Japanese and Korg35

[jhaible]

> You are probably right. I can't seem to get good hi-pass response out of
it
> the way it is. No problem with lo-pass though.

Have you tried a larger resistor ? 2.2 or even 10 meg ?

> Did you leave the diode
> pair as it was? If so is the distortion a quality of the ms35? 

The diodes are crucial to provide soft clipping at high Q and
self oscillation. (The same should apply for the MS-50 filter.)
I doubt that distortion was considered a quality in any of these early
filter designs. The SNR is terrible for sure, but with raw VCO signals
you can go into high distortion region without problems, and so
you get adecent SNR, because you don't "count" the distortion 
to the total noise here.
IMO building synth modules that sound good is not that much about
using components that provide high SNR on paper. It's all about 
having controlled behaviour at the upper end of the dynamic range.
Soft clipping, no slew limiting, no opamp outputs getting stuck at the
supply rails etc.

> Am I missing
> something by trying to make it sound like the ota version? 

I don't know (It's high time for me to build a OTA version !).
Many people claim they hear no difference between old and new MS-20's,
but I wouldn't buy that. But the most prominent effect, the filtered input
signal and the sine of the self oscillation interacting with each other
on the nonlinearity of the limiting diodes, should be similar on both
versions.

> Nah .. the ms50 clone I built has the "thump" syndrome and it's tricky to
> set it up too.

I can't really comment on that, because I haven't built a diode ring filter
yet.
But at least a diode ring is a symmetrical circuit (other than the 2
transistor
filter), and it would have zero feedthru with perfectly matched diodes.
Sure, every imperfection would cause thumping, too. But the transistor
version doesn't even try to be symmetrical from the start.

I was thinking about a symmetrical version of the transistor filter, with
two pnp's in parallel to the npn's, and a npn emitter follower to 
control the pnp bases. Then the two emitter followers would have to be
driven with inverted CV's. Haven't thought that to it's end, not simulated
or tried. Even if it worked, there would be a lot of matching required -
probably not worth the effort.

> If you were only going to build one sk type filter the
> 13600/700 ota version would probably be the best choice it's a lot easier
> to deal with 

Yes, that's how I'd see it as well.

> BTW the ms50 diode ring must be matched or it doesn't work well at all.
> Thats why I used the 3046. It probably should have the limiting diodes
too.
> When it breaks into oscillation it just kills me the way it is. I'm glad
I
> built it but I like the ms20 much better...... maybe most of all.

I highly suggest you add these limiting diodes (see above). Might make
it less dangerous, but more pleasant. I'd be interested how it compares
to the OTA version when the self oscillation is tamed, too.

JH.