[sdiy] linearize OTA for VCF?

[Terry Michaels]

Message text written by "Ian Fritz"
>Terry --

Thanks. With the circuit you have described, I believe that the input and
the output will clip at the same point when Iabc is .5 mA. But the input
always clips when Iin = Id, so if Iabc is less than .5 mA the input will
clip before the output does.

I believe one generally needs to design so that the input structure doesn't
clip. This leaves a bit of a dilemma. Should the input be set to clip when
the nominal system level (+/-5V, +/-10V or whatever) is reached, or at a
higher level to allow some overhead? The most conservative approach (when
designing a filter with feedback, say) would seem to be to design so the
input structure cannot clip until the input signal reaches the rails. This
of course will eat into whatever improvements in distortion and S/N the
linearizing diodes give you.

  Ian<

Hi Ian:

I set the input clip level equal to + and - 12.5 volts, this is the maximum
voltage swing available in other circuits (mixers, etc.) where the signal
passes through an opamp, which will deliver nominally + and - 12.5 volts in
a system with + and - 15 volt supply rails.  A quick bench test of my
CA3280 based VCA reveals the following.  If DC is applied to the control
voltage inputs so the VCA is full on (unity gain, Iabc = 0.5 mA), the
output clips when the input reaches + and - 12.5 volts.  The output swings
+ and - 12.5 volts also.  If I reduce the control voltage to, say 20% of
full on (Iabc = 0.1 mA), it still takes + or - 12.5 volts at the input for
clipping to show up at the output.  The output level is now about + and -
2.5 volts though.  If I apply enough control voltage to drive the VCA to a
gain of X2 (Iabc = 1.0 mA, the VCA has an internal limit at that point),
the output clips when the input reaches + and - 6.25 volts, the output
swings + and - 12.5 volts.  Since I don't usually run the VCA's at a gain
greater tha X1, they don't clip before other modules do.
 
Since my VCO's generate + and - 5 volt waveforms, I basically have X2.5 of
headroom.  This seems to work out well, if I mix the full output of 2
VCO's, l would expect the instantaneous sum of 2 VCO waveforms would not
cause clipping in the VCA's, mixers, etc.

Terry Michaels