Unconventional Op-Amp Clipping

I have been scratching my head for a few weeks about the behavior of my circuit (attached). In this circuit I give a arbitrary known input signal to drive a current through R1 which is a thermistor. I then use an instrumentation amplifier to measure across points A and B to calculate the thermistor resistance. Unfortunately when I give the circuit a positive voltage it will clip, but for some reason the negative voltage does not clip. I have looked at the datasheet and confirmed that the opamp can produce the necessary current ($\pm$10mA) and I have given sufficient supply voltage ($\pm$12V). I have observed that the clipping does go away with low input voltages i.e. ($\pm$1V). I believe I am missing something trivial but I just can't seem to find it. I have also observed similar behavior with the other ADA4077 opamps in my circuit. I have removed the instrumentation amplifier and isolated the circuit completely and measured the waveforms using an oscilloscope observing the same behavior. I would really appreciate the help guys. Thanks in advance.

The circuit of interest.

Left: The waveform of the input (Yellow) and the voltage measured between A and B (Red). Right: The waveforms of the input and the measured wave at A to ground(Green). The measured waveform at A to my understanding must be really close to zero so this was most intriguing.

ADA4077 DATASHEET

• If you replace the "thermistor" with a resistor, is normality restored? (just in case that isn't a thermistor). Also, link to the ADA4077 datasheet in the Q. – Brian Drummond Aug 17 '17 at 14:01
• "and the voltage measured between A and B " this sounds a bit confusing, where did you put what clip? the GND clips are internally connected and connected to earth usually, thus if you put GND to A and the probe to B then you measured your signal referring to A not to GND. – PlasmaHH Aug 17 '17 at 14:03
• In the datasheet the common mode voltage is not clearly given but where it was mentioned it was not symmetrical e.g. ranging from -3.8 to 3 V. Maybe you are experiencing some problem with that? Especially if you turn down the voltage and the problem goes away. – Arsenal Aug 17 '17 at 14:09
• Note that A should always be at 0V, and any bias from there will be very small. You don't need a differential amp. – Scott Seidman Aug 17 '17 at 14:22
• @Arsenal -- common mode voltage should not be an issue, as the input terminals should both be at 0V. – Scott Seidman Aug 17 '17 at 14:25

2 Answers

The ADA4077 does not let you get very close to the rails -- less than +4 V with a 5V supply. If your circuit requires an output of more than about 9.5V, it will not work. Change R2 to 20K or more.

My best guess would be that your -12V rail isn't holding up, once it hits its own current limit, B can't go any lower so A has to increase.