Connect with us

Distortion from audio power amp

Discussion in 'Electronic Design' started by Terry Pinnell, May 14, 2004.

Scroll to continue with content
  1. Don Pearce

    Don Pearce Guest

    The amp is oscillating because there is no dominant pole compensation.
    A small cap from collector to base of Q2 (the voltage amplifier)
    should make things much nicer.

    d
     
  2. Thanks, I'll try that tomorrow.
     
  3. legg

    legg Guest

    You could save everyone a lot of time by researching the source of
    your original schematic a bit more carefully. As drawn, there are a
    host of problems that people have been trying to point out. These may
    not have been present in the original source.

    This habit of documenting a 'problem' then posting it, without trying
    to examine and fix it first, yourself, is becoming tiresome. The
    circuit works no better than expected, as drawn, whether it's been
    around for 30 years, or not. Problems don't fix themselves. Errors can
    easily be carved in stone.

    RL
     
  4. Rich Grise

    Rich Grise Guest

    Well, probably R4 is to adjust out the crossover distortion, and
    R2 sets the closed-loop gain, which you've got set way too high.
    Of course, that could be an artifact of the open circuit between
    the C2-R7-Q1 node and the R9-R10-R4-Q4-R11-C1 node.

    Good Luck!
    Rich
     
  5. normanstrong

    normanstrong Guest

    Who designed this amplifier, and when? It looks like one that could
    have been designed 20 years ago. There are also several 'mistakes'
    in it that have to be fixed just to get it to work at all. IOW, this
    is not the way to design an amp.

    Norm Strong
     
  6. Terry Pinnell wrote...
    I think Don meant to place the cap from Q1's base to Q2's collector.
    But first change R7 to 47k, and R2 to 2.7k plus a 5k pot. Then use
    3.3pF compensation (for 1 MHz loop bandwidth). There should also be
    a 1k resistor in series with C3 for stability over the gain range.

    Thanks,
    - Win

    (email: use hill_at_rowland-dot-org for now)
     
  7. Rick

    Rick Guest

    Also, the difference between the base of Q4 and Q3 is 3*Vbe, so it
    would be better to bias Q3 and Q4 with a Vbe-multiplier rather than D1,
    D2 and R4.
     
  8. Have I been ignoring advice that 'people have been trying to point
    out'?

    The circuit follows the original in all significant respects. And I
    studied it rather carefully. Apparently more carefully than you
    studied my posts, or you wouldn't have made such an assumption. If
    you're seriously interested in the design issues, rather than tossing
    out patronising comments, take a look at the originals yourself.
    http://www.terrypin.dial.pipex.com/Images/AudioPowerAmpOriginal1.jpg
    http://www.terrypin.dial.pipex.com/Images/AudioPowerAmpOriginalText.gif
    Excuse me? If you're tired, I suggest a short nap. As for not trying
    'to examine and fix it first', how did you arrive at that second
    grossly inaccurate assumption?
    Wow, now I never would have thought of that!
    And your point is? Apart from trying (and failing) to appear
    important?

    If you have nothing constructive to offer, please don't contribute at
    all.
     
  9. No, but aim to do so tomorrow. Using current limiting from bench
    supply unit too.
     
  10. As mentioned earlier in

    Subject: Re: Trouble shooting push-pull amp
    Date: Thu, 13 May 2004 09:21:04 +0100

    it's about 30 years old. Comes from Babani Press series, Book #221,
    "28 Tested Transistor Projects", page 21. I'll check author's name
    (book is still in shed/workshop) and let you know tomorrow.
     
  11. Thanks. Actually, R2 sets DC level of output, intended to be half
    supply voltage, i.e. about 12V here. Gain is set by R1.

    Open circuit? Not sure I follow. I fixed the hair-line track, if
    that's what you mean, as reported in other thread yesterday:
    Subject: Re: Trouble shooting push-pull amp
    Date: Thu, 13 May 2004 20:49:37 +0100
    Message-ID: <>
     
  12. Terry Pinnell wrote...
    Terry, read my post and think about it. If R1 is set for low gain,
    yes it does attentuate the input, BUT it thereafter programs a very
    high ac gain, nearly open loop. The gain for an inverting opamp,
    which this acts like, is R_feedback / R_inout. R_feedback is R7 in
    your case. Note there's no R_in, so whenever the gain pot is set to
    zero, R_in = 0 and the amp's "gain" is actually *very* high. Weird.

    That's just one of a half-dozen serious problems with your circuit.

    Thanks,
    - Win

    (email: use hill_at_rowland-dot-org for now)
     
  13. It's not so much the current limiting I was concerned about, but
    with R11 sitting in the collector makes it, in my view, a sort
    of an unbalanced output stage. Since your output also shows a
    rather unbalanced behaviour, with the negative half periods of the signal
    being much better that the positive ones, my first try
    would be to give it as much balanced behaviour as possible, by
    moving that R11 to the emitter side. 2N3055 or cheap no brand
    look alikes have a lousy Hfe, sometimes as low as 15-20, but
    sometimes as high as 140. At least that is what I remember from
    testing a batch of cheap 2N3055's two decades ago ;-)
     
  14. "Winfield Hill" <> schreef in bericht

    [snip]
    Oliver Hardy would have said "Here's another fine mess
    you've gotten yourself into!"

    ;-)
     
  15. legg

    legg Guest

    Re-examine the circuit.

    Construct the circuit as indicated.

    this should include-

    Installing R1, as drawn, or placing it in series with old C1(new C3),
    to swamp out variability of source impedance or vol control setting on
    the gain of circuit.

    Used fixed value resistors as indicated for R2,R5, R6 and R7 in the
    documentation column addressing 24V operation.

    Only make changes after familiarizing yourself with the original
    circuit's function.

    Most of the suggestions posted wouldn't hurt.

    I assume you are fixing something made a long time ago. Better
    references existed at that time.

    The quasi-complementary output circuit (Q2 and onwards) was shown as a
    standard circuit in the GE and RCA transistor application manuals as
    early as 1960, using pnp germanium power transistors. These were
    possibly based on articles by H.C.Lin published around 1956.

    http://203.44.53.131/QuasiComp/Lin.zip

    other articles on same basic circuit at

    http://www.alphalink.com.au/~cambie/#Quasi

    The common references continued in the same manuals over the decades,
    to include the silicon npn variations in the same circuit. The parts
    used in the circuit from ETI'76 are identifiable in the basic
    references, by function, as are the functions of those parts that are
    omitted.

    RL
     
  16. Yes, but what is the reason for R11 (in collector bottom 2N3055).
    That transistor has high Z collector out, why add .5 Ohm? It would do
    nothing?
    JP
     
  17. OK, thanks, I missed that. Win made same point.
    The text suggests a 220k preset as an alternative to a fixed resistor.
    That option seemed more appropriate when I built this project, as
    exact supply voltage was not precisely 24V, and somewhat
    indeterminate. R5 and R6 *are* fixed. Original R7 is my R4, which
    again I made variable for same reasons of flexibiliy mentioned above.
    Yes, 1981 according to my notes.
    Interesting stuff, but couldn't find any close similarity in any of
    those circuits to this one.
    I'm impressed by your archives - my ETI mags were dumped many years
    ago!

    I do intend to implement most of the suggestions made, and will report
    back on the results.
     
  18. OK, I'll add a small series R to the R1 preset. But I still don't
    understand how Rin can ever be zero in practice. I had assumed it
    would be sufficient to prevent the sort of 'runaway gain' you warn
    about. Even on my bench test, my function generator (the PM5134 we've
    been discussing in another thread), has output impedance of either 50
    or 600R. I've been using the lower setting, but a gain of 1k/50 = 20
    hardly seems anything to get too excited about? And, although I'd have
    to check this, I think the actual previous stage (dog bark synthesised
    sound), would have Rs > 50.
    So - you don't like this circuit too much then? <g>
     
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

-