Time-staggered overdrive

[Rob Strand]

Simply changing the 68pf clock cap for maybe 82pf might address that, the clock frequency will still be high enough that the existing LPF should be fine.

I had this thought that one channel could be a true delay and the other channel could be an all-pass filter (a few in fact).  The all-pass filter would have a group delay equal to the delay chip to prevent nasty notches.   After that I realized the all-pass might defeat the effect of the analog delay.   Then I realized the short analog delay alone might sound like the all-pass!   Finally I decided I should stop thinking about it and build it (or at least do the experiment in DSP).

[ElectricDruid]

I'm interested to hear how you get on with this, Mark. It's something I've often wondered about too, and I've got as far as doing a test PCB for the delays and developing a couple of different distortion circuits. Like you, I was thinking of taking one signal and doing several different distortion tones to it and then delaying them a bit and mixing them back together, aiming for the sound of multiple guitarists playing very tightly. Things I want to discover by experiment are:

1) How long should the delay be? I was thinking "at the long end of chorus", so probably 20-40msec or so.
2) Does modulation of the delay help? A bit of detuning might help differentiate the sounds, but it might all get too much.
3) Can noise be kept under control? I was thinking of three channels, with two delayed. That's the added noise from three gain stages, plus two BBDs. Ouch. Or rather, "Shush!".

So...I'd be very interested to hear your take on this.

Tom

[bool]

I think all the answers you seek could be sucked out of the Scholz Rockman "chorus doubler" - and sure, small amount of modulation serves as a primitive "humanizer".

In reality, any "stupid" KISS guitar multi-take multi-tracking will sound "natural" vs. "synthetic, chorus-y" sound of a doubler-effect. Even if you beat-align the "stupid" guitar multis, each stroke will be slightly different in timbre, and result less "synthetic".

But there's one more thing to consider: any "multi-band" distortion will be somewhat "delayed" by nature -  any filtering will introduce some group delay.

[Mark Hammer]

But there's one more thing to consider: any "multi-band" distortion will be somewhat "delayed" by nature -  any filtering will introduce some group delay.

It likely will.  But as those who have studied the difference between time-delay and phase-shifting will know, it is one thing to unalign the phase of different parts of the spectrum, and another to move everything over by a fixed delay amount.

Now, will very short fixed delays (<5msec) yield an audible difference with mere group delay?  I hope to find out.

Got the board drilled and fully stuffed yesterday.  Some other things to attend to today, so the answer will have to wait a few days.

[pinkjimiphoton]

after rocking roland guitar synths for 20 some years, 5 ms latency is damn near unnoticeable.  the lower notes on some of the older ones, up to 28ms or so on the lowest notes. its noticeble and bothersome slightly that far out, but not insurmountable.

some digital amps have that much latency. <behringer "v-ampire" comes to mind>
and its not that big a deal. once ya get used to the tiny "seperation" its actually kinda groovy.

[Mark Hammer]

That's a different sort of latency, jimi.  And truth be told, if the patch has any sort of non-instantaneous attack time, one can easily withstand even longer latencies with a guitar synth.  I've mentioned the Casio MG-510 I have, in past, and the MIDI conversion has a bit of latency.  But whatever the latency is, even if I use MIDI thru to drive two tone generators, they will be triggered at about the same time.

This is a case of two ostensibly identical "sounds" (with different timbral qualities) being out of sync by a hair.  There will likely be some cancellations, but not that much.  Again, the objective is to mimic what it might be like to stand somewhere not in the exact middle between a pair of non-identical amps.

[EBK]

It occurred to me that use of two amps on stage might result in some very slight differences in ear arrival time, and perhaps some slight cancellations as a result. 

Depends on how they get to your ears, doesn't it?  At live stage volumes, is Billie Joe even listening to the output of his amps directly?  Or, are the amps mic'd up, and he is hearing them only through a monitor mix?

[Rob Strand]

2) Does modulation of the delay help? A bit of detuning might help differentiate the sounds, but it might all get too much.

Detuners and things like the Boss Dimension C are going the help with cancellation.    I wonder how an all-pass phaser set-up like a Dimension C would go?  Maybe not much point if you already have a delay.

3) Can noise be kept under control? I was thinking of three channels, with two delayed. That's the added noise from three gain stages, plus two BBDs. Ouch. Or rather, "Shush!".

If the BBD's are after the distortion boxes that might be minimal.

[pinkjimiphoton]

That's a different sort of latency, jimi.  And truth be told, if the patch has any sort of non-instantaneous attack time, one can easily withstand even longer latencies with a guitar synth.  I've mentioned the Casio MG-510 I have, in past, and the MIDI conversion has a bit of latency.  But whatever the latency is, even if I use MIDI thru to drive two tone generators, they will be triggered at about the same time.

This is a case of two ostensibly identical "sounds" (with different timbral qualities) being out of sync by a hair.  There will likely be some cancellations, but not that much.  Again, the objective is to mimic what it might be like to stand somewhere not in the exact middle between a pair of non-identical amps.

i really only meant that a 5ms delay to allow the processing shouldn't even really be noticeable to most players.
i still think making it envelope controlled to control the time delay would make it more likely to sound like two players or two sources playing the same thing with different tone, being able to move just a couple ms ahead or behind would likely make it much more organic sounding of an effect. the dynamic quality of air-mixing two signals may be better approximated that way i'd think. that way, not a chorus of flange, but a sorta envelope controlled comb filtering effect.
seems theres a lot of possible ways to skin this cat.
and it also seems it may be legit, in that to my ears at least, in dual amp live situations it seems like one amp will always favor some frequencies over the other, and you could use the envelope control to exploit that as well... some frequencies are "faster" than others... takes more time to produce a bass frequency <moving the whole cone more> than a treble one <less movement, more centered on the dust cover> with a speaker than a treble one, right?
i shut up now and lurk

[Mark Hammer]

Again, in terms of "when the sound starts", 5msec shouldn't be noticeable.  But as the Haas effect indicates, even very short inter-ear differences in arrival time are detectable.  Maybe not consciously, but we do make use of the inter-ear differences in identifying where a sound is coming from.

Now, would two slightly time-staggered signals, mixed down to mono, and ultimately heard the same by both ears, be anything special?  I don't know.  As noted earlier, blending of two different overdrives or distortions to a mono out, was explored by Boss in several pedals, and now by MXR in their Dookie.  I'm just curious about whether a teensy bit of time-staggering improves that formula.

[Rob Strand]

ut as the Haas effect indicates, even very short inter-ear differences in arrival time are detectable.  Maybe not consciously, but we do make use of the inter-ear differences in identifying where a sound is coming from.

Haas is based on binaural hearing.  There's quite strong evidence showing we can detect down to about 30us difference in arrival times between the two ears.   It works best for tick and impulse sources.  The 220us or so delay found in cross-feed systems can clearly move the perceived location to the left or right.  I've tried this with true delays and also group delay from an all-pass filter - the perceived location is quite similar.

When there's two different sound sources presented equally to both ears you have to apply the rules of masking (backward and forward).   This is more your case.  I suspect the longer delays might help.   (The cancellation thing is audible regardless of the delay.)

Long delays with all-pass filters produce some odd effects.   Even if you keep the Q's low enough to prevent audible ringing, you still hear a chirp sound.  When you have a true delay there is a gap of no sound but the all-pass filter fills that zone with a chirp.   The chip exists for a long time so it doesn't get masked.   For the shorted delays used in cross-feed systems the chip is very short and blends into the real signal.

I think all the answers you seek could be sucked out of the Scholz Rockman "chorus doubler" - and sure, small amount of modulation serves as a primitive "humanizer".

It's looking like a two input version of this type of thing would be the most successful way to do it.  You have two "voices" except they originate from different sources.

[bool]

There's quite strong evidence showing we can detect down to about 30us difference in arrival times between the two ears.   It works best for tick and impulse sources.  The 220us or so delay found in cross-feed systems can clearly move the perceived location to the left or right.  I've tried this with true delays and also group delay from an all-pass filter - the perceived location is quite similar.

Didn't Q-sound use allpass networks?

[Rob Strand]

Didn't Q-sound use allpass networks?

I haven't kept up with this stuff at all, but apparently they do,

https://patents.google.com/patent/US5440638A/en?oq=US5440638

I haven't read through the details so I don't know how much delay they are using.
That's an old patent the modern "Q-sound" might be different.

You can only go so far using all-pass filters before it starts chirping.    For all-pass smallish delays are fine.   The all-pass filter seems possible for positioning images in stereo. Something located on one side of you (90 deg) only requires about 750us interaural delay.

For me, the hanging issue is to see if designing all-pass filters for a flat delay over the entire audio-band pushes the chirp outside of the audio band.  The idea here is there is no dispersion of the audio signal so it acts more like a perfect delay.    For digital implementations that may require upsampling.

[Mark Hammer]

That Q-sound patent reads like my description of the stereo image enhancer.  I was always under the impression that the Q-Sound one had to license for studio use (the ways that Aphex Aural Exciters were initially licensed/rented) was digital.

[Rob Strand]

For me, the hanging issue is to see if designing all-pass filters for a flat delay over the entire audio-band pushes the chirp outside of the audio band.  The idea here is there is no dispersion of the audio signal so it acts more like a perfect delay.    For digital implementations that may require upsampling.

FWIW,  I tried the wide-band allpass idea and it is possible to get a delay and push the chirp (dispersion) outside of the audio band.    I did this in the digital domain (upsample -> allpass filter -> downsample).  The downsampling actually removes some of the chirp.   It requires a lot of stages and is totally impractical for analog circuits.   In fact even for the digital domain it's a bit crazy since you would just use a simple sample delay.   For fractional delays on top of the base delay you can use a short all-pass delay.