Instrumentation amplifier and common mode question

Question

I have some questions about the instrumentation amplifier (IA) and common mode feedback (CMFB). The supply voltage for all op-amps is ±5V and I am using TL074CN

Do the input resistors of the CMFB need to match the IA resistors, or does their relationship matter? Is it normal for the CMFB to output a non-zero DC level, or should it always be at 0V? For example, with V_in1 and V_in2 as sine waves (20mV amplitude, 50Hz, and 20mV DC offset), the CMFB output has an amplitude of 4-6.5mV and a DC level of 370mV. This remains at 370mV even when I set the input offset to 0mV.

Answer 1

The supply voltage for all op-amps is ±5V and I am using TL074CN

OK, but you do know that op-amps inherently have a small DC offset voltage equivalent to putting a circa 1 mV battery in series with the non-inverting inputs? On a good day these offsets may be subtractive and, on a bad day these offsets may be additive.

On an indifferent day early in the morning things may look good but, over the course of the day things can get worse.

So, irrespective of what you set the true input voltage to be, there will be an offset unless you exactly counteract that offset with the correct external voltage with the right polarity. And, the output of your strangely named CFMB stage magnifies that offset from both input amplifiers by it's circuit gain.

The above from the TLO74CN data sheet page 17

Answer 2

Basic idea

Is it normal for the CMFB to output a non-zero DC level, or should it always be at 0V?

The idea behind a common-mode feedback (CMFB) circuit (elements in red in your schematic) is to generate a voltage only when both input voltages change simultaneously and in the same direction (the so-called "common mode"). When they change simultaneously but in opposite directions, the voltage should not change.

So, this voltage acts in opposition to the input common mode voltage, thereby realizing negative feedback only in common mode. This suppresses the common-mode voltage variations at the outputs of the input op-amps.

Implementation

Do the input resistors of the CMFB need to match the IA resistors, or does their relationship matter?

In your circuit of an instrumentation amplifier, this idea is implemented by an op-amp inverting summer with equal weighted inputs (equal resistors R4). Obviously, there is no dependency between the resistance of these resistors and the resistance of the instrumentation amplifier resistors. The only requirement is that they are equal (R3 = R4 in the circuit below).

^{simulate this circuit – Schematic created using CircuitLab}

Operation

Initial state

Vdiff = 0 V, Vcm = 0 V

^{simulate this circuit}

Common mode

Imagine that Vin1 and Vin2 simultaneously increase. Their amplified replicas Vout1 and Vout2 increase with the same rate of change (the input op-amps act as followers in this mode). As a result, the common-mode feedback voltage Vcmfb (at the output of the op-amp in red) also increases. This arrangement ultimately results in the attenuation of common-mode input signals, although the specific mechanism is not explicitly depicted in the provided schematic.

Vdiff = 0 V, Vcm = 1 V

^{simulate this circuit}

Differential mode

If the input voltages change simultaneously in opposite directions, theit sum (the voltage Vcmfb at the output of the inverting summer does not change). As a result, there is no negative feedback, and the input voltage changes are not suppressed.

Vdiff = 100 mV, Vcm = 1 V

^{simulate this circuit}

Vdiff = -100 mV, Vcm = 1 V

^{simulate this circuit}

About the Vcmfb voltage

What is Vcmfb?

Is it normal for the CMFB to output a non-zero DC level, or should it always be at 0V?

OA1 and OA2 in common mode act as voltage followers - Vout1 = Vin1 and Vout2 = Vin2, and at the output of the inverting summer OA3, their (amplified) sum appears. So, "it is normal for the CMFB to output a non-zero DC level" (Vcmfb is not zero).

How is Vcmfb used?

To answer this question, you need to go to a lower level and look into the internal structure of the op-amp. I recommend you visit my question How do we explain the common-mode feedback in an intuitive way? and my answer to it where I have considered in detail the CMOS implementation of this clever trick. See also my answer to the question Need for common-mode feedback.