[time-nuts] Re: GPS receiver with reference clock input

[Bob Camp]

Hi

The first plot “wraps” at 1ms. That’s a pretty normal thing for one of these gizmos. It re-syncs
to the nearest GPS code frame tick ( likely the wrong term ….).  

Total drift is about 4.6 ms. Over a 9 hour period, that puts the internal clock at about 0.17 ppm
off frequency. It’s a reasonable outcome for the TCXO in these devices. 

On the second plot the key is the hard to spot “1e8” in the upper left corner of that plot. I’ve
never seen it on my plots so I can only guess that it’s a scale factor. There are numeric data
files in the package they send. You can confirm what it means from them. 

If the data is in nanoseconds x 1x10^8. The clock moves 0.33 seconds over the ~15 hour period. 
This puts you around 5.7 ppm off frequency. 

If the data is truly in nanoseconds, then the clock is off by 5.7 x 10^-14. At that level, the blue line 
is nowhere near that smooth. 

Bob



> On Jun 10, 2023, at 5:21 AM, HÃ¥kan T Johansson via time-nuts <time-nuts at lists.febo.com> wrote:
> 
> 
> Dear Keelan,
> 
> On Sat, 10 Jun 2023, Keelan Lightfoot via time-nuts wrote:
> 
>> Bob,
>> 
>> I did that experiment a while back. Running a static survey in my back yard
>> with the receiver exposed to the sun over the course of the day, the
>> internal clock offset was quite a mess, with 4 milliseconds of drift over 9
>> hours:
>> 
>> http://beefchicken.com/dump/clockoffsetbad.png
> 
> I think what is plotted is microseconds, not nanoseconds.  Since it looks like it wraps at 1000000.
> 
> If so, the blue curve is just a free-drifting clock?
> (4 second total drift)
> 
> Best regards,
> HÃ¥kan
> 
> 
> 
>> 
>> Once I acquired the receiver with the reference input, I used a fairly
>> recently calibrated HP 5316A with the OCXO option as a clock source, and
>> kept the receiver indoors in my underground lair where the temperature is
>> fairly constant. The results were much better, with -3 ns of drift over a
>> 16 hour survey.
>> 
>> http://beefchicken.com/dump/clockoffsetgood.png
>> 
>> The results are interesting, but in terms of the static survey solution,
>> (and to be painfully reductive), the terrible clock seems to have had
>> little impact on the final survey, with both surveys having identical error
>> ellipses. And the internally clocked survey had the odds stacked against
>> it; it was a shorter survey, and there were a number of trees blocking a
>> clear view of the sky.
>> 
>> What are the other applications? These receivers were specifically marketed
>> as CORS receivers, I imagine having a reference station running for years
>> with a clock with milliseconds of drift per day might have some negative
>> consequences...
>> 
>> - Keelan
>> 
>> On Fri, Jun 9, 2023 at 5:07 PM Bob Camp <kb8tq at n1k.org> wrote:
>> 
>>> Hi
>>> 
>>> Simple answer:
>>> 
>>> Feed the 10 MHz out of your Rb (or whatever). Send the data off to NRCan
>>> (or wherever). Get
>>> back a plot that lets you know how many parts in 10^-15 your standard is
>>> off frequency.
>>> 
>>> There are other applications. The more general answer is: to remove local
>>> clock issues from
>>> the solution.
>>> 
>>> Bob
>>> 
>>>> On Jun 9, 2023, at 7:37 PM, Keelan Lightfoot via time-nuts <
>>> time-nuts at lists.febo.com> wrote:
>>>> 
>>>> In my pile of Trimble GPS receivers, I have a small handful of 4000SSi
>>> CORS
>>>> receivers with a BNC clock reference input, and in the receiver firmware
>>> I
>>>> can switch the receiver over to the external 10 MHz clock. It's all
>>>> documented in the manual, but what I don't understand is... why?
>>>> 
>>>> I'm trying to reconcile my mental model of the GPS receiver, and in my
>>>> model, any critical timing in the receiver is all relative to the
>>> received
>>>> GPS signals, what would I gain from feeding the 10 MHz from a GPSDO other
>>>> than the more stable OCXO in the GPSDO? I imagine it might improve jitter
>>>> on the 1PPS signal...
>>>> 
>>>> Thanks,
>>>> 
>>>> Keelan
>>>> _______________________________________________
>>>> time-nuts mailing list -- time-nuts at lists.febo.com
>>>> To unsubscribe send an email to time-nuts-leave at lists.febo.com
>>> 
>>> 
>> _______________________________________________
>> time-nuts mailing list -- time-nuts at lists.febo.com
>> To unsubscribe send an email to time-nuts-leave at lists.febo.com
> _______________________________________________
> time-nuts mailing list -- time-nuts at lists.febo.com
> To unsubscribe send an email to time-nuts-leave at lists.febo.com