The principle of the WFC is the shadow mask camera (Dicke). A two-dimensional position- and energy sensitive detector (Multi Wire Proportional Counter) is placed behind an opaque screen with a pseudo random array of holes, a so-called shadow mask or coded mask. X-ray sources in the field of view cast shadows of the mask on the detector, each displaced according to the position of the source. A cross-correlation (Fenimore et al. 1978) carried out on ground, reconstructs the position and flux of all sources in the observed part of the sky. The WFC has a high sensitivity over a large field of view and at the same time a good angular resolution over the whole field (due to small mask holes). Figure 1 (click here) shows a schematic drawing of the WFC.
Figure 1: WFC instrument schematic drawing. A coded mask is placed 70 cm in front of a two dimensional position and energy sensitive detector. The total mass is 42.5 kg and the power consumption is 14.1 W
|Energy resolution at 6 keV
|20% over full area
|Field of view
|Angular resolving power
|Source location accuracy
|< 1 arcmin
|2000 c s-1
|(Crab = 270 c s-1)
The detector and the mask have sizes of cm2 and cm2, respectively. A stainless steel structure between detector and mask allows the X-ray photons to reach the detector through the mask only and fixes the position of the mask 0.7 m away from the detector. In front of the mask a 6 m thick kapton foil is placed as a thermal shield. The field of view is 20 square (FWHM) although sources can be detected with less sensitivity in a field of in total. The angular resolving power is 5 arcmin (FWHM), determined by the mask-to-detector distance, the size of the mask elements which is mm2 and the detector position resolution. The energy resolution is 20% at 6 keV and 22% at 22 keV. The source location accuracy is better than one arcminute, limited mainly by the satellite pointing stability and the statistical noise of the observed sources.
The limiting sensitivity is a few mCrab in 105 s but varies with the background X-ray flux and thus with the pointing direction in the sky. At high galactic latitudes the background will be dominated by the diffuse X-ray background (130 c s-1) plus the detector background (20 c s-1) while near the galactic plane the main contribution is from strong X-ray sources. Figures 2 (click here) and 3 (click here) show the minimum detectable source strength and the time needed to detect time variability. From early measurements it could be shown that for reliable source detection a confidence level higher than is needed. Detailed in-flight calibrations will clarify this issue.
Figure 2: Minimum detectable source strength for two sky regions. The sensitivity is on average higher at high galactic latitudes where the overall flux from X-ray sources is less
Figure 3: Observing time required to detect source variability for two sky regions, in less crowded regions the sensitivity for source variability is higher
The WFCs can be used in two scientific and three diagnostic modes. The scientific modes are the Normal (imaging) Mode (NM) and the High Time Resolution Mode (HTRM). The diagnostics modes are DM1, DM2 and the Background Analysis Mode (BAM).
In NM the time resolution is 0.5 ms for every block of 4 photon events. Each event is labelled with 26 bits, 10 bits each for x- and y-position, 1 bit for indicating the upper or lower detector half and 5 energy bits. The maximum event processing rate is 2000 s-1, limited by the instrument processor performance and the satellite telemetry rate.
In HTRM the time resolution is 0.25 ms per event while the maximum processing eventrate is 5000 s-1. In this mode no event position information can be produced, this mode is useful for timing analysis of strong sources.
For diagnostic purposes two Diagnostic Modes (DM1 and DM2) are available. In these modes no onboard processing takes place and all raw event data is available. For background calibration purposes a Background Analysis Mode (BAM) is implemented.
When the available telemetry in NM for the WFCs is not sufficient, the instruments can be programmed to reduce their telemetry. Processing of every n-th event will then be stopped at the input, where n is an integer between 2 and 4. If this reduction is not sufficient and the telemetry buffer becomes full, no more events will be processed until the buffer is read out.
In the WFC, coded mask and detector have the same size; this corresponds to the so-called ``simple" camera described by Sims 1980. This design does not allow complete coding, in contrast to the ``optimum" camera (Proctor et al. 1979), where mask and detector do not have the same size. However, the ``simple" camera has the following significant advantages over the ``optimum" camera:
The detector is an improved less heavy version of the COMIS (Coded Mask Imaging Spectrometer) detector. COMIS/TTM (Brinkman et al. 1983) is a coded mask camera instrument in the KVANT module of the MIR space station and is succesfully operating since 1988.
The detector configuration is a conventional two dimensional position sensitive Multi Wire Proportional Counter (Charpak 1985). The detector consists of two separate counters in one sealed housing: the main photon counter and the guard counter used for background rejection, see Fig. 4 (click here).
Figure 4: Schematic diagram of the WFC detector showing the configuration of the main and the guard counter. The guard counter serves to reject background and spurious events
Both counters are separated by a grounded wiregrid K1. The detector contains drift field electrodes in both drift regions, and wire grids in the amplification regions.
Because of weight constraints, the detector housing is made of titanium with a beryllium entrance window support structure (Fig. 5 (click here)). This structure was built by Electro Fusion Corporation of Fremont, U.S.A.
A 1 mm aluminium shield covers the titanium walls at the outside of the housing to minimise the AC magnetic field susceptibility of the detector. The detector is sealed with an indium gasket. The size of the detector entrance window is mm2. The actual window consists of a 150 m beryllium foil epoxied to the beryllium window support structure. The support structure subdivides the window into subwindows). The total detector sensitive area is 520 cm2.
Figure 5: Front view of the WFC detector, clearly visible is the beryllium window support structure. The detector housing is made of titanium. The size is
Figure 4 (click here) shows the lay-out of the grid system, which consists of five wire grids. These grids serve as anodes, sense grids, and electrical grounding (grids A1 and A2, S1 and S2, and K1, respectively). The wire grids are mounted with respect to each other with an accuracy of about 40 m. Table 2 (click here) gives dimensions and voltages for the gridwires. In general, the wires of a grid are interconnected. We now discuss the individual grids in more detail, starting with the anode grids.
The anode grid A1 is located half way down the Main Counter. It serves to obtain energy information on the impingent X-ray photons, and triggers the event processing. Three wires are not connected electrically to this grid, they function as the anode of a separate so-called Wall Counter.
Likewise, the Guard Counter possesses an anode grid A2. We note that the wires at the edges of both A1 and A2 have increased diameters. This lowers the local electric field, and therefore prevents high voltage break-down.
The sense grids S1 and S2 allow the position of the photon track to be determined in the x- respectively the y-direction. To this aim, they are positioned at both sides of the anode grid A1 of the Main Counter, while running perpendicular and parallel, respectively, to the anode wires of A1. A special interconnection scheme, using fine and coarse sections, serves to read the pulses from the sense grids, as shown in Fig. 6 (click here). In each sense grid, the wires are grouped into 48 fine sections. Four fine sections in turn make up one coarse section, so that there are 12 coarse sections in total. Every fine section is connected to the corresponding location within a coarse section.
All grids together have 35 signal outputs. The wires themselves are composed of gold plated tungsten and fixed with an accuracy of 10 m on ceramic macor strips with glue spots. The interconnection patterns for the fine and coarse sections have been deposited using thick-film techniques, the wires being connected with conductive epoxy.
The wire grid system ( mm2) has been chosen larger than the entrance window to minimise distorting boundary effects on the spatial and spectral response. In this way, secondary radiation originating from the housing walls and grid supports can be discriminated against. Furthermore, a field correction plate was installed at both sides of the grid system. These plates guide the field lines in both drift regions (see Fig. 4 (click here)) the direction perpendicular to the entrance window.
Figure 6: Interconnection diagram for the wires of the sense grids S1 and S2 (see Fig. 4 (click here)). The position readout scheme for one of the two detector coordinates is shown. The 12 coarse sections C1..C12 each contain 8 wires while the 4 fine sections F1..F4 each consist of 96 wires
The detector is filled at 2.2 bar with a mixture of Xe (94%), CO2 (5%) and He (1%). While the gas gain can be adjusted by changing the high voltage supply, the nominal gain value is kept at 15000.
In order to continuously verify the energy and spatial response, a number of inflight calibration sources are mounted to the detector window support structure. Nine ``pinhole" source containers containing 55Fe radio-active material are used to monitor the spatial response of the detector. Each pinhole source emits 6 keV photons in a beam with a diameter of < 0.2 mm FWHM and with an intensity of about 0.01 photon per second. Eight sources are located at the borders of the sensitive area, while one source is in the centre of the sensitive area. All sources are at positions where the spatial response is most sensitive to minute displacements of the individual grids and to electrical interference.
A ``cocktail source" holder with 55Fe and 109Cd radio active material emits a mixture of 6 keV photons (10 c s-1) and 22 keV photons (1 c s-1) respectively in a beam of 1 mm FWHM. This source is used to verify the energy response and the short term stability of the gas gain.
The following properties are desirable for an optimum detector:
To keep the mass down, the detector housing is of titanium while the window support structure is entirely made of beryllium. The gas pressure was chosen to be 2.2 bar. This allows an entrance foil thickness of 150 m and a sub window area of mm2. The width and height of the support bars was chosen to be 1.5 mm and 10 mm, respectively. To increase rigidity the window support central cross bars are made 22 mm high and 13 mm wide.
The spatial resolution of the photon detection is determined by the gas filling, the gas gain, the grid system and the electronic noise. This has been discussed in an earlier paper on the COMIS detector (Mels et al. 1988).
Photon detection in the amplification regions results in larger pulse heights than in the drift regions. In the upper part of the energy range the energy resolution is easily degraded by this effect. This is not caused by electron attachment but arises because the gain is affected by space charge in the avalanche (Mels et al. 1988). The charge density in this avalanche is higher for drift region events than for amplification region events due to the binning effect of the sense wires. This difference becomes smaller when the electric field in the drift region is increased. A drift voltage of 900 V was needed to obtain a resolution of 19% for a collimated beam of 22 keV photons.
There are 35 detector output signals, each of which is connected to its own readout channel. The electronics of such a detector readout channel consist of a charge sensitive preamplifier followed by a pulse shaping circuit.
The anode channels have fast pre-trigger outputs. The triggers are used to initiate an event handling cycle as described below in Sect. 2.3.2 (click here). They trigger the pile-up sensing circuit and serve for anti-coincidence purposes.
Each coarse section channel of the sense grids has a trigger output to set the corresponding bit in two 12-bits coarse address registers. The pulse height of the fine section signals and the anode A1 signal are digitised by a 12-bit ADC. These channels are equipped with a baseline restorer to decrease the sensitivity to microphonics and pile-up effects. The charge collection efficiency in these channels is about 40%.
The event selection and position computation is performed by the front end electronics and the event processor in real time. For each validated event a data packet is generated. The data is then formatted in blocks of fixed length by the communication processor and stored in the instrument memory. The data handling system reads the memory and stores its contents on the onboard tape recorder. Rate meter time profiles are included to enable monitoring of the count rates on the anode grids and other trigger outputs. With this data the following diagnostic parameters can be calculated: the instrument dead time, the high energy (> 30 keV) event rate, the background event rate and the rate of pile-up events.
Once per orbit the contents of the tape recorder is transmitted to the ground station. In case modified procedures of onboard data handling are required, all the software and look-up tables can be updated by uplink from the ground station.
By definition an event occurs in the main counter if either a photon or a particle ionises the gas in the chamber. The resulting cloud of electrons drifts through the sense grid to the amplification region. In case of detection of a photon with an energy below the 35 keV Xe -absorption edge, an avalanche occurs around one single anode wire or two neighbouring wires. If the charge impulse on the anode is sufficiently high, the event handling cycle is initiated.
First the frontend electronics read the detector signals. Subsequently the pulse height in the eight fine section and the main anode channel is stored in analog stretchers. The pattern of the triggered anode channels and the coarse section channels is latched into registers. At the end of this phase the event handling cycle is aborted if one or more veto conditions are satisfied. Veto conditions are:
If an event is accepted
by the front end electronics, the analog signals stored in the
stretchers are digitised and the event position is determined. By
comparing the charge induced on sensegrids S1 and S2, the event processor also
determines whether the event occurred above or below the anode grid
A1. This information is useful to improve the position resolution
at higher energies for inclined beams, and also serves to reject spurious low
(< 10 keV) in the lower drift region.
The point of absorption of a photon in lateral direction is derived from the centre of gravity Z of the charge distribution induced by the avalanche charge on the sense grids. The coarse section index is determined from the pattern of triggered coarse section channels. The position of Z within that coarse section is calculated using a simple linear combination of the charge values measured on the fine sections. The resulting value for the ``position estimator" has a non-linear relationship with the photon position. This is corrected for by look-up tables inflight.
The mask consists of a 0.1 mm thick stainless steel plate with a 2 m gold coating
on both sides.
It contains a pattern of mask elements of
mm2. Of the mask elements 33% are perforated, i.e.
X-rays. The mask has been produced by means of an etching process. To assure the structural integrity of the mask, the size of the open elements is mm2. The mask hole pattern is based on a so-called ``triadiç residue (in 't Zand et al. 1994) set and is shown in Fig. 7 (click here).
Figure 7: WFC coded mask pattern consisting of elements of mm2 of which 33% are transparent for X-rays
The arguments behind the choice of the coded mask parameters were presented by in 't Zand et al. (1994) and are summarized here.
The optimum open fraction of a coded aperture depends on the spatial
response of the instrument. For the BeppoSAX WFCs the optimum open fraction lies
between 0.25 and 0.33. The requirement of a high sensitivity to
X-ray bursts then led to the choice of 0.33 for the mask transparency. Compared with a more traditional choice of 0.50, not only the sensitivity improves but also the telemetry that is needed decreases with one-third.
A transparency of 0.33 can not be achieved with Uniform Redundant Array (URA) patterns which are based on quadratic or bi-quadratic residue sets. By extrapolation of these sets we have arrived at a ``triadiç residue set, where the non-zero shift autocorrelation is 3-valued with 2 values being the same.
A URA pattern in an ``optimum" design (see Sect. 2.1.1 (click here)) would have a 1-valued autocorrelation (i.e. flat sidelobes). In our camera design these autocorrelation features can however be permitted. Due to the shadowing effect of both the collimator and the detector window support structure, the coding of the image is never complete. The resulting coding noise can be largely removed during the image reconstruction process, see Sect. 4.2 (click here).