I am working with a custom built PCB that uses an AD5933 impedance converter to measure complex impedances from 1Hz-100kHz with logarithmically equidistant steps (this latter means reprogramming after each frequency). Due to the large frequency range and specific measure frequencies I use about 10 different MCLKs (depending on the region) and about 10-15 frequency codes for each thus having 125 measure points.
For the impedance magnitude results (after gain calculation) the standard deviation is negligible everywhere and the results mostly fit the ideal curve. But at some MCLK changes the results jump, but only there.
For the phase results (after compensating) the standard deviation is also small but the jumps are there (again they are at MCLK change points).
Has anyone faced a similar a problem, and any ideas what could cause it?
P.S. I have attached one Bode diagram.