[sdiy] Korg MS-20 VCO gives up below 3.5 kHz

lanterma at ece.gatech.edu lanterma at ece.gatech.edu
Sat Apr 30 22:16:55 CEST 2011

Howdy SDIYers,

My students got the Korg MS-20 VCO sawtooth core oscillating on the breadboard. It happily oscillates up to 20 kHz and more I think, with an 850 mV peak-to-peak swing . The trouble is at the lower end; as you decrease the current below 3.5-ish kHz, it just stops. If you increase the current to the oscillator, it happily restarts once it hits that 3.5-ish kHz spot.

We experimented with different capacitors - by increasing the 6200 pF integrator cap (actually they had a 6800 pf, closest they could find when they were building) to something around *ten times* that, they got the VCO to operate down to around 300 kHz. But that's a crazy high integrator cap value!

They took some DC voltage measurements with the multimeter at various points:


The measurements are in red. These measurements were taken just above the frequency at which the oscillator "gives up." Q2 and Q4 look like their base-emitter junctions are "on," with a 0.7 volt drop across Q4, and 0.8 volts drop across Q2. There's a 0.7 volt drop across D4. Q3 only measures 0.1 volt across its B-E junction, but I think it's responsible for resetting the oscillator, so I wouldn't expect it to look too "on" in a  DC measurement (I think). 

R16 gets toasty (but not too toasty), as one might expect with 12 volts dropped across such a small resistor. It's sort of at the edge of a 1/4 watt  rating.

At this 3.5+plus_a_bit kHz setting, there's 300 mV dropping across R14, which corresponds to 0.03 milliamps (aka 30 microamps).

So, freq = current / (C*Vpp) = 0.03E-3 / (6800E-12 * 850E-3) = 5.2 kHz, so rough calculations indicate things are in the ballpark.

If you change the voltage going into the expo converter to change the current coming out of it (well, I guess technically flowing into it since it's a current sink), you can get a smooth transition from 0 mV to 300 mV at that low end.

The oscillator just kind of wants to stop bothering down there. 

I should mention that in their experimentation they swapped the CA3083 for some regular 2N3904; didn't seem to make any difference. 


As always, any advice is greatly appreciated!

Juergen Haible suggested that the original transistors might be important; I can try getting some on Monday when Ack Electronics is open. But if someone spots something easier/faster to try, that would be fantastic...

- Aaron

More information about the Synth-diy mailing list