Author Topic: Instrumentation-amp: unequal gains for +Vsignal and -Vsignal?  (Read 2102 times)

0 Members and 1 Guest are viewing this topic.

Offline awallinTopic starter

  • Frequent Contributor
  • **
  • Posts: 694
Hi all,

I'm designing a precision (I hope!) temperature measurement circuit. I'm simulating/designing with NI Multisim.

The idea is to use AC excitation for a pt100 RTD sensor.
With 500uA through a 116 Ohm pt100 resistor I get a signal of 58 mV. With the same current reversed I obviously get a -58mV signal (these are the same to 6 digit precision in the simulator).

I've connected the resistor to an instrumentation amp, currently the AD8429, with a gain of about 76.
With a positive current I get +4,39930 V out, with the current reversed -4,41314 V out.
So the gain for the +signal is 75.85 and for the -signal 76.09
That's a difference in gain of about 0.3% (if I calculated correctly)

This is not because the current through the resistor would be unequal, the simulator shows +/-500uA with 6 digits precision.
Also not because the actual input to the instr-amp would change, probing the inputs I get exactly +58mV and -58mV on the input.

So I have to conclude that the instrumentation amp has unequal (DC-) gains for positive and negative swinging signals?!?

What am I missing?  :) Is there a better instr-amp I should be using?
 

Offline Paul Price

  • Super Contributor
  • ***
  • Posts: 1419
Re: Instrumentation-amp: unequal gains for +Vsignal and -Vsignal?
« Reply #1 on: June 12, 2013, 01:10:21 pm »
You are probably seeing the influence of ground loops of the power supply affecting your reading that you are switching the polarity of, not the D.U.T.

You should, if possible, try this same test with a totally isolated power source to switch polarity of, i.e. battery power.
« Last Edit: June 12, 2013, 01:13:46 pm by Paul Price »
 

Offline Kremmen

  • Super Contributor
  • ***
  • Posts: 1289
  • Country: fi
Re: Instrumentation-amp: unequal gains for +Vsignal and -Vsignal?
« Reply #2 on: June 12, 2013, 05:59:38 pm »
Is your REF input in the actual circuit solidly connected to the signal common? Any deviation from ground level there is going to shift the output (by design).
You can verify this by changing the gain and seeing if the difference is proportional to the gain or not. Any offset in REF will not follow the gain setting.
Nothing sings like a kilovolt.
Dr W. Bishop
 

Offline AndreasF

  • Frequent Contributor
  • **
  • Posts: 251
  • Country: gb
    • mind-dump.net
Re: Instrumentation-amp: unequal gains for +Vsignal and -Vsignal?
« Reply #3 on: June 12, 2013, 07:50:51 pm »
Could this be related to two most recent of Dave's "Fundamentals Fridays" (offset voltage, offset current for op-amps)? I don't recall the details but it sounds like you're describing a similar issue that he was talking about. If you haven't seen them, they might be worth watching.
my random ramblings mind-dump.net
 

Offline ignator

  • Regular Contributor
  • *
  • Posts: 206
  • Country: us
Re: Instrumentation-amp: unequal gains for +Vsignal and -Vsignal?
« Reply #4 on: June 12, 2013, 07:57:19 pm »
What about seebeck voltage?  Your down in the mud measuring microvolts, and this will affect your measurement.  Or is your connections not of dis-similar materials so you can rule this out.
 

Offline Rufus

  • Super Contributor
  • ***
  • Posts: 2095
Re: Instrumentation-amp: unequal gains for +Vsignal and -Vsignal?
« Reply #5 on: June 12, 2013, 09:21:04 pm »
Seebeck? Ground Loops? Solid connections?

As far as I can tell he is only simulating in Multisim. Possibly the amp model includes worst case error terms. The datasheet worst case gain error for G > 1 is 0.3% for the A grade part, his observed 0.3% error may not be coincidence.

 

Offline awallinTopic starter

  • Frequent Contributor
  • **
  • Posts: 694
Re: Instrumentation-amp: unequal gains for +Vsignal and -Vsignal?
« Reply #6 on: June 13, 2013, 05:16:00 am »

Thanks for your answers!
It may indeed be a feature of the instrumentation amplifier - not a bug.
Also, I will check out the links about input-offset-voltages and look into that.

FWIW in the simulator an instrumentation amp built from three discrete op-amps using AD8554 shows a much smaller gain error.

AW
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf