Connect 2 ch preamp (Reckless Abandon) through FX loop? And it squeals on BBoard

[matopotato]

I just finished breadboarding the Reckless Abandon: a 2 channel guitar preamp pedal made by @jonny.reckless (https://www.diystompboxes.com/smfforum/index.php?action=profile;u=20479) and described in this post: https://www.diystompboxes.com/smfforum/index.php?topic=125355.msg1193103#msg1193103.
It sounds very promising and I am excited about getting it soldered and boxed.

Being a 2 channel pedal I thought I could add a few switches to allow for: (neither A nor B), A, B, A=>B or B=>A options (see https://www.diystompboxes.com/smfforum/index.php?topic=129478.msg1251520#msg1251520 for a discussion on that) but it got way louder when stacked and the sound was not any better IMHO.
Might be curable with some components for the stacking cases, but I no longer think it is worth it.

I have two questions:
1. There is a high pitched squealy kind of (bi-)sound to it. I am hoping that after soldering, grounding and boxing everything properly this will be resolved.
But if anyone has any idea as to what the reason might be and possibly a remedy, I'd be very happy to try.
It squeals just as much on both channels, a bit less if bypassing both channels. It is fairly consistent with regards to knob turning, but no configuration takes it out though.

2. I tried it first into the Input of my amp, and later straight into the return of the FX, and it is clearly meant to be used as the pre-amp it is right into the power stage.
I have a pedalboard in front, and one in the FX loop. How can I add the Reckless Abandon to this mix in a useful way?
I am guessing one switch pedal for guitar going to front-board or to the Reckless. Then the Reckless from the Send, first in FX, through the FX board and back to return? Or another switch pedal so that only the Reckless is engaged in the FX? Like switching Reckless or FX loop to Return. Although that would mean two switches for that one pedal.
Ideally I would have liked it to work as any other pedal with all its mates in cooperation.

[amptramp]

I have a few thoughts on question 1.  This is a pedal with a huge amount of gain and breadboard wiring is notorious for feedback going all over the place.  Maybe it will clean up if you box it up but there are a few observations here:

The input stage has a gain of 11, the upper clipping stage has a gain of 0.4545 up to 45.9, the lower clipping stage has a gain of 4.545 up to 459 and the output stage has a gain of 30.3.  The worst-case combined gain is 1529847.  On a breadboard, this is inviting oscillation.

IC2 is used for both the input amplifier and the output amplifier.  This means the output after the product of all these gains will be on pin 7, just across and one down from the first stage input on pin 3.  Since there is a gain-bandwidth of 15 MHz, even small amounts of feedback are amplified at all frequencies.  You may find that high RF frequencies cannot be heard but will cause the dynamic range to be reduced because clipping will occur on the peaks of the oscillation riding on top of the audio.

There are no feedback lead capacitors across R19 and R20.  The capacitance to ground at the inverting input on both of these stages causes a feedback lag that has to be balanced by a feedback lead or the response will rise until it reaches the falling response of the op amp, causing a phase shift that can promote oscillation or at least a nasty transient response.  With a high bandwidth amplifier like the LM833, you can't afford weak bypassing capacitors or missing feedback lead capacitors.

Your VMID supply has a 100 µF cap for bypassing but electrolytic caps are notoriously bad for response at high audio frequencies and needs to be shunted by a film cap.

SW4B isn't really needed.  You can drive the clipping stages without having to select the inputs because the low output impedance of the first stage allows both stages to be driven, but the 220 ohm midband input impedance of the lower clipping stage is a bit too low - the LM833 is only characterized down to a 2K load.  With no input switching, you can delete R1, R6, R14 and C15.  The first stage output can go direct to C3 and C7.  Eliminating the switch and its leads may help.  R5 has to increase in value anyway and R2 should increase in value as well as this approaches the drive limit of the LM833.

[matopotato]

Thanks! A lot (for me) to take in a digest, so please bear with me...

I have a few thoughts on question 1.  This is a pedal with a huge amount of gain and breadboard wiring is notorious for feedback going all over the place.  Maybe it will clean up if you box it up but there are a few observations here:

The input stage has a gain of 11, the upper clipping stage has a gain of 0.4545 up to 45.9, the lower clipping stage has a gain of 4.545 up to 459 and the output stage has a gain of 30.3.  The worst-case combined gain is 1529847.  On a breadboard, this is inviting oscillation.

Yes, I think I could also hear this. Might be why there is little point in stacking the two stages, and also why it sounds better without the pre-amp of the amp itself.

IC2 is used for both the input amplifier and the output amplifier.  This means the output after the product of all these gains will be on pin 7, just across and one down from the first stage input on pin 3.  Since there is a gain-bandwidth of 15 MHz, even small amounts of feedback are amplified at all frequencies.  You may find that high RF frequencies cannot be heard but will cause the dynamic range to be reduced because clipping will occur on the peaks of the oscillation riding on top of the audio.

There are no feedback lead capacitors across R19 and R20.  The capacitance to ground at the inverting input on both of these stages causes a feedback lag that has to be balanced by a feedback lead or the response will rise until it reaches the falling response of the op amp, causing a phase shift that can promote oscillation or at least a nasty transient response.  With a high bandwidth amplifier like the LM833, you can't afford weak bypassing capacitors or missing feedback lead capacitors.

So should I add Caps between pins 1 and 2 and 6 and 7 of IC2? Or should they go in series with R19 and R20 respectively?
Any particular value to try?

Your VMID supply has a 100 µF cap for bypassing but electrolytic caps are notoriously bad for response at high audio frequencies and needs to be shunted by a film cap.

So I am guessing a Film Cap in parallel with C24? 100nF?

SW4B isn't really needed.  You can drive the clipping stages without having to select the inputs because the low output impedance of the first stage allows both stages to be driven, but the 220 ohm midband input impedance of the lower clipping stage is a bit too low - the LM833 is only characterized down to a 2K load.  With no input switching, you can delete R1, R6, R14 and C15.  The first stage output can go direct to C3 and C7.  Eliminating the switch and its leads may help.  R5 has to increase in value anyway and R2 should increase in value as well as this approaches the drive limit of the LM833.

OK, I am actually thinking of keeping the switch SW4B, but your suggestion is that is only really needed as SW4A? I thought both points were needed to avoid grounding points in the non used stage.
I am guessing R5 and R2 contribute to the character of the different stages, so if I want to keep them "different" and still raising them, what would you recommend me to try for new values?
Given that I actually do keep R1, R6, R14 and C15 as they are now.

Again, many thanks for this valuable input. Overall I kind of would like to keep what I can from the original character and if I can reduce the squeal with the ideas above, it would be welcome. If any smaller amount is left I will hope that final assembly helps.
Looking forward for some component values to try.

[matopotato]

Tried 2n2 across IC2 pin 6 and 7 so in parallel with R20 (R19) and the squeal went away (I think, there were several other things happening at the same time...), but at the expense of a very muddled tone. So I tried 100pF and no effect on the squeal. 330pF a slight effect, but tonal impact as well...
Not sure if the 100k should be in series instead and if both 1-2 pins and 6-7 need to have a cap at the same time...

[PRR]

...Might be why there is little point in stacking....

Amplifying audio is like enlarging an image. What happens if you blow-up an image over and over a million times bigger?

This is Anne(?) portrait size, 10X close-up, 30X close-up.


I tried 300X but my image program choked. I snuck up on it, but even at 300X there was hardly any Anne, and at 3,000X I just had blops of grain.


Did Santa bring you or your kid a telescope? Put it on the longest objective and the shortest eyepiece with a barlow boost in the middle. You really can't see anything. Microscope too.

Some gain is good. Gain upon gain upon gain is usually awful.

[matopotato]

...Might be why there is little point in stacking....

Amplifying audio is like enlarging an image. What happens if you blow-up an image over and over a million times bigger?

This is Anne(?) portrait size, 10X close-up, 30X close-up.


I tried 300X but my image program choked. I snuck up on it, but even at 300X there was hardly any Anne, and at 3,000X I just had blops of grain.


Did Santa bring you or your kid a telescope? Put it on the longest objective and the shortest eyepiece with a barlow boost in the middle. You really can't see anything. Microscope too.

Some gain is good. Gain upon gain upon gain is usually awful.

Point taken. I tried, discovered and abandoned the idea in this build as mentioned earlier.
My inspiration or idea came from King of Tone and similar dual pedals. I thought it/they were 2 gain stages that can operate each on its own, or stacked. If you want reversed order of stacking you need to flip some DIPs on the inside of a KoT. For KoT there are several variations, but not all combinations might not be available. (I have a RullyWow clone).
I was thinking that perhaps some limitations when stacking the Reckless might be needed, although KoT manages quite well the A, B or stacked options as far as I can tell.
So perhaps gain stage stacking might work in some cases after all.
And no, I do not have any children.

[amptramp]

I'm not going to attempt to quote your post that responded to mine because I usually don't get the quotes to line up, but here are some further comments from the tortured mind of a retired engineer:

1. If you have an input stage gain of 11, one volt of input will pretty much drive the signal to the maximum the op amp can deliver.  Most guitars can deliver this.  Many other clipper pedals work perfectly well with no gain stage ahead of the clipper.  Since you are clipping from the inverting input, you do need a first stage to keep the input impedance of the clipper at a reasonable level.  You already have a bandshaping network consisting of R11, R18, C17 and C18 and SW2, if you want to keep these values, R19 has to go down.  Otherwise, R11 and R18 have to go up and C17 and C18 have to go down by the same proportion.  It may be better to alter the bandshaping network and keep R19 high since it is one of the loads on the op amp output.  The op amp is only rated to drive 2K and this is the parallel value of R19, R14 and either R1 and R2 or R5 and R6.  R2 is a little bit too low and R5 is way too low to be driven by an LM833 because the input to the clipping stages is a virtual ground set to VMID.  The clipping op amp stages will drive the output so that the inverting and non-inverting voltages are the same, meaning the inverting input appears to be a fixed voltage.

When you raise R2 and R5, you can afford to have C2 and C7 go down by the same proportion.  The reduced leakage of the lower value capacitors means that R1 and R6 can be increased, reducing the load on the first stage.  An input stage gain of 2 would be adequate.  Even with the tone stack losses, the output stage gain of 30.3 is more than you need.  A gain of 11 should be adequate.

2. The feedback lead caps go in parallel with R19 and R20.  What you are trying to do here is counteract the input capacitance to ground from the inverting input such that the capacitance to ground times the input impedance (which includes the source impedance in parallel with R15 and R16 in series with R17) equals R19 times its feedback lead capacitor.  Usually, this is on the order of 10 pF to 100 pF.  This may not have eliminated the squeal but it will be necessary anyway since a high-frequency oscillation riding on the audio, when clipped, will compress the output because the clipping will occur on the high-frequency signal but the audio output will be the average of the audio and the clipped oscillation.  As an example, suppose you have a ±1.5 volt audio signal and one volt of oscillation superimposed on it.  When you clio with red LED's as shown, the clipping level will be about 1.88 volts.  It should not clip at all.  But the audio signal is actually added to the oscillation so the clipping will start once the audio goes above 1.38 volts because the oscillation goes ±0.5 volts above and below the audio and the audio output that you hear will begin to compress at 1.38 volts.  You will notice it but you would need a high-speed scope to actually see what is happening.

3. Yes, you do need a film cap across C24, the VMID filter.  You also need ceramic capacitors across C5 and C19, the op amp bypass capacitors because most electrolytics have a minimum impedance in the audio range then the impedance goes up above that frequency, meaning it acts as an inductor, not a capacitor.  You need a film or ceramic (or both!) capacitor across the electrolytics to keep the impedance low.  You have 80 db minimum and 115 db typical of power supply rejection at low frequencies in an LM833 but you have over 120 db of gain and the rejection drops off to under 80 db at 100 KHz.


Are you ready to carry on down the rabbit hole?

[matopotato]

I'm not going to attempt to quote your post that responded to mine because I usually don't get the quotes to line up, but here are some further comments from the tortured mind of a retired engineer:

Thanks, I really appreciate it. And I agree that the editor used in this forum might allow for some level of improvement.
I was just about to write about my testing thus far.

1. If you have an input stage gain of 11, one volt of input will pretty much drive the signal to the maximum the op amp can deliver.  Most guitars can deliver this.  Many other clipper pedals work perfectly well with no gain stage ahead of the clipper.  Since you are clipping from the inverting input, you do need a first stage to keep the input impedance of the clipper at a reasonable level.  You already have a bandshaping network consisting of R11, R18, C17 and C18 and SW2, if you want to keep these values, R19 has to go down. 

For the sake of the squeal or to bring the whole circuit down to a more reasonable level? Considering all the gain done all over the place...

Otherwise, R11 and R18 have to go up and C17 and C18 have to go down by the same proportion.  It may be better to alter the bandshaping network and keep R19 high since it is one of the loads on the op amp output.  The op amp is only rated to drive 2K and this is the parallel value of R19, R14 and either R1 and R2 or R5 and R6.  R2 is a little bit too low and R5 is way too low to be driven by an LM833 because the input to the clipping stages is a virtual ground set to VMID.  The clipping op amp stages will drive the output so that the inverting and non-inverting voltages are the same, meaning the inverting input appears to be a fixed voltage.

When you raise R2 and R5, you can afford to have C2 and C7 go down by the same proportion.

You mean C3 and C7? (Or C2 and C8)

The reduced leakage of the lower value capacitors means that R1 and R6 can be increased, reducing the load on the first stage.  An input stage gain of 2 would be adequate.  Even with the tone stack losses, the output stage gain of 30.3 is more than you need.  A gain of 11 should be adequate.

I have a little bit of electronics in my background, but not enough to have a feel for juggling the values. I am reading up on this as well, but it is a bit of a climb. So I mostly revert to picking up kits, sometimes adding a mod or two for improved control and versatility. (That I seldom use once dialed in...). So I am learning by experience and adding on theory as I climb in and out of holes. From rabbits, moles, groundhogs, meercats, termites... Feels like I've been down all sorts. By now I am mixing between kits vs schematics to vero-boards. I like to try out on a breadboard to see if the pedal is worth building, since I am literally running out of real estate.
So it will take me some time to get to grips with this, and I might not be ready to take this all the way. I am torn between the original design, that worked well for the original designer vs re-designing almost something new and different.
I plan to use this as a pre-amp and bypass the built in pre-amp in a Marshall DSL40CR. So directly from guitar to the pedal to FX board and FX return. Not sure if it matters with regards to gain levels or not.
Based on the earlier reply I experimented a bit. Mainly with R2 and R5. First I went with R5=> 2k7 and R2 => 27k in an attempt to keep their ratio and raise the 220R one up to the 2k area. (I now realize it is a bit more complex than that perhaps).
But this kind of took away most of the gain on the upper "R2" stage. So I started down to 2k2 and 22k but not much difference.
Then I thought to put in a trim pot instead for both (R2:20k and R5:2k5) and dialed them in to about 10k and 1k respectively. Here I felt that gain had a more useful interval on respective gain pot. And then I though, why not keep them in for the final build, then I can adjust if need be later on.
Still the volume(s) are hard to turn up, and the gain(s) felt like the last 25% or so didn't change much. So I thought to reduce the overall gain by changing to 50k instead, and then I tried Log and Lin and feel that Lin makes more sense. So I decided to trade the A100k gain pots for B50k ones.
And so I went to write in the thread, and now it is all a bit scrambled in my head...

2. The feedback lead caps go in parallel with R19 and R20.  What you are trying to do here is counteract the input capacitance to ground from the inverting input such that the capacitance to ground times the input impedance (which includes the source impedance in parallel with R15 and R16 in series with R17) equals R19 times its feedback lead capacitor.  Usually, this is on the order of 10 pF to 100 pF.  This may not have eliminated the squeal but it will be necessary anyway since a high-frequency oscillation riding on the audio, when clipped, will compress the output because the clipping will occur on the high-frequency signal but the audio output will be the average of the audio and the clipped oscillation.  As an example, suppose you have a ±1.5 volt audio signal and one volt of oscillation superimposed on it.  When you clio with red LED's as shown, the clipping level will be about 1.88 volts.  It should not clip at all.  But the audio signal is actually added to the oscillation so the clipping will start once the audio goes above 1.38 volts because the oscillation goes ±0.5 volts above and below the audio and the audio output that you hear will begin to compress at 1.38 volts.  You will notice it but you would need a high-speed scope to actually see what is happening.

I understand it will be hard for me to detect the direct impact of 10pF - 100pF, so I really would appreciate a recommendation and add my trust to that and go with it. I have a few 100pF and 2 of 82pF I could use. Or should they be of different values considering the input capacitance and impedance seem to differ between the "pin 1 and 2" location and the "pin 6 and 7" location.

3. Yes, you do need a film cap across C24, the VMID filter. 

I am guessing add a 100nF film in parallel with the C24 when comparing with other builds. This seems to be a common enough combination in the power stage albeit at the VDD in most cases.

You also need ceramic capacitors across C5 and C19, the op amp bypass capacitors because most electrolytics have a minimum impedance in the audio range then the impedance goes up above that frequency, meaning it acts as an inductor, not a capacitor.  You need a film or ceramic (or both!) capacitor across the electrolytics to keep the impedance low.  You have 80 db minimum and 115 db typical of power supply rejection at low frequencies in an LM833 but you have over 120 db of gain and the rejection drops off to under 80 db at 100 KHz.

I think the C19 and C5 are not electrolytics. According to @Johnny.Reckless:

Perhaps more surprisingly, the value of the decoupling cap on the op amp power supply pin affected the sound quite a lot. I ended up with 10uF multilayer ceramic caps there rather than the standard 100nF, which tamed the shrill harshness I was getting, and seemed to increase the amount of low end warmth.

So I managed to get such multilayer ceramic caps. (not too cheap IMHO...)

Are you ready to carry on down the rabbit hole?

Sort of...?
I liked the way it sounded in the clip made in the original thread, and I am adjusting a few things to fist try to reduce the squeal until boxing and soldering. And to some extent try and tame the enormous gain without changing its character too much.

Either way, I really appreciate you taking your time explaining and suggesting what to try.
Again, thanks!

[matopotato]

Did some more testing on R19 and R20. Put in two trimmers, and the squeal is reduced, but also the volume and gain, so it might not be a reduction as such. Yet.
I found R20 works with 20k and R19 on 50k.
I can't say much of the effect's tone is changed, besides the volume and gain is reduced.
For me to play around with the other pre- and post-gain shaping R's and C's is quite a challenge unless I get some specific values to try out.

I also gave my second question some more thought, how to introduce a pre-amp pedal, that does not approve of any other pre-amp in the setup.
I realize I will need one AB splitter, but it seems I would need two. Unless I am missing something.
The AB works "both ways".

[amptramp]

From your reply:

You mean C3 and C7? (Or C2 and C8)

Yes, I mean C3 and C7.  C2 and C8 seem to be OK.

[matopotato]

Thanks, any thoughts on the ceramic caps for the pin 1-2 and pin 6-7 on IC2? If it is not very audible, I would be glad to learn what value you would have picked.

About the C3 and C7, If I increase R2 and R5 about 5 times(-ish) what is the effect/consequence if I do not compensate for C3 and C7?

[matopotato]

...
About the C3 and C7, If I increase R2 and R5 about 5 times(-ish) what is the effect/consequence if I do not compensate for C3 and C7?

OK, I tried it and although I did not turn any knobs, I couldn't hear much of a difference. Perhaps ever so slightly smoother with Caps compensation.