LCLS_I Diagnostics Lessons Learned for LCLS_II

Josef Frisch

Diagnostics Functions

• Provide data to beam feedbacks• Provide experiments with beam parameters• Allow beam tuning• Diagnose problems– Need to work when the accelerator is NOT operating

correctly.– Need to be able to discover when the diagnostic IS the

problem. • Support machine physics studies– Need to measure things we haven’t thought of yet.

LCLS_I

• Seemed easy (if you weren’t there)• In reality things went wrong (COTR, Toroids, etc) but

we had backups• Machine operating modes are very different from

baseline (1nC, 800-8.3 KeV)– Now run 10pC to 250pC 480 – 11 KeV

• We will probably have a different set of problems in LCLS_II.

• Need to assume that LCLS_II may not operate with the parameters that we expect.

Stripline BPMs

• LCLS striplines were in general very successful– Resolution ~5um at 250pC: Sufficient to tune machine

down to 10pC.– Low power “coldfire” processor caused software

difficulties.• LCLS_II has the same requirements with the

additions:– Likely will want 2 bunches (~50ns separation)– May want to operate with lower charge for ultra-short

bunches.

LCLS_II Stripline BPMs• Developing technology allows higher speed digitizers and higher bandwidth

– Lower noise, or can tolerate cables with more loss– More powerful processors to allow multi-bunch decoding (electronics is OK).

• Expect to package in uTCA, but design and performance are not dependant on form factor.

• Will initially use existing cable plant in S10-20. Can upgrade later for lower charge operation if needed.

• Try to find a common processing frequency for all stripline lengths – will simplify 2-bunch operation.

• Try to find a processing frequency that is compatible with low transverse-wakes multi-bunch operation– BPMs are the only frequency in the system that are not harmonic with 2856MHz so

they may set the multi-bunch frequency– Lots of technical issues with selecting the frequency. – Frequency selection has minimal impact on the rest of the design.

Cavity BPMs• LCLS design had acceptable resolution• Reliability and drift issues were caused by a manufacturing

defect in the front end. • Waveguide seals were problematic– Problems with leaks.– Waveguide is lower loss than coax, but not needed – theoretical

noise is only a few nm.– Waveguides transmit high harmonics which caused problems

for front end electronics.• LCLS_II BPM requirements are the same but expect

somewhat different design

LCLS_II Cavity BPMs• LCLS_I cavity design worked

– Replace waveguide connections with coax. Additional loss is OK, we are far from theoretical noise limit.

• Use single sideband PC board electronics similar to that used for ATF2 cavities (20nm resolution)– Lower cost and more compact – Easy to fabricate more boards if there are problems.– Can package as uTCA RTM module

• Operate with unlocked digitizer clock and LO to simplify cable plant.– Need to demonstrate that this does not increase noise

• Collaboration? Pohang FEL project has almost identical cavity BPM requirements. – Pohang built the cavity BPMs used for the ATF2 project – very sucessful!– Have Pohang build the cavities, SLAC does the electronics?

Charge Calibration• BPMs provide a sum “TMIT” signal that has good stability,

but must be calibrated against an absolute measurement• Toroids and Faraday cups were provided for calibration but

neither worked correctly.• LCLS now relies on an old SLC toroid as its absolute

current calibration!• Why is absolute charge calibration important?

– Need to cross-correlate LCLS_I and LCLS_II operating charge: If the machines behave differently is it because they are operating at a different bunch charge?

– Important for understanding beam physics

Faraday Cups• Never worked properly for LCLS

– They were not CUPS!!!!– Secondary and scattered electrons were not contained – can cause

large errors in measured charge• Faraday cup design is well understood but requires significant

space. • A single fixed faraday cup in the spectrometer line can provide

charge calibration– Large space, does not need to move.– Beam losses are small through the accelerator

• May also have a faraday cup at the gun spectrometer to calibrate charge before the beam is accelerated.

Toroids• Should provide absolute calibration

– Both “by design”, and with a calibration loop• LCLS toroids were required to have high bandwidth to separate dark

current from beam– Very difficult design – didn’t work properly– Can turn beam on / off to separate dark current

• LCLS_II will use low bandwidth toroids to integrate beam current. • Will use a pair of (probably commercial) toroids to cross calibrate against

LCLS_I– Mount both toroids in LCLS_1 after BC1 to cross calibrate– Move one to LCLS_2 after BC1

• Only need 1 toroid for calibration, but keep BCS toroids to avoid re-design.– Can use low bandwidth toroids for BCS – but is it worth the change?

Profile Monitors

• Ce:YAG screens– Work well at gun energy, but saturate with the smaller

spots at high energy.• OTR screens– COTR light makes OTR screens unusable after OTR2. – Screens in dispersive regions can provide some

qualitative information• Wire Scanners– Only known way to measure high brightness beam

profiles.

YAG Screens

• LCLS Screens have good performance• Some mechanical problems with actuation

system– Not clear if this warrants a re-design

• Dump screen requires study– Large energy loss and energy spread from TW FEL

operation.

OTR Screens• Coherent OTR prevents useful measurements• No known solution

– Coherent enhancement is large >104

– OTR is partially coherent – can not assume fully coherent– Microbunching is not uniform – some parts of the beam are enhanced

relative to others– Microbunching extends into the near UV, expected to reach 100nm

• Problem expected to be worse for low charge beams.– With 1pC bunch FWHM is <100nm.

• May be possible to use XUV OTR light, but this needs extensive R&D.

Wire Scanners• Installed in LCLS as a backup, but now the primary profile

diagnostic• Significant tuning time is spent doing wire scans• In some locations it is possible to steer the beam across the

wire for high speed scans.• Two different high speed wire scanner designs under

development– High speed external linear motor– In-vacuum mover

• Can use LCLS_I design, but should switch to high speed scanners if available.

Spectrometers

• Dispersive regions in DL1, BC1, BC2, DL2, Dump– All except dump used in beam feedback– Dump energy loss used to measure FEL pulse energy

• BPMs used for centroid measurement• Wires used for profile measurement• LCLS_II changes:– Software should fit to beam orbit– Dump spectrometer may need modification (next slide)

Dump Spectrometer for TW operation

• Long tapered undulator to generate TW beams is under consideration for LCLS_II

• Energy spread and energy loss ~few %.• Can compensate for average energy loss with a current shunt

on the dump bend– The shunt can be current limited to preserve BCS requirements

• May still need to design the dump for larger energy acceptance.

• Large energy acceptance may make low power energy loss measurements difficult– Can use thermal sensor for calibrated measurements.

Transverse Cavities

• Only quantitative longitudinal profile measurement• TCAV0 works well, no changes required• TCAV3 works for high charge, but insufficient

resolution for low charge– Use X-band for LCLS_II ?– Eliminate and use dump TCAVs

• XTCAV being installed in the LCLS dump. – Will provide high resolution (~2fs) FEL temporal

measurements– Should include XTCAVs for LCLS_II dumps

Bunch Length Monitors• Pyroelectric bunch length monitors work well for LCLS• May be possible to simplify the design – but may not be worth the

engineering cost.• Want higher sensitivity detector for BC1 for operation with low

charge• Need to investigate widow options – need wider spectral range for

low charges. – Diamond will probably work – in use for THz system

• Multi-bunch operation requires fast response detectors – Need to test pyro detector response time– Fast response and good sensitivity may not be compatible– R&D required

Phase Cavities• LCLS Injector and Linac phase cavities work, but are rarely

used• Beam phase determined by accelerator structure phase

scans• Gun and laser phase sufficiently stable that scans are only

required 1/shift. • Phase cavity after each undulator required to provide

precision timing to the dump– LCLS system works– Evaluate upgrading to X-band for better resolution

• LCLS_II undulator cavity BPMs

Loss Monitors• Fiber loss monitors work well

– Will not have position resolution when used in regions with multi-bunch beams

• Undulator loss monitors– LANL “fork” design unnecessarily complex and has non-uniform response.– Can replace with Cherenkov radiator bars or scintillators – need to

evaluate whether we need top and bottom or just one per segment– Need BSA readout system – link nodes not designed for this– Need integrating monitors (fiber, RADFET) for accurate total dose

measurement. Remote readout is desirable.– Needs some R&D

• BCS for LCLS_1 / LCLS_2 interaction needs to be considered– Trip both beams on any loss?

Decisions:• Stripline BPMs:

– Study linac BPM frequency• Cavity BPMs:

– Demonstrate low noise with unlocked clock– Collaborate with Pohang on cavity BPMs?

• Charge Calibration:– Develop new toroid and Faraday cup design.

• Profile monitors– Improved mechanical motion on YAG and OTR– Faster wire scanner design– Fix COTR (very difficult)

• Spectrometer– Design broad band dump spectrometer

• TCAV– X-band in linac? TCAV after each undulator?

• Bunch length monitor– Develop fast mm-wave detector

• Loss Monitors– Develop improved undulator loss monitor