AW,2:Re: AW: Another VCF Finished

[Haible_Juergen#Tel2743]

>Actually the clipping is avoided only at frequencies considerably lower
>than the pole frequency. At the pole frequency approx. 30% of the signal
>appears at the input of the OTA and at higher frequencies gradually all
>of the signal appears at the input of the OTA.
>
>You can handle large signals without any attenuating resistors with this
>design, so signal-to-noise ratios can be tens of desibels better than with
>the other designs. Also low frequency distortion is lower with this design.

Very good point. Though it's not "ten's of decibels" IMO, the effect
is remarkable. It's like a built-in preemphase / deemphase system.
Many OTA-based phasers also take advantage from this.

> Wrong! What about the resonating high-pass filter in Korg MS-20 ?

The MS-20 High pass is a completely different story, as it is not a
multiple-Lag-with-overall-feedback design. (Think of it as a common
Sallen&key High pass with the resistors replaced by gain cells
and local feedback)
Anyway, resonance on the OTA-design Gene is speaking of *is*
possible to some degree. If memory serves, the FORMANT filter
(which seems to be very similar to the one which Gene built),
does provide resonance in HP mode. But resonance on these
High pass circuits is indeed much more critical in Hp than in LP,
and exactly due to the reasons which Gene pointed out.
(There is a discussion about this in another EN issue, where they
built a 4-pole HP around a 2040, and found that oscillation occured at
much lower Q's than in LP mode.)
BTW, the FORMANT was actually a design that was *switchable*
between LP and HP. (Speaking of the FORMANT 24dB VCF; not
to be mixed up with the more common FORMANT VCF, which was a
state variable design).
*Maybe* they included some rolloff in the feedback for HP mode
in the FORMANT - just guessing here - have to look it up at home.

> Oscillating occurs, when there is a phase shift of 180 degrees in the
> feedback loop and a gain of at least of one.

Yes. And as Gene pointed out, in LP this is the case only fore one
single frequency, while it is for many frequencies in the HP design.

JH.