There's been a lot of experimentation with the design in the attempt to upgrade to an OCXO, but the result is still unacceptably unstable. The problem is that I am not sure what to do about it.

The SC189Z based buck converter is performing well - it's down to 3 mV P-P of noise and ripple measured at the output filter cap. I think that's downright miraculous for a buck converter - particularly one that's that small (around a square cm of board space).

On the control voltage side of things, I've added an integration cap in parallel to the feedback resistor on the compression amp, and a low pass filter on the output. This was in reaction to seeing something like 6 mV P-P of oscillation on the CV pin. It's not entirely clear where that was coming from, but it's gone now.

Still, even with all of that, the stability of the first prototype is an order of magnitude worse than the current TCXO design. It's supposed to be the other way around.

I've ordered another oscillator just on the off chance that all of the experimenting has damaged the first one somehow. I've also ordered more boards with the current design tweaks. I'm going to try and build three new prototypes when they get back - one with a TCXO, one with an OCXO, and one with a particularly good (that is, expensive) OCXO. I could believe that the DOC020V doesn't offer any improvement over the DOT050V. After all, the data sheets show the same short-term stability specification. But the OH300 should be much better than both.