RNG: from Zener diode?

[Liam David Gray <lg2g+@andrew.cmu.edu>]

Tim,

I recall you or some other Cypherpunk mentioning that, by some means,
strong random numbers can be had by observing some property of Zener
diodes.

Could you explain this?  ...If feasible, I imagine exploiting this
phenomenon to make a "random number dongle," an inline device for a PC's
serial or parallel port, which could be polled by savvy software.  If
possible, and if someone hasn't done it already, I'd like to work up a
simple recipe (schematic, list of ingredients, and sample code) for
this--and get it put into our "crypto-anarchists' cookbook"--i.e., on
soda.berkeley.edu.  Perhaps a future version of PGP could take advantage
of such a random noise dongle, selectable by command-line switch.

This is pure speculation at this point.  Am I completely off-base?  In
any case, I'd love to hear more about it.

Liam
---
PGP public key available by arrangement   -- The cat is out of the bag.
Too much of a dreamer not to be practical -- Go have your own "valiant defeat."
Liam David Gray <lg2g+@andrew.cmu.edu>