Audiable pops when switching DC biased audio signal

Question

In my previous question, I have asked how to DC bias audio signal for a crosspoint audio switching matrix (Switch array (crosspoint) and DC biasing of audio signal).

I am using the following circuit to achieve this at the moment. I adapted the original circuit from the answer linked above to avoid picking the noise of the power supply and used external bias point generated using TLE2426 (based on the answer from this question: DC biasing audio signal).

^{simulate this circuit – Schematic created using CircuitLab}

The circuit works and the audio is properly passing it without any audible frequency range loss, but when turning the switch on or off, there is a loud pop audible on the output connected to the speakers.

I have done simulations of this circuit in LTSpice but there are no big spikes visible at the switching moment. I am a beginner in electronics and don't know where to start looking for the source of poping. I don't own a scope too.

Does anyone have any ideas on what in this circuit could be causing these pops?

EDIT 1:

I have added another 10k resistor to the output side. This has helped with the poping a bit. There is still audible pop though, too loud to be ignored.

I have tried lowering 2M2 resistors too, without any improvements.

The updated schematic below:

^{simulate this circuit}

I have chosen 10k resistors with trial and error as it seemed to my ear that when using 10k resistors, there was less poping than with 100k ones.

I am using these switches in the matrix (see my original question linked above) and if instead just connecting audio in to audio out, I connect another effect in the matrix, so the switches are audio in to effect in, and another switch effect out to audio out, where effect in and out are connected to audio effect, and then I switch between effect in the chain, and input directly to output, the pops are much louder.

It might be worth noting that I am using the switch matrix in latched mode, so all the switches are switched at the same time.

Any ideas what could be causing this popping or what should I look for?

Answer 1

A loud pop suggests a transient in which a large burst of current is being sunk to, or sourced from, your output (or equivalently, a sudden large voltage swing on the output.) This suggests that, when the output pin is connected or disconnected, the voltage on it suddenly swings. This makes sense for connection, because a floating output pin has nothing keeping it at a particular voltage, and leakage through the capacitor or elsewhere is likely to remove charge from it so that it's no longer at 12V by the time it's reconnected.

For disconnection, though, I'm a little bit puzzled about why there would be a pop. I wouldn't expect one if the switch is just leaving the output floating. Grounding the output would cause a big pop, but I don't see anything about that in the datasheet.

One thing possibly contributing to this is that your resistors to ground are extremely large. The related question you linked:

DC biasing audio signal

Has an answer suggesting 10k resistors to ground where you have 2.2M resistors. I would consider lowering those.

I don't know if this is orthodox, but I would consider connecting your DC bias to the output, as well as the input. If the switch is leaving the output floating when disconnected I would expect this to completely solve the problem. If it's grounding it, I would expect this to reduce the problem a bit, but not eliminate it.

Answer 2

A step function will also sound like a pop, you should see a step function in the output because the output cap is not charged prior to the switch being enabled. One solution is to pre-charge the output cap with the same bias. Connect a separate large resistor from the output pin to the same bias voltage used on the input.

Answer 3

Before you changed your circuit, you were biasing the input of the I2C switch to your reference voltage, but not the output. This means that with the switch open, the output would drift to wherever it wants -- probably ground.

Then when you switch it on, it was seeing the input, which was biased.

It should be enough to bias the outputs to your reference voltage with another \$100\mathrm{k}\Omega\$ resistor. You may want to try it on one channel to see if it works, then put them everywhere if you find joy.

It sounds like that didn't happen for you. Probably what's going on is that the fact that the audio is on and getting switched causes the pop.

I've had this same situation happen to me. The solution for me, back in the days of switches that you had to build up from individual components, was to turn the switch on gradually -- I'd run the gate voltage through an RC circuit, so the turn on event was spread over many audio cycles (something sub-audible, like 50ms, i.e., 1/20Hz is probably good).

This is not something that you can control with a switch operated by I2C -- the answer would either be to briefly mute the audio, then switch, then unmute, or to shop around and see if anyone makes switches specifically for audio that have a gradual switch-over.