SUMMARY FROM THE SVT WORKSHOP

1
SUMMARY FROM THE SVT WORKSHOP

• CDF Collaboration Meeting
• Stefano Belforte
• for the SVT group

2
The 1st SVT Workshop

• Goals
• finalize hardware design
• review finalize error handling protocol
• prepare for next year test (vertical slice)
• review mechanical alignment issues
• review where we stand on patterns issues
• what goes in the AM
• what do the Track Fitters do
• a dozen SVT workers attended in person 5 on
  video
• managers, spokespersons
• other people too (perspective SVTers, physics
  convenors, trigger experts)

3
Hardware Status Schedule

• All SVT boards (8 different kinds) are now fully
  specified.
• Few small details will be adjusted as
  implementation allows
• None is waiting external input, boards design
  rolls to completion by end of year
• Pre-production prototypes have started arriving.
• Will have one of each board by March.
• XTRP fanout will come later (not a problem)
• Production will start in June, installation
  completed by end of year
• 3 month window in March-June to assemble 1/12th
  of SVT and test full operations before production
  start
• vertical (phi) slice test
• not just a test a step in a continuos process to
  build full SVT hardware and software and
  integrate it in the Trigger/DAQ

4
Vertical Slice Test

• Board debug at home in Jan/Feb
• Vertical slice test is mainly infrastructure test
• we have now
• schedule, coordinators, list of tasks, estimate
  of manpower, plan for software development, more
  already done software then it looked, green light
  for crates, power, cpus etc., confidence of
  meeting the goals
• Goals
• Build a team of SVT experts (not just single
  boards)
• Extensive and intensive tests of the hardware
  (look for subtle bugs)
• Connect with SVX readout and Level 2 (XFT later
  on)
• Interface with R_C and demonstrate the SVT
  startup, control, error-handling and recovery
  procedures

5
Error Handling

• Years long debate. We now have a final agreement.
• Two kind of errors
• SVT_Error for internal use. Details spared.
• CDF_Error for error that requires full DAQ
  re-sync
• one month ago SVT does not pull CDF_Error, just
  passes error flags to Level2 in the data stream
• now each SVT board will pull CDF_Error as soon
  as it detects a severe error condition (if it
  can)

6
Mechanical Alignment

• SVX builders could not come, but William Trishuck
  managed to put together a talk from Canada that
  Giovanni showed us on video from Pisa ! Thanks to
  both for the effort !
• He reviewed April alignment fest and October
  special topic meeting
• SVT spec everything (including the beam) line
  up to 100 ?rad
• situation as hardware is being finalized
• SVX mechanics should be OK
• COT can not be moved, 300 ?rad expected at
  installation, may be OK but more studies are
  needed
• SVX adjustment (inchworms) is promising, wait for
  more tests with full mechanical weight
• potential worry 100?rad is a maximum range, not
  RMS
• but looks under control

7
Beam Alignment

• It was NOT a workshop topic
• Still surfaces every time
• Liason with Beam Division must be strengthened,
  agreement spelled out (verba volant, scripta
  manent)
• Our plan
• we will provide real time feedback on beam angle
  position
• we expect the beam to be put where we want it at
  each store
• we will not adjust dynamically to beam movement
• we expect the beam to be kept still in place
• The SVT hardware will NOT be able to adjust in
  real time to beam center movement

8
Increasing SVT efficiency

• It is has become a general request, but vague.
• Clear statements are welcome When ? How much ?
  Why ?
• Two sources for inefficiency
• incomplete pattern coverage (AM has a finite
  size)
• SVX single layer efficiency 0.9 0.95
• Two basic rules agreeded upon at the workshop
• we do not deal with 3-hit tracks
• if a track has 5 hits, use only 4 in the fit
• Solution require 4/5 SVX layers
• two methods possible (details spared )
• detailed comparison in progress to pick best
  approach (not hardwired anyhow)

9
How to do Majority Logic in SVX (4/5)

• Not a straightforward issue. Still we take it
  seriously.
• Several people in SVT are working on it.
• 4/4 always possible (baseline from TDR, well
  suited for B triggers)
• Going to 4/5
• can we do 4/5 ? Yes
• do we need hardware changes ? No
• is there a price tag attached ? Yes
• what is it ? Level 2 Time
• why ? Replicates (roads fits)
• how bad is it, and what should we do ? Next
  slide

10
SVT efficiency with 4/5

• Target (efficiency 1) is to find in SVT all
  tracks with 4 hits in SVXII
• This plot is for 20 GeV ISAJET bbar simulation

11
Track Replicates

• Same 4-hit combination may appear in more roads
  (up to 1.4)
• One 5-hit track fires up to five 4-hit
  combinations
• What to do with track copies ? Maybe Level2 can
  not handle them, so we need to build more SVT
  hardware...
• How many clones we have ? How long does it take
  to deal with them ?
• What will Level2 processor have to do with these
  tracks ?
• Cut on d ? Easy and fast.
• Find vertex ? Jet-probability algorithm ? How
  long does it take ?
• It all depends a lot on the kind of event (20Gev
  jet ? 2-track ?)
• Time to think in terms of integrated triggers, L1
  SVT L2
• if SVT takes longer ? less time for Level2
• proposals for triggers requiring higher SVT
  efficiency should include understanding of what
  to do in Level 2 processors

12
Ways out of 4/5 troubles

• Decide we do not need it. Back t o 4/4 (the SVT
  TDR)
• Take the hit Level 2 will have to wait.
• Limit 4/5 to high Pt tracks (e.g. displaced high
  Pt lepton for exotic) ?
• Threshold will be adjustable at any time (4 GeV
  looks a good one)
• Work in progress within SVT, but needs input from
  collaboration.
• Decide what to do according to which Level 1
  trigger fired.
• Looks interesting
• Too bad we have no way of doing now
• Not clear if it may ever be possible
• But Necessity is the mother of invention

13
Online software

• Various tasks needed to properly monitor the SVT
  operation have been listed. The list is long and
  will grow.
• Some will require standard monitoring a la
  TRIGMON
• Some will require our own special DAQ.
• Likely dedicated computing hardware as well
• The SVT workstation
• Definitely a lot of work
• Interface with Run_Control is starting to be
  addressed
• DataBase requirements are being layed down (looks
  OK)
• You will hear more about this as time goes by

14
online

• Here I will show the two cartoons with the
  hardware mountain and the software ocean

15
Conclusions

• It was a very intense meeting
• Some very positive results
• hardware finalized
• schedule organization for next year test
• start acting as a single group
• beginning to grasp how to handle 4/5
• throw back work to do to the collaboration
• start looking beyond hardware building
• Next step
• integrate more the various institutions
• put in place a structure to deal with common
  issues
• tackle software more aggressively
• add more collaborators