[sdiy] CA3080 switch driving voltage

[Jay Schwichtenberg]

I have been playing around with OTA VCOs looking into them for LFOs. I
started with Fig 15 in the LM13700 data sheet. Some other ElectroNote
references are Ian's all ready mentioned article in EN#112 but there is also
some info in AN#19 & AN#20. Here are some observations. Even though some
maybe obvious they are probably worth mentioning for archival purposes or
for some of the newbes.

I've tried several configurations for the Schmitt trigger: LM13700 in the
data sheet, LM393 and op-amp. Main issue with these is the +/- saturation
voltage of the devices is not completely equal. This affects the high
frequency symmetry of the oscillator. The solution is to add some sort of
limiter on the output of the Schmitt trigger which goes back in to the OTA.
You see this in a number of other oscillators based on integrators. This can
be diodes, zeners or a diode bridge/zener combo. You might even go as far as
matching components to get symmetry since I don't know how close diodes and
zeners are. Also don't for get a current limiting resistor between the
Schmitt trigger and the limiter.

At low frequencies the offset of the integrator OTA becomes relative
affecting the tracking. So it's a good idea to add offset nulling to the
integrator OTA.

If you're using the newer OTAs with diode inputs and limiting the output of
the Schmitt trigger use Id. Since the OTA is in a closed system and the
input to the integrator OTA is limited at the output of the trigger you
shouldn't be worried about saturating the inputs of the OTA. That should be
dealt with by providing the resistor divider on the OTA input. Adding Id
will allow higher voltages for better SN or if you keep the voltages down
better linearity.

Mumbling on.
Jay

> -----Original Message-----
> From: owner-synth-diy at dropmix.xs4all.nl
> [mailto:owner-synth-diy at dropmix.xs4all.nl]On Behalf Of jhaible
> Sent: Monday, September 15, 2003 12:24 PM
> To: diy
> Subject: [sdiy] CA3080 switch driving voltage
>
>
> I remember I've seen this somewhere (Electronots?), but
> I don't remember _where_.
>
> If you build a triangle VCO core,
> opamp integrator, 3080 providing the integration current,
> schmitt trigger to switch the 3080's current direction,
>
> what is the optimal voltage and impedance to drive
> the 3080 input, in order to get the best linearity
> of frequency vs. control current ?
>
> At first, you might think maximum overdrive with
> low impedance would be best (like 5V and 0 Ohm)
> to get immediate switching (no RC time constant
> with parasitic C's), but apparently such a configuration
> is "too much" and degrades the linearity of
> Iabc -> Iout operation. (Yes, speaking of switching!
> Maybe base current? Maybe too much common
> mode signal? Don't know ...)
>
> So, where is the optimum? I vaguely remember
> a resistance of a few hundred ohms, but I
> don't remember the voltage at all. Certainly
> more than 100mV, certainly less than 5V.
>
> JH. (Working on an improoved version
>         of the FS-1)
>
>