Article Index

The X-Gamma Spectrometer (XGS)

The X-Gamma ray Imaging Spectrometer (XGIS) comprises 3 units (telescopes). The three units are pointed at offset directions in such a way that their FOV partially overlap. Each unit (Figure 7) has imaging capabilities in the low energy band (2 -30 keV) thanks to the combination of an opaque mask superimposed to a position sensitive detector. A passive shield placed on the mechanical structure between the mask and the detector plane will determine the FOV of the XGIS unit for X-rays up to about 150 keV energy. Furthermore the detector plane energy range is extended up to 20 MeV without imaging capabilities. The main performance of an XGIS unit are reported in Table 2.



Energy band

2 keV – 20 MeV

# detection plane modules


# of detector pixel /module


pixel size (= mask element size)

5x5 mm

Low-energy detector (2-30 keV)

Silicon Drift Detector

450 μm thick

High energy detector (> 30 keV)

CsI(Tl) (3 cm thick)

Discrimination Si/CsI(Tl) detection

Pulse shape analysis

Dimension [cm]


Power [W]


Mass [kg]




2 - 30 keV

30 - 150 keV

> 150 keV

Fully coded FOV

9 x 9 deg2



Half sens. FOV

50 x 50 deg2

50 x 50 deg2 (FWHM)


Total FOV

64 x 64 deg2

85 x 85 deg2 (FWZR)

2π sr

Ang. res

25 arcmin



Source location accuracy

~5 arcmin (for >6σ source )



Energy res

200 eV FWHM @ 6 keV

18 % FWHM @ 60 keV

6 % FWHM @ 500 keV

Timing res.

1 μsec

1 μsec

1 μsec

On axis useful area

512 <cm2

1024 cm2

1024 cm2

Table 2. Top: XGS specifications. Bottom: XGIS unit characteristics vs energy range    


The detection plane of each unit is made of 4 detector modules each one about 195x195x50 mm in size detecting X and gamma rays in the range 2 keV – 20 MeV.. For each energy loss in the module, whatever procured by EM radiation or ionizing particle, the energy released, the 3 spatial coordinates and the of the interaction and time of occurrence will be recorded. The basic element of a module (Figure 8) is a bar made of scintillating crystal 5x5x30 mm3 in size. Each extreme of the bar is covered with a Photo Diode (PD) for the read-out of the scintillation light, while the other sides of the bar are wrapped with a light reflecting material convoying the scintillation light towards the PDs.


  • 7a
  • 7
  • 8
  • 9
Figure 7. Sketch of the XGIS Unit (top left). Sketch of the XGIS module (top right and bottom left). A module (bottom left) is made of an array of 32x32 scintillator bars with Si PD read-out at both ends (top right). Both PD and scintillators are used as active detectors. The PDs readout electronics consist of an ASIC pre-amp mounted near each PD’s anode while the rest of the processing chain is placed at the module sides and bottom. Bottom right: Principle of operation of the XGS detection units: low-energy X-rays in teract in Silicon, higher energy photons interact in the scintillator, providing an energy range extended to three orders of magnitude.


The scintillator material is CsI(Tl) peaking its light emission at about 560 nm. The PD is realized with the technique of Silicon Drift Detectors (SDD-PD) (Gatti 1984) with an active area of 5x5 mm2 matching the scintillator cross section. Crystals are tightly packed in an array of 32x32 elements to form the module. SDD and scintillator detect X- and gamma-rays. The operating principle (see Figure 7 bottom right) is the following. The top SDD-PD, facing the X-/gamma-ray entrance window, is operated both as X-ray detector for low energy X-ray photons interacting in Silicon and as a read-out system of the scintillation light resulting from X-/gamma-ray interactions in the scintillator. The bottom SDD-PD at the other extreme of the crystal bar operates only as a read-out system for the scintillations. The discrimination between energy losses in Si and CsI is based on the different shape of charge pulses.

While the electron-hole pair creation from X-ray interaction in Silicon generates a fast signal (about 10-ns rise time), the scintillation light collection is dominated by the fluorescent states de-excitation time [0.68 µs (64%) and 3.34 µs (36%) for CsI(Tl)] and a few µs shaping time is needed in this case to avoid significant ballistic deficit. Pulse shape analysis (PSA) techniques are adopted to discriminate between signals due to energy losses in Si or CsI. The results we obtained for the separation of the energy losses in the case of an 241Am source (Marisaldi et al. 2004) are shown in Figure 8 (left panel). As can be seen from Figure 8 (middle panel), the ratio of the two pulse heights is approximately constant for pulses of common shape and allows discrimination between interactions taking place in Silicon or in the scintillator. For gamma-rays interacting in the scintillator, combining the signals from the two PDs at the extreme of each bar allows to determine the energy and the depth of the interaction inside the crystal (Labanti et al. 2008b).


  • 10
  • 10a
Figure 8. Left: bidimensional spectum of a 241Am source showing interaction in Silicon (top line) and in the scintillator (bottom line). The output of two parallel processing chains of different shaping times (fast / slow chain) is plot one against the other. Middle: distribution of the ratio of the two processing chains for events in Silicon and in the scintillator. The large peak separation indicates the optimal discrimination performance. Right: XGIS unit main building bloks.


XGIS building blocks

In the XGIS HW, the main building blocks (see Figure 11 for one XGIS unit) are:

a) the mask and the FOV delimeter,

b) the scintillator detectors

c) the FEE in both its analogue part (with SDD-PD, ASIC1 and ASIC2), digital part (DFEE) and services (TLM, TLC Power supply).


Silicon Drift Detectors

The SDD-PD design will be derived directly from the Italian REDSOX R&D program funded by INFN, centered on SDD and related electronics studies. Different SDD geometries have been designed, realized and tested. These SDDs are designed at INFN Trieste and produced by Fondazione Bruno Kessler (FBK, Trento, Italy) that since 2013 is capable of processing 6’’ wafers. This kind of technology demonstrated a noise performance of 8.6 e- rms at room temperature (Bertuccio 2014). The SDD-PD design will be tailored for our specific application. An array of 4 SDD-PD has already been produced (see Figure 9) allowing a tight packaging of the scintillating crystals.


  • 12
  • 13
Figure 9. 2x2 SDD array, 5mm side each one designed by INFN Trieste and produced by FBK, Trento.


Front-End Electronics

The Analogue Front End Electronics (AFEE) design will meet the requirements of minimizing the overall noise readout of an SDD taking into account the limited power budget. For the whole XGIS the overall number of SDD-PDs to be read out is 24.576. The functions of the AFEE for a single detector are illustrated in Figure 10 (left panel). Due to the large number of elements, the technology used in the AFEE will be based on ASICs. Certain operations on the signal, as pre-amplification, must be done as close as possible to the SDD-PD to reduce the coupling stray capacitance. As a best solution, two different ASICs will be realized. ASIC-1 dealing with few SDD PDs signals and performing pre-amplification and buffering to transfer the signal to the rest of the processing electronics few tens of cm away, and ASIC-2 dealing with more signals, leading to 512 ASIC-1 and 128 ASIC-2 per module.


  • 14
  • 15
  • 16
Figure 10. Left: Scheme of the front end electronics we propose. Right - top: VEGA ASICs realised for the readout of 32 SDD channels. Right - bottom: single channel ASIC cell 200x500 μm2 in size.


The ASIC-1 and ASIC-2 design will be based on the VEGA-ASIC design developed at Politecnico di Milano and Università di Pavia for the readout of SDD devices in the REDSOX project frame (Figure 10, right panel) (Ahangarianabhari 2015). The main characteristics of one of the 32 VEGA channels, having a size of 200×500 μm2, are ~20 electrons rms noise when tested with large-area SDDs and 420 μW/channel consumption for the functions of preamplifier, programmable shaper, discriminator and peak stretcher. With this configuration, an energy resolution of 205 eV FWHM at 5.9 keV has been obtained (Campana 2014). ASIC-1s will be placed on the top and bottom of each XGIS module while ASIC-2s will be placed on the lateral side of each XGIS module.

The Digital Front End Electronics (DFEE) interface the FEE with the Instrument Data Handling Unit (I-DHU) and the Payload Power Supply Unit (PSU). Each XGIS module will have its DFEE whose main functions are the following: interface the stretched ASIC-2 output and provide Analog to Digital Conversion (ADC), event time tagging, detector and module identification.

Units Control Electronics (UCE) At XGIS units level (4 modules) the UCE will manage: a) low voltage (LV=3.3 V) and medium voltage (MV =180V) power supply post-regulation and filtering and distribution, b) TM and TLC interface and management.


Data Handling Unit (DHU) and its functions

The whole XGIS background data rate (3 units) towards the DHU is of the order of 6.000 event/sec in the 2-30 keV range and about 3.700 event/sec above 30 keV. Each event received by DHU will be identified with a word of 64 bits (4 for module address, 10 for bar address, 10 for signal amplitude of the fast top channel, 11 for signal amplitude of the slow top channel, 11 for signal amplitude of the slow bottom channel, 18 for time).

DHU functions will be:

  • discriminates between Si and CsI events.

  • For CsI events, evaluates the interaction position inside the bar by weighting the signals of the 2 PD (a few mm resolution expected). Combining this information with the address of the bar (5x5 mm in size) each module becomes a 3D position sensitive detector.

  • Exploiting the 3D capabilities background can be minimized .

  • It continuously calculates along the orbit the event rate of each module in different energy bands (typically 2-30 keV, 30 -200 keV and >200 keV) and on 5 different times scales (eg 1 ms 10 ms, 100 ms 1s 10 s).

  • In the 2-30 keV range and for each unit, it produces images of the FOV in a defined integration time.

  • It holds in a memory buffer all the XGIS data, rates and images of the last 100 (typical) seconds with respect to the current time.

  • Produce maps of the three unit planes with event pixel by pixel histograms in different energy bands (typically 32 with E width varying logarithmically) and with selectable integration times (min 1 sec).


XGIS and the GRB trigger system

XGIS will contribute to the THESEUS’s GRB trigger system in different ways:

a) Qualification of the SXI triggers. The primary role of XGIS is to qualify the SXI triggers as true GRB. The basic algorithm for GRB validation is based on an evaluation of the significance of the count rate variation, calculated as described below. The procedure will be the following:

  1. from the SXI direction given to the event, it is identified one of the three XGIS units in which the event has potentially been detected;

  2. look for an excess of the rates in the modules of this unit in the bands 2-30 keV and 30-200 keV with respect to the average count-rate continuously calculated by DHU.

b) Autonomous XGIS GRB trigger based on data rate. The autonomous GRB trigger for XGIS inherits the experience acquired with the Gamma Ray Burst Monitor (GRBM) aboard BeppoSAX and that acquired with the Mini-calorimeter aboard AGILE (Fuschino et al., 2008) and concerns all the modules of the 3 XGIS units. For each module the above energy intervals (2-30 and 30-200 keV) are considered for the trigger. The mean count rate of each module in each of these bands is continuously evaluated on different time scales (e.g., 10 ms, 100 ms, 1s, 10s ). A trigger condition is satisfied when, in one or both of these energy bands, at least a given fraction (typically ≥3) of detection modules sees a simultaneous excess with a significance level of typically 5 sigma on at least one time scale with respect to the mean count rate.

c) Autonomous XGIS GRB trigger based on images. For each XGIS unit, the 2-30 keV actual images will be confronted with reference images derived averaging n (typically 30) previous images, and a spot emerging from the comparison at a significance level of typically 5 sigma will appear. If one of the above trigger condition is satisfied, event by event data, starting from 100 sec before the trigger are transmitted to ground, the duration of this mode lasting until the counting rate becomes consistent with background level.


Telemetry requirements

XGIS for transient or persistent source observation.

For the study of transient or persistent sources different transmission mode will be selected starting from the photon list and the histogram maps of the units. Typically the TLM load will be maintained below 2 Gbit/orbit transmitting:

  • at low energies (<30 keV) pixel by pixel histograms in various Energy channels (e.g. 32 ch) with variable integration times (e.g. 64 sec).

  • above the 30 keV the whole photon list.

In particular observations (e.g. crowded fields) a photon by photon transmission in the whole Energy range will be selected for a total maximum TM load of 3 Gbit/orbit.

In the case of a GRB trigger all the information available photon by photon is transmitted with a maximum TLM load of 1 Gbit.


XGIS sensitivity

The 5 sigma XGIS 1 s sensitivity with energy in the SXI FOV, is shown in Figure 11, along with the XGIS flux sensitivity vs. observation time at a significance of 5σ in different energy ranges. In Figure 12, the FOV of the XGIS in the 2-30 keV band is compared with the SXI FOV, and the XGIS sensitivity vs. GRB peak energy is compared with that of other instruments.


  • 17a
  • 17b

Figure 11. Left: XGIS sensitivity vs. energy in 1 second. Right: XGIS sensitivity as a function of exposure time in different energy bands.


  • 17c
  • 17d

Figure 12. Left: FOV averaged effective area of the XGIS SDD and CsI detecting planes. Right: Sensitivity of the XGIS to GRBs in terms of minimum detectable photon peak flux in 1s (5sigma) in the 1 - 1000 keV energy band as a function of the spectral peak energy (a method proposed by Band 2003). As can be seen, the combination of large effective area and unprecedented large energy band provides a much higher sensitivity w/r to previous (e.g., CGRO/BATSE), present (e.g., Swift/BAT) and next future (e.g., SVOM/ECLAIRS) in the soft energy range, while keeping a very good sensitivity up to the MeV range.