RST Engineering wrote:
Roy ...

I've been playing around (ahem, excuse me, heuristically engineering) with
zener noise sources for a while using the same spectrum analyzer trick and
as yet I haven't been able to make the noise as "flat" across the passband
as I'd like. I've tried varying the bias, the voltage, and a few other
tricks, but as yet, no joy.

Can you shed some light on what you've found to make the noise power/voltage
fairly level across the band?

Jim




Some are very noisy. The noisiest I've seen have been ones in the 12 - 15
volt range when biased at considerably less than a mA. I've used one,
followed by a 50 ohm amplifier "pill" IC, as a broadband noise source to
see filter responses with a spectrum analyzer. The noise is easily visible
well up into the UHF region.

But all zeners generate some noise, so you have to use appropriate
filtering in sensitive applications. In my experience, though, band gap
references can be even noisier than a typical zener.

Roy Lewallen, W7EL



What I did for some project which needed equal amplitude
uncorrelated noise,is amplify the zener noise with a wide
band video opamp,with high pass and lowpass filtering.
Used the low pass as input for a zero cross detector,
delayed the zerocrossing 10 microseconds,and used that
for clock to a circulating bit in a shift register.
each parallel output of that register controlled a
sample/hold opamp,sampling the highpass signal.
Voila!! 8 audio frequency, non-correlated noise sources.
The zerocrossing clock was made this way,to avoid detectable
clock tones int the output.(2 to 20 microsec between
crossings)
the 10 microsecond delay was used to get a voltage at
the sample and hold opamp which was not correlated to the
zerocrossing.
If you need only one signal ,leave out the shift register,
and just use a 10 and a 1 microsec. oneshot for the s/h
opamp clock.
The application? A wind and engine noise generator for a
car simulator.