17/05/18

Bustape @ Oxford:

Electrical: - Plugin temrinal wired in, single phase 32A 240V, several standards 20ohm Vacuum: - you should not be able to tape it, place them under vacuum

• vacuum, pump, liters/min pumps 27cubicfeet/min
• size of pipes matter:
• get up to 0.75 ?
• 185 um thickness of the tape, 17x2 nominal coper on each side + 50 um base capton + 100um from the cover +- 10%
• 50 um core crapton + 25um glue
• Problem of enough flow rate, what is the rate? for one hour or so

testing: incoming reception test bare tape, cocure to carbon fibor skin, measire monitor stretching, integrating to the honeycomb and cooling pipes full electrical and mechanical tests, complete stage after forming the stave, be 99% sure

• plate that holds the tape, whole plate the same tickness as stave, silicon
• you put an open lid to maintain the tape in place
• we don't want cardboard between the tapes
• coreagated plastic ? that is regid

• handling with gloves

• Air conditioned room: constant temperature, +-1 degree for the stability of the tape

• temperature cycling: coolant temperature is -25C, not for the bustape testing

• Computer: network and database for the ITK production database, regularly upload after a test

• once the tape is tested, it is covered by a plastic blue tape to protect the tape. ticker tape for the incoming reception test, co-cure, then remove it for testing, then place a second thinner tape, you need a tooling to make sure it is applied uniformely

• second table that needs vaccuum connection to maintain it

• drawings for the rubber rollers, cost about 160pd.

• Check with Parker in Germany for stages

• Time: 15min total for the tape and film, few mm crudly cut protection tape around the bustape
• company: qualitape for tapes to protect bustape

• jig made things much better

• Electronics cabinet: 7.5m, Make sure it is not too long to not impact the tests, for exmple noise and capacitance

• Filtering: AC noise, driver @ 16kHz from mototrs, if you are trying to measure GOhms, picking up noise from the motor,
• keyflee dvm filter down to nonoAmps (see the part list)
• High voltage isolation: HV switches should not be more than 650V, test at double the opertional voltage

Space:

• 1x1m for the electronic cabinet, long table for the roller,
• sockets stick out of the wall, so the cabinet will come to the side
• Rack mount 9U, lighting controls for the ring, HV pass, laser pointers

Transportation:

• Libjana sends the tapes without foil under vacuum, get it at DESY and then coat it in plastic roller, then sends them off to the co-curring

• Potential problem of edges havign carbon falling off the edges

• Calibration how to calibrate camera system: Measure the smooth distortions before and after co-curring, carbon expands by few millimiters
• Calibration stages: 4 stages
• Safety concerns

Resolution of 10 microns

08/03/18

Visit Jens at RAL - Day2:

The module loading procedure on the stave:

• Determine the cone shaped loactors positions by getting the right and left side and recording it for all locators.
• Fit a line through the average of the left/right locator positions.
• This line sets up your x-axis in parallel to the stave
• Next determine the origine of your coordinate system. You define it as the intersection of the line you fitted with the edge of the holder of the cone locators. Now, the drawings tell you where the fiducials of every module should be with respect to the origin of this coordinate system
  • Note that some rotation is requied to be performed since the modules are not perfectly sitting in parallel but rotated
• Next step is to load the positions of all the fiducials into an excel spreadsheet that is in turn readout by LabView controller of the gantry. Now the gantry knows where all the module fiducials should be, you can say go to module 5 and right fiducial or left fiducial and it goes.
• Next step is to get the module from the nitrogen chamber, place it on top of the module loader, apply vaccum to hold the module on the table, turn the nobs to lift the small table under the module from the board tester, then pluck out the wire bonds very carefully
• Then rise even further and adjust the angular rotation you need to place the module at an angle
  • Look up Vernius scale of rotation used for precision rotation
• Get the bridge to place on top of the module assembly, there will be contact with the sucktion cups, undo the vaccum on the module table, and apply it unto the bridge. Hoop, the module jumps to the bridge.
• You unscrew the bridge and remove it very carefully from the small table and place it on top of the stave onto the bridge pins that lock it in place. Now the fun begins...
• The scope of the gantry knows where fiducial right/left of the module should be, you already put it in, all you do is select which module and which side (right/left). Once you get two fiducials on the upper two corners of a module, the module is correctly placed as it is all symmetric.
• Start adjusting the knobs on the bridge to get the fiducials to line up with the cross of the gantry scope in the desired position. This is the most TEDIOUS part as you need to make adjustments in x and y and also rotations as you have two x-axis
  • In fact this is a feature of the Jens' bridge, it is very hard to make the rotation of the two x-axis simultaneously, so what happens is you get diagnal or rotational motion
  • Also you need to make rotational adjustments, as the rotation that one gets from the table is only to get close to the desired precision, but the knobs get you the final angly you want. you just cannot do it just with the knobs as it will take too much time and i am not even sure you have enough room.
• Once you placed both sides of the module in the desired location, you are done with the placement. You put the module carefully on the drawer in the bottom. Then you move to the glue...
• The way Jens has the gluing step is to first cut a pattern on a double 75mm tick sheet, so 150mm tick glue at the end.
  • Place two sheets on top of each other, they stick to each other
  • enter the pattern drawing including the angle rotation in the program
  • cut the pattern via the special cutter/printer
  • remove the inside of where you want to fill with the glue and leave the rest
  • place an additional sheet just to hold the structure of the mold when you remove it from the original paper where it is stuck
  • Stick the pattern on the stave, slowly remove the "structure" sheet you added at the end for transportation and stability of the mold.
  • Make sure to remove all the air bubbles, and now you have a pattern on the stave
• Now that you have the pattern, you need to fill it with the glue, time to mix it up!
• There are two glues needed
  • Se4445 for the bulk module curing, it has two components black and white, you mix them up following:
    • Keep the two glues seperated, mix each seperately than shake them
    • get a clean container on a precise scale, reset it, then add 4grams of white and 4grams of black componenets
    • take them to the rotation machine and place it inside for 1 minute, it will get rid of all the bubbles and mix it well
    • once it is ready, take off the small cup, and place the container on a vaccum chamber, slowly apply vaccum to get all the bubbles and humidity out
  • Epotek E4110-PFC or sylver epoxy, same idea two components that need to be mixed with precise proportions and put into the rotation machine, and vacuum chamber
• Now that you have all your glues, you apply Se4445 on the left edges so that you use a tool to spread the glue across the cutout shape of the pattern you produced. you need to make sure you have enough glue to fill it out.
  • note that the advantage of the tape is that you protect the circuits around, it is dangerous to get conductivity on the stave where you are not supposed to... something to consider for the glue machine
• Next, you apply Epotek on the one dot where you want conductivity for the HV (i think?)
• Once done with the glue, you can take the module and place it back on top of it. in theory, the adjustment is already done. but now you do the final touches. the fact that you have glue on the surfaces, makes it easier to move the module arround.
• you get the two corner fiducials aligned with the center of the gantry scope, and in principle you are done! This process is tedious, at least with the manual placement of the module and screwing the knobs.
• Save the position of the center of the scope in respect to the fiducial markers.
  • if you have the two positions, in essense you know where the modules are on the stave. meaning you know where they should be (the center of the scope), and you know how far off they are. AS LONG AS YOU ARE WITHIN 50MICRONS, WE DON"T CARE.
  • now you need to account for some motion once you remove the bridge the next day after curing, but it should be well within the 50 microns
• Let the glue cure overnight and collect it on the next day.
• Record all pictures of the fiducials and the relative positions, also save the glue and lavel it. this is useful to go back to it in case there are problems with gluing one of the modules. then you can check the details.
• next use a laser for metrometry to see if the module bends after curing. typically it does. you want to have this information to give it to the detector alignment people. they need to know if the modules bend at all or curl.

Now let's talk about philosophy of the method:

• Jens' method requires fiddling with the adjustment of the knobs and can be improved by using a glue dispencer and that's it.
• In essense, with two people it can easily be done in half a day for one stave face
• Bernd's method needs software identification of fiducials, you cannot load two modules at once, and you still need to place the fiducials precisely on the gantry as the gantry needs to know where the modules are. Somewhat this is alliveated by the pattern recognition but still ...
• things to think about and discuss with sergio/ingrid

07/03/18

Visit Jens at RAL:

Fiducial size ~ 100 microns GLUE NAME: WHAT ARE THE SPECS IN NEWTON for Gantry: Check video of wirebonding from Zeuthen where each time you wire bond, the noise level goes up Testing of wirebonding? You will see noise if there is signal? When placing the modules, circumvent the modules that we already placed

• The number of stave modules to load is 400, 200 at RAL and 200 at BNL
• Jens plans to do 1 stave side per day (2 sides), say 2 staves per week, so complete in about 2 years
• The modules you are placing are first wirebonded to a test setup for quality assurance purposes. you have to remove the wires first, then place. It would help to have two people.
• The process goes as: in the mornining, slide the stave from the moving support to the gantry, align the gantry, load the module on the bridge, locate the fiducials, turn the nobs in the bridge to precisely place the module, remove the bridge, repeat the procedues for all modules in the stave, apply the glue pattern on all modules, as long as you don't blow on it the outer layer of the glue should be ok, then take the bridges with the modules attached to them via vaccuum and place them one by one. You still have freedom to adjust each module to make sure that the modules are aligned. Once done with the adjustments in half a day, you let it cure until the next morning (nearly 18hours), then you turn it over repeat the operation.

• Wirebonding: Once the modules are loaded on the stave on both sides, you put the stave on the transport sheet and you take the stave to the wirebonding machine. The first part is the alignment where you tell it about the geometry of your modules and where they need to be loaded (location of the chips, heigths,...). The bonding is a delicate process: the robot can make 5 bonds/second but when it gets to the complicated layouts where you have in the order of 4 different layers of bonds one on top of the other, it is more safe to bond 1/sec. You can know if we correctly wire bonded by doing electrical tests. If few channels are missed, then it is not a big deal in the grand scheme of things. The other thing is that you have a telescope mounted on the wire bonding machine so that you can take a look and check if everything is bounded.

• glue pattern: ```

  gantry x-axis, ^ z-detector | |

| |

| | ``` Along the x-axis on top of the cooling path to ensure thermal conductivity through the glue, along the z-axis in parallel to the two hybrids to ensure a solid support for the hybrids

• Fun facts

  • one side of the stave is for HV voltage and ground, the other side is for data (there are technicalities involved so that you don't run your data lines and power on top of each other)
  • The motion of the robot arm is quite powerfull, , Jens installed an interlock system so that once you enter the power is shutdown
  • Once you cut the power, then home is no longer accurate, you have to re-scan the whole surface area to locate "home"
  • There are two coordinate systems: Gantry and Stave/Petal, they need mapping one into the other. The gantry does not need to know anything about the geometry of the stave or petal, all it cares about is a configuration file that you load into it

• Question: Why do we need 4 loading sites for petal?

• In principle we can do 1 petal a day. Based on loading two petals on the same stage (check that we have enough room).

  • Set up two frames around the petal (do we really need the inner frame that Bernd has)
  • Align/calibrate to identify the coordinate system of the gantry
  • Precision placement of the module with its bridge (figure out the bridge pick up and placement)
  • Record the location
  • Apply the glue pattern
  • place the module
  • Repeat with the other modules on one petal, then move to the modules on the other petal
  • Here we complete one sides for two petals, let it cure overnight, and do the same on the other sides
  • So completed two petals in 2 days, 4 petals a week, we should produce all the petal in 3 years in one site or 1.5 for 2 sites or <1year for one site
  • if we need to produce 400 petals, this can be done in

• What will slow us down is the 2 separated R3/R4/R5 (which are apparently seperated because Infenion cannot do wafer radius cuts that large?)

  • We need to machine a tool that will pick up either both modules at the same time. In this case, we pick up one module, then pick up the second module with high precision to determine where it should be in respect to the first module. In this case you only need to align based on one of the modules, the other one will be already aligned
  • Or you machine a tool which can place one module and somewhat allow the other module to be inserted, seems harder
  • One note about the tools, they can be quite expensive in the order of ~20k per tool
  • Need to discuss with machining people at DESY to understand the process

• Material:

  • Computer:
  • Camera:
  • Lens: with light through, 2-3micron resolution, 5x lens
  • Splitter:

• Need to understand the optics of what we are trying to achieve keeping in mind 2 things:

  • Be able to identify the fiducials
  • Have enough clearance to freely move the robot without running into things > VERY BAD, can harm and damage things

• Contact these companies to get the lensing you need.
• Smart scope measurement?

18/02/18

Teams working on module loading:

• DESY:
  • Othmane Rifki
  • Sergio Diez Cornell
  • Ingrid Gregor
• SFU (Vancouver):
  • Kate Pachal
  • Francesco Guescini
  • Bernd Stelzer
• BNL:
  • Hannah Herde
  • Laura Bergsten
  • Prajita Bhattarai
  • Gabriella Sciolla
• Freiburg:
  • Liv Wiik
  • Marc Hauser
• Valencia:
  • Daniele Madaffari