• State Resolved
  • Locked Locked
  • Replies 8 replies
  • Subscribers 49 subscribers
  • Views 1214 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.

LMC6484: large offset in noninverting voltage follower configuration

Samuel Neis
Prodigy 50 points
Part Number: LMC6484
Other Parts Discussed in Thread: LMC6482

I have one gate of an LMC6484 acting as a noninverting voltage follower.  The input signal to the circuit is at about 1.6V, with 300kohm of equivalent thevenin resistance and attached to the opamp's positive input.  The opamp's output is connected directly to the negative input to form a voltage follower circuit.  the opamp output sees a load of about 2k.  The opamp is connected between 5V and 0V for power.  The circuit seems to have an offset error of 50mV, while the datasheet indicates it should be about 0.1mV.  As far as I can tell, the circuit is stable with the input at 1.60V and the output at 1.65V - that is, I don't see any oscillation going on, or any indication that the outputs are railing out.  Multiple boards wit this circuit on it have the same issue.  Other circuits on the same board, that use the other 3 gates of the quad opamp package, do not exhibit this problem.  They have lower input resistance so I suspect this may have something to do with that 300k resistance in front of this particular circuit.  I have tried tying another 300k resistance between the negative input and ground, to compensate for input bias currents but it has not had any effect.  Any idea why this circuit might cause the LMC6484 to have a larger-than-specified voltage offset?

  • 0
    Guru 140200 points

    Hi Samuel,

    what is connected to the output of OPAmp? Is there any capacitive load?

    300k source impedance is rather high. It can work like an antenna and "receive" HF signals like AM modulated radio signals. If the input stage of OPAmp is not fast enough, the radio signals can be demodulated at nonlinearities of the input stage. The result can be a considerable offset voltage. A small capacitance directly from the +input of OPAmp to signal ground can solve the problem.

    Kai

  • 0
    TI__Guru 51846 points
    Hello Samuel,

    The LMC6482 has one of the lowest bias currents in the industry (in the 10's of fA's at room temp). You should never see that much bias current at the input unless it is damaged.

    Exactly *how* are you measuring this difference?? Are you using the same meter/scope to probe both input and output?

    Scope probes and most DMM's present a 10Meg load to the measurement point.

    When probing the input pin, you are placing a 10M resistor load on the input, creating a voltage divider with the ~300k source impedance.

    Noodling out a voltage drop with a 1.6V source and 300k + 10M voltage divider, I get 1.55V - a 50mV drop.

    Use a second meter at the output and probe the input and see if the output moves.

    Using one meter, you will not see the output drop 50mV when probing the input. You need to watch both the input and output. That would also explain why the "lower" impedance sources are not affected as much.
  • 0 in reply to Paul Grohe
    Guru 140200 points
    Good catch, Paul!

    Kai
  • 0 in reply to Paul Grohe
    Prodigy 50 points

    Thanks, I had thought of this and tried measuring with 2 meters (though I found the issue while probing with only one).  With one meter on the output/neg input (they're shorted together to be a follower), I connected and removed another meter to the pos input.  The meter on the output/neg node did not change when the other meter was connected and removed.

    Oddly, when I connect just one meter between the two points (instead of each point referenced to ground) the difference goes away.

  • 0 in reply to Samuel Neis
    TI__Guru 51846 points

    Hello Samuel,

    Measuring no difference between input and output is showing that the amp is working correctly. The difference between the two points is <1mV and does not generate enough current through the meter (~100pA) to drop across the source resistance.

    I am assuming you are measuring the input between the +IN pin and GND (V- pin), and the output between the OUTPUT and GND.

    What exactly are you using to measure? What model DMM/Scope+probe? Handheld? Bench? Or is this some sort of computer DAQ card?

    Can you provide a schematic of your circuit? Showing items connected to the input and output.

    Did you watch for oscillations *while* probing the voltages with the DMM? It is possible that the high input capacitance of the DMM (usually >500pF)  is causing the amp to oscillate *while* you are probing it, creating the "offset". I have seen this many times. So a 15pF scope probe does not cause it to oscillate, but a 500pF DMM does.

    Try probing the output with the DMM in the AC voltage mode. It should be 0V if there is no oscillation. If you see more than ~100mV, it is oscillating.

    Or simultaneously watch the output with the scope while probing the output with the DMM and watch for an oscillation in the 10k-500kHz range (depending on actual cap load).

  • 0 in reply to kai klaas69
    Prodigy 50 points
    in the end it looks like both issues were actually a factor
  • 0 in reply to Paul Grohe
    Prodigy 50 points
    I added some capacitance to the pos input based on Kai's suggestion and then tried several of these measurement techniques again. This time it became apparent that the DMM (a fluke 189) was loading the positive input node down a little and that the output was just following it. So I think the cap fixed something that made all the interactions more predictable.
  • 0 in reply to Samuel Neis
    TI__Guru 51846 points
    Hi Samuel,

    Then most likely what was happening is that you were injecting environmental noise (line noise, switcher noise, etc) into the high impedance (300k) point. The inputs to DMM's are very noisy, and the DMM leads were acting like antennas - picking up local noise. The extra noise was integrated as the "offset".

    Touch the neg lead to ground, and just hold the positive probe in your hand...you will probably not see 0.000V. That "extra" noise voltage was added to the actual DC value on the (not loaded) input pin. Gott'a be careful measuring high impedance.

    Adding Kai's capacitor low-pass filtered (smoothed) the input - to the point where you can see the actual change due to the loading. Just be sure adding this capacitor does not cause any unexpected delays.

    The low output impedance of the output as not affected by the lead noise - so the output *was* reflecting the actual input voltage.

    Murphy lives!