RF System Models and Longitudinal Beam Dynamics T. Mastoridis 1, P. Baudrenghien 1, J. Molendijk 1, C. Rivetta 2, J.D. Fox 2 1 BE-RF Group, CERN 2 AARD-Feedback and Dynamics Group, SLAC T. Mastoridis LLRF 11, October 19th 2011 1
1 RF System models 2 Longitudinal diffusion due to RF noise 3 Coupled-bunch instabilities 4 Conclusions T. Mastoridis LLRF 11, October 19th 2011 2
RF System models: Motivation The RF station-beam interaction defines many important beam dynamics. As a result, determining the optimal settings for the RF/LLRF system to achieve both station and beam stability can be rather complex The theoretical study of the beam-rf interaction is difficult due to the complexity of the multiple feedback loops, the non-linear nature of the system, and the complicated multi-dimensional parameters space. A solely experimental approach would not only require a lot of machine time and suggest risks for system components, but also would not allow for an arbitrary variation of system parameters. Models and simulations [1] can help estimate the effect of RF configurations on beam dynamics, predict the beam behavior, determine optimal settings, and study alternative hardware designs. The ability to adapt to completely different machines (hadron, lepton) and beam dynamics of interest is also an important aspect. T. Mastoridis LLRF 11, October 19th 2011 3
RF Station/Beam Dynamics Interaction Model Klystron Polar Loop Driver Klystron RF cav. + Σ Digital RF Σ Feedback + + Analog RF Feedback + Σ + 1 Turn(comb) Feedback RF reference RF system models were initially developed at PEP-II (J. Fox, T. Mastoridis, D. Teytelman, C. Rivetta [2], [3]) to help push the current to higher levels and better understand/utilize the trade-off between RF station and beam stability Limit from coupled-bunch instabilities These models were updated for the LHC architecture and expanded for the beam dynamics of interest: Multi-bunch: coupled-bunch instabilities driven by the cavity fundamental Single-bunch diffusion due to RF noise, essential for a hadron collider with 10-20 hour long coasts Single-bunch stability driven by broadband impedance T. Mastoridis LLRF 11, October 19th 2011 4 Beam
1 RF System models 2 Longitudinal diffusion due to RF noise 3 Coupled-bunch instabilities 4 Conclusions T. Mastoridis LLRF 11, October 19th 2011 5
Longitudinal Emittance 2 The longitudinal emittance is a measure of the area in phase space occupied by the beam Intrabeam scattering and RF noise lead to emittance increase Energy lost to synchrotron radiation reduce the emittance Normalized Energy 1.5 1 0.5 0 0.5 1 1.5 2 4 3 2 1 0 1 2 3 4 Normalized Position The synchrotron radiation for protons in the LHC is practically zero (damping time of about a hundred hours at 3.5 TeV) As a result, the noise power spectrum of the RF accelerating voltage can strongly affect the longitudinal beam distribution Increased bunch length decreases luminosity and eventually leads to beam loss due to the finite size of the RF bucket T. Mastoridis LLRF 11, October 19th 2011 6
LHC RF system Power Supply Klystron Polar Loop 400.8 MHz RF Reference VCXO Synchro Loop LLRF boards Modulator Setpoint Beam Phase Loop Klystron Demod RF cav. V Phase Detector Beam Cavity Sum The Beam Phase Loop (BPL) is a narrow bandwidth loop updated once a turn, that modulates the RF reference to achieve damping of mode zero beam motion around the synchrotron frequency f s T. Mastoridis LLRF 11, October 19th 2011 7
LHC RF Noise Sources Power Supply Klystron Polar Loop 400.8 MHz RF Reference VCXO Synchro Loop LLRF boards Modulator Setpoint Beam Phase Loop Klystron Demod RF cav. V Phase Detector Beam Cavity Sum Two major noise sources for beam diffusion: The RF reference noise introduced during the modulation/demodulation process in the Cavity Controller. Intrinsic noise in baseband from the Cavity Controller feedback boards. Since the RF feedback impedance reduction is delay limited, the Cavity Controller includes very wide-band electronics (up to 100 MHz bandwidth components). The final RF feedback has a single sided bandwidth of 400 khz, extending over 35 f rev bands. T. Mastoridis LLRF 11, October 19th 2011 8
of to RF System models RF Noise Coupled-bunch instabilities Conclusions Performance limiting components at LHC /3 VCXO 1/f 4 law Not understood VCXO 1/f 2 law (resonator Q) In physics f s0 ~28 Hz and dipole mode zero sits here Imperfect TX and LLRF noise compensation Demodulator noise dth is ~ 300 khz (single sided), limited by the 650 ns loop delay ominates at low frequencies (below 200 Hz) f the driver noise is responsible for the 200 Hz to 20 khz range khz closed-loop BW, the spectrum is flat, dominated by the Closed Loop BW limit ~ 300 khz The Beam Phase 16.12.2010 Loop (BPL) reduces the noise around f s Cavity Phase Noise (dbc/hz) 40 60 80 100 120 140 160 180 10 0 10 1 10 2 10 3 10 4 10 5 10 6 10 7 Frequency (Hz) Phase noise PSD of the RF sum (through an 8-way combiner) of the cavity voltage seen by the beam (no interfering electronics). The accelerating voltage phase noise is dominated by the 400 MHz reference up to 300 Hz, the Cavity Controller at higher frequencies The intensity lifetime would have been less than an hour otherwise BPL OFF: Noise dominated by VCXO - correlated. BPL ON: Nominal Operation. Noise dominated by LLRF - uncorrelated. T. Mastoridis LLRF 11, October 19th 2011 9 BPL On BPL Off
Proton Measurements By varying the BPL gain, we could change the noise level around the synchrotron frequency and look at the result on the longitudinal beam emittance. Noise (dbc/hz) 40 60 80 100 120 140 G = 1125 G = 281 G=140 G=20 G=5 G=0 Bunch Length (ps) 150 145 140 135 130 125 120 115 Data G = 1125 G = 281 G=140 G=0 G=20 G=5 110 160 105 180 10 0 10 1 10 2 10 3 10 4 10 5 10 6 10 7 Frequency (Hz) RF station 6B2 noise spectral density with BPL gain 100 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 Time (s) Beam 2 Bunch Length with time T. Mastoridis LLRF 11, October 19th 2011 10
Ion Measurements For a more quantitative and accurate study, a technique was developed (talk by J. Molendijk yesterday) to inject noise of controllable amplitude in a narrow band around the synchrotron sidebands of a set revolution harmonic (k = 1 for these measurements) Injected Noise 10log(rad 2 /Hz) 70 80 90 100 110 120 70 dbc/hz 76 dbc/hz 82 dbc/hz 85 dbc/hz 88 dbc/hz 94 dbc/hz 100 dbc/hz Bunch Length (ps) 400 350 300 250 200 130 100 80 60 40 20 0 20 40 60 80 100 Frequency (Hz) Levels of injected noise around f rev + f s. Horizontal axis shifted by f rf + f rev 150 0 2000 4000 6000 8000 10000 12000 Time (s) Beam 1 bunch length growth T. Mastoridis LLRF 11, October 19th 2011 11
LHC RF Noise Threshold With these measurements and the relationship between bunch length growth rate and noise power spectral density it is possible to estimate a noise threshold of approximately -93 dbc/hz for a single cavity (SSB) for acceptable performance Acceptable performance set to dσ = 2.5 ps/hr (Intrabeam Scattering levels between dt 1 and 3.5 ps/hr at 3.5 TeV) The cumulative single-sideband noise power level per cavity is approximately -102 dbc/hz for LHC (assuming uncorrelated noise sources) 9 db margin [4] The goal is to estimate whether the RF noise can become excessive with the addition of the necessary loops for future high current operation If the noise threshold is crossed, these tools and measurements can be used to identify the sources of noise that are most damaging with the intent to selectively improve the responsible equipment T. Mastoridis LLRF 11, October 19th 2011 12
Design Analysis x z Processing DAC Dig Demod RF FB In An Demod Processing Digital FB Analog FB G analog G digital w RF FB Out y To better understand the contributions of the LLRF components, we terminated the input of the RF Feedback, switched the analog and/or digital path on and off, and adjusted the analog and digital gains As such, we identified the dominant LLRF noise contributions Digital path: differential amplifier driving the ADC and digitizing noise of the ADC Analog path: the large amplification stage after the analog demodulator Since the digital path is narrowband, in the end we have a single gain stage in the analog demodulator that dominates the RF noise contributions to beam diffusion! T. Mastoridis LLRF 11, October 19th 2011 13
1 RF System models 2 Longitudinal diffusion due to RF noise 3 Coupled-bunch instabilities 4 Conclusions T. Mastoridis LLRF 11, October 19th 2011 14
Coupled-bunch instabilities Impedance reduction is of fundamental importance at the LHC since there is no dedicated bunch-by-bunch longitudinal feedback system. The substantial bunch length leads to stability through Landau damping. The effective cavity impedance though depends strongly on the LLRF configurations. The system models and simulations allow us to include non-linearities in the system and more accurately reflect the system configuration in the growth rate estimation. In particular, the simulations employ the same parameter structure used at the station setting-up. 10 1 Growth Rate l (s 1 ) 10 2 10 3 Growth OTFB Off Damping OTFB Off Growth OTFB On Damping OTFB On 10 4 40 30 20 10 0 10 20 30 40 Mode Number T. Mastoridis LLRF 11, October 19th 2011 15
Growth Rate sensitivity to LLRF parameters One of the important features of the LHC time-domain simulation is the ability to study alternative configurations of the RF and LLRF system, without requiring time from the real machine. As such, it can be used to analyze the sensitivity of the modal growth rates to variations of the LLRF parameters. These studies provide insight on the limits of the implementation, on the operational margins, and on the parameters most essential to reliable operations. LLRF Parameter Adjustment Growth Rate Change Nominal Value - 0.033 - Cavity Detuning ±1 khz 0.038/0.028 +15/ 15% FB Gain ±3 db 0.028/0.043 16/ + 31% Loop phase ±10 0.23/0.19 + 590/ + 490% 1-turn FB Gain ±3 db 0.026/0.039 20/ + 20% 1-turn FB phase ±10 0.12/0.10 +270/ + 220% Table: Growth Rate Sensitivity on LLRF parameters. T. Mastoridis LLRF 11, October 19th 2011 16
Growth Rate sensitivity to LLRF parameters Effect of 5 degree loop phase rotation (RF Modulator) on effective impedance and growth rates T. Mastoridis LLRF 11, October 19th 2011 17
1 RF System models 2 Longitudinal diffusion due to RF noise 3 Coupled-bunch instabilities 4 Conclusions T. Mastoridis LLRF 11, October 19th 2011 18
Conclusions Conclusions/Future Directions The RF system models can help optimize the RF/LLRF configurations, estimate the effect of the LLRF implementations on critical beam dynamics, and provide insight on alternative hardware designs A noise threshold was set for the LHC for acceptable lifetime and a margin of operation was estimated for RF noise The dominant components for beam diffusion were identified With this formalism and RF simulation tools [1] we can design future RF systems and budget the allowed noise Estimates of coupled-bunch instabilities reveal significant margin of operation, but also great sensitivity to RF/LLRF configurations. T. Mastoridis LLRF 11, October 19th 2011 19
Acknowledgements Acknowledgements The CERN BE-RF group for hardware/firmware/diagnostic development, help with measurements, and more LHC operations for allowing these MDs SLAC collaborators Thank you for your attention T. Mastoridis LLRF 11, October 19th 2011 20
References [1] T. Mastorides et. al., Modeling and Simulation of the Longitudinal Beam Dynamics - RF Station Interaction in the LHC Rings", Proc. EPAC 2008, 23-27 June 2008, Genoa, Italy. [2] C. Rivetta et. al., Modeling and Simulation of Longitudinal Dynamics for Low Energy Ring-High Energy Ring at the Positron-Electron Project, Phys. Rev. ST-AB, 10, 022801 (2007) and SLAC-PUB-12374, February 2007. [3] T. Mastorides et. al., Analysis of Longitudinal Beam Dynamics Behavior and RF System Operative Limits at High Beam Currents in Storage Rings, Phys. Rev. ST-AB, 11, 062802 (2008) and SLAC-PUB-13287. [4] T. Mastoridis et. al., Radio frequency noise effects on the CERN Large Hadron Collider beam diffusion", Phys. Rev. ST Accel. Beams 14, 092802 (2011). T. Mastoridis LLRF 11, October 19th 2011 21