1. Various activity in NICT Space-Time Standards Laboratory including optical clocks and applications
Responsible for Japan Standard Time
Space-Time Standards Laboratory
Applied Electro-magnetic Research Institute
National Institute of Information and Communications Technology (NICT)
Tetsuya Ido

2. 80 ns / 0.6 year = 4 × 10-15

3. JST Kobe sub-station started since Jun. 10, 2018
Distributed generation of JST
Kobe sub-station is scheduled to begin time-keeping in June 2018
2 H-maser & 5 Cs clocks
Primary purpose is a backup of Koganei HQ against disasters
(JST has never stopped in more than 40 years) HM
Cs (5071A)
DT (Koganei-Kobe) (ns)
SD=2.0ns
SD=6.7ns
Time (day)
Operation mode:
Copy of Koganei HQ
Independent operation
JST as an ensemble of all clocks operated in 4 stations
・JSTも、セシウム原子時計の合成から作る原子時を元にしていますが、
・UTCとは違って、JSTはリアルタイムに連続に時を刻む実時計を持っています。
LF station 1
LF station 2

4. Statistics in services
Total wave clocks
　～ 50 (million)
　～ 63
　～ 78
　～ 93
　～ 107
Now almost one clock per one
Japanese citizen
Number of NTP access per day
　～ 170 million
　～ 300 million 　～ 1.57 billion
　～ 1.33 billion
Telephone-JJY
Access per month
140k→
100k→
80k→
Telephon-JJY
　～ 80k
　～ 100k/mon
　～ 140k/mon

5. Time scale steered by an optical clock

6. JST and International Atomic Time
NICT
institute k
BIPM
EAL (ensemble)
UTC(k)
H-Maser
>400 clocks
Local Ensemble
adjustment
TAI
Cs (or Rb) fountains
18 clocks
Phase Micro Stepper
adjustment
leap second
adjustment
UTC(NICT)
UTC UT
+ 9h
Interval of UTC-UTC(k)
is 5 DAYS in Circular T (BIPM monthly report)
Generating a real signal in institute k
Japan Standard Time (JST)

7. Optical clocks utilized for time scale
NICT
institute k
BIPM
EAL
UTC(k)
H-Maser
>400 clocks
adjustment (~1e-13)
TAI
Cs (or Rb) fountains
PMS
adjustment (~1e-15)
adjustment
leap second
UTC(NICT)
UTC UT
Japan Standard Time (JST)
PMS: Phase micro stepper

8. CIPM# (2015): (2)Hz

9. INTERMITTENT Evaluation of a HM for 6 monthsby Sr and UTC(NICT)
1×10-13
104 s measurement

10. Steering Linear drift estimation interval: T=25days
Number of OFS operation in T : N+1 >4
(once per week or more frequently)
One HM free evolution time: DT = T/N

11. Comparison against UTC & TT(BIPM16)
TA(Sr)-TT(BIPM16)
Clearly detect the frequency offset of UTC
Phase difference against TT(BIPM16) < 1ns after 5months
Stability 20 days
TA(Sr)-UTC
TT(BIPM16)-UTC

12. Supplying frequency standard @ 2um for Lidar experiment
Two laboratories (standard lab. & lidar lab.) don’t need to be physically close.

13. All-optical direct comparison between NICT & UT clock
Google map
NICT
UT (Hongo)
45 km
12 km
Air part
Otemachi
All-optical direct comparison between NICT & UT clock
Beat measurement
769 nm
698 nm
87Sr
56 m
NICT UT
24 km ×2
Optical
Carrier transfer
1538 nm laser
57km

14. Frequency difference & stability between distant Sr clocks
A Hz-level frequency difference is clearly visible over the time scale of minutes
nNICT - nUT (Hz)
Offset: predominantly
due to differential gravity shift
Obtained instability
consistent
5×10-16
Dick-effect-limited instability
UT :
NICT :

15. Requirement of frequency standard for LIDAR
Wavelength: /- 10 nm (CO2, R30 line)
Spectral width: 100kHz
Frequency drift: <10kHz/hour
Minimum operation time: 10 hours
Not difficult in Standard laboratory. But need to be transferred to the LIDAR site.
Option A: transfer 1.5um CW, and converts to 2um at remote site
Frequency comb operated in remote site
Option B: transfer 2um.
Large loss limits the distance of transfer
Option C: transfer 1.5um pulse, and optical spectrum broadened in remote site
Frequency comb unnecessary in remote site
Our choice

17. No frequency drift when seed laser is stabilized to comb.
Result
Slave Q-switch laser
Tm. Ho:YLF based
Pulse energy: 25mJ
Pulse width:150ns
Rep. rate: 30Hz
No frequency drift when seed laser is stabilized to comb.
Furthermore, it is JST traceable.
S. Nagano et al., Opt. Express 4, (2016)

18. Members Space-Time Standards lab. Remote-sensing lab. S. Ishii Sr
M. Aoki
R. Ohtsuka
K. Mizutani
S. Ochiai Sr
H. Hachisu
2mm transfer
S. Nagano
Time Scale
F. Nakagawa
Y. Hanado
JST operation team
Fiber transfer
M. Kumagai
Y. Li
Technical assistance
H. Ishijima
M. Mizuno
S. Ito

20. International comparison Analog line Optical line
　Dissemination service of Japan Standard Time
JST Generation and measurement system
satellite
H maser
PFS
calibration
International comparison
Optical frequency standards
JST
Cs clocks
Radio wave(JJY)
Telephone dissemination
NTP
service
Provision to time stamp service
Calibration
service
TAA
・次は供給サービスについてです。
TSA
Analog line　Optical line
Via GPS
internet
Time stamp
UTC 12:10:35
NICT
Carry-in
（5, 10MHz）
電波時計用
中継器

21. Advantage of “optical” steering
1 Operation
104 s of operation is sufficient to evaluate the scale unit of a HM at mid level
Short term fluctuation of HM may be compensated by an optical clock.

22. Optically steered time scale
TA(Sr)
No external reference
Reference to compare with: UTC, TT(BIPM16)

23. Conclusion
Realization of a real signal of “optically steered” time scale
Synchronization to the most precise time scale TT(BIPM16) in sub-ns level in 5 moths
Stability 20 days, same level as TAI using only one Sr and one HM
Simple analytic formula to estimate the quality of intermittently steered time scale
Seamless TAI calibration for six months, consistent with other PFS.

24. Steering by intermittent operation of Sr lattice clock
HM frequency and drift rate calibrated by Sr
Adjustment of PMS offset frequency every 4 hours
No servo to reduce the time offset UTC-T_Sr
Based on our frequency* obtained in 2015
Source osc.
(HM)
Phase micro stepper (PMS)
f(HM)
based on Sr
T_Sr
Intermittent operation more than once a week for 104 s continued
* Hz (Hachisu et al., Appl. Phys. B 34, 123 (2017))
… Hz (CIPM2015) →　… Hz (CCTF WGFS, Jun 2017)

25. Potential provision to Univ. in Kansai
TWCP for optical standards
1st in Univ. Osaka –NICT in 2014
IEEE Trans. Ultra. Ferr. Freq. Control 63, 2231 (2016).
2nd in TWCP NICT-Korea Lattice CLK comparison in 2017
Possible benefit of Kobe sub-station
Optical fiber (NICT(Kobe)- Osaka-Kyoto in future?)
Two-way carrier phase (TWCP) [+portable HM]
１）<1e-16 accuracy confirmed
With GPS Integer-PPP technique
Stability
Commercial transceiver hopefully available in late 2018
2）Sr(NICT)-Yb(KRISS) direct comparison using TWCP
mobile
In a lab.
Portable system physically carry-in to NICT-Kobe
GPS one carrier freq.
(< 1million yen)
TWCP
NICT-KRISS arxiv (2017)

26. Simulation in worse HM & infrequent OFS operation
HM2 (dash) noisier than HM1 (solid) (HM2 Flicker floor: 5e-16)
Blue: once per week, orange and green: once per two weeks (odd, even)

27. Evaluation of TAI scale interval by Sr
Effects
Uncertainty (10-17)
Sr systematic 6
Gravity
2.2
Hydrogen maser
deterministic 25
stochastic (dead time) 18
Phase measurement 5
UTC-UTC(NICT) link
20 (30 days average)
Sr frequency
9 (SD of mean in 14 measurements over 10 years)
(40, secondary representation of
the second)
Total
36 (53)
y(HM)
7 days
1 month t

28. Result: Evaluation of TAI scale interval by Sr
Sr standard frequency: Hz
(weighted average of 14 measurements performed in various lab over 10 years)
i = SYRTE-FO2, PTB-CSF2, NICT-Sr1
j = month in 2016
SYRTE-FO2
PTB-CsF2
NICT-Sr1
𝑑 𝑖,𝑗 − 𝑑 𝑗 /2 [10-16]
6.3
7.8
2.6
𝑑 𝑖,𝑗 − 𝑑 TT BIPM16 ,𝑗 /2 [10-16]
2.3
2.8
Mean of
3 evaluations in month j
TT(BIPM16) is calculated by using all PFS data including SYRTE-FO2 and PTB-CsF2.
So, it is a natural result that their value are close to TT(BIPM16).

29. Requirement for HM and OFS operation rate
𝐸 Δ𝜙 = 𝜀 𝑝 2 + 𝜀 𝐹 = 𝑇 𝑁 2𝑁+1 2𝑁+3 𝑁 𝑁+1 𝑁+2 𝜎 𝑝 ln 2 𝜎 𝐹
Linear drift estimation error
(from LSF)
Stochastic phase excursion
in Flicker noise (*)
𝜎 𝑝 =4𝐸−16
104 s)
𝜎 𝐹 =3𝐸−16
Flicker floor of Hadamard dev.
* D. Allan, IEEE Trans. Ultrasonic. Ferro. Freq. Cont., (1987)

30. Time scale NICT institute k BIPM EAL (ensemble) TAI UTC UTC(k)
H-Maser
EAL (ensemble)
UTC(k)
>400 clocks
Local Ensemble
adjustment
TAI
Cs (or Rb) fountains
18 clocks
PMS
adjustment
adjustment
leap second
UTC(NICT)
UTC UT
+ 9h
Interval of UTC-UTC(k)
is 5 DAYS in Circular T (BIPM monthly report)
Generating a real signal in institute k
Japan Standard Time (JST)

31. Dissemination by telephone line
number of accesses per month
Telephone JJY
Online traffic
NICT
Modem
Code Generation
Modem
Broadcasting
station, etc
Telephone JJY
・ number: (042)
300/1200/2400 bps, 8 bit, non-parity, SJIS
ID (Name?): TJJY
Time delay measurement by loop back

32. R & D on the way, and International activity
Distributed generation of JST
Kobe sub-station is scheduled to be ready in June 2018 with
2 H-maser & 5 Cs clocks
Firstly, as a backup of Koganei HQ
But also in Future:
JST as total ensemble of clocks in Koganei-HQ, Kobe, two radio-wave clock broadcasting stations
Activity for Asia-Pacific region
Mutual recognition agreement
Help GPS calibration in other institutes
APMP
International contribution
Contribution for the computation of TAI
・JSTも、セシウム原子時計の合成から作る原子時を元にしていますが、
・UTCとは違って、JSTはリアルタイムに連続に時を刻む実時計を持っています。

33. Estimation of linear drift rate
y(HM)
How long?
The frequency drift of the HM is estimated from recent calibration by Sr.
How long should we use for the linear fitting? t
3 sample variance is not affected by linear frequency drift.
The duration of the minimum 3 sample variance would be the best to obtain the drift rate
3-sample variance　（Hadamard variance)
1 6 𝑀−2 𝑖=1 𝑀−2 𝑦 𝑖+2 −2𝑦 𝑖+1 +𝑦(𝑖) 2
= 1 6 𝑁−3 𝜏2 𝑁=1 𝑁−3 𝑥 𝑖+3 −3𝑥 𝑖+2 +3𝑥 𝑖+1 −𝑥 𝑖 𝑁=1 𝑁−3 𝑥 𝑖+3 −3𝑥 𝑖+2 +3𝑥 𝑖+1 −𝑥 𝑖

34. Comparison between UTC and a real-time signal of TA(Sr)
TA(Sr): Sr-based time scale
Frequency difference
6.4 x 10-16
From Circular T 345

35. TAI (UTC) evaluation by Sr clock Japan Standard Time (JST) system
UTC(k)
Cs ensemble
BIPM
UTC(NICT) HM
100 MHz
measurement
Comparison Df
Monthly
Yearly
100 MHz
Circular-T
- Frequency differences -
・f(UTC-UTC(NICT))
in Circular T
UTC-UTC(k) 5 day interval
UTC-TT(BIPMyy) 10 day interval
・f(UTC(NICT)-HM)
in JST system
・f(HM-Sr)
in Sr operation
f(UTC-Sr)

36. Comparison between TT(BIPM) and TA(Sr)
| TA(Sr)-TT(BIPM) | < ±1.5 ns
Instability of TA(Sr)-TT(BIPM)
20 days
UTC-UTC(k) link dominant
@5, 10 days
Sr clocks able to evaluate the TAI scale (as PFS worldwide do) even in the intermittent operation
→D2L-A, Thursday in last session of Opt,. clock
TA(Sr) clearly detects that
UTC does not realize the SI second.
For 5 months, frequency standards in level were kept generated.

37. Summary
Generation of real time-scale signal for a half year with reference to an optical clock, which is operated intermittently once in a week.
Capability to provide frequency standards in level in 24 hours 7 days a week

38. Estimating the scale interval of TAI with an optical clock
EFTF-IFCS 2017
10th-13th July 2017
The Micropolis Convention Center, Besançon, France
National Institute of Information and Communications Technology (NICT)
Hidekazu Hachisu & Tetsuya Ido

39. Optical clocks utilized for time scale
T. Ido et. al., in A2L-B 1077
NICT
institute k
BIPM
EAL
UTC(k)
H-Maser
>400 clocks
adjustment
Cs (or Rb) fountains
TAI
PMS
adjustment
adjustment
leap second
UTC(NICT)
UTC UT
Japan Standard Time (JST)
PMS: Phase micro stepper

40. JST : Time scale generation
Clocks for generating UTC(NICT) :
Cs 5071A : 18 (ensemble timescale)
Anritsu H-Masers : 1(source) + 2(backup)
The behavior of UTC(NICT) :
| UTC – UTC(NICT) | < 20 ns.
Stability ~ 10-30d.
・Let’s move on each topic.
・This blue line shows the UTC-UTC(NICT), and red points show the additional frequency adjusting to trace UTC.
・frequency stability at month is almost 2e-15.
Accuracy
conservative:　5e-14
(employed for calibration service)
Standard deviation:　<4e-14
24hours, 7days a week

41. Optical clocks utilized for time scale
T. Ido et. al., in A2L-B 1077
NICT
institute k
BIPM
EAL
UTC(k)
H-Maser
>400 clocks
adjustment
Cs (or Rb) fountains
TAI
PMS
adjustment
adjustment
leap second
TA(Sr)
UTC UT
Japan Standard Time (JST)
PMS: Phase micro stepper

42. Comparison between UTC and a real-time signal of TA(Sr)
T. Ido et. al., in A2L-B 1077
TA(Sr): Sr-based time scale
0.74ns
TT(BIPM16): a time scale post-processed
incorporating all data from
Primary Frequency Standards
(PFSs) by 2016
The averaged frequency difference
0.74 (ns) / 5 (months) = 5.7 × 10-17
Ref.
Current sophisticated UTC(k)s are adjusted to UTC;
・UTC-UTC(PTB): <±5 ns
・UTC-UTC(OP): <±5 ns
Possibility to generate UTC(NICT) on the same level as these UTC(k)s
This is another evidence that supports our absolute frequency measured in 2015

43. Optical clocks utilized for time scale
T. Tdo et. al., in A2L-B 1077
NICT
institute k
BIPM
EAL
UTC(k)
H-Maser
>400 clocks
adjustment
Cs (or Rb) fountains
TAI
This work
PMS
adjustment
adjustment
leap second
TA(Sr)
UTC UT
Japan Standard Time (JST)
PMS: Phase micro stepper

44. [2] H. Hachisu and T. Ido, Jpn. J. Appl. Phys. 54, 112401 (2015).
TAI calibration 30
Time (day)
Fractional frequency 1
yHM
yUTC(k)
yPFS
yTAI
Continuous operation
Cs (Rb) fountain clock
Ex. 10days[1]
Optical clock 30
Time (day)
Fractional frequency 1
yHM
yUTC(k)
yOptical
yTAI
[1] SYRTE, Metrologia 53, 1123 (2016).
Need to operate for all through a month to reduce the satellite link uncertainty
Hard to evaluate for all through a month and continue to do every month
Intermittent operation [2] 30
Time (day)
Fractional frequency 1
yHM
yUTC(NICT)
ySr
yTAI
Optical clock (Our scheme)
104 s per one operation once a week
Five operations taken into account
to obtain one-month mean
[2] H. Hachisu and T. Ido, Jpn. J. Appl. Phys. 54, (2015).

45. [2] H. Hachisu and T. Ido, Jpn. J. Appl. Phys. 54, 112401 (2015).
TAI calibration
Available to evaluate a scale interval of TAI all through a month as PFS regularly does
Advantages
Once per week makes it easier !
During a long non-operation time (dead time) of clock operation;
・Capable to evaluate the systematics
・Possible to improve the system and use it
for other experiments
Intermittent operation [2] 30
Time (day)
Fractional frequency 1
yHM
yUTC(NICT)
ySr
yTAI
Optical clock (Our scheme)
104 s per one operation once a week
Five operations taken into account
to obtain one-month mean
[2] H. Hachisu and T. Ido, Jpn. J. Appl. Phys. 54, (2015).

46. Absolute frequency of 87Sr lattice clock
CIPM recommended # (2015): (2)Hz, Uncertainty: 5×10-16
->CCTF recommended # (2017): (2)Hz, Uncertainty: 4×10-16
Weighted mean of 5 data points
(25)Hz
Standard error: 6×10-17
[*]
CIPM frequency was somewhat too high…
We employed our result of the measurement in 2015, which was 5×10-16 lower.
Accuracy of 5.7× [*]
[*] H. Hachisu, et al., Opt. Express 25, 8511 (2017)
Accurate enough as a reference for evaluating scale interval of TAI with a level of 10-16

47. TAI (UTC) evaluation by Sr clock Japan Standard Time (JST) system
UTC(k)
Cs ensemble
BIPM
UTC(NICT) HM
100 MHz
measurement
Comparison Df
Monthly
Yearly
100 MHz
Circular-T
- Frequency differences -
・f(UTC-UTC(NICT))
in Circular T
UTC-UTC(k) 5 day interval
UTC-TT(BIPMyy) 10 day interval
・f(UTC(NICT)-HM)
in JST system
・f(HM-Sr)
in Sr operation
f(UTC-Sr)

48. Evaluation of TAI scale interval by Sr
Effects
Uncertainty (10-17)
Sr systematic 6
Gravity
2.2
Hydrogen maser
deterministic 20
stochastic (dead time)
18 (*)
Phase measurement
4.5
UTC-UTC(NICT) link
20 (30 days average)
Sr frequency
6 (using recent results of
SYRTE, PTB x2, NICTx2)
(40, secondary representation of
the second)
Total
35 (53)
y(HM)
7 days
1 month t
(*)Optimum prediction uncertainty:
τ· 𝜎 flicker ln2
D. W. Allan, IEEE UFFC 34, 647 (1987)
86400×7×3× 10 −16 ln2
Dead time per 7days: = 0.22 ns
Dead time per month: 4 times 7days dead time in a month (30 days)
0.22 ns × /(86400×30) = 1.8×10-16

49. Result: Evaluation of TAI scale interval by Sr
BIPM’s reports
Thin error bar: aggressive evaluation
Thick error bar: secondary representation of second
i = SYRTE-FO2, PTB-CSF2, NICT-Sr1
j = month in 2016
SYRTE-FO2
PTB-CsF2
NICT-Sr1
𝑑 𝑖,𝑗 − 𝑑 𝑗 /2 [10-16]
7.2
5.8
3.1
𝑑 𝑖,𝑗 − 𝑑 TT BIPM16 ,𝑗 /2 [10-16]
2.3
2.8
3.6
Mean of
3 evaluations in month j
TT(BIPM16) is calculated by using all PFS data including SYRTE-FO2 and PTB-CsF2.
So, it is a natural result that their value are close to TT(BIPM16).

50. Absolute frequency measurement of 87Sr with TAI link
Advantage:
The measurement uncertainty is not restricted by uB of the local PFS [1, 2]
[1] H. Hachisu, et. al., Appl. Phys. B 123, 34 (2017)
[2] H. Hachisu, et al., Opt. Express 25, 8511 (2017)

51. Absolute frequency measurement of 87Sr with TAI link
Advantage:
The measurement uncertainty is not restricted by uB of the local PFS [1, 2]
[1] H. Hachisu, et. al., Appl. Phys. B 123, 34 (2017)
[2] H. Hachisu, et al., Opt. Express 25, 8511 (2017)
July in 2016
(x10-17)
half year
Sr systematic
6.3
6.6
Gravity
2.2
Hydrogen maser
deterministic 25 12
stochastic (dead time) 18
7.3
Link
UTC-UTC(NICT) link 20
7.9
Phase measurement
4.5
1.8
UTC-SI second
systematic (uB) 14 15
other parts
5.6
Total 42 23
uB of each PFS is not correlated

52. Absolute frequency of 87Sr lattice clock
CIPM recommended # (2015): (2)Hz, uncertainty: 5x10-16
CCTF recommended # (2017): (2)Hz, uncertainty: 4x10-16

53. Summary
Evaluation of TAI (UTC) scale interval by an optical clock all through a half year using intermittent measurements Feasible to monitor the scale interval of TAI as regularly as Cs & Rb fountains Evaluation of 87Sr frequency with TAI link (using multiple PFSs) through 6 months to be (10) Hz, which is one of the most precise optical absolute frequencies reported so far
We thank Dr. Petit for providing TAI calibration data including late reports from institute k.

54. Time scale steered by an optical clock
National Institute of Information and Communications Technology
(NICT), Japan
Tetsuya Ido
Hidekazu Hachisu, Fumimaru Nakagawa,
Yuko Hanado

55. 80 ns / 0.6 year = 4 × 10-15

56. What’s the International Atomic Time?
NICT
institute k
BIPM
EAL (ensemble)
UTC(k)
H-Maser
>400 clocks
Local Ensemble
adjustment (~1e-13)
TAI
Cs (or Rb) fountains
18 clocks
adjustment (~1e-15)
adjustment
leap second
UTC(NICT)
UTC UT
+ 9h
“Tick” in UTC rarely happens,
ONCE IN 5 DAYS.
Japan Standard Time (JST)
Real signal generated in NICT,
which is Japan Standard Time

57. Optical clocks utilized for time scale
NICT
institute k
BIPM
EAL
UTC(k)
H-Maser
>400 clocks
adjustment (~1e-13)
Cs (or Rb) fountains
TAI
PMS
adjustment (~1e-15)
adjustment ?
leap second
UTC(NICT)
UTC UT
Japan Standard Time (JST)
PMS: Phase micro stepper

58. CIPM# (2015): (2)Hz

59. Steering by intermittent operation of Sr lattice clock
HM frequency and drift rate calibrated by Sr
Adjustment of PMS offset frequency every 4 hours
No servo to reduce the time offset UTC-T_Sr
Based on our frequency* obtained in 2015
Source osc.
(HM)
Phase micro stepper (PMS)
f(HM)
based on Sr
T_Sr
Intermittent operation more than once a week for 104 s continued
* Hz (Hachisu et al., Appl. Phys. B 34, 123 (2017))
… Hz (CIPM2015) →　… Hz (CCTF WGFS, Jun 2017)

60. Advantage of “optical” steering
1 Operation
104 s of operation is sufficient to evaluate the scale unit of a HM at mid level
Short term fluctuation of HM may be compensated by an optical clock.

61. INTERMITTENT Evaluation of a HM for 6 monthsby Sr and UTC(NICT)
1×10-13
104 s measurement

62. Estimation of linear drift rate
y(HM)
How long?
The frequency drift of the HM is estimated from recent calibration by Sr.
How long should we use for the linear fitting? t
3 sample variance is not affected by linear frequency drift.
The duration of the minimum 3 sample variance would be the best to obtain the drift rate
3-sample variance　（Hadamard variance)
1 6 𝑀−2 𝑖=1 𝑀−2 𝑦 𝑖+2 −2𝑦 𝑖+1 +𝑦(𝑖) 2
= 1 6 𝑁−3 𝜏2 𝑁=1 𝑁−3 𝑥 𝑖+3 −3𝑥 𝑖+2 +3𝑥 𝑖+1 −𝑥 𝑖 𝑁=1 𝑁−3 𝑥 𝑖+3 −3𝑥 𝑖+2 +3𝑥 𝑖+1 −𝑥 𝑖

63. Detrended to see the detail
Excursion of UTC(NICT) 7×10-15
HM scale steered by Sr deviates less than 3×10-15

64. Comparison between UTC and a real-time signal of TA(Sr)
TA(Sr): Sr-based time scale
TA(Sr) has bias frequency from UTC obviously

65. Comparison between UTC and a real-time signal of TA(Sr)
TA(Sr): Sr-based time scale
Frequency difference
6.4 x 10-16
From Circular T 345

66. Comparison between TT(BIPM) and TA(Sr)
| TA(Sr)-TT(BIPM) | < ±1.5 ns
Instability of TA(Sr)-TT(BIPM)
20 days
UTC-UTC(k) link dominant
@5, 10 days
Sr clocks able to evaluate the TAI scale (as PFS worldwide do) even in the intermittent operation
→D2L-A, Thursday in last session of Opt,. clock
TA(Sr) clearly detects that
UTC does not realize the SI second.
For 5 months, frequency standards in level were kept generated.

67. Members Sr H. Hachisu Time Scale F. Nakagawa Y. Hanado
Technical assistance
H. Ishijima
M. Mizuno
S. Ito
Special thanks to
1. G. Petit at BIPM for the fruitful discussions