Twin Peaks harmonic tremolo... in stereo?

[aion]

I've been reading up on the Twin Peaks harmonic tremolo and it sounds awesome. Ever the dreamer, I had to wonder: what would it take to add a second tremolo channel that's out of phase with the first one, so it could pan between them in stereo? Has anyone experimented with splitting and inverting the PWM of the tap tempo controller? I know there was some conjecture about it awhile ago, but I didn't know if anyone had actually tried it yet.

Anyway: one tap tempo, one set of knobs, but two outputs. If you only use one of the outputs then it behaves normally as in the original unit, but otherwise the LFOs are 180 degrees out of phase so it has the effect of panning between the two.

It could be wired up as either two independent channels (2 inputs, 2 outputs) with true bypass, or 1 in, 2 out with buffered bypass (in bypass, it still splits the signal like a Boss stereo pedal). I myself would probably not use it as a splitter since my delay already does that, but 2 in/2 out would be great.

Possibly it could support a toggle switch that would allow you to link the input of the first channel to the second so you only needed to supply one signal. This switch would have to force buffered bypass at the same time, which I think could be done via a 3PDT toggle (one pole to link the channels, two poles to move the bypass from the input jack to after the input buffer in both channels)

Besides the PWM inverting circuit and the second trem channel, it's just a matter of working out how to handle the bypass. Possibly two AMZ microcontrollers could be used, with one switch daisy-chained to two boards for 4PDT plus LED. Or a hardware 4PDT with millennium bypass if you don't want to go the relay route.

I'm thinking that tap-tempo panning could produce some very interesting sounds, especially if it's just subtle harmonic panning and not the full signal.

Would anyone else have any interest in this type of project? If I could get some help with the PWM inverter and someone could verify the schematic, I'd take care of designing the PCB and setting up a group buy.

[midwayfair]

This is actually fairly easy.

It's already out of phase -- that's what all that is around U3.

Take the output of U2B and instead of taking it to U1B, create another separate op amp stage that duplicates all the U1B stuff. That's your second "out" channel. You can fairly easily switch between the stereo and mono output simply by moving that connection back to U1B again.

[aion]

I do like the idea of not having to duplicate the LED/LDR stuff. That will work in each switch setting, even the standard tremolo?

What about for the 2 input, 2 output configuration? Would you also duplicate the input stage (C2 and earlier) and connect the R13 portion to that instead of C2?

[midwayfair]

What about for the 2 input, 2 output configuration? Would you also duplicate the input stage (C2 and earlier) and connect the R13 portion to that instead of C2?

That's a bit more complicated, because of the crossover frequency stuff going on in the middle of the circuit. Won't work without a bunch of changes.

[samhay]

aion - As you probably know, harmonic tremolo works by splitting the signal into a bass and treble component and then tremolo'ing each component 180 o out of phase with each other. A simpler stereo implementation that might be interesting would be to send the treble to one channel and the bass to the other (with the associated out-of-phase tremolo'ing). In the case of the Twin Peaks circuit, I don't think this would add any active components, and you could still sum them to get 2 conventional mono channels.
That may be what midwayfair was suggesting.

[aion]

What about for the 2 input, 2 output configuration? Would you also duplicate the input stage (C2 and earlier) and connect the R13 portion to that instead of C2?

That's a bit more complicated, because of the crossover frequency stuff going on in the middle of the circuit. Won't work without a bunch of changes.

Dang, I wondered about that. So are we back to the inverted PWM and a second audio path as the only way to accomplish the 2 in, 2 out setup?

[midwayfair]

What about for the 2 input, 2 output configuration? Would you also duplicate the input stage (C2 and earlier) and connect the R13 portion to that instead of C2?

That's a bit more complicated, because of the crossover frequency stuff going on in the middle of the circuit. Won't work without a bunch of changes.

Dang, I wondered about that. So are we back to the inverted PWM and a second audio path as the only way to accomplish the 2 in, 2 out setup?

Like I said, you already have the inverted PWM in the pedal. That doesn't change. You asked about adding a second output -- to do that so that both sides sound the same, all you have to do is duplicate the output section and, instead summing them in stereo mode, the right channel goes toward its own output.

If you want stereo in, stereo out, you need to completely separate both paths. In that case, you might as well redo the entire audio path so that they're not connected by that crossover pot in the center. Just put a low pass on the right channel at 3000KHz and a high pass on the left channel at 3000KHz and it'll be pretty similar (I think that was DRolo's crossover frequency, or maybe it was Samhay's ...). Or just make a basic stereo tremolo and EQ the amps.

The way I did a stereo tap tempo tremolo with harmonic was to make two independent audio paths, and then put a switch to put a capacitor between them at the output to create the crossover (and also switch in a treble cut in one channel). It sounds like this:
https://soundcloud.com/jon-patton-3/cosmo-tremolo

(You have to listen in stereo or it won't sound like anything. Which incidentally is a problem with anything you record using a stereo/panner pedal ... if most of your audience is listening in mono, you're going to sound very boring.)

[samhay]

Nice sound clip Jon, but listening with headphones does induce a little bit of sea sickness - I'm not sure this effect is for everyone.

[drolo]

I had been thinking about how I could (easily) modify the circuit without adding too many parts.
The obvious way is to route each of the audio path to a separate out instead of joining them. That's easy. But as you already mentioned, this would work fine in harmonic mode but in normal tremolo mode, you would have bass on one side and treble on the other side tremoloing at the same time, not very stereo-y...

I had thought that instead of using the rotary switch to select what LDR gets what signal, one could have a switch that cuts the high and low filter caps out of the circuit. I have not taken this tought any further so there might be a lot of other cans of worms doing this (loud pops for example and other switching config issues ...) Unfortunately, there are always new projects coming in the way of me spending more time figuring this out ... :-)

[aion]

But as you already mentioned, this would work fine in harmonic mode but in normal tremolo mode, you would have bass on one side and treble on the other side tremoloing at the same time, not very stereo-y...

Yep, I wondered if that would happen when using it in non-harmonic mode. My goal is that the stereo version will work in all modes, so even if it's volume only (standard tremolo), it'll pan between the two outputs. Pointless if it'll be mixed down to mono, of course, but you can just use a single output if you're in a mono situation.

We already have the inverted PWM, but to make this work I imagine at a minimum we'd need to run a second set of LDRs and a duplicated audio path. Am I right that we'd just need to duplicate the LDRs on each of the outputs from U3A and U3B, but then reverse the rotary switch wiring for the second set of poles and send that to our second audio path? There wouldn't be any issues driving 4 LEDs instead of 2, right?

We'd probably need a dual pot (or dual-concentric) for the Balance control to keep the channels matched up, and the rotary switch would of course need to be 4 poles instead of 2. Tayda does have a 20k dual pot which should be fine for the Balance pot unless we need fine-grained control over both of the channels, which I wouldn't think would be necessary.

Sorry I'm a little slow to grasp all this - I'm pretty new to LFOs and modulation-based stuff in general. My experience is solidly in the much-less-complex overdrive and compression fields. Gotta learn somehow, though.

[drolo]

Am I right that we'd just need to duplicate the LDRs on each of the outputs from U3A and U3B, but then reverse the rotary switch wiring for the second set of poles and send that to our second audio path? There wouldn't be any issues driving 4 LEDs instead of 2, right?

Nope, this will lead to issues. The OP amp will not be able to pull as much current. I already notice a difference in how much current it is able to provide if I change the 5532 I am using to something else. You notice a volume drop for example when switching from harmonic to normal trem, as the normal trem uses probably twice the current the harmonic mode uses. You would then have to use a transistor to drive at least each set of LEDs.

We'd probably need a dual pot (or dual-concentric) for the Balance control to keep the channels matched up, and the rotary switch would of course need to be 4 poles instead of 2. Tayda does have a 20k dual pot which should be fine for the Balance pot unless we need fine-grained control over both of the channels, which I wouldn't think would be necessary.

Well, if you would go that stereo route, you already have a quite complex pedal, you might as well have a balance pot for each section :-)

The balance pot is something I use a lot as it allows to shape the tone and feel of the unit. Sometimes I even turn down the depth completely and use the balance pot as an EQ :-)

[midwayfair]

Yep, I wondered if that would happen when using it in non-harmonic mode. My goal is that the stereo version will work in all modes, so even if it's volume only (standard tremolo), it'll pan between the two outputs. Pointless if it'll be mixed down to mono, of course, but you can just use a single output if you're in a mono situation.

My recommendation is still to do it the way I did in the Cosmo tremolo: Make the paths completely separate, full-range tremolo stages, and you make a switch to go to harmonic mode in the left channel only that does the following:
Left channel gets a treble cut at about 1KHz.
A capacitor creates a crossover frequency between the two outputs at 3Khz (or thereabouts).

There's not really a good way to get harmonic in both channels, and it's of questionable utility anyway. If you listen to the demo I posted, you can hear that you end up with filtery sounds in both channels when in stereo even though only the left channel has the correct cutoffs for harmonic tremolo, simply because you're still futzing with multiple signals. It sounds like four guitars.

We already have the inverted PWM, but to make this work I imagine at a minimum we'd need to run a second set of LDRs and a duplicated audio path. Am I right that we'd just need to duplicate the LDRs on each of the outputs from U3A and U3B, but then reverse the rotary switch wiring for the second set of poles and send that to our second audio path? There wouldn't be any issues driving 4 LEDs instead of 2, right?

I had a heck of a time driving three LEDs (one is the rate indicator). Drolo's setup is a bit more robust, so you might be able to drive four LEDs, but again, I'm not seeing the value.

Sorry I'm a little slow to grasp all this - I'm pretty new to LFOs and modulation-based stuff in general. My experience is solidly in the much-less-complex overdrive and compression fields. Gotta learn somehow, though.

No need to apologize, it's good that you want to learn about it. LFO effects are fun.

[aion]

Okay, here's my attempt at a schematic for the stereo version.

https://www.dropbox.com/s/98gglyzfi6bo13x/twinpeaks_stereo_schem.pdf

It's pretty deluxe, and includes the switching system based on the AMZ relay microcontroller. All the switching takes place on the PCB, so wiring should be a breeze, even though the PCB will likely be pretty big.

I'm waiting to hear back from Jack as to whether one microcontroller can trigger 2 relays in parallel, so I may end up needing to add a second microcontroller, but this is the optimistic scenario. I'm planning to design the board around a 125-BB enclosure, taller than the 1590BB so I can have enough room for 4 in/out jacks as well as an external tap input.

I have a couple of questions about it before I start designing the PCB.

1. Are my 2N5088 current followers hooked up correctly? Will these give it enough juice to power all four of the LEDs without any trouble?

2. Will the "Treble Only" and "Bass Only" modes work correctly with the way that I have the rotary switch set up? On Treble mode, the bass remains constant in both channels and the treble pans left to right, and the same on Bass mode? I have a hunch that something may be wrong here, but I don't know what or why. Originally I thought I needed to use a 4P4T rotary switch (one pole for each LED), but when I actually drew up the schematic I realized that it should probably only need one pole per PWM and I just need to invert the LED positions for the second channel. But I'm second-guessing myself now.

3. Am I taking the buffered bypass output from the right place - after the op-amp input buffer but before the 47n coupling cap?

4. Does the switching scheme make sense, and will it work as designed? The microcontroller triggers 2 relays for an effective 4PDT switch. The effect can be run in 2 stereo modes: 2-in/2-out, or 1-in/2-out. The 1-in/2-out mode must use buffered bypass, otherwise we're dealing with a passive splitter which is bad. So the 3PDT switch that links the channel 1 input to channel 2 also changes the circuit from true bypass to buffered bypass. In the schematic, the switches are shown in the true bypass, effect off positions.

5. As I'm designing the PCB, are there any considerations I should take? I'm planning to keep five mostly-isolated ground planes: for the power section, the microcontroller/relay section, the LFO/PWM section, and each of the channels. They'll be star-grounded at the input jack, or at least grounded together at a single pad on the PCB that will go straight to the input jack.

6. What opamps should I use for the audio path? The original schematic didn't specify, so I just threw NE5532s in there, but that may be overkill. Any reason not to use a TL074 quad for each channel to save a little bit of space?

[drolo]

I can't access dropbox here at work :-(.
I will have a look at your schematic this evening at home if I have some time. (unless you can host the image elsewhere, like Photobucket etc)

[samhay]

^I'm waiting to hear back from Jack as to whether one microcontroller can trigger 2 relays in parallel

If you buffer each of your relays with e.g. a BJT then you can drive as many of them as you want with a single uC.
http://electrosome.com/interfacing-relay-with-pic-microcontroller/

[aion]

Here's another link for the schematic:
http://aionelectronics.com/schem/twinpeaks_stereo_schem.pdf

[aion]

If you buffer each of your relays with e.g. a BJT then you can drive as many of them as you want with a single uC.
http://electrosome.com/interfacing-relay-with-pic-microcontroller/

That looks like it's specific to non-latching relays needing a constant voltage in order to stay open. Jack's uses a latching relay which only needs a pulse for a few milliseconds in order to switch. Does that change anything?

I guess if the switching current draw is too high for a single uC, I could do another current follower just like I did on the LEDs.

[samhay]

That looks like it's specific to non-latching relays needing a constant voltage in order to stay open. Jack's uses a latching relay which only needs a pulse for a few milliseconds in order to switch. Does that change anything?

I don't think so. The BJT (or you can use a MOSFET) is acting as a switch to ground. When the uC output pin pulls the base (or gate) to +5V, the switch opens.

If you are using a latching relay - which doesn't sound like a bad idea - then you can switch it without a uC. Try googling 'mictester "A Switching Scheme" '.

[drolo]

1. Are my 2N5088 current followers hooked up correctly? Will these give it enough juice to power all four of the LEDs without any trouble?

This should be OK. I have done it on other occasions and worked fine. Might be worth breadboarding first to test it. In some situations it works better to have the leds on the collector side. Not sure what the math is to back it up though

2. Will the "Treble Only" and "Bass Only" modes work correctly with the way that I have the rotary switch set up? On Treble mode, the bass remains constant in both channels and the treble pans left to right, and the same on Bass mode? I have a hunch that something may be wrong here, but I don't know what or why. Originally I thought I needed to use a 4P4T rotary switch (one pole for each LED), but when I actually drew up the schematic I realized that it should probably only need one pole per PWM and I just need to invert the LED positions for the second channel. But I'm second-guessing myself now.

My first thought when using a bass only or treble only position in stereo would have been to have the fix part on one channel and the tremoloing part on the other. I think you may want to test these situations to see which make more sense and are musically the most useful. The same with the harmonic mode, as Jon mentioned, it might be sickening to have the separated channels oscillating out of phase on different amps. Sorry that my answers are so vague ... It's a bit difficult to imagine what it will sound like exactly. Only a test will give you this answer.

3. Am I taking the buffered bypass output from the right place - after the op-amp input buffer but before the 47n coupling cap?

That looks OK

4. Does the switching scheme make sense, and will it work as designed? The microcontroller triggers 2 relays for an effective 4PDT switch. The effect can be run in 2 stereo modes: 2-in/2-out, or 1-in/2-out. The 1-in/2-out mode must use buffered bypass, otherwise we're dealing with a passive splitter which is bad. So the 3PDT switch that links the channel 1 input to channel 2 also changes the circuit from true bypass to buffered bypass. In the schematic, the switches are shown in the true bypass, effect off positions.

I'm not sure I completely understand how your switching goes. The switches labeled LINK are Jack switches ?
but indeed, there is no way around buffered bypass in  1-in/2-out mode.

5. As I'm designing the PCB, are there any considerations I should take? I'm planning to keep five mostly-isolated ground planes: for the power section, the microcontroller/relay section, the LFO/PWM section, and each of the channels. They'll be star-grounded at the input jack, or at least grounded together at a single pad on the PCB that will go straight to the input jack.

The most important thing is to keep the grounds and power rails used in the processor/LEDS/Relay part separate for the audio part.

6. What opamps should I use for the audio path? The original schematic didn't specify, so I just threw NE5532s in there, but that may be overkill. Any reason not to use a TL074 quad for each channel to save a little bit of space?

Anything should go for the audio path. I used TL074's

All in all it looks promising :-)
Just check what panning modes work best when doing some tests with a stereo setup before making a final decision.

[aion]

2. Will the "Treble Only" and "Bass Only" modes work correctly with the way that I have the rotary switch set up? On Treble mode, the bass remains constant in both channels and the treble pans left to right, and the same on Bass mode? I have a hunch that something may be wrong here, but I don't know what or why. Originally I thought I needed to use a 4P4T rotary switch (one pole for each LED), but when I actually drew up the schematic I realized that it should probably only need one pole per PWM and I just need to invert the LED positions for the second channel. But I'm second-guessing myself now.

My first thought when using a bass only or treble only position in stereo would have been to have the fix part on one channel and the tremoloing part on the other. I think you may want to test these situations to see which make more sense and are musically the most useful. The same with the harmonic mode, as Jon mentioned, it might be sickening to have the separated channels oscillating out of phase on different amps. Sorry that my answers are so vague ... It's a bit difficult to imagine what it will sound like exactly. Only a test will give you this answer.

Hey, any effect can be sickening depending on how you set the knobs It could be pretty crazy at fast/deep settings in stereo, but I'm thinking the harmonic stereo modes would sound really interesting at lower depth and rates - a little more subtle on the modulation. I still feel like the sound I have in my head is both channels doing exactly the same thing but out of phase, so both sides are pulsing all the time on each of the settings. Does my schematic show that behavior, or would it do something else?

4. Does the switching scheme make sense, and will it work as designed? The microcontroller triggers 2 relays for an effective 4PDT switch. The effect can be run in 2 stereo modes: 2-in/2-out, or 1-in/2-out. The 1-in/2-out mode must use buffered bypass, otherwise we're dealing with a passive splitter which is bad. So the 3PDT switch that links the channel 1 input to channel 2 also changes the circuit from true bypass to buffered bypass. In the schematic, the switches are shown in the true bypass, effect off positions.

I'm not sure I completely understand how your switching goes. The switches labeled LINK are Jack switches ?
but indeed, there is no way around buffered bypass in  1-in/2-out mode.

I probably could have explained it better. The input and output jacks would go directly to the "IN" and "OUT" pads for each channel. The "K1" and "K2" switches are the DPDT relays which are integrated into the board itself, so that's the bypass switch. Then the "LINK" switches are all a single 3PDT toggle (A, B and C are the poles), so that when you split the channel 1 signal to channel 2 (pole A), you force the bypass system to take the output from the buffer instead of the unbuffered input (poles B and C).

I originally planned to use switched jacks to accomplish this without a toggle, but decided that it would be pretty difficult (or maybe impossible) and that a toggle would probably be more user-friendly anyway, and easier to wire up.

Although, looking at it again, part of the reason it doesn't make sense is because I drew it up wrong. The "LINKA" switch obviously would need to come before the buffered bypass output so that the output has a signal to work with. Not my best moment! I'll get that corrected.