09-12-2018 06:41 AM
Hi all, anyone has any experience to share with regards to designing for extreme operating temperatures (‑55°C to 125°C). ? Are there specific differences in the style? Is timing performance harder to meet or it is the same? I read online that I should watch out for jitter rates, like system and clock jitter, besides the usual. Thanks.
09-12-2018 07:56 AM
running at 125 ambient,
is interesting how do you get the heat out to keep the silicon below 85 ?
running at -55 ambient
things like crystals and VCXO's don't always start up.
but after a few minutes of self heating from the other components they tend to start up.
and ensure you don't get condensation is interesting,
09-12-2018 08:32 AM
Even the industrial and automotive grade parts are only spec'ed to work at a junction temperature of 125C. Assuming that you have an ambient temperature of 125C and are using any power at all, your junction temperature will be higher than 125C. If you are a few degrees higher, the part will probably continue to work, but with a reduced lifetime and with worse timing.
Activation energy for most physical processes is an exponential function with temperature. At some point, not too much higher than 125, you are likely to hit the knee of the curve. If that happens, you may be measuring the lifetime of your part in minutes.
09-12-2018 10:04 AM
It should be made very clear...
Xilinx devices operate within the specified limits. Period. These are specified as maximum and minimum operating temperatures - the values depend on the commercial grade of your device (commercial, industrial, extended, space grade)
The second you go outside these limits, all bets are off. The tools have no mechanism for determining whether the design will operate outside the specified limits - it is simply impossible to do timing analysis outside the limits.
So, you have
- no guarantee of timing
- no guarantee of lifetime
- no guarantee of startup
- no guarantee pretty much of anything
If you choose to go this route, you will, by definition, have an unreliable design - it can fail in pretty much any way at pretty much any time...
09-17-2018 09:16 AM
You may find this article from the Xcell journal (Issue 94) useful. It talks about a design using Spartan 6 FPGAs in a high-temperature camera with temperature requirements exceeding 125°C. Unfortunately, the article does not really discuss timing issues.
Link to article: How to Extend the Operating Temperature of FPGAs
09-18-2018 02:13 AM
did not know thta source
there is also the pdf available