ADF4252 TPC22 labeled “Phase Noise (Referred to CP Output) vs. PFD Frequency, IF Side” on pp. 10. Usually the phase noise starts out high, close-in to the carrier and then decreases as you look at higher frequency offsets. This graph shows the opposite. It starts at -168 dBc/Hz at 10kHz and goes to -142dBc/Hz at 1MHz. Am I looking at this data incorrectly?

Hi Ed, This is not a phase noise graph, rather it is trying to show how the phase noise scales with increasing PFD frequency.

The standard PLL phase noise formula that is commonly used states that:

PLL_PN(dBc/Hz) = PLL_Normalised_Phase_Noise_Floor + 10log(PFD) + 20log(N)

For the graph we measured the PLL_PN on a spectrum analyser and then subtracted out 20log(N) to leave

PLL_Normalised_Phase_Noise_Floor + 10log(PFD)

this is what is plotted on the graph versus PFD frequency. This graph shows that the ADF4252 phase noise formula holds up pretty well

over a wide PFD frequency range. This was important as some older generation PLLs' phase noise would start to degrade

at higher PFDs which would show up as a deviation away from the expected 10log(PFD) slope.

The reason TPC22 flattens out at lower frequency, whereas TPC21 doesn't, is that you are hitting the IF PLL noise

floor at low PFD frequencies. The RF PLL noise floor is lower.