A proposal: "Polyphonic Inhomogeneity"

I just read the manual of the prophet 5. It is not clear what exactly the vintage knob does. In one sentence they talk about random fluctuations (so modulation?) and in another it is “differences from voice to voice”. I will give an update to this, when I receive my Take 5 hopefully around christmas.

I’m sure it’s a nice synth - Matt Johnson likes it :slight_smile:

the vintage knob is at 11:13

1 Like

Here’s how I look at both:

  1. Per voice offsets, which you will get on polyphonic synths with discrete voice cards or chips. These offsets are usually static (ish - I’m sure they’re not 100% never moving but it probably happens more over a longer period of time). Some synths like the Jupiter-4 or Oberheim OB series allow you to modify the voice cards to your liking by changing the calibration of each voice. This is also basically what the Vintage knob on my Prophet-10 Rev 4 seems to be doing: applying wider and wider offsets to tuning, envelope timing, filter cutoff, resonance, maybe even env depth. After 1 or 2 o’clock, things are wildly out of tune and the envelopes and filter are extremely sloppy. Even at minimum values though, you still hear things are not totally perfect between voices, but pretty tight.

  2. Drift. This is what you’d expect on a VCO as the temperature changes and will be more random but LFO-like in nature. You turn on the synth, and play, and everything’s out of tune and wonky. You let it warm up a few minutes, and now things are better. You wait an hour, and now you’re grabbing the oscillator fine tune to try to get things back or running a tuning procedure (or just letting things stay out).

Either way, I still think those types of features are best built into modules.


The deeper you dive…

This guy has written a paper about the topic (and called it Voice Component Modeling).

Excellent read and a lot of video examples.

  • mo

What you’re looking for might be Towers from unless modules. I am currently working on a small module collection that incorporates Towers and N1. Here’s a not so good and soundless preview of my 2HP versions. Basically it takes a Polyphonic Signal array and holds it… when Voltages are stored… you can exchange the input and use the new Signals as modulators or for exchanging stored values.

"CVBank is a clone of unless modules Towers - and except for its mode options brings the original 18HP down to 2HP. Towers is an advanced S&H which stores 16 Voltages per Row and keeps them until the user overwrites or resets them.

Plug a signal into its input, grab all ch. using the manual button, a mono trigger or use poly to grab specific channels. After the CV’s stored, remove the cable.

CV Bank or Towers has two modes for dealing with modification of stored signals:

  1. non-destructive Modulation
  2. destructive Updates of all or specific parts of the storage

The Input can now be used to feed another set of Signals into the module. Opening the attenuverter will mix the incoming signal with the storage (CW is additive, CCW is subtractive) - a CV in allows for precise per channel control if wanted. This method is non-destructive and closing the Attenuverter or disconnecting the Input will return to the original storage. By pressing the trigger or using the trigger in, voltages can be overwritten by the respective inputs. CV Bank has two identical rows for storing 2x16 signals.

SHSeq (working title)

SHSeq is a combination of one unless modules Towers row (top) and a 23volts N:1 polyphonic sequencer/switch.

S&H/Voltage Storage/Modulator

  1. Signal in (up to 16ch)
  2. Grab Button
  3. Voltage Range (0-10 or -10-10)
  4. MOD CV input
  5. Attenuverter/Mixer
  6. Trigger/Grab CV input
  7. Polyphonic output

Internal routing sends the output signal into the bottom section.

Poly Signal to Mono Sequence

  1. CV input for channel/step selection
  2. Step forward TR/GT input
  3. Step backward TR/GT input
  4. Step random TR/GT input
  5. Reset Seq. (to Step 1) TR/GT input
  6. Step output - holds CV of sel. CH
  7. Trig output (Pulses when ±rnd in)

SHSeq is compatible with 8Face & Transit. An exotic, versatile Interface that joins utilities, modulators, random signals and precise sequences together for all sorts of purposes."

i,e,; Get a set of voltages, save them in CV bank, run output into an offset and back into both CV bank inputs, set a little negative offset and save Bank 1, set a little positive offset and save Bank 2, remove offset and put Bank 2 output into Bank 1 input… turn the Attenuverter up and insert an… LFO or Gates or a Random source or send various modulation sources via a polyphonic cable and modulate the voltages channel specific.

Easiest might be to use it with Transit.

Whatever solution you come up with, you can probably save and automate it using this.


Having designed an analog poly synth, I can tell you what they do. The are horribly out of tune “naturally”, and yes, a big part of that is each voice being many cents sharp or flat. BUT - they all had a tune button (including the original prophet 5. Once you “tuned” it, all the voices had pretty much the same initial pitch and scaling. But they would quickly drift out of tune. As others have said, that drift is primarily thermal, with different voices being at different temperatures. So there didn’t tend to be a “this voice is always 3 cents sharp” but more each voice had a very small offset from each other to start with, then they would all slowly drift. Were/are there “micro drifts” that are faster? Maybe, I don’t know.

Most (all?) synths of that era would save their calibration into info non volatile memory so they would stay there. And once your synth was warmed up the drift was not too bad. The (very few) times I every played one in public, I would hit the tune button after each song, “just in case”. Once it was in tune each tune cycle just took a few seconds.

As far as what would sound good now? I guess people have different theories, or have taken more or less accurate measurements. They would certainly know more than I about that. I was just always trying to minimize it, not emulate it.


Thanks for your input, that seems to be a very good idea to use as a standalone module. I am very pleased with the Module from LindenbergResearch at the moment, but will try other ideas as well.

I am not talking about Pitch Drift. After reading and watching a lot more about the current Sequential Synths and the implementation of the vintage mode in the new Prophet5/10 rev4, Prophet 6 (there was a firmware-upate which turned the “slop” into the vintage mode) and now the Take 5 the voice-to-voice variation is not a simple modulation of pitch to simulate pitch-drift.

I am sure there are a lot of other implementations of the same concept with different ideas, but I was especially interested in the way Sequential imlemented this feature. The knob on the Prophet5 Rev4 demonstrates this idea as it goes from 4 to 1. The values correspond to the version of the Prophet5, if you would buy it today. So on 2 this means, if you would buy a used Prophet5 rev2 today, that would the best approximation of the sound you would get out of that. Yes, it is a bit of marketing mixed with analog/vintage vibe and I do not say I need it, I was looking for a way to simulate that in VCV.

So what is it? Constant (!) offsets for Pitch, most likely Attack, Decay, maybe Cutoff and Resonance per Voice.

Why? I don’t know, I am not an electrician. I assume some kind of deterioration of components (Leakage, old Capacitors, etc. ?).

Thank you for posting this . It has been a very interesting read. Actually this entire thread is very interesting.

1 Like

It seems pointless to argue about what the mystery knob does on the prophet. As you said, when you get one you can attempt to find out what it does.

btw, at least in the US, an “electrician” would not know the answer, either.

We’re lucky enough to have the service guide for the Prophet-5 online:

If i understand correctly, the TUNE only does the oscillators - not the envelopes and filters - that has to be done manually. and i wonder if the tuning algorithm tunes every note in the range for each oscillator.

This video is an interresting watch, if one wants to know more about tuning the original Prophet-5.

The bit about adjusting the filter envelope trimmers starting around 19:42 to 25:10 shows how small adjustments affect the sound, and is only one ingredient in the analog “vibe”.

And offcause one doesn’t know if it’s the Prophet-5 they want to approach with the vintage mode. Could be even older synths/organs.

Wouldn’t hurt to ask them, but they’re probably not going to release their source code or schematics.

You are right, “Electronic Engineer” - pardon my german lazyness when it comes to correct translations.

And why is it pointless to argue? Those are not my findings, but from the author of the article I posted above and what you can hear on the video I posted in my first post. I do not own a Prophet, nor will I ever…

BTT: At the moment I am trying to model the “Harmonic Frequency Jitter” he showed in the following video. I got it working in VCV but I can definitely not hear a difference.

That sound right. fwiw I have serviced Prophet Rev 1 and 2 “back in the day”. Also, the tuning code in the prophet was designed by (written by) the famous Dave Rossum. I paid him for some consulting in the early 80’s when I was working on the turning code for my Voyetra-8.

Anway, yes, I’m pretty sure you are right, the P5 only tuned to oscillators pitch. And I’m pretty sure it only tuned it at two points, as was the custom for all synths back then. So, yes, differences between the filters could be huge. and, yes, why not differences it the ADSR. They were analog in the prophet rev 1 and 2 I know. Maybe the LFOs were even analaog, I don’t know.

When I was working on the Voyetra-8 a few years later we originally had analog adsr and lfo, but there was just too much circuitry to fit in the box, so we used the main CPU to generate ADSR and LFO. In addition to tuning the VCO pitch at two frequencies, we also computer tuned the pulse width so that at 50% they would all be more or less square.

I think later analog polysynths, like the Matrix-12, might have also calibrated the filters, but I’m not sure about that.

1 Like

Sorry. You started the post with " And with christmas around the corner I am on the brink of ordering a Sequential Take 5". I didn’t realize you were joking.

I meant that it’s pointless just to make guesses. “I think it is old capacitors” “no, I think it’s drift” - citing people’s research is not pointless at all. It’s quite interesting.

Oh! I am one of those. But I still don’t know at all what makes the magic of these old beasts.


Perhaps one day the schematic will find its way to archive org … interresting stuff i think - but I’m only an electronics amateur…

Hm, maybe I don’t get it, perhaps something lost in the translation. Take 5 != Prophet 5? Or were you referring to see what the vintage knob on the Take 5 does, once I own it? Re-reading your post that seems to be likely now.

I know that you are one of the guys around here with huge knowledge and I always admire your input. But I am having a hard time in this thread here, where everybody is saying “Oh yeah, pitch drift, why should you want that?” or “That is pitch drift, but it wasn’t that bad, when the synth was tuned” or “Here is a module for what you try to do” - while the whole thing I wanted to achieve with this post was finding out if it would be an idea to “implement something like the vintage-knob on sequential synths directly into polyphonic modules, not just VCOs, but ENVs, VCFs, VCAs too” and then to further dig deeper into the actual implementation of the vintage knob to give a little bit more information.

1 Like