I'm assuming a similar setup like the later 492 for the following comments...
The issue seems to be one of accuracy. It won't lock in time when high accuracy is demanded by the cpu.
If like the 492 scheme, there are 4 oscillators used in 2nd if, a 2200 fixed from 100 MHz ref, a 16 to 20MHz, a 2182 and 719Mz which is used only for the middle bands.
It's like noise is getting into one of the oscillators, likely the 2182, or 18 hz or the error amp. After the cal, if no joy, check the 2182 error signal is clean. The later 27xx units may have the surface mount caps that fail and or leak. Checking inside rf shields is a good idea as some circuits have a dedicated 5v regulator under shields. I bet it's similar to the older design 492b etc.
