Random Bit Generator

I am not sure how many random bits could be extracted from one click of a relay. Once I tested reed relay for life expectancy; the bounce pattern looked very repeatable. BTW, the life term appeared to be not so great: somewhat 500K cycles or so.

Agreed. Typical RMS jitter ~ 1% of the period, and a big part of it is due to the unpredictable noise of the components.

One should be very careful with the assesment of the amount of the true random information. It is easy to make a RNG which could seem very solid from first glance, but it would fail to the basic analysis.

Vladimir Vassilevsky DSP and Mixed Signal Design Consultant

ly

So their web-site is broken. How come you didn't manage to work this out?

The sloppy work was yours - you picked up a link to an obviously corrupted data sheet, and present it to the user-group with a snide comment, rather than finding a non-corrupted data sheet.

Probably for the same reason that Philips stopped making the SD214 - too much like hard work for a relatively small market.

Fairchild would appear to have put an idiot in charge of the relevant bit of their web-site. These things happen. At least it isn't a nuclear reactor.

-- Bill Sloman, Nijmegen

If you timed the bounces to ns resolution, they would be pretty noisy. Temperature, vibration, coil voltage variations, all sorts of things would vary the timing and bounce details. Of course, using a relay for making random numbers is a whimsy, not a very practical idea.

I agree about reeds. Even at low switching levels, they aren't very reliable.

John

Does not matter..it cannot generate any (and all) given pattern more than once in a row..so it does not pass specs.

Not quite..the _probability_ of any repetition is related to the bitlength under observtion, factored into all of the other possible groups. Repeat N times, and the probability, starting near zero, gets so small that the engineering zero is most likely larger..

Mercury-wetted reed relays do last longer - about 100 million operations - and the contact resistance is pretty stable. They do need to mounted close to vertical - people did keep offering oerientation insensitive mercury reeds, but they never seemed to stay on the market for any length of time.

-- Bill Sloman, Nijmegen

Does anybody know that that means?

John

true

LSFR

imply.

What's wrong with a zener diode? Or an led and a photodiode? Or just a resistor?

John

No such luck. Personally, I'd suspect that you'd have relatives working for Fairchild, but there must be other genetically impoverished groups in the US (like the Jukes and the Kallikaks).

-- Bill Sloman, Nijmegen

Yes, and they had progeny.

Fortunately, that won't be so in your case.

So Slowman is a Kallikaks-disposition family? ...Jim Thompson

...

That might be an interesting problem, but it's not what I was trying to say.

The normal stable state for a LFSR is all 0s. You can change that to all 1s if you put an inverter after the XOR or on any of the inputs to the XOR. (XNOR vs XOR)

Having the stable state be all 1s might be handy. For example, if your FFs have a reset but not a set, or if you want to trap the stuck-at state and you have a wide AND gate but not a wide OR gate.

...

Too simple? Do you want audio-band white or pink noise as noise, or do you want to turn it into a digital bitstream? How would you go from a zener to a digital bitstream? I have a couple of ideas, but how do the big boys do it?

Thanks, Rich

Comparator.

John

Well, Zener diodes have a large-ish junction area, and breakdown isn't completely uniform, so you can get semistable 'hot spot' formation. Look at a Zener 'way down in the low current range on a curve tracer, sometime. The trace goes fuzzy, BUT the fuzz has sharp boundaries. Those boundaries are nonrandom.

An LED and photodiode are excitable by cosmic rays, stray light, maybe even acoustic input; the output amplification gets contamination from its power supply ripple; the LED has poor ageing stability (well, SOME do).

A resistor: well, the metal film types are lower in noise than the carbon, but carbon is piezoresistive, so it picks up acoustics. The output amplification is a problem here, too.

So, my preference is for a multi-stage AGC type amplifier, with nil for its input, with the input impedance determining a Johnson noise background. Bandwidth is several MHz, the high-gain stages are all at the IF where good bypassing of the power ripple is easy. Even if some weather phenomenon does excite the antenna, it's still likely that the least-significant- bit end of the as-digitized signal is pure thermal noise. UHF TV digitizer, every pile of old computers has one...

So why are zeners used as high-quality RF noise diodes?

Zeners can make relaxation oscillations at very low currents. So don't do that.

All silly. PD shot noise is an excellent source of noise. And if high currents, or light leakage, degrade a circuit, don't do that.

Cosmic rays?

All resistors of a given value generate exactly the same Johnson noise. Anything else would violate conservation of energy.

Does anybody here use carbon resistors as microphones?

Unless you're amplifying the noise of the front-end semiconductors. That noise can have bad ststistics.

John

Standard old undergraduate experiment, has a plastic block with a photodetector (it's a scintillator), and looks for flashes; when two flashes occur at a short-time spacing, it's likely that a meson has been captured (first flash) and created a muonic atomic orbital, which then decayed (second flash). The timing statistics show a strong signature at a couple of microseconds (muon decay time).

All my LEDs have a plastic block attached. I'd distrust randomness in the detector. I'm also averse to the kinds of trickery that gets good signal out of a (highly capacitive) sensor like most of my photodiodes.

Shot noise is randomness in the arrival of photons. That's trustworthy.

John