The Official Sequential/DSI Forum

Bug or Feature? Aux Envelope

Bug or Feature? Aux Envelope
« on: November 28, 2021, 08:42:14 AM »
Hi!

I am now a proud owner of a Take 5 and I try to understand the synth. So far I am very happy with the sounds and the sound design process, but I ran into something today I don't understand.

I designed a Sound with env1 set to aux (env2 is filter+amp). I used the aux-envelope to modulate noise: Noise fully CW. env-amount -127, A0%, D10%, S50%, R50% - That way I get a short noisy attack and a key-off noise sound as well, depending on the release of the aux and amp envelope.

When I hold two notes and then release the first note I play, the noise level is increasing as if the amp envelope on the first voice is not going through the release-stage or as if the release-stage from the aux-envelope of the second voice is also triggered by the release-stage of the first voice. Is this a bug or intended behaviour? I now that noise is monophonic, but the aux-env is polyphonic and should not overlap between voices.

Kind regards,

- mo




kpatz

Re: Bug or Feature? Aux Envelope
« Reply #1 on: November 28, 2021, 09:43:11 AM »
Noise level is a monophonic modulation destination, you can see this in the matrix when you select Noise as a destination, it only shows an M.

So, not a bug, but having it be a poly destination would be nice, but it may be a hardware limitation (only one CV channel to control noise level for all 5 voices perhaps)  Actually, all the mixer levels are mono mod destinations.


Re: Bug or Feature? Aux Envelope
« Reply #2 on: November 28, 2021, 02:07:42 PM »
Ok, makes sense, i am controlling mixer volume and not an individual VCA. Thanks for clarifying.
I am a bit disappointed in the mostly monophonic mod destinations. Even OSC1Shape is monophonic, when modulated by the aux-envelope while LFO2 modulates it polyphonic. That is strange...
Pitch, FM, Cutoff, Panning and mod-amounts are essentially the only polyphonic destinations which can be modulated by the aux-envelope (not counting envelope-parameters).
Hm...

kpatz

Re: Bug or Feature? Aux Envelope
« Reply #3 on: November 28, 2021, 03:01:08 PM »
Ok, makes sense, i am controlling mixer volume and not an individual VCA. Thanks for clarifying.
I am a bit disappointed in the mostly monophonic mod destinations. Even OSC1Shape is monophonic, when modulated by the aux-envelope while LFO2 modulates it polyphonic. That is strange...
Pitch, FM, Cutoff, Panning and mod-amounts are essentially the only polyphonic destinations which can be modulated by the aux-envelope (not counting envelope-parameters).
Hm...
I was just trying it out on mine.  Oddly, Osc1Shp shows both M and P, but it only functions monophonically, even when using LFO2 as the source (you can confirm this by modulating LFO2 rate with Key Number... the LFO corresponding to the first key hit affects the shape of all the voices).  Osc2Shp shows M only.  (EDIT: see my edit at the bottom of this post... I found a twist).

Although each oscillator's individual waveforms have their own VCAs that control both shape and overall mix level, I'm guessing these VCAs across the 5 voices share DAC channels and thus CVs, so all 5 voices have to have the same mix/shape blen at any given time.  As there is just one noise source, it may use a single VCA that is fed into all 5 voice mixers to reach the filter, or it may be per voice with a shared CV as the VCAs used for waveform blend and mix are built into the SSI2130 VCO chips and the mixing occurs on-chip.

The filter is much the same way.  Only cutoff is a poly mod destination... resonance and drive are monophonic, probably due to CV sharing as well.

Many analog synths that aren't designed to be multitimbral work this way, sharing CVs across voices for voice parameters that are either not modulateable or only modulateable monophonically.  On some, like Jupiter-8 or OB-Xa, there's two sets of CVs for each half of the voices, allowing for splits/layers.  The Rev4 Prophet 10 is like this too... the additional 5-voice card has at least some of its control voltages independent of the other 5 voices.

I guess this means we'll never see multitimbrality added to the T5 in a firmware update.

PLOT TWIST:  Osc1Shp shows both M and P, and I just figured out why.  PWM is polyphonic on Osc 1, so if you turn the shape knob around 3 o'clock and use LFO2 to modulate just within the pulse-width range of the shape knob, the modulation IS polyphonic, on Osc1 only.  Osc2's PWM is monophonic, and shape only shows M.  So, pulse width on Osc1 must have separate CVs per voice, but Osc2 uses a shared one.

TL;DR version:  Wave shape modulation is monophonic on both oscillators, but pulse width is polyphonic on Osc1 only.
« Last Edit: November 28, 2021, 03:37:15 PM by kpatz »

Re: Bug or Feature? Aux Envelope
« Reply #4 on: November 29, 2021, 02:23:22 PM »
PLOT TWIST 

Your plot twist made me playing around with OSC1Shape a bit more and it is even more confusing. I think you are right, that only the PWM-Part is polyphonic, which leads to funny modulation. But: The LFO2 (and the envelopes too) are working over the full range of the shape. That leads to the weird behavior, that the first half (sine to somewhere between saw and square)  is modulated monophonic and then the polyphony kicks in and modulates the second half polyphonically. Just set OSC1Shape to Sine and modulate it with an Envelope with 127 amount. Then set a slow attack and decay with sustain and release at zero and play two notes. Depending on the time between the two notes you can here the point where they reach the maximum or not, because too much time between them and the first decay is already in the monophonic section when the second voice is still on the attack stage, which immediately puts the second voice in line with the first.

I am not that familiar with analog poly-synths, I own a few VAs and analog mono-synths and played with modular a bit, so this may be a completely common thing among analog polys. And it's nothing I am too worried about because that seems to be an uncommon modulation. But I think this is a "nice to know" and may help with modulating PWM polyphonic in the "right" way.

kpatz

Re: Bug or Feature? Aux Envelope
« Reply #5 on: November 29, 2021, 03:43:35 PM »
That leads to the weird behavior, that the first half (sine to somewhere between saw and square)  is modulated monophonic and then the polyphony kicks in and modulates the second half polyphonically.
If you think of PWM and shape as two separate things, which they are in the hardware, just sharing a knob in the UI, it should be easier to wrap your head around.  The blend between sine, saw and pulse is controlled monophonically, while the pulse width is controlled polyphonically.
Quote
I am not that familiar with analog poly-synths, I own a few VAs and analog mono-synths and played with modular a bit, so this may be a completely common thing among analog polys. And it's nothing I am too worried about because that seems to be an uncommon modulation. But I think this is a "nice to know" and may help with modulating PWM polyphonic in the "right" way.
If you look at a more basic analog poly, like the Prophet 5, or a Juno, that lacks a full mod matrix, certain parameters are set by a knob but can't be modulated by an envelope or LFO at all.  For example, resonance.  Or, something like pulse width might only be able to be modulated by a single global LFO (on a Juno... I know the P5 has the polymod section).  For these instances, it's more cost effective to share a single control voltage for all the voices for that parameter.  Of course, things that need to have polyphonic control for the synth to be polyphonic, have separate CVs per voice, like oscillator pitch, filter cutoff, and output VCA level, envelopes etc.

The Take 5 seems to be designed in this manner.  If you look through all the mod destinations in the matrix, only a few are polyphonic... osc pitch, osc1 shape (PWM only), filter cutoff, envelope parameters, output volume, LFO2 rate, etc.  Some analog polysynths are built with every parameter individually addressable on every voice, like the Alesis A6 Andromeda, Moog One, and the Arturia Polybrute, but these synths were designed from the ground up to be mega modulation monsters and/or multitimbral.

Of course, with digital and software synths, the only limitations are in the code, CPU and memory capacity.

I have a hunch the decision to make PWM poly on Osc1 only wasn't just a cost cutting decision, but a flexibility decision as well.  By making each oscillator different this way, you have the choice of having mono or poly PWM, just by choosing which oscillator to use (or use it on both, in which case you can have poly on one and mono on the other).