Nessun titolo diapositiva

1
Software integration aspects for the
commissioning of Atlas m spectrometer
S. Miscetti et al. (INFN-LNF)
LNF Atlas sw meeting LNF 17-Oct-2005
2
Commissioning phase with Cosmics in situ

• In the next months a lot of activity will be
  devoted not only to the physical
  
• installation of the chambers but also to the
  not trivial effort in
  
• - debugging all hardware a.s.a.p.
• - defining a standard running time and running
  types for DAQ in USA15
  
• - debugging acquisition with the first set of
  final MRODs
  
• - defining a first functioning trigger scheme
• - defining a first proto-calibration scheme
• - trying a first offline reconstruction on the
  data
  
• It is nice to say that sector 13 will be (it is
  ) the first place
  
• where we can exercize, learn how to do all of
  this and make a real step
  
• forward for the commissioning of the whole
  detector.
  
• I will just describe here few starting comments
  of what has been done so
  
• far and where we can / should collaborate /
  integrate with other people.

3
RC implementation used today ..

• In the following I will recall the different
  places where we can readout
  
• the MDTs (and/or) the RPCs (at least so far
  )
  
• At all times, we used the so called ROD crate
  DAQ developed by E.Pasqualucci
  
• which is able to read out many MRODs adding
  then the missing ROB, ROS and
  
• EB formats.
• The different places we are working on are
• SX1 to debug with noise-pulsing the chambers
  already tested in BB5
  
• i) MDT CSM readout via GOLA card
• ii) RPC pulsing test under
  implementation (no gas flowing here)
  
• UX5 at the bottom of ATLAS pit. Now fully
  dedicated to RPCs
  
• USA15 the place where final RODs will be
  located. It is at the level of the LHC
  
• pipe. Here most of the readout boards should
  converge.
  
• For what I have understood the DCS HV main
  readout should also go here.
  

4
Noise and pulsing tests

• Noise runs are easily implemented by 10 kHz
  free-pulsing with a NIM oscillator.
  
• Pulse tests for MDT have been implemented in July
  2005 by
  
• sending a 3-5 V, 20 ns wide pulse in the HV
  chain.
  
• (1) Best performance seen at 3 V. When using 5
  V a lot of x-talk present.
  
• (2) Other pulse tests tried at ROMA3
• (3) Other pulse tests possible by manipulation
  of TTC directly at the
  
• TDC level.
• SX1 pulse of kind (1) needed only to check
  MDT status
  
• However tests of dependence on T vs Q possible
  together with a check of
  
• the scale calibration of the TDC.
• For above reasons and for the possibility of
  being used in the final setting ..
  
• the pulse technique number (3) much more
  promising to check T0 stability
  
• (TO BE TRIED! Implementation possible in SX1,
  USA15)
  

5
Noise and pulsing tests
From P.Branchini presentation at Roma3 .
In Atlas sara possibile utilizzando il sistema
TTC e programmando opportunamente lo stream di jt
ag. In laboratorio a Roma non abbiamo il sistem
a presente al CERN ed abbiamo un csm0 che consent
e di programmare i mezzanini utilizzando Jtag ma
non distribuisce impulsi ai mezzanini.
Di conseguenza a prima vista non sembrerebbe pos
sibile mettere su un sistema di impulsaggio con
il daq che ce a Roma3. Quindi lo abbiamo fat
to.
6
Reading out more than a CSM link

• There is a lot of confusions when accessing data
  for the terminology used.
  
• The CSM is the Service Module of the MDT where
  a TTC fibers
  
• goes in input to provide the trigger and an
  optical fiber of readout
  
• transfer the data over threshold from the
  TDCs.
  
• However in the MDT partitioning scheme
  (ATLAS-DAQ-2003-023)
  
• there is continuous reference to CSM link
  number indicating the position in the MROD where
  the optical readout link of a given CSM is
  inserted.
• The MROD itself is identified by the SourceId
  word.
  
• The SourceID in Hexadecimal is divided in 4
  fields
  
• Reserved Word ModuleType SubDetector
  ModuleId
  
• ModuleType 0,1,2 identify the ROD, ROB, ROS
  fragment
  
• SubDetector define Barrel or others (eg 61
  MDT barrel)
  
• ModuleID is two (three) bytes ex. 09, 0A ..
  Identifying the crate and the board
  
• Position in the crate. For the moment being we
  just use the slot position

7
Reading out more than a CSM link

• In USA 15 we have readout 3 chambers connected to
  the same MROD board
  
• in three different MROD inputs.
• By dumping the data we find MRODtype x 6109
• The words readout from the different readout
  links from the CSM are collected as a list with a
  BOL (Beginning of link) and EOL (end of link)
  header and trailers. In this way a single CSM (ie
  an MDT) is selected.
• For each CSM the TDC words are then stored as a
  timing list connected to a given TdcId and
  ChanId.
  
• For a given MROD there are from 6 to 8 CSM
  optical links.
  
• For a given CSM link the number of TDC is
  related to the number of mezzanine cards.
  
• In USA 15 we have 2 MRODs with 6 CSM links per
  board.
  
• In total we can read out (at the moment) up to 12
  csm links or 12 MDT chambers

8
Cosmic running with Scintillators

• While the RPC system is still not gas-flushable
  we have organized
  
• a first readout in USA15 by setting a simple
  coincidence of two
  
• scintillators. S1, S2 each of dimensions 0.2x3
  m2
  
• Each scintillator is readout by two PMs (A,B).
  We plateau the PMs in SX1.
  
• We make them light-tight and we test the Rates
  of (S1a S1b) , (S2a S2b)
  
• and of the 4-fold coincidence when overlapping
  them along the longitudinal
  
• dimension.
• We measure 100 HZ , 30 HZ for the two-fold ,
  4-fold coincidence as expected.
  
• When in UX5 we measured 30 HZ , 1 HZ for the
  same configuration.
  
• We then set a 125 m long RG58 cable to bring the
  trigger from UX5 to USA15.
  
• We finally correct for the delay by setting the
  correct timing in the pipeline
  
• readout in the mezzanine cards.
• We took data with cosmics both in the pit or
  from USA 15 reading out
  
• BML1(2)(3)A13.

9
GNAM vs MDT2ntu

• When trying to analyze these data we found that
  the offline tools existing at
  
• BB5 were not sufficient. Indeed also at SX1 we
  were able to analize with
  
• MDThistos by S.Zimmerman only one chamber at
  the time.
  
• We discussed with Fabio/Ludo and all other LNF
  people how to improve this.
  
• The first available and easy solution was to
  use GNAM
  
• GNAM is able to read data from memory when
  RodDaqCRate runs and also
  
• to read them from a file using a file_server
• The connected MDT-library is already well
  developed and able to treat everything
  in the correct way also for the whole detector.
  
• However to modify it or add any kind of specific
  selection is a little bit
  
• too complicated and expert oriented.
• As a first step, we therefore modified the code
  of MDThistos

10
MDT2ntu format
11
GNAM spectrum
12
MDT2ntu fast analysis
13
MDT2ntu fast analysis
14
MDT2ntu display
15
StorageReader implementation

• The first scheleton of data interpretation in a
  stable way w/o the usage
  
• of the whole Athena framework has been done by
  M.Barone by implementing
  
• the Storage Reader for MDT.
• Its advantage will be evident for the debugging
  period on
  
• the whole detector since
• 1) It is as flexible as GNAM and can analyze
  different chambers at the
  
• same time.
• 2) It can be used by simple modification of
  Root Macros
  
• 3) It reads directly the file w/o any need of
  a file server.
  
• Discussion is in progress with Roma 1 people to
  include in the same way
  
• also the routines for the RPC decoding and for
  monitoring L1 trigger.
  

16
Towards athena

• It is obvious that an important point will be
  that of reading out everything
  
• with Athena and to be able to create the same
  Root Tree/Ntuple in this
  
• framework.
• The real advantages will be
• - To start working on the calibration
  algorithms
  
• - Look also at tracking (Moore)
• - Ev.Display

17
Towards Athena calibration
1 ) The cosmics rate we observe now is not great
(0.2 2 Hz/m2) but it will be
enough to start looking at least to
T0,Tmax stability and to get a first rough
estimate of the R-T relations.
2 ) Illumination of the whole chamber has to wai
t for the RPC trigger. 3 ) Discussion with Rom
a3 people started. - I am trying to organize
a meeting during the muon-week of 28-30 November
to start collaborating on this. - I
have sent an e-mail yesterday to D.Orestano to
discuss how to use Calib on these data
. The response seems to be very positive. Any
volounteers? 4 ) Insertion of T0 calibration vi
a pulsing is relevant to decide how much effort
needed for the partial event reconstruction
on ROI selection at L2 (E.Pasqualucci).
18
Tracking ?

• MOORE tracking with cosmics is slightly
  problematic since the TDC stops
  
• (ie the trigger) is not phased-locked with the
  machine clock (25 ns of
  
• additional jitter).
• This should be an interesting effect to analyze
  when the final trigger will
  
• be in place. A good review of the tracking
  status with cosmics has been
  
• done by C.Guyot at the software meeting.
• The Timing-effect will be tested with real data
  vs simulation in the
  
• coming months.
• The effect of this and of the rough calibration
  on HLT triggering
  
• has to be evaluated.
• We are in contact with HLT people to organize a
  strategy to DEFINE
  
• how to use TrigMoore for the cosmics muon
  trigger for data taking of
  
• the whole spectrometer.

19
Overall readout and software scheme ??
links
L1 trigger
MDT
RODS
HV
L2trigger
Fee Mezzaninemapping
EVB Emulator Or Final
StorageReader
Partial EB At L2
Calibration Geometry Mapping
HLT
Athena
Calibration Stream