Ground Loop Buzz with no electrical connection

[BurnedResistor]

Hey All!

I am bulding an audio relay step attenuator.
Got the pcb in an assembled, and the attenuator itself works great.

Schematic is included.

I though I would be smart and avoid ground loops between different input devices, so I kept the grounds of the different devices seperated and only switch in the ground of the device I have selected with the relays.

Because the whole setup consists only of passives and relays there is no electrical connection between the audio signals/audio ground and control signals/digital ground: The attenuator is fully isolated from all digital circuitry, power supplies, etc. I liked this, because I though this would keep any noise coming from the control circuitry out of the signal (Except of course any noise due to capactive coupling between traces)

To test the circuit I had the output connected to a headphone amp, the input connected to the headphone output of my laptop and the stm programmer hooked up. Everything worked great.

Until I unplugged the programmer. I was getting a ground loop style buzzing in the headphones. Not loud, but very much audible and annoying.

I realize that while the Programmname was connected the signal ground and digital ground where of course tied together, but are now completely isolated again. If i electrically connect the two grounds togehter on the board the buzzing goes away.

Similarly, if I hook up my phone as the input (which is floating ofc) I get a very similar bussing. Connecting the grounds together gets rid of the problem.

I am at a loss. I don't understand what is going on.

When I hook up the phone directly to the headphone amp all is good.

Now - I could just tie analog ground to digital ground at the power supply with a net tie, but I would rather understand what is going on. Is there anything I am doing wrong?

Could there be some kind of large swing between the two circuits that is coupling through, and by tieing them together I am getting rid of that?

Any Ideas?

[Andy Watson]

(Except of course any noise due to capactive coupling between traces)

I think you have the answer right there!
I notice that your control circuit(s) is powered by "9 V AC" on a power jack - does this mean that it is powered via a wall-wart type box? With equipment that is powered via a transformer you have to remember that there will be a few, perhaps tens of pico-farads capacitance to a mains voltage level signal - combine this with your signal path impedance in the 1-10k ohm range - it's probably going to leak through!
Cure it either by grounding your control circuit and/or finding a transformer with a shield connection.

[BurnedResistor]

(Except of course any noise due to capactive coupling between traces)

I think you have the answer right there!
I notice that your control circuit(s) is powered by "9 V AC" on a power jack - does this mean that it is powered via a wall-wart type box? With equipment that is powered via a transformer you have to remember that there will be a few, perhaps tens of pico-farads capacitance to a mains voltage level signal - combine this with your signal path impedance in the 1-10k ohm range - it's probably going to leak through!
Cure it either by grounding your control circuit and/or finding a transformer with a shield connection.

Yup powered by a wall plug type transformer

That would explain the behaviour I am seeing!

The headphone amp is grounded and so the reason the buzzing went away when I connected the two together was not inherently caused by them being connected, but by my digital circuit being grounded through the headphone amp!

Will try and report back.

I am not quite sure I understand what the reason for the noise is though:

With equipment that is powered via a transformer you have to remember that there will be a few, perhaps tens of pico-farads capacitance to a mains voltage level signal - combine this with your signal path impedance in the 1-10k ohm range - it's probably going to leak through!
Cure it either by grounding your control circuit and/or finding a transformer with a shield connection.

I understand that the relatively high impedance of the signal path makes it very susceptible to noise coupling in.
Some capcaitance between transformer sides makes sense, but what exactly is being coupled through? The 50hz swing of the mains? Noise on the mains?

and how does grounding my circuit change this?

[BurnedResistor]

Yup,

grounding the control circuit gets rid of the problem (with a seperate wire rn)

I am unsure however on how to best implement this 

I would like to stick with a barrel jack, is there something like a grounded 9Vac adapter? (Or DC, does not really matter?)