Continuation of this question, apologies for the schematic. Rwire and Roh are depicted in balanced state.


This is a common Constant Temperature Anemometer circuit, however the diff amp output (Vbridge) keeps on saturating. I think the problem is on the INA118 output (Vina). A voltage divider analysis at the bridge would show that if Rwire < Roh / 10 then Vwire < Voh and hence Vina = G(Vwire - Voh) would have different polarity with Vbridge. Similarly, when Vwire > Voh then Vina would have the same polarity with Vbridge.

Measurement table

The table above shows a couple of measurement I did and it can be seen Vwire > Voh for all cases, but Vina have different polarity with Vbridge everytime. This differs with the analysis, I thought the diff amp may be messing with the inamp output but not sure why. Assuming that the components are not broken (I checked but maybe I'm wrong) what else could cause this?

  • \$\begingroup\$ Break the loop and get each part to operate with a linear gain offset as expected. Do not continue until fixed. U1 output is wrong. I’m not familiar if this the optimum cct for a HWA in Const Temp mode as there is no fixed Vref for calibration on INA -IN \$\endgroup\$ Feb 13, 2020 at 3:22
  • \$\begingroup\$ Why such a complex circuit when a single op-amp and BJT would do the job? \$\endgroup\$
    – Andy aka
    Feb 13, 2020 at 10:52
  • \$\begingroup\$ Something is very wrong between Voff and Vninv. That schematic cannot generate those voltages. Also, as Vninv > Vinv you can never turn Q1 on (Vout = +V) \$\endgroup\$
    – user16324
    Feb 13, 2020 at 10:53
  • \$\begingroup\$ @Andyaka Yes it can and that would be my other option. The reason I used this schematic for the time being is because I need the Voff to optimize the systems response. The other option I mentioned is using a single inamp and feeding Voff directly to the reference pin. The 2 sounds similar, but there's a slight difference in the gain and offset formula. \$\endgroup\$ Feb 14, 2020 at 9:04
  • \$\begingroup\$ @BrianDrummond now that you mentioned it, I think you're right. No way a voltage divider between -3.11V to GND can yield -4.55V.... \$\endgroup\$ Feb 14, 2020 at 9:08

1 Answer 1


You have to constrain the output of TIP31 to be higher than 0V by (at least) a few mV, otherwise this analog computer can find the wrong solution. BTDTGTTS.

  • \$\begingroup\$ Yet the INA has major offset unlike 40uV spec x100 \$\endgroup\$ Feb 13, 2020 at 3:14
  • \$\begingroup\$ It seems even after reducing Roh from 100 to 40 the Voh inverting input is always lower than Vwire the NON-INV input while the output Vina ought to always be +Ve yet is not.. . This suggests something non-linear perhaps noise on inputs \$\endgroup\$ Feb 13, 2020 at 3:21
  • \$\begingroup\$ @TonyStewartSunnyskyguyEE75 Yes this is bugging me as well. Would the diff amp mess around with the INA output? \$\endgroup\$ Feb 13, 2020 at 3:23
  • \$\begingroup\$ Could be oscillating, check that for sure. But the output when negative will heavily load the amplifier so it could just be not keeping up. \$\endgroup\$ Feb 13, 2020 at 3:27
  • 1
    \$\begingroup\$ I also probed those points with oscilloscope and it is no oscillating. \$\endgroup\$ Feb 13, 2020 at 3:31

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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