[time-nuts] Rb vs.Crystal OCXO

Magnus Danielson magnus at rubidium.dyndns.org
Sat Apr 26 16:45:07 UTC 2014

On 04/26/2014 06:34 PM, Said Jackson wrote:
> Shane,
> The trade off for most applications is as follows:
> Rb has much faster stabilization time after power on. Ocxos suffer from retrace, that can take hours to days to get rid off. Retrace could cause a frequency shift of several ppb or more from say 15 minutes after power on compared to 10 hours after power on
> Retrace on a good Rb is very low, turn on a PRS-10 and after 10 minutes or so it will be stable and drift much less than 1ppb over the next days typically. 10x to 50x less retrace than on a good ocxo
> Is possible. This is important when you do not have a GPS to remove retrace error from the ocxo.
> If you run without gps (holdover) the best docxo will start to drift more and more after a day. Rb will stay stable for months or years. Important for base station applications where the amount of drift determines how much time can pass before a repair crew has to be sent. They do not want to send crews over the weekend for example because it could cost double overtime pay.
> Because the loop BW of a Rb is larger than a GPSDO (say 10Hz vs 0.001Hz) a typical Rb will have higher ADEV noise close in than a really good GPSDO due to the loop steering noise being additive to the ocxo noise. This is why a GPSDO can have significantly lower phase noise below 10Hz.
> But it depends on the Rb. For example the CSAC which works on the Rb vapor cell principle actually improves noise close in as Rick explained because it has a fairly low cost tcxo and the vapor cell thus is more stable than the tcxo by itself. On a PRS-10 one can see a steering hump below 1 Hz (around 20s or so depending on the selected loop time constant) that probably would not be there without the loop steering..
> Most of the time Rbs are used because they require much less calibration, have much less g (tilt) sensitivity and much less initial retrace/warmup error. In the case of the CSAC they also have more than an order of magnitude less power consumption than a good Ocxo (0.12W vs 1.7W on a typical docxo)

The PRS-10 have a nice little trick in it, it stores the previous OCXO 
steering value, so on power-up it sets the OCXO to this and that gives 
it about the right frequency and only once the rubidium have heated up 
it locks to it. That gives a relatively quick stable signal for 
starters, which is quite quick.


More information about the Time-nuts_lists.febo.com mailing list