Opamps - to Buffer or not to Buffer, that is the question

I've attached a circuit of a circuit I am now supporting. I am pushing to have many things redesigned and one of the things I want to push is to remove U12A, the buffer.

Just to summarize, this is a diff amp with a low pass filter built in.

The reason is, that the non-idealities of the opamp will just be x2 now, and the final output just goes into an ADC so is charging up 12pF or so. The output impedance of U12B will be the same as U12A (if I change the resistor values of course). The most concerning non-ideality is the offset voltage, which can be up to +-10mV, so with two stacked up could be up to +-20mV now.

Given it is a diff amp, I realize an instrumentation grade may be better, but I am not focusing on that topic for now.

Is there something I am overlooking, or am I justified in my assertion that many Engineers tend to overdue it on the opamp unity gain buffers? I realize they have their place, but this does not appear to be one to me.

Thank you in advance.

The diode D8 will limit the output swing on the first op amp to ~ 0.6 V above Ref3V. The 1 kohm resistor limits the output current of the amplifier in the case where the voltage is being limited by the diode. A lower resistor will increase the current in the voltage limited state and could overdrive the op amp. This is not good for overall frequency response. The unity gain buffer does not have the additional current to output, so it can provide all of its current to the ADC throughout its range of operation.

There must be some reason that the engineer decided to limit the voltage into the ADC. If the circuit has changed and the diode protection can be removed, then the unity gain amplifier can be removed. Otherwise, I think it is best the leave the unity gain amplifier in the circuit.

If offset is a really big issue, why not find an amplifier with a lower offset?

It is also a filter, from what I'm seeing. A 2 stage low pass.

It is a filter yes, but that all occurs in the first op-amp. The opamp has 40mA of drive capability and the circuit only goes above 3V in a fault condition to a max of 5V (supply rail). So 2/100 = 20mA. It should still be acceptable, as long as 3-4 of these faults don't occur at once because Ref3V as shown is also just an op-amp able to sink 40mA.

I guess a 3V zener with a 100R limiting resistor may be too slow to actually actively clamp at ~3.4V (it is a schottky as it is now) as zener currents pick up quite slowly above their breakdown voltage. Would probably only begin to overpower the 100R current at around 4V+.

Do you guys agree with my reasons for feeling a zener inadequate?

I didn't realize you were talking about the second op amp. What would you do with it? It will be in the circuit one way or another, and unless you have a need for another op amp somewhere else there is no reason to change. If you do change you would likely leave it as a buffer and connect the input to some reference point.

Since there isn't any real way to tell why they have a zener (or Shottky) there it is hard to make the judgement. Since there is already a op amp isolating that point from whatever I'm wondering if anything is needed, but if it is, why?

You can always just try removing the buffer, test the circuit, and see if it does what you need it to do. Test your highest frequencies and values up close to your Ref3V. Ideally, it is nice to have some separation in function; however, that doesn't mean it will not work without the additional part.

If your concern in circuit redesign is the offset voltages, you can try the OPA350, which is similar to the OPA353 but has 1mV offsets.