The prototype boards with the SC189Z buck converter came back and their performance is bad. The ADEV in preliminary tests is an order of magnitude higher than the same circuit with an LDO.

I do think I am going to stick with it this time and chase down the problem and squash it. The idea of being able to upgrade to an OCXO is too attractive for me to ignore at this point. But the problem is that it requires an extremely high current, but also extremely stable power supply. That's going to be a big challenge.

One thing I'm trying as well is to provide a separate LDO for the analog section of the board. I don't have high hopes, however. The issue is that the control voltage input to the oscillator is only meaningful relative to the supply voltage. That means that if there's switching hash on either one the result won't be as good as we want. That test is being run right now, though, so I could be proven wrong about that.