The hadron calorimeter (HCAL) consists of three subsystems. A barrel calorimeter, HB/HOB measures particles with h < 1.4. An endcap calorimeter, HE/HOE, measures particles in the range 1.4 < h < 3.0. A forward calorimeter, HF, measures particles in the range 3.0<h<5.0. The barrel calorimeter part inside the coil is called HB (Hadron Barrel). The two layers of the calorimeter outside the coil use the coil and the muon iron as the absorber material and are called HOB (Hadron Outer). Similarly the Hadron Endcap calorimeter (HE) has a small one layer section using the muon iron as an absorber and is called the HOE calorimeter.
HB/HOB and HE/HOE are tile-fibre sampling calorimeters, i.e. they use scintillator tiles and fibre readout to sample the energy deposition of hadrons in the copper and steel absorber. Fig. 6.1 shows a generic diagram of the optical readout system of these calorimeters. The scintillation light is collected using wavelength shifting (WLS) fibre embedded in a groove in the tile. The groove follows a "sigma" (s) pattern on the tile, as shown on Fig. 6.2. Outside the tile, the WLS fibre is spliced to a clear fibre. Clear fibres are connected the readout device via an optical cable using mass-terminated connectors.
Fig. 6.1: Generic diagram of the optical readout system for HCAL
Barrel and Endcap calorimeters.
For the HB, where the readout is located inside magnetic field of 4 tesla, the readout devices are Hybrid Photo Detectors (HPDs). The readout for the HO calorimeter is located in the region where the magnetic field is approximately 1 kG. Here either HPDs or shielded photomultiplier tubes can be used as a readout devices. The current default design calls for using HPDs everywhere to minimise the number of optical readout systems.
The inner HB calorimeter consists of 18 sampling layers as shown in Fig. 6. 3 for the r-f view and in Fig. 6.4 for the r-z view. The innermost layer (Layer 0) is located between the steel supporting plate of the ECAL calorimeter and the inner HCAL endplate, i.e. before any copper plates. The thickness of the scintillator for Layer 0 is 9 mm. The sampling layers between the inner and outer HCAL endplates (Layers 1 through 16) are placed every 5 cm of copper. They are instrumented with 4 mm thick scintillator, Kuraray SCSN-81. Layer 17 is placed between the HCAL outer endplate and the cryostat and will be instrumented with 9 mm thick scintillator. The outer HOB calorimeter consists of two sampling layers (three layers at h=0). The first layer of HOB is placed after the solenoid magnet and the second layer of HOB is placed after the first 20 cm thick muon steel absorber. The HOB layers are instrumented with 1 cm thick Bicron BC408 scintillator. All HB and HOB scintillator layers are read out by 0.94 mm Kuraray Y11 multiclad WLS fibres.
The HB/HOB calorimeter is segmented into three longitudinal readout segments. Layer 0 is separately read out as segment H1. This segment measures the hadronic component of the shower which was produced in the EM crystal calorimeter. Since hadronic energy in ECAL is undermeasured by the crystal, the ratio of energy in H1 to total Hadronic Energy (H1+H2+HOB) can be used in an algorithm to improve the resolution of the combined EM+HAD system. The second readout, H2, consists of 17 layers (1 to 17). The third readout, the HOB, consists of the two layers (or three layers at h = 0) located after the solenoid. The HOB segment corrects for the high energy tails measured by H1 and H2. The HOB and HOE calorimeters will also be used by the muon system to identity muons.
Fig. 6. 4: r-z view of a HB 20 wedge. The inner and outer steel plates are not indicated on the drawing. The dimensions are stated assuming the inner HB radius of 1806.5 mm.
The scintillators are inserted from the large h end of the HCAL wedge into the 9 mm gaps between the copper plates. Fig. 6. 5 shows a typical scintillator tray assembly for the a single layer of the HB calorimeter. Each readout layer consists of three elements: Side-Left, Middle and Side-Right scintillator trays which are inserted to separate copper slots. The neighbouring tiles in each tray are optically separated by grooves filled with white epoxy glue. The epoxied multi-tile sub-assemblies cut from a single piece of scintillator are called megatiles. Several megatiles are put together between two black plastic cover plates to form a scintillator tray. The bottom cover plate is 0.95 mm thick. The top cover plate is 1.9 mm thick. The routing of the fibres is inside the top cover plate.
The top cover plate has additional grooves. One of them is for a 1.3 mm diameter source tube where a remote motor will drive a pointlike radioactive 137Cs g source embedded in a tip of a wire. This source will be used for calibration both during construction and data acquisition. All layers will have "occasional" source tubes for use during checkout without a B field, while the endcaps are withdrawn. In addition, two layers in each tower (layer 0 in H1 and layer 9 in H2) will have a source tube permanently coupled to one of source drivers capable of operation with the magnetic field on.
Another groove is for a quartz fibre that will be used to inject either UV or blue light into the tiles. The quartz fibre will be inserted into the same two layers (one in H1 and one in H2) which will have permanently accessible source tubes. The light injection to tiles (or to photodetectors separately) can be done much more frequently. Light injection and g source calibration will allow us to monitor gain changes in the photodetectors as well as changes in the light output of the tile/WLS fibre system.
As a part of quality control each scintillator tray assembly will be scanned by a collimated radioactive g source in a 1.6 m x 5.7 m megatile scanner. Following this measurement, the megatile will be scanned with the radioactive wire source. The collimated source provides an absolute excitation of each tile, and its ratio to the wire source excitation (which depends on tile size and other solid angle effects) provides a permanent data base to enable the accurate transfer of testbeam calibrations to the assembled calorimeter, via the wire source.
The HB/HOB calorimeter designs are close to the design of the CDF Plug Upgrade Calorimeter. Since all the components of the CDF Plug Upgrade Calorimeter are built and tested, many of the design and production issues for the CMS calorimeter are well understood. The HCAL group has built testbeam calorimeters using the tile-fibre method for 1994, 1995, and 1996 testbeam runs at CERN. Hence, the HCAL group is experienced in building tile-fibre calorimeters. The optical system of the HE calorimeter has similar design. Each scintillator tray covers a 10o wedge. There are 19 sampling layers in each Endcap module.
The testbeam results show that the HB calorimeter can attain the
target energy resolution of 100%/
5%. In
terms of component performance, the 5% constant term requires
that the variation of the tile-to-tile light yield be better than
10 % and that the rms of intra-tile transverse uniformity be better
than 4% (even if the nonuniformity is the same for all tiles in
a projective tower). An overall tile-to-tile light output variation
of 10 % contributes approximately 3% to the constant term because
the hadron shower is typically spread out over more than 10 layers.
R&D on prototypes and QA/QC tests on completed calorimeter
counters indicate that this performance level has been met.
The exact magnitude of the induced constant term depends on the calibration method. If all towers of the HCAL are to be measured in a test beam, or with a source/cosmic ray muon to sufficient accuracy, than a "local" calibration can be made, which means that all towers in HCAL have the same mean. If this cannot be accomplished, than the towers have means which vary due to the tile manufacturing tolerance. Clearly, in this "global" case, one has a larger induced error for HCAL taken as a whole, since the energy mean responses are not all the same.
The results shown in Fig. 6. 6 indicate that a 10% gaussian error on the individual towers induces a 2.5% local and a 3.2% global constant error in the fractional energy response. Since the H2 test beam data imply a 5% constant term, the additional error, folded in quadrature, causes a small and acceptable increase in the HCAL energy resolution.
Fig. 6. 6: Induced constant term in the fractional energy error in the HCAL (y axis) as a function of tile manufacturing quality (fractional rms of light yield, x axis). The star symbols correspond to global calibration case and the open circles correspond to the local calibration case.
We require the calorimeter to have enough light to achieve 100%/
5% and to determine at 3 s level if
a muon traversed a tower. A minimum ionising particle going through
17 layers of the H2 readout section of the calorimeter with 5
cm copper sampling corresponds to an energy of 2 GeV. We require
a minimum light yield of 1 pe/mip/tile. Therefore, the photostatistics
contribution to the fractional resolution with a light yield of
17 pe/muon in 17 layers is 45% /
, or half
of the 100 %/
expected from sampling.
Presently, electronic noise in HPD (at the gain of 2000) corresponds to approximately 2 photoelectrons. Thus 17 pe/muon is also the minimum number for the HPDs and electronics to see a muon. For a factor of 25% in safety, we should start with at least a level of 20 pe/muon. This light yield level has been achieved in CDF, which was able to get more than 1 pe/muon per tile, or more than 17 pe/muon in 17 tiles. In CMS initial tests show that we can achieve a light yield of 2 pe/mip/tile.
Fig. 6. 7 shows the radiation levels over the detector. The loss of light from radiation is another source of variation of the light yield of tiles. Since the maximum allowed tile to tile variation is 10%, we want the light loss from radiation to be less than 10%. For the HB calorimeter, the radiation damage (reaching 20 krad at radius of 198 cm at h=0 and 30 krad at h=1.1) is expected to be small over 10 year lifetime of the detector, assuming a total of 5105 pb-1 of integrated luminosity. We will monitor the decrease to 2-3% as a function of time, which is much better than our margin of safety.
Fig. 6. 7: Radiation levels in the CMS detector. Fluence of hadrons E>100keV in cm-2 s-1 (upper plot) and radiation dose in Gy (lower plot) in the HB/HE region. The dose values have been smoothed by taking weighted running averages over neighbouring bins. Values are given for 5 x 105 pb-1. The intermediate (dashed) contours in the fluence plot corresponds to 3.16 x 10n. The dotted lines indicate the geometry.
The lateral segmentation must be fine enough for detection of
narrow states decaying into pairs of jets. A segmentation of 
=
0.087 x 5o is sufficient for good di-jet separation
and mass resolution. The calorimeter must have sufficiently good
time resolution (
< 5 ns) to determine
the beam crossing of electrons and jets of physics interest. We
require that 99% of the light generated by a particle going through
a tower be collected in two beam crossings (each crossing is 25
nsec).
The length of the HB scintillator trays vary from 3.7 m to 4.3 m. The central tray vary in width from 33 cm to 50 cm and the side trays vary in width from 16 cm to 24 cm. The smallest tile in the HB calorimeter is 15 cm x 17 cm and the largest tile is 24 cm x 40 cm. The HE trays are 10o wedges called Pizza Pans that have the shape of a trapezoid with the tiles located 35 cm to 280 cm radially away from the beam.
The scintillators with the WLS fibres are the active medium of the hadron calorimeter. In order to attain the required resolution and maintain this value over the 10 year lifetime of the detector, the scintillators and WLS fibres have to meet stringent material specifications. Energy of the particles in the calorimeter is measured by the amount of light that reaches the photo detector. The criteria are developed so that a minimum ionising particle will produce nearly the same amount of light no matter where it crosses the scintillator in a tower. The baseline for the HB/HE detectors is the Kuraray SCSN81 scintillator (polystyrene based plastic) and the Kuraray Y11-250 double clad (WLS) fibre. The HOB and HOE tiles are substantially larger in size with the lower light yield. For these tiles, a higher light yield is required so that minimum ionising particles can be observed. The baseline is BC408, for which the plastic base is polyvinyl toluene (PVT) which produces a factor of two more light compared to SCSN81, and the thickness is 10 mm. The WLS fibres are the same as for the HB/HE calorimeters.
The scintillator material should have good mechanical and thermal properties. Its hardness and deformation temperature should be high to allow for fast sawing and machining with only cold air stream cooling with no lubrication. A minimum machining speed for a 0.88 mm groove 0.4 mm deep is 250 cm/minute with no signs of melting.6.2.1.2
The light yield is directly proportional to the scintillator thickness. The nominal value for layers 1-16 is 4.0 mm. Since the scintillator together with the plastic and other materials that form the scintillator tray has to fit inside the 9 mm slots in the calorimeter, the thickness cannot increase. A tolerance is: Thickness = 4.0 mm ±0.4 mm. Within a single scintillator sheet the thickness tolerance is better ± 0.2 mm. The thickness for layer 0 and layer 17 is 9 mm. The HOB and HOE are constructed from thicker scintillators and the nominal value is 10 mm ± 1 mm.6.2.1.3
The light yield should match WLS fibres doped with K27 dye (such as double clad Y11 or BCF91A), and produce light signal that is equal or exceeds the baseline scintillator and fibre. Fig. 6.8 shows an ADC spectrum of a photomultiplier tube signal from a minimum ionising particle traversing a 19 cm by 26 cm tile. The dimensions of this tile correspond to the size of HB tile in layer 7, tower 10. The light from the tile was collected with 0.83 mm WLS fibre spliced to 1.5 m long clear fibre. The fibre was then connected via a 65 cm long optical cable to a photomultiplier tube. Note that the light path included two optical connections and the WLS and clear fibres have length expected for the actual readout in the CMS detector. The light yield of this tile (4 mm SCSN-81) is approximately 2 photoelectrons/mip. The measurement was performed using a 2 MeV electron gun. We require that light yield of scintillator would be at least as high as of the sample used in this measurement.
Fig. 6. 8: ADC spectrum of a PMT signal from tiles using collimated 2 MeV b source. The upper plot shows a spectrum for a calibration tile used to establish ADC-to-pe scale. The lower plot shows the spectrum of a 19 cm x 26 cm tile read out with optical cables corresponding to the actual length for the CMS. The light yield of the tile is approximately 2 photoelectrons/mip.
The attenuation length of scintillator must be long enough, so that the light yield of larger tiles (24 cm x 40 cm in HB) is not degraded relative to the light yield of smaller tiles (15 cm x 17 cm in HB). In addition, the uniformity of the light yield across the tile is also dependent on the attenuation length of the scintillator. In order to keep the uniformity of tiles within few percent, the effective attenuation length must be 80 cm or larger.
The typical irradiation dose in the barrel at shower maximum in 10 years of operation at design luminosity is about 300 Gy (30 krad) at h = 1.1. Most of the commercial scintillators and fibres will survive these rates with minimum degradation. The irradiation doses at shower maximum in the end caps are substantially higher reaching a value of about 0.4 Mrad (4 kGy) at h = 2.0 and about 2.4 Mrad (24 kGy) at h = 3.0. There are no commercially available scintillators that would survive the radiation dose in the region of 2<h< 3 without a loss of 50% in light yield.
Fig. 6. 9 shows the relative light yield of a 10 cm x 10 cm tile/fibre assembly as a function of radiation dose. The light yield versus radiation dose follows an exponential form [exp(-Dose/D0)]. This light loss is due to the reduction of the scintillator and WLS fibre attenuation lengths. The D0 values for the fibres are about 20 Mrad for a 40 cm fibre, while that of scintillators vary from about 3 to 10 Mrad for a 15 cm x 15 cm tile. In order to keep below the 10% damage limit, the combined (scintillator and fibre) D0 should be greater than 6.5 Mrad for h <2. The only commercial scintillator that meets this value at this time is SCSN81.
Therefore to be able to partially correct for the radiation damage of HE scintillator trays, the HE calorimeter will have multiple readout sections in the h>2 region. Since most of the light yield is at shower maximum, where the damage is also maximum, a simple correction (first order) as a function of tower h can correct the light yield loss. As the light yield reduction is dependent on the tower depth, a full correction is not possible.
Fig. 6. 9: Relative light yield of tile/fibre assembly as a function of radiation dose. No radiation damage was noticed for doses below 0.1 Mrad.
In order to compensate the radiation damage of the scintillators the calorimeter must be divided longitudinally. The drop of light yield can be corrected by adjusting the weights for corresponding readout segments. In addition, the absolute light yield for minimal ionising particle must be at least 1 photo electron to minimise contribution of photostatistics to the HE calorimeter energy resolution.
Taking into account the longitudinal dose distribution after 10 years of operation for the most loaded scintillator presented in Fig. 6. 10, and the scintillator SCSN81 degradation vs dose shown in Fig. 6.9, we have the longitudinal distribution of the scintillator light yield reduction after 10 years of operation, presented in Fig. 6. 11. The minimal number of the longitudinal divisions follows from a 20 % uniformity specification within each readout segment. Thus, the first read out is the first layer (after 8 cm plate), the second read out is the next 4 layers and the third read out is the last 14 layers.
Fig. 6. 10: Radiation dose (in Mrad) in HE versus calorimeter depth for ten years of LHC operation (assuming 1034/cm2 sec). Note that this level is twice that quoted for the first ten years of LHC operation.
Fig. 6. 11: Reduction of light yield for HE scintillator versus depth layer for ten years of LHC operation at full design luminosity
The radiation doses for HOB and HOE are negligible and pose no problems.
It has been known for some time that magnetic fields increase the light yield of scintillators. Fig. 6.12 shows the light yield increase of various scintillators that are commercially available. The light yield increase saturates at about 2 tesla. Therefore for the CMS detector, where the scintillators are within 2 to 4 tesla field, this corresponds to a simple overall correction term.
This section addresses the quality control of scintillators produced by manufacturers. This section will form the basis of writing the scintillator specifications.
The scintillators are to be delivered in rectangular plates with saw cut edges. The two diagonals of the rectangle are required be of equal length with a tolerance of 5 mm. The plates are to be covered on both sides with adhesive, protective paper cover sheets. The cover sheets must be easy to remove and must not damage the scintillator. Samples of scintillators will be machine cut and grooved at the speeds specified above. The cuts must look clean and show no obvious sign of melting. The machined grooves should also show no melting or friction welded chips. The plate thickness will be sampled at uniformly spaced points to verify that it meets the thickness tolerance. The light output will be measured on various samples of grooved standard tile (such as 15 cm x 15 cm) with a standard WLS fibre to show that it produces at least sufficient light to meet or exceed the baseline elements of SCSN81 and that the light yield variation from plate to plate is within the tolerances described above. The attenuation length will be measured of sufficient number of samples to ensure that it meets the specifications. Several samples from each batch of plates will be irradiated by either electrons or photons to 0.5 Mrad (5 kGy). The light yield loss for this level of radiation should be less than 10%. For the lifetime requirement that the scintillator overall light yield does not degrade
in 10 years by more than 25%, the manufacturer should supply data that the type of scintillator has been used in high energy experiments, under conditions similar to this experiment and that the scintillators have shown no degradation, and are chemically, mechanically and optically stable and the surface crazing with time, if any does not reduce the light yield by more than a few percent.
Each plate from the manufacturer will be identified by a unique code and a data base will be produced that will identify each megatile as to the plate code, date delivered, date and machine that cut and grooved it, plus any other information that will be necessary to keep a lifetime history of each scintillator tray.
The baseline is a double clad 0.94 mm fibre doped with 250 ppm of K27 green dye. Each tile within a specific tower will have a WLS fibre of the same length, diamond cut at each end. This length corresponds to the green WLS fibre length required by the largest tile in each tower (i.e. layer 17). Note that tiles in towers 17 and 16, and some in tower 15 will have no splices since they are close to the connector at the edge and will have to be treated in a special manner. One end of the WLS fibre will be mirrored by aluminium sputtering and dipped in a clear coating for protection. The other end will be fused to a clear fibre of the same type and diameter. The second end of the clear fibres for one row of tiles in a scintillator tray will be placed in an optical connector and diamond cut polished. Two fibres produced by industry could meet these specifications, namely BCF91A-MC by BICRON Corporation and Y11-250 by Kuraray. The non-S type fibres of Kuraray are the default design. The non S type fibre is the simplest to polish and splice. However, the S-type fibre does have better flexibility. We will be doing R&D on getting and testing a slightly more flexible form of the non S-type fibre. Note that Kuraray can produce fibres with properties between S type and non-S type fibre. S-type denotes a slow extrusion process and non-S type is a faster extrusion speed.
The current design for CMS uses 0.94 mm wavelength shifting fibres. We have measured transmission of splices, light yield and attenuation lengths for 1 mm fibres. These tests indicate that 1 mm fibres have similar performance to 0.83 mm fibres, which were used in the CDF Plug calorimeter. However, the ball groove can be cut much faster for 1 mm and 0.94 mm fibres. The design calls for 0.94 mm throughout the system from tile to connector to optical cables to the decoder box and the HPDs. This size was chosen because unlike the 1 mm case, the 0.94 mm option does not require any change in the current HPD design, as the 0.94 mm fibre can fit within the current active area of the HPDs.
The WLS fibre should have an attenuation length of at least 1 m
and the clear fibre and attenuation length of at least 6 m when
measured with a green sensitive (enhanced) bialkalai photo cathode.
Fig. 6. 13 and Fig. 6. 14 show the measurement of light attenuation
in Kuraray and Bicron multiclad WLS fibres. Fig. 6. 15 shows
attenuation of light in a clear multiclad fibre.
The set-up consisted of a 8 m long piece of scintillator (similar to BC404). The tested fibre was inserted into a long groove extending through entire length of the scintillator, pushed up against a light mixer and connected to a R580-17, green extended tube. The length of the fibre outside the scintillator was 7.5 cm. The scintillator was excited using a movable strontium source rides in a plastic enclosure. The DC current from the photomultiplier tube was measured with a picoammeter.
Fig. 6. 13: Attenuation of light in 1.0 mm Kuraray multiclad WLS Y11. The fitted value of attenuation length in the range of 0.3 m < x < 1.0 m is equal to l=2.3 m. The light yield ratio, LY(1.0m)/LY(0.3m)=0.71. Note that CMS will use 0.94 mm diameter WLS fibres.
Fig. 6. 14: Attenuation of light in 1.0 mm Bicron multiclad WLS Y11 fibre. The fitted value of attenuation length in the range 0.3 m < x < 1.0 m is equal to l= 1.5 m. The light yield ratio, LY(1.0m)/LY(0.3m)=0.65.
Fig. 6. 15: Attenuation of light in 0.83 mm Kuraray multiclad clear fibre used in CDF Plug Upgrade. The fitted value of attenuation length in the range 0.3 m < x < 3.0 m is equal to l = 6.8 m. The light yield ratio, LY(3.0m)/LY(0.5m)=0.70.
The radiation damage, at levels encountered in the CMS detector, results in an additional reduction of the fibre attenuation length. For the HB, this is not a problem for the WLS fibres made by Kuraray and BICRON as levels are small and WLS fibre lengths are short. For the HE, the fibres in the high radiation area are short and the light loss is negligible.
The light yield from WLS fibres coupled to scintillators have been measured to be roughly proportional to the diameter of the fibre. The specifications of the core diameter is 0.94 mm with the double clad adding 60 to 120 microns. The concentration of the K27(250 ppm) should be uniform enough for all batches to keep the variation in light yield and attenuation length to a minimum. The core diameter and K27 concentration variations should be small enough so that overall light yield does not vary by more than 3%.
The testing of fibres for light yield will be based on samples from various production batches to assure that it meets the 3% tolerance. These fibres will be cut a standard length of 80 cm, diamond cut each end and inserted inside the groove of a standard 15 cm x 15 cm scintillator and excited by a same radioactive source.
To check attenuation length of WLS fibres, fibres will be cut to 1.5 m length and diamond cut polished. Each end of the fibre will be attached to a green extended photomultiplier tube and placed in an attenuation measuring box. The fibre will be excited by the scintillator chosen for the detector. The DC current from the photomultiplier will be recorded as function of light excitation position. Most of the green light travels in the core of the fibre. A small portion of the light travels inside the clad (with a very short ~20 cm attenuation length), this causes the light yield as a function of distance to be a double exponential with a long and short components. The attenuation length, measured in the region. between 0.3 m to 10 m away from photodetector. should be greater than 1.0 m.
The attenuation length of clear fibre will be measured by splicing a WLS fibre to it and inserting it to a scintillator tile. By cutting pieces of clear fibre off (and resplicing it to the WLS fibre) we will be able to plot light yield at the other end of clear fibre as a function of clear fibre length and fit it to an exponential function. We will use the clear fibre length in the range 0.5 m to 4.0 m and require that attenuation length is greater than 6 m.
Both WLS and clear fibre samples will be irradiated and attenuation length measured. At 0.5 Mrad (5 kGy) the reduction of light yield for a typical tile at shower maximum should be less than 3%.
There are two different types of quartz fibres in the HF detector. In the region |h| > 4, radiation-hard fibres with fused silica core and a fluorine-doped silica cladding are used. In the lower pseudorapidity region, |h| < 4, where the radiation levels are less severe, fused silica core and a plastic cladding fibres are planned to be utilised. Both types of fibres are 345 micron in outer diameter with 300 micron diameter core.
Fibre lengths will nominally vary from 2.5 to 3.5 meters depending on the location of the towers with respect to photodetectors. The optical transmission of <1 dB/m is required to minimise loss of light. Both end of the fibres will be polished. The OH- level in the fibre core material is typically 1000 ppm. To avoid breakage during construction, we require short term mechanical stability at minimum bend radius of 1.5 cm or less. The tensile strength is required to be above 12 kg. The required fibre core noncircularity is less than 5% and the desired core concentricity error is 3%.
Table 6. 1 summarises the necessary dimensional characteristics for fused silica core fibres. Type I is identified as quartz-quartz (QQ) fibres in order to indicate the core and cladding materials. The core is fused silica and the cladding material is fluorine-doped silica. The buffer provides radiation hard coating and structural stability, e.g. polyimide. Type II fibre has the identical core but the cladding is made out of a radiation hard synthetic material (polymer) and less costly and is identified as QP to indicate the core (quartz) and the cladding (polymer) materials. The numerical apertures of the fibres that are used in prototypes are 0.22±0.02 for QQ and 0.35±0.02 for QP for visible light.
|
|
|
| |
The extreme radiation levels where the HF will have to operate put stringent requirements on the acceptable levels of degradation in fibres. The expected total dose at EM shower maximum at |h| = 5 reaches few hundred megarads in ten years and in the body of the HF, the total accumulated dose varies by four orders of magnitude.
The single most important reason for choosing quartz as active medium is its inherent radiation resistance. High-purity quartz (suprasil) has been reported to withstand radiation levels up to 30 Gigarads with the transparency of small samples in the wavelength range 300-425 nm changing by less than 2%[2]. The radiation hardness of quartz fibres depends on its core material, in addition, on the properties of the low refractive index cladding material. With fluorine-doped silica cladding, the amount of light changed by less than 10% up to levels of 20 Gigarads[2]. Radiation damage studies with intense photon sources resulted in about 50% transmission loss in the visible band after 1 Gigarad of dose for 1 meter length fibres[3]. The performance degradation of the HF due to the expected radiation damage is addressed in detail elsewhere in this document. The radiation hardness of QQ and QP fibres are being carried out with protons, electrons, photons and neutrons by the members of HF group at different facilities. Fig. 6. 16 shows the radiation damage of QQ fibres under photon radiation.
Fig. 6. 16: Radiation damage characteristics
due to photons of fibres with quartz core and fluorine doped quartz
clad with polyimide buffer (Type I) as a function of total dose
and wavelength is shown above.
There are a number of quartz fibre vendors that can manufacture what is needed for the HF. Several standard measurements and tests are performed by some manufacturers.
a) Dimensional tests (clad and core): on-line using a laser micrometer. In the drawing stage, 100 % of the fibre is tested and continuously monitored.
b) Dimensional tests (core, clad and buffer: spool ends are viewed under a microscope to verify dimension. This test is carried out in sampling mode.
c) Proof test: 100 kpsi, 100% of fibre is tested as being drawn,
d) Proof test: 100 kpsi, 100% of fibre is tested while spooled,
e) Optical attenuation (100% of fibre).
A set of non-standard tests are also required to be performed by the manufacturer in sampling mode;
a) Tensile test/Weibul Plot. 60 m of fibre for test is needed,
b) Thermal cycling in the ranges of -50 C to 80 C,
c) Attenuation length vs bending radius,
d) Attenuation length vs temperature, and
e) Numerical aperture (NA) test.
Before fibre insertion into the absorber matrix, we plan to carry out our set of quality tests for physical dimensions, tensile strength, bend radius, light attenuation length and radiation damage for each draw batch and archive fibre samples in order to be able to trace back problems should they appear in a later stage. The total length of a batch depends upon the size of the preform and these sizes typically vary between 40 cm to 120 cm. For each batch, we will archive 100 m of fibre, in addition to the tests we plan to carry out per batch.
The HB/HOB barrel hadron calorimeter has a large number of towers: 2448. The 18 layers in the inner HB calorimeter consist of 42624 tiles. Each tile is 5 degrees in j and 0.087 in h. The wedges are 20 degrees in j (4 tiles each). There are 18 j 20_ segments and 36 wedges (2 wedges for each j section, one for positive and one negative h). To simplify production and assembly into the copper absorber, the scintillator plastic for scintillator trays are packaged into a mechanical units called a scintillator trays. These scintillator trays contain the scintillator, readout fibres and optical connector at the end of the unit which the light can be accessed.
The HB scintillator trays are divided into two types of trays. One type is a side scintillator trays (Left and Right Side Tray) and the other type is a middle scintillator tray (Middle Tray) (see absorber).
A top view of the scintillator trays for layer 8 is shown in Fig. 6. 5 and the cross section of the scintillator tray is shown in Fig. 6. 17. The unit begins with a 0.95 mm thick, black polystyrene, Bottom Cover Plate. Then comes a sheet of black, opaque, .05 mm Tedlar which will wrap around the scintillator to provide an opaque layer of light tightening. Then the 4 mm SCSN81 scintillator plastic is covered on both sides with 0.15 mm thick, white, reflective Tyvek paper. The surface of the scintillator tiles are grooved to hold the WLS fibres. The black Tedlar sheet is wrapped completely around the scintillator and the top Tyvek. One side of the Tedlar sheet is taped onto the upper Tyvek sheet, and the other is taped to the Tedlar thus completing the wrap. Above the scintillator plastic is a 1.9 mm thick, black polystyrene, top cover plate (Fibre Routing Plate). Both the top and bottom plastic plates have the same transverse size as the scintillator. The fibres rise out of the scintillator through a 3.2 mm x 25.4 mm slot in the Fibre Routing Plate into 1.6 mm deep grooves on the top side of the same Fibre Routing Plate. The fibres are mass terminated at the edge of the tray into an optical connector. A mass terminated optical cable transports the light from the tray to the photomultiplier tube readout. At the photomultiplier tube readout, an equivalent of a cable "patch panel" assembles fibres from different longitudinal tiles of a tower onto the HPD designated for that tower. Source calibration tubes, 1.27 mm~O.D., are placed on the top of the Fibre Routing Plate. There is a calibration tube every 5o, so that a calibration tube cross every tile. The megatile unit is held and compressed together by a set of 4.77 diameter screws and rivets (screw-nut). The top is a flathead 4-40 brass screw. The bottom currently is a commercial rivet (screw-nut). The rivets may need to be redesigned so that they need not be taped in, and snap in more easily by putting in a slight bevel. Each tile has two screws and rivets compressing the plastic sheets and pushing the Tyvek against the scintillator.
Table 6.2 shows the thickness materials in the scintillator tray in the order which they appear in the cross section. The 0.15 mm black polyester tape hold the fibres in place. It also wraps around the sides of the pan and adheres to the bottom of the pan. Hence in the table it is shown twice. The top and bottom plastic cover plates will have maximum thickness of 2 mm and 1 mm respectively. The scintillator manufacturer can supply scintillator with thickness variation of 10%. The manufacturer can supply black plastic cover planes with a thickness variation of as low as 5%. Therefore, we specify that the top plate should be 1.9±0.1 mm and that the bottom plate should be 0.95 ±0.05 mm. The scintillator provides the largest thickness variation of the package. The nominal thickness of the total tray for HB is 7.6 mm. Table 6.2 below shows a maximum thickness of 8.31 mm to fit in a 9 mm gap. Note that 0.25 mm aluminium venetian blind type springs that hold the tray against the outer copper plate need to be added.
| Thickness (mm) | Max Variation (mm) | |
| Polyester tape | 0.15 | 0.03 |
| Top plastic | 1.9 | 0.10 |
| Tedlar | 0.10 | 0.00 |
| Tyvek | 0.15 | 0.05 |
| Scintillator | 4.00 | 0.40 |
| Tyvek | 0.15 | 0.05 |
| Tedlar | 0.05 | 0.00 |
| Bottom plastic | 0.95 | 0.05 |
| Polyester tape | 0.15 | 0.03 |
| TOTAL | 7.6 | 0.71 |
| Venetian blind alum | 0.25 | 1.25 |
| Available gap | 9.00 |
To reduce the number of separated scintillator tiles and to give the scintillator trays more mechanical rigidity, individual scintillator tiles for a scintillator tray are glued together. The scintillator for a scintillator tray is composed of two or three subassemblies, called megatiles. Each megatile consists of many h tile divisions bonded together by thin channels of opaque white-reflective epoxy that also provides optical isolation, as shown in Fig. 6.18. The cross section of the ball groove holding the WLS fibre is shown in Fig. 6.19.
Table 6.3 below gives the radial position, length and width of
scintillator trays. The tolerances on the sizes of both, copper
slots and scintillator trays were included, when defining the
size of each scintillator tray. The wedges are 20 in j.
The j boundary of the wedge is 1 mm
away from the nominal 0 and 20 edge. The j
boundary of the wedge is covered by a 0.5 mm thick metal sheet.
The j edges of the trays are 1 mm away
from inner edges of the copper slots. Hence, the outer edge of
side scintillator trays are 2.5 mm away from the nominal 0 or
20 boundary. The inner edge of the side trays and the edges of
the middle trays are 1 mm away from the nominal 5 and 15 boundary.
The centre of the middle tray is in the nomial j
location. The q = 90 degree line is
1 mm away from the edge of the wedge. The scintillator tray sits
1 mm inside the wedge at 90 line and 3 mm inside the wedge at
the 53 line. Hence, each scintillator tray has at least 1 mm of
nominal gap on all its edges.
The light (collected by the Kuraray Y11(250 ppm) WLS fibre) is
transported to the HPD using a series of clear fibres. Within
the scintillator tray, the fibres are Kurarays 0.94 mm multiclad,
non-S type fibres. The tip of the WLS fibre inside of tiles has
been polished, aluminised, and protected with a thin polymer coating.
The other end is spliced to a clear non-S-type multiclad Kuraray
fibre. This splice is a heat fusion splice and is covered by a
2.54 cm long, clear plastic ferrule (FEP shrink tubing, 0.05"
OD). The light transmission across this splice is 92% with a 2%
rms. These optical fibres are routed from the tiles to optical
connectors at the edge of the megatile via grooves in the black
Fibre Routing plastic Plates. This groove is 1.6 mm deep. These
grooves and fibres are covered with black polyester tape to secure
and to protect the fibres. The grooving of the black plastic sheet
is done with the Thermwood x-y milling table. The WLS-to-clear
splice is kept in a straight section of the Fibre Routing Plate.
At the edge of the scintillator tray, the clear fibres from the tiles are terminated optical connectors. Fig. 6. 20 shows the design of the optical connector. The fibres and optical connectors for the scintillator tray are constructed as one unit and tested before installation of the fibres into the tiles. These fibre-connector units are called pigtails. A pigtail consists of the WLS fibre which goes in the tile spliced to a clear fibre. A pigtail contains all the fibres for 5 degrees of a scintillator tray. The clear fibres of a pigtail are sandwiched by kapton tape make the pigtail easier to handle. The connectors are held on the outer edge of the black plastic cover plates. Fig. 6.21 shows a view of the layout of cables in the HB/HE gap.
Optical cables in the form of a flat, light-tight bundle of fibres (0.94 mm) carry the light from the tray's connectors to a router box near the HPDs. The optical cables consist of 18 optical fibres sandwiched by a Tedlar cover. The number of holes used for tile readout varies from 14 to 17. The eighteenth spare hole is used for optical light injection. The Tedlar cover seals the optical cables from light. The cables have the same optical connectors at both ends. Within the router (fibre patch panel) box, optical signals are sorted from layers into calorimeter towers. This sorting is performed using 0.94 mm fibres which connect fibres in layer connectors to tower based photomultiplier tube light mixers. The default fibres for the cables and descrambler are Kuraray, multiclad S type fibres. The use of the optical connectors allows for the segmentation of the optical path into three parts, the scintillator trays, cable, and router box. The design provides maximum protection for the fibres at all stages of assembly and installation. In addition different parts of the optical system can be assembled at various institutions and assembled at a central location.6.5.1.2
To achieve good intra-tile transverse uniformity, the tile's surface reflectivity is kept uniform and a uniform response fibre groove pattern is used. This avoids the use of a complex optical mask on the surface of the tile. To keep the tile's surface reflectivity uniform, its reflective wrapping material must be held evenly against its surface. Two rivets per tile performs that function. The transverse uniformity of the tiles is "tuned" by adjusting the depth of the fibre groove inside the tile. The optimal groove depth for 4 mm thick tiles is 1.5 mm. At this depth, the uniformity is fairly insensitive to the placement of the fibre groove relative to a tile's edge. The uniformity of individual tiles is expected to be better than 4%.
The light yield of tiles within a projective tower must also be kept reasonably uniform at the photomultiplier tube readout. The longitudinal front-to-back light yield variation must be kept below 20% for the variation not to effect the resolution of the tower. Several factors contribute to longitudinal light yield. Since towers are projective, the area of tiles in a tower increases from the front to the back. Larger tiles at the back have smaller light yields. We have built tiles which span the size of the tiles used in the CMS calorimeter and measured the light yield from these tiles. We have determined that the light is a function of L/A, where L is the length of WLS fibre in the tile and A is the tile area. Using this function we can predict the light from an individual tile. The clear fibres in the tray and the cables attenuate the light with attenuation length of approximately 7.5 m. The back tiles suffer less attenuation from clear fibre going from the connector at the edge of scintillator try to the HPD since length of the optical transmission cable decrease from the front to the back of the calorimeter. This effect partially compensates for the lower light yield (from L/A) of the larger back tiles.
Fig. 6. 22 shows an expected light yield (in number of photoelectrons per minimum ionising particle) of H2 readout section of Hadron Barrel calorimeter as a function of h tower number. The plot indicates that we the light yield exceeds the requirement of 20 pe/mip.
Fig. 6. 22: Expected light yield (in number of photoelectrons per minimum ionising particle) of H2 readout section of the Hadron Barrel calorimeter as a function of h tower number. 6.5.1.3
As stated earlier, variation in the radial position of the scintillator in the groove causes variation in the response of the tile. Hence, the scintillator tray should be pushed up against the upper edge of the copper slot. This will be done by a 5 cm piece of 0.3 mm brass venetian blind type section (acting as a spring) which is the length of the slot. The 5 cm cross section will be bent to form a venetian blind spring. The venetian blind spring will be slid in under the scintillator tray and push the tray up against the copper. The total force of the brass piece applies to the tray should be 2-3 times the weight of the pan. A middle pan weights about 20 kg. Hence, about 50 kg should be applied to the pans. As a middle pan will have two strips pushing on the pan each strip should apply about 25 kg.6.5.1.4
This scheme requires some additional R&D and at the present time is only in the conceptual design stage. One extra fibre in a connector is used for laser calibration. UV laser light is sent through this plastic fibre to the connector at scintillator tray. One question is whether this plastic section will change its attenuation to UV light with radiation. At the tile, the HB design (which is different from HE) is for a 1 mm plastic clad quartz fibre to be put into the connector after the connector has been machined with the pigtails. The fibre will routed in a straight line to the opposite end of the tray in a groove in the top cover. It will turn around and go in a straight line close to the middle of the tiles (parallel to the source tube). There will be two 1 cm holes drilled in the black cover plates, Tedlar and Tyvek so the quartz fibre can enter the tile. The quartz fibre goes into one hole, passes through a 2.5 cm long groove on the top of the scintillator, exits out of the second hole and continues to the next tile. In the region of the hole, the cladding will be stripped from the quartz fibre, and the fibre etched with acid or scratched to form a diffuse surface. The length which is etched determines the amount of light injected to the tile. The UV light will exit the quartz fibre into the tile, and convert to blue light. If a blue LED is chosen instead of a UV laser an additional 1 cm white circle of paint must be painted on the bottom of the tile to reflect more blue light towards the green fibres. The quartz fibre's end is cut at 45o and painted black to avoid reflection. This geometry brings the UV light to all the tiles approximately at the same time as a real particle from the interaction point.6.5.1.5
The source tubes are placed in the top of the 2 mm thick black plastic cover plate, in a groove nominally 1.52 mm deep. The source tube generally go at constant j over the centre of each tile, but must make an S-bend (as gently as possible) to exit the edge of the pan near the optical connector and aimed purely in the z-direction (Fig. 6. 5 and Fig. 6. 23). The source tube groove drops gently to the bottom of the top plastic layer at the pan edge, and flares slightly in phi to provide tolerance where the source tube is inserted into a special cone coupler for transition from a 1/8" OD low-friction acetyl plastic tube. The metal tube is locked in place with a 2-56 nylon-tipped set screw; the nylon tube is locked by a paraxial 3-48 screw whose threads intrude into the 1/8" socket.
Most megatile layers in the assembled calorimeter will be accessible only when the endcap is retracted. At that time all layers of each tower will be measured, in order to check the integrity and uniformity of each tower and transfer test beam module calibrations to the main calorimeter. The occasional access tubes will be brought up in the channel which brings up the fibres to an access panel near the coil.
Layers 0 and layer 9 will be permanently coupled, via the acetyl tubing, to source drivers installed in the 8 cm gap at the back of a few wedges. One source driver, with a 35-foot (10.6 m) source wire can access up to 6 wedges, for a total of 6 installed source drivers in HB. The plastic tubes will make a gentle bend in the r-z plane rising no more than 5 cm from the end surface of the copper, and running in the same protected channel as the optical cables. The tubes will then, at the back of each wedge, bend (with radius no tighter than 12 cm) into the j direction towards the source driver, which has a channel indexer to select one of 72 different tubes. There are 144 permanent tubes in each half barrel.
The compact source driver design using non-magnetic motors (most probably small commercially available air powered motors and piezoelectric air valves) is under development, and appears feasible at this time. The motors are claimed to have a long operating life and not to require a lubricated air stream. The extra-low-friction tubing and the NICOTEF-coated source wire have already been developed for the CDF endplug calorimeter upgrade. NICOTEF is a proprietary coating containing nickel-sulphide, ptfe, and phosphorus, with Rockwell hardness 40, chemically plated onto the source wire by the Nimet Corporation.
The HE calorimeter has a slightly different design of the optical system, as both the cost and machining capabilities are different in US and Russia. An HE scintillator tray covers 10_ in j. Each 10_ wedge is subdivided into 20 towers. Each scintillator tray contains tiles as separate units. This simplifies the production of megatiles. The rigidity of the structure is provided by spacers placed at the edge of the megatile. This is possible due to staggered structure of the absorber plates. The tiles and covers are attached by pins through the scintillators. The cover plates are made of duraluminum because it is cheaper than plastic plates and provides better rigidity. Both the front and back cover plates have the same thickness, 1 mm. Between one of the plates and the scintillator there is 1.5 mm wide gap, fixed by spacers, for the optical fibres paths.
Fig. 6. 24 and Fig. 6. 25 show the design of the tray. The tray layers are as follows: The first layer is 1 mm thick duraluminum cover plate, next a sheet of Tyvek reflective paper with holes cut in them for connecting screws. Brass spacers, 7 mm thick, are screwed along the edge of the duraluminum cover plate. The tiles are placed on the Tyvek paper and covered by another sheet of Tyvek with holes for screws, WLS fibres and quartz fibre reflectors. WLS fibres are inserted into scintillator grooves through the holes in the Tyvek and kept in the place by small spacers with holes. The edges of the scintillator plates are covered with reflective paint (BC 620). Insertion of Tyvek strips between the scintillators is an option we are considering. The optical connectors with glued fibres are fasten to the lower cover plate. The proposed QF light injection system is different from the HB system. Here the light injection fibre is fanned out to many small quartz fibres, separate one for each tile. In the gap between the scintillators (covered by Tyvek) and the upper cover plate is where the quartz fibres and stainless steel tubes are placed for the laser and the wire source calibrations respectively. To equalise the timing of the laser calibrations all fibres inside the tray have equal length. Therefore, reels with quartz fibres (delay lines) are also placed in this gap. The quartz fibres are glued to reflectors (to feed the UV light into the scintillators) which in turn are glued to the Tyvek. The upper cover plate has rivet nuts (1.5 mm thick) opposite the holes in the lower cover plate. The screws from the lower cover plate go between the scintillators and are screwed into the nuts. The surfaces of the cover plates are covered by resistive layer (galvanisation) to prevent chemical reaction with the brass plates. The trays are positioned on the absorber plates by screws on the outer surface of the HE.
Fig. 6. 24: Design of the scintillator tray for HE; cross sections and front view without upper plate.
The inner barrel calorimeter (HB) is only 5.12 interaction length (l) deep and hence will be unable to contain the higher energy showers completely. The outer calorimeter is designed specifically to sample the tails of hadronic showers, in particular for those showers which developed deep inside the calorimeter. This is necessary to improve upon the missing ET resolution and also to achieve the design resolution of 100%/E5%. Outer calorimeter can also be used in identifying and triggering on muons. In order to achieve these twin objectives, HO should have the capability to identify single muons. A light level of about 7 photo electrons (pe) due to 'single' muon may be necessary to identify muons at 3 s level with the HPDs. Taking into account the long term degradation of the scintillator light yield due to the hostile radiation environment in which the detector will be located, it may be necessary to start with about 15 pe from 'single' muon using the two HO layers combined. Due to the large size of the scintillator tiles for the same h-f segmentation as in the inner calorimeter and the limitation in the number of fibres that can be embedded in the tiles, we opted to use 1 cm thick BC408 scintillators for the two layers of outer calorimeters as a base line design. An additional 5 mm of space is needed for two plastic sheets on either side of the scintillators to anchor the individual tiles as well as to route the fibres and also to accommodate the stainless steel covers for the scintillators. Thus including the plastic and stainless steel covers, the thickness of the individual layers will be 15 mm.
Each of the five muon rings have 12-fold symmetry along f with each f sector covering 300 in f and 2.53 meters along z. One such 30 f sector of the muon ring maps onto 6 calorimeter towers along f. The ideal size of the tiles for the two outer calorimeter layers in the f direction can vary between 0.40 m to 0.448 m depending on the location of the tower within the sector. These sizes are listed in the first row of Table 6. 4. However for a more realistic estimate of the tile sizes, we need to take into account the dead zones occupied by the stainless steel support beams. The two end tiles of the front layer are shortened and one end tile of the back layer has to be split into two trays due to the location of a support beam. The resultant sizes of the tiles along f are given in row 2 of Table 6. 4.
There are additional constraints on the tile sizes along f due to the frame structure needed to hold the scintillators in place. As will be described later, several scintillator tiles will be packed together into a single mechanical unit called a scintillator tray. A single tray will be one f slice wide (5 in f) and will cover the entire span of a muon ring along z i.e. 2.53 m. These trays will be resting on sheet metal C-channels welded to the inner and outer faces of the return yoke YB1. These C-channels will be of .5 mm wall thickness. An additional space of about 4 mm on either sides of the C-channels may be required for smooth insertion of the trays. To make these additional space for support structure, the tile sizes along f direction will be constrained further. The final sizes of the tiles along f are given in row 3 of Table 6. 4.
| Layer 1 | Layer 2 | |
| Tile length (m)
(ideal case) | 0.406 0.400
0.400 0.406 0.419 | 0.434 0.428
0.428 0.434 0.448 |
| Tile length (m)
(constrained by the return yoke support structure) | 0.367 0.419 0.406
0.400 0.400 0.377 | 0.434 0.428 0.428
0.434 0.448 0.123 |
| Tile length (m)
(constrained by the return yoke and tray support structure) | 0.354 0.408 0.396
0.391 0.391 0.367 | 0.257 0.426 0.419
0.419 0.426 0.559 |
| Dead space along f (%)
# of tiles along f | 5.8
6 | 4.4
7 |
In the z direction we are constrained by the ring boundaries. The HO scintillators have to be terminated at these boundaries as the space between successive rings will be used as service path and will not be available for sampling. This will have the following constraints on the tile sizes along z.
- The tile size along z can vary from 34 mm to 661 mm.
- In layer 1, h tower # 4 will split between rings 0 & 1 (also 0 & -1) and h tower # 10 will split between rings 1 & 2 (also -1 & -2).
- In layer 2, h tower # 9 will split between rings 1 & 2 (also -1 & -2).
- h tower number 14 in layer 1 will be truncated at the outer edge of Ring 2 (and -2).
- h tower numbers 3, 4 and 13 in layer 2 will also be truncated at the edge of rings 0, 1 (-1) and 2 (-2) respectively. So these tiles will also be shorter along h.
Table 6. 5 summarises various tile sizes along z.
| Ring | Layer 1 | Layer 2 | |||||
| Tower | hmax | Size(m) | Tower | hmax | Size(m) | ||
| 0 | 1
2 3 4 | 0.087
0.175 0.262 0.274 | 0.400
0.402 0.408 0.058 | 1
2 3 | 0.087
0.175 0.262 | 0.427
0.431 0.410 | |
| 1 & -1 | 4
5 6 7 8 9 10 | 0.349
0.436 0.524 0.611 0.698 0.785 0.791 | 0.160
0.430 0.446 0.465 0.487 0.514 0.034 | 4
5 6 7 8 9 | 0.349
0.436 0.524 0.611 0.698 0.747 | 0.274
0.460 0.477 0.497 0.522 0.306 | |
| 2 & -2 | 10 11 12 13 14 | 0.873 0.960 1.047 1.134 1.171 | 0.390 0.578 0.617 0.661 0.290 | 9
10 11 12 13 | 0.785
0.873 0.960 1.047 1.116 | 0.124
0.582 0.619 0.660 0.551 | |
The radial position of the front faces of the two HO layers are 4.57 m and 4.89 m respectively. They are located on either side of the return yoke YB1. The radial thickness of each layer of outer calorimeter is only 15 mm. A "no go zone" of 5 mm separates the two HO layers from the either surface of the YB1. Similarly there is a 10 mm "no go zone" between inner face of layer 1 and outer face of MB1. A "no go zone" of 10 mm is also kept between the outer face of layer 2 and the inner face of MB2. The radial thickness of the HO layers should not exceed 15 mm allocated for them. It is however allowed to utilise the "no go" zone between the HO layers and the YB1 in order to fix the mechanical holding structure for the two HO layers.
Fig. 6. 26 show one 300 f sector for each of the two layers for the three rings with individual scintillator tile boundaries mapped onto it. It is clear that some of the tiles are very narrow as they were truncated due to location of supporting beams or due to inter-ring gaps.
We have decided to combine these narrow tiles with their nearest neighbour as single tiles as follows:
- All the 12 tiles of 5.8 cm width on either side of ring-0 for layer 1 will be combined with their neighbours in the z direction.
- 6 tiles of 3.4 cm width on the outer edge of ring 1 and -1 will similarly be combined with their neighbours in the z-direction.
- All the 6 tiles of 12.4 cm width on the inner edge of ring 2 and -2 will be combined with their neighbours in the z-direction.
- All the 12.3 cm wide tiles along the f edge of each 30 f sectors in layer 2 will be combined with their nearest neighbour along f in all five rings.
With these minor modifications and adjustments one requires 14 ( 2) different sizes of tiles for layer 1 and 13 ( 2) for layer 2 for each f slice. The total number of tiles required for the two layers is 3888. Table 6. 6 summarises the number of different tiles required.
| Layer 1 | Layer 2 | |
| # of tiles per sector
Total # of tiles
Max. tile size (mm
Min. tile size (mm | 168
2016
419
367 | 156
1872
571
267 |
Based upon these estimates the total requirements of scintillators for the outer calorimeter are summarised in Table 6. 7.
| Layer 1 | Layer 2 | |
| Position along R (m)
Width of a 300 f sector (m) Length of a f sector along Z (m) Area of a f sector (sq m) Total area of 5 x12 sectors (sq m) | 4.570
2.449 2.536 6.211 372.65 | 4.890
2.621 2.536 6.646 398.74 |
In order to simplify the installation of the two HO layers, several scintillator tiles will be packaged into a single mechanical unit called the scintillator tray. A scintillator tray will cover the entire length of a muon ring along z. In the f direction, it will only be one tile wide. Each tray will contain 4,5 or 6 tiles depending upon its locations. These tiles will be wrapped completely first with TYVEK paper (for light reflection) and then with Tedlar sheets (to stop light leakage). This whole package will be sandwiched between a 1 mm thick plastic sheet on one side and a 2 mm plastic sheet on the other side. The 1 mm plastic is used to anchor the tiles in their respective positions using 6 BA countersunk screws passing through the tiles and bolts embedded inside the plastic. The 2 mm plastic on the other side will have channels grooved into it to route the fibres from individual tiles to an optical connector located at the edge of the tray to access the scintillator light. Finally, this whole assembly of scintillators, plastics and fibres will be placed in an tray of length 2.536 m and height 13.8 mm made of stainless steel plates of 0.3 mm thickness. This tray will be covered with a stainless steel plate of 0.3 mm thickness along the whole length of 2.536m. The bottom stainless steel tray and the top stainless steel plate of this assembly will be anchored to the plastic-sheet-covered scintillator detector by countersunk 6 BA brass screws and special form of nuts. These nuts will have a cylindrical shaft with 6 BA threads which ends in a thin circular plate. This thin circular portion will be projecting outside the stainless steel plate facing the muon chamber and the corresponding head of the holding screw will be outside the steel plate facing the magnet return yoke. we will need 360 scintillator trays for each of the two layers.6.5.3.1
Scintillator trays for the outer barrel calorimeter (HOB) will
be one f slice wide (50
in f). However along z (h)
direction , they will cover the entire span of a muon ring i.e.
2.53 m. Mechanical design for these trays and their support structure
has already been discussed in detail in chapter 4. In brief, the
trays would be resting on sheet metal C-channels of 0.5 mm wall
thickness. An additional space of 4 mm on either side of the C-channels
is required to ensure smooth insertion of the trays and also to
account for the skin thickness of the trays. The actual separation
between two successive scintillator tiles in the f
direction will therefore be about 8 mm. There will be six
trays for a 300 f sector
, each ideally covering 50 in f.
Although all the trays will be of length 2.53 m along z, their
width along f will vary from one to
other. There will be 9 different sizes of trays each containing
either 4, 5 or 6 tiles along z. The actual inner dimensions and
the number of trays required for each size are given in Table 6. 8.
In total, there will be 720 trays corresponding to a total active
area of 733 m2 containing 3888 scintillator tiles of
different sizes.
| Tray width (m) | Area of each tray
in (m2) | Number of
trays | Total area
(m2) |
| Layer - 1 | |||
| 0.359 | 0.90 | 60 | 54 |
| 0.411 | 1.03 | 60 | 62 |
| 0.397 | 1.00 | 60 | 60 |
| 0.392 | 0.99 | 120 | 119 |
| 0.367 | 0.93 | 60 | 56 |
| Layer - 2 | |||
| 0.259 | 0.65 | 60 | 39 |
| 0.426 | 1.08 | 120 | 129 |
| 0.420 | 1.06 | 120 | 128 |
| 0.563 | 1.41 | 60 | 85 |
| Total | 720 | 733 |
A top view of a typical scintillator tray is shown in Fig. 6. 27 and its cross sectional view is shown in Fig. 6.28. The trays are similar in design to those used for the inner barrel calorimeter and will be packed and assembled using similar techniques. However, unlike the inner barrel, the HOB scintillators are 10 mm thick, thereby increasing the overall thickness of the tray.Table 6. 9 shows the thickness of various materials in the order in which they appear in the tray.
All the scintillator tiles in a tray will be made out of a single piece of scintillator by cutting a straight groove of 0.9 mm thickness and 9.5 mm depth between successive tiles. These grooves will be filled with an opaque, white epoxy to provide rigidity and optical isolation. At the bottom of these straight grooves is a bridge of 0.5 mm thick scintillator, which forms an enclosed area into which the epoxy will flow. In order to reduce the cross talk of light between the tiles , these bridges will be marked with a black marker on the outside. This technique is similar to that used for the inner barrel trays.
Groove design on a tile
Fig. 6. 29 shows the top view of a typical HOB scintillator tile. The light from the tile is read out using WLS optical fibres which are held inside the tile using circular grooves. The grooves are similar in design to that in an HB tile i.e. each with a circular part inside the scintillator of diameter of 1.35 mm and a neck of 0.86 mm width. The base line design is to use 0.94 mm double clad, non-S type, Y11 fibre of Kuraray. The HOB tiles are larger in size than the tiles in the inner barrel part of the calorimeter. As a result, they will give much less light if we use a single sigma groove running around the perimeter of the tile. The large length of the WLS fibre required for such a readout scheme will further reduce the light output as the attenuation length in WLS fibre is around 2m. In order to collect sufficient light from these large sized tiles, we plan to put 4 separate WLS fibres in a single tile in separate grooves. Each tile will be divided into 4 quarters. Each quarter will have a sigma groove as shown in Fig. 6. 29. Fig. 6. 30 shows a clasp of the sigma groove at the fibre insertion point. In a tile, the straight sides of the rectangular grooves are located at a distance of 2.5 mm from the edge as well as from the centre of the tile. The corners of the grooves are rounded and have 31.8 mm bend radius. This will prevent any damage to the fibre at the bend and also ease the process of fibre insertion in the narrow grooves. For smaller truncated tiles, instead of 4 fibres, we need only 2 fibres for equivalent light yield.
Fig. 6.31 shows the cross sectional view of an HOB tile for a better perspective of the groove design. It looks like a circular hole attached to a narrow neck. With such a design it will be possible to keep the fibre inside the groove provided the neck of the groove is narrower than the fibre diameter. Since the fibre diameter is 0.94 mm, the neck diameter is kept at 0.86 mm. In order to insert the fibre in the groove, one end will have a rounded shape of size 1.8 mm which will narrow down to 0.86 mm (neck size) within a distance of 9 mm.
The light collected by the WLS fibres inserted in the tiles will be transported to photo detectors located outside the muon rings using clear fibres. The captive end of the WLS fibres located inside the grooves will be polished, aluminised and will be protected with a thin polymer coating. The other end of the WLS fibre will come out of the tile through a slot (3 mm X 25 mm) made on the 2 mm thick black plastic cover sheet. This end of the WLS fibre will be spliced to a clear non-S type multiclad Kuraray fibre. The clear fibre will then be routed along the Z direction through 1.6 mm deep guiding grooves made on the outer side of the 2 mm plastic sheet to an optical connector located at the edge of the tray. Each tray will have two optical connectors mounted on either side of the tray. All the fibres from a tile will terminate on the connector located nearest to it. Since a tray will have a maximum of 6 tiles, there will be at the most 24 fibres per tray, 12 fibres per connector.6.5.3.2
The HOB calorimeter is designed specifically to sample the tails
of hadronic showers, specially for those which develop deep inside
the detector. This will enhance the energy
resolution of HCAL. In addition to this, it is expected to have the capability to identify and trigger on muons. A yield of 4 pe/layer is necessary to identify muons at 3 sigma level if HPD is used as the light readout element. Another consideration is to have uniform response within the tile. Both the requirements are fulfilled by
a) choosing higher scintillator thickness (10 mm) so as to have higher light output (higher photo electron yield)
b) 4 groove pattern for large tiles as shown in Fig. 6. 29 and 2 groove pattern for small tiles so as to have uniform response within the tile.
Light output from two layers of the same tower is pooled together at the decoder box and recorded by one readout element. Different response in the light output of these layers would affect energy resolution of the system. Responses of these layers are kept the same within 7 % approximately, using the techniques described in chapter 6.5.1.6.5.3.3
Since HOB will be placed in a region where the magnetic field will be negligible, there will be no need to compensate for any field effects.6.5.3.4
Though further R & D is required for this system, a spare hole would be provided on the optical connector to accommodate a quartz-fibre. Also, an additional groove would be made on the 2 mm black plastic as described in chapter 6.5.1.6.5.3.5
A radioactive source will move along the z-direction and scan the central portion of every tile on the tray. Data collected during this scan will be used for calibrating the tiles. A tube for the passage of the source will be laid on a groove on the 2 mm plastic sheet by the side of the fibre routing groove. There will be one source tube per tray.
The HOE scintillator covers the pseudorapidity region of =1.2 to =1.5. It is meant to provide coverage for 7 (towers in the gap region between the HCAL Barrel, HB and the HCAL endcap, HE. Part of the HOE will be attached to the ME/1/2 muon chambers and be installed with the muon chambers while the rest will be attached directly to the YE/1 iron. The layout of HOE is given in Fig. 6.32.
The portion of the HOE scintillator that will be installed on the back faces of ME/1/2 (inside the muon chamber frames) covers 5.5 towers.
Due to the overlapping coverage of the ME/2, ME/-2 chambers, the same coverage is afforded to the HOE tiles in this region; nearly 100 percent. The muon chambers will provide the necessary structural integrity and support for the HOE tiles. An entire HOE package which covers the back of a muon chamber will be referred to as a megatile. Each megatile in this region will cover 10 degrees in f or two f towers.
In this scheme, the clear fibre that is spliced to the green Y-11 WLS fibre in the tile will carry the signal to the HPDs.
The construction of these megatiles will be the same as the construction of the HCAL barrel megatiles but the scintillator will be 10 mm thick BC 408 for greater light collection. An additional 5 mm of packaging will complete the megatile to give it an overall thickness of 15 mm, similar to the HOB tiles.
As the megatiles will be effectively trapped by the muon chambers, access will be limited to lengthy shutdowns and scheduled muon chamber maintenance periods.
In order to provide the remaining 1.5 (tower coverage over the "trough" of the HCAL barrel - endcap gap, additional tiles will be placed in the region between ME/1/2 and ME/1/3 where the "z-stops" for the endcap detector reside. These tiles will be fixed to the YE/1 iron before the muon chambers are in place. Due to the congested nature of this space, the tiles will only cover approximately 80 percent of the area. These additional tiles will be attached to the muon iron in a method similar to that of the HOB tiles.
Thin steel I-beams will be tack welded to the YE/1 iron which will hold the tiles in place. The supports and tiles must be placed on the YE/1 before the ME/1/3 muon chambers are installed. The reason for this is that the endcap cables go over this region of HOE tiles and under the ME/1/3 muon chambers. The complete effect of these cable paths has not yet been determined, but regardless of the layout, there will be some additional loss of coverage.
All the tiles in this region will be trapped by cabling from the inner detectors, so access will be limited at best. The HPDs will be accessible during shut down periods when the endcaps will be retracted.
Each tower in HOE will have 4 optical fibres going to a single pixel in a multi-pixel HPD which resides in the photodetector box of the HE Endcap. There are 4 fibres per tower for a total of 4032 fibres. There are 504 towers per end, 1008 towers total, with one readout per tower for a total of 1008 readouts.
Fibre insertion into the absorber matrix is one of the critical parts of HF construction. The past experience in constructing the EM module suggests that about two man-years would be required for manual fibre insertion per detector. Fibre bundling for each tower and bundle installation behind the absorber would certainly require additional time. A programmable semi-automatic fibre insertion tooling is presently contemplated for this repetitive task. A robotic arm would pick an already cleaned fibre and insert it into a groove by the required length and move by groove-to-groove spacing to repeat the same task until a tower is completely finished. A manual intervention would be required if an insertion or another problem is encountered. The engineering aspects of this design are currently under study.
Fibre insertion will be done in a clean tent where a positive air pressure will be present to avoid dust particles accumulating in the grooves and on the fibre bundles over time.
One of the requirements for the absorber matrix is such that all the grooves are tested with a steel wire gauge for clear passage before the insertion of the quartz fibres. After this test, dry pressurised air is blown into the grooves to clean out possible burrs and dust particles. This procedure will reduce time consuming interventions if a robotic arm is used.
Once a tower is completed, a visual inspection from the far end (front face of HF) of the calorimeter will be conducted to make sure the uniformity of fibre insertion lengths. A light source will be used to inject light into each fibre from the same end to make sure that there was no fibre breakage during fibre installation. The acceptable rate of failure is one fibre in thousand.
The fibres will be bundled to form towers at the back of the absorber. The fibre bundles will be made to form thin ribbons in order to minimise optical pickup noise from background radiation. At the very end of the bundle, fibres will be closely packed into cylindrical ferrules for mechanical mounting into the photodetector housing.
There are three types of fibre bundles that emerge from each tower; long fibres that run the entire length of the absorber (EM section) will be bundled separately from the medium length fibres (HAD section). The short fibres (TC section) will form yet another bundle. There are two different sizes of towers, the smaller ones (5 cm by 5 cm) and the larger ones (10 cm by 10 cm) depending on the eta region. For TC, superimposed towers will be formed in 20 cm by 20 cm square sections.
EM fibres will alternate with HAD fibres in the absorber, i.e. every other fibre will go either to EM bundle or HAD bundle. TC fibres will be inserted 30 cm into the absorber in the same groove as the EM fibres but at every second groove.
The experience with the previous prototypes has provided information
in distinct ways; the fibres that constitute a tower are bundled
at one end first and the free ends are inserted into the absorber
as in the case of EM prototype that was built in 1996. During
various stages of prototype construction, fibre bundles were glued,
cut and polished several times and this experience proved extremely
valuable. A new type of a wire saw that is recently introduced
into the market makes cutting and polishing fibre bundles easier
as shown in Fig. 6.33. The samples that are cut in this fashion
require less time and effort to polish the bundle adequately.
Fig. 6. 33: The top photograph illustrates the surface quality of the fibre ends after they have been cut with a diamond saw. Note that the core, cladding and the buffer are clearly visible. The bottom photograph the fibre ends after they have been polished.
Once a crack-free cut is established with the ferrule, wet polishing procedure starts with 33 micron grit silicon carbide. This is followed by a set of finer aluminium oxide polishing powders; i.e. 18, 12 and 3 micron grit sizes. The final polish is done using a 2 micron cerium oxide powder until the fibre ends are clearly reflective.
Before the ferrules are mounted into the holding grid, the polish will be visually inspected under magnification. In case of unacceptable polish (cracks, scratches, glue smears, etc.), the polishing procedure is repeated until an acceptable result is accomplished. After polishing, the bundle will be cleaned with a cleaning agent to remove small particles and dust and a protective cover will be placed over the ferrule. The failure rate of 1 cracked or deeply scratched fibre per hundred in a bundle is acceptable.
Each megatile sub-assembly is constructed from a single plate of scintillator. The processing of the scintillator plates is done over several steps. First, the protective paper removed from top side of the plate. The thickness of the plate is measured at two points at the edges using a micrometer. All machining operations are done on this top side only. The protective paper is left on the bottom scintillator until just before the scintillator pan is assembled. Since the separation grooves are cut so that only 0.25 mm of material remain, the scintillator plate with separation grooves machined onto it is fragile. However, as the protective paper on the other side is left intact on the bottom side, it gives additional structural support to the plate when it is handled. In addition, the paper prevents surface scratching when the scintillator is moved about.
The plate is positioned on the Thermwood x-y milling table. Next, reference holes are drilled along the edges for realignment in later operations. A "long reach" 0.90 mm end mill is then used to cut the tower separation grooves 3.75 mm into the scintillator (the last 0.25 mm is left uncut). The outer boundary of the megatile is not cut, just the inner tile separation grooves. The fibre grooves are not milled along with the tile separation grooves because of a risk of epoxy seeping into the fibre grooves from the separation grooves in the epoxying operation. This is because the fibre grooves are 3 mm away from the separation grooves, and a tape seal in such a small gap is not robust enough.
The scintillator plate is then removed from the milling table and white, opaque, epoxy is injected into the separation grooves. This is done by first taping over the grooves and then injecting epoxy into the channels. The epoxy is cured at room temperature for one day. The scintillator plate is put in a oven which is at 38 C for one day to harden the epoxy. The epoxy provides optical separation of the tiles, mechanical support, and a reflective surface at the tile edges. The plate is taken back to the milling machine and the megatiles are re-registered to the milling machine's co-ordinate system. The milling machine cuts the fibre groove routing, machines the rivet holes, and cuts the edges of megatile from the scintillator plate. The 1.52 mm deep fibre grooves are routed with a 1.35 mm end mill. The groove's circular shape, Fig. 6. 19, is cut with a 1.14 mm ball mill. Since the neck of the groove is smaller than the diameter of the fibre, the fibre is trapped in the groove.
The megatile is removed from the milling table. A fibre is inserted into each groove to insure that each fibre groove is clear. The megatile edges are painted with white TiO2 paint to provide a reflective surface on the outside edges of the megatile. In addition, the side of the separation groove with the leftover 0.25 mm of scintillator is "painted" with a black marker pen. This reduces the adjacent tile-to-tile crosstalk to an acceptable 1 ±.6% per side. Fig. 6. 18 shows the mechanical configuration of the separation groove. The construction produces a large megatile that contains individual tiles which are optically isolated but are mechanically one unit.
Each scintillator tray consists of two or three scintillator megatile sub-assemblies. The largest piece of scintillator that manufacturer can supply is 2 m by 1.1 m. In order to minimise the total amount of scintillator used, the tiles are cut from the scintillator in the following way for layers 1-16. The scintillator plates are 2m long. We start with layer 1 tower 1 on the edge of a plate. Next the Thermwood goes to layer 1 tower 2. It continues sequentially until the next tile does not have enough space to fit on the plate. The next tower is on a new plate. At the end of the layer, the Thermwood goes to tower 1 of the next layer. If this tower fits on the old piece of scintillator, it puts it on the old piece of scintillator. Otherwise, it goes to a new plate of scintillator. The process proceeds to the last tower of layer 16. In this procedure, megatile sub-assemblies from separate layers can be cut from the same piece of scintillator. This procedure reduces the amount of scintillator we need to buy and reduces the cost. The thickness for layer 0 and 17 are different from the rest. Hence, a separate size plate is devoted to layers 0 and layer 17. The scintillator for 20o will be cut from a single plate. Each plate will have two side megatiles and one middle megatile cut from it.
The top and bottom black plastic plates are grooved on the Thermwood milling machine. The code for the milling machine is prepared from the database. Reference holes are drilled along the edges for realignment in later operations. Next the Thermwood cuts the grooves and holes on the black plastic. The Thermwood can cut plastic which is as long as 3.5 m. However, the longest scintillator tray is 4.5 m. Hence, after the cutting operation is over the plastic is repositioned on the milling machine using the reference holes. The rest of the plastic is machined to form a 4.5 m long single piece.
The milling of black polystyrene plastic cover plates are independent of the scintillator machining. However, the black plastic cover plates must be produced at the same pace at which the scintillator megatiles are cut. The scintillator, plastic (polystyrene), white paperlike reflector (Tyvek), light tightening black wrapping (Tedlar), and top and bottom black cover plates (black polystyrene) are assembled into a partially finished megatile-tray unit. This is the megatile-tray pre-assembly step. The fibre insertion is done later for the final finished scintillator tray pans.
Table 6. 10 shows the material needed to produce one wedge and the full calorimeter. 2052 scintillator trays are needed for the HB calorimeter. Note that before 36 wedges for HB are constructed, a single wedge pre-production prototype will be built.
| Material | 1 wedge | 36 wedges |
| Scintillator, SCSN81, 4 mm (m2) | 73.5 | 2683 |
| Scintillator, SCSN81, 9 mm (m2) (layer 17) | 5.5 | 164 |
| Scintillator, SCSN81, 9 mm (m2) (layer 0) | 4.2 | 110 |
| Black plastic, 2.0 mm(m2) | 67.7 | 2170.0 |
| Black plastic, 1.0 mm(m2) | 67.7 | 2170.0 |
| Reflective paper: Tyvek 0.15 mm(m2) | 180. | 4340.0 |
| Black wrapping: Tedlar, 0.04 mm(m2) | 210. | 4882.5 |
| Y11 WLS multiclad fibre, 0.94 mm(km) | 1.5 | 53.9 |
| Clear multiclad fibre, 0.94 mm(km) | 2.5 | 91.1 |
| Clear 18 fibre cable, 0.90mm(1.2x15mm)(m) | 53. | 1900.0 |
| Source calibration tubes, SS(0.050"OD)(km) | 0.3 | 10.7 |
| Epoxy: TiO2 loaded resin(kg) | 15. | 560 |
| Rivets | 2400 | 86400 |
| Polyester tape, .15 mm(km) | 0.5 | 16 |
The scintillator thickness of layer 0 is 9 mm. The scintillator thickness of layer 1-16 is 4 mm. The thickness of layer 17 is 9 mm. Since, the width of the scintillator pans is the smallest for the innermost layer and largest for the outermost layer, we can reduce cost by ordering plates with different width. We have decided to order scintillator plates of 3 different width for layers 1-16. Table 6. 11 lists the sizes of the plates and the number we need to order.
| scintillator | scintillator | scintillator | 1 wedge | barrel |
| thickness(mm) | length(mm) | width (mm) | total | |
| 4 | 2000 | 1085 | 13 | 475 |
| 4 | 2000 | 930 | 14 | 514 |
| 4 | 2000 | 800 | 12 | 435 |
| 9 | 1600 | 860 | 4 | 119 |
| 9 | 2000 | 692 | 3 | 79 |
The scintillator and plastic cover plates will be cut in Lab 8 at FNAL. The time to cut the scintillator and plastic cover plates for HB are based on the amount of time it took CDF to cut the scintillator and plastic cover plates for the CDF Hadron Plug Upgrade. We start by trying to estimate the amount of time CDF took to complete its Thermwood machining. These time estimates include both cut time and set-up time. The cut times were estimated form the total length of cuts and the cut speeds that were used. From the comparison of the CDF estimate and the actual CDF times we get a fudge factor. The fudge factor is an estimate of how much our time estimates are off. Next using the same methods, we estimate how much time it will take to cut the CMS plastic cover plates and scintillator. We have measured the machining speeds for the CMS grooves. We then multiply the CMS time by the fudge factor to get the CMS production time. The total times should be fairly accurate.
The actual time CDF spent was 1 calendar year. That calendar year consisted of 2 consecutive Thermwood shifts , 16 hours, on one Thermwood machine. The actual cut time available was 12.5 hours. The 3.5 hour overhead is due to turning the machine on, turning the machine off, cleaning the machines, and preparing the machines for the next day. Lab 8 has two Thermwood machines, and the technician had to prepare both machines for the next day. The CMS estimates will assume that we get the same machine time/day as CDF did.
We estimate that it should have taken CDF 0.58 years to cut the scintillator, 0.22 years to cut the plastic cover plates , and 0.13 years for set-up for each layer. We are off in our CDF estimate by 7%, as it took CDF one year of cutting time. Hence ,we multiply our CMS time calculations by 1.07. For CMS, we obtain the cut times of 1.44 years for the scintillator, 0.54 years for the top plastic cover plates and 0.34 years for the bottom plastic cover plates. We estimate 0.22 years for set-up time for each layer, for a total of 2.55 years of production cutting.
Next we estimate start-up times for the cutting. It took CDF 0.5 years from the time it tried to start cutting to full production. We give the same estimate, but we break it up into two components. We estimate 0.25 years of start-up for the preproduction wedge and .25 years for start-up for the full production. Hence, total barrel cutting time will be 2.80 years.
For the preproduction prototype we assume 0.25 years for start-up time. We also assume 0.22 years for the set-up time and transition between layers. These were the same times which we used to calculate the total production time. For the cutting time we take the total cut times and divide by 36. This gives us a total time of .55 years for the preproduction prototype.
Scintillator tray production cannot go faster than the Thermwood time, but with enough people hired it should go at the same rate. The assembly steps need to assemble the CMS megatiles are very similar to CDF. CDF needed about 3 people to do the assembly up to the fibre assembly. Hence, we estimate the 3 people are needed to do the preassembly up to the fibres. In addition we estimate this part of the project needs at least one full-time production manager.6.7.1.1.1
An important quality control item in this step of the production is tracking the quality and uniformity of the scintillator plates used for each megatile. As part of the SCSN81 scintillator purchase agreement for CDF Upgrade project, Kuraray marked each plate delivered with an ID number that specified the production history of the plate. This information is included in the manufacturing protocol for the megatiles. Plates from different production batches are also tested and the results included in the protocol. Both the attenuation length and the light yield are monitored using tests described in the scintillator specification document. With this protocol, the megatile performance having to do with scintillator quality can easily be monitored as the megatile production proceeds.
Several checks of the scintillator plates are done. From each plate a 1 cm by 1 cm piece of scintillator is cut out of the sheet. This piece will be stored with the information as to which scintillator tray it is associated with. If a problem is found with the scintillator tray this piece can be retrieved and measured with a bismuth source. A fraction of the 1 cm by 1 cm pieces will be measured to determine the overall scintillator quality. The thickness of the plates will be measured in three places. Any plate with a the thickness outside the tolerance will be rejected. However, it is very important that the thickness of the plates not exceed 4.5 mm. To ensure they do not exceed 4.5 mm, a gauge with a 4.5 mm gap will be made. This gauge will be run down the edges of the plates. Any plate with a point that is thicker than 4.5 mm will be rejected.
After the scintillator is machined, a fibre is passes through each groove to insure each groove is clear. From knowledge gained from manufacturing the megatiles for CDF, we know that the dominate variation in the response of the scintillator trays is due to the variation of the fibre groove of the scintillator. The cutting of the scintillator gives a 5.5% rms for the light output and the distribution is gaussian with no tails. There are no manufacturing problems from cutting which give rise to low light yield tails. This information was determined by manufacturing 20000 tiles in 594 scintillator trays for CDF and measuring them. Hence, a visual inspection of the scintillator is sufficient to determine the quality of the megatile. The visual inspection determines the following: The scintillator is clean with no scratches, the edges are painted white, the fibre groove is clear, and the epoxy fills the separation groove.
The difference in grooves between low light yield tiles and high light yield tiles cannot by determined by inspecting the grooves. It can only measured when the trays is completely assembled with fibres. Hence, a sample of the preassembled pans will be stuffed immediately with fibres and measured. Production schedules will dictate whether all the pans can immediately stuffed with fibres.
The preassembly of the scintillator pan is done very quickly after that parts are available. This preassembly checks the whether the plastic is milled correctly.
Each scintillator tray will contain a traveller. The traveller will contain information about the tray at each step in production. It will contain information about the scintillator pieces used in the tray. Using the data from Kuraray, we can reconstruct information such as the thickness of the scintillator, the batch of the fluors, etc.6.7.1.2
The preassembly of the scintillator pan done very quickly after the scintillator megatile sub-assemblies are prepared. This protects the megatiles from damage. First the rivets are snapped into the rivet holes in the bottom plastic plate. The plastic is laid on a table and the sheet of Tedlar is put down over the plastic. The Tedlar has holes in the position where the rivets are. Next the Tyvek is put down over the Tedlar. The scintillator is put down over the Tyvek and positioned with the rivets. a piece of Tyvek is placed over the scintillator. The bottom sheet of Tedlar has been cut big enough so that it can fold around the edge of the scintillator. The sheet on one side is taped onto the top sheet of Tyvek. The sheet on the other side is folded around the scintillator and is taped onto the Tedlar. This forms a light seal for the scintillator. Next the top black plastic cover is put on the Tedlar. 4-40 flat head screws are put through the holes in the black plastic and are screwed into the rivets. A electric screwdriver with controlled torque is used to tighten the screws. The optical fibres are not part of this assembly. After this assembly, the scintillator is well protected and can be safely stored or shipped anywhere. The preassembly trays are stored in boxes awaiting the fibres.6.7.1.2.1
Quality control
Quality Control of the preassembled tray is visual. We look at them to determine that all the tasks are performed. A traveller is checked off for the steps for the preassembly. The thickness of the scintillator pan must not exceed 8 mm. A special gauge will be made which has a air gap of 8 mm. The tray will be measured with the gauge along all edges to ensure the tray thickness does not exceed 8 mm. A sample of the preassembled trays have fibre installed and they are measured them with the scanner after the production.
The technology of production of the tiles was developed and tested for more then thousand tiles by SDC and CDF groups. New results were obtained during R&D at CERN by CMS collaboration. All the information and experience which we have give us confidence that the tile will satisfy the CMS requirements.6.7.1.3.1
The WLS fibres are cut to length. A template will enable to fibres to be cut to the correct length. Next both ends of the fibres are polished. Next the one end of the fibre is mirrored. The clear fibres are cut using a template to a length two inches longer than the length the fibre will be in the connector. The fibres are spliced together with an automated fusion splicer. The set of 17 polystyrene fibres + one quartz fibre are then assembled into a connector.
The pigtails are made using a plastic template with fibre grooves set to the correct length of the fibre run in the actual megatile. At one end of the template, a connector is
secured onto the template. The template for the side pigtail has location for the connector offset from the fibres. Fig. 6. 21 shows the curve the pigtail must make at the connector. First, each spliced WLS+clear fibre for a specific tower is inserted into its hole in the connector insert and then laid into its corresponding groove on the template. As there are tick marks in the template at the location of the splice for each fibre groove, it is clear if a fibre from a wrong tower is laid into a groove. After the fibres are in place, kapton tape is put on the bottom and top of the pigtail. This retains the shape of the pigtail and keep the pigtail flat. For the pigtails for the side trays another operation must be done to put in the curve near the connector. The kapton tape to hold the fibres is put on up to 30 cm (6 inches) away from the connector. A total of 15 cm of fibre away from the connector is left free of tape. The fibres are inserted in to connector. The connector and fibres are rotated to put in the bend that the pigtail fibres need for the side trays. The connector is then inserted into its position on the template and the fibres near the connector are taped in place. Hence, the fibres will easily follow the pattern they must follow in the scintillator tray. The fibres are secured with kapton to the connector and then glued. After the glue cures, the connector insert is faced off with a diamond cutter to form a clean, uniform optical surface. The combination of the fibre fusion splice, the WLS fibre mirroring, and the optical connector produce a light transmission rms spread of 3.5%.
We assume that the same steps that were needed for CDF to assemble the pigtails will be used by CMS. In order to estimate the assembly time for the pigtails we have used assembly times taken from CDF. The steps needed for building the fibres are the following: engraving the connectors, cutting the fibres, polishing the fibres, coating the mirrors with epoxy, splicing, cutting the protective tube used on the fibres, laying out the pigtail, gluing the pigtail, polishing the connector after gluing and testing the pigtail. Since the pigtails are bigger for CMS we estimate that it will take 20% longer to do the following steps: laying out the pigtail, gluing the pigtail, splicing the fibres, and testing the pigtail. We assume that it takes 10% longer to polish the pigtail. With the above estimates , we estimates it takes 4.9 man hours to make a pigtail for 5 degrees. This gives 350 hours needed to make the pigtails for a wedge. With 7 man hours in a day, this amounts to 50 days. For the entire barrel, it takes 12600 man hours. With 240 days in a year, this amounts to 7.5 man years of production
It took CDF about 6 months of start-up time for the pigtail production. During this period, one full time CDF physicist and 2 technicians worked on fibre R & D. We estimate that it will takes 0.30 years of start-up for the preproduction prototype and 0.30 years for the barrel.
For the preproduction prototype we estimate we need 1 production manager and 2 technicians. The time is 0.3 years of start-up + 0.2 year = 0.5 years of production. With 4 technicians and a production manager, it will take 0.3 of start-up + 1.9 years = 2.2 years to produce the pigtails for CMS.6.7.1.3.2
Unlike the scintillator, low light tails in the tile response
can come from bad fibres. Almost all of the problems in the tile/fibre
assembly occur from bad splices or WLS fibre problems. Therefore,
fibres are checked and tested prior to use. Several fibres from
every batch of Kuraray are visually inspected for defects. If
defects are found in these fibres then all the fibres of the batch
are inspected. For WLS fibres, a sample is scanned with the UV
scanner to assure that the light yield and attenuation length
are within specifications. The fibres for each tower (fixed length
green) are cut to length. The WLS fibres are cut in bulk,
polished, and mirrored on one end. Each batch of mirrored WLS fibres has several control fibres which are checked to assure uniformity in the mirroring. The reflectivity of the mirror is measured by measuring the mirrored fibre with the UV scanner. Next, for a small sample of test fibres, the mirror is cut off, and the fibre is remeasured. From these two scans the reflectivity is calculated. We expect a reflectivity of 0.85 ± 1.5%.
Each week the splicing quality is checked. This is done by taking a WLS fibre, cutting it in the middle of the fibre, and splicing it there. Next this fibre is scanned with the UV scanner. By measuring difference in light output across the splice the transmission across the splice is measured. The transmission across the splice expected to be 92% with an RMS of 1.8 %. In this measurement, the cladding light is removed by putting black tape on the fibre cladding before photodetector.
After the WLS and clear fibres are spliced and assembled into fibre-connector assemblies (pigtails), all fibres for all pigtails tested. They are tested in an automated UV-scanner box that is controlled by a PC. Those that are out of specifications are either reworked, or rejected outright. The results of these pigtail scans are saved in a data base for future reference.6.7.1.3.3
The main quality assurance tool is the UV fibre scanner. The light yield of all the fibres in a layer is compared and all bad fibres are replaced.6.7.1.4
The source tubes in the megatiles will be 18 gauge thin wall stainless steel hypodermic tubing, needle-grade fully hardened. The nominal OD is 1.27 mm to 1.32 and the ID is 0.965 mm. The source-carrying "wire" is 22 gauge stainless steel hypodermic tubing, 0.71 mm OD, with a bullet-shaped enlargement of approx. 0.833 diameter closing the active end of the tubing. As discussed elsewhere, the source wire is given a NICOTEF antifriction coating. All this is done before the active element is loaded and a keeper wire inserted, followed by closure of the inactive end of the source "wire".
The metal source tubes will be cleanly finished at one end, probably by EDM cutting. Each tube is crimped, or crimp-cut to length, and laid into the black plastic groove. The clean-cut end is secured in the pan-edge coupler with a nylon-tipped 2-56 set screw. The tubing is then taped in place with 0.1 mm thick polyester backed clear tape. The depth of the groove in the black plastic is nominally identical to the OD of the metal source tubing. The width of the groove is at least 0.065 mm to provide tolerance (especially against kinking) and the groove will flare near the edge of the pan to provide tolerance going into the pan-edge tube coupler.
The tube should end near the edge of the last tile (at h=0),
but should end approximately 1 cm short of the end of the groove
in the plastic, for tolerance and to accommodate some degree of
thermal contraction of the plastic (the coefficient of thermal
expansion of plastic is some ten times that of steel). The pans
should be protected from large thermal excursions at all times.
6.7.1.4.1
After cutting and before closure of one end, each metal tube will be flushed and/or blown-out, and probed with a 0.89 mm diameter wire to guarantee clearance for the source wire.
The tube must be securely fastened to the coupler. Tightening of the nylon-tipped set screw must be done firmly but not excessively, or the tube will be distorted or dimpled.
Test: The coupled tube must resist a pull of at least 3 kg-weight before being laid into the groove. If it fails, the set screw will be backed off and the tube recoupled more tightly.
Test: The tube and coupler must also still freely pass the 0.89 mm probe wire. Any tube failing this probe will be either repaired or replaced. Satisfactory repair can be made by carefully forcing an approximately full-diameter probe into the tube. The person doing the coupling will do the tests immediately. This will provide rapid feedback. The probe test will be repeated by another person, as a cross check.
It is important that the distance of the tube to the scintillator, and the degree of embedding of the tube in the plastic, not change with time, or the accuracy of the collimated source to wire source ratio will be degraded. We will rely on QC/QA of the depth of the groove in the plastic, and the tight manufacturing tolerance of the steel tubing. We believe that taping the tube into a nominally same-depth groove will locate it adequately. We will rely on QC/QA of the megatile assembly and riveting procedure. We will rely on the support-springs which locate the megatiles to the back of the copper slot, to help keep the megatile packages in a compressed state.
The coupling of the plastic tubes to the pan edge and to the indexer of the wire source driver will be checked by a 3 kg weight pull test.6.7.1.4.2
Travellers will accompany batches of tubes, and will have checkout lists detailing procedural steps, including probing and testing. Similarly, tube installation and the testing thereof will be made part of the travellers which accompany megatiles.
The plastic tubes connecting pan source tubes to the source driver indexers will be colour coded at both ends, to help prevent scrambled couplings.6.7.1.5
In the megatile final assembly, the pigtails are installed and the connector piece of the pigtail is attached to the scintillator tray with rivets. The fibres in a pigtail are stuffed into their corresponding scintillator tiles, and the rest of the fibre laid and secured in its black plastic cover routing groove. The fibres are taped over with 6 mil polyester tape. To insure uniform compression, all rivets are torqued to fixed, specified level. 6.7.1.5.1
After installation of the optical fibres in the scintillator tray, the tray is put through a QA/QC test to assure that the light yields from each tile are within specifications. Light yields from tiles are required to be ±20% of nominal. Those that are not within specifications typically have fibre damage in either the splice or WLS fibre. The QA/QC test is there to detect and correct these problems. Light yield results from these tests are also stored in a data base for future reference.
Each scintillator tray will be scanned by the Megatile Scanner using a photon source. If any individual tile light yield is outside of tolerance, due for example to broken WLS fibre, corrective action will be taken. The light yield information of each tile of each tray will be saved in computer files. There are 18 wedges in each half barrel of HB, so there are 36 identical wedges. Each layer in a wedge includes one two-j wide middle (M) tray, and two singe-j wide trays. Therefore, there are 36 identical middle, side-left and side-right trays for each layer. The mean light yield of all the layers can be matched by ordering each layer based on its mean light yield, and matching the lowest and highest mean light yield layers, respectively in corresponding wedges. Three megatile scanners will be constructed, one for HB, one for HE and the last one for HOB. The HOE tiles will be scanned by the HB megatile scanner. Each megatile scanner has a photon source that scans the megatile in x-y motion and stores the information in a computer. The HB scanner has a scan of 450 cm110 cm.
The relative tile-to-tile light yields from completed megatiles for CDF is shown in Fig. 6. 34. As the plot is shown for large scale production, we expect the same result for the CMS tiles. The correlation between the light yield variations measured here and the pigtail light yield variations are shown in Fig. 6. 35.
Fig. 6. 34: a) Relative light yield of fibre/connector assemblies (pigtails) before insertion into megatiles. The rms of this distributions is 3.5%. b) Relative light yield of individual tiles after the final assembly of fibres into megatiles. The rms of this distributions is 6.5%.[4]
Fig. 6. 35: Correlation between average fibre light yield (x axis) and average tile light yield (y axis). CDF data.)
The light yield measurements are taken with a collimated 137Cs g source positioned over the centre of each tile. This is taken on an automated x-y scanning table controlled by a PC. The optical readout of scintillator trays, is done using mass-terminated optical cables which go to an optical patch panel which takes the light to a Hamamatsu R580-17 PMTs. The photomultiplier tube gains are monitored by two systems. One is a set of reference SCSN81 tiles (read out with Kuraray Y11 fibres) permanently connected to the PMTs. These are scanned by the g source to provide a tile/fibre reference system. The second monitor is a set of NaI (with Am241 "light pulsers" directly mounted on the face of the PMTs. The megatiles, the x-y scanner table, and the optical system are within a large dark box. Measurement errors are 1% or less and the PMT gain monitoring system tracks the gain to better than 1%.
In addition to the collimated g source measurements, pointlike source measurements are taken using the source calibration tubes and a 137Cs wire source. The correlation between the collimated and point source measurements is good to ~1 % for a given size and shape of tile, indicating that the source tube locations, especially their heights above the scintillator, are very reproducible. There is a ~20% systematic variation of the tile response to the pointlike source as a function of tile size and shape. Large tiles have greater path lengths for the g rays. On the other hand, the collimated g source uses a lead cone such that the direct g radiation falls entirely within a tile. This makes the collimated source response less dependent on source height and tile size. Since only the wire source can enter the assembled calorimeter, a data base is maintained of both the pointlike and collimated source responses.
Using a similar volume integral calculation for the pointlike source responses, the calculated pointlike to collimated response ratios (averaged, for each tile, over all megatiles in a layer) track the measured ratios with a rms of better than 2%. The calculation attempts to model the actual tile geometry in detail. For the final calibrations, the use of measured ratios is preferred because they should reflect any variations in source tube placement, etc., from one megatile to another.
Below are the measurements taken for quality control and quality assurance during production of the pizza pans.
a) Record information on each plate of scintillator from the manufacturer. Take and save samples of the scintillator as megatiles are made.
b) Measure the light yield of control samples from each batch of scintillator sheet. Measure attenuation lengths from batches.
c) Inspect each batch of optical fibres, and measure the attenuation lengths and light yield from samples of each batch.
d) Measure the combined light yield and transmission of each fibre-connector assemblies (pigtails).
e) Measure the light yield of each assembled megatile with a collimated and a wire (pointlike) 137Cs g source.
The key to production quality control is contained in Steps 4 and 5. Production information kept in a data base includes:
a) The information sent from the manufacturer about the scintillator pieces.
b) Measure of the light yield and attenuation of the scintillator control pieces.
c) The results of the UV pigtail scans.
d) The results of the collimated and wire source scan of the megatiles.
6.7.1.5.3
The total calendar time for the final assembly is determined the amount of time it takes to cut the plastic cover plates and scintillator. Production can not go faster than this time. Hence, we will assume that enough technicians will be hired to keep up with rate of production of the preassembled trays and pigtails. We estimate the number of technicians need for final assembly by looking at what CDF needed. CDF needed one technician to put in fibres, one technician to test the pans and one technician to do the final assembly and do other jobs that were necessary. Hence we estimate we need a production manager and three technicians for the final assembly.
The end cap hadron calorimeter must have about 27500 scintillation tiles (4 mm thick) with different size and configuration. Only 1/36 part of them are identical. The mean size of the tile is about 15x15 cm2. The total area of the plastic scintillator is about 620 m2. The total length of the tiles edges and key shape grooves is about 16.5 km. Cutting and grooving of the tiles will be done with high speed milling machines. Technology tested ensures success of key shape groove production at revolution speed 20000 rpm and at the speed 20 cm/min via two milling passes. The edges cutting speed is 40 cm/min via one milling pass. So the mean size tile will be fabricated during 8 min. To finish all the 27500 tiles it will be needed about 1.3 year with one milling machine working 8 hours per day. We plan to use at least two milling machines.
For HOB, as described earlier, there will be one tile per tray in the f direction. But in the z direction, a tray will cover an entire muon ring. This means that one tray will contain 4-6 tiles. The individual tiles will be part of a big piece (similar to a megatile for HB). Deep grooves will mark the z boundaries of these tiles. Due to physical constraints there will be several different sizes of the tiles (see section 4.5.1 for the table of sizes). All these sizes will be kept in a data base, which will be referred to while machining the scintillator plates. The procedure of making the scintillator plates will be similar to that for HB, as described in section 6.7.1. However, since the thickness of the scintillator for HO is 10 mm, the grooves that separate the tiles optically will be 9.5 mm deep, leaving 0.5 mm material at the bottom. The scintillator plate is fragile at this point of the production and therefore has to be handled with care. A CNC machine of appropriate capacity will be used for milling the grooves. A vacuum bed will hold the scintillator in place with rubber gaskets during grooving. The position of the scintillator on the vacuum bed will be marked accurately. This way, the scintillator piece can be brought back to the same position if it is removed.
Grooving will be done in two stages. The deep, separator grooves will be milled first with a 0.89 mm(width) cutter. These grooves will be 9.5 mm deep leaving only 0.5 mm material at the bottom. The scintillator plate will then be taken out and the deep grooves will be filled with epoxy as described for HB in Section 6.7.1. Until the epoxy dries the scintillator plate will be fragile and has to be handled with care. After the epoxy dries, the scintillator plate will be put back on the vacuum bed again in the same position. Then the sigma grooves will be made with a keyhole type design and the procedure will be as described in 6.7.1.1. The length of the grooves will vary according to the sizes of the tiles. These lengths will also be kept in the data base.
The 2 mm piece of black polystyrene plates which will be used for routing the fibres coming from the scintillator tiles, will also be grooved by the CNC machine. There will be one 2 mm and one 1 mm polystyrene plate corresponding to each tray. Therefore these plates will also be of different sizes as given in section 6.5.3.1 Each of the 2 mm plates will be grooved for fibre routing, passage of source tube and quartz fibre routing. The sizes of grooves for each plastic cover plate will be kept in the data base. The CNC machine will be programmed for different grooving using the data base for both the cover plate and the scintillator. Holes will be drilled at regular intervals on the scintillators and the cover plates for rivets which will hold the whole assembly together.
HOB will have 360 trays per layer, i.e. 720 in total. Each tray will be roughly 5 degrees in f (0.41 m) and 2.536 m in the z (h) direction. Scintillators will be ordered in pieces having roughly the dimension of a tray i.e. 0.41 m by 2.6 m. Table 6. 12 below shows the material needed for the production of 720 HOB scintillator trays.
| Material | Quantity |
| Scintillator, BC408, 10 mm (m2) | 770 |
| Black Plastic, 2.0 mm thick (m2) | 770 |
| Black Plastic, 1.0 mm thick (m2) | 770 |
| Reflective paper: Tyvek 0.15 mm (m2) | 1570 |
| Black wrapping: Tedlar, 0.05 mm (m2) | 1650 |
| Y11 WLS multiclad Fibre, 0.94 mm (Km) | 15. |
| Clear Multiclad Fibre, 0.94 mm (Km) | 30. |
| Clear 18-fibre Cable, 0.94 mm (1.2 x 15 mm)(m) | 1600.0 |
| Epoxy: TiO2 loaded resin (Kg) | 100 |
| Kapton tape: 0.15 mm (Km) | 4 |
In order to give a realistic time estimate for the production of the two layers of HOB, an estimate has to be made of the amount of grooving necessary. The following is an estimate of the total length of grooves in HOB for fibre laying.
If one assumes each groove to be roughly of the size 0.18 m x 0.25 m in layer 1 and 0.21 m x 0.28 m in layer 2 then the total groove lengths will be 6842 m and 6912 m for layers 1 and 2 respectively. Similarly, for the tiles with two grooves, the total groove length will be 126 m and 138 m for layers 1 and 2 respectively. Thus, we will have to mill about 14 km to make all the grooves in the scintillator. This will be the major portion of the production work.
The length of the separator grooves can be estimated similarly. Every tray has typically 6 tiles and hence 5 deep separator grooves. Therefore for 720 trays total number of deep grooves will be 3600. Each deep groove will be roughly 0.4 m long (j length of a tile). Therefore, total length of the deep grooves will be 1.44 km.
The 2 mm polystyrene plate will have two wide pathways for fibre routing and one groove each for laser signal and radioactive source tube. This can be considered as four grooves, each of length 2.56 m. The total length of grooves on 720 such polystyrene plates will be about 7.4 km.
The following production times are based on simple calculations that break down the total operation into steps whose timing has been measured. The times are given for the machining of all the tiles in the two layers of HOB. A CNC day is one 8 hour shift. A year is 200 working days. A summary of the HOB labour time is presented in Table 6.13.
| Scintillator | One tray |
| 0.94 mm ball groove | 0.5 days |
| Separation groove | 0.1 days |
| Total | 0.6 days |
| Top black plastic | |
| Grooving | 0.2 days |
| Total CNC time | 0.8 days |
Therefore for 720 trays we need 600 CNC days, i.e. 3 years in real time.
Quality Control during production of scintillator plates
The scintillator pieces will first be examined for any damage, like scratches or breaks. After some superficial cleaning it will be sent for grooving. Next will come fibre laying and packing. A quality control sheet will be filled at each stage of the production on each piece of scintillator. The information on these sheets will be then be entered into a data base in the computer.
The following steps will be taken to check the quality of the tiles.
- Record information on each plate of scintillator from the manufacturer.
- Measure the light yield of control samples from each batch of scintillator sheet. Measure attenuation lengths from batches.
- Inspect each batch of optical fibres, and measure the attenuation lengths and light yield from samples of each batch.
- Measure the combined light yield and transmission of each fibre-connector assemblies (pigtails).
- Measure the light yield of each assembled scintillator tray with a collimated and a wire (pointlike) 137Cs gamma source.
The following production information will be kept in the data base.
- The information sent from the manufacturer about the scintillator pieces.
- Measurement of the light yield and attenuation of the scintillator control pieces.
- The results of the UV pigtail scans.
- The results of the collimated and wire source scan of the scintillator trays.6.7.3.2.1
Fibre cutting/polishing/splicing/assembly
Essentially the same steps will be followed for HOB as described for HB in chapter 6.7.1. Two fibre splicing machines will be set up along with a single fibre polisher. These will be used to polish the WLS fibres and splice them to clear fibres.6.7.3.2.2
Quality control
A UV scanner will be set up which will be able scan and measure the attenuation length of several fibres simultaneously. Fibres will be tested at regular time intervals to monitor the uniformity of their performance.
All the connectors, one at each end of the optical cable, one at the pigtail exit and one at the HPD Box entrance have essentially the same design. Fig. 6. 20 is a schematic sketch of these connectors showing 18 fibre holes, the alignment pin holes, the dimensions w, the width, h, the height, d the depth and the location of mounting holes. The alignment of two joined connectors is maintained by two pin/screw elements within the alignment holes which bracket the 18 fibre holes. The alignment holes are different sizes to allow only one possible pin numbering. We have chosen to number the fibre holes 1-18 and require the 1 pin be next to the largest of the alignment holes. The overall arrangement is sketched in Fig. 6. 36. With this arrangement the two ends of the optical cable may be interchanged without danger of mixing up the fibre numbering.
The connectors on the ends of the cable will use the mounting hole region for attachment to a strain-relief boot and a light tight cable jacket. Similar optical connectors will be used for HB, HE, HOB and HOE calorimeters with only the channel count modified for HE and HOE. The overall sizes of the connectors for the other calorimeters vary according the mounting needs, but the basic design of the fibre holes and alignment holes will be the same.
After studying various prototypes and materials it was decided to have the connectors made by injection moulding. We could have a reliable, precise product (within tolerance) at a reasonable price. The material chosen was acrylonitrile-butadiene-styrene (ABS) plastic which has a shrinkage during injection moulding of less than 0.5%. Delrin, another possible candidate material, was found to have a shrinkage of 2% and was rejected.
In order to do the injection moulding a stainless steel mould is required. Aluminium moulds will not produce the required number of connectors. In the following we describe the tests that were made which lead to these decisions.
In the past we found that alignment tolerances have been met for moulded connectors as long as the moulds are initially calibrated and monitored occasionally throughout production. Measurements made on 1100 moulded connectors (with 0.835-to-0.835 mm fibre matching), produced for the D0 preshower detector determined light transmission to be 81% with sigma=3%. An air gap was used between connectors. Fibres were illuminated by green LED bargraphs with diffusers. The diameter step-up design would improve transmission slightly (85%) however light loss is dominated by a 9% transmission drop due the air gap. Transmission through connectors have been monitored for two years and found stable to ± 1%.
The connectors will be injection moulded from a mould made of steel. The useful life of an aluminium mould is about 1000 pairs of connectors. The construction of the mould will be farmed out to an external vendor. Production of the connectors could take place at the vendor. However, we are also contemplating doing the production on an existing commercial quality moulding machine located at UIC.
Clear fibres and ribbons will be held in place with BC-600 epoxy and then fly-cut using a diamond to polish the ends at the P3 facility at FNAL. Approximately ten connectors can be polished simultaneously in one twenty minute run on the P3 facility.
Transmission depends on careful fibre-to-fiber alignment. The holes for the fibres are fixed during creation of the mould and misalignment must be kept below 50 microns in order to keep the transmission high. Initial alignment will be checked with measurements of the mould and measurement of the first moulded pieces. Several connectors will be assembled with clear multiclad fibre and the transmission directly measured. Green LEDs are used as light sources and the output of silicon photodiodes are measured with a picoammeter.
Other quality factors which must be monitored include thin layers
of plastic blocking holes and bubbles in the connector material
which can affect strength and optical isolation characteristics.
Both are obvious upon visual inspection.
The entire production run can take less than a week in a commercial facility so most monitoring throughout the production run will actually take place after the run and if a problem is discovered, a new run will be performed.
We have produced injection moulded connectors for the D0 upgrade central fibre and preshower detectors. We have the expertise in mould design and injection moulding so connectors for the HCAL prototypes will be produced in house. However, commercial vendors with larger injection moulding machines will be required for production runs for the final detector.
The quartz fibre bundle are formed by loosely gathering together the fibres emerging from the back of the calorimeter using PEEK plastic tie-wraps. The fibre bundle is turned between 75-90 and routed towards the outer radius of the calorimeter. The first row of PMT are located at about 100 cm in radius from the beam, whereas the first fibres emerge from the calorimeter absorber matrix at about 7 cm from the beam. The bend radius of the fibre bundle is at least 200 times the fibre radius, or ~8 cm minimum. The length of a typical bundle is about 1 m, including the bends. The fibre bundles terminate in a highly polished, hexagonally close-packed bundle held in place by a snugly-fitting stainless steel ferrule (cylindrical) collar (from 6 mm-18 mm in diameter, depending on eta) and about 7 cm long) which is epoxied in place. The fibre bundle light guide is bent a second time at the ferrule end, so that the fibre ferrule is oriented parallel to the collider beam (along z), pointing towards the photomultiplier windows, through mirrored air light guides. The ferrules all terminate in holes in a rigid annular plate, held by a pressure collet around the conical holes. The steel plate, about 1.4 m in outer radius, is oriented perpendicularly to the beam direction.
The purpose of the air-core lightguide is to:
a) mix the light,
b) save costs of fibres,
c) provide a method to enable heavy shielding around the PMT to be conveniently penetrated, and
d) avoid Cerenkov background generated by mips in a solid light guide.
The lightguide consists of a hexagonal cross-section hollow regular parallelopiped mirror which is at minimum 3.5 times longer than the useable photocathode diameter. This minimum length ensures that the meridonal light emerging from the fibres (NA=0.22=sinq) centred in the hexagon illuminates the entire cathode with only at most 1 bounce, while only allowing the bulk of the skew rays to have only at most 2 bounces from the mirror. The major diagonal of the hexagon is 1.5 mm less than diameter of the photocathode (15 mm) for a tolerance. The maximal length of the light guide is <35 cm, chosen as a compromise between transmission and shielding.
The shape of the mirror - square or hexagonal - will be optimised before final design in order to maximise transmission, provide good mixing, and minimise needed photocathode area. In the test beam, hexagonal mirrors provided good performance and were a better match to round PMT photocathodes, as described below.
The mirror material has been tested and consists of Alzac (Alcoa Metals), an aluminium sheet material (75-1,000 µm typical available thicknesses). The purified aluminium substrate has been anodised with a special non-porous anodisation, designed for mirrors used in indoor and outdoor commercial lighting fixtures, which survive outdoors in temperate climates. The anodisation produces a sealed film of boehmite, a form of transparent amorphous sapphire (alumina) that is a typical alumina ceramic, and in film coatings highly resistant to radiation damage[5,6]. The total reflectance is guaranteed to be 95% at normal incidence in sunlight. (Note: this process is not normally used for optical mirror protection because the surface finish is not able to reach the wavelength tolerances necessary for high quality imaging - the reflectance has a small ~ few % diffuse component at normal incidence. In the quartz fibre transmission case, at nearly grazing incidence, this is largely irrelevant and contributes to the mixing.) In the HF case, the minimum angle to the normal is about 75°, where the reflectance in the blue (440 nm) averages above 98% in bench tests. These bench tests used a blue LED and the same quartz fibre used in the calorimeter. The light cone from the quartz fibre was oriented onto the mirror material at variable angles and the resulting light measured with Si photodetectors. The mirror materials were fashioned into light guides for bench and beam tests.
In test beam and bench tests, a 1 m long similar cylindrical mirror
~2.5 cm in diameter transmits 65% of the light injected at the
emission angle of the fibres (about 15° max). It is because
the emission from the fibres is <13° from the axis of
the fibre that good light collection is possible using specular
reflection. Fig. 6. 37 shows the design of a slightly tapered
light guide used in the test beam with the prototype calorimeter.
Fig. 6. 38 shows the measured transmission through these ~ 1m
long guides. Fig. 6.39 shows the ADC distribution of the response
to 80 GeV electrons in the test beam through a 1 m long air light
guide, confirming the 65% measured in bench tests and calculated
with ray tracing.
Fig. 6. 38: Measured light transmission through a conical 2.8 cm-> 4.5 cm aperture mirrored air light guide using a quartz fibre as the light source vs the angle of the fibre WRT the axis of the air guide. At ~12° (~78° to the surface) about 65% of the light is transmitted.
Fig. 6. 39: The ADC distribution of the response to 80 GeV electrons in the test beam through a 1 m long air light guide, confirming the ~65% transmission measured in bench tests and calculated with ray tracing.
For the test beam, hexagonal light mixers as shown in Fig. 6.40, were built using aluminised plastic. Measurements confirmed a reflectivity similar to the Alzac materials. The data shown for the test beam was taken mainly through these guides.
For the HF design, we have specified a short air light guide. This guide is ~31 cm long and 15 mm in diagonal. A photon emitted from the compressed phase space of the fibre will bounce at most 5 times in this guide, and be transmitted at a level of 70% with 93% reflectivity; the average photon will have ~4 bounces and be transmitted at 75%. At 95% reflectivity, typical for the (grazing) angle of incidences >78 emitted from the fibre bundle, the average transmission is >80%.
Full radiation damage tests will be conducted over the next 2 years. If any reason should be found to reject these mirrors, aluminium mirrors overcoated with a film of radiation hard magnesium fluoride, silicon monoxide, or all-metal nickel or rhodium plated mirrors with a reflectance of >90%, are alternates which are known to survive high radiation levels[7,8].
The minimum acceptable transmission requirement of the air light guides is 50%, which are met by the guides constructed heretofore. This requirement is set so that the least count in a hadronic physics tower is not greater than 5 GeV at h=3. The average energy of a particle at an average pT of 0.5 GeV, low for LHC , is 5 GeV.
The mirrors will be constructed in an assembly line with a precision slitting apparatus to cut the mirror material with a bevelled edge (60°). 3-sided halves of the hex mirror structural element are constructed from 3 mm thick bent and machine-finished steel. CERN-spec epoxy will be used to fasten the mirror skin to the structural elements. The non-reflective seam between the 6 mirror elements must be less than 250 µm in width. The manufacturing process will be designed to ensure that the mirror dimensions are the same with 1% from mirror to mirror, a tolerance of about ±0.15 mm. The flatness tolerance is ±0.1 mm across each panel of the hexagon. The design parameters are inherently not critical, as the number of mirror bounces is small, so that dimensional changes as high as a few % have negligible effect, so long as the output area is within the photocathode area.
The first 10% of all mirrors will be tested fully with angle (0°-15°) and transverse scans using fibre light sources. Transmission less than 2% from nominal will be rejected. All others will be tagged for selection. If most of the first mirrors pass testing, all mirrors will be visually inspected, and put in a very simple go/no-go jig, made of fibre light sources and a photodiode array connected to a simple yes/no discriminator consistent with the 10% full testing. In any case, the mirror and PMT assembly will undergo a separate optical test later in the assembly process.
The light from the scintillator pans are brought to the HPD box
with optical cables. These optical cables consist of 18 clear
0.94 mm, s type Kuraray fibres with optical connectors at both
ends. The optical fibres are covered with Tedlar to make the optical
cables light tight. Nissei Opto Co. takes the clear fibre and
makes the 18 fibre cables covered by Tedlar. Nissei made a 10
fibre cable consisting of 0.9 mm clear fibre for CDF.
Fig. 6. 3 is a view of the CMS hadron barrel at the large h boundary looking in the direction of the beam. It shows the routing of the optical cables from the scintillator pans to the HPD box. The optical cables and source tube are enclosed in a hadron cable channel consisting of 1 mm aluminium. The hadron cable channel surrounds the cables from layer 0 to the photodetector box. The channel protects the cables and forms a light tight seal for the cables and optical connectors.
The optical connectors and source tubes connectors protrude into the 53 degree crack. This is done for 2 reasons. If the optical connector were inside the copper, then the top of the optical connector must lie below the top of the scintillator pan. Since the optical connector is 4 mm high and the top plastic is 2 mm high, we would have to route out the scintillator where the connector is. The fibre for this tile would have to move in 4 cm way from the edge of the tile. The best location of the source tube connector is outside the copper. The source tube connector connects the source tube in the pan for the tube for the driver. Since the tube in the pan is on top of the plastic, the source tube connector must stick above the scintillator pan. Hence, if the connector is inside the copper the copper must be milled out where the connector is. The source tube connector must be 8 mm in depth. Hence, the scintillator must be removed where the source tube connector is if it is inside the copper. This degrades the uniformity of the edge tiles.
Fig. 6. 21 shows a top view of the scintillator tray at the end of the megatile. The figure gives a detailed view of the routing of the cables and source tubes at the pan.
Fig. 6.41 is a clasp of this region looking in the direction of the beam. The clear fibres for the pigtail for the middle tray are straight to the connector. while the clear fibres for the pigtail for the side tray are curved. The curve enables the optical connectors for the middle trays and side trays to overlap in phi. This reduces the amount of space need to route the cables in the 53 degree gap. The source tubes cannot cross the optical fibres on the pan. Hence, at the edge of the pan source tubes route out on both sides of the cables. As shown in the figure, the width across the optical connector and source tube connector is 6.05 cm. To further reduce the space needed to route the cables and source tubes in the 53 degree crack, the optical cables are offset with respect to each other, see Fig. 6.41. Therefore, the optical and source tube connectors take up 6.05 cm in the j dimension for routing. The thickness of the hadron cable channel is 1.1 mm and 3 mm of tolerance separate the edge of the connectors from the inner edge of the hadron cables channel aluminium. The hadron cable channel is 7 cm wide at the lowest radial point. At the photodetector, the cables for the side and middle layers plug into different columns. The total j width needed for 10 degrees is 11 cm at the photodetector box. Hence, the hadron cable channel will start out with a j width of 6.85 cm and grows to a width of 11 cm at the photodetector box.
Fig. 6.42 shows a r-z view of the routing of the cables
along the 53 degree crack. The optical connector on the pan and
connector for the source tube protrude into the 53 degree crack
by 1.9 cm. A 0.7 cm cable connector connects to the pan connector.
The total protrusion of the optical connectors is 1.9 cm. The
1.9 cm protrusion of the optical connectors is 1.5 cm normal to
the 53 degree line. The optical cables bend to follow the 53 degree
line. We have verified that the optical fibres are not damaged
if their bend diameter is greater than 3.8 cm Taking the minimum
bend radius for the optical cables as 3.8 cm, the optical cables
bend to be a minimum of 2.9 cm normal to the 53 degree line. The
cables are 0.11 cm thick. Near the photodetector box, the cable
bundle will be about 28 cables high, and so the cable bundle is
be 3.1 cm thick. 7.0 cm of space perpendicular to the 53 degree
line will be needed to route the cables at the photodetector box.
The permanent source tubes start bending approximately 1.5 cm (in z) from the pan and 1.2 cm (in the direction perpendicularly across the HB-HE gap) from the copper. With a minimum bend radius of 10 cm, these source tubes will rise to be 5 cm away from the HB copper perpendicular to the 53 degree line. Hence, the cover for the hadron cable channel will be 5 cm perpendicular away from the HB 53 degree line At layer 0 the channel contains both the optical cables and the source tubes, with the height of the channel determined by the source tubes. At layer 15 the channel will be 7 cm away from the 53 degree line, with the height determined by the optical cables.
Each tray (covering 100 in ) has two optical connectors (10 fibres in each), two connectors-mixers for laser calibration and two connectors for radioactive source tubes. Optical cables from the trays go to decoding boxes. In each tray the optical connectors are shifted in such a way that all optical cables can be laid in four layers only not to exceed the space allocated for HE cables, as shown of Fig. 6.43. The quartz fibres in protective skin go from connectors-mixers on the trays to connectors-mixers where the laser light is fanned into 36 fibres corresponding to 36 trays in each layer, see Fig. 6.44. The plastic tubes from radioactive source tubes go around the HE in each layer and end up at the same (protruding from the HE cover plate. The basic idea is shown in Fig. 6.45 and Fig. 6. 46. In this way it will be easy to connect them to control system to check the performance of active elements without interfering with other cables when the calorimeter is moved outside of the magnet coil during a shut down period of collider run.
Light from individual tiles in a tray is brought through optical fibre cables to the decoder box, located above the outer most layer of muon station. In each tray there are 4, 5 or 6 tiles depending upon its location. Generally there are 4 fibres per tile (some smaller tiles will have 2 fibres), making a maximum of 25 fibres per tray (24 fibres from the 6 tiles and an additional fibre for transporting laser light to the scintillator tiles). The standard HCAL optical connectors could accommodate 18 fibres. Thus two such connectors per tray will be used. These two fibre connectors will be fixed at one side. Since there are 6 trays in each 30o j sector, 12 pairs of optical cables (6 pair from six trays of Layer 1 and remaining 6 pairs from layer 2) runs vertically upwards and transport the light to a decoder box located at the outer edge of the muon rings. A 30 mm wide and 5 mm thick corridor in the middle of every 5o j sector on one side of each muon ring is required to route these cables from the tray edge to the decoder box. The total number of calorimeter towers in a 30o j sector for ring 0, 1 and -1 is 36. For ring 2 and -2 the corresponding number is 30. There will be a maximum of 8 fibres for each tower (combining layer 1 and 2).
The umbilical cord of cables and cooling hoses is laid in the 50 cm by 50 cm trench between the transporter tracks. The cables are looped and bound into a bicycle-chain type cable tray. This avoids the triple depth trench which would be required if the bundle were folded back on itself beneath the detector in the running position.
In the garage position the cable bundles are fully extended, while in the operational position approximately 2 meters of slack must be accommodated. This extra 2 meters remains in the trench during beam operation. In intermediate positions 2-7 meters of slack must be accommodated. Since this amount of bundled cable cannot be accommodated in the trench, the trench will be uncovered and the cable bundle looped above it during horizontal moves of the detector. The flexible cable tray is attached to the support structure and will move with it during vertical moves.
The scintillator pans will be assembled and tested at Fermilab. They will be boxed in wooden boxes at Fermilab and shipped to CERN. The scintillator pan is fairly robust. We have shipped 2 testbeam modules to CERN with no damage to the pans or optical cables. The pans will be shipped to Building 168 where the pans will be installed in the wedges.
At CERN a small sample of pans will be tested with a megatile scanner. This will verify that no damage has taken place. If a small sample of pans are fine, then all the pans will not be scanned.
The pans will be removed from the box and the 'venetian blind', which pushes the pan up against the top of the slot in the copper absorbers will be taped on. The copper wedges will be in Building 168 with the slots for the pans parallel to the floor. The pans will be put on a stretcher and the stretcher will be put up to the slot of the wedge. The pan will be slid into the slot.
HE modules will be manufactured in Russia and shipped by rail to CERN for installation.
HOB modules will be manufactured in India and transported by ship/rail to CERN for installation.
HOE modules will be shipped to CERN along with the HB modules.
We anticipate that HF modules will be assembled in Hungary and shipped by rail to CERN.
The optical system is sealed. Other services will be covering the optical system. Hence, we will have no access to the optical system. We anticipate that no access is needed. If we need to access optical system, other services can be removed and we can get at the optical cables and scintillator pans.
During the access period we will do a full source scan of all the tiles. This will determine whether any part of the optical system has deteriorated from radiation or other sources. The expected radiation levels are small enough that we expect the light output from the system to decrease no more than 7% over 10 year lifetime of the detector.
One end of HOB modules will be untrapped, so trays could be removed, if needed.
HOE will be attached to the Endcap muon chambers and could be only repaired when muon chambers are removed.
Alignment of the detector and maintenance of the optical systems
will be carried out from the platform support of the detector
or in the garage position. At beam height, access is provided
by a portable scissors. This device is stored in the garage during
beam operation to protect the hydraulics from radiation. The electronic
modules in the racks are accessed and maintained easily in the
garage position. In the case of a need to access the electronics
racks when the detector is in the elevated beam position, the
portable scissors will be used. The PMT boxes are protected by
endplug shielding in operational position. Access to the PMT boxes
requires no removal of heavy shielding elements.
References
[1] P. Gorodetzky, Rad. Phys. And Chem. 41 (1993) 253.
[2] J. B. Chesser, IEEE Trans. Nucl. Sci. 40/3 (1993) 307.
[3] V. Gavrilov et al., CMS TN/94-324.
[4] CDF End Plug Hadron Calorimeter Upgrade, P. de Barbaro at al, IEEE Trans. on Nucl. Science, vol 42, #4, p510-517.
[5] A. Holmes-Sidle and L. Adams, Handbook of Radiation Effects, (1994) Oxford U.P., Oxford, U.K.
[6] J. Wong and C.A. Angell, Glass: structure b y spectroscopy (1976) Dekker, Basel.
[7] Optical Radiation Corp., Azusa, CA, 91702, USA.
[8] Optic-Forms, Inc. Temecula, CA 92590, USA.
6.1 OVERVIEW 2736.1.1 HB design 2746.1.2 Design Requirements of the HCAL optical system 2776.2 SCINTILLATOR SPECIFICATIONS FOR HB/HE/HOB/HOE 2796.2.1 Scintillator specifications 2796.2.2 Radiation damage. 2816.2.3 Magnetic field 2836.2.4 Quality control. 2836.3 FIBRE SPECIFICATIONS FOR HB/HE/HOB/HOE 2846.3.1 Attenuation length 2846.3.2 Quality control 2866.4 QUARTZ FIBRE SPECIFICATIONS AND REQUIREMENTS (HF) 2876.4.1 Dimensions 2876.4.2 Radiation damage 2886.4.3 Quality control and assurance 2896.5 SCINTILLATOR TRAY DESIGN HB/HE/HOB/HOE 2896.5.1 Hadron barrel scintillator tray design 2896.5.2 HE scintillator tray design 2986.5.3 HOB scintillator tray design Error!6.5.4 HOE scintillator tray design 456.6 HF OPTICAL SYSTEM 486.6.1 HF fibre insertion tooling 486.6.2 HF fibre bundling, cutting and polishing 486.7 MANUFACTURING (HB/HE/HOB/HOE) 506.7.1 HB manufacturing 506.7.2 HE manufacturing 616.7.3 HOB manufacturing 616.8 OPTICAL CONNECTORS AND CABLES (HB/HE/HOB/HOE) 656.8.1 Connector and cable design 656.8.2 Connector production 666.8.3 Quality control 666.8.4 Quality assurance 666.9 HF AIRCORE LIGHT GUIDES 666.9.1 Design description 676.9.2 Performance requirements 716.9.3 Quality control 716.9.4 Quality assurance 716.10 CABLE AND FIBRE LAYOUT 716.10.1 HB layout 716.10.2 HE layout 756.10.3 HOB cable layout 776.10.4 HF layout 786.11 SHIPPING/INSTALLATION 786.11.1 HB shipping/installation 786.11.2 HE shipping/installation 796.11.3 HOB shipping/installation 796.11.4 HOE shipping/installation 796.11.5 HF shipping/installation 796.12 ACCESS, MAINTENANCE AND OPERATIONS 796.12.1 HB/HE access, maintenance and operations 796.12.2 HOB Access, maintenance and operations 796.12.3 HOE Access, maintenance and operations 796.12.4 HF Access, maintenance and operations 80
mm)
mm)
661
160
706
274