LCLS Energy Jitter Status in 2012 Franz-Josef Decker 24-Oct-2012 thanks to: J. Turner, R. Akre, J. Craft, A. Krasnykh, M. Nguyen, W. Colocho, for helping.

LCLS Energy Jitter Status in 2012 Franz-Josef Decker 24-Oct-2012 thanks to: J. Turner, R. Akre, J. Craft, A. Krasnykh, M. Nguyen, W. Colocho, for helping to reduce jitter over the years 2 Outline LCLS Jitter Status in 2012 History of L1S jitter How does a high voltage change of L1S modulator propagate History of DL2 energy jitter (or different energies, peak currents) -Jitter is 120% (high E) to 250% (low E) of energy spread [0.040.12%] What is needed for seeding Different techniques to localize and quantify jitter 3 L1S phase jitter over the years 0.35 deg to 0.03 deg LCLS Jitter Status in 2012 Sample images HV=300kVBC1: E =250 MeV Un-SLEDed, HV=340kV ? 4 L1S Phase and Amplitude Calibration vs. High Voltage Phase -22 deg and Amplitude +145 MeV (BC1: E = 250 MeV) 5.0 deg / % voltage change 2.5 MeV / % voltage change [expected: 5.2 deg, 1.74 MeV (sqrt 2?)] 0.1% 0.5 deg, 0.25 MeV Now at deg, 115 MeV gives: E= MeV MeV MeV [0.42 (phase) (ampl)] 0.66/220 = 0.3% more energy in BC1 R56 = mm deg in L2 -36 deg deg: 5 GeV 4.97 GeV % (less energy) [R56 = mm deg or deg in L3, (L2 phase can be compensated with +18 deg in L3)] LCLS Jitter Status in 2012 V/V = 100 PPM (0.05 deg) 0.057% at 5 GeV 5 Expected vs Measured Slope BC2 E vs BC1 E Expected: / 0.3 = -1.9 Measured: /0.38 = 0.16 corrcoef = rho LCLS Jitter Status in 2012 6 DL2 Energy vs BC2 Energy Expected: 5 GeV / 12.3 GeV = 41% Measured: 36% / = 48% [ = 0.044% / 0.091%] (peak current increases jitter 20%) [over-compression reduces jitter] LCLS Jitter Status in 2012 7 Energy Jitter History at DL % at 3.5 GeV 3.0 times 0.04 % at 13.5 GeV 3.85 times LCLS Jitter Status in 2012 8 Effect of Peak Current Best energy jitter 0.03 %, long bunch, high energy LCLS Jitter Status in 2012 9 LTU Wire E/E 12.3 GeV Hard X-ray 3.5 GeV Soft X-ray LCLS Jitter Status in % Jitter % 250 % = % = % Jitter % 140 % % Good: % + % = % 10 How does the Energy Jitter Influence Seeding? Intensity on Direct Imager after Simulation: Jitter paints the KMONO [SASE: 0.07%] vs LTU x distribution 0.052/1.25 = 0.042% [seeded sometimes 0.030% 2] LCLS Jitter Status in 2012 11 Average Intensity and Variation, Seeded after KMONO 70% for 0.042% energy jitter, variation 40% rms 90% for 0.020% energy jitter, variation 10% rms LCLS Jitter Status in * 0.020% energy jitter requires V/V = 80 PPM or 0.04 at L1S [= 0.020/1.2*13.6/5 / *100 PPM] 12 L1S phase is still the worst offender of energy jitter [12% of jitter power at 120 Hz,] but 35% at 60 Hz But other devices are close by: 60 Hz 120 Hz L1S 24-1 L0A L0B LCLS Jitter Status in 2012 13 Main Causes of L1S Jitter A)RF: a)Multi-pacting in load and reflecting back up and DOWN b)Single pulse break down (?) Both avoided by going un_SLEDed (after 220 MeV in BC1) B)Modulator HV: a)Thyratron not switching off cleanly, affecting next pulse Avoid by going to 300 kV (costs ~30 MeV, 190 MeV in BC1?) or MANY other approaches LCLS Jitter Status in 2012 14 L1S Modulator Modifications over the Years Try to match modulator to klystron impedance (high micro- perveance helps) Change klystron and a few thyratrons, and ranging them Run at 300 kV instead of 340 kV Shorter PFN Tail end clipper Mystery fan Delay charging PFN after thyratron firing [ended in fire 120Hz] Measuring and correcting HV (ala SCALA) going on now Other ideas: capacitors with lower inductance LCLS Jitter Status in 2012 15 Techniques to Localize and Quantify Jitter 1.Methods to get data: 1.BSA (Beam Synchronous Acquisition) 2.Correlation plot 3.get_wf (waveforms) 4.L1S_jit (Jitter vs time) 5.Jitter CUD, FTP from matlab, 6.LLRFgui, KLYSTRONgui, 2.Analyzing data: 1.BSA_jitter_fft (FFT, corrcoef, slope, ), svd 2.corr_jitt 3.get_wf_an14 4.Plot History LCLS Jitter Status in 2012 16 BSA Data (Power Spectrum in DL2 Energy BPM) Feedback on: 0.054% Feedback off: % (even with big 60 Hz component, without 0.040%) LCLS Jitter Status in 2012 17 Correlation Plot and BSA [Transverse and FEL Intensity] BPM:IN20:371:Y is 75% correlated with VCC x position So laser causes at least 56% (=0.75^2) of 371 Y jitter (Is 42 Hz coming from laser OR MCOR power supply?) LCLS Jitter Status in 2012 From injector laser ? 5 Hz data 18 Jitter from Waveforms, e.g. L1S 300 raw waveforms, with RMS*1000 (red) revealing some elevated jitter inside the pulse, reflected signals show early multi-pacting Insert Presentation Title in Slide Master 19 DL2 and BC1 Pie Charts Much of DL1 explainedRest of BC1 is big (maybe 60 Hz shows more), Laser_P is big, but gets compressed, BUT FBs sees it LCLS Jitter Status in 2012 20 Localize Jitter vs z Concentrate on source: 16.7 Hz Make IFFT of FFT-data with only that frequency, then plot max-min vs z, fit oscillation LCLS Jitter Status in 2012 BPM_pvs(100) = 'BPMS:BSY0:61' 21 SVD for 1 st BPM in DL2 Insert Presentation Title in Slide Master 60 Hz L0A_A L0B_A last pulse missing, not used L0A_A L0B_A last pulse missing, not used L0A_A L0B_A MDL 476 Thales Ref L0B_A DL2 BC2 22 Summary L1S makes ~36% of DL2 energy jitter power: 0.6^ ^2 = = 1.0 Energy jitter is 0.04% for hard x-ray, 0.12% for soft [0.18%] Lowering BC1 energy to 190 MeV, might avoid current modulator instability Hard seeding need about 0.02%, [soft seeding ?] Presented different methods to track down and quantify jitter LCLS Jitter Status in 2012