Professional Documents
Culture Documents
ir
https://www.tarjomano.com/order
Steven L. Koontz Member IEEE, Paul A. Boeder, Courtney Pankop, Brandon Reddell
The shielding depths usually assumed for ISS design and IV. ON-ORBIT OBSERVATIONS: TOTAL IONIZING DOSE
verification are substantially smaller than the actual shielding ISS is performing well within expectations with respect to TID
depths. For a typical point internal to the ISS pressurized to electrical, electronic, and electromechanical (EEE) parts and
elements, detailed Sector Shielding [12], and CAD based [13- materials interior to structural shielding. With few short term
15] analysis of the as-built ISS structure reveals a cumulative exceptions, ISS has been flying at altitudes between 350 and
shielding depth distribution function with a median (50%) 400 km during the past 6 years, well below the 500 km
value of between 50 and 60 g/cm2 and ranging from 10 to 100 specified for the worst-case radiation design environment in
g/cm2 at the 10% and 90% points respectively. Similarly, the SSP 30512. TID accumulated to date is well below the
cumulative shielding depth distribution of ISS avionics performance degradation threshold (1 krad (Si)) for even the
enclosures external to the pressurized elements typically most sensitive ISS EEE parts and will remain so for the life of
ranges from 4 g/cm2 to 20 g/cm2 with a median value of 10 ISS. Ionizing radiation dose measurements, conducted within
g/cm2 [15]. It should be noted that no secondary particle the habitable volume using thermo-luminescent dosimeters and
design environment is specified in SSP-30512 Rev. C, though crew personal dosimeters, range from 5.1 to 9.5 µGy (H2O)
some analysis of secondary neutron and proton effects was (0.51 to 0.95 mrads (H2O)) per hour, depending on location in
completed for ISS optoelectronic devices labile to the habitable volume, corresponding to an annual dose range
displacement damage produced by structural secondary and of 45 to 83 mGy (H2O)/year (4.5 to 8.3 rads (H2O)/year or
Earth-albedo neutrons [16]. In the analysis presented below, about 2.1 to 3.8 rads (Si)/year) [17-19]. Somewhat higher
the shielding mass distribution of avionics boxes mounted dose rates were observed in the Service Module at 248 to 302
external to the pressurized elements are used with no µGy (H2O)/day or 9 to 11 rad (H2O)/year [17c]. For most of
additional contributions from the truss or pressurized element the subject dosimetry measurements cited here, ISS has been
structure which either have relatively low mass and/or subtend flying at altitudes below 400 km during the year 2001. The
only a small solid angle as viewed from the avionics box variation in TID with location in the habitable volume is
location. For avionics boxes mounted inside the pressurized expected as a result of variations in the average line-of-sight
elements, the avionics box mass distribution is simply added to shielding depth with location in ISS [12-15]. Estimates of
the pressurized element mass distribution with no TID internal to the ISS pressurized elements, using the worst-
consideration of local geometry. case design environment, ranged from 8 to 30 rads (Si)/year
before application of the 2X design margin [12], depending on
III. THE RADIATION DESIGN ENVIRONMENT: TOTAL IONIZING
DOSE
location in the pressurized elements. Direct measurements of
the ISS TID in rad (Si) using the DOSTEL instruments [17a]
The ISS TID radiation environment is specified for 500 km at with a location and viewing direction comparable to the 8 rad
solar maximum and includes trapped protons defined per (Si) design environment estimate, gave results between 5.1 and
AP8MAX and trapped electrons defined per AE8MAX. The 6.6 rad (Si) per calendar year.
contributions of solar and galactic cosmic rays and secondary
particles are not included. Shielding mass effects are included V. THE RADIATION DESIGN ENVIRONMENT: SINGLE EVENT
in the form of look-up tables for Al shielding mass in two EFFECTS
simple geometries, as determined with the well known Two different natural single event environments, for design
Shieldose model combined with the SSP-30512 natural and verification, are specified in SSP-30512 Rev. C [1]. The
environment [1]. Dose estimates at specific points in more nominal SEE environment is based on the AP-8MIN model for
complex ISS configurations can be made by combining the trapped protons and the CREME86, space weather index,
CAD or Sector [12-15] based shielding mass distributions with M=4, solar minimum model for galactic cosmic rays. The
the Al shielding look-up tables in SSP-30512. Other minor nominal SEE environment is specified for a 500 km altitude at
contributions to TID, such as galactic or solar cosmic rays and solar minimum so as to define a global worst-case
x-rays, uncertainties in the trapped radiation models, environment for both trapped protons and GCR [1]. The
contributions from Solar Energetic Particle (SEP) and other extreme SEE environment is also specified for 500 km. The
space weather events, and degradation rates of exposed labile extreme SEE environment for ISS design and verification is
materials, are addressed through the application of a 2x dose based on the October 1989 SEP event; a 99th percentile worst-
design margin to the 500 km design environment. The ISS case extreme SEE environment with respect to both energetic
radiation design environment represents a conservative, low proton and energetic heavy ion fluxes [20-23].
cost solution for ISS hardware design and verification. The
selection of 500 km as a design point altitude is, in itself, a Methods for testing and analysis of EEE parts in support of
worst-case assumption because ISS operates between 350 km ISS avionics performance verification in the nominal and
and 460 km. TID rates increase by a factor of about 3X extreme SEE environments are described in detail in SSP-
between 300 and 500 km, largely as a result of the altitude 30513 Rev. B (2). The Scott Effective Flux Approach (SEFA)
structure of the South Atlantic Anomaly (SAA). [24], using a charged particle incident angle cut-off of 80o
(normal incidence is 00) and an LET cut-off of 110
MeVcm2/mg, is used to calculate worst case estimates of on-
orbit upset rates for ISS design and verification purposes.
111
Downloaded from https://iranpaper.ir
https://www.tarjomano.com/order
Petersen [25] has shown that the ISS SEFA methods can The upset rate for a particular orbit, and an assumed 100 mils
overestimate or underestimate predicted on-orbit SEU rates (0.71g/cm2) of Al shielding, is then calculated using an orbit
compared to the Integral Rectangular Parallelepiped (IRRP) specific rate coefficient, Corb, that has the units upsets per bit-
method for some classes of device geometry and relative day [26] or,
radiation hardness. For the ISS 1Mx4 DRAM considered in
Rate (upsets/(bit day)) = FOM x Corb.
detail below, an overestimation of about one order of
magnitude is expected. In subsequent work, Peterson has The effect of shielding mass can be approximated using the
demonstrated that a generalized Figure of Merit (FOM) following equation [26],
approximation can produce useful SEU rate predictions (often
comparable to much more complex detailed IRPP with Corb/shld = 2Corb - 0.5Corblog(t),
shielding particle transport calculations) for a wide range of where t is a shielding thickness in mils.
EEE part types, shielding mass environments and orbits [26].
Barak, Reed and LaBel have analyzed this surprising result 1.0
Below, SEFA and FOM SEU rate calculations are compared 0.8
External MDM-10 Model
MDM-10 Model in USL Endcone
with observed on-orbit SEU rates for an ISS 1x4M DRAM.
FOM requires heavy ion and proton test data in combination 0.5
with the shielding mass distribution function. The DRAM die 0.4
for the specific 1Mx4 DRAM (Texas Instruments 44400, 0.9 0.3
micron design rules, 13.7mm x 6 mm die size) used in the ISS 0.2
the basis of results of TID and SEE testing conducted by the 0.0
0.1 1 10 100
European Space Agency (ESA) [28] on the essentially Equivalent Aluminum Thickness, T (g/cm2)
identical 4Mx1 device. The normal-incidence heavy ion test Figure 1: MDM 1Mx4 DRAM Structural Shielding Distributions.
data can be fit to the well known Weibull function, as shown
below, allowing calculation of SEU cross section, σ, as a S1-1 MDM-10
S0-1 MDM-10
function of ion effective linear energy transfer, L. P1-1 MDM-4
where σ (s) = 3.00 x 10-7cm2/bit is the saturated SEE cross- S1-2 MDM-4
S0-2 MDM-10 P1-2 MDM-10
section at large values of L, σ (L) = the SEU cross-section for
any value of L, L = the Linear Energy Transfer (LET) of the US Laboratory Module
112
Downloaded from https://iranpaper.ir
https://www.tarjomano.com/order
VII. ON-ORBIT ISS MDM SEE PERFORMANCE for occasional loss of telemetry signal (LOS) periods the actual
The performance of the ISS command and data handling SEU data collection time turns out to be closer to 135 days.
system MDM units is of special interest in light of the critical LOS is most likely between longitudes of +55 and +80 east as
nature of MDM function. During the first 4 years of flight, the is apparent in Figure 4. This LOS region corresponds to a
16 MDMs in the US Lab module have displayed only 3 region where TDRSS coverage cannot always be guaranteed
performance anomalies that may be attributable to SEE due to scheduling issues.
processes (see Figure 3). The three possible SEE performance The expected increase in SEU counts in the South Atlantic
anomalies showed no correlation with SEP events or flight Anomaly (SAA) and the high latitude or “horn” regions of the
through the South Atlantic Anomaly, though two of the three map are obvious. Table 1 shows the relative number of upsets
occurred at latitudes near +51 degrees. MDM performance in the SAA region, the high latitude horn regions, and the
anomalies are driven by functional lockup of the 80386SX remainder of the orbital ground track, i.e. the mid latitude,
microprocessor and the 82370 controller. An average US Lab non-SAA region. All 6 external MDMs are displaying nearly
MDM anomaly rate of 4.4 x 10-3 per day (6 per 4 years) was the same upset rate and very nearly the same distribution of
predicted using an assumed shielding mass of 14 g/cm2 and an upsets in the three major geographic regions, except for MDM
assumed critical bit fraction of 20 percent of the total bit count S0-1, which shows a lower SEU rate in the SAA region and
in the critical devices. MDM P1-1, which shows a significantly higher rate overall.
90 With the exceptions of MDM P1-1 and MDM S0-1, the overall
Solar Particle
Exposure Region
MDM Anomalies
APS Anomalies
ACBSP Anomaly
AUAI Anomaly
variation between MDMs is within that expected from Poisson
60 11/11/03
process counting statistics. The SAA region makes a relatively
4/1/02
small contribution to the total number of soft upsets, most of
which occur in the horn and mid latitude regions.
Latitude (degrees)
30
9/27/04 ISS
12/22/01
Latitude
Limits
+/-(51.6)
0
2/6/04
5/20/03
3/11/05 12/26/01*
-30
6/16/03
8/8/01
11/11/03
-60 South Atlantic
Anomaly
Solar Particle
Exposure Region
-90
-180 -150 -120 -90 -60 -30 0 30 60 90 120 150 180
* >100 MeV SEP event in declining phase at time
Longitude (degrees) of anomaly, GOES >100 MeV flux level ~1 pfu.
During the first 4 years of flight the observed anomaly rate for
the 14 US Lab MDMs is 2.1 x 10-3 per day (3 per 4 years), a
factor of 2 lower than the pre-flight verification estimate. No
MDM anomalies were observed during any of the 23 >
100MeV solar energetic particle events (extreme SEE
environments) that occurred between April 2001 and June
2005.
ISS MDMs are equipped with error detection and correction Figure 4: Aggregate ISS External MDM DRAM SEU Map;
(EDAC) capability. The EDAC detects and corrects single- [155 days calendar time with 135 days data acquisition time after
accounting for loss-of-signal (LOS) periods]
bit-per-data-word errors but only detects errors involving more
than one bit per data word. No true, single word, double bit Table 2 shows a comparison of the observed average (of all 6
errors have been detected to date, though 13% of the DRAM external MDMs listed in Table 1 and Figure 4) external MDM
SEU events observed to date involve 2 or more bits per 1 DRAM SEU rate for the 135 day data acquisition time with
second telemetry reporting cycle. DRAM refreshes are estimates made using the SEFA and FOM SEU rate methods.
initiated every 8 µsec and take 8.2 seconds to complete so that The median value of the shielding mass distribution function
at any time there are 125,000 refreshes in progress. The (10 g/cm2 Al) for external MDMs as shown in Figure 1 was
multiple upsets are most likely the result of a single charged used for both the SEFA and FOM estimates. The SEFA
particle event affecting physically adjacent DRAM cells that method overestimates the rate by a factor of 13, close to the
are not part of the same data word because the probability of one order of magnitude overestimate expected from inspection
two independent events in 1 sec is equal to the square of the of Figure 5 of Reference 25, given the DRAM heavy ion cross
single event rate and is therefore minute. section and sensitive volume depth. The FOM method is in
remarkably good agreement with the observed rate,
DRAM SEUs for 6 of the external MDMs are plotted as a overestimating by only a factor of 3.5 using the median
function of SEU latitude and longitude in Figure 4. SEU data shielding mass.
was collected for 155 calendar days; however, after accounting
113
Downloaded from https://iranpaper.ir
https://www.tarjomano.com/order
Table 1: ISS External MDM DRAM soft upsets corresponding to Figure at low shielding mass, and 2) overestimating the mitigating
4; total count and breakdown by geographic region.
effects of the higher shielding mass by neglect of secondary
Lat./Lon. SAA “Horns” Mid- Total particle production in structural shielding mass.
Region Latitude
Table 4: Effects of integrating over the shielding mass distribution
P1-1 73 129 97 299
MDM-4 MDM On-orbit SEU SEFA SEU FOM SEU
Count Count Count
P1-2 69 124 90 283 (SEU/238 days) (SEU/238 days) (SEU/238 days)
MDM-10
Lab-1 MDM
S0-1 41 121 96 258 (Figure 1; 488 1560 593
MDM-10 USL Endcone)
S0-2 72 105 84 261 S1-1 MDM
MDM-10 (Figure 1; Ext. 488 5721 1626
MDM-10)
S1-1 81 97 78 256
MDM-10 [ISS Internal/External MDM DRAM soft upset rates (corrected for
multiple bit errors) and comparison of on-orbit rates with predictions of
S1-2 69 108 74 251 SEFA and FOM models. MDM-10 and MDM-16 are nearly identical for
MDM-4 DRAM SEU shielding mass evaluation purposes]
[SAA region: Lat. 15 south to 45 south; Lon. 15 west to 70 west; North
Horn: Lat. 35 to51.6 north: Lon. 20 east through 0 to 160 west; South VIII. SUMMARY AND CONCLUSIONS
Horn: Lat. 35 to 51.6 south; Lon. 0 through 180 east to 135 west]
Detailed consideration of the effects of both the natural and
induced ionizing radiation environment during ISS design,
Table 2: ISS External MDM DRAM soft upsets; a comparison of on-orbit
rates with predictions of SEFA and FOM models development, and flight operations has produced a safe,
efficient manned space platform that is largely immune to
Average on-orbit SEU SEFA SEU Count FOM SEU Count
deleterious effects of the LEO ionizing radiation environment.
Count per MDM 10 g/cm2 Shielding 10 g/cm2 Shielding
10 g/cm2 median shielding (SEU/135 days) (SEU/135 days) The assumption of a small shielding mass for purposes of
(SEU/135 days) design and verification has been shown to be a valid worst-
268 3574 949
case approximation approach to design for reliability, though
predicted dependences of SEE effects on latitude, longitude,
SEP events, and spacecraft structural shielding mass are not
Table 3 shows a comparison of the observed on-orbit rates for observed. The FOM method over predicts the rates for median
internal and external MDMs along with SEFA and FOM shielding masses of 10g/cm2 and 40g/cm2 by a factor of 3 to 4,
estimates for the corresponding median shielding mass. while the SEFA method overestimated by about one order of
Obviously the SEFA and FOM methods are overestimating the magnitude as expected. The IRPP, SEFA, and FOM methods
mitigating effects of shielding mass when median shielding for estimating on-orbit SEU rates all utilize some version of
mass is used as the basis of estimate. the CREME 86/96 treatment of energetic particle interaction
Table 3: Effects of median shielding mass: with structural shielding, which has been shown to
underestimate the production of secondary particles in heavily
MDM: Median On-orbit SEU SEFA SEU FOM SEU shielded manned spacecraft [4-6 and 29-31]. The need for
Shielding Count Count Count
(Fig.1) SEU/238 days) (SEU/238 days) (SEU/238 days) more work directed to development of a practical
understanding of secondary particle production in massive
Lab-1: 40 g/cm2 488 966 468 structural shielding for SEE design and verification is
Lab-3: 40 g/cm2 490 966 468 indicated.
2
P1-2: 10g/cm 536 6309 1673 In contrast, total dose estimates using CAD based shielding
S1-1: 10g/cm2
488 6309 1673 mass distributions functions and the Shieldose Code provided
a reasonable accurate estimate of accumulated dose in Grays
[ISS Internal/External MDM DRAM soft upset rates (corrected for
multiple bit errors) and comparison of on-orbit rates with predictions of
internal to the ISS pressurized elements, albeit as a result of
SEFA and FOM models using median shielding values for estimate] using worst-on-worst case assumptions (500 km altitude x 2)
that compensate for ignoring both GCR and secondary particle
Using the shielding mass distribution functions displayed in production in massive structural shielding.
Figure 1 instead of the median value of the distribution
function does not produce an important improvement in the
accuracy of the SEU rate estimates, as is shown in Table 4.
114
Downloaded from https://iranpaper.ir
https://www.tarjomano.com/order
115
Downloaded from https://iranpaper.ir
https://www.tarjomano.com/order
[28] Harboe-Sorensen, R., Muller, R., Daly, E., Nickson, B., Schmitt, J.,
Rombeck, F. J.; “ Radiation pre-screening of 4 M bit dynamic random access
memories for space application,” in RADECS 91, Proceedings of the First
European conference on Radiation and its Effects on Devices and Systems,
09/09/91 to 09/12/91, La Grande-Motte, France, IEEE Catalog Number
91TH0400-2.
[29] O’Neill, P. M., Badhwar, G. D.; “Single Event Upsets for Space shuttle
flights of New General Purpose Computer Memeoy Devices,” IEEE
Transactions on Nuclear Science, 41, (5) October 1994.
[30] Shinn, J. L., Cucinotta, F. A., Wilson, J. W., Badhwar, G. D., O’Neill, P.
M., Badavi, F. F.; “Effects of Target Fragmentation on Evaluation of LET
Spectra from Space Radiation in Low-Earth Orbit (LEO) Environment: Impact
on SEU Predictions,” Proceedings of the 32nd Annual International Space
Radiation Effects Conference, Madison Wisconsin, July 17-21, 1995.
[31] O’Sullivan, D. O., Zhou, D., Semones, E., Heinrich, W., Flood, E.; “Dose
Equivalent, Absorbed Dose, and Charge Spectrum Investigation in low-Earth
Orbit,” Advances in Space Research, 34 (2004) pp. 1420-1423.
116