Whiny Tubescreamer

[gaussmarkov]

Hmm. I haven't tried it with a resistor yet, but shorting it out does nothing. However, when I touch the wire coming from R8 (the blueish wire on the layout), the squeal stops and I can turn the other two pots and there is no squeals.

by "touching" the wire from R8, do you mean that you are grounding pin 1 of the RC4558D?  it seems to me that ought to stop the signal cold:  no squeals because there is no signal going to the rest of the circuit.

[Ortiz]

No, actually when I ground it with the circuit, then the entire signal stops. When I touch it, only the squeal stops.

[gaussmarkov]

No, actually when I ground it with the circuit, then the entire signal stops. When I touch it, only the squeal stops.

so when you "touch" the wire you are just placing your finger on the pvc insulation of the wire?  sometimes, for example in fuzz circuits, people need to use shielded hook up wire to prevent oscillations.  i have not heard of a need for this with the TS circuit.

i don't think that touching the pot casing or shaft should affect the circuit.  i suggest that you remove that pot to simplify matters.  i am not sure why simply shorting the pot didn't help.  but i would still remove it to eliminate a complication.  after taking it out, you could measure the resistance across the wiper and each outside lug to check whether that the pot is doing what it is supposed to do.  (in theory, you could also check without disconnecting the pot.  i would disconnect the voltage source first.)

(edits in red)

[Ortiz]

No, actually when I ground it with the circuit, then the entire signal stops. When I touch it, only the squeal stops.

so when you "touch" the wire you are just placing your finger on the pvc insulation of the wire?  sometimes, for example in fuzz circuits, people need to use insulated hook up wire to prevent oscillations.  i have not heard of a need for this with the TS circuit.

i don't think that touching the pot casing or shaft should affect the circuit.  i suggest that you remove that pot to simplify matters.  i am not sure why simply shorting the pot didn't help.  but i would still remove it to eliminate a complication.  after taking it out, you could measure the resistance across the wiper and each outside lug to check whether that the pot is doing what it is supposed to do.  (in theory, you could also check without disconnecting the pot.  i would disconnect the voltage source first.)

I guess I should've said this but I did remove the pot. I cut it off and twisted the two wires together to short it.

When I say "touch" I mean I'm touching the stripped end of the wire.

Edit: I'll just replace the pot and see what that does.

[gaussmarkov]

When I say "touch" I mean I'm touching the stripped end of the wire.

Edit: I'll just replace the pot and see what that does.

ok.  here's a prediction:  replacing the pot won't fix anything.  if it's not working without the pot there, there's no reason (yet) why putting in a new pot should help.  if it does, that's new info (at least to me  ).  on the other hand, if the pot is working properly, it should not hurt either.  as far as we know, the pot you installed earlier was not working properly as wired.

i am guessing that you want to replace the pot because everything else seems to work when you touch the pot side of R8.  but cutting out the old pot and twisting the two wires together is like having a proper pot dialed completely counter-clock-wise (ccw).  so you have already accomplished what putting in a new pot will do.

i am unsure whether you are working on this circuit in an enclosure at this point.  this is a concern because sometimes things touch each other or the enclosure when they should not.  for this reason, people often suggest keeping the circuit out of the enclosure for all this debugging. if your pcb is in an enclosure, how about removing it?

[didier]

i have built the tubescreamer from GGG.

no problems with the layout or so i can assure you that...

i had some problems with the wires though, i used some crappy wire that cracked off if i were to wiggle it around for 3 times, i also found a problem with the drive that had to do with a wire that had almost come off but not entirely...

wires are sneaky bastards!

[Ortiz]

Yes, I took it out of the enclosure when you asked your first question.

[gaussmarkov]

Yes, I took it out of the enclosure when you asked your first question.

great!  here's what i do next.  it's not sophisticated, because i don't know much about the sources of oscillation other than those already mentioned in this thread.  and it's tedious.  but it works for me.  and knowing that the circuit works when everything is done correctly keeps me motivated.

first i use my dmm to check continuity.  i print out a copy of the layout and use a pen to check things off as i go.  i start with a net, for example the ground net that runs around the sides and bottom of the layout, and check for continuity between the leads of components that are supposed to be on that net.  for example, i would check the continuity between the south lead of R22 and the south lead of R3.  sometimes leads are not accessible from the top, like C12, those i check on the solder side.

after confirming that a net is properly hooked up, i use my dmm to check to make sure that that net is not connected to any other net.  i take any lead on the net that i have confirmed and check for continuity between that lead and a lead on each net that i have not already confirmed.  so i might take the south lead of R22 and check for continuity with the green hookup wire coming from the 3pdt switch (or the north lead of C1), the emitter of Q1, the base of Q1, and so on.  be sure to include off board components in this check.

once a net has passed both continuity/discontinuity checks, i don't touch it again.  if one of the tests fails, obviously, then there is something to fix.

i admit that a lot of this is redundant for someone with more knowledge.  the fact that your pin voltages check out implies that a lot of this is unnecessary.  but i don't have the knowledge yet to know where to take shortcuts.

second, i use my dmm to check the values of the resistors (including pots) and the capacitors and the directions of the diodes.  again, some of this is unnecessary, but necessary for me.  and every mistake that i have made has been uncovered in this way.

sometimes someone like R.G. Keen comes along with greater analytical skill, who can give us more focused strategies.  but while we're hoping for that, i figure it's good practice and bound to turn up a problem.

[Ortiz]

Well thank you so much for your help so far! I'll get started as soon as possible.

Unfortuately, based on past experience, it's probably somehing dumb.

[gaussmarkov]

Well thank you so much for your help so far! I'll get started as soon as possible.

Unfortuately, based on past experience, it's probably somehing dumb.

you are most welcome!  i wish i could pin point the problem faster.  and i get to learn, too.

re sthg dumb:  totally! everyone has been there, and will be again.

[gaussmarkov]

we could test our understanding of things by making some (not so) educated guesses.    and there is a chance that we will get some feedback on our thinking.  here's some thoughts of mine:

• it's probably not something around Q1.  the voltages all check there.  and the circuit is so simple there that it looks like the voltages tell the whole story.  same logic for Q2.
• the feedback loops for the op amps are prime suspects.  for pins 1 and 2 we have D1, D2, C3, C4, R6 and R7 to consider.  for pins 6 and 7, R17 and the tone stage.  that tone stuff seems worth particularly close attention.


• and also the things in between the op amps, because they also hook into the feedback loops:  R8, R16, and C5.

that list still contains a lot of the circuit.  so it doesn't narrow things down much.

here's an odd fact (that R.G. explained to me) that i would experiment with (because i am grasping for straws).  the south lead of R6 could alternatively be hooked up to ground, instead of Vref, as long as C3 is in place.  you can see that there is even a pad in the GGG layout for exactly that alternative arrangement.  this change should not make any difference.  if it does, then something is wrong with the first feedback loop.

incidentally, if you have not read this yet, consider it required reading for what we are attempting to do:  The Technology of the Tube Screamer by R. G. Keen 

[gaussmarkov]

i decided to work harder at understanding what causes oscillation/squeal in a tube screamer circuit.    and to take my own advice and reread R.G.'s The Technology of the Tube Screamer. here's what i have learned: two common reasons for oscillation in noninverting op amp setups (see R.G.'s article) like those found in the tube screamer are found in the feedback loop and/or the power supply to pin 8.

observations found:  C4 (the 51pF cap) parallel to the gain pot rolls off highs and, in particular, prevents oscillation.  second, the power supply filtering provided by C12 (the 100uF cap) also works to prevent oscillation.  this cap is often preceded (but not in the tube screamer layout) by a low (say 10-100 ohm) resistor.

inferences drawn:  especially check C4 and C12.     if you cannot measure their capacitance, then replace them.

the second op amp's gain is harder for me to understand, because the tone pot is hooked up across the inverting and noninverting inputs.  but it seems possible that if R17 is a significantly higher than its 1K spec then the gain could be much higher than it should be and you might get oscillation.  alternatively if R11 were too small, same possible problem.  ditto C9.

the power supply issues suggest another question:  are you using a battery to power this circuit?  it seems that ripple from a poor DC adapter could also be a source of problems, unless C12 is adequately managing that.

so that narrows things down a bit ... 

[Dave Eason]

Sheilded cable from the jacks to the board sometimes gets rid of some problems.
You can never have enough bypassing capacitors too, according to craig anderton,
and he's right - especially with batteries.  The analog man site had a section up demonstrating the ripple and hum and tonal effects of different batteries, it's quite suprising. 
I suppose you could possibly have another HF filter, like the one with the 51pf at the core across the gain stage,
acorss the tone stage, that may help roll off the higher frequencies present.  Or possibly a smaller capacitor shunted to ground in there somewhere, to get rid of more unwanted HF.

Personally, I don't think the buffers built around Q1 and Q2 are totally necassary; without them still gives you a fairly good
effect.  The JRC4558D has a high input impedance anyway.  I know people use combined buffer/gain stages like this
with success without the transistor stage.  There's a good slimmed down tube screamer at http://www.runoffgroove.com/
called "Tubereamer".
Sorry to be no help really!

[gaussmarkov]

thanks, dave!  it's helpful to get a response that suggests we are on the right track. i actually understood everything that you said   or maybe more accurately, i recognized everything that you said. 

for example, R.G.'s article has a schematic for the "Son of Screamer" that has the transistors removed.  SysCrusher just built it as his first stompbox (see this thread) and thinks it's awesome.

[Ortiz]

Well I've gone over it with a multimeter and everything seems normal.

What you said about C4 is interesting seeing as how I replaced the 51pF cap with a 47pF cap since I didn't have one available and was told on this forum that it would only have a negligible change in tone. Are we still sure this is the case?

I connected R6 to ground like you suggested and there's a difference. Now while the tone pot is all the way up, which is when there was squealing before, there is no squealing. I rejoiced but found out that there is squealing when the drive pot is turned up (I reconnected the drive pot), unless the tone pot is all the way down.



(I'm horrible at describing things.)

[gaussmarkov]

Well I've gone over it with a multimeter and everything seems normal.

What you said about C4 is interesting seeing as how I replaced the 51pF cap with a 47pF cap since I didn't have one available and was told on this forum that it would only have a negligible change in tone. Are we still sure this is the case?

I connected R6 to ground like you suggested and there's a difference. Now while the tone pot is all the way up, which is when there was squealing before, there is no squealing. I rejoiced but found out that there is squealing when the drive pot is turned up (I reconnected the drive pot), unless the tone pot is all the way down.



(I'm horrible at describing things.)

the 47pF cap is a perfectly fine sub.  in fact, virtually everyone makes that sub AFAIK.  i'll give some thought to what happened with R6.  it shouldn't have made any difference.  the fact that it did may be a helpful clue.

[gaussmarkov]

hmm.  now i am wondering what the voltages are on the pins when R6 goes to ground instead of Vr.  can you check?

[Ortiz]

No change in voltages.

[gaussmarkov]

No change in voltages.

well, i suppose that's good--because that's what's supposed to happen.

i have a virtually complete breadboard layout of your circuit.  i don't seem to have the 25k pot handy, so i will have to fake that.  or i will pick one up tomorrow.  maybe we can make some useful comparisons early next week?  i'm booked up for the rest of this week.

maybe someone else will jump in with some new thoughts.  it seems like we need to figure out where the oscillation is occurring.  i wonder if an audio probe would be helpful for this?  see aron's debugging page if you are not familiar with this.  if the oscillation is happening in the second op amp, then i suppose that you would not hear a squeal at pin 1 of the first op amp even though you hear it at the output of the whole circuit.

[Dave Eason]

In the TS9 "R6" goes to ground, and not the Vbias supply.  I've always wondered what the need is to bias here as well as the input signal to this stage.  Surely if the input is sitting idle at half the supply, this pin could be grounded with no affect?  The feedback path in the tone stage goes to ground on the TS808, and not Vbias. 

I'm currently trying to make a "tubescreamer style" overdrive, albeit with more gain, using the MAX1044 DC-Dc converter, so no DC biasing or coupling is needed.  I've always felt there're problems when the power supply is linked to the signal somewhere along the line, leading to various audible hums and percularliar sounds and so forth.

If there is oscillation in the power supply, it will leak into the signal somewhere down the biasing circuit.

Going on a tangent; I know someone who left an Allen and Heath mix wizzard in the rain and when we thought it was dry and plugged it in, the EQ controls acted as oscillators when the gain was turned up, and the frequency controls changes the pitch - I wasn't sure why, but It all points to the fact that a short somewhere will be the most obvious case of some self-oscillation.