VCDO / RC4151

[Joachim Verghese]

Michael B. wrote:

> The purpose for the 4151 was to ensure that the sawtooth would reset
> with the sharpest possible edge, and have no high-frequency pitch
> droop.

Not quite -- this is how I see it:

The expo converter sinks current through the integrating cap,
resulting in a rising ramp at the op-amp output. When the ramp
reaches the treshold set by pin 6 of the 4151 (+5V), a one-shot
in the 4151 is triggered. Instead of "hard-resetting" the cap
voltage back to zero, we apply a charge pulse to the integrator
input to bring the output back to zero. This is done by injecting 
a constant current through pin 1 for a constant period of time
(Q = I * t). The current is set by the resistance from pin 2 to
ground, and the on-time is set by the RC network on pin 5 (approx.
20 microseconds in this case).

Now, since the expo converter output is active during the retrace
phase (as well as during the ramp phase), a portion of the charge
pulse from the 4151 is "stolen" by the expo converter. This is why
the VCO is virtually free from HF-mistracking -- the sawtooth doesn't
get reset all the way to zero volts with higher control currents.

The 20 us retrace ramp is not really fast enough for really bright
timbres, so pin 3 of the 4151, being an open collector output, is
used to force the output signal to zero during the retrace phase,
thus giving the waveform a sharp (albeit "superficial") retrace edge.

Also notice how the exponential converter is dimensioned. Normally,
this type of expo converter would be biased for an oscillator
frequency of approximately 1 kHz at zero control voltage, because
this is where the circuit is least sensitive to temperature changes
(equal transistor BE-junction voltages). The Chroma VCO, however,
oscillates at *maximum* frequency with zero control voltage applied,
and as you increase the CV the frequency falls. As far as I can tell,
this scheme is necessary because the unipolar control voltage is
software-scaled at the CV DAC, and not at the 18mV/oct input of the
exponential converter.

> Who knows?  Maybe this one will fill the bill for clocking a stable,
> wide-range VCDO. (after changing some of the cap values to smaller
> ones)

I've actually used this design with success for a VCO-driven
digital tone generator. There are some limitations though -- the
oscillation frequency can't be made much higher than a little
over 100 kHz. This is because the minimum retrace/one-shot time
can't be much shorter than a few microseconds. Linearity is
excellent below this maximum limit, though.

As far as ICs are concerned, National's LM331 is very similar to
the RC4152. The RC/XR4151 doesn't have as good linearity specs,
but costs a fraction of the 4152.

Have a nice weekend, everyone.

-joachim