What methods are you using to do this particular test? (Phase noise system , good spectrum analyzer )
Problem with measuring this on a spectrum analyzer is the measurement is only as good as the noise floor on the spec an. I have the phase noise option on our E4446A but it only measures down to -120 dBc/Hz. Rohde & Schwarz held a phase noise seminar here not too long ago and they have some nice equipment that measures down in the -160's.
There are a number of options available. The E5052A/B are pretty nice and available for about $30K and up used. We also own an HP 4352B VCO PLL Analyzer. Look at the specs to see if it meets your needs. And also the HP 89441A has a Phase Noise setup (need to get a white paper to see how to do it). It is okay for down to around -137dBc/Hz or so. I really like the E5052A/B boxes. They are fast and have a low noise floor for doing SSB Phase Noise.
We use the E5504B. very good system, but the E5052B is better. Depending on the freq you are testing,you may also need an E5053A to get to 26.5GHz
E5505A, we test a lot of the ESG/MXG/PSG models and can get down in the -160 or better at 100 hz to 100 kHz offsets by using the 3 source analysis assuming the generators are up to it like a PSG w/UNY option in Low Noise Mode. It takes about 15 min/msmt to get that
R&S FSUP is also good, probably as good as you can get out of what is basically a spectrum analyzer platform.
I've seen (and used) several of the Agilent/ Keysight systems (PSA w/phase noise personality, E5052B/E5053A, E5505A i think it is). They're all good depending on the application, however the E5505A is really only as good as your signal sources using the three source technique Bryan mentioned. To get down in the 'dirt', that'll cost you over $300k, just for the sig gens.....
Thank-you for all the input. We have the PSA4445 Series but confidence level is low.