leonardo rossi – atlas plenary- stockholm 10/07/061 inner detector: highlights and concerns a lot...

Leonardo Rossi – ATLAS plenary- Stockholm 10/07/06 1

Inner Detector: highlights and concerns A lot of progress on SCT & TRT integration. Pixel production and disk integration is well advanced. Barrel

integration suffers from low mass cables cracks. Services installation has greatly improved but still on critical

path.

Concerns in Feb06. and now: TRT

SCT

Pixel Stave cooling pipe corrosion

IDGEN Service installation

TRT

SCT

Pixel Low mass (type0) Al cable

cracks

IDGEN Service installation


ID Barrel integration

Barrel SCT

Barrel TRT

Went smoothly (1mm clearance), then connecting all services took ~5 weeks (~1/8 TRT (=9 RODs) and 1/4 SCT (=12 RODs)) + ~2w to check and adjust all and bring to life.

Very important test to verify the delicate connection process and the overall operation (up to data taking and analysis of cosmics run)


Aim of the combined test in SR1

The operational aspects: Gain experience with detector operation Test combined detector supply systems

Development of standalone & combined monitoring tools Commission and test combined readout and trigger

Commission offline SW chain with real data

The detector performance aspects: TRT performance with SCT inserted and powered

Test 4 SCT barrels together and operation with TRT Checks of grounding for SCT and TRT

Test synchronous operation and check for X-talk and noise Collect cosmics for efficiency, alignment & tracking studies


TRT & SCT see cosmics, they work together without any noise increase

~450k cosmic events taken (livetime 57%)

Efficiencies as expected, but for cabling error in 2 modules

Few highlights (details in Maria Costa talk tomorrow)


A hic-up

During the combined test, on May 3rd, we had a failure of the liquid cooling for the TRT electronics while running unattended. TRT DCS failed to switch off the upper TRT sector Power Supplies after a “missing coolant” alert the temperature on some on-detector read-out chips went up to 120C for hours.

One wire broke Luckily no other degradation of performance has been

observed after restoring normal conditions (indeed the results shown are obtained after this incident). But the effects could have been much worse.

We then went through a review process, see: http://indico.cern.ch/conferenceDisplay.py?confId=a062480 to evaluate changes and precautions to avoid similar incidents to happen again and we are now implementing the recommendations of the reviewers (including overheat interlock).

http://indico.cern.ch/conferenceDisplay.py?confId=a062480


ID Barrel installation in the pit

Now ID Barrel is already un-cabled and should be ready for pit installation on August 14

Installation coordinators have been named (Apsimon for SCT and Romaniouk for TRT). They have defined a detailed set of Work Packages for installation, connections and sanity checks up to the acceptance sign-off.

The installation process can only begin when all the services for the ID barrel will be tested and the pipes also cleaned. It foresees 5w for the TRT (connect and test) followed by 11w for the SCT, including 2w at the end for final tests (enough? sign-off means EC will be installed and then Barrel inaccessibility for long…).

Prior to the ID Barrel installation there will be a full transport + insertion test (from SR1 to cryostat) to be done with a dummy load on a transport trolley similar to the one to be used for IDB.

Plus the solenoidal B-field measurement


Inner cryostat equipped with services

Waiting for the ID Barrel, services have been installed. This is now quite advanced (3/4 done) and the work inside the cryostat is expected to ~end by mid July (when B-field measurement should begin)

Other activities not needing access to cryostat bore (e.g. pipe connection to racks) will continue.

Artwork done under enormous schedule pressure


Pipework from Z=0 to racks

~1200 pipes (~5000 connections, most crimping plus 800 welds)

Must increase the speed to meet IDB installation (end August)

tested pipes

0

200

400

600

800

1000

10-Mar 9-Apr 9-May 8-Jun 8-Jul 7-Aug 6-Sep 6-Oct

total # 1034300 pipes in 20 days

14/7

9/8

9/9


Solenoidal B-field measurement (see W. Kozanecki talk)

Has to be done before ID Barrel installation, 2w (last of July+1st of August) and interferes with services installation.

Machine ready and installed in parking position inside cryostat.


Evaporative cooling system Just one (maybe the most complex) of the

ancillary systems necessary to operate ID Barrel (and later EC and Pixel).

Main cooling plant and control

PLC

Heater control

Heater power supplies USA.L2

Heater power supplies US.L2

Distribution racks x 4

US15

USA15 UX15 platforms

On-detector loops capillaries

and HEX’s

heaters

Loop-Back


Loop-back tests are foreseen in the cryostat prior to ID Barrel installation to verify (+clean) the circuits.


Liquid 16BarVapour 2Bar

Heate

r loa

ds x

5

TE bulkhead

Loop-back tests are foreseen in the cryostat prior to ID Barrel installation to verify (+clean) the circuits. 5 full circuits with heat loads (mimicking detector), simplified for all (clean + test).


EC integration - SCT

ECC arrived at CERN on Feb 23 and completed acceptance tests on March 24 (500 points visual inspected, continuity tests, leak tests,…)

Cont. tests found 50 new disconnects due to unplug at the Temporary Interconnect, which required repair (now partially done).

April 10 moved on cantilever tool and during May it was fixed into its “integration” position


June: the Inner Thermal Enclosure was inserted (min clearance 2mm) sealed and grounded

Then the front membranes were added (separate sheets for gas and grounding)

Tests from mid-July to Aug 4th then integration with TRT begins and plan to be ready for combined tests from early September to early November

RFI: beginning of December


ECA arrived at CERN on Apr 21st and completed acceptance tests including all disk running cold on June 22nd


Number of dead channels remains constant and at the 0.2% level (typical value for both Barrel and EC SCT).

Nr of dead channels

0

200

400

1A 2A 3A 4A 5A 6A 7A 8A 9A

disk number

total dead channelsdisktestbox

total dead channelscylinder

total dead channelsCERN

VCSEL failure is not negligible, 4% of the modules is readout through the redundant scheme (1 link out of 2 working). This number is stable

Integration steps of ECA follows ECC with ~1 month delay (now getting shorter).


TRT EC

ECC Harnesses installed and under testing. Active gas and cooling running. Faraday shields installed, CO2 safety valves in place. 0.5% dead channels.

Moved to ID trolley 28.06.

First tests with self trigger allows to see cosmics, complete SCT & TRT test will follow from September


ECA FE electronics installed and tested. Active gas leak tightness OK, single cooling circuits leak tested: OK, 0.3% bad channels.

HV tests ~OK problems with 2 wheel sectors,

repair once on integration trolley

Integration with SCT should follow (for both TRT EC) according to SCT schedule.

ECC combined test in SR1 test area, ECA will follow

N.B.: 1 broken fuse found for EC A after ~1 year(out of 15000); replaced


Pixel progress

The Pixel End Cap integration is proceeding well.

PixelECC at CERN and passed the acceptance (leak check and connectivity) tests. Dead channels at few per mille level.

PixelECA ready for transport. Still need some repair at CERN (problems with 3 modules)

We are preparing for cosmic tests with PECC to:

test whole DAQ chain commission a lot of service SW,

from DCS to heavy usage of CondDB and survey data.

and offline SW (detector description, track reconstruction,...)


In practice


Barrel progress Module production is done

IZM part finished with 97% yield after rework

AMS part should end in a couple of weeks with ~90% yield

1744 modules are needed and we’ll have ~2300 good bare modules. This allows to produce more staves to minimize # of “inserted”.

Stave cooling pipe corrosion has been solved using three paths Change the pipe (for staves not yet loaded with modules) Insert a smaller pipe (for already loaded stave) Build new staves (to compensate for losses in the above)

all progressing well


Stave loading

61 staves with the new pipe already loaded, 25 still to go (should end in September).

35 of the 42 staves already loaded have now inserted pipe and 13 are qualified (the rest is ongoing, a couple of staves have been “lost”)

Loaded Staves

y = 0.6424x - 24928

y = 0.3792x - 14651

0

20

40

60

80

100

120

01/08/04 09/11/04 17/02/05 28/05/05 05/09/05 14/12/05 24/03/06 02/07/06 10/10/06Date

No.

of

Sta

ves

Loaded Staves old Pipes

Original half-shell need

Loaded Staves new Pipe

Trend

Linear (Trend)

Linear (Loaded Staves newPipe) 2.7 Staves/Week

4.5 Staves/Week

Strategy is to have mixed bi-staves in Layer2 (less cooling needed after irradiation) and non-inserted elsewhere. Therefore only 26 “inserted” are needed (and extra modules become necessary).


Also creating a pool of the “best 30 staves” to be devoted to the B-layer (=22 staves) which is more critical for precise vertexing.

Barrel integration started in late April and 2 bi-staves were loaded on their support frame (for layer 2).

Tests showed that two low mass (Al) cable were broken, this stopped integration and started investigations on the possible causes.


The low mass (or type0) cable problem

The barrel Al cables are made of 100 m wire for signal and 300 m for power (21 wires in a bundle), both wire-bonded on a small printed board + mini connector (~1m, modulePP0).

The initial interpretation was that the cable was stressed during manipulation and strain relief and better practice should solve the problem.

Visual inspection and electrical tests after stress on all production started and found that ~50% of the 2000 cable produced are bad or likely to evolve bad.

Indeed the insulation was found to have cracks and, being the insulator much stronger than a 100 m Al wire, any stress would concentrate on the wire thus breaking it (sooner or later), this is not true for the thicker power wires.


The statistics of the faults is: Green = good Yellow = broken insulation but only

on 1 or 2 power wires Red = broken insulation on signal wires Black = electrical failure

Then the initial interpretation true but partial: some cables are already damaged in the production process (that’s why we did not see with home made prototypes...). Unfortunately our QA did not spot early this problem.

The problem shows-up just after a heat shrink (example)

Al shines through insulator crack


Last week visit to firm in Taiwan of our experts analyzed production process and found ways of reproducing the problem

Wires are bent and then immersed in 400C NaOH for stripping. Excessive bend and high T can damage the insulator New production process should give 50 cables this week and

then 130/w (1000 cables to produce ~ 2 months). After analysis of the 50 cables we’ll already know if the

problem is solved, but we’ll continue detailed QA (including visual inspection, which is very manpower consuming)


We already restarted integration of bi-staves last week with the green+yellow cable set and we are now back to the mid-April time frame…

i.e. we have lost ~3 mo in the already critical barrel integration process.

We’ll than have a TC review on July 18 (chair A. Clark) to: Evaluate if the technical solutions are ok Evaluate what can be done to speed-up the schedule (without

risk)

Once the work packages will be defined there may be a call for help to non-pixel groups in case this is deemed necessary to meet the April 2007 installation date (preliminary analysis indicate there is no time for pixel installation in the 2008 shut down).


Some about software: New Tracking progress with v.12

σ(1/pT)σ(d0) sinΘ x σ(z0)

Resolutions as a function of || (single , 100 GeV/c):

New Tracking improvement from 11.0.41 to 12.0.0


NewTracking now at level of other trackers

Total multiplicity per event (P>1 GeV):xkalman iPatRec newTracking40.84 39.24 37.13

Primary track efficiency / fake rate :xkalman iPatRec newTracking

Barrel 98% / 4.1% 98% / 0.4% 98% / 0.3%Transition 99% / 5.2% 99% / 0.3% 99% / 0.9%Forward 98% / 4.2% 99% / 1.6% 98% / 1.9%

Number of hits per track Pixel/SCT/TRT :xkalman iPatRec newTracking

Barrel 2.75/8.07/26.26 2.81/7.92/26.242.94/8.06/26.83

Transition 2.80/8.08/24.032.89/7.90/24.942.99/8.00/23.92

Forward 3.03/8.68/12.683.03/8.31/13.023.34/8.57/12.78


Conclusions

ID is progressing well toward integration and installation, but the next weeks and months are very critical as we have a very condensed installation schedule.

The ID barrel installation will be a major milestone as it should also include bringing to life and safely operate complex ancillary systems (evaporative cooling, PS system, gas systems) which have not been commissioned yet. The experience in SR1 is positive, but problems cannot be excluded.

Pixel Barrel has a schedule risk that we are trying to minimize


….but we finally hope to succeed…

…and keep the ATLAS logo high

leonardo rossi – atlas plenary- stockholm 10/07/061 inner detector: highlights and concerns a lot...

Documents

sct trt integration

id barrel installation

trt performance

tomorrowtrt sct

sct barrels

atlas plenary

combined test

trt electronics