1 CERN-NTU-HIP Production Readiness - PowerPoint Presentation

Slide1

1

CERN-NTU-HIPProduction ReadinessRachel Bartek◊, Po-Hsun Chen◊, Anna Elliott-Peisert†, Francesco Fiori◊, Jaakko Härkönen§, Rong-Shyang Lu◊, Panja Luukka§, Stefano Mersi†, Mercedes Miñano Moya◊, Jui-Fa Tsai◊,Esa Tuovinen§◊National Taiwan University†CERN§Helsinki Institute of Physics

Slide2

Outline

Status of module components and their testingStatus of HDI test and modules assembly jigs, cold box, X-ray test setupBump bonding status: production and test experience production speed and schedule (as agreed with Advacam)Module assembly experienceModule qualification and calibration experience, test throughput including temperature-cyclingDB experienceModule production schedule including early production with TBM08b2

Slide3

Component Yields

3ComponentYield✧CommentsSensor0.7CiS yieldDicing yield0.93Good sensor on wafer to good sensor ready to bump-bondROC0.9ROC yield from previous productionsBare module0.8All 10 bare modules were grade A quality for pre-productionHDI+TBM0.9One TBM replaced on full module due to faulty TBM in pre-productionAssembled module0.97 assembled modules

All numbers except ‘sensor’ from previous production

Slide4

Number of Components Required

NeededYieldTotal NeededReceived as of Feb 2015ExpectedCommentsROC(v2.1respin)33920.6552182400(12 wafers)March 2015Test at PSISensor2120.47451225To be orderedTest by CiS then HIPTBM08c2120.812620June 2015Test at MOSISHDI2120.81262262+Have allTest at Catania, then CERNBase strips4240.9

4720Production Feb/MarchMaterial at PSI4

Half of Layer 3: 352/2 = 176 + 20% spare = 212 modules

ROC yield: 0.9x0.8x0.9=0.65

Sensor yield: 0.7x0.93x0.8x0.9 = 0.47

HDI+TBM yield: 0.9x0.9=0.81

Base strip yield: 0.9

(due to module assembly yield)

Slide5

Other Components

ReceivedTotal RequestedExpectedCommentsTemporary cables100250 @Catania100250 @CataniaHave allWill received HDIs with cables from CataniaModule handles24250April 20158 needed to be re-surfaced, were cleaned at CERNHDI handles3232Have allWill rotate handles as HDIs built into full modulesModule adapters215TodayModule inserts3250TodayAllowing for broken molex connectors. Already broke 3HDI foam storage1111 (6 HDIs per foam)Have allESD safe, will reuse as HDIs built into full modulesModule foam storage39

50 (5 modules per foam)OrderedESD safe, capacity for 250 modulesHDI shipping boxes3535

Have allShip 8 HDIs to CERN from Catania, boxes returned with empty HDI handles

5

Slide6

Materials Needed To Complete Production

3.5 more batches of sensors14 more ROC wafers262 TBM08cWith temperature circuit?472 base strips225 module handles13 module adapters18 module inserts11 ESD safe foam molds for module storage6

Slide7

Test, Assembly Tools/Jigs Status

HDI test (Catania/CERN):Test setup and SW readyPerformed a cross calibration exercise October 2014 between CERN and CataniaBare module test (HIP/CERN):Test stand in Helsinki almost readyJig will be ready by the end of this week, the commission SWTest setup and SW ready at CERNStill trying to improve grounding to avoid noiseModule assembly (CERN):Two of four assembly lines readyAssembled 8 pre-production modulesModule qualification (CERN):One cold box that can test four modules at onceX-ray box with temperature controlled volumeTests one module at a time and 4-target moveable stage7

Slide8

Bare Modules (bump-bonding)

Bare module pre-production at Advacam:10 digital modules delivered to CERNAll of grade A qualityBare module test experience at CERN:Digital probe card, noise filter circuit in placepXar and Labview control test standProduction speed at Advacam:10 bare modules per weekTest speed at CERN:Two days to test 10 modules75 min per bare module8

Slide9

Module Assembly

Modules assembly at CERN:Two fully commissioned lines8 full modules and 5 prototype modules producedChecklist for quality assurance during assemblyThroughput 2 modules per dayWill commission other two lines and have throughput of 4 modules per day if need be before extra gluing shifts need to be addedWill glue two sensor modules per week and one full module per week as well until TBM08c delivered9

Slide10

Full Modules

Wire bonding done by CERN bonding labFull module test bench next to assembly stand:IV, PixelAlive, BumpBondingWill keep this test until production becomes boring and monotonousStored in ESD safe foam in temperature and humidity controlled clean roomIs storage at RH of 50% good enough for HDIs?10

Slide11

Cold Box Test

Module qualification at CERNThroughput 4 modules per day in cold boxSW: pXar, elComandante, and MoReWebStill missing module adapters for simultaneous test of 4 modulesCurrent test stand commissioned for 2 modules at a timeCold box qualification sequence:IV + Full test at +17°CIV + Full test at -20°C10 Thermal cycles between +17°C and -25°CIV + Full test at -20°CIV + Full test at +17°CAverage time per sequence: 9 hours11Temp (°C)

Slide12

X-ray Test

Module calibration at CERNCommissioned for Fluorescent light measurements (Zn, Mo, Ag, Sn) used for Vcal calibrationUses MoReWeb12High rate efficiency measurementCross check with bare module test for missing bump-bondsStill waiting for MoReWeb compatibility

Slide13

Database

Sensor data:Test results have been loaded in the databaseBare module test:Parameters being put in place by database developersHDI:Test results have been loaded in the database13Module assembly:Assembled modules have been entered into the databaseCold box module qualification:Test results have been loaded in the databaseX-ray module calibration:Not available in database yet

Slide14

Manpower Coverage for Production

10 modules/week for 25 weeks (7 months)May 2015 to November 201514Bare module testingHDI testingAlignment and gluingFull module testingThermal cyclingX-ray calibrationExpert: Setup and commission the test MercedesRachelRachel, Stefano, MercedesRong-ShyangJui-Fa, FrancescoPo-Hsun, Rong-ShyangTime estimate~1 hour/module20 min/ HDI1 hour/ 3 modules1 hour/ module9 hours/ 4 modules2-3 hours/moduleTime estimate(10 modules/week)12 hours3 hours4 hours10 hours21 hours30 hoursCommentMight not need to test allMight not need to test all4 hours

Time estimate does not include debugging~1 hour effective work for 4 modules, the rest “should” be automatic30 solid hoursAvailable personnelMercedes, Anna, Strasbourg student (Jun-Aug)

Rachel, Anna, Strasbourg student (Jun-Aug)Rachel, Stefano, Mercedes, Anna

Rong-Shyang

, Andreas,

Mercedes,

Jui-Fa

, Stefano,

Strasbourg student (Jun-Aug)

Jui-Fa

, Francesco, Strasbourg student (Jun-Aug)

Shared by Mercedes,

Rong-Shyang

, Po-

Hsun

, 1.5 day each

Slide15

Production Schedule CERN-NTU-HIP

Pre-production 201410 bare modules built at Advacam and tested at CERN8 modules assembled by CERN1 bare module given to HIP to test their bare module test stand1 module damaged during assembly, being used for alignment2 modules given to other production centers6 modules tested at CERN for qualification and calibrationBare module production 2015 (295 pieces):March through October 2015 with 10 modules per weekWill receive 30-39 bare modules before MayHDI test 2015 (262 pieces):Test 20-30 HDIs with TBM08b in before MayTest remainder as received from Catania in late spring and summerModule qualification (236 pieces):Test one full module per week with TBM08b until TBM08c deliveredTest 10 full modules per week at full speed15

Slide16

Conclusion

CERN-NTU-HIP is ready to start production of Pixel Phase I Upgrade modulesOnce base strips have been received16

Slide17

Back Up Slides

17

Slide18

Detailed Sensor Calculation

CiS production yield = 0.7 an average from Panja’s table of 73% if one takes all good sensors or 67% if one disregards wafers with only 1 good sensorDicing yield = 0.93 Starodumov’s Good sensor on wafer-> good sensor ready for BBBump bonding yield = 0.8 Starodumov’s diced good sensor + good ROCs -> good bare moduleOverall yield is 0.7*0.93*0.8 = 0.52To make 250 good bare modules we need 250:0.52 = 480 sensors, or 160 wafers, or 6.4 batches of 25 wafers. With three batches in hand CERN needs 3.5 more. We will discuss at the meeting if someone can share a batch with us. Otherwise we order 4 batches.18

Slide19

Other Materials II

ReceivedTotal RequestedExpectedCommentsJigs44Have allAligning jigs for productionGlue stamps26OrderedRecently finalized stamp designAraldite 201112?< 1 week from orderReorder from CERN store when < 2 tubes leftGlue nozzles16?< 1 week from orderReorder from CERN store when < 7 nozzles left19

Slide20

Pre-Production Review

10 bare modules built at Advacam and tested at CERNAll grade A8 modules assembled at CERN1 bare module given to HIP to commission their bare module test stand1 module damaged during assembly, being used for alignment2 modules given to other production centers before full qualification5 modules tested at CERN for qualification and calibrationTwo bare modules damaged during bare module testTwo modules lacked glue under HDI bonding pads for wire bondingTwo given to other centers before full qualification and calibration20

Slide21

Pre-Production Low Yield

Training assembly personnelUsed for aligning the gluing jigsHandled much more than foreseen in productionSome were assembled more than once ripped off the HDI to glued second HDI on sensor moduleCommissioning the test benchesAll of them went through the bare module testing many timesThe goal of the pre-production exercise was to spot the weak and delicate points of the assembly and to write a quality assurance document, which will guide the operators. Document draft already madeChecklist for assembly and HDI test made21

Slide22

Pre-Production Review

Bare ModuleOrder MadeDamaged During AssemblyCommentsAssembly LocationCurrent LocationMoReWeb11M0329Glue between ROCsNeeded new TBMPSIGiven to Aachen25M0336Scratched during bare module testFails I-V curve, 3 ROCs scratched, Jig incidentCERNCERN33M0284Bump bonds missing in middleGrade C cold box testCERNCERN42M0360Glue between ROCs

PSITo PSI for TBM09 tests5Vacuum incident, damaged corners

Given to HIP for bare module test stand

6

4

M0285

Scratched during bare module test

Grade A

CERN

CERN

7

6

M0283

Taped to module handle for bonding

Grade B cold box test, see slide 23

CERN

CERN

8

7

Glued to vacuum

head

Could not be completely

bonded due to glue mess

CERN

Using to align

jigs

9

8

M0297

Jig incident? Was heated

No

Vcal

fit ROCs 4, 6, 10, 11 and 15, Cold Box Grade B

CERN

CERN

10

Being used to investigate

Digital probe card failure in bare module

test stand

Flat

cable fault

CERN

22

Grading algorithms still need to be discussed

Slide23

PH Optimization

M0283 and M4509 fail at same ROC for PH optimizationResult in Grade BGrading algorithms need to be discussedpXar and MoReWeb23M0283M4509

Slide24

Links to Test Results

X-rayhttp://pccmspixel186.cern.ch/pixel_dev/DATA/Overview/Overview.htmlCold Boxhttp://pcpixelcb.cern.ch/pixel_dev/MoReWeb/Results/M0297 needed timing scan at -20°CHDI testhttp://cmspixelprod.pi.infn.it/cgi-bin/view.cgi?objName=Test_Hdi_Electric24