Mid-Side FX Loop Pedal

Hello Wigglers, I need some help on a design please!

I’ve been wanting to design a mid-side encoder/decoder (like the Worng Electronics LRMSMSLR) in guitar pedal format, something I can use to nicely interface mono pedals in my all-stereo FX chain – I play piano, so two mics is always my starting point. I’d also like to add parallel wet/dry controls on the FX loops, and a tone control and saturation circuit on the mid-side returns before the decoder.

Attached is a draft of the circuit I cooked up. I just know there’s going to be some things wrong, even before I try breadboarding and testing, so I would love some tips an advice.

Things that are probable concerns:

  • Gain stages, especially around the m/s encoder and decoder.
  • Correctly biasing the signal, especially around the op-amps.
  • Do I need the caps to GND on the op-amp negative power pins, since its just powered by +9v and 0V?
  • Impedance for playing nice with guitar pedals.

Overall, my main goal is that this fits in and plays nice with my guitar pedal setup, so it needs to be powered by DC+9v centre-negative, and have the right impedances etc. I’ve not designed a circuit this complex for pedals, I’m more used to eurorack modular designs, so I’m sure I’ve portably made a tonne of mistakes!

Currently all the op-amps are TL072, but I’ll probably change some to TL074 so save on space and caps – I’ll probably make that decision based on the PCB design when I get there, but any advice on which would be great.

I’ve attached the Kicad files, if anyone wants to jump into it!
Kicad Files

Any help and advice would be hugely appreciated :slight_smile:

Here’s a list of the resources I used for this design:


No definately not. The diagram shows them (e.g. C25, C28, C30 etc.) connecting GND to GND, which has no electrical effect at all (only a monetary effect).

I’ve browsed the schematic and without checking its intended function there are a few things that strike me as odd. Maybe they are not, but I wanted to mention them anyway:

  • On the RHS of R34 and R33 are 2 connection points that lead to nothing, you may want to check those.

  • Why are R45 and R46 in parallel to a potentiometer each? Is this to change the characteristic of the potentiometer? Otherwise you could leave them out.

  • R63 will cause quite a voltage drop even if there is only a small current flowing through it. Is it meant to be 220 kilo Ohm?

  • You use a +VDC and a VDC net label, which carry different voltages which may lead to confusion. Maybe you could change one of the label names? Maybe call +VDC something like VOFFSET or VGND / something relating to the fact that it is used as a virtual ground?

Great, thought so, just thought to check.

re your points:

  • item good pick up, not sure what happened there!
  • not entirely sure… they were in the design for the saturation circuit that I based that section on, I assumed they were to bias the op-amp output before the next bit of the design (the high shelf in the Saturation circuit)
  • my mistake, a copy-paste error, should have been 100R.
  • yeah I would have preferred something like VREF or VGDN, but Kicad wouldn’t give me the option? I jus mustn’t have the symbol in my library and I doesn’t let you change the label names on power symbols…

Thanks again for all the feedback! New version attached with your advice and some other things.

1 Like

Right, you can’t just change the labels on power symbols; you can, however, create your own power symbols — or, more easily, bring up a power symbol in the symbol editor, do “save as” into your own symbols library, and modify that. But you have to have the symbol name match the name of the power input pin.

HOWEVER… I’ve only looked at this very briefly, but you’re not using +VDC as a power rail, right? It’s just a reference voltage? In which case, it’d be more appropriate NOT to use a power symbol for this — just use a label (global, local, or hierarchical). Because power symbols are power input pins in disguise, and that’s not what you’re using this for (I think). Save the power symbols just for things like ±12 V rails and outputs of voltage regulators.

1 Like

That’s still problematic. It’s in series with the 9 V power supply on the way to a whack of a lot of TL072 power pins, so just as a guess maybe you’ll be drawing 30 mA? 0.030 x 100 = 3: You’d drop 3 volts across that resistor, and VDC will be 6 V, rather than the 9 V you might think it’s supposed to be. (That’s ignoring the voltage drop across the series diode.) And the power dissipated is 90 mW — if it’s 50 mA rather than 30 then it’s 250 mW — so if you’re serious about having a 100R resistor there it’d better be rated for more than 1/4 watt. I’m not at all sure what your reason for having that resistor there is, but normally you wouldn’t have more than about 10R in series on the power rail.

ah okay that makes sense. You’re right that it’s not a power rails, it’s just a +4.5V reference voltage. So yeah, I’ll make a different symbol that’s not the “+” power one.

This is the design I’ve based the power section on, which has solved hum issues in past designs, but maybe it won’t work for this one?

If the consensus is that the 100R isn’t going to work, I’ll drop down to 10R.

The other thing I’ve wondered, is there a benefit (besides number of components) to using TL074 instead of the dual op-amps?

1 Like

Ohm’s law is ohm’s law. You have 14 op amps, for a TL07x it’s 1.4 mA each, that’s 20 mA. I was guessing 30 mA overall. So the voltage drop across 100R is at minimum 2V.

A TL07x will work only if the inputs are within a range smaller than the supply limits — typical value is 3 V above V- up to V+. So with 9 V and ground you have 3 V to 9 V, or -1.5 V to +4.5 V relative to your VDC+. Now if you’re dropping 2 V across that resistor your input swing is down to 3 V to 7 V; VDC+ drops to 3.5 V so the swing relative to that is -0.5 V to +3.5 V. Or 0 V to +3 V if the drop is 3 V. (That’s “typical” but the worst case spec is V- + 4 V to V+ - 4 V, which leaves you with nothing. That’s not likely though.) Output voltage swing is typically V- + 1.5 to V+ - 1.5, so with a 3 V drop that leaves you with ±1.5 V relative to VDC+. If you’re okay with that, fine, but I’d go with something better than a 1/4 W resistor as I said above.

You could use TL06x which is 200 µA instead of 1.4 mA per op amp. I think TL07x has better noise performance, though. I’ve never used TL06x.

Just dropping to 10R would increase your cutoff frequency by a factor of 10, so you wouldn’t want that. Could you use 470 µF and 10R?

The second stripboard you show has four 100R resistors and 47 µF caps in parallel, splitting out four 9 V rails for a lower current draw and hence a lower voltage drop on each. So you could consider that too. But I’m not sure it’s really any different from using 100R/4 and 47 µF x 4 on a single rail.

They’ll behave the same, so it’s just the smaller number of components. Depending on the circuit TL072s might have an advantage of being easier to route.


Okay I’ve worked on it a bit and made a lot of revisions!