Fast analog sw. as VC resistor

Grant Richter grichter at
Thu Jan 9 20:59:41 CET 1997

>A log pot covers 15% of it's resistence in the first 50% of it's travel,
>and 85% in the last 50%.

Uh guys,

This is a great discussion, I've really enjoyed following it. I think there is one
point that I don't recall being mentioned.

Let's just talk about the pulse width modulator feeding the switch. Assume
that this is a simple comparator, one terminal is being fed from a linear pot
set up as a voltage divider between 0-10 volts. The other terminal is being
fed by a sawtooth wave running between 0-10 volts. 

Now if the capacitor generating the sawtooth is being fed by a current source,
the sawtooth ramp will be linear and the pulse width will be a linear function of
the pot position (more or less).

However, if the capacitor generating the sawtooth is fed by a resistor to the
supply, the sawtooth ramp will not be linear. It will be the common "convex" RC charging
curve, which is an exponential function already. The pot position will not linearly
relate to pulse width.

If the capacitor generating the sawtooth is fed from a voltage controlled current source,
and the output of the sawtooth is fed back to the voltage controlled current source, then
any function between "convex" exponential to linear to "concave" exponential can
be realized depending on the polarity and amount of feedback.

This would let you tailor the knob position to response curve without the complexity
of breakpoint generators.

I built a VCF controlled in this way out of National LM339's and LF11331  switches
running at 50 kHz switching frequency. VCF's work very well this way
because the filter poles act as there own smoothing filters.

Just my $0.02

More information about the Synth-diy mailing list