Brief reminder

1
Lessons from LEP

• Brief reminder
• LEP Challenges beam related
• LEP challenges controls
• What helped in the way of controls
• Control group myths
• LHC?
• Conclusions

2
LEP - The largest particle accelerator to date
1989 First turn 1989-1995 The
Z-years (precision studies) 1996-1999 The
W-years (precision studies) 2000 The
Higgs-year (almost a discovery?) Nov
2000 Start of dismantling
Circumference 27 km Energy range 20 104.5
GeV
3
LEP challenges

• Multi-cycle injection
• Stability of lines, steering
• Accumulation resonances, coherent tune shifts,
  wigglers, radiation in experiments, etc. etc.
• Ramp between 22 GeV and 104 GeV
• Tune, chromaticity and orbit control
  (particularly the start), resonances, bunch
  length, wigglers
• Squeeze between ? 20 cm and ? 5 cm.
• Tune, chromaticity and orbit control
• Physics
• Beam-beam, control of tune, chromaticity, orbit,
  beam crossings, coupling, lifetimes
• Background optimisation - collimation
• Continual optimisation to maximise delivered
  luminosity.

4
LEP challenges

• 27 km of equipment and instrumentation to keep
  running
• 700 or so power converters,
• 1000s of magnets 8 of which superconducting
• 20 or so electrostatic separators
• Huge RF system
• Lots of Collimators
• Kickers, beam dumps
• 250 BPMs, BCTs, Q-meter, BST, profile
  measurements, beam loss monitors etc
• A few interlocks
• Communication with the experiments

All held together with a rudimentary control
system
5
1989 - commissioning

• 14th July first beam
• 23rd July circulating beam
• 4th August 45 GeV
• 13th August colliding beams

These people are to blame for what followed
6

LEP difficult teething

• Fractured high level control system
• It was slow (even in 2000 it took 15s to acquire
  a closed orbit)
• Poor measurement facilities
• Beam instrumentation lived in a world of its own.
  Very little integration.
• Essential signals not available e.g. no beam
  lifetime, for example
• Poor data management
• Inflexible communication with experiments
• No easy way of closing the measure/correct loop
• Poor and unreliable, incoherent data acquisition
  systems
• After commissioning and 2 years of operations we
  were faced with just wanting to get the beam up
  the ramp occasionally. Operations a real struggle
  (turn around was around 7 hours back then)

7
1999
253 pb-1
BORING!
8
Performance
LEP2
LEP1
Continual improvement even for same peak
luminosity!
9
So what went right?

• Things clearly got a lot better
• Turn around
• Injection efficiency
• Transmission through the ramp and squeeze
• Performance, in spite of limited current and the
  huge RF system

10
Settings management

• Need to drive machine through a reproducible
  cycle
• Handle all modifications to settings in a
  sensible way
• Be able to roll back some or all changes
• Exert control in a appropriate way

11
Settings

• Took around 20 minutes to fill LEP, a lot of
  fiddling around with tunes, orbit and stuff.
• Changes need to be incorporated to ramp in an
  appropriate way without screwing the ramp up
• Ramp (always a rocky start) and squeeze
• Driven by current functions downloaded to power
  converters (and RF and separators)
• Control in terms of Tune and chromaticity and
  corrector strengths
• Plus stops in ramp or squeeze, fiddle around and
  then carry on

12
Parameters
Ability to control beam in terms of appropriate
parameters

• Trim synchrotron tune, calculate total voltage
  change, trim total voltage.
• Trim tune calculate changes in Kqf, Kqd, Iqf,
  Iqd, and send to hardware - where in fact the
  current is delivered by 8 power converters
• Trim integrated B-field in wiggler, Calculate
  associated orbit correction, calculate associated
  optics change, calculate current changes in
  wigglers, wiggler compensation coils, orbit
  correctors and insertion quads.
• Plus user-definable KNOBS, e.g. orbit bumps,
  betasqueeze etc etc

For either functions in the ramp or at steady
state provide trim history, rollback,
consistency etc and the ability to carrying on
ramping
Note The ability to set a current is not
considered sufficient.
13
Combined controls/operation project
Controls provided fellows support for the old
system
Redesigned and re-implemented high level control
system (on-line ORACLE controls database)
Successfully solved serious data management
problem
14
Trim History
All changes recorded on database. Rollback of any
or all systems possible
15
Databases
KEY FEATURE - THE USE OF DATABASES
Extremely useful, providing as they do
Consistency, back-up, support of relational
model, access mechanism, a lot of neat stuff,
data management, etc, CENTRAL RESPOSITORY

• Measurement database
• beam, equipment, experiments, max rate 0.25 Hz,
  years worth of history,
• Controls database
• All settings, machine parameters, configuration,
  optics etc
• All trims are recorded
• Logging database
• many years, sparser than measurements plus
  environment etc etc
• RF logging database

16
Series of applications accessing data via database
ORACLE database provides central repository
System dependent black boxes pushing data up at
appropriate rates
Experiments communication system
17
DATA EXTRACTION - JAVA GUILS
18
DATA EXTRACTION ? POST RUN ANALYSIS
With historical data on the database, reasonably
easy to extract and analyze off-line
19
Statistics
Data hauled from database automatically at end of
fill
20
Fixed Displays
Generic data driven application dynamic SQL
backgrounds, radiation, beam-beam tune shifts,
bunch currents, angle and positions, beam sizes,
luminosities from various sources...
21
Typical view of LEP control room
Fixed Display
Operator
22
Scans tie it together
Dedicated program standard calls to perform trims
Measurements from Database

• Separator voltages
• Beam sizes
• Beam angle
• results

23
Tools e.g. Dataviewer
Generic display manipulation of data
24
Dedicated Video (FAST) Signals
Data sampled at slower rate ? database
Vertical beam sizes v. useful for luminosity
optimization Life without lifetimes - impossible
25
1000 Turns - ? Beating

• BIG FILES!!! - Dedicated applications - Brain
  power

26
Turn around

• Semi-automatic sequencer
• Reproducibility
• Reduced scope for error

Typical 2000 turn-around 45 minutes
27
Optimisation

• Reproducibility
• Golden orbits including corrector settings
• Procedures
• Easy to perform measurement procedures
• Coupling, beta, dispersion
• Fast signals
• Beam sizes, luminosity from lifetime
• Intellectual Property rights
• DFS

28
eventually the Q-loop
29
RF System special mention
30
And other neat stuff
104.0 GeV
103.3 GeV
Mini-ramp
Beam lifetime 9 hours 3 hours quantum
lifetime
31
LEP COULD BE OPERATED BY ONE MAN!
32
Analysis and design
In both LEP and SPS a major design effort was
under taken

• Method adopted (SASD)
• the whole thing (e.g. operate LEP)
• spent a long time in the analysis phase
  understanding the requirements
• Data analysis database design included
• pragmatic but serious
• CASE tool
• documentation, focus, continuity and support of
  the method
• Operations heavily involved
• Same team throughout
• An appropriate user view

Vital in providing the generic tools outlined
above
33
Interface(s) to Equipment

• Specialised groups power converters, RF, beam
  instrumentation, kickers, separators, vacuum,
  dedicated expertise (electronics, controls,
  hardware)
• Both the SPS and LEP efforts accepted the
  existing equipment interfaces and buried the
  access to them in black boxes (encapsulation)
• However its clear that the front-end system can
  compromised the high-level
• By not supplying appropriate functionality
• The LEP low-level power converter s/w was
  probably its saving grace, but fixed approach to
  the start of the ramp, no RT knobs
• By being slow
• By making life bloody awkward
• Of course, we also have accept what ever the
  control group offers in the way of communication
  mechanisms. Again buried in the black boxes.

34
Components

• Access system/Machine interlocks always a
  problem
• Alarms dedicated system with a lot to deal with
  worked well
• Timing Extremely important, few problems with
  the mtgs but hardware infrastructure worked well
• Gateways, networks, front-ends flakey at first
  but things settled down
• Reboot tool always useful
• Interlocks not many

35
Control Myths

• The operator wants a unified standard MMI
• Maybe a similar look and feel but
• Speed, reliability and functionality are much
  more important
• A standard API to the equipment is vital
• Not a user requirement
• And we can deal with it if it isnt
• Accelerator consists of devices with properties
• Not invented here
• Some nice wheels (e.g. dataviewer) re-invented.

36
Conclusions

• Databases are useful
• But need managing and designing properly
• And everything doesnt have to forced onto them
• Settings management is key.
• Control in terms of the right parameters is
  vital.
• Fast, reliable signals of key beam parameters are
  vital.
• Key feedback systems must be made to work ASAP
• Anticipate closing the loop
• Heterogeneous equipment access is a bit of a pain
  but hey
• If it aint ready in time, other solutions will
  be found (e.g. middleware)
• Integration of, and maturing of BI took a long
  time.
• Committed, aggressive, tool using individuals on
  the machine are, like, totally invaluable
• Things evolve
• We eventually ended up with some really cool
  stuff

37
Conclusions
We didnt solve all the problems and what we
ended up with was a bit of a mish- mash but it
did enable us to effectively exploit the machine.
The eventual efficiency with which LEP could be
operated, even in the final years at the
performance limits of the hardware system, was in
large part due to the integration of a well
designed control system using commercial
databases.
38
What did we miss?

• Scripting environment
• On-line model, the interface to MAD was a pain
• Configurable feedback/control loops.
• Machine protection/interlocks
• Speed
• Rigour

39
LHC - Operating challenges

• Super-conducting magnets
• multipoles, snap-back, persistent currents, key
  beam parameters affected, strong dependence on
  magnetic history
• High energy, high intensity beams
• extremely low tolerance to beam loss, quench
  protection collimation mandatory at all times
• Machine design
• 2 rings, 8 sectors, bits of the ring in common,
  cross-talk between the rings, small mechanical
  aperture, large energy swing ? large range in
  magnets and power converters.
• Beam dynamics
• Wide range of optics Beta 18 m to 0.5 m
• Dynamic aperture, limited by non-linear fields
  from magnet imperfections or beam-beam. Problem
  at injection where non-linearities are large and
  beam has large emittance. Tight constraints on
  beam parameters
• crossing angle, beam-beam effects
• Intra-beam scattering, synchrotron radiation,
  instabilities, electron cloud, PACMAN bunches,
  ghost bunches. beta beating very tight orbit
  tolerances

40
LHC
Inject, ramp, squeeze, collide

• Integrated measurements control. Already see
  the need for tight coupling between controls,
  equipment, beam instrumentation, magnetic
  measurements.
• Timing synchronization (of measurements as
  well).
• Speed
• Protection/Interlocks
• High-level data management
• keeping track of trims, feed-forward, history,
  control in terms of relevant parameters,
  reproducibility, databases, analysis and design
• Feedback
• Requisite low-level functionality in equipment
  controllers
• Cross-system communications
• Demands on controls high
• high degree of automation, surveillance,
  post-mortem, diagnostics etc..
• powerful, flexible, rigorous, real-time,
  integrated, coherent, fast, safe, available on
  time

41
Last slide
Databases, settings management,
reproducibility, logical architecture LEP
Real-time control Feedback Speed
Flexibility PEP II
Rigour, diagnostics protection
LHC