1. TC 25th October1 LHC Real-time requirements What is real-time? What is latency? The time between asking for something to be done and it being done The violation of one or more temporal constraints causes a loss of functionality. If it takes too long it don’t work. It ain’t necessarily speed.

2. TC 25th October2 Implications Front-ends Field bus Gateway Network Workstation Widely available industrial standards: ATM, WorldFIP, LynxOS, PowerPC which would satisfy LHC requirements (and in fact are already in use, planned to be used, or being used in prototypes.) E.g: Full ATM for LHC: 2M CHF Power converters Beam Instrumentation Reference magnets Real time To allow real time control/access of: But note: things are moving quickly out there

3. TC 25th October3 Why bother? 150 units of Q’ Large but reproducible, will clearly have to be measured Energy during injection plateau Orbit 1. Multipole errors

4. TC 25th October4 Control: assumptions Snap-back and low energy provide the challenge Nominal is the worse case... The “multipole factory” takes care of at least 80% of the effect i.e. –magnetic measurements –non-linear models –on-line reference magnets Characteristic frequency of snap-back slow & not changed by feed-forward from “multipole factory” Corrections to: horiz. orbit correctors (b1) & tuning quads. (b2) sextupole (b3) and decapole (b5) spool pieces Characteristic frequency is low 70 s

5. TC 25th October5 Beam parameter changes

6. TC 25th October6 Beam based feedback max rate db3/dt  0.067 units/second, 3 - 5 units Q’/second  1 unit per second given the 80% from the magnet factory (NB out of tolerance in 1 second) Chromaticity is NOT an easy measurement and is not likely to be available for the physics beams (emittance growth). Available for pilots  reproducibility (again…) Tune & orbit we can do but b3: caveat...

7. TC 25th October7 So we have large multipole swings during the injection plateau and snapback... Beam stability will depend totally on –feed-forward from the reference magnets –feedback on the key beam parameters tune, orbit, energy, chromaticity Have to make a phase jump in our expected performance of a control system: –at present it is in the stone-age –have a look at synchrotron light sources to get an idea of what is possible... We will need similar to deal with multipoles, tight physics constraints & the low tolerance to beam loss.

8. TC 25th October8 LEP Orbit Maximum sampling frequency 0.067 Hz LHC Orbit Proposed sampling frequency 10 Hz The stone age...

9. TC 25th October9 RT Requirements Reference magnets –feed-forward of corrections to machine Feedback-loops –Global Orbit –Local Orbit - e.g. orbit stability at collimators –Tune –Chromaticity Real-time knobs –typically tune, orbit bumps –extension of feedback, operator close loop, need... Display –Orbit –Beam loss (not for abort, but display & possible feedback) –Luminosity, Beam sizes, Lifetimes… Timing like actions (possibly) –e.g post-mortem trigger

10. TC 25th October10 RT requirements II Response limited by PC/magnet

11. TC 25th October11 Work in progress Power converters: –WordlFip 2.5 Mb/s + PowerPC/LynxOS gateway –already planning RT capabilities up to 100 Hz BPM/BLM –work in progress but use of RT functionality assume ATM prototyping –SPS Q-loop –LHC orbit correction simulation Requirements elucidated: –LHC report 221 “Requirements for real time correction…” –CO/OP forum –LHC-CP workshop –here...

12. TC 25th October12 SL-BI Two classes of instrumentation: global & local The global systems e.g. orbit and beam loss will provide the main control challenges:- Orbit: 10 Hz RT connection to PCR BLMs: 10 Hz RT connection to PCR (NOT to decide on beam abort) Local systems such as tune & chromaticity will be involved in RT feedback Proposed Architecture VME World FIP for RT Ethernet for off-line retrieval & Post-mortem Under discussion

13. TC 25th October13 Gateway MCCS Server ~80 Gateways ~1700 Digital Controllers Up to 30 Digital Controllers Per WorldFIP fieldbus RT LHC Network Major system - all magnets - set, ramp, trim etc. etc. Offering RT control at 100 Hz (whether you need it or not!) Digital controllers:- CERN design, manufactured by industry. Prototypes for String II. Gateways:- standard SL-CO - Power PC running LynxOS Fieldbus: WorldFIP from CEGELEC Timing: via an IRIG-B receiver card Power Converters

14. TC 25th October14 Reference magnets C MCCS Gateway Multipoles Factory DB I SM18 Magnet Test Benches WorldFIP fieldbus Real-time LHC controls network FB Power Converter Real-Time LHC Control System Instrumented Magnet 3-10Hz

15. TC 25th October15 SL-CO-FE - Real Time Investigating RT architectures –Front-ends, Networks –Prototype WorldFIP, FIP running LynxOS, ATM SPS as test-bed –Q-loop –LHC orbit correction simulation

16. TC 25th October16 Conclusions Demonstrable need for real-time control at the LHC Essential requirements established. Foresee dual system: –real-time for purposes outlined above –non-deterministic but high bandwidth for non-critical monitoring, command flow, post-mortem etc. etc. Accept RT is necessary, form WG under auspices of LHC- CP, ask them to investigate possible architectures… Much work already done in prototyping possible system, however technology choice can wait. Final architecture will clearly take into account such things as: –cost –reliability –maintenance

17. TC 25th October17 Acknowledgements Michel Jonker Pedro Ribeiro Thijs Wijnands Quentin King Alan Burns Oliver Bruning Luca Bottura Robin Lauckner