New dual VCA module troubleshooting

Hello hello everyone and sorry for being so late to get back to you all - I hit a wall after being in the studio working on several electronic, electromechanical, and programming problems and basically burned out.

Took a bit of time to ride my bike, plant some plants, make some music, cover a LMNC song “Ride” with my orchestrion (“Musica Machina” - chucked it up on our Instagram - it was a hoot!), restore an antique wooden table.

I’m back and recharged and back on projects so I’ll return to this in the next day or so after I’ve had a fresh look at it and followed advice you’ve all kindly given here.

Today: the wild ride that has been the clock multi-divide/phase shift with dynamic updates for Arduino

1 Like

So here’s where I’m at:

First, my scope had a faulty probe so that was throwing some suuuuuper helpful spanners in the mix. I’ve now replaced that and getting proper readings.

Second, I still see a higher offset voltage when “closed” than when open. Using the trimmer on the long tailed pair of transistors I don’t seem to be able to change this at all. Whenever the VCA is trimmed to be closed, even just closed, it will jump up in voltage.

Third, you can see when the gate opens we thump down to a lower bias as we output the waveform. This is the pop that I hear when opening and closing the VCA.

I’ve matched the transistors that are paired on each channel using a DMM that has a transistor Hfe tester on it to get them quite close (within 10 of each other), and despite that, shouldn’t this be different on each of the two channels? It appears to be about the same on each.

I did try with another TL074 I bought at a different time in a different batch, but I’ll check that again in case something is going on.

I’m sort of hitting a wall here in terms of what to prod to test - as I’ve mentioned I’ve built this same circuit pretty much as a single channel with a TL072 before, so the basics should be pretty rock solid.

Here’s the waveform (ignore the glitch to the right - that’s what my scope does when you go long) as the gate closes on it:

1 Like

So having a closer look at the original Yusynth schematic here, and ignoring that I’ve built this one before, I’ve had a look at the conditions on the long tail pair that feed the op-amp.

As you can see in the diagram below only showing the long tail pair, you can see that if the trimmer is dead centre at the top (the transistors are a perfect match in a perfect world) there’s a misbalance in resistance acting on them to the value of 21kohms leaning to less resistance towards +12V…

What’s going on here?

buuuuutt… also we’re slightly charging the base of the transistor that feeds into the mix from the CV area so that we’re pulling down slightly to counter that. So what might cause this offset I see when the amp is “closed”?

I hope nobody minds me speaking this out so that it may help others in future:

I can see a difference between my original build single channel VCA and this dual channel: the transistors.

Though they’re still BC547, the label on the ones I’ve used in this dual channel build allude to be “C” but they’re strange poorly painted ones I’ve picked up somewhere over the years.

On the original build they’re neat fresh looking BC547 , but with no denominator, meaning they’re ungraded. They’re matched, but ungraded.

A quick test on the hFE tester on my multimeter shows me the poorly painted ones are an amplification of just over 400 or so, while the neat ones I originally used are over 800 which means the ones I have that are causing me grief are actually BC547B, while the working originals are BC547C.

Being that it’s roughly double amplification, that’s a huge difference, so let’s have a look at what happens when I match some up, and replace them on this board.

well well well - seems Dr Stupid paid my brain a visit in these last couple of weeks. Lesson learned - try to focus on fewer projects simultaneously…

My other VCA that I’d built on this design that I kept referring to as working properly was actually seemingly a different beast to the one I was recalling. At the time it was peak COVID and supplies (and suppliers) were difficult to source here in Australia. I wasn’t able to source the TL072 without epic wait times (seriously a part for my Amiga 500 took over 3 months to arrive from Germany) so I went with what seemed like a fairly close match with LM358s in building Eddy Bergman’s design. It tuned nicely and performs well.

Being that I tried to pull from original Yusynth designs that this was based on which used a TL074 I decided to go back to that.

This is why things seemed to go so differently with what was basically the same design but multiplied by 2: I was using a different op-amp at the centre. That’s not to say Eddy’s design doesn’t work, but maybe it’s far more fiddly or performs differently than I expected.

Right now I just dropped an LM324 in the place of the TL074, and bingo we’re off to the races and tuning seemed to go super easily getting the initial level to just clip that output to zero volts.

So. What have I learned? I’m not really sure. It’s frustrating. It should really work, but doesn’t.

Interestingly the suggestion by @analogoutput to try different TL074’s definitely yielded different results. Actually alarmingly different results each time. I had 3 different batch sets here and even inside the same batch things were different. These were from a trusted supplier, so they’re not eBay stuff or too mismatched. Perhaps there’s allowed to be a large range of variance in TL074s.

But yes… It is now working.

I’ll update the stripboard layout in case anybody would like to use it and upload it here. I’ll bash it into a PCB design now too. (and add proper buffering for those input LEDs when I put them back in play)

Phew! I can now put it through its paces and actually play some music now thankfully.

Also happy to take any further suggestions or comments - it’s always important to never stop learning