Why does the spectral waveform change in Continuous Wave mode?

Recently, I encountered a thorny problem. Neither the software nor the hardware has changed, but 10% of the cards produced do not measure distance or the success rate of measuring distance is only half of the original. I found after test that the waveform of the continuous wave mode is the same as that of a good card when it is first powered on, but the waveform changes a few decades after it is powered on. The UWB schematic diagram of the card is as follows:




The waveform of continuous wave mode in the first ten seconds after power on is as follows:

The waveform in continuous wave mode is as follows ten seconds after power on:

Who knows what causes the spectrum waveform change?
Thank you in advance!

Hi @rainDecaWave ,

It could be due to the PLL lock time is too long so you see a unlocked CW signal at the first 10 seconds and after it settles. Do you always see the first picture for the bad boards that you have or it similarly settles after a 10 seconds?

Kind regards,
Emre