JH VCHADSR up and running!

[jorgen.bergfors at idg.se]

>>I now have your (JH's) VC envelope generator fully working on a beautiful etched
>>circuit board. And it has my punch feature too!

>Congrats !

Thanks!

>>Adding the punch feature was no problem at all. It worked correctly on the first try. To get the variable trig >>level, I had to add another tranny. Otherwise the threshold was lowered when the control voltage was >>increased. That wasn't quite to my liking.

>>These are the min and max times that I got:

>>Hold time: 0,5 ms -- 45 s
>>Attack time: 0,4 ms -- 35 s
>>Decay time: 1,8 ms -- circa 60 s
>>Release time: 1,8 ms -- circa 60 s
>>Trig level: 0,5 -- 10,2 volts
>>With punch activated, the attack time is 0,3 microseconds.

>>One slight problem is that the settings are very logarithmic. At "12
o'clock" the A, D and R are only 25 ms. >>So you might get the impression
that nothing happens in the first half of the range. But in reality there
is no >>dead range. It is just that the resolution at the short times is
very high. I suppose this is a good thing, but >maybe the resolution at the
longer times is too low?

>Yes, that's a typical problem of VC envelopes: the CV is truely
>exponential, and not a crude approximation
>like "logarithmic" potentiometers. Only that these log pots have the
>"right" feeling, and not the true expo stuff :-(
>You can increase the range of this envelope into several minutes time
><constants, but the pot range
>problem becomes worse. For emulating a "normal range" (like 2ms to 10s), you >can use a 10k pot and
>a 8k2 resistor from
>the wiper to the cw (+10V) end. For larger range, some diode function
>waveshaping might be helpful.

Well, I think it is okay the way it is. Maybe not ideal, but it is an advantage to have high resolution at the short end of the scale.

>>I omitted all trimmers and instead selected fixed resistors as follows:
>>R16: 68k
>>R58: 56k
>>R59: 56k
>>R56: 56k
>>R50: 56k to + and 47k to -
>>R53: 300k
>>R15: 150k
>>Do you think the trimmers are necessary because of tolerances in the
>>parts? I.E. if I build another one with >these resistors, will the min and
>>max times be about the same?

>The trimmers for minimum time constants would be necessary, IMO. Not those
>for the max range.
>Yes, that's because of the vbe tolerance. You might not need any trimmers
>with two matched pairs.
>(npn pair and pnp pair)

Oh, well. I used matched transistors on my prototype. I have measured Vbe on all my transistors and written it right on them. So I just have to pick some more with the same number from the box, when I build my four VCADSRs.

>>Because of the rather low impedance of the sustain input, the voltage from
>the pot gets loaded down. I >suppose you need to have a trimmer in series
>>with the sustain pot, to get exactly 10V max sustain level. I connected a
>>3k9 resistor to +15V from my 10k pot, but this might not always get the
>>right voltage, as pots >have a rather high  tolerance.

>The whole circuit is *not* intended to work with +15V (and series
>resistors) for the pots. Using
>a buffered +10V supply (one opamp ...) is crucial here. You mentioned the
>problem with the
><sustain voltage already. The impedance is rather high in the sustain
>(decay) phase of the
>envelope. It's only low impedance at attack and release phase. So a 10k
>linear pot to 10V
>will work without much nonlinearity of the sustain CV. But if you want to
>drive several envelopes (in a polyphonic synth), you need a buffer stage. >Well, with a DAC and S/H
>multiplexing of a programmable polysynth you need a buffer anyway ...
>The A/D/R CVs need a 10V supply, too. The unpleasant behaviour of the knobs
>you mentioned
>gets worse with +15V and series resistor. The 8k2 resistor trick won't work
>anymore.
>If the CVs go above 10V, the simple electronic switches will limit the CV,
>too.

Good solution. I'll put in an opamp to get 10 volts for the pots. It's also cheaper than adding a trimmer for the sustain voltage.

>>All in all, I'm really satisfied with this circuit. I'm going to use it
>>for my quad EG, and throw out the modified Doepfer board.

>>Joergen

>I knew my circuit would have *one* good side effect, at least ! (;->)

>JH.