the POPCORN machine (a different burst box?)

[iainpunk]

we dont have that brand here in the netherlands, so i won't be going with that. i want to do something really pop-art-ish, since its a bit of a weird pedal anyway.

cheers

[ElectricDruid]

i was talking to my friend some more and i think im going with transistor noise, a shit-tonne of gain, and a Schmitt trigger to get really stable high and low levels.

This seems like a complicated way to generate digital noise, which is what you'll get at the end - noise with only two levels. That's exactly what a digital noise source generates "natively", and which we usually filter somewhat to turn it back into analog noise.

Note what you're proposing won't change the frequency of the noise, just its character. You'll have something that basically still sounds white (because your ears can't hear the difference between broadband analog noise and broadband digital noise). It doesn't turn into popcorn noise just by virtue of having only two states.

Given that the first thing the circuit

the way im mixing the random square with the clean signal is with this contraption:




instead of just adding pops, it also inverts the signal. this limits the amplitude of the 'pops' to the wave's amplitude, and also the bias offset of the input signal determines the amount of DC pop. the phase inversion wouldn't be to noticeable as it happens at such low rates.
this is just a proof of concept, a mock-up basically.

I don't understand what's going on here well enough to be able to comment intelligently, sorry. I don't even understand where the various signals shown are coming out from. However I don't see anything particularly unusual and it still looks like a "long way around" to get the result.

One thing that does jump out is that the first thing the circuitry on the RHS does is filter the incoming digital noise source (if I've understood you correctly) thus turning it back into an analog source - so (I think?) you could save yourself a *lot* of steps there.

Kipper is right too - StompLFO's random waveform will give you random pulse edges at a selectable rate (the classic S&H signal). If you need the rate to be more random, you could feed a second StompLFO to the FreqCV of the first, or even try feeding back the output to the Freq CV on a single chip. I've never done that, but it should be interesting, at least!

[PRR]

....This seems like a complicated way to generate digital noise, which is what you'll get at the end - noise with only two levels.....

Two voltage levels. But the transitions, the Time, is quantized randomly.

[ElectricDruid]

....This seems like a complicated way to generate digital noise, which is what you'll get at the end - noise with only two levels.....

Two voltage levels. But the transitions, the Time, is quantized randomly.

True. But if the clock frequency is high, that's effectively true for a digital noise source too.

[iainpunk]

I don't understand what's going on here well enough to be able to comment intelligently, sorry. I don't even understand where the various signals shown are coming out from. However I don't see anything particularly unusual and it still looks like a "long way around" to get the result.

One thing that does jump out is that the first thing the circuitry on the RHS does is filter the incoming digital noise source (if I've understood you correctly) thus turning it back into an analog source - so (I think?) you could save yourself a *lot* of steps there.

the transistors are basically an analog xor gate. it takes the relative difference of the 2 signals, and subtracts that form the VCC. the square wave goes between 0v and VCC/2, and the signal is biased at VCC/4, so the difference between the signal and the square wave is the same DC either way, but the phase is 180 inverted. a slight offset in the bias of the guitar signal generates slight DC pops, so controlling the bias controls the pop-level when no signal is present, which gets added to the difference between phase flipping blip.

the incoming noise's way of generation is kind of arbitrary for this proof of concept circuit i posted, as its to show how im mixing the popcorn with the signal. i totally forgot to show how bias offset changes the popcorn amplitude, although it does show the phase flip blips.

hope this explains the signals

....This seems like a complicated way to generate digital noise, which is what you'll get at the end - noise with only two levels.....

Two voltage levels. But the transitions, the Time, is quantized randomly.

True. But if the clock frequency is high, that's effectively true for a digital noise source too.

yes, this is a bit of a long way round to create digital noise, but the thing is, its the easiest way i can intuitively come up with that creates randomly timed long-ish intervals (10 to 0.1 Hz range fo bandwidth). i can generate analog noise with only 2 transistors (and some passives), and filter that analog noise to a lower bandwidth in the sub-sonic range, and then turn it in to squares with the Schmidt trigger.

cheers

[ElectricDruid]

Ok, I get what you're doing now. Thanks for the explanations, both of you. Sorry if I was slow to pick it up.

The non-quantised time that PRR drew attention to is interesting. Clearly with some firmware programming, you could write something that would generate this eventual popcorn digital signal directly, and probably with a time quantisation in usecs (e.g. not something you'd ever hear or care about). But *without* just sitting down and writing it from scratch, I agree, it's tricky to see a simpler way. The slower digital sources generally achieve longer times by reducing the sample rate, and hence increasing the time quantisation. Sometimes that doesn't matter - it can be cool to have a trigger signal that might occur or might not, but if it does, it'll always be on the beat. But this isn't that situation.

Some variation on Noise->Integrator->Comparator is probably going to be what you finish up with. I was wondering if you could short-circuit this by feeding a random control voltage to a 555 timer to get random intervals, but then I realised that the 555 timer is basically the integration and comparator steps, except in one IC. So it's fundamentally still the same idea, although it might use a few less bits.

[Rob Strand]

Ok, I get what you're doing now. Thanks for the explanations, both of you. Sorry if I was slow to pick it up.

I get how it works but I can't see the motivation for inverting the signal.

The discontinuity at the inversions are obviously where the clicks are heard.   However,  if you feed the comparator output through a pulse generator (differentiator, monostable, or any of those through a filter to tune to tone) that will also produce a pop on the edge.  In this case the original signal passes through and the pop noise is just mixed in, like mixing in hiss from a noise source.

[PRR]

If you have ever heard true Popcorn Noise, you won't be nostalgic. It's just crap. Very small crap too. Would go unnoticed on stage.

As for "making" it, today I would be real tempted to just record a long loop (a minute?) in an MP3 player.

[iainpunk]

Ok, I get what you're doing now. Thanks for the explanations, both of you. Sorry if I was slow to pick it up.

I get how it works but I can't see the motivation for inverting the signal.

i just think its a cool circuit, with a certain kind of elegance to the benefits it provides. i can turn down the popcorn to only pop when there's signal present, by biassing exactly in the middle, and moving away from the middle adds more DC pop, which is audible also when not playing.

PRR, that MP3 method is tempting indeed, but its not the end result that matters, its the fun of making a weird ass pedal. im finally using a few circuits i wanted to but never have had the need to before, makes me feel good.

cheers

[ElectricDruid]

As for "making" it, today I would be real tempted to just record a long loop (a minute?) in an MP3 player.

Lol, that's definitely cheating!

It's funny how often it's been done in the history of pseudorandom noise though - many times the easiest way is just a look-up table of "random" values. Read one, move on to the next. As long as the table is quite a bit longer than the timescale you're interested in, you'll never notice.

[Rob Strand]

i just think its a cool circuit, with a certain kind of elegance to the benefits it provides. i can turn down the popcorn to only pop when there's signal present, by biassing exactly in the middle, and moving away from the middle adds more DC pop, which is audible also when not playing.

It is pretty cool looking.

[iainpunk]

i just think its a cool circuit, with a certain kind of elegance to the benefits it provides. i can turn down the popcorn to only pop when there's signal present, by biassing exactly in the middle, and moving away from the middle adds more DC pop, which is audible also when not playing.

It is pretty cool looking.

thanks, i didn't **actually** design it, i just changed a TTL XOR gate to work in the linear region. biggest problem is that it has quite low input impedance, as do all common base amplifiers.
its a cool modulator as well, it sits inbetween an octave up and a ringmod when you provide two signals. for best modulation sound, the signals should somewhat match amplitude wise.

if one signal is ''silent'' and both signals are biased the same, it kind of gates due to Ube needing to be exceeded, and it creates an octave up effect. ill upload a more robust, higher input imedance, and no gating version when i have draw it up

cheers

[iainpunk]

the input signals can best be biased at VCC/4, which leaves 2.25v of headroom on either sides of the wave, which might be a little low for some applications.
it can lose signal amplitude when modulating (dependent on a whole bunch of factors), you might be tempted to increase the resistor from VCC to both transistors, but this diminishes headroom, its probably better to boost prior to or after this ''modulator''

it works best with heavily compressed or oscillator signals.

cheers

[Rob Strand]

thanks, i didn't **actually** design it, i just changed a TTL XOR gate to work in the linear region. biggest problem is that it has quite low input impedance, as do all common base amplifiers.
its a cool modulator as well, it sits inbetween an octave up and a ringmod when you provide two signals. for best modulation sound, the signals should somewhat match amplitude wise.

Thanks for the background.  That's some good out of the box thinking - I like it!