lbnl leo greiner , eric anderssen, giacomo contin,

L. Greiner 1PXL Detector Progress – July 2013

STAR HFTSTAR HFT

LBNLLeo Greiner, Eric Anderssen, Giacomo Contin,Thorsten Stezelberger, Joe Silber, Xiangming

Sun, Michal Szelezniak, Chinh Vu, Howard Wieman

UT at AustinJerry Hoffman, Jo Schambach

IPHC StrasburgMarc Winter CMOS group

STAR PXL Detector Progress Report

2PXL Detector Progress – July 2013L. Greiner

STAR HFTTalk Outline

• Detector assembly, integration and installation for engineering run.

• Engineering Run results and lessons learned.• Progress on sensor production, thinning probe testing

and yield.• Progress on production of final detector.• Timeline for installation in run 14.


STAR HFTAssembling sensors into ladders – Assembly

• We use precision vacuum chuck fixtures to position sensors and assemble ladders.

• Sensors are positioned manually with butted edges. Acrylic adhesive mechanically decouples sensors from the cable and prevents CTE difference based damage.

• Weights taken at all assembly steps to track material and as QA.

Reference pins for cable/sensor alignment



• Hybrid cable with carbon fiber stiffener plate on back in position to glue on sensors.

• The cable reference holes are used for all aspects of assembly of ladders and sectors.


STAR HFTCompleted PXL ladder

Completed ladder in anti-static carrying box with follower and check off list.


STAR HFT

• Ladders to sectors


STAR HFTLadders to sectors


STAR HFTLadders to sectors

Sector in optical metrology machine

• Sensor positions on sector are measured and related to tooling balls.

• After touch probe measurements, sectors are tested electrically for damage from metrology.


STAR HFTsectors to detector half

• Sectors are mounted in dovetail slots on detector half.• Metrology is done to relate sector balls to each other and

to kinematic mounts.


STAR HFT

• PXL detector mechanics


STAR HFTPXL insertion mechanics

Interaction point view of the PXL insertion rails and kinematic mount points

Carbon fiber rails

Kinematic mounts


STAR HFTPXL insertion mechanics

PXL detector half with complete insertion mechanism


STAR HFTEngineering Run

Left detector half being inserted

Right detector half being inserted


STAR HFTEngineering Run

PXL eng detector inserted, cabled and working in 1 day access


STAR HFTPXL engineering run statistics

• Installation of 3 sector engineering detector on May 8, 2013. Run ended June 10, 2013.

• 510 GeV p-p running

• 2 weeks to get detector integrated with DAQ, slow controls, trigger, online monitoring, offline event processing. Set up initial thresholds and bias settings.

• PXL tracking data requires low multiplicity. Data taken during short (~1 hr) periods at the end of RHIC fills.

• Approximately 10 M events in low multiplicity environment. (trigger >10 multiplicity in TOF)

• ~600 GB of low multiplicity PXL data.

• ~ 180 GB high multiplicity PXL data (taken late in run at high prescale value to minimize PXL data rate)


STAR HFTInstalled sectors

Measured ~ 1 week after first beam. There were only 5 bad sensors before installation.


STAR HFTOffline event displays

Hit pattern on sensor over one run

Possible heat issue?


STAR HFT

Raw hits/sensorFor 2 sectors

Inner Ladder (1) shows ~ 4 times as many hits as outer ladders.

Offline event displays


STAR HFT

Sector hit display

Correlation plot for TPC trackpointing to outer PXL ladder (shows pointing resolution of un-calibrated TPC)

Offline event displays


STAR HFTLatch-up and SEU (high luminosity)• DATA SET 1: June 8, 2013 (10 hours) • DATA SET 2: June 9, 2013 (5hrs) only sector 2 and 7, luminosity was only ⅛ of the

nominal value; • DATA SET 3: Third data set: June 9, 2013 (4hrs) only sector 2 and 7, sensors in

pattern mode

Set Ladder Sector 2 Sector 4 Sector 7

LU cnt LU cnt LU cnt

Set 1 L1 19 13 10

L2 0 1 0

L3 6 1 0

L4 0 0 0

Set 2 L1 8 0 3

L2 0 0 0

L3 76 0 0

L4 0 0 0

Set 3 L1 31 0 5

L2 2 0 0

L3 7 0 0

L4 0 0 1

Latch-up Data Summary

Latch-up events were observed only on the digital power supply (no latch ups on the analog power supply nor in the MTB.)


STAR HFT

Summary of SEU count in DATA SET 1:

Latch-up and SEU

• These are only JTAG register errors.• Memory errors tested by pattern mode sensor

output are still under analysis.

Set Ladder Sector 2 Sector 4 Sector 7SEU cnt SEU cnt SEU cnt

Set 1 L1 27 13 92L2 1 0 4L3 0 4 2L4 3 5 4

SEU is measured as corrupted sensor JTAG configuration registers and as corrupted data from pattern registers.


STAR HFTSelected Lessons Learned

Mechanical

•Sector mechanical conflict in the inner ladder driver boards.

•Fixturing and sector tube fabrication process.

Electrical

•Power issues (already discussed)

•Diode temperature measurement.


STAR HFTSector Driver board conflict

Driver board mechanical conflict.

We ran with sectors in non adjacent positions for the engineering run.

•Sector tube and inner driver board have been re-designed. •Inner ladder boards are now all low profile components.•Sector tubes made smaller to increase inner layer radius to beamline. (~3mm)

new


STAR HFTSector tubes and assembly fixturing• Sector tube steps – inner radius is too shallow and stress fractures the sensor in the upper corner during the gluing process.• Fixturing for assembling ladders to sectors had an oversized cutout for surface component clearance. This gave an unsupported section at the end of

ladder with the edge of the support behind the edge of the last sensor. This would fracture the sensor during the gluing process.


STAR HFTDiode temperature measurement

DACs

Buffers

Temperaturesensor

DACs 7mA

Analog buffers 18mA

• Each sensor has a integrated temperature sensing diode.

• The diode is located as shown and is in a “hot” area of the sensor.

• The cathode is brought out to a bonding pad whilst the anode is internally wired to sensor ground.


STAR HFTDiode temperature measurement

The response is ~ 2 mV / degree C

Silicon bandgap temperature sensor:

Problem: The ground potential at the various locations of the ladder fluctuates depending on the current draw of the sensors on the ladder which depends on the occupancy and other factors. The effect of this fluctuation is larger than the expected diode response.

Solution: We measure the temperature using a 100 µA current source and a I2C ADC on the MTB for 2 sensors per ladder. We will modify the circuitry (one wire patch) to allow us to measure the ground potential on the sensor being measured. This will result in only measuring one temperature per ladder unless we add another ADC to the MTB and another patch wire.

Suggestion for future: bring both anode and cathode of the diode to bonding pads for flexibility.


STAR HFTOther issues• Tracking with PXL hit points is not yet completed.• TPC calibration needs to be done to allow for tracking and efficiency measurements.• Possible radiation damage to sensors, under investigation. (beam excursions /

dumps after installation could have damaged sensors as some stopped working after the detector was inserted)

• The automated threshold setting scripts worked but we need to update this for the full detector.

• We discovered some sensors that had problems, even when probe tested. We need to carefully map probe testing error modes and properly propagate this to the selection process.

• Efficiency measurements based on integrating the correlation function plots is a work in progress. The method is approximate at best and currently yields 60-70% efficiency.

• Some oscillations were seen with the remote sense voltage regulators for the ladders before installation (despite working fine at LBNL). We modified to local sense but this may have caused other problems (low supply voltage?, VCLP,) This, along with current draw, will be investigated after the detector returns to LBNL.


STAR HFTIn progress

From Hao Qiu on July 4, 2013


STAR HFTProgress in production

• Sensor yield

• Aluminum cable ladders

• Status of production readiness

• Schedule for production


STAR HFTMechanical thinning yield for DRIE Batch wafer #

good sensors percent

comment

1 7 37 77 dirty

1 8 - - Being probe tested

1 9 48 100 Mixed/dirty

210 21 44

Wafer released early, 70 um thick

211 30 63


2 12 0 0 Entire wafer lost

213 35 73


214 22 46


3 15 47 98 Mixed/dirty

3 16 48 100 clean

3 17 48 100 mixed

3 18 48 100 clean

4 3 48 100 Mixed

4 20 48 100 Mixed

4 22 48 100 Mixed

4 25 48 100 Mixed

4 19 48 100 Mixed

4 24 47 98 mixed

4 21 47 98 mixed4 1 48 100 mixed4 23 46 96 mixed4 6 48 100 mixed4 5 47 98 mixed4 2 48 100 mixed4 4 48 100 mixed

• We have been working with Aptek industries for back thinning and polishing of DRIE trenched wafers.

• These are the results of the first batch of 25 wafers from AMS.

• Visual microscope inspection for chips, cracks, etc.

• The processing method for DRIE wafers is being fine tuned, but appears to be good now.

• The yield for the most recent batch thinning of 13 wafers was 99.2 %. We expect this result to be representative for further processing.

• The cost of thinning 25 wafers whilst preserving wafer position information is $6375 ($1200 for preserving wafer position)

Overall yield 87%


STAR HFTPXL Sensor selection

• Automated interface to a database (18 config + 68 result parameters)• Sensors are binned according to performance

Example of wafer maps for Ultimate-2 wafers with DRIE processing (ion etching)

~65% - 70% yield


STAR HFTAluminum cable ladder• First Al cables from CERN PCB shop.• Needed rework to allow wire bonding.• Validated through bonding, soldering and resistance (voltage drop)

testing to allow fabrication of batch to meet schedule.• The first cable loaded with sensors is shown below.• Testing is beginning.• Based on schedule, only the inner ladders will be fabricated on Al

conductor cables for the run 14 detector.


STAR HFTStatus of PXL production parts

Ladders25 wafers of sensors in hand40 Cu cables loaded in hand100 outer driver boards in hand40 inner driver boards arrive in July20 Al conductor cables ship July 1

Fine wire cables 25 in hand – production in progress

MTBs

10 in handPossible minor modificationFull production in August timeframe.Non-critical path



VHDCI Cables All cables in handHalf modified to be non magnetic

PXL (SSD) motherboard

5 production motherboards in hand.Design validated in engineering runTested for use with SSD.Order balance in the next 2 weeks.



Sector tubes Re-design in progressFabrication in August timeframe

D-tubesInsertion mechanics

Minor modifications to existingNew pieces in August timeframe

• The effort is geared to the first production detector to be installed for run 14. The spare detector will follow.

• Other pieces – power supplies, TCD interfaces, control PCs, etc. are in hand.


STAR HFTSimplified Production Schedule• March 2013:

– Deliver the engineering run detector to BNL. installed in May

– Aluminum conductor cable fabrication begins.begun in June

• April 2013:– Re-design as needed for mechanical and/or electrical changes based on eng run experience.

– Fabricate parts for mechanical. – in progress

– Place RDO board orders. – non critical path, to be done in the next 2 weeks

– Production sensors arrive, test and send out for thinning and dicing.

• May-June 2013:– Production dicing and thinning. – Production ladder assembly. in progress

– Final electronics fabricated – non critical path, in progress

– Mechanical assembly of detector parts. – now scheduled for July - August

• July-August 2013:– Sector assembly – now scheduled for August-September

– Metrology - now scheduled for September

– Final detector assembly – Now scheduled for late September

• September 2013: – Delivery of primary PXL detector – now scheduled for early October


STAR HFT

• end



Early optimizationTesting results (ladder with 2 bad sensors)

Assembled ladder on FR-4 handling piece

Sensor noise performance only minimally degraded with all sensors running


STAR HFTOffline event displays

Hits per cluster3-4(Measured 2.5 hits/cluster for normal incidence beam test 120 GeV pions.)

lbnl leo greiner , eric anderssen, giacomo contin,

Documents