design decision questions

[Gerd v. Egidy]

Hi,

I've just seen a link to the OneRNG Kickstarter and since I'm interested in hw 
rngs for some time, I looked at your design and compared it with other ones I 
collected for some time.

I'm curious why you chose the dual transistor solution which gives you 
non-50:50 results. Why didn't you use a zener diode with amplification instead? 
If you do it right, the diode is automatically driven to the correct bias 
level to give the max. noise.  Here is a schematic showing what I mean:
http://jtxp.org/tech/XR232circuit_v21_LoRes.png
(just as a reference: this is linked from http://jtxp.org/tech/xr232web.htm, 
you'll find a detailed description there, but in German).

Have you considered using other random sources than avalanche and radio noise? 
Or adding them as additional sources onto OneRNG?

Here are projects using other methods I consider interesting:

http://jtxp.org/tech/xr232usb.htm
Unfortunately in German only, but Google translate should do the job. This 
relies on small imbalances within two elements of a LM393 comparator.

https://github.com/waywardgeek/infnoise
This one uses modular multiplication / FireBug.

Both methods can be implemented with few simple and cheap components, 
microcontroller and usb connection usually being the most complicated parts.

Kind regards,

Gerd