Connect with us

Single power rail opamp

Discussion in 'Electronic Design' started by Neil Bernard, Nov 15, 2003.

Scroll to continue with content
  1. Neil Bernard

    Neil Bernard Guest

    Hi I want a balanced output using two opamps (for audio) using a single rail
    power supply +9v any ideas?
     
  2. TS922?
     
  3. 1) use a transformer; or
    2) use DC-blocking capacitors.

    If you do #1, you won't need the second opamp, and you'll have better
    impedance balancing.

    Or, 3) don't bother with the so-called "balanced output". See my earlier
    post responding to a question on balanced output by Ban. Since what matters
    for audio is balanced *impedance*, not balanced *voltage*, all you really
    care about is that the + and - signals have precisely the same source
    impedance. See whitepapers by Bill Whitlock on the Jensen Transformer web
    site.

    Or, 4), make a "virtual ground" at 1/2Vcc by taking a voltage divider and
    then amplifying it with an opamp. But there's no good reason to use that
    approach given that the other three are better.
     
  4. Neil Bernard

    Neil Bernard Guest

    Like the transformer idea but trying to make this a small as possible. I
    like the idea of a voltage divider with an opamp output buffer do you think
    this will cause any problems?
     
  5. Is there some reason you don't like the balanced-impedance,
    unbalanced-voltage approach? It really is quite valid.

    If you're trying to run from batteries, doing the virtual ground buffer is
    going to waste some juice; and it takes an extra opamp.

    What, exactly, are you trying to accomplish here? What problem are you
    trying to solve?
     
  6. Neil Bernard

    Neil Bernard Guest

    Trying to understand what you mean by balanced impedance?

    The application is going to be a Micro controlled sine osc using a digital
    pot for amplitude control. this is for ID and testing of levels on MIC and
    LINE level circuits. I need the output to be balanced but alos need the
    unit to balanced
     
  7. Okay. Reading some of the whitepapers on Jensen Transformers' and Rane's
    web sites will help you understand how balanced audio works. But here's the
    short version.

    The point of a balanced input on a device (e.g., on a mic preamp) is to be
    able to detect a differential signal in the presence of common-mode noise.
    You assume that there is some external noise source that is coupling to your
    signal lines, and you try to set things up so that the signal will couple
    equally into both lines, and then you subtract one from the other to get the
    difference.

    There are two key issues here: first, getting the interference signal to
    couple equally to both lines; second, doing the subtraction.

    Doing the subtraction entails having good common-mode rejection in the
    receiver. Ideally you can measure a 1mV difference between two lines even
    though both of them have got a 120V common-mode signal on them, for
    instance. Note that this has NOTHING TO DO WITH THE TRANSMITTER - it's all
    about the design of the receiver. However, it's easier to achieve if you
    can assume identical source impedances on both signal lines, so the
    transmitter does affect it some.

    Getting the interference signal to couple equally is the big one. You've
    basically got a voltage divider: whatever capacitive or inductive factors
    are coupling the signal from the interference source to your signal lines on
    one side, and then whatever impedance from the signal line back to ground.
    The source impedance is a lot lower than the load impedance, and they're in
    parallel electrically, so the source dominates. Imagine that one of your
    signal lines has a 10 ohm impedance to ground, and the other has 20 ohms to
    ground. Then the interference signal is going to couple twice as strongly
    to the 20 ohm line! Obviously this is going to mean that no matter how good
    your common-mode rejection is at the receiver, there will still be
    differential signal and thus interference.

    So, it is very important that your transmitter have equal impedance to
    ground from both signal legs, at all frequencies (not just frequencies of
    signal, but frequencies of interference). Keep in mind that opamp circuits
    may have low output impedance at frequencies where the opamp has gain, but
    at frequencies past their gain bandwidth, they may have fairly high
    impedance, several hundred ohms or more.

    Now, what about balancing the output voltage? Well, it helps for efficient
    power transmission. But you don't care about power transmission, because
    you're not the phone company. And it helps to reduce crosstalk. But your
    application isn't sensitive to crosstalk, as far as I could tell. And, it
    does help a bit in the event that one or the other signal line shorts to
    ground, in a real-world signal transmission application. But it does NOT
    HELP with regard to noise, and it is NOT REQUIRED as long as the receiver is
    balanced. You do end up with less signal (because half of it looks like
    common-mode to the receiver), but then, you only took half as many
    amplifiers to get there.

    So, in summary: use an opamp for your + output, and play some tricks (see
    Jensen web page) to get the output impedance constant and low at frequencies
    above where the opamp has gain. Then, simply connect your - output to
    ground, THROUGH AN IMPEDANCE MATCHED NETWORK that gives it the same source
    impedance as the + output at all frequencies.

    -walter
     
  8. Ban

    Ban Guest

    Walter Harley wrote:
    || ||| Trying to understand what you mean by balanced impedance?
    |||
    ||| The application is going to be a Micro controlled sine osc using a
    ||| digital pot for amplitude control. this is for ID and testing of
    ||| levels on MIC and LINE level circuits. I need the output to be
    ||| balanced but alos need the unit to balanced
    ||
    || The point of a balanced input on a device (e.g., on a mic preamp) is
    || to be able to detect a differential signal in the presence of
    || common-mode noise. You assume that there is some external noise
    || source that is coupling to your signal lines, and you try to set
    || things up so that the signal will couple equally into both lines,
    || and then you subtract one from the other to get the difference.
    ||

    I had a similar question and decided to use the SSM4142S by AnalogDevices.
    This part you can get for free as a sample, mine are supposed to be shipped
    beginning of December. The advantage with this part is the crosscoupled
    feedback from the outputs, so if the connected equipment is unbalanced, the
    remaining line doubles its voltage and you have the same output. This is
    just as a transformer would react.
    I do not advise you the transformer choice, because for testing purposes you
    do not need that 120dB rejection and since you want to power the tester from
    batteries there will be no issue of hum either. Better you save those 80
    bucks and have a small instrument that fits inside a pocket.

    ciao Ban
     
  9. We still don't really know much about the OP's application. (For instance,
    it was my surmise that it's to be battery-powered; I don't believe he has
    said so.) We don't know whether he's planning on making one or 10000, we
    don't know whether it's to be handheld, we don't know what the budgetary
    issues are, we don't know what kind of testing he intends to use it for, and
    thus we don't know anything about the required performance.

    But, *presuming* that he's trying to make the equivalent of a Neutrik
    Minirator (which is an excellent device, and very affordably priced), the
    SSM chip could be a good option. Or, he could just use DC blocking caps
    like most other folks who build single supply devices...

    How good is the impedance matching of the SSM's outputs at high frequencies?
    What's its quiescent supply current?

    -w
     
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

-