RF Upgrades & Experience At JLab. Rick Nelson

Transcription

1 RF Upgrades & Experience At JLab Rick Nelson

2 Outline Background: CEBAF / Jefferson Lab History, upgrade requirements & decisions Progress & problems along the way Present status Future directions & concerns

3 CEBAF at Jefferson Lab Design 4 GeV, 200 ua 3 Experimental Halls Present (pre-upgrade) 6 GeV, 200 ua Upgrade 3 Experimental Halls 11 GeV, 200 ua 3 Experimental Halls 12 GeV, 200 ua 4 th Hall D only

4 From 6 to 12 GeV Upgrade magnets and power supplies CHL-2 Upgrade Existing Halls

5 RF Upgrades Original 42.5 cryomodules/338 SC cavities 340 klystrons: 5 to 6.5 kw CW, 1497 MHz Upgrade C25, C50 cryomodules 10 cryomodules, 80 SC cavities 80 klystrons: 13 kw CW, 1497 MHz New designs for klystrons, power supplies, circulators, controls

6 Energy Content (Norm.) Key RF Requirements 10 new zones of RF power for new accelerating structures: 1497 MHz Operating Gradients Required >17.5 MV/m RF Power per cavity 13 kw saturated Regulation requirements (table) Cavity Q L 2x Phase Stability (rms) Amplitude (rms) Master Oscillator EPICS IOC Ethernet LLRF Controls Fast ( <1sec) Ethernet High Voltage Power Supply Klystron Slow (>1sec) 0.5º 3.0º 4.5x10-4 NA Detuning (Hz) 0.0 Superconducting Cavity Cavity de-tuning curve

7 1 per cavity (existing configuration) Minimum impact of failures How Many RF Sources? 1 per zone or 1 per linac Larger impact on faults High power splitters High power amplitude and phase control required with high precision. Additional controls and high power modulators found to be more $$ than individual RF sources. Single LINAC upgrade shown

8 CEBAF Klystrons Compared Parameter Power Old Spec 5 to 8 New Spec Actual Units KW Center frequency MHz Bandwidth, -1dB MHz Bandwidth, -3 db MHz 0.5 db incremental gain at 4 10 meets kw Efficiency (at rated power) 32 > % Gain 38 >42 >50 db Harmonics meets dbc Beam voltage 11.6 < kv DC Heater voltage typ V DC Modulating anode Yes Yes Yes Isolated collector Yes Yes Yes Cavities/Resonators Focus PM EM ~900 Watts

9 Each system powers 8 klystrons (as before) Resonant mode switcher design (15-20 KHz) 4 separate supplies. Each feeds 2 klystrons Minimizes klystrons taken offline due to power supply failure Controlled as a unit Each adjustable to -15kV 15 A total Design adapted from electrostatic precipitator application (higher volts/lower amps & in oil) units in the field at award Highly tolerant to load faults HV DC Power Supply Lower stored energy than T-R, fast turn off on fault, series resistor limits output current (no crowbar)

10 Additional Views HV Deck (4 per system, on rollers) Rear View

11 Typical RF Installation All zones installed and commissioned with beam

12 Tunnel Connections Waveguide installation

13 RF Commissioning Selected Data Commissioned w/ Beam Commissioned w/o Beam Zone/ Cavity Gradient Cryomodule Energy Gain Beam Current Zone/ Cryomodule Gradient Cavity Energy Gain Beam Current SL24 SL NA NA NA NA NA NA NA NA Total Total SL25 SL NA NA NA NA NA NA NA NA Total Total

14 Good & Bad Results Not all according to plan usual problems Delivery delays on several key components Klystron & general WG close to schedule HOM filters, isolators, solenoid power delayed Revisions and rework One contract cancellation Multiple installation delays with starts & stops Budget problems Reassignment of workforce Rework and reinstallation

15 Waveguide Isolators 13 kw CW, full reflection 0.2 db insertion loss 21 db isolation (any phase & power) Water cooled PM only - no TCU Operates adjacent to others Awarded to Ferrite (who supplied 350+ units for CEBAF) Full power testing at JLab Vendor capabilities missing Several rounds of testing with sliding short Using FA klystron at L-3 At JLab using 2 x 6.5 kw and 13 kw klystron

16 Events Initial tests looked OK and first lots were installed Initial tests into WG shorts not as good Results not repeatable/consistent Performance different for distance to short Two rounds of measurements & adjustments to understand the fix Next production units still variable

17 Testing

18 Sensitive to Match & Phase Isolation affected by Ferrite temperature Magnetic field strength These may be adjusted to maintain good performance Less field needed at higher temperature Solutions include TCU, active field control (VSWR) Automatically handled in small units Reflected phase Match (all ports) 2 of 3 need to be good for high isolation Load OK, short bad, klystron needs to be good (but not easily measured)

19 Early Test Results " short iso Iso-J " Iso-J1196 Iso-J1204 iso 1210 iso 1229 Iso-J1219 Iso-J " Iso-J " Iso-J " Iso-J " Iso-J "

20 Early Test Results

21 Isolation (db) Isolation vs. Short Position Iso-J " Iso-8.5" Iso-10" Forward Power (kw)

22 Temperature Initial measurements done steady-state Find position for lowest isolation Changes observed at turn-on Concerns for off-resonance conditions at turn-on Avoid tripping on high reflected power 8 units per RF zone; different distances to cavities

23 power (dbm) Isolation vs. Heating/Time forward-0 reverse-0 isolation : : : : : : : : :09.1 Elapsed time (seconds) RF heating of ferrite resulted in significant changes over (short) time

24 Adjusting Magnetic Fields Isolation response at 10+ kw vs Magnet Count : : : : : : : : :09.1 isolation-0 isolation-1 isolation-2 Isolation-3 isolation-4

25 Isolator

26 Tuning

27 Resolution Vendor reworked dome new domes & rebuilds Results were still inconsistent Something being missed during setup Ultimately all setup the same, but fine-tuned at Jlab 100% re-tested at high power Reflected power well below threshold for klystron damage or performance degradation Solution meets operational needs including credible fault conditions All 84 units modified, tested, reinstalled Extra work since all units were installed/removed/reinstalled.

28 Isolation (db) Isolation vs. Short Position Iso-J Iso-J Short Position (inches)

29 HOM Filters Uncertainty of need Originally not needed so not ordered Directions changed, but only 2 of 8 caviteis expected cavities (later testing confirmed) Normal procurement process Final filters essentially identical to what we had from multiple purchases Small tweaks to reduce fundamental absorption Manufacturing relied on external shops (as before) Vendor a small concern, limited resources & staff Fabrication subcontracted (metal fab, Iridite, dip brazing) Dummy spool pieces installed in other positions Had expected this to come in last

30 Klystron HV PS Performance has been good overall Switcher design and controls work well - good reliability DSP-based controls with hardware safety interlocks Code changes needed to address timing issues 1 unit tested OK, but all 4 might trip external breaker Extended step-start to deal with high inrush & breaker trips Possible race condition for contactor control vs. status reporting (several contactors changed but seem to be OK) Control transformers (480:120) shorted out Loose connections/loosening connections Contactors, IGBT Suggestion: check everything carefully! No similar problems with old supplies, but a lot less connections DC power guys regularly check transistor connections New doesn t mean perfect -- especially after x-country trips A couple noticeable events

31 Installation Challenges Funding shortages resulted in work reassignments & delays Techs reassigned to dismantle other systems Start/stop/start not efficient and required relearning

32 Maintenance Issues New systems to be learned and maintained New systems to be checked closely Old systems getting older and more frail Significant PM planned for summer

33 A Year for Water Brazing issues and water leaks Multiple new components, nuisance problems Both believed to be of similar origin but different suppliers Pressure tested (but not long enough) Trapped flux dissolved out resulted in small leaks on a few pieces Circulator load assemblies New loads built, and testing refined Solenoid leaks on plumbing Longer pressure testing with hot water HOM dip braze excesses Control transformers shorted (purchased by vendor)

34 Ongoing Circulator (old style) Reliable for a lot of years, but load failures becoming more frequent LC DI water 15 years+ erosion and leaching Self-rebuilding w/o retuning Same load back to its circulator New circulator loads won t experience this failure mode

35 Water Flows Downhill All LCW was turned off during extended down (~1 year) Circulator load seals lost their seal Water in select waveguides (not our selection) Water level horizontal run Bleed hole was for air

36 Summary All new LINAC RF has been installed and commissioned Operating requirements met, though staff are still learning differences from old systems Maintenance activities scheduled for summer down old and new systems Lobby to purchase spares with new equipment Costs more later and may be too late Monitor procurements closely, start early