@neopticus If you are serious about wanting a reference to set others by then you need a GPSDO that's on all the time. You mention using a rubidium to run by and calibrate that by GPSDO every month or so. If that's the case then you need a very good GPSDO and it would have to be on all the time to reach a resolution that would exceed that of the rubidium.
Proper timing GPSDO's such as Trimble Thunderbolts, allow you to tune their behavior to your location. i.e. you can set an elevation mask to minimize multi-path interference, change the loop time constant and damping factors. You also need a clear view of the sky for as many satellites as possible to get the best possible accuracy.
A good GPSDO will also profile their own OCXO's so that when sats are obscured and they go into holdover, the oscillator is steered in the best possible way to maintain the highest possible accuracy until enough sats are visible again to steer things properly.
And if precise time is your thing it a good GPSDO allows you to insert a propagation delay for the length of cable between the unit and the antenna to get that last 60 odd ns out of the way.
Also it can take a month for a GPSDO to attain 1E-12 or better accuracy and keep it.
You have to ask yourself... "What level of accuracy for 10MHz or 1PPS do I want ?"
There are many proper GPSDO's for reference & timing available on eBay for such purposes without spending too much money.
e.g.
http://www.ebay.com/itm/HP-SYMMETRICOM-Z3805A-GPS-Disciplined-Oscillator-GPS-Frequency-Time-Receiver-/171293077196
http://www.ebay.com/itm/HP-Z3815A-GPS-Frequency-Time-Receiver-GPS-disciplined-oscillator-10-Mhz-1PPS/171293069062
http://www.ebay.com/itm/FEI-PicoSync-GPS-Time-Frequency-Engine-GPS-Disciplined-Oscillator-10MHZ-PP2S/181389532308
http://www.ebay.com/itm/HP-Z3816A-GPS-Frequency-Time-Receiver-10-Mhz-1PPS-/271906200825
just to name a few.
TSL, thanks for your advice. I was aware that the professional systems for timekeeping are more accurate, but since they are too insensitive (to be be used in my apartment), they were ruled out pretty fast. As such, the list of eBay items you gave is not is not useful in my case. But I appreciate the effort, and I think many budget restricted (but not open sky restricted
) members here will appreciate it too!
I should have been more precise (no pun intended) in my description of the use case: I mentioned checking the rubidium (as in "is it on the same ballpark as GPSDO?" or "not-drifted or completely failed") but not necessarily calibrating it to its highest possible accuracy.
As I've understood, Rb standards are accurate in 10e-11 level, and to calibrate them an accuracy of at least 10e-12 is needed, which would then require 24h running, high end GPSDO with at least few weeks uptime, as you mentioned.
Now, to my use case: Here are some examples of the OCXOs I have in my possession:
HP 53132A option 010: Temp stability: <2 x 10e-9. Turn-on stability 30 min: <5 x 10-9. Aging: <1.5 x 10e-8/month
CMU200 option B12: Temp stability < 5 × 10e–9. Turn-on stability 2h: ±5 × 10e–9. Aging ±3.5 × 10e–8/year.
HP 8657A option 001: Temp stability: <7 x 10e-9. Aging < 3 x 10e-8/month.
Calibrating these would need accuracy of just 10e-10, so anything better than this is basically just money thrown away, right?
I read the review of BG7TBL 2014-12-09 by Tim Tuck (link on the first page) and there he states: "After a few hours the frequency settles further to within a Milli Hz or so of 10MHz."
This would be accuracy of level 10e-10 to e-11 and well enough for my needs, or is there something I have failed to grasp here?