3086 Tremolo (needs some help)

[duck_arse]

I've been caught by the schmitt/integrator osc a couple of times on the breadboard. I see the voltage divider referencing the, in this case, IC1-d (+), and think to add a bypass electro. and blow me, the osc don't work. so from the bypassed divided, I add a resistor - 100k? - to the (+), and away it goes, happy as larry. if larry was an oscillator. that section needs to be able to bash the reference about, or it won't work.

so if I was doing this circuit in my own maddening manner, I'd delete the oscillator, and the IC2. use any dual oppie for what is now IC1-b and IC1-a section - deleting R26 and C13. better - use them as a supply filter to the new oppie, 47R and 47uF, say.

then use a LM358 or TL0xx low-dual for the osc, again with its own supply R/C filter. and lay it out so that the supply arrives on the circuit board in the middle of the two sections, so audio doesn't share any grounds or supply runs with oscillator. then string two resistors between yer [now double filter] isolated osc supply and ground, connect that to what is currently "B" on IC1-d.

IC1-a "B" would go to the same points as "C" bias in the audio section,  which would go to the opamp doing the rail-split w/ bypass resistors at its input. and I don't think it would tick.

[anotherjim]

Can only say I've never known the function generator circuit to fail to run given the build is correct. As the integrator is inverting, no matter what state it powers up in, it ought to start going the opposite way. If it doesn't switch reverse from that, the Schmitt trigger has a problem of either too much positive feedback and/or too little feedback from the integrator. As the circuit is originally for opamps with bipolar supplies where Vref is a real ground - you can't get stiffer Vref than that.

[soggybag]

Thanks again for all of the advice. I've been trying a few of these ideas. Here is what I have observed.

- I swapped C1 10µ for a 220µ added a 100n cap across C1, and added a 100n cap across C2. This had no effect on the ticking.
- Observing the the ticking on the scope its about 100mv (if I'm reading the scope correctly)
- I'm not observing any ticking with the scope at the power supply.

The trimmer R33 seems to dial out the ticking but runs out of range. Measuring the voltage at R12 I get 7.01v to 4.74V. Seems If I could bring the voltage R12 IC3 Pins 5 and 9 down a little lower it might clear up the ticking?

I'm guessing here from observation because I'm not really understanding what this trimmer is doing here.

Measuring the voltage at IC3 Pins 5 and 9 as I adjust the trimmer the range is about 7.9v to 7.1v.

Would it be possible to bring the voltage at pins 5 and 9 lower by adding a resistor to ground from the uncontested pin 1 of the trimmer R33?

[anotherjim]

Extra supply filter caps are really to protect against less than perfect power supplies. The 100nF ceramics in particular help kill switching noise from SMPS types - which were not around when these old circuits were designed.
Tweaking the R33 trimmer may suggest the ticks come via Ref C.
C16 is to slow the snap action of the Schmitt trigger built on IC1-c. The speed of this snap action is the clicking you hear.
You can try making C16 a little bigger starting with 47nF (assuming you have 33nF at the moment).

[ElectricDruid]

Designing the LFO onto the same chip as op-amps that are dealing with your reference power supply voltages looks like a really daft idea to me (IC1/TL064). Stompboxology might have over-engineered some things, but it looks to me like they engineered some problems in first, and then had to go to great lengths to remove them again.

Ultimately, the schmitt trigger/integrator LFO is absolutely famous for ticking. It's not impossible to stop it, but it causes *way* more than its fair share of problems!

[soggybag]

Is it possible that the noise could be caused by the LFO feeding through? I don't understand this VCA circuit. The notes say "adjust R33 for minimum feed through".

With no input I can see the LFO at the output jack. As I turn R33 the wave form shrinks. R33 runs out of range before the LFO is removed. It looks I'm at 0 resistance so I can't add a larger value here.

Is possible that adjusting R12 and R13 or tinkering with R33 could allow for a greater range of adjustment of the "feed through"?

Help me understand what is happening here. R33 is an adjustable resistor with 4.5v coming and R13 has 9v on one side. How do I calculate the voltage at the R12 R13 junction?

[soggybag]

I got some improvement by hanging R13 to 7k5. This seems to give more range to R33. The thumping which I suspect was the LFO feed through the VCA.

I'm not sure why this works because I don't have any idea how this VCA circuit works. If anyone has explanation I'd like to hear it.

There's a little bit of noise now that sounds like a tiny hi-hat playing at the tempo of the tremolo. I suspect this is noise from the input, because it goes away when turn the guitar volume down.

[anotherjim]

Did you read my comment about C16?

[soggybag]

Thanks Jim, I caught your comments about C16. This cap softens the edge of the square wave.

I think the thumping was from the LFO feeding through the VCA section. Last night it seemed to be cleaned up.

When I left off last night there was a little noise (sounded like white noise) beating at the rate of the tremolo. I'm guessing this is coming from the input of the circuit because it goes away when I turn the volume down on my guitar!

I'm using humbuckers, and my guitar isn't usually noises. The whole circuit board, jacks and all is just sitting on my desk out of the box so that could be part the problem.

[PRR]

> I don't have any idea how this VCA circuit works.

Point D wobbles. R8 feed a wobbling current to Q2&Q3. This varies their transconductance. Q4,Q5,Q6 is a pretty sloppy 20k current to voltage converter.

[anotherjim]

When you turn the guitar volume down the volume pot shorts the signal wire to ground. This has the side effect of quelling any radiated noise that was picked up by your circuits input. You may also notice the noise is stronger at some intermediate volume settings and less again at full volume when the signal wire is directly connected to a pickup coil which has a comparatively low impedance to ground.
Anyway, it could point to layout problems - parts or wires carrying the input signal too close to parts/wires carrying LFO signals.
You haven't done something undocumented like add an LFO rate indicator LED have you?

[soggybag]

Thanks for the insight Paul. This one is over my head.

I changed R13 to 7k5 and it seems to have fixed the LFO feeding through the VCA. Should I also change R19 to match?

Thanks again Jim. No I haven't added a rate LED. There could definitely be some issues with my layout. This is the first pass. I'll need to look more closely at the layout for the next version.

[ElectricDruid]

I don't understand this VCA circuit.

As a summary, I'd call this a "discrete OTA". It does what a 3080 does, except built with separate transistors. Unfortunately  (for it/us) that hits the performance significantly since now every transistor is not on the same wafer and not the same temperature and etc etc...there is much more divergence between the parts that there would be on an integrated circuit.

Honestly, I'm not sure why they bothered. When Stompboxology was written, it's not like there was a shortage of OTAs.

[soggybag]

This came from an issue of Stompboxology where the theme was "going discreet" and projects were built mostly around transistors. The four transistors in the VCA/OTA section are all contained in a CA3086 transistor array. They mention the matching in the article.

I have a few boards to experiment with, I wanted to try using some individuals and see how well they work. The 3086 and 3046 seem hard to get these days.

[bluebunny]

FYI, Alfa is producing a 3046.  You can get them through Tom's website, or Erica Synths (and likely elsewhere).

[soggybag]

Thanks for pointing that out.

Looks like the Druid has 3080 and 571 also! They're all about $5, if I could get this to work with 4 2n3904's the cost would be $0.50!

[ElectricDruid]

Yeah, the Alfa AS3080E *is* a bit expensive for a single OTA. On the other hand, it's a really nice chip - the spec is as good as the selected "A" version of the original CA3080, and then better again on several key parameters. And adding the linearisation diodes of the LM13700 on a spare pin is a nice touch - it stays pin-for-pin compatible with the original, but you have an extra feature if you want it.

Doing it all with 2N3904's would be nice, I agree. I'm starting to see the point of that Stompboxology "going discrete" article, even if I still don't see the point of using a transistor array chip instead of an OTA chip! That seems a bit like cheating as far as "going discrete" is concerned!!

[soggybag]

I'll keep those AS3080E's in mind. I still have a few old 3080's in my stash. Having OTA on an 8 pin DIP is really great for stompbox projects.

I have been planning todo a PCB for the John Hollis Frobnicator. I built this a long time ago and it sounded good. The biasing was a little weird with those 3mm LEDs.

The funny thing about Stompboxology is that the project here Tremolo-Matic XXII in that same issue are Tremolo-Matic X and XVIII. So Boscorelli designed at least 22 tremolos between the Stompbox cookbook and the Stompboxology newsletters (there is a Mo'Tremolo issue) all around 2002 and then disappeared. He's never posted here, or on any other forum.

He was up to Distort-o-matic XI, the designs were pretty original, though maybe impractical compared to is really successful. You have to give the guy was prolific! And, enigmatic.

[anotherjim]

One of my most frustrating failures was what I thought would be a surefire tremolo. A dual opamp LFO controlling a vactrol attenuator between 2 opamp buffers. The LFO was in its own chip but I couldn't get rid of the ticking! This was despite the optical control not using any direct electrical connections to the audio path from the LFO. In the end, I realized I'd made the mistake of having all the opamps close together when they should have been as far apart as possible. Layout is all!

I scrapped the board and made a replacement with a phase shift oscillator. This is ideal for tremolo in my opinion. You have a sine wave LFO that has a more natural sweep than the opamp triangle. It has a narrower frequency range compared to the opamp function generator but it happens to cover the needs of a tremolo effect IMHO.

[ElectricDruid]

I like it, Jim. A phase shift oscillator is a good idea. That avoids the ticking caused by the schmitt trigger by designing it out.