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.

Trouble with ina122 measuring voltage over an 100A current shunt

Other Parts Discussed in Thread: INA122

Hi, i have trouble measuring low voltages (0 to 100mV) with INA122.

As input, I have several shunts (some are copper bus bars) with different values and 100A full rectified (pulsating 100Hz, not filtered) passing through them.

With the INA122 I measure the voltage drop across them, and then get their value in micro Ohms.

So, I am measuring 0-100mV full rectified (not filtered) with INA122, after the IA I have 2 Sallen Key filter i series to get the "mean value" then to the microcontroller ADC.


The input is a shielded twisted pair (I have 100A (100Hz) and their noise in the same box...), I have 2 5k6 1% resistors from both inputs to ground, the INA122 is powered single supply 24V, and then a voltage divider to the sallen key filters. Also I have the reference pin (5) with a trimmer to adjust voltage reference.

Gain resistor are 2k2+220 1% (gain approx 87).


I cant get values to repeat, or to be consistent. I recently read about (V-)+0.15, so I´ll change gain resistors.

Any help would be great

  • Which are the voltages at pin 2 & 3 referred to GND (pin 4) of INA122?

    Are you shure that one of the two (pin 2 pr 3) is not negative referred to GND (out of the Common-Mode Voltage Range)?
  • Pin 2: full rectified not filtered peak 460mV mean value 200mV, negative spike -50mV
    Pin 3:full rectified not filtered peak aprox 480mV mean value 214mV, negative spike -50mV

    on behalf of the second question I think they are not, as the 2 voltages are positive (save for the spikes)

    what is strange (for me, at least) is that when I measure resistance of 100uOhm (expected input mean value 10mV) I get for example 2V output, and when I measure 10uOhm (expected input mean value 1mV) I can get 1.8V or 2.2V. I am using Vref 12V, (and in some tests 5 or 6V)
  • What if you inject a direct current instead of a pulsating one?

    In any case I suspect that the negative spike -50 mV (out of the Input Common Mode Range) "disturbs" the INA122

    It takes a little time to draw a schematic to explain my idea so I'll post tomorrow

    See you later

  • Thanks for your reply, the issue is that I need to use pulsating current.
    Because its a micro ohm meter, I rectify 100A (low voltage, 1.4V, two diodes transformer with middle point) and works as a "current source" for resistances up to 1000uOhm. I cant inject direct current beause cap forfiltering that, would be huge.
    When I have the current flowing though the resistance (the 100A) I measure voltage, with AI, and Sallen Keys take the mean value, hence I have the resistance in uOhms.
  • IMHO, what I would point out is that the input circuit is working beyond  of the Input Common-Mode Voltage Range.

    Please look at the following schematic

    Let's assume

    Shunt: 100 uOhm
    Shunt DC current: 100 A
    Voltage across the shunt: 10 mV

    I we measure the voltages (referred to GND - heavy green line - pin 4 of INA122) at the 2 inputs of INA122 we get:

    pin 3: +5 mV
    pin 2: -5 mV

    Voltage at pin 2 (negative) is out of Input Common-Mode Voltage Range as spec'd in the data sheet of the IC

    If you have a "100A full rectified (pulsating 100Hz)" current in the 100 uOhm shunt which are the peak voltages at pins 3 & 2?

    Assuming roughly 150 Apeak current we get:

    pin 3: +7.5 mVpeak
    pin 2: -7.5 mVpeak (out of IC specs)

    I propose 2 solutions (one of the two dotted ligth blue line)

    1) Short circuit R17: this way the Input (Common-Mode) Voltage at pin 2 goes 0 (zero).

    This way the imbalance of the circuit is minimal as pin2 & 3 are "short circuited" by the shunt (100 uOhm)

    OR

    2) (Better) "Rise" R16 & R17 from GND (pin 4 of INA122)  inserting a "few volts" between GND and the R16 & R17 "common point"

  • Thanks, I´ll try the second (I have some 3.3V in the PCB near).

    I dont think the first option would work in my circuit because I have next of the shunt, a Hall efect IC to measure current, and then the ground of the "power circuit" is attached to the ground at PCB.

  • Hello,

    This is very likely an input common-mode range issue. Try lifting pin #4 of the INA122 from the pad and applying a negative voltage (e.g. -5V) to the pin. Make sure it does not touch the pad otherwise the -5V source will short to ground. Also remove R22 and monitor the input pins and output pin using an o-scope. Please post the waveforms of the input pins and output pin.
  • Hello, I supplied the INA122 with a computer power supply with +12V and -12V, pin5 with 0V ref (GND), and removed R22.

    These are the oscilloscope waveforms for current across a 100uOhm shunt, later I´ll put over a 10uOhm copper bar.

    osciloscopio.xls

  • Here are the oscillographs for a copper bar of 10uOhm. 10uohm.xls

  • Hello Martin,

    Thanks for the additional information. For each spreadsheet, I added a curve to the Pin 6 tab. It's simply (Pin3-Pin2)*Gain where Gain=5+(200k/(2.2k+220))=87.645V/V.

    I expected the results to be very similar to what was measured at the output (Pin 6). You will notice that this is definitely not the case. There appears to be a significant amount of noise and a lot of error in general. There are a variety of possible factors for this, including noise from the power supplies and/or reference voltage or EMI. This assumes that the data you sent was taken literally at the pins of the INA122.

    In order to start eliminating variables, please disconnect the INA122 from the signal source. It is unclear how to do this from your schematic. Nonetheless, once disconnected, try injecting a known precision differential voltage (dc) and then measure the output. The output should correlate with the INA122 transfer function. In addition, you may want to investigate noise on your power supplies and reference (which I believe is currently connected to ground). Finally, make sure that your layout has power supply bypass caps (e.g. 0.1uF) placed as close to the power supply pins of the device as possible.

    martin.zip

  • Pete, thanks for your answer.

    The readings where indeed taken on the INA122 pins, but the wires that "bring" the voltages are 1meter long (although twisted). The gain is not correct because I changed the gain resistance and forgot to mention the new gain 190.87=5+200k/1076. If you change the value In the excel concerning the 100uOhm shunt, It gets somehow similar. On the 10uOhm copper bar no.

    Next to your advise of using the dual supply I "lifted" voltage reference up to 2V, and then with the 2 sallen key filters connected I began to get better results. Waveforms are similar to those on the spreadsheets I sent, but after the filters I get dc values which I can calibrate. Its not the best, but results are repeatable and "very" linear in the range of 0-115uOhm.

    I have ceramic 0.1uF near the INA122. I´ll put the o-scope on supply and reference voltages. But I think the problem is the interference generated by the 100A currents and transformer.

  • Marco I tried rising R16 and 17, but got better results with the dual supply. Nonetheless I want to thank you for your answers.
  • Hello Martin,

    thank you for your appreciation

    If I may make an addition I suggest to add a low pass filter in front of the INA122 in order to cut the noise, as in the following schematic

    2 x 1 KOhm and 0.1 uF

    Martin Rodriguez51 said:
    Marco I tried rising R16 and 17, but got better results with the dual supply. Nonetheless I want to thank you for your answers.

  • I´ll try it and let you know.

    Thanks!