PEP II RF Systems by pengxiuhui

VIEWS: 4 PAGES: 27

									     LLRF upgrades and new
         Comb-2 project
                     Mike Browne




    D. Brown, R. Chestnut, J Dusatko, M. Laznovsky,
              W. Ross, W. Roster, C Yee


1                         MJB, PEP-II MAC Review, December 13-15, 2004
                    Outline

    • System Performance Summary
    • System Issues
    • LLRF Module Status
        •   IOC
        •   CLOCK
        •   IQA
        •   AIM
        •   RFP
        •   COMB
    • Other Status
    • COMB-2 Project


2                     MJB, PEP-II MAC Review, December 13-15, 2004
             RF Fault Summary for PEP-II RUN 4
    RF fault analysis over 313 days (09/22/2003 - 07/30/2004)  1190 faults
          Average of 3.8 aborts per day (2.9 without non-RF events)
                          Compiled by Dmitry Teytelman & Dan Van Winkle




         LLRF?
          LLRF



          LLRF




3                                     MJB, PEP-II MAC Review, December 13-15, 2004
                      LLRF System Issues
    • The Following Issues Surfaced During RUN 4:
       – Station Phase Drifts       - Lead to 47 Aborts
       – COMB Module Overflow Faults          - Lead to 34 Aborts
       – Various VXI Module Failures (Clock, COMB, RFP, IQA)

    • RUN 4 Also Saw Some System Improvements:
       – Upgraded to new RFP Modules in all stations
                     » 4 with new fixes
       – Upgraded to new IOCs in all stations
       – Upgraded to new IQAs for all cavity IQA locations
           • Used 8 channel scanning mode to obtain Fault data for all
             channels.
       – Tested new Arc Interlock Module new HVPS monitor channels
           • HVPS DC Voltage and current monitors
           • HVPS Ripple and fast Current Transformer Monitor
           • The Current Transformer will be helpful in analyzing station
             performance and trips
                     » tested on HR8-1

4                                         MJB, PEP-II MAC Review, December 13-15, 2004
                        LLRF System Issues
    • Phase Drifts
       – Isolated to 3 leading sources:
            • RFP Module Circuit Sensitivity
                     » Fixes Described later
            • LLRF System Rack Temperature Instability
            • Clock Module RF Amplifier Failures


    • COMB Overflow Faults
       – Developed Initial SW fix (“Band-Aid”)
       – Developed HW FIX to Resolve Problem

    • VXI Module Failures
       –   Component Aging
       –   Connector Wear & Tear
       –   Random Failures on Some Modules
       –   Swap & Repair as we found them


5                                        MJB, PEP-II MAC Review, December 13-15, 2004
                       LLRF Module Status - IOC
    • Old NI 68030/40MHz IOCs were replaced with KSC PPC 350MHz
      Units
    • The old IOCs were seriously performance limited
       – (80% – 90% loading)
    • A MAJOR development effort was undertaken to adapt the new IOC
      for LLRF: This work was performed by John Dusatko
       –   Modified 63 LLRF VXI Modules last summer to fix design issue
       –   Reverse-Engineered & Modified KSC IOCs to fix design bugs
       –   Developed Special SW / Fixes to KSC SW as well
       –   Had to adjust some LLRF system SW task timing for increased speed
            RESULT: Tremendous Performance Increase < 8% loading (typ)
            As of 7-13-2004 ALL LLRF Station Have Been Upgraded

    • Have 6 Spares / 2 destined for New LLRF Stations / Looking into
      getting more
    • Motorola Has Informed us that the MVME2400 will go away in 2005!
      (We have a recommended substitute)




6                                          MJB, PEP-II MAC Review, December 13-15, 2004
    LLRF Module Status - IOC cont.
        Station with OLD IOC:




        Station with NEW IOC:




7                  MJB, PEP-II MAC Review, December 13-15, 2004
                      LLRF Module Status - CLOCK

    •   Current System contains a mix of (8) old (R1) and (3) new (R2) modules
    •   Saw Several R1 Clock Failures This RUN
         –   Mainly Due to RF Amplifier Death (selected different device for repairs)
         –   Some Bad RF Connectors as well (Wear & Tear)
         –   Bad Modules Were Swapped and repaired
    •   Amplifier Failures Prompted a system-wide measurement of Clock RF
        output power levels.
         – We have replaced all old style RF Amps on all remaining Clock modules
           the during the downtime
    •   Originally only built (5) New Modules for Added Stations / Building qty = 10
        more R2 Clock Modules
    •   We suspect that some old CLKs are contributing to Phase Drift Problem
         – Considering replacing ALL clocks with R2 versions at some point




8                                              MJB, PEP-II MAC Review, December 13-15, 2004
                     LLRF Module Status - IQA

    • New Version (R3) of IQA was designed, built & installed (Built
      23 / 20 Installed) The new version has an eight channel mode,
      for displaying fault histories of all channels.
    • System now contains a mixture of old and new modules
    • IQAs have been performing fairly well / no major problems
    • We have seen several older modules with both intermittent and
      solid connector failures.
    • Discussed creating a Design Proposal for new version (more
      robust / enhanced features / simplified design/very few
      connectors)
       – This is a long-term item:
           • Not Terribly Critical at this point




9                                           MJB, PEP-II MAC Review, December 13-15, 2004
                    LLRF Module Status - AIM

     • More version 2 AIMs were built & installed in all stations
       during the summer down time.
         – Last run the system has mix of (3) old (R1) & (8) new (R2)
           modules

     • All stations were their HVPS channels wired to the new AIMs.




10                                    MJB, PEP-II MAC Review, December 13-15, 2004
                              LLRF Module Status - RFP

 •   New RFPs were installed in all 11 LLRF stations during last run.
 •   Phase Stability Improvements:
      – Discovered that several RFP circuits were contributing to phase drift and
        operating point errors
           • Reduce circuit impedance at coefficient input of the Gilbert-cell multipliers -reduced
             offset 5 times
           • Made the comb multiplier signal capacitive-coupled.
           • Changing amplifier operating point to reduce D.C. drift by a factor of 2
           Replaced old difference amplifier in multiplier circuits with lower drift amplifiers (>20x
             improvement)
           • Found bias current offset problems causing large “Tune” mode offset.

      – Developed corrections and tested  Successful: RFPs are now more stable, far
        less sensitive to temperature variation, smaller operating errors
      – (4) RFPs were tested in PEP-II at the end of Run 4.
      – All other RFPs have been modified during the summer downtime
 •   Other Work
      – Mike Laznowsky rewrote the Calibration SW to accommodate new functions and
        fix standing problems. His new code is much cleaner, compact and runs 10
        times faster than the old version. 2 minutes verses 20.




11                                                 MJB, PEP-II MAC Review, December 13-15, 2004
                             LLRF Module Status - COMB


     •       We saw many COMB overflow faults this run
         –       (this has been a known design flaw)
     •       Overflow typically causes spurs at ½ 136KHz rate that perturb the beam
     •       Created Two Fixes
         1.     SW “Band-Aid”: alert operators to manually clear fault
         2.     HW Fix: Add-on daughterboard -- Corrects Design Flaw / Overflow is now
                handled correctly (D. Teytleman & D. Van Winkle)
     •       A prototype of fix #2 was created and tested: good results -overflows
             stopped
         –      A Production version of this Daughterboard has been created, fabbed and
                assembled (qty = 30)
     •       All COMBs were upgraded with these daughter boards during the
             downtime.
     •       Meanwhile a new COMB-2 Module is being designed (W. Ross)
         –      More on this later



12                                             MJB, PEP-II MAC Review, December 13-15, 2004
                    Other Status - LLRF System Cooling

     •       In Investigating LLRF Phase Drift, we decided to look at the LLRF
             rack cooling system
     •       Discovered several issues:
         –      ELMA VXI Crate airflow design is poor
         –      A/C Units not operating efficiently / temperature not well-regulated /
                most compressor control relays were welded ON
         –      Overall thermal design is sub-optimal
     •       Devised Several Fixes
               1.   Insulate The LLRF Rack
               2.   Improve airflow steering in rack
               3.   Enhance the A/C unit with PID Loop / Heater / Blower Control
               4.   Modify the VXI crate for better airflow
     •       Have Tried first three fixes on LR4-5 and items 2 and 3 on LR4-3
         –      Results show that the temperature is now much more stable
                        »    See figure
     •       We plan to upgrade all LLRF stations at some point but, we did
             not do this during the summer down time due to limited resources


13                                             MJB, PEP-II MAC Review, December 13-15, 2004
             Other Status - LLRF System Cooling




     Ambient & Crate Temperature For Station with Enhanced Cooling System



14                                     MJB, PEP-II MAC Review, December 13-15, 2004
              Other Status – Phase Reference System

            RF Phase Reference Distribution System is being
             improved to reduce phase drift.
                          This work is being done by Bill Roster
     •       We have and fabricated new Tx boards for the phase
             controllers and Rx boards for the distribution units.
         –      These boards have been installed in all controllers and
                distribution units in the system.
     •       We are adding new RF detector boards with higher
             gain to improve the overall gain stability in the control
             path. These will be installed this week.
         –      We will monitor the system over the next few weeks to see
                whether if the system is more stable.




15                                       MJB, PEP-II MAC Review, December 13-15, 2004
        COMB-2 PROJECT
             W.Ross , M. Browne

     PRINCIPAL ENGINEER AND TASK LEADER



            C Yee, M. Laznovsky
               Major Contributors




16                    MJB, PEP-II MAC Review, December 13-15, 2004
     COMB-2 PROJECT OUTLINE
     • FEATURES
     • BLOCK DIAGRAM
     • DSP‟ FPGA FIRMWARE
     • „DSP‟ FPGA UTILIZATION
     • PROJECT TASKS
     • PC PARTS PLACEMENT
     • PC BOARD LAYER STACK
     • PROJECT TASK LIST
     • PRODJECT SCHEDULE
     • PROJECT RESOURCE LIST


17                    MJB, PEP-II MAC Review, December 13-15, 2004
                         COMB-2 FEATURES
     • Combined I Channel and Q Channel Processing in a Single Module
     • Fixes Over-flow and VXI interface bugs present in old Combs
     • Flexible „DSP‟ FPGA design:
          The filtering algorithms can be changed in the future by re-programming
          the In-Circuit-Programmable Initialization ROMs. (J-tag port)
     • Initial DSP Functionality:
           Initially, Comb 2 Module will be firmware programmed to perform the full
           2-dimensional Comb Filter and Equalization processes, implementing
           the additional I to Q and Q to I filters which were not available in the
           earlier version Comb Modules.
     • Inter- module Analog Connection:
          I/Q signals between the RFP module and the Comb Module will be
          differential analog via the VXI Backplane local bus lines for compatibility
          with current RFP
     • Serial Data Link for a Possible Future Digital RFP Module
          Via an Infiniband connector on the front panel. The total serial bit-rate
          using 8/10-bit coding is 1.4 GBits/Sec in each direction. This link will be
          tested but not implemented in the initial design.

18                                       MJB, PEP-II MAC Review, December 13-15, 2004
     COMB-2 BLOCK DIAGRAM




19           MJB, PEP-II MAC Review, December 13-15, 2004
         COMB-2 DSP‟ FPGA FIRMWARE
     Design implemented in the „DSP‟ FPGA Firmware:




20                            MJB, PEP-II MAC Review, December 13-15, 2004
           COMB-2 „DSP‟ FPGA UTILIZATION


                                     „DSP‟ FPGA
                                      1Kx18 bit            3.125   10.312
                  Logic    18x18                                            Avail
     FPGA Type                         BRAM       Power-   Gbps    Gbps             Price
                  Cells    Mults .                                          wks
                                       Blocks      PCs     Ports   Ports
     XC2VP30-7C   30,816    136         136         2       8        0       5      $1,000




       The initial design implementation uses approximately 30% of the
       resources of the XC2VP30. We are using the 896 pin BGA
       version of this part and the “– 7” the fastest speed grade. This
       should allow considerable flexibility for any future upgrade.




21                                         MJB, PEP-II MAC Review, December 13-15, 2004
     COMB-2 PC PARTS PLACEMENT




22              MJB, PEP-II MAC Review, December 13-15, 2004
     COMB-2 PC BOARD LAYER STACK

                      • 16 LAYER BOARD
                      • 10 TRACE LAYERS
                      • 6 POWER/GND
                        BURRIED CAPACITANCE
                        LAYERS 2/3, 8/9, 14/15



                          BC-24 MATERIAL




23               MJB, PEP-II MAC Review, December 13-15, 2004
                       COMB-2 PROJECT TASK LIST
     1.   FUNCTIONAL SPECIFICATION DEVELOPMENT AND APPROVAL
     2.   INITIAL ALGORITHM / ARCHITECTURE DESIGN FOR DSP PROCESSES
     3.   DEVELOPE INITIAL VHDL FIRMWARE FOR “DSP” FPGA
                                            (to determine required DSP FPGA size)
          3.1 VHDL CODE WRITING - “DSP” FPGA”
          3.2 SIMULATION OF VHDL SUB MODULES
              AND TOP LEVEL MODULE – EPD                      (to debug VHDL code)
          3.3: VIRTEX2-PRO PART IMPLEMENTATION - XILINX ALLIANCE
                                             (to determine part size for „DSP‟ FPGA)
          3.4: MATLAB SIMULATION OF DSP SUB-MODULES
               (to assure transfer functions, sufficient dynamic range, decimal point
                                                alignment, and overflow management)
          3.5 MODIFY VHDL CODE AS REQUIRED - FINALIZE VHDL CODE

     4.   DEVELOPE VHDL FIRMWARE FOR “VXI -INTERFACE” FPGA
          4.1 VHDL CODE WRITING - “VXI -INTERFACE”
          4.2 FUNC. SIMULATE - IMPLEMENT (FIT) FPGA - TIME SIMULATE

     5.   EPICS SOFTWARE DRIVER AND DIAGNOSTICS PANEL DEVELOPMENT

                   XX COMPLETE            XX IN PROCESS             XX NOT STARTED

24                                                   MJB, PEP-II MAC Review, December 13-15, 2004
                  COMB-2 PROJECT TASK LIST CONT.
     6. DESIGN BOARD LEVEL SCHEMATICS
         –   6.1: SCHEMATICS DESIGN - EPD
         –   6.2: SIMULATE AT SCHEMATIC LEVEL - EPD
     7. WRITE MODULE DIAGNOSTICS SOFTWARE
     8. PARTS ACQUISITION (all parts for prototypes, some parts for full
             production)
     9. PC BOARD LAYOUT – PROTOTYPES
     10. PC BOARD FABRICATION - PROTOTYPES
     11. FRONT PANEL LAYOUT AND FABRICATE
     12. PARTS KITTING AND PC BOARD ASSEMBLY - PROTOTYPES
     13. HARDWARE PROGRAMMING, TESTING, TEST PLAN DOC - PROTOTYPES
     14. SOFTWARE DRIVER TESTING AND INTEGRATION INTO LLRF SYSTEM
     15. MODIFY DESIGN AND PC BOARD LAYOUT AS REQUIRED ( for production)
     16. ADDITIONAL PARTS ACQUISITION (for production)
     17. PC BOARD FAB (for production)
     18. PARTS KITTING AND PC BOARD ASSEMBLY (for production)
     19. PRODUCTION HARDWARE/SOFTWARE TESTING (production modules)


                 XX COMPLETE       XX IN PROCESS       XX NOT STARTED
25                                          MJB, PEP-II MAC Review, December 13-15, 2004
     COMB-2 PRODJECT SCHEDULE




                                     W. Ross 11/15/04
26             MJB, PEP-II MAC Review, December 13-15, 2004
               COMB-2 PROJECT RESOURCE LIST
                                         SLAC, ESD, LLRF GROUP

              FUNCTION                                                 RESPOSNIBILITY FOR
                                            DECISION AUTHORITY
                                                                           DELIVERABLES
 PROJECT LEADER                                W. ROSS, M. BROWNE              W. ROSS

 BOARD LEVEL FUNTIONAL SPECIFICATION                 GROUP                     W. ROSS


 DSP ALGORITHM DESIGN AND ARCHITECTURE              W. ROSS                    W. ROSS

 FIRMWARE DESIGN                                    W. ROSS                    W. ROSS

 MATLAB SIMULATIONS                                 W. ROSS                    W. ROSS

 SCHEMATICS DESIGN                                  W. ROSS                    W. ROSS

 PC BOARD LAYOUT                                 C. YEE, W. ROSS                C. YEE

 PC BOARD FAB MANAGEMENT                             C. YEE                     C. YEE

 PARTS ACQUISITION                                   C. YEE                C. YEE, J. RUSSELL

 PARTS KITTING AND MODULE ASSEMBLY                   C. YEE               C. YEE, (ASSOC. TBD)


 MODULE HARDWARE TESTING                            W. ROSS              W. ROSS, (ASSOC. TBD)

 SOFTWARE DEVELOPMENT                        MIKE LAZNOVSKY, W. ROSS       MIKE LAZNOVSKY

 SOFTWARE TESTING                            MIKE LAZNOVSKY, W. ROSS       MIKE LAZNOVSKY




27                                               MJB, PEP-II MAC Review, December 13-15, 2004

								
To top