Breakout Session Controls Timing and Event System S
Breakout Session: Controls Timing and Event System S. Allison, M. Browne, B. Dalesio, J. Dusatko, R. Fuller, A. Gromme, D. Kotturi, P. Krejcik, S. Norum, D. Rogind, H. Shoaee, M. Zelazny July 10 -12, 2007 LCLS Lehman Review Patrick Krejcik pkr@slac. stanford. edu
Outline Overview and HW Block Diagram Performance of timing system for injector commissioning Outstanding Issues Task List BC 2 installation Status & Long-Haul Fiber Issues July 10 -12, 2007 LCLS Lehman Review 2 Patrick Krejcik pkr@slac. stanford. edu
Timing and Event Architecture New LCLS Timing System ~ Linac main drive line Low Level RF FIDO 119 MHz SLC MPG 360 Hz P E I LCLS N SLC O events V E events C G T * EPICS Network *Micro. Research P S m N T P E B T July 10 -12, 2007 LCLS Lehman Review F A N Beam Synchronous Acquisition fiber distribution I E O V C R* Precision<10 ps D E V TTL-NIM convert. SLC Trigs 3 Digitizer LLRF BPMs Toroids Cameras GADCs SLC klystrons Patrick Krejcik pkr@slac. stanford. edu
Hardware Block Diagram 2007 Commissioning Modulator Triggers Existing Control System RF Timing Future MPS Beam Path Timing Crate F F P E 360 Hz A A N V Fiducial 119 MHz Clock I O N N E G 1 2 T C Acq and BPM FEE Calibration Triggers E I BPM V O Crates R C Beam Rate, Beam Path E V. . . R 3 1 1 1 8 … July 10 -12, 2007 LCLS Lehman Review Acc and Standby Triggers E V LLRF R Crate 1 E V R 2 E I V O R C 3 Fiber Distribution: Timing Pattern, Timestamp, Event Codes TORO FEE Trigger F E C I Laser Steering A V A O N R M C Crate PADs and PACs C A M 8 I O C 8 F E I Toro A Farc V O N Crate R C 4 4 Triggers Profile E C C I Monitor V A A O R MM C Crate E V … R 1 1 2 1 4 … Patrick Krejcik pkr@slac. stanford. edu Triggers C A M 7 C A M 8 I O C 4
Progress since Oct, 2006 Installation in sector 20 and 21 commissioned ! Successfully integrated it with the SLC Timing System Finished hardware bench testing Finished PMC-EVR driver EVG sequence RAM programming at 360 Hz, conditional logic is fairly basic, but does the job. Event pattern records and timestamp distribution on the EVR IOCs Finished cabling plans and documentation Finished beam-synchronous acquisition July 10 -12, 2007 LCLS Lehman Review 5 Patrick Krejcik pkr@slac. stanford. edu
IN 20 Installation Diagram shows IN 20 / LI 20 Event System installation Top-Level shows connections to SLC Timing, MDL The EVG Crate is shown with its fanouts to all subsystems July 10 -12, 2007 LCLS Lehman Review 6 Patrick Krejcik pkr@slac. stanford. edu
Tallies for 2007 Commissioning # EVRs = 31 (mostly PMC) # IOCs with EVRs = 28 # EVR Fanouts = 4 # Hardware Triggers = 120 and counting All TTL except 2 NIM triggers for QDCs Most require short cables (except LLRF) EVR with clock now available for BC 2 installation All acquisition electronics using either internal clocks, clock output from the RF timing system, or other external clocks July 10 -12, 2007 LCLS Lehman Review 7 Patrick Krejcik pkr@slac. stanford. edu
Event System Functionality Event Generator IOC: Send out proper event codes at 360 Hz based on: PNET pattern input (beam code and bits that define beam path and other conditions) Add LCLS conditions such as BPM calibration on off-beam pulses , diagnostic pulse etc. Future – event codes also based on new MPS and user input Send out system timestamp with encoded pulse ID from PNET Send out PNET pattern to be used by SLC-aware IOCs Manage user-defined beam-synchronous acquisition measurement definitions (event definition or EDEF) Check for match between user EDEFs and input PNET pattern at 360 Hz and tag matches in outgoing pattern July 10 -12, 2007 LCLS Lehman Review 8 Patrick Krejcik pkr@slac. stanford. edu
Event System Functionality, cont Event Receiver IOC: Set trigger delays, pulse widths, and enable/disable via user requests (not yet done on a pulse-by-pulse basis) Set event code per trigger (triggering done in HW when event code received) Receive event pattern 8. 3 msec before corresponding pulse Perform beam-synchronous acquisition based on tags set by EVG in the event pattern Perform beam-synchronous acquisition for the SLC-aware IOC based on the PNET part of the event pattern Process pre-defined records when specific event codes are received (not yet available – bug!) July 10 -12, 2007 LCLS Lehman Review 9 Patrick Krejcik pkr@slac. stanford. edu
EVG Event Time Line – 4 Fiducials 360 Hz Fiducial F 0 (n=0) Time (msec) 0 1. 0 HW starts sending R 0 event codes, starting with fiducial event code Receive Fn+3 PNET, determine Fn+3 LCLS P 0 pattern, and advance pipeline (n-2 ->n-1 ->n) Send LCLS pattern Set Event Codes in Alt RAM based on the last patterns for Fn+1 120 Hz BEAM July 10 -12, 2007 LCLS Lehman Review L 0 E 1 F 3 (n=3) F 1 (n=1) 2. 8 R 1 F 2 (n=2) 5. 6 R 2 8. 3 9. 3 R 3 P 1 P 2 P 3 L 1 E 2 L 2 E 3 L 3 E 4 B-3 B 0 10 Patrick Krejcik pkr@slac. stanford. edu
Trigger Event Time Line – 1 Beam Pulse (B 0) Record processing (event, interrupt) Hardware Triggers Triggering Event Codes Event Timestamp, Start pattern records, and BSA ready Receive pattern for 3 pulses ahead Kly Standby Beam Acq Kly Accel Trigger Fiducial Event Received Fiducial B 0 F 3 Time (usec) July 10 -12, 2007 LCLS Lehman Review 0 18 0. 3 100 500 11 1023 Patrick Krejcik pkr@slac. stanford. edu
Trigger Control Display July 10 -12, 2007 LCLS Lehman Review 12 Patrick Krejcik pkr@slac. stanford. edu
Trigger Event Assignment Display July 10 -12, 2007 LCLS Lehman Review 13 Patrick Krejcik pkr@slac. stanford. edu
EVG Displays July 10 -12, 2007 LCLS Lehman Review 14 Patrick Krejcik pkr@slac. stanford. edu
Issues The Event System HW is performing as expected At the first order - still in its infancy Outstanding problem (presumably software) where CPU hangs when EVR interrupts are enabled. Some IOCs running without interrupts (hardware triggers but without timestamps, BSA, or event code IRQs). Documentation and help screens still to be done. EVG IOC doesn’t know when there’s no beam, only when beam is possible – limits trigger conditions. Software not yet in place to handle hardware and communication errors. July 10 -12, 2007 LCLS Lehman Review 15 Patrick Krejcik pkr@slac. stanford. edu
LCLS Event System for Beam Synchronous Data Acquisition All pulsed readback devices BPMs Cameras RF phase and amplitude Bunch length monitors Loss monitors Can be read simultaneously on a single pulse, throughout the machine With a repetition rate of 120 Hz July 10 -12, 2007 LCLS Lehman Review 16 Patrick Krejcik pkr@slac. stanford. edu
Beam Synchronous Acquisition (BSA) EVG Event Definitions (EDEFs) are used for BSA EDEF includes: Beam code SLC PNET bits and LCLS Modifier bits No. of pulses to average, no. of pulses to measure Client or System name Permanent System EDEFs that are always active: 1 Hz, 10 Hz, Full, Single Shot, Feedback Available to all clients Client applications may reserve a Client EDEF Client receives EDEF # assignment on demand Client proceeds to fill out definition, then sets “GO” July 10 -12, 2007 LCLS Lehman Review 17 Patrick Krejcik pkr@slac. stanford. edu
User Defs Slots for 20 possible Event Definitions allocated Fixed Defs July 10 -12, 2007 LCLS Lehman Review 18 Patrick Krejcik pkr@slac. stanford. edu
User interface for defining an Event Def July 10 -12, 2007 LCLS Lehman Review 19 Patrick Krejcik pkr@slac. stanford. edu
Example of Data Fanout 20 Buffers One for each EDEF 2800 samples per buffer July 10 -12, 2007 LCLS Lehman Review 20 Patrick Krejcik pkr@slac. stanford. edu
Current Activities Software - fix hang problem, add HW error detection and recovery Next installation phase is BC 2 (sectors 21 to 30): 60 EVRs to be ordered - 34 VME, 26 PMC – critical path 5 fanout modules (1 every 2 sectors) SW - only DB and display work planned Long Haul Fiber Test Integrate new LCLS MPS and Master Pattern Generator (MPG): Make EVG aware of MPS trips Rate-liming logic including beam “burst” and “single-shot” modes Send out PNET pattern to EVG and CAMAC controls (for modulator triggers) Beyond BC 2: BSY/LTU/Undulator Procurement and Plans Changes to EVR firmware to change behavior when multiple triggers on a single pulse Generic trigger delay scan package Bunch charge change on pulse-by-pulse basis? More complex conditional logic for event codes (EVG) and pulse-to-pulse changes in trigger attributes? July 10 -12, 2007 LCLS Lehman Review 21 Patrick Krejcik pkr@slac. stanford. edu
Challenges in extending the LCLS timing system beyond the injector biggest challenge is Long-Haul distribution of timing data via the Fiber-Optical (F-O) Links Reasons MRF Event System is designed around Multi-Mode SFP (Small Form-Factor Pluggable) F-O links which allow a maximum fiber length of ~300 meters Our requirements include runs of several Kilometers (at least) Cannot daisy-chain the event F-O links: exceeds the jitter budget Temperature effects on long fibers – drift, but how bad? July 10 -12, 2007 LCLS Lehman Review 22 Patrick Krejcik pkr@slac. stanford. edu
Long-Haul Fiber Distribution Proposed Solution: New HW and some testing Single-Mode, pin-compatible SFP modules are commercially available (Agilent ACFT-57 R 5) Allows us to go ~10 KM Plugs into our event system HW Does not solve temperature-induced phase-drift problem Solve this by: Running Long Fiber in temperature-controlled environment Re-Syncing EVRs to a locally-distributed 119 MHz source Long-haul fiber test: Function of Single-Mode SFP Modules Drift, Trigger time jitter, Phase Noise July 10 -12, 2007 LCLS Lehman Review 23 Patrick Krejcik pkr@slac. stanford. edu
Long-Haul Event Fiber System Test What it will determine: Function of Single. Mode SFP Modules Drift, Trigger time jitter, Phase Noise Compare two EVR’s triggers & RF Clocks One EVR Short-Haul Other EVR Long-Haul Also compare to reference RF Clock July 10 -12, 2007 LCLS Lehman Review 24 Patrick Krejcik pkr@slac. stanford. edu
- Slides: 24