VCO troubleshooting

[hububalli]

Anyone get any advice as to how to troubleshoot this one? All seems to be working ok apart from the inverted peak the wave forms.

Saw


Triangle


Sine

[gabbagabi]

hi,
your Saw is the problem,
the discharge time is to long - and because of that you have the peaks in the tri and sine wave.

so the good news: your waveshapers are working

you "only" need to investigate why the discharge takes so long

[hububalli]

[Marco Piatti]

I'm having the same imperfection in all 3 VCOs I Made, have you figured it out yet?

[JovianPyx]

Dunno where to find the schematic, but whatever is supposed to discharge the integration capacitor has too much resistance. This will also cause CV tracking to be worse as the frequency increases. This can easily be caused by error in picking resistors, such as using a 100K instead of a 10K somewhere (that value is just an example, not actual). It could also be caused by a poor solder connection.

[Marco Piatti]

Thanks Jovian, that could probably explain my tracking probem, I'll check on it

Here's the oscillator schematic
http://yusynth.net/Modular/Commun/VCO/new/VCO-V2.gif

[hububalli]

Any luck with this one? Recently got some free time again so want to try to get this VCO fixed.

If have checked all the parts again and everything seems correct. Not tried re-soldering that area so will give that a go.

[hububalli]

After poking around a bit I noticed this on pin 6 of U7 (yellow).



Earlier in the circuit around R22, there is a massive spike, but at this point I am pushing my abilities a bit as to how to investigate further.

[JovianPyx]

In the VCO schematic, the discharge is implemented by FET Q1. It's resistance when "on" is too high to discharge C1 quickly enough.

There are a few ways to deal with this. One is to supply a gate voltage (at discharge time) that opens Q1 more. Someone with the circuit and scope might be able to help there. Another way is to reduce the size of C1. If C1 is half it's current size, it will discharge twice as fast. However, that will cause the VCO to oscillate an octave higher. That might be mitigated by changing the CV input circuit, but I'm no help there - I'm sure it can be done, but I just don't know how.

Another thing is that there is no such thing as a sawtooth VCO that can reset the integration cap instantly. All one can do is try to make the reset time as short as possible. That said, there is no way to completely eliminate the glitch in the triangle output. What I have read from modular users and designers is that it if you start with a Triangle core and then use a waveshaper to create sawtooth from the triangle, you will have a perfect triangle and sawtooth with no glitches.

I've played with using a saw passed through a waveshaper to make tri and the temptation is to attempt to filter the glitch out. The problem with that is that the filter will do more than remove the glitch - it will also cause the ramps of the triangle wave to become S shaped curves (also called splines). And depending on the filter, it can also cause the triangle output amplitude to drop as it's frequency is increased. Even filtered, the glitch is still there unless the filtering is so heavy that the output becomes a sine.

[gabbagabi]

you are righth that there will be never a perfect triangle out of a saw,
but what we see here is a little bit to bad for a yusynth vco

pin6 of ub7(b) is the virtual ground (inverting input) usually u should not measure anything cos it is (virtual)ground, but i would guess this is not the problem

the spikes of the second scope is the signal that drives the discharge-transistor,
would be interesting to see the full spike --> yellow channel on 10V/div

can you check if c11 is really 47p = only a "47" is printed on not 471 or 472 ...
same with C10 should 10pF = only a "10" should be there

may u are also willing to post a pic?

[hububalli]

Thank you for the explanation JovianPyx, it is interesting and helpful to know these things.

I can confirm C11 and C10 are marked '47' and '10' for sure.

Here is a shot of the spike which was taken from between R22 and D1. Measured about 17.5 V at its highest.

[img]

Here is the board. Please forgive the rather poor workmanship!
R32 & R49 are matched 100k pairs to make up a 50k as I couldn't get hold of any.

[hububalli]

Sorted it!

Was being caused by U6 TL074. I got given a load of them and tried various ones but they all had pretty much the same problem. Some better than others but still giving what you see in the scope shots.

I needed to make an order up to £20 so chucked in a couple of Tl074s and boom, works perfectly now.

There is still a very tiny spike and I can just hear it on the sine wave but I guess that as discussed, that is to be expected?

Thanks for all the help!

[ixtern]