FFT averaging technique in debugging audio circuits

Discussion in 'General Electronics Chat' started by Pickup Artist, Oct 15, 2009.

  1. Pickup Artist

    Thread Starter New Member

    Jun 17, 2008
    I've designed an audio preamp application, with no input signal connected, I'm hearing a constant high pitched sound at the output when I connected headphones.

    I put a scope on the output with no input signal connected and I could see a 6Khz ripple with more than a 100mV pp. I eventually found the fault with my design and it was due the IC I'm using to generate a negative power supply rail. Anyway, I managed to put in a nice ripple rejection hack onto the PCB and calmed it down, to the point where I can't see it on the output anymore with the scope.

    However I can still hear that high pitched sound at the output (but much reduced).

    I can't see the transient waveform on my scope because it's too low and below the ADC noise threshold generated by the scope itself.

    To see if I'm still getting a 6Khz ripple on the output, I did an FFT where the scope software averages all the FFTs over many captures. Attached is what I saw, low and behold there is a spectral peak at 6Khz but I can't believe that I would hear that because the average voltage magnitude is around 0.5mV. (ignore the large spike near 0Hz and the 0.1mV or so overall offset as that's just noise from the scope)

    If you stuck headphones onto a 0.5mV 6Khz signal do you think it would be audible? I guess it depends on the headphone impedance, these are pretty high quality cans that are used in recording studios for monitoring.

    I thought, lets get that sucker down to less than a few mV and nobody will be able to tell, but I'm amazed I can still hear the damn thing and I don't think my hearing is particularly sensitive either. I'm wondering if what I've hearing is that 6Khz spike or some other problem.
  2. MikeML

    AAC Fanatic!

    Oct 2, 2009