Refugee wrote:... The only way I can see for this happening is by the capacitor between the two halves of the bridge having less leakage than before ...
I would hope that any
capacitor (other than an electrolytic) would have zero
leakage! A capacitor with less
than zero leakage might seem desirable but it is practically and theoretically impossible to achieve!
Is your 2% error consistent throughout each band or does it vary from end to end? If so, have you considered that it might be a mechanical problem whereby the pointer is slightly offset from its datum mark - like a radio where the pointer position has moved on the cord?
If the problem is electronic, a low reading suggests slight loss of capacitance
somewhere - considering that the error is only 0.5% out of spec, it won't be much!
Another thing to consider is how accurate is your DFM? And how do you check its calibration?