I'm using an instrumentation amplifier (INA118) to measure the signal from a full bridge strain gauge (tension only, all 4 resistive elements are strain gauges, slow changing (DC) signal). I'm using a single +3.3 V supply (V+ = 3.3V, V- = 0V). The common mode voltage has been set to 2.35 V based on the data sheet specs, and the gain is set to 1000. I have measured the load vs output and it is linear as expected.

I'm having issues figuring out the best way to compensate for the voltage when there is no load. The offset voltage is 0.9 mV, which when amplified is 0.9 V. This takes up a large part of the useable ADC range.

One of the methods I've found so far revolves around putting a shunt resistor in parallel with one of the legs of the strain guage, or trimming the output of the bridge with a potentiometer. The perceived issue with this is that the instrument needs to operate between -20 Celsius to +40 Celsius, and the resistor compensation will not be adequate for that entire temperature range.

Another of the methods I've found is setting Vref on the INA118 to compensate for the offset voltage. This seems like it would be fine if the offset voltage was negative, but since the voltage is positive and I'm using a single supply, it seems like I won't be able to compensate this way. Is this correct?

The last thing I've thought about trying is lowering the gain and getting a higher resolution ADC, and then doing the offset compensation in software. It would be more desirable to use the ADC that is on the microcontroller, but it seems that this option has the most temperature stability.

My question(s): Which of the options above would be the best given the temperature range constraints? Am I missing any other options? Is my understanding in each of the scenarios correct? Any input is much appreciated.

  • \$\begingroup\$ is this a 2 leg strain gauge which needs an external mid scale reference, or a 4 leg gauge which is a whole Wheatstone bridge? Sounds like the latter but it's not clear. Perhaps post a schematic. \$\endgroup\$ – Neil_UK Nov 23 '17 at 9:04
  • \$\begingroup\$ 2 or 4 lead makes a difference about where you can put trimming resistors. Beware using resistors of different tempco to trim, a strain gauge may well have a designed tempco different from the nominally zero of general purpose reisstors. \$\endgroup\$ – Neil_UK Nov 23 '17 at 10:05
  • \$\begingroup\$ Hmm, could this problem be solved using an indirect current-feedback instrumentation amplifier? Just guessing. \$\endgroup\$ – Marko Buršič Nov 23 '17 at 12:14
  • \$\begingroup\$ @Neil_UK this is a 4 leg strain gauge that forms the whole wheatstone bridge. \$\endgroup\$ – Bales Nov 24 '17 at 0:56
  • \$\begingroup\$ Does you application required accurate relative or absolute readings? Normally one uses an autocal on startup. Have you recorded any unload hysteresis? Where is the T vs Offset data? \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Nov 26 '17 at 0:18

I would try to use indirect current-feedback instrumentation amplifier. Like MAX4208/4209 or AD2420 or any other....instead of using the classic three opamp instrumentation amplifier. These amplifiers are made to cope with the issue you have.


The traditional 3-op-amp in-amp architecture, shown in Figure 2, has a differential gain stage followed by a subtractor that removes the common-mode voltage. The gain is applied on the first stage, so the offset is amplified by the same factor as the signal of interest. Thus, the only way to remove it is to apply the opposite voltage to the reference (REF) terminal. The main limitation of this method is that adjusting the voltage on REF cannot correct the offset if the first stage of the amplifier is already saturated

enter image description here

Indirect current feedback (ICF) in-amps, such as the AD8237 and AD8420, make it possible to remove the offset before it is amplified. Figure 3 shows a schematic of the ICF topology

enter image description here enter image description here

enter image description here enter image description here

  • \$\begingroup\$ Thank you for the suggestion. The link was a big help - I was still struggling after reading the datasheets for the AD8237 to see how to apply a negative compensation. The image in the article here helped me figure it out. The use of the DAC to set the offset compensation seems most amenable to a "TARE" functionality - is there anything else to consider when going that route? \$\endgroup\$ – Bales Nov 26 '17 at 2:49
  • \$\begingroup\$ @Bales Perhaps checking how to drive the REF input. Some of them require to be driven by low impedance output (opamp), which is the case for most 3 opamp instrumentation amplifier. \$\endgroup\$ – Marko Buršič Nov 26 '17 at 7:58
  • \$\begingroup\$ Looking through the datasheet for the AD8237, it says that unlike other instrumentation amplifiers technologies Vref doesn't have to be low impedance. I've tested this and it is working well. Thanks again! \$\endgroup\$ – Bales Mar 1 '18 at 6:41

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.