SVF 1V/Oct calibration and resonance

[eljay]

I noticed while calibrating my recently built SVF that I seem to be encountering some sort of threshold issue with the resonance signal being produced. I have set the resonance pot to max as required and got a 1V/Oct input from my mididac. I am using a scope to observe the resonant waveform and measure the frequency. What I have observed is that the resonant waveform appears unstable below about 480hz and slowly collapses so there is no resonant waveform visible. To get the filter to resonate again I have to turn the frequency pot up high and or input a high note to the mididac (say C6 or 7V from my midiDAC). I'm thinking this isn't correct behaviour as I encountered it during the 1V/Oct calibration process and that suggests measuring between 880hz and 440hz to set the trimmer. I have set the trimmer by using 880 to 1760Hz but would like to know if I have a problem I need to resolve.

I have used ceramic capacitors for the 470pF C2 and C7, I did notice they were 5% not the 1% specified, so measured the value of a number of them in the packet and selected two that read 470pf and 471pf on my capacitor tester, are these causing the problem?

I should probably state that this is the 5U version and I am building the full 2U width version.

[Synthbuilder]

The number one reason for the SVF not oscillating at lower frequencies is the value and type of capacitors. It may be worth trying out some different capacitors in there.

Check the resonance switch is set to HIGH too.

The next thing to perhaps check is the offset voltages. This should be done before any calibration of scaling.

It may be also worth trying a different LM13700.

[eljay]

Unfortunately I don't have any 470pf polypropylene caps so will have to order some, The resonance is set to high and I have calibrated the Offset voltages, I have other LM13700s so I'll give that a try as well.

Update : I tried another LM13700, resonant oscillation is now working as expected. Is the previous LM13700 faulty or just a slightly different tolerance that won't affect it's use in an alternative situation?

[eljay]

Further update : Now the module has been powered up for several minutes the problem has re-appeared with the new LM13700, Nothing appears to be getting warm, let alone hot, and yet the behaviour has clearly changed since first power up when oscillation down to 40hz could be seen.

[terjewinther]

General observation: Yes, LM13700 (and other transconductance ICs) can often have different tolerances. Not necessary broken, but fit for another circuit. I have had oscillators not working, but swapping fully working ICs around made them work again. If you can, try out some other ICs.

[eljay]

Ok, just to confuse matters further.

I have another 1U wide variant of the SVF that I built 3-4 years ago, I've never really noticed any problems with it, and I must admit I've never turned up the resonance to max and put it into self-oscillation mode, although I did calibrate it when I built it and didn't notice any issues at the time.

I have just repeated the same calibration tests on the second SVF from 4 years ago, after a short period of time this also exhibits the same behaviour but works for a short period as expected after power up. This version has the original 1% Polypropylene capacitors. Both SVFs use the Issue 5.1 PCB.

So I'm thinking this isn't a capacitor problem as two different implementations one with polypropylene the other with COG capacitors, demonstrate the same symptoms. I find it hard to believe that LM13700s bought several years apart could both demonstrate an identical tolerance problem. So am I seeing things or is there another possibility? I have checked the offset trimmers on both boards and they are correctly set. I have adjusted the Tune trimmer so that both are the same and have a 0V bias going into R30.

The transistors on the earlier board are a BC550 and a BC560, on the later board they are BC549 and BC559, they are held together with a tie-wrap to provide temperature stability.

[Synthbuilder]

Well, this is interesting behaviour.

The lack of resonance is normally caused by differences in the behaviour of the two integrator sections. This warming up thing I have not seen before but it could be that one side of the LM13700 is getting warmer than the other side and changing the transconductance in one half just enough to stop the oscillation at low control currents.



I will investigate and get back to you.

[eljay]

When being used as a filter, rather than when being observed with a scope during calibration, the SVF sounds good. The resonance pot adds resonance as you would expect and seems to track nicely across the keyboard. Perhaps I am just observing a quirk in the calibration process that is not seen when the filter is actually in use. I notice from my notes when I calibrated the original SVF that I alternated between 880hz and 1760hz to calibrate the 1v/oct so presumably the problem didn't manifest itself at the higher frequencies.