My previous post was for an analog MUX using discrete MOSFETs and a flying capacitor for voltage measurement of the cells in a battery pack. That seems to be a workable design and I am going to get some prototype PCBs made to prove the concept. But I also considered a design using a DG409 (for 4 cells) or two DG408s (for 8 cells). To do that properly involves an instrumentation amplifier or differential amplifier that can handle inputs at the supply rails. The closest I found was INA326 which handles 100 mV above and 20 mV below, but supply is limited to 2.7-5.5 V. There are some "over the top" OP-amps such as the LT1636/7 which can handle 44V supply and inputs above the rails.
However, I made a simulation of a cell monitor circuit that uses a matched pair of PNP BJTs (BCM857) and it looks like it can read the top 3 of a 4 cell pack within about 5 mV. There is some common mode voltage effect, of about 0.1% (3mV for each 3V cell), but this could be compensated for by the microcontroller (PIC) that will do the ADC readings. The measurement draws only about 50 uA when sampling, and essentially zero when the MUX is disabled. There will probably be some error from the degree of matching (should be 2 mV max), and temperature, but since the devices are in the same package there should not be too much effect, and it may be possible to sense temperature and compensate for that as well.
I may be overlooking something, but this seems like a good way to accomplish this sort of measurement. let me know if you see any "gotchas"!
Thanks,
Paul