I have a two-wire Pt100 element and a 5V ADC and want to measure the temperature. I have made a wheatstone bridge like below and connected it to an instrumentation amp. No matter the resistance of the Pt100 the output is always ~0.3V. Knowing little about electronics, I suspect this is because the negative terminal is clipped as the op amp is trying to bring it below 0V.

One solution would be to use -5V instead of 0V as Vs-, however I do not have -5V available.

How do you use an instrumentation op amp in such a scenario?


simulate this circuit – Schematic created using CircuitLab

  • 1
    \$\begingroup\$ What is the voltage measured at the - and + terminals with a DMM? What is the gain of the instrumentation amp set to? If PT100 > 100Ohm, the difference at the opamp will be in the range of a few mV, this is too close to the rail for the output of the instramp unless you are setting sufficient gain. With the bridge configured as it is, you can bias your opamp input up to 2.5V and measure the signal with 2.5V as the zero level. \$\endgroup\$ – crasic Aug 10 '15 at 23:48
  • 3
    \$\begingroup\$ I've heard of instrumentation amps and op-amps, but not "instrumentation op-amps". Can you be more clear exactly what part you're asking about? AD821AN doesn't seem to be a real part number. \$\endgroup\$ – The Photon Aug 11 '15 at 0:58
  • 1
    \$\begingroup\$ You're running your bridge way out of balance. You really want the two inputs to be much closer to the same voltage, and operate the inamp at a much higher gain in order to see the changes. \$\endgroup\$ – Dave Tweed Aug 11 '15 at 1:11
  • \$\begingroup\$ @ThePhoton I meant "instrumentation amp". Thank you for pointing this out. \$\endgroup\$ – Friend of Kim Aug 11 '15 at 10:23
  • \$\begingroup\$ @DaveTweed What do you mean it is out of balance? When the Pt100 is at 100 Ohms the difference is 0V as I see it. \$\endgroup\$ – Friend of Kim Aug 11 '15 at 10:24

You need to respect the amplifier input common mode range (and output range). If your amplifier (I'll assume it's an AD8221 inamp) had large power supply voltages your bridge configuration would be fine. You're putting about 0.7mA through the RTD element, which is reasonable for a Pt100. Your full-scale span will be 100mV, so you need sufficient gain in the inamp to give you full scale span for the ADC. If it's 2.5V, then G = 250.

You might want to use a bit lower than 100 ohms if you want to cover 0 degrees C for sure.

The problem is that the valid range of the AD8221 looks like this with +/-5V supplies:

enter image description here

There may be other instrumentation amp types that can work acceptably with a single 5V supply (you would probably have to decrease the 100R resistor a bit to offset the minimum output above zero). Also note that it can't get to 5V output.

Another option is to create a -5V supply using an ICL7660 or similar part.

  • \$\begingroup\$ This is the amp I am using. So the only way to achieve the result I need is to create a -5V supply (or finding a different amp)? \$\endgroup\$ – Friend of Kim Aug 11 '15 at 9:49
  • \$\begingroup\$ You could change bridge to bring the common mode voltage up near 2.5V, but the RTD would no longer be grounded and the output is only guaranteed to swing from 1.1V to 3.9V under those conditions. It would make the linearity a bit worse too. \$\endgroup\$ – Spehro Pefhany Aug 11 '15 at 10:44
  • \$\begingroup\$ How would you do this? \$\endgroup\$ – Friend of Kim Aug 11 '15 at 23:04
  • \$\begingroup\$ Reduce the 6.8K resistors to 3.4K and put a 1.7K resistor between ground and the bridge (roughly). \$\endgroup\$ – Spehro Pefhany Aug 11 '15 at 23:20
  • 1
    \$\begingroup\$ The output swing is limited by the supplies, and by saturation of internal nodes with this instrumentation amplifier. The vertical lines on the diagram I posted are where the output voltage is limited by the supply, but the others are related to the input voltage and supplies. The horizontal lines are where the input common mode voltage limits are exceeded. \$\endgroup\$ – Spehro Pefhany Aug 12 '15 at 14:14

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.