[time-nuts] To improve a Tbolt?

John Miles jmiles at pop.net
Fri Nov 21 22:00:36 UTC 2008


> On your rubidium comment, one needs to be a little careful about
> expectations. For short tau (say 0.1 to 10 seconds) your average
> cheap eBay surplus telecom Rb will have far less performance
> than a good OCXO. Yet mid-term (say 10 to 10^4 seconds), a
> Rb-GPSDO will win. However, long-term the LO makes much less
> difference since GPS always wins. And GPS aside, clearly the
> holdover performance of Rb will blow away OCXO.
>

Good point, there's a big piece of the curve where simply upgrading a
quartz-based GPS standard to drive an Rb would not be a win.  Two pieces,
actually, since Efratom (at least) doesn't seem to care much about phase
noise.

> So it all depends on ones need. I guess my main point is that a
> typical rubidium-based TBolt is not necessarily, just because it's
> "atomic", automatically better than a stock TBolt at every point.
>
> If you'd like to test an FRS or LPRO version of a TBolt for me let
> me know. I'd rather see your plots than my words.

I could graft a rubidium source onto the original Thunderbolt platform I
used for the OCXO tests, but I currently don't have a good way to look at
Allan deviation compared to the TSC/maser setup you've got.  I can measure
the phase noise easily enough but it'd be better to get it back to you for
longer-term tests, especially if the loop parameters need to be adjusted
iteratively.  That would be a heat-death-of-the-Universe process on a
3048A...

-- john, KE5FX




More information about the time-nuts mailing list