Concept of Drawers Thoughts on Drawers and mini-Drawers: Small story Tilecal upgrade meeting (’08 July 4th) François Vazeille Small story Geometric constraints Mechanic constraints Electronic constraints User constraints Short comparison of Drawers and mini-Drawers The elements which drove the present concept of 2 Drawers/Super-Drawer Not independent from the design evolution of the electronics and vice versa.  Conclusion 1

Small story Drawer Birthday: in 1993, just before the first Calorimetry panel on the choice between scintillating fibres and tiles at a time when the beautiful concept of the longitudinal Tile arrangement had no access solution to the PMTs.  New idea of a movable object (welcoming the Front End electronics) sliding on supports (also) welcoming the bundles of WLS fibres. - Support: Girder Rings. - Movable object: Drawer. Faulty elements replaced in place moving the Drawer or extracting the Drawer towards a refurbishing area. Fair impact on the ATLAS detector: no space taken outside Tilecal and no interference with other subdetectors. Comment: Since the very beginning, Tilecal was a good guy, not much requesting (See the conclusion) It was the same, putting the LVPS inside the fingers. 2

Geometric constraints Main constraints: 1. Relative positioning of PMTs with respect to the Girder Rings. 2. The space for the FE electronics. 3. The PMT orientation with respect to the main stray magnetic fields.  Consequence #1: the cross section shape of the Girder. First design Black: Girder Red: PMT orientation Eliminated Chosen  Consequence #2: number of Drawers already reduced by a factor 2.  Consequence #3: the maximum free space is roughly 175x175 mm2. New constraint: to have universal Drawers in sizes, so knowing that a LB Girder is roughly 2 times longer than an EB Girder.  Consequence #4: A Drawer cannot be longer than 3 meters. 3

Mechanic constraints Ideal Drawer from the geometric point of view (and also for electronics/cabling/pipes…: see next sections): 3 m long object with a maximum cross section of 175x175 mm2. Mechanic constraints A welding somewhere  The final design of Girders (Not a square). The Girder tolerances and the true free space inside the Girder. Space taken by Girder Rings The size of PMT Blocks (hosting the PMTs) and electronics boards, cables, pipes, cable trays…  Maximum cross section of 165x165 mm2 for the whole system. The tolerances for positioning the PMTs with respect to the fibre bundles, along the Girder (1 mm in the worst case) and in front of the Bundle to have a 1 mm wide gap between mixer and bundle + the tolerances on the Girder Ring positioning…  The number of channels (PMT Blocks) in the LB and EB Modules.  The stiffness of the matrix and the heat conductivity for cooling… For more information on sizes and tolerances: see Clermont-Fd engineers. 4

• Short summary of envelops: 175 maximum 165 effective 160 retained for safety • Drawer matrix made of Aluminium (The best heat conductivity). 5

 The recovery of the Girder Rings alignment from an outside insertion tool. Basket … 3 m 3 m also? The stiffness of the Basket containing the Drawer and the 2 m possible displacement of the End Caps for access. Concept of a Super-Drawer made of 2 Drawers about 1.5 m long with a flexible link to make easier the insertion inside the Girder.  The flexible link induces a new tolerance. Comment: pulling is easier than pushing. 6

Electronic constraints The best working and reliability need to minimize the number of: - Boards: Tests at the production sites, working in Drawers and Data Base Id. - Stacked Boards (HV side: 2 stages, Readout side: up to 5 with Mezzanine) - Connectors: Power supplies (LV and HV) and analog/digital signals, Fibres and Cooling circuitry. - Cables: Power supplies, analog/digital signals, Delay and Trigger... - Fibres. - Pipes for leakless water cooling. Because of: power drops, sources of HV trips, quality of signals, water leaks … reliability and costs + maintenance. To certify the maximum of components at the same time at the assembly site as an autonomous whole system, true also later. To minimize the number of Drawers: The ideal Drawer would be 3 m long, but because of mechanic and access constraints: Choice of 2 Drawers making a 3 m long Super-Drawer. 7

 Other constraints: • 7 types of Drawers (0 to 6) because of the Tile quality. • To pay attention in particular on: - The very tight space (inside cable trays for example) for routing. The 3in1 Trigger cables: To minimize the number of Adders having cables coming from 2 different Drawers. Adder The routing in between the Drawers, very difficult, in particular at the readout side (with access from the opposite side). • The space taken by flexible pipes of the cooling circuitry. Heat loss outside Stiff Stiff Less Cooling 8

User constraints First users: Clermont-Fd people  production and certification of Drawers. • No special constraints … except the requested manpower over 3 years and the fragility of some components. • The assembly from the zero level requested a detailed procedure at every level to prevent mistakes, but corrected during the certification tests. Second users: same people (+ Georges second part)  CERN insertion/certification • As above over 3 years, in various sites (180, 175, pit). • But handling more difficult, despite dedicated tools  mainly because of access in not foreseen conditions or even impossible: ATLAS did not respect fully the foreseen requests.  At least 2 people, but more in some cases.  Alignment of Girder Rings not always good. • Repairs not always easy because of the same access difficulty. • Special Drawers needing more works, manpower and time  No Finger or special Fingers, with insertion from the opposite side in some cases. 9

 Third users: the refurbishing team over a shorter available time (1 year). • Possible thanks to the sharing of tasks and manpower, and also because the works did not start from zero. • Same access difficulties as above, + many difficult refurbishing operations in-situ and others at the surface.  The next users will continue the refurbishing and start the maintenance. The difficulties will be the same and even increased if ATLAS decided to not follow the access specifications. 10

Short comparison of Drawers and mini-Drawers  The ideal 3 m long Drawer was not realistic.  The nearest design was a 3 m long Super-Drawer made of 2 Drawers. It is obvious than having mini-Drawers (3, 4 … Drawers to make a Super-Drawer) should go away from the ideal solution: here is a comparison of advantages/disadvantages. Geometric aspects: the same for every solution. 11

2. Mechanic aspects  Cross section envelop • A priori, it is not a concern in this comparison. But there is a proposition to move the 3in1 cards from the PMT Block to the Mother Boards. • In my opinion - The present design inside the PMT Block has many advantages: • Direct connections PMT/Divider/3in1 without any cables  Reliability and pick noise minimization for signals. • Benefit of the iron shielding of the PMT Block. • Benefit of the cooling given by the iron shielding. • PMT Block working and tested as an autonomous object. • Access not very easy but solved. - The design “on the Mother Board: • Loss of all the above qualities. • If there are put as Mezzanines: additional stacking level  Space and cooling problems. • Access needing to remove boards above: Digitizers, Interface (in some cases) and to handle cables and connectors  New failure risks. 12

… come back to mini-Drawers  1 2 2: 1.4 m long (Present design) 1 2 3 3: 0.93 m long -  (1 link more) 1 2 3 4 4: 0.70 m long – 2   Positioning of PMTs with respect to fibre bundles Mini-Drawers: More difficult  2 reasons - Tolerances increased at the inter-Drawer interface. - New increase in the case of flexible links Comment: the flexible link is an obligation! Reminder: - With the present design: loss of some fibres (broken) recovered by increasing the accuracy on the links. - Theoretical tolerances/bundles increased by the replaced fibres. Standard bundle Added fibre 13

…  Pulling and pushing operations: More Drawers  Increased tolerances.  Possible blocking in the pushing operation. 1 2 1 3 2 … Cooling circuitry: More Drawers  Flexible part longer.  Stiff part smaller. Cooling less efficient Heat loss increased 1 2 3 Efficiency decreased More Drawers  More Cooling connectors  Tightness more difficult  Possible working problem of the Cooling plant.  In the worst case: Possible leaks (When several leaks occurs despite LCS). 14

3. Electronic aspects And completely new electronics for everything (PMT Blocks?)  Time, manpower … and funds. More Drawers More Boards  More tests at the production sites, bigger Data Base for Drawers…  More replacements in case of problems including other boards for accessing  new failure risks.  More cables/fibres/pipes and connectors at various levels: known as being the main problem till now:  new failure risks: Voltage drops, loss of communication, damaged signals…  The reliability of the electronics will be not increased, despite it should be new electronics.  The cost is increased.  The Drawers are more interdependent  more failures expected. The number of types of Drawers become bigger: now is 7, new: how many types?  New study requested (not easy).  The assembly will request to pay more attention.  Drawers less and less universal.  More adders will share several Drawers  Failure risks increased. Adder Reminder of Standard case 15

4. User aspects What do bring mini-Drawers?  The handling is easier, including the transport.  They can fit a smaller access space … if ATLAS changed the Tilecal requests (See later). But the insertion/extraction is not better and even worst (section mechanics). 16

Conclusion The only interests of mini-Drawers are their transport and handling, (but not in the insertion/extraction operations) and less space requested to ATLAS. Other points (Mechanics, electronics) are worst and even strongly spoiled. The assembly task of Drawers is different from a refurbishing task, because starting from zero: protocol with tests all along, plus certification (short term tests, long term tests).  Redoing everything means to have the funding (very big)! That does not mean that the present design should not be improved, in particular in the handling aspects about tools (Baskets and others) and in the use of requested (and not often available) ancillary tools (Hoist, rod support to pre-position the basket…as foreseen, not made), and other points at various levels (As proposed in the previous talk or others). Moreover, ATLAS must guarantee to maintain the access gap as it was foreseen in the specifications. 17