© GMV Property - All rights reserved
PULCHRON: A NEW PULSAR-BASED TIME SCALE REALIZATION
NAVISP Industry Days
ESTEC
January 23/24, 2019
© GMV Property – All rights reserved
Pulsars: Highly Magnetized Rotating Neutron Stars
Image property: ESA/XMM-Newton/L. Osklnova/M. Guerrero; CTIO/R. Greundl/Y.H. Chu
© GMV Property - All rights reserved
Pulsars: Rotating Stars• Pulsars are highly compact, rapidly rotating and strongly magnetized
neutron stars• They have very regular rotational periods• Beams of electromagnetic radiation originate at the magnetic poles of the
star are swept around the sky as the star rotates• Radio beams can be detected by radio-telescopes on Earth• In the fastest rotating “millisecond pulsars”, we can measure the Time of
Arrival (ToA) with accuracy below 1 us, and down to 30 ns for the best pulsars -> benefits of pulsar timing can only be seen in the long term
• Despite their high stability, each pulsar has an arbitrary rotation period and drift -> they cannot be used to implement or measure the definition of the second
• First results published already in the 1980’s
CREDIT: Bill Saxton, NRAO/AUI/NSF
© GMV Property - All rights reserved
Atomic Time Scales
• Terrestrial timekeeping is based on an ensemble of atomic clocks, located at different timing laboratories; data from such clocks are provided to the BIPM, which combines them to generate different time scales:
_TAI (International Atomic Time): continuous atomic reference time scale aligned to the SI definition of the second
_UTC (Universal Time Coordinated): adaptation of TAI, including leap seconds, aligned to earth rotation within 1 s
_TT (Terrestrial Time): yearly post-processing of TAI, to form the most stable time scale on earth; mainly used for scientific purposes due to its high latency; TT is BIPM’s recommended reference for any pulsar time scale analysis.
from ESA, atomic clocks located at ESTEC
© GMV Property – All rights reserved
Pulsar Time Scale Generation
• Previous research regarding pulsar timing (PT) realize a paper time scale using pulsar measurements and compare it to terrestrial time scales TT and TAI
• Very complex “astronomical” software and models are needed to model the pulsar rotation and signal propagation, and conversion of ToA to the solar system barycenter; TEMPO2 software is the reference one
• PT seems to agree better with TT than with TAI
BIPM’s TT-TAI as solid lines, quadratic model removed; PT-TAI in dots. G. Hobbs et al., “Development of a pulsar-based time-scale”, Mon. Not. R. Astron. Soc. 427, 2780–2787 (2012)
© GMV Property - All rights reserved
Pulchron Design
• Three main blocks are identified in the PulChronarchitecture:
_PMCS: Pulsar Measurement Collection System_PFSU: Pulsar Frequency Standard Unit: physical
pulsar time scale realization_PACE: Pulsar-Augmented Clock Ensemble: a-
posteriori “paper” time scale mixing pulsar data and GNSS station and satellite clocks from ODTS
• Integrated within the Galileo Time and Geodetic Validation Facility (TGVF) at ESTEC
• TEMPO2 software provided by the University of Manchester
• Time transfer between the radio-telescopes and the PulChron is done using GPS receivers (at ESTEC, GNOR station)
© GMV Property – All rights reserved
Pulsar Measurement Collection
• The Pulsar Measurement Collection System (PMCS) is in charge of gathering measurements from five different radio-telescopes in Europe and provide such measurements to the PulChron server
• An agreement with the European Pulsar Timing Array (EPTA) has been reached, and measurements from them are delivered monthly under a Service Level Agreement (SLA). Data collection activities are led by the University of Manchester
Effelsberg, Lovell, Westerbork, Nançay and Sardinia radio telescopes conform the EPTA
© GMV Property – All rights reserved
Pulsar Data: Monthly SLA
• The quality metric (Q) is the nominal uncertainty of the optimally weighted arithmetic mean of the data; this is based on the formal error of the data (after any error factors or quadrature errors are included), and represents the maximum possible uncertainty in the clock signal from these data.
Metric Value
Typical Number of pulsars 18
Minimum number of pulsars 12
Maximum quality metric (Q) 200 ns
Data structure and format Pulsar Files
Frequency of delivery 1 per calendar month
Delivery date 01:00 UTC on the 15th day of the calendar month following the dataset month
Delivery method sFTP to TGVF server
SLA Summary
Code: NAVISP1-DD-GMV-006-007
Date: 15/11/2018
Version: 1.3
Page: 14 of 21
Pulsar Timescale Demonstration ESA UNCLASSIFIED
PMCS Design and SLA Implementation
Data dataset was generated
Name of person and institute that generated the dataset
Origin of the dataset (e.g. European Pulsar Timing Array)
Number of pulsars in the dataset
Number of observations in the dataset
Quality metric of the dataset.
The quality metric is the nominal uncertainty of the optimally weighted arithmetic mean of the data. This is based on the formal error of the data (after any error factors or quadrature errors are included), and represents the maximum possible uncertainty in the clock signal from these data. This is given by:
,
where, for each ToA i, i is the uncertainty on that TOA after application of any corrections, i.e. this is
the uncertainty value included in the residual file.
The format of the dataset summary file shall be plain text, in the format shown below: PULCHRON Epoch ID: 201903
Data start: 2019-03-01
Date end: 2019-03-31
Date generated: 2019-04-10
Generated By: M. Keith, University of Manchester
Data Origin: European Pulsar Timing Array
Number of pulsars: 18
Number of observations: 584
Quality Metric (ns): 123.1
6.4. PULSAR PARAMETER FILES
The pulsar parameter files will be provided in a format suitable for use with TEMPO2. The format of
these files is a series of key-value pairs, with additional flags indicating if the parameters should be included in the fit. These files will generally be the same each epoch, though they may be updated if it
is decided that this will improve the performance of the PULCHRON. The par file may contain corrections to the measurement uncertainties to account for pulsar jitter or instrumentational effects. The format of these files is described in Section 5.1.1.
6.5. TIME OF ARRIVAL MEASUREMENT FILES
Time of Arrival measurement files will be delivered as plain text files in a format compatible with the
widely used TEMPO2 software. In particular, the files contain the observing frequency (in MHz), the measured TOA as an MJD (with a precision of 10-14 days), the formal uncertainty on the TOA in microseconds, the site code and any flags required by the par file. The format of these files is
described in Section 5.1.2.
6.6. CLOCK CORRECTION FILES
For each site code in the TOA files, an observatory clock correction file shall be provided which
contains values of UTC(GPS) – UTC(SITECODE) computed at least once per day over the period that that site code appears in the TOA files. UTC(GPS) is an estimate of UTC derived from a GPS receiver,
and UTC(SITECODE) is timeframe in which the TOAs were made (e.g. UTC(JBDFB), UTC(EFF), etc.) These clock correction files shall be provided in a format compatible with the TEMPO2 software. The
format for these files is described in Section 5.1.3.
Q = 26 ns
© GMV Property – All rights reserved
Pulsar Data: Monthly SLA
• Typical recent Q is of the order of 20-30 ns; improvement by a factor 3 with respect to past EPTA data with a similar number of pulsars tracked.
Month/Year Nmr. Of Pulsars
Quality Metric (ns)
May/18 18 16.5
June/18 17 17.2
July/18 14 20.0
August/18 14 31.5
September/18 17 21.9
October/18 18 26.2
November/18 18 37.2
December/18 15 38.8
January/19 17 36.5
February/19 16 23.7
March/19 17 27.3
April/19 13 72.9
May/19 17 26.2
June/19 16 17.1
July/19 16 25.4
© GMV Property – All rights reserved
Installation at ESTEC
Hydrogen Maser
TGVF: GNSS monitoring
© GMV Property – All rights reserved
Pulsar Timescale versus UTC (ESTEC)
December 2018
© GMV Property – All rights reserved
Comparison against UTC
PT – UTC = [PT–UTC(ESTC)] – [UTC-UTC(ESTC)]
© GMV Property – All rights reserved
Comparison against UTC (Zoomed)
© GMV Property – All rights reserved
Frequency Stability
© GMV Property – All rights reserved
Summary and Outlook
• A physical pulsar time scale realization has been running operationally at ESTEC during the last year
• We exploit the short and medium term stability of a Hydrogen Maser, and steer it towards the result of the pulsar measurement processing
• The quality of the recent pulsar measurements looks 3x better than in the historical datasets from previous years
• We have observed the impact of the clock behavior in such an architecture, where a small jump/change in the nominal frequency behavior will change the rate w.r.t. UTC timescale and could impact the stability
• The generated time scale is largely independent of atomic time, and thus comparison against the most stable time scales, such as TAI/UTC and TT, will be of great interest
• As the Pulsar timescale does not run at the frequency of the second because of unavoidable synchronization inaccuracies, increasing phase deviation are observed with respect to UTC. The stability once removing the deterministic trend is comparable to the most stable timescales
© GMV Property – All rights reserved
Lessons learned
• Due to the relatively high uncertainty of pulsar measurements, a long experimentation period would be required to confirm the value of pulsar time scale (several years to one decade), but the PULCHRON time scale already provides a completely independent reference time scale suitable to certain applications
• For the future, the use of several atomic clocks in parallel is deemed suitable, in order to detect jump in the comparison between them. Another solution could be to use GNSS time to monitor the expected behavior of the clock and its correct steering
• Finally, it was suggested that longer dwelling times might be needed for the steering of the clock, to better extract the information of the pulsar measurements