I took 5 spectra, 4 with the Agilent at high freq from 1kHz to 5 MHz and 1 with the SR785 from 100 mHz to ~10kHz, stitched them together, and calibrated them. 

The original spectra were all in Vrms/rtHz, and the two spectrum analyzers were not in perfect agreement when it comes to calibration.  Their levels were about a factor of 2 off from one another, see the last plot... the SR785 is the pink line that reaches low frequency.  I put them together anyways. 

The Marconi I tuned down to a relatively low (for us) 3 kHz/Vpeak FM demod.  This was a mistake, the temp fluctuations are still off the charts and I had to babysit the PLL all dang night.

I took a PLL OLG measurement by doing the CLG = OUT2/EXC measurement, opening the loop, doing it again to get the excitation gain (should be the same as the preamp gain which was x10), then combining it to get the OLG between 100 Hz and 10 MHz.  I need to fix our PLL OLG, whatever I'm measuring this is not right b/c it flattens out at 10 kHz.  Also I'm still using the physical measurement to calibrate the spectrum instead of some 1/f fit.  This limits our trusted spectrum calibration to between 100 Hz and 10 MHz... at very low frequencies the gain of the PLL should be higher. 

When I talk about PLL OUT2/EXC, I mean that I put in EXC into the B input port on the SR560 preamp, and read OUT2 from the 50Ω output.  My OUT2/EXC measurement confirms the leveling out at 10 kHz.  I am probably saturating something.

Despite these glaring issues I calibrated the spectrum anyways, recalling the 3 kHz/Vpeak marconi settings, and got plot one.  I'm posting this because this is not a completely crazy spectrum, even if the PLL OLG is weird.  I'll iron this out.  We need a good Start of September Spectrum so that when we get to the End of September and we're done venting and putting in temp controls and PMCs we gotta know that it wasn't for nothing.