AW: Japanese and Korg35

[jhaible]

Hi,

don't be surprised, I'm back again, sadly I had to cancel my voyage.
But making the best of it, I could spend a lot of time with the soldering
iron.
So meanwhile I got a complete dual VCF module that can fairly pass
as a MS-20 filter clone. I think it is quite close, possibly with one
restriction which I am going to explain below.

> I listened to the CD "Synthesizer von Gestern" "Synths of the past"
> again, yes one can easily find out which track is MS20 and PSxxx.
> These guys really sound different!

Yes indeed ! 
Matthias Becker used to describe it (like many other filters) as "surely
not as good as a Moog filter" (my words - not an exact quote), which
probably was the general feeling around the time when he wrote his book,
and which influenced myself amongst many other "second generation"
synth enthusiasts.
If our latest asumptions are true, this he would even be right from a
circuit designer's point of view. From a musical point of view, the
Korg sound has a right on its own side to side with other classics,
meanwhile.

> So a clone of the filter and VCA would be nice.
> Unfortunately there is this Korg35 problem.
> Nowone seems to really know what this is...
[...]
> Any Korg employees there ?

We cannot know, but I think we came pretty close last week.
I'd love to hear from 1st hand sources, too, of course !!

I have made some tests with the circuit I distributed last week, and a
filter
built around two real Korg-35 ICs, "almost" side by side. Almost, that is
I didn't use the same signal and modulation sources. The new circuit was
hooked up to saw waves from the Synthi clone, and a Saw LFO for
modulation, while the -35 based module was connected inside the JH-3
modular.
The sound and general behaviour is very similar.
One point of scepticism was the high CV feedthru of the simple transistor
circuit. Think of it, the whole CV is more or less added to the signal
to get the desited modulation. At least with large CV's. (see below)
BTW, medidtating about the working principle, I thought that the
transistors
in the signal path might as well conduct in reverse direction (emitter
becomes collector, and the other way round - a technique often used
for electronic switches - what do you think: does this make sense here,
too ?)
Back to the comparison (sort of). As long as the modulation strength 
keeps the cutoff frequency roughly inside the audio range, the "thump"
from the CV feedthru is similar on the -35 original, too. This might
indicate that the -35 IC really doesn't have a symmetrical circuit (like
the moog ladder etc.), but rather has a simple structure as shown last
week.
One should keep in mind that the MS-20 does not really have sharp
SAW waves for modulation. It's a highly asymmetrical triangle, but still
has a relatively slow slew rate. I don't know how fast the encelopes are.
Slow transisents and AC coupling of the audio path will mask the CV
feedthru. (BTW, some Korg synths that also use the simple VCA have a
capacitor directly at the transistor base, probably to slow down fast 
transisents.)

This whole thumping only becomes unpleasant when the the cutoff frequency
is modulated way below, or way above the audio range. On the bottom end,
a resistor across the capacitor at the opamp input is crucial. But Tom,
your
100k across 1nF would kill the bass range wouldn't it ? 10Meg seem about
right,
IMO. On the high end, if you modulate above the audio range, you see a
heavily
increased thumping on the scope as well. This leads me to the conclusion,
that
Korg might (!) have limited the voltage range on the transistor bases. If I
got that
right, the CV input of the -35 IC is not a few-millivolt-summing node like
my circuit,
but works with higher Voltage swing. Maybe they included a resistor divider
inside the IC, and made the external CV summing opamp do the limiting job
(+/- 15V rails). 
Remember the argument of having the MS-20 envelopes centered around the 
sustain voltage, and not 0V being the off-value ? I vaguely remember there
was
an argument in a Korg manual that this makes better use of the whole cutoff
range
of the VCF. Now try to see this in the light of the benefits of CV limiting
on this type
of filter in general !

If these speculations should turn out to be true, a real Korg-35 clone
should have
some CV limiting.
But by accident I found that the "unlimited" version has benefits when it
comes
to audio rate VCF modulation. One saw VCO into the signal input, and a
second saw
VCO into the CV input gives to usual audio rate FM effects known from the
Minimoog and the Prophet 5. With relatively low modulation strength, that
is !
But with increased modulation depth, the simple 3-transistor-filter is
capable of
beatiful FM type sounds that go far beyond the general Minimoog stuff.
Beautiful
is the only word to describe it. That's why I probably leave the circuit in
it's
"unlimited" version.

As I wrota above, meanwhile I have a complete dual VCF on veroboard, each
filter
switchable to LP and HP (depending on which input is chosen), and even some
notch filter type when both inputs are connected. (Have to check the actual
frequency response with Spice, but at least it sounds like a Notch)

The circuits is very similar to the one I sent out last week. I found that
it was
quite susceptible to hum because of the high impedance, so I changed the
whole impedance scheme, using 10n and 33n capacitors, a 1 Meg resistor
across the opamp input, and a 20k emitter resistor in the CV buffer.
If you look at last weeks circuit, you can easíly figure this out, so I
won't
make a new drawing.

As I said, I don't know how close I am to the real Korg-35, but I am really
fond of this filter.

Hope you found this interesting, despite the high speculation amount.

I still have one big question. When Korg already had Diode Ring filters
(in the MS-50 for sure, and probably in some of the Traveller filters
as well), why would the go back to an asymmetrical filter for the
custom chip they were going to use in so many products ??
The diode ring should have much less CV feedthru even with
unselected diodes, wouldn't it ?

JH.