1 - Istituto Nazionale di Fisica Nuclearepeople.na.infn.it/~palladin/mice_proposal1.doc · Web viewIt has been chosen to use MS Word to edit the proposal. Word is easy to use, but

Proposal for an International Muon Ionization Cooling Experiment1

Editor’s note:It has been chosen to use MS Word to edit the proposal. Word is easy to use, but it is also easy to create a mess. Please observe the following very short guidelines in order to make the editor’s life much easier.

Figures, Tables, Equations If you insert figures, please have the original file at hand, as we may need it. To insert captions for tables or figures please use InsertCaption and then the

according item (table or figure). All headings, tables and figures are automatically numbered. Please do no enter any

of these numbers by hand. To insert a cross-reference in the text, please use InsertCross-reference. There is no convenient way to automatically number equations in word, so you have to do this by hand. Please use (chapter.number).

To create Equations, please use InsertObjectsMicrosoft Equation 3.0

Page format, printing Under no circumstances, change the FilePage Setup. To print on a letter-

sized printer, use the PrintOptions… dialog.

Symbols, greek letters To insert symbols or greek characters, use InsertSymbol. This may seem more

work than writing the letter and choosing the Symbol font, but it avoids a principal source of errors.

References For references, please use the Harvard style. Just write [Nam02] in the text. At the

end of the text, simply collect all references. You do not need to sort them. Please use a [TAB] key between the reference and the author. Example:

[Nam02] N. Name: A paper on writing, CERN Neutrino Factory Note 79 (2002)

Headings, styles, copy&paste To add headings, subheadings etc please use the styles which are normally located at

the upper left corner of your screen, just below the File menu. Please use the styles Heading 1, Heading 2, Heading 3 accordingly. Other styles used are caption, reference. The “normal” text should always use the style Normal.

Please do not use any “manual” formatting except Bold or Italic. Please do not use the Format->StyleModify command. When you have trouble with copying/pasting text from other word documents or you

get unexpected results, try EditPaste Special and select Unformatted Text.


When asked: please select “reapply”


Proposal:

An International Muon Ionization Cooling Experiment (MICE)

(List of authors)


1 General motivation (S.Geer + K. Long), 6 pages

1.1 Neutrinos and our lives (PR motivation)

1.2 Neutrino oscillations and the role of NuFact (Particle physics motivation)

1.3 other physics at NuFact complex and Muon collider

1.4 Neutrino Factory description (the accelerator)

1.5 Role of cooling in Neutrino Factory and muon collider, the need for MICE (Acc. physics motivation)


2 Cooling (Kaplan), 5 pages

2.1 Principle of Ionization cooling

2.2 Cooling channel design

2.3 Possible extensions (gas cooling and ring coolers)


3 The International Muon Ionization Cooling Experiment (Blondel, Hanke, Palmer), 8 pages

3.1 Layout and principle

3.2 Experiments to be performed

3.3 Expected performance and statistical precision

3.4 Systematic errors and resulting requirements

3.5 Possible development of the experiment in time


4 The magnetic channel (Green/Rey), 8 pages

4.1 Layout and optics (Palmer / Fernow)

4.2 Design of coils (Green/Rey)

4.3 Services (->move to infrastructure chapter 10)

4.4 Magnetic measurements (Rey)

4.5 Return flux and constraints

4.6 Responsibilities cost and time line


5 The RF system (Li, Haseroth), 8 pages

5.1 Cavities

5.2 Windows

5.3 Power supplies

5.4 Achievable performance

5.5 Services and safety (->move to infrastructure chapter 10)

5.6 Responsibilities, cost and time-line


6 The absorbers (Cummings/Ishimoto), 8 pages

6.1 General design (incl. safety constraints)

6.2 Windows

6.3 Thermodynamics (heat exchanger, etc..)

6.4 Accommodating different length and materials

6.5 Responsibilities, cost and time-line


7 Beam and beam line (Drumm), 6 pages

7.1 Layout and location

7.2 Expected performance and time structure

7.2.1

7.3 Time line


8 Detectors (Bross/Palladino), 15 pages

8.1 Overall description, functionality and redundancy

8.2 8.2 TOF and trigger

Draft 0.2 Maurizio, Mauro, Alberto 7/10/2002

Our present preliminary baseline design of the detectors devoted to the measurement of the complete set of muon parameters is sketched below, along with a few variants. The desired active area of the detectors is a circular region with a diameter of 300 mm.

8.2.1 Scintillators for trigger, timing and particle identification by time-of flightThe present baseline design foresees three fast scintillator equipped TOF stations . The first two stations (TOF I and TOF II) , spaced by about 10 meters, will provide the basic trigger of the experiment, in coincidence with the beam particle signal. These two stations will have precise timing (between 50 and 100 ps) and will provide muon identification by time-of-flight (TOF). The second of these stations will also provide the muon timing (vis a vis the RF phase) necessary for the measurement of the input emittance.

The coincidence with a third scintillator station of similar nature (TOF III), downstream of the second measuring station, will select particles traversing the entire cooling channel. The variation of emittance due to losses will thus be distinguishable from the one truly due to the action of the cooling channel. This station will of course also record the muon timing for the measurement of the output emittance.

In the present baseline design the three TOF stations are 12x12, 40x40 and 40x40 cm2

respectively. The two largest stations (TOF II, TOF III) can be equipped with 8 scintillator slabs, to make a 40x6x2.5 cm3 plane (Y), by staggering and superimposing them at the edges for about 1 cm to allow cross-calibration with impinging beam particles. Bicron BC-404 (with 1.5 ns decay constant and 1.7 m attenuation length) is the most suitable choice for the scintillator material. The smallest plane (TOF I) could be made of two crossed planes (X-Y), each of two 12x6x2.5cm3 slabs, using BC-420 plastic scintillator, which is even faster than BC-404 but with a shorter attenuation length. Each slab, after a plexigas lightguide, is read at the two extremes by a fast photomultiplier. The analog signal is then fed, after a RG-213 cable 1 , to an active splitter chain that divides the signal 25% to the ADC line2 and 100% to a leading edge discriminator followed by the TDC line3. The time-of-flight measurement is achieved combining leading-edge time measurements (from the TDC) with pulse-height informations for time-walk corrections (from the ADC).

TOF I could be equipped with Hamamatsu R4998 (20 mm useful diameter, 0.7 ns rise time,160 ps transit time jitter) PMT’s.

1 Used instead of standard RG-58 coaxial cables, to minimize distortion2 the use of CAEN V792 QDC is foreseen3 the use of CAEN V775 TDC, 12 bits, 32 channels modules is foreseen


Being the fringe fields of the solenoids of the cooling channel up to 1 Tesla, the choice of the PMTs of TOF II and TOF III is a critical issue. One option is to use the same R4998 PMTs but with multiple mu-metal shielding (in principle capable to reduce the transverse component of the magnetic field of a factor around 106) and a suitable design of the light guides. An alternative option is the Hamamatsu R5505 fine mesh PMT (17.5 mm useful diameter, 1.5 ns rise time, 350 ps transit time jitter) which can operate beyond 1 Tesla but with a reduced gain (1.8 104 against 5 10 5). Similar solutions were successfully adopted in the BESS experiment 45 . Figure 1 shows the results of a simulation of the fringe field of one solenoid of the cooling channel done with the Poisson-Superfish program 41. In the position where TOF II and TOF III will be put stray magnetic fields up to 1 Tesla can be expected.

Figure 1. Simulation of the longitudinal component Bz (a) and the transverse component Br (b) of the stray magnetic field at the exit of the solenoid of the cooling channel (z= 0 cm).

Studies with several scintillator counters equipped with both types of PMTs will be performed at INFN Milano and INFN Padova. and in the free air bore of a large superconducting solenoid facility 42 at the INFN Milano LASA laboratory with fields up to 4 T. A testbeam at CERN PS is also foreseen, before taking a final decision. Budget requests for the initial phase of these tests have already been submitted and granted by INFN.

For the time inter-calibration of a single detector plane we propose to use cosmic rays, with a dedicated setup for the trigger (as done in the Harp experiment 43) or to use beam passing


particles in the overlap region of two nearby counters. The time monitoring of the system will be done with a laser-based system, similar to the one presently used in the Harp experiment [44] and shown in figure 2. Studies are under way to assess if the expensive laser system 4 used in the Harp experiment could be refurbished.

Figure 2. Layout of the laser calibration system of the Harp experiment at CERN PS. A similar layout is foreseen for the three walls of the MICE TOF system. A fast laser pulse is injected into the scintillator bars via optical fibers, giving the TDC STOP, while the TDC START is obtained by beam splitting the laser pulse to a fast PIN photodiode (30 ps risetime).

Performance ranging between 50 and 90 ps intrinsic resolution has been already published for TOF planes of similar or slightly bigger dimensions[45], [46]. A 50 ps resolution fits quite well the requirements of this experiment. Assuming that the electronics and calibration system of the TOF wall of the HARP experiment can be reused, a prelimary cost estimate for scintillator, PMT’s and general reconfiguration would amount to about 200 k€ (details areShown in table I)..

4 SYLP0 by Quanta Systems srl, Milano, Italy

Item Cost (KE)

Detector system

40 PMT + mu metalScintillator (BC-404 and BC-420)Light guidesMechanics

100 8

7 5

Calibration system

Laser Optical system Cosmic rays setup

80 (*) 15 5

Electronics

QDCTDCScalers, MT, NIM modulesDelay boxesActive splittersCrates, HV systemdiscriminatorsHV cables, signal cables

10 10(*) 10(*) 5 (*) 5(*) 10 20 10(*)

Total 300(180 excluding items marked as (*))


Table I. Cost estimation for the TOF system (only capital investment). Items marked (*) can be eventually recuperated from the HARP experiment.

This simple technique for muon identification appears adequate if the contaminations of pions and electrons in the beam are indeed as low as presently estimated.

[41] Reference Manual for the POISSON/SUPERFISH Group of Codes, Los Alamos National Laboratory Report LA-UR-87-115.

[42] E. Acerbi et al., INFN/TC-92/23, October 1992 M. Bonesini et al., INFN/AE-94/14, May 1994[43] G. Barichello et al., “ The Harp Tof Wall Construction and Test “ , INFN/AE-02/02,

July 2002[44] M. Bonesini et al., “ Construction of a Fast Laser-based Calibration System for the

Harp Tof counters wall “, INFN/AE-02/02, July 2002[45] BESS Collaboration, NIM A455 (2000) 596-606;

S. Geier et al., Proceedings of the 26th ICRC, Salt Lake City, 1999, 3, 117.[46] NA49 Collaboration, NIM A430 (1999) 210-244, NIM A451 (2000) 406-413;

NA52 Collaboration, Nuclear Physics A 590 (1995) 347C;Pamela Collaboration, 27th International Cosmic Rays Conference, OG 2215, Hamburg, August 2001, http://wizard.roma2.infn.it/pamela/fram_des.htm

http://wizard.roma2.infn.it/pamela/fram_des.htm


%\documentstyle[12pt,psfig,epsfig]{article}\documentclass[12pt]{article}\usepackage{psfig}\usepackage{epsfig}

\overfullrule=0pt\hbadness=10000\vbadness=10000\tolerance=16000

\textwidth 6.0in\textheight 8.5in

\topmargin 0.0in

\evensidemargin 0.30in\oddsidemargin 0.30in \begin{document}

%\include{hep_symbols}\hfill {MICE Note} \\\large\begin{center}{\bf Cherenkov Detector for MICE} \\[0.1in] Don Summers, Lucien Cremaldi, Romulus Godang \\ ( University of Mississippi ) \\

%\date{\today} \\[0.2in]

\end{center}\normalsize

\subsection*{Introduction}\label{sec-intro}

Cherenkov radiators can provide particle identification for the MICEexperiment. Upstream a simple $C_6F_{14}$ radiator can provide ($\pi-\mu-e$) separation for many momenta if pulse height information is used as well as thresholds [see D. Bartlett et al., NIM A260 (1987) 55]. This can beat down any backgrounds that might be left over from the time of flight detector.

The ($\mu-e$) separation in the downstream detector provides a morechallenging problem. Muon decay has a lot of phase space. Electrons arewidely dispersed. Again a Cherenkov counter is used. The radiator is aerogelwith an electron threshold of about 2 MeV/c and a muon threshold of about 400MeV/c. The majority of electrons hit the aerogel, but a fair number miss. With a Belle style detector [R. Suda et al., NIM A406 (1998) 213] withphotomesh PMTs on each aerogel block, all hits register. With a light guide


some high angle hits are lost. In either case, one must rely on finding kinks in the tracking system to do some of the electron ID.

The strength of the Cherenkov detector is to veto forward decays. MICE mustbe able to tell the difference between forward decays that have almost thesame momentum as the parent muon and muons that just did not get cooledproperly. Some tracking may be needed near the Cherenkov detector, even if itis just a scintillator hodoscope.

If X-rays from the RF cavities are below 2 MeV they should not be lead toelectrons that give Cherenkov light. A larger concern is RF noise pickup byphotomultipler tubes. For 201 MHz, the skin depth$(SQRT(2./(2.*PI*FREQ*CONDUCT*U*U0)))$ of pure aluminum is 6 microns. So 100microns of aluminum (17 skin depths) might provide an adequate RF shield. This is the thickness of a Coca Cola can. Copper would be easier to solderto form a sealed container and has a skin depth of 5 microns.

\subsection*{Aerogel Cherenkov Detector}\label{sec-aerogel}

Mississippi proposes to work with the Universite Catholique de Louvain onCherenkov particle ID. The task would be split between the twoinstitutions.

For downstream ($\mu-e$) discrimination, we propose to use a particle IDsystem similar to that adopted by the Belle collaboration at KEK in Japan. It uses silica aerogel blocks with fine-meshphotomultiplier tubes (PMTs) to work in a magnetic field. For details of whatthe detector elements look like see Fig.~\ref{fig:belle_design} and Fig.~\ref{fig:belle_simulation} respectively. \begin{figure}[htbp]\centerline{\psfig{figure=aerogel_belle_97.eps,height=3.2in}} \caption{The aerogel Cherenkov counter design [R. Suda et al., NIM A406 (1998) 213]. \label{fig:belle_design}}\end{figure}

\begin{figure}[htbp]\centerline{\psfig{figure=aerogelmc_belle_97.eps,height=3.0in}} \caption{The aerogel Cherenkov simulation of light generation and collection for single event [R. Suda et al., NIM A406 (1998) 213]. \label{fig:belle_simulation}}\end{figure}

We propose to use two PMTs per aerogel block. We have been making maps of


the solenoidal fringe fields. The phototubes must be in a field less than 1.5 Tesla. The layout would be designed to make it difficult for a particle to pass through both PMTs. Rare cases of double passage could be excluded from analysis. The solenoid opening is large and tracks at the edges in fringe fields are not parallel to tracks in the center. Muon decay will give transverse kicks to some of the electrons. Proximity collection of light will work if a track passes though a silica aerogel block. Electrons give light, muons do not.

Upstream, the beam is smaller and more parallel. We propose to use a singlecell of $C_6F_{14}$ liquid to do ($\pi-\mu-e$) discrimination [L. Cremaldi and D. Summers, "MUON Cooling Experiment - MC Simulationsof Beam Particle ID," MUCOOL Note 0221 (2001)]. The PMTswould be mounted directly on the vessel holding the liquid. Pulse heightinformations used to aid in the discrimination. We have tested a C6-F14Cherenkov detector with cosmic rays in Mississippi as noted in above reference.This is a fairly simple detector that would supplement the time of flightsystem to help beat systematic errors down to a low level.Table~\ref{tab:Cherenkov_thresholds} shows the Cherenkov thresholds forelectrons, muons, and pions.\begin{table}[h]\begin{center}\vspace*{1mm}\renewcommand{\arraystretch}{1.1}\tabcolsep=0.5mm\begin{tabular}{lcccccccc} \hline \hlineMaterial & Boiling & Density & X$_0$ & Length & Refractive & Electron & Muon & Pion \\ & Point $^0$K & g/cm$^3$ & mm & mm/15pe & Index $n$ & MeV/c & MeV/c & MeV/c \\ \hlinePolystyrene & & 1.03 & 424 & 4 & 1.581 & 0.42 & 86 & 114 \\Quartz & 2500 & 2.20 & 123 & 4 & 1.458 & 0.48 & 99 & 132 \\Water & 373 & 1.00 & 361 & 5 & 1.33 & 0.58 & 120 & 159 \\$C_6F_{14}$ & 329 & 1.68 & 206 & 6 & 1.244 & 0.69 & 143 & 189 \\$C_2F_6$ & 195 & 1.61 & $\sim$200 & 7 & 1.222 & 0.73 & 150 & 199 \\LN$_2$ & 77 & 0.81 & 471 & 7 & 1.205 & 0.76 & 157 & 208 \\LD$_2$ & 24 & 0.18 & 7540 & 10 & 1.128 & 0.98 & 202 & 267 \\LH$_2$ & 20 & 0.071 & 8900 & 12 & 1.112 & 1.05 & 217 & 287 \\LNe & 27 & 1.206 & 240 & 14 & 1.092 & 1.16 & 241 & 318 \\


Aerogel & 2500 & 0.30 & 995 & 16 & 1.075 & 1.30 & 268 & 354 \\Aerogel & 2500 & 0.20 & 1490 & 24 & 1.050 & 1.60 & 330 & 436 \\Aerogel & 2500 & 0.15 & 1990 & 31 & 1.038 & 1.84 & 379 & 501 \\Aerogel & 2500 & 0.10 & 2985 & 46 & 1.025 & 2.27 & 470 & 620 \\Isobutane & 261 & 0.0027 & 169300 & 581 & 1.0019 & 8.29 & 1710 & 2260 \\\hline \hline\end{tabular}\end{center}\caption{Cherenkov thresholds ($p = \gamma\beta{m}$; $\beta = 1/n$) for electrons, muons, and pions. The refractive indices for $C_6F_{14}$ and $C_2F_6$ arefor 350 nm and are approximations based on linear extrapolations [T.~Ypsilantis and J.~Seguinot, NIMA343 (1994) 30]. DELPHI [http://wwwcn.cern.ch/$\sim$reale/richmain.html] and SLD use $C_6F_{14}$ as a liquid Cherenkov radiator. BELLE at KEK usessilica aerogel [R. Suda et al., Nucl. Instrum. Meth. A406 (1998) 213]. \label{tab:Cherenkov_thresholds}}\end{table} The dimension of the Cherenkov detector is 90 x 90 x 15 $cm^3$ wall of the aerogel counters. The Cherenkov detector consist of 49 aerogel tiles and each tile is 15 x 15 x 15 $cm^3$ with two Fine Mesh PMTs (Hamamatsu).It includes 5 mil window of polyethylene as an entrance.The layout of the Cherenkov detector is shown in Fig.~\ref{fig:ckov_array}.\begin{figure}[htbp]\centerline{\psfig{figure=bbox_array.eps,height=7.0in}} \caption{Possible layout of a downstream Cherenkov detector. \label{fig:ckov_array}}\end{figure}

\subsection*{Beam Decays Spectrum}\label{sec-beamdecay}

In order to understand the muon beam decay spectrum,we studied the decay of muons that may or may not fire the Cherenkovdetector. In most of the time the 200~MeV/c muon beam could decays intoelectrons a long with both its neutrinos and muon neutrinos.The decay is $\mu^- \rightarrow e^- \bar{\nu_{e}} \nu_{\nu}$.(The charged conjugate modes are applied throughout this note).

The majority of the electrons decay into a small angle with respect to the beam direction, however $2\%$ of such decays make an angle larger than $120^{\circ}$ and about $5\%$ of the them make an anglelarger than $90^{\circ}$ shown in Fig.~\ref{fig:cosine_angle}.\begin{figure}[htbp]\centerline{\psfig{figure=bbox_cosine_angle.ps,height=9.0in}} \caption{200~MeV/c muon beam cosine angle decay distribution.


\label{fig:cosine_angle}}\end{figure}

We also studied the total momentum of electrons versus itscosine angle distribution. Most of the electrons which have a low total momentum make a larger angle as we expected. Fig.~\ref{fig:cosine_vs_momentum}shows the corresponding of the total momentum of the electrons versus itscosine angle with respect to the beam direction. The electrons that madean angle larger than $90^{\circ}$ have a total momentum about 30~MeV/c. \begin{figure}[htbp]\centerline{\psfig{figure=bbox_cosine_vs_momentum.ps,height=9.0in}} \caption{Total momentum of the electron versus its cosine angle distribution with respect to the beam direction. \label{fig:cosine_vs_momentum}}\end{figure} \subsection*{Cherenkov Detector in GEANT4 Simulation}\label{sec-geant4}

We have coded the aerogel Cherenkov detector into MICE software simulation package.CKOVTrackerGeom class is coded for the description of the Cherenkovgeometry and materials. The logical and the physical volume ofthe Cherenkov detector described by CKOVTracker and CKOVSD classesrespectively. We use CKOVHitBank class as an interface for recordingthe hits information of the photons into the Cherenkov system.It records the position and time of each step, the energy deposition of each step, the momentum and energy of each track as well as its step length.

All the classes have been coded into the CVS MICE repository. We also tested the code in order to integrated them with the other subdetector system. The code has a flexibility on using some of the aerogel mixture that consist of $62.5\%$ of quartz and $37.5\%$of water with corresponding to its index of refraction as well asits density.Parameters of the aerogel are given in Table~\ref{tab:aerogel_para}.\begin{table}[htbp]\centering{}\begin{tabular}{||c|c|c||} \hlineTYPE & $\rho$ ($g/cm^3$) & $n$ \\ \hline \hlineAerogel101 & 0.04 & 1.01 \\ \hlineAerogel102 & 0.08 & 1.02 \\ \hlineAerogel103 & 0.12 & 1.03 \\ \hlineAerogel104 & 0.16 & 1.04 \\ \hlineAerogel105 & 0.20 & 1.05 \\ \hline\end{tabular}\caption{The aerogel parameters coded into CVS MICE repository. \label{tab:aerogel_para}}\end{table}

\subsection*{Advanced Model of the Rear Analysis Solenoid}\label{sec-solenoid}


In order to advance our study we have modeled the rear analysis solenoid with a length of 1.8 meters and central field value of $B_0$=4 Tesla. We have implemented a full field calculation to help us understand the particle trajectories and patternrecognition problems involving muons and decayed electrons in the rearCherenkov detector. Muons are injected into the solenoid entrance with a$\sigma_R$=10 cm and $\sigma_\theta$=0.15 spread as a starting point. A longitudinal momentum of $P_l$=230 MeV/c and spread $\sigma_{P_t}$=30 MeV/cwas used. See Fig.~\ref{fig:model_solenoid}.\begin{figure}[htbp]\centerline{\psfig{figure=decayed_muons.epsf.eps,height=7.0in}} \caption{Model of rear solenoid and aerogel counter. Muons are shown decaying at z=1.5 m near the last tracker plane. \label{fig:model_solenoid}}\end{figure}

\subsection*{Particle Tracking Simulation}\label{sec-tracking}

In our simulation four tracking stations are located in the finalsolenoid at z=0.15 m, 0.65 m, 1.15 m, and 1.65 m. At each tracker location werecord the x and y space point. For muon without decays these hits form a tightspiral of order a few cm [$P_t$=0.3~q~ B(T)~R(m)]. Electrons from decay willexhibit a pulled hit and may be easily vetoed by the tracker in most cases.

Decays near the rear of the solenoid or present the greatest danger ofmisidentification. In the figure show forced muon decay at z=1.5~m,just before the rear station. In some cases the hit on the final trackerplane will lie within the chi-square of fit limit and escape the particle idsystem undetected, possibly signaling an emittance increase. Fig.~\ref{fig:muon_track} shows the x and y projection of tracker hits for muon track. Note the small cluster size.\begin{figure}[htbp]\centerline{\psfig{figure=muon-track.epsf.eps,height=5.0in}} \caption{X-Y projection of tracker hits for muon tracks. Circles indicatetight reconstruction patterns. \label{fig:muon_track}}\end{figure} and Fig.~\ref{fig:elec_track} shows the x and y projection of tracker hits for electrons from decay near the final tracking plane. The cluster pattern is pulled by the decay electrons in some cases but not all. \begin{figure}[htbp]\centerline{\psfig{figure=e-tracking.epsf.eps,height=5.0in}} \caption{X-Y projection of tracker hits for electron tracks. In two of four cases well separated electron hit is recorded.


\label{fig:elec_track}}\end{figure}

It may be advantageous to swim tracks from the final tracker station tothe particle id entrance making a light yield prediction. Pixelation of theparticle id system will give some advantage. The optimum level of pixelation is under study.

\subsection*{Response of 7 x 7 Aerogel Array}\label{sec-respond}

Response of the 7 x 7 aerogel array to electrons and muons has also beensimulated. Aerogel of index n=1.05 with block lengths of 10~cm wereused. Cherenkov wavelength spectrum, PMT quantum efficiency, and a $50\%$light collection efficiency are folded into the photoelectron count.A muon momentum distribution of average 230~MeV/c was used.The electrons and the muons are well separated in this case.(Fig.~\ref{fig:ckov_respond}).\begin{figure}[htbp]\centerline{\psfig{figure=photoelectrons.epsf.eps,height=5.0in}} \caption{Photoelectron response of electron and muon tracks. \label{fig:ckov_respond}}\end{figure}

\end{document}

8.3 Tracking detectors

1.1.1 Tracking device (option I) Scintillating fiber tracker


9 Fiber Tracker Option for Tracking Spectrometer

Although at face value it would appear that the tracking issues for the MICE spectrometers are relatively straightforward, the large X-ray background from the cooling-channel RF cavities makes tracking single muons problematic. Measurements on prototype cavities indicate background rates in the range 1–10 MHz/cm2 of area covered by the fibers. These high background rates have led us to decide that an excellent technology for the tracking detectors is scintillating fibers. A multi-plane scintillating-fiber tracker is envisioned, each plane consisting of three sets of fiber doublets at 120° to each other. The fiber doublet structure is shown in figure F.1a and the u-v-t fiber plane in figure F.1b.

Figure F.1 (a) Fiber doublet structure. The pitch to diameter ratio is > 1. (b) u-v-t readout plane

This structure will give efficient space-point reconstruction, since hits in any two of the three doublets suffice. Each doublet consists of two overlapping layers of round fibers, giving 100% geometrical coverage over the face of the detector and an approximately uniform thickness of scintillator, independent of where a particle crosses the doublet. Each detector plane must have an active area whichcompletely encloses a 30 cm diameter circle. Scintillation light from the fiber will be piped to the photodetectors using fiber light guides of a length not to exceed 3 m. The photodetector of choice is the Visible Light Photon Counter (VLPC), due to its high quantum efficiency (85%) and high gain (50,000). We plan to follow the design of the D0 Fiber Tracker’s (reference) VLPC system. To minimize the effects of the background X-rays, we plan to use the smallest fiber that still yields enough light for efficient tracking. Based on results from D0 (figure F.2), we believe that 350 m fiber diameter is acceptable, given the short active length (30 cm) of the MICE tracking planes and the relatively short light guides as compared to those used in the D0 Fiber Tracker. We are currently prototyping fiber ribbons made with both 500 and 350 m fiber. The light yield from these prototypes will be measured with a 32 channel VLPC system using cosmic rays.

Fiber Diameter 350 micronsFiber Pitch/Diameter in ribbon 1.2

Views per plane 3Radiation Length per Plane (% Xo) 0.35

Planes per spectrometer 5Total Number of fibers 43,000 (approximately)

Estimated Light yield per singlet 8 peTable F.1. Fiber Tracker Parameters


Figure F.2. Light yield in photoelectrons for each singlet layer as a function of the length of the fiber light guide. Data from D0 using 835 micron scintillating fiber

The success of MICE depends critically on its tracking capabilities. In order to cope with the high background rates from the RF cavities, we assume 5 fiber-tracking planes upstream and 5 downstream of the cooling apparatus. Each plane will consist of the triplet-doublet structure described above, using 350-m-diameter fibers. To cover the 30-cm-diameter aperture requires a system of 45k channels (including spares), given a fiber-ribbon pitch-to-diameter ratio of 1.2. Figure F.3 illustrates a trough holding the 5 planes.

Fig. F.3: Representational drawing of the 5 planes of one of the fiber tracking systems

Readout Electronics


We plan to use the D0 VLPC system design for MICE. Our current plan envisions building 45 D0 type 1024 channel VLPC cassettes. No new engineering will be done and we plan to use available infrastructure at Fermilab for the VLPC cassette production. A new order for VLPCs will have to be initiated with the vendor, however. We will also be able to use the D0 cryogenics design for MICE. Our current design calls for two D0-type VLPC cryostats each holding approximately 22k channels (22 cassettes) worth of VLPC readout. One cryostat will service the upstream tracker and one the downstream tracker. This configuration allows us to minimize the fiber light guide lengths thus maximizing light yield and reducing fiber costs. Each VLPC cassette uses two 512 channel readout boards. The existing boards currently being used in the D0 experiment provide analog and trigger bit (digital) information from the VLPCs, temperature and bias control, and calibration. D0 is currently working on a new front-end design that will allow for higher readout rate and includes for possibility for adding timing information (with 1-2 ns precision) for each hit. MICE plans to work with D0 personnel on this effort with the hope that a single final design can be used for both experiments. A schematic of the Front-end readout system is given in figure F.4


Figure F.4. Schematic of VLPC Analog Front-End Electronics (AFE)


Cost and Schedule

We have based our cost estimate on the experience of the D0 experiment. Costing for the ribbons, support structure, waveguides, etc. follow the MICE design as outlined above. The readout costs follow directly from D0 numbers. Table F.2 gives a breakdown of the full M&S costs for the fiber 43k channel fiber tracker. This table does not include manpower which is estimated to be approximately $1.5M.

Table F.2. Full Fiber Tracker Cost Estimate.

M&SCOST

WBS DESCRIPTION EST1.1.2 FIBER TRACKER 4,096,1041.1.2.2 Scintillating Fiber Ribbons 111,0001.1.2.2.1 Scintillating Fibers 35,0001.1.2.2.1.1 Scintillating Fibers 25,0001.1.2.2.1.2 Scintillating Fiber QC 10,0001.1.2.2.3 Connectors 25,0001.1.2.2.4 Aluminization 6,0001.1.2.2.5 Ribbon Machine/molds 15,0001.1.2.2.6 Ribbon QC test box 15,0001.1.2.2.7 Ribbon laying & fixturing 10,0001.1.2.2.7.1 Ribbon Laying Fixturing 10,0001.1.2.2.7.1.1 Ribbon Laying Fixturing Parts 5,0001.1.2.2.7.1.2 Ribbon Laying Fixturing Shop Chrgs 5,0001.1.2.2.8 Ribbon Connectorization 5,0001.1.2.3 Clear Waveguides 94,0001.1.2.3.1 Clear Fiber 50,0001.1.2.3.2 Connectors A 01.1.2.3.3 Connectors B 01.1.2.3.4 Sheathing 2,0001.1.2.3.5 Factory setup 2,0001.1.2.3.6 Test/QC system 15,0001.1.2.3.7 Fabrication/testing 25,0001.1.2.4 Visible Light Photon Counter Cassette 2,500,0001.1.2.4.1 Production cassettes 2,485,0001.1.2.4.1.1 VLPC Hybrids 1,570,0001.1.2.4.1.2 Cassette Mechanical Parts 320,0001.1.2.4.1.3 Cassette Electrical Connectors 120,0001.1.2.4.1.4 Cassette Flex Circuits 410,0001.1.2.4.1.5 Cassette Internal Fiber 65,0001.1.2.4.2 VLPC Test Cassettes 15,0001.1.2.5 Visible Light Photon Counter Cryogenics 216,6991.1.2.5.1 Test Cryostat 15,0001.1.2.5.2 Final Cryostat System 122,6781.1.2.5.2.1 Raw Materials 22,000


1.1.2.5.2.2 Machining and Construction 65,0001.1.2.5.2.3 Intercepts, Shields, Insulation, Vacuum 11,8191.1.2.5.2.4 Top Lid Heater and Controls 17,8121.1.2.5.2.5 Relief and Check Valves and Fasteners 6,0471.1.2.5.3 Helium Distribution System 67,0211.1.2.5.3.1 Detector Valve Box, Bayonet Cans 45,0211.1.2.5.3.2 Transfer Lines 22,0001.1.2.5.4 Cryostat Installation 2,0001.1.2.5.4.1 Staging Stands, Roll in Plates 1,0001.1.2.5.4.2 Movement Mechanisms 1,0001.1.2.5.5 Temperature Controller 10,0001.1.2.5.5.1 Cryostat Sensors 5,0001.1.2.5.5.2 Cassette Controller Mods, Power Supp. 5,0001.1.2.6 Calibration System 9,0001.1.2.6.1 Optical Panels and LEDs 3,0001.1.2.6.2 Support Structure for Optical Panels 1,0001.1.2.6.3 Drive Electronics 5,0001.1.2.7 Mechanical Support, Services 34,4051.1.2.7.1 Support cylinders 15,0001.1.2.7.1.1 Cylinder Materials 5,0001.1.2.7.1.2 Mandrels 3,0001.1.2.7.1.3 End Rings 2,0001.1.2.7.1.4 Construction Tooling 5,0001.1.2.7.2.1 OD Rings 1,7051.1.2.7.2.2 Disks 5,7001.1.2.7.2.3 Fixturing 10,0001.1.2.7.3 External Supports 2,0001.1.2.8 Detector Assembly 6,0001.1.2.8.1 Waveguide install fixturing 2,0001.1.2.8.2 Survey fixturing 2,0001.1.2.8.3 Tracker installation fixtures 2,0001.1.2.9 VLPC Electronics 1,125,000

The schedule for construction of the fiber tracker is given is figure F.4. It assumes that work on the tracker starts immediately.


Figure F.5. Construction Schedule for the Fiber Tracker



Low RF Background Scenario

If the backgrounds from the RF cavities are low, we can reduce the cost and channel count of the fiber tracker by multiplexing the fibers. It is our working assumption that we will still use 350 micron scintillating fiber in the tracker even if the RF backgrounds are low. In this case the amount of material presented to the beam remains the same, the sensitivity to backgrounds remains the same, and the tracking resolution of the detector remains the same. If backgrounds are low, we would multiplex the readout of the scintillating fibers into the VLPCs. With 350 micron fiber we can multiplex 7, 350 micron fibers into 1 VLPC pixel. This is show in figure F.6.

Figure F.6. Mapping of 7, 350 micron fibers onto 1 VLPC pixel.

There would be about a 10% loss in the outer ring of fibers due to a slight mismatch, however. This implementation of the fiber tracker does require an additional component. A 7:1 multiplexing wavguide would have to be added to the system if we wanted to retain the capability of reading out each fiber individually if the backgrounds became large. The trackers would not have to be changed, just the waveguides, and of course, the addition of more readout channels. This scenario has a number of nice features. First, if the backgrounds are low, we can run in this configuration and reduce the system channel count (and therefore cost) substantially. If the experiment is pushed to run at high RF fields and backgrounds become a problem, the trackers themselves would not have to be rebuilt. The cost savings with 7:1 multiplexing is approximately $3M. The fiber tracker systems would then cost about $1M exclusive of manpower.

1.1.2 Tracking devices (option II) TPG


DRAFT of the TPG section for the proposal (for comments)

1. General principleThe reasons for a TPC with GEM amplification as a tracking device for MICE can be summarized in a few points:

A TPC presents a light and uniform material along the beam line, yet producing a large number of points on each track.

The GEMs, manufactured by standard photolithographic processes, are less expensive than the traditional electron multiplication for TPCs (usually wire chambers); at the same time, the GEMs are able to minimize the ion feedback into the drift volume, so reducing the risk of uncontrolled field distortions.

By choosing an appropriate gas mixture, it is possible to make the detector transparent to the X-ray background emerging from the RF cavity. This allows to counterweight the disadvantage of the long integration time.

2. Operational modeA sketch of the envisaged device is shown in figure 1 for the downstream spectrometer. An identical one will be installed in the upstream spectrometer. The operational principles of the device are as follows. The sensitive volume of 1m long and 30 centimeters diameter is situated in the homogeneous magnetic field region of the spectrometer solenoids. The electric field provided by a field cage surrounding the sensitive volume is parallel to the guiding magnetic field. The charges produced by ionization of muons are collected on the far side of the chamber with respect to the cooling section, so as to ensure a minimum of material. The charges are amplified by GEM foils, and read-out on a plane of pads from which the signal is shaped in preamplifiers and digitized by flash-ADCs. The total length of the chamber corresponds to about 120 samplings at a drift velocity of 1.7 cm/s, so that the device provides for each track 120 2D points times three coordinates. The chamber will be filled with low mass gas (a Helium dominated mixture) thus reducing multiple scattering and offering very small conversion probability for X-rays resulting from the cavities dark current.


Fig. 1 View of the downstream spectrometer of MICE with a TPG as tracking device.

2. Characteristics and performance of the chamberA complete definition of the operational parameters of the chamber will only be possible after a full scale test to be performed in spring-summer 2003 using the HARP TPC magnet, field cage, digitizing electronics and gas system at CERN. The proponents of the project have been previously involved in the conception, construction and operation of the HARP TPC, and can provide a large fraction of the electronics. Operation of drift chambers with a helium-based mixture is quite customary, and performance figures will be given here for a 90%He, 10% iso-butane mix at atmospheric pressure. A drift voltage of 500 V/cm provides a drift velocity of about 1.7cm/s. The maximal potential of -50 kV is situated in the plane of the field window and needs to be degraded in less than 5cm of insulator in the outward direction of the solenoid, this can be done with solid insulator such as Teflon, and in less than about 50 cm in the direction of the liquid hydrogen absorbers, this can be done with a suitable gas, e.g. low pressure N2, or with vacuum. The most probable number of primary ionization electrons along a minimum ionizing muon is calculated to be 12 per cm. These electrons will drift toward the GEMs with a transverse diffusion of 1.4 mm.z[m], and a similar longitudinal diffusion. The GEMs are made of 50-micron thick foils with holes of 70 microns diameter at a pitch of 150 microns, as shown in Fig.2. They introduce a small additional diffusion; each primary electron will produce a “spray” of about 0.5 mm to 1mm transverse diameter on the readout plane. This parameter is not well known and will be measured in the forthcoming tests. The read-out board described below has a pitch of 450 microns, contributing also very little to the resolution. If the exposure time of MICE during the RF pulse is of the order of


500 microseconds, the available ADCs allow digitization in 1024 time slots of 500 ns second each. Each track is then sampled in 120 time slots of 500 ns, 0.85cm long, containing 10 primary electrons each, giving 120 points with a spatial resolution, in the worst case, of the order of the pitch of the readout plane. In a perfect chamber this would give a transverse momentum resolution of ? 0.2 MeV/c ?.(to be verified, action Mario)

Fig. 2 Photograph of the GEM showing the 70 microns diameter holes at a pitch of 150 microns, and description of the electric field lines in the Gas Electron Multiplier (GEM) foils.

Fig 3: The TPG read-out: left: the 3 GEM foils providing amplification to the hexaboard. Right: the hexaboard structure with a third of the pads (blue) connected in strips at 30o, one third at 120o (red), and one third at 90o (green). The amplification in each GEM will depend on the high voltage, but it is planed to work at an amplification level of 20-50 per GEM, giving a total signal of 104 to


105 electrons on the pad plane per primary electron. These will be read-out by a hexaboard as described in Fig.3. The signal is distributed among several individual hexagons. The hexagons could in principle be readout individually, but for cost reasons the hexagons will be connected to form strips in three orientations Each primary electron will thus give signals in at least three projections. The strip signals, will be collected from the pad plane and send over flat cables of 16 channels each to the preamp boards. In the present design, each preamp board will collect 48 channels. The signal will be shaped to a length commensurate to the sampling frequency of 500 ns, and then sent to the FADCs. The FADCs from the HARP TPC will be used. .Prototype preamp boards in “HARP” style and adapted to MICE geometry have been produced and delivered. The channel count is 667 per coordinate, a total of 4000 for two TPGs.

Fig 4. Description of the elements of the TPG.


Fig 5. Simulation of 100 microseconds of data taking with the upstream MICE TPG. Vertical axis is the strip number in each of the three projections (u,v,w at 30, 90 and 150 degrees). Horizontal axis is the time slot number. Each time slot is 500 ns. The muon rate is assumed to be 0.1 per ISIS proton bunch, i.e. 3 per microsecond.

4. Sensitivity to backgroundTwo kinds of backgrounds can be detrimental on the performances of the TPG.The first kind is the pollution by RF noise, both in the experimental area and in the beam pipe. Thanks to the mechanical construction, most parts of the detector are screened by thick aluminum parts, which will be duly grounded. The only parts that cannot be enclosed in massive metallic pieces are the GEMs: RF noise coming from the cavity may, in principle, reach the readout plane and produce fake hits. However, even if the GEMs are very light, the thin Cu layer deposited on them is already an effective screen for 200MHz EM waves, since its thickness is equivalent to the skin depth. Even if we are confident in the immunity of the detector to RF noise, specific tests will be performed with a small prototype chamber to make sure that all the


details of the RF immunity, from the readout plane to the FADCs, will be understood.The second kind of noise will come from the X-ray emission from the cavity. Conversion of X-rays may produce fake hits and, ultimately, endanger the ability to reconstruct a clean set of tracks.As we have seen, one time-slice corresponds – with the present indication of gas mixture and field cage potential – to about 0.85cm. In terms of conversion probability in the gas, this amounts to about 10-4 for X-ray energy of 10KeV. If one makes the hypothesis of a primary X-ray rate of 1GHz 10KeV photons over the whole diameter of the detector, each 0.85cm slice will integrate a few conversions. This is perfectly tolerable, since each slice is read out in 1800 coordinates. Softer X-ray spectra may of course be more dangerous, but we estimate that the rate may still be tolerable at 5KeV.More precise understanding of the influence of the X-ray background will be possible when an X-ray emission spectrum will be measured.

5. Preliminary timescale for tests in 2003

The tests of a TPC with GEMs read-out will be carried with several aims:First, the capacity to shield the detector against RF electromagnetic radiation will be tested on a small chamber built in Frascati, equipped with electronics similar to that which will be used for the final detector. This will be done at CERN with the help of the RF group who will provide a tunable antenna radiating at 200 MHz.

Then the exact performance of the read-out system, diffusion properties in the gas and especially in the GEMs will be tested in a 0.7 T magnet with the HARP solenoid and field cage. This will be a test of a full size readout board, in which an already sizeable amount of electronic channels will be involved (600). If the test is successful the readout board could be part of one the final MICE trackers.

At the same time the response of the TPG and in particular of the GEMs to RF field emission should be tested. The effect of photon conversions in the TPG gas is easily calculable, their effect on the GEMs themselves is not. The 88 MHz test cavity at CERN will be powered at the beginning of 2003. Alternatively, one could bring a small chamber to the labG at Fermilab. This may be a critical issue and should proceed quickly.



The exact schedule for construction of the final detectors is being worked out.

2. CostsFrom preliminary estimates based on the construction for the tests in 2003, one can estimate the cost of the readout and amplification system to be less that 100K$ per tracking station. Based on the cost of the HARP field cage, whose design is, for the moment, being considered for the MICE TPGs, we estimate building the two field cages may account for 200K$.Ancillaries like the gas and HV system may account for another 100K$.

\documentclass[11pt]{report}\usepackage{epsfig}\begin{document}\section{Tracking studies for the MICE TPG}The crossing of muons in the MICE TPG has been simulated, mainly to assessissues related to hit occupation, and 3-D reconstruction with the loss ofinformation due to the strip readout. We have considered 1024 time slotsof 500 ns, for a total acquisition time of 512 $\mu$s. It was assimedthat muons are produced by the accelerator every 330 ns, with Poissonprobability having average $\mu=0.1$. Therefore, for every acquisitionsession about 160 muons enter the detector, half of them fully in theacceptance region of the TPG. Being the drift time about 2 cm/$\mu$s,the most distant ionization will take 50 $\mu$s to reach the exaboard,about 1/10 of the total acquisition time. For each time slot, on average16 tracks (or about 10 in the fiducial region) are present.The track hits are recorded by the three strip planes, so each 3-D helicoidaltrajectory will be seen as three sine functions in the time vs strip plane(see figure \ref{fig:strips}).\begin{figure}[tbhp] \begin{center} \epsfig{file=pics/globalstrips.ps,width=8cm,bb=0 134 560 670} \caption{Particle trajectories as sees from the three strip planes as a function of acquisition time} \label{fig:strips} \end{center}\end{figure}

The trajectories described in this picture are not pointlike, since the effectsof transverse diffusion of 150 $\mu$m$\sqrt{l(cm)}$ has been considered,summed in quadrature with additional 500 $\mu$m to account for the spray in


the GEM. Figures \ref{fig:utit} we see that two in different time slotstracks can actually partially merge due to diffusion.\begin{figure}[tbhp] \begin{center} \begin{minipage}{0.45\textwidth} \epsfig{file=pics/utit.ps,width= 5 cm,bb= 0 150 530 660} \end{minipage} \begin{minipage}{0.45\textwidth} \epsfig{file=pics/utit2.ps,width= 5 cm,bb= 0 150 530 660} \end{minipage} \caption{Signal distribution on the three strip planes, for two consecutive time slots} \label{fig:utit} \end{center}\end{figure}

Muon momentum can be independently measured on the three planes from a fitof the parameters of the sinusoidal oscillation. In particular, if wewrite the sine function as\[f = \alpha \sin(\beta x + \gamma) + \delta\]the following relations hold: $P_t=0.3 B \alpha$, $P_l=0.3 B/\beta$, whereB is the magnitude of the magnetic field; so the amplitude and the frequencyof the oscillations are directly connected to the two components of themuon momenta, and can be measured quite independently.\parOf course, due to diffusion, a proper clustering of the signal has tobe performed independently on the three strip planes, to get the center-of-gravity of the track position. Assuming perfect pattern recognition,the precision of the fit is about 15 KeV/10 MeV in $P_t$ and 150 KeV/200 MeVin $P_l$, on each independent strip plane. Combining the three planes,a further resolution improvement of about a factor 2 can be obtained.\parAnother important issue in dealing with a detector like the TPG concernsthe possibility of actually reconstructing the 3-D points from theirprojections on the strip/time planes. Neglecting overlaps, in a single timeslot, on each strip plane there will be a number of hits equal to that ofthe tracks producing a ionization signal. For $n$ tracks, there will be$n^2$ (x,y) coordinate combinations satisfying the observed hits forboth u and v strips. Out of these, only $n$ will be real, the others being``ghost'' hits, that should be eliminated by requiring coincidence withhits in the t strip. Our simulation shows that, in absence of noise butalso not exploiting any information relative to the height or width ofthe hits, the number of fakes can be kept at the level of few per cent, andnot prevent a proper 3-D reconstruction of the muon tracks (see figure\ref{fig:views} for 3d views and projections on the (x,y) plane ofthe acquisition session whose strip projections were shown in figure\ref{fig:strips}).


\begin{figure}[tbhp] \begin{center} \begin{minipage}{0.45\textwidth} \epsfig{file=pics/view2d.eps,width= 5 cm} \end{minipage} \begin{minipage}{0.45\textwidth} \epsfig{file=pics/view3d.eps,width= 5 cm} \end{minipage} \caption{3d view (left) and its projection on the (x,y) plane (right)of the events whose readout was shown in figure \ref{fig:strips}.The upper plots show the hit position as computed from the knowledge ofthe muon helix coordinate, while in the lower plots this quantity isinferred from strip information only.} \label{fig:views} \end{center}\end{figure}

\end{document}

9.1 Particle ID

9.1.1.1.1.1.1.1 The calorimeter option

As the state of the art in high resolution electromagnetic Pb-scintillating fiber calorimetry we take into account a KLOE-like design which provides adequate energy resolution and superior timing accuracy with respect to bulk scintillators.

9.1.1.1.1.1.1.2 Structure and Layout

The proposed calorimeter is built by gluing 1 mm diameter blue scintillating fibers between 0.5 mm thick grooved lead plates (fig. 1a), representing an uniform and symmetric Pb-scintillating fiber structure with a regular fiber to fiber spacing of 1.35 mm (fig. 1b). When layers are superimposed, fibers are located at the vertices of adjacent quasi-equilateral triangles, forming a homogeneous and compact structure (fig. 1c) which has a fiber:lead:glue volume ratio of 48:42:10. The value of the Pb-fiber ratio has been chosen slightly different from the optimal value of 1:1, which should be the best compromise between a high sampling fraction (to minimize the energy deposit fluctuations in the active cell) and a high density, to leave 10% volume for filling around the fibers with an epoxy. The resulting


composite, which has a density of 5 g cm-3 , a Moliere radius of 3.5 cm , a radiation length X0 1.5 cm and a sampling fraction of 15% for a minimum ionizing particle, gains considerable stiffness and can be easily machined to an appropriate shape required for the final assembly.

This “spaghetti” design offers the possibility of a fine sampling and results in optimal lateral uniformity of the calorimeter. Moreover in the proposed structure fibers run mostly transversely to the particle trajectories. This reduces sampling fluctuations due to channeling, i.e. showers developing along the fibers’ direction, an effect particularly important at the low energies of interest. Finally the very small lead foils thickness ( 0.1 X0 ) results in a quasi homogeneous structure and in high efficiency for low energy electrons.

At first we propose to build a calorimeter module of 6060 cm2 cross section , 15 cm thick, consisting of 125 lead and scintillating fiber layers for a total of 58.000 fibers by using the facilities of the LNF workshop formerly used for the construction of KLOE electromagnetic calorimeter ( Ref. 1).

The calorimeter is structured in 4 layers of 16 cells each; the dimension of a single cell is 3,753,75 cm2 and 60 cm long (fig. 2a).In order to reconstruct the coordinate of the impact point along the fibers two options are open: by time differences reading out the cells at both ends – at least in the first layer - or arranging the four layers with fibers alternatively oriented in x and y directions (fig. 2b). In the first case we realize a compact structure 15 cm thick (fig. 2c and 2d) , in the second one we have to manage a system of 4 discrete layers 3,75 cm thick each.

9.1.1.1.1.1.1.3 Building components

The grooved lead foils are obtained by rolling 0.5 mm thick foils into special shaping machine. The grooving rollers result by the assembling of 13 disk-like pieces, 50 mm thick and 400 mm in diameter ( fig. 1d), made of hardened steel and ground to shape by a sintered diamond tool; the rollers are fixed by means of ball bearings on a very rigid frame and are aligned and checked with a set of micrometers. The achievable thickness uniformity is around few tens of m and the grooves can deviate from a straight line by less than 0.1 mm per foil length.

As far as the fiber choice concerns we can profit of the KLOE experience; we propose to build our detector with the optical fibers Pol.Hi.Tech. type 046 (2), emitting in the blue-green region and satisfying, with reasonable cost, a minimum set of specifications. Special care has to be devoted to handle the scintillating fibers, as they result to be damaged by the ultraviolet component of natural and artificial light which affects and deteriorates both the light yield and the attenuation length.


The mechanical assembly starts by gluing the first lead foil on a 2 cm thick aluminum supporting plate. After the first layer has been set, a small amount of non aggressive epoxy glue (3) is distributed on the plane; a layer of fibers is then positioned in the grooves , then glue is spread again before positioning next lead foil. This procedure is repeated until a stack of 10 planes is formed. Before the glue stars curing, a uniform pressure is applied for half hour on the calorimeter surface to allow a more uniform distribution of the glue. The stacking, alternatively of lead and fibers planes, is repeated until a thickness of 14 cm is reached. The resulting calorimeter composite has a density of 64 fibers/cm2. Since the timing resolution depends on the calorimeter light yield, great care has to be put in maximizing the efficiency of the light collection system and insuring uniform photocathode illumination. Also for this aspect we can profit of the KLOE experience realizing light guides which consist of a tapered mixing part, where the quadrangular entrance face pass smoothly to its inscribed circle, and a Winston cone concentrator. (4) The latter part has to match the area of the calorimeter element to the sensitive photocathode face. In absence of residual magnetic field, a suitable read out choice is the 1-1/8 inch Hamamatsu phototube, type R1355 with a transit time spread less than 1 ns, already used in the HARP experiment. In this case the area concentration factor is 2.65 with an efficiency in light collection of 85 %.In Fig.3 a sketch of the assembling procedure is shown.

9.1.1.1.1.1.1.4 Front end electronics

Each PM signal has firstly to provide information about energy deposit. This could be accomplished by using conventional ADC, like the VME module type 792, produced by the firm CAEN and already used in the HARP experiment. A clean minimum ionizing particle selection, which guarantees a non biased cross calibration of all the calorimeter readout elements, can be obtained both from cosmic rays and muon beam tests.A rough measurement of the impact point on the calorimeter can be derived from the ratio of two sides energy deposits in the fired cell of the first layer. Alternatively, if the single side read-out option has been chosen with a system structure made of 4 distinct layers, the information of the coordinate along the fiber has to be obtained by the geometric crossing of the fired cells.Moreover, if we want better spatial resolution ( 1.5 cm ) or the global trigger requires a fine coincidence with the calorimeter, a more precise timing information has to be provided. This means that each photomultiplier signal - of the first layer , at least - has to be split in order to be read out by a TDC after a constant fraction discriminator. At the end the calorimeter event time will be defined as the mean between the times of the two readout sides obtained using all the cells belonging to the shower and weighted with their own energy. Timing accuracy is mostly limited by the light yield, the emission time distribution of the fibers and by the transit time of the photomultiplier and it depends also on the shape and amplitude of the signals and on the type of discriminators used.


9.1.1.1.1.1.1.5 Energy and timing resolutions

The energy resolution depends weakly on the z position because the fluctuations of the relatively large number of photoelectrons (Npe) is negligible with respect to the sampling fluctuations. The time resolution instead has a strong z dependence since it depends essentially only on Npe. The light output Npe/mip (photoelectrons per minimum ionizing particles) can be calculated by the knowledge of the gain of phototube together with the ADC gain and response for mip.

The fibers Pol.Hi.Tech. type 046, which have decay constant of 2.5 ns and attenuation length is 3.5 m, with a bialkaly photomultiplier, yield about 80 90 and 65 70 photoelectrons, per minimum ionizing particle crossing a 3.53.5 cm2 cell of composite described above, respectively at distances of 20 cm and 1 m from the photomultiplier. (5) The KLOE calorimeter test beam data (5) show that distributions of total energy, that is the calibrated sum of energy over all fired channels, are gaussian with almost no tails. The calibration constants are obtained from cosmic rays and 400 MeV/c pions data, both of which are (to a first approximation) minimum ionizing particles.

The energy response of the calorimeter, independently of the incident angle, for electrons and photons in the energy range 20-300 MeV, is linear, as shown in fig.9 of ref. (5) with a slope of 37 MIP/GeV. On the other hand the simulation of visible energy, deposited by a minimum ionizing particle in a 3.53.5 cm2 readout element of active material results to be equivalent to an average energy deposit of 3.24 MeV. It results a sampling fraction of 12% for electromagnetic shower and therefore we can take into account that the energy release of a MIP in one elements corresponds to the energy released by a 27 MeV photon.

Considering the KLOE EmCal performances (6) obtained for electromagnetic showers of similar energies at DANE, the Frascati -Factory, the expected energy resolution of the Electron Identifier proposed for MICE is and it’s fully dominated by sampling fluctuations. As far as the time resolution concerns, the experimental KLOE result, obtained with modules 4 meters long, is t and it has to be taken into account as top reference for possible contribution of the calorimeter to the MICE timing and/or trigger system. Moreover, in case of double side read-out, the timing resolution of the calorimeter gives a direct measurement of the resolution of the Z coordinate of the impact point along the fiber through the relation z t vf where vf 17 cm/ns is the light speed in the fiber.


9.1.1.1.1.1.1.6 Particles Identification

The MICE timing requirement has in fact a strong influence on the overall precision needed for the global time measurement and special attention has to be dedicated also to the particle identification at the downstream end of the cooling apparatus. A combination of time of flight with both energy deposition and profile measurements can efficiently identify electrons from muon decays reaching a rejection factor of 10-3 . In fact the response of the calorimeter for muons, in the momentum of interest of MICE, shows that they are mostly punching through with respect to electrons which leak basically all their energy in the first two layers.

Moreover, as shown in the KLOE test beam with pions and muons of 200-250 Mev (5), the calorimeter can provide some identification by energy deposition patterns integrating the TOF information and contributing consequently to reject background pions coming from the upstream part of the cooling channel.At the end, as auxiliary time measurement device with a precision better than 200 ps, for energy of impinging particles greater than 100 MeV, the calorimeter can also contribute to locate the primary particles at least within the correct quarter-period of the RF cycle.

9.1.1.1.1.1.1.7 Tests and calibrations

Being the KLOE building technique well known and tested, a specific R&D program on the construction is not needed a part for what regards refurbishing and tuning of the tooling machinery.As far as the calibrations concerns preliminary cosmic rays test are foreseen together with tests of detector response to low energy electrons.For this purpose the natural place where such measurements could be performed is at the Beam Test Facility in the DANE complex of LNF. (7)The BTF is a beam transfer line which has been designed in order to optimize the operation mode in which single electrons are produced for detector calibration purposes. The pulse duration is 10 ns , the maximum repetition rate is 50 Hz and the common available energy range for electrons and positrons is 50 540 MeV

9.1.1.1.1.1.1.8 Construction Schedule

The INFN technical manpower will be engaged for lead foils grooving, drawing of light guides, design and realization of the mechanical support for photomultipliers and patch panels. The precise schedule for the preparation of the grooved lead plates has to be negotiated with the Pisa colleagues of the AMS experiment. Actually and for the close future, they are the main users of the lead-grooving machine at LNF.


Construction of light guides , production of fibers , gluing of layers of lead with fibers , machining of the assembled module and final gluing of light guides have to be commissioned to external firms. The complete realization of the project can be can be brought to conclusion in one year from the approval of the INFN financial committee.

9.1.1.1.1.1.1.9 Costs

In the case that the entire read out system (photomultipliers, voltage dividers, cables, front end electronics and VME crate) has got back from HARP, the cost of the MICE calorimeter - in the basic configuration of 606015 cm3 - is restricted to the purchase of lead foils, fibers, glue, light guides and relative workings and it amounts to 50 k Euro.

9.1.1.1.1.1.1.10 References

[1] The KLOE Collaboration , The KLOE Detector Technical Proposal , LNF-92/019 (1992)

[2] http://village.flashnet.it/users/polhitec/index.htm[3] Bicron Optical Cement BC-600[4] W.T. Welford, R. Winston, The optics of non-imaging concentrators, Academic

Press, New York, 1978 and H. Hinterberger, R. Winston, Rev. Sci. Instr. 37 (1966) 1094

[5] Antonelli et al., Nucl. Instr. and Meth. A354 (1995) 352[6] M. Adinolfi et al., Nucl. Instr. and Meth. A482 (2002) 364[7] http://www.lnf.infn.it/acceleratori/btf/

Proposal of a downstream e- identifier for MICE

Version 2 - 11 Nov. 2002Gh. Grégoire

Institute of Physics- University of LouvainB-1348-Louvain-la-Neuve

Contents.

A. Global features and physics constraintsB. Detailed description

1. Aerogel radiator


2. Light collecting tubes3. Photomultiplier4. Mechanical layout

C. Optical performances and electron detection efficiencyD. Cost estimates

A. Global features and physics constraints

The identification of electrons and muons downstream the solenoid magnet is strongly influenced by their momentum and spatial distributions (Figure 1). The particle simulations presently available show that both electrons and muons have broad and continuous momentum distributions (0< pe < 300 MeV/c and 0< p < 400 MeV/c).

Figure 1. Momentum distributions of muons and electrons.

In my opinion, a threshold Čerenkov detector looks as the simplest device to perform this identification. As far as possible it should be designed to give no response for muons while having in principle 100 % efficiency in detecting electrons.

In the momentum ranges given here, we cannot use solid or liquid Čerenkov radiators (with index of refraction n> 1.2) since the higher energy muons (p > 160 MeV/c) would then generate Čerenkov photons in such materials. On the other hand, we cannot use (pressurized) gas radiators (with n-1 0.005) as part of the decay electrons (pe < 5 MeV/c) would give no light and would be confused with muons. This is the basic argument to propose aerogel radiators (1.01 < n < 1.06). An adequate value of the index of refraction is discussed later in this document. However the optimum index of refraction should obviously be determined following more detailed particle simulations unavailable at the moment.

Although the tracking devices inside the solenoid certainly help in decreasing misidentifications, we prefer to base the present device on a genuine "standalone" operation.

The stray magnetic field of the solenoid should not be neglected and partly justifies the location of the photodetector(s) away from the solenoid. In addition, electrons or muons


along the solenoid axis should not hit these photodetectors as they could generate spurious Čerenkov photons in the glass envelopes. As an example, a fully relativistic electron generates about 40 photoelectrons from Čerenkov light produced in a 2 mm thick glass sheet.

The gross geometrical features of the threshold detector are determined by the angular and spatial distributions of particles at its entrance window. The presently available simulations (P. Janot's files evaluated 50 cm downstream of the solenoid) do not show any correlation between angles, positions and energies of the incoming particles. The geometrical aperture of the proposed design was chosen to be twice as large as requested by Janot's data: our aim is to take into account the defocusing of particles in the stray magnetic field of the solenoid. (Figures 2 and 3).

Figure 2. Particle spatial distributions in a plane perpendicular to the beam axis. (Ref.

P. Janot)

Figure 3. Particle spatial distributions in a plane perpendicular to the beam axis. (Ref.

P. Janot)

Taking into account a defocusing due to the stray magnetic field the RMS "beam spot" diameter is expected to be about 60 cm. The position distribution has a roughly gaussian shape. The divergence of particle directions has also a gaussian shape with a FWHM of about 60 degrees. These features constrain the geometrical aperture (about 60 cm in diameter) and the optical acceptance of the device.


B. Detailed description

Sketches of the essential parts of the system are shown in Figures 4 and 5.

Figure 4. Meridian cut (side view) through the one pair of PM tubes.


Figure 5. Perspective view of the optical parts.

The optical characterization of the whole setup has been performed by accurate 3D-optical ray tracing, taking into account realistic surface properties of the mirrors (spectral and angular reflectivities), the bulk scattering inside aerogel materials, transmittance of the window and the typical quantum efficiency of standard photomultipliers. The overall light collection efficiency reaches about 80%.

Several items are discussed in the following points.

1. Aerogel radiator

The aerogel radiator used here has an index of refraction n=1.02 and a total thickness of 10 cm. The transverse size is 90 cm x 90 cm. It is constructed with blocks of aerogel of typical dimensions 20 cm x 10 cm x 2.5 cm. (These dimensions depend from the manufacturer).

The thickness is chosen on the basis of the small photoelectron yield with low energy electrons (Fig. 6).

Figure 6. Semilog distribution of the number of photoelectrons induced by Čerenkov light in a 10 cm thick aerogel slab with n=1.02 (assuming a

100% light collection efficiency).

The choice of the appropriate index of refraction for the radiator is governed by the relative light yield of electrons and muons. For electrons the most probable number of photoelectrons is about 40, while for muons only the high-energy part of their energy spectrum will bring some contribution.

It is safe to fix a detection threshold of 10 photoelectrons since the photon collection efficiency and unavoidable losses have still to be taken into account. With these conditions,


the detection efficiencies of electrons and muons versus the index of refraction are shown in Figure 7. The detection efficiency i is defined as

with i = electron or muon

Figure 7. Detection efficiency for electrons and muons versus the index of refraction of the radiator.

It is clearly seen that some compromise is needed here: a higher index improves the electron detection efficiency, but also increases the rejection of muons. To get a quantitative estimation, we could for instance define a figure of merit F(n)

which would reach a maximum for the best compromise between the electron detection and the "muon losses". (Figure 8).

Figure 8. Figure of merit of the global detection efficiency as a function of the index of refraction of the radiator.

A broad maximum is clearly seen around n = 1.04 to 1.06, independently of the relative populations of electrons and muons. Further refinements to still optimize the index should be undertaken when more particle files will be available.


The published literature (Ref. Buzykaev at al. NIM A433(1999)396 and BELLE internal reports) shows that aerogel strongly scatters violet and blue light as shown in Figure 9. These properties are included in the present proposal.

Figure 9. Internal "transmission" (i.e. "non-scattering") of a 10-cm thick piece of aerogel versus wavelength in the visible region.

In order to increase the light output from aerogel, one could imagine stacks of 25-mm thick aerogel tiles interspersed with wavelength-shifting foils as is done for the AMS and Phenix experiments. In this way, one benefits: a) from the conversion to visible light of the UV-part of Čerenkov emission spectrum, and b) from the much better transparency of aerogel to larger wavelength. This improvement is left for further studies and simulations.

Aerogel is quite hygroscopic and has to be maintained at all times in a dry atmosphere to keep its optical properties. Our design foresees the whole setup to be vacuum tight, although it should always be operated very near to the atmospheric pressure. It is best to fill the containment vessel with some inert dry gas (nitrogen or helium).

This explains the need for an optical window isolating the photodetector volume. We plan to use a 7-mm thick glass sheet made from Schott B270 material.

2. Light collection walls

For reasons of availability, cost and handling, all side surfaces potentially hit by light are constructed from aluminized PMMA plastic sheets or (more expensive) optical glass plates (again 7 mm thick Schott B270).

In this study the reflectivity of the aluminized plastic sheets is taken to be about 90% (as used for the Čerenkov detector of the HARP experiment). It is however possible (A. Braem, private communication) to benefit from recent developments based on a multilayer of Aluminium, Magnesium fluoride MgF2 and Hafnium oxide HfO2 to reach reflectivities as high as 96% in the interesting visible domain (250 to 500 nm). (Figure 10).


Figure 10. Reflectivity of a multilayer Al + MgF2 + HfO2.

A plane mirror tilted at 45 degrees reflects light at 90 degrees to the optical beam axis towards the photodetectors. It can also made from aluminized PMMA plastic or Schott B270 optical glass.

3. Photomultiplier

The proposed system uses 20-inch diameter photomultipliers Hamamatsu R3600-02 (same as those at Kamiokande) with a standard bialkali photocathode. This particular type has the advantage of having a rather large gain 3 106 adapted to the low light yields of aerogel radiators.

As indicated before the PM's are mounted in volumes separated from the radiator section. It allows easy access to the PM's without disturbing the dry environment of the aerogel and avoiding possible damages to the mirrors.

The optical simulations show there is no need for a Winston cone.


4. Mechanical layout

A perspective view of the containment vessel is shown in figure 11

Figure 11.Perspective view of the containment vessel.All dimensions are in millimeters.

It is planned to construct the different parts with Nickel-plated soft steel sheets 5 to 15 mm thick. Once again this is based on arguments of weight, costs and machinability. The use of soft steel is also beneficial as it shields the photodetector from the stray field of the solenoid. Whenever possible all parts should be welded and separately checked against leaks.

The entrance and exit windows (not shown here) for the particles could be made with relatively thin foils, typically 0.5 mm stainless steel.

The assembly is reasonably straightforward with four different pieces, prepared separately and bolted together with O-rings at the appropriate places:

- the radiator box- a "central" box containing the large tilted mirrors and their orientation controls. This

box also supports the glass windows. It is obtained by welding flat sheets of steel.- two assemblies containing the PM tubes, their mumetal shielding and the HV

circuitry.

Due to the large opening of the system, detailed studies are needed for the shielding of PM's from the stray magnetic field. A heavy soft iron wall (with adequate large opening for the particles) is foreseen in front of the device (not shown in Figs. 4, 5 and 11). Should this be insufficient, one solution would be to place the PM's farther away from the beam axis.


C. Estimated optical performances and detection efficiency.

The simulation of the optical system with all features defined above (reflectivities, bulk scattering in aerogel, diffusive layers inside the aerogel box, transmission in the window...) is performed with Zemax v.19_08_02.

A graphical summary is shown in figure 12 as a false color intensity distribution on the photocathode of the R-3600 photomultiplier. The overall light collection efficiency exceeds 77% assuming no polarization dependent reflections.

Figure 12 Intensity distribution on the photocathode of the R3600-02 photomultiplier. The light collection efficiency exceeds 77%.

No polarization effects were taken into account.

A consequence of the overall light collection efficiency is that the theoretical threshold of 10 photoelectrons stated above corresponds to 7-8 photoelectrons generated in the photomultiplier. Finally assuming additional photon losses to about 25%, we are left with the actual detection of at least 5 photoelectrons.


The electron detection efficiency versus momentum is obtained as follows:

a) for each electron, the theoretical number of Cerenkov photoelectrons is computed taking into account the effective thickness of aerogel traversed. Remember that the nominal aerogel thickness was chosen to be 100 mm. But the effective thicknesses traversed by the electrons depend on their angle with respect to the beam axis.

Asymptotically at high energies, the number of photoelectron levels to a constant value as expected for ultra relativistic particles.At the low momenta, a high light yield is observed due to electrons exiting the solenoid at very large angle, and traversing larger thicknesses of the radiator.

b) next we assume that the electronics associated to the PM tube has a threshold corresponding to 5 photoelectrons.This value takes into account a possible noise level and incorporates some contingency for unpredicted photon losses.

The ratio of events with light yield above threshold to all incident electrons is considered as the electron detection efficiency (Fig. 13). As an indication, the statistical uncertainty at 50 MeV/c is about 7%: it is due to the limited sample used.

Figure 13. Detection efficiency for electrons versus their momentum assuming a light collection efficiency of 77% and a detection threshold of 5 photoelectrons.


D. Cost estimates (in kEuro - no VAT)

Item Supplier Delivery (after order)

Costs (no VAT)kEuro

Note

Aerogel Matsushita 100 6Airglass 5

Glass plates Schott 9-11 weeks

14

Cutting + Packaging TBD* 1Aluminization A. Braem TBD* 2.5 4

Supports TBD* 4

Containment vessels(raw steel plates + flanges)

0.5

Machining + Welding Pegard 12 wks 15 1Ni plating Pegard 4

Assembly + Tests 4

5 Photomultipliers R3600-02 Hama' 40 2+ve HV supply CAEN ?

Mumetal shielding Imphy ?Mechanical supports TBD* 5

Slow controls (sensors...) ? 1Transports/Shipping TBD* 1Miscellaneous TBD* 1 3

Total 105 7Contingency 10% 10

TOTAL 202

Notes.

* = TBD : to be defined

1. Pressure and temperature sensors with their respective controllers.Measurement of the absolute water vapor content inside the vessel.

2 Including one spare photomultiplier.Beware that it requires a positive HV supply.


3 Administrative and management costs.

4. Private communication with A. Braem and Cl. David(CERN 11.09.2002)Price is given for HARP-like layers Al + MgF2 giving a reflectivity of about 90% in the visible range.

It is proposed for MICE to use an Al + Hf2O3 coating: it is much stronger mechanically, has a somewhat narrower bandwidth (250 to 500 nm) but with 96% reflectivity.

5. Airglass aerogelStandard Slabs of 60 x 60 x 2 cm3

6. Matsushita Electric WorksMasaru Yokoyama

1048 Kadoma, Kadoma-shi, Osaka 571 Japan email: [email protected] tel: *81-6-909-7383 fax: *81-6-904-7104

Here is an example of what they can make (this dates from 1998): Size: 110mm*110mm*10mm-thickness Refractive Index: 1.015, 1.03, and 1.05 Transparency: Transmittance at wavelength 550nm >90% (thickness=10mm)Price (FOB) 50,000 yen/liter (for orders >100 liter)

100,000 yen/liter (for orders <100 liter) = 843 Euro/literThey can make aerogel having refractive index of 1.015, 1.025, 1.03 and 1.055.It is also possible to make aerogel-having indices between 1.01 and 1.055 optionally.

The aerogel having an intermediate refractive index will be cost 20% or 30% higher than normal index. The tolerance will be +/-0.02, so the tile will have the value of (n - 1) to within 6%.

7. BELLE modules 180 modules Aerogel and 3" PM R6233 cost 40MYen.(June 2002). This is about 1900 Euro/module.Ref. http://utkhii.px.tsukuba.jp/~highpt/2002

To cover the same area 600 x 600 mm as MICE, we would strictly need 36 BELLE modules, corresponding to a cost of 36 x 1900 Euro/module = 68400 Euro.But to get some overlap between the modules let's assume we need about 50 modules at a cost of 50 x 1900 Euro/module = 95000 Euro.Remark. The BELLE's modules do not seem to have any magnetic shielding.


9.2 DAQ, on-line monitoring and reconstruction in each sectio : responsibilities, costs and time line


10 DAQ

10.1 Introduction

10.1.1 Beam structureThe ISIS beam delivers a spill of 6000 bunches 100ns long, separated by 220ns inter-bunch, and with a repetition rate of 50Hz. The RF power system cannot stand a duty-factor of more that 10 -3; using it in the most efficient way delivers 850s flat-top at 1Hz, during which about 2600 bunches will be reaching the experiment. The storage capabilities of the digitizer boards may put limits on the duration of such a flat-top: as an example, the FADCs (s at the design sampling frequency) used in the TPG will need to operate the cooling channel at 2Hz (2 times 150s rise-time followed by 350s flat-top), or at 1Hz keeping only 500s of data.For the purpose of evaluating the data rates, it has been taken into account that only 1/4 of the incoming particles will fall into the acceptance after the diffusing plates, and that only 1/6 of them will be in phase with the RF, so being able to travel down the cooling channel.

10.1.2 Working Hypothesis The beam time-structure is such that there is not enough time for particle-by-particle

readout. For this very reason, the digitizer boards like ADCs and TDCs will have to buffer data during the ISIS flat-top, waiting for read-out by the DAQ at the end of the spill.

A 850s flat-top at 1Hz is used for the Sci-Fi tracker option (IS THIS CORRECT?), while a 350s flat-top at 2Hz is considered for the TPG option.

The maximum beam rate will be 1 particle per bunch. This is a limit case, since it will often provide more than one particle in the same 100ns time-window.

One “calibration” (RF off) cycle will be taken per normal (RF on) cycle The front detector station will have to measure both out-of-acceptance and out-of-RF-

phase particles The rear detector station will not receive any particle when RF is OFF.

10.2Detectors and Data Volumes

10.2.1 Trackers

10.2.1.1 SciFiWe assume here that one particle will make about 60 hits in the fiber tracker station. Roughly assuming one data word per hit, this leaves us with 240 Bytes per particle and per station. The front station will see about 2600 such particles during the 850s flat-top, corresponding to about 600KB. If one additional 850s interval is used for calibration purposes with RF OFF, the total data rate due to beam particles will be 1.2MB/s.The data volume due to beam particles in the rear station can be neglected, since it will contain only 1/24 of the particles when the RF is ON, and none when OFF.The picture may, however, be changed by the possible presence of noise due to X-ray conversions in the fibers. If one considers a high hit rate due to noise, like 1MHz/fiber in the plane near the RF cavity, a total of about 15MB/s per tracker station has to be added to the overall data rate. Taking into account all the sources, the data rate from the fiber trackers is considered to be about 35MB/s.


10.2.1.2 TPGTo ascertain the ability of the DAQ to handle the data volume produced by the TPG, the limit-case of 50% occupancy in the detector has been assumed. Each TPG tracking station accounts for 1800 readout channels, sampled 1000 times every 500ns. If 50% of the samples are over the ADC threshold, 0.9 million hits per TPG are produced. Thanks to the on-line data compression, this will fit in 0.3 MWord, or 1.2MB per TPG and per cycle.The total rate from the trackers, in the case of the TPG option, will not exceed 10 MB/s.

10.2.2 TOFThe beam particles will produce 4 hits in the front station’s TOF planes, and 2 in the rear station’s one. Since, for every hit, both ADC and TDC words will be stored, a minimum of 32 Bytes will be produced in the front station, the rear one being negligible. The total rate, including the cycle taken in RF OFF mode, will be about 0.5MB/s.

10.2.3 e-IdEven if the e-Id may represent several hundred channels, the fact that it is placed in the rear station makes its data rate completely negligible.

10.3TriggerFrom the point of view of the DAQ, in normal data taking conditions the trigger must provide start-of-spill and end-of-spill signals. Timing signals of the 100ns-long bunches are, even if useful for other purposes, not needed at the DAQ level, since there is anyway no time to read out in-between bunches.The trigger system must, however, provide enough flexibility to allow the detectors to run in stand-alone mode for setup, test and calibration purposes.

10.4Readout and Network Structure

10.4.1 Crates and Readout BusAll the digitizer boards will be held in VME crates, either 6U or 9U. As it can be seen from the table, no crate is exceeding the maximum sustained rate attainable on a standard VME bus.Subsystem # VME crates Data volume Data volume/crateTPG 3 front + 3 rear Up to 10MB/s 1.5MB/sSciFi (?) Several (?) Up to 35MB/s 5MB/s (?)TOF 1 front + 1 rear Up to 0.5MB/s 0.25MB/se-Id 1 rear (same crate as TOF?) - -

10.4.2 Readout ProcessorsEach crate will be equipped with a VME processor, which is a Single-Board Computer provided with a VME interface and a network connection. The task of the processor will be to read data out of the digitizer boards via the VME bus, and to pass them over the network to the Event Builders. The evaluated data rates indicate that a standard 100Mbit-Ethernet interface will be fast enough.

10.4.3 Switched NetworkAs it is nowadays common, the DAQ system will make use of a switched Ethernet network for transferring data to the Event Builders and the Storage System, to synchronize and control the DAQ processes running on several nodes, and to perform online monitoring tasks. A “backbone” Gbit-Ethernet switch will directly connect the Event Builders and the Storage


System, and offer “down” links for at least two lower-performance switches, dedicated to VME readout processors and control/monitoring workstations.

10.4.4 Event BuildingTo cope with the total rate of the experiment and to allow for redundancy, the DAQ will comprise at least 2 Event-Builder (EVB) computers, each one equipped with Gbit-Ethernet interfaces. Their role will be to assemble event fragments produced by the many VME readout processors, make consistency checks, write assembled events to the Storage System, and serve live data to the monitoring tasks. The possibility to give the EVB system enough parallelism and computing power to perform on-the-fly data compression will be considered and tested, in the effort of minimizing the storage cost.

10.4.5 Storage SystemThe more widespread low-cost storage solutions for scientific laboratories are large PC cases holding several large-capacity E-IDE hard disks, connected to the PCI bus by hardware-RAID controllers, and configured as ~1TB available capacity. The performances of these redundant storage systems are rated at 30 to 50 MB/s, so that two of such systems may be able to cope with the data rate from MICE. These figures have to be taken with care, since the online Storage System will not only provide space for storing runs, but will also serve data for online-analysis work. For this reason, the hardware will have to be tested for concurrent sequential read/write access of large files before being considered compliant with MICE rates.MICE will be able to cool, at most, 100 muons per seconds. To achieve a statistical precision of 10-3, one has to take data for 1000 seconds, during which the detector can produce up to 40GB of data. However, it is clear that the total data volume is very much dependent on the X-ray noise rate. In the absence of the latter, the total volume (hence the storage cost) would be much reduced.Storing about 100 runs may need up to 5TB. The availability of a centralized storage system at RAL would help in reducing the online Storage System complexity and avoid the need of a high-performance backup system.

10.4.6 Operating SystemThe online cluster, including the VME processors, will use the Linux operating system. Since no read-out operations are foreseen during the spill, it is unlikely that Linux real-time extension will be needed on the VME processors; this matter will be subject to tests.

10.4.7 Need for on-line analysis?

10.5Costs The following budget table is compiled under a few assumptions: MICE will have to buy the VME crates. The cost of storage is evaluated at today’s prices, which will most likely drop in the

next year. The VME processors can be recovered from the HARP experiment at CERN,

provided they meet the MICE performance needs, to be assessed by testing.Items Number Cost in KEuroVME crates 10 (?) 100VME processors 10 (?) 40 (0 if recovered from the HARP experiment)100Mbit switches 2 5Backbone switch 1 10


Event Builders Up to 4 15Mirrored disk buffers 5x1TB 40Tape unit ?Workstations + monitors (control, monitoring, on-line analysis)

20 50

TOTAL 220 to 260 (depending on availability of VME processors)


11 Radiation on detectors (Norem/McKigney), 6 pages

11.1General understanding and characteristics

11.2Measurement results and effect on detectors

11.3Projected effect on performance of MICE (with help of simulation team)

11.4Further measurements to come


12 Implementation of MICE at RAL (Ivaniouchenkov, Black), 6 pages

12.1Hall layout

12.2Services, cryogenics, etc

12.3Safety issues

12.4 Interference with ISIS

12.5 Installation plan, time-line, costs and requirements


13 Summary of responsibilities, costs, funding and grand time line (Zisman, Edgecock), 6 pages


14 Summary and requirements from RAL (NN), 3 pages


Appendices (if needed)


15 References

Documents

1 - Istituto Nazionale di Fisica Nuclearepeople.na.infn.it/~palladin/mice_proposal1.doc · Web viewIt has been chosen to use MS Word to edit the proposal. Word is easy to use, but