Feedback Synthesizer Design Help Request Eurorack

Cwejman, Livewire, TipTop Audio, Doepfer etc... Get your euro on!

Moderators: Joe., lisa, luketeaford, Kent

Post Reply
belzib
Learning to Wiggle
Posts: 5
Joined: Tue Mar 03, 2020 8:18 pm

Feedback Synthesizer Design Help Request Eurorack

Post by belzib » Tue Mar 03, 2020 8:56 pm

I'm attempting to design my own synthesizer using eurorack modules. I've been playing with eurorack for a few years, but this is the first time I have a system with a specific purpose in mind. That purpose is to make music based on a system of negative feedback loops. The post below is super long, but I think it is quite clear and systematic. There are a few places where I think I'm suffering from not knowing enough about the underlying electronics or utility modules available. Maybe I'll post a tl;dr version later if no one responds to this, but I'm not sure how to do that. In order to understand the problems, you need to understand the overall aim. With that said, there are specific problems that need to be solved (especially those described in the third to last paragraph) and some of the solutions I've proposed might not be viable.

Negative-Feedback Control Synthesizer

System 1 generates a random signal (gate, CV amount, audio rate) which is fed into system 2 (which may be a single module or a collection of modules) which affects some active parameter of system 2. An active parameter is some feature of the system which already has an effect on the output of the system independently of the signal received from system 1. System 2 either outputs an audio rate signal affected by the active parameter directly to an output module or provides a CV or gate affected by the parameter to some further system which can produce an audio rate signal. System 2 must also produce some signal which can be fed back into itself and thereby “correct” the system, i.e., cause the parameter affected by the random source to revert to the desired, previous state.

It’s desirable that (a) the error-correcting signal have some controllable amount of delay so that one may control the time it takes to correct the parameter either directly or via some automated signal, (b) the error-correcting signal is generated by a system that allows one to control the amount of correction, (c) it is possible to invert the correcting signal from a negative control signal to a positive signal causing feedback oscillation, and finally (d) multiple feedback control loops may be implemented within the synthesizer with opportunities for loops to constrain, modify, control or otherwise interact with one another.

Possible implementation

System 1 consists of a random CV, gate, and audio module (e.g. the Make Noise Wogglebug) or collection of modules which perform these functions separately. It would be aesthetically pleasing if the random source we a product of the environment external to the synthesizer by, for example, using a module which converts biometric data into control signals or one that converts ambient sound into control signals. Functionally, however, this is superfluous; randomness is randomness. System 2 consists of (1) an oscillator where the active parameter is the pitch of the output waveform. If the random effect of system 1 is a change in the pitch of the oscillator, the output of the oscillator should also be used for corrective feedback as it will vary systematically with the random input. There are therefore three additional requirements. System 2 must also consist of (2) some mechanism which is able to detect this difference in output, (3) some mechanism for producing a corrective signal sensitive to this difference, and (4) a way of applying the signal to the oscillator to correct the parameter.

Module types which may be able to address requirement (2) include comparators and slope detectors. Comparators trigger a 5v gate whenever an input signal passes a set threshold. Slope detectors trigger a 5v gate whenever and input value increases, decreases or remains constant. Such modules may not be able to handle audio-rate signals as inputs, however. A possible solution is, therefore, to first feed the audio signal through an envelope follower before patching it through to the comparator or detector.

For (3), the most natural solution would be to use the output of the oscillator which is to be controlled. There are two problems. First, the oscillator cannot be corrected using an audio rate signal. The solution may again be an envelope follower, but it’s unlikely it would produce the right kind or amount of signal. Second, the correcting signal must be calculated based on the difference between the signals original value and the value resulting from the random disturbance of system 1. There must be a mechanism for calculating this difference. That value must then be inverted and fed back into the system as described in (4) below. The inversion can be accomplished using an attenuverter module. I’m not sure what module could be used to calculate the difference between the original signal and the randomly derived signal.

For (4), the main problem is that in most cases, any specific parameter of a module is controlled by only one input. If, for example, the random source is controlling the pitch of the oscillator via a 1v/octave input, then the corrective signal cannot apply. A potential solution to (4): the DPO is a dual oscillator with two separate 1v/octave inputs per oscillator. The first oscillator can be made to follow the pitch of the second. The first oscillators 1v/octave receives the random signal. The second oscillator receives the corrective signal to its 1v/octave input, along with a 5v gate to the follow CV control input which triggers the first oscillator to follow the second only when the first departs from the desired state.

Problems

Ideally (1) wouldn’t be an oscillator but some module which generates control signals and gates. This would allow for far greater flexibility. To that end and also to address the outstanding problems in (3), my first guess is to experiment with sample and hold modules, slew limiters, logic modules, offset/attenuators, and maybe phase-locked loop modules.

User avatar
sir stony
Common Wiggler
Posts: 250
Joined: Mon Feb 25, 2019 6:09 pm
Location: Bavaria

Re: Feedback Synthesizer Design Help Request Eurorack

Post by sir stony » Wed Mar 04, 2020 2:57 am

I have a feeling this concept sounds much more complicated than it really is, but I'll wait for the optional "tl:dr" version of this anyway...

The point that I'm struggling with is the idea of some "correcting" signal. I'd say instead of correcting the system, you need the system to be stable in itself and just need to reduce the amount of modulation injected into it when things get too wild so it will return into its stable state. There are ways to detect some aspects technically, but if the sound is really undesirable can only be decided by your own human factor, after all, and making this an automatic process might prevent you finding the unexpected shades of awesomeness in your system. 8-)

belzib
Learning to Wiggle
Posts: 5
Joined: Tue Mar 03, 2020 8:18 pm

Re: Feedback Synthesizer Design Help Request Eurorack

Post by belzib » Wed Mar 04, 2020 12:36 pm

Yeah, the basic concept is just that there is some parameter that randomly departs from a desired state and then a system that automatically causes the parameter to revert back to the desired state. You're definitely right that I myself could play the role of detection and correction, but I can also play the role of an LFO (or at least a slow one!). It's just nice to have these things automated sometimes, especially because it then allows you to manipulate the parameters of the automated system. But really, my interest comes from biology and cognitive science where the use of negative feedback control is ubiquitous. So it's sort of a theoretically interesting project for me more than anything.

User avatar
Pelsea
Veteran Wiggler
Posts: 667
Joined: Thu Apr 19, 2018 11:46 am
Location: Santa Cruz CA
Contact:

Re: Feedback Synthesizer Design Help Request Eurorack

Post by Pelsea » Wed Mar 04, 2020 2:09 pm

I've done something like that in the distant past. We had a Gentle Electric pitch to voltage convertor in the UCSC studios (Carl Fravel was my predecessor in my job there.) We generally used it to track instruments like flute, but it was also a lot of fun in feedback setups. For instance with a microphone picking up the speaker playing the patch output, combining the GE with keyboard control at the oscillator would produce strange arpeggios.

Gentle Electric is not active, but Disting does pitch and envelope to voltage. It even has a slew rate control on the pitch algorithm.
To get the difference between two voltages, invert one and mix them. A Disting can do this too.
You should also look into phase locked loops-- I was able to get something like what you are describing with a synchrodyne.

Random->SyncVCO->PLL clk, STO->PLLin, PLLout->STO

It produces anything from a slow portamento to a chaotic mess depending on the PLL track speed.
Books and tutorials on modular synthesis at http://peterelsea.com
Patch responsibly-
pqe

User avatar
DanPacific
Wiggling with Experience
Posts: 356
Joined: Thu Aug 02, 2012 1:43 pm
Location: Gothenburg, Sweden

Re: Feedback Synthesizer Design Help Request Eurorack

Post by DanPacific » Wed Mar 04, 2020 2:38 pm

There's quite a few useful utilities you can look into. Some may be obvious, but I'll mention them anyway. Envelope followers, attenuverters and VCAs will let you set up a patch where it'll be easier to control feedback systems; patch a copy of the signal though an envelope follower, invert it, and let it control a vca placed within the feedback loop – this way an inverted, "mirrored" image of the amplitude controls how much is fed back into the loop. This patch can be applied to a lot of different things and is fun to experiment with. I also warmly recommend comparators and similar. Check out Ladik's comparator J-120 and the J-110 Derivator – both great for feedback patches. I also warmly recommend logic modules, like the Doepfer A-166.

User avatar
nostalghia
Ultra Wiggler
Posts: 904
Joined: Sun Apr 07, 2013 2:46 pm
Location: central usofa

Re: Feedback Synthesizer Design Help Request Eurorack

Post by nostalghia » Wed Mar 04, 2020 2:40 pm

Addressing this part of your post only:
"For (4), the main problem is that in most cases, any specific parameter of a module is controlled by only one input. If, for example, the random source is controlling the pitch of the oscillator via a 1v/octave input, then the corrective signal cannot apply. "

Solution for this (if I understand what you're wanting to do correctly, didn't read the entire post in depth) is to use a mixer module that is suitable for CV, not just audio (look for "DC coupled" mixers). This will let you combine or blend two or more inputs (control signals in this case) then patch the mix output into your destination module's single input.

The mixer module could be anything from a simple "unity mixer" (all inputs are combined at equal level, so any variation in input signal strength or polarity would have to be controlled at the source-unity mixers have no level controls) to a simple 3 or 4 input mixer with knobs for the relative level of each going into the single mixed output, to a larger, more versatile, multi-channel mixer with multiple outputs which can be used independently or cascaded together, many featuring the ability to invert (polarity) as well as attenuate each input, and/or add an offset voltage.

Examples:
Unity Mixer
Basic mixer w/level control per input
Multi-channel mixer w/inversion and offset

Edit-previous post covered similar ground while I was still typing...
Edit 2: forgot to mention Matrix Mixers/VCAs-very useful in feedback oriented patching. Doepfer A-138m, 4ms VCA Matrix are good examples.
Last edited by nostalghia on Wed Mar 04, 2020 3:45 pm, edited 1 time in total.
“If you look for a meaning, you'll miss everything that happens.” - Andrei Tarkovsky
https://soundcloud.com/nostalghia70s

User avatar
sir stony
Common Wiggler
Posts: 250
Joined: Mon Feb 25, 2019 6:09 pm
Location: Bavaria

Re: Feedback Synthesizer Design Help Request Eurorack

Post by sir stony » Wed Mar 04, 2020 3:01 pm

Good shot, Pelsea! 8-) I was just this second starting to type an answer that would point at PLL when yours popped up. The Doepfer 196 might be more hands on than a Disting, tho.
So, PLL is a way of control over frequency, cyclewise.
Then there's the amplitude. The direct approach here would be a compressor, (check out the WMD compressor - not the mscl) and wired in well, that can do wonders, like using the env output into a vca to reduce the signal or cv that caused the source to inflate the volume in the first place. Using a slope limiter with that output (or a cv compatible delay, check out the mungo d0) can make the regulating circle softer and less immediate.
You may also want a filter or even better, two. One to manipulate the sound itself, and one to focus modulation or manipulation. As you aim for feedback sounds, you wouldn't necessarily need filters that go into self oscillation. The Eowave Fluctuations Magnetiques can do well here, which I use for such things, too. For more trickery with feedback and self-modulating system concepts, take more looks at the ADDAC modules 215 and 216, and the Klavis Flexshaper. What did you intend to use as non-random sound source? I'm not getting tired to suggest the Twin Waves, as it can offer a sound source and a modulation lfo at the same time, and even sports quantizers. Probably my most used module altogether.

belzib
Learning to Wiggle
Posts: 5
Joined: Tue Mar 03, 2020 8:18 pm

Re: Feedback Synthesizer Design Help Request Eurorack

Post by belzib » Thu Mar 05, 2020 9:19 pm

Thanks for these suggestions! This is very helpful and makes me feel like I’m generally on the right track.

“Solution for this... is to use a mixer module that is suitable for CV, not just audio (look for "DC coupled" mixers).”

I had never thought of CV mixing, even though I have the Intellijel Panar II which (I think) can do this.

“To get the difference between two voltages, invert one and mix them. A Disting can do this too.”

This is why I'm in the humanities. I really overthought (or I guess underthought) that one. The Disting being the Expert Sleepers module? I had suspected that a module like that one would have the functions I want buried somewhere inside it. I'm also curious if you still have any of the feedback recordings you did back in the day. I'd be happy to hear.

In any case I’m definitely going to check out the Disting, PPLs, comparators and so on, and then try to implement the basic idea. Then I’ll probably recheck some of this advice to work out the kinks. Of course I’ll have to sell a few things first, but I’ll let you know how it works out!

Oh, and the Make Noise Wogglebug is the only random source I have on hand currently, so I'll start with that. I'll definitely check out your recommendation though. (On the other hand, I recently described the project to my brother. He wants to hook up a Geiger counter to a piece of uranium and hack the Geiger counter to output to the synth to achieve true randomness. This seems completely unnecessary to me, but he does have the resource and the know how. So we'll see about that!)

User avatar
moremagic
Super Deluxe Wiggler
Posts: 1083
Joined: Wed Jun 23, 2010 8:45 pm
Location: NC
Contact:

Re: Feedback Synthesizer Design Help Request Eurorack

Post by moremagic » Thu Mar 05, 2020 9:59 pm

many mixers can subtract one signal from another, all you need is to invert the signal you wana subtract then mix them (a lot of mixers can do both at once)

desolationjones
Wiggling with Experience
Posts: 401
Joined: Sat Jan 21, 2017 5:55 pm

Re: Feedback Synthesizer Design Help Request Eurorack

Post by desolationjones » Thu Mar 05, 2020 10:07 pm

Wogglebug has a couple PLLs in it! The earlier model is easy to patch as a pitch extractor but I forget how :doh: Walker could tell you though.

belzib
Learning to Wiggle
Posts: 5
Joined: Tue Mar 03, 2020 8:18 pm

Re: Feedback Synthesizer Design Help Request Eurorack

Post by belzib » Fri Mar 06, 2020 11:47 am

Thanks moremagic. I did a search on modulargrid: is it a polarizing mixer you have in mind that could do both at once? That would seem to do it.

User avatar
sir stony
Common Wiggler
Posts: 250
Joined: Mon Feb 25, 2019 6:09 pm
Location: Bavaria

Re: Feedback Synthesizer Design Help Request Eurorack

Post by sir stony » Fri Mar 06, 2020 1:04 pm

That was exactly my point, suggesting the addac 216.
Sure, the Disting is great and can fill in for an insane amount of different uses, I have one as well, but the interface is so non-wiggly... :zombie:

belzib
Learning to Wiggle
Posts: 5
Joined: Tue Mar 03, 2020 8:18 pm

Re: Feedback Synthesizer Design Help Request Eurorack

Post by belzib » Fri Mar 06, 2020 1:15 pm

Aha. Thanks sir stoney. That looks like a good option too. I'll just have to think about it in terms of price. The menu diving is a big turn off with the disting and other multi-fuction modules, but the value is attractive.

Post Reply

Return to “Eurorack Modules”