I'm constructing a synth using various vco's and touch switches instead of a keyboard and it works quite well. For each musical note, I use a switch and potentiometer to decide the voltage for the V.C.O. I then tune the "instrument" using a guitar tuner.

The problem is I get a slight drift in the tuning (up to half a semi-tone).

The power supply is home-made +/- 12v regulated, on the multimeter it reads 12.00 or 11.99, there's an occasional .01v variation.

I'm using various V.C.O's including xr2206, 4046 and good old 555's and they all drift slightly (increasing the frequency). I have a separate polyphonic octave using 40106's and they also drift in the same direction (increasing the frequency)

Do I need to introduce some component that's going to drift in the opposite direction to equalize the drift?, or should I use secondary regulators or zeners?

I give the contraption 10 minute to warm up before tuning, I don't mind that, but I would like it to be stable every time after a short warm-up.

Any advice - most welcome.

 

Hi Spence,

What kind of capactor are you using as timing element? You need to use ones that are stable with temperature. i.e. NPO or COG type. Use Google.

Regards,
Ifixit

 

You could use 12 xtal oscillators for the 12 notes, and use a binary divider IC on each oscillator that will not only reduce the osc Hz down to the audio range but also give you all the octaves you need.

ie if you inject a high semitone into a binary divider IC each pin of the IC makes the same semitone but at a different (lower) octave.

 

Hi Spence,

What kind of capactor are you using as timing element? You need to use ones that are stable with temperature. i.e. NPO or COG type. Use Google.

Regards,
Ifixit

That'll be the problem, I'm using ceramic. Unfortunately it seems I haven't done enough research before starting, still, I can replace the caps without much trouble.

 

You could use 12 xtal oscillators for the 12 notes, and use a binary divider IC on each oscillator that will not only reduce the osc Hz down to the audio range but also give you all the octaves you need.

ie if you inject a high semitone into a binary divider IC each pin of the IC makes the same semitone but at a different (lower) octave.

Good idea, I wanted to make something different, using a variety of different oscillators to get some 'old fashioned' synth sounds, without having hundreds of pots to twiddle. (A synth with pre-programmed sounds, if you like) I had planned to have a divide by 2 and 5 counter to create sub harmonics, but I hadn't considered using xtals.

 

In the 1970's, most electronic organs, and some synthesisers, would use these : http://synthdiy.com/files/2001/mk50240.pdf which, when use with standard CMOS dividers, would generate nice stable tones.
Not available any more of course, although I have seen some ingenious ways of copying them, like the circuit in the 'tos' attachment below. You would need 12 of them though!!

If by 'old fashioned synth sounds' you mean things like this: http://www.youtube.com/watch?v=HBcqQpv6C70&feature=relmfu
then I'm afraid the 'divider' type generator approach won't do very well.

These old analogue synths used genuine 'free running' VCOS, alongside voltage controlled filters (VCF) and voltage controlled envelope generators (VCA) to get their famous 'fat' sound, and it's not easy to fake it any other way!

But as you've already discoverd, keeping VCOs in tune can be a major headache!

My 1980's Moog Prodigy uses a clever way of keeping the oscillators stable. Although most analogue VCOs exploit the exponential transfer characteristics of transistors, to simplify the external control circuitry, transistors used in this mode are incredibly temperature sensitive.

So, in addition to using the suggestion you made earlier about 'opposite' compensation, Moog used a transistor array, which they used differentially, and in addition deliberately heated the chip, by drawing current through an unused transistor, to keep it well above 'ambient' all the time.
Worked quite well, but, I'm sorry to say, needed a lot of discrete components to work well!

Take a look at a typical Moog oscillator in my second attachment. It's not the actual Prodigy oscillator, but similar in concept.

I'm guessing you're not intending to go in that direction, at this stage??