The primary functions of the cylindrical drift chamber [13], which
subtends a
solid angle of 
sr for kaon decay products of momentum P> 150 MeV/c,
are to provide a momentum measurement in the 1-T magnetic field and to
achieve good tracking between the target and the range stack. Momentum
resolution of 
(rms) is required to obtain the necessary level
of background rejection related to 
and 
decays.
The chamber is required to have a short memory time in order to operate
successfully in the presence of high rates of kaons and pions in the beam.
The chamber wires are arranged in five superlayers of 36, 40, 50, 60, and
70 cells, respectively.
The wires of layers 1, 3 and 5 are strung axially, and those
of layers 2 and 4 are strung between feedthroughs offset by one cell,
corresponding to stereo angles of approximately 
to enable axial position measurements. The configuration of wires
and wire diameters within
a unit cell is shown in Fig. 7a.
Figure 7: (a) Schematic showing the wire geometry in a drift cell.
(b) Trajectories of ionization electrons to the sense wires in a 1.0-T magnetic
field.
Cells within a layer are identical, but
cell size varies from layer to layer between 12-mm and 17-mm half width.
The cathode-wire plane, consisting of
19 Be-Cu wires strung 2.54-mm apart to a tension of 150 g , is shared by
neighboring cells. The sense-wire plane has two Be-Cu end guard wires and
eight 20-
m-diameter gold-plated tungsten anode wires spaced
5.08 mm apart at a tension of 50 g and staggered 
254 
m
from the midplane to resolve the left-right ambiguity locally within the cell.
The end wires on both anode and cathode planes are of larger diameter as
shown in Fig. 7a to optimize electric field uniformity and to
maintain surface fields <20 kV cm
on the cathode wires when
those on the sense wires are about 250 kV cm
.
This arrangement without focussing or potential wires in the anode plane
provided a high electric field in the drift region of 
kV cm
,
and reduced the Lorentz angle.
To provide the correct potentials at the cylindrical boundaries of
the chamber, thin copper-plated Mylar foils were attached to the
graphite-epoxy support cylinders and set at the appropriate voltages.
The trajectories calculated for a typical cell,
of ionization electrons in a 1-T magnetic
field are shown in Fig. 7b.
Only the six middle anode wires are read out in order to avoid the cell-end
distortions in the electric field and to reduce the effects of the Lorentz angle
which is about 25
for a drift velocity of 5 cm 
s
.
The chamber is operated at atmospheric pressure with a
gas mixture of Ar : C
H
= 50 : 50. The argon fraction is
bubbled through ethanol at 0
C.
The chamber occupies the radial region between 95 mm and 432 mm with a
total length, including preamplifiers, connectors, and cables of 650 mm.
The active volume, 508-mm long, is enclosed between 9.5-mm-thick
precision-machined Al endplates carrying the slots to receive the
feedthroughs which position and support the wires. Each feedthrough
is made up of a machined comb of Vespel epoxied into an
injection-molded Ryton insert which is epoxied into the slot.
The overall mechanical tolerance for the location of each wire
was measured to be 
20 
m. The endplates are supported
by inner (80-mg cm
thick) and
outer (94-mg cm
thick) cylinders made of graphite-fiber epoxy, which was
built up of four (inner) and five (outer) pre-impregnated layers
with fiber orientation alternately axial and 
. The endplates were
prestressed prior to stringing with a system of temporary tie-rods
which were adjusted and then removed as wires were installed, taking up
the tension load. The total endplate loading due to wire tensions is 880 kg.
The outer cylinder was
installed after stringing was complete. During stringing, its compression
load was taken up by a set of temporary posts. Subsequent replacement of
individual broken or faulty wires has been done with the outer cylinder
in place. The chamber is supported in the detector by a 3.2-mm-thick
aluminum cylinder which extends from the magnet
pole and is attached at the circumference of the downstream endplate.
Contact to each wire for high voltage (HV),
ground, or onboard preamplifier,
is via a printed edge connector card fitted into the feedthrough, to which
the wire was soldered after tensioning. For the cathodes, HV is distributed
to one third of a layer from one of 15 Bertan 1755N HV power supplies.
For each of the 15 sections of HV distribution the
voltage is graded in a resistor chain to compensate for the variation in cell
width with radius in order to form a uniform drift field
throughout the cell. The external regions
of both endplates, on which the HV distribution and preamplifiers are
mounted, are sealed and dry N
is circulated to control temperature and
to eliminate HV breakdown due to atmospheric humidity.
Each sense-wire plane has six preamplifiers mounted on a card plugged onto
the downstream feedthrough and grounded locally to the endplate. A typical
preamplifier gain is 10 mV 
A
with a risetime of 3 ns and power
dissipation of 24 mW. The positive-ion-induced crosstalk of signals
between adjacent wires of 
15% is reduced to
by a compensation resistor
network which splits the input charge at the front ends of the preamplifiers.
Each preamplifier drives 34 m of 50-
coaxial cable which degrades
the signals to about half amplitude and risetime of about 7 ns
when they reach the post-amplifier in the counting house.
All cables, including signal, HV, preamplifier power,
and test pulse, exit the magnet and go through a bulkhead
connection at the downstream end of the detector.
The post-amplifier-discriminator circuits are housed, 24 per module,
in a DIN 41494 standard sub-rack system (Euro-crates).
The input signals are decoupled from ground by transformers
and split for separate analog and discriminator processing. The gains of the
amplifiers are 7--10 for the analog output and 100 for the discriminator
path. Pole-zero cancellation is employed to minimize fall time and reduce
the pulse tail, due to positive ions and to the 34-m cable, to a level
of 
. The discriminator produces a time-over-threshold output with
a 10-ns minimum width and feeds an LRS 1879 Fastbus pipeline TDC for
drift time (up to 350 ns) measurement.
The drift velocity v, Lorentz angle 
, and time offset (pedestal)
are determined iteratively by fitting tracks in the chamber.
For v, zero-magnetic-field data were used.
Position resolution obtained
from 
data, shown by the residuals from track fits
in Fig. 8,
Figure 8: Typical local position resolution in the drift chamber
as a function of drift distance for stereo and axial layers.
varies with drift distance and layer between 130 and 250 
m
for the axial layers. The z (axial) resolution obtained from the
stereo layers is between 2.2 and 4.2 mm.
The momentum resolution, after correcting for the
energy loss in the target, is shown in Fig. 9.
Figure 9: Momentum for 
events corrected for the energy loss
in the target system.