Strange problem with mixing console

Pierre

Banned
2004-11-26 6:32 am
Paris
Hello all.
I have implemented a couple of Class-D ampls, for those who don't know my work so far, they are based on a triangle generator, so they are natural-sampling PWM ones. The power stage is a half-bridge driven by a IR2113. Feedback is taken after a 2nd order filter (have a look at "Help with feedback" thread).

The thing is that I have experienced a weird problem. Until now, I was testing with the soundcard output of a laptop. Everything ok.
But now I use the output of a Gemini PS626 mixer. It has a vu-meter, that I think is measuring directly the output terminals.
Well, if I connect it to my amp, when I turn the balance to one side only and put some watts out of the amplifier, a large amount of signal is couple into the other channel, making horrible peaking/noise. The funny thing is that if I disconnect the "inactive" output RCA at the mixing console and leave it "open", the signal is also coupled. How can that be?

My only clue is that it must have higher output impedance and that some signal is reflected from the amp input back to the console, but how can that couple to the other channel then?

If I disconnect both outputs, no coupling between channels is seen at the console.

Both amplifier modules share a power supply, but also do the same if they use separete supplies (but common GND).

I am quite astonished by the situation, if anyone can point to the right direction it will be very appreciated.

Best regards.
Pierre
 

Pierre

Banned
2004-11-26 6:32 am
Paris
Well, I think I have discovered the problem: it is due to EMI. Connecting my oscilloscope probe to the inactive channel, I see a big quantity of switching noise entering to it from the other channel. Now I have to discover the severity of the problem and where this noise comes from (radiated or conducted). My theory is that it is conducted through, because it doesn't change when I move the input wires around the rack case.

The problem is more noticeable with the mixing console due to its higher output impedance (400 ohm vs. less than 32 ohm in a pc soundcard output).

Perhaps I should put any chokes or LC filters to reduce EMI, but I don't know exactly where they could be more effective: in the input of the voltage regulators for the modulator section? between power and signal GND planes?

Any idea will be very appreciated. I will post any advance I make in the issue.

Best regards,
Pierre
 

Pierre

Banned
2004-11-26 6:32 am
Paris
In fact most of the switching noise has disappeared as soon as I have wired the amp inputs to the RCA connectors with shielded twisted pair.
Some residue still remains (about 50mVpp) when I look at the input with nothing connected and hence with no signal.

My front-end is a balanced line preamplifier. This should act as the buffer, and has 1.5gain at each input. Here is the schematics. This opamp is in the same package as the error amplifier.
 

Attachments

  • input_amplifier.gif
    input_amplifier.gif
    5.7 KB · Views: 102