Running in partial assist mode, I've found that when stepping frequencies between 11.9G and 12G the FCAL_EN procedure can't seem to decide if it should use VCO4 or VCO5 (sometimes switching back and fourth every frequency step), making the lock time 140-210us (rather than the 20-30us almost everywhere else).
The datasheet specifies, VCO4's upper limit is 12G (exactly), while VCO5's lower frequency limit is 12G (exactly), but FCAL on the device sometimes chooses VCO5 for a frequency of down to 11.9G. The datasheet also specifies that VCO4 should be used from 11.9G to 12.1G (which confusingly contradicts the table above). I have also found that even with *NO ASSISTANCE* FCAL_EN will choose to start using VCO5 at around 11.9Ghz , which is below the specified range.
My partial assistance settings (VCO_DACISET_STRT/VCO_CAPCTRL_STRT/VCO_SEL) are read from the device after a lockup, stored and written back again later.
After running an initial calibration sweep, and saving the results of each FCAL, I can set VCO_SEL_FORCE, and the lock timing for the 11.9-12G range returns to the 'normal' 20-30us.
For my application consistent lockup times are ideal, but being very sure the frequency is stable at the current setting is more important. I have verified these behaviors on both of my sample boards. I assume there is something going on between VCO4/5 given the behavior and the note after table 6 in the datasheet.
My questions are:
1: Will the device be damaged after long term use, allowing the FCAL to choose a VCO and run it outside its specified frequency range?
2: Is there any reason I shouldn't force VCO_SEL (after a calibration sweep), but allow FCAL to tune the other VCO parameters (CAPCTRL/DACISET) to its liking?
3: Would it be necessary to periodically re-run the calibration sweep to choose a new VCO_SEL after continuous running for a period of time? [temperature drift/phase performance?]
