The detection of the 
decay photons is a crucial part of the
experiment. In order to meet the precision requirements for the
measurement, the electromagnetic calorimeter
must have excellent energy resolution as well as good space resolution
and fine
granularity for the rejection of the overwhelming 
background. In addition, the photon detector must perform reliably at rates
of the order of 1MHz and must provide a
time resolution better than 1ns to cope with the rate of the tagging
counter.
The electromagnetic calorimeter which has been designed for the NA48 experiment
consists in a quasi-homogeneous liquid krypton (LKr) detector with
a 
2cm 
longitudinal tower structure. Its transverse
dimension
is about 2.6m and its thickness is 125cm, corresponding to about
27X 
of liquid krypton. The calorimeter has also a projective geometry
with a maximum aperture angle of 10mrad.
The density of the liquid krypton is 2.41g/cm 
at T=120 
K,
the Moliere radius is 4.7cm and the radiation length is 4.7cm.
The total krypton volume in the cryostat is about 10m .
Figure 4: The LKr electrode structure. One fourth of the detector is shown.
The electrodes are made of 40 
m thick copper-beryllium ribbons oriented
along the beam axis. They have a width of 18mm and are attached
at both ends by
a front-plate and a back-plate made
of Stesalit 4411W. The position accuracy
of the electrodes inside the calorimeter is insured by five 5mm thick Stesalit
plates
with 10mm 
18.5mm holes to let the ribbons go through. These slot masks
have been precisely machined to guarantee a local accuracy in the positioning
of the ribbons better than 50 m and an absolute transverse position better
than 100 m/m. In addition, they are alternatively
displaced by +/-50mrad with respect to the beam axis in order
to produce a zig-zag geometry of the cells. This feature provides a good
mechanical positioning of the ribbons and allows to smear signal losses
when particles impinge very close to the electrodes location. The amount of
passive material in front of the calorimeter is about .85X .
The drift cell size is defined by a central anode and by two cathodes at a distance of 10mm on each side of the anode. A 2mm gap separates adjacent layers of cells in the vertical direction. Figure 4 shows a schematic view of the calorimeter electrode structure.
To achieve excellent energy resolution and speed, the initial current
read-out technique is used. Signals collected from each anode
are sent to preamplifiers located on the back-plate, in the liquid krypton.
These are silicon JFET charge
integrating amplifiers with a restoring time of 150ns.
The signals are then differentiated outside the calorimeter
by transceivers located directly on feedthroughs
and sent via twisted-pair cables to a shaping amplifier, 10m away.
The shaping amplifier output signal has a width of 80ns and
a 3 
undershoot which lasts few microseconds corresponding
to the drift time of electrons across the cell.
To efficiently cover a dynamical
range of 3.5MeV to 50GeV for the energy deposited in a single cell, a
4-gain switching amplifier is used with respective
gains of 1, 2.9, 7.1 and 17.7. The output signals of the shaper amplifiers
are digitized by 10-bit FADCs running at 40MHz in a fully pipeline mode.
To improve the time measurement of the detected photons, scintillator
fibers have been inserted vertically in the liquid krypton at a depth
of 9.5X .
The fibers have a diameter of 1mm and are ganged in bundles of about 25.
These are accurately positioned between ribbons.
The scintillator fibers are instrumented with Hamamatsu-1355 photomultipliers
also located in the liquid krypton. The time resolution achieved with
these scintillator counters is better than 250ps.
The performance of such a calorimeter has been recently measured with a prototype [4]. The energy resolution achieved with an electron test beam is
The position resolution obtained at 25GeV is 1mm and the time resolution better than 300ps.
