Quantum Random Number Generator module breakout

[diracshore]

Hi all,

I have been looking into building a hardware RNG for some time now. One project I thought was promising uses decay from a radioisotope, americium-241 from a disassembled fire alarm paired with the CCD from a webcam to visualize Cherenkov radiation as a chaotic source.

http://www.inventgeek.com/Projects/alpharad/overview.aspx

The open source software to facilitate this,

http://www.lavarand.org

This solution is cheap but it cannot operate for very long before the decay blinds the CCD. And putting a source of alpha particles inside a computer as he suggests is rather unwise. http://www.siliconfareast.com/soft-error.htm

Here is a solution from a different open source project I have not yet attempted,

http://warmcat.com/_wp/whirlygig-rng/

But it was this solution by a Swiss firm called id Quantique that really caught my attention. It can output 4Mbits/sec from a small module which houses a semi transparent mirror acting as a beam splitter and two single photon detectors to extract entropy using Born's Postulate.

http://arxiv.org/pdf/1004.1521v2

I have acquired an engineering sample of one of these modules and now need to create a breakout board. The application note below contains a simple schematic for extracting the random bitstream using a 74HC164.

Now are there better ways of extracting the stream ? Will I get the full 4Mbits/sec from that circuit ?

Datasheet
http://www.idquantique.com/images/stories/PDF/quantis-random-generator/quantis-oem-specs.pdf

Application note
http://www.idquantique.com/images/stories/PDF/quantis-random-generator/quantis-appnote.pdf

Whitepaper
http://www.idquantique.com/images/stories/PDF/quantis-random-generator/quantis-whitepaper.pdf

[Balaur]

... And putting a source of alpha particles inside a computer as he suggests is rather unwise. http://www.siliconfareast.com/soft-error.htm

Actually no, there is no danger that a alpha particle emitted by stand-alone source to affect packaged ICs. In this context, alpha particles from an Americium source have penetration depths of tens of microns, and will not reach the sensitive areas of the circuit.

Alpha particles cause Soft Errors (or more appropriately-called Single Event Effects) when radioactive contaminants from the packaging materials, bonding, balls, etc ejects alphas very close to the sensitive areas (usually transistor drains).

So, what's wrong with the recommended application note? It looks like a very elegant and simple solution (just one chip). And yes, the 4Mbit/s looks obtainable (or a new random byte each 2 microseconds).

However, I'm quite interested on why and how you will use so many random numbers. Care to share?

Cheers,
Dan

[diracshore]

Ok, thats good to know about the contamination. But I still think decay based RNGs are a somewhat crude solution.

What was bugging me with the example schematic was just that they described it as simple. Which it is and thats good but I was unsure if it was just proof of concept as opposed to the optimal solution. They sell a USB version of this module at a significant markup and I think perhaps it would be unwise for them to make it too easy to replicate breakout function. There is a 500 euro difference from what I remember between the OEM module and a OEM module in an enclosure with a USB output. I was assuming there was a reason, some necessary engineering or additional components that would explain the price difference.

The entropy is necessary for some work I am doing with genetic algorithms and stochastic programming. Its just a recurring dependency I am trying to solve.

Thanks !