Audio Amp problem

Discussion in 'General Electronics Chat' started by PRS, May 7, 2009.

  1. PRS

    Thread Starter Well-Known Member

    Aug 24, 2008
    989
    35
    I have an audio power amp that oscillates at a high frequency -- about 500 - 1000 kHz. This cannot be heard, of course. I input an audio frequency sinewave from my function generator and I can't see it on the oscilloscope but I hear it from the speaker. The scope will not show it. It only shows that high frequency wave. Not only this, but I believe the output is distorted by a mixing of this unwanted signal with my input signal. Does anyone know what is going on here?
     
  2. beenthere

    Retired Moderator

    Apr 20, 2004
    15,815
    282
    A schematic might help. One thing to look for is a capacitor in the feedback path that has opened up. The high frequency attenuation is often through a low value resistor and a series-connected pair of electrolytic capacitors.

    No audio amp should be outputting at above 500 KHz. Bet the heat sinks get pretty warm.
     
    Last edited: May 7, 2009
  3. PRS

    Thread Starter Well-Known Member

    Aug 24, 2008
    989
    35
    Thanks, Beenthere, I'll look around for bad components. This never happens on paper, you know. It's when you build it that stuff happens!
     
  4. dsp_redux

    Active Member

    Apr 11, 2009
    182
    5
    Like beenthere said, I'd look for an open capacitor in the feedback loop first. If it's still present, can't you just filter the unwanted frequencies with a LPF? That's what I'd do and see if that signal was really the problem or where it comes from.
     
  5. studiot

    AAC Fanatic!

    Nov 9, 2007
    5,005
    513
    Is this amp one you are designing and building or an existing one that you are trying to service?

    In other words was the amp once stable and need a faulty component(s) replacing or do you need to introduce a pole into the response curve to bring about stability?
     
  6. PRS

    Thread Starter Well-Known Member

    Aug 24, 2008
    989
    35
    This amp is my own design and is prototyped on copper-clad board. I made pads (nodes you might call them) by scraping off the copper with a knife then tested the each pad for discontinuity with its adjacent pads. Then I soldered the parts on. Granted, it's ugly, but I've used this method of construction before with success. I think I inadvertenly made an oscillator. When the load was an 8ohm resistor all was well; it wasn't until I tried a real speaker that the problem came up. Thanks.
     
  7. Darren Holdstock

    Active Member

    Feb 10, 2009
    262
    11
    If I'm reading it correctly that the HF oscillation at the amp output is at quite a high magnitude, then how come the resistor in your zobel network isn't smoking? If you haven't got a zobel network, then adding one might help. It's always good to have an amp that's intrinsically stable with no load, but the frightening truth is a lot of amps rely on the zobel network to provide some load at HF to maintain stability when no external load is applied.
     
  8. bertus

    Administrator

    Apr 5, 2008
    15,645
    2,344
  9. studiot

    AAC Fanatic!

    Nov 9, 2007
    5,005
    513
    If you have mounted your components as indicated, they are likely to have quite long leads.

    Long leads often leads (ha ha) to parasitic oscillation. Try some base/gate stopper resistors of about 100 ohms in the high gain positions. Ferrites over the leads can also help.
     
  10. PRS

    Thread Starter Well-Known Member

    Aug 24, 2008
    989
    35
    Darren, thanks. I think you have something here. I will experiment with zobel networks. Until now, I had never even heard of them. I'll let you know what happens.
     
  11. PRS

    Thread Starter Well-Known Member

    Aug 24, 2008
    989
    35
    Bertus, thanks for the links. They helped me understand some things. I've never run into this problem, but the idea that a speaker includes a coil hit me just yesterday. I took apart a ruined speaker yesterday and looked at its innards. The coil is not very big, but it as your linked article pointed out it has an inductance of 1mH. That's quite large in my book. If it were to interact with a tiny output capacitance that could form part of an oscillator. And this must be the case because the unwanted oscillations only show up when the speaker is connected. I'll play with it and let you know what happens. Thanks again.
     
  12. PRS

    Thread Starter Well-Known Member

    Aug 24, 2008
    989
    35
    Zobel worked great! Thanks Darren, Byrtus, dsp, and studiot! ;)
     
Loading...