Tool/software:
I am seeing the LMX2820 unlock at some frequencies when performing a random tuning test. We know there is a problem with the doubler in autocal so at the moment we stick with frequencies that don't need the doubler to troubleshoot the unlock issue. We find that at some frequencies,(~5600MHz, ~6800MHz) typically with N in the range of 13.75 to 16.75, unlock can occur.
When the previous N is 18.75 and in a locked state, N is then set to 16.75, and R0 is written, the unlock can be observed quite frequently. (On another board this happens when N is changed from 18.75 to 13.75)
When this happens, the Vtune is scoped and observed to be oscillating with an almost sinusoidal wave centered around a Vtune of 1.8V with an amplitude of around 0.35Vp-p and a frequency of 2 - 3 MHz. The resultant output is an FM signal centered at 6.735GHz. Ref is around 298.5MHz and R divider is 1. MASH order 1.
When R0 is written a second time without writing to the other registers, it is then able to lock. Vtune then settles to a steady 1.36Vdc and the output is a clean 6.792GHz.
What could be the problem? Why does it tend to get stuck in an oscillatory state and requires a 2nd writing of R0 to get the Vtune out of the oscillation?
What can I do to prevent this unlock from happening?
Loop filter is stable with 75 degree of phase margin.
Thank you.
