1. Report PRR on time calibration
Workshop on Time Calibration, Amsterdam,
Report PRR on time calibration
Juande Zornoza (IFIC)

2. introduction Reviewers: Calibration team:
Michael Punch (APC, France)  CTA expert on WR
Dawn Williams (University of Alabama, USA)  Time calibration coordinator in IceCube
Ralf Wischnewski (DESY, Germany)  One of main contributors to time calibration system on Tunka (based on WR)
Calibration team:
Marco Circella (former technical project manager)
Giorgio Riccobene (coordinator of calibration)
Mieke Bouwhuis (editor of TDR)
Juande Zornoza (coordinator of time calibration)
Support (=doing the work!):
Rosa Coniglione
Karel Melis
Martijn Jongen
Alex Creusot

3. procedure Meetings (vidyo and face to face) Report by reviewers
Reaction by the collaboration on the report
Our counter-report (reviewed by the SC) sent to Scientific and Technical Advisory Committee
we are here

4. history of meetings
December 22nd: Kick off meeting with reviewers (vidyo)
January 23rd: Meeting with reviewers (vidyo)
February 20th: Internal meeting
February 22nd: Internal meeting
February 24th: Meeting with reviewers (face to face, at CPPM)
April 18th: Internal meeting after report

5. history of meetings

6. Structure of report 1. Design 2. Procedure 3. Verification
1.1 Requirements
1.1 Clock synchronization
2. Procedure
2.1 Dark room calibration
2.2 Round trip time measurement and stability
2.3 PMT-PMT (intra-DOM) timing measurements
2.4 Recalibration after HV tuning
2.5 Documentation
3. Verification
3.1 Verification with nanobeacons
3.2 Verification with muons
3.3 Verification with laser beacon
3.4 Documentation
4. Summary of committee findings and recommendations

7. Summary of the report (1/2)

8. Summary of the report (2/2)

9. Requirements Summary Remarks
Required precision determined by the requirement of not worsening the intrinsic precision due to chromatic dispersion (1.5 ns) and PMT TTS (1.3 ns)
Angular resolution not impacted by errors at the level of 1 ns
Above several nanoseconds, worsening is observed
Therefore, time resolution should be kept at 1-2 ns
Remarks
Efforts to study and present this effect are commended
We urge the collection of acoustic data and examination of the time calibration data with updated geometry

10. Clock synchronization
Summary
Clock signal is distributed over a fibre optic network, based on a customized version of White Rabbit (to reduce the number of uplink channels)
In the KM3NeT implementation, the master clock signal is broadcasted to all base units
This approach need custom WR firmware
Disadvantage: calibration must be done in the dark room on a completely-equipped line
Advantage: return signals can be multiplexed onto a standard Ethernet switch
Remarks
Cost reduction not clearly expressed in quantitative terms
Firmware is frozen or has to be re-developed
We suggest to explore costs and benefits of standard WR approach. This may also avoid the need to calibrate completely equipped lines
A standard WR switch could be introduced at the DU bases, with all DOMs synchronized to it. A splitting of the bulk-data stream from DOMs to shore can be done via normal switches. A potential issue of this is the reliability of hardware off-shore

11. Dark room calibration Summary Remarks
Calibration is done in the dark room with a laser signal distributed over even fibres (offset PMT – DU base)
Time offsets in the same DOM are calibrated in the sea using K40
The laser time calibration is done for a single PMT in each DOM. The calibration of the rest relies on K40. A potential problem could arise if the PMT becomes inoperable soon after deployment, before in situ K40 calibration is done
It is essential that the HV is kept the same in situ as measured in the laboratory to have the same transit time
Remarks
It would be prudent to measure at least two PMTs as reference
Check if there is any change in HV from 20ºC to 10ºC

12. Round trip time measurement and stability
Summary
RTT includes the effect of chromatic dispersion, the asymmetry in the shore station and the asymmetry in the junction box due to different uplink and downlink paths.
RTT has been tracked for half a year for two ARCA lines. Large jumps of the order of tens of nanoseconds are attributed to changes in the fibre optic network. Smaller changes (1-2 ns) occur coherently in both DUs, correlated to day/night temperature changes (onshore).
Remarks
For one of the next DUs, we recommend a test base unit with two clocks in parallel to crosscheck long term performance and stability
Changes in the detector or network should be well documented

13. PMT-PMT (intra-dom) timing measurement
Summary
Light from K40 is used to calibrate PMTs in the same DOM
The mean of the distribution of time differences of hits give the time offset
The width of that distribution gives the TTS due to PMT TTS and electronics

14. Recalibration after hv tuning
Summary
PMT transit time depends on the HV applied.
The potential loss of gain with time of PMT will be compensated with an increase in the PMT HV, which will change the transit times.
The procedure foreseen to carry out such recalibration consists of changing first the HV of one half of the PMTs, then recalibrate with K40 and change the second half.
Remarks
It is recommended to put back the HVs of the first half of PMTs, then to modify the HVs of the second half and compare with the unchanged first half using a new set of K40 measurements
It is proposed to test the effect of several re-tunings with a toy MC. Also, the effect of several iterations should be tested in the already installed lines and/or in the laboratory
It is suggested to install SiPM sensors (transit time of ps) on future DOMs to observe light from nanobeacons and laser beacons

15. Documentation (of procedure)
Summary
The time calibration constants are stored in a plain text file (“detector file”)
Remarks
The file is apparently not self-documented (no column headers)
Other options like XML or JSON should be considered
Evolution of detector files should be documented

16. Verification with nanobeacons
Summary
Nanobeacons are LED flashers installed in each DOM
Intra-DOM calibration with K40 and nanobeacons agree within 0.7 ns
Remarks
Verification of intra-DOM calibration with nanobeacons should be part of the standard procedure

17. Verification with nanobeacons
Summary
Nanobeacons are LED flashers installed in each DOM
Intra-DOM calibration with K40 and nanobeacons agree within 0.7 ns
Remarks
Verification of intra-DOM calibration with nanobeacons should be part of the standard procedure

18. Verification with Muons
Summary
Atmospheric muons can calibrate in principle the whole system chain
It assumes though a perfect knowledge of DOM acceptance and water properties (and trigger, reconstruction, etc. dependence)
It is therefore not recommended to rely solely on muons
Remarks
A study of systematic effects mentioned above should be done with MC
Intrinsic limitations of muon fits should be studied
Effect of systematic uncertainties (water optical properties, etc.) should be studied

19. Verification with laser beacon
Summary
The laser beacons described in the TDR has not been deployed yet
Ultra-short (sub ns) flashes can illuminate up to 50 m vertically and 200 m horizontally
It only requires knowledge of DOM/beacon positions
It can verify the calibration system on a short time scale and in a way independent of other effects
Remarks
We regards this as the most reliable and straightforward way to verify inter-DU time calibration
We make the strongest possible recommendation for the deployment of the laser beacon
Precision in laser location is needed
Several laser beacons will be needed

20. Documentation (of Verification)
Remarks
We encourage KM3NeT to collect long term stability data in the verification phase
Changes to the system should be captured in a sustainable way for long-term documentation

21. Summary of the report (1/2)

22. Summary of the report (2/2)

23. Some counter-remarks/thoughts 1/2 (Mieke, martijn, karel, marco, jd…)
Some more work not required by the committee will be anyway useful to do, like more extensive analysis on the effect of time miscalibration in the performance of the detector
Custom WR could not be big problem in terms of need of frequent changes (+KM3NeT pluggin for WR, + workshops)
Calibrating the complete line is advantageous
Broadcast approach has been proved to work (but scalability, cost…)
In any case, recommendation about standard WR could be adopted for Phase 2 (still under discussion, but not the default option. It is one of the topics of the workshop: scalability, cost, reliability, redundancy…)
Hardware changes do not imply firmware changes

24. Some counter-remarks/thoughts 2/2 (Mieke, martijn, karel, marco, jd…)
SiPMs: Do we need a new sensor in addition to the 31 we have?
The header of the detector file contains a tag number which is interpreted by the software according to the SVN version
Nanobeacon calibration is (basically) already part of the standard procedure
Muons are probably better than what the committee feels (we have proven that they are consistent with other systems, they are cheap and fast…)
Hardware changes do not imply firmware changes

25. Next steps (with names!)
Response document
Marco will coordinate the edition of the document
Tasks already agreed in the Marseilles meeting
Check on effect of temperature on HV tuning (Alex?)
Check with MC and lab on HV tuning procedure (Alex? Negligible?)
Other tasks
Study of systematic effects with muons (Nafis)
More studies on effect of timing on performance (Rosa)
Cost analysis of Full vs Custom WR options (Tommaso?)
Test base unit with two clocks in parallel

26. conclusions Positive reaction of referees
We have learnt a few things in the mean time
Still work to do!
Report
Implement changes
Thanks to all for the work done until now and for the next efforts
This starts (as a matter of fact continues) NOW with the discussions in this meeting