[Cryptech Core] Alpha documentation

[Fredrik Thulin]

On lördag 28 maj 2016 kl. 13:58:42 CEST Peter Stuge wrote:
> Fredrik Thulin wrote:
> > > > What I do know is that we can't spend time on in-depth validation
> > > > before Berlin =).
> > > 
> > > I think it would be about a day or two worth of effort.
> > 
> > To do that particular test, or all in-depth tests?
> 
> All - some is already done with the SDRAM test, but stressing the
> power supply is also pretty important.
> 
> One important test is to exercise FPGA internals while monitoring
> voltages, another is to exercise IOs both on FPGA and on STM32.

I did some power tests at least. We have our old FMC testbench program that 
writes and reads 32 bits of data 100,000 + 100,000 times from the ARM to a 
test core in the FPGA and validates it gets correct answers back.

In my test I made the delay between test runs 1 ms (something observable on 
the scope I figured) and a 1 s pause every 10 runs to see if I could see when 
the test was running or not.

The 1.8V rail as measured on FB5 has jitter of around 100 mV peak-to-peak.

The 1.0V rail as measured on FB7 has jitter of around 100 mV peak-to-peak.

The 3.3V rail as measured on FB6 has jitter of around 200 mV peak-to-peak.

I couldn't spot the 1 ms pause between test runs on any of the voltage rails. 

On the 3.3V rail I could see a slight difference in the jitter shape during 
the 1 s pause between every 10 tests. Hard to capture, but somewhat visible in 
the "working" and "sleeping" images.

The 1s pause is not visible to me at least on the long-term sampling (10 tests 
take about 6 seconds, so there should be a change for half a horizontal square 
every three squares, with 2s time scale). 

The short black line to the right of the 2s image is the "scan line". I think 
I found a bug in my scope where freezing the display doesn't really work at 2s 
resolution, so I had to let it run while saving the image.

See the attached scope pictures and please share your thoughts of the test 
results and also feel free to ask me for any (quick =) ) additional 
measurements.

/Fredrik
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Alpha-rev02-1V0-FB7.png
Type: image/png
Size: 43368 bytes
Desc: not available
URL: <https://lists.cryptech.is/archives/core/attachments/20160530/9aaf0fe9/attachment-0006.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Alpha-rev02-1V8-FB5.png
Type: image/png
Size: 40531 bytes
Desc: not available
URL: <https://lists.cryptech.is/archives/core/attachments/20160530/9aaf0fe9/attachment-0007.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Alpha-rev02-3V3-FB6.png
Type: image/png
Size: 34960 bytes
Desc: not available
URL: <https://lists.cryptech.is/archives/core/attachments/20160530/9aaf0fe9/attachment-0008.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 3V3-jitter-sleeping.png
Type: image/png
Size: 70783 bytes
Desc: not available
URL: <https://lists.cryptech.is/archives/core/attachments/20160530/9aaf0fe9/attachment-0009.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 3V3-jitter-working.png
Type: image/png
Size: 69329 bytes
Desc: not available
URL: <https://lists.cryptech.is/archives/core/attachments/20160530/9aaf0fe9/attachment-0010.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 3V3-jitter-2s.png
Type: image/png
Size: 71601 bytes
Desc: not available
URL: <https://lists.cryptech.is/archives/core/attachments/20160530/9aaf0fe9/attachment-0011.png>