DACs driving VCO(s)

[Paul Schreiber]

The argument on DACs driving VCOs always centers around: how good is good
enough?

If you are driving 1 VCO, then the answer is "just OK". Why? The ear, in
hearing just
1 pitch, was accuracy that "falls off" for higher piched tones. Can you hear
the difference
between 1000 and 100.323232 Hz? Probably not.

BUT.... you CAN hear the difference if you have 3 VCOs in your patch, all
driven from a common
CV, and you tune to a *zero beat* at 110Hz then try to play 880Hz. If 1 VCO
is off about 0.4Hz you
will hear that beating. What's that error?

0.4/880 = .05% error

So, (assuming you have those wonderful MOTM-300 VCOs that can track this
well to start with!), that
corresponds to a voltage tolerance of:

83.3 x .05/100 = about 50uV!

So even if we can live with 10 times this (say a 4 Hz beat) that is a 500uV
error and that what I use
in previous posts. My Kenton Pro2000 averages about 206uV error over 7
octaves.

Of course, there are arguments for "musical sloppiness" but I don't go down
that road much. I prefer
"controlled, added sloppiness" to "built-in sloppiness" in most cases.

As other people have pointed out, in a "synthesizer system" many factors
come into play. Take the
average resistive-string keyboard. Most use 1% resistors. So assuming a
perfect current source
(meaning no drift with temperature) you can be off quite a bit. Moog
switched to SIP networks, because
in general the resistors within the network will be within 0.3% of *each
other* (but not network-to-network!!)

A simple upgrade is to buy 0.1% resistors and swap them out (they are about
25 cents each). But I've
never seen a synth with a temperature-compensated current source!

I guess the point is: when you design a *system*, think about 'taking
control' of each section. With all these
modern parts available, the limitations of older designs are laughable. Why
waste time and effort with some
complicated gain/range switching around a 8-bit DAC for a MIDI-CV converter
when shoving a $10, 14-bit Maxim part
in there solves every issue?

Paul Schreiber
www.synthtech.com/motm  <<see the MOTM-300 VCO specs here