Highest Reliability SRAM

What I'm doing here at NASA is using an ACTEL rad hard RTAX series FPGA for my logic design, then utilizing it's built in SRAM to avoid using external SRAM. The RTAX is rad tolerant by design, and by using EDCC and memory scrubbing techniques, becomes rah-hard. I don't know off the top of my head if they have a part with 4M of SRAM, but you can check it out on their website.

Another technique is to use conventional SRAM, storing 3 copies of everything, or using a very good EDCC, and a rad-hard FPGA to scrub and correct the data. This operation runs continuously, in the background to other operations.

I used that second technique some 35 years ago in one of my MCA design where data integrity was required over years of performance. It worked well. Initially I used PROM for program and logid as they needed solid >1W power to blow the fuse in thim. Thy could not fail in program. SRAM memories I used to check integrity online on continuous bases. Data was in file structures. I used that Intel 8085 RadHard uC some time in 1978 or about to that time.

I agree this technique is good one but makes the system logic bit slow. My worry has been high energy alpha particles and multiple hit or burst of them. I am not worries about high energy Photons but that of charge particles. They can kick the memory out.

Static memory inside or outside will have same effect. Problem is also with edge triggred data registers that can get event triggered. I use lots of current in the logic to keep the level stable. I am not so sure of the CMOS logic that works with pC charge if that can save its soul from burning out.

Did you do some single event research on those FPGA? Some people put too huge a price on RadHard memories so not too sure if that is worth at all. I did some test of EMP on memories long ago and most of the EPROMs were blown and their connections were burnt. Only those could survive were also RadHard.

This is what they say:

• SEU-Hardened Registers Eliminate the Need for
Triple-Module Redundancy (TMR)
– Immune to Single-Event Upsets (SEU) to LETTH >
37 MeV-cm2/mg
– SEU Rate < 10-10 Errors/Bit-Day in Worst-Case
Geosynchronous Orbit
• Expected SRAM Upset Rate of <10-10 Errors/Bit-Day
with Use of Error Detection and Correction (EDAC) IP
(included) with Integrated SRAM Scrubber Featuring
– Single-Bit Correction, Double-Bit Detection
– Variable-Rate Background Refreshing
• Total Ionizing Dose Up to 300 krad (Si, Functional)
• Single-Event Latch-Up Immunity (SEL) to LETTH >

104 MeV-cm2/mg

I haven't had a chance to actually test these in our radiation lab. but will sometime in the next 6 months or so.

You might look at aeroflex, too:

http://ams.aeroflex.com/ProductPages/RH_4Msram.cfm

No-no that is not a good story.

High energy gamma will not harm the memory as it will hardly deposit any energy. It is the heavy charge particle that can go in will do lot worst. Try electron stripped Oxygen or Nitrogen atom or nucleus then you will see its worst effect. These people areport directional flux when they irradiate the sample. Change it 90 degree and then it will fail easily.

Perhaps if cell is distributed and remain disconnected from each other and can isolate the charge damage then it is very much possible to have a better RadHard scheme. This can be done in multilayer memories or distributed mutiple charge storage tank memories.

Do you have data on cosmic ray heavy charge particles that are reaching earth and are up to 300km from earth surface?

I believe our radiation facility tests with protons, but that's really all I know about it. I'm working on the CEV spacecraft, and our radiation environment is the worst, i.e., interplanetary space beyond low earth orbit, so we'll be testing for that environment, but I don't have any details on it at this point.

Most of the heavy charge particles can be stopped by external shield and body of the aircraft. That may be the reason that gamma and electron may be of great interest. Protons will come through neutron irradiation of hydrocarbons and water.

http://www.klabs.org/richcontent/fpga_content/DesignNotes/rtax-s_ax/index.htm