First off, I'm putting together a battery stack monitor for a 20-string pack with an absolute maximum common mode just south of 250V, where my end goal is 16 single ended measurement points for some data acquisition hardware in a lab environment.
I attempted to operate these AD8479s with a single 15V supply, and with the reference pins tied to ground, the output rails either high or low (seemingly randomly between chips). If I power the device with a split 9V supply, with the reference pins tied to the virtual ground, the output follows the differential on the inputs pins perfectly. The problem is, as I have narrowed through experimentation, there must be some non-zero reference voltage on the REF pins when the V(-) pin is tied to ground for single supply operation.
So my question then is: what exactly are the limitations of this chip with regards to single supply operation? What limits are imposed upon the reference voltages when the negative supply is tied to ground for single supply operation?
I have used the LT1990 for this exact same purpose in the past, and it functions exactly as expected in single supply mode. I was hoping to utilize the AD8479 in order to leverage the higher common mode rejection ratio as well as my academic pricing at Mouser (Mouser doesn't carry LT parts).