Currently, I am working on an amplifier design concept selection which will amplify thermocouple output signal 80uV/degree Celsius. I have found many instrumentation amplifier circuit design using single supply as well as dual dual supplies. If there is anyone who has good experience on this, may I ask what is the primary advantage of using a dual supply signal conditioning circuit over a single supply circuit, assuming that we are using the same Op-amp and same instrumentation amplifier configuration.
-- If your input signal is always of a single polarity (as it is for a thermocouple), other than a DC offset somewhere (not too hard to do) there's not much difference in what your circuit will look like.
-- In general, dual supplies are easier when you have multiple amplifiers in a chain and want DC coupling (as you might for a thermocouple) because getting all the offsets to track correctly becomes a pain.
-- The more-or-less converse: With AC coupled designs (which is not your case), regardless of whether the inputs are of one or both polarities, the single supply design is often only trivially harder than a dual supply design.
-- With dual supplies you have to come up with that second power supply, which won't please people if your little amplifier circuit is the only thing on a big board that needs it.
Basically... dual supplies makes the design easier; how much so (and how much harder a single supply design is) depends a lot on the application. When starting out a design, if you're using something like SPICE most people start with ideal op-amps first (which implies dual supplies) and then modify it as needed to make it work with single supplies, swap out the ideal op-amps for a real model of the chip they intend to use, etc.
I was thinking of the case where, e.g., amp 1 (say, low noise, fixed gain) uses the pseudo ground directly but then a second amp is going to provide its own pseudo ground to insert a DC offset or somesuch and the offset is referenced to the first pseudo ground voltage -- in that case you might introduce various small new voltage offsets in the process.
I suppose it's only a problem at high gains -- where millivolt offsets somewhere end up being volts by the time you're done.
Yep. Heck, if you just need AC coupling at the output, you can still use single supply op-amps.
Much like in linear regulators, you probably don't want a lot of gain in the virtual ground circuit. Thus you would probably not want an op amp that has two gain stages since they take longer to settle.
Basically you care that the voltage is stable under transient conditions more than you care about the actual voltage. Some argue all you need are rail splitting resistors and a good buffer chip. See page 12: