QC and Process Control

1
QC and Process Control

• G. D. Patel

2
QA

• There exist various QA criteria for the different
  components that go into the manufacture of a VELO
  module plus a well defined procedure for their
  processing.
• My role is to ensure that the selection criteria
  are correctly applied and the procedures
  followed. To aid in this process I have designed,
  written and implemented a web based database
  system that holds all the information relating to
  all the components.
• All components are entered into the database upon
  arrival and the subsequent procedures in the
  processing of each component is defined by the
  database.

3
Database Components
The database is designed to store all relevant
information on the LHCb Velo detector in a
central, web accessible location. It is
implemented with an Apache web server running
with a php interface to a MySQL database. Data
input is restricted by password protection, but
information retrieval is unrestricted. Informatio
n is stored on cables, chips, pitch adaptors,
sensors, pedestals, substrates, hybrids and
modules. All changes made to the database are
logged and can be traced back to a specific
user. All test data are stored and online
analysis is possible.
4
Cables

• Visual inspection and log in database.
• Send for population.
• Electrical tests.
• PASS/FAIL logged in database.

5
Chips

• Arrive in 11x11 waffle boxes, position in box
  identifies chip.
• Maximum of 120 chips, but gaps indicate rejected
  chips.
• Used chips have a link to the module on which
  they reside.
• Link to original chip test data at Max Planck web
  site is provided..

6
Pitch Adaptor

• Pre-selection at CERN.
• Visual inspection and log in database
• Must be no breaks or shorts in any of the 512
  strips.
• Detected by a capacitance measurement of every
  strip.
• Data logged in the database.
• Algorithmic code implemented via the web server
  informs the operator if the pitch adaptor meets
  the QA requirements.

7
Web based code can be run before the data is
logged in the database. This allows for
re-measurements on the probe station before
logging the final data.
Problem channels identified and classified as
shorts or breaks.
PASS/FAIL indicated
Data presented graphically
8
Pitch Adaptor DB Summary
9
Sensor

• Metrology
• Must be no damage and shape must be with in
  limits of nominal.
• Shape metrology measurements from the Smartscope
  are logged in the database.
• Algorithmic code implemented via the web server
  informs the operator if the sensor shape meets
  the QA requirements.

10
Shape measurements analysed and compared with
nominal shape. Any edges oversize or undersize
outside of the specified tolerance are flagged
and a software PASS or FAIL issued. Final
decision on accept or reject taken by me after
discussions.
Software decision
Oversize
Undersize
11
Sensor

• IV Currents
• Total current at maximum volts must be less than
  100 µA.
• Data logged in database in text format.
• Code displays maximum current in red on summary
  page if over limit.

12
Sensor

• Bad strips
• Measure capacitance per strip, must be within 20
  of expected value.
• Measure current per strip, must be less than 1
  µA.
• Algorithmic code implemented via the web server
  analyses the raw data from the probe-station and
  informs the operator of the numbers of strips
  which fail the capacitance and current criteria.

13
Summary of numbers of channels that failed
(capacitance or current tolerance)

• Problem channels identified.
• Failed channels in red
• Warnings in green.
• This analysis available before
• the data is logged in the database

Graphjcal presentations of data
14
Sensor

• All the test information is displayed on the
  sensor summary page.
• Physicist must look at the sensor summary page
  and the data and make a QA decision on the final
  status of the detector.
• Acceptable sensors are categorised as OK, GOOD
  or VERY GOOD.
• I inform Micron if a sensor is to be rejected.

15
Sensor DB summary
Sensor thickness Type Hybrid IV CV CAP/S TCUR/S CURV CONTOR Status QA Flag Update
2440-10D 300/303 r-04 x/ 0- 1 0 Update
2439-23D 300/302 r-04 x/ 6- 1 0 Update
2436-18A 300/305 phi-04 x/ 14- 1 0 Update
2439-20E 300/302 r-04 x/ 0- 138 -16 3 0 Update
2439-06A 300/299 phi-04 / 2- 107 -5 3 0 Update
2439-19E 300/298 r-04 x/2.42 20 2-2 0 121 -15 6 0 Update
2439-11C 300/302 phi-04 x/ 2- 111 0/1 -3 0 Update
2439-09A 300/302 phi-04 x/ 2- 97 0/2 -3 0 Update
2439-04C 300/301 phi-04 x/2.40 20 5-5 0 137 -9 6 9 Update
2436-05E 300/304 r-04 x/2.92 50 1-1 0 134 -9 6 9 Update
2436-05D 300/304 r-04 x/4.10 50 1-1 0 141 -12 6 9 Update
2392-16D 300/312 r-04 x/12.89 70 18-18 0 107 -7 6 7 Update
2392-02B 300/307 phi-04 x/47.42 60 6-6 0 105 -5 6 8 Update
2392-02A 300/308 phi-04 x/1.94 60 0-0 0 100 -2 6 9 Update
2391-24D 300/309 r-04 x/4.69 70 2-29 3 122 2 -5 -5 Update
2391-04E 300/306 r-04 x/14.78 60 3-3 0 110 -2 6 8 Update
2313-07B 300/311 phi-04 x/264.00 60 13-4 1 106 20/0 6 -10 Update
2317-11D 200/201 r-04 x/11.09 30 7-10 224 1/1 -3 7 Update
2313-07A 300/310 phi-04 p-07 x/9.00 80 21-13 365 19/0 7 100 Update
16
Sensor DB Summary
17
Substrate

• TPG and CF laminated in-house.
• Metrology data logged in database
• Sent to Stevenage for circuit to be laminated on
  the two side.
• Logged in the database on return

18
HybridStages 1-4

• Metrology measurements to determine curvature and
  bowing.
• Those out of tolerance are heat treated between
  metal plates.
• Additional metrology measurements on the heat
  treated hybrids.
• Hybrids with acceptable QA are sent out for
  population.
• Rest are rejected

19
HybridStages 5-7

• Populated hybrids are electrically tested to
  ensure specified functionality.
• Electrical test failures are rejected if they
  cannot be repaired.
• The rest are cleaned and inspected.
• Additional metrology measurements on those that
  pass electrical tests.
• Those that are too far from flatness required are
  rejected.

20
HybridStage 8

• The hybrid is now ready for the gluing of pitch
  adaptors and chips.
• The database has a hold at this point and no
  further work is allowed unless specifically
  cleared to proceed by Tony Smith or Themis Bowcock

21
HybridStages 8-9

• Batches of pitch adaptors and chips to be used
  taken from store.
• Glue pitch adaptors to R side.
• Visual inspection and fix problems.
• Glue chips to P side.
• Visual inspection and fix problems.
• All pitch adaptors and chips used are logged in
  the database

22
HybridStages 10-11

• Glue pitch adaptors to P side.
• Visual inspection and fix problems.
• Glue chips to P side.
• Visual inspection and fix problems.
• All pitch adaptors and chips used are logged in
  the database

23
HybridStages 12-16

• Bond back ends of R side chips.
• Visual inspection, repair problem bonds and log
  in DB.
• Bond back ends of Phi side chips.
• Visual inspection, repair problem bonds and log
  in DB.
• Bare chip test data to electrically test that all
  chips are working. Results via web database
  interface.
• Replace any problem chips and repeat the cycle
  till all chips OK.

24
(No Transcript)
25
HybridStages 17-19

• Front end bond R side chips to pitch adaptors.
• Visual inspection, repair problem bonds and log
  in DB.
• Front end bond Phi side chips to pitch adaptors.
• Visual inspection, repair problem bonds and log
  in DB.
• Bare chip test data to electrically test that all
  bonds are working. Repair if necessary.

26
HybridStage 20

• The hybrid is now ready for the gluing of
  sensors.
• The database has a hold at this point and no
  further work is allowed unless specifically
  cleared to proceed by Tony Smith or Themis Bowcock

27
HybridStages 20-25

• Glue R and Phi sensors to hybrid. Log sensors
  used in the database.
• Metrology measurements to confirm the alignment
  of the sensors.
• Make bias bonds and check IV for the sensors.
• Bond R and Phi sensors to pitch adaptors.
• Quality of ALL bonds are logged and database
  informs operator of known problem channels on
  the sensors which are NOT to be bonded in terms
  of the bond numbers programmed in the bonding
  machine .

28
Information on known problem channels on the
sensor are presented to the bonding machine
operator. The bonds which are NOT to be made in
the current pass are shown in RED
Rest of this page allows status of every bond in
this current pass to be entered and logged in the
database.
29
HybridStages 20-25

• Laser tests on all 4048 channels of R and Phi
  sensors.
• Correlate problem channels detected by laser scan
  with known bad channels and problem bonds.
• Investigate anomalies and resolve.

30
HybridStages 26-32

• Glue complete hybrid to pedestal.
• The final module number is defined by the number
  engraved on the invar feet of the pedestal base.
• Full module metrology.
• Plug and clamp cables.
• Burn-in and thermographs
• Final metrology.
• Package for shipping.

31
OutstandingTasks

• A further selection of chips within the ones we
  receive may be required. Plan to implement web
  input so that Aldo Saavedra can input this
  information directly into the database.
• Front end bonding has recently been implemented
  on the 710 as a backup to the 8090. The bonding
  sequence and bond wire numbering is different and
  a separate input page for bonding problems has to
  be implemented and both machines supported.
• Algorithmic analysis of laser test data and cross
  correlation with known problem channels is still
  to be implemented.
• Analysis of burn-in data still to be implemented.
• Database software and data store to be
  transferred to a more powerful server machine.

32
Conclusion
The web based database system has proved
invaluable for the logging and tracking of all
the many components and their QA test data and
results which are required for the VELO
project. I am sure that the inbuilt process
control will be an essential tool for the
management of the production phase giving a
constantly updated overview of the status of
every module. Additional features will be added
that give numerical summaries and provide lists
of known bad channels for every module for input
to the conditions database