AW: Re: discrete SSM2040-style filter

Haible_Juergen#Tel2743 HJ2743 at denbgm3xm.scnn1.msmgate.m30x.nbg.scn.de
Wed Jan 8 00:18:00 CET 1997


Hi Joachim (and list) !

> Great description of your 2040 "clone"! Seems to be very similar to the
> one I built last year.

Wow, you did the same thing! Must have missed it - last thing I remember
is the stuff with clipping diodes on existing otas ...

> Yeah, this must be the greatest problem with having several discrete
> gain cells in cascade. I tried using EP2015/EN2016 arrays to get rid
> of temperature sensitivity, but the problem is that these devices
> don't boast particularly high gain, so offset and CV rejection becomes
> a problem. I ended up using matched ZTX109C/ZTX212C pairs, and the
> results were somewhat similar to those you got, but I'm still not
> satisfied with the offset performance. Requires some tweaking.

CV rejection is not a problem in my filter. This was the great (positive)
surprise, that the total offset is nearly independend of control current!
It's only temperature gradients that make problems. So I guess I'd just
have to use the same heat-conductive epoxy stuff ARP and Emu used
back then (??)

> Me too!! In fact, I designed a complete 2VCO-VCF-VCA circuit using
> only discrete devices in the signal chain. Well, the final output
> stage uses a (high performance) op-amp, but everything else is discrete
> and DC-coupled.

Very similar to what I intend to have in my PolySynth. The 4pole
path shall be completely discrete. Only the 2pole path would have
3080's and opamps. That is, if I stick to the idea of using the
SEM filter without changes. Would even include a 741 buffer
stage. But I am still not sure about this. Another ideat that haunts me
is using the Wasp design for 2pole responses. It doesn't have
opamps at all, but unbuffered CMOS inverters as integrators and amplifiers.
Unfortunately, component values are very unreadable on my wasp
schematics copy.

> Instead of having a level control for each VCO, I have a single
> VCO A - VCO B mix control, and a separate "Filter Drive" control to
> vary the amount of distrotion. Increasing Filter Drive decreases
> final VCA gain, so that the overall amplitude remains nearly
> constant. In a similar vein, the VCO waveform amplitudes are "averaged"
> so that the VCO level remains constant through all saw-pulse-tri
> combinations.

I used the Mix/Drive concept on my JH-2 synth, but I did not like it very
much for one reason: A single VCO with a certain amplitude always sounds
louder that a mix of two VCOs with half amplitude. So turning the knob
from single VCO to dual VCO gave the impression of *removing* something
instead of adding something. But of course this is just a matter of taste,
similar to the "feeling" of loudness of different waveforms, versus their
amplitude (which will be affected by clipping).

JH.

PS.: I'd love to see the schematics of your all-discrete synth!



More information about the Synth-diy mailing list