[time-nuts] Advise on building a DIY GPSDO?

[Nick Sayer]

I’m coming at this thread a bit late, but I’m part-way down the DIY GPSDO design path and I’d like to share my looking-back perspective, for what it’s worth.

I think it’s good to start with some sort of goal in mind. The problem with that statement is that as a beginner, I didn’t fully comprehend the vocabulary as well as I (think) I do now.

If you share what you come up with, you can be confident that there will be criticism of your design. That criticism will seem to you more harsh than it will likely have been intended. Distill the criticism to as many testable design changes as you can and try them out - hopefully (as much as practicable) one at a time in isolation so you can see the impact on the stability.

That takes one to the next challenge… You must be able to measure your design’s performance. This means getting ahold of a TIA that’s at least an order of magnitude better than your performance target. If you can’t trust your measurements, you’ll never be able to tell if your design is doing better or worse as you make changes. But a TIA isn’t all… You need to also be able to measure power supply stability (noise and ripple) and things like that, so that means a good scope and proper short-lead scope probes for doing it (Dave Jones has a good video on this topic over at eevblog.com). You’re going to need to be able to get performance logs from your device while it’s running so you can see what your firmware is doing. You’re going to want to do this while your device is in its enclosure (an OCXO will perform much, much better in an enclosure than not in one).

As a beginner, I advise designing from new components. This, in principle, gives you some hopes of getting support from the component manufacturers. Connor Winfield has been very helpful in answering questions, as has Semtech (for their SC189 switching controller). Design with the datasheet in one hand and your pencil in the other. :) For OCXOs, pay particular attention to the manufacturer’s recommendations for board layout and chassis considerations. The manufacturers want your designs to work - they sell more of their stuff that way. Take heed of their advice. :) Designing from new will very likely take you into the world of surface mount. If you haven’t done it before, SMD is easier than you think. It didn’t take me long before I came to view SMD as actually easier and faster than through-hole.

Most of all, the people here are by far the most helpful resource I’ve found. Even those whose criticism seemed harsh at first have had an impact on the design I have today. And I believe I can say without fear of contradiction that every change has been an incremental improvement. And credit for a lot of that goes to people subscribed here.