3.1　 framework of software

The relationships between the four parts are shown in Fig. 2. The Controller_module consists of two parts: DAQ_Control and HV_Control. DAQ_Control is responsible for sending data acquisition instructions and setting data acquisition parameters. The corresponding instructions are transmitted to the shared memory, received, and executed by the DAQ. HV_Control is responsible for setting the working parameters of high-voltage devices. Similarly, the parameters are written into the shared memory and read by HV_Control. After HV_Control receives the working parameters from the shared memory, the high-voltage module DT5533 is initialized with the setting value, and the corresponding voltage is applied to the detector. Each time HV_Control completes a voltage addition operation; it feeds back the status parameters to the Controller_module. The control function of the Controller_module is realized by calling the shared memory class and the control GUI class defined in the software. The parameter passed to the shared memory class is the defined structure, which includes the working parameters of the high-voltage devices, the data acquisition instructions, and the corresponding feedback parameters. The implementation of HV_Control is based on the shared memory class and HV_class defined in the software. The HV_class calls the CAENHV Wrapper library, which is a software library for power supply control, issued by CAEN to realize the control of DT5533.

Fig. 2Full-screen View

Frame diagram of software

The data acquisition module (DAQ_module) receives commands from the Controller_module via shared memory. When the DAQ_module receives the start command, it controls the VME bus to acquire the data. After each data acquisition loop ends, an increasing voltage signal is fed back to the HV_module which waits for the next instruction from the Controller_module. Once the trigger logic triggers a physical event for data acquisition, the events are recorded in the register of the front-end electronics (V785, V775, MADC-32,etc.), and then read by the external first-in first-out (FIFO) memory. Finally, the data in FIFO are transmitted to the PC via optical fiber and stored in the hard disk in binary format for future offline analysis and query. Simultaneously, the raw data will also be transmitted to shared memory for online monitoring through the write buffer. Alternatively, the data can also be acquired by a digitizer [27]. The input pulses, which come from detectors coupled with a charge-sensitive preamplifier (such as HPGe, silicon, etc.) or detectors coupled with PMT, SiPM, etc.,. are digitized into many samples by the flash ADC. Based on the sampling frequency of the digitizer, users can set the number of sampling points according to the width of the waveform by the configuration file. For example, the sampling frequency of DT5720 is 250 MS/s which means sampling a point every 4 ns if a waveform with a width of 1μs, 250 points were sampled. Each sampled data set can be analyzed to extract the information of time, energy, and so on using the software. Because of the differences in the interface, the base class of the data acquisition modules in the software is divided into two types: the base class of the module with the VME interface and USB interface. The modules with VME_interface, such as ADC (V785, MADC-32, etc.), TDC (V775, MTDC-32, etc.), and QDC (V792, MQDC-32, etc.) are inherited from the base class of the VME interface. The flash ADC contains both modules with a VME interface and a USB interface, which are inherited from two different base classes. Although the flash ADC is divided into two categories, the basic functions are the same. The obtained data are transmitted through the shared memory class, and the parameters passed to the shmget function in the shared memory class are transferred as an array of float type or int type.

The Online_module is mainly composed of two parts: monitoring and analysis. Periodically (every 10 s in the present test system), raw data stored in the front-end electronic registers (e.g., the event number, time stamp, and each channel’s energy) can be monitored by online processes through the shared memory mechanism (Online, Fig. 2). The received raw data has a defined data structure, such as the existence of a header(including channel, ID, event count, etc.), datum, and end of event mark. Therefore, to achieve an online monitoring function, it is necessary to call the RawData class (defined in the software) to extract the necessary information from the raw data. In addition to the real-time display of the spectrum of the obtained data, the Online_module can automatically select the analysis algorithm to search for the peak position and set the fitting parameters to fit different spectra according to the radioactive source used in the test. The analysis function of the Online_module first converts the binary data into a ROOT format file after the data acquisition ends, and then analyzes the ROOT file using the corresponding algorithm. During detector unit tests, all data acquisition parameters and analysis processes are saved as a log file for future queries. Furthermore, the analysis results of energy resolution, peak position, and other measurements will also be stored in the disk in txt format.

3.2　 Graphical user interface (GUI)

To facilitate user operation, a graphical user interface(GUI) was designed. The GUI is divided into two parts: the main control interface and the online monitor and analysis interface. The GUI was compiled based on ROOT under the Linux system. As shown in Fig. 3, the Controller_module serves as the hub for instruction transfer and parameter adjustment. It is composed of two parts: DAQ_Control and HV_Control, which are associated with the DAQ_module and HV_module, respectively.

Fig. 3Full-screen View

Interface of Control’s GUI:The figure shows the main control interface, through which all the acquisition commands and working parameters can be completed.

The DAQ_Control part of the GUI shown on the left side of Fig. 3(a), contains the function of setting the data acquisition parameters. Users can run the start/stop command via the button in the top "control" region. In the middle "setup" region, users can set the file name header and running time of every data acquisition. The high voltage value added to the detector is automatically used as a supplement to complete the file name. All raw data were stored in binary format. The information region shows a simple log of the test system. The HV control part shown on the right side of Fig. 3(b), allows users to control the high-voltage channel on/off in the left option box, as well as select the current monitoring range and accuracy. The right part of the HV control shows two high-voltage channel control interfaces. The control interface of each channel displays the default working parameters (modifiable) of DT5533, the monitoring voltage, and current values of the detector. The bottom of each control interface was used to set the voltage range and step interval of the detector test.

The main interface of the Online_module contains an online monitor and analysis function. The monitor part reads the raw data from the shared memory in real time. The obtained raw data are decoded first, and then transmitted to the online monitor for spectrum display. The monitored spectrum was updated every ten seconds. Moreover, the monitored channels of each module can be selected in the online GUI. There are also some buttons with specific functions on the online GUI interface. For example, the Online_module can realize an automatic data analysis function through the button "Analysis" on the online GUI. The details of the automatic data analysis function are presented in the next section.

4.1　 Structure of CsI detector and test

The CsI(TI) crystal of each detector unit of the detector array is a hexagonal prism with a length of 200 mm, and the distance between the opposite sides of the hexagon is 55 mm. The crystal was produced by the Shanghai Institute of Ceramics, Chinese Academy of Sciences, and covered with 1 mm polytetrafluoroethylene(PTFE) and 2 mm thick aluminum as shielding materials. The photomultiplier tube connected to one side of the crystal was R1828-01 (Hamamatsu). The detailed construction of the detector unit can be found in Ref. [19].

To verify the feasibility of the test system, two types of radioactive sources, ¹³⁷Cs and ⁶⁰Co, were used to test the performance of the CsI(TI) detector unit. During the test process, the measurement voltage range was set from 1600 to 2100 V, and the measurement interval was set to 100 V. The acquisition time for each condition was set to 10 min. When the Controller_module sends the start instruction to the DAQ_module, it enters the acquisition loop and waits for the feedback signal from the high voltage module (HV_Control). The HV_Control module receives the working parameters from the shared memory. After receiving the corresponding parameters, a high voltage was added to the detector at an initial value of 1600 V. During the voltage raising process, HV_Control feeds back the control status parameters of DT5533 control in real time. Until the voltage-raising operation is completed, DAQ_module starts to control the VME bus to obtain a set of data under 1600 V. When the running time reaches the set time of 10 min, DAQ_module ends the acquisition loop and returns a raising-voltage signal to HV_Control again. After receiving this signal, HV_Control will raise the voltage to the next target value at a measurement interval of 100 V and feedback the status parameters to control in real time. Then, the Controller_module sends the start command again, and the DAQ_module controls the VME bus to obtain another set of data under 1700 V. The above steps are repeated until the test system obtains the data under 2100 Vat.

Figure 4 shows the energy spectrum of ¹³⁷Cs and ⁶⁰Co obtained by CsI(TI) detector under many different voltages, which are acquired by ADC module V785 or MADC-32. As mentioned in Sect. 3.1, the digitizer, such as DT5720, has been added into the test system. Therefore, the data can also be acquired by digitizer. In this study, the trigger signal of the digitizer is generated by a self-trigger mode. The raw waveform is recorded as 300 samples according to the actual width of waveform output by the CsI detector. Although a single raw data file obtained by DT5720 will take up a large amount of storage space, the advantage is that it retains all the information of the output signal of the detector. Several algorithms have been integrated into the test system to extract energy and time information from sampling data. For example, users can extract the time information via the method of constant fraction discrimination (CFD) or leading-edge discrimination (LED). Since the spectrum obtained by DT5720 is consistent with that obtained by V785 and MADC-32, it is no longer presented in this paper.

Fig. 4Full-screen View

Energy spectrum of CsI(TI) detector under ¹³⁷Cs and ⁶⁰Co radioactive sources. (a)¹³⁷Cs 1600 V-2100 V; (b) ⁶⁰Co 1600 V-2100 V.

4.2　 Data online analysis

During the test process of the detector arrays, a great amount of data is acquired resulting in large files. This can result in a time-consuming and tedious offline data analysis. Because the spectrum of radioactive sources used for detector tests is usually known [30], it is possible to program automatic peak searching and spectrum fitting according to different radioactive sources. Users then only need to declare the appropriate radioactive source in the file name, and the Online_module can select the corresponding algorithm to process different raw data.

Typically, the raw data obtained by the analog-to-digital conversion module (ADC, TDC, etc.) are a series of integer values, called channel numbers, which are determined by the resolution of the module. This needs to be converted into the corresponding energy by an energy calibration. Generally, the calibration function of the scintillator detector is nonlinear, and the calibration functions of the silicon detectors and HPGe detectors are linear. Therefore, at least three calibration points are required for the energy calibration. In this study, ¹³⁷Cs and ⁶⁰Co sources were used to test the performance of the CsI detector. There are two peaks in the γ-spectrum of ⁶⁰Co, and the corresponding energies are 1.173 and 1.332 MeV, respectively. There is one peak in the γ-spectrum of ¹³⁷Cs, and the corresponding energy was 0.662 MeV. It meets the minimum requirement for energy calibration.

For online monitoring and analysis, the raw data must first be decoded, and then the binary file converted into the ROOT file. After this, the Online_module fits and searches for the peak of the spectrum to obtain the channel numbers corresponding to the peaks of the two types of radioactive sources under various voltages, as shown in Fig. 4. Generally, the fitting methods for the ¹³⁷Cs and ⁶⁰Co spectra are different. Because the energy spectrum of ¹³⁷Cs has an exponential background and a single Gaussian peak, the fitting algorithm is carried out using a single Gaussian plus an exponential function. However, for the spectrum of ⁶⁰Co, the fitting algorithm adopts double Gaussian function plus an exponential function because of the existence of double Gaussian peaks and an exponential shape. The corresponding fitting algorithm is integrated into the software. Users only need to declare a radioactive source in the file name.

After fitting the spectrum, the calibration curve of the two sources under different voltages can be obtained, as shown in Fig. 5. The horizontal axis represents the channel number, and the vertical axis represents the energy. When the γ-ray energy was lower than 1.332 MeV, there was a good linear relationship between the energy and channel numbers. As presented above, the calibration curve of scintillator detectors is usually nonlinear. However, because the tested CsI(TI) detector is a large-size detector, high-energy γ photons can deposit all their energy in a CsI(TI) crystal. This makes the calibration curve of the CsI(TI) detectors linear. A detailed discussion can be found in found in [19]. Figure 6 shows one of the results of the analysis. Based on this figure, it can be inferred that a maximum voltage limit can be added to the detectors to prevent the DAQ from exceeding the effective range to excessive working voltage.

Fig. 5Full-screen View

Energy calibration on different voltage: 1600 V-2100 V

Fig. 6Full-screen View

Curve of peak position versus voltage

To compare the performance of the detector under different voltages, it is necessary to calculate the absolute energy resolution under different conditions. As shown in Fig. 4 and Fig. 5, the spectrum fitting and energy calibration under different voltages were completed. Therefore, the FWHM of each peak can be calculated according to the Gaussian fitting parameter using the formula FWHM = 2.355σ, and then it can be converted into energy according to the calibration curve. Figure 7 shows the absolute energy resolution of the CsI(TI) detector under different voltages. It can be inferred from the figure that, when the γ-ray energy is lower than 1.332 MeV, the energy resolution improves with an increase in voltage. In addition to the graphs shown above, other analysis results, such as the energy calibration function, Gaussian fitting parameter σ, absolute energy resolution, and relative energy resolution, will be recorded in the txt file, which is convenient for users to archive and query.

Fig. 7Full-screen View

Absolute energy resolution with a gain of 3.3