Instrumentation amplifier: why subtractor block?

[kirikinny]

Hi, I'm looking at a circuit of a classical instrumental amplifier having 3 opamps, 2 for buffer, with Rg to adjust the gain. The output from this buffer + gain circuit is then feed into a third opamp, in a circuit that is the 'basic' differential amplifier, but whose function now that Rg is included in the buffer circuit, is only to match the resistances to have no common mode gain.

(this is the circuit I'm referring to: http://upload.wikimedia.org/wikipedia/commons/e/ed/Op-Amp_Instrumentation_Amplifier.svg)

My question is: why do we need the opamp on the right? If we take the output directly from the buffer circuit (we take voltage difference between the 2 outputs of the 2 opamps on the left), we obtain the same gain, and no common mode gain even if there is a resistors mismatch, right? because: Vout = (V2-V1) (1 + Ra/Rg + Rb/Rg), being Rg the gain controlling resistance and Ra Rb the other 2 resistors in the buffer circuit.

Thanks

 

[dlgoff]

I found this explanation for 3 Op Amp Instrumentation Amplifiers.

The differential amplifier part is often essential when measuring sensors. Why? A sensor produces a signal between its terminals. However, for some applications, neither terminal may be connected to the same ground potential as your measuring circuit. The terminals may be biased at a high potential or riding on a noise voltage. The differential amplifier rescues the signal by directly measuring the difference between the sensor’s terminals.

 

[mjhilger]

I'm not following the question. Are you asking why go through the right opamp which provides a single output referenced to ground as opposed to just looking at the TWO outputs from the buffers? If so, let's say we are reading the value with an A/D and the value is not stable or we are reading very near the Nyquist limit, using the third opamp allows a single A/D to determine the value, where without it you would have to take two readings and subtract in software; but that would only be true assuming the signal stayed the same during the two measurements.

 

[kirikinny]

Ok, I see, I thought that, if you just need to measure the output, you connect the ground of the picoscope to one of the two ouputs and then measure the other one. But I guess if you want to insert the circuit inside a bigger circuit you could not do that

 

[dlgoff]

...why do we need the opamp on the right?

Besides input biasing and noise, by adding an op-amp to the differential amplifier's output, you can make a current amplifier.