Anyhow, I unpacked the GPS disciplined oscillator and the two distribution amplifiers and set them up on the bench in a temporary configuration for testing purposes over the next few days but I've got a location in mind for them for permanent installation when the time for that comes.


I got out a new power bar to accommodate the three power adapters. The way the plug prongs are oriented with respect to the rectangular adapter casing is annoying because they would block several adjacent sockets in a power strip. If they were rotated 90 degrees, you could space them closer together. Also, there is such a thing as too bright when it comes to indicator LEDs. These are painful to look at when you're trying to see if the middle one of the three is on.
The test jig is a Hewlett-Packard 5335A counter that has one of the high stability oven temperature controlled oscillator options on it to monitor the frequency of the GPSDO teed off to a Tektronix TDS640A to observe waveform. Either of those machines can provide a 50 ohm load termination so no external terminator was needed to maintain matched impedance throughout the entire system. It's going to be interesting to see how close the counter and GPSDO are to each other or if there's a significant discrepancy between the 10 MHz output of the GPS disciplined oscillator if it can lock up and stabilize, and what the counter claims the frequency of the output is.
The first test was to patch the test jig into the 10 MHz output of the GPS disciplined oscillator and verify it was working. It was initially off by about 26 Hz according to the 5335A until the lamp indicating GNSS lock came on and the oscillator settled in - it settled in fairly quickly and it was only out by a cycle with respect to the counter when I started taking pictures. I was able to get this picture of the counter just as the GPSDO was still settling in:


The waveform matched what the specifications of the GPSDO said to expect so it was time to move the connection feeding the test equipment over to one of the distribution amplifier outputs and connect the distribution amplifiers.

When I bought the equipment, I had thoughts of using a tee connector to effectively Y cord the output of the GPSDO to feed both D/A inputs in parallel but with a fallback plan to temporarily daisy chain one D/A from the other if paralleling the two distribution amplifier inputs loaded down the GPSDO output and order a third distribution amplifier to act as a top level D/A to take the GPSDO output and feed it to the other two D/As. Luckily, this wasn't necessary. Connecting one distribution amplifier to the GPSDO and then adding the second one didn't cause any problems and didn't cause any shifts in output from the D/A to the test equipment.


The GPS disciplined oscillator got a GNSS lock and settled in very quickly and it and the HP 5335A are very close to each other. The original idea I had for genlocking the test equipment to a master reference was to use the 5335A's 10 MHz reference out as the timing source because it has the optional timebase I mentioned earlier. Given how close it is to the GPS disciplined oscillator, I'd have no problem with substituting it as the master clock source to one or both distribution amplifiers if the GPSDO ever runs into problems and I have to wait for a replacement to come in.
I also put up the scope's frequency measurement function:

It's interesting to note how much further out the frequency given the scope is compared to the counter so it'll be interesting to see if that improves by feeding a 10 MHz reference into the scope from either the GPSDO or the counter's reference out and then measuring the other to see if it tightens up similar to how the GPSDO and counter are.

Finally, here's a picture of the two tests that are going to run overnight (minus the scope - there's no need to run up the CRT hours unnecessarily since it's frequency stability I'm most interested in). The 8903B is still running and it's numbers haven't changed much so far. I did try toggling the the 80 kHz low pass filter on and off and it does appreciably lower the THD+N vs. the analyzer input wide open.