For background rejection, it is desirable to locate the vertex of the
decay to <5 mm transverse to the beam and to identify other charged
particles in the target in time
coincidence with either the incident 
or with the outgoing 
.
Inert material in the target was minimized to achieve good particle
identification from energy-loss measurements of the stopping kaon
and of charged decay products, and to
maintain good photon detection efficiency. To meet these specifications
plastic scintillating fibers were chosen [12].
The good time resolution necessary both for background
rejection and for triggering purposes was obtained by using fast PMTs.
The 2-mm-diameter fibers were drawn in 200-m lengths by Optectron Corp.
(Les Ulis, France) with a core of S-101 styrene scintillator
(refractive index n= 1.59), and a vinyl-acetate (n= 1.46) cladding.
In the core, butyl-PBD
and POPOP shift the wavelength peak to 440 nm. The 25-
m cladding prevents
damage to the clad-core interface where total internal reflection up to
6
captures 8% of the light.
A 1000-Å Al coating was sputtered onto the fibers by Vacotec
(Reconvilier, Switzerland) to provide a
protective layer for the fragile vinyl-acetate cladding, and to prevent
possible crosstalk of light from one fiber to another.
The fibers were cut into 3.12-m lengths and one end was
polished for testing of light output using a 
Ru source.
Similar tests were repeated at each stage of subsequent
fabrication to maintain quality.
The typical fiber had an attenuation length of about 2 m
but variations of more than a factor of two were observed.
A bundle of six fibers was epoxied together in a triangular aluminum mold
leaving 1 m free of epoxy at one end for routing to a PMT.
Fibers were matched for grouping so that variation in light
output from fiber to fiber in each bundle was kept to <10%.
The sizes of the triangular bundles were kept
identical to a tolerance of 
m to enable precise interlocking
to form the target. The resulting overall target is 75% active,
the inert fraction being mostly epoxy.
An Al mirror was vacuum evaporated onto the polished upstream ends of the triangles in order to reflect scintillation light back towards the PMT mounted at the downstream end. This increased the response for light generated close to the mirror by a factor of 1.6. The Al mirror coating was about 1% transmitting in order to admit light from a light-emitting diode (LED) calibration system. The layout of the target triangles, looking downstream, is shown in Fig. 5.
Figure 5: Schematic showing the arrangement of the scintillator-fiber
triangles and the surrounding I- and V-counters.
The unepoxied downstream region of each bundle was fitted with a black teflon sleeve, a brass mounting collar, and an UVT acrylic light-mixer block glued onto the six polished fiber ends for coupling to the PMT. At the target center is a single fiber coupled to its own PMT.
Each of the 379 bundles is coupled to a 10-mm-diameter
Hamamatsu R1635-02 PMT in a miniature
spring-loaded assembly.
The PMTs have high-gain first dynodes for good one-photoelectron
sensitivity. The relative quantum efficiency was determined for each
PMT and matched to the brightness of each triangle. After the target
was installed in the detector, average signals from muons
in 
triggers were used to set gain values.
Typically, the PMT readout yields
one photoelectron per mm path for minimum ionizing particles.
Each PMT signal is connected to a 
10 amplifier
feeding a splitter-filter box. From there, signals are distributed to the
ADC and TDC systems,
and to the energy-sum units used in the fast trigger
to require an energy-sum pulse of over 5 MeV coincident
with the kaon beam logic.
For each triangle, the TDC data are
recorded for 
500 ns with respect to the event.
A signal for the number of triangles hit is fed to the trigger system
as an
estimator of the range of the decay particle in the target in order
to reject 
decays. The
threshold is set below the pulse height from a single
photoelectron.
Fig. 6a shows an event display of the target ADC information,
Figure 6: Detail display showing the target elements struck by a stopping
and the 
from its decay. (a) shows the calibrated energy in MeV
deposited in each element and (b) shows the time of the hit in ns.
looking downstream, for a typical 
event. The TDC information
for the first hit in the time window 
50 ns for the same event is
shown in Fig. 6b.
The time accuracy is limited by the small number
of photoelectrons seen in each triangle and the intrinsic time response of
the scintillation material.
The leading edge time from each triangle, obtained from the first
photoelectron, can be weighted by the energy
to arrive at the best estimator for the kaon or pion time. The
resulting time resolution from 
calibrations is 700 ps.
The time information is crucial to efficient background rejection
because only hits that are coincident in time with
either the K or 
need be considered. Tracks that occur at
other times are indicative of accidentals from other beam particles and need
not necessarily cause the event to be rejected.
The time difference between the K and 
is used to
eliminate in-flight decays.
In off-line analysis a pattern recognition program sorts the hit triangles into
clusters of stopped kaons, charged decay products connected with a stopped
kaon, separate photon conversions, and random backgrounds by using the target
data for position, energy, and time.
Candidate kaons or decay products are identified, to
first order, using the energy-loss information which is corrected
for the 
25% inert material. K decay
products within the detector tracking acceptance are close to
minimum ionizing and travel within 30
of the perpendicular
to the fibers and therefore deposit <2 MeV per triangle. The stopping
kaons, however, travel parallel to the fibers with a large 
and
typically
deposit >5 MeV per triangle. Each kaon excites up to 5 triangles,
and the pion excites up to 25 triangles, depending upon the origin of
the vertex in the target. Energy and path-length information from the
target are used to correct the total
energy and range determination for charged decay particles.
Hit triangles that are disconnected from the
kaon or pion tracks but still in time with the event
can be used to identify photon
conversions from 
decay in the target.
The spacial resolution in the xy plane is 1.8 mm (rms) transverse to a track in the target, consistent with the dimensions of the triangles. The resolution on the vertex position longitudinal to a track is 4.0 mm (rms). The longitudinal resolution is dominated by the size of the cluster of triangles hit by the stopping K, since the pion energy is not resolved within the stopping cluster. Using the vertex determination, a correction for the energy loss of the pion in the target is applied to the energy, range and momentum measurements obtained from the range stack and drift chamber.
In addition to the hexagonal array of scintillator fibers, the target assembly
includes two sets of six conventional plastic scintillators which surround
the target as shown in Fig. 5. The first set
consists of the I-counters, 240-mm long 
6.4-mm thick, which define the fiducial stopping region by tagging charged
decay products after a kaon stop and before they enter the drift chamber.
Including the I-counters, the target assembly in the fiducial region
is 127-mm across, from flat to flat.
The second set, the V-counters (1960-mm long 
5-mm thick),
overlaps the downstream end of the I-counters by 6 mm and helps to
distinguish hits downstream of the fiducial region of the target
fibers which serves
to veto various background processes. The I- and V-counters are viewed
through UVT lightguides by EMI 9954KB PMTs.