Re: [AH] the snappiest envelope ever

From Florian Anwander
Sent Tue, Nov 11th 2008, 12:44

Hi Phil

> So if I am right it would mean that the fastest setting of the A-140 would make the full envelope's duration faster than the sole attack of the Cwejman envelope ?
No. The minimal allover time of the envelope consists of 50uS Attack, 
50uS Decay, 50 uS Release -> 150uS.

> Do you think this is plausible or could this be a mistake on the Doepfer site which should read 0,5 ms instead of 50us for the full envelope duration ?
Yes it is true. I sat together in the "Cafe Freiheit" in Munich with 
Dieter Doepfer, Andreas Merz, and some other well known synth freaks not 
to name here, when we talked about values like this one. I also own the 
schematics and the cap/resistor-values (incl the internal resistance of 
the switching CMOS) go for a value like that. I also modified two of my 
A-140's because I think the fast-value is too fast.

And another comment to this topic in general. I often see, that people 
often do not really mean "envelope speed", when they say "snappy" (or in 
  other relations "punchy").
Analogue envelopes usually are based on loading/unloading an capacitor, 
which is an exponential process. This has the consequence that the 
impression of an voltage-envelopes characterstics is mostly related to 
1.) the modulation destination it serves (exponential or linear) and 2.) 
the modulation depth of this destination.

The problem exponential and linear is quite easy to understand: If I 
modulate an exponential input with the exponential envelope, the 
"characteristics of the overall modulation will be double-exponential, 
which may cause faster/snappier sounds. The rule by thumb for 
VCA-modulation says: for envelope modulation an linear VCA is 
recommended (the envelope already offers the required exponential 
characteristics), for any other modulation source like LFOs, Velocity, 
Audiomodulation a exponential VCA is recommended.

More difficult is the topic modulation amount.
"Exponential" means: most of the action happens at the start of the 
action. E.g. the attack goes very fast to ~80% of the overall range, but 
takes much longer for the remaining ~20%. Now think of an Lowpass-VCF 
with positive envelope modulation. The envelope is a typical Spike, with 
zero attack quite short decay, and zero sustain. It should sound 
something like "psiiiaaaoouuu".
Lets now say the cutoff frequency is at 1000Hz (which is already quite 
bright) and the envelope amout is turned up fully,  so the modulated 
cutofffrequency reaches 20000Hz at the envelopes maximum. Now the fast 
80%-part of the envelope (the "snappy" portion) happens when the cutoff 
goes down from 20000Hz to 2500Hz. Unfortunately you don't hear much 
difference between these two values. It will be between "very bright" 
and "still quite bright". The real audible difference in sound will 
happen between 2500Hz and 1000Hz, but unfortunately this change will be 
done by the much slower 20%-part of the envelope.
The result is: though the envelope is infact fast (= could provide 
snappy sounds), you will recognize it as slow, because of the envelope 
amount and the base-offset of the modulated parameter.