I understand perfectly. Nothing random at all about your absolutely predictable pure horseshit. Looked like for a while there, you had become something more mature. I'll surely update that assessment... again.
I don't see much difference between his electronics and what you claim will work.
He is making a lot of very good measurements, which is the real issue. Very few people have the equipment or the skill to make these measurements.
Your pdf doesn't say anything about how random the noise is, and it says nothing about wild variations with changes in current with some zeners while others remain predictable.
Ritter was showing frequency spectrum, autocorrelation graphs, and fit to Gaussian curve. They are all analog measurements.
Once you get into digital, there are many readily available tests to see how random the signal is. Die-Hard is a good place to start.
Prove it.
You can't just look at a scope waveform and say it is Gaussian.
[...]
He was working with other sources besides zeners.
Jim Williams wrote a number of articles on measuring low level signals. He came up with the same solutions as Ritter.
You have often complained about the RFI from a local radio tower and how it got into every scope measurement. So you now tell us you measure Johnson noise on the bench with no EMI problems? I don't see how you can say these issues are silly.
So tell us how you measure Johnson noise on the bench and out in the open with no shielding.
Crosstalk to microvolt level signals is a severe problem.
Until you can provide measurements that analyze the noise properly, your claims don't mean very much. Scope photos do not qualify.
Well, other than being right?
No, that would be a major project, and there's no upside for me here.
-- John Larkin Highland Technology, Inc picosecond timing laser drivers and controllers jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
It demonstrates that zeners have pretty much three distinct regions: spikes at low current, popcorn type behavior at mid current, then increasingly gaussian wideband noise at higher currents. This is a well-known pattern. Moral: keep the current up.
If a zener is only going to cause state jumps in a digital scrambler, we don't care much about its statistical excellence. It only needs to inject some physics-based randomness.
He missed the time-domain waveforms, which show what is actually happening.
Go to the Noisecom or Noisewave sites; they make noise diodes and have specs. Noise figure meters use zeners, too.
Yeah, like bandgap references.
Altoids can, or Danish Butter Cookie can for bigger stuff. The big problem with high impedance measurements is of course 60 Hz pickup.
I never said on the bench with no shielding. We do get down to PPM noise levels, on PC boards, and amps dominated by Johnson noise, with switchers on the same board, but that's products in a rackmount box.
Then don't let it happen.
-- John Larkin Highland Technology, Inc picosecond timing laser drivers and controllers jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
Actually No, Die Hard is not a good place to start. Die Hard is intended to find faults with PRNG's. Only a few of Die Hards tests are valuable for a TRNG.
[..../] IRONY
-- Easy doesn't mean right, since if you take out the low frequencies you color the noise, which makes it less random. John Fields Professional Circuit Designer
What we want to take out is 1/f noise, which makes the comparator output *less* random. Big blobs of low frequency noise will rail the comparator for relatively long intervals, hiding all the good high frequency stuff.
But all we want is a random source to make the LFSR keep jumping states. That source needn't be statistically great.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.