I am trying to phase lock 10 MHz (ref) and 6.8 GHz (rf) using ADF4107 and the eval board (EV-ADF411XSD1Z).
Instead of placing components on the eval board to build a loop filter, I am instead passing the charge pump to an external integrator/servo.
When I do this method with a higher reference frequency (213 MHz) with ADF4007 and its eval board (EVAL-ADF4007EBZ1), it works very well. In this image, the blue signal is an oscilloscope trace of the Charge Pump output from when the reference signal is below the match point with the RF signal, in the PFD. The green signal is from when the reference is above the match point. Integrating and feeding back a correction signal using the servo produces a stable phase lock and the RF follows the REF. (the charge pump is intentionally biased around 0)
In contrast, using the ADF4107 and programming it for the proper frequencies does not result in similar behavior, and a phase lock using a servo is not working - it's highly unstable, unreliable.
Using the ADF4107, the blue signal is the charge pump output from when the ref (10MHz) is below the RF (6.8GHz), green is when the ref is above. If the polarity is changed in the programming, the above/below condition for these signals flips. Basically, there doesn't seem to be a useful Charge Pump signal from one direction - the 'sign' doesn't change when changing the reference frequency to match them in the PFD.
Also attached are the settings I'm using to program the SDP-S board with the software (ADI PLL Int-N).