We have a few applications where current leakage on the integrator capacitor (i.e. the capacitor in series with resistor right at the output of the charge pump) has caused problems with locking onto the higher frequencies of the application. We think it is because the charge pump cannot correct anymore after certain range and the leakage of the integrator capacitor on the loop filter causes the loop to go out of lock and not able to maintain lock for more than one PFD cycle. Is this a scenario you have encountered before?
We have on application where the use of ceramic capacitor for the integrator circuit caused this problem because a batch of capacitors had a manufacturing defect where the had excessive microvoids leading to internal cracking of the plate with minimal mechanical stress. Have you seen this scenario before?
We have another application where the use of tantalum capacitor for the integrator circuit caused this problem because the capacitor was not able to self-heal based on the applied voltage leading to increase leakage which would have otherwise been "healed". Have you seen this scenario before?
Any recommendations or tidbits on how critical the leakage current specification for the integrator capacitor? Any rules-of-thumb? Any prior experience with tantalum capacitor "self-healing" on this type of application?