Phase 1 FPIX module assembly status Kirk Arndt

Phase 1 FPIX module assembly status Kirk Arndt Purdue University for CMS FPIX Mechanical Group S. Kwan, C. M. Lei, S. Los, G. Derylo (Fermilab) G. Bolla, D. Bortoletto, I. Shipsey, Y. Ding, V. Noe-Kim, D. Snyder (Purdue) Phase 1 Pixel Upgrade Workshop - Aug 2012 Pixel Modules Parallel Session 1

Phase 1 FPIX module Flex cable strain relief (built into module end holder) Flexible Printed Circuit SMT connector (SMK CFP 8730 -0101 F) HDI Flat flex cable 75 cm length (short “connector saver” pigtail flex will be used for testing during assembly) Module end holder (with thru hole for #00 or M 1. 2 screw) FPIX sensor Phase 1 Pixel Upgrade Workshop - Aug 2012 2 x 8 ROCs Pixel Modules Parallel Session TBM 2

Flex cable and strain relief insertion sequence Threaded insert in module end holder (for engagement with screw thru hole in blade) Connector cover flips and locks easily with only a small force to connect flex cable Phase 1 Pixel Upgrade Workshop - Aug 2012 Pixel Modules Parallel Session 3

Flex cable strain relief trial It works! Phase 1 Pixel Upgrade Workshop - Aug 2012 Pixel Modules Parallel Session 4

Semi-automated module assembly HDI flex and TBM 2 x 8 Bump. Bond Module Status • For HDI-to-sensor gluing, beginning trials using a stamp on the ‘pick-and -place’ machine to improve the uniformity of epoxy dispensing (similar to the process used by BPIX). • Early results on stamping Araldite 2011 epoxy on glass slides look promising… Phase 1 Pixel Upgrade Workshop - Aug 2012 Pixel Modules Parallel Session 5

Update: module assembly and testing at Purdue/Nebraska • Completed reverse engineering of custom front-end tools on loan from UCSB. Designs sent to Nebraska for machining of the tools for Nebraska and Purdue gantry systems. • Aerotech software upgrades in progress to make Purdue and Nebraska gantry controls identical. • Completed Lab. VIEW program at Purdue with all functionality required for pixel module assembly as baseline code for commissioning Nebraska system (see video at http: //www. youtube. com/watch? v=q. Ms_w 89 dn. Xc) • Plan to adopt the PSI cooling box for FPIX module testing…received drawings and parts list from Andrei Starodumov (…need to follow-up with Philipp Eller for the changes that ETHZ made to the design) • Parts for two boxes will be purchased by Kansas University and machined and assembled at Nebraska…one will go to Purdue. Phase 1 Pixel Upgrade Workshop - Aug 2012 Pixel Modules Parallel Session 6

Wirebond encapsulation past and future Phase 1 Pixel Upgrade Workshop - Aug 2012 Pixel Modules Parallel Session 7

Vibrations From CMSFPIX Tech Board Meeting – Dec 1, 2006 (full report in backup slides) click here for the movie • Vibrations are detected all the way down to 3 x 8 pixels pulsed in all ROCs that correspond to a current of ~8 m. A in the single Bond being investigated. • More pixels implies more current and so larger amplitude. • The frequency detected is ~14 KHz and obviously the wire oscillates also when pulsed at 7 Khz and 3. 5 KHz, but the amplitude goes down Phase 1 Pixel Upgrade Workshop - Aug 2012 Pixel Modules Parallel Session 8

Phase 1 Pixel Upgrade Workshop - Aug 2012 Pixel Modules Parallel Session 9

FPIX Plaquette Encapsulation • The encapsulant (Dow Corning Sylgard 186) is removable. Wirebonding can be done again in case of failures • The 2 -part encapsulant is mixed, poured into a syringe and degassed in a vacuum or centrifuge • The syringe is connected to an air powered fluid dispenser and inserted into a holder on a 3 -axis stage (motion along rows of bond feet is motor-controlled). • Shape of encapsulant beads is determined by the rate of volume dispensed and the rate of motion of the syringe • Mixed encapsulant should be used between ~0. 5 and ~1. 5 hours after mixing. Fixture cure at room temp. is ~8 hours, full cure in 48 hours. • Estimate ~3 hours to encapsulate 6 plaquettes (including mixing and clean-up time). • See section 3. 3 “Wirebonding and encapsulation” in Assembly and qualification procedures of CMS forward pixel detector modules, Nucl. Instrum. Meth. A 638, Issue 1, 11 May 2011, Pages 55 – 62, doi: 10. 1016/j. nima. 2011. 02. 106 Phase 1 Pixel Upgrade Workshop - Aug 2012 Pixel Modules Parallel Session 10

FPIX Encapsulation result ROC and VHDI wire bond feet were potted in separate encapsulant beads for each ROC Phase 1 Pixel Upgrade Workshop - Aug 2012 Pixel Modules Parallel Session 11

Bead dimensions height ~100 um height ~150 um width ROC bead ~600 um width VHDI bead ~900 um Phase 1 Pixel Upgrade Workshop - Aug 2012 Pixel Modules Parallel Session 12

Phase 1 FPIX encapsulation • Use 3 -axis motion-controlled “robot” and programmable dispenser controller for increased accuracy and repeatability • Would be nice to add camera to robot for pattern recognition alignment of dispensing pattern-to-module wirebonds • Will try conical dispensing nozzle (shown above) with 50 micron ID orifice (used 100 micron ID syringe needle in the past) Phase 1 Pixel Upgrade Workshop - Aug 2012 Pixel Modules Parallel Session 13

Backup slides 14

flex cable strain relief design Gold plated area 12. 8 mm X 2. 5 mm, Non-bendable region 15

9. 6 o angle up from plane of connector 16

Step 1: Place Flex cable in front of connector; Bend at ~5 mm from end. Step 2: Insert Flex cable into connector; Bend at 7. 5 mm from end. Step 3: Snap in the cap (handle not shown) Flex cable strain-relieved. 17

18

• The insertion direction is reversed so that friction locks the plug when the cable is pull accidentally from the free end. • Locking feature = deflection tip at the front end of the plug. • Need experience with prototype. 19

• Having trouble finding a manufacturer of a “textured” stamp similar to the stamp used by BPIX (on the left). • Closest I’ve found is shown on the right. 20

Purdue status Report G. Bolla, P. Merkel, I. Shipsey, K. Arndt, G. Arndt, D. Graves, D. Bortoletto, C. Mc. Kinney, I. Childres and more …. • The main topic is encapsulation or not, and if so how CMSFPIX Tech Board Meeting – Dec 1, 2006 21

Encapsulation or not • Approach – Understand what is going on with measurements • Can be very time consuming, but fun as well – Make an educated decision – Develop the tools and procedures to get the job done 22

• • Magnetic field How do we get it at Purdue? – – Cheap Handy and easy to use Safe Compatible with • • • Investigated various options: – Multiple magnets available in the Phys. Bldg. up to 2 T • • • installing a Plaquette in the field with the ~right orientation Watching the bonds with high-mag optic Close to the DAQ (cosmo based thx Rutgers people) Similar to what Atac-et-al used What a pain » » » it might trip on anything Schedule with other users Not really my style Make a SHOPPING LIST – Bought some nice permanent magnets ($150 for the whole thing) • • Neodymium Iron Boron (Nd. Fe. B) Magnets 2 inches diameter and 2 inches thick (a beauty but I had to spend 2 hours on the phone with the vendor for safety reasons; He thought I could not handle them properly and I was going to be killed by them) – Safety first so Kirk (the wise guy) guy helped and set up a safe frame for the new toys – Bought some first surface mirror to get the bond image out to the stereoscope • • 6$ a piece (I bought 10 of them) Handy 9 mmx 9 mm I can put them anywhere and they are compatible with a small gap between the magnets 9/SQRT(2)=~6. 4 mm between the magnets (~ 0. 4 inches for the Yankees) – Coupled our lab digital camera to the stereoscope and take movies and pictures 23

Mapping the field with a GAUSSmeter Mapping of the B-field. The field is above 0. 9 T within a radius of 2 cm. The grid of the map in in 5 mm steps. This matches pretty well the predictions done when we made the shopping list. 24

Frequencies of interest are the ones associated with the beam cycle/s 1. f(88. 924 ms) is 11. 24 KHz 2. And many others (higher than…. ) Current variations (Source: Roland H. at the upgrade workshop) Source: http: //sl. web. cern. ch/SL/sli/new_filling. gif Magnitude of the current spikes Per BOND: 8 m. A/2 bonds= 4 m. A in a single ROC (inside a Plaquette) From VHDI-to-HDI 8 m. A*N_ROC/8 (8 m. A*N_ROC/4) in a 2 x. N(1 x. N) Plaquette: 2 x 5 and 1 x 5 is 10 m. A, 2 x 4 is 8 m. A 2 x 3 is 6 m. A 1 x 2 is 4 m. A Width of the spikes: 3 ms (t 5), ~1 ms (t 3 and t 4), ~0. 2 -0. 3 ms (t 1 and t 2) 25

Idig versus time Tektronix TCP 202 AC/DC Current Probe • The current I measure is the one flowing on the VHDI to HDI (fanout board) bonds. • I cannot find a way to measure directly the current in the ROC bonds • Borrowed a current probe from EE. • Replaced the Vdig connection between the Plaquette and the FANOUT module with a wire (bypass the connector) • This allows for monitoring of the current to the Plaquette with extreme precision in time and about 1 m. A resolution 26

• This is hard to Trigger get in house READOUT of 6 ROCs I have chosen to set the ROCs With the wrong WBC in order NOT to have a long, long READOUT Cal inj – Get smart (Gino You should call for help) 27

Precision of the method • Clearly capable to detect the <1 m. A during readout (0. 75 m. A from the PSI 46 V 2 Manual) 28

Using the ROC as a pulse tuner • Depending on the pattern and number of pixel you inject you can get different shape pulses on the digital current 29

Vibrations click here for the movie • Vibrations are detected all the way down to 3 x 8 pixels pulsed in all ROCs that correspond to a current of ~8 m. A in the single Bond being investigated. • More pixels implies more current and so larger amplitude. • The frequency detected is ~14 KHz and obviously the wire oscillates also when pulsed at 7 Khz and 3. 5 KHz, but the amplitude goes down 30

Remarks • The expected current variation induced by the abort gaps (inactivity) of the LHC on the digital power supply of the ROC are reproducible (Inverted) by injecting charge in multiple pixels in a Plaquette with a good tuning capability • With a 0. 9 T magnetic field wirebonds can be seen to oscillate with current pulses as small as 8 m. A (we cannot exclude that at lower current the wire vibrates with an amplitude lower than our detection threshold) – This translate to <2 m. A for a 4 T field • Should we worry also about the AOUT bonds and maybe some others? • The per BOND current swing expected in LHC are up to 10 m. A for FPIX • Encapsulation (or something else) is mandatory! • We started a process of learning how to encapsulate the bonds in the plaquettes (next 4 transparencies) – Great help from the expertize and equipment left over by the CLEO detector construction 31

32

Encapsulation • The encapsulant (Sylgard 186) is removable. Wirebonding can be done again in case of failures. • The 2 -part encapsulant is mixed, poured into a syringe and degassed in a vacuum. • The syringe is connected to an air powered fluid dispenser and inserted into a holder on a 3 -axis stage (motion along rows of bond feet is motor-controlled). • Shape of the encapsulant bead is determined by the rate of volume dispensed and the rate of motion of the syringe. • Mixed encapsulant should be used between ~0. 5 and ~1. 5 hours after mixing. Fixture cure at room temp. is ~8 hours, full cure in 48 hours. • Estimate ~3 hours to encapsulate 6 plaquettes (including mixing and clean-up time). 33

Encapsulation trial result ROC and VHDI wire bond feet are potted in separate encapsulant beads for each ROC. 34

Bead dimensions height ~100 um height ~150 um width ROC bead ~600 um width VHDI bead ~900 um 35

Checking the outcome • • Does it still work? YES Any effect on the Analog OUT (different capacitance due to the different dielectric on the bonds) NO • Yet to be done: • • Thermal cycling • I cannot see a problem with it (with the geometry of the BEAD that Kirk achieved). • Will do this w-e Vibration tests • Will get done after CERN 36

Conclusions (encapsulation) • As a group we believe there is no question left and the encapsulation is mandatory • We have a way to encapsulate the bonds on plaquettes that is reliable and fast enough – A few crosschecks to be done • If a decision is taken, we will start encapsulating the bonds before the next shipment. 37