17 Dec
2003
17 Dec
'03
11:17 p.m.
You might find this article instructive: Herschell F. Murry, "A General Approach for Generating Natural Random Numbers," IEEE Transactions on Computers, December 1970, p. 1210. A fairly recent patent uses your approach of two oscillators: No. 4,855,690 by Dias, assigned to Dallas Semiconductor Corp., "Integrated Circuit Random Number Generator using sampled output of variable frequency oscillator." I'd suggest using Johnson noise; reverse-biased diodes generate noise which is pink. Ive built devices using amplified Johnson noise, squared up with a comparator, then averaged by a D flip-flop. The preliminary results look pretty good. Please post your results here -- and good luck. John A. Thomas jathomas@netcom.com