Hi all,
I build a pepper recently and I'm having a big noise floor in the module audio output. If I plug the us, run and then stop the project the noise reduces a lot. Most of the noises sync almost perfectly with the blue lights in the Bela. If someone has some advice or thoughts on this please let me know,

best!

How are the jumpers at the back? Does the issue present itself even without any jacks connected to the inputs? How are you using the analog inputs in your project, if at all? What programming language are you using?

Hi, the jumpers at the back are:

  • 12v w/regulator
  • 10v to pots from rack
  • all the inputs in analog
  • SDA-SCL
  • amplified and audio w/rack 12v

Yes the issue happens with or without jacks in the inputs.
I'm not using the analog inputs in my project for now, just the audio inputs, audio outputs and the analog outputs.
I'm using pd, I've also tried the Bela examples for pepper and the noise remains in every patch I tried. I will try with examples from another language just in case.

Thank you very much for your reply

there may be something wrong with the amplifier stage of the audio output. Did you install the NE5532 in the socket labelled IC1 ? Is the orientation correct?

Hi, I'm having issues removing the front panel because of the pots, I'm pretty sure I have checked the micro and the position is right. I wonder if it would even work with the NE upside down? I will try to check if that's the problem, it seems there is something wrong with the noise of the Bela getting into the audio signal, the noise pretty much stops when I stop running the patch.

thanks again!
best!

    a_G because of the pots,

    you should remove the knobs and nuts and it should come off easily.

    a_G . I wonder if it would even work with the NE upside down?

    It wouldn't "work" but I am not sure what the failure mode would be.

    a_G the noise pretty much stops when I stop running the patch.

    That may be expected even if the failure was on the Pepper PCB. It depends on the failure.

    Did you solder the 4x1 socket header on the Bela cape that is mentioned on step 24 of the build guide ?

    Hi, thanks for the reply!
    I checked the NE5532 and the orientation is correct. The socket header is also correct, I'm not sure what to check from here, I would try to check if changing something from the software anything different happens,

    thanks again

    Could you try separating the Bela cape from the Pepper PCB and then use the audio adapter cable that came with the Bela starter kit to connect headphones to the Bela OUT connector? Is the noise still there?

    Can you also send a picture of your Bela cape ?

    I've just tried the Bela cape alone, with the headphones it sounds cristal clear 😢 , I used pd and c++ examples
    from the Bela examples.

    I've just connected the audio input to the Bela just to test the same granulator patch I was using in the Eurorack and the noise appeared. The same noise, but this time without the Eurorack, only the Bela. When I unplug the audio input conector the noise goes away.

    are you connecting your Pepper to the host computer both via USB and some other via (e.g.: is a soundcard in the loop?). In that case, you may be hearing the result of a ground loop caused by the audio and USB grounds both being shared. You can easily verify that by starting a project from the IDE, verifying the noise is there and then leave it running while disconnecting the USB from the computer: does the noise stop?

    I power the Bela only with a 5V power supply via USB. When I start the Bela with the audio output and the patch starts (the delay example from Bela examples) everything is ok, I don't hear any noise, when I plug the conector for the audio input the noise starts. The patch works well because I can hear the delay working, but when I connect the input cable is when the issue appears.

    So the issue seems to be that the noise is coming in through the input rather than being generated at the output? What is connected to the input cable?

    Nothing, I'm testing the signal noise, I'm just connecting the output to hear the Bela, when I plug the Bela connector the noise happens. I cannot upload this video but I'll send you a dropbox link so you can see it. When nothing is plugged and the patch is working everything is ok, no noise floor at all. When I connect the Bela cable without no audio input, just the cable that comes with Bela an awful noise starts. You can even hear the delay working. I switched the cables and the same thing happens.

    That's pretty much expected behaviour: the input is floating and your fingers are causing interference. That interference would disappear if you were to connect something to the input or short it to ground.

    Let's get back to basics. Connect the board to Pepper and, run a simple Pd project such as this:

    [osc~ 300]
    |
    [*~ 0.5]
    | \
    |    \ 
    [dac~]

    Does that have any noise?

    Ok, I'll follow your toughts, that works perfectly whitout noise, what do you think it would be better to do now?

    Do you have anything to patch into the input to see if the noise is being picked up from the input? In that case you could try:

    [adc~]
    |    |
    [dac~]

    with the input shorted to ground vs with something plugged into it.

      Now with Bela detached from pepper that patch with a balanced audio input (I'm using a TRS output from a drum machine) works perfectly too, no noise at all.

      OK, now put it into Pepper and test this patch:

      giuliomoro
      [adc~]
      | |
      [dac~]

      Note that when using the audio inputs on Pepper the expected input voltage range is approximately 10Vpp instead of the 1.8Vpp that it expects when running without Pepper, so you'll need a correspondingly larger signal to do a like-for-like comparison.

      ok! I've test it with my modular system, I think that's what you mean with the 10Vpp right? The noise is still there, now I'm comparing the pepper with the mutable instruments beads (which is also digital, just to have a digital noise floor reference).

      I think the most annoying thing is that there are specific signals that are present in the noise, you can see them in the spectrogram, the noise floor is still louder (maybe a bit more than the double of the beads noise floor) but it could be gated or something with a mixer, but that cluster of static signals are really present. I've also test it with an analog vco and when you go to a lower pitch you can still hear the high noise signals with the vco.

      beads

      pepper