[time-nuts] carrier phase tracking GPS receiver

[Dr Bruce Griffiths]

Didier
Didier Juges wrote:
> Hi Bruce,
>
> Dr Bruce Griffiths wrote:
>
>   
>> Bon soir Didier
>>
>> You are confusing the disciplining precision of the oscillator with 
>> measurement of its short term stability via the GPS receiver in this case.
>>
>> Bruce
>>
>>   
>>     
> Bon après midi to you, I believe, and if my Clock program is any good, 
> you just had lunch :-) Do you have Daylight Savings Time?
>
>   
We are currently observing 1 hour daylight saving so I had lunch about 2 
hours ago if I remember correctly.
> Yes, I realize that I am very confused, but at the end of the day, the 
> two have to come together. It does me little good to know how much off 
> my oscillator was, some time in the past. That information has to be 
> available in time to steer the oscillator, in real time, because what I 
> am looking for is a precise oscillator, not a history lesson (like the 
> BIPM view of UTC, where it takes a month for the paperwork to be 
> processed so that everyone knowns by how much they were off last month :-)
>
> You said:
>
> Typically with a good local oscillator you can do even better than that, 
> around 1E-11 in 1 sec is achievable and has been achieved.
> Even with the on board TCXO typically 3E-11 or so in 1s is achieved.
>
>
> Is this 3E-11 1) measured on the 1 PPS output, 2) measured on the TCXO, 
> or 3) the interpretation, by external computing of the carrier phase 
> data, of the quality of the timing prediction from the GPS receiver?
>
> If it is 3), how can I use that information to make a 3E-11 @ 1sec (or 
> anything near that) stable GPSDO?
>
> I think I am beginning to understand, but it's slow coming...
>
>
> Didier
>
> _______________________________________________
> time-nuts mailing list
> time-nuts at febo.com
> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>
>   

The PPS output doesn't exhibit this stability as its position is 
quantised to the nearest timing clock transition.

The 3E-11 is inferred from the SV carrier phase tracking data from a set 
of receivers used as part of a geodetic survey network.
This is mainly due to the TCXO short term instability.
Note that it is significantly better than the spec for the Rakon TCXO used.

You need to make your own ionospheric delay corrections using the the 
code and carrier phase pseudo range variations or from an equivalent 
real time source
You also need to ensure that the effect of the SV motion is accurately 
corrected.

Once you have corrected for the SV motion, the ionospheric delay and a 
host of other small but significant effects, then the changes in 
measured carrier phase are due to receiver local oscillator offset, 
which is why the TCXO must be replaced by a 10MHz signal derived from 
the OCXO you wish to discipline.

There are a host of other things to consider such as cycle slipping in 
the carrier phase tracking loops - the receiver will detect these.

The carrier phase (actually carrier beat phase) measurements are just 
like the output of an extremely high resolution (< 1ps) phase detector 
and are used in the same way to control the OCXO EFC DAC or offset DDS 
after suitable statistical filtering to keep the OCXO on track.

In fact it is possible to discipline the OCXO frequency so that it has a 
known offset from its nominal frequency. This is useful with older OCXOs 
which have drifted out of their adjustment range. A frequency standard 
with a fixed known offset is just as useful as a frequency standard with 
zero offset. A DDS based narrow range offset generator can always be 
used to produce a low phase noise signal with an offset of 1E-14 or less 
even though the OCXO may have an offset as much as 1E-5.

The first step is to log carrier phase and code range data from the 
receiver using its standard TCXO, then analyse the data so that you can 
see/understand what is going on.

Then use the 10MHz signal derived from your TCXO as the receiver local 
oscillator, again log the carrier phase and code range data to check 
that the performance is indeed improved.

Once you understand what's happening you can experiment with satellite 
orbit modelling, ionospheric corrections etc using the logged data to 
get your algorithms correct.

Then try closing the loop in real time. Remember you will need a high 
resolution (24bit optically isolated) DAC with good short term stability 
(low noise and low tempco) to adjust the OCXO EFC voltage. You will also 
need a processor with adequate performance, not a PIC, to execute the 
algorithms.

The final result is that you will then know your antenna phase centre 
location to within a few cm and your OCXO will have an Allan deviation 
of 1E-11 or less for Tau from 1 sec to 1 day. Even better performance is 
possible with a really good OCXO (FTS1200, FTS1000, OSA8607 etc.). See 
Quartzlock A8-MX GPS-BVA datasheet for some idea of what's possible if 
you have a good enough OCXO and GPS receiver coupled with good algorithms.

Bruce