Some time back I acquired a KS24019 shelf which was manufactured to provide 15MHz and 1PPS signals for CDMA mobile telephone systems.
It comprises two boxes:
- The RFTGm-II-Rb which contains a Datum LPRO 10MHz Rubidium Oscillator (10MHz)
- The RFTGm-II-XO which contains a 10MHz OCXO and a GPS receiver
and a 6-way signal splitter which takes its input from whichever of the two boxes is currently providing the 15MHz signal (IOW it has two inputs and six outputs). The 1PPS signal is routed out from the RS-422 9 pin ports on the front.
In normal operation the unit operates using the Rubidium as the active unit and fails over to the GSPDO in case of failure. It is however possible to swap over to the XO as the primary. The GPS is used to discipline *both* the OCXO and the Rubidium Oscillators.
There's plenty of pictures of the exterior of these out on 't web, so I won't bore you with those.
Starting with the RFTGm-II-Rb general interior shot:

Now a close up of near the connector into which the LPRO plugs:

You'll notice the 51R1 resistor to ground, a cap to its left, a 20K resistor to ground, and the 47K5 resistor to +5V to centre the 10MHz signal on about 1.5V.
The 10MHz signal is then fed into pin 9 of the Altera EPM7032LC44-15 IC at the bottom of the pic.

On the other side of the IC the 15MHz signal comes out from pin 37 via jumper (E1), L100, and fixed cap (C101) and trimmer cap C102 in parallel to the input of a 15MHz bandpass filter in the centre of the picture.
The output from the bandpass is then fed through a couple of amplification stages, and finally through a Mini-Circuits PLP-18-11 lowpass filter (silver part just behind the SMA socket at the RHS of the first picture).
So the plan is to remove the 15MHz bandpass filter

then to take off 10MHz from the right end of the 51R1 resistor shown in the second picture
using a 470pF radial ceramic capacitor, via a short run of coax to a 20dB Pi attenuator1 made up of two 60R4 and one 249R surface mount two 1K21 and one 4K99 0805 resistors (input and output impedance 1K0 approximately) mounted on perf board to a daughter-board
2 where the bandpass used to be with the output feeding in to the first amplifier (U100) via C103.
I'll make another post with a look inside the RFTGm-II-XO.
Footnote:
1. The signal at the output of the LPRO is .55V
rms. So as to avoid clipping and other distortion, the signal level at the input to U100 should not exceed about 55mV
rms (150mV
pp). Prior to modification this is well over-driven with a signal of about 1V
pp.
2. More details in a later post - the OCXO variant has 5V CMOS signal - need to cope with either type.