We have problems with the LMX2364 . PLL in the circuits creates a variable frequency between 9 and 10GHz and we have discovered that the lost nail in PLLens output ( that controls a discrete VCO) . These spikes have been difficult to explain and there are no signs of any supply voltages that something is strange, the only correlation we have yet to find is that the nails appears to occur just after switching many bits in the internal registers .
The frequency selected varied in something that can be likened to a ramp , thus continuously changing the output frequency so that the voltage output from the PLL exhibits a ramp. Then you go back to the starting frequency and awaits the next ramp. PLL reprogrammed during trace, at a frequency that is 41.667 kHz ( with 31 bit messages over SPI) . It registers R4 as an reprograms RF_N and RF_FN to get different output frequency , according to the data sheet should not be any problems doing this " A single word write to this register ( R4 ) is all that is required to power up and tune the RF synthesizer to the Desired frequency " ...
Below is a typical appearance of such a spike ,
Unfortunately you can see the nail bad in the upper part ( the boom ) but lower part shows the zoomed-in . The frequency of oscillation that arises is recognized as the bandwidth of our external filters.
One can see here that when you are sending data where you change the input from 16FF84 to 17004 (which means ten bits change value at the same time ) so the nail occurs when this change is clocked into . Since 8-11 bits changed in the message about 15 times per scan so happens that in a majority of the sweeps ...
PLL is configured with a digital lock detect output on the selectable pin FTEST / LD and sometimes detects circuit itself that it lost sync at that spike but not always.
As we can not explain this phenomenon with external influences such as watches, ripple in the supply voltage or noise starts it feel like there is a danger that it is caused by signal integrity problems internally in the PLL .