MS50 LP-Filter

[Martin Czech]

Someone offered a Korg-MS50 filter schematic some mails ago.  I printed
that out, but what is this crazy circuit with this diode ring? How do
people get such crazy ideas? Maybe this is the smae as the "ringmodulator
ring-idea".  I couldn't understand this circuit at all so I sat down and
figured it out.

Now I would like to hear if I'm right or wrong.

Assuming that you have the schemo in front of you.

Amazing, they did everything the other way arround! Just to avoide
any similarity to a normal cacade? Affraid of infringing Moogs patent?

The filter control voltage is differential amplified into a positive
and negative voltage (IC 7 a and b, there's a mistake : two times a
?).  These are fed into two diodes of the CA3019 diode array in order
to convert the voltage into an exponential current (pins 4 and 1). This
is the bias current for the diode ring. Note that the control current
is differential in contrast to the normal ladder aproach.  The signal
comes from amp ic6b via coupling capacitor into pin 6. The diodes
between pin 6 and 5 or 6 and 8 serve as current controlled resistors
(see my last mail "diode ladders"). But unlike a normal cascade, the
first stage capacitor is not differential mounted, there are two
capacitors c11 and c12 to ground (almost, there's a 100 Ohm to
ground).  This is necessary because the audio signal is not
differential, again in contrast to the normal ladder. So the signal at
pin 5 and 8 is lowpass filtered in the first stage.  The second stage
is comprised by diodes pin 5 to 2 and pin 8 to 8 and cap c13. The
signal is amplified by 221.8/1.8 (amp IC4f), the amplitude is limited
to four times 0.6V by the diode feedback network D5-D12.  So far this
is a two stage filter, 12dB/oct. The feedback or resonance is
controlled via pot vr10 (bypassed by R50 to get some comfortable
nonlinear characteristic). The feedback is fed via R48 and C15 into the
first stage. This is why C11 and C12 don't go to ground but to R46 with
100Ohm.  C15 seems to provide more feedback for higher frequencys,
maybe to overcome some bandwith limitations of IC4f?

I've done some spice simulation which shows that all this explanation
makes sense.  But : I don't understand the substraction circuit arround
IC6b. It seems always to be the same signal as "Peak", but 6dB larger.
Is this substraction to limit the amplitude for high resonance in order
to get a more "even" signal under all resonance settings? Depending on
the setting of VR10 the circuit sometimes gives a phase reversed
Peak-Signal, sometimes not.

Any suggestions? 

However, a very different circuit. I wonder if it also sounds so
different.  CA3019 is no longer, the other Harris diode arrays seem to
be not suitable, but it should be possible to select 6 matching diodes
out of my box of hundreds of 1N4148.

If the winter comes ....



m.c.