BIVBlog #26: An auditable random noise generator

If you rather trust a hardware random number generator you can build and audit yourself, rather than buying some trusted (according to the vendor) platform module, this series may be for you. In this episode I explain how the Zener/avalanche diode based noise generator I use works.


Table of Contents

00:00:00
Randy Bush’s lightning talk at RIPE-68 on the the Cryptech project
00:01:05
Why another hardware random number generator (HWRNG)?
00:01:40
My talk on the limits of cryptography at the 2014 Easterhegg in Stuttgart/Germany. (Don’t let yourselves get deterred by myself being introduced in German, the talk is in English.)
00:02:20
End user auditability as the ultimate goal
00:03:20
Auditable and general purpose components
00:04:15
Diversity as a goal
00:05:00
Full disclosure architecture and design
00:05:36
Why random number generators can’t be blackbox-tested
00:07:00
Tamper resistant ( and “tamper proof”) hardware random number generators vs. auditability
00:08:08 
Possible but not so practical random sources
00:10:43
The avalanche effect in semiconductor junctions
00:11:22
Zener/avalanche diodes
00:13:28
The hardware components it takes…
00:15:09
My current (pre-)prototype
00:18:25
How I got here: Failures and design decisions
00:30:42
Looking at the circuit with an oscilloscope
00:31:48
My current noise generater circuit
00:34:45
Comparing first and second stage output
00:36:05
A call for support

About

Long term IPv6 evangelist/book author/trainer/consultant and generic Unix guy (*BSD, Linux, Solaris, and about a dozen more).

3 Comments

  1. Kasper Pedersen

    To make the second transistor get out of saturation faster (and possibly making a BC547 or 237 suitable), you can borrow the trick that made the 74LSxx series much faster than the 74xx series.

    LS stands for low power schottky. What they did was add a schottky diode with anode on the base, cathode on the collector.
    When you then give the transistor base current, instead of the transistor going into saturation, then, when the collector drops to about 0.4V (the base is at 0.66, the schottky forward voltage higher), the schottky steals the base current, and the transistor does not go into saturation.
    It works exactly the same as without the schottky, only without the saturation recovery time.

    You want a small signal schottky, like a bas85/bat85 or smaller.

  2. Richard Bowden

    Can you make a tiny white noise generator with earplug and volume control?

  3. Benedikt Stockebrand

    Hi Richard,

    white noise for acoustic purposes has little to do with the crypto grade output of the generator here; I assume it is possible to do something like you want (I assume it is to deal with a tinnitus?) but the real problem with that sort of issue is with the size and power consumption.

    If you want an in-ear solution this is really tricky. If on the other hand you want something with a form factor similar to an MP3 player, then actually just recording white noise (you can get that from an old, analog TV set if nothing else) and putting that on an MP3 player should actually do the trick quite nicely.

    Cheers,

    Benedikt

Leave a Reply

Your email address will not be published.