I have an LMK04832 fed by a 10 MHz source through a balun into CLKIN0. The input buffer mode is set to bipolar, so I can't use the LOS detectors. For backup, I have an LVDS signal (also at 10 MHz) into CLKIN1, this has worse phase-noise than the main input. I can switch between CLKIN0/1 using the CLKinSEL_MANUAL bits in R 0x147.
I would like to use the holdover function so the part will force the VCXO to be stable in the event of a dropout. With R 0x150=37 (HOLDOVER_EN=1, exit mode DLD, PLL1_DET=1, VTUNE_DET=1, CLKin_SWITCH_CP_TRI=1), I see that when I switch from CLKin0 to CLKin1 (or vice versa), the device enters holdover for a few seconds, then regains lock. If I disconnect the CLKin0 cable, however, PLL1 goes unlocked, but HOLDOVER is never asserted. Why doesn't the device enter holdover even though PLL1 is out of lock in this case? I tried fiddling with the DAC trip values, but no luck.
Another question about holdover: When I allow my system to stabilize, the DAC value is normally about ~640 counts. On start up, the DAC is at 512 and takes some time to reach 640. If holdover is enabled during this time, DLD will not be asserted since there is no phase match between the ref clock and wherever the VCXO gets tuned, so the device will never leave holdover. When using bipolar inputs (and therefore no LOS signal), is there a reliable way to exit holdover when the DAC value is not close to it's 'normal' one?
Finally, is there a penalty to using MOS inputs in my case? I use Bipolar following the data sheet recommendation for differential inputs.
Thanks in advance.
