RF Control in SNS Linac Implementation and what has been learned - - PowerPoint PPT Presentation

▶
rf control in sns linac
SMART_READER_LITE
LIVE PREVIEW

RF Control in SNS Linac Implementation and what has been learned - - PowerPoint PPT Presentation

Oct 12, 2023 49 likes •495 views

Hengjie Ma, 1/18/2007, JLab RF Control in SNS Linac Implementation and what has been learned Hengjie Ma 1-2007 O AK R IDGE N ATIONAL L ABORATORY 1 Hengjie Ma, 1/18/2007, JLab SNS Linac RF Control implementation and understanding Outline

slide-1
SLIDE 1

OAK RIDGE NATIONAL LABORATORY

1 Hengjie Ma, 1/18/2007, JLab

RF Control in SNS Linac

Implementation and what has been learned

Hengjie Ma 1-2007

slide-2
SLIDE 2

OAK RIDGE NATIONAL LABORATORY

2 Hengjie Ma, 1/18/2007, JLab

SNS Linac RF Control – implementation and understanding Some Observations

  • Lorentz Force detuning.
  • 5/6pi mode excitation.

Issues & Improvements

  • Control stability
  • Cavity filling techniques.
  • Feed forward control

algorithms

SNS LLRF System Overview

  • Configuration
  • Phase reference schemes
  • Mechanical construction

Implementations

  • Chosen controller type
  • Performance analysis
  • Digital implementation

Test & Operation Results

  • Control bandwidth, error
  • Linac-wise performance

Outline

slide-3
SLIDE 3

OAK RIDGE NATIONAL LABORATORY

3 Hengjie Ma, 1/18/2007, JLab

1999 The Spallation Neutron Source

slide-4
SLIDE 4

OAK RIDGE NATIONAL LABORATORY

4 Hengjie Ma, 1/18/2007, JLab

Summary of Beam Parameters Achieved in Commissioning

degrees rms 4 3 CCL1 bunch length MeV 952 1000 Linac Output Energy msec/Hz/% 1.0/60/3.8 (DTL1 run) .050/1/.005 (CCL run) 0.85/0.2/0.017 (SCL) 1.0/60/6.0 Linac Pulse length/Rep- rate/Duty Factor Units Achieved Baseline/ Design Parameter Protons/pulse 5x1013 1.5x1014 Extracted protons/pulse Ions/pulse 1.3x1014 (DTL run) 1.0x1013 (CCL run) 8.0x1013 (SCL run) 1.0x1014 (Ring run) 1.6x1014 Linac H-/pulse mA 1.05 (DTL1 run) 0.003 (SCL run) 1.6 Linac Average Current mA > 38 38 Linac Peak Current π mm-mrad (rms,norm) 0.3 (H), 0.3 (V) 0.4 Linac Transverse Output Emittance

slide-5
SLIDE 5

OAK RIDGE NATIONAL LABORATORY

5 Hengjie Ma, 1/18/2007, JLab

slide-6
SLIDE 6

OAK RIDGE NATIONAL LABORATORY

6 Hengjie Ma, 1/18/2007, JLab

Thales CPI CPI & Thales Thales E2V & Thales Vendor 4 4 81 4 7 Installed Accumulator Ring MEBT Rebunchers SCL CCL RFQ, DTL Application 1 100 kW 1 & 2 MHz Tetrode 3 20 kW 402.5 MHz Triode 14 550 kW 805 MHz Klystron 5 5 MW 805 MHz Klystron 5 2.5 MW 402.5 MHz Klystron Spare Peak Power Frequency Type

The SNS utilizes 100 RF systems for acceleration and bunching

slide-7
SLIDE 7

OAK RIDGE NATIONAL LABORATORY

7 Hengjie Ma, 1/18/2007, JLab

SNS HPRF Configuration

RFQ DTL CCL to SCL from CCL Medium Beta SCL High Beta SCL

1 3 2 4 6 5 1 2 3 4 6 1 2 3 4 5 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

NC Linac

XMTR A XMTR 1 XMTR 2

K K K K K K K

XMTR 3 XMTR 4 XMTR 5 XMTR 6

SNS Linac Layout

XMTR 1 XMTR 2 XMTR 3 XMTR 4

K K K K

XMTR B XMTR A XMTR B

High Beta SCL

XMTR A XMT B XMTR A XMTR B XMTR RFQ XMTR A XMTR B SCL Klystron HV Tank 550 kW klystron XMTR A XMTR B XMTR A XMTR B

SCL ME18 SCL ME21 SCL ME12 SCL ME15 SCL ME9 SCL ME5 SCL ME1 ME1 ME3 ME5 ME1 ME2 ME3 ME4

slide-8
SLIDE 8

OAK RIDGE NATIONAL LABORATORY

8 Hengjie Ma, 1/18/2007, JLab

RFQ, DTL and CCL RF Systems Seven 402.5 MHz, 2.5 MW klystrons power the RFQ and DTL Four 805 MHz, 5 MW klystrons power the CCL

slide-9
SLIDE 9

OAK RIDGE NATIONAL LABORATORY

9 Hengjie Ma, 1/18/2007, JLab

SC Linac RF System Layout

High Beta 4-Cavity Cryomodules 3 Klystrons per HV Tank HVCM Cooling Manifolds

Typical SC RF Layout by HVCM

CAV CAV CAV CAV CAV CAV CAV CAV Load K K K K K K K K K CAV CAV CAV CAV SCR

Transmitter Racks LLRF Racks

CIRC Tunnel Klystron Gallery Load CIRC

6-1-05 McCarthy

K K K

Transmitter Racks LLRF Racks Cooling Manifolds

slide-10
SLIDE 10

OAK RIDGE NATIONAL LABORATORY

10 Hengjie Ma, 1/18/2007, JLab

Eighty-One 805 MHz, 550 kW klystrons power the superconducting Linac Transmitter racks & LLRF for support of six klystrons. Three klystrons mount to a single oil tank.

slide-11
SLIDE 11

OAK RIDGE NATIONAL LABORATORY

11 Hengjie Ma, 1/18/2007, JLab

Each Rack in the Superconducting Linac Contains LLRF Hardware for Two RF Systems

Typical LLRF control rack installation in the superconducting Linac. The VXI crate contains:

  • Input/Output Controller: PowerPC running VxWorks
  • Utility Module: Decodes events from Real Time Data Link
  • Timing Module: Generates RF Gate timing signal
  • Two FCM/HPM pairs
slide-12
SLIDE 12

OAK RIDGE NATIONAL LABORATORY

12 Hengjie Ma, 1/18/2007, JLab

SNS Linac RF Control – system overview

System configuration and phase reference scheme

R e f e r e n c e L i n e

Host (“IOC”)

RF / IF

Host (“IOC”)

RF / IF

Host (“IOC”)

RF / IF

Host (“IOC”)

RF / IF

Time : Calibration pulse : RF pulse

slide-13
SLIDE 13

OAK RIDGE NATIONAL LABORATORY

13 Hengjie Ma, 1/18/2007, JLab

SNS Linac RF Control – system overview

FCM Electrical Specs.

  • I-Q sampling interval:(1+1/4)*2pi

(Fs=40MHz)

  • IF inputs: 2, 50MHz
  • IF ADC resolution: 14-bit
  • RF inputs: 2, 805/402.5MHz
  • Channel analog bandwidth:1MHz
  • High-speed output DAC:1
  • DAC resolution: 14-bit
  • Output Frequency shift range:+/-645kHz
  • FPGA: XC2V1500, SDRAM: 128Mb
slide-14
SLIDE 14

OAK RIDGE NATIONAL LABORATORY

14 Hengjie Ma, 1/18/2007, JLab

The Field Control Module (FCM) consists of a motherboard and three daughterboards

Analog Front End (AFE) Down-converting channels: Incident and Reflected RF (402.5 or 805 MHz) IF channels: Cavity and Reference (50 MHz) Digital Front End (DFE) Four 14 bit, 40 MHz ADC channels One Virtex II FPGA (XC2V1500 – 1.5M gates) RF Output (RFO) Clock & PLL circuitry One 14 bit, 80 MHz DAC Up-Conversion to 402.5/805 MHz Filtering

slide-15
SLIDE 15

OAK RIDGE NATIONAL LABORATORY

15 Hengjie Ma, 1/18/2007, JLab

SNS Linac RF Control – implementations

Field Control Module (FCM):

P-I feedback + AFF control

Waveform plotting

slide-16
SLIDE 16

OAK RIDGE NATIONAL LABORATORY

16 Hengjie Ma, 1/18/2007, JLab

SNS Linac RF Control – implementations

Digital implementation:

from continuous (1) to discrete time (2)

( ) (

)

⎭ ⎬ ⎫ ⎩ ⎨ ⎧ ⎟ ⎟ ⎠ ⎞ ⎜ ⎜ ⎝ ⎛ − ⋅ + ⋅ ⋅ ⋅ + ⋅ ⋅ − ⋅ − =

− − − − − 1 1 s i p 2 s 1 c 2 c

Z 1 Z T K 1 K Z C Z C S 1 ) Z (1 (Z) G ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ + ⋅ ⋅ ⋅ = S K 1 K R (S) H (S) G

i p f c

⎭ ⎬ ⎫ ⎩ ⎨ ⎧ ⎟ ⎠ ⎞ ⎜ ⎝ ⎛ ⋅ + ⋅

∑

= − N n n n

a Z 1 Kp

slide-17
SLIDE 17

OAK RIDGE NATIONAL LABORATORY

17 Hengjie Ma, 1/18/2007, JLab

SNS Linac RF Control – implementations

Digital implementation: data flow

Kp·Cos(θ) __ Kp·Sin(θ) Is, Qs Kp·Sin(θ) ___

  • Kp·Cos(θ)

Qs,-Is

  • Kp·Cos(θ)

__

  • Kp·Sin(θ)
  • Is,-Qs

Kp·Cos(θ) __ Kp·Sin(θ) Is, Qs

slide-18
SLIDE 18

OAK RIDGE NATIONAL LABORATORY

18 Hengjie Ma, 1/18/2007, JLab

SNS Linac RF Control – implementations

Digital implementation: Development system

RTL Compilation Iverilog / ModeSim Synthesis XST / Synplify PAR Xilinx ISE End-to-End System Behavioral Simulation Machine Code Gen. : drivers GUI/Control Automation software development Machine Code gen.: Netlists Verilog Modeling: Functionalities Verification Test Lab.

O K ?

NO YES YES NO

Previous VHDL flow current llc system

slide-19
SLIDE 19

OAK RIDGE NATIONAL LABORATORY

19 Hengjie Ma, 1/18/2007, JLab

SNS Linac RF Control – implementations

Controller Performance: Stability

Effect of loop delay

With no loop delay With 1us loop delay

slide-20
SLIDE 20

OAK RIDGE NATIONAL LABORATORY

20 Hengjie Ma, 1/18/2007, JLab

SNS Linac RF Control – implementations

Controller performance: Steady-state error

Constraints between control gain, stability and steady-state error.

Example 1: 1us delay, cavity BW=10kHz (NC) Kp=4, Kp=100 Kp=5 Ki=0 Ki=0 Ki/Ts=10kHz stable unstable stable Er=20% Error=0

slide-21
SLIDE 21

OAK RIDGE NATIONAL LABORATORY

21 Hengjie Ma, 1/18/2007, JLab

SNS Linac RF Control – implementations Example 2: 1us delay, cavity BW=500Hz (SC) Kp=100, Kp=400 Kp=100 Ki=0 Ki=0 Ki/Ts=150Hz stable unstable stable Er=1% Error=0

Controller performance: Steady-state error

slide-22
SLIDE 22

OAK RIDGE NATIONAL LABORATORY

22 Hengjie Ma, 1/18/2007, JLab

SNS Linac RF Control – implementations

Controller performance: Transient response

Exam the response of a P-I controller to a step disturbance (beam loading) at the input of the plant (cavity). Rearrange so that the disturbance Becomes the input.

slide-23
SLIDE 23

OAK RIDGE NATIONAL LABORATORY

23 Hengjie Ma, 1/18/2007, JLab

SNS Linac RF Control – implementations

Controller performance: Transient response

Comparison: P-I control (dipole tune, delay=1us) vs. performance benchmark : a 2nd-order deadbeat tune, (Ts=5us, Wn*Ts =4.82, alpha = 1.82, undershoot 0%,

  • vershoot<0.1%, Trise100% =6.58*Wn )
slide-24
SLIDE 24

OAK RIDGE NATIONAL LABORATORY

24 Hengjie Ma, 1/18/2007, JLab

Basic cavity model:

Is it reasonable to use a simplistic P-I controller of LTI design in a nonlinear application ?

Yes, provided in a linearized operating condition.

SNS Linac RF Control – implementations

⎥ ⎦ ⎤ ⎢ ⎣ ⎡ ⋅ ⋅ = ⎥ ⎦ ⎤ ⎢ ⎣ ⎡ ⋅ ⎥ ⎦ ⎤ ⎢ ⎣ ⎡ −

i r L 1/2 i r 1/2 1/2

I I R ω V V ω Δω Δω ω

slide-25
SLIDE 25

OAK RIDGE NATIONAL LABORATORY

25 Hengjie Ma, 1/18/2007, JLab

SNS Linac RF Control – implementations

Operating feedback control with assistance of cavity filling and adaptive feedforward.

Cavity filling Cavity filling + FF (with no beam) (with beam)

slide-26
SLIDE 26

OAK RIDGE NATIONAL LABORATORY

26 Hengjie Ma, 1/18/2007, JLab

SNS Linac RF Control – implementations

The Issue of loop phase change in system

LFD and Klystron phase change

Availability of 60 degree phase margin under LTI assumption

slide-27
SLIDE 27

OAK RIDGE NATIONAL LABORATORY

27 Hengjie Ma, 1/18/2007, JLab

SNS Linac RF Control – Some test and operation results

5 10 15 20 25 30 35 40 45 50

  • 200

200 400 600 800 1000 1200 1400 1600 1800 Time (usec) Amplitude of cavity field and LLRF drive changes(ADC counts) Transient Response of SNS LLRF to 12.5% Step Function Change in Set-Point Input ← Dead time (delay from DAC to ADC) = 1.0728 usec. ← Field overshoot peak level = 22.5589 %, measured Field overshoot peak time = 3.295 (us), measured Control bandwidth ωb = 167.9263 kHz @ ξ=0.707 Total Loop Gain K = 10.0809 for ξ=0.707 Settling time Ts = 4.2705us, measured, 95% rise Cavity pole Kc = 16.6579 kHz Integral zero Ki = 12.6 kHz System ID: DTL

LLRF:FCM1:

FWD power: 500kW LLRF drive base offset: 2523.1627 Field base offset: 5185.4878 Cavity field, ξ=0.707 LLRF drive, ξ=0.707 Cavity field, ξ<<0.707 LLRF drive, ξ<<0.707 Cavity field, ξ>>1.0 LLRF drive, ξ>>1.0

Control Bandwidth measurement:

By the step response (set-point as the input)

slide-28
SLIDE 28

OAK RIDGE NATIONAL LABORATORY

28 Hengjie Ma, 1/18/2007, JLab

500 1000 1500 2000 4000 6000 8000 10000 12000 Time (usec) Amplitude (ADC counts) Transient Response of SNS LLRF to 20% Step Function Change in FF Input System ID: SCL

LLRF:FCM12a:

Field gradient: 9 MV/m Cavity field Forward wave Reflected wave 20 40 60 80 100 120

  • 800
  • 600
  • 400
  • 200

200 400 600 800 Time (usec) Change in Amplitude (ADC counts) Transient Response of SNS LLRF to 20% Step Function Change in FF Input ← Dead time (delay from DAC to ADC) = 1.519 usec. ← Field overshoot peak level = 2.4405 %, measured Field overshoot peak time = 13.0633 (us), measured Control bandwidth ωb = 59.2564 kHz @ ξ=0.707 Damping factor ξ = 0.7634 used for test Total Loop Gain K = 104.8081 for ξ=0.707 Settling time (95%) T

s = 3T = 8.0576 (us), for

ξ=0.707 Cavity pole Kc = 0.56538 kHz System ID: SCL

LLRF:FCM12a:

Forward wave base offset: 3349.7845 Reflected wave base offset: 3334.4912 Field gradient base offset: 11420.5685 Cavity field Forward wave Reflected wave

SNS Linac RF Control – Some test and operation results

Control Bandwidth Measurement:

By transient response to a step in FF waveform.

slide-29
SLIDE 29

OAK RIDGE NATIONAL LABORATORY

29 Hengjie Ma, 1/18/2007, JLab

500 1000 1500 2000 4000 6000 8000 10000 12000 Time (usec) Amplitude (ADC counts) Transient Response of SNS LLRF to Beam Loading System ID: SCL

LLRF:FCM12a:

Beam Current: 10 mA Cavity field Forward wave Reflected wave 20 40 60 80 100 120

  • 1500
  • 1000
  • 500

500 1000 1500 Time (usec) Change in Amplitude (ADC counts) Transient Response of SNS LLRF to Beam Loading Rise time :11.0092 us Settlling time :19.4495 us Field dip-peak :0.29167 % Field dip-residual :0.11954 % Beam Current: 10 mA Field trace magnification: x 10 Field gradient base offset: 11387.064 Forward wave base offset: 3410.0756 Reflected wave base offset: 3251.8126 System ID: SCL

LLRF:FCM12a:

Cavity field Forward wave Reflected wave

SNS Linac RF Control – Some test and operation results

Transient Response to Beam Loading

slide-30
SLIDE 30

OAK RIDGE NATIONAL LABORATORY

30 Hengjie Ma, 1/18/2007, JLab

SNS Linac RF Control – Some test and operation results

A demo of FCM’s stability and dynamic range

  • Operating point:

2M V/m (13% of the nominal value)

  • Relatively heavy

beam loading, >900% rf power increase.

slide-31
SLIDE 31

OAK RIDGE NATIONAL LABORATORY

31 Hengjie Ma, 1/18/2007, JLab

A snapshot of RF errors throughout Linac

No beam, AFF turned off, field error is maintained under 1%, 1 deg.

SNS Linac RF Control – Some test and operation results

slide-32
SLIDE 32

OAK RIDGE NATIONAL LABORATORY

32 Hengjie Ma, 1/18/2007, JLab

A snapshot of RF errors throughout Linac

With 25mA beam, AFF turned ON, the RF control is still able to keep the field error under 1%, 1 deg.

SNS Linac RF Control – Some test and operation results

slide-33
SLIDE 33

OAK RIDGE NATIONAL LABORATORY

33 Hengjie Ma, 1/18/2007, JLab

A resonance of ~2kHz has being seen on many medium-beta cavities, which has created some difficulty to the rf control, mainly in the choice between the control error and gain setting.

Interaction between rf and cavity vibrations

SNS Linac RF Control – Some observations

slide-34
SLIDE 34

OAK RIDGE NATIONAL LABORATORY

34 Hengjie Ma, 1/18/2007, JLab

100 200 300 400 500 600 0.8 1 1.2 1.4 1.6 1.8 x 10

4

Time (usec) Amplitude of cavity field (ADC counts) Cavity 2kHz resonance vs. field gradient - full view System ID: SCL

LLRF:FCM07a:

10 MV/m 10.5 MV/m 11 MV/m 11.5 MV/m 12 MV/m 12.5 MV/m 13 MV/m 13.5 MV/m 14 MV/m 14MV/m 10MV/m 1.12 % total

Interaction between rf and cavity vibrations

The magnitude of the resonance grows as the gradient increase.

SNS Linac RF Control – Some observations

slide-35
SLIDE 35

OAK RIDGE NATIONAL LABORATORY

35 Hengjie Ma, 1/18/2007, JLab

Interaction between rf and cavity vibrations

Measured Lorentz force detuning of all running cavities

SNS Linac RF Control – Some observations

  • The 2kHz oscillating

shows up on all MB cavities, but not HB

  • nes.
  • The peak of LFD is
  • nly one half of the

cavity BW. (Calculation using in closed-loop control condition.)

( ) ( ) ( )

Φ Ψ ρ ω ω r r Φ Ψ ρ ω Δω r Φ r

/ / /

− ⋅ ⋅ = ⋅ − − ⋅ ⋅ = ⋅ − ⋅ = ⋅ ⋅ − + sin sin

2 1 2 1 2 1

& & & y ω x Δω j ω x

1/2 1/2

slide-36
SLIDE 36

OAK RIDGE NATIONAL LABORATORY

36 Hengjie Ma, 1/18/2007, JLab

Interaction between rf feedback control and 2kHz resonance. Oscillation coupled from phase to amplitude.

SNS Linac RF Control – Some observations

slide-37
SLIDE 37

OAK RIDGE NATIONAL LABORATORY

37 Hengjie Ma, 1/18/2007, JLab

200 400 600 800 1000 1200 1400 1.8 1.805 1.81 1.815 1.82 1.825 1.83 1.835 1.84 1.845 x 10

4

Time (usec) Amplitude of cavity field (ADC counts) Effectiveness of FF on compensatging "2kHz" cavity ringing System ID: SCL

LLRF:FCM07a:

Iternation: 1 Iternation: 10 Iternation: 60 at end of AFF learning at beginning of AFF learning

Adaptive feed forward control not only compensates the beam loading, but also effectively damps the 2kHz resonance.

SNS Linac RF Control – Some observations

slide-38
SLIDE 38

OAK RIDGE NATIONAL LABORATORY

38 Hengjie Ma, 1/18/2007, JLab

5/6π mode sideband excitation

Exists in both open-loop and closed-loop control condition, but is contained. Normally, the peak magnitude is under 3% of the field level. Feedback OFF Feedback ON

SNS Linac RF Control – Some observations

slide-39
SLIDE 39

OAK RIDGE NATIONAL LABORATORY

39 Hengjie Ma, 1/18/2007, JLab

SNS Linac RF Control – Issues and improvements

Issues & Improvements

  • Control stability
  • Cavity filling schemes
  • Feed forward algorithms.
slide-40
SLIDE 40

OAK RIDGE NATIONAL LABORATORY

40 Hengjie Ma, 1/18/2007, JLab

  • System performed well and met the needs

during the commissioning and recent operation runs.

  • Hardware platform has flexibility and

capacity to allow further expansions.

  • Strong software support has offered ease and

convenience of operations.

  • Further development is necessary in the

aforementioned areas. SNS LLRF project is a collaborative effort among ORNL, LBNL, and LANL.

SNS Linac RF Control – Summary

slide-41
SLIDE 41

OAK RIDGE NATIONAL LABORATORY

41 Hengjie Ma, 1/18/2007, JLab

Power Upgrade Project

Power Upgrade Project requires increased RF power – Beam energy increases from 1000 to 1300 MeV – Average macropulse current increases from 26 to 42 mA Total peak RF power requirement (structure + beam) – 42.8 MW baseline – 71.5 MW upgrade Additional voltage requires more cryomodules Additional beam loading requires increased RF power everywhere Design Goal – maximize re-use of existing systems – maintain compatibility with existing systems SNS Linac RF Control – Looking ahead

slide-42
SLIDE 42

OAK RIDGE NATIONAL LABORATORY

42 Hengjie Ma, 1/18/2007, JLab

Power Upgrade Project

  • The RF Systems Upgrades include:
  • Installation of 36 additional Linac RF stations
  • Installation of 2 HEBT RF stations including cavities
  • Upgrades to 4 existing Ring RF stations

SNS Linac RF Control – Looking ahead

slide-43
SLIDE 43

OAK RIDGE NATIONAL LABORATORY

43 Hengjie Ma, 1/18/2007, JLab

Power Upgrade Project The 36 additional RF systems will essentially be duplicates of the 81 systems already installed. Layout: – 36 klystrons – 6 transmitters – 3 high-voltage converter-modulators (HVCM) – 36 low-level RF control systems Klystrons: Baseline Parameters – Baseline klystrons rated for 550 kW peak RF power with 9% duty factor at 75 kV cathode voltage – Stations 1-48 limited to 69 kV (445 kW) due to 12 klystrons per HVCM – Stations 49-81 may operate at full power due to 11 klystrons per HVCM Klystrons: Upgrade Parameters – Cathode voltages will increase to 83 kV for high-beta stations (37-117) Peak RF power increases to 708 kW Provides minimum margin of 37% (need 517 kW max without margin) – Similar to baseline klystrons, which are very robust and have been factory tested to 105 kV, 1.3 MW (one test) SNS Linac RF Control – Looking ahead

slide-44
SLIDE 44

OAK RIDGE NATIONAL LABORATORY

44 Hengjie Ma, 1/18/2007, JLab

Milestones for RF Systems, Power Upgrade Project

Oct 06 Apr 08 Oct 08 Oct 10 Aug 08 May 09 Feb 10 Apr 08 Mar 09 Jun 08 Dec 09 Feb 08 Jun 09 Dec 09 Oct 08 Oct 10 Mar 11 Begin design activities Award klystron contract Receive 1st klystron Complete klystron installation Award transmitter contract Receive 1st transmitter Complete transmitter installation Begin waveguide installation Complete waveguide installation Begin HEBT RF system installation Complete HEBT RF system installation Award HEBT cavities contract Begin HEBT cavities installation Complete HEBT Cavities installation Decide extent of Ring RF upgrade Complete Ring RF upgrade Commissioning Early Dates Milestone