Connect with us

anyone familiar with this topology?

Discussion in 'Electronic Design' started by justin, Sep 11, 2004.

Scroll to continue with content
  1. justin

    justin Guest

  2. I originally saw it as a reader's idea in Electronics,
    about the late 70's to early 80's.

    "Getting Power and Gain out of the 741-type op amp."

    By Pedro P. Garza, Jr. General Electric Co, Apollo
    and Ground Systems, Houston, Texas.

    I later pinched the idea and pushed it to a 1250VA
    amplifier (8+8 Mosfets in the o/p stage, fan cooled),
    used in a test jig to stimulate wound components at
    full power.

    Winfield Hill also tried out the idea last year, with
    a high voltage power amplifier. I don't know what the
    final outcome was.

    Interestingly, neither of us felt comfortable with that
    capacitor from output back to the opamp output pin. It
    gives wide bandwidth but an uncomfortable feeling of an
    oscillator, just waiting for an excuse.

    I used a capacitor from the opamp output pin to 0v,
    which Win later analysed and remarked that it formed a
    nice and tidy Zobel Network with the opamp output Z.
     
  3. Oh dear...you have my sympathies...
    Personally, I think this type of method is a truly dreadful way to
    design an amp. There are so many issues with it. The output devices bias
    is completely undefined, unless it set to essentially zero. The transfer
    function is not well controlled or modelled, in addition to poor hf
    response. Its really grungy, like finger nails down a blackboard sort of
    thing. Those that use this type of circuit should be shot on sight, and
    have their entrails thrown to the vultures.

    In general, putting an op amp in the feedback loop is bad news. It justs
    gets you extra poles that you could avoid with a discrete design.
    Ahmmm...

    Kevin Aylward

    http://www.anasoft.co.uk
    SuperSpice, a very affordable Mixed-Mode
    Windows Simulator with Schematic Capture,
    Waveform Display, FFT's and Filter Design.
     
  4. Tim Wescott

    Tim Wescott Guest

    I'm generally deeply suspicious about using parts outside of their
    original intent -- what happens when the process (or temperature)
    changes, or purchasing cleverly buys someone else's chips to save a few
    pennies?

    Certainly this may do for a clever one-off, but I'd do more than
    hesitate before putting it into production.
     
  5. Pooh Bear

    Pooh Bear Guest

    Yup, seen that design concept.

    Give it a wide berth. It's full of problems. Not even fitted with emitter
    resistors for DC bias stability on the output devices !

    Truly horrid. It belongs back in the early 80s too.

    Graham
     
  6. Kevin Aylward wrote...
    It worked out very well, Tony. Remember the schematic I sent you?
    With the crummy circuit in the gif, perhaps, but not with a sensible
    circuit. In fact it's remarkably easy to define the class A current.
    And control the crossovers, because one is using current-source mode
    to drive the output, with drive circuitry pegged to the rails, rather
    than an awkard floating bias circuit.
    It has a well behaved transfer function, although the load figures in.
    Which in the end may limit bandwidth, but allows good output linearity
    and is especially well suited for current drive amplifiers.

    In the case of Mark Alexander's current-feedback design (Analog Devices
    AN-211), a much wider bandwidth was achieved than with common circuits.
    Why don't you tell us how you really feel?
    Actually, not here, because with the opamp acts as a voltage-current
    converter, via the opamp power-rail cascode transistors. The current
    transfer function is determined by resistor ratios: the high-voltage
    rail resistor, the opamp's output load resistor (and capacitor!), and
    the Darlington output-transistor's emitter-degeneration resistor.

    Right, Tony, it looked very bad.
    Right, a beautiful addition suggested by Tony, nicely enhancing the
    transfer function.
     
  7. I wouldn't hesitate at all, to put it in the bucket...

    Kevin Aylward

    http://www.anasoft.co.uk
    SuperSpice, a very affordable Mixed-Mode
    Windows Simulator with Schematic Capture,
    Waveform Display, FFT's and Filter Design.
     
  8. Genome

    Genome Guest

    The proper way to do it is to give the op-amp a local gain of about 3 and
    ask it to drive a low value resistor. Then you take feedback from the power
    stage output to the op-amps output.

    Add some emitter degeneration for the darlingtons.

    Use a TL074. Implement a Vbe multiplier with two of the op-amps driving each
    other and scale things through the bias chain. Use one as a DC servo and use
    one as the main amp.

    SEE ABSE

    DNA
     
  9. I've seen this concept before -- an op-amp whose range is extended by
    putting it in the middle of a swinging power supply. I think I've even seen
    it in a TI book, but I'm not sure.
     
  10. Jim Thompson

    Jim Thompson Guest

    Oh, Bull Puckey!

    In the early '70s I built many a hybrid power amplifier that way,
    using, of all things, a 741 as the core.

    And a variant that allowed a low dropout regulator, with regulated
    zener current.

    In the '60s this was done at Motorola with a standard NPN chip
    (P-substrate) and a PNP chip (N-substrate) so that both good NPNs and
    PNPs were attained.

    ...Jim Thompson
     
  11. Roy McCammon

    Roy McCammon Guest

    I inherited the sustaining on such a design. Its been rock solid
    and free from problems for years. I have to admit that although I
    understand it in the dc-low freq sense, I don't understand it's
    stability criteria sufficiently well to be comfortable.
     
  12. John Larkin

    John Larkin Guest


    I do this as a transimpedance stage in all my NMR gradient amps. These
    are inherently current-output amps, so the opamp rail currents
    (cascoded, as shown here) then drive output current mirrors, more
    precise versions of the PNPs and NPNs used here. I use power mosfets
    in closed loops as the mirrors. The biggest one I do this for is +-180
    volts, 100 amps (18 kilowatts) peak out.

    I don't know who did this first, but it's a very old trick. It's in
    National appnote AN-31 by Robert Dobkin, February 1978.

    John
     
  13. Fred Bloggs

    Fred Bloggs Guest

    The type of output is what is regulated by feedback-and that looks like
    voltage to me. That 4n7 at the output looks more like a wide band
    negative feedback loop around the OA output stage to cascode to
    Darlington at high frequencies.
     
  14. John Larkin

    John Larkin Guest


    I sell a million dollars or so a year of amps that use this as a key
    stage. It works beautifully to solve an otherwise very nasty
    current-splitting drive problem.

    John
     
  15. Ken Smith

    Ken Smith Guest

    Some op-amps have a nearly constant no-load supply current that is fairly
    low. If the op-amp's supply current is not well controlled or is at all
    large, this circuit will be hopeless.

    This sort of circuit only works where you don't care about the distortion.
    The output section is running class C. If you try for AB operation the
    slop in the supply current spec will make your design unreproducable.
     
  16. John Larkin

    John Larkin Guest

    Right; I don't do that. I just use the opamp as a unity-gain follower,
    with an extra resistor from output to ground. This output resistor
    defines the transimpedance. So the input signal voltage becomes opamp
    supply rail currents, with essentially perfect crossover behavior.

    I'm driving inductive loads, and I want my output impedance to be as
    high as possible, and this topology does that nicely. But for audio or
    motors or whatever, I'd go for a topology with low open-loop
    impedance, so loop dynamics wouldn't be too load dependent. But then I
    don't design audio!

    John
     
  17. Jim Thompson

    Jim Thompson Guest

    This thread sets the record for so many wrong statements about a
    circuit configuration.

    It's class-B, not class-C, but class-AB would indeed be difficult
    (without some extra servoing).

    The open-loop dead-band can be made reasonably small, and distortions
    way under 1% are easy to attain... based on hands-on experience with
    this configuration ;-)

    Except for the OpAmp Q-current, there is really no difference between
    this approach and the Sziklai connection (AofE, p95) commonly used in
    power amplifiers.

    ...Jim Thompson
     
  18. John Larkin

    John Larkin Guest

    I run this, I suppose, AB, in that I run a modest quiescent current, a
    few percent of peak, in the output fets. The opamp supply currents
    drive mosfet-based current mirrors. As the signal goes positive from
    zero, the P-side fets increase their current from the idle value, and
    if the sig goes negative the N-side fets increase. There's really no
    crossover distortion, basicly perfectly sharp corners around the
    zero-signal point. That's a little hard to think about, but it works.

    My amps settle to under 10 PPM in 50 usec (limited partly by current
    shunt eddy effects) and have noise and distortion of a couple of PPM.

    This circuit paid for my house and is putting The Brat through
    Cornell, so I have to defend it.
    If I were given unlimited funding to design the perfect current
    splitter custom linear IC for my gradient amps, I'd probably wind up
    with the opamp I'm using.

    John
     
  19. Ken Smith

    Ken Smith Guest

    I still maintain that "the output section is running class C" by which I
    mean
    that the op-amp's output current needs to be greater than about

    1.4V/680 - 1.4mA = 0.66mA

    in either direction to turn on one of the output transistors. This means
    that there will be a dead band while the internal nodes of the op-map
    slew. If the dead band was small enough to be called zero, I'd agree that
    it is a class B output.
    I have used basically the same configuration and yes the distortion can be
    small well below the frequency limit of the circuit. As you go up in
    frequency, the dead band time does not decrease as fast as the period does
    so the distortion gets much worse.
     
  20. John Larkin wrote...
    Hear, hear. I haven't seen your circuit, but I attest to the
    viability of the architecture; very powerful if properly used.
     
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

-