Top Priorities
- Comparisons of data should be to reference set rather than to "ideal case"
- Monitoring and Event display - get up and running routinely
- Requires configuration to be regularly added to database!
- Online TIM monitoring - get going with ROD & BOC stuff.
- Histogramming robust against module failure
- next on the list
- Error mask and s/n number sending
General Stuff
- NPtGainSummaryWriter? should only output most serious defect on that chip.
- Rod with no modules should not be an error.
- Recovery of Analysis and Archiving after re-start (as is done for FittingService)
- SctNames? should be configurable, and should contain support for re-analysis of data
- Make sure this is done in a thread-safe way
- Propagate changes into services
- Persistifying changes to configuration in a heirarchical way.
- e.g. RxThresholds?, but should be done in a general way
- This needs some thought!!
- Have chatted to Shaun Roe and Paul Bell about this - needs further thought
- Have done global saving of configuration at end of run.
- Scripting and default display for GUI
- Some way of putting non-consecutive scans into an analysis
- include continuous 3-pt gain sequence?
- Make system stand-alone boot-strap. (StrobeDelay needs 2.0fC point).
- On hold for the moment
In Progress
- Running scans with Tim sequences.
- Boc Setup Tests
- Previous bug may have been fixed by DSP upgrades - need to check if this still exists.
- Problems with 10PtGain and 5PtGain
- see NPtGainProblems
- Possibly related to fit accuracy/precison/iterations in NagFitStrategy?, which is now bug fixed [2006-06-09 AJB]
- Cond DB interface for data archiving.
on hold
- Non-sequential scan points
- investigate load/config separation.
- Actually monitor the BOC temperature monitoring information.
- Currently goes to IS unchecked.
- What about the ROD temperature.
- Now going to be done through DCS
lower priority
- Histogramming of Raw data on-rod
- Currently done with API. Very slow with many modules (poor scaling!)
- Find out what's wrong with ROD after raw scan
- need to fix standard dump first
- Accumulate test - talk to Peter
- Display improvements
- Display fit parameters vs channel (vt50, width, etc)
- Investigate threading the histogram readout
Finished - to be tested
- Apply the results of a test (config-updater style) from old data.
- Set configuration from saved data.
- Needs gui method
- Check how well the mapping check scans work - CheckingMappings.
- Sort out RxDelay when more than one rise and fall
Really finished
- TimingInSCT - try with scintillator on H8 test box?
- Set up config/histogramming faster - see CalibrationController and ConfigurationPerformanceImprovements
- Sort out problems with DoubleTriggerNoise - are we histogramming the correct trigger? We really think so now.
- CounterTest? - write test which makes sure that counters actually count (basically an analyis of RawScan)
- Speed up initialisation and histogram setup
- ModifyABCDVarRod? rather than modifyABCDVar?
- Get the analysis chain off the roller-coaster
- Define a path in extended TestData object
- Changes in IS mean we need to update IS wrappers etc done 25/5/2005.
- QuickNMask? test in cvs and analysis modified to work with it
- Looks ok to me
- SynchTriggerNoise should be a threshold scan
- Hopefully ok
- sort out noise numbers for short endcap modules
- done but not tested.
- Analysis of DoubleTriggerNoise
- Bilge taking this along with analysis of OccupancyPerEvent histograms.
- UniqueID? class should be used in code rather than string representing UniqueID?, to allow for re-analyzed UniqueID?'s.
- Many classes such as ScanResult? and TestResult seem to create a UniqueID? on demand from run, scan and module :(
- Problem with stuck-cell/dead-cell pipeline test. On module on dead fibre at oxford, we have no pipeline defect when no data is coming from chip M8.
- was due to dead cells not failing pipeline test.
- MarkSpaceRatioTest done and tested on modules in the box
Analysis of Barrel 3
- Double Trigger Noise -- Bilge doing this at DoubleTriggerAnalysis
- Synch Trigger Noise -- Alan Phillips doing this
- Slopes in gain, offset, noise -- See SlopesOnBarrelThree
- Check error levels and "problem" flag
- RxThreshold for Kamil -- data in his hands.
- Add normalised occupancy plot for noise occupancy test
- to avoid confusion over "bands" due to different # triggers
- Should now be done in head.
- Coordinate migration towards online 21
- merge back head - done
- look at online 22
- Check and fix BugFixesForOmniMigration
- ChrisLester has taken over most of this.
- Oxford now on CVS head as default.
- Test error counting and event counts in both noise and cal-loop scans
- Done - works
- Reorganise Oxford machines for higher performance.
- 2 1U rack-mounted CPU's on order - should have arrived 24/1/05
- NFS directory structure given to John M
- Working - but Pete G may still be tweaking these.
- Update John Hill's TimModule? changes in our CVS branch for TIM3C
- What's wrong with SynchTriggerNoise