[sdiy] Popcorn Noise Generator?

[Tim Daugard]

Sorry to be late for the discussion. I was shut down by a hurricane
(again.)

> "Popcorn" Noise. A peculiar additional ac noise mechanism is sometimes
seen
 . . . Pulses
> can exist with widths of many milliseconds where the constant baseline
of
> random ac noise abruptly steps up by a large enough factor (say, 5:1)
to
> trace out a rectangular pulse,

>From this and other sources, it appears that popcorn noise should be
simulated by a waveform switching between two states. The noise is
actually the sound of the transistion (as discussed by all of you.)

> loudspeaker. Popcorn noise is caused by a defect that is dependent on
the IC
> manufacturing technique.

It occurs in more than IC's. In fact research led to references as
Random Telegraph Signals (RTS) and much more infromation under that
title.

> This ac noise results from abrubpt changes in the input current
between the
> normal level and a second, qusistable level that differs by
approximately

It appears that popcorn noise is a result of trapping and releasing
charges. If the charge is one electron, the amplitude is reasoanbly
constant.

> 100 pA. Evidence of popcorn noise is closely monitored by the
manufacturer.

One electron across the resistance of one metal / semiconductor atom?
(Beyond me at the moment. . . )

> however to guarantee that it is not present requires special, time
consuming

Looking at a Motorola document, It appears the use an integrater to
count the number of pulses, and reject thoses parts that have too much
popcorn noise.

> Testing and manufacturing techniques, I'm certain, have come a long
way in
> 21 years so I'd bet this is not as prevalent as it used to be.

I suspect the finer geometries of newer processes have reduced the
number of traps.

The analogy is popcorn - at low temperatures, nothing happens. If temps
rise fast enough the popcorn pops. I suspect the amplitude of each pop
is about the same. Popcorn too small is not mature enough to have the
popping property. Too mature popcorn has already seeded.

So I suspect the way to create popcorn noise is use a noise source -
filtered by a 10 to 100 Hz lowpass filter - fed to a comparator
(compared to a very slow LFO may be real good) and use the output of a
comparator to clock a flip-flop.

My two smallest local currency coins.

Tim Daugard
AG4GZ 30.4078N 86.6227W Alt: 3.7 M
http://home.sprintmail.com/~daugard/synth.htm

Who is going out to drag tree branches, trunks, leaves to the curbs -
but luckily no house parts (yet, Emily is next) Four named Atlantic
storms so far and 3 have hit within 150 miles. The closest so far was 39
miles with a 40 mile hurricane wind radius.