\author{M. D. Mestayer$^a$, Your Name Here$^b$ }
\maketitle
\abstract
{\ \ \ \ \ Experimental Hall B at Jefferson Lab (formerly CEBAF)
houses CEBAF's Large Acceptance Spectrometer (``CLAS") which is
based upon a six-coil superconducting toroidal magnet.
The magnet coils divide the detector azimuthally into six ``sectors".
Within each sector are located three types of drift chamber, for a total
of eighteen chambers (~34,000 instrumented sense wires).
We describe the design, construction, installation, commissioning and
calibration of the chambers using recently obtained electron scattering
data.
\renewcommand{\thesection}{\Roman{section}}
\renewcommand{\thesubsection}{\thesection-\Alph{subsection}}
\renewcommand{\textfraction}{0.1}
\renewcommand{\topfraction}{0.9}
\renewcommand{\bottomfraction}{0.9}
\section{CLAS Tracking System Design }
\subsection{Design Overview}
*** 6 sectors in toroid - independent tracking - bend radially -
-good acceptance - shadow area - mom. resln. of .1 -1% needed -
200 micron resln. consistent with mult. scatt. - shape of chambers -
figure - overall detector showing particle trajectories ***
The CLAS detector is based upon a 6-coil toroidal magnet which bends
negatively (positively) charged tracks in toward (or away from) the beam
axis (``z axis").
The six coils naturally separate the detector into six independent tracking
areas, the six ``sectors". Within each sector,
track position information in three radial locations (near the target,
near the point of maximum track sagitta within the B field region,
and outward of the high-B region) allows us to reconstruct the
initial angles and curvature of charged particles which traverse
the chambers.
We decided on a design with three separate chambers in each of the six sectors.
Six ``Region 1" chambers surround the target in a region of low magnetic field.
Six ``Region 2" chambers are somewhat larger and are situated between
the magnet coils in a region of high field, while the six ``Region 3"
chambers are large devices located radially outward of the magnet.
Figure \ref{f:clasview} shows a cross-section of the CLAS detector cut
on a horizonal plane through the beam line and target.
\begin{figure}[p]
\vspace{4in}
\caption[CLAS Plan View]{Horizontal cut through CLAS detector at beam
line elevation showing a charged particle traversing the drift chambers.}
\label{f:clasview}
\end{figure}
Our physics design goals were to
\begin{itemize}
\item reconstruct the trajectories of
all charged particles which do not strike the magnet cryostat,
\item which have momenta greater than $100 MeV/c$,
\item and which lie within the polar
angle range of $8^{\circ}$ to $142^{\circ}$
\item track resolution goals
were $\delta p/p \le 0.5 \%$ (FWHM) and $\delta \theta ,\delta \phi
\le 1$ mrad.
\end{itemize}
To achieve these goals, the chambers are required
to measure particle trajectories at three positions along the track
to an accuracy of $100 \mu m$ (in the bend plane) and to be constructed out of
materials which will contribute less than $1 \%$ of a radiation length
to multiple scattering.
\subsection{Superlayer Design }
*** figure - superlayer layout, with isochrones ***
The toroidal magnetic field was built such that it deflects charged particles
toward (or away) from the beam axis but leaves the azimuthal angle unchanged.
For maximum sensitivity to track curvature the
wire direction is perpendicular to the bend plane, i.e. approximately
parallel to the B field.
To minimize multiple scattering, we have avoided an arrangement
in which the whole tracking volume is densely filled with wires and chamber gas.
Instead, the wires are grouped into three regions radially, to be separated
in the future by helium-filled gas bags.
For tracking redundancy we grouped the wires from each reagion
into ``superlayers" of six layers
each. Thus, each ``region" is a separate chamber
consisting of two superlayers, with each superlayer consisting of six layers
of sense wires; see Figure \ref{f:superlayout}. Because of space constraints, there are
only four layers in the first superlayer of Region 1.
\begin{figure}[p]
\vspace{4in}
\caption[Wire Superlayer Layout] {Sense, field and guard wire locations within
a ``superlayer" arrangement. }
\label{f:superlayout}
\end{figure}
The two superlayers in each chamber are of two types, ``axial" and ``stereo".
Wires in the ``axial" superlayer are normal to the mid-plane of
the sector, while wires in the ``stereo" superlayer are oriented at
an angle of 6$\degg$ w.r.t. the axial direction.
A resolution of $200 \mu m$ per layer combined with the $6^\circ$ stereo angle
will achieve the design resolution in each region of
$100 \mu m$ in the bend plane and 2 mm in the coordinate along the
wire.
Each sense wire layer consists of 128 wires in Region 1, and of
192 wires in Regions 2 and 3. This leads to maximum drift distances
of 0.8, 1.3? and 2.3 cm in the three regions respectively.
For the eighteen chambers (three regions times six sectors),
there will be a total of about 34,000
instrumented sense wires.
\subsection{Cell Type}
*** figure - cell with isochrones ***
The cell type which we have used
is a ``small-cell" hexagonal arrangement; that is,
each sense wire is surrounded by six field wires
in an hexagonal arrangement.
Each superlayer begins with and ends with a layer of guard wires whose
potential is adjusted to reproduce the electric field configuration
given by an infinite grid of hexagonal cells; i.e. the electric field
at the sense wire surface is the same for all sense wires,
independent of layer number.
Figure \ref{f:isocrhones} shows the cell layout and
a track which traverses one superlayer,
viewed along the wire direction. The figure shows a single event display of
a cosmic ray track. The hit wires are indicated with circles where the
radius of the circle is the distance of closest approach of the track as
calculated from the time of the hit.
\begin{figure}[p]
\vspace{4in}
\caption[Isochrones] {Plot of equal-time contours for a give sense wire cell
in Region 3. }
\label{f:isochrones}
\end{figure}
We have chosen the hexagonal configuration for several reasons.
In a typical CLAS data sample, there will be tracks traversing
the chambers at widely varying angles.
In contrast to jet cells, which are optimized for tracks with a particular
direction, hexagonal cells offer a reasonable
approximation of the ideal circular cell in which the drift-time relation
is independent of entrance angle.
In addition, the `brick-wall' pattern of the layout enables easy resolution
of the left-right ambiguity, gives a high degree of segmentation and thus a
good high-rate performance,
and is also quite effective at discriminating against out-of-time tracks.
Finally, the hexagonal layout is effective at reducing the electrostatic
forces on the wires, allowing the use of low tension on the wires.
We have modelled the electrostatic and drift time properties of
the hexagonal cell design. To accomplish this, we have extensively used the
drift chamber simulation
program, GARFIELD\cite{GARFIELD}.
Figure 4 shows the
isochrone contours ($100 ns$ intervals)
for a representative cell in Region 2 at a
typical magnetic field strength of 1 Tesla. Note that the contours
are circular close to the wire but become hexagonal
near the outer boundaries of the cell. This illustrates the necessity of
knowing the track's entry angle in order to deduce from the drift time
the distance of closest approach to the sense wire.
In a later section we discuss the function used to describe the time to
distance correlation.
\subsection{Gas Choice}
*** fig- CED event w/ background, velocity vs. E/P for AR/CO2 ***
The requirements of low multiple scattering, ability to handle
high rates, good spatial resolution and
modest dE/dx capability lead to our choice of gas type.
Typical electron scattering events at CEBAF
are expected to have a low
multiplicity (1 to 5 tracks per event) but to occur at a rate as high
as 1 MHz for our design luminosity of $10^{34} s^{-1}cm^{-2}$.
Although the occupancy from real, in-time tracks is low, our estimates of
electromagnetic and hadronic backgrounds necessitates a fast gas.
\begin{figure}[p]
\vspace{4in}
\caption[Event with background] {Typical event with background}
\label{f:bkgrnd event }
\end{figure}
Our preliminary choice of gas was a
$50/50$ argon ethane mixture\cite{Sauli}.
In this mixture,
the primary ionization clusters are spaced about $300 \mu m$
apart; for a track very close to the sense wire this will give an
inherent spread of about $150 \mu m$. The diffusion term
is about $100 \mu m/\sqrt{cm}$ of drift. In addition, the effect of
time slewing in the electronics is expected to be of order $1 ns$,
or about $20 - 50 \mu m$ for spatial resolution.
From the above estimates, we concluded that resolutions of about
$200 \mu m$ per wire should be achievable.
We planned to
reduce multiple scattering in the Region 2 chamber
(where it has the greatest effect on momentum resolution) by using
an admixture of helium\cite{Boyarski}
in the argon-ethane mixture; probably at
the level of $50 \%$ helium. In a
test of a prototype drift chamber,\cite{Brookhaven}
we showed that tracking resolutions
are not considerably worsened by using an admixture of helium.
Although our choice of a large helium admixture
is motivated by the desire to reduce multiple scattering,
a beneficial side-effect is that the dependence
of the drift velocity on
the (non-uniform) magnetic field of the toroid is greatly reduced.
In 1996, safety concerns mandated that we not use a flammable mixture.
Additional concerns about small gas leaks and the proximity of many
photomultiplier tubes argued against Helium mixtures.
In trying to find a non-flammable alternative to Ar:Ethane, we studied
Argon-C02 mixutes finally settling upon a $90/10$ argon:CO2 mixture.
This gas has several drawbacks: lower velocity at low electric fields,
increased multiple scattering relative to Helium mixtures, and a short
voltage plateau due to lack of quenching. Nevertheless, the mixture
provides adequate tracking resolution and reasonable collection times.
\subsection{Wire Choice}
The use of small wire diameters and low wire tension
makes possible the design of thin endplates
which maximizes solid angle coverage.
The dual requirements
that the wire tension be minimized and that the gas gain
be a few times $10^4$ require that we use a small diameter ($20\mu m$)
sense wire and an operating voltage of about 2.5 kV. For these conditions,
the electric field at the
surface of the sense wire will be approximately $280 kV/cm$.
In addition, our design
must ensure that the electric field at the surface of the field wires
remains below 20 kV/cm to minimize
conditions causing cathode deposits.\cite{Kadyk}
Since there are roughly twice as many field as sense wires, this
requires that the diameter
of the field wires be about seven times the diameter of the sense wire;
in this case, about $140 \mu m$ diameter.
The choice of wire is $20 \mu m$ diameter gold-plated tungsten for
the sense wire and $140 \mu m$ gold-plated aluminum for the field wires.
Gold-plating gives chemical inertness as well as a smooth surface finish.
We have chosen tungsten because of its
durability whereas aluminum was chosen for the field wires
because it has the longest radiation length of any
practical wire material and thus minimizes multiple scattering. In
addition, the low density of aluminum means that the field wires
can be strung at lower tension
than a more dense wire and still keep the same gravitational sag.
This minimizes the forces on the endplates.
\subsection{On-Chamber Electronics Design}
*** space constraints, signal distribution,
gain, amp./cable type, power distribution
fig - shadow area ***
\begin{figure}[p]
\vspace{4in}
\caption[Shadow Area] {Cut perpendicular to cryostat showing the chamber
endplates and electronics which lie in the shadow of the cryostat coil.
The shadow region is the volume which is outside the cryostat but which is
not in direct sight of beamline. }
\label{f:shadow}
\end{figure}
\subsection{Off-Chamber Amplifier/Discrimination Design }
*** gain, rise-time, trigger logic, calib. pulsing ***
\subsection{Off-Chamber TDC Design}
***accuracy needed, wire decoding, calibration circuitry ***
\section{Construction }
\subsection{Chamber Construction and Stringing }
\subsubsection{Region 1}
*** endplate design/feedthroughs, box design/gas bag,
stringing/tension control, electronic board installation/grounding, gas
distribution ***
\subsubsection{Region 2}
*** endplate design/feedthroughs, box design/gas bag,
stringing/tension control, electronic board installation/grounding, gas
distribution ***
\subsubsection{Region 3}
*** endplate design/feedthroughs, box design/gas bag,
stringing/tension control, electronic board installation/grounding, gas
distribution ***
\subsection{On-Chamber Electronics}
*** fabrication, installation ***
\subsubsection{SIP Amplifiers}
\subsubsection{Signal Translation Boards}
\subsubsection{High Voltage Translation Boards}
\subsection{Cables}
\subsection{Amplifier-Discriminator Boards}
\subsection{Trigger-Calibration-Interface (TCI) Boards}
*** figs - pulser calib concept ***
\begin{figure}[p]
\vspace{4in}
\caption[Pulse calibration schematic] {Schematic of pulse distribution system
used to cross-calibrate the fixed delays of all sense wire channels. }
\label{f:pulsecalib}
\end{figure}
\subsection{Multiplexer Boards }
*** figs - signal de-muxing ***
\begin{figure}[p]
\vspace{4in}
\caption[Signal Multiplexing] {Schematic of method to encode wire number
information into TDC pulse widths. }
\label{f:mux}
\end{figure}
\subsection{TDC System}
\subsection{High Voltage System}
\subsection{Low Voltage System}
\subsection{Gas System}
\section{Installation and Survey}
\subsection{Region 1}
*** transportation, six-sector assembly, installation, survey ***
\subsection{Region 2}
*** transportaion, installation, tension transfer, survey ***
\subsection{Region 3}
*** transportation, installation ***
\section{Commissioning}
\subsection{Initial Check-out}
***accidental disconnects, HV shorts, dead channels, super-hot wires ***
\subsection{Setting Discriminator Thresholds}
***specific noise studies, noise vs. disc. scans ***
\subsection{Setting High Voltage Values}
*** efficiency plateaus ***
\subsection{Study of Noise Sources }
*** shadow tracks, border noise, shotgun noise ***
\subsection{Monitoring Performance}
*** quick response to malfunctions ***
\subsubsection{Raw Data Monitoring}
*** time windows, de-mux'ing, multi-edge noise ***
\subsubsection{Hit Monitoring}
***occupancy plots ***
\subsubsection{Times Monitoring}
*** time distributions, time difference ***
\subsubsection{Tracking Monitoring}
*** trks/evt, hits/trk, residuals ***
\subsection{Data Acquisition Issues}
*** FASTBUS, CODA ***
\section{Tracking/Calibration}
\subsection{Charged Track Reconstruction Overview}
*** find e, find start-time, trks' momentum, beta from TOF ***
\subsection{Time Delay Calibration}
*** relative fixed time delays wire-to-wire ***
\subsection{Drift Time Determination}
*** flight time, signal prop. time, time walk ***
\subsection{Time to Distance Calibration}
*** non-linear, angle-dependent ***
\subsubsection{Function Parameterization}
*** constraints, figures ***
\begin{figure}[p]
\vspace{4in}
\caption[Distance versus time plot] {Scatterplot of tracks' distance
of closest approach to a wire versus the corrected drift time from that
wire.}
\label{f:xvst}
\end{figure}
\subsubsection{Calibration Procedures}
*** fit data for T0, TMAX, etc. ***
\subsubsection{Laser Calibration Chambers}
*** scale TMAX ***
\subsection{Results: Resolution, Efficiency, Backgrounds}
\end{document}