Connect with us

OP-Amp disable

Discussion in 'Electronic Design' started by Syd Rumpo, Mar 13, 2013.

Scroll to continue with content
  1. Syd Rumpo

    Syd Rumpo Guest

    I have an existing device with low-noise differential amplifier at the
    front end. It uses two LT6200 op-amps in inverting configuration and
    has a gain of ten.

    Occasionally, that's too much gain, and an attenuator is pugged in in
    front of the amplifier.

    I want to make the gain electronically switchable without hurting the
    noise performance, so I want to avoid analog switches.

    These op-amps have a disable pin. It seems that if I disable the op-amp
    then the signal goes straight through the input and feedback resistors
    and on to the next stage without gain. That's handy as I can control
    the disable pins with a spare uC output.

    As long as the signal doesn't exceed the supply at the op-amp pins, that
    should work fine. Shouldn't it?

    Cheers
     
  2. Syd Rumpo

    Syd Rumpo Guest

    Of course, the polarity will change, but that's no problem.

    Cheers
     
  3. The 'disable' configuration inverts (or rather doesn't invert) the
    signal, right?
     
  4. Guest

    To me, the "shutdown output leakage" spec (100nA, typ) says the
    output is safely tri-stated in shutdown.

    It's not clear for the inputs-the internal schematic isn't detailed
    enough. You might try it--drive a shut-down part thru 10k, check for
    input clamping / distorted output.
     
  5. A little relay?

    What's the source resistance and feed back resistance?

    George H.
     
  6. Syd Rumpo

    Syd Rumpo Guest

    Yes, that's firmware-fixable.

    [Actually, with a gain of 1, you could make a low component count
    synchronous rectifier. For this chip, the shutdown turn off and on
    times are 180ns]

    Cheers
     
  7. Syd Rumpo

    Syd Rumpo Guest

    Harder to retrofit compared to a track cut and short piece of wire.
    This uses 100R source, 1k feedback.

    <snip>

    Cheers
     
  8. So when the opamp's disabled you've got ~1.1k of source imepdance...

    That 'defeats' the nice low noise opamp... what's after the LT6200?
    Is it low noise too?

    George H.
     
  9. Syd Rumpo

    Syd Rumpo Guest

    Noise at unity gain doesn't really matter. There are PGA stages after
    this which get wound right down when the input is high. Normally, the
    input is a low level from a very long cable, but a short cable is used
    for lab testing which then gives too high a level for the fixed gain
    front end.

    The LTspice simulation works - often things like shutdown aren't modelled.

    Cheers
     
  10. Interesting thought.

    One could moderately light up a good LED with the bias current of that
    op-amp.. 50uA worst-case... wow. There is a heavy price to be paid for
    the high GBW and <1nV/sqrt(Hz) noise performance.

    You might want to have a close look at figure 1-- the /SHDN pin
    clearly is shown to affect (probably kill) the front end bias.. there
    is also a back-to-back diode pair effectively across the inputs. The
    latter could cause issues if the signal level is high enough.

    http://cds.linear.com/docs/en/datasheet/62001ff.pdf
     
  11. Syd Rumpo

    Syd Rumpo Guest

    <cut and paste reply to Phil Hobbs>

    Noise at unity gain doesn't really matter. There are PGA stages after
    this which get wound right down when the input is high. Normally, the
    input is a low level from a very long cable, but a short cable is used
    for lab testing which then gives too high a level for the fixed gain
    front end.

    Cheers
     
  12. Syd Rumpo

    Syd Rumpo Guest

    Good spot, I'd clocked the supply rail diodes but hadn't noticed the
    diode pair across the inputs. That will kill it, damn.

    Now where did I put that relay catalogue?

    Cheers
     
  13. Syd Rumpo

    Syd Rumpo Guest

    That's worth a try, thanks.

    Cheers
     
  14. Frank Miles

    Frank Miles Guest

    Or use a resistor from the input {it must have been low impedance from
    your other comments} to the noninverting input, avoiding a second analog
    switch.
     
  15. That's what I was thinking. 100k should be enough to stop the input
    flapping around.
    Say, if used as a +1/-1 gain stage for a lockin, I don't really care
    about a 1% |gain| error, do I?

    Or, to say in another way, if I had a +1/0 amp, it'd still work fine
    in a lockin, I'd just give up 1/2 the signal.

    George H.
     
  16. Hmm, scratches head... The input is through R5, I assume.

    George H.
     
  17. Frank Miles

    Frank Miles Guest

    If this for an opamp that is turned off - don't see how. Either the amp
    is OFF, and the NI input is at the same potential as the Inv input (and
    small input offsets shouldn't matter) _OR_ the amp is ON, the NI input is
    grounded via analog switch - so the resistor doesn't matter. Of course
    this assumes that the Ron*Coff product is low enough to justify these
    assumptions.

    The other technique for switching between inverting and noninverting uses
    the technique that you describe, akin to the well-known 1-pole all-pass
    configuration. A SPDT analog switch would be ideal for this, either
    connecting the NI input to the overall input (+1 gain) or ground (gain=-
    Rf/Ri). I hadn't thought of that before - it could well be better than
    the 'shutdown' method as it would result in reduced offset changes and
    maintain impedance buffering, not to mention avoiding using an opamp in
    ways not fully specified on the data sheet.
     
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

-