Hardware Random Generators: Transistors vs Diodes

[EinsteinKreuz]

So which method works better for random number generation(in terms of flatter frequency spectrum): transistor breakdown or Zener noise?

I want to know whether to use a Zener diode or a Transistor when building a white noise generator. And FTR I intend to use Germanium because of its higher electron mobility and lower barrier potential.

But what about covered photodiode dark saturation current? Has that even been tested as a white noise source(Germanium photodiodes are much noisier than Silicon photodiodes).

 

[meBigGuy]

This article addresses some of the issues you will face.

http://www.ciphersbyritter.com/NOISE/NOISRC.HTM

 

[EinsteinKreuz]

This article addresses some of the issues you will face.

http://www.ciphersbyritter.com/NOISE/NOISRC.HTM

The LM336 Zener diode this fellow used is of course made of SILICON rather than Germanium. I'd be interested to see the performance spectra for a Germanium diode. But I have yet to see any date for Germanium photodiode dark saturation current. I think I'll try the Germanium transistor breakdown noise approach.
But this page is quite dated! Much progress has been made in HRNGs(mainly in terms of electro-optical devices).

 

[meBigGuy]

Not my field. Maybe someone with current knowledge will respond.

 

[Svein]

Also check out linear feedback shift registers (https://en.wikipedia.org/wiki/Linear_feedback_shift_register). They generate the maximum possible random sequence for a given number of bits.