Make your own free website on

  GRAAL project

    General information

    1.  Summary

    2.  Technical description

    3.  Results

    4.   For more in depth information see the detailed manuscripts


The lowering of the energy threshold in ground based gamma-ray detection requires large mirror areas to collect Cherenkov light emitted in air showers effectively. GRAAL is one of totally  four projects  worldwide (another France and two others in the United States)
which attempts to reach this aim via the use of existing fields of solar  mirrors (heliostats). It uses an innovative technique - described in more detail below - which allows costs savings by about a factor 5 - 10 with respect to similar experiments both in investment and operations. The aims of GRAAL is to explore the power of the ``light-sampling'' technique - where the intensity and precise arrival time of the Cherenkov light is measured at many points on the ground. This approach - the only possible one when using heliostat fields for technical reasons - is an alternative  to the ``imaging'' technique, where the precise direction of the Cherenkov photons is measured at one or a few points on the ground.

GRAAL is presently taking data in each clear and moon less night and the basic possibilities and problems of the light-sampling approach when using heliostat fields have been understood.

Technical description, properties of the light-sampling technique

The solar-power plant ``Plataforma Solar de Almeria'' (PSA)  located at 505 m above sea level in southern Spain was chosen for the experiment. 63 out of a total of 225 solar mirrors (``heliostats'') of the field ``CESA-1'' are used for GRAAL.  Each heliostat has a mirror surface of 39.6 sqm which leads  to a total mirror area of  2500 sqm making GRAAL the worldwide largest Cherenkov-light collector. The Cherenkov light, generated in the particle showers induced by the  primary cosmic radiation, is concentrated by the mirrors onto a detector in the central tower .  After test measurements in 1998,  the final version of  GRAAL  was set up at the PSA in 1999. The final - remotely controlled - setup went into operation in 2000. In case the control program detects technical problems, a remote operator is informed via mobile telephone and can take corrective action via the internet. The detector consists of four large light collectors, so called ``Winston cones'', which concentrate incoming light onto large-size photo multipliers. These cones have windows of 108 cm diameter and are mounted inside a protecting hut at 70 m height in the central  tower of the solar field. Each cone points onto a different region of the heliostat field and collects the  signals of 13-18 heliostats onto a single photo multiplier. The windows size of the collectors limits the angle  acceptance angle of the detector to a rather small 0.6 degrees (full angle). For the individual control of the heliostats used by GRAAL a dedicated computer program was developed by the Spanish groups.  Each heliostat sends a signal of about 4 ns width and with an  amplitude determined by its received light to one of the cones. Due to the different distances of the individual heliostats from the collectors the signals reach the read-out electronics at different times and are individually registered by flash ADCs. This principle allows the detection of light from 63 heliostats with only 4 data-taking channels. It is this principle that leads to the cost saving mentioned in the introduction.

The delay between the individual signals depends on the shower direction. The arrival times are measured with a precision < 500 ps. This makes it possible to reconstruct the shower front in  time and thus measure the shower direction independent on pointing direction. The energy threshold is limited mainly by the night sky background (NSB);  the direct star light or scattered light  generates noise signals in the PMT´s. With a trigger electronics consisting of  charge integrators followed by  threshold discriminators  and a 3 out of 4 coincidence the true showers are selected with a  rate of  2 to 4 Hz, depending on weather conditions and zenith angles.

GRAAL takes data continuously since August 1999.  Losses of  observation time due to technical problems were only a few percent. The main source of losses were unstable weather conditions due to the relatively low elevation and closeness to the sea of the PSA and dew formation on the mirrors. The latter problem was improved by the installation of  an dew prevention procedure in spring 2000.

From a comparison of pointing direction and reconstructed direction from the timing analysis we find an absolute precision of  the pointing < 0.05 degrees for both zenith and azimuth angles. From Monte-Carlo simulations we get an angle resolution for Gammas (from MC) = 0.35 degrees at zenith angle 10 degrees and  azimuth angle 45 degrees and an energy threshold for the detection of gamma-ray (defined as energy of particles contributing the maximal count rate of analyzed showers) of ca. 250 GeV.

Detailed comparison of data and Monte Carlo simulations show that the mentioned small angular acceptance - practically unavoidable  when using solar mirror fields - leads to time structure of proton induced showers which is nearly identical to the one of gamma induced showers. This makes gamma-hadron separation - crucial both for enhancing the sensitivity and inferring absolute fluxes - difficult. Moreover, the small acceptance angle leads to a bias even the determination of the  incoming direction for proton induced showers.

Results, status of project

The Cherenkov light could be measured reliably and precisely, all time-lines and specifications of the original  technical proposal were met without cost overruns.  The current operation of GRAAL is cheap due to the complete remote control (via internet) of  the experiment, with the assistance of only the  regular night operator of the PSA (for field operation).

For confirmation of the principle function of the detector the well known Crab pulsar was observed in spring 2000 for about 64 hours in total. After filtering the data with weather criteria only  7 h and 10 min of good data remained for ON and the same time for  OFF position (total of ca. 140000 events taken). From these data an excess of  4.5 sigma in the central region (< 0.7 degrees deviation from source direction) - corresponding to a Gamma flux of 1.7/min - was measured.  Furthermore data on the potential sources 3C 454.3, 3EG 1835+59 and a ``pseudo source'' (for test purposes) was taken. None of these sources showed a significant excess  compared to the OFF position.  Data taking continued until end of 2001 when the
project was finished as planned. The final reports for the project (see publication section on main page) were the first documents to explain why the heliostat-field approach to gamma-
ray astronomy is not competitive to the imaging approach. The other heliostat-field projects, with total costs about an order of magnitude higher than GRAAL have still to supply
such information at the time of the most recent update of this page (July 2002).