Comparison of Cesium Fountain Clocks in Europe and Asia

Transcription

1 APMP/TCTF workshop 214,Daejeon, Korea Comparison of Cesium Fountain Clocks in Europe and Asia Aimin Zhang National Institute of Metrology(NIM) Sep.2,214

2 Outline Introduction Setup of PFS comparison Comparison data processing Numeric Results Uncertainty evaluation Conclusion 2

3 Introduction In recent years, the Cesium fountain clocks have been developed and used as primary frequency standards (PFSs) to realize the SI unit of time, the second. Nowadays the uncertainties of fountain clocks have reached the level of 1-15 ~ Data of some PFSs have been regularly used by BIPM to generate TAI. It is necessary to compare the PFSs to verify their accuracy and reproducibility. NIST PTB Multiple NMIs GPS NICT PTB Two-Way Satellite Time and Frequency Transfer () and GPS carrier phase (GPS CP) time transfer techniques have been proved to be able to compare remotely located PFSs with uncertainties in the order of

4 Introduction 4

5 Introduction Use of technique Time transfer Frequency transfer network Europe-USA Europe-Asia Asia/pacific rim link between Asia and Europe A link between Asia and Europe was established in 25 Since 21, the geostationary satellite AM2 has been used At present, the participating laboratories of this link include NICT, NIM, NPLI, NTSC, PTB, SU and TL 5

6 6

7 ground station at NIM 7

8 Introduction In May 213, six Cesium fountain clocks were compared fountains NMI Name CSF(NIM) NIM NIM5 CSF(NPLI) NPLI CsF1 CSF1(PTB) PTB CSF1 CSF2(PTB) PTB CSF2 CSF1(SU) VNIIFTRI CSF1 CSF2(SU) VNIIFTRI CSF2 and GPS CP technologies were used 8

9 Setup of FPS comparison GEO GPS NPLI NIM PTB VNIIFTRI-SU UTC(SU) CSF1 CSF1 UTC(PTB) H-maser CSF2 H-maser CSF2 9

10 GEO NPLI UTC(NPLI) UTC(NIM) NIM H-maser H-maser CsF1 NIM5 1

11 Setup of PFS comparison 5 MHz HM(k) FDA CSF(k) 1) : [UTC(i)-UTC(j)] 2) GPS RINEX : [UTC(i)-UTC(j)] UTC(k) 1 pps 5 MHz FDA Station 3) Phase difference between H- maser and local UTC: [UTC(k) - HM(k)] PDA 5 MHz GPS receiver Phase comparator 4) Frequency difference between fountain and H-maser: [CSF(k)- HM(k) ] The comparison setups at NIM, NPLI and PTB At VNIIFTRI, the reference of the GPS receiver was from the H-maser not the UTC(SU). The fountain clock comparisons were built on daily mean frequency differences between the fountain pairs. 11

12 Data processing Fountain measurement data measurement data [CSF(i)-CSF(j)] TW = [CSF(i)-HM(i)]-[CSF(j)-HM(j)] + [UTC(i)-UTC(j)] TW - [UTC(i)-HM(i)] +[UTC(j)-HM(j)] H-maser measurement data [CSF (i)-csf (j)] GPS = [CSF (i)-hm (i)]-[csf (j)-hm (j)] +[UTC(i)-UTC(j)] GPS -[UTC(i)-HM(i)]+[UTC(j)-HM(j)] VNIIFTRI-SU comparison by GPS CP GPS measurement data in Rinex file [CSF (SU)-CSF (j)] GPS = [CSF (SU)-HM (SU)]-[CSF (j)-hm (j)] + [HM (SU)-UTC (j)] GPS + [UTC (j)-hm (j)] all terms in brackets are frequency differences. 12

13 Data processing For, the measuring procedure, data processing and formats for exchange of data are standardized by Rec. ITU-R TF The operational use of two-way satellite time and frequency transfer employing PN codes The Europe-Asia measurements were performed once per hour five minutes for each pair of stations The measurements repeated during 12 hours every day between 11: and 23: UTC due to the limited transponder working time. To calculate daily frequency differences, we chose the phase measurement data between 17:-18: UTC (roughly in the middle of 11:-23:) for the comparison. y( ) x( t ) x( t) 13

14 Data processing In the same way, GPS CP results were obtained every five minutes and averaged over one hour The relative frequency differences per day were calculated using the time differences at the 17th hour from two consecutive days. y( ) x( t ) x( t) For GPS CP processing, all the RINEX files were concatenated and NRCan_PPP software with IGS files were used. 14

15 Data processing To compare with the and GPS CP measurement data, the frequency difference of fountain vs. maser and frequency difference of H-maser vs. UTC(k) were calculated from (x) 17:3 UTC to (x+1) 17:3 UTC. In this way, all the measurement data were processed nearly with the reference epoch from 17:3 UTC. 15

16 Results Experiment period:may 7 ~ May , PTB, VNIIFTRI-SU and NPLI: data during 21 days VNIIFTRI fountain CSF1 was not operational on May 16, 213, the missing daily value for that day was interpolated by averaging over the data before and after the missing day. NIM: data during 13 days 16

17 Frequency difference(1e-15) Frequency difference(1e-15) Frequency difference(1e-15) Frequency difference(1e-15) Frequency differences between the fountains and H Masers 1754 NIM5-H271 5 CSF(NPLI)-HM ,419 56,424 56,429 56, ,419 56,424 56,429 56,434 56, CSF(PTB)-H8 CSF2 CSF CSF(SU)-H81 CSF1 CSF ,419 56,424 56,429 56,434 56, /1/1 17

18 Frequency Difference Frequency Difference Comparison Results 2 x 1-14 CSF1(PTB)-NIM5 2 x 1-14 CSF2(PTB)-NIM5 1 GPS PPP 1 GPS PPP ,419 56,424 56,429 56, ,419 56,424 56,429 56,431 18

19 Frequency Difference Frequency Difference Frequency Difference Frequency Difference Comparison Results 2 x 1-14 CSF1(PTB)-CSF1(SU) 2 x 1-14 CSF1(PTB)-CSF2(SU) GPS PPP 1 GPS PPP ,419 56,424 56,429 56,434 56, ,419 56,424 56,429 56,434 56,439 2 x 1-14 CSF2(PTB)-CSF1(SU) 2 x 1-14 CSF2(PTB)-CSF2(SU) 1 GPS PPP 1 GPS PPP ,419 56,424 56,429 56,434 56, ,419 56,424 56,429 56,434 56,439 19

20 Frequency Difference Frequency Difference Frequency Difference Frequency Difference Comparison Results 2 x 1-14 CSF1(SU)-NIM5 2 x 1-14 CSF2(SU)-NIM5 1 GPS PPP 1 GPS PPP ,419 56,424 56,429 56, ,419 56,424 56,429 56,431 2 x 1-14 CSF1(PTB)-CSF(NPLI) 2 x 1-14 CSF2(PTB)-CSF(NPLI) 1 GPS PPP 1 GPS PPP ,419 56,424 56,429 56,434 56, ,419 56,424 56,429 56,434 56,439 2

21 Frequency Difference Frequency Difference Frequency Difference Comparison Results 2 x 1-14 CSF1(SU)-CSF(NPLI) 2 x 1-14 CSF2(SU)-CSF(NPLI) 1 GPS PPP 1 GPS PPP ,419 56,424 56,429 56,434 56, ,419 56,424 56,429 56,434 56,439 2 x 1-14 NIM5-CSF(NPLI) 1 GPS CP ,419 56,424 56,429 56,431 21

22 Comparison Results values in 1-15 Pair of fountains GPS CP PTB CSF1-NIM PTB CSF2-NIM.5.1 SU CSF1-NIM SU CSF2-NIM PTB CSF1-SU CSF PTB CSF1-SU CSF PTB CSF2-SU CSF PTB CSF2-SU CSF PTB CSF1-NPLI PTB CSF2-NPLI NIM-NPLI..5 SU CSF1-NPLI SU CSF2-NPLI For SU the difference between comparisons by GPS CP and was larger than other labs, which suggests that there was a local systematic difference between the two techniques which is possible due to its incorrect calibration of GPS 22

23 Uncertainty evaluation Uncertainty sources: Uncertainty due to the PFSs u A and u B, Uncertainty due to the time transfer links u L, Uncertainty due to the direct comparison system used in the phase difference measurement between H-maser and UTC(k) Compared with the other two sources, the uncertainty from the internal measurement system is in the 1-17 level and could be neglected. The combined uncertainty : 214/1/1 23

24 Uncertainty due to the PFS The uncertainty of the fountains in 1-15 was evaluated and provided by each lab. fountains u A (21 d) u A (13 d) u B CSF(NIM) CSF(NPLI) CSF1(PTB) CSF2(PTB) CSF1(SU) CSF2(SU) The operating time for nearly all the Cs fountain clocks is over 9%, the contribution of dead time could be ignored. T. E. Parker, COMPARING AND EVALUATING THE PERFOMANCE OF PRIMARY FREQUENCY STANDARDS: IMPACT OF DEAD TIME, IFCS 211,

25 Frequency difference MDEV Uncertainty due to the time transfer links Differencing the data obtained using and GPS CP eliminates the influences of the time references The modified Allan deviation (MDEV) of the double-differences reflects the combined link frequency transfer uncertainty of and GPS CP. One example: 8 x SU-NPLI 1-14 FREQUENCY STABILITY(SU-NPLI) Tau Sigma e e e ,419 56,424 56,429 56,434 56,439 (a)double frequency difference Averaging Time, Days (b)frequency instability If we assume that the instabilities of and GPS CP are independent and that they contribute equally to the combined uncertainty, then the 1-day uncertainty of the link for each of the two techniques u A is calculated from the MDEV (1d) divided by square root of 2. T. Parker, P. Hetzel, S. Jefferts, First comparison of remote Caesium fountains, Proc. IFCS,

26 Uncertainty due to the time transfer links The 1-day uncertainty is extended to the relative link uncertainty u L for days according to the equation: u L : link uncertainty, u A : time transfer uncertainty in seconds, τ: the interval in days, for links with NIM, =13d, for other links, =21d, τ : the effective interval in days over which the u A value is determined. It was suggested by Tom Parker and approved in the 17 th CCTF meeting, with the assumption that the link noise was mainly from white PM and flicker PM. T. E. Parker, Report to the 17th Session of the CCTF, Conference report, CCTF/6-13,

27 Uncertainty due to the time transfer links values in 1-15 Labs Mod y (1d) u L PTB-SU PTB-NPLI SU-NPLI PTB-NIM NPLI-NIM SU-NIM

28 Results and Combined uncertainty values in 1-15 Pair of fountains GPS CP u PTB CSF1-NIM PTB CSF2-NIM SU CSF1-NIM SU CSF2-NIM PTB CSF1-SU CSF PTB CSF1-SU CSF PTB CSF2-SU CSF PTB CSF2-SU CSF PTB CSF1-NPLI PTB CSF2-NPLI NIM-NPLI SU CSF1-NPLI SU CSF2-NPLI

29 Conclusions The remote comparison among multiple PFSs could be implemented by and GPS CP techniques. In this campaign, the relative frequency differences among all the six PFSs are within 1-sigma(1 ) of the corresponding combined standard uncertainty in low 1-15 level. 29

30 Thank you! 谢谢! 3