[Cryptech Tech] fyi: secure TRNG thread on crypto list

[Benedikt Stockebrand]

Hi folks,

time to catch up on the list again...

=JeffH <Jeff.Hodges at KingsMountain.com> writes:

> just an fyi/fwiw. I'm not a hardware guy or crytographer so I can't
> really eval whether this thread is worthwhile for you folks to peruse
> or not, nor whether this Bill Cox guy might be someone to reach out
> to, YMMV, etc..
>
> [Cryptography] The world's most secure TRNG
> http://www.metzdowd.com/pipermail/cryptography/2014-September/023051.html

I've taken a bit of a look at the message and just for the record: 

The analog section of the arrgh design is meant to deal with noise
injection through the power input.  Basically, the voltage used for the
avalanche breakdown is using a buffering capacitor which is kept at the
breakdown voltage; the only influence the supply voltage has here is
that it determines how fast the capacitor reaches that voltage during
power-up.

The amplifier stage is only dependent on its input voltage with regard
to its absolute voltage output; its output is however rail-to-rail
(well, sort of...), so again, changes in the supply voltage don't
actually affect its output at the logic level.

And of course, as an additional precaution (largely because I don't
trust USB, but we've already had that) I've used LC filters on the power
supply inputs to deal with external noise.

As far as crosstalk between traces is concerned: With the voltage
levels, currents and frequencies used this shouldn't be a problem.
However, I will consider putting some shielding ground traces between in
a future design.


Cheers,

    Benedikt

-- 
Benedikt Stockebrand,                   Stepladder IT Training+Consulting
Dipl.-Inform.                           http://www.stepladder-it.com/

          Business Grade IPv6 --- Consulting, Training, Projects

BIVBlog---Benedikt's IT Video Blog: http://www.stepladder-it.com/bivblog/