The mechanical vibrations should be
decoupled by the link, just as in the case of the Cryotiger this should be prevented that the gas
lines transfer compressor vibrations.
ACKNOWLEDGEMENTS
This research is supported by the Dutch Technology Foundation (STW), the Institute for
Biomedical Technology (BMTI), Philips Medical Systems and Signaal USFA. We would also
like to thank Ricor/Air Liquide for making a cooler available for test purposes.
REFERENCES
1. Ziegert, K., Selbig, D., Soltner, H., Trahms, L., Hanrath, P. and Stellbrink, Ch.,”Evaluation of a new
fragmentation score of the QRS complex for risk stratification using a high-Tc-SQUID-
magnetocardiography-system”, The Official Journal of the Int. Society for Holter and Noninvasive
Electrocardiology, Inc.; A.N.E., Vol. 3 (3), Part 2, (1998), pp. 61.
2. Crowe, J.A., Woolfson, M.S., et al., “Antenatal assessment using the fECG obtained via abdominal
electrodes”, Journ. of Perinatal Medicine, Vol. 24 (1) (1996), pp. 43-53.
3. Peters, M.J., Stinstra, J.G., Quartero, H.W.P., and ter Brake, H.J.M., “The foetal
magnetocardiogram.”, Res. Adv. In Biomedical Eng., Vol. 1 (2000).
4. Vrba, J, “Multichannnel SQUID Biomagnetic systems” in: Applications of Superconductivity, ed.
Weinstock, H., NATO ASI Series E: Applied Sciences Vol. 365, Kluwer Academic Publishers,
Dordrecht, The Netherlands.
5. Vrba, J., “SQUID magnetometry”, La Physique au Canada, Vol. 36 (1980), pp. 3-9.
6. Rijpma,, A.P., ter Brake, H.J.M., Peters, M.J., Bangma, M.R. and. Rogalla, H., “Cryogenic aspects
of a fetal Heart Monitor based on High-TC SQUIDs”, to be published in Adv. Cryo. Eng.
7. IGC APD Cryogenics Inc, 1833 Vultee Street, Allentown, PA 18103-4783, United States.
8. AirLiquide, Division des techniques avancées, B.P. 15, 38360 Sassenage,France.
9. Rogalla, H., Superconducting Electronics. Cryogenics 34 (ICEC suppl) (1994), pp. 25-30.
10. Lakeshore Cryotronics Inc, 575 McCorkle Blvd., Westerville, OH 43082-8888, USA.
11. Bartington Instruments Ltd, 10 Thorney Leys Business Park, Witney, Oxford, OX8 7GE, England.
12. AE sensors BV, PO Box 9084,3301 AB Dordrecht, The Netherlands.
13. Bangma, M.R., Rijpma, A.P., de Vries, E., Reincke, H.A., Holland, H.J. ter Brake, H.J.M. and
Rogalla, H, “Interference characterisation of a Cryotiger to be used in a SQUID-based fetal heart
monitor”, submitted to Cryogenics.
14. ter Brake, H.J.M., van den Bosch, P.J. and Holland, H.J., “Magnetic noise of small stirling coolers”,
Adv. Cryo. Eng., Vol. 39, Plenum Press, New York (1993) pp. 1287-1295.
15. ter Brake, H.J.M., Hogenkamp, J.E.M., Ulfman, J.A. and Flokstra, J., “Low evaporation hybrid
cryogenic system for superconducting devices”, ICEC10, Colan, H., Berglund, P. and Krusius, M.,
eds., Butterworth, Guildford (1984), pp. 58-61.
16. van den Bosch, P.J., Holland, H.J., ter Brake, H.J.M. and Rogalla, H., “Closed-cycle gas flow
system for cooling of highTc dc SQUID magnetometers”, Cryogenics 35 (2) (1995), pp. 109-116.
17. Hohmann, R., SQUID-System mit Joule-Thomson-Kuehlung zur Wirbelstrompruefung von
Flugzeugfelgen, PhD thesis, Institut fuer Angewandte Physik, Justus Lieblig Universitaet Giessen,
Germany (1999).
18. Personal communication with Ajay Khatri from APD Cryogenics (2000).
19. Personal communication with A. Abedi (May 2000).
20. Kawecki T.G. and James, S.C., Cryocoolers 10 (1999), pp. 43-54.
21. Hill, D., “Throttle Cycle Cooler Vibration Characterisation”, Cryocoolers 9 (1997), pp. 737-745.
Vapor Precooling in a Pulse Tube Liquefier
E.D. Marquardt, Ray Radebaugh, and A.P. Peskin

Boulder, CO 80303
ABSTRACT
Experiments were performed to study the effects of introducing vapor into a dewar where a
coaxial pulse tube refrigerator was used as a liquefier in the neck of the dewar. We were
concerned about how the introduction of vapor might impact the refrigeration load as the vapor
barrier in the neck of the dewar is disturbed. Three experiments were performed where the input
power to the cooler was held constant and the nitrogen liquefaction rate was measured. The first
test introduced the vapor at the top of the dewar neck. Another introduced the vapor directly to
the cold head through a small tube, leaving the neck vapor barrier undisturbed. The third test
placed a heat exchanger partway down the regenerator where the vapor was pre-cooled before
being liquefied at the cold head. This experiment also left the neck vapor barrier undisturbed.
Compared to the test where the vapor was introduced directly to the cold head, the heat
exchanger test increased the liquefaction rate by 12.0%. The experiment where vapor was
introduced at the top of the dewar increased the liquefaction rate by 17.2%. A computational
fluid dynamics model was constructed of the dewar neck and liquefier to show how the
regenerator outer wall acted as a pre-cooler to the incoming vapor steam, eliminating the need
for the heat exchanger.
INTRODUCTION
Liquefaction plants have always used some form of recuperative cryogenic refrigerator (i.e.
Joule-Thomson, Claude, or Brayton cycles) to provide the cooling. The reason for this is the
higher efficiencies that can be obtained with recuperative systems over regenerative systems (i.e.
Stirling, pulse tube, or GM cycles) in larger scale cryogenic refrigerators. There is now increased
interest in small to medium liquefaction plants (20 W to 1 kW refrigeration capacity) for local
cryogen liquefaction in the aerospace, military, and commercial sector. As refrigeration capacity
is reduced, regenerative systems become comparable with and then surpass recuperative systems
in efficiency, making them attractive choices for small and medium scale liquefaction plants.
One advantage of recuperative systems is the ability to continuously cool the vapor through
the recuperative heat exchanger. An ideal regenerator cannot accept heat so the regenerative
systems must be multi-stage or remove all the heat at the lowest temperature, reducing
efficiency. It has been shown by Radebaugh et. al.1 that non-ideal regenerators are able to accept
some heat at any point. We performed three experiments to determine whether this effect could
be used to improve liquefaction efficiency of regenerative systems.

EXPERIMENTAL SETUP
Figure 1 shows the experimental setup with the pulse tube in the neck of a dewar. The pulse
tube was fully characterized2 in a vacuum chamber so that refrigeration loads could be
determined from the input power. It had a nominal refrigeration capacity of 18.8 W at 90 K (20%
of Carnot) and produced 14.5 W at 77 K (19% of Carnot) with 223 W PV input power. Nitrogen
was used as the test fluid to be liquefied. In all experiments, the input PV power was held
constant at 223 W. The nitrogen mass flow rate was measured to hold the pressure in the dewar
constant.
In the first experiment, the nitrogen was introduced through a 1.5 mm tube directly to the
cold end. This did not allow the insulating gas in the neck of the dewar to be disturbed. The
second experiment introduced the nitrogen at the top of the dewar. The final experiment
introduced the nitrogen through a tube leading to a heat exchanger and then through another tube
leading directly to the cold end. The heat exchanger was clamped to the 50 mm long regenerator
30 mm down from the warm end. It was a split design allowing it to be clamped anywhere along
the regenerator’s length. The contact thickness was 1.65 mm and the heat exchanger contained a
channel 25 cm long to insure adequate heat transfer between the fluid and heat exchanger.
RESULTS
Figure 2 shows a typical cool down of the dewar with the inlet at the top of the dewar. The
temperature locations are shown in Figure 1. Under steady state operation, the temperature
profile in the neck of the dewar changes dramatically as the warm fluid flows passed the neck.
Figure 3 shows the wall temperature profiles in the dewar neck for both the cryocooler and the
dewar walls. The dewar wall temperatures were measured at the three positions shown in
Figure 1. Experiments 1 and 2 show linear temperature profiles, as expected, along the wall
Figure 1. Experimental setup.

VAPOR PRECOOLING IN A PULSE TUBE LIQUEFIER 805
Figure 2. Typical cool down with fluid inlet at the top. Temperature locations are given in Fig 2.
within the experimental errors associated with the placement of the thermocouples. The dewar
wall temperature in experiment 3, as shown in Figure 3, is interpolated from the measurements at
the three locations given in Figure 1. The cryocooler wall temperatures along the regenerator for
experiment 3 are derived from our theoretical model and are displayed to show the boundary
conditions the gas is exposed to as it travels to the cold head.
The dewar had a measured heat leak of 1.1 W. The enthalpy required to liquefy nitrogen at
0.1 MPa from 300 K is 428 J/g. This results in 7.13 W·min/g energy required for nitrogen
liquefaction from room temperature. The measured mass flow rates for the 3 experiments were
1.92, 2.25, and 2.15 g/min. Table 1 shows the refrigeration power required for the liquefaction
given these mass flow rates. Ql, Q2, and Q3 represent the refrigeration load required for the
three different experiments respectively.
Given the 1.1 W dewar heat leak and the 14.5 W of cooling power available, the energy
required for the case where nitrogen is directly applied to the cold head, Ql, agrees well with the
calculated value. The case using the mid-stage heat exchanger, Q3, shows that by accepting
some energy at a mid-point, we can increase the liquefaction rate by 12%. Our theoretical model1
predicts a 15% increase in the liquefaction rate. The case where nitrogen was introduced at the
top of the dewar shows that the regenerator continuously pre-cools the fluid since we only have
14.5 W of cooling power available and with the dewar heat leak, we would require 17.15 W if
heat was only absorbed at the cold head. We have increased the liquefaction rate by 17.2%
without using any special heat exchangers or flow channels. The model predicts a 23% increase
if the fluid is continuously pre-cooled by the regenerator.
The liquefier figure of merit, FOM, can be determined by
Figure 3. Temperature profile in the dewar neck.
where is the ideal work required for liquefaction, is the actual work input, is the
liquefaction rate, is the ambient temperature, is the entropy difference from ambient to
the saturated liquid state, and is the enthalpy difference from ambient to the saturated liquid state.
Assuming the efficiency of the compressor, is 85%, the total electrical input
power is 262 W. Table 2 summaries the experimental results.
CFD MODELING
We have numerically modeled the best case where fluid enters at the top of the dewar and
the cryocooler cold head is placed within the neck of the dewar. Using a commercial CFD
package, we have determined the temperature profile and streamlines within the dewar neck
shown in Figure 4. These results show how the regenerator pre-cools the fluid before it reaches
the cold head. The predicted fluid temperature along the regenerator closely matches the
regenerator wall temperature shown in Figure 3.
To study the flow and temperature profiles inside the dewar neck, we used a finite element
numerical model. We used a 2-dimensional model, discretisizing the domain into 13,350 4-node
elements. Constrained temperatures were input as boundary conditions along the cryocooler
walls and at the fluid entrance. Adiabatic conditions were assumed for the outside walls of the
dewar. Compressible flow was modeled by assuming a Boussinesq equation of state for the
density. A function for the volume expansion was input for a range of densities for vapor and
liquid states. Fluid properties were taken from the NIST12 database3 for nitrogen.
VAPOR PRECOOLING IN A PULSE TUBE LIQUEFIER 807
Figure 4. CFD results showing temperature contour and streamlines with in dewar neck.
CONCLUSIONS
We have shown that heat transfer to the regenerator can be used to improve liquefaction
rates in regenerative cryocoolers with no special attention given to the fluid flow paths.
Improvements of 17.2% were shown over then case were all the liquefaction energy is removed
at the cold head. Regenerative systems do not require any complex recuperative heat exchangers
where a leak could cause mixing of the working fluid and liquefaction fluid.
REFERENCES
1. Radebaugh, R., Marquardt, E.D., Gary, J., and O'Gallagher, A., ''Regenerator Behavior with
Heat Input at Intermediate Temperatures," Cryocoolers 11, Plenum Publishers, New York
(2000), in press.
2. Marquardt, E. and Radebaugh, R., ''A Pulse Tube Oxygen Liquefier," Cryogenic
Engineering Conference Vol. 45, Plenum Press, (1999), in press.
3. Lemmon, E.W., McLinden, M.O., Peskin, A.P., and Friend, D.G., Thermodynamic and
Transport Properties of Pure Fluids and Mixtures (NIST12). NIST: Gaithersburg, MD.
Terrestrial Applications of Zero-Boil-Off
Cryogen Storage
Salerno, L. J.1, Gaby, J.2, Johnson, R.3, Kittel, P. 1 , and Marquardt, E. D. 4

1
NASA Ames Research Center
Moffett Field, CA, USA 94035
2
NASA Glenn Research Center
Cleveland, OH, USA 44135
3
NASA Kennedy Space Center
Kennedy Space Center, FL, USA 32899
4
ABSTRACT
Storing cryogenic propellants with zero boil off (ZBO) using a combination of active
(cryocoolers) and passive technologies has recently received a great deal of attention for long-
term space missions. This paper will examine a variety of potential near-term terrestrial
applications for ZBO and, where appropriate, provide a rough order of magnitude cost benefit of
implementing ZBO technology.
NASA’s Space Shuttle power system uses supercritical propellant tanks, which are filled
several days before launch. If the launch does not occur within 48-96 hours, the tanks must be
drained and refilled, further delaying the launch. By implementing ZBO, boil off could be
reduced and pad hold time extended by a factor of eight.
At NASA’s John F. Kennedy Space Center, vented liquid hydrogen storage dewars
lose 650 kg (500 gal)/day through boiloff. Implementing ZBO would eliminate this, saving
$625,000 per year.
Overland trucking of from the supplier to the launch site via roadable dewars results in
a cryogen loss of 10% per tanker (1300 kg (1000 gal)/tanker). If this loss could be eliminated,
the savings would be approximately $30,000 per year.
Within the superconductivity community, there is skepticism about using coolers, based
upon reliability concerns. One approach would be to design a hybrid system including a smaller
dewar to hold the cryogen for a short time (approx 1 month) and a cooler sized for continuously
re-liquefying the boil off. This approach would provide a system with both the high reliability of
a stored cryogen combined with the low maintenance and small size of a commercial cryocooler,
and could greatly benefit not only high temperature superconducting power applications, but
cellular phone base stations or any commercial application that cannot afford a system failure.

INTRODUCTION
Storing cryogenic propellants with zero boil off (ZBO) using a combination of active
(cryocoolers) and passive technologies has recently received a great deal of attention for long-
term space missions1-5. Eliminating the need for storage margin to compensate for cryogen boil-
off and the concomitant smaller size of tankage and insulation translates to greater available
payload mass. This paper will examine a variety of potential near-term terrestrial applications for
ZBO and, where appropriate, provide a rough order of magnitude cost benefit of implementing
ZBO technology.
Commercial application of ZBO dewars will only be successful if they make economic
sense to implement. To examine the benefits, we must make some assumptions of cost. We have
assumed energy cost to be 0.04 $/kW·hr, based upon industrial rates, and the bulk cost of liquid
nitrogen, oxygen, and hydrogen to be 0.113 $/liter, 0.176 $/liter, and 0.288 $/liter respectively.
Using these figures, we can estimate the cost savings per day as a function of the percent of
Carnot efficiency that the liquefier achieves. Figure 1 shows the results.
Because the cost of energy is only a fraction of the total cost of buying liquid cryogen, the
rest being transportation and other overhead, higher boiling point cryogens will benefit more
from local liquefaction, since less cooling is required. Advanced low cost cryocoolers (GM type)
can achieve efficiencies of approximately 9% and 11% of Carnot at 20 K and 80 K, respectively.
This makes them marginally effective for nitrogen ZBO dewars and reasonable for oxygen ZBO
dewars. A typical liquid oxygen dewar with a capacity of 26000 kg (6000 gallons) may have a
boil off rate of 1% per day.
With an advanced cryocooler operating with an efficiency of 16% with oxygen, we can save
per day. If we can buy a cryocooler for $15,000, the payoff
will occur in under 2 years.
Figure 1. Cost savings based on liquefier efficiency.

APPLICATIONS OF ZERO-BOIL-OFF CRYOGEN STORAGE 811
Hydrogen requires a liquefier with at least a 17% efficiency just to break even. It may be
that hydrogen dewars could benefit from reduced boil off by cooling the radiation shield but that
is beyond the scope of this paper. All ZBO applications will benefit from low cost higher
efficiency cryocoolers such as Stirling and pulse tubes.
Space Shuttle Power Systems

NASA’s Space Shuttle power system uses supercritical propellant tanks as part of the Power
Reactant Storage and Distribution (PRSD) system. These tanks are filled several days before
launch. Zero loss storage of cryogenic fuel cell reactants on board the Space Shuttle is a potential
application for demonstrating zero loss technologies. The fuel cells use oxygen and hydrogen to
provide power and drinking water for the crew. The reactants are stored as supercritical fluids in
vacuum insulated tanks with active vacuum ion pumps to maintain a very low vacuum in the
annular space prior to loading.
The oxygen tanks are 84.9 cm (33.43 inches) in diameter with a volume of (11.2
cubic feet) and a working pressure of 6.2 Mpa (900 psia). The hydrogen tanks are 105.5 cm
(41.5 inches) in diameter with a volume of (21.4 cubic feet) and a working pressure of
1.7 Mpa (250 psia). Heat leak into the tanks is estimated to be 8.1 watts for oxygen and 3.6
watts for hydrogen. Figures 2 and 3 show a photo and a schematic diagram of a tank. The tanks
are similar, except for size and that the hydrogen tank contains a vapor cooled shield.
The tanks are loaded with normal boiling point (NBP) liquid approximately 48 hours before
the planned lift-off. The number of tanks is dependent upon the mission power and duration, but
we have assumed five of each in this paper. The tanks are hydraulically pressurized to the
operating pressure. Excess pressure is vented overboard. This boil-off limits the amount of time
the orbiter can be on the launch pad ready to go. Depending on the mission, these constraints
can severely limit supportability of the shuttle. Typical hold time is 1 to 4 days before the
reactants must be reloaded. Launch delays can be the result of many different situations, weather
at the launch pad, weather at an emergency landing site, or technical problems with one of the
many orbiter systems. How quickly the vehicle will be ready for another launch attempt depends
greatly on the fuel cell reactant supply.
Figure 2. Photo of Shuttle PRSD tank.

Figure 3. Schematic diagram of Shuttle PRSD tank.
Once the tanks fall below the minimum reactant mass to support a mission, the fuel cell
tanks must be partially drained and filled with fresh reactants. These ‘turnaround’ operations
require a 3-day delay in launch, costing tens of thousands of dollars per day.
By configuring these tanks to a zero-loss configuration, these delays can be drastically
reduced or eliminated. A cold head in the reactant manifolds or in each tank would compensate
for the environmental heating on the system, eliminating the reactant mass loss that occurs while
the tanks are locked up. Adding a 15% cooling reserve to the tank heat leak requirement yields a
cooling system of 4.1 watts for each hydrogen tank (21 watts total) and 9.4 watts for each
oxygen tank (47 watts total).
The coolers are only required to operate during ground hold. The launch mass can be
minimized by mounting the compressor remotely, off the shuttle. A launch disconnect
mechanism must be provided to disconnect the compressor(s) from the cold heads prior to
launch. Since the cold heads must remain on the shuttle during launch, orbit and reentry, cold
heads meeting all shuttle safety and vibration standards are required.
The first approach is to provide a cold head in each tank to completely eliminate the boil off.
A heat switch is also needed to reduce the parasitic load when the cooler is off, since the cooler
will only be operated while the shuttle is on the pad. A cooler (or multiple coolers) of 21 watt
capacity at approximately 25 K is needed as well as a cooler (or multiple coolers) of 47 watt
capacity at approximately 105 K. Hydrogen cold heads that are space flight suitable and provide
on the order of 4 watts of cooling at 25 K are presently in the development stage only
Another approach would be to cool the vapor cooled shield in the hydrogen tank to reduce
the parasitic losses. This would allow use of less expensive and presently available coolers (80
K vs. 20 K) and potentially simplify the system by requiring only a single compressor on the pad
with a single disconnect line. While this is actually a reduced boil off configuration rather than a
zero-boil off case, if the losses could be cut significantly, the increase in hold time could be
sufficient to prevent draining and refilling the tanks in the majority of instances. If a shield
cooler were installed on each tank, roughly 7.8 watts at 80 K would be required. This approach
would drop the tank heat load to 0.5 watts per tank, instead of 4.1 watts, and would extend the
storage lifetime by a factor of eight. The cost of a 10 W, 90 K cooler system, including
compressor, cold heads, plumbing, and heat switches is on the order of $200 K per tank, or $2M
total. If a $90,000 turnaround delay cost is assumed, the cooler system cost is recoverable after
20 launch delays have been averted. This analysis does not include recertification costs, which
are likely to be on the order of several million dollars. However, one of the options being
considered is replacement of the existing shuttle PRSD tanks to assure flight capability through
the year 2020. This will require recertification, and the delta cost to certify the cooler system at
that time will be minimal.
NASA John F. Kennedy Space Center Launch Site Facilities
The nation’s Space Shuttle launch pads LC 39A and 39B at NASA’s John F. Kennedy
Space Center (KSC) each contain a (850,000 gal.), vacuum insulated, liquid hydrogen
storage tank (Figure 4). Each tank loses approximately 650 kg (500 gal.) of hydrogen a day
through boil-off, or 474,500 kg total per year. In addition, each over-the-road tanker truck used
to replenish these storage tanks loses an additional 1,300 kg (1000 gal) per tanker in transferring
liquid to the tank over a 4 hour transfer period. Typically, 450 hydrogen tanker trucks are off-
loaded at KSC each year, losing a total of 585,000 kg. With the storage tank losses, the yearly
loss is approximately of hydrogen. Cooling the transfer lines and eliminating the
storage tank boiloff losses would result in an annual saving of $625,000 in propellant costs, with
the current eight shuttle flights per year. Increasing flight rates will result in even more savings.
The cost savings are based on a cost to KSC of $0.21 per liter of KSC has a 10 yr contract
and uses several million liters per year. Lower usage customers may pay $1.15 per liter or more.
Figure 4. hydrogen storage tank at NASA John F. Kennedy Space Center.

The liquid hydrogen storage tanks at KSC were manufactured in the late 1960’s to support
the Apollo space program and have never been taken out of service. Making modifications to
the pressure vessel itself is impractical due to the launch schedule at KSC. An add-on system
that can be tapped into the storage tank vent system and deliver liquid back into the transfer
manifold system provides the least operational impact for launch operations. An auxiliary
storage tank of (50,000 gal.) with a series of cold heads in the tank will provide adequate
storage to capture the main tank boil-off. This approach would allow for a tank to tank transfer
once every 30 to 45 days and also correspond to an over the road tanker truck delivery to further
minimize transfer losses. Figure 5 shows the concept for this system. Transferring from the new
tank to the existing tank would also precool the line to minimize losses from the roadable
dewars.
To capture the daily boil-off and provide the surge capability necessary to handle tanker
transfer operations, two cooling systems are required in the design. The first, or ‘daily’ cooling
system, will provide 800 watts of cooling. When tankers are off-loaded the additional heat load
must be accounted for with additional cold heads. When tankers are offloaded, the storage tank
pressure must remain below gage pressure of 173 kPa (25 psig) and be back to less than a gage
pressure of 13.8 kPa (2 psig) within 64 hours to be ready for the next tanker off-load opportunity.
The second, or ‘surge’ cooling system must provide 9,000 watts of cooling, thus making the use
of a commercial liquefier a suitable option.
Using multiple coolers, which are staged would provide both reliability and cost advantages.
To handle the ‘surge’ capability, commercial coolers totalling 9 kW capacity (588 kW power
consumption) are available at a cost of roughly $3,000,000. For the ‘daily’ cooling requirement,
additional coolers totalling 800 watt capacity (52 kW power consumption) cooler would cost
$270,000. Adding the $350,000 cost of the new tank, connecting lines, transfer line cooling loop,
etc. results in an equipment cost of $3,660,000 per facility, or $7,320,000 total. Assuming that
the surge capacity of 9 kW is required once per 30 days for a period of 64 hours (energy required
of 452,000 kw-hr), and that the 800 watt daily cooling is required continuously (energy required
of 455,000 kw-hr), the total power consumption is 907,000 kW-hr. Using $0.04/kW-hr, the
annual utility expenditure then is $36,300 per facility, or $72,600 total. Because $625,000 per
year is saved by utilizing ZBO, and $72,600 is spent in energy costs, the savings is $552,000 per
year. The initial equipment cost is $7,320,000, so the payback period is 13 years.
Figure 5. ZBO concept at NASA KSC launch site.

Figure 6. Typical roadable dewar (photo courtesy Air Products).
Readable Dewars
Overland trucking from the supplier to the launch site via roadable dewars results in a
cryogen loss of 10% per tanker (1300 kg (1000 gal)/tanker). If a typical roadable transport dewar
is considered (Figure 6) of 14,300 kg (11,000 gallons), this boil off saving amounts to
approximately $33,000 per year.
The cooling power required to prevent this boil off is about 1 kW at 20 K. Commercial
coolers have efficiencies of roughly 9-11% of Carnot, so using a figure of 10% gives an input
power of 140 kW to achieve the 1 kW required cooling, or 188 horsepower (hp). Even if 20% of
Carnot efficiency could be achieved, nearly 100 hp would be required. Providing an additional
diesel engine on board the rig does not seem feasible from a cost, maintenance, or weight
standpoint.
Other Applications
Within the superconductivity community, there is skepticism about using coolers, based
upon reliability concerns. One approach would be to design a hybrid system including a smaller
dewar to hold cryogen for a short time (approx 1 month) and a cooler sized for continuously re-
liquefying the boil off. This would provide a system with both the high reliability of a stored
cryogen combined with the low maintenance and small size of a commercial cryocooler. This
would eliminate the need to refill the dewar continuously and, with remote monitoring of the
cryocooler status, a time period of one month would be available to replace a failed cryocooler or
simply refill the dewar until routine maintenance is scheduled.
A similar system could be implemented for cellular telephone base stations. Cellular
systems presently have demonstration facilities involving cryocoolers only. Moving to a hybrid
system by adding a small dewar could potentially reduce maintenance costs and more
importantly, avert unexpected and negatively perceived downtime.
CONCLUSION
This paper has discussed several potential applications of zero boil off cryogen storage that
would be feasible here on Earth. Applying ZBO to large cryogenic storage facilities could
potentially save millions of dollars per year across the nation. Advances in cryocooler efficiency
with concomitant increases in reliability and reduction in mass and cost will further enable ZBO
applications that are not feasible presently. For example, the roadable dewars discussed would
benefit greatly from efficiency increases. If a cooler efficiency of 20 W/W could be realized, 20
kW of power would be required, or merely an additional 27 hp on board the rig.
In addition, fuel cell systems for mass transportation would be a candidate for ZBO
technology if lightweight, efficient coolers could be made inexpensively.
Zero Boil Off Cryogen Storage is a new and innovative technology, which will provide a
great advantage not only for facilities storing large amounts of cryogens, but for any commercial
application that cannot afford a system failure.
REFERENCES
1. Salerno, L. J., and Kittel, P., “Cryogenics and the Human Exploration of Mars”; Cryogenics 39,
(1999) pp. 381.
2. Kittel, P., Salerno, L. J., and Plachta, D. W., “Cryocoolers for Human and Robotic Missions to
Mars”; Cryocoolers 10, Plenum Press, New York, (1999) pp. 815.
3. Plachta, D. W., and Kittel, P., “Hybrid Thermal Control Testing of Cryogenic Propellent Tanks”;
presented at CEC-99 (Montreal, July 12-16, 1999), Adv. Cryo. Engrg., Ed. Q.-S. Shu, v. 45 (Kluwer,
New York, 2000) to be published.
4. Kittel, P., and Plachta, D. W., “Propellant Preservation for Mars Missions”; presented at CEC-99
(Montreal, July 12-16, 1999) , Adv. Cryo. Engrg., Ed. Q.-S. Shu, v. 45 (Kluwer, New York, 2000) to be
published.
5. Plachta, D., Kittel, P., and Alexander, R., “Launch Vehicle Mass Savings for Zero Boil-Off Cryogen
Storage Approach Applied to Human Missions to Mars”, presented at Mars Society Convention, Boulder,
CO Aug. 12-15, 1999
Proceedings Index
This book draws upon the work presented at the 11th International Cryocooler Conference, held in
Keystone, Colorado, June 20-22,2000. Although this is the eleventh meeting of the conference, which
has met every two years since 1980, the authors’ works have only been available in hardcover book
form since 1994; this book is thus the forth hardcover volume. Prior to 1994, proceedings of the
International Cryocooler Conference were published as informal reports by the particular government
organization sponsoring the conference — typically a different organization for each conference. Most
of the previous proceedings were printed in limited quantity and are out of print at this time.
For those attempting to locate references to earlier conference proceedings, the following is a
listing of the ten previous proceedings of the International Cryocooler Conference.
1) Refrigeration for Cryogenic Sensors and Electronic Systems, Proceedings of a Conference held at
the National Bureau of Standards, Boulder, CO, October 6-7,1980, NBS Special Publication 607, Ed.
by J.E. Zimmerman, D.B. Sullivan, and S.E. McCarthy, National Bureau of Standards, Boulder, CO,
1981.
2) Refrigeration for Cryogenic Sensors, Proceedings of the Second Biennial Conference on Refrigera-
tion for Cryogenic Sensors and Electronics Systems held at NASA Goddard Space Flight Center,
Greenbelt, MD, December 7-8, 1982, NASA Conference Publication 2287, Ed. by M. Gasser, NASA
Goddard Space Flight Center, Greenbelt, MD, 1983.
3) Proceedings of the Third Cryocooler Conference, National Bureau of Standards, Boulder, CO,
September 17-18,1984, NBS Special Publication 698, Ed. by R. Radebaugh, R Louie, and S. McCarthy,
National Bureau of Standards, Boulder, CO, 1985.
4) Proceedings of the Fourth International Cryocoolers Conference, Easton, MD, September 25-26,
1986, Ed. by G. Green, G. Patton, and M. Knox, David Taylor Naval Ship Research and Development
Center, Annapolis, MD, 1987.
5) Proceedings of the International Cryocooler Conference, Monterey, CA, August 18-19, 1988,
Conference Chaired by P. Lindquist, AFWAL/FDSG, Wright-Patterson AFB, OH.
6) Proceedings of the 6th International Cryocooler Conference, Vols. 1-2, Plymouth, MA, October
25-26, 1990, David Taylor Research Center Report DTRC-91/001-002, Ed. by G. Green and M.
Knox, Bethesda, MD, 1991.
7) 7th International Cryocooler Conference Proceedings, Vols. 1-4, Santa Fe, NM, November 17-19,
1992, Air Force Phillips Laboratory Report PL-CP-93-1001, Kirtland Air Force Base, NM, 1993.
8) Cryocoolers 8, proceedings of the 8th ICC held in Vail, Colorado, June 28-30, 1994, Ed. by R.G.
Ross, Jr., Plenum Press, New York, 1995.
9) Cryocoolers 9, proceedings of the 9th ICC held in Waterville Valley, New Hampshire, June 25-27,
1996, Ed. by R.G. Ross, Jr., Plenum Press, New York, 1997.
10) Cryocoolers 10, proceedings of the 10th ICC held in Monterey, California, May 26-28,1998, Ed.
by R.G. Ross, Jr., Kluwer Academic/Plenum Publishers, New York, 1999.
817
Author Index
Abbondante, N. 587 Collaudin, B., 567 Haynes, C.V., 577

Abedzadeh,S., 131 Crook, M., 55 Heeg, B., 621
Abhyankar, N.S., 17, 35, 45, Heiden, C., 281
699 Dadd, M.W., 169, 175 Helvensteijn, B.P.M., 475
Adachi, Y., 381 Davey, G., 175 Henry, D., 759
Ade, P.A.R., 577 Davis, T.M., l,27, 163, 175, Hill, N., 169
Ahart, M., 541 707, 729, 749 Hiratsuka, Y, 119
Anderson, J.E., 631 DeBarber, P.A., 621 Hirsch, M., 587
Arai, O., 465 deVries, E., 793 Hofmann, A., 221
de Waele, A.T.A.M., 309 Höhne, J., 597
Baik, J.H., 249 Devpura, A., 689
Bailey, P.B., 169, 175 Holland, H.J., 551, 793
DiPirro, M., 587 Hoyt, C.W., 631
Baker, G., 55 Dolan, F.X., 489 Hozumi, Y., 363, 371
Bangma, M.R., 793 Drummond, J.R., 759 Hristov, V.V., 577
Barr, M.C., 259 Duband, L., 561, 567
Benschop, A.A.J., 111 Hummon, M.R., 79
Berenschot, E., 551 Edwards, B.C., 631 Ichikawa, M., 243
Berry, D.J., 769 Elwenspoek, M., 551 Iida,T., 371
Bhandari, P., 531, 541 Epstein, R.I., 631 Ikuta, Y., 95
Bhatia, R.S., 577 Evtimov, B., 145, 155 Ilyn, M.I., 457
Blankenship, S., 27 Féger, D., 55 Imai, F., 119
Bock, J.J., 577 Flint, E.M., 739 Inoue, T., 243
Boiarski, M., 513 Fountain, T.L., 27 Izenson, M.G., 481
Bowman, R.C., Jr., 531, 541 Frederking, T.H.K., 401
Bradley, P.E., 189 Jackson, M., 587
Fujimoto, S., 213, 465 Jaco, C., 131
Bradshaw, T.W., 55 Fujishiro, H., 443
Breedlove, J.J., 489 Jamotton, P., 567
Brisson, J.G., 605 Gaby, J., 809 Jeong, E.S., 345
Bugby, D., 707, 729 Gan, Z.H., 281, 291 Jeong, S., 345
Bühler, M., 597 Gardeniers, J.G. E., 551 Jiang, N., 301
Burger, J.F., 551 Gary, J., 409 Jiang,Y.L., 301
Gibbon, J.A., 481, 489 Johnson, D.L., 155
Canavan, E., 587 Gifford, P.E., 205, 387 Johnson, R., 809
Castles, S., 649 Glaister, D.S., 63, 505, 649, Jones, W.C., 577
Ceridon, K.M., 393 769, 775
Champagne, P.J., 145 Glenn, J., 577 Kallman, J.P., 699
Chan, C.K., 163 Godden, J., 131 Kanao, K.,95
Chang, H.M., 345 Gong, L., 265 Kang, K.Y., 119
Charles, I., 229 Gong, M.Q., 523 Rang, Y.M., 213, 465
Chase, S.T., 577 Grabowski, M., 587 Kashani, A., 475
Chen, G.B., 281, 291, 301 Gschneidner, K.A., Jr., 433, Keating, B.G., 577
Cheuk, C.F., 169 449, 457, 475 Khatri, A., 513
Chuang, T.C., 401 Gully, W.J., 63, 505, 649, 775 Kiehl, W., 769
Church, S.E., 577 Kirn, M.G., 345
Chung, W-S, 125 Hackett, J., 759 Kirn, S-T, 125
Clappier, R., 145 Halouane, A., 317 Kirn, S-Y, 125
Colbert, R., 131, 163 Hanes, M., 87 Kirkconnell, C.S., 69, 259
Collaco, A., 145 Harvey, D., 131 Kittel, P., 475, 809
819
820 AUTHOR INDEX
Kobayashi, H., 273 Okamura, M., 443 Sirbi, G., 577

Korf, H., 139 Olson, J.R., 145, 155 Sixsmith, H., 481
Kotsubo, V., 145, 155, 649 Onishi, A., 381 Slaymaker, P.A., 613
Kovalenko, V., 513 Oodo, T., 213, 465 Smith, J.L., Jr., 393
Kuo, D.T., 659, 665 Oonk, R., 775 Stack, R.G., 769
Kurihara, T., 213 Orlowska, A.H., 55 Stouffer, C., 707, 729
Suzuki, Y., 95
Ladner, D.R., 189 Paine, C.G., 531 Swift, W.L., 481, 489
Lange, A.E., 577 Pan, H., 221
Le, J.P., 681 Panek, J., 587 Takamatsu, K., 371
Ledbetter, J.D., 1,749 Park, J-J, 125 ter Brake, H.J.M., 551, 793
Lee, H-K, 125 Pearson, D., 531 Theiß M., 597
Leong, J., 577 Pecharsky, A.O., 433, 449, Thummes, G., 281, 597
Lewis, M.A., 419 457, 475 Thürk, M., 327
Liang, J.T., 265, 337, 523 Pecharsky, V.K., 433, 449, Tijani, M.E.H., 309
Lieber, M.D.,775 457, 475 Tishin, A.M., 457
Little, A., 55 Peskin, A.P., 803 Tomlinson, B.J., 1, 17, 27, 35,
Liu, L., 265 Pfotenhauer, J.M., 249 45, 55, 175, 427, 505, 621,
Lloyd, J.L., 621 Phelan, P.E., 689 673, 707, 719, 729, 749
Lody, T.D., 665 Philhour, B.J., 577 Tuchinskiy, L., 427
Longsworth, R.C., 783 Phillips, C.L., 605 Tuttle, J., 587
Loutfy, R., 427 Podtcherniaev, O., 513 Tward, E., 163, 169
Luo, E.C., 523 Poncet, J.M., 229
Lybarger, G.W., 699 Price, K.D., 35, 69, 183, Umehara, I., 381
259, 649 Unger, R.Z.,79
Maeda, Y., 119 Prina, M., 541 Urbancek, V., 183
Maguire, J., 235
Mand, G.S., 759 Qiu, L.M., 291, 301 Wade, L.A., 531, 541
Marechal, J-C, 317 Wagner, R., 327
Marland, B., 707, 729 Raab, J., 131, 163, 169 Waldauf, A., 327
Marquardt, E.D., 409, 681, Radebaugh, R., 189, 409, Wang, C., 205, 387
803, 809 419, 681, 803 Wang, Z., 689
Martin, C.M., 199 Ravex, A., 229 Watanabe, N., 95
Martin, J.L., 199 Rawlings, R.M., 103, 155 Whitehouse, P.L., 499
Martin, M.L., 281, 699 Reilly, J., 35, 45, 649 Wiedmann, Th., 139
Reincke, H.A., 793 Willen, G.S., 719, 739
Mason, P.V., 577
Matsubara,Y., 119, 213, 273 Renna, T, 145 Winn, P., 235
McCormick, J.A., 481 Richards, J., 769 Wiseman, R.B., 79
Meijers, M., 111 Rijpma, A.P., 793 Wright, G.P., 505
Metzger, A., 155 Rogalla, H., 551, 793 Wu, J.F., 523
Miller, S.A., 449 Ross, R.G., Jr. 155, 637
Mills, G.L., 613 Rühlich, I., 139 Xiao, J.H., 189
Minehara, E.J., 381 Rumbles, G., 621
Russo, J.T., 259 Yamaguchi, T., 443
Miskimins, S., 103 Yamazaki, S., 443
Mord, A.J., 613 Salazar, W.E., 11 Yanagitani, T., 465
Mullié, J.C., 111 Salerno, L.J., 809 Yang, L.W., 337, 353
Murakami, M., 363, 371 Sargeant, A., 55 Yoneshige, C.H., 17, 45, 281,
Murayama, K., 119 Sato, A., 465 699
Nakane, H., 443 Sato, K., 381 Yoshida, S., 401
Nakano, A., 371 Satoh, T., 381 Yoshizawa, S., 443
Nam, K., 345 Schmauder, T., 327 Yu, J.P., 291, 301
Nast, T.C., 145, 155, 649 Schmelzel, M.E., 531 Yuan, J., 235
Nellis, G.F., 481, 489, 499 Seidel, P., 327 Yuan, S.W.K., 659, 665
Nguyen, T., 163 Seppenwoolde, J.H., 551
Sheik-Bahae, M., 631 Zagarola, M.V., 481, 489, 499
Nicholson, J.D., 449
Nogawa, M., 243 Shiraishi, M., 363, 371 Zeegers, J., 309
Numazawa, T., 213, 443, 465 Shirron, P., 587 Zhang, L., 265
Sidi-Yekhlef, A., 235 Zhou, S.L., 273
O'Gallagher, A., 409 Simmons, D.W., 63, 505 Zhou, Y., 337, 523
Oellrich, L., 221 Simon, Y., 317 Zhu, S., 243
Subject Index
Acoustic streaming, visualization of, 371 10K hybrid Stirling/J-T cooler, 505
Acoustic wave cryocooler for -60°C, 309 12K Stirling for 6K Hybrid J-T, 63
ADR (see Magnetic refrigerators) 6K J-T for NGST Instruments, 775
AIM (AEG Infrarot-Module), 139 HIRDLS Stirling integration, 769
Air Force Research Lab (AFRL): optical cryocooler research, 613
Astrium 10K Stirling cooler, 55 Bearings:
Ball 10K hybrid Stirling/J-T cooler, 505 flexures in tactical coolers, 103, 111, 155
char. and endurance update, 17 flexures in long-life compressor, 169
cooler contamination lessons, 649 BEI (see BAE Systems)
fail-safe experiment stand, 699 Brayton cycle cryocoolers (see Turbo-Brayton
failure analysis of Creare SSRB, 673 cryocoolers)
failure analysis of TRW 3503, 673 British Aerospace (BAe) cryocoolers (see Astrium
program requirements overview, 1 cryocoolers)
Raytheon 35K 1.2W PSC cooler, 35 California Institute of Technology, 577
Raytheon 60K 2W PSC cooler, 45 (see also Jet Propulsion Laboratory)
reliability initiatives, 27 CEA/DRFMC (France):
technology for cryogen storage, 749 sub-Kelvin coolers for space, 561, 567
vibration reduction in compressors, 175 GM-type pulse tube refrigerators, 229
Aisin Seiki 4K pulse tube, 243 Chinese Academy of Sciences, Cryogenics Lab:
American Superconductor Co., 235 4K GM/PT hybrid refrigerator, 265
Ames Research Center (NASA): mixed-gas auto-cascade J-Ts, 523
regenerator materials for 20K PT, 475 multi-bypass PT refrigerator, 337
zero-boil-off cryogen storage, 809 Chiyoda Corp., 363
Ames Laboratory (Iowa State Univ.): Compressors:
erbium regenerator materials, 433 Oxford/Hymatic linear for PT, 169
properties of polycrystalline 457 flexure bearings in, 103, 111
regenerator materials for 20K PT, 475 vibration reduction in, 175
rare earth powder manufacturing, 449 Conductance, measurements of:
Anti-stokes fluorescence: beds of metal spheres, 419
in Los Alamos optical cooler, 631 silver filled epoxy junctions, 689
using molecular dyes, 621 Contamination:
in Ball optical cooler, 613 space cooler contamination lessons, 649
APD Cryogenics: temp. dependence of outgassing, 659
J-T cryostats for cryosurgery, 783 COM DEV, 759
small throttle-cycles with mixed gases, 513 Creare Cryocoolers:
Applications (see Integration with cryocoolers) 4-35K Brayton cooler developments, 481
Arizona State University, 689 failure analysis of Creare SSRB, 673
Astrium cryocoolers (formerly BAe and MMS): flexible Brayton cryocooler model, 499
10 K 2-stage Stirling cooler, 55 life and reliability of Brayton coolers, 489
50-80K MOPITT cooler performance, 759 Cryomech, Inc:
Atlas Scientific, 475 4K pulse tube, 205
BAE Systems (formerly BEI): 20 to 40K GM cryorefrigerators, 387
cooler contamination study, 659 CSA Engineering, 739
life test results, 665 Daikin Industries:
Ball Aerospace cryocoolers: 2. 4K two-stage PT using 213
cooler contamination lessons, 649 80K-1W pulse tube for HTS filters, 119
821
822 SUBJECT INDEX
regenerator material for sub-4K, 465 Hybrid cryocoolers:

Database of cryo material properties, 681 4K GM/PT hybrid refrigerator, 265
DC flow, effect on PT efficiency, 371 Ball 6K hybrid J-T/Stirling, 63
DRS tactical Stirling coolers: Ball 10K hybrid J-T/Stirling cooler, 505
1.75W-77K performance characterization, 155 Raytheon Stirling with PT expander, 259
gamma-ray pulse tube cooler, 155 thermoacoustic driven PT, 301
with flexure springs, 103, 155 Hymatic Engineering Co., 169
Dynacs Engineering Co., 17, 45, 673, 699
IGC-APD Cryogenics (see APD Cryogenics)
Ecole Normale Superieure, 317 IGC-Polycold Systems, 513
Eindhoven Univ. of Technology, 309 Imperial College, London, 621
Electric field emissions (see EMI/EMC measure- Integration of cryocoolers with:
ments) bolometers, 577
Electromagnetic interference (see EMI/EMC cryosurgical applications, 783
measurements) external cold-end fluid loop, 393
Electronics: fail-safe experiment stand, 699
cooler drive and control, 145 flight electronics, 145
EMI/EMC measurements: gas liquefiers, 79, 199, 749, 803, 809
for SQUID-based heart monitor, 793 gimbal thermal transport system, 707
Erbium regenerator materials (see Regenerators) heart monitor, 793
European Space (ESA-ESRTC) activities, 567 heat pipes (see Heat pipes)
Exergy analysis, 69 HTS applications (see HTS applications)
Experiment stand, fail-safe, 699 long-term cryogen storage, 749, 809
Polatron GHz receiver, 577
Flexure spring bearings (see Bearings) space experiments (see Space experiments)
Friedrich-Schiller Univ., Jena, 327 space instruments (see Space instruments)
SQUIDs, 793
Gas-gap heat switches (see Heat switches) thermal switches (see Heat switches)
Georgia Tech Research Institute, 27 vibration isolation systems, 719, 739
Gifford-McMahon Cryocoolers: Iwate Univ., 443
4K GM/PT hybrid refrigerator, 265
Cryomech 20 to 40K single-stage, 387 J-T cryocoolers:
optimum intermediate temperatures for, 401 APD J-T for cryosurgery, 783
Sb regenerator compounds for 4K, 443 Ball 6K J-T for NGST Instruments, 775
Sumitomo, operation below 2K, 381 Ball 6K hybrid J-T/Stirling cooler, 63
with external cold-end flow loop, 393 Ball 10K hybrid J-T/Stirling cooler, 505
Gimbal thermal transport system, 707 mixed-gas auto-cascade J-Ts,523
Glenn Research Center (NASA), 809 small throttle-cycles with mixed gasses, 513
Goddard Space Flight Center (NASA): Jet Propulsion Lab:
cooler contamination lessons learned, 649 cryocooler reliability prediction, 637
continuous magnetic refrigerator, 587 gamma-ray PT cooler, 155
Create 4-35K Bray ton developments 481 Planck sorption compressor element, 531
flexible Brayton cryocooler model, 499 Planck sorption performance pred. tools, 541
life and reliability of Brayton coolers, 489 TES pulse tube cryocooler, 131
Johnson & Johnson Co., 189
Heat switches: Joule-Thomson Cryocoolers (see J-T cryocoolers)
JPL design and reliability of, 637 Justus-Liebig Univ., Giessen, 353
Swales development and testing of, 729
Heat pipes: Kennedy Space Center (NASA), 809
for waste heat removal, 707 Kogakuin Univ., 443
Heat conduction (see Conductance measurements) Korea Adv. Institute of Science and Tech., 345
Heat transfer in pulse tubes, 345
High temperature superconductor applications LG Electronics 65K-5W PT, 125
(see HTS applications) Life estimation method, 665
HIRDLS cryocooler integration, 769 Life test results:
Hong Ik Univ., 345 BAE tactical coolers, 665
HTS applications: Creare SSRB turbo Brayton, 673
coolers for, 79, 95, 87, 119, 139, 235, 291 TRW 3503 pulse tube, 673
fetal heart monitor using SQUIDs, 793 Liquefier:
RF filters for telecommunications, 119, 125 89K-20W PT for oxygen liquefier, 199
SUBJECT INDEX 823
vapor precooling with PT, 803 LosAlamos/UNM, 631

zero-boil-off cryogen storage, 809, 749 Imperial College London, 621
Lockheed Martin Adv. Tech. Center (Palo Alto): Outgassing, temperature dependence of, 659
35K LADS Stirling cooler, 317 Oxford University:
cooler contamination lessons, 649 high-efficiency compressor, 169
cooler electronic controller, 145 vibration reduction in compressors, 175
gamma-ray PT cooler w/DRS comp., 155
miniature PT coolers for space appl., 145 Planck sorption coolers, 531, 541
Lockheed Martin Astronautics (Denver), 189 Polatron, sub-Kelvin coolers for, 577
Lockheed Martin Commun. and Power, 145 Properties, database of cryogenic, 681
Los Alamos National Laboratory: Pulse tube cryocoolers:
optical cooler development, 631 2. 4K two-stage Daikin using 213
4K two-stage Aisin Seiki, 243
Magnetic refrigerators: 4KCryomech PT405, 205
continuous ADR for below 50 mK, 587 4K GM/PT hybrid refrigerator, 265
PT-cooled ADR for below 96 mK, 597 20K two-stage Karlsruhe, 221
regenerator materials for (see Regenerators) 50K-60W single-stage for HTS, 235
refrigerant materials for (see Refrigerants) 65K-5W LGE air-cooled PT, 125
Massachusetts Inst. of Technology: 65K-0.3W miniature Lockheed, 145
dissipation by metal bellows, 605 70K-67W high cooling power, 327
G-M cooler with external flow loop, 393 80K-1W with Ricor compressor, 317
Materials: 80K-1W Daikin for HTS filters, 119
properties database, 681 80K-1.5W JPL/LM gamma-ray cooler, 155
conductance of (see Conductance) 89K-20W for oxygen liquefier, 199
refrigerants (see Refrigerants) 95K Raytheon hybrid Stirling/PT, 183
regenerator (see Regenerators) GM-type PT devel. at CEA, 229
Matra Marconi coolers (see Astrium coolers) NIST/Lockheed miniature flight, 189
MER Corp., 427 TRW 3503, 673
Meisei Univ., 443 TRW 6020, 17
Mesoscopic Devices, 199 TRW 95K high efficiency, 169, 163
MetroLaser, Inc., 621 TRW TES, 131
Miniature sorption coolers, 551 Pulse tube theory and investigations:
Mission Research Corp., 749 2.4K achieved using 213
MITI, Mech. Engin. Lab, 363, 371 4K 2-stage development, 243
Mixed gasses for refrigerants (see Refrigerants) 20K 2-stage development, 221
Molecular dyes, 621 50K-60W development for HTS, 235
MOPITT cooler performance, 759 70K-67W four-valve development, 327
Moscow Power Engineering Inst., 513 80K-1W with Ricor compressor, 317
Moscow State Univ. (Russia), 457 combined cooling and freezing cycle, 291
Multistage optim. using exergy analysis, 69 dc flow visualization, 371
exergy analysis for multi-stage optimization, 69
Nat'l Inst. of Standards and Tech. (see NIST) gas dynamics, numerical study of, 363
NGST, coolers for, 775 GM-type PT development at CEA, 229
Ninon University: He and mixtures as working fluid, 281
2.4K two-stage PT using 213 heat transfer in PT, numerical study, 345
80K-1W pulse tube for HTS filters, 119 hybrid Stirling /PT expander, 259
using 3He for sub-4K PT, 273 multi-bypass pulse tube, 337
NIST: Oxford/Hymatic compressor for PT, 169
database of cryo material properties, 681 PT design for GM-type compressor, 249
heat conduction through metal spheres, 419 regenerator materials for 20K, 475
NIST/Lockheed miniature flight PT, 189 shuttle loss in pulse tubes, 353
regenerators with heat removal, 409 sub-4K operation using 3He, 273
zero-boil-off cryogen storage, 809 thermoacoustic driven pulse tube, 301
vapor precooling in PT liquefier, 803 vapor precooling in PT liquefier, 803
Nat'1 Res. Inst. for Metals (Japan):
regenerator materials for sub-4K, 465 Queen Mary & Westfield College, 577
regenerator prop, of Sb compounds, 443
RAL (see Rutherford Appleton Laboratory)
Optical cooler developments: Rare earth compounds (see Regenerators)
Ball Aerospace research, 613 Raytheon Corp., Philadelphia, 401
824 SUBJECT INDEX
Raytheon Systems (formerly Hughes Aircraft): Stirling cryocoolers:

cooler contamination lessons, 649 AIM space cooler program, 139
35K 1.2W PSC Stirling cooler, 35 Astrium 10K Stirling cooler, 55
60K 2W PSC Stirling cooler, 45 BAE Systems tactical life tests, 665
95K high efficiency hybrid Stirling /PT, 183 Ball 1.5W-55K HIRDLS, 769
exergy flow for multi-stage optimization, 69 Ball 6K hybrid J-T/Stirling cooler, 63
two-stage hybrid Stirling /PT expander, 183, 259 Ball 10K hybrid J-T/Stirling cooler, 505
Refrigerants: DRS linear tactical coolers, 11, 103
mixed gases in J-Ts, 513, 523 DRS 1.75 W 77K cooler, 155
PT with He and mixed gases, 281 exergy analysis for multi-stage optimization, 69
3
He for use below 4K, 273 linear for tactical weapon systems, 11
Regenerators: MMS/Astrium 50-80K MOPITT, 759
behavior with intermediate heat removal, 409 Raytheon 35K 1.2W PSC, 35
erbium-based materials for, 433, 475 Raytheon 60K 2W PSC, 45
GAP material for sub-4K, 465 Raytheon hybrid with PT expander, 183, 259
in 2.4K PT, 213 reliability prediction methodology, 637
with 3He in l.65K GM, 381 Signaal-USFA high reliability, 111
heat conduction through metal spheres, 419 STI improvements for HTS, 87
with 3He in 1.64K GM, 381 Sumitomo 70K-6W for HTS, 95
properties of polycrystalline Gd 2 In, 457 Sunpower M87 with 7.5W at 77K, 79
properties of Sb compounds for 4K, 443 temperature dependence of outgassing, 659
parallel microchannel, 427 Storage, long-term of cryogens, 749, 809
rare earth powders, manufacturing of, 449 Sub-Kelvin coolers:
Reliability of cryocoolers: continuous ADR, 587
1st-order life estimation method, 665 dissipation in bellows for, 605
AFLR reliability initiatives, 27 pulse tube precooled ADR, 597
AFRL char, and endurance update, 17 sorption coolers for space, 561, 567
cooler contamination lessons, 649 Sumitomo Heavy Industries:
inspection results on TRW 3503, 673 70K-6W Stirling for HTS appl., 95
inspection results on Creare SSRB, 673 GM cooler below 2K, 381
JPL reliability prediction methodology, 637 Sunpower, Inc. cryocoolers, 79
life and reliability of Brayton coolers, 489 Superconductor applications (see HTS
temperature dependence of outgassing, 659 applications)
Requirements, space cooler overview, 1 Superconductor Technologies Inc. (STI), 87
Reverse-Brayton coolers (see Turbo Brayton coolers) Superfluid helium as working fluid, 605
Rutherford Appleton Laboratory: Swales Aerospace:
10K 2-stage Astrium Stirling, 55 cryogenic thermal switch, 729
gimbal thermal transport system, 707
Signaal USFA high reliability coolers, 111
Silver-filled epoxy conductance, 689 Taiyo Toyo Sanso Co., 401
Software: Technology Applications, Inc., 719,739
Turbo Brayton cooler model, 499 Texas Instruments (see DRS cryocoolers)
Sorption cryocoolers: Thermal conductivity (see Conductance
microminiature, 551 measurements)
Planck hydride compressors, 531 Thermal switch (see Heat switch)
Planck performance prediction tools, 541 Thermoacoustic:
Polatron sub-Kelvin, 577 refrigerator for -60°C, 309
sub-Kelvin for space application, 561,567 thermoacoustic driven PT for 138K, 301
Space experiments: Throttle-cycle (see J-T cryocoolers)
International Space Station, 139 Toshiba Corporation, 443
NIST/Lockheed Martin miniature PT, 189 TRW cryocoolers:
Space instrument missions: 3503 pulse tube cooler, 673
FIRST, 567 6020 pulse tube cooler, 17
HIRDLS, 769 95K high efficiency PT, 163
MOPITT, 759 Oxford/Hymatic compressor, 169
NGST, 775 TES pulse tube cooler, 131
Planck, 531, 541, 577 Tsukuba Magnet Laboratory, 213, 465
TES, 131 Turbo Brayton coolers:
Stanford University, 577 failure analysis of Creare SSRB, 673
Starmet, rare earth powders, 449 flexible Brayton cryocooler model, 499
SUBJECT INDEX 825
life and reliability of, 489 Vibration:

developments for 4 - 35K range, 489 isolation system, 719, 739
reduction in compressors, 175
Univ. of Colorado, 577
Univ. of California, UCLA, 401 Worcester Polytechnic Institute, 587
Univ. of Giessen, 281, 597
Univ. of Karlsruhe, 221 Yokohama National Univ., 381
Univ. of New Mexico, 631
Univ. of Toronto, 759 Zero-boil-off cryogen storage, 749, 809
Univ. of Tsukuba, 363, 371 Zhejiang Univ.:
Univ. of Twente, 551, 793 PT with freezing cryogen, 291
Univ. of Wisconsin, 249 PT with He and mixed gases, 281
US Army Night Vision, 11 thermoacoustic driven PT, 301

Thom Davis, B. J. Tomlinson, Jim Ledbetter (Auth.), R. G. Ross Jr. (Eds.) - Cryocoolers 11-Springer US (2002)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Thom Davis, B. J. Tomlinson, Jim Ledbetter (Auth.), R. G. Ross Jr. (Eds.) - Cryocoolers 11-Springer US (2002)

Uploaded by

Copyright:

Available Formats

CRYOCOOLERS 11

A publication of the International Cryocooler Conference

KLUWER ACADEMIC PUBLISHERS

Ronald G. Ross, Jr.

CONFERENCE CO-CHAIRS PROGRAM COMMITTEE

PROGRAM CHAIRMAN ADVISORY BOARD

Government Cryocooler Development and Test Programs 1

Status of Programs for the DoD Family of Linear Drive Cryogenic

Air Force Research Laboratory Cryocooler Characterization and

Air Force Research Laboratory Cryocooler Reliability Initiatives ........... 27

Space Stirling Cryocooler Developments 35

Characterization of Raytheon’s 60 K 2 W Protoflight Spacecraft

Development of a 12 K Stirling Cycle Precooler for a 6 K Hybrid

Thermodynamic Optimization of Multi-Stage Cryocoolers ................ 69

Long-Life Tactical and Commercial Stirling Coolers 79

Performance and Reliability Improvements in a Low-Cost Stirling

Development of a Long-Life Stirling Cryocooler ........................ 97

Flexure Springs Applied to Low-Cost Linear Drive Cryocoolers ............ 103

High Reliability Coolers under Development at Signaal-USFA ............. 111

Long-Life Commercial Pulse Tube Coolers 119

Development of a 5 W at 65 K Air-Cooled Pulse Tube Cryocooler ........... 125

Space Pulse Tube Cryocooler Developments 131

The AIM Space Cryocooler Program ................................. 139

Miniature Pulse Tube Cryocooler for Space Applications ................. 145

Gamma-Ray Pulse Tube Cooler Development and Testing ................ 155

High Efficiency Pulse Tube Cooler ................................... 163

High Performance Flight Cryocooler Compressor ....................... 169

Vibration Reduction in Balanced Linear Compressors .................... 175

95K High Efficiency Cryocooler Program .............................183

Design and Test of the NIST/Lockheed Martin Miniature Pulse Tube

Low-Cost Pulse Tube Liquefier for In-Situ Resource Utilization ............199

GM-Type Pulse Tube Coolers for Low Temperatures 205

Experimental Study of a 4K Pulse Tube Cryocooler ..................... 213

Developments on Single and Double Stage GM Type Pulse Tube

Two-Stage 4K Pulse Tube Refrigerator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243

Compressor-Specific Design of a Single Stage Pulse Tube Refrigerator ....... 249

Hybrid Cryocoolers Using Pulse Tubes 259

Numerical Study of a New Type of 4K GM/PT Hybrid Refrigerator . . . . . . 265

Thermally Actuated 3He Pulse Tube Cooler .......................... 273

Investigation of Helium and Nitrogen Mixtures in a Pulse Tube

Pulse Tube Refrigeration with a Combined Cooling and Freezing

Experimental Investigation of a Pulse Tube Refrigerator Driven

Design, Development, and Operation of a Thermo-Acoustic

Pulse Tube Analysis and Experimental Measurements 317

Investigation of a Single Stage Four-Valve Pulse Tube Refrigerator

Analysis and Experimental Research of a Multi-Bypass Version

Experimental Study of the Heat Transfer in Pulse Tubes ................. 345

Shuttle Loss in Pulse Tubes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353

Numerical Study of Gas Dynamics Inside of a Pulse Tube Refrigerator . . . . . . 363

Visualization of DC Gas Flows in a Double-Inlet Pulse Tube

GM Refrigerator Developments 381

High Efficiency, Single-Stage GM Cryorefrigerators Optimized for

Remote Cooling with a G-M Cryocooler by Use of Cold Electro-

Optimum Intermediate Temperatures of Two-Stage Gifford-McMahon

Regenerator Analysis and Materials Developments 409

Measurement of Heat Conduction through Metal Spheres ................ 419

Innovative Technology for Low Temperature Regenerators ............... 427

Ductile, High Heat Capacity, Magnetic Regenerator Alloys for the

Low Temperature Properties of HoSb, DySb, and GdSb .................. 443

Manufacturing Considerations for Rare Earth Powders Used in

Magnetothermal Properties of Polycrystalline Gd2In . . . . . . . . . . . . . . . . . . 457

New Regenerator Material for Sub-4 K Cryocoolers ...................... 465

Turbo-Brayton Cryocooler Developments 481