[sdiy] 4-pole filter concept

[David G. Dixon]

I've been thinking about a different sort of four-pole filter for the last
couple of weeks, and have finally done some simulations to confirm that it
should work pretty well.  I was disappointed that the standard 4-pole
configuration under self-oscillation gives sine waves at each pole which are
-3dB from the previous one, and the first one always hits the rails.  My
goal was to develop my own version of the Mankato filter (the schematic for
which I've never seen, as it's apparently a closely guarded state secret),
but with voltage-controlled resonance.  To this end, I've strung together
four LPF blocks based on LM13700s, making use of the internal buffers.
After much experimentation, I found that the LPF circuit published in the
LM13700 datasheet was basically optimum in terms of giving large bandwidth
and low THD.  Inbetween each block I've put another LM13700 VCA with
automatic gain control (again, more or less as suggested in the datasheet).
Stable oscillation is easily achieved.  And, with the judicious selection of
resistor values, the AGC limits the amplitude of the resulting sine waves
such that they stay more or less constant, even if the resonance (VCA gain)
current is cranked up well beyond the oscillation point, although the THD
does suffer slightly.  In other words, this should give very nice sine waves
at +/-5V (or whatever level is desired).  Finally, unity-gain inverters at
each pole ensure that all eight 45-degree quadrature waves are available, a
la Mankato.

I'm going to breadboard this tomorrow (if I have time) to confirm that it
works like the simulations suggest (circuits virtually always do, I find).
I'm wondering, has anyone else tried this?  Is there a similar circuit out
there that I'm not aware of?  Am I in for major disappointment when I
finally get this thing working?

Dave Dixon