So I'm airing a bit of dirty laundry, hoping for some wisdom. I re-did a pcb for a high impedance Instrument Amp. (INA121) The first incarnation I posted about here a month or so ago. (Dang thing oscillated at highest gains.) I re-did the pcb two ways. The first was just to add a ground plane to both sides of the pcb (ground plane was originally only on the bottom) to try and reduce the capacitive coupling between output and input. (I also rerouted a few traces to keep high level signals away from input.) On the other one I changed to a SO-8 package and tightened up the design. (moved int-amp closer to input.) Also with ground planes everywhere.
Both pcb's work OK, but to compare the frequency response of each channel (+ and -) I looked at the pulse response, (and used the DSO to invert the (-) signal.)
So here is the TH (Dip-8) signal, (1 G ohm input impedance)
A little bit of difference before ~1ms and then the traces lie on top of each other.
Then the SO-8
No weirdie's at the beginning, but it's like there are two different time constants.
At lower input resistances things are similar... The pulse response of the two channels is closer for the DIP layout.
Oh these pics are all with a gain of 1,000, at lower gains (100 say) there barely a discernible difference in the two channels.
George H.