ID status ATLAS week June 2004

1
ID status ATLAS week June 2004

• Try to give overall update since February
• Will use slides foreseen for the LHCC Monday (one
  hour talk) so some I have to skip several slides
  in the following
• Will cover TRT, SCT, PIXEL status in that order
• Summary
• A large amount of engineering, integration,
  installation work is ongoing
• Strong effort on services and services supports
• Gas and cooling work
• SR1 integration preparation
• Installation and commissioning planning
• Will not cover today (there will be more in the
  LHCC Monday about these subjects)
• Use slides from many sources but in particular
  Fido Dittus, Leonardo Rossi and Mike Tyndel

2
The ATLAS TRT electronics ASICs first

• The ATMEL contract problems had delayed the
  analogue chip by several months, but all 143
  wafers were finally delivered.
• DC wafer-level tests gave good results so hope
  for final yields as planned (order 50) - around
  45 needed for complete TRT.
• All chips were then packaged and labeled
• Early tests indicated low yield, but this was
  eventually traced to other problems
• All chips tested by now, but first half will be
  re-tested (lt15 sec per chip)
• Yield between 50 and 60 - should have enough
  chips even for C-wheels!
• DSM digital chip made in one iteration with high
  yield, 92 wafers delivered (2x ordered)
• 8000 chips tested high yield
• Remaining chips to be tested next (lt11 sec per
  chip), before re-testing first half of ASDBLRs

3
TRT Front End Boards - Barrel

• Active Roof Boards
• 1BL and 1BS production PCBs received, component
  mounting in progress
• Qualified Assembler (ACAMAS)
• Noise looks good and insensitive to clock,
  details of grounding, etc (120 DAC counts 2 fC
  or 250 300 eV)
• Edge-plated boards, ground attachment to space
  frame with RF springs
• Final prototypes of 2F, 2B and 3B exist
  (installed in testbeam) 3F design in review

4
TRT Barrel boards - cooling

• Temperature differences from Reservoir
• Cooling Plate lt 1 degree C.
• DTMROC 5 degrees C
• ASDBLR 20 degrees C on average
• Tension Plate 10-12 degrees C on average
• Thermal RTV performed better than thermal foam in
  our limited tests.

5
(No Transcript)
6
TRT Front End Boards End-cap

• Manufacturing quality of boards under evaluation
  some concerns, in particular for boards from
  SEISYSTEM
• Have ASD boards for 4 A-wheels 2 B-wheels
• Have DTM boards for 2 wheels
• More boards needed for end-cap integration by end
  of August

7
TRT Barrel Module acceptance
8
TRT Barrel Integration

• Barrel Support Structure
• Inner Cylinder end-frames assembled within
  tolerances
• Copper ground layer mounted
• RF finger installation nearing completion
• SCT rails installed load test completed
• Released for Module installation
• Integration procedures tested during preparation
  of testbeam setup
• Module installation starting in 1-2 weeks

RF fingers
9
TRT end-cap wheel assembly
10
(No Transcript)
11
TRT WEB circuits (around 2900 needed)

• Production stopped after observation of
  de-lamination problems in Sep.03
• 200 existing WEBs were reinforced with glue
• now being assembled into wheels
• Another 100 WEBs were already assembled into
  wheels
• no problems with gas tightness or mechanics
  observed

Delamination problem solved end 2003 !
12
Web production

• 50 of needed A-type webs accepted at CERN
• 70 of needed B-type webs accepted at CERN
• Webs now at all stages in the pipeline
• Webs should not delay wheel assembly anymore, if
  production continues as now
• Expect some slow-down during summer
• Quality needs permanent attention and large
  efforts

13
Wheel delivery schedule EC side C OK side A
very critical
B-wheelassemblyat JINR
A-wheelassemblyat PNPI
14
TRT End-cap Integration

• Two wheel stacking apparatus built(A-wheels
  B-wheels) and tested
• Wheel stacking alignment procedures to be
  validated
• Starting July-August aim to finish first stack
  end 2004
• Second EC (side A) very critical due to late
  arrival of last wheels

15
SCT Barrel Endcap modules

• SCT Endcap Module, 462/252 built/good yield
  promising (84) testing still behind (163 not
  yet tested of the above)
• The most critical part is thermal baseboards
  (spines) from Russia, new agreement in place to
  speed up and it is helping, but still no stock
  very constructive help for IHEP
• Have for some time worried about some of the W12
  sensors (see next slides)

• SCT Barrel Module, 1635/2270 completed/started
  approaching 80
• All clusters in production, yield acceptable

Hybrids now flowing in sufficient numbers, pitch
adapter and de-lamination problems solved
16
CIS detectors

• The ATLAS Endcap SCT uses silicon sensors from HP
  and CIS(The barrel uses HP)
• There are 5 different shapes (see table)
• The W12 CIS sensors for the inner ring were
  oxygenated to improve radiation tolerance.
• Some W12 CIS sensors exhibit breakdown below the
  specified 500volts after being stored in dry
  nitrogen
• The effect is now understood as being caused by
  charge in the oxide interface. It depends on
  humidity (see fig) time
• A fast screening test was devised in MPI and
  used to classify the detectors.

17
CIS detectors

• The screening test is done using a fast ramp in
  normal wet air and measuring the current at 150
  volts (the initial operating voltage)
• 48 have Ibias lt 1 mamp
• 36 have Ibias gt 5 mamp
• 200 HP detectors have been ordered by MPI to
  replace the 36. There is an option to order an
  additional 130 detectors after a more detailed
  study is completed to fix a safe Ibias ltcut in
  the range 1- 5 mamp

18
CIS detectors

• Note
• In general modules have a better behaviour than
  the individual sensors i.e. after wire-bonding
  the breakdown voltage increases.
• Modules with breakdown can usually be trained
  and brought into an LHC-like operating regime
  safely provided they are biased slowly initially
  then kept biased.
• The effect goes away after irradiation (see
  following slides)
• The effect is not seen on the other shapes of the
  CIS detectors (25 screened)
• 90 have Ibias lt 1 mamp
• 3 have Ibias gt 5 mamp
• Efforts are underway to ensure that there is no
  impact on the schedule

19
SCT macro-assembly getting all components
together
Barrels produced, first two at RAL (B3 and B6).
Disks also going well (PCI/NIKHEF) , all
received, first disks now at RAL.
Barrel services and disk services next critical
items, now the first disk and barrel have been
transfered to Liverpool, Oxford. Several
technical issues related to harness robustness
(now solved) and pipe connection qualification
(now done), brackets (now repaired see next
slides) and currently B4/B6 harness designs (also
next slides) have introduced very serious delays.
Large increase of manpower has been implemented
to recover...
The completed barrels and disks are sent to
Oxford Liverpool/NIKHEF respectively (next
slides).
20
SCT the bracket problem

• Barrel module brackets coming off.
• Discovered end February
• All barrels produced (B3,B4, B5 and B6), B3B6 at
  RAL, B4B5 still in Geneva at that time.
• B3 had all harnesses on, waiting for the cooling
  loops (8, each for 48 modules). Expected
  completion by end March.
• The bad surprise - 8 of the module support
  brackets (glued) have come away on B3
• Then also 2 on B6 (next in line)
• This a fatal problem feared to become worse with
  time, and repair was needed of all brackets (two
  per module).
• Situation today B6 repaired (1 month delay), B3
  repaired (3 month delay), B4B5 done in Geneva
  before sending to RAL without introducing delays.
  B5 has now been sent to RAL after repairs.

21
SCT B6 bracket repair (some details)

• Pad reinforcement G. Barbier / E. Perrin
• Mechanical reinforcement.
• Screw M1x 3 stainless steel, washer.
• Longnut drop of glue to secure nut
• Work started on repairing B6 on the 7th April
  and completed on 26th Apr
• FSI installation started 30th Apr
• Installing rails, inserting spindle pedestals,
  building service cages expect to mount first
  harness 10th May

Material added per module - 0.18 g of st.
Steel - 0.15 g of PEI or PEEK This represents
0.029 X0 if spread over module surface of 120 x
60 mm.
Screw Longnut
22
The tooling Miniature drilling
machine Template Guide vacuum
channel Procedure (Here shown on
B5) Dismount brackets and 3rd mounting
points Position template on a pair of pads
Backup slide
23
Drill the 4 holes dia. 2mm Mount the four
screws, washer and nut assembly
Backup slide
Pads are re-inforced
Four nuts seen from inside
24
SCT B3 repair (some details in slide from end
April)

• Started 20/4/04. De-bonded pads being repaired.
• ¼ of services opened at each end. Access is
  possible to all locations.
• Repair details
• Remove 8 brackets/hr/person drill 6
  brackets/hr (3 people)
• Drill in two passes, pad outer skin, then inner
  skin.
• Insert nuts glue 6 brackets/hr (4 people)
• Replace 8 brackets/hr (2 people)
• Remove/replace FSI/DCS
• Test bracket/barrel isolation
• Test harnesses (not yet started)
• As of 1800 30th April 2004,
• 816 holes drilled though pads outer skin (max
  1536) 53
• 728 holes drilled though inner skin (max 1536)
  47.5
• 14 brackets missed due to proximity of FSI 9
  part-drilled 5.9
• 174 brackets repaired (max 384) 45.3
• 135 brackets re-assembled (max 384) 35.2.
• 19 brackets checked electrically (max 384)
  4.9.
• Complete drilling, long nuts, brackets etc. ( 2
  weeks) ? 19/05/04
• Test up to 64 harnesses (almost in parallel) ?
  20/05/04

25
SCT the B4 and B6 harness problem

• However, a new problem was identified end of May
  
• A mistake was made in translating the B3 harness
  design to B6. The rotation required was applied
  around the 1st connector pin and not around the
  module centre. This should have been picked up
  earlier if correct procedures had been followed
  and the detailed design information was fed back
  to the project engineer for checking.
• As a result there is a mismatch of 2.8mm between
  the so-called dogleg and module (see picture).
• The Constraints
• Barrel-6
• The barrel is complete with repaired pads and
  brackets in place.
• The harness production for B6 is complete
• Harness mounting on B6 has started 24 out of 112
  mounted and tested (as of today these are taken
  off again).
• The 6-block sections for B6 cooling are in full
  production. 48/112 6 block sections are
  available. All other parts ordered or available.
  The 6-block sections for B6 are useable on B5.
• Planned date for B6 delivery to Oxford was
  mid-July.
• Barrel-5
• The barrel is complete with repaired pads and
  brackets is sent to RAL.
• The harness production status is that 55 out of
  96 are at RAL and tested. The remainder could be
  ready by end June
• The 6-block sections for B6 will be diverted to
  barrel-5
• Barrel-4
• The barrel is under repair in Geneva and the
  remounting of brackets is now on hold. A few
  weeks work needed to complete it.
• 25 out of 80 harnesses assembled

26
SCT the B4 and B6 harness problem

• Repairs
• Transition PCB needed on the dogleg (i.e harness)
  side
• Use a transition piece which brings the harness
  connector back to the expected position. The
  transition PCB would be double-sided 0.5 - 1.7 mm
  thick and within the dogleg boundaries.
• The process is now being engineered in Taiwan.
  The assembly steps involved
• Remove the dogleg connectors
• Clean the connector pads
• Attach interface PCB
• QA involving thermal cycling to check quality of
  the soldering
• Technical risk
• Establishing the reliability of the new solder
  interface (first results at Oxford encouraging)
• Yield during assembly the need to rework/supply
  extra harnesses in Taiwan
• Impact on schedule
• Estimated date to develop the technique and
  prepare tooling and have first results on dummy
  doglegs is 2 weeks end June. If these are
  successful trials will be carried out on
  individual doglegs for B4. Again, if these are
  successful complete B4 harnesses will be modified
  in Taiwan. If this solution can be developed and
  applied to B6 the start of B6 assembly would be
  delayed by 2-3 months (while B5 will start
  slightly earlier)

27
SCT harness problem contned, and B3 completed

• Other solutions have been studied (mechanical
  shifts, dogleg re-design and re-manufacture,
  bending kaptons without changing connectors but
  by introducing small PCB in between, etc) but
  have either very high schedule or cost impacts,
  or high technical risks, or all of the above ..
  and are currently less favoured.
• Overall impact is serious Since the ATLAS week
  in February B3 has lost 2-3 months (due to
  bracket problem) and B6 in the best case a
  similar amount (due to bracket first and harness
  problem second).
• B5 and B4 are less affected.
• It will take until end July to understand the
  consequences fully when the repair procedures,
  time estimates and yields are proven.
• All this is putting increased pressure on the
  module mounting site at Oxford.
• There the DAQ/DCS/cooling and cleanroom
  facilities are all geared up and the next
  weeks/months will be very critical .

28
SCT EC macro-assembly

• Reminder Disks machined at NIKHEF, disks
  prepared with services at RAL/NIKHEF, modules
  mounted in Liverpool and at NIKHEF
• First disk now at Liverpool, and two more being
  fitted with services currently at RAL.
• Also NIKHEF aim to have their first disk ready
  for module assembly by Aug/Sept.
• First modules on disk first half of July
• Several EC items remain schedule critical
  (modules, services on disks, modules onto disks,
  etc).

29
Pixel status
The rad-hard electronics being out of the list of
the critical parts, we now face the challenge to
produce the large number of modules needed with
the necessary yield and in 15 months. According
to our schedule (in part validated by prototype
mounting of modules on stave and sectors) the
critical part is now the module fabrication and
qualification, not the mounting on the local
support neither the local support fabrication
(that has started long ago). We have qualified
3 sites for the module production (Bonn, Genova,
LBNL) and three more will shortly follow. The
bumping start-up has been a bit longer than hoped
for due to a the need for a key machine repair
(now done) in one of the 2 bump vendors. We
obviously have to also look carefully at all the
mechanical compatibilities between parts coming
from many sources, thats why we plan 2 dry tests
(one in Cern and 1 in LBNL) to check all parts
before the end of the year. Patch panels are
quite complex, design is well underway and all
the necessary tests and mock-ups has been done,
but fabrication should to be followed with
attention. In particular the optoboards on PP0
are close to the critical path and we are only at
the preproduction phase (we did implement a
close follow-up of this item at last steering
group) Optical fibre bundles have been ordered,
cables order will be done soon (need to know the
final length calculation). Lets see now where
we are with the module fabrication (all plots
updated on May 31st).
Disk module
30
Sensors
Production of sensors are now proceeding well,
including new delivery (90 sensor tiles) of On
Semiconductor much better than the 1st one and
now fully tested. A 3rd delivery (300 tiles) is
on their way to the labs and should finally
confirm production quality.
Test labs Dortmund New Mexico, Prague, Udine
31
Front-end electronics
102 8 wafers already delivered (out of 246
ordered) and 54 tested. 48 wafer deliveries are
scheduled each 2 months and therefore expect to
have all in hand before the end of the year.
Testing speed (LBNLBonn) very rapidly at cruise
speed and agrees with schedule
32
Flex hybrids
Two batches of gt1000 flexes delivered and most of
them populated with components.
33
Modules (qualified and ready to go on the
experiment)
The production of pixel modules still in the
learning curve (some sites need qualification,
only one of the two bump vendors started
production). We must work on this issue and have
the curve ramp-up in the next couple of months.
34
When can we load pixel modules on stave/sectors?
The loading site qualification is proceeding well
(with weekly phone meetings). Already proven that
we are able to mount without damage (3 sectors
and 3 staves already mounted and tested and
burned-in and tested again) and that the thermal
interface and the precise positioning is under
control. We are now going through a
cross-calibration between the loading sites to be
sure that the procedures and the reference system
is well understood by all. Shown below are
thresholds and noise measured on 13 modules on a
stave, before loading (flex), after burn-in
(burn) and after loading on stave (stave). All
data are well within specs. Some noise increase
is an artefact of the fitting procedure and is
now being addressed (anyhow still well within
specs).
STAVE Problem with S-curve fit
35
Loading
Shown below is a source scan of a module on a
stave (see components and also some detail of the
flex circuit, but no bump damage due to the
mounting process).
We plan to start loading of production modules to
production stave and sectors before the end of
July.
36
The ATLAS Inner Detector overview