Son of Screamer cuts-out when room light is switched

[highwater]

I have been playing with a Son of Screamer on breadboard for a month or two, and finally got some LEDs to try as clippers, and decided to turn the (incandescent) lights off in the room to see if they light-up at all... and the volume promptly takes a nose-dive. I have two SPDTs, one at the input, the other at the output, to switch between the SoS and an Orange Squeezer that is on the same breadboard, sharing the jacks. If I flip the output (with or without switching the inputs as well) to the OS and back, the SoS works again. The OS has no such issue, and they both have TL072 opamps in them. The output of the tone section of the opamp (before the output cap) sits at 3.9v when it's working, and drops to 3.0v when I switch the lights off.

It's powered by a fresh (~9.1v) battery, and running through a JFET booster (on perf, powered by it's very own battery) into a solid-state Crate amp.

My apartment's electrical wiring is dodgy to say the least - the labels on the breaker panel have no relation to the circuits they control, there's bad grounds on a couple sockets, polarity reversed on one... needless to say I won't plug anything but a cellphone charger into any of those.

I'm certain that the root cause of the problem is with the house electrics, but I'm at a loss to figure-out how the heck this is knocking-out my "pedal", being as it's run off a battery and there is no audible pop when the light is switched. Is it likely to be something stupid like a failing cap? Or is this actually as weird a problem as it seems to me to be?

[amptramp]

The LED's are also going to act as photodiodes.  If you turn the light off, the volume should go up, not down.  If switching the output without switching the input causes the SoS to work again, the switch may be the culprit.

Also, what are you powering the breadboard with?  There may be an AC variation if your wiring is as bad as you say.

[antonis]

Just cover the LEDs tight (with a kind of tape or so..) and try with the lights on...

[GibsonGM]

....or cover the board with something fairly light-tight.

But, my vote is on something on your end, as there is no electrical connection to your mains without a wall wart or something...

[Shoeman]

Ha!  Reminds me of the old RD350 bike days..installed a Piranha ignition that used an LED and photo-trans.   Ran fine in the garage once timed up.  Pushed it outside for an afternoon test run, made it 100 ft, turned a corner and it quit.  Kick kick kick, no go.  What the heck.  Pushed it back, and it fired.  I was kneeling by the ignition while the rider started it.  Step back, it dies.  Finally figured out that the sunlight was fooling the ignition with the points cover removed.  Put on the cover and it ran fine everafter.

[antonis]

Put on the cover and it ran fine everafter.

As fine as could run such a "crash hunter"...

( Off topic - but I couldn't resist to... )

[highwater]

It doesn't make any noticeable difference whether the lights are actually on or not - once it cuts out, toggling the output off and then back on fixes it regardless of whether the lights are actually on. Just to make sure, though, I pulled the LEDs and it still did the same thing.

Here's where it gets even weirder, though - yesterday, turning the lights *off* made the volume drop, and turning the lights back on had no effect (in both cases, when done with the circuit working correctly). Today, it's turning the lights *on* that kills it, and turning them off has no effect.

I'm using the circuit at the bottom of RG's Technology of the Tube Screamer, with the unlabeled resistor in the power supply being 100ohms. The Orange Squeezer is using the same battery, but has it's own separate vbias. I have the 4k7 resistor/0.05 cap in the first stage's feedback switched relative to that schematic (cap->resistor->vbias), and a few of the caps are the values from Runoffgroove's Tube Reamer. The vref bypass cap is 220uF (the largest cap I have), and removing it entirely has no effect. All polarized caps are oriented correctly.

DC Voltages when the circuit is operating correctly are as follows:
Battery: 9.45v
Opamp:
1: 4.15v
2: 4.15v
3: 3.99v
4: 0.001v
5: 4.25v
6: 3.36v
7: 3.25v
8: 8.25v (the drop relative to battery voltage is entirely across the unlabeled series resistor)

Opamp when symptom is present:
1: 4.13v
2: 4.18v
3: 3.97v
4: 0.001v
5: 4.10v
6: 3.35v
7: 3.25v
8: 8.24v

Both sets of readings were done with the tone control pot turned fully toward the pin 5 end. Interestingly. turning the tone pot fully towards the pin-6 end *temporarily* eliminates the volume loss, but the loss returns when the tone pot is rotated away. Could this suggest high-frequency oscillation?

Edit to add: as mentioned above, I'm using a TL072 opamp.

[highwater]

Also, I was under the impression that it is supposed to be mandatory that all opamp circuits have a DC path to ground on the non-inverting input, but this schematic (and the circuit as built) has every ground path for pin 5 blocked with caps. Is this an error on the schematic, or are there exceptions to that rule? If that *is* an error, should I just add, say, a 1meg resistor from pin 5 to ground?

[slacker]

Ground from the opamps point of view is the 4.5 volts reference voltage. Pin 1 of the opamp sits at 4.5 volts DC and connects to pin 5 through the 1k resistor, so that the DC path to ground.

[Keppy]

Your voltages are suspicious. Your opamp +/- inputs should match exactly, and if they don't then the outputs should be all the way at the rails. If your readings are correct, then either your opamp is bad or your meter's impedance is affecting your voltage readings. Pins 2 & 3 should match, as should 5 & 6. In fact, in this circuit 1-3 and 5-7 should ALL match since the only DC path to ground or Vref is at pin 3. Your voltages are not possible with accurate metering and a functional opamp.

Try a different opamp or a different meter. While you're at it, meter both sides of your caps, to make sure DC isn't leaking in or out, and check continuity from circuit ground to jack ground to amplifier chassis.

As far as the tone pot, it should be quieter turned to the pin 5 end as a result of reduced treble. Moving the wiper to pin 6 boosts treble. If it's affecting more than treble, you might have an incorrect component value somewhere in the tone section.

[anchovie]

Here's where it gets even weirder, though - yesterday, turning the lights *off* made the volume drop, and turning the lights back on had no effect (in both cases, when done with the circuit working correctly). Today, it's turning the lights *on* that kills it, and turning them off has no effect.

Do you have to get up and sit down again to operate the light switch? Might be that something is loose and the act of you moving gives it a gentle nudge.

[PRR]

> opamp +/- inputs should match exactly

True; but as you say " your meter's impedance is affecting your voltage readings". This happens a lot in our circuits.

If the input is biased with say a 1Meg resistor, and you poke it with a typical 10meg volt-meter, it will read 10% down *while poked*. (OK, 9.09% down.)

If the other input gets bias from a 10K resistor it reads only 0.1% down.

So if you have 4.5V reference, find 4.5V on one input, and 4.0V on the other input (through a 1Meg resistor), this is *probably* correct.

If you have two meters, watch the +In and the Out. The output may be at 4.5V until you poke the +In, then it drops to 4.0V (same as the +In), and comes back to 4.5V when you stop poking +In. This confirms it is just meter-poke loading, not the actual operating condition.

Traditional electronic meters are 10Meg or 11Meg input loading. However many general-purpose meters are lower: 3Meg and 1Meg are common. (This reduces "phantom voltage" on wires which are truly un-connected but run near live wires.) And the $10 needle-meters are often 10K on the 10V range (which makes them horrible for electronics poking).

[highwater]

First-off, I just tried a JRC4558D, and could no longer reproduce the problem. An NE5532 was also issue-free. All five of my TL072s have the same issues, however, so it's probably something related to their design rather than a faulty chip.

Also, with the 4558, the pin voltages measure much closer, so the differences between them are quite likely to be due to meter loading. I only have the one meter though (a Bluepoint DSMC683), and I can't find impedance specs for it, so I can't say for sure. Am I correct in assuming that opamps (and, for that matter, circuits in general) with higher input impedance will be more susceptible to meter-loading?

Do you have to get up and sit down again to operate the light switch? Might be that something is loose and the act of you moving gives it a gentle nudge.

No, I can reach the switch sitting at my "bench", and although I can kill the sound completely in several ways by poking at the breadboard, I haven't managed to cause this particular problem that way. Oh, how I wish that were the reason, though.

Ground from the opamps point of view is the 4.5 volts reference voltage. Pin 1 of the opamp sits at 4.5 volts DC and connects to pin 5 through the 1k resistor, so that the DC path to ground.

D'oh! I tend to forget that opamps aren't limited to AC.

Ha!  Reminds me of the old RD350 bike days..installed a Piranha ignition that used an LED and photo-trans.   Ran fine in the garage once timed up.  Pushed it outside for an afternoon test run, made it 100 ft, turned a corner and it quit.  Kick kick kick, no go.  What the heck.  Pushed it back, and it fired.  I was kneeling by the ignition while the rider started it.  Step back, it dies.  Finally figured out that the sunlight was fooling the ignition with the points cover removed.  Put on the cover and it ran fine everafter.

A friend of mine had a car that would stall-out when he got about a block away from his house. As it turned-out, there was a HAM-radio guy living in a house right by where it always quit running - as far as I know, he never figured-out what it was interfering with (maybe ignition, maybe anti-theft, maybe something else), he just took a different route to the main road and never had any more problems.


For all practical intents and purposes, it seems the solution is to just use the BJT-input opamp instead, with the bonus that I'll never have to worry about getting lynched over using the "wrong opamp". I slightly prefer the sound of the TL072 (fighting words in some other parts of the internet ), but the difference is so small that I'm still satisfied, and it's not like I did a proper blind A/B test anyway.

...I've never been one to be satisfied with merely solving a problem unless I also understand it, though, so I'm still curious as to the *why*. My JFET input opamp has issues, but my BJT input opamps don't. The input resistance of the 072 is orders of magnitude higher than the 4558 (10^12 vs 5*10^6), it's slew rate higher (assuming the 4558 datasheet is listing slew at unity gain like the 072 sheet does, 8-13V/us vs 1V/us, and in any case the 5532's 9V/us at unity gain eliminates it as a possibility anyway), and they have similar specs for everything else (GBP in particular) that isn't *way* above my pay grade. My theory (also above my pay grade) is that I'm getting high-frequency oscillation that eats the gain without being audible, and if that is correct, it would necessarily be happening in the tone-control half of the opamp, as the character of the distortion wasn't affected, only the volume. That sufficiently lowering the resistance the output and inverting input see before the cap to ground cures the problem seems to fit with what little I *do* understand about oscillation, but is contradicted by what I understand about that type of tone control (i.e. that treble is *boosted* in that case).

Is there a relatively simple way to test for oscillation without an oscilloscope? Am I way off the mark in thinking that it even*might* be oscillation?

[Keppy]

My theory (also above my pay grade) is that I'm getting high-frequency oscillation that eats the gain without being audible, and if that is correct, it would necessarily be happening in the tone-control half of the opamp, as the character of the distortion wasn't affected, only the volume. That sufficiently lowering the resistance the output and inverting input see before the cap to ground cures the problem seems to fit with what little I *do* understand about oscillation, but is contradicted by what I understand about that type of tone control (i.e. that treble is *boosted* in that case).

Is there a relatively simple way to test for oscillation without an oscilloscope? Am I way off the mark in thinking that it even*might* be oscillation?

Sounds at least sort of plausible to me, though I wouldn't have thought there to be enough gain in that stage to cause such oscillation.

Strap a cap across the 1k feedback resistor to cut HF gain. .0047uF should cut only supersonic frequencies. If that fixes/changes it, you have your answer. Alternatively, and with no input signal, try metering the opamp output for AC, or measure the output frequency if your meter has that function.

It's also worth audio probing the + inputs and outputs of the opamps just to verify what you think you know about where the problem is.

[highwater]

Audio probe results: Volume loss heard at pin 7, but pins 5 and 6 are the same volume either way. The other five pins are also unaffected. No output at all from pins 4 and 8, so that's a good sign at least .

AC measurements: 0v at pins 1, 2, 4, 5, 6, 7, 8. 0.002v at pin 3, along with an increase in audible noise when pin 3 is touched. I guess that confirms meter loading, but seems pretty-much useless besides.

0.0047u cap across the 1k resistor between pins 3 and 5: *audible* oscillation, somewhere near an F3, kinda nasally sounding. After doing so I realized you said *feedback* resistor, but I may as well mention it anyway since I tried it.
0.0047u cap across the 1k resistor between pins 6 and 7: Volume still drops.

[highwater]

I found a relevant post from a couple years ago:

but I've heard that the TL072 is quite prone to oscillation at unity gain.

I use TL07x parts at unity gain all the time, in volume. They are specified for unity gain. However, we are now well past the era when one could trust an electronic part to be what the label says it is. Are you certain that these are real TL07x parts? Sorry - I have to check.  And any high bandwidth, high gain part will oscillate if you give it half a chance. There are periodic moans about the TL07x series I see, but if they were really that bad, they would not be used in such huge quantities. *I* wouldn't use them in even my tiny quantities if they were that bad.

All opamps are most prone to oscillation at unity gain. This is the gain with the most feedback available to power oscillation. But if you have bona fide TL07x parts and they really will not do unity gain, return them and get another brand. Be sure you can demonstrate this independent of your layout. If all of them oscillate in a socket jumpered for unity gain from a battery, you have a case.

If the same part does not oscillate when lashed up with wires and sockets, but does in your layout, then I'd say you have to examine the layout again. Doing all the conventional-wisdom processes in a layout does not ensure success, it only avoids the problems which have been found by other people before.

I'm quite new to analog design but I've seen that people often add low value caps in the feedback path to correct the phase margin (or is it simply for attenuation?). Is this the best solution?

Adding a cap across the feedback resistor is not something that necessarily corrects phase margin. What it does is to increase feedback with increasing frequency above the rolloff of the cap and any parallel resistances. This may cut gain below unity before the phase hits oscillation, but it's not a slam-dunk, especially as the nodes you really want to use a phase compensation cap on are buried inside the IC package. That's why opamps used to bring out compensation pins in the bad old days.

If your opamp is not unity gain stable, using a cap from output to input can sometimes cause oscillation. As someone who's new to analog design, you have a long - but very interesting and fun - path ahead of you. If you've been extruded through some kind of linear controls/feedback training, go read the books by Walt Jung and Jerald Graeme on opamp applications.

That seems to pretty-much confirm that (a) my theory is at-least plausible (b) despite (and maybe even because of) the low gain of this stage, and (c) that the cap across the feedback resistor didn't work doesn't necessarily prove anything.

It's getting a bit late to try this tonight, but the jumper wires to where the tone pot is mounted are rather unseemly in their routing, so tomorrow I'll try re-routing them as best as I can, and see if that makes a difference.

[Keppy]

 Yeah, the things I gave you to try could have proved oscillation, but couldn't really disprove it. Like the post above says, the feedback cap doesn't always work. I also have no idea if your meter is capable of measuring AC at the frequencies we're talking about or is optimized just for 60Hz mains, but it seemed worth a shot. A positive result would have proved oscillation, but a negative one proves nothing. I'm out of ideas now, aside from just poking at the layout. Which is cool (though not helpful), because when you figure it out I get to learn something.

[PRR]

> I can kill the sound completely in several ways by poking at the breadboard

Always a bad sign (unless you mean leaving your finger on a spot, which is always going to upset things). Loose breadboard connections. Does not fit the '558 vs 072 discrepancy (unless your 072s have skinny pins).

Oscillations usually change when finger or pencil-tip is "near" the hot-spot. On a Champ with too-tight tonestack wiring, I could get all different obvious squeals by moving a pencil among the wires, and "cure" it (barely) by pushing wires around. On the faster (than a tube) chips the squal could be supersonic with no direct sound, but the over-excited circuit could de-bias itself to a low-gain condition.

This is obviously a fluke fault. If on breadboard, IF the sound is OK, I would not be over-concerned until it was wired up solidly in a box.

> assuming the 4558 datasheet is listing slew at unity gain like the 072 sheet does

Internally-compensated opamps have the same slew-rate at any gain. All of these rates are faster than a 9V guitar thing needs.

For external compensation ('709, '301, maybe '5534), if you change the gain, you should change the compensation, which changes the slew. But that's so annoying that ext.comp. went out of style in the 1970s (except specialized work). Also the lame little '308, which would be forgotten except it was used in some Old Classic.  

[karbomusic]

Is it possible you are seeing oscillation on the output (instability via capacitive loading) that you can't actually hear but eats up current? I had a circuit with that problem not long ago, fixed by adding a 100R in series on the opamp output. It was also somewhat intermittent by poking around etc, but it turned out it only went into oscillation with the output at full and that was one of the things poking around would toggle (on the breadboard). The main symptom was unexplained volume drop, when I threw it on the meter you could see the mA nearly triple when it occurred.

[highwater]

Is it possible you are seeing oscillation on the output (instability via capacitive loading) that you can't actually hear but eats up current? I had a circuit with that problem not long ago, fixed by adding a 100R in series on the opamp output. It was also somewhat intermittent by poking around etc, but it turned out it only went into oscillation with the output at full and that was one of the things poking around would toggle (on the breadboard). The main symptom was unexplained volume drop, when I threw it on the meter you could see the mA nearly triple when it occurred.

Yup, that did the trick, I think. I'm always hesitant to consider an intermittent problem fixed without giving it plenty of time to rear it's ugly head (I've been bitten on the tuchus a few times), but a minute-long "light-switch rave" didn't quiet it down, so I'm reasonably confident, especially since the theory and fix both add-up. I was getting a little over 1ma extra current draw with symptoms present, BTW.


I typed-up the following before reloading to see the 100R suggestion, so I may as well post it for the sake of posterity:

> I can kill the sound completely in several ways by poking at the breadboard

Always a bad sign (unless you mean leaving your finger on a spot, which is always going to upset things). Loose breadboard connections. Does not fit the '558 vs 072 discrepancy (unless your 072s have skinny pins).

Oscillations usually change when finger or pencil-tip is "near" the hot-spot. On a Champ with too-tight tonestack wiring, I could get all different obvious squeals by moving a pencil among the wires, and "cure" it (barely) by pushing wires around. On the faster (than a tube) chips the squal could be supersonic with no direct sound, but the over-excited circuit could de-bias itself to a low-gain condition.

I was being facetious, I simply meant that there are plenty of things I can short out - the whole thing is only about 15 rows on the board, including the tone and gain trimpots (the volume is just a 100k resistor to ground ATM). In retrospect, a tight layout on breadboard has many downsides and few virtues... I'm probably going to end-up being one of those crazy people who tries to stuff a submini Marshall clone into a 1590A without resorting to RG's "real tube tone in a stompbox" trick.

This is obviously a fluke fault. If on breadboard, IF the sound is OK, I would not be over-concerned until it was wired up solidly in a box.

That's definitely reassuring - my work is closed for the next two weeks, so I should be able to get some proper building done. I'm actually not even completely sure that I won't change to a different tone control anyway.

> assuming the 4558 datasheet is listing slew at unity gain like the 072 sheet does

Internally-compensated opamps have the same slew-rate at any gain. All of these rates are faster than a 9V guitar thing needs.

For external compensation ('709, '301, maybe '5534), if you change the gain, you should change the compensation, which changes the slew. But that's so annoying that ext.comp. went out of style in the 1970s (except specialized work). Also the lame little '308, which would be forgotten except it was used in some Old Classic.  

That makes sense - I wonder why TI is explicitly mentioning that it's at unity gain, though... can the inputs have their own slew-rate limits that could theoretically become important at gains substantially *less* than unity?