Noise sources: multi-oscillator vs. semiconductor noise?

[Thor Lancelot Simon]

I am working on a semi-experimental hardware RNG project.  I note that
the publically available designs, e.g. http://willware.net:8080/hw-rng.html
or http://world.std.com/~reinhold/waynesrngcomp.gif or the others
listed at http://www.std.com/~reinhold/truenoise.html all use semiconductor
junction noise as the source -- diode or transistor avalanche noise.

But all the modern commercial designs with which I'm familiar (which
include the Intel, the Hifn, one Motorola design, and some others that
are not publically documented) are all multiple-oscillator designs, in
which some number (usually 2 or 3) of undisciplined oscillators of close
design frequency drift against one another, and an ADC is used to sample
the resulting output waveform.

I cannot find any public, rigorous discussion of why such a design might
be preferable to the semiconductor noise type of design -- but I have to
assume the people designing the commercial sources have all converged on
similar designs for _some_ reason.

Can someone point me to a discussion of the advantages or disadvantages
of either design type in the literature?  I am not interested in the
theoretical advantages of other, costlier sources such as radioactive-decay
or "more direct" (than junction noise) quantum or thermal noise sources;
I just want to understand why all the public domain designs are of one
type, and all the commercial designs of the other.


-- 
  Thor Lancelot Simon	                                     tls at rek.tjls.com

  "We cannot usually in social life pursue a single value or a single moral
   aim, untroubled by the need to compromise with others."      - H.L.A. Hart

---------------------------------------------------------------------
The Cryptography Mailing List
Unsubscribe by sending "unsubscribe cryptography" to majordomo at metzdowd.com