Forum Springer Astron. Astrophys.
Forum Whats New Search Orders

Astron. Astrophys. 334, 299-313 (1998)

Previous Section Next Section Title Page Table of Contents

2. The WATCH experiment aboard GRANAT

The GRANAT satellite was launched on 1. Dec. 1989 and the WATCH experiment is composed of 4 WATCH units mounted in a tetrahedral geometry, each being designed with a very large field-of-view (4 steradians). One of the WATCH units has the Sun in its field-of-view and there exists almost two years of solar observation, covering the solar data gap between the end of the HXRBS/SMM observations (December 1989) and the launch of COMPTON (April 1991). As the GRANAT orbit is highly eccentric and has a long period ([FORMULA] 96 hours), it allows long uninterrupted observations between passages through the radiation belts of the Earth. Telemetry was on an average done every 24 hours on 3 days out of 4 in each GRANAT orbit.

2.1. Technical description

The WATCH detectors are based on the rotation-modulation-collimator (RMC) principle which allows to get some information on the arrival direction of the photons. The first modulation collimator was designed by Oda (1965) and consisted of two plane grids of parallel wires separated by a short distance and placed in front of a detector. The Oda technique was thereafter improved by letting the two planes of parallel 50 [FORMULA] transmission grids rotate synchronously around a fixed axis perpendicular to the grids. This allows to modulate the X-ray flux received from a point source. This technique referred to as rotation-modulation-collimator (RMC) was proposed by Mertz (1968) and further developed by Schnopper et al. (1968, 1970) and Willmore (1970).

The WATCH experiment is based on the RMC-principle, but the second grid of the collimator has been replaced with two interleaved grids of X-ray detectors (NaI(Tl) and CsI(Na)) (Lund 1985). The advantage of this method is that by adding together the count rates in the two scintillators, the X-ray burst time profile can be measured, unaffected by the modulation pattern. The ratio between the count rate measured in one scintillator and the total count rate provides the normalized instrumental modulation pattern which can be extracted independently of the time variations in the source flux (Lund 1981). The interleaved scintillator grids of the WATCH detector are a circular mosaic of 22 scintillator (NaI and CsI) strips, each having a width of 5 mm and a thickness of 2 mm. The modulation grid is mounted 5 cm above the scintillator surface and is made up of rods of tantalum with an equal width and a spacing of 5 mm thus providing a spatial resolution of   5.7[FORMULA]. To provide a sufficient mechanical strength of the grid as well as to absorb fluorescent X-rays produced in the tantalum, thin layers of copper and of carbon fiber epoxy were added to the tantalum grid. The total area of the detector system is
95.0 cm2 equally divided between the NaI and CsI scintillators. The two scintillators are viewed by a single photomultiplier tube through a lead glas optical window and have an X-ray entrance window of 25 micron Aluminium which prevents optical light from entering the scintillator and allows the X-rays to pass with little absorption above 6 keV. The signals from the two types of scintillators can be separated electronically, due to the different decay characteristics of the scintillator materials (0.23 µs for NaI(Tl) and 0.63 µs for CsI(Na)).

For the study of solar bursts, basically two count rate channels and two modulation channels from the NaI scintillator were used. The time accumulation for the count rates and modulation patterns is based on the rotation velocity of the modulation grid/scintillator system which is slightly variable with time but which is recorded. The time accumulation for the normal count rate data is approximately 6.5 seconds (8 rotations) and for the modulation patterns approximately 13.9 minutes (1024 rotations). Bursts (triggers) having rise times less than two rotations causes the WATCH instrument to enter burst mode. The accumulation time for the count rate is of one rotation, approximately 0.8 seconds. For bursts triggers with time scales of 2 to 32 seconds the observing program enters the transient mode. In this mode modulation patterns with 16 rotation time accumulation are collected ([FORMULA] 13 seconds).

2.2. The WATCH modulation pattern

When the modulation and scintillator grid of WATCH rotate together, the illuminated fraction of the scintillator will vary as (Brandt 1994):

Eq. (1): [FORMULA] = saw[(([FORMULA] L)/d) tan([FORMULA]) cos(x- [FORMULA]) + [FORMULA] ]

where saw[ ] is the symmetric triangle sawtooth function normalized between 0 and 1 with period 2 [FORMULA] (saw(0)= saw([FORMULA])= saw(2 [FORMULA])=0.5, saw([FORMULA] /2)= 1 and saw(3 [FORMULA] /2)=0.0), L is the distance between the modulation grid and the scintillator surface, d is the width of the tantalum rods of the grid, x is the rotation phase in radians [0,2 [FORMULA] ], [FORMULA] is the off-axis angle of the source in the WATCH field-of-view (the angle between the rotation axis and the incident radiation), [FORMULA] is the phase angle of the source within the field-of-view (the spin axis of the solar viewing WATCH unit on GRANAT is located in the symmetry-plane of the satellite's solar panels and as a consequence the solar image never deviates much from the [FORMULA] =0 line in the field-of-view). [FORMULA] is the measure of the relative position of the grid and assembly of the scintillator strips in units of [FORMULA] /d ([FORMULA] = 0 corresponds to half of each scintillator strip being exposed when the instrument is illuminated head-on). For WATCH on GRANAT [FORMULA] is approximately equal to 0. In reality the modulation grid is not infinitely thin and the finite thickness of the grid must be taken into account when attempting to model the shadow pattern. For a detailed analysis of how the real modulation pattern was simulated, see Brandt (1994).

One rotation is divided into 256 equal angular increments. Thus the modulation patterns consist of 256 values of count rates that are integrated over a given accumulation time. They are used in determining the positions and strengths of the X-ray sources in the field-of-view. The upper part of Fig. 1 shows the time profile of a solar event in the 10-30 keV energy range. The modulation pattern has been accumulated over the time indicated by the two vertical lines. The modulation pattern (bottom) is centered at phase bin= 128 ([FORMULA] = [FORMULA] 0) as expected for the Sun. Fig. 2 illustrates how the simulated solar modulation pattern (left side) using Eq. 1 compares with the observed modulation patterns normalized to the maximum count rate (right side) for different incident angles [FORMULA]. It can be noticed that for all incident angles [FORMULA], the modulation pattern for the Sun is always centered at phase bin 128 ([FORMULA] = 0). The other X-ray sources will be identified by their different modulation patterns (as e.g. the Crab Nebula).

[FIGURE] Fig. 1. The time profile of a solar event (top) with corresponding modulation pattern (bottom). The background estimated in the countrate time profile before and after the burst is indicated as a dotted line in the modulation pattern.
[FIGURE] Fig. 2. The simulated solar modulation pattern (left) for different incident angles compared with the measured solar modulation patterns normalized to the maximum countrate. The background measured in the countrate time profiles before or after the burst is indicated as a dotted line.

2.3. Energy calibration of the NaI scintillator

Before launch the WATCH experiment was calibrated using radioactive sources. However a major gain shift after launch prevents us from using this prelaunch calibration. Furthermore, the radioactive source (Cd-109) associated with each detector was found to be too weak compared to the background to be used for calibration after launch. Therefore the energy calibration of the WATCH-0 detector was performed in the following way: 1) Modelling the detector and 2) Using observations of the Crab Nebula source for the energy calibration. The purpose of the energy calibration is not to perform a complete spectral analysis but to get an estimate of the energy bands in which count rates were accumulated and of the number of photons corresponding to a given count rate.

The X-ray detection efficiency depends on the X-ray absorption coefficients of the scintillators and the transparency of the Aluminium entrance window. The detection efficiency of the NaI scintillator in WATCH-0 is computed by using an analytic calculation of the energy loss probabilities in the detector (program developed by R.A. Schwartz, private communication). It takes into account the probability of photoelectric effect, Compton scattering and fluorescent escape in the scintillator as well as "broadening" due to the spectral resolution of the scintillator. The effect of the final thickness of the modulation grid for an incident photon with angle [FORMULA] is to reduce the effective area by a factor g([FORMULA]) compared to the geometric projected area. Simulations of the mean value of the modulation patterns as a function of [FORMULA] give the following correction factor: g([FORMULA])= 2 [FORMULA] (0.5-0.121 [FORMULA] tg([FORMULA])) for 2.90 [FORMULA] [FORMULA] [FORMULA] [FORMULA] 51.80 [FORMULA]. Also included in the computation is the absorption by the Aluminium window and the thermal blanket located in front of the WATCH instrument. This blanket reduces the transmission efficiency of incoming photons. Before launch it was found that the transmission of 5.9 keV X-rays with blanket was about 60 percent of the transmission without the blanket. An indirect way of taking into account this passive material in front of the instrument is by multiplying the Aluminium thickness by a factor which compensates for this decrease in transmission efficiency. It was found that a value (f) of 1.63 gives a good estimate of the total transmission efficiency at 5.9 keV.

The simulation of the scintillator is then performed using the previous description. The electronic chain of the detector amplifies the output signal of the scintillator with a certain gain which changed during the 2.5 years of the observations (Castro-Tirado 1994). This implies that the effective energy ranges of the count rate channels changed during this period in relation with the changes of the gain. To determine the effective energy channels of WATCH-0, a standard X-ray source must be used. During the 2.5 years of observation the Crab Nebula was at times in the field-of-view. Its coordinates in relation to the orientation of the experiment are known, so that one can identify this source in the background by its modulation pattern. The low- and high-energy count rates can thus be calculated using the two corresponding modulation patterns. It is found that the Crab count rates in WATCH-0 changed over the 2.5 years simultaneously to the changes of the voltage of the experiment (from May 1991 to November 1991). After this period, the voltage returned to its previous value. Three observing periods must then be defined corresponding to different values of the WATCH-0 energy channels.

For each period the Crab incident photon spectrum given as:

Eq. (2):dN/d(h [FORMULA])= 2.02 [FORMULA] 10-2 [FORMULA] (h [FORMULA] /20)-2.09
             ph cm-2 s-1 keV-1 with h [FORMULA] in keV

is introduced into the response of the detector. The count rates produced in the low- and high-energy channels are computed for different limits of the output energy bands and compared with the observed count rates. The effective energy ranges found from this procedure are indicated in Table 2.


Table 2. Effective energy ranges of WATCH-0

The knowledge of the effective energy ranges of the channels allows to estimate the number of counts that will be registered in each given channel. The right side of Fig. 3 illustrates for periods 1 and 3 the expected count rates for a non-thermal photon spectrum I(h [FORMULA])= A [FORMULA] (h [FORMULA] /10) [FORMULA] with A=1 and h [FORMULA] is the energy of the incoming photons in keV. The harder the spectrum (the lower the spectral index [FORMULA]) the more counts are expected in the high-energy channel and the lower the ratio between the two channels. The left side of Fig. 3 shows for periods 1 and 3 the expected count rate as a function of temperature (T) for a given thermal photon spectrum I(h [FORMULA]) defined by the temperature T of the emitting plasma and its emission measure (EM) with EM= 1047 cm-3. For low temperatures, count rates are recorded mainly in the low energy channel. With increase in temperature the expected number of counts will increase much faster in the high-energy channel compared with the low-energy one.

[FIGURE] Fig. 3. Right side: The expected count rates in the low and high energy channels and the ratio for a non-thermal spectrum as function of the spectral indices. Left side: The expected count rates in the low and high energy channels and their ratio for an incident thermal photon spectrum as a function of temperature (see text for details).
Previous Section Next Section Title Page Table of Contents

© European Southern Observatory (ESO) 1998

Online publication: May 12, 1998