[Cryptech Tech] Noise board on Novena

[Bernd Paysan]

Am Dienstag, 23. September 2014, 19:22:06 schrieb Benedikt Stockebrand:
> Hi again,
> 
> Bernd Paysan <bernd at net2o.de> writes:
> > For things that connect to a PC, however, I prefer UART, because USB to
> > serial is very cheap and easy to get (and supported out of the box by the
> > OS),
> ...if we're talking FTDI chips, yes.

The PL2303 has also good out-of-the-box support.

> > whereas USB to SPI requires a specific FTDI device and a special library.
> 
> FTDI has something like that?

Yes, the FTDI 2232D can be used for SPI, JTAG, and even I²C (though probably 
not multi-master I²C, but I've never tried I²C with the FTDI).

>  I've only found the Microchip MCP 2210
> for that job.  But yes, one of the reasons why so far I've stayed away
> from them is that they need a suitable driver (kernel or userspace,
> whatever).  At which point I decided I'd rather consider the USB capable
> PICs instead and do away with that extra chip altogether---maybe next
> year...after I've made myself comfortable with FPGAs, of course...

The disadvantage of USB is that it is way more complicated... if you implement 
USB on an FPGA, I'd suggest you use a microcontroller core+USB combination.

> > When I use UART protocols, I make sure that the first byte you will send
> > over the UART is odd, so the device can do automatic baud rate adjustment
> > (like "AT" modem commands - 'A' is odd: 65).  UART has the advantage of
> > needing just two lines (RX and TX; when you talk to hardware, you don't
> > need flow control, the hardware will always be fast enough to respond),
> > and there's no mode inflation (when you use 8N1).
> 
> Well, there's the issue with the speed; if you use automatic baud rate
> detection then you have to make sure you discard that first bit.

Actually, the first bit you use to measure the baud rate is the start bit.  
The next bit has to be a 1 to get an edge at the end of the start bit, so you 
can do that measurement.  The bit itself will be sampled correctly (you sample 
bits in the middle of the cycle).

> And as far as flow control is concerned: I actually kind of use that
> with the arrgh board.  Eventually it might be useful to shut down the
> step-up converter and save some power.  But of course that's an optional
> thing, and I'm not even sure I want that extra complexity anyway.

If you use the "entropy only for initial seed" approach, shutting down the 
diode current may be a good idea.

-- 
Bernd Paysan
"If you want it done right, you have to do it yourself"
http://bernd-paysan.de/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part.
URL: <https://lists.cryptech.is/archives/tech/attachments/20140923/a3af3257/attachment.sig>