2.1 DAQ board

As shown in Fig.2, the DAQ board, sized at 40 cm×36 cm, features a field-programmable gate array (FPGA) and 24 custom interfaces, with some ancillary components essential to support them. It talks to FEEs directly via the 24 custom interfaces including 16 FEE interfaces on the top and bottom edges and 8 Hit interfaces on the left and right edges. Communication signals (scientific data, trigger, and command-monitoring) between FEEs and the DAQ board are integrated in the FEE interfaces. Thirty-two “Hit” signals, which are generated by the BGO FEEs when corresponding layers are fired, are connected to the FPGA trigger module via the Hit interfaces to generate a trigger signal by the coincidence logic. Thus, different events of incident particles can be extracted by different coincidences of hit signals. With a USB2.0 interface chip (CY68013A, from Cypress Corp.) controlled by the FPGA, data transfer from the DAQ board to a host computer is achieved. With the trigger rate of less than 100 Hz during ground-based tests and the data package size of approximately 4 k bytes for each event, the maximum data rate for the BGO calorimeter is about 3.2 Mbps, which is far less than the USB 2.0 bandwidth. A USB-to-serial bridge controller (FT230XS, from FTDI Corp.) is employed for setting configuration parameters from the host computer to FEEs, and sending BGO FEE status data to the host computer. The trigger, control and communication logics are implemented in a single Altera FPGA device, EP3C40F484.

Fig. 2Full-screen View

(Color online) Photograph of the DAQ board

Figure 3 shows the block diagram of the interconnection between the FEEs and the DAQ board. Each of the interface signals has two channels for redundancy, with “A” channel as the main signal while “B” channel as the backup.

Fig. 3Full-screen View

Block diagram of the interconnection between BGO FEEs and the DAQ board

The DAQ has three trigger modes for different acquisition conditions. The BGO trigger mode, used in cosmic-ray test, is triggered by specific coincidence logic of hit signals, which can be configured by the GUI software. The Inner trigger mode, used in FEE calibration test, is generated by FPGA logic itself. The External trigger mode, derived from an external logic pulse, is useful for board-level tests of BGO FEEs and debugging tests for BGO calorimeter.

Coupled with a 20 MHz clock signal, the scientific data are transmitted from FEEs to DAQ board via a user defined serial protocol. After being locally synchronized and de-serialized, the data are checked by cyclic redundancy check (CRC) detection logic in case of transmission errors. In order to readout all the 16 FEEs after each trigger, the readout logic in the FPGA is designed with the daisy chain concept. Using a “token” signal, the data packages of all the FEEs are written to the USB interface chip one by one, while each FEE channel can be individually configured to stand-by mode for flexibility.

Each BGO FEE has three kinds of monitoring data: the positive (+2.5V) and negative (–2.5V) supply currents of the front-end chips (VA160/VATA160, from IDEAS Corp.)^[11-13], the values of temperature sensors, and the status registers of the FPGA logic. After the calorimeter being assembled in the satellite, the temperature information and FPGA status are checked every 16 seconds, while the current information is checked every second. In ground tests, the DAQ board, which is remotely controlled by the GUI software via a UART port, checks the monitoring data periodically. In this case, the host broadcasts commands with a specific identity to all FEEs through the UART port, however, only the FEE which matches the identity will respond back to the host.

2.2 User Interface software

The custom user interface software is designed with LabWindows/CVI virtual instrument development platform and software test automation technology^[14,15]. The graphical interface, including control buttons, message display windows, and configuration text-boxes, and the hierarchy structure of the software as well, are shown in Fig.4. Applications talk to the DAQ board or other instruments with NI-VISA library and CVI serial port control library. Besides design considerations for long-term ground-level cosmic-ray test, more features are integrated to make the software suitable for various applications, for instance, PMT or BGO FEE performance^[16] tests. It provides a guarantee for the smooth progress of the development and construction of the BGO calorimeter before launching.

Fig. 4Full-screen View

(Color online) The GUI interface (left) and its software hierarchy (right)

2.3 Performance tests

The performance tests of the calorimeter include pedestal test, FEE calibration test^[17], and ground-level cosmic-ray tests. All the tests are carried out frequently, so the readout system should be automated and easy-to-use.

The calorimeter contains 2016 channels, of which 1848 are signal channels, while the others are floating as spare ones. The pedestal test is carried out under the “Inner trigger” mode which is periodically generated at about 100 Hz by the DAQ board. The statistic features of the pedestal for each channel (usually obeying an ordinary Gaussian distribution) can be used as the first diagnosis index throughout the ground-based tests before launching.

The FEE calibration is conducted by remotely controlling a DAC-based calibration circuits under the “Inner trigger” mode. The circuits are designed on each FEE, with a function of generating calibration pulses with programmable charges. The calibration goes automatically. As a result, gain differences among all channels and the drifts over temperature can be conveniently evaluated.

The ground-level cosmic rays, in most cases composed of muons of several GeV, are used to calibrate the minimum ionization particle (MIP) response of each BGO bar, which is a fundamental parameter for the energy reconstruction of the calorimeter. The ground-level cosmic-ray tests are carried out under the “BGO trigger” mode. After pedestal subtraction and gain correction, MIP calibration^[18,19] can be done directly. Light yield of a BGO crystal depends on temperature significantly, and during the cosmic-ray tests, the temperature effect of each BGO is investigated in the temperature-cycling test and thermal-vacuum test^[20,21], so as to evaluate the temperature coefficients.

3.1 Description of the experiments

A series of environmental tests were carried out for the qualification and flight models. For instance, thermal-vacuum test was used to verify performance of the calorimeter in a simulated vacuum and thermal environment similar to space. The BGO calorimeter, sized at 0.9 m×0.9 m×0.5 m, was installed in a heating cage, with a supporting structure of proper thermal insulation design. The Φ5-m vacuum tank, with an inner wall LNF (liquid nitrogen freezing) system, named as “space environmental simulator”, is evacuated to 10^–3 Pa (Fig. 5, left). The calorimeter temperature could be adjusted from –15°C to +30°C by the heating cage (on the outer surface of shielding panels) for multiple cycles, while the temperature range was from –25°C to +40°C for the qualification model.

Fig. 5Full-screen View

(Color online) The flight model in a vacuum chamber (left) and schematics of the readout system (right), for the thermal vacuum test

The readout system for thermal vacuum test of the flight model is shown schematically in the right of Fig.5, which is similar to the set up for other environmental experiments. As the kernel part of the readout system, the DAQ board receives all the hit signals with RS422 level from the BGO calorimeter, and then processes them with coincidence logic and generates a trigger signal, to indicate a cosmic-ray particle (a muon in most cases at ground level) penetrating several layers of BGO bars. When a trigger is received, the FEEs digitize the signals in all the channels, and send the data packages to the DAQ board via low voltage differential signal (LVDS) level based on a user-defined serial protocol.

The DAQ board also broadcasts commands to FEEs, and receives status information from FEEs as well. All the signals between the DAQ board and the BGO calorimeter are transmitted by twisted cables (for RS422 signal) or shielded twisted cables (for LVDS signal), which have a total length of nearly twenty meters from the control room to the vacuum tank, and go through sealed connectors in a vacuum flange.

The readout system worked stably during the 15 days of uninterrupted operation in a thermal vacuum test.

3.2 Results

It took nearly half a year to accomplish the assembly and ground-based tests for both the qualification model and the flight model of the BGO calorimeter respectively. A series of experiments were conducted, each lasted for days or weeks.

Fig.6(a) depicts the histogram of the pedestal noises of all the FEE channels of the flight model with the width of the Gaussian function for per channel, in which the X-axis is the pedestal noise (in RMS), and one ADC bin is 1 fC. Obviously, the spare ones with lower input capacitance have a lower noise level (approximately 2 fC).

Fig. 6Full-screen View

(Color online) Width distribution (a) of the pedestal noises for all the FEEs channels; typical energy spectrum (b) for ground-level cosmic rays of a channel; and temperature variations at four monitoring points inside the calorimeter (c) in the thermal-vacuum test (the discontinuities were caused by manual re-power operation due to experimental requirements)

When ground-level cosmic rays pass through the calorimeter, the energy deposit in a BGO crystal bar is digitized and MIP energy from an obtained energy spectrum is determined. Fig.6(b) shows a typical energy spectrum from one end of a crystal bar. The results demonstrate that the maximum equivalent noise charge for all the FEE signal channels are approximately 10 fC in RMS, which is much less than the charge of the MIP signal and meets the requirements.

In the long-term experiments, the temperature and current information of the calorimeter was real-timely monitored and recorded. Fig.6(c) shows temperature data from four thermal-sensors inside the calorimeter in a two-week thermal-vacuum test of the flight model. The left parts of the curves depict the thermal-balance process, while the right parts stand for the vacuum thermal-cycling process. The experiment lasted for nearly two weeks, ruptured by several discontinuities due to manual re-power operation planned by the experimental program.