[time-nuts] GPS and national standards in frequency

[Bruce Griffiths]

Marco A. Ferra wrote:
> Thanks Mike, I'll give it a close look.
>
> Sincere regards,
> Marco 
>
> -----Original Message-----
> From: time-nuts-bounces at febo.com [mailto:time-nuts-bounces at febo.com] On Behalf Of Mike S
> Sent: terça-feira, 25 de Março de 2008 1:53
> To: Discussion of precise time and frequency measurement
> Subject: Re: [time-nuts] GPS and national standards in frequency
>
> You might want to read this: http://tf.nist.gov/service/gpscal.htm
>
> At 09:27 PM 3/24/2008, Marco A. Ferra wrote...
>   
>> Dear time-nuts
>>
>> My name is Marco Ferra.  I'm doing research on an electrical metrology 
>> laboratory in Portugal, and I'm conducting a study on how to establish 
>> tracebility to frequency using one (or more) GPSDO's.  I came across 
>> febo.com's site and this mailing list after searching the net.  Let me 
>> tell you what I'm doing.
>>
>> I have put side-by-side two GPS receivers, one HP58503 (now owned by 
>> Symmetricom I believe) and a Fluke 910R, both with it's own antenna.
>> Both antennas are separated by about 100 meters.  I'm tracking the 10 
>> MHz output of both GPS receivers with a Fluke PM6681R Counter.  I'm 
>> using the internal Rubidium oscillator of the counter and using 
>> channels A and B for the measurements.  Each 10 seconds, I'm recording 
>> the frequency on channel A, B, and the phase (in degrees) between the 
>> two signals.  I have resolution on the phase by 1/10th of degree.
>>
>> With the phase, I can get the time differente in nano-seconds between 
>> the two signals:
>>
>> =phase(degress)/(360*10000000)*10^9
>>
>> I'm putting 10000000 Hz has a nominal frequency for the calculation, 
>> I'm not sure if should use the frequency recorded on channel A or B.
>>
>> Then I calculate the time difference, the TIE (ns), the Delta Tie (ns) 
>> accumulative, and then the Delta f (it's equal to Delta TIE (ACC) / 
>> measurement time (could be 10, 20, 30 secs)).  Please take a grasp at 
>> page 4 of 
>> http://www.pendulum-instruments.com/Assets/download/accurate_calibratio
>> n
>> _article.pdf
>>
>> Given these two GPSDO's, and given the very small Delta f, I can assume 
>> that both GPSDO's are functioning proprely (could I indeed?).  But I'm 
>> not sure how I can get tracebility to national standards, to NIST, for 
>> example, on the 10 MHz output frequency.  Perhaps it's difficult, but I 
>> need a solid proof that my 10 MHz output is traceble to a primary 
>> laboratory.
>>
>> I have seen that you all have a vast experience in this area, and 
>> perhaps we could work together to gather good results and achive this 
>> goal.
>>
>> Sincere regards, and thank you for reading, Marco
>>
>>     
You will do much better if you divide the frequencies to be calibrated 
down to 1Hz and time stamp the 1Hz signal transitions and the leading 
edge of the leading edge of the PPS output of a modern GPS timing 
receiver. Record the timestamps along with the associated GPS receiver 
sawtooth corrections. This is what NIST does in their remote frequency 
measurement systems. The recorded data is analysed along with NISTs  own 
measurements using a similar setup driven by one of their caesium 
clocks. A time stamp resolution better than 1ns is necessary to avoid 
significantly degrading the performance achievable with a modern timing 
receiver.

Using an indirect method such as employing a GPSDO (particularly one 
using a legacy timing receiver) just adds another layer of uncertainty 
to the traceability chain.

Even better resolution is possible by making GPS carrier phase 
measurements, however this requires that the GPS receiver local 
oscillators etc, be locked to the frequency source being characterised.

Since neither NIST nor your local NMI control the GPS system 
traceability using GPS requires that both you and your NMI make such 
measurements.
If you are not too far away from your NMI then both you and your NMI 
will experience essentially the same GPS timing fluctuations due to 
common mode effects such as GPS SV transmission fluctuations and 
ionospheric delays.

Bruce

Bruce