Connect with us

Noise leaking from PSU to mic amp

Discussion in 'Electronic Design' started by Daniel Kelly \(AKA Jack\), Oct 11, 2004.

Scroll to continue with content
  1. Hi,

    I have just built a PCB which holds a switched power supply and a microphone
    amplifier. The power supply runs a camera whilst the audio amp is powered
    directly from the 3.7v lithium polymer battery. The problem is that the amp
    picks up a lot of noise from the powersupply. So much noise that the amp is
    unusable at the moment. If I turn off the power supply then the amp runs
    perfectly. Please could I ask your advice as to how best to limit this
    noise? I've had a few ideas myself:

    1) The PCB is doublesided and the back of the PCB is a ground plane. This
    ground plane is continuous across the entire PCB so the powersupply and the
    amp both share the same ground plane. If I separate the ground plane by
    cutting the copper with a knife at the boundary between the two circuits,
    will this limit some of the noise that's leaking from the power supply to
    the amp?

    2) Use better shielding on the PSU's inducer.

    3) Build a separate PCB for the amp (something I don't really want to do
    because I'm quite short of time).

    Things I've tried already (but with no success):

    1) Running separate power leads from the PSU and amp to the battery

    2) Putting coupling capacitors all over the place.

    3) Shorting the length of the microphone cable to the absolute minimum.

    4) Shielding the entire PCB with grounded silver foil

    5) Shielding only one circuit with grounded silver foil.

  2. Very difficult to diagnose ( well for me) on the net, but
    try reading AN70 by Jim Williams on the Linear website, it has a good
    section on selecting inductors, and sniffer for locating nasties


    Serious error.
    All shortcuts have disappeared.
    Screen. Mind. Both are blank.
  3. Don Pearce

    Don Pearce Guest

    You need to revise this board radically, because you are trying to
    make it do two things which are fundamentally incompatible - deal with
    large, high voltage switching signals and tine signals from a

    First - as you suggest - slice across the ground plane so that the
    amplifier and power supply are completely separated. Now link the
    ground across to a single point, which should be the ground point for
    the mic connector. You will still get some pickup until you have
    attended to the grounding around the input, and that is a bit too
    detailed to go into here, but get some references to star point
    grounding in order to do this properly.

    Pearce Consulting
  4. Bob Masta

    Bob Masta Guest

    Just a thought, but if you could increase the switcher
    frequency to move the noise above the audio range,
    you could use a simple low-pass filter on the mic amp.
    Probably best used as a last resort after all the "good
    design practice" stuff fails.

    Best regards,

    Bob Masta

    D A Q A R T A
    Data AcQuisition And Real-Time Analysis
  5. That's the first thing I would do. It might help a lot, but not totally
    cure the problem.
    Well, if the circuit doesn't work, then it doesn't matter now little
    time you have. Use a saw and cut the circuits apart.
    Try adding some toroid chokes to the incoming power leads, both + amd -.
    Welcome to the world of analog preamp meets switched power supply. ;-?

    It will probably take a combination of isolation techniques to solve
    your problem.
  6. Ban

    Ban Guest

    Did you by chance use the autorouter to make this board? And aren't you a
    digital guy maybe?
    I have done a mike preamp and a complete DSP with fast parallel flash
    memory on a supertiny board for a cellphone with 2 switching supplies et al
    on 2 sq. inch, and no crosstalk at all. I suspect your problem can come from
    the voltage supply for the electret mike which must be heavily filtered and
    a wrong reference point for the mike preamp, which should be the positive
    supply. A ground plane is not always the reference, a complete understanding
    of the function is needed, determining which way currents are flowing into
    and out of the circuit.
    You will probably need a redesign, this cannot be just updated like a
    software, sorry for that.
  7. Nico Coesel

    Nico Coesel Guest

    No, simply limit the bandwidth of all audio amplifying elements to
    25kHz (or less if permissable) and you'll be fine.
  8. Hi everyone,

    Thanks loads for all your feedback. You're all right - I've designed the
    board badly.

    I tried separating the ground planes, using a star configuration ground and
    putting the mic on its own supply but none of this worked. So I went ahead
    and cut the PCB in half! Which worked fine! It's a bit of a crappy
    solution but it's worked! Now I can propperly sheild the two PCBs from

  9. Tim Shoppa

    Tim Shoppa Guest

    You note in idea 1) grounding, and good grounding is essential, but
    I'm 99% sure that what you're seeing is magnetic coupling into a loop
    in the microphone preamp. Making the preamp circuitry be small would be
    a Good Thing. But with an inductor millimeters away I think this may
    not be enough.
    Shielding the magnetic field from an inductor is rarely economically
    possible. Yes, you can buy mu-metal, but that's a very expensive
    and clunky afterthought. (I once worked with a mu-metal shielded
    phototube that we nicknamed "the planet crusher" because of the
    extreme weight of all the shielding!)

    Toroidal inductors are preferred but not a cure-all (they still leak).
    Geographic separation, if you can be flexible about PCB shape, may help
    you a lot since magnetic fields decrease like 1/(distance cubed). Simple
    orientation of the inductor may help by a factor of ten or so.
    Those would maybe help with electrostatic radiation, but you're
    seeing magnetic radiation.

  10. Great, thanks a lot for your really detailed reply.

  11. AKA Jack\ wrote...
    Two comments: You may or may not be suffering from any magnetic
    pickup, but to the extent you are, sheilding won't help. Second,
    to the extent separation and shielding do help, the same result
    can be achieved on one PCB by adding two separated ground planes,
    and a small over-sheild with tabs soldered into the ground plane.
  12. Tim Shoppa wrote...
    To the extent that Daniel / Jack is suffering from magnetic pickup,
    we can point out this pickup is proportional to the cross-section
    area (whose plane is at right angles to the magnetic vector), so
    reducing this area (e.g. keeping signal paths immediately adjacent
    to ground return paths, etc.) will help dramatically. Second, keep
    in mind the effectiveness of twisted-wire pairs in rejecting magnetic
    pickup; each half-turn's pickup is canceled the opposite polarity of
    the next half turn. It's at least theoretically possible to cancel
    all the pickup entirely in an analogous manner, with a loop of input
    wiring arranged to add in its pickup with the opposite polarity. :>)
  13. martin griffith wrote...
    The small magnetic-field sniffer probe section in Jim's appnote is
    a meaty and excellent source of information. It's appnote E101 by
    Bruce Carsten Associates, Inc., 6410 NW Sisters Place, Corvallis,
    Oregon 97330 541-745-3935, You can make
    your probes from the drawings he provides, but they sell several
    versions for $300 to $500, Looks like good investments.
  14. Hi,

    Thanks for the reply.

    Yes, I tried separating the ground plane before I cut the board apart, but
    that didn't seem to work (although maybe I didn't separate the ground planes
    far enough - I only made a thin stanley knife cut).

    Next time I design a PCB like this, I will do as you suggest.

Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day