LHC Era Core Control Application Software

1
LHC Era Core Control Application Software

• Mike Lamont CERN
• For the LSA project team
• 13th October 2005

2
EVOLUTION

• REPLICATORS
• MUTATION
• COMPETIVE ENVIRONMENT

3
Given 3 Billion Years
Can come up with this
But probably not this
4
Alternative
INTELLIGENT DESIGN
(implies existence of intelligent designer)
Might come in handy here
5
CULTURAL EVOLUTION
MEME Replicator - conveys the idea of a unit of
cultural transmission
Good ideas live on (but then so do some bad ones)
We have run a 27 km accelerator before
6
Challenge

• We have to whizz 2 beams of 100,000,000,000 high
  energy protons in opposite directions around a 27
  km ring
• Through two very small, very cold, very dark
  holes
• Squeeze em down to 16 microns, get them to
  collide and keep them colliding for something
  like 10 hours
• While keeping our losses down to ridiculously low
  levels

AND WE GOT A DO IT WITH OUR CONTROL SYSTEM
7
The controls challenge

• EQUIPMENT
• Collimators/TDI/TCDQ etc.
• Beam Dump
• Power converters,
• Kickers
• RF, TFB, LFB
• Spectrometers compensation
• INSTRUMENTATION
• Distributed systems
• BLMs, BPMs,
• Standalone
• BCT, BTV, AGM, BIPM, BWS, Schottky..
• Tune, Chromaticity, Coupling
• Luminosity monitors
• Radiation Monitors
• MAGNETS RMS, errors
• MACHINE PROTECTION
• VACUUM, CRYOGENICS, QPS, EE
• EXPERIMENTS

Settings, functions, monitoring, display, post
mortem, control, acquisition, concentration,
archiving, alarms, interlocks
?
?
Driving the machine through the cycle Magnet
errors, crossing angles, snapback, ramping,
squeezing, colliding, orbit, parameter control,
optimisation etc. etc. HIGHLY DESTRUCTIVE
BEAM TIGHT TOLERANCES
?
8
LSA Core

• Analysis of the requirements of accelerator
  operations
• Set of software modules for accelerator controls
• Reference architecture
• Organization (See Lionel Mestres talk on
  Tuesday)

9
Strategy Memes

• Understanding what you need to do can be useful
• Analysis, LEP
• Pragmatic (rather than ideological)
• Simplicity
• Implement once Iterate
• Modular
• Factor out commonality
• Share Reuse Test
• TI8, SPS, LEIR, LHC hardware commissioning

10
Data Memes

• Controls is data driven and data is the master
• Data is Key, Data Stays
• Fast access
• Well organised
• Good data model

11
LSA Data model

• Well designed
• Its evolved!
• Same for TL, SPS, LEIR and LHC
• Optics
• Parameter Model
• Settings
• Configuration
• Etc.

12
LSA Core - Modules

• Optics
• Characteristics of the machine, layout, twiss
• Settings Management
• Functions or values for all parameters (physics
  to hardware)
• Settings Generation
• Generation of the functions based on the optics
• Trim
• Coherent modification of functions, revert
• Hardware Exploitation
• Equipment Control, Measurements
• Operational Exploitation
• Sequencing, Super Cycle changes

13
E.g. Settings and Trim

• LSA Core provides complete Settings Management
  and Trim facility
• Defines parameters and their relationships
• Defines setting per context (Cycle)

14
(No Transcript)
15
(No Transcript)
16
Orbit Steering
Using OPTICS, TRIM, DRIVE, ACQUIRE modules
17
Tool Memes

• Dont re-invent
• But the tool has to be appropriate to your
  requirements
• Tools have to be maintained
• Loss productivity if they get in your way
• fancy features that arent required
• Using a Ferrari as wheel barrow
• Can be great if they dont e.g. Spring
• Can be too heavy for the job (EJBs)
• But can turn out to be indispensable
• Relational Database (Oracle)

? Lionels Mestre talk
Understand your tools (and maybe read the
documentation)
18
Database Meme

• Databases
• Its a database not a data dump
• Dont use generic data models
• Design to perform dont tune to perform
• Keep It Simple
• Use supplied functionality e.g. integrity
• Understand your tool

See Tom Kyte
19
Interface meme

• JAPC
• One API for all equipment access
• See Vitos talk - coming next
• Use Case based interface
• To match specific application requirements

Generic doesnt always equal flexible
20
Only solve the same problem once

• Fixed Displays
• Logging
• SDDS
• Post-Mortem
• Alarms
• Common Components
• Dataviewer
• Beans

21
Steal Meme
Intelligent Laziness

• If someone has already solved the problem
• Make sure youve got the source code
• Say Thank You

22
SDDS
THANK YOU ARGONNE!
23
Conclusions

• LSA Core
• Analysis good data model - commonality
• Modular
• Pragmatic
• Sound architecture
• Appropriate use of available tools, reuse where
  possible
• Complex problem space no one answer
• Open to technological evolution
• User needs are paramount, not an ideology
• Remain close to your users - collaborate
• Remember - some people are religious

LETS EVOLVE!
24
If all else fails
Study the problem, understand the technology,
hire an experienced architect, and choose to
build systems based on appropriate technologies.
25
Thanks to

• Lionel Mestre
• Greg Kruk
• Delphine Jacquet
• Eugenia Hatziangeli
• Francois Chevrier
• Markus Albert
• Lasse Normann
• Guy Crockford
• Vito Baggiolini
• Anthony Rey
• Jorg Wenninger

The Team Meme