I am thinking noise differences due to zener knee device to device
differences there is no "matching". Nothing in datasheet that gives
pk-pk min/max or its distribution. Its a statistical process, not deterministic.
As far as any other differences the 1 uF coupling cap tolerance and leakage
one part to another is poor. The tolerance is+/- 20%, so V divider effect
with 1 Mohm R will effect both offset and G in signal path. leakage can be
in uA range = volt(s) of offset due to 1Mohm, and bias current of OpAmp.
Regards, Dana.