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?

It's not particularly flat over the whole band, and I haven't attempted
to make it be. My interest has been mostly in relatively narrow band
filters, or the shape of a main filter rolloff. The noise is adequately
flat over the bandwidths I've been interested in. It's been a while
since I've fooled with it, but as I recall, the shape of the noise
spectral distribution changed all over the map as I changed the diode
bias. I imagine that it changes with temperature and with individual
diodes, too. So any circuit used to flatten it would only work for a
particular diode, current, and probably temperature. The bottom line is
that it's probably a lousy way to try to generate a flat broadband noise
spectrum.

Roy Lewallen, W7EL