• State Resolved
  • Locked Locked
  • Replies 7 replies
  • Answers 1 answer
  • Subscribers 48 subscribers
  • Views 283 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

LMP2012QML-SP: Op-amp output ripple

Mark Jones
Prodigy 10 points
Part Number: LMP2012QML-SP
Other Parts Discussed in Thread: LMP2012, LMP2022

Hello, we've been looking into the output ripple on our parts which use the LMP2012 op-amp.  The output ripple is measured to be about 18kHz, is there any theory on why this is?  I noticed in the datasheet the chopper frequency is 35kHz but could it possibly drop to 18kHz?  Or would input capacitance contribute to this? thanks.

  • 0
    TI__Guru 53316 points

    Hi Mark,

    The clock frequency is typically 32-36kHz. It's a RC oscillator (so don't expect crystal timebase accuracy).

    The clock frequency is ~35KHz, but it is divided into phases to drive the input switches, so it is possible to see "sub harmonics", particularly if you have mismatched high impedance sources (>100k).

    The chopper switches on the inputs have blow-by charges when they switch - injecting charge spikes into the input. Any input capacitance will integrate these chargers and cause an offset or ripple (depending on the time constant). Think of it like a 35kHz pulsating current noise.

    And if you do not perfectly balance the input capacitance, one input may accumulate larger charges. So if your sensor has a 100k source impedance on the + input, and the feedback resistor equivalent is about ~1k ohms on the inverting input, you would get larger ripple or offset on the positive input.

    This is similar to using equal resistances to balance out the bias current on a bipolar input - but now you are dealing with capacitance balance (equal time constants). You can try to drive yourself nuts trying to balance the capacitance accurately..

    See section 8.1.2 in the LMP2022 datasheet for more info. Also see John Caldwell's note on chopper noise effects:

     e2e.ti.com/.../Chopper-Noise.pdf

  • 0 in reply to Paul Grohe
    Prodigy 10 points

    Hi Paul, thanks for getting back so quickly.  So our design feeds a 1000/1500 ohm wheatstone bridge output into the positive and inverting terminals of the chip.  I believe the effective impedance of these inputs is Rbridge/2.  Would a high resistance (~80k) feedback resistor still cause a larger ripple?

  • 0 in reply to Mark Jones
    TI__Guru 53316 points

    Hi Mark,

    The  impedance that the inputs see would be dominated by the Thevenin bridge resistance, or 1k//1.5k. So the 80k feedback resistor is swamped by that impedance and is not an issue. So it sounds like it is not due to the high impedance. 

    How are you measuring the ripple? Scope? FFT?

  • 0 in reply to Paul Grohe
    Prodigy 10 points

    Hi Paul, I'm using a scope.  And i should mention we have 3.3nF capacitors across the inputs, in+ to ground, in- to ground.  During the investigation, adding two 3.3nF capacitors in parallel from in+ to ground dropped the output ripple to around 2.5kHz.  Not sure if that helps.  Just trying to come up with an explanation of why we are seeing the ripple present.

  • +1 in reply to Mark Jones
    TI__Guru 63650 points

    Adding a large capacitance on the inverting input terminal is typically a very bad idea as this may lead to circuit instability caused by delay in the feedback loop.  Please show the detailed schematic to confirm whether the circuit has sufficient phase margin to prevent oscillation.  Also, adding to Paul's comments, there is typically two-clock cycles for each auto-correction - thus, 36kHz clock would result in 18kHz offset correction frequency. 

  • 0 in reply to Marek Lis
    Prodigy 10 points

    Unfortunately I can't send the schematic but I think this gives us enough to go off of.  Thanks for the help!

  • 0 in reply to Mark Jones
    TI__Guru 63650 points

    No problem.  Good luck!