Sept 16-17, 2003 Linac Coherent Light Source Stanford Synchrotron Radiation Laboratory Stanford Linear Accelerator Center Controls Timing Workshop Patrick Krejcik, LCLS LCLS Timing Requirements - and how they were arrived at Patrick Krejcik September 16-17, 2003 Accelerator physics requirements Scientific user requirements Compatibility and operational requirements Accelerator physics requirements Scientific user requirements Compatibility and operational requirements

Sept 16-17, 2003 Linac Coherent Light Source Stanford Synchrotron Radiation Laboratory Stanford Linear Accelerator Center Controls Timing Workshop Patrick Krejcik, LCLS Accelerator physics requirements Stating the obvious Timing and RF phase control are synonymous in LCLS From an accelerator physics point of view: Ensuring the correct electromagnetic field in the device at the time the electron bunch passes by LCLS bunches have sub-picosecond duration Bunch timing tolerance is also sub-picosecond Controls perspective can distinguish between fine resolution (e.g. phase locking) and coarse resolution (choosing an RF bucket) And synchronization with AC power (timeslot) Operational perspective distinguishes between stability and tuning (e.g. feedback) Stating the obvious Timing and RF phase control are synonymous in LCLS From an accelerator physics point of view: Ensuring the correct electromagnetic field in the device at the time the electron bunch passes by LCLS bunches have sub-picosecond duration Bunch timing tolerance is also sub-picosecond Controls perspective can distinguish between fine resolution (e.g. phase locking) and coarse resolution (choosing an RF bucket) And synchronization with AC power (timeslot) Operational perspective distinguishes between stability and tuning (e.g. feedback)

Sept 16-17, 2003 Linac Coherent Light Source Stanford Synchrotron Radiation Laboratory Stanford Linear Accelerator Center Controls Timing Workshop Patrick Krejcik, LCLS Resolution versus stability Resolution can be coarse or fine Coarse resolution means putting the bunch (or observing) the correct S-band RF bucket Coarse timing needs to be reproducible after power dips and other resets Fine timing, or phase control Pulse-to-pulse stability, or jitter, is the performance criteria Longer term drift can be corrected on time scales ~10 sec. Resolution can be coarse or fine Coarse resolution means putting the bunch (or observing) the correct S-band RF bucket Coarse timing needs to be reproducible after power dips and other resets Fine timing, or phase control Pulse-to-pulse stability, or jitter, is the performance criteria Longer term drift can be corrected on time scales ~10 sec.

Sept 16-17, 2003 Linac Coherent Light Source Stanford Synchrotron Radiation Laboratory Stanford Linear Accelerator Center Controls Timing Workshop Patrick Krejcik, LCLS LCLS Principal Components – excluding DC devices Gun laser Laser diagnostics Gun RF Linac L0, L1, X1, L2, L3 RF Electron bunch diagnostics Gated readback of devices Active devices using pulsed laser synchronized to e-beam RF deflecting cavity Pump-probe laser for user experiments X-ray diagnostics Gun laser Laser diagnostics Gun RF Linac L0, L1, X1, L2, L3 RF Electron bunch diagnostics Gated readback of devices Active devices using pulsed laser synchronized to e-beam RF deflecting cavity Pump-probe laser for user experiments X-ray diagnostics

Sept 16-17, 2003 Linac Coherent Light Source Stanford Synchrotron Radiation Laboratory Stanford Linear Accelerator Center Controls Timing Workshop Patrick Krejcik, LCLS Allocation of major components of the LCLS rf system

Sept 16-17, 2003 Linac Coherent Light Source Stanford Synchrotron Radiation Laboratory Stanford Linear Accelerator Center Controls Timing Workshop Patrick Krejcik, LCLS Why not just ask for infinite precision? Timing jitter is a performance parameter for the user Shouldn’t we at least aim to demand the best available? Ultimately limited by inherent stability of SLAC klystrons Measurements on SLAC klystrons show they have jitter of ~0.1 deg. S-band Simulations used to determine the expected performance given this level of jitter Timing jitter is a performance parameter for the user Shouldn’t we at least aim to demand the best available? Ultimately limited by inherent stability of SLAC klystrons Measurements on SLAC klystrons show they have jitter of ~0.1 deg. S-band Simulations used to determine the expected performance given this level of jitter

Sept 16-17, 2003 Linac Coherent Light Source Stanford Synchrotron Radiation Laboratory Stanford Linear Accelerator Center Controls Timing Workshop Patrick Krejcik, LCLS linac  phase  0.1 deg-S rmslinac  phase  0.1 deg-S rms linac  voltage  0.1% rmslinac  voltage  0.1% rms DR phase 0.5 deg-S rmsDR phase 0.5 deg-S rms Charge jitter of 2% rmsCharge jitter of 2% rms linac  phase  0.1 deg-S rmslinac  phase  0.1 deg-S rms linac  voltage  0.1% rmslinac  voltage  0.1% rms DR phase 0.5 deg-S rmsDR phase 0.5 deg-S rms Charge jitter of 2% rmsCharge jitter of 2% rms …and bunch arrival time variations… 0  0.26 psec rms Simulate bunch length variations… 82  20 fsec rms Pulse-to-pulse jitter estimates based on machine stability P. Emma

Sept 16-17, 2003 Linac Coherent Light Source Stanford Synchrotron Radiation Laboratory Stanford Linear Accelerator Center Controls Timing Workshop Patrick Krejcik, LCLS Tolerance budget (p tol ) for <12% rms peak-current jitter (column 3) or <0.1% rms final e − energy jitter (column 4). The tighter tolerance is in BOLD text and both criteria, |ΔI/I 0 | < 12% and |  ΔE/E 0  | < 0.1%, are satisfied if the tighter tolerance is applied. The voltage and phase tolerances per klystron for L2 and L3 are  N k larger. ParameterSymbol|ΔI/I 0 | < 12% |  ΔE/E 0  | < 0.1% Unit mean L0 rf phase (2 klystrons) 0 S-band deg mean L1 rf phase (1 klystron) 1 S-band deg mean LX rf phase (1 klystron) xx X-band deg mean L2 rf phase (28 klystrons) 2 S-band deg mean L3 rf phase (48 klystrons) 3 S-band deg mean L0 rf voltage (1-2 klystrons) V0/V0V0/V % mean L1 rf voltage (1 klystron) V1/V1V1/V % mean LX rf voltage (1 klystron) Vx/VxVx/Vx 0.25 % mean L2 rf voltage (28 klystrons) V2/V2V2/V % mean L3 rf voltage (48 klystrons) V3/V3V3/V % BC1 chicanes B1/B1B1/B % BC2 chicanes B2/B2B2/B % Gun timing jitterΔt0Δt psec Initial bunch charge Q/Q0Q/Q %

Sept 16-17, 2003 Linac Coherent Light Source Stanford Synchrotron Radiation Laboratory Stanford Linear Accelerator Center Controls Timing Workshop Patrick Krejcik, LCLS Distributions of core-slice-averaged values for the beam and FEL for 227 seeds

Sept 16-17, 2003 Linac Coherent Light Source Stanford Synchrotron Radiation Laboratory Stanford Linear Accelerator Center Controls Timing Workshop Patrick Krejcik, LCLS Scientific user requirements Maintaining saturation in the FEL Provide femtosecond timing for pump-probe experiments Time stamp arrival of FEL pulse w.r.t. an optical laser pulse i.e. synchronize user laser with linac RF reference Again, coarse timing of RF bucket (see OTR diagnostic) And jitter at subpicosecond level Maintaining saturation in the FEL Provide femtosecond timing for pump-probe experiments Time stamp arrival of FEL pulse w.r.t. an optical laser pulse i.e. synchronize user laser with linac RF reference Again, coarse timing of RF bucket (see OTR diagnostic) And jitter at subpicosecond level

Sept 16-17, 2003 Linac Coherent Light Source Stanford Synchrotron Radiation Laboratory Stanford Linear Accelerator Center Controls Timing Workshop Patrick Krejcik, LCLS Compatibility and operational requirements Timing system also has a supervisory role Software inputs distinguish between different beam pulses (beam codes) and controls repetition rate LCLS upgrades are foreseen for bunch trains to multiple users, with differing beam parameters Timing logic also includes MPS and PPS functions Single bunch dumper, or gun veto Timing must coexist with controls of PEP II beams in adjacent beam lines, through the same micros. Linac configuration for LCLS must be revertable to fixed target operation (interleaving not yet foreseen) Timing system also has a supervisory role Software inputs distinguish between different beam pulses (beam codes) and controls repetition rate LCLS upgrades are foreseen for bunch trains to multiple users, with differing beam parameters Timing logic also includes MPS and PPS functions Single bunch dumper, or gun veto Timing must coexist with controls of PEP II beams in adjacent beam lines, through the same micros. Linac configuration for LCLS must be revertable to fixed target operation (interleaving not yet foreseen)

Sept 16-17, 2003 Linac Coherent Light Source Stanford Synchrotron Radiation Laboratory Stanford Linear Accelerator Center Controls Timing Workshop Patrick Krejcik, LCLS Operational tuning requirements LCLS is a single pass machine, not a storage ring Pulse-to-pulse jitter is a performance parameter It must be diagnosed And minimized through tuning Diagnostic devices*, especially BPMS, need single pulse readback All BPMS, from gun to laser, should readback on the same pulse (single pulse orbit display) All BPMS should have 120 Hz buffered data acquisition capability * other gatable devices include FEL powermeters etc. LCLS is a single pass machine, not a storage ring Pulse-to-pulse jitter is a performance parameter It must be diagnosed And minimized through tuning Diagnostic devices*, especially BPMS, need single pulse readback All BPMS, from gun to laser, should readback on the same pulse (single pulse orbit display) All BPMS should have 120 Hz buffered data acquisition capability * other gatable devices include FEL powermeters etc.

Sept 16-17, 2003 Linac Coherent Light Source Stanford Synchrotron Radiation Laboratory Stanford Linear Accelerator Center Controls Timing Workshop Patrick Krejcik, LCLS Operational tuning requirements: beam based feedback Pulse-to-pulse stability is set by the hardware Long-term drifts are to be corrected by feedback Feedbacks control (experience with SPPS) Transverse position and angle launch in the undulator Energy at the injector, chicanes, undulator Beam phase Bunch length Feedbacks should operate as fast as possible 120 Hz Pulse-to-pulse stability is set by the hardware Long-term drifts are to be corrected by feedback Feedbacks control (experience with SPPS) Transverse position and angle launch in the undulator Energy at the injector, chicanes, undulator Beam phase Bunch length Feedbacks should operate as fast as possible 120 Hz

Sept 16-17, 2003 Linac Coherent Light Source Stanford Synchrotron Radiation Laboratory Stanford Linear Accelerator Center Controls Timing Workshop Patrick Krejcik, LCLS Some concepts The timing system has 3 levels of inputs An RF clock harmonic of 2856 MHz RF A 360 Hz clock, phase locked to the AC power A software beam code from a master patern generator (MPG) The timing system has 3 levels of inputs An RF clock harmonic of 2856 MHz RF A 360 Hz clock, phase locked to the AC power A software beam code from a master patern generator (MPG)

Sept 16-17, 2003 Linac Coherent Light Source Stanford Synchrotron Radiation Laboratory Stanford Linear Accelerator Center Controls Timing Workshop Patrick Krejcik, LCLS RF frequency choices The PEP II timing system is derived from the 8.5 MHz damping ring revolution frequency The 56 th harmonic, 476 MHz is transmitted on the main drive line (MDL) In the crate a PDU counts cycles of the 4 th subharmonic at 119 MHz to allow step changes of 8.4 ns The PEP II timing system is derived from the 8.5 MHz damping ring revolution frequency The 56 th harmonic, 476 MHz is transmitted on the main drive line (MDL) In the crate a PDU counts cycles of the 4 th subharmonic at 119 MHz to allow step changes of 8.4 ns

Sept 16-17, 2003 Linac Coherent Light Source Stanford Synchrotron Radiation Laboratory Stanford Linear Accelerator Center Controls Timing Workshop Patrick Krejcik, LCLS RF frequency choices – LCLS gun laser A local oscillator isolates the laser from noise on the MDL and from phase jumps during PEP II injection It relocks to the MDL between PEP II cycles LCLS beams are on a different time slot to PEP II The choice of local oscillator frequency is driven by compatibility with the (unused) 8.5 MHz and by the length of available commercial laser cavities A local oscillator isolates the laser from noise on the MDL and from phase jumps during PEP II injection It relocks to the MDL between PEP II cycles LCLS beams are on a different time slot to PEP II The choice of local oscillator frequency is driven by compatibility with the (unused) 8.5 MHz and by the length of available commercial laser cavities

Sept 16-17, 2003 Linac Coherent Light Source Stanford Synchrotron Radiation Laboratory Stanford Linear Accelerator Center Controls Timing Workshop Patrick Krejcik, LCLS RF frequency choices – LCLS gun laser CDR quotes laser frequency of MHz This is the 36th sub-harmonic of 2856 MHz. For comparison, the damping ring 8.5-MHz revolution frequency is the 336th sub-harmonic of 2856 MHz. The sector-0 master oscillator VCO and the LCLS VCO frequencies are therefore in the ratio 6:56. Design would be simplified if laser could be made to operate at 119 MHz Technology is improving SPPS Ti:Sapphire pump-probe laser is now operating at 102 MHz, te 12 th harmonic of 8.5 MHz CDR quotes laser frequency of MHz This is the 36th sub-harmonic of 2856 MHz. For comparison, the damping ring 8.5-MHz revolution frequency is the 336th sub-harmonic of 2856 MHz. The sector-0 master oscillator VCO and the LCLS VCO frequencies are therefore in the ratio 6:56. Design would be simplified if laser could be made to operate at 119 MHz Technology is improving SPPS Ti:Sapphire pump-probe laser is now operating at 102 MHz, te 12 th harmonic of 8.5 MHz

Sept 16-17, 2003 Linac Coherent Light Source Stanford Synchrotron Radiation Laboratory Stanford Linear Accelerator Center Controls Timing Workshop Patrick Krejcik, LCLS Timing and rf distribution in sector-0 and sector-20 of the linac

Sept 16-17, 2003 Linac Coherent Light Source Stanford Synchrotron Radiation Laboratory Stanford Linear Accelerator Center Controls Timing Workshop Patrick Krejcik, LCLS Functional Requirements Maximum Link Length2 Kilometers Timing Stability (Long Term)< 5 picoseconds Timing Jitter RMS< 0.5 picoseconds RF Phase Stability (1 second)< 0.07 deg. S- band RF Phase Stability (Long Term)< 5 picoseconds RF Phase Jitter RMS< 0.07 degree S- band Phase Transmission Frequency2856 MHz Timing Resolution (Normal)350 ps (S-band bucket) Timing Resolution (with vernier)1 ps Required Frequency Stability3x10e-9 Maximum Link Length2 Kilometers Timing Stability (Long Term)< 5 picoseconds Timing Jitter RMS< 0.5 picoseconds RF Phase Stability (1 second)< 0.07 deg. S- band RF Phase Stability (Long Term)< 5 picoseconds RF Phase Jitter RMS< 0.07 degree S- band Phase Transmission Frequency2856 MHz Timing Resolution (Normal)350 ps (S-band bucket) Timing Resolution (with vernier)1 ps Required Frequency Stability3x10e-9

Sept 16-17, 2003 Linac Coherent Light Source Stanford Synchrotron Radiation Laboratory Stanford Linear Accelerator Center Controls Timing Workshop Patrick Krejcik, LCLS Control system components at each klystron station

Sept 16-17, 2003 Linac Coherent Light Source Stanford Synchrotron Radiation Laboratory Stanford Linear Accelerator Center Controls Timing Workshop Patrick Krejcik, LCLS Schematic of two adjacent nominal sectors showing distribution of rf power to the klystrons

Sept 16-17, 2003 Linac Coherent Light Source Stanford Synchrotron Radiation Laboratory Stanford Linear Accelerator Center Controls Timing Workshop Patrick Krejcik, LCLS Klystron phase stable to <0.1 deg. S-band over ~10 sec. Pulse-to-pulse phase variations, and histogram, measured at PAD of a single klystron shows degree S-band rms variation over 17 seconds. Pulse-to-pulse relative amplitude variations measured at the PAD of a single klystron shows 0.06% rms variation over 2 sec (horizontal axis is in 1/30-sec ticks).

Sept 16-17, 2003 Linac Coherent Light Source Stanford Synchrotron Radiation Laboratory Stanford Linear Accelerator Center Controls Timing Workshop Patrick Krejcik, LCLS 0.5 deg. S-band klystron phase variation over several minutes Phase variations measured at the PAD of a single klystron over a period of minutes. Each point is an average over 32 beam pulses.

Sept 16-17, 2003 Linac Coherent Light Source Stanford Synchrotron Radiation Laboratory Stanford Linear Accelerator Center Controls Timing Workshop Patrick Krejcik, LCLS Long term stability dominated by RF phase drifts Measurement of phase variations seen along the linac main drive line over a period of several days. Measurement of the phase variations between two adjacent linac sectors over a period of several days