AW: Osc reset time

[Haible Juergen]

	> Do you know of anywhere I might find a schematic of such an expo
convertor ?

Korg Polysix, for example. (It's multiplexed for all 6 voices,
but the basic principle is there.)

	>Do I just use a normal expo convertor with the output current
passing thru a
	>resistor and extract the voltage across the resistor ?

Yes. But I'd keep the Vce of the expo converter constant,
and feed it into a virtual GND node. Opamp with resistor in
feedback path, collector of expo transistor directly connected
to opamp "-" input. 

	>What constitutes a stable linear current source ? Will a transistor
in the
	>feedback path of an op-amp do ?

That's how it's usually done, and how I would start, too.
voltage input is "+" input of opamp, opamp output drives
transistor base, transistor emitter tied to opamp "-" input,
resistor from the same connection to GND to set the
I / V conversion ratio. You can switch in separeate
resistors here to vary footage. 100k, 200k, 400k.

The Korg MS-20 shows a slightly different approach for
the current source (schemos are on the web, too): They
built a standard expo converter (for the linear current source,
that is ! Remember they already have an expo Keyboard 
voltage ...), but only used the "linear FM" input of this expo
converter for the keyboard. Footage range was set by
a set of 100k / 200k / 400k / 800k resistors, too.
(all built from a string of 8 equal 100k resistors, of course)
They only use the expo input for uncritical stuff (in terms of
linearity and tempco), like modulation and tuning.

(side note: The theorethical sonical benefits of linear VCOs
are completely spoiled in the MS-20 by the huge interlock of
the Oscillators. On the Polysix, you can hear it in unison mode,
but alas here it's 6 VCOs and not just 2 at one time. If you
have the chance to play a Trident - that was the instrument
that triggered it all for me.)

Now the general drawback of these linear VCOs was that
you *had* to make the footage switching, because if you
wanted to cover the full range with one expo CV you came
in conflict with offset voltages, and more important, offset
voltage drift.

My assumtion (still untested) is that we would have much less 
trouble today with low offset opamps. I think that it would still
be difficult to cover ultra low LFO range and audio range with
one CV, but audio range down to fast LFO range might work.
(It does in my Frequency Shifter VCO, but then I have only
built one single unit.)

	>> I'm curious if the finite resistance of the switches has an
effect
	>> on the waveform. 
	>
	>Me too. The on resistance for the 4066 is 80 ohms max. Given that
in a
	>normal osc the cap would be connected to the +v supply, how much
extra
	>resistance does the switch present in addition to the output of the
supply
	>regulator ?

I'd expect some glitch on the saw waveform, but the (usually
more critical) triangle wave should be quite smooth by definition.

	> And what sort of impedance does a typical current source have ? 

Depends on how you build it. from 100s of kOhms to 100s of
MegOhms if you build a cascode. But that wouldn't interfere
with the switches' on-resistance, anyway, would it ?

	>Come to think of it, what sort of series resistance does a
capacitor have ?

MilliOhms. Magnitudes lower than the CMOS switches.

JH.