B030402_22V10 Heavy Ion Single Event Effects Test Results for Three Candidate 22V10 Reprogrammable Logic Devices Scott Kniffin1, Chris Palor1, and Mike Xapsos2 1. NASA/Orbital Sciences Corporation 2. NASA/GSFC TEST DATE: March 4-6, 2002 REPORT DATE: March 18, 2002 I. INTRODUCTION This study was undertaken to determine the radiation-induced single event transient (SET), single event functional interrupt (SEFI) and single event latchup (SEL) sensitivity of three programmable array logic devices. The device specifics are as follows: the Cypress PALCE22V10 is a flash erasable, reprogrammable CMOS PAL device, the Atmel ATF22V10B is a high performance EE programmable logic device (PLD) and the Lattice GAL22V10/883 is a high performance E2CMOS PLD Generic Array Logic (GAL) device. Heavy ion testing was done at Brookhaven National Laboratory’s Single Event Upset Test Facility. The output of the device was monitored for radiation-induced errors. II. DEVICES TESTED The 22V10 is a fully reprogrammable 132x44 AND-array, suitable for any number of logic functions. All three devices were configured/programmed identically. Outputs 1-4 count from 1 to 15. Output 5 is a Carry Out, which senses the end of the count to 15, drives the Output Low, then back to High. Output 6 is a Flip-Flop, toggling from 0 to 1 with each clock pulse. Outputs 1-6 are compared to a reference device under test (DUT) (an Atmel device with the identical program loaded) to look for SET. Output 7 is High, Output 8 is Low; both are compared with an exclusive NOR to look for SEL. The Cypress and Atmel devices typically drew about 70-80mA ICC and the Lattice drew about 20mA ICC. This was due to differences in chip architecture. The output of the DUT being irradiated was compared to a reference part (Atmel) to determine the error induced by radiation. There were a total of six DUTs, two from each manufacturer (DUTs C1, C1, A1, A2, L1, and L2). The DUTs were irradiated with several different ions. The devices were delidded before exposure. DUTs C1 and C2 were marked PALC22V10D-10DMB 5962-8984106LA Q712201 THA0134 on the top and 125928 Thailand on the bottom, DUTs A1 and A2 were marked C 9B0127 A 5962-8984106LA ATF22V10B-10GM/883 on top and 9B2398-19423K 1 Philippines-F 9B0127 on the bottom, and DUTs L1 and L2 were marked 5962-8984106LA GAL22V10D-10LD 883DDC D0B0043B on top and 0043 5F36AM1 Philippines on the bottom. III. TEST FACILITIES Heavy Ion Test Facility: Brookhaven National Laboratory Single Event Upset Test Facility. Flux: 8.52x103 – 1.14x105 particles/cm2/s. Particles: linear energy transfer (LET) B030402_22V10 Ion LET (MeV·cm2/mg) O-16 2.57 Si-28 7.88 Cl-35 11.44 Ti-48 18.6 Ni-58 26.69 Br-79 37.43 I-127 59.39 Table 1: List of ions used and their LET at normal incidence. IV. TEST METHODS Case Temperature: ambient in a vacuum Test Hardware: A custom test set was used to supply a nominal input level to the DUTs and monitor the DUT output for changes resulting from the radiation exposure. Each radiation event occurring on the output was captured in a custom error capture software package running on a laptop computer or with the HP5316B Universal Counter. The compare DUT determined if there was a difference in outputs given by the test DUT and the reference DUT, sending the error code to the software. Additionally, the High and Low of the test DUT were checked with an exclusive NOR gate. Upon an error condition, a “1” was sent to the universal counter. An HP6624A power supply provided power to all devices. Software: Customized LABVIEWÒ software provided a user interface to control signals to the DUTs. The software also automatically monitored the DUT output and generated an SET/SEL file history. Test Techniques: Tests were performed on the DUTs to measure the SET/SEL susceptibility as a function of particle LET for the specific application described in this report. Any deviations from the application described should be reviewed carefully. The test setup did not allow for monitoring or changing the temperature, therefore temperature effects were not considered. The power supply was set at +5V with nominal current. The DUTs were driven at 1MHz. The DUTs were programmed as given above and placed on to a test board as shown in Figure 1. The setup allows for continuous monitoring of all parameters by the software. The device was deemed to have experienced an SET when the output of the Test DUT did not match the output of the Reference DUT as compared by the Compare DUT (See Fig. 1.) or there was a momentary loss of one of the rails as detected by the counter as an occasional single count on the display. The device was deemed to have experienced a SEFI if the counter goes into runaway (the exclusive NOR gate comparing Outputs 7 and 8 give continuous errors) without the device drawing excessive current. The device was deemed to have experienced a SEL if the power supply current exceeded a specified amount, set in the software. The number of deviations during an irradiation was tabulated. The software captured all events. B030402_22V10 FIFO and FIFO downloader program: 1)if no difference between the REF & DUT then FIFO will read 000000 (w/ QA as MSB and T3 as LSB) +5V RESET U1 1 2 I1/CLK 23 3 I2 O1 22 4 I3 O2 21 QA 5 I4 O3 20 QB 6 I5 O4 19 QC I6 O5 7 18 QD RESET U6 U3 8 I7 O6 17 CO 1 I8 O7 R 9 16 T3 U1 11 6 9 10 I9 O8 15 HI 1 C1 5 D0 Q0 10 QA 11 I10 O9 14 LO 2 I1/CLK 23 X1 3 2 4 D1 Q1 11 QB 13 I11 O10 3 I2 O1 22 X2 4 1D 5 3 D2 Q2 12 QC I12 4 I3 O2 21 X3 7 6 27 D3 Q3 16 QD 5 I4 O3 20 X4 8 9 26 D4 Q4 17 CO I5 O4 D5 Q5 Q1 TEST DUT 22V10 6 19 X5 13 12 25 18 T3 7 I6 O5 18 X6 14 15 24 D6 Q6 19 8 I7 O6 17 XTOTAL 17 16 2 D7 Q7 13 9 I8 O7 16 18 19 D8 Q8 1 MHz CRYSTAL 10 I9 O8 15 7 21 I10 O9 XI EF 11 14 74LS273 8 13 I11 O10 23 FF 20 I12 FL/RT XO/HF RESET U1 1 15 1 1 RE 2 I1/CLK 23 COMPARE 22V10 XNOR BNC is actually 22 WE I2 O1 RST 3 22 2 outside of I3 O2 4 21 QA Chamber 5 I4 O3 20 QB to Freq Counter IDT7207L 6 I5 O4 19 QC 7 I6 O5 18 QD to pins 2,4,6,8,10,13 RESET 8 I7 O6 17 CO write to FIFO 9 I8 O7 16 T3 of COMPARE 22V10 7404 4 I9 O8 S 10 15 1 2 3 C1 5 11 I10 O9 14 1 2 6 Sends a Pulse to read FIFO 13 I11 O10 1 1D I12 Reset R to the 74ls244 JP1 +5V Connect to RESET 74LS74 Middle Enable 0 REFERENCE 22V10 MSB 21 1 EN 1 LSB 22 2 Reset Circuit 21 1 +V1 23 3 Switch is 22 2 24 4 23 3 25 5 outside the C1 C5 to EF of 24 4 CAP CAP NP 26 6 vacuum FIFO 25 5 Gnd 27 7 74LS244 chamber 26 6 28 8 27 7 +V2 29 9 QD 2 18 28 8 30 10 LOW CO 4 1A1 1Y1 16 MSB 29 9 C2 CAP NP 31 11 DATA T3 6 1A2 1Y2 14 30 10 32 12 8 1A3 1Y3 12 Middle 31 11 CAP 33 13 QA 11 1A4 1Y4 9 32 12 +V3 34 14 HI QB 13 2A1 2Y1 7 33 13 35 15 DATA QC 15 2A2 2Y2 5 LSB 34 14 C3 C7 36 16 17 2A3 2Y3 3 35 15 CAP NP 37 17 2A4 2Y4 36 16 CAP +V4 38 18 1 37 17 39 19 19 1G *you will need the FIFO Downloader 38 18 C4 C8 40 20 2G Program to get the data from the FIFO* 39 19 CAP NP 40 PIN to LAPTOP 40 20 CAP Figure 1: Schematic of 22V10 test circuit. V. RESULTS A. Cypress PALCE22V10 The DUTs showed SEFI and one non-destructive SEL at an LET of 7.9 MeV·cm2/mg (Si-28 at normal incidence). Typically, the SEFI occurred simultaneously with ~110-130 errors from Outputs 1-6 while the counter monitoring the High and Low rail of the device counted literally every clock cycle as one of the rails had failed. Each time the device recovered after a reset. Nominal current was 76mA; during one run, DUT C2 latched and the power supply current went to 110mA and recovered after a reset. B. Atmel ATF22V10B The DUTs showed no SEL up to an LET of 84.39 MeV·cm2/mg (I-127 at 45° from normal incidence). One DUT was also tested at an LET of 84.1 MeV·cm2/mg (Au-197 at normal incidence) with no SEL. There were a few SET mismatch errors, with the cross section tending to increase slightly with increasing LET (see Fig. 2). The mismatch threshold LET was 13.21 MeV·cm2/mg (Cl-35 at 30° from normal incidence). Additionally, the software can detect timing errors when they occur separately from other mismatch errors in Outputs 1-6.