Loopers, grounding and oscillation

[remmy]

I have made myself a bypass looper and approached it quite pedantically regarding the ground wiring as I wanted to avoid possible hum.  I have wired it so there are four separate grounds (the sockets, LEDs, switches and DC socket) which all meet at the output socket which is grounded to the enclosure.

Having tested the looper yesterday I am pretty happy with it, nice and quiet but I have one situation which is causing a shed load of noise.  When I have three certain loops on (fuzz face, rat and klon) there is oscillation, and I would not normally use all three together but the oscillation has grabbed my attention as when the same three pedals or in series without using the looper there is no oscillation.

So the issue is down to the looper but I am not sure if there is any more I can do regarding keeping the grounds separated, I have added a picture below to show how everything is connected.  Oh and the sound doesn't go away if I disconnect the power supply.

Is there anything different I can try to avoid the oscillation, and what would be causing it?  I am looking for an understanding of why the looper has this effect rather than a solution (though that would be great) as the issue isn't really one that is going to cause me any hassle.

[R.G.]

Grounding issues require changing your thinking about electronics and wires.

Wires are NOT zero-ohm highways for electrons. They are low-value resistors.

Being resistors, they generate a voltage in accordance with Ohm's law when current passes through them. This voltage may be small, but even small voltages can be amplified up into ugliness when there are high gain amplifiers involved.

There are an infinite number of ways to avoid grounding issues, but only one way that can be known ahead of time to do that: star grounding, and it's derivatives. All the others require tinkering and testing for the specific setup.

There are multiple kinds of grounds. One is a shield ground. This keeps the external sea of electromagnetic interference out. You want this to be your external metal box. Another is what I call sewer ground. This is the return path for all the "used electrons" going back to the power supply. Another is "reference ground", a point of zero volts that your circuit uses to see what those wiggles on the signal line really mean, or a quiet source of zero volts to keep the opamp outputs standing still, not dancing around in defiance of the signal's instructions.

The current through signal grounds causes voltages that are the sum of the currents returning back to the power supply. If that's one stage, all is probably well, as this is much less than the actual signal wobbles in the circuit. If two or more circuits share a sewer ground wire, the voltage in the sewer ground wire jumps around in coincidence with the sum of the returning currents. With various signals in the circuits having different phases and sizes, and with switched circuits like LEDs, this can be a pretty noisy place, relatively speaking.

The trouble starts when you connect a reference ground to a sewer ground out at the noisy end. Now the circuit using reference ground to mean "0.00000000V" instead sees its reference wobbling. Since the incoming signal doesn't see that same wobbling, the reference ground voltage is impressed on the input as a signal. (Note that "common mode signal rejection" is the degree to which circuits can ignore this ground dance. It varies with circuits.) If the circuit has little common mode rejection, the ground noise is amplified. High gain distortion pedals can amplify it a lot. If this causes perturbations in the ground current, perhaps after another few pedals up the gain some more and perhaps change the phase a bit, you can get the ground noise being amplified and feeding itself - an oscillator.

If you think about it, the way to keep the voltage across a resistor (including a wire!) down to zero is to make the current through it zero. One good way to keep the reference voltages clean is to not let current flow through them, and that can be done by running a separate reference wire from the circuit back to the circuit One True Ground point. This one idea is the seed from which star grounding grows.

For simple transistor circuits for which separating reference from sewer ground for that one  circuit is not possible, limiting the currents on sewer+reference ground to be only that circuit is almost as good. Any errors from sewer are merely a reflection of the circuit's activity itself, and not in a way that causes oscillation. For circuits which can be set up to have differential inputs and which have high common mode rejection, you can use the ground from the incoming signal for a reference, and only the actual signal vs ground will be seen, and all will be well.

We generally get away with incredibly bad grounding practice in pedals because they are small, close together, and use small currents in an isolated way. Setting up a looper with many pedals invalidates that.

And now I'm finally back to your looper.

If I were doing this (and I have before  ) I would do a couple of things. First, decide where a One True Ground will be. This can be the power supply ground, output plug, some place. The actual place can be one of several ones, as we'll see. Isolate every single input jack from the case, to allow you to run a ground wire from the jack to the circuit it serves. This keeps any currents on the case from being cross-injected into the circuit inputs. Connect the case to the One True Ground point, and connect no other wires or connectors to it. Now the case is an RF and mains hum shield, and no other currents can flow in it. It's a pure shield.

Connect the power supply ground on individual wires from the power supply ground return to each circuit, one per circuit. In your case, this is just the LED's, and they don't have any "gain", so you can run one wire from the power supply to all LEDs. But keep this "ground" away from your signals. If you run a ground wire to the switches/LEDs, do not connect that ground wire to the send/return jack grounds except at the One True Ground. Isolate the signal-handling wiring on the switches from the LED-switching parts.

Finally, you're probably using a single external power supply daisy chained to the external pedals.  This constitutes an external ground path in parallel with the signal reference ground on the send/return shields. The right way to do this is to shield the cables only at the pedals and use the power ground connections as reference ground, but that carries sewer ground issues of its own.

So there... it's a step toward understaing, and not a solution. You did ask.   

[slacker]

Make sure you don't get so focused on looking for a problem with your grounding that you forget to investigate other causes, for example it could be the proximity of the signal input and output wires that's causing the problem.

[Processaurus]

Make sure you don't get so focused on looking for a problem with your grounding that you forget to investigate other causes, for example it could be the proximity of the signal input and output wires that's causing the problem.

I agree, I've had the squeals with a couple high gain pedals in a true bypass loop, and the problem is that the pedals have a high input impedance, and extremely high gain when they're cascaded, and then the output wire comes back and might be very close to the input, so the input picks up some of the output and it loops around a couple times a second and amplifies each time, causing the squealing.

These problems can often be solved by physically separating the input and output wiring as much as possible, routing them away from each other.  You might even plug everything in and get the pedals squealing and open the the looper up and try moving the wires around and see if anything changes.  They need to come together at the 3pdt stompswitch, nothing to be done about that, but you can put the LED indicator on the middle pole of the switch, so the ins and outs are on opposite sides.

If that doesn't work, shielded cable for the inputs is a good next step.  One last thing that can work is lowering the impedance shown to the first effect, by having a simple buffer preceding the first switch, and putting the buffer physically close to the jack where the guitar comes in.

I assume you ground the inputs to the effects when they're bypassed?  Another interesting problem with true bypass loopers, they'll easily oscillate if the high gain effect's input is floating in bypass, and you can hear the squealing bleeding into the bypassed signal, just by being in the same box as the squealing stuff.

[remmy]

So there... it's a step toward understaing, and not a solution. You did ask.   

Thank you RG for such a detailed explanation, it took me a few reads (I'm not the quickest on these things) but I get it now 

So am I right then that I should approach the wiring as per the diagram above but only have the LED grounds on a bus from the one true ground, with every other ground connect to the OTG in the star configuration?

Finally, you're probably using a single external power supply daisy chained to the external pedals.  This constitutes an external ground path in parallel with the signal reference ground on the send/return shields. The right way to do this is to shield the cables only at the pedals and use the power ground connections as reference ground, but that carries sewer ground issues of its own.

Yes it will be daisy chained to the same supply, I did wonder if this would have an effect.  When you say the above do you mean that I should make up the patch cables to be shielded at the pedal end and have the sleeve lugs of the send and return sockets in the looper all going to the DC ground?

[remmy]

Make sure you don't get so focused on looking for a problem with your grounding that you forget to investigate other causes, for example it could be the proximity of the signal input and output wires that's causing the problem.

I assume you ground the inputs to the effects when they're bypassed?  Another interesting problem with true bypass loopers, they'll easily oscillate if the high gain effect's input is floating in bypass, and you can hear the squealing bleeding into the bypassed signal, just by being in the same box as the squealing stuff.

Yes I have used this layout for the wiring http://www.singlecoil.com/tb-strip/dia1.html

I did try using shielded cable for the input and output wires, but not the loop send and return wires.  I figured they wouldn't require it as the unused loops would be shunted to ground, but I guess if the loop is open then there cold be crosstalk?