Laser Calibration System for Time of Flight Scintillator Arrays
A. Denniston, E.P. Segarra, A. Schmidt, A. Beck, S. May-Tal Beck, R. Cruz-Torres, F. Hauenstein, A. Hrnjic, T. Kutz, A. Nambrath, J.R. Pybus, P. Toledo, L.B. Weinstein, M. Olivenboim, E. Piasetzky, I. Korover, O. Hen
LLaser Calibration System forTime of Flight Scintillator Arrays
A. Denniston, E.P. Segarra, A. Schmidt ∗ , A. Beck , S. May-Tal Beck , R. Cruz-Torres, F. Hauenstein ,A. Hrnjic, T. Kutz , A. Nambrath, J.R. Pybus, P. Toledo, O. Hen Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA
L.B. Weinstein
Old Dominion University, Norfolk, Virginia 23529, USA
M. Olivenboim, E. Piasetzky
School of Physics and Astronomy, Tel Aviv University, Tel Aviv 69978, Israel
I. Korover
Nuclear Research Center Negev, Beer-Sheva 84190, Israel
Abstract
A laser calibration system was developed for monitoring and calibrating time of flight (TOF) scintillatingdetector arrays. The system includes setups for both small- and large-scale scintillator arrays. Following test-bench characterization, the laser system was recently commissioned in experimental Hall B at the ThomasJefferson National Accelerator Facility for use on the new Backward Angle Neutron Detector (BAND)scintillator array. The system successfully provided time walk corrections, absolute time calibration, andTOF drift correction for the scintillators in BAND. This showcases the general applicability of the systemfor use on high-precision TOF detectors.
Keywords: scintillator, calibration, laser system, neutron detector, TOF
1. Introduction
Time of flight (TOF) detectors are used to mea-sure the time it takes for a particle to travel a givenpath length. This can provide particle identificationby measuring the TOF of particles with a knownmomentum, determine the momenta of known par-ticles, or determine the coincidence of multiple par-ticles detected from a single interaction. Typically, ∗ Corresponding Author
Email address: [email protected] (A. Schmidt) Present address: George Washington University, Wash-ington, D.C. 20052, USA Also at: Nuclear Research Center Negev, Beer-Sheva84190, Israel Also at: Old Dominion University, Norfolk, Virginia23529, USA Also at: George Washington University, Washington,D.C. 20052, USA
TOF detectors consist of a plane of scintillating de-tectors whose signal arrival time can be comparedto a reference time.Large-scale TOF detectors often use many sep-arate scintillators to cover large solid angles withgood spatial granularity. Each scintillator requiresdedicated light detectors and readout electronics.This in turn requires the precise absolute and rel-ative calibration of all scintillators in the detector.While data from over-constrained reaction channelscan be used for these calibrations, a dedicated cal-ibration system based on an external light sourceis often more flexible, more convenient, and moreprecise. A pulsed laser system that delivers simul-taneous light pulses to each scintillator with knownamplitude can quickly establish a full set of calibra-tions and correct any time-dependent drifts.
Preprint submitted to Elsevier May 22, 2020 a r X i v : . [ phy s i c s . i n s - d e t ] M a y igure 1: The backward angle neutron detector (BAND) installed in Hall B at Jefferson Lab. Such a laser system was developed for the back-ward angle neutron detector (BAND), a TOF arrayconsisting of 140 scintillator bars [1]. BAND, pic-tured in Figure 1, was recently deployed in exper-imental Hall B at the Thomas Jefferson NationalAccelerator Facility (JLab). BAND was designedto detect backwards recoiling neutrons over the mo-mentum range of 200–600 MeV /c , specifically spec-tator neutrons from the deep inelastic of electronsoff of protons bound in deuterium. This techniqueof spectator-tagged DIS allows the determination ofthe proton’s nuclear modification as a function ofvirtuality, with the goal of elucidating the relation-ship between the EMC Effect and the short-rangecorrelations between nucleons [2]. BAND’s designgoal is to measure neutron time of flight with aresolution better than 300 ps, making timing cali-brations at the level of 100 ps paramount.The required specifications for the calibration ofTOF arrays, including but not limited to BAND,are discussed in the following section. The lasersystem discussed in this article provided critical cal-ibrations for BAND and served as a proof of con-cept for application in other large TOF scintillatorarrays.
2. Design criteria
The exact requirements for the calibration ofscintillator arrays depend on the specific demandsof the detector. However, some general assump-tions are valid for most plastic scintillating detec-tors. Their time resolution should be at the level ofa few hundreds picoseconds (ps) or better. There-fore, the calibration system should provide shortpulses with a pulse width up to a few nanoseconds(preferably <
3. Laser System
The primary laser chosen for the system is a TeemPhotonics STV-01E-140 Nd:YAG pulsed laser, pro-ducing light with a wavelength of 355 nm in a sin-gle longitudinal mode [3]. The laser is optimized tocreate pulses less than 400 ps in duration, with arepetition rate between 10 Hz and 4 kHz.The laser couples directly to a fiber-optic ca-ble, and the pulses are contained within fiber-opticcables throughout the entire system until enter-ing the scintillators. This laser complies with thethree aforementioned demands: (1) UV wavelengthto mimic scintillation from ionizing radiation, (2)short rise-time of the laser pulse, and (3) closed op-tical route for safety.Each laser pulse has an energy of 1 microjoule( µ J), significantly smaller than that required byfree-space gas lasers, like those employed in theTOF calibration systems of the former CLAS [4]and BLAST [5] experiments. Even after splittingthe light into 400 output fibers, this was still morethan enough per pulse to mimic the light responseof scintillators to high-energy particles. Test benchstudies showed that a laser pulse of 25 pJ was ap-proximately equivalent to the light produced by cos-mic rays in a BAND scintillator module. A second, less expensive, 405 nm ThorlabsNPL41B pulsed laser [6] was also used for initialtesting and comparative table-top studies. Thelaser was tested to ensure the wavelength was suf-ficiently short to excite standard plastic scintilla-tors. The laser pulses have a minimum width of6 ns, with a repetition rate of up to 10 megahertz(MHz). Each pulse has an energy of 1.5 nJ, sev-eral hundred times smaller than the primary 355nm laser. This laser is not fiber-coupled, requiringthe implementation of a Thorlabs PAF-X-18-PC-AFiberPort coupler [7]. Two Thorlabs NB1-K08 mir-rors [8] are used to ensure that the incoming laseris aligned with the fiber axis of the coupler.Both lasers are single-mode, which is incompati-ble with the fiber distribution system described be-low. We therefore added a Newport FM-1 fibermode scrambler at the laser exit to produce a sta-ble multi-mode distribution from the single-modelaser pulse [9].
The primary laser is operated using a Teem Pho-tonics MLC-03A-DP1 laser controller [10] and aSiglent SDG 1032X pulse generator [11], whichserves as the trigger for the laser controller.Both the laser controller and pulse generator arecontrolled by a Raspberry Pi computer that alsomonitors the temperature readout from the laseritself. The laser controller’s safety interlock systemprevents it from being energized if the laser enclo-sure is open. The setup is depicted in Fig. 2b and 3.To control the laser system, the Pi connectsto the network via Ethernet and provides a GUIthrough an HTTP server (see Figure 4). This al-lows the Pi to directly connect to the Internet forremote control. The GUI allows control of the trig-ger pulse settings, the attenuator level, monitoringthe laser temperature, and powering up/down theentire system.
After a light pulse leaves the laser in an opticalfiber, it is processed using various optical compo-nents that split and distribute the light, as shownin Figure 2a and discussed below.An initial 90:10 fiber splitter is used to divide thelaser pulse into a reference pulse and a pulse to bedistributed to the detector.The 10% reference pulse is sent to a ThorlabsDET025AFC/M fiber-coupled silicon photodetec-tor [12]. The signal from this photodetector is used3 iber Coupled Laser Mode Scrambler Variable Attenuator90%10%Reference Photodiode Fiber Distribution System Detector ...
ScintillatorPMT PMTScintillatorPMT PMTScintillatorPMT PMTTo discriminator (a) Laser system
Raspberry Pi Pulse Generator Laser Controller Network Variable AttenuatorInterlock System Fiber Coupled LaserPulse to LaserLaser Temp Sensor \\ (b) Laser controller system A D C T r i gg e r M a n a g e r Fan In Fan Out DiscriminatorScintillatorPMT PMT DelayReference Photodiode Discriminator Data Acquisition VME Gate Generator PMT SignalADC GatePMT TimeReference TimeTDC triggerL1A InputL1A Response T D C Fiber Optic Cable (c) Electronics setup
Figure 2: Schematic diagrams of the laser system, laser controller, and electronics. Red indicates optical components. Blackindicates electronic components. Orange indicates electronic components downstream from the reference photodiode. Blueindicates detectors. aser AttenuatorPhotodiodePulse generatorPi Laser controllerHV 90:10 splitter P y bu s | / / | C L A S C o ll a bo r a t i on M ee t i ng F i be r D i s t r i bu t i on S ys t e m (cid:135) La s e r pu l s e e x i t s l a s e r bo x t h r ough op t i c a l f i be r (cid:135) F i be r i n s e r t s t o x s p li tt e r (cid:135) S p li tt e r ha s r o w s o f ou t pu t f i be r s (cid:135) E a c h r o w c on t a i n s ou t pu t c onne c t o r s (cid:135) f i be r s f a ll w i t h i n de s i gna t ed i n s e r t i on l o ss r ange Fiber input
Splitter
Fiber outputs
Figure 3: The laser system in its rack-mountable case (left). The 1 ×
400 fiber splitter (right). for the TDC reference time and trigger (see Sec-tion 4).The remaining 90% of the pulse is sent to a cus-tom motorized attenuator by Oz Optics [13]. Theattenuator is used to vary the light pulse intensitysent to the detectors. It is operated by a dedi-cated 12V power supply and consists of two bladeswith a precise ultrasonic computer-controlled actu-ator, providing nearly continuous transmission from100% to 1%. The attenuated pulse is then sent tothe fiber distribution system.
To distribute the laser pulse to each scintillatorbar with similar amplitude and common timing,a custom SQS Vl´aknov´a Optika 1 ×
400 splitter isused. The splitter contains a series of three multi-lens arrays (MLAs). The first two shape the incom-ing laser pulse to a tophat profile. The third MLAfocuses this tophat beam into each of the 400 out-put fibers, which are equipped with LC connectors.The splitter is shown in Figure 3 (right).The uniformity of the splitter was tested by themanufacturer using a 405 nm laser. The resultingpower distribution is shown in Figure 5a as the per-cent deviation from the median power. Over 95%of the fibers had less than 20% variation in powerfrom the median fiber, meeting the stability guar-anteed by the manufacturer. Fibers outside of this range were not used for the calibration system andare not included in Figure 5a.An Lfiber 1 × × × × × × ± × After the splitter, the light from each fiber needsto be directed into the individual detectors. Ge-ometrical restrictions usually prevent placing thefiber such that if shines directly into the face of thescintillator. Specifically for BAND, following pre-vious work [4], the fiber is placed parallel to thescintillator between its surface and the optical re-flector wrapped around the bar [1].5 igure 4: GUI for remote control of the laser system. These panels show settings for the laser controller, the pulse generator,and optical attenuator, which can be adjusted remotely, without any need to enter the experimental area.
Several different fiber terminations were tested inan attempt to optimize the light yield and stabilityof pulses entering the detector. These included: • Fiber severed at 90 ◦ relative to cable direction. • Fiber severed at 45 ◦ relative to cable direction(so-called “side fire” termination) • Cylindrical diffuser tip used to diffuse light uni-formly over several centimeters. • Spherical diffuser tip used to diffuse light froma localized point. • Fiber exposed (no jacket or cladding) for lastfew centimeters prior to termination, allowinglight to escape the fiber.The various termination methods were tested usingthe secondary laser and attenuator. The attenuatedlaser pulse was split using the 1 ×
4. Scintillator test setup
A test-bench scintillator detector was set up totest and optimize the performance of the system.It consisted of a typical BAND scintillator [1] withdimensions of 7 . × . ×
200 cm , with an opticalfiber attached as discussed above. It is read out bytwo Hamamatsu R7724 PMTs [16] positioned onboth ends of the scintillator.The electronics setup used to process the PMTsignals is shown in Figure 2c. It consists of a LeCroy428F linear fan-in fan-out (FIFO) module, a LeCroy623B octal discriminator, a Philips Scientific 7946 N u m b e r o f c h a nn e l s (a) D e v i a t i o n f r o m m e d i a n p o w e r [ % ] (b) Figure 5: (a) Power distribution of the 1 ×
400 splitter output fibers. The horizontal axis indicates percent deviation from themedian power. (b) Power distribution of the small-scale splitter output fibers. The vertical axis indicates percent deviationfrom the median power. gate generator, and finally a CAEN V1290A TDCand CAEN V792 ADC [17, 18, 19, 20, 21]. TheTDC and ADC are in a VME crate that interfacesthe electronics with a computer recording the data.The photodiode reference signal is sent to theFIFO module, which provides one output to theADC and the other to the discriminator. The dis-criminator outputs are used for the TDC referencetime and the Level One Accept (L1A) of the VMEtrigger manager. The L1A output serves as a com-mon stop for all TDC channels and triggers the gategenerator to produce the gate signal for the ADC.The signals from the scintillator PMTs are alsosent to the FIFO modules, allowing the same PMTsignal to be processed for both the TDC and ADC.One FIFO output is sent to the discriminator, withthe output sent to the TDC. This allows the sig-nal arrival time of each bar to be compared to thearrival time of the reference signal. During calibra-tion with the laser system, the time difference ∆ t between the signal from the PMT and the signalfrom the reference photodetector is measured:∆ t = t P MT − t ref (1)A second FIFO output is sent to the ADC, with adelay added such that the signal coincides with theADC gate. The ADC signal allows a determinationof the energy deposited by the signal, used for atime walk correction (see discussion below). Thedigitized TDC and ADC signals are recorded on the computer for offline analysis. As the sensitivity of the PMTs extends below 300nm, they are capable of detecting light directly fromthe laser itself in addition to the induced scintilla-tion light. This effect would be problematic for cali-brations, since it would lead to a difference betweenthe PMT response to laser-induced and particle-induced scintillation light.We tested this effect using a 400 nm longpassfilter, which blocked light from the laser, but trans-mitted light produced by scintillation. We com-pared the PMT signals from a short 5 cm scintil-lator with and without the filter. We found no de-tectable difference, implying that the PMTs wereresponding to scintillation light, rather than directlaser light. A similar test was carried out using the405 nm laser and a 410 nm longpass filter, with thesame results.
5. Performance
A typical ∆ t spectrum measured with the lasercalibration system is shown in Figure 7. The meanand variance of a Gaussian fit to the spectrum yieldthe mean value of ∆ t and time resolution, respec-tively. Note that the time resolution is on the orderof 100 ps.7 Transmission(%) A D C c h a nn e l side-firediffuser Transmission(%) A D C c h a nn e l side-firediffuser Figure 6: Comparison of PMT measurements for the side fire and cylindrical diffuser terminations for the reference PMT (left)and scintillator-coupled PMT (right). Points represent the average of multiple measurements, with dashed lines indicating theminimum and maximum measurements used in the average.
One of the primary uses of the laser system is toallow a time walk correction (TWC) for the TDC.Leading-edge discriminators apply a fixed thresh-old value to an input signal, above which the inputsignal is transformed into a logical NIM pulse out-put. The timing of the output pulse correspondsto the time the input pulse passes the threshold,and thus become later for smaller pulses. Whena discriminator is used with a TDC, this leads to
Figure 7: ∆ t spectrum measured with laser system. timing measurements that depend on the height ofthe pulse. This is referred to as “time walk”, andmust be corrected to achieve better time resolution,independent of the signal amplitude.The TWC is obtained by varying the attenuationof the laser pulses delivered to the scintillator, effec-tively varying the amplitudes of the discriminatedwaveforms. This data can be used to produce atime walk curve (see Figure 8a) showing how tim-ing measurement from the TDC changes as a func-tion of the ADC signal A . The curve is fit with thefunctional form∆ t = 1 c A + c + c A + c (2)where the ∆ t is defined by Eq. 1. The c i are freeparameters of the fit. For each event, this fit isused to correct the ∆ t measurement as a functionof the corresponding ADC signal. The time walkcorrected curve is shown in Figure 8b. The strongdependence of ∆ t on ADC signal is eliminated withthe correction. This calibration is performed indi-vidually for each scintillator in the detector. Byrunning the laser at a sufficiently high repetitionrate ( ∼ kHz), this calibration can be performed in amatter of seconds.In addition to the TWC, this quick calibrationalso establishes an absolute reference time for all8 DC [a.u.] · t [ n s ] D (a) ADC [a.u.] · t [ n s ] D -2-1012345678 (b) Figure 8: (a) Initial time walk curve with with fit, showing the ADC dependence of the ∆ t measurement. (b) Corrected timewalk curve, showing no ADC dependence. TDC channels. Because the fiber optic cables fromthe splitter are all of equal length, each scintillatorreceives its laser pulse at the same time. Therefore,all cable delays between the PMTs and electronicscan be quickly accounted for by taking data withthe laser system.
6. Drift corrections
Over the course of an experiment, the perfor-mance of detectors and electronics can change ordrift due to effects such as temperature fluctua-tions, radiation damage, or normal wear and tear.Because the laser system maintains fixed timingthroughout the experiment, it can be used to cor-rect the effect these drifts have on timing measure-ments.To demonstrate this correction, the laser systemwas run at 500 Hz over the course of 1 hour. Tosimulate changes in timing due to drifts in the de-tectors or electronics, a delay was attached to theoutput signal of the PMT. Initially, the delay wasset to zero. At times (10, 30, 40) minutes, the delaywas set to (0.5, 1.0, 1.5) ns.Figure 9 shows how the delay changed the ∆ t measurements over the course of the hour. The res-olution of each ∆ t measurement is approximatelyconstant in time. However, if the measurements areintegrated over time, the total resolution becomeslarger as the delays are increased. The laser sys-tem can be used to correct for this change by sub-tracting the average ∆ t minute by minute. Figure10 shows that before the first delay, the total inte-grated resolution is nearly the same as the minute to minute resolution, approximately 0.1 ns. As de-lays are added, the total integrated resolution in-creases. After the correction is made, the total in-tegrated resolution remains the same as the minuteto minute resolution.
7. Laser signal stability
It was observed that the amplitude of the laserpulse delivered to each scintillator changes overtime. The cause of this drift was isolated to the1 ×
400 splitter. When the splitter is removed fromthe system (and the attenuator is used to repli-cate the amplitude reduction of the splitter), theamplitude remains constant. Moreover, the effectwas found to vanish when an identical test was per-formed with the 405 nm laser.As previously mentioned, the pulses from the 405nm laser carried significantly less energy than the355 nm laser pulses. To correct for this energy dif-ference, pulses from the 405 nm laser were com-pared to attenuated pulses from to the 355 nmlaser. Despite the differences in pulse length andfrequency, the PMT response from the induced scin-tillation is largely the same. The ADC was mon-itored over the course of 1.25 hours at 500 Hz totrack the changing amplitude of the system. Figure11 shows how the ADC changed for one scintillatorover the course of the run. The setup with the 355nm laser shows large wandering of the ADC ampli-tude. In the setup with the 405 nm laser, the ADCamplitude remains constant. The conversion fromADC channel to MeVee was calibrated using a Cosource.9 igure 9: ∆ t as a function of time. Each blue data point isobtained from the ∆ t spectrum over one minute of measure-ments. The central values and uncertainties of each pointare obtained from a Gaussian fit to ∆ t , as shown in Figure 7. Figure 10: ∆ t resolution as a function of time. The bluetriangles show the resolution of each individual ∆ t measure-ment. The green stars show the integrated ∆ t resolution.The red circles show the integrated ∆ t resolution, correctedfor the drifts in ∆ t .
8. Conclusion
A laser calibration for TOF scintillator arrayswas developed. The system was tested with botha 1 × ×
400 fiber-optic splitter, allowing ap-plication to both small- and large-scale scintillatorarrays. The laser system was successfully imple-mented in BAND to provide a time walk correction,absolute time calibration, and TOF drift correction.The ability to continuously monitor the laser pulseheights during data taking enables off-line correc-tions of both energy measurements and spacial dis-tributions of the detected particles. The system
Time [hours] A m p li t u d e [ M e V ee ]
355 [nm]405 [nm]
Figure 11: ADC signal as a function of time for the atten-uated 355 nm laser (blue triangles) and 405 nm laser (redcircles). currently has amplitude instabilities on the level of ±
2% (section 3.3) to ±
30% (section 7). Whilethis did not have any significant negative impact forBAND, which measures neutron momentum purelyfrom time of flight, this must be improved for thesystem to be deployed in detectors where amplitudecalibrations are also critical. One potential applica-tion is the LAD Experiment planned for JLab’s HallC [22], which will use the Large Acceptance Detec-tor (LAD) to detect recoiling spectator protons. InLAD, amplitude information from the scintillatorswill be crucial for reducing accidental coincidencebackgrounds. Therefore, it is a goal for future im-plementations to improve the amplitude stability ofthe system.This research was supported by the U.S. De-partment of Energy, Office of Science, Office ofNuclear Physics under Award Numbers DE-FG02-94ER40818 and de-sc0020240, the Israel ScienceFoundation under Grant Numbers 136/12 and1334/16, the Pazy Foundation, and the IsraelAtomic Energy Commission.
References [1] E. P. Segarra, et al., The CLAS12 Backward Angle Neu-tron Detector (BAND).URL https://arxiv.org/abs/2004.10339 [2] O. Hen, G. Miller, E. Piasetzky, L. Weinstein, Nucleon-Nucleon Correlations, Short-lived Excitations, andthe Quarks Within, Rev. Mod. Phys. 89 (4) (2017)045002. arXiv:1611.09748 , doi:10.1103/RevModPhys.89.045002 .
3] Teem Photonics STV-01E-140 pulse laser, .[4] E. Smith, et al., The time-of-flight system for CLAS,Nucl. Instrum. Meth. A 432 (1999) 265–298. doi:10.1016/S0168-9002(99)00484-2 .[5] D. Hasell, et al., The BLAST experiment, Nucl. In-strum. Meth. A 603 (2009) 247–262. doi:10.1016/j.nima.2009.01.131 .[6] Thorlabs NPL41B pulsed laser, .[7] Thorlabs PAF-X-18-PC-A FiberPort coupler, .[8] Thorlabs NB1-K08 mirror, .[9] Newport FM-1 fiber mode scrambler, .[10] Teem Photonics MLC-03A-DP1 laser controller, .[11] Siglent SDG 1032X pulse generator, https://siglentna.com/product/sdg1032x/ .[12] Thorlabs DET025AFC/M fiber-coupled silicon photodetector, .[13] OZ Optics custom attenuator, .[14] Lfiber fiber optic splitter, .[15] Thorlabs fiber optic splitter, .[16] Hamamatsu R7724 PMT, .[17] Teledyne LeCroy 428F linear FIFO, https://teledynelecroy.com/lrs/dsheets/428.htm .[18] Teledyne LeCroy 623B octal discriminator, https://teledynelecroy.com/lrs/dsheets/623.htm .[19] Phillips Scientific 794 gate generator, .[20] CAEN V1290A TDC, .[21] CAEN V792 ADC, .[22] O. Hen, L. Weinstein, S. Wood, S. Gilad, In MediumNucleon Structure Functions, SRC, and the EMC effect,Jefferson Lab experiment E12-11-107 (2011)..[22] O. Hen, L. Weinstein, S. Wood, S. Gilad, In MediumNucleon Structure Functions, SRC, and the EMC effect,Jefferson Lab experiment E12-11-107 (2011).