[time-nuts] Disciplining Rubidium

[Bruce Griffiths]

Tom Van Baak wrote:
> Bruce,
>
> Can you explain this a bit more? I know you often mention
> this point. But it seems there must be something more to it;
> otherwise we all would have seen low-cost carrier-phase
> GPSDO products on the market over the past 15 years...
>
> Instead there are only a few, they are all very expensive,
> and none (?) of them use standard OEM GPS receivers.
>
> I ran a carrier-phase Ashtech Z12T here for a while. It was
> my understanding that the reason it performed so well was
> a combination of carrier-phase tracking, L1/L2 choke ring
> antenna, phase stabilized cables, and dual band receiver.
>
> It also required an external free-running 20 MHz laboratory
> reference (I used a 4x multiplier off a cesium or maser).
> Further, one used it by collecting raw RINEX data and daily
> sent the batch files to be post-processed for two weeks.
>
>   
Whilst you need to do this to use precise satellite orbit data and 
repair cycle slips for the ultimate performance.
However for slightly lower performance, especially when disciplining a 
frequency standard as opposed to a time standard this is not necessary.
The very fact that  a commercial carrier phase disciplined standard 
using a single frequency L1 receiver is available surely attests to that.
These devices do not seem to require either phase stabilised cables or 
use of a choke ring antenna (they appear to use a quadrifilar helix 
antenna).
However they do use a local oscillator and mixer to downshift the 
carrier frequency before transmission over the antenna cable.
The same local oscillator reference plus a mixer then upshifts the 
carrier frequency again.
The local oscillator only needs relatively low short term phase noise 
and hig short term stability as long term (>> antenna cable delay) local 
oscillator phase errors cancel out.
Whilst this system has advantages in reducing the cable attenuation I'm 
not convinced it improves the phase shift stability when the fact that 
the local oscillator signal is transmitted up the cable is taken into 
account.
A relatively low frequency reference is transmitted up the cable, where 
a frequency multiplier or harmonic mixer is used.
The reason given for using a custom single channel GPS receiver (which 
uses no custom parts) is to ensure continuity of component supply for 
several decades.
This receiver periodically switches from one satellite to another.
When discipling a frequency standard carrier cycle slips arent as 
important (provided you can detect them) as when making position or 
equivalently time measurements.
With a fixed position receiver one can take advantage of the fact that 
the antenna's position is very stable, at least in the short term.
If one samples the carrier phase data at a high enough rate then a 
sequence of intervals is available when no carrier cycle slips have 
occurred.
The carrier phase differences over these time intervals can be used to 
estimate the local oscillator frequency error.
A multichannel receiver tracking several SVs should ensure that time 
intervals where cycle slips occur for all tracked SVs are relatively 
infrequent.

> It would be interesting to know how much each of these five
> pieces contributed to its overall performance. My hunch is
> cheap OEM timing receiver carrier-phase measurement
> alone is not enough.
>
> /tvb
>   
One could easily test that assertion by replacing the receiver's local 
oscillator with a source locked to a high stability source such as a 
hydrogen maser, logging the receiver carrier phase data and then later 
analysing it.

Bruce