CMSUK ECAL Regional Centre meeting, 26-Jul-1996
CMSUK ECAL Regional Centre meeting, 26-Jul-1996
Minutes of UK CMS/ECAL Regional Centre Meeting; 26 July 1996, RAL
------------------------------------------------------------------------
Present: K Bell (RAL), R Brown (RAL), D Cockerill (RAL),
J Connolly (RAL), G Davies (ICSTM), L Denton (RAL),
P Flower (RAL), R Head (Bristol), H Heath (Bristol),
P Hobson (Brunel), B Kennedy (RAL), L Lintern (RAL),
B Smith (RAL), M Sproston (RAL)
------------------------------------------------------------------------
Minutes: B W Kennedy / RAL
------------------------------------------------------------------------
1) Tasks to be undertaken at Regional Centres and tasks to be
undertaken at CERN.
DC noted that the ECAL design was still nowhere near being finalised,
making planning very difficult. He had attended a meeting at CERN in
June which claimed to be "deciding" the assembly sequence. DC had not
been invited to this meeting, and learned of it by chance. He was not
invited to stay by the participants (Denes, LeBeau, LeCoq, Loos,
Rondeax), but was sent a copy of the minutes, which were in French
(though the official language of CMS is English).
From the minutes, several issues emerged. The current design is based
around a 2*6 submodule, despite unanimity among the engineers that
this unit is not strong enough. A 6*6 unit is felt to be much more
desirable mechanically. Crystals will be tested for size, optical
properties, and decay time. It is at present undecided whether
electronics and cooling services should be distributed to individual
crystals or to a submodule. JC commented that this had to be done at
the submodule level, as it would take too much time (and therefore
money) to work at the level of single crystals. The electronics group
anticipate a single box containing the VFE and FERMI, to be installed
at CERN, not at the RCs. It was not clear how the required manpower
can be arranged at CERN to do this job. The minutes raised the idea
of building a control module for long-term tests outside CMS; this was
felt to be sensible.
JC made three further comments on the ECAL design and construction.
Anything (eg modules) sent from UK to CERN must be fully tested
beforehand, as it will be very difficult to carry out repairs and/or
rebuilds at CERN, for reasons of money, manpower, and management.
The alveolar structure does not provide rigidity, but takes its
structural strength from the rigid backplane. Consequently the VFE
must be built in to the submodule at the RCs - it will not be possible
to "plug in" the VFE at CERN after the submodule is completed.
There are two cooling systems in the ECAL: the electronics generates
heat, but is rather insensitive to temperature; the crystals do not
generate heat, but must be held at constant temperature. These two
systems must be thermally insulated from each other.
2) Mechanical design, and consequences for the UK institutes.
JC described the structure of submodules (2*6 at present), modules (=
baskets) ad supermodules.
Submodule contains: 12 crystals, with wrapping and glued APDs, preamps
and line drivers; temperature monitoring, power and cooling services,
and mechanical fixations. JC noted here that, with 10**5 crystals to
be assembled, a one-minute task performed on each crystal requires one
person-year in total.
Module structure: stiff backplate holding 12*4 submodules (if 2*6
crystals); enclosed in basket; front of basket supports submodules,
and has entry points for optical fibres.
Supermodule structure: services and signal cables connected to outer
(high-eta) end; 4 backplates connected to U-beam support system;
backplate fitted with electronics, power, cooling, and auxiliary
services; connected by U-beam to HCAL.
Assuming the ECAL will be assembled over 5 years by 4 RCs:
100000 crystal /4 => 25000 crystals for UK RC
=> 2100 2*6s
=> 400 2*6s/year
=> 8/week at peak
=> 2/day + 1 day for testing and housekeeping.
Most of the work on a 2*6 submodule is on the rear assembly. The 2*6
approach requires approximately 10* less work than the single-crystal
scheme advocated by the Lyon group.
Staff needs:
1 General manager (including University liaison)
1 floor manager + 2 skilled operators
6 hired staff (3 skilled + 3 general)
4 PPD support staff (database and testing)
Facilities needed at centres:
Secure clean areas
Environmental control
3-4 areas for reception/reception tests/assembly/flatpack testing/
packing/transport bay
DC pointed out that the gluing of APDs is an additional task since
the PPESP proposal, but the total manpower requirement is unchanged.
JC replied that the original estimates allowed some flexibility for
such unforeseen changes as the design evolved.
JC then returned to the mechanical strength of the current design.
There were serious risks: FEA has shown that the 2*6 with 100um walls
is not strong enough, and its present shape would need the
inter-basket gap to be 12-14mm at high eta. The first attempt to
place a loaded basket in the most stressed position (3 o'clock) will
put about 500 crystals at risk. According to the FEA studies, the 6*6
submodule is strong enough, and would give a much safer design.
3) Status report from Bristol.
RH reported that there is competition for space at Bristol. The CMS
group now has about 30m2, and another 80m2 may be available soon if a
strong bid can be made. It is possible that there may be additional
storage areas outside this 80m2. It was agreed that 30m2 was too
small, and the extra 80m2 was needed.
RH then asked a number of questions regarding the facilities needed
for module assembly.
Is a dedicated cooling water system needed ? JC replied that a
complete basket generates about 2kW, and a closed-loop cooling system
will be needed.
What about a clean room ? JC said that a 100000-class clean area will
be good enough, corresponding to a normal clean lab.
Temperature control ? The assembly and testing areas will need to be
temperature-controlled, but there will be no need for special
measures in transit (unless CMS decide to run the ECAL at eg 0
degrees, in which case things get more difficult).
Safety considerations ? The crystals do not present particular
problems, however debris from a broken crystal should be treated as a
lead compound and disposed of appropriately - in particular it should
not be flushed down the sink, to avoid water pollution.
What testing needed during assembly ? JC stressed that we need to
achieve yields of ~98%, so that it was vital to test at every stage
before adding further value. Submodules should be acceptance tested,
and ideally each basket should be continuously monitored during
assembly, and rigorously tested when complete.
PH made the point that faulty 2*6s would not necessarily be
transported to Brunel. Brunel has undertaken to: (a) batch test
crystals and submodules, possible to destruction; (b) manage quality
control and fault-finding procedures. Diagnosis and repairs will in
general be made at the assembly centre, with intervention from Brunel
as required.
JC and PH agreed that systematic batch testing is needed to ensure
that consistent standards are maintained over time. All items should
be routinely acceptance-tested; if an item fails the production line
should be stopped until the problem is understood and, if necessary,
corrected.
4) Status report from Brunel.
PH described Brunel's tasks during ECAL construction.
i) Acceptance testing of 25000 APDs and (pre)amps.
ii) Batch testing (~0.1%) of crystals and APDs
iii) Quality Assurance/Quality Management for UK RC.
JC asked whether Brunel expected to carry out batch tests of eg
completed submodules, in addition to routine production tests. PH
replied that they did, but it was not clear whether such tests would
take place physically at Brunel or at the other centres. DC suggested
that longitudinal scans could be made on a sample of crystals in the
ISIS test beam. PH emphasised again that it was not appropriate for
Brunel to run a repair service. JC remarked that acceptance-testing
should cover only those properties of the crystals which have been
specified to the manufacturer in the contract.
PH said that there is no ready-made "right" solution for QM, but there
are established practices which can be adopted. He suggested running
a "pilot project" not on the RC critical path, such as QA/QM for the
ISIS test beam. In discussion of this idea, JC pointed out that the
vital feature of the ECAL construction is yield, and this is not
typical of industrial experience and not relevant to the test beam
work. DC said that ISIS has the advantage of being accessible to all
of the groups. PH will propose a suitable project at the next
meeting. JC remarked that it would be useful to have a quantitative
model of the production process, to explore the effect of failures.
5) Status report from IC.
GD reported that 90-100m2 of space will be available from the end of
1996, with some attached storage space. It will take about 6 months
to refurbish this area to lab standards, at an estimated cost of
22 kpounds. JC and RH felt that this figure was rather optimistic,
based on experience at RAL and Bristol. It is not known yet where the
funds will come from. CMS can only pay for capital items purchased
specifically for the experiment, but not for improvements to the lab
facilities.
6) ACCOS machine.
GD had news from CERN on the ACCOS machine. The Johanssen machine was
favoured, as it was 200kFf cheaper than the Zeiss system, and offered
similar facilities. The final decision is expected to be made at a
meeting at CERN on 4/9/96, which GD will attend. The order should be
placed by the end of September. The filter system should be ready by
August. Annecy expect to take delivery of the first machine in
December 1996. The basic price will be about 400kFf, with an
estimated total cost including add-ons and manpower of 150-200kSf (nb
Swiss francs). RB commented that the total amount of money
allocated for the UK RC in 1997 was about 200kSf.
GD agreed that he should keep BS informed of developments, as the UK
ACCOS machine will be installed at RAL. PH asked how the ACCOS
machines themselves will be tested. GD replied that no definite plans
had yet been made.
7) Timescale for startup and running of the regional centre.
DC reported that according to Hans Hofer, no production crystals would
be ordered until at least 1997, and production would not start until
at least 1998. The proposed date for pp collisions in LHC is summer
2005. The UK regional centre should be starting up in 1997-1998. JC
and RB were concerned that we should not lose impetus because of the
delay in ordering crystals. The UK RC should be ready to receive
pre-production crystals, for assembly into test-beam prototypes. BS
will devote most of his time from now on to RC planning rather than
the details of the ECAL design.
JC remarked that Ecole Polytechnique and Lyon are not RCs, so the UK
group is in a strong position to influence the design to ensure that
it is actually buildable. A final drawing of the detector is unlikely
before spring 1997, and it is not possible to commit significant
expenditure until we know exactly what we will have to build. He
asked the physicists to begin thinking about a series of tests on the
modules during assembly to make sure that the yield does not fall to
unacceptable levels - this should be addressed at the next meeting.
8) Requisition money needed for the UK Regional Centre
DC showed the money expected to be available for the RC in the next
three years, after prototyping expenditure, assuming that the proposed
funding levels are achieved.
Year 97/98 98/99 99/00
RC Amount (kpounds) 65 240 397
Total funding (kpounds) 500 1030 1200
(Proposed)
The planned breakdown of expenditure for the ECAL and trigger allows
130kpounds for RC infrastructure and 400kpounds for hired staff for
assembly. This looks like a problems given the cost of the ACCOS
machine. RB suggested that some savings might be made on prototyping
in 1998-99 to release more money for the RC. In addition, he asked
that the 340kpounds allowed for prototyping should be split into
separate sums for RC tooling (which appears in CORE) and ral
prototyping (which does not).
RB said that the UK is supposed to pledge a figure for common fund
contributions for 1997 at the next Resources Review Board. After
discussions with George Kalmus he has circulated a draft reply to
Derek Imrie, Geoff Hall, Jim Virdee, and Greg Heath. This confirms
the 150kSf already agreed, but not the extra 300kSf which will be
requested if approval is given for an early start to magnet
construction.
CMS assume that 200kSf will be set aside for the UK RC in 1997, and we
must show that this is possible within the expected funding. The
difference in CERN and UK financial years is helpful here. 35kpounds
will be left in 1996-97 for the RC, which with the 65kpounds allowed
in 1997-98 comes to 100kpounds, or about 200kSf at the permitted
exchange rate to 2.0674Sf/pound.
RB reported a comment from GEorge Kalmus that if the funding proves
inadequate to run a Regional Centre, money invested in it will be lost
to CMS. It is clear though that the ACCOS machine will be needed in
any event, so that expenditure would not be wasted.
9) Computing Resources
HH described the resources needed for the CRISTAL database. A new
document (CRISTAL Prototype II) will soon be available describing
CRISTAL in detail.
The central database wil be held at CERN. Regional Centres must run
the CRISTAL software, and be able to store enough data for their own
work. The minimum specification for a machine to run CRISTAL are 65MB
or RAM and 2.4GB of disc space.
The UK RC will be distributed, which causes complications. There are
three options.
a) Each centre runs CRISTAL - extra cost
b) One CRISTAL machine with network access from elsewhere
- possible risk from network failures.
c) CRISTAL software treats centres differently - needs extra coding
The computer supplied with the ACCOS machine is assumed to be powerful
enough to run CRISTAL - in principle it should run on most UNIX
machines, but not all platforms can be supported. Several licences
will be needed.
o O2 or Objectivity - CERN will negotiate a common price
o ORBIX - run-time licence only, about $100.
o WEB browser - free to Universities, expensive for RAL.
It was agreed that the ACCOS computer will be used initially as the
CRISTAL server (option (B) above). If problems arise an extra server
could be bought later (expected cost about 15kpounds).
Jean-Pierre Vialle has been investigating light-pens for reading
barcodes to identify parts, operators, etc. Several pens would be
needed per centre (at ~100 pounds each), with PCs to run the software.
AOB)
RB raised the question of the secretary for CMS UK management
committee. In previous experiments, this job had been undertaken by
members of PPD, but there was no reason in principle why the secretary
should not come from a University, and volunteers were invited. GD,
PH, and HH will consult their colleagues. Volunteers were invited to
contact DC, as RB will be on holiday until mid-August.
Automatic conversion from plain text file