Unwanted noise gate in power amp distortion circuit

[pokus]

Hey there,
I'm trying for weeks now to solve a problem with a circuit I've built, but couldn't find a way.
The circuit I've modified is the acapulco gold one's. I removed everything before the first 1uF coupling cap and put an non-inverting opamp stage there. The gain can be set very high around 2000. The Input impedance is 1M. The bandpass in the feedback path is from 720 Hz to 21 kHz. The opamp I used for it is a lm308. There is also an passive tone stack in front of the volume control.



It sounds great when playing straight forward power chords and else. But the problem is that there is something like a noise gate when the strings ring out. When playing single strings sometimes the tone becomes really fuzzy(not the good fuzz) and then breaks off.
I've built it all on breadboard and there is also a lot of oscillation going on there and sometimes it happens that the oscillation noise stops the circuit from creating the noise gate and when I'm playing a single note then, it sounds just nice.(except of the oscillation noise)
So it seems to me that the lm308 can't provide enough output to drive the power amp at lower volume correctly or at least the way I want it to  But why? What else could it be? Or is it even possible that it works just fine when it's boxed?

Really appreciate any kind of help!

[PRR]

The '308 will not directly replace a '386 without new bias circuits.

Show the plan of what you think you built.

Some DC voltage readings are always good.

[pokus]

Alright. I drew a quick layout of the active part of the circuit. I also measured 10V at both ICs V+ pins. And 5V at the bias point connected to the 1M.
The second opamp is a LM386N-1. I also tried out a bunch of other ones(JRC386BD, N-3,...) but the N-1 gives me best result.




Edit: I forgot to draw the center pin connection of the pot, pin 2 and 3 are connected, pin 1 is the one going to the output of the opamp

[PRR]

Neaten the layout, shield the input lead, and box it. Is it more stable?

[thermionix]

Is the 47R too small?

[pokus]

But hasn't the RAT pedal a even smaller resistance in its feedback path?
I will try out a higher resistance and see if it sounds better. I also read that a 10uF can be placed at the bypass pin of the LM368 connected to ground. Even if I don't know why, I'll give it a go.
Can it makes sense putting a faraday cage around the breadboard or is the main oscillation/noise coming from components interacting with each other?

[Rob Strand]

It sounds great when playing straight forward power chords and else. But the problem is that there is something like a noise gate when the strings ring out. When playing single strings sometimes the tone becomes really fuzzy(not the good fuzz) and then breaks off.

It's probably oscillating because you don't have the Zobel networks on the outputs of the amplifiers.  See figures 16 and 18.  They have the 10 ohms an 0.05uF cap.
http://www.ti.com/lit/ds/symlink/lm386.pdf

You might want to look up Paul Marossy's post regarding the LM386 around Christmas 2017.  It's about the bypass caps.   (There might have been another thread as well.)

https://www.diystompboxes.com/smfforum/index.php?topic=120343.msg1127274#msg1127274

[thermionix]

But hasn't the RAT pedal a even smaller resistance in its feedback path?

Ah yeah, the Rat has a 47R there also.  It was just a wild guess on my part.  Going from memory, most circuits I recall have something over 1K.

[pokus]

Ok thanks for the advices! I put in a 10R resistor and accidentally a 10n cap and it sounded horrible. Then I connected bypass pin to ground through a 10uF cap and the oscillation was completely gone. But the unwanted noise gate is still there and the tone breaks off even quicker now. When I replaced the 10n cap with the 50n one, oscillation was there again.

And now the confusing stuff began. 
I put in some LEDs(symmetrical clipping) with a 1k resistor right after the LM308's output and the single notes really sounded better now, not perfect but they didn't break off that quickly, unless the sound changed. But as I bypassed the circuit, the LEDs start to light up alternatingly.
After reconnecting the DC jack they now light up constantly when it's bypassed and adjust to the volume(volts) when the signal goes through the circuit.
This has confused me even more, if that's possible 

[thermionix]

Use caps to isolate the LED clippers from Vref.  Sounds like you have them between 4.5VDC and ground, lighting them up.  Also look at the OCD, where the (Mosfet) clippers "ground" to Vref.

Edit:  I might not be following you totally.  What I was imagining would only light up one of the LEDs.  Can you draw a complete schematic of what you've built?  And post pics of the build?

[Rob Strand]

Ok thanks for the advices! I put in a 10R resistor and accidentally a 10n cap and it sounded horrible. Then I connected bypass pin to ground through a 10uF cap and the oscillation was completely gone. But the unwanted noise gate is still there and the tone breaks off even quicker now. When I replaced the 10n cap with the 50n one, oscillation was there again.

That is all pointing to oscillation problems.    While it is oscillating it's probably not even worth considering what it sounds like because the sound will vary depending on what the oscillation is doing.  You need to completely remove the oscillation first.

You need to add bypass caps on your power rails.   See what Paul did and follow it exactly.

[kusi]

the non-inverting pin of the second opamp has to go to +4,5V, not gnd.

[iainpunk]

the non-inverting pin of the second opamp has to go to +4,5V, not gnd.

No. The lm386 has a ground reference input and a vcc/2 referenced output. It s made for ease.

Thinking about that, therr is some leakage out of both input pins, which demand there is a path to ground, otherwise the voltage on the pins will keep rising ever so slightly, essentially gating the input of that opamp.
Try a 56k from ground to the node between the 4.7uf and the 38k, so the leakage has a place to go.

[PRR]

> therr is some leakage out of both input pins, which demand there is a path to ground,

There are internal 50K resistors. The induced offset is negligible.

As you say: "designed for ease".

[pokus]

The 47uF across v+ pin and gnd pin did the job!!!
At least almost  the noise gate now only comes in when you mute your strings, also single strings sound great until a certain point when they ring out. But the result towards my original circuit is just great!
The zobel network and the cap from bypass pin to gnd now seem to be doing nothing to the tone when there's the 47uF cap across pin 4 and 6.
Unfortunately there's still a lot of noisy stuff going on and it also has added some constant pink noise.
I guess it's time now to put it on vero and box it to see how it acts when it is shielded.

Thanks to everyone for taking care and especially for the link, all of that really helped! Maybe I will come back to this subject after I've boxed it and it's still that noisy.     Further ideas are still welcomed.


PS: I discarded the idea with the LEDs, but to make it clear: I put them behind the coupling cap not direct behind the IC like I mentioned.

[Rob Strand]

The 47uF across v+ pin and gnd pin did the job!!!

The zobel network and the cap from bypass pin to gnd now seem to be doing nothing to the tone when there's the 47uF cap across pin 4 and 6.

That makes a lot more sense.
There's no doubt Paul Marossy's efforts would help someone one day.

Unfortunately there's still a lot of noisy stuff going on and it also has added some constant pink noise.

That could be a result of the opamps and the amount of gain in the circuit.   If you increase the 72p or the 8.2n caps you might be able to remove some noise without affecting the tone too much.  It will only be a small improvement.  You could also try a small dose of low-pass filter at the output of the LM386.

[pokus]

Hey again,
so I boxed the vero and also put some single ic adapter pins in the circuit so I still can make some little tweaks.
The good news is, the oscillation and noise problems are completely gone. Even the 47uF cap across the LM386 power pins isn't necessary anymore.
BUT ..there is still a kind of a noise gate in there when you hit the strings really hard and then mute them with your hand. The sound is completely gone for 1-2 seconds and then comes back in with a buzzing swelling noise.

It's really not dramatically and sometimes kind of useful but I want to get rid of that effect and the thing that bothers me most is that I have no idea why this phenomenon occurs.
I tried changing the chips, even replaced the LM386-1 with a -4 and -3, put bigger caps in the power supply, bypassed the 47R from +9V, ran it at 12V and changed some F and R values, but as the tone changes the noise gate is always there. I always thought that the LM308 maybe can't provide enough output to drive the LM386 correctly at low volumes/voltages, but as the muted guitar strings sound(typical pickup noise,..) comes in after a few seconds it shouldn't be that.
The only thing left that I can think of is that the LM386 usually drives a low impedance speaker, but in my case there is this a classic fender tone stack right behind of it. May that be a reason? But isn't the Acapulco Gold's LM386 also directly driving the high impedance input stage of your amp?

As you see I'm really lost with that problems and could really use some help 
Maybe I find a way to upload some audio here, so you can hear exactly what i mean.

[pokus]

Ok I think I can close this thread now. It works!!

For everyone who is sharing interest, as mostly it was some of the easiest and last things I had thought of.
The coupling caps were just too big, so the ICs had problems to charge them. I put in some way smaller values and now it works just fine. No oscillation, no noise gate.

Thanks to everybody who has helped me on my journey!   
Really hope I can leave this closed now.