The HP8640B is an excellent signal generator for RF testing. It has not been produced for many years but continues to be an important part of many labs. They are known for there extremely low phase noise which is important when testing receivers.
I received mine over a year ago and have always been concerned with what I considered to be poor phase noise performance. A very simple test is to tune a receiver to the generator output frequency and then use a PC sound card program to look at the frequency spectra. With my generator the image would be triangular in shape rather than a narrow spike. Furthermore the output would wander up and down in frequency by 10's of Hz (depending on the band). I compared my output to others and did some automatic logging of the frequency over many hours to verify that the output was not stable. I did other testing and corresponded with other HP8640B owners and decided the results that I was observing were not typical.
To make a long story short I discovered that there was low frequency noise on the Varactor tuning line. If I removed the A7 board and powered the Varactor with a battery I saw the expected stable 'spike'. By tracing back into the circuit (and doing some LTSpice simulations) I eventually came to the counter board. This circuit generates the correction voltage for the varactor tuning diode to 'nudge' the cavity back to the correct frequency. The driver is a transistor configured as an emitter follower. This transistor is driven by a FET. I was able to measure noise on the transistor but not at the FET. According to the schematic that I had there is a resistor connecting the two devices. How can a resistor generate so much noise?
I removed the board and discovered that my board (1977) actually had a variable resistor between the FET and the final transistor. The purpose of the variable resistor is to set the anode voltage at a midpoint so the correction could go equally in the positive and negative directions. I gave this variable resistor a quick squirt of DeOxit and rotated it back and forth a number of times.
Once I reinstalled the counter board the generator powered up with an extremely clean and stable output. It continous to run with a very stable and clean output.