1. O RBIT F EEDBACK EXPERIENCE AT CTF3 Davide Gamba davide.gamba@cern.ch 3 October 2014 Octoberfest at JAI – Royal Holloway University of London

2. Outlook CTF3: the CLIC Test Facility at CERN. Why an orbit feedback. A general feedback algorithm and its implementation. – An application case. Some issues: – Beam Jitter. – Energy profile along the pulse. Dispersion measurements/corrections. Summary.

3. CLIC: A multi-TeV e + e - linear collider study. Based on two-beam acceleration scheme. Besides the two colliding beams, it requires two parallel high intensity Drive Beams for RF power production.

4. CLIC: A multi-TeV e + e - linear collider study. Based on two-beam acceleration scheme. Besides the two colliding beams, it requires two parallel high intensity Drive Beams for RF power production. T WO B EAM A CCELERATION X-BAND H IGH G RADIENT X-BAND H IGH G RADIENT D RIVE B EAM R ECOMBINATION F ULL L OADED L-BAND A CCELERATION F ULL L OADED L-BAND A CCELERATION L ONG -H IGH CURRENT (4 A) B EAM L ONG -H IGH CURRENT (4 A) B EAM

5. CTF3: the CLIC Test Facility at CERN. F ACTOR 2 R ECOMBINATION F ACTOR 2 R ECOMBINATION D RIVE B EAM G UN D RIVE B EAM G UN F ACTOR 4 R ECOMBINATION F ACTOR 4 R ECOMBINATION P ROBE B EAM P HOTO -I NJECTOR P ROBE B EAM P HOTO -I NJECTOR T WO B EAM T EST S TAND (TBTS) D OGLEG E XPERIMENT D ECELERATION T EST B EAM L INE (TBL) P HOTO I NJECTOR T ESTS XBOX EXPERIMENT XBOX EXPERIMENT CR 84 [m] DL 42 [m] Building 2001 Building 2013 Building 2010 Building 2003 CLEX area 1.5 GHz – 4 A – 1.2 μs 3 GHz – ~8 A – 4 x 140 ns 12 GHz – ~30 A – 140 ns 1.5 GHz – <1 A – <150 ns

6. CTF3: some pictures.

7. Steering necessities. Final aim: during recombination, different section of the initial train take different paths. O NLY ONE SECTION OVER TWO IS DELAYED INTO THE D ELAY L OOP T HEY DON ’ T NECESSARILY COME BACK ON THE SAME TRAJECTORY. I NITIALLY WE ONLY HAVE A LONG TRAIN “ DIVIDED ” INTO 8 SECTIONS, EACH 140 NS LONG P ROJECTED EMITTANCE BLOW - UP. I NCREASED LOSSES. P ROJECTED EMITTANCE BLOW - UP. I NCREASED LOSSES. I N THE C OMBINER R ING ALL THE 8 SECTIONS CAN END UP IN HAVING DIFFERENT ORBITS

8. Steering necessities: emittance growth Injection orbit error (Δx = 0.5 mm) After different number of turns the centroide of the ellipses in phase-space is moving accordingly to the tune of the ring. “recombined” emittance increased of 10%. CR: 4x

9. Orbit correction: the algorithm B EAM LOSSES D ATA SYNCHRONIZATION B EAM LOSSES D ATA SYNCHRONIZATION M ODEL ACCURACY R ESPONSIVITY AND RELIABILITY OF THE CONTROL SYSTEM S TRENGTH CALIBRATION R ESPONSIVITY AND RELIABILITY OF THE CONTROL SYSTEM S TRENGTH CALIBRATION BPM RESOLUTION D ATA SYNCHRONIZATION N OISE C ALIBRATION L INEARITY BPM RESOLUTION D ATA SYNCHRONIZATION N OISE C ALIBRATION L INEARITY

10. Feedback implementation C ONTROLS : F EEDBACK G AINS W EIGHT ON CORRECTION MINIMIZATION P OSSIBILITY TO REDUCE OVERALL CORRECTION STRENGTH C ONTROLS : F EEDBACK G AINS W EIGHT ON CORRECTION MINIMIZATION P OSSIBILITY TO REDUCE OVERALL CORRECTION STRENGTH RM VIEW AND O RBITS HISTORY C ORRECTORS SETTINGS AND EXPECTED CORRECTION C URRENT ORBIT AND EXPECTED AFTER CORRECTION BPM S WEIGHTS

11. An example: orbit closure in Combiner Ring (CR) Nominally 4 trains of bunches at 3 GHz are injected and interleaved with each other in CR, generating a short 12 GHz pulse with high current (up to 32 A). For the orbit closure we use a shorter pulse circulating for many turns. O N TWO CONSECUTIVE TURNS THE ORBIT MIGHT BE DIFFERENT. A SHORT PULSE (140 NS LONG ) I S INJECTED IN CR A SHORT PULSE (140 NS LONG ) I S INJECTED IN CR P ROCEDURE : 1.W E TAKE THE DIFFERENCE BETWEEN FIRST AND SECOND TURN ORBIT AS OBSERVABLE TO MINIMIZE. 2.W E MEASURE THE R ESPONSE M ATRIX OF CORRECTOR MAGNETS INSIDE OR OUTSIDE CR ON THIS OBSERVABLE. 3.W E PROCEED WITH THE CORRECTION. P ROCEDURE : 1.W E TAKE THE DIFFERENCE BETWEEN FIRST AND SECOND TURN ORBIT AS OBSERVABLE TO MINIMIZE. 2.W E MEASURE THE R ESPONSE M ATRIX OF CORRECTOR MAGNETS INSIDE OR OUTSIDE CR ON THIS OBSERVABLE. 3.W E PROCEED WITH THE CORRECTION.

12. An example: orbit closure in Combiner Ring (CR) B EFORE CORRECTION A FTER CORRECTION A FTER CORRECTION H IGH W EIGHT BPM S P ATTERN VERY SIMILAR TO NOMINAL X- DISPERSION … T O BE INVESTIGATED. P ATTERN VERY SIMILAR TO NOMINAL X- DISPERSION … T O BE INVESTIGATED.

13. Issue: beam jitter and energy profile We are affected by a natural jitter of the energy due to RF power compression. Beam energy along the pulse is also not always flat. Thanks to generality of our feedback tool we can use it to shape the RF power of last linac structures to try to compensate the second effect. Jitter, as expected, doesn’t change…. …But we can use it… Looking at horizontal beam position along the pulse in a dispersive pickup.

14. Dispersion measurement from Jitter Developed an application to passively/actively measure dispersion by beam jitter. O NE CAN MEASURE : C ORRELATED JITTER ( PASSIVE ) L INEAR FIT ON INJECTOR BEAM CURRENT INDUCED JITTER ( ACTIVE ) L INEAR FIT ON MAIN BENDING SCALING ( ACTIVE ) C AN BE USED AS ONLINE DETECTION OF MACHINE DRIFTS. O NE CAN MEASURE : C ORRELATED JITTER ( PASSIVE ) L INEAR FIT ON INJECTOR BEAM CURRENT INDUCED JITTER ( ACTIVE ) L INEAR FIT ON MAIN BENDING SCALING ( ACTIVE ) C AN BE USED AS ONLINE DETECTION OF MACHINE DRIFTS.

15. Dispersion measurement from Bending Scaling Scaling only the main bending magnets one can have (at first order) the nominal dispersion (depending only on quadrupoles strength and beam energy). An example: CLEX experimental area. D ESIGNED DISPERSION SHOULD BE ZERO T WO B EAM A CCELERATION T EST A REA I NCOMING DISPERSION SHOULD BE ZERO D OGLEG DISPERSION SUPPRESSION WITH QUADRUPOLES

16. Dispersion measurement from Bending Scaling Before correction (Dark green line) O NLY IN THE FIRST BPM DISPERSION IS EXPECTED !

17. Dispersion measurement from Bending Scaling Manually changing quadrupoles used for dispersion correction.

18. Summary What has been done: - Developed and tested a generic linear feedback application. - Smart way to measure the response matrix: it can work in quasi-parasitic mode and/or during orbit correction. - Feedback is currently used for orbit correction and energy stabilization. - Dispersion measurements from jitter and bending-scaling. What we are working on: - Injector stabilization. - “Jitter Free” Steering (Dispersion Free Steering). - Dispersion Target Steering. Main limitations: - Beam losses induce fake signals of BPMs due to move of beam centroid and noise on acquisition. - Reliability of Beam Instrumentation. - Very important to “know” the machine (i.e. many hours in the control room). - Non linear dispersion may give some contribution in beam position and jitter measurements.

19. Thank you. Davide Gamba davide.gamba@cern.ch 3 October 2014 Octoberfest at JAI – Royal Holloway University of London

20. CLIC: Some parameters. Center of mass energyE cm 3000GeV Main Linac RF Frequencyf RF 11.994GHz LuminosityL5.9 x 10 34 cm -2 s -1 Linac repetition ratef rep 50Hz No. of particles / bunchNbNb 3.72 x 10 9 No. of bunches / pulsekbkb 312 Bunch separationΔt b 0.5 (6 periods)ns Bunch train lengthτ train 156ns Beam power / beamPbPb 14MW Unloaded / loaded gradientG unl/l 120 / 100MV/m Overall two linac lengthl linac 42.16km Total site AC powerP tot 415MW Length of PETSL PETS 0.213m Nominal output RF Power / PETSP out 136MW Wall plug -> RF efficiencyη ACRF 58.6% RF -> drive beam efficiencyη bRF 93% drive beam -> RF efficiency (HDS input)η decRF 65% RF -> main beam efficiencyη bRF 27.7% Wall plug to main beam power efficiencyη tot 7%

21. Issue with CR BPIs: an example

22. Orbit closure after Delay Loop (DL) First stage: match orbits of delayed and not-delayed trains. U SE CORRECTORS IN D ELAY L OOP F ORCE THE SAME ORBIT IN THE NEXT T RANSFER L INE (TL1) In principle only two correctors with the right phase advance are needed. – Aperture limitations may impose to use more correctors.

23. Preliminary result – 1 st run 2013 “Orbit matching” in TL1

24. Jitter free steering The first area where we have dispersion is at the end of Drive Beam linac, where a chicane is installed. N OMINALLY DISPERSION FREE N OMINALLY NON ZERO DISPERSION N OMINALLY NON ZERO DISPERSION

25. Jitter free steering Using the same feedback application and measuring dispersion as jitter (std). Change correctors inside chicane to reduce jitter (i.e. dispersion) downstream. Acquiring 30 beam pulses for each iteration. Main disadvantage: RM measurement and correction takes longer time.

26. Emittance growth by energy spread Simulations ongoing to study effects of energy spread (up to 0.7 %) in the different lines. Here an example for a single passage in the Delay Loop: emittance increase ~200%