AW: Pink noise circuit??

[Haible_Juergen#Tel2743]

>     When I was a kid in high school, I remember building a 2 transistor
>     pink noise generator.  I think it was nothing more than an NPN
>     transistor "biased" to 18 volts across the collector and emitter and
>     the base was not connected.  This caused breakdown noise that was
>     amplified by the other transistor.  It produced really clean and
>     random noise and didn't have that periodic thump that IC based
>     pseudorandom pulse generators produce (like the MM5837 IC).  So today
>     I can't find the hardware I built or the schematic.

Yes, the method works well. Just use the reverted BE as a zener and
amplify the noise.

Hints:
Take care of *large* coupling capacitors if You also want to derive slow
random voltages from the noise source.
You might need *lots* of gain - make sure that Your amplifier has enough
gain in the upper frequencies!
What You produce is almost *white* noise. For pink noise You need an
additional 3 dB /8ve filter. This can be approximated by several RC
networks. Cheap filters use two RC 's, good ones 5 or 6 ones (Moog
Modular).
But IMO, who cares for exact pink noise? *Any* filtering of white noise
should be interesting!

Problems:
(Why manufacturers are tempted *not* to use this method)
You either have to trim the gain of Your noise amplifier (good) or select
a transistor for a certain noise level (better).
Both is no problem for a single unit, but troublesome for mass production.
Once I have seen a design that used a LM13600-based ALC circuit (I
think it was in some Jupiter synth, but I'm not sure).

JH.