I am using a high-common mode differential amplifier (AD629) to measure several DC 10 to 15 voltages stacked together. Since the AD629 is not the cheapest amplifier in the world (though not as expensive as an instrumentation amplifier), I want to multiplex one to measure every signal and save power, probably with a solid-state relay (SSR) since they have a relatively long lifetime (number of cycles) relative to electro-mechanical relays. Any mismatch in the SSRs will reduce the common-mode rejection ratio. Since this is unavoidable, so I would like to calibrate the unrejected portion out, as well as any other bia voltages that may enter the signal later down the circuit path before it is sampled.
What is the best way to do this? My guess is to use a high precision voltage reference on the input. If the reference is not connected to the common-mode voltage, it only calibrates to some of the biases, but not anything due to the common-mode voltage. Would it be possible to connect the reference circuit's ground to the common-mode voltage as a virtual ground?
Then a new problem is introduced, which is how to power the reference using the common-mode voltage as a virtual ground? I could use a voltage regulator with the common-mode voltage as ground and the top of the signal as power, but I do not want an input capacitance to the regulator impacting the signal. Besides, isn't drawing unnecessary power from the signal of interest frowned upon, even if it is essentially a DC supply voltage?