[Moved] DC Crosstalk/Coupling

[tlee1]

Hey,

I am using the circuit shown below, just a basic input select (there are multiple inputs, only one shown) and then an noninverting opamp. I have a problem that there is some DC (50-200mV - it varies) on the trace marked as INPUT_BUS, causes pops when the input relay is switched. The DC cannot be coming from the source, as it is still there when there is no input connected or the relay is in the NC position. Likewise, I believe it is not the opamp bias current leaking through C2 as it is still present when C2 is lifted from the board. The relay is powered with 5v and the DC on the input bus is greatly reduced when the 5v supply is disconnected from the board. This obviously leads me to the conclusion that the DC is somehow crosstalking over from the 5v traces to the input bus, but my understanding is that crosstalk is an AC phenomenon and cannot happen with DC. Can anyone describe the mechanism by which the DC is appearing, and how to mitigate this?

Thanks

 

[KlausST]

Hi,

is it possible to connect the 1u poly in front of the relay?

***
Maybe there is a creepage current caused by flus residuals.

Or it is capacitive coupling from the relay coil to the relay contacts.

Also possible that you use the same current path for signal and relay coil.
Or you use same power supply ofr relay and OPAMP.
Even a missing diode across the relay coil may cause this problem (high voltage peak)

A lot of possibilities. Can yo shw us a picture of the PCB.

Klaus

 

[tlee1]

Hi Klaus, thanks for your suggestions, I have answered your questions below:

Hi,

is it possible to connect the 1u poly in front of the relay?

No, the main point of the poly cap is to stop the opamp's bias current going through the relay, so putting the cap before the relay would negate this effect. Also as there are actually 5 seperate relay inputs, this would require 5 caps instead of 1.

Maybe there is a creepage current caused by flus residuals.

Or it is capacitive coupling from the relay coil to the relay contacts.

Also possible that you use the same current path for signal and relay coil.
Or you use same power supply ofr relay and OPAMP.
Even a missing diode across the relay coil may cause this problem (high voltage peak)

The opamp and relay have seperate supplies, and seperate grounds which only meet at the PSU star point. Don't appear to be any missing relay diodes.

A lot of possibilities. Can yo shw us a picture of the PCB.

Klaus

In the fist picture I have highlighted the 5v trace in yellow and the input trace in grey, but bear in mind that the 5v is on the bottom layer whilst the input trace is on the top layer

Thanks

 

[FvM]

Having different grounds ("PE" versus "0V") suspicious. Are you sure that there's no DC difference between grounds.

If not, a DC voltage should be expected at the X1 input, which would well explain the "pop". In this case, place a capacitor between X1 and relays.

 

[tlee1]

Having different grounds ("PE" versus "0V") suspicious. Are you sure that there's no DC difference between grounds.

If not, a DC voltage should be expected at the X1 input, which would well explain the "pop". In this case, place a capacitor between X1 and relays.

Having separate grounds is pretty common in audio, one is a clean 0v reference for audio circuitry, PE is the chassis ground, and there is also a dirty digital/relay ground not shown; this is to reduce common impedance coupling. They all join together at the PSU, and I have checked there is no measurable voltage between them.

I understand that if there was a DC offset on whatever is connected to the input then that would cause a pop, however that is not what my problem is. The issue is that when I have nothing connected at all there is still DC present on the input trace. Even when I removed the components as mentioned in my first post, so I basically have a trace that is floating at both ends and tied to ground with a 100K resistor, there is still DC on it :bang:

Thanks

 

[D.A.(Tony)Stewart]

Scope probe ground may influence the circuit ground shift. I suggest you use two probes balanced such that A-B on any signal is a flat line then measure all signals and grounds again using grounds for both probes coupled by a medium size cap ie 0.1uF is 1sec RC with 10M Probe.

 

[KlausST]

Hi,

the NE5532 has 200nA typ. bias current. At -Vin there is low impedance, so about no additional offset voltage.
But at +Vin there is only a 100kOhms DC path. now multipöy this with 200nA then you may expect 20mV offset voltage.
On the other hand this does not explain a "pop".

I recommend to equal R1 and R2. If there is a creepage current in the input, then (now) there is a 100kOhms DC path to GND when sitch is closed, but a 2k2 DC path to GND when switch is open.
.. but where should the current come from?

Relay supply: Is there a capacitor to stabilize the relay power? I peronally would try to route the relay power and the relay return path through the same connector.

...

To be true i don´t see a real mistake in your circuit.

Klaus

 

[FvM]

Having separate grounds is pretty common in audio, one is a clean 0v reference for audio circuitry, PE is the chassis ground.

Yes, but if the the input signal reference is PE, than the connected amplifier will either use the same ground or implement a differential input.

There's no acceptable reason why the floating INPUT_BUS should have a DC offset, can be either wrong measurement, circuit wiring fault, defective 1 µ capacitor or polluted PCB.