50Hz (and 100Hz) interference may masquerade as phase noise; even the very best phase detectors have some amplitude to phase conversion, and the cartridge itself is mildly non-linear, so can do this as well.
A simplistic example may help. Imagine the phase detector is just measures the positive going zero crossing time of the 3.15kHz sine wave - I know the real one is better than this, but bear with me. If you get interference that adds to the signal, the crossing is early; if it it subtracts it is late. So simple additive interference can masquerade as phase modulation. Now, the algorithm in use is lot better than timing of zero crossings, but the effect remains at some level.