[sdiy] Square to Tri Conversion

[Jason Nanna]

I've been working on a similar scheme, though I hadn't thought of
incorporating a f-v converter.  The only related design i could find
was this

http://www.theremin.us/Circuit_Library/converter.htm

though I was interested in the communications circuit it references, I
couldn't locate a source.  So instead of tracking frequency to drive
the VCA, it uses averaging to produce an error signal.  And instead of
a leaky integrator it's servoed.

I breadboarded the circuit as-is, and a big problem I encountered was
the way the transistors generate a DC bias that needs to settle every
time the frequency changes. A nice feature of this circuit is the
pulsewidth of the input is reflected in the shape of the output,
becoming saw-ish or ramp-ish!  Thought about a way to use an OTA
(lm13700) as the input stage, and servo that instead so it only has to
adjust for a "fixed" DC offset for a given signal.  Tried simulating
it but can't figure out yet why my design won't work.  Also adding a
derivative component to the gain error might be a good way to go, as
the settling time is a tradeoff for tracking range and it ought to be
minimized.

If you used a SSM2164 as the current source are there any
considerations for using the input as a dc-cancelling summing node?

Are there advantages of using F-V conversion over this technique?
that would be a feedforward technique, right?  So the linearity of the
F-V conversion would impact the amplitude accuracy?


On Tue, Apr 4, 2017 at 12:04 PM, David G Dixon <dixon at mail.ubc.ca> wrote:
> The "sine wave" being generated by that 3RC circuit is not very good, with 4
> or 5% THD at best.  A sine wave from a good sine VCO would be less than 1%,
> and a self-oscillating filter sine can be 0.1%.
>
> Plus, this circuit only works at one frequency.  It would have to be
> redesigned for other frequencies.
>
> I just designed a circuit that works OK.  You simply take the square wave
> through a linear VCA (like a linearized 2164) and into a Miller integrator.
> The integrator must be slightly "leaky" to keep the resulting triangle wave
> centred about 0V -- maybe a 1nF cap and a 1M to 3.3M resistor across the
> feedback -- the bigger the resistor, the straighter the sides of the
> triangle wave, but the slower the response.  Then, the gain of the linear
> VCA must be proportional to the frequency of the square wave.  This can be
> achieved with something like an LM331, or a simple charge pump circuit will
> do (a couple caps, a diode, a transistor, a resistor).  The problem is that
> the "homemade" F2V converter might not have a very good response time, so
> the triangle wave's amplitude will be all over the map if the frequency
> changes suddenly.
>
> I'm still working out some quirks, but I'll report back when I have
> something worth building.
>
>
> ________________________________
> From: Synth-diy [mailto:synth-diy-bounces at synth-diy.org] On Behalf Of Elaine
> Klopke
> Sent: Tuesday, April 04, 2017 3:33 AM
> To: synth-diy at synth-diy.org
> Subject: [sdiy] Square to Sine
>
> Does this actually work?
> http://www.learningaboutelectronics.com/Articles/Square-to-sine-wave-converter-circuit.php
>
> I would assume that being 3 RC networks in series means quite the drop in
> volume.
>
> Also, in a related article they show two networks in series converting the
> square wave to a triangle.
>
> All of the converters being dependent on the frequency of the input wave
> being roughly the same as what the RC networks are tuned to.
>
> Thanks guys!
>
> Sent from my iPhone
>
>
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at synth-diy.org
> http://synth-diy.org/mailman/listinfo/synth-diy
>