Sidechain splitter/buffer - clean effects mixer

[Phoenix]

Hi everyone.

Thought I'd share something I've been working up. It'll buffer and split your signal so that you can mix a dry tone with another external effect.
It's a splitter/buffer and mixer with differential eq on the dry/wet paths, so when you turn the tone control in one direction, the dry signal is more bass, less treble, and the wet signal is more treble, less bass, and vice versa when turned the other way, with both paths flat with the control at 12 o'clock. This means that you can mix a distorted treble with a clean bass or similar, which should be particularly useful for bass.
There's a switch to make this eq pre-or-post send.
The center frequency of the eq can be switched between 3 different settings (I haven't actually bread-boarded this yet, so they're just guesses about what might sound good).
There's a phase switch to avoid problems with other effects.
Given that I think this will most likely be used with distortions/overdrives/fuzzes, and most of them have a fair bit of boost, there is a clean gain  control on the dry path to help with balancing the volume of the two signals without having to heavily attenuate the wet path.
The last control is a blend using an MN taper pot, so at the center detent both paths will be full volume, and turned in either direction it will attenuate one of the paths.

This has suffered a bit of feature creep, and so it needs 7 op amps as it currently stands, and I haven't figured a way to get it down to 6. Two quad packages leave me with a spare, so may as well put it to use as a Vref buffer. If anyone has any ideas to ditch an op amp, or any other comments, that would be greatly appreciated!

[garcho]

nevermind

[TejfolvonDanone]

The output impedance is really low for both send and out. You might want to consider to have a high impedance output for send to be able to blend in a low input impedance circuits (like a Fuzz Face) with their best sound.
Also the EQ seems to attenuate a little bit even when it's flat so you might want to compensate with the buffers U1C and U1D.
All in all great work!

[Phoenix]

The output impedance is really low for both send and out. You might want to consider to have a high impedance output for send to be able to blend in a low input impedance circuits (like a Fuzz Face) with their best sound.

Thanks for the suggestion, but I think that comes down to the same problem as using any buffered effect before a low impedance circuit like a Fuzz Face, I think rather than add more feature creep to this, you'd be better off using something like the AMZ pickup simulator if you're planning on using a low input impedance pedal. You still don't get to keep the guitar's volume control interaction with the effect, but at least it'll mostly work.

Also the EQ seems to attenuate a little bit even when it's flat so you might want to compensate with the buffers U1C and U1D.
All in all great work!

Ah yes, I'd forgotten about that in my haste. I've toyed around with a few versions of this concept before and always remembered to add some makeup gain for the tonestack, but this is the first time I've thought of the pre/post send eq option. I've updated the schematic, but I think I'll probably end up changing the location of the makeup gain on the dry signal path to the input op amp rather than post eq for better noise performance. On the wet path I think the makeup gain should stay with the op amp immediately following the eq so that the path is "unity gain" regardless of whether the eq is pre or post send. Really need to get this on a breadboard. Also probably need to add some more AC coupling and bandwidth limiting, that'll come next...

Thanks for the comments!

[TejfolvonDanone]

This has suffered a bit of feature creep, and so it needs 7 op amps as it currently stands, and I haven't figured a way to get it down to 6.

I think you can use only one of the two input buffers if you make sure that you don't exceed their output current capacity. In this case you can get rid of U2D and you end up with 6 op-amps. But you have to decrease R23 and R24 a little more so your reference voltage will be more stable. I would stick with the 2x4 op-amp configuration it doesn't seem to be such a big problem to have a stable reference. As Archimedes said: "Give me the place to stand, and I shall move the earth." Also you can't put the make-up gain before the dry tone stack.

A thing about the EQ: through C18/C5 (if the switch SW2B is in the left position) or C19/C4 or both the to signal path are linked. If the EQ is post send and you have a really high output return signal this might be troublesome.

I can't really find the two 2V Zeners' purpose in the feedback loop of U1D. Looks like limiting the signal but i don't really know why you would do that. Can you please explain it to me?

[Phoenix]

I think you can use only one of the two input buffers if you make sure that you don't exceed their output current capacity. In this case you can get rid of U2D and you end up with 6 op-amps.

The idea behind the parallel input buffers is so that if the eq is set to post-send, the response of the eq on the dry side won't be effected by the connected effects input network. If say a fuzz face was connected and only a single input buffer were used, the eq on the dry path would be in parallel with the input network of the connected effect, which would wreak havoc.

But you have to decrease R23 and R24 a little more so your reference voltage will be more stable. I would stick with the 2x4 op-amp configuration it doesn't seem to be such a big problem to have a stable reference. As Archimedes said: "Give me the place to stand, and I shall move the earth."

I know, those values are only so high because Vref is buffered, so may as well minimize current draw. If Vref were unbuffered the DC resistance of the voltage divider would need to be lowered.

Also you can't put the make-up gain before the dry tone stack.

Well maybe "make-up gain" is technically the wrong term in that case (better called pre-gain), but with parallel input amps I don't see any reason why I can't include some gain on U1A and not U1B.

A thing about the EQ: through C18/C5 (if the switch SW2B is in the left position) or C19/C4 or both the to signal path are linked. If the EQ is post send and you have a really high output return signal this might be troublesome.

That's true, I hadn't considered that. May need to change SW1 to a 4PDT or 4 pole rotary.

I can't really find the two 2V Zeners' purpose in the feedback loop of U1D. Looks like limiting the signal but i don't really know why you would do that. Can you please explain it to me?

Yes, that's exactly my intention. My choice of gain pot is based on an assumption of a minimum 100mV peak input signal, to take full advantage of the voltage headroom. However most pickups will put out more than that, and guitarists are wont to abuse things, so I'd rather (relatively) graceful diode clipping than ugly rail clipping if the gain pot is at max with hot pickups.

Thanks for the input, keep it coming!

[TejfolvonDanone]

Yes, that's exactly my intention. My choice of gain pot is based on an assumption of a minimum 100mV peak input signal, to take full advantage of the voltage headroom. However most pickups will put out more than that, and guitarists are wont to abuse things, so I'd rather (relatively) graceful diode clipping than ugly rail clipping if the gain pot is at max with hot pickups.

In that case you limit the available input range too much. So much you really don't have any headroom (which is the nominal input minus the maximum input without clipping). With 200mV peak-to-peak (100mV peak) input you can have maximum 26 gain which is exactly at the point of clipping with the Zeners. In fact it might clip taking the tolerances into account (the Zeners and the R5 is smaller than the nominal and the pot is bigger). But as you said it's less than what a normal guitar output and you will end up with clipping the dry path with near maximum boost settings. (I'm assuming that's not what you intended.)
So you want to create an idiot proof device. You have to take the BIGGEST signal you expect and offer a circuit which WILL NEVER clip at that level. So you expect a 100mV peak input and you offer 26 gain you have to create an amp which can output MORE than 2.6V without clipping. With 9 V supply you will have a rail clip at about 3.5V. Changing the Zeners to at least 2.4V won't do any harm. Also you could just dial back the gain to 21 (200k pot). With it you can have a more standard pot values (the 1-2-5 for every decade) and reasonable headroom. Both decreasing the gain and increasing the Zener voltage would have the best result.
Remember that Zeners work like regular SI diodes when forward biased with about 0.6-0.7V forward drop (at these current level).

The idea behind the parallel input buffers is so that if the eq is set to post-send, the response of the eq on the dry side won't be effected by the connected effects input network. If say a fuzz face was connected and only a single input buffer were used, the eq on the dry path would be in parallel with the input network of the connected effect, which would wreak havoc.

Well maybe "make-up gain" is technically the wrong term in that case (better called pre-gain), but with parallel input amps I don't see any reason why I can't include some gain on U1A and not U1B.

I hadn't consider the low impedance input devices in this case. Point taken.

Putting (as you noted correctly) pre-gain to the U1A without paralel input buffering makes your send signal slightly larger than you would have with paralel buffering or with the make-up gain. It wouldn't be devastating but isn't necessary.

That's true, I hadn't considered that. May need to change SW1 to a 4PDT or 4 pole rotary.

Or a FET switching system with a less complicated (and less expensive) switch.

[Phoenix]

In that case you limit the available input range too much. So much you really don't have any headroom (which is the nominal input minus the maximum input without clipping). With 200mV peak-to-peak (100mV peak) input you can have maximum 26 gain which is exactly at the point of clipping with the Zeners. In fact it might clip taking the tolerances into account (the Zeners and the R5 is smaller than the nominal and the pot is bigger). But as you said it's less than what a normal guitar output and you will end up with clipping the dry path with near maximum boost settings. (I'm assuming that's not what you intended.)
So you want to create an idiot proof device. You have to take the BIGGEST signal you expect and offer a circuit which WILL NEVER clip at that level. So you expect a 100mV peak input and you offer 26 gain you have to create an amp which can output MORE than 2.6V without clipping. With 9 V supply you will have a rail clip at about 3.5V. Changing the Zeners to at least 2.4V won't do any harm. Also you could just dial back the gain to 21 (200k pot). With it you can have a more standard pot values (the 1-2-5 for every decade) and reasonable headroom. Both decreasing the gain and increasing the Zener voltage would have the best result.
Remember that Zeners work like regular SI diodes when forward biased with about 0.6-0.7V forward drop (at these current level).

I'm not exactly certain I understand your point... If I've got an input of 100mV peak, with the gain control turned to its maximum of A=26, the output should be just at the threshold of clipping as set by the zener diodes. If the input signal is larger than 100mV peak, then you wouldn't turn the gain control up to its maximum. But, if you turn it up anyway, it'll diode clip rather than rail clip. You're probably right that I can increase the zener voltage, as I mentioned I haven't breadboarded this yet, the 2V was just a conservative "placeholder" choice.

By the way, 250k pots are exceedingly common, at least in my part of the world, much more common (and cheaper) than the closest preferred number value of 220k, which is much more commonly available in Europe than the US or Australia, where 250k is the norm. 200k in my experience is rarer still than either 220k or 250k.

Putting (as you noted correctly) pre-gain to the U1A without paralel input buffering makes your send signal slightly larger than you would have with paralel buffering or with the make-up gain. It wouldn't be devastating but isn't necessary.

To clarify, what I was saying was that using the parallel input buffers, I could put pre-gain in the dry path at U1A, unity gain buffer for U1B, and makeup gain for the wet path at U2A. Then both paths would be unity (ignoring any gain or attenuation between send and return) when the gain control on the dry path is set to 0.

Or a FET switching system with a less complicated (and less expensive) switch.

A 4 pole 3 position rotary switch is pretty cheap too...

[TejfolvonDanone]

Looks like what is cheap and not is really different where we live. 
I can find pot with 1-2-5 for every decade for less than 1$ and they don't even sell any other value.

I'm not exactly certain I understand your point... If I've got an input of 100mV peak, with the gain control turned to its maximum of A=26, the output should be just at the threshold of clipping as set by the zener diodes. If the input signal is larger than 100mV peak, then you wouldn't turn the gain control up to its maximum. But, if you turn it up anyway, it'll diode clip rather than rail clip. You're probably right that I can increase the zener voltage, as I mentioned I haven't breadboarded this yet, the 2V was just a conservative "placeholder" choice.

If you have 100mV peak input you almost not clip it at maximum gain. But you might in the worst case scenario with the component tolerances.
The purpose of the booster is to increase the volume of the dry signal but don't alter it in any other way. In order to this can be achieved you have to set a nominal voltage with which it will work fine. The nominal voltage is the level of input signal you think the device will get (for a guitar let's say 200mV peak). You also have to set a headroom which means if you increase the nominal input by the headroom you get also just a boost and no clipping. This has to hold true for every possible setup of the booster.
Considering that someone will end up using it with a high output pickup 100mV peak nominal isn't the best because depending on what kind of guitar the user has your full range of controls aren't accessible or work deferently. With a low output pickup you have just a booster and with a high output pickup you get a distortion but less boost.

So leaving the Zeners completely you would end up with a booster which still can't handle a 200mV peak signal. In order to be able to handle a signal like that you also have to decrease the gain to 17.5. You might get away with 20 if the op-amp doesn't clip at 3.5V. You can't be sure about that. But if you are closer to the rail clip point the less you will get that diode clip.

TL;DR: A clean booster have to boost and shouldn't do anything else. Operation outside the nominal range (+ headroom) should give you clipping. It doesn't have to be a "nice" clipping because you operate where it shouldn't be operated. If you plug in 230VAC straight form the wall wart into the input you shouldn't be surprised that you get a toaster instead of a clean boost.

So my point is you set the goal to 200mV leave the Zeners the gain to 16 (maybe 11) and you still get about 33mV input headroom (which is 16.5% of the nominal). No clipping in the this region and it covers all of the guitars.

[Phoenix]

Yes, what is available to us is very dependent on location.

Ok, you're talking about a fundamentally different design philosophy than I follow with regards the booster.
Your philosophy is that a booster should not be able to clip under any circumstances, so you would minimize the range of gain to suit the largest expected input signal.
My philosophy is that a booster should be able to make full use of its headroom even for the smallest reasonably expected input signal. So if the input signal is only 100mV peak (200mV peak-to-peak), and the available headroom is 3.5V peak (7V peak-to-peak), then there must be a gain of 35 available. This does mean that if a higher input signal is seen, that the gain control will run out of effective range before reaching maximum, and as guitarists are wont to abuse things like gain controls, when it DOES overload (and guitarists will always find a way to overload inputs), it should clip in a graceful manner. Diode clipping is much more graceful for most opamps than rail clipping and possible signal inversion and recovery time. especially so with a TL07x. Some rail-to-rail op amps and other specialist examples do behave well when exceeding their rails, but I'm not using them here, so is a moot point.

Being that guitar pickups can range in output voltage from around 100mV peak to 2V peak, taking your design philosophy to the extreme, you'd only be able to include a gain of 1.75 for the guy with active pickups intended for metal styles to be absolutely sure that the dry path is never overloaded, which wouldn't be very useful to someone who wanted to use a vintage Strat with weak magnets and extra-light gauge strings. I think it's reasonable to expect that if someone doesn't want clipping, they can turn down the gain control.

Anyway, I understand your position, but I can tell you know that we are not going to see eye-to-eye on it. We'll just have to agree to disagree.

[TejfolvonDanone]

Instrument level is -20 dbu. That is a little more than 200mV peak. As a guitar effect builder i expect instrument level. 2V input is not what i expect. So i create the effect to be able to handle the 200mV signal (and a little more) and not the 2V.
My philosophy is more like i set a maximum input and any control would work the same in that range. This means yes my available gain would be smaller.

The problem with your philosophy is that you say it's a booster everybody expects it to boost and the same amount for every guitar. What if i need 26 boost to my high output guitar? It's not reasonable if you offer 25 gain but not for everyone. As a user when someone says "this offers x gain" and i grab my active pickup guitar (which still doesn't put out a lot more than 200mV) and i end up overdriving the booster at around 12 of the gain knob i will be really angry.

I'm not forcing you to do what i say and you have to follow my philosophy. So we will just disagree on this i think.

BTW i have a phaser and my active guitar overdrives it when i play really hard. Sometimes i just curse the designer. Sometimes i just abuse it and push it really hard (with even putting a booster before it). It has a rail clipping and it can be useful.

[garcho]

The problem with your philosophy is that you say it's a booster everybody expects it to boost and the same amount for every guitar.

Says who? I don't expect ANYTHING to be the same for every guitar, i don't think i know any guitar players who do.
A "booster" pedal is not a Neve console or a Quad power amp or a Primare preamp. Why would a guitar player need to know the exact gain for a "booster"? Why would it have to be the same for every guitar? Do you keep the exact same rig but only change your guitar - between active and passive - at the same gig? How many guitar players do that? How many guitar players need to do that? There's usually a more graceful, artful solution to a musical problem than throwing gear at it.
I think you're making good points AND getting bogged down in EE issues. This is do-it-yourself-stomp-boxes, not do-it-for-other-people-stomp-boxes. If it works for you, then great! If it doesn't work for the other guy, the other guy should buy a breadboard and learn how to read a schematic. 

I'm not forcing you to do what i say and you have to follow my philosophy

how would anyone be able to do that? 

[karbomusic]

But, if you turn it up anyway, it'll diode clip rather than rail clip.

Yep, I've done that... The LEDs clip just before the opamp does at 9VDC PSU.



The switch is there so I can take the diodes out of the picture. IMHO, when it comes to a clean boost one should be able to run at a higher voltage anyway such as 12VDC or even 18VDC because 9VDC is pretty limiting comparatively when we understand that the initial transient of a hard hit passive guitar note will hit up to ~1.2 volts. However, any boost would sound better if it breaks up gracefully at its limits. I like the Neve reference because what makes them so great is how they overdrive and sound when pushed too hard; also gain knobs exist for adjusting to the conditions, like guitars with varying outputs.

[TejfolvonDanone]

how would anyone be able to do that? 

In the true and tested way: first insult your intellectual capabilities second include insult with the other's mother maybe with a sexual intercourse then use caps lock. The bonus is to correct the other's grammar ignoring your mistakes. Works every time 

I think you're making good points AND getting bogged down in EE issues.

You are right. I tend to do that. But this approach somehow really bugs me. And i assumed that not the OP would use it. Which is also weird to say the least.

Do you keep the exact same rig but only change your guitar - between active and passive - at the same gig?

I do that. Modern LP style guitar with active pickup + a hollow body with P90 and a humbucker with series/split/paralel switch. But i don't gig a lot.