SciELO - Scientific Electronic Library Online

 
vol.66 número4Configuración y montaje de instrumento para la caracterización magnetoeléctrica de compuestos cerámicos multiferroicosThermal properties and degradation kinetics of epoxy-γ -alumina and epoxy-zinc oxide lightweight composites índice de autoresíndice de materiabúsqueda de artículos
Home Pagelista alfabética de revistas  

Servicios Personalizados

Revista

Articulo

Indicadores

Links relacionados

  • No hay artículos similaresSimilares en SciELO

Compartir


Revista mexicana de física

versión impresa ISSN 0035-001X

Rev. mex. fis. vol.66 no.4 México jul./ago. 2020  Epub 31-Ene-2022

https://doi.org/10.31349/revmexfis.66.475 

Research

Instrumentation

Triggerless charge measurement system for fast characterization with avalanche photodiodes and photomultipliers

G. Tejeda Muñoza 

M. del C. Grados Luyandoa 

A. Fernández Télleza 

M. Rodríguez Cahuantzia 

M. I. Martínez Hernándeza 

B. de Celis Alonsoa 

aFacultad de Ciencias Físico Matemáticas, Benemérita Universidad Autónoma de Puebla, 72570 Puebla, México.


Abstract

Particle detectors based on the response of sensitive materials (plastic scintillator, saturated gas, etc.) usually need characterization and test procedures before final installation. Cosmic ray particles are normally used to perform those preliminary tests which include a detailed inspection of the readout electronics and the acquisition of the charge distribution of cosmic ray detection. The Data Processing Interface we propose is implemented in an field-programmable gate arrays (ALTERA family), and is designed to acquire digitized signals from light sensors, of which photomultipliers and avalanche photodiodes are the most common ones. The architecture is based on a medium density field-programmable gate array that continuously reads the data coming from a 10 bit, 40 MHz analog-to-digital converter. Input data is stored in a dual port memory designed to search for valid pulses and compress them by removing data below a programmable voltage threshold. The interface can produce two types of data packets, non-zero and empty packets. Data in non-zero packets is compressed with a lossless technique and marked with a start of data, a time stamp, valid data and data size information for reconstruction purposes. Empty events are generated when the maximum waiting time for a valid pulse is exceeded and information is added to preserve time continuity.

Keywords: Front-end electronics for detector readout; analogue circuit electronic; FPGA data processing

PACS: 07.05.Hd; 29.40.Wk

1.Introduction

The Data Processing Electronics Interface described here is designed to acquire and process 10-bit signals for avalanche photodiodes (APDs) [1] with a metal resistor semiconductor and a Hamamatsu R6095 photomultiplier (PMT) [2] integrated in a plastic scintillator particle detector. The interface’s architecture contains several blocks to avoid typical perturbations due to temperature variation in electronics, detectors, and induced noise. The data processor system is able to perform a continuous read out, identify a valid signal pulse, reject glitches, and compress the data without any loss of information. The signals are compressed in the data compression block which removes all data below a predefined voltage threshold and identifies a pulse when three consecutive data values are over the threshold. Each pulse is tagged with a time stamp, number of samples, and event number for reconstruction purposes. All pulses are formatted and stored in a multi-event buffer with a bandwidth of 200 Mbytes/s. In continuous mode, the maximum input data rate is 40 MHz, where all the signals coming from the analog-to-digital converts (ADCs) are stored in a dual port memory. The maximum read frequency for the dual port memory is 300 MHz. The interface can generate a calibration signal that allows to detect errors in any of the blocks. The system has been tested with particle detectors based on APDs and photomultipliers (PMTs) triggered with atmospheric muons. The interface is designed to process large data volumes without any loss of information. The output signal is the same regardless a PMT or an APD is connected to the input of the interface. The maximum rate expected for a single channel is about 260 bits/event with up to eight channels allowed in a single FPGA interface.

2. Materials and methods

A front-end electronic card (FEE) is designed to digitize the signals from the PMTs and APDs. This FEE, in Fig. 1, is composed by a charge preamplifier, a 4th order shaping amplifier and a 10 bit 40 MHz ADC.

Figure 1 Front-end card to read out the APD and PMT detector signals. 

Up to eight FEE cards can work in parallel connected to a single interface card. Since the output data are 10-bit wide, all of the FEE signals would require 80 inputs and 8 dedicated output from the field-programmable gate arrays (FPGA), which are the maximum number of I/O’s on it. The 8 dedicated outputs are required to distribute the clock signals for each FEE to synchronize the operation and data transfer without jitter issues. The data processing interface card we designed is a 9.5×9 cm2, 8 layer PCB. The card is equipped in the top side with four header connectors for up to 185 I/O, 3 power supply connectors, one 100 MHz SMD oscillator, one I2 C configurable oscillator, three power regulators, a configuration memory, and a microcontroller for the USB interface control. The system is based on an ALTERA Cyclone V FPGA (5CEBA2F23C8N). An 8 layer PCB (see Fig. 2] was specifically designed to mount the FPGA, the components to process the digitized signals and a microcontroller to drive the USB interface for data transfer.

Figure 2 Interface Board. The FPGA is at the center and the microcontroller is on the right. 

The firmware of the interface consists of the five main blocks shown in Fig. 3: a) the Ring Buffer is a dual port memory [3] of 8192×10 bits with a write frequency of 40 MHz and a read frequency of 80 MHz; b) the Pulse Finder is a pulse scanner, it is used to identify a valid pulse and reject the signals without valid information; c) the Data Format receives a valid pulse and adds a time stamp, data length and data number. This block also generates an empty event to preserve time continuity whenever no pulse is processed for a long time; d) the Multi-Event Buffer is based on a first-in first-out (FIFO) [4] memory where the pulses processed by all channels are temporarily stored before they are sent to an external permanent storage system, the read-out frequency is determined by the FIFO memory, the maximum transfer speed is 200 Mbytes/s; e) the Main Control synchronizes the operation of the blocks, receives the configuration parameters, and distributes the sequences to start and stop the data processing.

Figure 3 Schematic diagram of the Interface Board firmware. There are three clock domains controlled by an internal Phaselocked loop (PLL) in the FPGA. All the blocks run in a clock domain two times faster than the writing frequency of the Ring Buffer to assure there is always free space for continuous data acquisition. 

2.1.Ring Buffer

The Ring Buffer is a dual-port memory designed to continuously store all the digitized signals in order to avoid loss of information. The read access speed is always twice the write access speed to ensure availability of memory space to continuously write data. The maximum read and write access of the dual port memory is 300 MHz but we are running only up to 80 MHz. There is one Ring buffer for each channel while the maximum number of channels allowed is eight.

2.2.Pulse Finder

The Pulse Finder compares each channel’s data from the Ring Buffer with a programmable threshold generating a hit and identifying a signal whenever it detects three consecutive samples over the threshold. This condition allows to discriminate valid pulses from any glitch.

2.3.Event format

The Data Format block constructs the internal data format for valid pulses and empty events. The data format for a valid pulse is composed of a start of event value that identifies each pulse, the channel number, a time stamp with respect to the previous pulse, the payload, the data length, and the number of data. For an empty event, the data format is composed of a start of data value, the channel number, time stamp, the data length, and the data number. The empty events are inclu-ded only if the maximum waiting time for a valid pulse has elapsed; the purpose of this is to preserve time continuity and to identify the exact time when a muon is detected with respect to the internal clock of the FPGA. Figure 4 shows the internal data format for empty and valid events.

Figure 4 On the left, the data packet for a valid pulse and, on the right, the data packet for an empty event. 

2.4.Multi-Event Buffer

The Multi-Event Buffer block is an 8192×10 bits FIFO memory that formats and prepares the data from all active channels to be sent to a permanent storage system. All data is stored including empty events. The memory is read out at a maximum transfer speed of 200 Mbytes/s through a USB port driven by a microcontroller [5].

2.5.Main Control

The Main Control block receives and distributes the configuration settings for the data processing. The distributed parameters are the active channels, the reference threshold, the start and stop data process commands and the auto-test procedure command. The Main Control block synchronizes the operation of all the modules and triggers the sequences to process the data when specific signals are generated by each module.

3.Results

Digitized signals from cosmic rays in plastic scintillating detectors [6] are used as data source to test the interface; these detectors are the same used for the cosmic ray detector (ACORDE) [7]. The digitized signals come from a 10 bit 40 MHz ADC. The analog to digital conversion is activated by the Main Control Block. The maximum input signal rate that the system can handle is 10 KHz for both PMTs and APDs. The data processing interface is able to acquire digital signals through the implementation of the following functions in the FPGA’s firmware: clock distribution, configuration for the data acquisition, data formatting, and transfer to an external storage system. The system can also generate a test signal for calibration purposes. The signals from the ADC are 650 ns wide and are sampled at a rate of 20 MHz. Figure 5 shows the user interface of the acquisition system.

Figure 5 Data acquisition user interface for system configuration and control. 

The connection button between the interface and the data acquisition system (DAQ) system is indicated with the number 1 in the figure, number 2 indicates the discriminator voltage input field for all the acquisition channels, number 3 is the button to enable the data storage in a specific file, and number 4 is the graphic area to display all the data from the FEE cards.

Besides processing and storing all the digitized signals, the maximum value of each signal is also measured and recorded. Figures 6 and 7 show the amplitude distribution for APD and PMT signals, respectively. For APDs the threshold was set to 4 ADC counts, the duration of the test was 57 minutes and the optimum bias Voltage 35.5 V. For the PMT, the discrimination threshold was 4 ADC counts, the duration of the test was 70 minutes and the optimum operational Voltage was 700 V.

Figure 6 Amplitude distribution of the APD signals. 

Figure 7 Amplitude distribution of the PMT signals. 

A test signal of amplitude 714 ADC counts is stored in the firmware. One hundred thousand events with this test signal were injected to the FPGA during a period of 14 hours. The amplitude distribution of these events is shown in Fig. 8 where it is clear that the system does not register events of an incorrect amplitude.

Figure 8 Reconstruction of the maximum value for the internal test signal in the firmware. 

4.Discussion

We can observe that the threshold level was the same for the PMT and the APD. It is usually expected for APDs to be noisier than PMTs; nevertheless, the FEE cards are able to drive the signals from both detectors. When an interruption occurs in the data transfer, the processing interface pauses the operation, only to be resumed when the resources become ready. During this pause period, the processing interface does not interrupt the data acquisition from the ADCs preventing any data loss. In addition, if the USB data transfer is stopped, the interface is capable of taking data for 1 more minute, since the Ring Buffer is able to write data even if the reading part is paused. Finally, the power consumption of the card in standby mode is 100 mA while for each active channel the card consumes 80 mA. Only one 5 V symmetrical power supply is needed to power the system.

5.Conclusions

The FPGA based Processing Interface we designed and built was successfully used to read-out, discriminate, digitize, and transfer signals produced by PMTs and APDs coupled to scintillating plastic detectors. This board is the first prototype based on a FPGA for application in cosmic ray and high energy physics. This interface is an excellent, easy-to-use option for scintillation detectors characterization. Its reduced size makes it ideal for applications where the space and access to the detector area are restricted. The Processing Interface is a novel way for reading out and store data from plastic scintillator based detectors. It is also well suited to test and verify the correct performance of a specific section in a multichannel or single channel detectors. The Pulse Finder module included in the firmware makes an external trigger detector unnecessary. A limitation of this prototype is that only eight channels were implemented for multichannel applications though this is clearly a drawback that can be overcome.

References

1. A. Akindinov, et al., Detection of Ligh Pulses Using an Avalanche-Photodiode Array with a Metal-Resistor-Semiconductor Structure, Intrum. Exp. Tech. 48 (2005) 355, https://doi.org/10.1007/s10786-005-0063-5. [ Links ]

2. Hamamatsu Photonics, “Photomultiplier Tubes”, accessed June 15, (2019). https://www.hamamatsu.com/resources/pdf/etd/PMT_TPMZ0002E.pdf. [ Links ]

3. G. Yijun and W. Zuo, Mapping N-Port Memory with Dual-Port Array, in Proceedings of the World Congress on Computer Science and Information Engineering, Los Angeles, 2009, edited by IEEE Computer Society (CSIE, Los Angeles, 2009), p. 444, https://doi.ieeecomputersociety.org/10.1109/CSIE.2009.889. [ Links ]

4. S. Windmann and J. Jasperneite, An FPGA based FIFO with efficient memory management 2015. in Proceedings of the IEEE 20th Conference on Emerging Technologies and Factory Automation (ETFA), Luxembourg, 2015, edited by IEEE (IEEE, Luxembourg, 2015), https://doi.org/10.1109/ETFA.2015.7301585. [ Links ]

5. Universal Serial Bus, “Specifications”, accessed May 7, 2019. https://www.usb.org/document-library/usb-32-specification-released-september-2 2-2017-and-ecns . [ Links ]

6. P. Akhilesh et al., A simple technique for gamma ray and cosmic ray spectroscopy using plastic scintillator, Nucl. Intstrum. Methods Phys. Res. A 824 (2016) 606, https://doi.org/10.1016/j.nima.2015.06.051. [ Links ]

7. E. Cuautle, et al., The ALICE cosmic ray detector, in Proceedings of the 30th International Cosmic Ray Conference, Mexico City, 2008, edited by R. Caballero et al. (UNAM, Mexico City, 2008), p. 1201. [ Links ]

Received: January 21, 2020; Accepted: March 27, 2020

Creative Commons License This is an open-access article distributed under the terms of the Creative Commons Attribution License