High Common mode op amp shunt compensation

Question

The High Common mode Voltage difference op amp AD629 Datasheet on page 12 says that the use of a large sense resistor skews the output. Why is this so? I can't wrap my mind around any other reason than that one side or the other would typically tie to either the source or sink of the power supply.

Is there any advantage of using one of these IC's versus using an instrument amp with input resistor dividers? Would the instrument amp with dividers suffer from the same error?

I am researching an isoamp replacement for an input stage for a galvanic oxygen sensor, which in this case is a small current source converted to voltage by a 2.2k resistor(0-200mv dc). This is a floating voltage other than water contact through the membrane, and I have doubts about whether this would require a 2.2k rComp as explained in the datasheet Figure 36.

Answer 1

These parts are designed for situations where you have a high common mode voltage, such as the example 200VDC.

The AD629 is a a differential amplifier (not an instrumentation amplifier) so the input impedance is relatively low. In the case of a 2K shunt, the common mode rejection is compromised by an imbalance of 2K/380K or about 0.5% so a 200V common mode voltage leads to a 1V error in the output.

There is a current of about 200V/380K flowing from both sides of the shunt resistor (about 520uA) into the amplifier inputs, so your supply is providing an extra mA or so just to feed the amplifier inputs.

An instrumentation amplifier may be better, however you have to keep the inputs within the common mode voltage range of the instrumentation amplifier, which is typically less than (sometimes much less than) the supply rails.

If you are measuring a small current with a small common mode voltage then the AD629 is probably inappropriate. It would be better, if possible, to measure the current on the low side, which requires only an op-amp and does not involve trying to reject a high common mode voltage, with the resulting extreme sensitivity to resistor ratios, which leads to drift and error.

If you provide your own resistor dividers on the inputs to an instrumentation amplifier then you have to consider the stability of the ratios and the increased noise of the (already usually noisy) instrumentation amplifier. And you still have the effect of the shunt resistor imbalancing the ratios.