It is a question I have been asking myself for a couple of weeks, ever scince I got a requirement for a low offset drift high impedance true differential amp with high CMRR. I have a solution in the form of the $3 AD627, but I cannot really say that I am **satisfied**. There are some good cheap low voltage single supply solutions as well, but of course they don't let your inputs float around much!
Can it be possible that a such a device costs more than a 24bit ADC? The obvious conclusion is that it is the laser trimming that costs, but keen to get to the bottom of this I went back to basics and blew the dust off my AOE. Now in ch.7 we have the run down on auto-zero techniques, chopper stabalized amps and blow off into differential amps and finally instrumentation amps....and things just dont add up. The first thing that crossed my mind is why don't FET input amps switch around thier inputs to get a zeroing effect. Ha! there is an example, the ICL7605 which uses a capacitor to pass a voltage to a chopper op amp. The text notes that the drawback is the noise, but I would expect that 16 years on somebody would have come up with such a device with a low pass buil in. Better still, why not make an ADC with this sort of front end built in....go and pick up a voltage somewhere and then eg. use it in a capacitive SAR, rotating it in opposite polarities each conversion. Alas, not only can I not find such a device, I can't even find a ICL7605.
The second thing I find curious is the theoretical **advantage** of the instrumentation amplifier over a differential amplifer with simple buffers on each input. The text notes that one of the snags of the differential amp is that tight resistor matching is required to achieve high CMRR, whilst in the instrumentation amp configuration this is not necessary. OK, but let's have a look at the 'INA' range that the text quotes...the laser trimmed differential amp devices cost less than the true instrumentation amps. How come? Of course the snag with the instrumentation amps is the low impedance, of course it is not a problem in some apps such as thermocouples but if your measuring a bridge, well it is a disadvantage. Were it not for price the true instrument amp would win hands down over the diff amps in nearly all applications....if nothing else they are also easier to protect. So the fact that a range of instrument amps also includes lowwer cost low impedance diff amps suggests to me that there is **something** in the instrument amp that makes them costly.
Clearly I have missed something here, or perhaps there is some new technique that eclipses all requirements and instrument amps are considered legacy?