lo phase noise

When we talk about the VCO or PLL phase noise, we rarely specify explicitly it's the PM or AM noise. And simulation and measurement always give the total of the both.

However, when LO is used to perform frequency translation with RF via a mixer, it is not a straightforward multiplication (which is exactly what we've assumed in the analysis).
In the context of a Gilbert mixer, LO just commutates the switches sharply during its cross-zero points, making PM noise on LO directly get into the IF. Once that's done and one switch is on and the other off, AM noise on LO hardly finds its way into mixer (somewhat like cascode noise).

Maybe there is some other AM-PM mechanism (if there is, can anybody tell me?). But at least, can we say that AM PN is greatly suppressed during the hard switching?
Do we truly overestimate system degradation (SNR, ACS) introduced by the measured phase noise?

Exactly what "system" are you talking about? A QAM or OFDM comm system would have a very different anser from a Radar system.

If you consider that the noisy carrier is a clean carrier plus noise (and that the carrier level is much stronger than noise), the in-phase component of noise is esentially AM, and the quadrature component is esentially PM.
When one speaks about phase noise, it is the quadrature (PM) component of noise. The in-phase (AM) component is theoretically lost in a limiter or a mixer, as you pointed out. It is usually weaker than phase noise in oscillators and it is neglected in phase noise analysis.

Regards

Z