VCO Exponenetial Converter Issues

[robertlong13]

Hello everybody. I'm a long-time electronics enthusiast, but brand new to modular synthesis. I just bought a massive collection of electronics components at an estate sale, and decided I'd try to breadboard a VCO (the plan is to eventually make my own cheapo eurorack synth using nothing but perfboard, pine, and some 3d printed parts).

I'm building the Yusynth VCO-V2 using mostly parts I had on hand. For now, it's just the sawtooth part, none of the waveshaping yet until I get the core working. To my astonishment, I was able to get it working... kind of. At low frequencies, the waveform starts to sag more and more, instead of being a straight line.

Around 10kHz, it looks good


But down at 800Hz, it's not


The waveform looks like that of a capacitor discharging through a resistor, which led me to believe that, at low currents, my exponential converter isn't acting as a constant current source. To confirm this suspicion, I used the uA measurement on my multimeter, and with the coarse knob adjusted very low, I was drawing about 5uA when connected to +5V, 12uA when connected to +12V, and 0.3uA when connected to 0V. So at that particular knob setting, expo converter is acting more like a 1M resistor than like a constant current source. At first I thought this might be somehow related to my using hand-matched 2n3904s, but then I found some BC547Bs, got two that matched and had the same symptom.

Thinking that the issue is with the very low currents, I changed R13 to 47k to give me something like 100uA reference current on the left transistor instead of 5uA. Then I had to increase C8 to make up for the higher currents (to a whopping 10nF), and change C10 to 100pF to give the capacitor a chance to fully discharge when the comparator triggers. This solved my low frequency issues entirely. They look great now. Unfortunately, and as I expected, my high frequency waves are starting to distort.

I searched around, and it turns out someone else has had the same symptoms I'm having. It's on this forum, under the title "Yusynth VCO problem tracking at low frequencies", but it looks like he never figured out what was going on.

Anyone have any ideas on what might cause this? It's on a breadboard, so there's obviously the chance I screwed something up, but I've checked and triple checked, and probed every suspect I can think of, but seem to have exhausted all my ideas.

Thank you very much (especially to Yves for designing, publishing, and explaining everything so well)

[blue hell]

Are you sure that your scope's input is set to DC (and not to AC)?

[Grumble]

yes, its on DC

---

39596B4D-BF70-4AC8-9564-D4A5F46C4355.jpeg
Description:
Filesize:	58.63 KB
Viewed:	3664 Time(s)

---

[blue hell]

ah .. oaky, DC .. still it looks like AC coupling somewhere .. or a badly trimmed probe maybe.

[robertlong13]

Gah!!! I figured it out on my drive to work this morning and just got a chance to test my new theory. It should have struck me as suspicious that there seemed to be almost exactly a 1MOhm bleeding from my capacitor. Like an idiot, I was probing the capacitor directly, and not the buffer amp output. Guess what my scope's input impedance is...

In all my life I've never actually had to deal with probing anything that had microamp-levels of current in it, and I've been able to live in the happy oblivious assumption that scope and multimeter impedance is just "infinite" for my practical purposes. I've had to be careful of probe capacitance before, but completely forgot about probe resistance.

It's working perfectly now (and technically was before, whenever I wasn't probing)!

Thanks for taking a look. Yes, the trace has a DC offset that's making it look like it's set to AC coupling. It's caused by the fact that this dumpter-reclaimed power supply I'm using isn't isolated from earth (but the ground isn't shorted to earth either though). It's weird, the ground in my circuit is about -0.7V referenced to earth, so I couldn't ground my circuit to my scope (it would cause overloads on my negative rail). I did try plugging channel 2 into ground and plotting the difference between CH1 and CH2, and it really was just a DC offset, so that wasn't screwing anything up, just adding a bit more noise than is really there. I'm building a power supply for my new DIY case tonight, so that'll solve that problem.

Thanks again everyone.

[blue hell]