Due to lightning within 60nm.
SO.......we spent the rest of the day going over our system, trying to figure out where the "missing" 20dB went from the other day when we tested it, and why the remote spectrum analyzer was acting all wacky.
We checked every connection (SMA and TNC) in the system, ruled out any network errors or timing problems, powered down/powered up, and rebooted everything, to no avail.
Finally, I decided to connect a different spectrum analyzer to the cable coming out of the switch matrix, on the off chance that the rack-mounted spec-a, just back from calibration, had a problem.
The Type-N connector soldered to the semirigid cable came off in my hand as I removed it from the input of the spec-a.
Well, at least I found it........