[sdiy] analog loses another niche

[Mike Pepper]

>
http://spectrum.ieee.org/computing/hardware/intel-makes-a-digital-coin-tosser-for-future-processors
>
> ”Historically, RNGs have been analog,” says Greg Taylor, director of
> the Circuit Research Lab. ”But porting to smaller technology nodes
> [with analog devices] requires a lot of fine-tuning that is
> unnecessary with digital versions.”
>
> Be interesting to see how they rigged the metastability to be a 50/50
shot.
>
>
> Tom "And I, for one, welcome our new randomly digital overlords" Corbitt

FYI, related circuit for a random bit generator and references:
http://web.jfet.org/hw-rng.html


A lot of professional RNG's generate the bit stream by feeding a comparator
with the ouitputs of two noise source. It goes high when one is greater than
the other, and low in the reverse condition. This helps even out bias in any
one circuit. The first version of this I know of was incorpprated in the
UK's selector for the state 'premium bond' system, (a kind of lottery bond,
where you lend the government money, but the dividends are allocated as
prizes, randomly selected. You keep your investment.) The device was called
ERNIE (Electronic Random Number Indicating Equipment). The original used
random noise from neon tubes. It was designed in part by Tommy Flowers, now
known to be the man behind the design of the WW2 code breaking machine
"Colussus".

http://objectwiki.sciencemuseum.org.uk/wiki/ERNIE_1

The latest version uses a PC with an Intel RNG built into the chipset.
Interestingly, this uses the same basic approach, though the noise is now
thermal, and is used to 'skew' an oscillator, (IIRC!)

http://www.lightstraw.co.uk/gpo/posb/ernie/ernie4.html

http://www.intel.com/assets/pdf/manual/298029.pdf

||\/||ike