I am using an AD8226 instrumentation amplifier as the pre-amp in a circuit which uses a micro fuel cell oxygen sensor to measure oxygen content in a gas stream. The pre-amp is configured in single ended supply mode, with a 5 V supply. Gain resistor was chosen to give a gain of ~80.

enter image description here

The output of the in-amp is then passed through an active low pass filter and to a microcontroller's ADC, but I don't believe this is relevant to the problem. Everything discussed here was done with the preamp outut (O2_PRE node in the diagram) floating and only connected to a multimeter.

When testing this circuit with the sensor I found that the gain was nice and linear - up until a certain point where the gain dropped off (this point was at about 20mV input/1.6V output).

To eliminate any issues with the sensor I replaced the sensor with a voltage divider made up of a fixed reisistor and a multi-turn linear pot:

enter image description here

With the resistor network I observed the same problem (see further down for a plot of the inamp output).

To further eliminate any issues with the surrounding circuitry I wired directly up to the AD8226, using my bechtop supply as VCC. Giving the circuit seen here:

enter image description here

This showed the same behaviour as seen in this plot:

enter image description here

('mainboard' refers to the output when the voltage divider is used to drive the original circuit, 'breadboard' gain is slightly lower as I used a 630R gain resistor)

The AD8226 data sheet specifies an output voltage swing of 0.1 V to +VS−0.1 V in single ended mode. I am measuring the output into a high quality multimeter (ie. high impedance), although I did observe the same results when I added a load resistor of 20 K. I have repeated the results with multiple AD8226 devices.

Can someone explain why the gain would be falling off above a differential input of ~20 mV given that I seem to be staying well within the output limits?

  • \$\begingroup\$ What is the DC output level? Also when you say 20mv in, is that 20mv DC offset, or +/-10mv square wave, or 20mv rms AC? \$\endgroup\$
    – user16324
    Commented Dec 28, 2012 at 12:06
  • \$\begingroup\$ The point where gain drops off is at 20mV DC differential between the inputs of the inamp (noting that the negative input is at ground in my voltage divider tests). This equates to 1.6V DC on the output. \$\endgroup\$
    – Matt B
    Commented Jan 2, 2013 at 2:41

1 Answer 1


Most likely a common-mode voltage violation. See Figure 15 of the datasheet. Unfortunately it does not give you the limits for a single power supply, but shows you that common-mode and output voltage limits go hand in hand.

Actions that may solve your problem:

  • Use dual power supply.
  • Lift inputs above ground.
  • Use a lower gain setting.


As pointed out by Dave Tweed, the datasheet does have information on common-mode and output voltage ranges for single supply. See Figure 13 on page 10 of the datasheet (your case is the blue triangular area). Your common-mode input is practically ground, so as soon as your output exceeds about 0.7 V, the limit is exceeded, and your device will not behave as specified. This will happen even if you change Vref to 2.5 V: the output limit will be 3.2 V for a common-mode input voltage of ground to become invalid.

enter image description here

  • 1
    \$\begingroup\$ Figure 13 shows the limits for single supply, G=100, Vref=0. When the output voltage rises above about 0.7V, the input common-mode range no longer includes ground. \$\endgroup\$
    – Dave Tweed
    Commented Dec 28, 2012 at 13:47
  • 1
    \$\begingroup\$ Yup-- split your 50K on the high side of the sensor into 2 25K, one on either side. \$\endgroup\$ Commented Dec 28, 2012 at 15:33
  • \$\begingroup\$ page 20 of the datasheet explains in more detail how you can clip the amplifiers inside without exceeding input or output ranges. +1 \$\endgroup\$ Commented Dec 28, 2012 at 18:39
  • 1
    \$\begingroup\$ 100% correct. I overlooked that when selecting the part. Fortunately the AD8227 has similar performance but a much wider output range at 0V common mode. Scott's suggestion also works great but is more difficult to achieve in the final design (without adding extra parts to bias the sensor up, which would mean re-spinning the PCB). \$\endgroup\$
    – Matt B
    Commented Jan 4, 2013 at 7:50

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.