Tracker Week, CERN, 20-24 Oct 2003 23 Oct 2003Tracker Week - Module ProductionSalvatore Costa - Catania Executive summary from Bonding WG Meeting 21 Oct.

Tracker Week, CERN, 20-24 Oct 2003

23 Oct 2003Tracker Week - Module ProductionSalvatore Costa - Catania

Executive summary from Bonding WG Meeting 21 Oct 2003

Salvatore Costa

INFN Catania

23 Oct 2003Tracker Week - Module Production 2 Salvatore Costa - Catania

Summary of recent TIB&TOB prod


Module Pull Tests (1)


Module Pull Tests (2)

PA Test Areas

0.00

0.20

0.40

0.60

0.80

1.00

1.20

24 25 26 27 28 30 33 369 1043 1044 1045 1046 1047 1048

Module ID

SD

Firenze To Pd Santa Barbara

PA Test Areas

6.00

7.00

8.00

9.00

10.00

11.00

12.00

24 25 26 27 28 30 33 369 1043 1044 1045 1046 1047

Module ID

min

max

mean


Module Pull Tests (3)PA-Sen 50th wire

0.00

2.00

4.00

6.00

8.00

10.00

12.00

14.00

25 26 27 28 30 33 46 47 48 369 363 370 342 341 340 339 1043 1044 1045 1046 1047 1048

Module ID

min

max

mean

Firenze To Pd Santa BarbaraBari Pisa

PA-Sen 50th wire

0.00

0.50

1.00

1.50

2.00

2.50

3.00

25 26 27 28 30 33 46 47 48 369 363 370 342 341 340 339 1043 1044 1045 1046 1047 1048

Module ID

SD


Module Pull Tests (4)Sen-Sen 50th wire

6.00

7.00

8.00

9.00

10.00

11.00

12.00

1043 1044 1045 1046 1047 1048

Module ID

m in

max

mean

Santa Barbara

Sen-Sen 50th wire

0.00

0.10

0.20

0.30

0.40

0.50

0.60

1043 1044 1045 1046 1047 1048

Module ID

SD


Highlights from Center ReportsAACHEN Just completed pilot run for TEC with 10 R6 frames. Bonded and tested the 10 frames

within two days. No problems with the P.A to sensor bondings. Had 1-2% failures (~5-8 lift-offs) during automatic bonding sensor-to-sensor. Investigating this with Hesse&Knipps.

BARI Got 48 Hyb in Sep. Bonded APV-PA for 11. New PAs pose no bonding problems.CATANIA Both Hughes are back in operation and equipped with fresh wire, after fixing hw damage

suffered in moving from old to new Dept. Building Will begin bonding Test Structures, Modules after this TK WeekCERN “Whiskers” still a problem in Bonding APVs to Planar PA with Delvotec).It appears that K&S

8090/8060 do not have this problem. So, UCSB OK. FNAL has accepted to help in bonding TEC 2,5,6). Hybrid bonding at CERN is otherwise

OK. If FNAL does pick up 20% of PA-APV bonding, will need to bond about 24 hybrids per day instead of 35/day.

PISA Showed Pull test results for Sensor qualification on 32 new Test Structures (16 STM and

16 Hamamatsu).– Over 898 channels total pull tested, FBL or SBL occurs on 16 strips (1,78 %)– An overall summary of 199 Test Structures (145 STM and 54 Hamamatsu) indicates

that over 5560 channels total pull tested, FBL or SBL occurs on 118 strips (2,12 %)


Highlights from Center Reports Reported on mechanical stress tests of proposed (now adopted) Hyb cable stiffener. 3 out of 4

Modules were retrofitted with stiffener and all 4 tortured with many different types of mechanical stress. Subsequently tested, the 3 with stiffener showed normal electric behavior, while in the one without 6/13 lines were lost, besides exhibiting evident cracks at optical inspection.

SANTA BARBARA Received first Hybrids with Planar PA's, bonded, and didn't see any of the "whiskering"

problem that Alan had found. Received first 6-chip hybrids and bonded 8 of them. Also bonded 5 of the 4-chip hybrids. (All

the stereo hybrids received were already bonded). Bonded 2 6-chip modules, 3 stereo modules. All modules have been fully tested with ARCS and LT system

– Both pulsed and continuous LED tests taken– ALL of 42 strips designated in DB as bad-CAC and isolated had regular test results

(passed) – No problem seen with these channels

Also worked (Tony) on adding stiffener to Hybrid cable test also OK.STRASBOURG Working on Just received 15 modules of the ring 7.VIENNA Will need to adapt jig as soon as they receive Modules with new frameZURICH Received 4 Modules which all failed test after ground bonding


Alan’s Comments on mod to TEC modules to avoid bond breakage

• TEC module design modified, adding a ceramic piece to be glued between Sen & PA spacer (and Sen-Sen). Ceramic piece and glue joint are exactly where one wants to support the module for bonding. Thus, bonding jigs must be modified. Alan would like to know how and how it works.

• TID modules may also have the same risk of bond breakage so some solution will have to be devised for them as well. Alan would like to know that solution and how it works.

Tracker Week, CERN, 20-24 Oct 2003

23 Oct 2003Tracker Week - Module ProductionSalvatore Costa - Catania

Which strips should the Bonders skip

because of known Sensor defects

(a question from Bonding WG to Sensor experts)


Recipe for strips to leave unbonded

• React only to bond immediately upstream of sensor with bad strip

• Skip (= leave unbonded)– All bad IDIEL (considered pinholes)

(IDIEL_1_SEN_.POSITION_OF_BAD_STRIPS)

– All Isolated bad CAC (believed to have high chance to develop into pinholes with irradiation) (CAC100HZ_1_ SEN_.POSITION_OF_BAD_STRIPS)

– All but lowest in a bad CAC chain (believed to represent shorts) (CAC100HZ_1_ SEN_.POSITION_OF_BAD_STRIPS)

CAC Example: 3 34 35 36 37 skip isolated 3, skip all but lowest in 34-37 chain, or: bond only 34


Isolated bad-CAC strips

Recently questioned, because:• They should fail the Pinhole search with Karlsruhe’s LED

system • But some Italian Labs and UCSB have bonded them

anyways in this initial Module Prod for investigation purposes and found no evidence of pinholes:

• >>>>>>>>>>>> LED test NEVER failed <<<<<<<<<<<<We wanted to learn more on bad I_DIEL & bad C_AC strips I’ve performed a comprehensive study of the bad

strips…and… as a result of this study, we are no longer confident on

any part of our current rule to skip bad strips So we are turning to Sensor experts for guidance!


Bad strips study

• In DB there are relevant data, in Tables IDIEL & CAC100HZ, for 2427 Sensors

• For 7 Sensors, data are nonsense 2420 Sensors are included in this study• For each of 2420 sensors, download from DB:

1) I_DIEL value (in nA) for all strips

2) List of bad strips for I_DIEL

3) C_AC value (in pF) for all strips

4) List of bad strips for C_AC



Bad strips study

Accounting: out of 2420 Sensors…1. Bad-I_DIEL strips are 2898 in total (~1.2 per Sensor in average)2. Bad-C_AC strips are 10251 in total (~4 per Sensor in average)

1. Of these, 3707 are ‘isolated’, 6544 are ‘chain’ (‘shorted’)2. Out of the 3707 isolated bad-C_AC strips:

1. 1146 (30%) are also flagged as bad-I_IDIEL2. 2561 (70%) are not flagged as bad-I_IDIEL

Plot Distributions of:1. I_DIEL for good-I_DIEL strips (expect ~0)2. I_DIEL for bad-I_DIEL strips (expect >0, but… )3. ‘relative’ C_AC for isolated bad-C_AC strips (expect small values, <1)3. ‘relative’ C_AC for chain bad-C_AC strips (expect integer multiples, 2, 3,…, but… )4. IDIEL for isolated bad-C_AC strips which are also in bad-I_DIEL list5. IDIEL for isolated bad-C_AC strips which are not in bad-I_DIEL list


I_DIEL for good strips (1)

• Plotted for 869,158 strips

(approx 1700 Sensors)• Extends to ~1A

but peaks near 0 as expected




(approx 1700 Sensors)• 857,754 (99 %) are within

1 nA




(approx 1700 Sensors• 742,388 (85 %) are within

0.1 nA• ~560,000 (65%) have

I_DIEL=0.00• There are a few negative

values


I_DIEL for bad-IDIEL strips (1)

• 2400 Sensors• 2898 bad-

IDIEL strips• (~1.2/Sensor)• Extends to

1.2x107, but most are much lower



• 2400 Sensors• 2898 bad-IDIEL

strips• (~1.2/Sensor)• 0<I_DIEL<1.2x107nA• 1576 (54%)

1A<I_DIEL≤1mA



• 2400 Sensors• 2898 bad-IDIEL strips• (~1.2/Sensor)• 0<I_DIEL<1.2x107nA• 1576 (54%)

1A<I_DIEL≤1mA• 797 ( 27%)

0.1nA<I_DIEL≤1A• A peak at ~100 nA

?



• 2400 Sensors• 2898 bad-IDIEL strips• (~1.2/Sensor)• 0<I_DIEL<1.2x107nA

– 1576 (54%)

1A<I_DIEL≤1mA– 797 ( 27%)

0.1nA<I_DIEL≤1A

• 476 (16%) have ‘normal’

(=as 85% of good strips) I_DIEL < 0.1nA !!!

Bond or not?


Relative C_AC for Isolated bad-CAC strips (1)

• 2400 Sensors• 10251 bad-C_AC strips• (~4/Sensor)• 3707 Isolated• (~1.5/Sensor)• Extends from

-10000< <2000

but most are much lower

C_AC

<C_AC>


Relative C_AC for Isolated bad-CAC strips (2)

• 2400 Sensors• 3707 Isolated bad C_AC

strips• (~1.5/Sensor)

-10000< <2000

• 3566 (96%)

-2 < < 2

• Some negative values• Peaks at ~0 and ~0.6

Bond or not?

C_AC

<C_AC>

C_AC

<C_AC>


I_DIEL for Isolated bad-CAC strips (1)


strips• (~1.5/Sensor)• 1148 (30%) also in bad

I_DIEL list• I_DIEL Distributions are

similar to the whole bad-I_DIEL ones in all scale ranges (as expected)




strips• (~1.5/Sensor)• 1148 (30%) also in bad


similar to the whole bad-I_DIEL ones in all scale ranges (as expected)

Picked in bad IDIEL list



• 2400 Sensors• 3707 Isolated bad-

C_AC strips• (~1.5/Sensor)• 2561 (70%) not in bad


similar to the whole good-I_DIEL ones in all scale ranges


• Out of the 2561 not in bad I_DIEL list:

• 2556 (~100%)

0 < I_DIEL < 1nA• 2358 (92%) have ‘normal’

0 < I_DIEL < 0.1nA• Only 193 (8%) have

0.1nA < I_DIEL < 1nA

(not shown)

Bond or not?



Relative C_AC for Chain bad-CAC strips (1)

• 2400 Sensors• 10251 bad-C_AC strips• (~4/Sensor)• 6544 Chain• (~2.7/Sensor)• Extends from

-200< <1000

but most are much lower

C_AC

<C_AC>




strips• (~2.7/Sensor)• -200< <1000

• 6508 (99%) -2 < < 8

• Expected integers (2,3,4…) Some do, but most have <1 with distrib ~ Isolated:

Our rule that assumed them to be all shorts might be too naïve!

C_AC

<C_AC>

C_AC

<C_AC>



• 2400 Sensors• 6544 Isolated bad C_AC strips• (~2.7/Sensor)• -200< <1000

5889 (90%)

-2 < < 2

Neighbors, but ‘isolated’? 619 (10%)

2< <8

There is a peak at 2.5 Actually shorted?

C_AC

<C_AC>

C_AC

<C_AC>

C_AC

<C_AC>


Summary & Request for guidance

1. Strips in bad-IDIEL list: 16 % do not seem to have a bad I_DIEL: bond or skip?2. Isolated strips in bad-CAC list:

– They all seem to have bad C_AC (‘relative’ C_AC mostly <1)– 30 % are also in bad-IDIEL list– 70 % are not in bad-IDIEL list – 92% of those not in bad-IDIEL list have indeed good I_DIEL, 8% slightly

high I_DIEL, almost none outrageously high I_DIEL– Those tested with LED pinhole search, all pass it

We wonder if we should bond these3. Chain strips in bad-CAC list:

– Only 10% of these seem to actually be shorted (‘relative’ C_AC >1)– Most (90%) behave just like the isolated ones

We wonder if we should bond or skip these or treat them in a more sophisticated way, such as: compute relative C_AC, then If <some threshold (1?), treat them the same way as the isolated If > , apply the current rule for shorts i.e. bond only the 1st in chain

In all cases we ask Sensor experts for guidance

Tracker Week, CERN, 20-24 Oct 2003 23 Oct 2003Tracker Week - Module ProductionSalvatore Costa - Catania Executive summary from Bonding WG Meeting 21 Oct.

Documents