Weird problems with circuit.

Started by FingerBlisters, January 22, 2021, 03:43:43 PM

Previous topic - Next topic

FingerBlisters

Hello everyone, I hope you're well.  I've got a distortion pedal here and it's suffering from some problems I can't track down. Nothing ground breaking and here's the schematic:



What we've got is a 1.5m pull down, 10n input cap, then VREFs all biased through a 1m then the gain pot section, into another gain section to drive the clipping didoes. Then there's a little tone stack and out again through a slight gain stage which cuts some treble before heading out.

Problems:


- Way too much treble. Insane amounts. I installed a 10nf at C10 which helped somewhat but when at lower gains, it muddied it all up.
- Voltage on pin 8 of both ICs is around 10.8v, all other pins (except 4 which is 0) are half that. Does swing mildly. 15 minutes prior it was 10.5v. It didn't get much higher than 10.8v though but still, it should be 9 (ish).
- Farty 'blatty' low end, makes it sound like a mis-biased fuzz. Not tight.

Am I missing something glaringly obvious why this is happening? I've run the numbers and can't fathom why, theoretically, any of this should be happening, especially the voltage being what it is when it's coming from a regulated 9v supply.

bushidov

#1
I redrew the schematic so I could read it. Not anything wrong with yours, just was trying to make sense of it.

"A designer knows he has achieved perfection not when there is nothing left to add, but when there is nothing left to take away."

- Antoine de Saint-Exupéry

FingerBlisters

Hey that's beautiful. Thanks for doing that.

If you're using Eagle, how do you make the shortcuts (vref, +9v etc)? It's something I can't figure out.

Another update:

I had the pedal unplugged and in front of me for a bit.

I plugged it back in with a DMM hooked up. The voltage started off at 9.4v (measuring pins 8 of the op amps) and slowly but steadily climbed until it topped out around 10.8v. This makes me think that maybe one of the electrolytics is dying causing the swing?

antonis

#3
Hi & Welcome..  :icon_wink:

1. Delete C13
(op-amps don't like high capacitive loads..)

2. > then VREFs all biased through a 1m <
Not both TL072B..
(there isn't any need for them 'cause they aren't inside signal path..)

3. Make C7 3.9nF (or 4.7nF), R8 22k (or 18k) and (optionally) R7 47k and C8 10nF..

P.S.
It's kind of strange complaining for "way too much treble" when R5/C5  Low Pass filter calls for 708Hz corner frequency..
"I'm getting older while being taught all the time" Solon the Athenian..
"I don't mind  being taught all the time but I do mind a lot getting old" Antonis the Thessalonian..

FingerBlisters

#4
cheers for that. C13 is filtering the vref line, just as c10 here:

https://www.pedalpcb.com/docs/AngryCharles.pdf

Thanks for the suggestions on the tone stack. That doesn't look too different to how I have it already though so could you explain the thinking behind the changes? I'm probably missing something.

FingerBlisters

P.S.
It's kind of strange complaining for "way too much treble" when R5/C5  Low Pass filter calls for 708Hz corner frequency..

I know, that's what I'm trying to work out.

iainpunk

maybe try changing C2 from 100n to 1n, it will take out some bottom end and cleans up the fuzziness in to distortion.
and changing C5 to 1n might prevent treble from getting clipped and creating even more treble.

cheers, Iain
friendly reminder: all holes are positive and have negative weight, despite not being there.

cheers

FingerBlisters

Right now the 100n/2.2k should be working as a high pass filter for the gain pot. Hopefully it's only boosting frequencies above 723hz (like a ts808, Marshall guv'nor etc). By making that a 1n it'll lower that filter to 7hz which is gonna add a bunch of flub and probably overload the next stages.

Interestingly the voltage has climbed to 11.5v... I'm really confused where that climbing is coming from considering it's a 9v supply (truetine cs7). That's got my stumped. Perhaps one of the ground connections isnt what it should be? What causes voltages to rise steadily like this?

PRR

Quote from: FingerBlisters on January 22, 2021, 05:04:03 PM...strange complaining for "way too much treble" when R5/C5  Low Pass filter calls for 708Hz corner frequency..

C2 R3 are a 750hz bass-cut. Some bass cut is expected in distortion, so we don't drown in mud, but it is a compromise. Some distorters use a less steep bass-cut.

  • SUPPORTER

Keppy

The only thing I can thing of that would cause V+ to rise gradually when you power up is C12. That cap has to charge up when you first plug it in, and until it's fully charged V+ won't be stable. All that is entirely normal, though. The time to become stable depends on the impedance of the power supply.

I agree about moving C13. I know you saw it in that other design, but it's better to filter the voltage THEN buffer it, instead of buffering an unfiltered input and trying to drive a big cap to ground to filter it. You can just move that cap to the R12/R13 divider and be better off, though you'll probably also want to reduce its value since that's a higher impedance node. If your low end sounds like a mis-biased fuzz, please start your troubleshooting by removing the bias filter cap if only to rule it out as the cause.

Your supply voltages don't seem too extreme. Often a new battery or an unregulated wall wart puts out well over 9v. What supply are you using, and have you measured the voltage it's putting out on its own apart from your circuit? Regardless, I doubt this has anything to do with your tone problems.

There's not a lot of room between the two filters in the first gain stage. R3/C2 is ~720Hz, C3/GainPot is ~1920Hz at max gain. You're focusing on a pretty narrow band at high gain, and it's middle/upper midrange. There's a similar dynamic between C4/15k ~320Hz and C5/R5 ~720, except the focus there is about an octave lower than the first stage, putting it right about where the tone control's mid scoop is.

Combined max gain of the first two stages is (250k/2.2k) * (680k/15k) = over 5,000. This is a ton of gain, but as mentioned above it's pretty frequency-limited. It's more frequency-limited as the gain goes up, so if you like the low gain settings better than the high gain settings (but still want high gain), maybe reduce overall max gain, but loosen the frequency restrictions a bit. This should give you less gain in the narrow midrange, but a bit more at other frequencies.

Opamp clipping is pretty harsh, which might be the source of your treble trouble. Lots of designs do just fine with the opamps clipping, but I haven't seen one with this much gain. Maybe try some TS-style soft-clipping diodes in the feedback loop of the second gain stage to prevent the opamp itself from clipping so hard. This can sometimes reduce harshness without sacrificing crunch, since you're clipping with diodes to ground after that stage anyway.

I think maybe your BMP-style tone control has too high of an impedance for the inverting stage that follows it. If you put the control all the way to the bass side and ignore caps and paths to ground, you end up with 63.3k series input resistance to the opamp (R6 + (R7/TonePot in parallel) + R9). This gives a gain REDUCTION through the opamp, which I suspect is not what you intended. It's also different on the treble side of the pot because the series resistance is reduced to just R6 + R9, giving you a gain of two. I think if you put a non-inverting buffer following the tone control you'll have much better bass/treble balance.

Anyway, hopefully some part of that helps you get it how you like it!
"Electrons go where I tell them to go." - wavley

FingerBlisters

That's a hell of a response, thank you.

I figured out my dmm had a dying battery and as soon as I installed a fresh one the voltage was reading stable at 9.4v with zero creep.

I think you're right on the gain. It was about right with just 2 stages, then the tone stack and then out, but the stage at the end has made it crazy. It clips the diodes on about 15% of the dial.

The crazy treble still puzzles me but I think the gain has a lot to do with it. I've found that basically installing a 10nf cap at the end works but isn't the ideal solution.

And yeah the intent was to force the gain into the upper mids where the snarl is. Tomorrow I'll shift the gain down a lot and see what occurs but I think it's on the right track.

Thanks for the insights about the buffering too, I can use the spare stage down there to buffer the tone stack like you suggested. It'll be more use doing that than what it's currently doing.

If you've any suggestions on how to better engineer the circuit in front of the tone stack I'm all ears! Funny how the simplest circuits can be the most difficult to dial in.

iainpunk

QuoteRight now the 100n/2.2k should be working as a high pass filter for the gain pot. Hopefully it's only boosting frequencies above 723hz (like a ts808, Marshall guv'nor etc). By making that a 1n it'll lower that filter to 7hz which is gonna add a bunch of flub and probably overload the next stages.
i think you maths is off here, and also the theory...
100n/2.2k is also incorrect, since the whole feedback path interacts with that capacitor, not only the 2.2k. at max gain, 252.2k/100n gives 6.3Hz. using 1n gives 630Hz.
making impedance in the 'termination path' (the path from (-) to ground) larger, lowers the gain, making the impedance smaller boosts the gain!

a smaller cap has larger impedance for low tones, preventing them from getting boosted
a larger cap has smaller impedance for low tones, boosting them more and adding flub.
i *strongly* recommend you try this stuff out IRL, just have 2 caps ready and switch them in and out to hear how it sounds. you'll see that the bass drops with a smaller cap.

cheers, Iain
friendly reminder: all holes are positive and have negative weight, despite not being there.

cheers

FingerBlisters

Thanks for explaining. I was getting the theory from the electeosmash analysis of the marshall guv'nor (except with a larger gain pot here).

I'll try out what you said. Can't hurt!

iainpunk

electro smash quite often has mistakes and problems on their site. take what they post with a grain of salt.

cheers, Iain
friendly reminder: all holes are positive and have negative weight, despite not being there.

cheers

bushidov

QuoteIf you're using Eagle, how do you make the shortcuts (vref, +9v etc)? It's something I can't figure out.
I grab GND from supply1, here:


I grab 9V from supply 2, here:


I had to make my own symbol and such for the 4.5V. Below is my library I built for EAGLE that I use at work. It would be under ADAPCO-misc
http://www.cyclone-radio.com/2021-01-23-lib.zip
"A designer knows he has achieved perfection not when there is nothing left to add, but when there is nothing left to take away."

- Antoine de Saint-Exupéry

FingerBlisters

Quote from: iainpunk on January 23, 2021, 01:55:32 PM
electro smash quite often has mistakes and problems on their site. take what they post with a grain of salt.

Fair enough. Thanks for explaining. I'll socket that cap and the feedback resistor after football today and see what's going on. Cheers again.

cheers, Iain

anotherjim

You really do need to move C13. It will be better on the input of the supply splitter (pin5) and only needs to be 10uF there.

FingerBlisters

Quote from: anotherjim on January 23, 2021, 05:06:08 PM
You really do need to move C13. It will be better on the input of the supply splitter (pin5) and only needs to be 10uF there.

I can do that, so from the 4.5v node between the 47k pair?

FingerBlisters

Quote from: Keppy on January 23, 2021, 04:10:36 AM
The only thing I can thing of that would cause V+ to rise gradually when you power up is C12. That cap has to charge up when you first plug it in, and until it's fully charged V+ won't be stable. All that is entirely normal, though. The time to become stable depends on the impedance of the power supply.

I agree about moving C13. I know you saw it in that other design, but it's better to filter the voltage THEN buffer it, instead of buffering an unfiltered input and trying to drive a big cap to ground to filter it. You can just move that cap to the R12/R13 divider and be better off, though you'll probably also want to reduce its value since that's a higher impedance node. If your low end sounds like a mis-biased fuzz, please start your troubleshooting by removing the bias filter cap if only to rule it out as the cause.

Your supply voltages don't seem too extreme. Often a new battery or an unregulated wall wart puts out well over 9v. What supply are you using, and have you measured the voltage it's putting out on its own apart from your circuit? Regardless, I doubt this has anything to do with your tone problems.

There's not a lot of room between the two filters in the first gain stage. R3/C2 is ~720Hz, C3/GainPot is ~1920Hz at max gain. You're focusing on a pretty narrow band at high gain, and it's middle/upper midrange. There's a similar dynamic between C4/15k ~320Hz and C5/R5 ~720, except the focus there is about an octave lower than the first stage, putting it right about where the tone control's mid scoop is.

Combined max gain of the first two stages is (250k/2.2k) * (680k/15k) = over 5,000. This is a ton of gain, but as mentioned above it's pretty frequency-limited. It's more frequency-limited as the gain goes up, so if you like the low gain settings better than the high gain settings (but still want high gain), maybe reduce overall max gain, but loosen the frequency restrictions a bit. This should give you less gain in the narrow midrange, but a bit more at other frequencies.

Opamp clipping is pretty harsh, which might be the source of your treble trouble. Lots of designs do just fine with the opamps clipping, but I haven't seen one with this much gain. Maybe try some TS-style soft-clipping diodes in the feedback loop of the second gain stage to prevent the opamp itself from clipping so hard. This can sometimes reduce harshness without sacrificing crunch, since you're clipping with diodes to ground after that stage anyway.

I think maybe your BMP-style tone control has too high of an impedance for the inverting stage that follows it. If you put the control all the way to the bass side and ignore caps and paths to ground, you end up with 63.3k series input resistance to the opamp (R6 + (R7/TonePot in parallel) + R9). This gives a gain REDUCTION through the opamp, which I suspect is not what you intended. It's also different on the treble side of the pot because the series resistance is reduced to just R6 + R9, giving you a gain of two. I think if you put a non-inverting buffer following the tone control you'll have much better bass/treble balance.

Anyway, hopefully some part of that helps you get it how you like it!

Awesome post. Thanks for taking the time.

Re: tone stack impedance. I could rework the values pretty easily, some kind poster above even suggested values based around using 22k resistors. Hopefully that would lower the impedance of the stack to a level that the following opamp would be happy with?

But then by doing that some of the low end is removed from the sweep (why I have 100ks in there as is) which isn't fun.

One thing I have done is beef up the 33n to a 220n and the 150n to a 330n to allow more low end through to hopefully make the dynamic range wider.

Keppy

Glad you figured out the meter issue. I haven't used mine in awhile, and I hate to think what state the battery might be in. :icon_eek:

The issue with the resistance of the tonestack isn't the overall resistance, it's that the amount of series resistance is larger on the bass side than the treble side. Since input resistance is part of the gain equation of inverting opamp stages, that means the gain will always be different for the two sides of the pot. Bass will always be lower. The more I think about it, the more I think this is the source of your treble issue, rather than filtering in the gain stages.

The solution (if, y'know, I'm right about any of this :icon_razz:) is to make the input impedance of the last stage so high that the difference in input resistance is small by comparison. Maybe try an R9/R10/C10 combination of 470k/1M/50pf and see if it makes a big difference.
"Electrons go where I tell them to go." - wavley