Eliminating ground loop: help please

[wellshuxley]

Hi folks, I'm trying to reduce noise in a build I'm working on, and I believe I have traced the issue down to a ground loop. When I lay out the ground connections visually, the issue jumps out, as my grounding scheme quite literally forms a loop - it's almost like they named them "ground loops" for a reason! 

Some context on my build and issue:

• The complexity in this grounding configuration stems from a big old inductor that I'm using in my build, which has its own chassis ground pins. The inductor is wobbly and unstable without these pins soldered in, and the manufacturer advises keeping the chassis ground separate from audio grounds.
• I also have audio grounds and power section grounds separated.
• I'm using uninsulated I/O jacks (REAN Neutrik NYS229), thus the I/O ground connections to the chassis.

In the attached screen shot, I have laid out my current ground nodes, ground "types" (or groups), and the existing connections between them. I've also marked 2 connections that I think, if severed, will break my ground loop and resolve my ground noise issue.



So, my questions are: am I on the right track here? Are there any issues I've overlooked or misunderstood thus far, or any opportunities for improvement I've missed?

Thank you in advance for your input!

[PRR]

> connections that I think, if severed, will break my ground loop

Do you have wire-cutters?

[Clint Eastwood]

Hi,

I would also cut the connection between the chassis grounds of the inductor and the chassis, and make a new connection from the chassis grounds to power ground or dc in. Basically, as I understand it, you should avoid multiple currents flowing criss cross through the chassis back to the power supply.

[wellshuxley]

> connections that I think, if severed, will break my ground loop

Do you have wire-cutters?

Haha I wish! The connection between the chassis and audio grounds is on the board, so changing it will be more complicated than a wire snip. The input jack's ground could be simply snipped, though.

[r080]

What sort of noise? Is this AC powered?

Edit- stupid question. Can't see pictures on one of my computers.  Clearly marked DC input.

[amptramp]

I don't think the cuts you have shown are going to do what you want them to do.  You show an audio ground separate from the power ground but unless you are running differential circuitry, there is always going to be a link from power to signal ground.  If the inductor uses ground terminals to hold it to the board, these should be connected to the audio ground.  The chassis ground should make contact at one point to the signal ground although most of the time you can get away with input and output jacks grounded to the chassis and this may offer redundancy in case one connection from jack to chassis goes to high resistance due to corrosion.

You show two separate shapes for chassis ground but chassis ground has no real use on the circuit board, so there should be just one chassis ground connection through one or both jacks.

[wellshuxley]

What sort of noise?

Good question - I'd describe it as a noticeably higher noise floor relative to when the pedal is disengaged, with no other distinctive characteristics.

[wellshuxley]

You show an audio ground separate from the power ground but unless you are running differential circuitry, there is always going to be a link from power to signal ground.

Good point! No, I am not running any differential circuitry. The audio and power grounds are on separate ground planes, which are (or will be, according to my planned next iteration) only connected via small wires joined at the DC input ground. Does this approach (separate ground planes, yet still ultimately connected) provide any benefits, or am I just overcomplicating the design without improving it?

If the inductor uses ground terminals to hold it to the board, these should be connected to the audio ground.

Interesting, this advice runs counter to the advice in the manufacturer (Cinemag)'s literature. Can you provide some more context on why this approach is better? Maybe Cinemag's advice is relevant in a pro audio context, but not so much for pedal building. If there is no need for a "separate" chassis ground group, then I can probably simplify the design significantly by just using one catchall ground plane.

You show two separate shapes for chassis ground but chassis ground has no real use on the circuit board, so there should be just one chassis ground connection through one or both jacks.

To clarify, the section in my diagram labeled "Chassis" is meant to indicate the literal chassis/enclosure itself, whereas the "Chassis grounds" group points to the inductor chassis and its pins.

[...] although most of the time you can get away with input and output jacks grounded to the chassis and this may offer redundancy in case one connection from jack to chassis goes to high resistance due to corrosion.

That's a good point regarding redundancy between the two jack grounds - thank you!

[anotherjim]

I've never used any fancy hi-grade audio transformers/inductors but the cheap audio transformers I've used in various projects have bend-up fixing tags that cause no trouble soldered to 0v.

noticeably higher noise floor relative to when the pedal is disengaged, with no other distinctive characteristics.

usually points to a circuit design being just noisy in the first place. 

[r080]

I've never used any fancy hi-grade audio transformers/inductors but the cheap audio transformers I've used in various projects have bend-up fixing tags that cause no trouble soldered to 0v.

noticeably higher noise floor relative to when the pedal is disengaged, with no other distinctive characteristics.

usually points to a circuit design being just noisy in the first place. 

Personally, I start to worry something is not working if I engage a new pedal and don't hear an increase in noise. Granted, we don't know what you are building.

It might be helpful to see some schematics and pictures of the build.

[wellshuxley]

Personally, I start to worry something is not working if I engage a new pedal and don't hear an increase in noise. Granted, we don't know what you are building.

It might be helpful to see some schematics and pictures of the build.

The pedal is an adaptation of a Pultec passive EQ. It's just input gain -> JFET source follower (as buffer + mojo source) -> passive filter stage -> output gain. The power section has an NE555-based voltage doubler. Unfortunately, I don't have photos of the version I'm debugging (I'm currently out of town), but here's a gut shot of the last iteration of this pedal:



As you can see in the photo, my last version was done on perfboard, with a haphazard rat's nest of wiring... and it was super quiet!

The version I'm working on now is a PCB layout (my first). The only significant differences between the designs are the relative positioning of components, the use of ground planes, and some of the ground connections.

[anotherjim]

If you want to be absolute with the ground scheme then here's the approach I've learned to take.

Use connector jacks that isolate the ground from the chassis.

Pick the signal input TRS jack as the primary ground node. Most pedal PSUs don't actually connect to the house AC safety ground, so the most likely source of a safety ground is via signal cable screen to something that is AC powered and grounded.

Wire the DC input negative directly to the input jack ring terminal. On the board, find the primary power supply rail filter cap and make it the centre point of a star ground scheme. Use a thickish wire to connect this point to the input jack screen terminal.

Route the power supply component 0v connections to the star point and nowhere else. Place the PSU circuit on one side or corner far away from the audio input.

Divide the audio path into blocks separated by coupling caps. Connect component grounds together and then route that ground to the star point.

Fit a bolt to the chassis/enclosure and use a solder tag to run a wire from the chassis to the star point.

Finally, a charge pump PSU is a noise generator, the 555 based ones more so. If you must use that, fit plenty of capacitance on input and output and add low inductance ceramic disc capacitors in parallel with the electro's.

[antonis]

On the board, find the primary power supply rail filter cap and make it the centre point of a star ground scheme.

I'd suggest for that cap to be physically placed between PSU negative terminal and star ground point, due to heavier current handling than rest of the circuit "returns"..
In the way of PSU GND -> Filter/Reservoir cap GND -> Any other power GND (e.g. Relays, Rail decouple caps, e.t.c.) -> Audio GND (Star point)