• State Resolved
  • Locked Locked
  • Replies 10 replies
  • Subscribers 50 subscribers
  • Views 275 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.

OPA4188: OP-AMP Gain error at specific input voltage range

Douglas Grover
Prodigy 20 points
Part Number: OPA4188
Other Parts Discussed in Thread: OPA2188, OPA2197, OPA2192, OPA4196

I'm using the OPA4188 to measure voltage drop across a .006 Ohm shunt resistor.   My test setup is a programmable supply that I can vary from 0.6V to 5V which connects to 4 separate .006 Ohm shunt resistors all connected in parallel,  the other side of the shunt resistors is connected to an electronic load that I vary from 1 to 17Amps.   The circuit is using this QUAD op-amp to measure voltage across each shunt resistor with a nominal gain of my circuit setup to approx 25.   Schematic just shows one of the 4 channels in the package but all 4 are configured the same way.   V+ is biased with 15V and V- is biased with -3V.

My first test was to set a fixed current of 7Amps so expect roughly 1.75A of current through each shunt resistor.   I then vary my shunt voltage source from 1 to 5V in 500mV increments.    Measuring my actual Vdrops across the shunt resistor and getting the ratio of the op-amps output voltage to confirm I'm getting my gain of approx 25.    This works well from 1 to 2V and then from 3.5V to 5V,   but for some reason at my 2.5 and 3V points my gain increases to as high as 37.5.    This is very consistent across 5 different prototypes I have.

Vdrop on .006 Ohm shunt op-amp output Calc Gain
Shunt Source Voltage CH-A CH-B CH-C CH-D CH-A CH-B CH-C CH-D CH-A CH-B CH-C CH-D
1 0.0115 0.0106 0.0106 0.0096 0.303 0.284 0.284 0.249 26.34783 26.79245 26.79245 25.9375
1.5 0.0115 0.0106 0.0106 0.0096 0.2956 0.2653 0.263 0.2422 25.70435 25.0283 24.81132 25.22917
2 0.0115 0.0106 0.0106 0.00965 0.2979 0.2672 0.2643 0.2426 25.90435 25.20755 24.93396 25.1399
2.5 0.01155 0.0106 0.0106 0.0096 0.338 0.398 0.355 0.275 29.26407 37.54717 33.49057 28.64583
3 0.01155 0.0106 0.0105 0.0097 0.311 0.349 0.337 0.245 26.92641 32.92453 32.09524 25.25773
3.5 0.01155 0.0106 0.0105 0.0096 0.3033 0.2654 0.2613 0.2425 26.25974 25.03774 24.88571 25.26042
4 0.01155 0.0106 0.0105 0.0096 0.3055 0.267 0.263 0.2428 26.45022 25.18868 25.04762 25.29167
4.5 0.01155 0.0106 0.0105 0.0096 0.3074 0.2671 0.263 0.2429 26.61472 25.19811 25.04762 25.30208
5 0.01155 0.0106 0.0105 0.0097 0.3093 0.267 0.2629 0.243 26.77922 25.18868 25.0381 25.05155

My 2nd test was to hold the voltage steady at 2.5V since I knew I had issues here and then sweep through my a current range of 1 to 17A total or approx .25A to 4.25A per shunt resistor.    In this setup I see on CH-B and CH-C op-amps that I get a distinct shift in calculated gain as soon as the current goes above approx 1.5Amps for their respective shunt resistors.   Before the bump and after that bump the slope of shunt current vs op-amp output stays the same but just get's an offset as I approach 1.5Amps.

Sorry for the length of this but really wondering if anyone can see why operation becomes abnormal at 2.5 - 3V range and then within that voltage range why I would see a shift in the plot once the drop in my shunt resistor get's over about 10mV.   I should mention that I've looked at the shunt resistors voltage source and did not see any significant change in noise or ripple voltage across the voltage and current loading ranges that I tests.  Thanks in advance

  • 0
    Guru 140200 points

    Hi Douglas,

    hhm, why are the voltage drops across the 0.006R shunts different, if they are in parallel? Shouldn't you see identical voltage drops?

    Kai

  • 0 in reply to kai klaas69
    Prodigy 20 points

    Hi Kai,   After the parallel shunt resistors they are not immediately tied back together but go through another board that has varying trace/plane lengths before being shorted to each other.   I'm not expected a perfect match in the current splitting.   Thanks, Doug

  • 0 in reply to Douglas Grover
    Guru 140200 points

    Hi Douglas,

    not having the shunts close to the OPAmps can cause stability issues. Are you sure that the OPAmps are not oscillating? Have you probed the outputs of OPAmps with a scope? Don't directly touch the outputs with the scope probe or DVM, though, but insert an isolation resistor first.

    Have you checked the circuit with only one OPAmp? Or with a totally different quad-OPAmp (no OPA4188)?

    Could the electronic load have an impact on the measurement? Have tried to programm the current by a simple resistor instead of an electronic load?

    Have you tried to mount small caps in parallel to "R1032" and "R1034"?

    Kai

  • 0 in reply to kai klaas69
    Prodigy 20 points

    Hi Kai,

    I should clarify,  the shunt resistors are very close to the op-amp, it's the output of the shunt that has longer distance before they are all shorted together actually on a different board.  This connection point where they short is not as balanced so I expect I can the current path to vary some.    I've checked as you mention with a scope and output looks very stable.  I have ordered a couple new op-amps to try out and hopefully will resolve, would just love to know root cause as well though.   I've not tried a purely resistive load but I have tried various small caps in parallel with R1032 (did not try R1034),  these tests only seemed to make my error greater.

    Thanks,

    Doug

  • 0 in reply to Douglas Grover
    TI__Guru* 93105 points

    Hi Douglas,

    It is possible that this may the result of the OPA2188 bias currents changing with the common-mode voltage change (datasheet Fig. 7). The OPA2188 is a chopper op amp and the input bias current does exhibit periodic switching transients. They can actually become transformed to a voltage offset when the bias currents flow through the high value circuit resistors.

    We often recommend that the resistors applied in a chopper amplifier not exceed 10-kilohms, and prefer even lower values be used. If you could try rescaling the R1031 through R1034 resistors to 1/10th the value and the retest, you may find that the non-linearity being observed improves.

    Another possible way to see if it is the chopper/ bias current effects is to replace the OPA2188 with a non-chopper such as the OPA2192/OPA2197 and see if the circuit performance becomes as expected.

    Regards, Thomas

    Precision Amplifiers Applications Engineering

  • 0 in reply to Thomas Kuehl
    Prodigy 20 points

    Hi Thomas,  Thanks for the reply.   I did neglect to mention that I tried changing my values at one point with a 1K / 25K combo but still had the same issue.   I did just order OPA4196 as it was in stock and I don't believe it has the chopper circuit.    Hopefully that will clear things up. 

  • +1 in reply to Douglas Grover
    TI__Guru* 93105 points

    Hi Douglas,

    Twenty-five kilohms would probably still be too high. Indeed, when you get the OPA4196 in place and tested we would really like to see what results you obtain.

    Best Regards, Thomas

    Precision Amplifiers Applications Engineering

  • 0 in reply to Thomas Kuehl
    Prodigy 20 points

    I may have misunderstood, you want all of the resistors to be below the 10K.     I'll also give that a try as well as the new op-amp.   Thanks again for the support.  I'll make sure to report my results

  • 0 in reply to Thomas Kuehl
    Prodigy 20 points

    I did try using r value of 365 Ohm and 9.09K Ohm but still saw the same issue as I had with the higher value resistors.   I then tried the OPA4196 both with my lower value resistors and the original value resistors and this did resolve the issue I was seeing.   In addition to the OPA4196 I also tried another op-amp that did not have the chopper input and it also worked for me.   Thanks for all the support!

  • 0 in reply to Douglas Grover
    Guru 140200 points

    Hi Douglas,

    Douglas Grover said:
    I should clarify,  the shunt resistors are very close to the op-amp, it's the output of the shunt that has longer distance before they are all shorted together actually on a different board.

    When I said the shunt must sit close to the OPAmp, I actually meant that every single signal path a portion of chopped input bias current of OPA4188 can flow through must sit close to the OPAmp. This means all feedback components including the terminals "VIN" and "VOUT". If any cabling is involved leaving the terminals "VIN" and/or "VOUT" a chopper OPAmp is not well suited.

    You can try to provide a path for the chopped input bias current to signal ground by adding filter caps from each terminal of shunt to signal ground like "C864",. But this will only work (if at all) when they sit close to the OPA4188. 

    Keep in mind that a chopper OPAmp can only work well when the circuit is very compact. If any cabling at the inputs is involved a chopper OPAmp is no good choice.

    Kai