Front-end control system and precise threshold configuration of the v-Angra experiment
Mariana L Migliorini, Antonio Fernandes Jr, Joao C Anjos, Pietro Chimenti, Igor A Costa, Luis F G Gonzalez, Germano P Guedes, Ernesto Kemp, Herman P Lima Jr, Guilherme S P Lopes, Amaro S Lopes Jr, Rafael A Nobrega, Igor F Pains, Iuri M Pepe, Dion B S Ribeiro, David M Souza
PPrepared for submission to JINST
Front-end control system and precise thresholdconfiguration of the ν -Angra experiment Mariana L Migliorini, a Antônio Fernandes Jr, a João C Anjos, b Pietro Chimenti, c Igor ACosta, a Luis F G Gonzalez, d Germano P Guedes, e Ernesto Kemp, d Herman P Lima Jr, b Guilherme S P Lopes, a Amaro S Lopes Jr, a Rafael A Nobrega , a Igor F Pains, a Iuri M Pepe, f Dion B S Ribeiro, f David M Souza a a Universidade Federal de Juiz de Fora, Juiz de Fora, Brazil b Centro Brasileito de Pesquisas Físicas, Rio de Janeiro, Brazil c Universidade Estadual de Londrina, Londrina, Brazil d Universidade Estadual de Campinas, Campinas, Brazil e Universidade Estadual de Feira de Santana, Feira de Santana, Brazil f Universidade Federal da Bahia, Salvador, Brazil
E-mail: [email protected]
Abstract: The ν -Angra experiment aims to estimate the flux of antineutrino particles coming outfrom the Angra II nuclear reactor. Such flux is proportional to the thermal power released in thefission process and therefore can be used to infer the quantity of fuel that has been burned duringa certain period. To do so, the ν -Angra Collaboration has developed an antineutrino detector anda complete acquisition system to readout and store the signals generated by its sensors. The entiredetection system has been installed inside a container laboratory placed beside the dome of thenuclear reactor, in a restricted zone of the Angra II site. The system is supposed to work standalonefor a few years in order to collect enough data so that the experiment can be validated. The detector’sreadout electronics and its environmental conditions are crucial parts of the experiment and theyshould work autonomously and be controlled and monitored remotely. Additionally, thresholdconfiguration is a central issue of the experiment since antineutrino particles produce low energysignals in the detector, being necessary to carefully adjust it for all the detector channels in order tomake the system capable of detecting signals as low as those generated by single photons. To thisend, an embedded system was developed and integrated to the detection apparatus installed in thecontainer at the Angra II site and is now operational and accessible to the ν -Angra Collaboration.This article aims at describing the proposed embedded system and presenting the results obtainedduring its commissioning phase. Corresponding author. a r X i v : . [ phy s i c s . i n s - d e t ] J u l ontents ν -Angra measurement system 12 Front-end control system 3 Introduction
The use of embedded systems for remote control and monitoring of experiments has been increasingin several areas of research. Its application facilitates management and reduces cost of scientificexperiments, allowing measurement conditions to be controlled and monitored even over greatdistances. In the context of experimental particle physics, this need is even more present giventhat in many cases the access to the experiment site is restricted. The Neutrinos Angra ( ν -Angra)experiment [1–4] is part of a global nuclear-fuel safeguard effort [5], coordinated by the IAEA(International Atomic Energy Agency). Antineutrino detectors can be used to measure the thermalpower of reactors and are sensitive to the burn-up process of nuclear fuel. The ν -Angra experimentaims to develop a surface antineutrino detector to be used as a tool for monitoring the fuel burn-upprocess in nuclear reactors, making it possible to infer the amount of nuclear fuel used in eachenergy production cycle. The detector is currently in operation next to the dome of the Angra IIreactor in Angra dos Reis, Brazil, and tests and data analysis are in progress. The detection systemwas mounted inside a container in a restricted area, 25 m away from the reactor core. ν -Angra measurement system The measurement system installed in the container was designed to work autonomously. A localserver provides connection to the external network through IP tunneling and has access to all thesubsystems that make up the system. A program named
Run Control running in this local server isresponsible for organizing and executing data acquisition runs through the control and monitoring ofthe whole system. This program uses the TCP/IP protocol as a means of exchanging messages andcommands with the experiment subsystems. Figure 1 shows a schematic diagram of this system.– 1 – igure 1 . Schematic diagram of the ν -Angra system mounted few meters away from the Angra II nuclearreactor dome. In addition to the local server and the detector itself, the measurement system is composedof the following subsystems: sensor readout electronics (front-end modules) [6], acquisition andtrigger electronics [7, 8], data storage, high-voltage system from CAEN model SY4527 [9], a devicewith pressure and temperature sensors (BMP180) [10] and an embedded system. This embeddedsystem, highlighted in figure 1 (in black), and its threshold configuration procedure are the centralsubject of this article. Before going into details about them, it is important to describe the generalcharacteristics of the readout and acquisition electronics of the experiment.The detector is filled with water using the Cherenkov effect as the main process for particledetection. It can be divided into two subsystems: an inner detector called TARGET that defines afiducial volume for the search of antineutrino events, and an outer detection system called VETO,used to identify events produced by natural radioactivity particles. More details are given in Ref.[1, 2]. The whole detector is instrumented with 40 Photomultiplier Tubes (PMTs) model R5912 byHamamatsu [11, 12], 32 installed in the TARGET detector and 8 in the VETO, operating at a gainof 1 × electrons per photoelectron. Each PMT generates impulse signals that are sent to thefront-end modules, which are composed of Amplifier-Shaper-Discriminator (ASD) circuits. Theoutput of the amplification/shaping stage is subdivided into two branches, one that is sent to theacquisition system where signals are digitized and stored in-board to wait for a trigger decision,and another that is sent to the input of a discriminator, also part of the front-end board, where theimpulse signal is compared to a voltage level (threshold) to generate logic pulses which are finallysent to the trigger system. The two adjustable parameters of a front-end channel are its discriminatorthreshold and the pedestal (or offset) of the signal which is sent to the acquisition system. Each front-end module contains eight independent channels for reading out PMT signals, and each channelhas its own threshold and pedestal levels that can be configured independently of each other. Tocontrol all these pedestal and threshold values individually, each front-end module contains fourAD5645R Integrated Circuits (ICs) [13] connected to a single I C bus. Each AD5645R CI providesfour Digital-to-Analog Converter (DAC) channels of 14 bits. The embedded system uses customroutines to adjust and monitor all the 80 DAC channels and to monitor atmospheric pressure and– 2 –emperature of the experiment. These two tasks are motivated below. Atmospheric pressure and temperature: the internal volume of the container should be main-tained at a temperature below 25 o C for reasons of stability and protection of the components that arepart of the system. In addition, atmospheric pressure measurements can also help in understandingthe detector’s acquired data since the rate and energy of background events caused by cosmic rays[14] are influenced by it. Discriminator threshold and signal pedestal: threshold adjustment is of paramount importancedue to the fact that antineutrino events are in a low energy region, which may generate impulsesignals of low amplitude in some channels of the detector [2]. In order to detect antineutrinos withgood efficiency, the threshold of each channel must be adjusted to a level slightly above the electronicnoise signal. Given its importance, an accurate and precise threshold configuration procedure hasbeen proposed and implemented in the context of this work. Pedestal adjustment is used to give agreater dynamic range for the signals produced by the front-end circuit since they saturate wheneveran amplitude greater than 1 . V is reached at the output of the front-end amplification/shapingstage [6]. A negative pedestal should therefore be used for this purpose. Finally, threshold andpedestal values should be continually monitored to ensure that the established settings remain validafter weeks/months of detector operation. In this section, the modules that make up the embedded system designed to control and monitorimportant parameters of the ν -Angra experiment will be described. The front-end control system makes use of a low cost embedded system composed of a SingleBoard Computer (SBC) based on a
Raspberry Pi 3 Model B . Such a system is accessed by Ethernetand is connected to six front-end modules and to a BMP180 device via I C protocol. The BMP180is composed of a piezo-resistive sensor, an ADC, a control unit with E PROM and an I C serialinterface. A mezzanine board based on the TCA9548A IC [15] ( I2C MUX in figure 1) offers an I C bi-directional switch supporting eight channels which might operate in Standard mode (100 kbit/s)or
Fast mode (400 kbit/s). Such board provides several I C channels which are required to carryout the communication between the SBC and the external devices. Figure 2 on the left shows aschematic of this system.The application layer implemented in the embedded system can be divided into four groups asshown on the right of figure 2: (1) definition and management of TCP/IP messages ( TCP/IP API );(2) control and monitoring routines; (3) state machine; and (4) definition of I C commands( I2C API ). A TCP/IP interface was implemented locally in the SBC by means of a
Python scriptwhich uses the fundamentals of the
Socket
API [16]. The I C interface was implemented using the smbus module for Python allowing access to the system’s I C devices. Finally, a state machine,as represented in figure 3, was implemented to be used by the Run Control in the organization andexecution of data acquisition runs. – 3 – igure 2 . Hardware (left) and application layer (right) schematics of the embedded system.
Figure 3 . State machine for data collection management and description of its commands.
The main routines implemented by the control system are: (1) temperature and pressure measure-ments; (2) configuration and monitoring of signal pedestal and discriminator threshold values; and(3) threshold scan procedure. They are described below.
Atmospheric pressure and temperature measurements:
Pressure and temperature measure-ments are stored in a local database, in the embedded system. Each measurement is accompaniedby date and time in a format known as
Unix Timestamp . The
Run Control can configure themeasurement rate and access this database via TCP/IP whenever needed.
Pedestal and threshold configuration and monitoring: different values of pedestal and thresh-old may be necessary to create a greater understanding of the operational characteristics of thedetector and of the collected events. To make it possible, different configuration files, containingall the values of pedestal and threshold, were created (e.g.
PHYSICS1 , PHYSICS2 and
TEST ). Ac-cording to the message received via TCP/IP, the values specified in one of these files are uploadedinto the front-end registers to set pedestal and threshold values. From this moment on, registervalues can be monitored with a rate that can be defined by the
Run Control .– 4 – hreshold scan procedure: this routine was implemented to allow precise adjustment of thresh-old and noise monitoring for each detector channel. In this routine, the threshold is swept downwardsfrom an initial value well above the pedestal level and for each new threshold value, the output signalrate of the front-end discriminator is measured to form a signal rate vs threshold curve as illustratedin figure 4. It should be taken into account that, for this routine, the trigger system measures thenumber of times the noise signal crossed the threshold level, considering only the signal risingedge. The threshold scan produced curve (on the right side of figure 4) tends to be bell shapedwhose characteristics depends on the bandwidth of the noise signal and on the discriminator timingresponse [17].
Figure 4 . Representation of the threshold scan procedure and its resulting curve at right.
In this procedure, therefore, the threshold moves from top to bottom and as it approaches thesignal pedestal, the discriminator signal rate increases due to the presence of noise. The peak valueof the resulting curve can then be used to calibrate the channel pedestal relating it to a correspondingthreshold value. When this routine is applied with the detector turned on, it is expected to see othercomponents besides electronic noise due to the interaction of particles in the detector and theoccurrence of dark current pulses [18]. Section 3.2 will present a study of this technique in orderto understand better its characteristics and possible applications.
A test setup was assembled in laboratory to serve as a basis for the development and testing of theembedded system and its routines. Figure 5 shows a schematic diagram of this setup that includes,in addition to the embedded system, a computer, a dark box with a PMT (same R5912 model usedin the ν -Angra detector) powered by a Hamamatsu C9525 power supply [19], a front-end module,an FPGA module and an oscilloscope, besides the sensors provided by the BMP180 circuit.The signals generated by the PMT are sent to the front-end module, which in turn sends itsanalog output signals to the oscilloscope and its logic signals to the FPGA module. The lattercommunicates with the computer using the UART protocol through an USB-TTL converter circuitwhile the oscilloscope is controlled via TCP/IP. The FPGA module emulates the Trigger electronicsof the experiment while the oscilloscope replaces the acquisition module. The embedded systemcontrols the discriminator threshold and signal pedestal values of the front-end module and measuresthe atmospheric pressure and ambient temperature through the BMP180 device via I C . Therefore,– 5 – igure 5 . Schematic diagram of the setup assembled in laboratory for development and validation of thecontrol system and its routines. this setup emulates part of the measurement system installed at the Angra II nuclear power plant site,making it possible to develop the embedded system routines and test them before implementationin the experiment.Finally, the single photoelectron (SPE) spectrum has been measured based on the front-endoutput signal in terms of peak amplitude in mV units. The high-voltage of the PMT sensor wasset to 1510 V, which was the value provided by its manufacturer to make it to work with a gain of1 × , and a light-emitting diode was placed in front of the PMT cathode and properly pulsedby a waveform generator. The acquired data were fitted by a function made up of a sum betweenan exponential and a Gaussian function. The former models the noise component, while the lattermodels the SPE spectrum. The mean value and standard deviation of the SPE spectrum wereestimated at 79.5 ± ± Figure 6 . SPE spectrum measured at the front-end output. – 6 –
Results
This section is divided into three topics: (1) temperature and atmospheric pressure test measure-ments, (2) study of the threshold scan procedure performed in laboratory, and (3) its application tocalibrate and configure the detector channels at the nuclear reactor site.
In one day, the local atmospheric pressure fluctuates mainly due to the temperature variation andthe force of gravity exerted by the moon. Pressure fluctuations occur throughout the day, peakingaround 10AM and decreasing around 4PM. Figure 7 (left) shows that the installed pressure sensorhas sufficient resolution to detect this oscillation. This measurement was performed next to thedome of the Angra II plant, resulting in an average pressure of approximately 1 . × Pa .This measurement was compared with data collected in the region of Angra dos Reis by theNational Meteorological Institute (INMET) [20]. It can be seen from figure 7 on the left that themeasurements have the same oscillation period, as expected, with a very small difference betweenthem, which can be justified by the different locations where the sensors were installed. The samemeasurement was performed in the city of Juiz de Fora, Minas Gerais, at an average altitude of678 m, resulting in an average atmospheric pressure of 0 . × Pa . According to INMETdata, in Juiz de Fora the atmospheric pressure usually varies from approximately 0 . × Pa (summer) and 0 . × Pa (winter).The temperature sensor was installed next to the detector’s readout electronics, inside thecontainer and therefore under the influence of air cooling. Figure 7 (right) shows the relatedmeasurement for a full day of detector operation. As can be seen, the temperature peaked aroundnoon, reaching a value of 23.6 o C . Note that the high frequency oscillations of the curve occur dueto the air conditioning on-and-off process. These measures will allow the ambient temperature ofthe laboratory to be monitored continuously during the operation of the experiment. P r e ss u r e ( ⁵ PA ) INMETSensor BMP180 T e m pe r a t u r e ( ° C ) Figure 7 . Atmospheric pressure and temperature measurements considering a full day of detector operation.
In this section the threshold scan procedure will be studied in different PMT conditions; with itturned-off to generate signal with electronic noise only, and turned-on to check for other signal– 7 –omponents. For testing purposes, the pedestal level was set to 300 mV and the threshold valueswept in decreasing mode. The initial threshold value was programmed to be well above the pedestalvalue, that is, far from the noise region. For each new threshold value, the event rate at the outputof the front-end discriminator was measured by the FPGA module (see figure 5), forming a graphof signal rate vs threshold as described previously in section 2.2.For a first test, two noise situations were created. One with the front-end channel inputconnected to a 50 Ω termination (no use of PMT) and another with the PMT signal cable connectedto the front-end input but with the PMT power cable floating. In this last configuration, thehigh-voltage cable undergoes electromagnetic interference increasing the noise level arriving to thefront-end circuit. Figure 8 shows the signals at the output of the front-end for both cases. Thenoise standard deviation were measured to be 3.5 mV and 5.8 mV, respectively. Figure 9 showsthe resulting threshold scan curve for those cases. This curve can be used to calibrate the thresholdin relation to the signal pedestal and to measure and monitor noise throughout the experiment dataacquisition period. A m p li t ude ( V ) T i m e ( s ) T i m e ( s ) A m p li t ude ( V ) Figure 8 . Noise at the output of the front-end channel when its input is terminated with 50 Ω (left) and whenits input is connected to the PMT signal cable but with its high-voltage power cable floating (right). -20 -15 -10 -5 0 5 10 15 20 Threshold (mV) D i sc r i m i na t o r ou t pu t r a t e ( s - )
50 TerminationHV Connected
Figure 9 . Threshold scan curves for the two noise situations shown in figure 8. These curves were centeredat zero from their peak values. – 8 –n a different scenario, used to understand the threshold scan procedure in a situation wherethe PMT is turned-on, the PMT power cable was finally connected to the high-voltage power supplywhich was configured to provide a voltage of 1510 V setting the PMT gain to 1 × electronsper photoelectron, as used in the experiment. Figure 10 shows the resulting threshold scan curve.In addition to the electronic noise component, shown in figure 9, it is possible to identify twoother components: one in the region between 40 mV and 120 mV and another after 120 mV . Theformer occurs due to dark current events [18]. Its causes include thermionic emission, field effectsand leakage currents. The characteristics of this process depend particularly on the compositionof the cathode and, throughout the usual range of supply voltage, the average amplitude of thesignals generated by this effect varies proportionally with the gain of the PMT [21]. Therefore, theidentification of the dark current events could be used to monitor the gain variation of the detectorPMTs. The latter is related to cosmic ray events that generate photons through Cherenkov effect[22], producing signals with a wide range of amplitude values. ⁰ ⁴ ⁵ ⁶ D i sc r i m i na t o r O u t pu t R a t e ( s ¹ ⁻ ) HV Connected
COSMIC RAYSDARK CURRENT
Figure 10 . Threshold scan curve obtained with a PMT powered with 1510 V and connected to the front-end input. The internal graph corresponds to one of the measurements presented in figure 9, but with thehorizontal and vertical axes changed to allow a direct comparison with the curve shown in the main frame.
The main error sources of the threshold scan measurement are the DAC integral nonlinearityerror which can reach, in the worst case, 4 × LSB or, equivalently, 1.2 mV for the AD5645R chip[13], and its gain error which can get to a value of -0.05% of the Full Scale Range (FSR) for atemperature of 100 o C, also considering the worst case. The front-end electronics uses a FSR of 5 Vfor its DACs, causing the gain error to reach a value of up to -2.5 mV. At a scale of 100 mV, whichrepresents a value closer to that used for the front-end thresholds, this error reduces to -0.05 mV. Thethreshold scan procedure cancels the DAC error component known as offset error and, therefore, itshould not be taken into account. Finally, the uncertainty given by the threshold scan voltage stepis approximately ± The threshold scan procedure was applied to all the ν -Angra detector channels. The resulting curvesare shown in figure 11 (left). Before the threshold scan procedure was available, the experimentthresholds were set assuming that the pedestal values were about the same for all the front-endchannels. Consequently, a single threshold value of -340 mV was set for all the detector channels inorder to ensure that all threshold values were outside the noise region. However, as shown clearlyby the threshold scan curves, a relatively large variation of the effective thresholds (threshold minuspedestal value) was observed, with minimum and maximum values of approximately 25 mV and50 mV. These values correspond to 0.31 ± ± ± ± ± ± Figure 11 . Threshold scan curves before (left) and after (right) centering them in relation to their peak valuesfor the TARGET detector channels.
Figure 12 allows to analyze the impact of setting all the effective thresholds to 25 mV. It showsthe energy distribution (in ADC counts) for two channels of the TARGET detector where it ispossible to observe that more signals with amplitude in the SPE region, below 10 ADC counts,were detected due to the increased efficiency in detecting SPE events (the ν -Angra ADC resolutionis 9.8 mV/count). – 10 – igure 12 . Energy spectrum (in ADC units) for two of the TARGET detector channels before and afterconfiguring their thresholds with the threshold scan procedure. A measurement of the probability density function of occurring an event with energy higherthan a given threshold, taken for different threshold values, is shown in figure 13. Two situationswere considered: before and after using the threshold scan procedure to set the thresholds. Onlyevents which fired five or more channels were considered. Such coincidence requirement is usedby the ν -Angra experiment to reject events that might occur due to dark current pulses. As canbe seen, setting the thresholds of the front-end electronics discriminators based on the thresholdscan procedure increases the number of low energy events around the SPE amplitude region byapproximately 28%. Figure 13 . Measurement of the probability density function of occurring an event with energy higher than agiven threshold, for different threshold values, considering 60 hours of data acquisition. Only events whichfired five or more channels were considered.
The results presented in this section validate the proposed threshold configuration procedureshowing that it is able to increase the probability of detecting low energy events, in the regionof interest for antineutrino detection, providing a way to improve the overall performance of the ν -Angra detector. – 11 – Conclusions
The main objective of this work was to describe the development of an embedded system aimed atthe control and monitoring of fundamental parameters of the antineutrino detector of the ν -Angraexperiment. The results obtained during the development and commissioning phases and theimportance of the developed system in the context of the application were presented. The calibrationand configuration of the detector’s thresholds were treated as a central issue in this article sincethe events of interest for the experiment are at low energies, making it important to maximize theexperiment’s efficiency in detecting SPE signals. This embedded system has been in use at thenuclear reactor site since the beginning of 2019 and all the detector channels have been configuredusing the threshold scan procedure. It was able to set all the thresholds of the inner detector channelsto 25.0 ± ± ± ± Acknowledgments
The authors would like to thank the Ministry of Education, the Ministry of Science, Technologyand Innovation (MCTI) and the Funding Authority for Studies and Projects (FINEP) for supportingthe Neutrinos Angra experiment whose initiative took place in 2007. We are also thankful to theFoundation for Research Support of the State of Rio de Janeiro (FAPERJ) and the Foundation forResearch Support of the State of Minas Gerais (FAPEMIG) which, in 2011, approved the projectTEC-APQ-02287/11 financing part of what was implemented at the Federal University of Juiz deFora (UFJF). Finally, we thank the UFJF University for the excellent structure and support providedfor the execution of this project.
References [1] J. Anjos et al.,
Status and perspectives of the Neutrino Angra project , NuInt12: eighth InternationalWorkshop on Neutrino-Nucleus Interactions in the Few-GeV region , Rio de Janeiro, Brazil, 2012.[2] J. C. Anjos et al.,
Nuclear and Particle Physics Proceedings
Nuclear Instruments and Methods in Physics Research A
830 (2016) 206-213[4] H. P. Lima Jr. et al 2019 JINST 14 P06010.[5] N. S. Bowden 2008 J. Phys.: Conf. Ser. 136 022008.[6] T. I. Dornelas et al 2016 JINST 11 P07018.[7] H. P Lima Jr, L. Andrade, L. Gomez, and R. Gama,
Data acquisition with optimal pulse amplitudeestimation for a neutrino detection experiment , CBPF Technical Notes , vol. 4, no. 2, pp. 34-41, 2014.[8] J. M. S. Souza et al.,
A FPGA-based trigger and DAQ system for Neutrinos-Angra detector , Workshopon Circuits and Systems , Belo Horizonte, Brazil, vol. 1, pp. 1-6, 2016.[9] CAEN,
SY4527/SY4527LC Universal Multichannel Power Supply Systems 16/10 slots Mainframes ,Rev. 20, 2020.[10] Bosch Sensortec,
BMP180 Digital pressure sensor , Rev. 2.5, 2013. – 12 –
11] Hamamatsu Photonics,
Large photocathode area photomultipliers tubes , 2019.[12] Y. Zhang et al 2016 JINST 11 P06016.[13] Analog Devices,
Quad 12-/14-/16-Bit nanoDACs with 5 ppm/°C On-Chip Reference - I2C Interface ,Rev. F, 2002.[14] P. N. Diepet al.,
Dependence of the cosmic muon flux on atmospheric pressure and temperature , Communications in Physics , vol. 14, no. 1, 2007.[15] Texas Instruments,
TCA9548A Low-Voltage 8-Channel I2C Switch with Reset , Rev. H, 2019.[16] W. J. Chun,
Core Python Programming , Prentice Hall , 1 ed., 2000.[17] L Anderlini et al 2013 JINST 8 P06001.[18] J. A. Baicker,
Dark Current in Photomultipliers , IRE Transactions on Nuclear Science, vol. 7, no. 2-3,pp. 74-80, 1960.[19] Hamamatsu Photonics,
Compact bench-top type regulated DC power supply C9525 series , 2007.[20] Instituto Nacional de Meteorologia (INMET). . Accessed: 2020-05-21.[21] K. Lung et al.,
Nuclear Instruments and Methods in Physics Research A
696 (2012) 32-39[22] M. A. Mostafá,
The high-altitude water cherenkov observatory , Brazilian Journal of Physics , vol. 44,no. 5, pp. 571-580, 2014., vol. 44,no. 5, pp. 571-580, 2014.