I am building a Wheatstone bridge for strain gauge measurement. Due to component deviations a wheatstone bridge will often show an offset which needs to be trimmed:

enter image description here

If we assume that we're using 5% resistors, the worst case would be that R1=950ohm and R2=1050ohm, making the reference equal to 1.575V (when Vcc=3V), which will through the measurement way off - expecially when amplified 1000 times. So my question is, what do you think is the best way to trim such an offset digitally?

I want to be able to trim and calibrate the bridge with a microcontroller.


Mostly, when it comes to full bridge strain gauge circuits we use an instrumentation amplifier with the reference input connected to a DAC output. This is pretty common in industry and allows nulling of offsets via software at any point. Pretty useful for pressure gauge's too.

  • \$\begingroup\$ That sounds like a good idea - but what about the DAC noise? Wouldn't that become a problem if we don't want to use very expensive components? \$\endgroup\$ – Jolle Apr 10 '15 at 8:39
  • \$\begingroup\$ DAC noise is usually high frequency and so a 100 ohm resistor and a 10uF capacitor filtering the DAC output gets rid of noise substantially. Also if the DAC is remaining fixed for long periods at one output level there will be no clock feedthru and no noise. \$\endgroup\$ – Andy aka Apr 10 '15 at 9:28
  • \$\begingroup\$ OK, makes sense! I like this solution a lot. It can be done relatively cheap (especially if the MCU has a DAC internally), and I will always be able to calibrate the offset without diminishing my ADC range. \$\endgroup\$ – Jolle Apr 10 '15 at 10:53

First of all you should really get better resistors. 5% with a strain gauge is just ridiculous.

To "trim and calibrate" your measurement you should just leave your gauges be, i.e. leaving them in a known state, measure the output and save it in a variable in your micro. This value should then be subtracted from all subsequent measurements. If you can you should measure the offset periodically, but if you want to fight just that 5% and nothing else a once in a lifetime measurement should do.

Please note that measuring Vr1r2 with respect to ground might appear appealing since you don't need to know the gauges state, but hey you are using a Wheatsone bridge to contain things like temperature drift and whatever, measuring your offset with respect to ground just defeats the bridge purpose.

Really, get better resistors.

  • \$\begingroup\$ +1 Better resistors are the first defense- 75mV offset multiplied by a gain of 1000 will "probably" saturate his amplifier. \$\endgroup\$ – Spehro Pefhany Apr 9 '15 at 19:28
  • \$\begingroup\$ You are right - the 5% was just to make a point. I intend to use 1% but it might still be a problem. Worst case then is 1.515V - and with the amplification I'll get an offset of 1,5V, and there went half of my range :) .. I'm considering a solution with either a digital potmeter or a DAC maybe. \$\endgroup\$ – Jolle Apr 9 '15 at 19:39
  • \$\begingroup\$ @Jolle any reason to not use a manual one-time adjustment? This only needs to (and only should be) done once. Every time after that, you should use digitally measured offsets. This allows you to correlate drift. \$\endgroup\$ – helloworld922 Apr 9 '15 at 19:45
  • \$\begingroup\$ @helloworld922 No, nothing wrong with that, but how would you do this adjustment? I want as close to full-scale as possible after this on-time calibration. Should I replace the resistors to it is right? or install a manual potmeter? or? \$\endgroup\$ – Jolle Apr 9 '15 at 19:54

Another option is to simply not worry about it. Put your bridge signals into a differential buffer and low pass filter, and straight into a 16+ bit differential ADC. At that point, you can simply subtract off the offset, and there is nothing to saturate. If it becomes necessary, you can turn the buffer into an amplifier, and you won't need nearly as much gain as a single ended configuration.


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.