[s-t] needle in haystack digest #3 (fwd from Nick.Barnes at pobox.com)

[Tim May]

On Thursday, November 6, 2003, at 09:56  PM, Riad S. Wahby wrote:

> "Major Variola (ret)" <mv at cdc.gov> wrote:
>> At 08:22 PM 11/6/03 -0800, Tim May wrote:
>>> I heard ten years ago that the National Semi fab on-site was a lowly
>>> 2-micron fab. Which was enough for keying material.
>>
>> And rad-hard circuits for their buddies at the NRO.
>
> Probably not on a CMOS process, though.  For the most part,
> rad-hard==bipolar, even nowadays.
>
>

Most ULSI today is BiCMOS, but Intel, Harris, and a bunch of others 
were making rad-hard CMOS nearly 20 years ago. The 80C86 rad hard part 
was and is used in a lot of  critical apps.

True enough, a project I consulted on picked the AMD 2901 for the 
Galileo Jupiter mission, and it was bipolar.

And of course the concern with shrinking geometries has moved from 
"suntan" effects (long exposure) to SEUs. And here the advantages 
mostly are with SOI (as they were with SOS and SOI when I started 
working on SEUs in 1977).

--Tim May