Choices for a simple transparent 5 to 10 band EQ (for vocals)

[Mark Hammer]

Nice work, and appreciate the ongoing followup.

I hope the two legs of C16 are not touching as they appear to be in the photo.

[jfrabat]

It is an optical illusion from the angle.  They are about 1/4 inch apart (a bit less, but more than enough to not make contact)...

[Rob Strand]

5. This is the one that is throwing me off...  Everything is working, but if I boost to max the highs, and select the highest frequency, I get a high pitched tone coming out.  It only happens in the last pot, and only at max boost with the highest frequencies.  In the last Wein Bridge I used 10K instead of 12K resistors in the other 3; should I go back to 12K in the highest frequency?

It's probably some marginal stability issue and the unit is oscillating.    It could also be made worse if the input and output wires are not kept far enough apart.

There's a few common things that help.  In weird cases they can make things worse.

Assuming the problem is only *caused*  by the last stage:
1) The place to start would be to add 100pF across R7.
     This is a very common fix for these problems, so if it works leave it at that.
2) Next would be to add a 100pF across R12.   You can try it with and without the cap across R7.
     I'm not 100% happy about this one if you don't need it.

Assuming the real problem is throughout the whole unit and is exposed by setting the boost pot to maximum:
3) Add 100pF across R7, R5, R9, R30
     Again this isn't an uncommon configuration.

I'm hoping (1) is good enough and the other items are just things to try.

[jfrabat]

1) The place to start would be to add 100pF across R7.
     This is a very common fix for these problems, so if it works leave it at that.

Just in parallel to R7? 

[Rob Strand]

Yes parallel.

[Rob Strand]

You can see the same sort of thing used on the Boss MT-2 pedal,


(The 330R resistor could be helping but I wouldn't go that far at this point.  Many working circuits don't have these.)

[jfrabat]

It did not work...  Will try with the other options.  Starting with 100pF across R7, R5, R9, R30...

[Rob Strand]

Will try with the other options.  Starting with 100pF across R7, R5, R9, R30...

Worth trying.  In fact even if the caps across R7, R5, R9, R30 don't work it might be worth leaving them in for now as they help reduce secondary issues.

Another thing you can try for an overall problem is a 100pF across R29 (1M).

It's weird the prototype worked.     Maybe the 10k on the last band was enough to push it over the edge.

[jfrabat]

It's probably some marginal stability issue and the unit is oscillating.    It could also be made worse if the input and output wires are not kept far enough apart.

There's a few common things that help.  In weird cases they can make things worse.

Assuming the problem is only *caused*  by the last stage:
1) The place to start would be to add 100pF across R7.
     This is a very common fix for these problems, so if it works leave it at that.
2) Next would be to add a 100pF across R12.   You can try it with and without the cap across R7.
     I'm not 100% happy about this one if you don't need it.

Assuming the real problem is throughout the whole unit and is exposed by setting the boost pot to maximum:
3) Add 100pF across R7, R5, R9, R30
     Again this isn't an uncommon configuration.

I'm hoping (1) is good enough and the other items are just things to try.

Tried 1, 2 and 3...  No change.  I did decide to leave the caps in (well, not 2; that one I took out, but the others) for good measure, though.  I also tried changing R4 and R22 to 12K, thinking that if I cut the top frequency, it would go away, but still same issue.  It is not a HUGE issue, as I doubt I will be boosting the top frequency for either vocals or guitar, but it would be nice to have it working properly...

Will try with the other options.  Starting with 100pF across R7, R5, R9, R30...

Worth trying.  In fact even if the caps across R7, R5, R9, R30 don't work it might be worth leaving them in for now as they help reduce secondary issues.

Another thing you can try for an overall problem is a 100pF across R29 (1M).

It's weird the prototype worked.     Maybe the 10k on the last band was enough to push it over the edge.

You know, I am not sure I tried max gain in highest frequency.  So maybe it was there in the BB...  And the breadboard had 10K resistors for that one.  One difference I did notice when building the OCB is that R28 was 100 Ohm instead of 100K in the BB (I fixed that), but that change should not make this happen.

[Rob Strand]

Tried 1, 2 and 3...  No change.  I did decide to leave the caps in (well, not 2; that one I took out, but the others) for good measure, though.  I also tried changing R4 and R22 to 12K, thinking that if I cut the top frequency, it would go away, but still same issue.  It is not a HUGE issue, as I doubt I will be boosting the top frequency for either vocals or guitar, but it would be nice to have it working properly...

At least you gave it a shot.

You know, I am not sure I tried max gain in highest frequency.  So maybe it was there in the BB...  And the breadboard had 10K resistors for that one.  One difference I did notice when building the OCB is that R28 was 100 Ohm instead of 100K in the BB (I fixed that), but that change should not make this happen.

Well, it wouldn't hurt to try increasing R28 to 1k.   

Another thing which can affect things is grounding the chassis of the pots.

It's seems like a tricky problem, where is the cause is a little deeper than the common problems.

If you are really keen you could try putting a TL072 in locations IC1 and/or IC2.   The TL072's are a little less susceptible to weird oscillations that the NE5532's.

[jfrabat]

Well, it wouldn't hurt to try increasing R28 to 1k.   

I already put everything in the enclosure, and it takes a little persuading to get the board to line up just right, so I think I will just leave it as is...

Another thing which can affect things is grounding the chassis of the pots.

If you are really keen you could try putting a TL072 in locations IC1 and/or IC2.   The TL072's are a little less susceptible to weird oscillations that the NE5532's.

These I could try; the grounding of the chassis of the pots (the dual gang ones, at least) is simple enough without having to take the whole thing apart.  Same with the chip change.  I used the NE5532's because they are supposed to be cleaner than the TL072 (remember this is mainly for vocals, not guitar).  Is my understanding correct?

By the way, here is the "final" (not really, because I need to change the knobs, and they have yet to arrive!) product working:

[Rob Strand]

These I could try; the grounding of the chassis of the pots (the dual gang ones, at least) is simple enough without having to take the whole thing apart. 

I'm assuming the metal enclosure is connected to ground?    If that's the case the pots should already be grounded.

I used the NE5532's because they are supposed to be cleaner than the TL072 (remember this is mainly for vocals, not guitar).  Is my understanding correct?

If you don't clip the signal they should be pretty much the same.    I used a lot of NE5532s for HiFi stuff but the idea is not to clip in the first place.   When clipping they do have minor differences, perhaps very slightly in favor of the NE5532.    If the TL072's fix the problem I'd be more inclined  to go with them.  After listening to the video I don't think that's the problem!

That oscillation sounds quite angry.   I would fix it.  It really sounds like outright oscillation.   I suspect you might even be able to tune the pitch with the frequency pot.   One possibility is frequency cap or frequency pot tolerances are causing the problem.   The way some of those EQs work is they push the gains close to oscillation.   That would be compensated for by:
- reducing R12 to 9.1k
- OR, increasing R21 to 51k
- OR perhaps increasing only R22 to 12k.

If you have a wrong resistor value or cap value on that band it could also cause that problem.

You could test the idea by adding a large resistor or pot in parallel with R12.  Adjust the value until the oscillation goes away.  Then measure the added resistor value with a multimeter.   After after that just solder a fixed resistor of that value or slightly lower in parallel with R12.

When you measure the added resistor value it will give you a good idea how close the gain is to the point of oscillation.   If it is quite low I'd really start looking hard for wrong part values.

[jfrabat]

I'm assuming the metal enclosure is connected to ground?    If that's the case the pots should already be grounded.

Ah, good point.  The input and output jacks should be grounding them anyway.  And since the enclosure is not even painted (I just polished it a little), the contact should be pretty good.

If you don't clip the signal they should be pretty much the same.    I used a lot of NE5532s for HiFi stuff but the idea is not to clip in the first place.   When clipping they do have minor differences, perhaps very slightly in favor of the NE5532.    If the TL072's fix the problem I'd be more inclined  to go with them.  After listening to the video I don't think that's the problem!

OK, I understand.  I will keep the 5532s then.

That oscillation sounds quite angry.   I would fix it.  It really sounds like outright oscillation.   I suspect you might even be able to tune the pitch with the frequency pot.   One possibility is frequency cap or frequency pot tolerances are causing the problem.   The way some of those EQs work is they push the gains close to oscillation.   That would be compensated for by:
- reducing R12 to 9.1k
- OR, increasing R21 to 51k
- OR perhaps increasing only R22 to 12k.

Well, I measured caps and resistors prior to installing.  Caps are within 1nF from spec, and resistors were also very close.  At least the Wein bridge ones, anyway (those were the ones I measured).

If you have a wrong resistor value or cap value on that band it could also cause that problem.

You could test the idea by adding a large resistor or pot in parallel with R12.  Adjust the value until the oscillation goes away.  Then measure the added resistor value with a multimeter.   After after that just solder a fixed resistor of that value or slightly lower in parallel with R12.

When you measure the added resistor value it will give you a good idea how close the gain is to the point of oscillation.   If it is quite low I'd really start looking hard for wrong part values.

I will try increasing R12 and will report back.  I may even go back to 10K in R4 and R22 (although, to be honest, the highest frequency is already pretty high, so it is not absolutely necessary)

Before replying, I decided to check all the resistors in the last stage; all measurements are taken with the resistors in the board (I know you are not supposed to measure with them in, but for simplicity sake I did it like that).  Here is what we have:

R3: 6.6K (should be 10K) - Not in series, so not sure what it should read; color code correct (Brown, Black, Black, Red)
R4: 11.73K (should be 10K) - This one I replaced with 12K, but will bring back to 10K
R7: 6.65K (should be 10K) - Not in series, so not sure what it should read
R11: 98.1 (should be 100) - OK
R20: started at 7K and rises; waited until 14K and got bored, but it was still going up (should be 47K; color code matches R21, so it seems to be correct)
R21: 46.7K (should be 47K) - OK
R22: 9.87K (should be 10K) - OK

Here is R20 and R21:

[PRR]

Lost track what plan you are working to, but....

> R3: 6.6K (should be 10K)

I have not seen a -33% tolerance resistor since before I was born.

That sure smells like 10k with a 20k across it. You can't measure soldered-in resistors without studying everything they are connected to.

[jfrabat]

Lost track what plan you are working to, but....

> R3: 6.6K (should be 10K)

I have not seen a -33% tolerance resistor since before I was born.

That sure smells like 10k with a 20k across it. You can't measure soldered-in resistors without studying everything they are connected to.

As I said, that circuit is not in series; it has other routes, so the resistance varies with a bunch of other resistors, so I wont even start to calculate.  Besides, colo code is right...

[ElectricDruid]

Is that R20 soldered properly? From the photo it looks like the solder hasn't flowed into the hole completely. Sometimes that's no problem if it's well soldered on the bottom, but other times it indicates a bad joint.

[Rob Strand]

That sure smells like 10k with a 20k across it. You can't measure soldered-in resistors without studying everything they are connected to.

That one might make sense because the boost/cut pot and R7 appear in series.
Then you get the same effect when measuring R7, which also shows up in the measurements.

I had a look at how close the circuit is to creating a wein-bridge oscillator on its own:

As far as an oscillator goes U1B has a gain of 2, U1A has a gain of 1.21, then the Wien network should have a gain of 1/3.   The total loop gain is 2 * 1.21 * (1/3) = 0.81.     It would need to be about 1 for it to oscillate on its own.

A loop gain of 0.81 looks like it should have enough slack in it to fend off some degree of tolerances.    If there was a wrong part  then that's an obvious cause but the measurements don't look too bad.  If the tolerance were way off in a worst-case combination it would struggle to get to a loop gain of 1.  Under normal circumstances, the mis-tracking of the pot taper would be the most likely candidate for the gain to be off, perhaps helped along with capacitance tolerances.

One possible key point is the upper band only oscillates when the frequency pot is at the upper frequency settings.  Please correct me if I am wrong here.  If the band oscillates at the lower frequency setting and at maximum boost that would definitely point to a part value or  other circuit issue.    The fact it *only* occurs near at the upper frequency setting might mean the opamps are involved.

For the sake of sanity, I definitely think it is worthwhile at least trying a TL072 for U1.   That would at least confirm the opamps are involved.

As far as a work around I can only suggest the tweaks I made in Reply#111.     Note the idea of changing R22 to 12k is to leave R4 at 10k.   What that does is reduces the loop gain a bit at the high frequency settings only, which should deter oscillations by a small amount.     The other suggestions directly reduce the gain and are likely to be more successful.

[jfrabat]

That sure smells like 10k with a 20k across it. You can't measure soldered-in resistors without studying everything they are connected to.

That one might make sense because the boost/cut pot and R7 appear in series.
Then you get the same effect when measuring R7, which also shows up in the measurements.

I am proud to have realized this on my own!  I am finally learning here!   

As far as an oscillator goes U1B has a gain of 2, U1A has a gain of 1.21, then the Wien network should have a gain of 1/3.   The total loop gain is 2 * 1.21 * (1/3) = 0.81.     It would need to be about 1 for it to oscillate on its own.

A loop gain of 0.81 looks like it should have enough slack in it to fend off some degree of tolerances.

Haven't learned THIS much yet, though!   

One possible key point is the upper band only oscillates when the frequency pot is at the upper frequency settings.  Please correct me if I am wrong here. 

You are not; what you say is correct, it only oscillates with the frequency pot at max and the boost close to max (a hair before max, it starts oscillating).  If the frequency pot is, say, ad the 12 o'clock position, and I go to max boost, there is no oscillation.

For the sake of sanity, I definitely think it is worthwhile at least trying a TL072 for U1.   That would at least confirm the opamps are involved.

That is simple enough to do; I will give it a shot and report back.

[jfrabat]

For the sake of sanity, I definitely think it is worthwhile at least trying a TL072 for U1.   That would at least confirm the opamps are involved.

Tried it, and no fixing the issue.  The TL072 does exactly the same thing.  So it is not the OpAmp.  Must be something else.  I also soldered all the suspicious resistors from the top, and still, no change.

Also, noticed something odd; when I connected it to my interface, the tone comes in even earlier; as soon as I boost even a little (while at max frequency), the shrill starts.  Maybe I could just not hear it in the speaker, not sure...

[Rob Strand]

Tried it, and no fixing the issue.  The TL072 does exactly the same thing.  So it is not the OpAmp.  Must be something else.  I also soldered all the suspicious resistors from the top, and still, no change.

It was definitely worth trying.

Also, noticed something odd; when I connected it to my interface, the tone comes in even earlier; as soon as I boost even a little (while at max frequency), the shrill starts.  Maybe I could just not hear it in the speaker, not sure...

That's not good.    Perhaps the load of the interface on the last opamp is making the oscillation come in earlier, or perhaps making it come in at a frequency you can now hear.

Given the load might be having an effect it might be worth increasing R27 (schematic on reply #77) from 100R to 1k.

I'm starting to run out of ideas.   At this point, there's a couple of things I would probably try if it was my unit:

- Change the caps on say band #2 so the frequencies are the same as band #4.   Band #2 is currently working and the idea is to see if there is some inherent problem with the design when frequencies are high.

- Built a single band on the bread-board and try to debug it.

It is still worth checking over the board, tracks, soldering.   Maybe check the frequency pot is actually working and is soldered correctly.   If one gang is broken it could cause this problem.