Design Philosophy and Overview
A conceptual sketch of MXGS is shown in Fig. 2. It comprises a Low Energy Detector (LED) and a High Energy Detector (HED).
The LED detector plane (green in Fig. 2) consists of pixelated Cadmium-Zink-Telluride (CZT) detector crystals that measure photons with energies from 20 to 400 keV. A hopper-shaped collimator defines the 80° × 80° fully coded field of view for the LED. A passive graded shield surrounding the LED housing (1 mm stainless steel, 0.5 mm Sn, 0.25 mm Ta and 2 mm Pb), except top and bottom, shields the LED against the ambient \(\gamma \)-rays as well as against the fluorescence photons and radioactive decays in surrounding materials. The shield, also used in the Low Energy Gamma Ray Imager, LEGRI on MINISAT-01 (Dean et al. 1991), is efficient for photon energies below 200 keV.
A coded mask provides the imaging capability of the MXGS LED. The mask is a random pattern of square-formed holes mounted on the entry of the hopper shield and photons below 400 keV can only enter the LED through the holes. The mask is covered towards the inside with a Kapton foil which will stop electrons with energies up to 200 keV but will allow photons down to 15 keV to enter the detector. From the penumbra pattern created by the flux of photons from a distant point source, their direction of arrival can be determined. The area of the holes is 50% of the total entry area of the detector. The coded mask is covered towards the outside with a Silver coated Fluorinated Ethylene Propylene foil, which absorbs visible light and UV radiation. The front end of the MXGS LED is covered with a 0.175 mm thick Arlon PCB for contacting with the high voltage supply. The mask structure is also used for support of a weak radioactive source (109Cd), which will be used for the in-flight calibration of the LED.
The MXGS HED comprises 12 Bismuth-Germanium-Oxide (BGO) detector bars each coupled to a photomultiplier tube (PMT). HED is sensitive to photons with energies from 200 keV to >20 MeV. The HED is mounted behind the LED and will effectively shield the LED against radiation coming through the rear of the assembly. Three weak 22Na radioactive sources are mounted in between the CZT detector plane and the BGO array. These sources are used to perform in-flight calibration of the BGO detectors.
The Detector Front End Electronics (DFEE) of the LED is mounted on two sides of its detector plane and for the HED behind the detectors in the housing that contains the HED. Behind the HED housing is the box with Power Supply Unit (PSU). It includes a High Voltage Power Supply (HVPS) and Low Voltage Power Supply (LVPS). The MXGS Data Processing Unit (DPU) is mounted in its own compartment below the electronics box.
Figure 3 shows the functional block diagram and the internal interfaces of MXGS. X and soft \(\gamma \)-rays, which enter through the opening of the collimator and the coded mask, interact in the CZT crystals of the LED. The energy of the incoming photon is thereby converted to electrical charges. These are drifted to the detector readout electrodes by an electric field, produced in the detector by the LED HVPS. The detector signals are amplified by the preamplifiers, ASICs, and subsequently digitized and time stamped by the Field-Programmable Gate Array (FPGA) of the LED Detector Assembly Unit (DAU). A large fraction of the \(\gamma \)-rays with energies above 300 keV will pass through the LED, but will be absorbed in the BGO crystals of the HED. The light produced in this process is read out by the HED PMTs, digitized and processed by the FPGA of the HED DAU. Both the data from LED and HED data are passed on to the DPU, which inspects the data for occurrence of short bursts. A trigger signal will be generated in case of burst detection. The DPU will also format the detector data for telemetry. The full set of parameters for an event consists of arrival time, position in the detector (LED: pixel-ID, HED: BGO-bar) and pulse height (proportional to photon energy). When a burst is detected, all events in a 2-second time interval around the trigger time will be transmitted for further on-ground analysis.
The LED is a position sensitive pixel detector, which together with the coded mask in front of the LED allows for determining the direction to the source for the detected radiation. The imaging feature of the MXGS is a unique capability, which will be exploited for the first time in connection with TGF research by MXGS. The MXGS is bi-directionally linked to the MMIA instruments. A TLE or a lightning event detected by the MMIA will also cause a 2-second data dump of the MXGS events. MXGS, in order to monitor its performance, will continuously collect spectra of the ambient environment and of the calibration source (see “MXGS On-Board Software and Data Processing”. These spectra are transmitted to ground at least once per orbit.
The LED and HED Detectors
The LED and HED are both assembled in four Detector Assembly Units (DAUs).
The LED detector plane consists of pixelated Cadmium-Zink-Telluride (CZT) detector crystals that measure photons with energies from 20 to 400 keV. CZT is a semiconductor with a high quantum efficiency for hard X- and \(\gamma \)-rays due to the high atomic numbers of the material. It can be operated close to room temperature (typically +10 °C) and does not require cryogenic cooling as required by conventional (thick) Si and Ge detectors. CZT detector systems require modest high voltage, and coupled to ASICs, these systems are compact and ideally suited for space experiments with a low read-out noise and high energy-resolution. Detectors with pixelated readout are well suited for imaging purposes. Each LED DAU consists of 4 chains of 4 Detector Modules (DMs), which are read out separately. Each CZT DM is \(4 \times 4 \times 0.5\) cm3, divided into \(16\times16\) pixels. This gives 4096 pixels per DAU and 16384 for the entire LED. The CZT crystals were provided by REDLEN, Canada. Each CZT DM is integrated with two ASIC-XA-1.82 units, provided by IDEAS, Norway. One DM and the entire LED are shown in Fig. 4, left panels. The total geometrical area of LED is 1024 cm2.
The HED detector plane consists of Bismuth-Germanium-Oxide (BGO) scintillators and is sensitive to photons with energies from 200 keV to >20 MeV. BGO has been used in other space missions, as Fermi, for detecting high energy photons. One HED DAU consists of 3 separate BGO bars each one connected to a photomultiplier tube (PMT), which gives 12 BGO/PMT detectors in the HED. The BGO bar is \(15 \times 5 \times 3.2\) cm3 and each BGO/PMT is read out separately. The BGO bars were provided by Saint-Gobain and the PMTs (R6231-01 MOD) by Hamamatsu. One HED DAU and the entire HED are shown in Fig. 4 right panels. The total geometrical area of HED is 900 cm2.
The PSU and the LED and HED Read-Out Electronics
The LED and HED units are self-contained read-out units, each controlled by an anti-fuse FPGA from Microsemi (RTAX 2000). Each unit has a dedicated data and control interface with a bit-rate of 18 Mbit/s connected to the common DPU.
Power is provided by the common Power Supply Unit (PSU), which consists of four identical blocks, each dedicated for powering one LED DAU and one HED DAU. Each PSU block supplies the two DAUs with appropriate LV and HV rails and also provides the PSU controller with analog telemetry channels and command interface. This allows the DPU to turn on and off individual supplies and set high voltage output value and to monitor all output voltages and currents together with temperatures at the critical points of the PSU and the DAUs. One LED DAU consumes about 6.3 W, summing up to 25.3 W for the LED. The HED layer consumes about 7.5 W including HV supply. Including also the consumption of the PSU controller 0.6 W, DPU 5 W and power supply loss, MXGS total power consumption is 70 W.
In order to deliver low-noise voltages to all DAUs (HED and LED) requires careful separation of ground connection to avoid any ground loops and resulting noise and interference. The LED detector is particularly sensitive to ASIC power rail noise and HV ripple. Therefore, the analog ground, the digital ground and the high voltage are galvanic separated. In addition, all power supplies inside the PSU are galvanically isolated from the primary power lines, and the analog, digital and high voltage grounds of each DAU are connected in single point inside the DAU.
The signals from the 64 CZT DMs of the LED are read out by 128 XA-ASICs from IDEAS (two XA-ASICs pr DM). Each ASIC is self-triggering and will give as an output: a trigger signal, the pixel address and its signal amplitude when a gamma photon is detected (see Fig. 5). From the ASICs we get analog energy signals and digital addresses conveyed as current signals, which need to be converted to voltages and digitized. It is possible to connect the ASIC output current signals from several ASICs in a chain, and then the current signals will be added together. In our system 8 ASICs are connected, forming one chain. There are 4 chains in one DAU. For the energy signals a12-bit ADC (10.5 ENOB) is used per chain for digitization. Finally, we interface to the FPGA where the so-called SCience Data Packages (SCDP) are assembled and forwarded to the DPU. A SCDP is a 48-bit sequence containing a 10-bit event energy, 14-bit pixel address, 20-bit time stamp of 1 μs resolution and various flag bits.
The output of the HED PMTs are voltage signals, which are digitized to 12-bit (10.5 ENOB) at a rate of 36 MHz. Continuous sampling at this rate allows for digital noise filters and advance triggering, permitting for example pile-up handling within the FPGA. A voltage pulse is accepted as a valid event if the voltage value is above a configurable threshold and is captured after peak-detection. A new pulse can be detected approximately 300 ns after peak detection, giving a typical “dead time” per PMT/BGO of 550 ns, since the pulse peaking time is about 250 ns. There is a time-over threshold counter per second available as an instrument housekeeping parameter, allowing detection of paralysation, which may happen if there is a high number of counts within the dead time. The 48-bit HED data packages consists of 12-bit event energy, 2-bit detector address, 27-bit time stamp of 27.8 ns resolution and various flag bits. The read-out stream may contain normal events, pile-up events (labelled fast events, which also contain valley data, and will be explained below) and overflow events. Overflow events are sent if the maximum ADC value is exceeded and energy bits are replaced with duration of the overflow. It is also possible to debug the detectors in-flight by configuring sample data mode, which captures 256 raw data-samples pre-triggered from the detection of an event.
Timing Accuracy, Dead-Time and Pile-Up Effects for LED and HED
Although the MXGS is designed with relatively fast detector response and fast read-out electronics, we cannot exclude dead-time and pile-up effects. For an average TGF, we expect to have about 70 counts in the LED detector and about 200 in the HED detector. Depending on the duration of the \(\gamma \)-ray flash this may or may not result in pile-up and/or dead time effects.
The temporal resolution for LED is 1 μs given by 20-bit timestamps and for the HED it is given by 27-bit time stamps resulting in a temporal resolution of 28.7 ns.
In the LED the dead-time for each XA-ASIC is 1.4 μs, which means that each XA-ASIC can only record one hit in this interval. In our system 8 XA-ASICs (4 DM) are connected in one chain. If there is a hit in a different XA_ASIC in the same chain within this 1.4 μs window it will be recorded as a multi-hit. This can happen because the chain is not allowed to reset before new hits occur. These events are distinguishable, but their energies are added together and their address is not correct and can therefore not be used for imaging. As long as a new hit occurs on the same chain within 1.4 μs of the previous hit this pile-up effect will continue, resulting in a significantly underestimated recorded count rates for high actual count rates. However, the instrument housekeeping data contains a dedicated counter for total number of hits including multi-hits, therefore a state of paralysation is detectable.
For HED, it is the peaking time of the filtered pulse (250 ns) together with a time window (300 ns) after pulse peak-detection that defines the dead-time of 550 ns. This will also be the systematic delay of the correct time. If more than one hit occurs within this dead-time only one count will be recorded, and the time stamp and energy will be a mixture of the two and dependent on the relationship of the two pulses. This is illustrated in Figs. 6A and 6B. If a new hit occurs after this dead-time of 550 ns it will be recorded with correct time but the energy will be the sum of its energy and the tail of the previous pulse, so-called tail-pile-up. This is illustrated in Fig. 6C. Since we record the valley between these two pulses, it is possible to reconstruct the real pulse height of this tail-piled-up pulse.
Trigger signals between MXGS and MMIA have a resolution of 1 μs and the relative timing accuracy between MMIA and MXGS is 10 μs.
The Data Processing Unit (DPU)
The MXGS DPU (Fig. 7) provides overall control, data acquisition, data processing, telemetry formatting and monitoring functions of the MXGS instrument.
The DPU is based on the radiation hard Xilinx Virtex-5 FX130T FPGA with one LEON3 (soft core) processor. The LEON3 core from Gaisler operates at a clock frequency of 50 MHz. Integrated into the DPU PCB is an internal power supply which delivers internal voltages of 1.0 V, 2.5 V, 3.0 V and 3.3 V from the 28 V delivered from the Data Handling and Power Unit (DHPU). In Fig. 8 the block diagram of the DPU is shown.
The DPU provides control of LED, HED and PSU. It collects and processes event data from the LED and HED and processes the event data, specifically searching for TGFs. It performs additional event processing related to secondary science objectivities and engineering monitoring (e.g. background event monitoring, detector pulse-height spectra, etc.). With the DPU transmitting and receiving event flags (triggers) between MMIA and MXGS, a TLE or lightning detected by the MMIA will also cause a 2-second data dump of the MXGS events and vice versa.
From the DHPU a 1 Hz Time Correlation Pulse (TCP) is received and encoded onto a DPU generated 1 MHz clock signal that is passed to both LED (CZT) and HED (BGO). This signal is used by LED and HED to time-tag the data with a resolution of 1 μs and combined with UTC this signal enables a relative timing accuracy of 10 μs between the MMIA and the MXGS.
The DPU also collects HK data and status data from the LED, HED and PSU and measures external temperatures of the MXGS instrument. The top-level interface diagram of the instrument DPU is shown in Fig. 9.
The DPU receives commands from the ASIM DHPU. These commands will include instrument mode change and configuration commands, data request commands and memory loads. The DPU then sends command acknowledgements back to the DHPU.
When requested, the DPU transmits data to the DHPU and these data will consist of both science and HK data. The data are processed and formatted by DPU for the ASIM DHPU before TM downlink.
The Collimator and Field of View
A hopper-shaped collimator mounted on top of the LED detector layer defines the LED Field of View (FOV) and LED Fully Coded Field of View (FCFOV). The main purpose of the collimator is to shield the LED array from X-rays coming outside the Earth horizon, called the Cosmic Diffuse X-ray Background (CDXB). The hopper walls upwards from the detector array boundary at an angle of +−40° inclined from the nadir axis define the MXGS FOV of 138° × 138° (Earth angular size from the ISS) and the MXGS FCFOV of 80° × 80°. TGFs observed within FCFOV secures an accurate imaging, while marginal TGF imaging is also possible for strong TGF at any Earth surface position. The hopper walls made in Aluminum (3 mm) and Tungsten (0.1 mm) provides a good shielding up to 60 keV (CDXB peak). Figure 10 shows the collimator geometric envelopes and the collimator flight model with the coded mask on top during MXGS integration. The collimator also has the function of shielding the detectors from Sun optical and infrared radiation. The coded mask is further described under the section “Imaging”.
The Engineering Budget
Mass, dimension, power consumption and data rates for MXGS are as follows:
The MXGS has a mass of nearly 147 kg and external dimensions of (\(774.9 \times 746 \times 523.6\)) mm. Those dimensions include radiators, Loop Heat Pipes and Axial Groove Heat Pipes, but exclude the Multi Layer Insulation.
The total power consumption of the MXGS, including LED, HED, PSU and DPU is ∼70 W.
The DHPU will collect up to 240 MByte of science and HK telemetry data from the MXGS per day corresponding to a mean data rate of 23 kbps.