A few years back I built a phase locked loop circuit very similar to that of Tom Polk which is linked here: http://www.tompolk.com/radios/frequencysynthesizer/freqsynth.html My circuit topology was the same but some of the component values were different and this is not meant to be critical of Tom's design since his may or may not have the problem noted below.
When on the air with this PLL I noticed a background noise which sounded a bit like a wind whistle in a car and was especially apparent when listening to the carrier on a receiver set to CW mode. Typically, such noise is caused by phase jitter (phase noise) and this was confirmed by viewing the lock signal on pin one of the 74HCT4046 with a scope. There was over 1 microsecond of peak to peak jitter and on the spectrum analyser there appeared a strong subharmonic of the carrier at about 580 kHz.
I did a bit of homework and found this site: http://www.sm0vpo.com/ which is linked from Tom Polk's site. On this site follow projects> synthesizer circuits> improved CMOS synthesizer to find a schematic of a similar circuit. The author shows a schematic with a different loop filter connected to pins 9 and 13 of the 4046 which he claims narrows the skirt on the spectrum for the synthesizer. Since the phase jitter I saw would result in a wider skirt I thought it worth while to try this loop filter so it was installed in my PLL and the noise is now gone. The jitter is now less than 40 nanoseconds peak to peak and is inaudible on the transmitter signal and there is no longer a subharmonic in the spectrum.
If you are using or going to build this circuit it seems beneficial to use the loop filter shown on Harry's page.