We use AD623 to amplify a 350ohm full bridge strain gauge. Recently due to the global shortage, we switch to AD8220 which was available at the time. Both bridge and AD8220 are running off a single supply of 0-5V. The problem I have is that for the first 1000 identical PCBs we made, the output on 25% of them is limited to 1 to 3V. The AD8220 is set up to take differential inputs and according to the datasheet, the diamond shape output limitation should be valid on common voltage mode only. Otherwise, the output swing mentioned in table 2 which is 0.2 to 4.80 is very misleading.
Being connected to a full-bridge strain-gauge, both input voltages are very close to 2.5V across the operational range which is within the limited input range of the chip.
Please help me understand why the chip is behaving differently from the datasheet and why its characteristics are changing randomly. After many tests, I am pretty confident it's down to the chip itself.
I can mitigate this in the new boards by reducing the gain and compensating that in the next stage but we have a lot of rejected devices due to this issue already. I have used LT6370 instead of AD623 without seeing this problem but AD8220 seem unpredictable.
Please see the attached schematic.