Moog f->V

[Martin Czech]

I'm currently taking a deeper look into the Moog patent No. 4 280 387
"Frequency Following Circuit".

It's basically an active  peak detector, quite in a clever way.
The principle is to detect the charging current of a peak detector,
this will only give a pulse for the largest peaks.


I'm simulating only one stage so far, but it works surprisingly well
with a set of harmonics, even if the fundamental is very weak.

If the signal is very noisy or with higher frequency dirt, this will
lead to errors in detecting the right time of the max peak, but the
behaviour is still better then a simple zero crossing detector.

Vocal like waveforms seem to work ok, but this depends on breath noise,
as the circuit is sensitive to noise.

Perhaps better results could be achieved if the bleed resistor of the
detector could be adapted, this is not possible in the original drawing,
cause the bleeder is also the opamp feedback...

Right now I see the result of a passive circuit..., diode, cap, and
bleeder R, like a normal peak detector, another R in the ground path
of the cap to detect the loading current, voltage across this R is
the output.

By adjusting the bleeder one can control the high cutoff, thus noise rejection. Torturing this little circuit with this signal (spice)

vi in k1 sin(0 1 300)   <- fundamental 300 Hz 1V
v2 k1 k2 sin(0 2 600)
v3 k2 k3 sin(0 3 900)
v4 k3 k4 sin(0 4 1190)  <- maltuned harmonic 4V 1190 Hz
v5 k4 0  sin(0 0.5 11111) <-hf noise 11111 Hz

gives quite good performance for this voice like example with noise.
Of course, the phasing of the mistuned harmonic is too much
for the detector, because after a while all the peaks have equal
amplitude for a while. A cascaded structure could perhaps cope with that...

Now I use three cascaded structures like described above.
Of course some linear amplification is needed from stage to stage.
Cascading helps to diminish false detections, but can't really cope
with phasing effects.

So , I guess the Moog patent works well for voice or flute, but may have
serious trouble with strings tones, because of their phasing waveforms.

This is all simulation, I should try it in reality!


m.c.