[sdiy] tuning oscillators
Magnus Danielson
cfmd at swipnet.se
Sun Mar 11 23:05:09 CET 2001
From: Dave Krooshof <krooshof at xs4all.nl>
Subject: [sdiy] tuning oscillators
Date: Sun, 11 Mar 2001 17:49:09 +0100
> As I mentioned here before, I'm taking a coarse in electronics.
> And naturally my attention is in oscillators and filters. I constantly
> run into very dull schemes though, as those techies seem to want
> very stable oscillators, I guess they are more interested in clocks.
> So all of the schems in my books (about half a metre now) seem to
> use fixed parts for the frequency determing circuits.
>
> They all start of with RC which is kind of tuneable by the R, but not that
> nicely in a synth as far as I managed. The Schmittriggers are OKish, but
> not my favourite soundwise (in my tests).
> And it seems one needs to turn two pots to set the frequency.
> Then they go into LC circuits, that bring two hardly tunable components.
> To make it worse, they then go into christals, and I feel I'm drifting again,
> for a christal is most untunable.
>
> All this would have been OK when I wanted to build an organ with 12 fixed
> oscillators and a fist full deviders to make octaves.
> But as I want constant frequency control over several oscillators, this is
> not at all the direction I need to go.
>
> So I was wondering...
> What are the basic principals to make a voltage controlled oscillator?
There are many principles, but the most popular for VCOs are the
combination an exponential converter with current output and a current
controled oscillator (CCO) giving sawtooth output. These can be
voltagetweaked over very large ranges, I've found my ASM-1 to go as
far as above 20 octaves. Check out the ASM-1 VCO at
http://home.swipnet.se/cfmd/synths/friends/stopp/
which might be an inspiration to you.
The CCO core is essentially an resetable intergrator and a reset
detection curcuit. The current is being pulled from the cap by the
current input and the buffer ensures to hold the current input/cap
point at the virtual ground by raising the other end of the cap. The
reset curcuit compare the integrator output with a reference voltage
(divided out of the positive power, actually not without importance)
and when the output reaches the reference voltage will the reset
curcuit enable a JFET which will short out the integratorcap so that
the ramp is reset. The reset time will introduce a frequency error
which worsen for higher frequencies, by inserting a resistor in series
with the integrator cap one can cancel the frequency error exactly if
the correct resistance is in use. The cost of this correction is that
the waveform is slightly missformed at high frequencies. Reduction of
reset time is thus also a design-issue, so that compensation does not
introduce as much waveform distorsion.
After the CCO core, the sawtooth is postprocessed by waveshapers to
get the waveforms of your choise.
Many, many other principlas may be used, all have their pros and cons.
> What would you use as a frequency dertermining unit, and what
> are the principals to be able to tune it over a several oktave range?
I think I coverd that in my brief commentary above. Please ask more
questions if you so need.
> I'd like to underline that I'm looking for concepts rather then concrete
> schems, as I want to be able to understand it to the point that I can come
> up with schems myself.
Well, pop the questions here, there are plenty of people able to help
you out on that track. I enjoy answer these questions, using my
restricted knowledge.
> And between brackets, I happen to feel 12tones in a tempered scale is
> kind of passe, so I'm not so much interested in 1volt/oktave standards.
Well, as long as you want exponential scale, the 1 V/octave is as
usefull as anything else.
> Ps. I asked Michel Waisvisz how he dealt with this problem, and he told
> me a nice story about a capacitor made of plastic bags and metal plates,
> a thing you could put under your arm like the bag of ulean pipes, and
> sqeezing it altered the capacity. Inspirering, isn't it?
Hmm... smelly caps is not a good sign ;)
Cheers,
Magnus
More information about the Synth-diy
mailing list