Cryptic words about TB-303... ?

[xhunaudio]

I just read this sentence about the TB-303 :

What makes the 303 unique amongst non-modulars is that the Envelope positively, then negatively modulates the Filter Cutoff (it's bipolar). These features are undoubtedly responsible for the 303's "squelchiness"

Anyone knows what "the Envelope positively, then negatively modulates the Filter Cutoff (it's bipolar)" means exactly ?

Since I think the author is not talking about the ENV MOD knob (it is not bipolar), what TB-303 "action/property" is he describing ?

Thank you for any hint...

[Music Engineer]

i think, it means, when you crank up the envmod, the excursion of the cutoff goes up more and then goes down more with respect to the baseline cutoff

[xhunaudio]

Thank you for your feedback Music Engineer,

ahh ok, so the EG, instead of a typical excursion, going from 0 to 1 to 0 (for example), on the TB 303 it goes from 0 to +1 to -1.

I hope the example above is clear, it's not easy to explain with a few words...

[chk071]

I guess that is what produces the typical "quack" when cranking up the env mod amount. Comparable to opening the attack a bit on the filter envelope on a "regular" synth.

[mystran]

The service manual actually explains this: increasing the envelope modulation also decreases the cutoff, in order to better keep the sweep in a range where it's more audible.

[imrae]

That's really interesting, and explains why it usually takes two hands to do good-sounding envelope-depth sweeps. We should be using macro knobs for this!

[urosh]

ahh ok, so the EG, instead of a typical excursion, going from 0 to 1 to 0 (for example), on the TB 303 it goes from 0 to +1 to -1.

Nope. It has DC offset. Envelope it self goes from 0 to 1 and back to 0, but some negative DC is added so it goes from -x to 1-x and back to -x.

[xhunaudio]

Thank you for all your feedbacks !

Interesting, so finally it is :

cutoff_knob + (mod_env * bipolar_eg)

where the bipolar_eg goes from -1 to +1 to -1 (and considering it to have symmetrical positive, negative ranges).

[mystran]

cutoff_knob + (mod_env * bipolar_eg)

Well the 303 hardware implementation is more like:
cutoff_knob + mod_env * (unipolar_eg - bias_constant)

Also I don't think the "bias_constant" is anywhere near half the EG swing, more like a 3rd or a quarter or something, although I can't remember the exact details right now.

[xhunaudio]

Found !

TB 303 service manual, page 8. It's about 1/3.

Thank you for the hint.

[mystran]

TB 303 service manual, page 8. It's about 1/3.

Right.. about 1/3rd puts it into the general ballpark of having the nominal cutoff roughly after one envelope time-constant.

[antto]

it's what i call "parameter coupling" ...
the cutoff and envmod params on the 303 are coupled in a non-standard way, but it's very cleverly done, so it's intuitive
there some more tricky things going on with the accent and resonance, but that's beyond the envmod question

the cutoff frequency equation is like:
Fc = (Vcutoff) * exp(Venvmod + Vaccent + TM3)

where Venvmod is a signal that contains the main envelope generator scaled by the envmod knob
note here that the scaling happens after effectively subtracting 31.37% of the "height" of the envelope, so this is like where the negative bias comes from, and the envmod knob itself is wired in such a way that the envelope can never be turned down to 0

the actual envelope is a Decay-segment only, but its decay time is modulated between whatever the Decay knob points to or the shortest setting via the sequencer (accent)

[xhunaudio]

Thank you, so the Accent directly affects the Cutoff frequency too ?

It's commonly stated that the Accent controls exclusively the filter Resonance and the VCA...

What's the TM3 ?

[mystran]

Thank you, so the Accent directly affects the Cutoff frequency too ?

It's commonly stated that the Accent controls exclusively the filter Resonance and the VCA...

Accent has no effect on resonances as such. Rather the resonance knob affects how the accent contribution to the cutoff is shaped. Essentially the higher the resonance setting, the more the accent contribution to the cutoff is smoothed, where as with low resonance it more or less just adds some extra envelope modulation. The way it's implemented also means that the accent modulation of the cutoff with high resonance is stronger for later accents in rapid enought succession, so it's all kinda special really.

[xhunaudio]

Yes, the gimmick circuit too... The 303 is really an uncommon, "black magic" device