The new rubidium clock arrived today from SRS. It is a FS725 and the calibration sheet, manual and control program on floppy are in the filing cabinet in the ATF.
At the moment we are doing some testing of the clock, beating it against a second rubidium clock at the 40m. The setup is that we have the 10MHz output feeding into a mixer (it is a 13dBm mixer at the moment but we should change this). Each output is transformer coupled to avoid ground loops. The output of the mixer goes through a low pass filter (10MHz I believe) and into an SR560 with a gain of 1 and DC coupling. This signal is going into the channel mc_drum1 of the ADC.
The clocks are drifting relative to each other with several minutes period. We need to make the measurement of phase noise close to the zero crossing of the beat so that we are sensitive to phase and not amplitude. We did a few things to try to get the clocks locked in frequency.
First we tried locking the clocks using the freq adjust input on the back of the new clock. We took the mixer output (giving ~500mV for pi radians phase difference) and used an SR560 to feedback. It wasn't clear whether this was doing anything or not. The frequencies did get closer over ~30mins, but this may just have been because the new clock is still aging (and also warming up). After this time it appeared not to be getting much closer.
Next we removed the calibration sticker to get at the manual frequency adjust underneath it, and gave 3 clockwise turns (0.0025Hz per turn). There was no immediate (or noticeable) change to the beat frequency.
We have now left them with the newer clock synced using the 1pps output from the old clock. The sync light is on, but the frequencies were not getting any closer when I left them. We're taking data and we'll look at it again tomorrow to see if they are getting closer. There is some long time constant used for this locking loop (~2hours) that will make this take quite a long time. We can shorten this time constant using the software and RS232 input if we need to.