time calibration of lhaaso-wcda engineering array...nal, it depends on the relative variation of the...

33RD INTERNATIONAL COSMIC RAY CONFERENCE, RIO DE JANEIRO 2013THE ASTROPARTICLE PHYSICS CONFERENCE

Time calibration of LHAASO-WCDA engineering arrayXIAOHAO.YOU1,2 , BO.GAO2, ZHIGUO.YAO2, MINGJUN.CHEN2, BIN.ZHOU2, HUICAI.LI3,2 , HANRONG.WU2

FOR THE LHAASO COLLABRATION1Hebei Normal University, Shijiazhuang, 050016, China2Institute of High Energy Physics, Chinese Academy of Sciences,Beijing, 100049, China3Southwest Jiaotong University, Chengdu, 610031, China

[email protected]

Abstract: The Water Cherenkov Detector Array of LHAASO is a high-sensitivity gamma-ray and cosmic-raydetector. The main purpose of it is to survey the northern sky for very-high-energy gamma ray sources. Theshower direction is reconstructed by the hit time so that the precise time calibration between PMT channels areessentially very important. In this paper, firstly we present a calibration method based on LED and optical fibers,then the setup of the calibration system in the engineering array is introduced. At last, the test results of futureplans are given.

Keywords: WCDA, Water Cherenkov, Gamma-ray, Cosmic Rays.

1 IntroductionA large high altitude air shower observatory (LHAASO)[1]is to be built at Shangri-La, Yunnan Province, China. Oneimportant physics goal of LHAASO is to survey VHE(≥ 100GeV ) gamma ray sources in northern sky by us-ing its water Cherenkov detector array (WCDA). For theground-based air shower detector arrays, the direction ofthe origin particle is reconstructed by measuring the hit-ting time of the secondary component which recorded bythe detector cells. According to LHAASO-WCDA, the sig-nal is the Cherenkov photons generated by particles in thewater received by the PMT. To reach a high direction re-construction precision, a reliable detector time calibrationis required.

2 The LHAASO-WCDA time calibrationThe LHAASO-WCDA[2] is to built as a water pond, withan area of 90,000m2and an effective water depth of 4.5 m.The pond will be divided into 3,600 detector cells by blackpolyethylene (PE) curtain which can avoid the “cross-talk”effect between neighbouring cells when a particle passingthrough the detector. Each cell detector is 5m × 5m in areaand a 8-inch PMT is fixed on the ground at the center. Con-sidering the electronics and DAQ design, 9 cell detectorsforms a group, 4 neighbouring group forms a cluster.

2.1 Time calibration requirementThe primary goal of the timing calibration system is tomonitor the time stability of the PMTs and related read-out/digitization electronics. For the gamma ray source suchlike Crab, the main signal reaches to the ground can gener-ate a very small shower. Due to the trigger design of WC-DA and simulation, a number of multiple hits is around 20when a shower is recorded. To achieve an angular resolu-tion ≤ 0.1◦, a simple simulation is done. In the simulation,20 detectors are selected randomly in a cluster (6×6), eachdetector is given the same systematic error from 0.06ns to0.35ns, and the shower front is fitted by plane function, the

result shows that the systematic error needs to be less than0.12ns (see in Fig.1).

Time precision [ns]0.05 0.1 0.15 0.2 0.25 0.3 0.35

]°T

het

a E

rro

r [

0.02

0.04

0.06

0.08

0.1

0.12

0.14

0.16

Theta Reconstruction Error

Fig. 1: Relationship between reconstruction direction errorand detector time resolution.

2.2 Time calibration methodAs the LHAASO-WCDA structure mentioned above, it isvery difficult to implement a single light source calibrationsystem such as a laser combines optical fibers, as the fiberlength will be longer than 200 m, it is very hard to preserve.So a new time calibration method is proposed. As it isshown in Fig.2. each cluster contains two sets of opticalfibers which are in the same length of 40 m, illuminatedby one LED. One set of fibers is implements in the cluster,the other set of fibers, part of them are implemented in thecluster, others are implemented in the neighbouring cluster,the whole WCDA are chained cluster by cluster to achievethe goal of whole detector array time calibration.

3 Time calibration of LHAASO-WCDAengineering array

To study “cross” calibration method, a time calibrationsystem prototype has been made for LHAASO-WCDAengineering array[3] (See in Fig.3). The prototype mainlyconsists 3 parts: LED, LED driver and optical fibers.

33RD INTERNATIONAL COSMIC RAY CONFERENCE, RIO DE JANEIRO 2013

Fig. 2: Time calibration scheme.

ED+MD

ED

ED

Charge Calibration Setup

5 m

5 m

1

4

2 3

65

987

4 6 5 1

078

3 2 9

4 channel

pulser

VCC

+170V

Driver I

Driver II

Driver III

LED I

LED II

LED III

Fiber bundle I

Fiber bundle II

Fiber bundle III

30m x 5

30m x 5

45m x 10

Delay

Fig. 3: Time calibration system prototype of LHAASO-WCDA engineering array.

3.1 LEDThe LED used in the time calibration system prototype ismanufactured by GreeThink Corp, which diameter is 5 mmand light emitting wavelength is around 470 nm, Fig.4 isthe wavelength distribution of the LED.

WaveLength [nm]200 300 400 500 600 700 800 900 1000

Co

un

ts

3.4

3.6

3.8

4

4.2

4.4

4.6

4.8

GreenThink_5mm

Fig. 4: LED wavelength.

3.2 LED driverIn order to obtain a stable, fast and narrow pulse, an easyLED driver has been made .It is a simple and stable , andused successfully in the engineering array. The circuit isbased on an avalanche transistor (2N5551), with a DCpower supply of 180 V, 1 A. The driver is triggered by afast positive pulse signal (rise time less than 5 ns). Theoutput signal is adjust to -9V in amplitude and 5ns in width.Fig.5 is the driver circuit and simulated output signal byusing Multisim.

3.3 Optical fiberThe optical fibers is the light transmitter, the fiber used inthe system is PMMA plastic fiber, the bending radius andtemperature effect of the fiber is tested in the lab by usingoscilloscope and PMT XP2012B manufactured by Photo-nis Corp., the results are shown in Fig.6, the signal risetime does not depend on the bending diameters of the fiber,whereas the noticeable change of the signal transmissiontime starts at diameter 10 cm; The temperature of pool bot-

Tektronix Oscilloscope-XSC1 Print Time:March 13, 2013 14:29:46Tek Acq Complete M Pos: 5.8ns

1

V11- 1HCsn5 MV01 1HC10kHz

Traces: 1,

Fig. 5: Left : LED driver circuit; Right : simulated signalby Multisim.

tom ranged from 5◦C to 10◦C and the temperature coeffi-cient of fiber and PMT is about 3‰in all.

Fig. 6: Top : fiber bending effect; bottom : temperatureeffect. (Red points mean the fiber and PMT are in thetemperature oven together; black points mean only fiber inthe temperature oven.)

4 Calibration resultsThe online time calibration of engineering array is set un-der in shower mode (trigger multiplicity = 3) with the PMTthreshold set to 1/3 PE, the LED frequency is set to 5Hz,every data taking time is set to 10 minutes. The single timecalibration result is got through Gaussian fitting. Three val-ues are concerned (see in Fig.7): short fiber time offset be-tween test PMT and reference PMT (S.T); long fiber timeoffset between test PMT and reference PMT (L.T); timeoffset long fiber and short fiber of one PMT (∆T ).

4.1 Single measurement precisionSingle measurement precision represents the accuracy ofcalibration system prototype, PMT, cables, electronics sys-tem. Four principal influence factors are identified : LEDlight respond time, TDC accuracy (0.5 ns), T.T.S of PMT


Ti me [ns]-400 -300 -200 -100 0 100 200 300 400

Am

pli

tud

e [

V]

-0.8

-0.6

-0.4

-0.2

0

0.2

Ti me [ns]-400 -300 -200 -100 0 100 200 300 400

Am

pli

tud

e [

V]

-0.5

-0.4

-0.3

-0.2

-0.1

0

0.1

0.2

S.T L.T

ΔT_ref

ΔT_test

Fig. 7: Example of LED signals between test channel andreference channel.

in the case of multi-photoelectron and the light pulse timewidth depending on the length of fiber.

σall =√

σ2LED +σ2

T DC +σ2f iber +T T S2 (1)

Fig.8 demonstrates the results of single measurement be-tween one test channel and reference channel of short fiberand long fiber, and Fig.9 demonstrates the results of timedifference between long fiber and short fiber in test chan-nel and reference channel separately.

/ ndf 2χ 58.18 / 25

Constant 5.88± 85.46 Mean 0.49± 93.52 Sigma 0.408± 8.831

-1400 [ 5.6 ns ]3

T∆

0 50 100 150 200

En

trie

s

0

10

20

30

40

50

60

70

80

90

/ ndf 2χ 58.18 / 25


Delta Time of F EE/ CH 0/3

/ ndf 2χ 81.42 / 31


-1400 [ 5.6 ns ]9

T∆

-50 0 50 100 150

En

trie

s

0

20

40

60

80

100

/ ndf 2χ 81.42 / 31


Delta Time of F EE/ CH 0/9

Fig. 8: Left : Time offset measurement between short fiber;Right : time offset measurement between long fiber. Theprecision is about 0.9 ns (short fiber) and 1.3 ns (longfiber).

/ ndf 2χ 31.25 / 19Constant 7.3± 117.3 Mean 0.37± 13.67 Sigma 0.325± 7.918

T [5.6ns]∆-40 -20 0 20 40 60

Ent

ries

0

20

40

60

80

100

120 / ndf 2χ 31.25 / 19


Time Difference of FEE/CH 0/3-0/9 / ndf 2χ 15.82 / 12

Constant 4.92± 90.62 Mean 0.5± 2853 Sigma 0.30± 10.31

T+2800 [5.6ns]∆2780 2800 2820 2840 2860 2880 2900 2920 2940

En

trie

s

0

10

20

30

40

50

60

70

80

90 / ndf 2χ 15.82 / 12

Constant 4.92± 90.62 Mean 0.5± 2853 Sigma 0.30± 10.31

Time Difference of FEE/CH 0/3-0/9

Fig. 9: Left : time offset between long fiber and short fiberof reference channel; right : time offset between long fiberand short fiber of test channel.

4.2 Long-term stabilityThe long-term stability is crucial in detector running, thetime calibration of engineering array has been done in 2months, Fig.10 shows the operating state of two test chan-nels which illuminated by 2 separate LEDs. The signalamplitude of the short-fiber is relatively larger than that ofthe long-fiber for the 2 test channels. The time stability ofchannel is not heavily dependent on the amplitude of sig-nal, it depends on the relative variation of the signal ampli-tude. But the charge distribution of signal varied less than

C ]

°R

oom

Tem

pera

ture

[

05

101520253035

LG C

harg

e [ A

DC

bin

]

290292294296298300302

[ n

s ]

off

S.T

1.82

2.22.42.62.8

3

[ n

s ]

off

L.T

8.89

9.29.49.69.810

10.210.410.6

181 182 183 184 185 186 187 188

T [

ns]

∆

265.5

266

266.5

267

267.5

C ]

°R

oom

Tem

pera

ture

[

05

101520253035

LG C

harg

e [ A

DC

bin

]

410

415

420

425

430

[ n

s ]

off

S.T

3.43.63.8

44.24.4

[ n

s ]

off

L.T

13.213.413.613.8

1414.214.414.614.8

181 182 183 184 185 186 187 188

T [

ns]

∆

268

268.5

269

269.5

270

Fig. 10: The running status of 2 PMTs illuminated by sep-arate LEDs in one week. From top to bottom is :1)roomtemperature;2)LED light intensity ;3)short fiber time off-set between reference PMT and test PMT;4)long fiber timeoffset between reference PMT and test PMT;5)time offsetbetween short fiber and long fiber of one PMT.

5% in the operating time , so it does not have a serious im-pact on time calibration. The long-term monitoring resultsare shown in Fig.11. As it can be seen, the average accura-cy of short fiber channel is 0.12 ns, but the long fiber chan-nel is 0.22-0.23 ns. The difference is caused by the influ-ence of different fiber length broadening pulse width andsignal overshoot. During operation in the site,the impactof environmental change on the calibration system causeabout 0.1 ns error and this also characterize the stability ofcalibration system for the change of external environmen-t. Optimizing the signal delay and uniforming the outputlight can further improve the measurement accuracy.

/ ndf 2χ 15.52 / 7Constant 23.4± 889 Mean 0.003± 2.405 Sigma 0.0020± 0.1255

0 1 2 3 4 50100200300400500600700800900 / ndf 2χ 15.52 / 7

Constant 23.4± 889 Mean 0.003± 2.405 Sigma 0.0020± 0.1255


8 9 10 11 12050

100150200250300350400450



265 266 267 268 269050

100150200250300



2 2.5 3 3.5 4 4.5 5 5.5 60100200300400500600700800



12 12.5 13 13.5 14 14.5 15 15.5 16050

100150200250300350400450 / ndf 2χ 79.43 / 31



268 268.5 269 269.5 270 270.5050

100150200250300350


Fig. 11: Top: The distribution of short-fiber time offsetmean value between test channel and reference channel;middle: The distribution of long-fiber time offset meanvalue between test channel and reference channel; bottom: The distribution of time offset mean value between short-fiber and long-fiber.

5 ConclusionThe principle of the LED and dual fiber calibration havebeen proven feasible and the precision can reach expecta-


tions. According to the experience of LAWCA time cal-ibration, PMT time performance need to more strict test.We will focus on improving reliability of the calibrationsystem as well as the design of the spectrometer in the fu-ture.

Acknowledgment:The authors express their gratitude to theessential support of X.F. Yuan, G. Yang in the installation, debug-ging, and maintenance of the detector. This work is partly sup-ported by the National Natural Science Foundation of China (NS-FC) under the grant No. 10975046.

References[1] Cao, Z. A future project at tibet: the large high altitude air

shower observatory (LHAASO), CHINESE PHYSICSC,2010,34(2):4,249-252.

[2] Zhiguo Yao for the LHAASO collaboration, Proceedings of32nd ICRC, Beijing(2011).

[3] Mingjun Chen for the LHAASO collaboration, Proceedingsof 33rd ICRC, Rio de Janeiro(2013).

time calibration of lhaaso-wcda engineering array...nal, it depends on the relative variation of the...

Documents