You are on page 1of 10

1 962 SOMER: REDUNDANT PULSE AMPLIFIER 7

REDUNDANT PULSE AMPLIFE:R'


Toivo A. Somer
Research Laboratories Division
The Bendix Corporation
Southfield (Detroit), Michigan

Summary satellite electronics and operate in an ambient


of -80°C ±2000C. For improved signal-to-noise
A redundant transistorized pulse amplifier ratio, pulse preamplifiers are placed near the
is described, composed of a remote preamplifier photodetectors.
operating at -80°C and a main pulse amplifier.
Active parallel redundancy is used in the pre- Specifications
amplifier as well as in the main amplifier. Com-
mon feedback between the parallel redundant The redundant pulse amplifier was designed
amplifiers stabilizes the voltage amplification in to satisfy the following requirements:
spite of component failures in one signal path.
Delay-line pulse shaping is employed to reduce (1) To operate for a year in an electronic
noise and to eliminate the detrimental effects of system whose required probability of
pulse pile-up. success is 0.95.

Introduction (2) To be sensitive to charge pulses of


4 x 10-14 coulombs.
Background and Scope
(3) To be capable of fast recovery from
With the advent of satellites and rockets overloads, 1000 times the threshold.
that must operate unattended for long periods of
time, the electronic circuit design must meet not (4) To handle random pulses at a rate
only the functional but also the reliability require- of 4000 per second with less than one
ments. These reliability requirements are not percent coincidence loss.
always met by the use of premium quality circuit
components of the type developed for the (5) The preamplifier must operate reliably
MINUTEMAN and POLARIS missile programs, at -80 ± 20°C ambient temperature.
but require the use of special techniques to make The pulse amplifier should give satis-
the circuits capable of tolerating catastrophic factory operation in 20 ±400C environ-
component failures and inevitable component de- ment.
gradation. In this report a pulse amplifier is
described whose life expectancy has been extended In view of the complexity of the problem, i.e., the
by the use of active parallel subcircuit redundancy. amplifier sensitivityl requirement, the severe
The pulse amplifier can tolerate considerable environmental conditions, and the long life ex-
circuit parameter degradation and catastrophic pectancy, it was decided to design first a pulse
component failures. amplifier capable of satisfying all the require-
ments with the possible exception of probable
The function of the instrument is to amplify survival for one year. However, in developing the
the pualses from the photomultiplier to a level basic amplifier, the principles for improved relia-
suitable for pulse-height discrimination. The bility were employed in selection of the circuit
photomultipliers are remote from the main configuration and circuit components.

*The infornation in this paper is based on a subcontract performed Basic Circuits


by the Bendix Corporation, Research Laboratories Division, for
the Princeton University Observatory under Contract AF19(604)- Unit Amplifier
4972 of the U.S. Airforce Electronic Systems Division. This paper
is a summary of the redundant pulse amplifier section of the The term "Unit Amplifier' used in this
Bendix Final Report No. 1919 submitted September 30, 1961. section applies to both the preamplifier and the
Orig. T SO' rec 'ci bv P'S 3/30,/, revised r rec'd 7/19/62.

Authorized licensed use limited to: The University of Utah. Downloaded on September 29,2022 at 22:53:29 UTC from IEEE Xplore. Restrictions apply.
Q IRE TRANSACTIONS ON NUCLEAR SCIENCE November
amplifier since these circuits are identical. The PULSE
PRE-
MAIN
PULSE
AMPLIFIER AMPLIFIER
Unit Amplifier consists of three direct-coupled A IN
PHAT DELAY LINE
transistor stages. The first two stages utilize A. MULTIPLIER PULSA

the common emitter configuration (Ql and QZ), 'IN RI

and the last stage is an emitter follower (Q3), CIN = TOTAL CAPACITANCE ASSOCIATED
WITH THE PREAMPLIFIER INPUT,
CONSISTING OF THE PREAMPLIFIER
PULSES FROM P.M. T, RIN C IN INPUT CAPACITANCE, PHOTOMULTIPLIER
APPEARING ACROAN OUTPUT CAPACITANCE AND SIGNAL
M AND R
IN NN CABLE CAPACITANCE
_ - R lN EFFECTAVE PREAMPLIFIER
TOTAL
TIME INPUT RESISTANCE
T. RI CN
PULSES AFTER A.IN
C. PREAMPLIFIER ANM \EIN - IF RIN CIN AT
PULSE SHAPING CIN
_§L~~~~~~~~TM _[I_ -1O0t

Fig. 3-Pulse amplification and the associated wave


forms.

and it can be shown that the closed ioop voltage


amplification (Eout /Ein ) can be approximated by:

E
R 1 +R F
out
(1)
E R
in F
Fig. 1-Basic unit amplifier circuit diagram. The error associated with Equation (1) is small
if open loop voltage amplification G is high; hence
as shown in Figure 1. This type of amplifier is considerable variation of circuit parameters can
characterized by high open loop voltage amplifica- be tolerated.
tion, proportional to the product of the hFE of the
first two transistors. The open loop voltage Adequate stability of bias conditions is pro-
amplification is sensitive to transistor parameter vided by the large emitter resistor Rl, RZ and
variations caused by temperature, aging, supply R3, with RZ and R3 bypassed for signal fre-
voltage, and manufacturing. Gain stability over quencies.
the wide temperature variations encountered in
the satellite application and for the required long The required signal threshold of 4 x 10
life, is accomplished by the liberal use of nega- coulombs per event imposes the need for low
tive voltage feedback. noise in the preamplifier. Noise is reduced
first by electrostatic shielding and by eliminating
Referring to Figure 2, the voltage appearing all ground loops. Employing a transistor with
across the resistor RF is fed back to the input, high hFE and a large fab at small collector cur-
rents in the preamplifier first stage, will further
reduce the noise.

Pulse Shaping
The output of a photomultiplier is in the
form of a charge increment AQ which is de-
posited on the effective input capacitance of the
preamplifier. As depicted in Figure 3, each
charge increment AQ causes a voltage step AEin
followed by an exponential decay. To prevent the
amplifier from overloading, delay-line pulse
shaping is used in this application because of its
advantages over other pulse-shaping methods.3' 4
A typical delay-line pulse-shaping network is
shown in Figure 4, together with the associated
Fig. 2-Basic unit amplifier block diagram. wave forms.

Authorized licensed use limited to: The University of Utah. Downloaded on September 29,2022 at 22:53:29 UTC from IEEE Xplore. Restrictions apply.
1962 SOMER: REDUNDANT PULSE AMPLIFIER 9
Redundancy for Improved Reliability
General Redundancy Considerations

The first step towards a reliable circuit de-


sign is the use of highly reliable, conservatiovely
rated, circuit components and a sound circuit de-
A. sign. After designing a reliable basic circuit,
redundancy may be used for further increasing the
probability of system survival,
RINCIN
Using the published failure rates of com-
ponents in Table 1, it is apparent that the reliability
of the basic circuit will be dominated by the
number of transistors. The application of a re-
dundancy technique to this circuit will then depend
on the best method of improving the circuit

-V INPUT FROM PREAMPLIFIER


tolerance to transistor failures.
V - REFLECTED PULSE
-P - SHAPED PULSE, EQUAL TO THE
VECTOR SUMOF V +[-VI
The two most appropriate schemes are:
DELAY TIME
(1) Circuit redundancy, which entails
-

the use of two identical amplifiers


Fig. 4-Pulse shaping network and the associated (three transistors each) with
wave forms. common input and output circuits.

Table 1-Failure Rates

** STARTING PRNEO
MINUTE MARTIN ELECTRO POINT PRNCETON
MAN CO. TEC; STL PONT, (HUFNAGEL)
AUTONETICS
RESISTORS 0.00002 0.003 0.025 0.0015 0.04 0.011
CAPAC ITOR
SMALL VALUE 0.00001 0.0025 0.005 0.001 0.001 0.0011

ARGEITORUE
CAPAC ITOR
0.00012 0.012* {

TANTALUM
CAPACITOR 0.010* ll

TRANSISTOR 0.00360 0.05 0.045 0.018 0.07 0.03


POWER
TRANSISTOR 0.01 0. l*
DIODES 0.000035 0.02 0.30 0.0036 0.02 0.0045
ZENER
00100 4
DIODES 0.00140 0.014l
RECTIFIERS 0.00150 0.015*b
DELAY LINES* 0.00010 0.001*
TRANSFORMERS 0.00002 0.001* l l

NOTE: Failures in percent per 1000 hours =failures per 105 unit-hours.
*Indicates estimates by Bendix to fit general pattern.
Referred to as "commercial failure rates" in the text.

Authorized licensed use limited to: The University of Utah. Downloaded on September 29,2022 at 22:53:29 UTC from IEEE Xplore. Restrictions apply.
10 IRE TRANSACTIONS ON NUCLEAR SCIENCE November

minimum effect on the bias points


of the other signal path. With stage
redundancy, this type of indepen-
dence is much harder to achieve,
if not impossible.

(3) According to experimental data,


lower noise is obtained with cir-
cuit redundancy.
Fig. 5-Redundant pulse amplifier block diagram. Functional Redundant Pulse Amplifier
(2) Stage redundancy, which consists The complete redundant pulse amplifier is
of coupling the three transistor composed of four identical Unit Amplifiers in a
stages to three identical stages parallel- redundant configuration. Active redun-
and providing cross-coupling at dancy is utilized on the subcircuit level. This is
each stage. accomplished by two independent preamplifiers
within a common feedback loop, and also by two
Stage redundancy offers the most significant independent main amplifiers within a common
reliability improvement if collector-to- emitter feedback loop, as shown in Figure 5. The feed-
shorts are neglected. Circuit redundancy does back resistor shared by the preamplifiers main-
not match the reliability improvement of stage tains the composite preamplifier output for most
redundancy, but does tolerate collector-to- of the failures occurring in one or the other pre-
emitter shorts of the transistors. amplifier. Constant gain is maintained similarly
by the feedback resistor shared by the main pulse
The choice between the two approaches must amplifiers. External to the preamplifier and
then depend on what percent of the transistor main amplifier feedback loops, the preamplifier
failure rate is determined by the collector-to- inputs are isolated from each other as a protec-
emitter shorts. This is complicated, due to the tion against preamplifier input shorts. The pre-
limited data published on the breakdown of amplifier outputs are likewise isolated from each
transistor failures by failure mode. other. Identical protective measures are em-
ployed in the redundant main pulse amplifier.
From the analysis of the two redundant
circuit configurations, it was concluded that
circuit redundancy should be employed until there Redundant Pulse Amplifier Circuit
is sufficient statistical proof (in the form of
actual transistor failure data) to indicate that The redundant pulse amplifier circuit dia-
shorts between the transistor electrodes (base, gram, composed of four Unit Amplifiers, is shown
emitter, collector) constitute only a minute frac- in Figure 6. The basic Unit Amplifier circuitry
tion (much less than 2 percent) of the total failure is unaltered except for the minor circuit changes
modes. or additions required to produce the redundant
configuration. The signal from the photomultiplier
Circuit redundancy was used in these pulse is coupled to the preamplifier input, Ql and Q2,
amplifiers for the reasons listed below: through Cl, R1 and CZ, RZ. The capacitors Cl
and C2 provide dc decoupling, and the resistors
(1) Circuit redundancy results in an Rl and R2 provide signal decoupling, thereby
amplifier which maintains its preventing signal shunting, should one
normal operating characteristics preamplifier input short. The feedback resistor
even when short circuits do occur R51 is now shared by the two redundant preampli-
in transistors or in passive cir- fiers. Under normal operation, both preamplifiers
cuit components. contribute equally to the feedback voltage appearing
across R51. However, if for any reason the con-
(2) The two redundant circuits of tribution to this feedback voltage from one pre-
the pulse amplifier are completely amplifier decreases, the other preamplifier will
independent from each other. Any assume the load and thus make up for the decrease.
failure in one signal path has the As a result, considerable signal current variation

Authorized licensed use limited to: The University of Utah. Downloaded on September 29,2022 at 22:53:29 UTC from IEEE Xplore. Restrictions apply.
1962 SOMER: REDUNDANT PULSE AMPLIFIER 11

r-
- -

REDUNDANT PRE AMPLIFIER - -


REDUNDANT MAIN PULSE AMPLIFIER
LI R49 LS
I R7
C5 R9 il a
R
i R 8.5 V

CCs Cii
I k C3
.R3
I I~~~~~~~a
I~~~~~0
Ql

II I I~~~C
Rc ,R13 tRass
R15
'-R21
RU
C9 R19 13
adI

T- -8.5 V

1-OUTPUT

|
TOTAL CAPACITANCE

_ _,
F

Fig. 6-Redundant pulse amplifier circuit diagram.

1.05

1.00"-
I--

0.95-
L
0.0 a- CONSTANT PULSE INPUT
I-a- AQIN 2 2.7 xIO13 COULOMBS
120 PULSES PER SECOND
0.90- l 260C AMBIENT TEMPERATURE
LUJ
C:3

Szz
O0 0
0.85

0.80 -

OD
8
0.75- (L
0 5 10 15 20 25 30 35 40 45
DC SUPPLY VOLTAGE (VOLTS)
SUM OF THE PLUS AND MINUS SUPPLY POTENTIALS

Fig. 7-Redundant pulse amplifier gain with DC supply voltage variations at room temperature.

Authorized licensed use limited to: The University of Utah. Downloaded on September 29,2022 at 22:53:29 UTC from IEEE Xplore. Restrictions apply.
12 IRE TRANSACTIONS ON NUCLEAR SCIENCE November
can be tolerated in one preamplifier, including a indicates that the redundant pulse amplifier gain
complete failure! is constant within +0.5 and -1.0 percent for supply
voltage variations from +4.5 v to +20 v for the
The input from each preamplifier is coupled positive supply and with the negative supply volt-
independently through separate coaxial cables to ages of -4.5 v and -20 v respectively.
the main amplifier housing. Protection against
short-circuiting the output of either preamplifier The preamplifier gain variations at low
is provided by the delay-line terminating resistors temperatures are shown in Figure 8. During
R23 and R24. Constant signal voltage is impressed these tests, the main pulse amplifier ambient
upon the delay line for constant signal input, temperature was maintained at +65 C, and the
provided that at least one of the preamplifiers is preamplifier ambient temperature was varied in
operating. The resistors R25 and R26 protect the
delay line from being shunted in case of an ampli-
fier input short. The capacitors C15, C16, C17,
and C18, provide dc decoupling. The gain control
for the main amplifier is the same as for the re-
dundant preamplifier.

Experimental Test Data

Temperature and Supply Variation Effects

The 12 GE Type 2N338 transistors used in


all tests were randomly selected from units whose
direct current transfer ratio was limited to the
range of 67 < h < 82.
FE-=
The redundant pulse amplifier parameters
.F
are summarized in Table 2. Many of these para-
meters would be improved by a factor of two if no
redundancy were used. Hence, the price that is
paid for extended life expectancy is a slight
initial performance degradation.

The redundant pulse amplifier voltage ampli-


fication as a function of dc supply voltage at room
temperature is shown in Figure 7. This curve

Table 2-Redundant Pulse Amplifier Parameters

INPUT CAPACITANCE........................ 40 MICROMICROFARADS AQIN 10 14COULOMB


x

INPUT RESISTANCE .......... 70 KILO-OHMS 120 PULSES/SEC

OUTPUT IMPEDANCE .... ...... 320 OHMS


Fig. 8-Redundant preamplifier gain with supply
OVER-ALL VOLTAGE GAIN (250C) . 20 voltage and ambient temperature variations.
PULSE RISE
TIME.....02MCOEND
.
(AFTER DELAY LINE) (10% - 90%) the range from -1000C up to -60 0C. (It should be
PULSE FALL TIME noted that the preamplifier and the main amplifier
(AFTER DELAY LINE) (10o - 90%). .25 MICROSECONDS used in these temperature tests were constructed
OVER-ALL AMPLIFIER RESOLUTION TIME from standard +5 and *10 percent carbon resis-
WITH QIN <4 x i10-3 COULOMBS . 2 MlCROSECONDS
tors and metalized paper capacitors with a temper-
OVER-ALL AMPLIFIER RESOLUTION TIME ature rating from -300C to + 850C.) The pre-
WITH QIN . 2 x 10-11 COULOMBS ---....4 MICROSECONDS amplifier gain is constant within 44 percent over
EQUIVALENT PEAK INPUT NOISE . 1 x 10X14 COULOMBS the temperature range from -1000C up to -600C,
TOTAL POWER CONSUMPTION X8_ without any special transistor selection, except
(AT +8.5 VOLT SUPPLY) .260 MLLIWATTs that 67 < hFE <. 82. By selecting high hFE
0.

Authorized licensed use limited to: The University of Utah. Downloaded on September 29,2022 at 22:53:29 UTC from IEEE Xplore. Restrictions apply.
1962 SOMER: REDUNDANT PULSE AMPLIFIER 13

Table 3-Redundant Pulse Amplifier Output with Simulated Transistor Failures

NORMAL E0 = 150 MV ALL TRANSISTORS WITH 67. hFE 82


PREAMPLIFIER AT -85°
AMPLIFIER AT +650
D-C SUPPLY VOLTAGE IS VARIED +2V FROM
THE NOMINAL 17 VOLTS
TYPE OF
OF EMITTER-BASE-COLLECTOR C

FAILRE
FAILURE
OPEN BASE OPEN COLLECTOR OPEN EMITTER
_____________SHORT SHORT |

Eo (MV) WITH Vs Eo (MV) WITH Vs Eo (MV) WITH Vs Eo (MV) WITH Vs


TRANSISTOR AS A PARAMETER AS A PARAMETER AS A PARAMETER AS A PARAMETER
15 VOLTS 17 VOLTS 19 VOLTS 15 VOLTS 17 VOLTS 19 VOLTS 15 VOLTS 17 VOLTS 19 VOLTS 15 VOLTS 17 VOLTS 19 VOLTS

Q1 160 160 160 150 150 150 160 160 160 150 150 150

Q3 148 148 148 152 152 152 150 150 150 150 150 150

Q5 152 152 152 150 150 150 152 152 152 150 150 150

Q2 155 155 155 145 145 145 145 145 145 130 130 130
Q4 130 130 130 125 130 145 130 130 130 190 290 320

Q6 130 130 130 120 120 120 130 130 130 130 130 130

Q7 150 150 150 150 150 150 150 150 150 150 155 160
Q9 150 150 150 200 210 220 145 147 150 140 145 150
Q1l 150 150 150 150 150 150 150 150 150 150 150 150
Q8 150 150 150 150 150 150 150 150 150 150 155 160
Q1O 150 150 150 200 210 220 150 150 150 150 155 160
Q12 150 150 150 150 150 150 150
o
150 150 150 150 150 OD

(hFE > 80) and low noise transistors for the pre- to have deteriorated outside the transistor selec-
amplifier first stage, the voltage amplification tion criteria during these tests. Consequently,
will be constant within ±2 percent from -100OC up the preamplifier with the odd-numbered transis-
0
to -60 C and with simultaneous supply voltage tors exhibited somewhat inferior gain stability
variations. Under similar conditions the equiva- during simulated failures when compared to the
lent noise will be less than 2 x 10-14 coulombs at preamplifier with even-numbered transistors.
-100 C and at the nominal supply voltages. Disregarding the results of the preamplifier with
odd-numbered transistors, the redundant pulse
Transistor Failures amplifier output is stable within about ±3 percent
for 85 percent of the simulated catastrophic
In Table 3 the redundant pulse amplifier out- failures.
put is shown for the various indicated transistor
failures. During these tests the redundant pre-
amplifier was maintained at -85 C and the re- Reliability
dundant main amplifier at +65 C with the input
constant at AQin = 2.75 x 10-13 coulombs per Background
pulse. For simplicity, all failures were intro-
duced one at a time. However, various open and As an introduction to the art of reliability,
short combinations were examined, resulting in a definition should satisfy the cursory reader.
similar gain stability. One of the odd-numbered Reliability is the probability that a component,
transistors in one of the preamplifiers was found device or a system will perform its prescribed

Authorized licensed use limited to: The University of Utah. Downloaded on September 29,2022 at 22:53:29 UTC from IEEE Xplore. Restrictions apply.
Table 4-Pulse Amplifier Reliability Calculations

GENERIC FAILURE RATE MINUTEMAN FAILURE RATES COMMERCIAL FAILURE RATES


PROBABILITY OF FAILURE PARTS COUNT/CIRCUIT Z COMPONENT F R/HOUR Z COMPONENT FR/HOUR
UNIT/HOUR COMMON COMMON R COMMON
PARTS MINUTEMAN COMMERCIAL PREAMP. AMP. CIRCUIT PREAMP. AMP CIRCUIT PREAMP. AMP. CIRCUIT
2x10~~ ~ 10
~~~10 -100 -10 -10 -10
R ESISTORS 2 x
Y 10 | 150 x 10 11 14 3 |2 | 8 x1
ZxI Z810 6x106 1650X10410 210ox10 450x1O
CAPACITORS
SMALL VALUE 1 100 1 1 100
CAPACITORS -
LARGE VALUE 12 120 6 9 7Z 108 720 1080
TRANSISTORS-SWITCH 360 1800 3 3 1080 1080 5400 5400 H

CHOKES 10 100 2 2 1 20 20 10 200 200 100


z
DELAY LINE 10 100 1 10 100 coI
-10
-10 -10 -10 ~~~~ 0
-10
~~~~~-10
EFR CIRCUIT FAILURE/HOUR 1195x10 | 1236x10 26x10 |8070x10 18780x10 450x10
z
* -~ ~ ~~~ ~~~~~~~~~ -6-6
t FR CIRCUIT FAILURE/YEAR 1050x1x 1085x10 |2.8x10 7080X10 7700x ~~~~~~~~610-6-6 | 570x 10 -6 z
6 6
ft FR) REDUNDANT CIRCUIT FAILURE/YEAR 1.15x10 1. 18x10 50.Zx 1O-6 59.4 x10-6
|Q TOTAL OF CIRCUIT PROBABABILITY ZZ 8 X 10 6 570 10 6
OF FAILURE PREDICTION FOR ONE YEAR 2.18 x57x z
MISSION TIME (1 CHANNEL) 1.15 50.2 C)~

tTj
25.13x 10 679.6
Q PROBABILITY OF FAILURE (6 CHANNELS)
Q = 6 (Q) 150.78 x 10-6 4147.6 x 10

~~~4
t = 0.876 x 10 HOURS FOR MISSION TIME OF ONE YEAR
* ~-tFR
Q = 1 - E (PROBABILITY OF FAILURE IS EQUAL TO
ONE MINUS PROBABILITY OF SUCCESS)
WHERE t 0
FR<<
t F toQ
R Z
0

CD

CT,
CD
*1

Authorized licensed use limited to: The University of Utah. Downloaded on September 29,2022 at 22:53:29 UTC from IEEE Xplore. Restrictions apply.
1962 SOMER: REDUNDANT PULSE AMPLIFIER 15
function for a specified period of time under (3) Redundant Pulse Amplifier,
given conditions. Commercial Parts (complete
circuit, Figure 6):
Most work on reliability assumes an expon- P(year) = 0.99932.
ential distribution:
-F t (4) Redundant Pulse Amplifier,
P(t) = e r (2) MINUTEMAN Parts:
P = 0.999974.
Here: P(t) = reliability of device. This is the (year)-
probability that the device will Since the pulse amplifier constitutes only a
operate successfully for a speci- small fraction of the complete satellite electronics,
fied period of time t. the above indicated reliability, P(year)- 0.999974,
is satisfactory if the whole electronics is to meet
t = desired lifetime of device in hours, the 0.95 reliability goal.

e = base of natural logarithm. Acknowledgements


F = sum of the failure rates of all The author is grateful to Mr. Robert R. Hoge,
r
the components in the system whose recommendations were helpful in the
in percent per 1000 hours. development of the redundant pulse amplifier, and
to Mr. Allen E. McCardell, who performed all
For further details in reliability theory and the the reliability studies and calculations.
various approaches to the reliability calculations,
the reader is referred to references 5, 6, 7, 8, 9
and 10. References
Redundant Pulse Amplifier 1. E. Fairstein, "Considerations in the Design
Reliability Calculations of Pulse Amplifiers for Use with Solid State
Radiation Detectors," IRE Trans. on Nuclear
The redundant pulse amplifier reliability is Sciences, pp. 129-146; January, 1961.
calculated using the MINUTEMAN and commercial
failure rates. Since no attempt has been made to 2. R. F. Shea, "Principles of Transistor Cir-
correlate the component failure rate confidence cuits," John Wiley and Sons, Inc., New York,
level with the circuit (system) confidence level, N. Y., pp. 438-452; 1953.
the numerical results should not be taken as
absolute values,10 but rather as indicators of the 3. W. C. Elmore and M. Sands, "Electronics,"
relative improvement on reliability, i.e., redun. McGraw-Hill Book Co., Inc., New York, N. Y.,
dant circuit using MINUTEMAN parts as compared pp. 127-160; 1949.
to the redundant circuit using commercial parts.
4. F. S. Goulding, R. W. Nicholson and
The various reliability predictions can be J. B. Waugh, "A Double Delay Line Linear
calculated using the failure rates from Table 1 in Amplifier Employing Transistors," Nuclear
conjunction with Equation (2). These calculations Instrument and Methods, pp. 272-281;
are tabulated in Table 4. From Table 4 the reli- September, 1960.
ability predictions are summarized below for 5. "Life Test Analysis Reports," Texas Instru-
convenience: ments, Second Quarter 1959 through Second
Quarter 1961.
(1) Nonredundant Pulse Amplifier,
Commercial Parts (upper half 6. D. J. Davis, "An Analysis of Some Failure
of circuit as shown in Figure 6): Data," Journal of the American Statistical
P = 0.984. Association, vol. 47, No. 258; June, 1952.
(year )
7. Robert Lusser, "Predicting Reliability,"
(2) Nonredundant Pulse Amplifier, Research and Development Division, Ordnance
MINUTEMAN Parts: Missile Laboratories, Redstone Arsenal;
P(year) = 0.9978. October, 1957.

Authorized licensed use limited to: The University of Utah. Downloaded on September 29,2022 at 22:53:29 UTC from IEEE Xplore. Restrictions apply.
16 IRE TRANSACTIONS ON NUCLEAR SCIENCE November
8. Vernon L. Grose, "Reliability Planning in 10. R. 0. Frantik, "Some Research Activities
Space Systems, "Missile Design and Develop- Pertaining to Reliability," Sandia Corporation,
ment; September, 1960. Reprint, SCR-5; February, 1958.
9. R. Lusser, "Unreliability of Electrons--
Cause and Cure," Research and Development
Division, Ordnance Mis siLe Laboratories,
Redstone Arsenal; November, 1957.

Authorized licensed use limited to: The University of Utah. Downloaded on September 29,2022 at 22:53:29 UTC from IEEE Xplore. Restrictions apply.

You might also like