Newsgroups: sci.crypt
From: (Wray Kephart)

Subject: RNG's
Organization: CTS Network Services (CTSNET/crash), San Diego, CA
Date: Wed, 2 Feb 1994 12:47:44 GMT
Summary: random number sources
Keywords: rng, prng
Lines: 38

I've been browsing this newsgroup for a month or two and have seen posts
for PRNG's, hardware RNG's, and recently a Nuclear RNG discussion.

Random phenomena in electronics are fairly common, e.g., diode noise or
oscillator phase jitter to name two.
The difficulty is not in locating a suitable source of random electrical
events, rather it is the measurement of the events without perturbing the
'system' or introducing skew into the data as a result of making the

One example that comes to mind was a project to interface a digital random
sequence generator chip to a PC. I buffered the chip and connected it to
the Rx line of the RS-232 port. When a random 1 or 0 was required, you
read the serial port. WRONG! The chip output was random however there
existed a _slight_ difference between the rise and fall times of the buffer.
This gave the data a slight skew towards 1's (slower rise, faster fall). 
The message was: asynchronous sampling of a random event is not necessarily
valid depending on your measurement system/interfaces. Be careful.

I've been working with development, debug, and validation of random
sequence hardware for several years. My advice to anyone seriously requiring
a source of truly random numbers is to buy the commercial version of a
'randomizer' chip. Last I looked, they are manufactured by both Rockwell
and Hughes. The price (in quantity) was around $150.00 for Class B devices.
I believe the specs for randomness are not as stringent as for the
military versions.

---  The opinions expressed above are not my own  ---
---     If you want to know what I think          ---
---            it will cost you                   ---
---> Technical and data = <---
---> General & Business =   <---
--->           Personal =  <---