KEN LaBEL
CODE 735.2
NASA/GSFC
Greenbelt, Md. 20771

                     TABLE OF CONTENTS

I.        INTRODUCTION

II.       TEST TECHNIQUES AND SETUP
          A.   Facility Usage
          B.   Test Hardware, Software, and Control
          C.   Device Test Procedures
          D.   Ion Beam Usage

III.      RESULTS AND DISCUSSIONS
          A.   SEEQ DQ28C256-250 EEPROM Results
          B.   Signetics 82HS641A PROM
          C.   Honeywell Transmitter and Receiver

IV.       SUMMARY

V.        ACKNOWLEDGEMENTS


I.   INTRODUCTION

The objective of this study was to determine the threshold linear
energy transfers (LETs) and cross-sections for single event upset
(SEU) and latchup due to heavy ions for several different devices
under consideration for use by the Small Explorer program.

The device types tested were:
     SEEQ Technologies DQ28C256 (32k x 8 EEPROM),
     Signetics 82HS641A  (8k x 8 Bipolar PROM),
     Honeywell HFE4811-013 (TTL Integrated Fiber Optic      
     Transmitter),
     Honeywell HFD3801-002 (TTL Integrated Fiber Optic Receiver).

Relevant characteristics of each device is summarized in the
following table:

Device Type    Mfg.      Date Code Technology 

DQ28C256       SEEQ      9044      CMOS/EPI
82HS641A       Signetics 9025A     Bipolar
HFE4811-013    Honeywell TBS       TBS
HFD3801-002    Honeywell TBS       TBS

Both the SEEQ and Signetics devices were delidded in order to
accommodate penetration limits of the test facility.

II.  TEST TECHNIQUES AND SETUP

A.   Facility Usage

The test facility used was the Brookhaven National Laboratories
(BNL) Single Event Upset Test Facility (SEUTF) between Nov. 9 and
10, 1990. This setup utilizes a Tandem Van De Graaff accelerator
suitable for providing varying ions and energies for SEU testing.
The test devices are mounted on a device-under-test (DUT) board
inside a vacuum chamber.

The SEUTF uses a computer-driven monitor and control program to
provide a user-friendly interface for running the experiments.
Additionally, support was provided by engineers working for Dr.
Stassinopoulis under the consortium that runs the SEUTF. Hard
copies of the test results were also made available.

B.   Test Hardware, Software, and Control

The hardware configuration is shown in Figure 1. The DUT board
consisted of six zero-insertion force (ZIF) sockets as well as four
transistor-type sockets. The ZIF sockets were populated with three
each of the EEPROMs and the PROMs, while the transistor sockets
were populated with two each of the optical transmitters and
receivers. This DUT board was mounted via six screws to the
mounting frame inside the chamber.

The DUT board was connected to the chamber adapter plate by six
three foot long cables with 40-pin Berg-style connectors on both
ends.

From the adapter plate, three foot cables were again used to
connect the chamber to the test equipment. One of these cables was
attached to the test circuitry for the integrated optical devices.
The other test equipment (described immediately below) only had a
single connector input to it, therefore cables had to be swapped in
order to test a different device.

To test the PROMs and EEPROMs, a Computer Service Technology Triple
Crown TC-700 multi-function instrument was used. This device
allowed for read operations of the PROM and EEPROM as well as
write/programming operations for the EEPROM. This tester interfaces
with a PC card and software for controlling these test operations.
Custom software was developed in order to take into account the
timing delay imposed by total of six feet of cable between device
and tester. Read operations for an entire device took less than 10
seconds, however a separate computer command had to be given in
order to store the file to disk. SEU detection (other than via a
checksum error) had to be done off-line after the testing. Write
operations to the EEPROMs took approximately 5-7 minutes in order
to program the entire device. The device then had to be read to
check for errors, then reprogrammed to determine if any permanent
errors occurred. Finally, the device had to be erased to ensure
integrity of the cells for the test.

To test the optical devices, Jim Cooley of Code 735 designed a
simple circuit that provided a 100 Khz square wave as input to the
fiber optic transmitter. This signal was then converted to an
optical signal, transmitted via optical fiber cable to the fiber
optic receiver. The receiver then converts this signal back to a
TTL square wave. The square wave is then compared (using as well a
small delay circuit to account for cable skews) to the original
square wave input. An upset was defined as the failure of
comparison of the two signals. The circuitry also counted the
number of errors that occurred.

C.   Device Test Procedures

1.   SEEQ DQ28C256-250 EEPROM

a.   Read Cycle:    The device was written with a known
checkerboard pattern while the ion beam was off. The ion beam was
then opened and the device read sequentially from low address to
high address, and the results written to a data file. The device
was then read a second time (without the ion beam) to determine if
the SEUs were in the memory cell or in the latches to the cells.

b.   Write Cycle:   The test device was written with a known
checkerboard pattern while being irradiated. The part was then read
without the ion beam on and stored to a data file. If a checksum
error occurred, the data file was interrogated and the device
rewritten (without the ion beam) and read again to determine if any
permanent damage occurred.

2.   Signetics 82HS641A PROM

The read cycle testing was essentially the same as for the EEPROM.

3.   Honeywell Integrated Optical Devices

These devices were irradiated in turn and the described test
circuitry was utilized to determine if any errors occurred.

D.   Ion Beam Usage

The following table summarizes the ions used for testing.

ION     ENERGY (MeV)  LET (MeV*cm2/mg) at 0 deg

F-19       140            3.38
Cl-35      208            11.5
Ni-58      268            26.5
I-127      320            59.7

Additional LET values were attained by changing the angle of
incidence of the beam to the device. This is seen clearly in the
data in Figure 2.

III. RESULTS AND DISCUSSIONS

A.   SEEQ DQ28C256-250 EEPROM Results

1.   During read operations, we were unable to cause the occurrence
of an SEU. I.e., no upsets occurred at LET values up to 72.9 (the
highest tested).

2.   During write operations, SEUs occurred at every LET value
tested starting at an LET of 3.38. Nearly every time an SEU
occurred, the byte affected became an FF hex as opposed to the
programmed AA or 55 hex. Since the device was erased prior to each
write operation, it has been hypothesized that these locations were
skipped in testing due SEUs in the address or control circuits
inside the devices or even in the 64 byte buffer that buffers
writes to the device. A few  of the errors that occurred were
single bit flips. The asymptotic cross-section based on the number
of bytes that had errors was determined to be greater than 10E-2
cm2 per device. Figure 3 illustrates the device cross section as a
function of LET values.

Additionally, permanent device errors occurred starting at an LET 
of 59.7. What occurred was the inability to rewrite the correct
bit pattern to the address where the error occurred. Once this
error occurred, we replaced the two devices we observed the damage
in and perform no further testing at higher LET values since they
rendered the device less than fully functional. The threshold for
this occurrence is between 53.2 (where no permanent errors
occurred) and 59.7 (where permanent errors did occur). Figure 4
illustrate the permanent device error cross section as a function
of LET.

3.   No latchup was observed on any of the above tests.

B.   Signetics 82HS641A PROM

1.   No SEUs were observed at LET values up to a value of 72.9.

2.   No latchup was observed at LET values up to a value of 72.9.

C.   Honeywell Transmitter and Receiver

We ran multiple tests with these devices at varying LETs and were
unable to make the device upset or latchup. However, the devices
were packaged and not delidded, so it is unlikely that the ions
were actually penetrating the device package. Therefore, the
results are suspect.

IV.  SUMMARY

In summary, the Signetics PROMs are recommended (given previous
total dose studies) for usage as are the SEEQ EEPROMs during read
operations. It is not recommended, pending further investigation,
to use the SEEQ EEPROMs for in-flight programming.

We were unable to properly test the Honeywell Integrated Optical
devices and plan to perform a retest with a modified/delidded
package to ensure accurate results.

V.   ACKNOWLEDGEMENTS

We would like to thank Dr. Stassinopoulis and his staff for all the
aid in preparing and performing the above tests. The test report
writer would also like to recognize the Code 735 members who took
part in this series of tests: James Cooley, Donald Hawkins, Mark
Flanegan, and Brian Smith as well as the support of Jackson and
Tull.