Transcript SNS Experience with a High-Energy Superconducting Proton Linac J. Galambos CARE-HHH-APD Beam 07 Workshop 1-5 October, 2007

SNS Experience with a
High-Energy
Superconducting Proton
Linac
J. Galambos
CARE-HHH-APD Beam 07 Workshop
1-5 October, 2007
The SNS Linac
 SNS is a pulsed, accelerator driven spallation neutron
source
 It is driven by a high power linac
 1.5 MW baseline (constructed device)
 3 MW upgrade power (ongoing project, CD-0 approved)
 It is the first high power or high energy superconducting
proton linac
 80% of the acceleration is provided by superconducting cavities
 We are writing the book on tuning it as we go
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
2
A Brief History of the SNS Superconducting
Linac
 Fall 1998: SNS receives approval and funds to build SNS (copper
linac)
 Summer – Fall 1999: Y. Cho leads a task force to investigate the
feasibility to change to a Superconducting Linac
 Dec. 1999 – Jan. 2000: SNS advisory panel and DOE review panel
approves the change, Jefferson Lab joins the SNS collaboration
 June 2003 - First cryo-module delivered to Oak Ridge
 August 2004– First cryo-module cool-down
 March 2005 – July 2005: High power RF tests
 August 2005: Beam commissioning
 Jan.- Feb. 2006: Support Ring Commissioning
 Oct. 2006 – present: Support neutron production runs
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
3
Superconducting
Linac
 Designed an built by Jefferson
Laboratory
 SCL accelerates beam from 186 to
1000 MeV
 SCL consists of 81 cavities in 23
cryomodules
 Two cavities geometries are used
to cover broad range in particle
velocities
 Cavities are operated at 2.1 K with
He supplied by Cryogenic Plant
Medium beta cavity
High beta cavity
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
4
Linac RF Systems

Designed and procured
by LANL

All systems 8% duty
factor: 1.3 ms, 60 Hz

7 DTL Klystrons: 2.5 MW
402.5 MHz

4 CCL Klystrons: 5 MW
805 MHz

81 SCL Klystrons: 550
kW, 805 MHz

14 IGBT-based
modulators
81 SCL
Klystrons
High Voltage
Converter Modulators
 2nd largest klystron and modulator installation
in the world!
DTL Klystrons
CCL Klystrons
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
5
Layout of Linac RF with NC and SRF
Modules
402.5 MHz, 2.5 MW klystron
805 MHz, 5 MW klystron
Warm
Linac
RFQ
DTL
(1)
(6)
805 MHz, 0.55 MW klystron
CCL
(4)
86.8 MeV
2.5 MeV
186 MeV SRF, ß=0.61, 33
cavities
379 MeV
SRF, ß=0.81, 48 cavities
1000 MeV
SCL from
CCL
Linac
1
(81 total powered)
•SCL has 81 independently powered cavities
Many parts to keep running
Many values to set w.r.t. the beam
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
6
Cavity Limitations I – Field Emission
Electrons emitted from
high field surface
Radiation ~ constant
throughout the RF pulse
PM Radiation
detector
RF waveform
 The primary cavity gradient limitation
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
7
Cavity Limitations II – “Cavity-Coupler
Interaction”
Radiation waveform
Electron probe signal
Some cavities have both
effects
 Another gradient limitation – not completely understood
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
8
Cavity Limitations III - Collective behavior
(clear indication at higher rep. rate)
•Amplitude and phase setpoints of one cavity
affect heating at other places
•Need to find setpoints that are friendly to
neighboring cavities
Example:
CM13 individual limits
19.5, 15, 17, 14.5 MV/m
CM13 collective limits at 60 Hz ; 14.5, 15, 15, 10.5 MV/m
a
b
c
d
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
9
SCL Sub-component Concern I – HOM
Coupler
 HOM couplers added as
insurance even though
probability that they are
needed was very low
 HOM feed-through is
susceptible to damage (FE,
MP interactions +
fundamental mode
coupling)
 Some cavities are limited by coupling of fundamental power
coupling (stray field + filter not set properly).
 We would not include HOM filters if we were starting over
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
10
SCL Subcomponent Concern II – CCG
Vacuum Gauges
 Used to monitor vacuum activity at the
coupler window – interlock input
 Often takes a long time for them to
“wake up” when turning on a cavity
 Erratic signals often observed when
they do “wake up”. Inconsistent with
electron probe signals
~5K
CCG
Flange
Temperature
Coupler
Temperature
 The CCGs do not limit the performance of any cavities, but they
do complicate operation.
 Moving towards interlocking on electron detectors and we have
developed procedures for cavity turn on to avoid non-physical
signals.
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
11
SCL Subsystem Concern III – Piezo Tuner
 Piezo tuners added to
cancel Lorentz detuning
 Never have been used
 Some have broken –
rendering the cavity
useless
3
 We are removing them.
 Lesson – keep the design
as simple as possible
Piezo tuner
1. CAD Model
2. Tuner + Bare Cavity
3. Reality – Pretty Complicated
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
12
Accelerating gradients and statistics
30
10 Hz individual limits
60 Hz collective limits
25
15
10
5
CM19; removed
23
b
22
c
21
a
21
d
19
c
20
b
18
a
18
d
17
b
15
d
16
c
14
b
15
a
13
c
12
a
12
d
10
a
11
a
9a
8a
7a
6a
5a
4a
3a
2a
0
1a
Eacc (MV/m)
20
Cavity number
Design gradient
Average limiting gradient (collective)
Average limiting gradient (individual)
Large fundamental power through HOM coupler
Field probe and/or internal cable (control is difficult at rep. rate >30 Hz)
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
13
Accelerating gradients and statistics (II)
25
25
Collective Limits at 60 Hz
Collective Limits at 60 Hz
Individual Limits
Operating setpoints at 60 Hz
20
no. of cavities
no. of cavities
20
15
10
5
15
10
5
0
0
8
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
5
6
7
8
9
Ea (MV/m)
10 11 12 13 14 15 16 17 18 19 20 21
Ea (MV/m)
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
14
SCL Cavity Amplitudes
81
81
77
77
77
73
73
69
69
65
65
61
61
57
57
53
53
49
49
45
45
41
41
37
37
33
33
29
29
25
25
21
21
17
17
13
13
99
555
30
35
35
30
30
25
25
25
20
20
20
15
15
15
10
10
10
555
000
111
E0
(MV/m)
(MV/m)
E0
E0 (MV/m)
(MV/m)
Cavity
Second
Run Run
First Design
Run
Ring Commissioning
cavity
cavity
 Strategy is to run cavities at their maximum safe amplitude limit
 Need to be flexible – SRF capabilities change, not near the design
 Linac output energy is a moving target
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
15
Tune-up Strategy
 RF group provides allowable RF gradients
 Calculate expected output energy
 Devise an appropriate RF and Quad tune
 Longitudinal: ~ constant focusing
 Transverse: scale design values with Br
 Local optimization for matching sections
 Step through each cavity to set the phase relative to the
beam
 Scale downstream transfer line and accumulator ring
magnets with beam energy
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
16
Setting the Phase of the SCL Cavities
Example SCL Phase Scan
BPM Phase Difference
Black line = measurement fit
Dot = model
Red = cosine fit
 A beam based measurement must be done to initially set each cavity RF
phase setpoint
 Scan the cavity phase of a cavity 360, and observe the resultant change in
the Time of Flight (TOF) between 2 downstream detectors
 Compare this difference with a model calculations.
 Gives the input beam energy, cavity voltage and RF phase offset calibration
 Need good relative phase measurements from the detectors (~ 1degree!)
 Scan each cavity sequentially
RF Cavity
Phase
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
17
Drifting Beam Method to Determine Cavity
Phase and Amplitude Setpoints
proton beam 422MeV, 15mA, 40us
simulations from the model with
superimposed measured noise
signals measured with/without
cavity detuning
100
80
Signal Phase (deg)
Signal Phase (deg)
100
60
40
0 Hz
200 Hz
20
400 Hz
80
60
40
0 Hz
200 Hz
20
0
400 Hz
0
0
20
40
t (us)
60
80
100
0
20
40
t (us)
60
80
100
 Allow beam to drift through a cavity
 Beam excites the cavity, and by comparing to a model one can
calibrate the LLRF phase and amplitude readings
 Phase prediction ~ 1 degree, amplitude ~ 4% compared to phase
scan technique
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
18
SCL Cavity Phase Setup Times are Getting
Shorter
 August 2005: 48 hrs
 560 MeV, initial run, > 20 cavities off
 Dec. 2005: 101 hrs
 925 MeV, turned on all planned cavities
 July 2006: 57 hrs
 855 MeV
 Oct 2006: 30 hrs
 905 MeV, used established cavity turn
on procedure
Power
cavities on
sequentially
 Jan. 2007: 6 hrs
 905 MeV, beam blanking used, which allowed all cavities to be on
during the tuning process
 The procedures used to setup the superconducting linac
have matured, and the setup time is now minimal
 Still exists a need for fast recovery from changes in the SCL
setup
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
19
SCL Tune-up – Linac Energy Gain is
Understood and Predictable
Predicted - Measured Energy Gain
10_01_2006
1_16_2007
1.00
0.00
Energy Gain per Cavity Prediction
Error
-1.00
30
21
31
41
51
61
71
81
15
10
 Energy gain per cavity is predictable to a few
100 keV and distributed about 0.
5
 Final energy is predictable to within a few MeV
-1
-0
.8
-0
.6
-0
.4
-0
.2
Cavity
1_16_2007
20
0
1
11
10_07_2006
0.
4
1
Frequency
25
0.
8
-2.00
0.
6
LLRF Cable
0
0.
2
D E (MeV)
2.00
Energy Gain Error (MeV)
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
20
Scaling Method for Cavity Fault Recovery
New cavity phases
 Use beam measurements for original beam arrival times
New Beam Energy
 User inputs changes to the SCL RF setup
 Model predicts changes in the beam arrival times (RF phase
setpoint changes), sends them to the machine and predicts the
new beam energy
 Takes < 1 second to calculate and apply the new SCL setup
 However – we have applied this technique to recover from “events” that
take hours / days to evaluate and proceed
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
21
Application of the Cavity Fault Recovery
Scheme (I)
6
Phase Change
Measured Error
200
4
100
2
0
0
-100
-2
-200
-300
-4
-400
-6
1
Measured Error (deg)
Phase Change (deg)
300
7 13 19 25 31 37 43 49 55 61 67 73 79
Cavity
 In the spring 2006, 11 cavities had to be either turned off or have their
amplitudes reduced for safe operation, 1 cavity was returned to operation
 The fault recovery scheme was applied “all at once”
 Phase scan spot checks indicate the scaling was within 4 degrees
 No detectable change in beam loss
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
22
Application of the Cavity Fault Recovery
Scheme (II)
30
20
-500
10
-1000
0
-1500
-10
-2000
-20
-2500
-30
D Amplitude (MV/m)
D Phase (deg)
0
Phase
Change
Amplitude
Change
1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81
cavity
 In April 2007 the SCL was lowered from 4.2K to 2 K to facilitate 30 Hz
operation.
 About 20 cavity amplitudes changed.
 The fault recovery scheme restored beam to the previous loss state.
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
23
Some SNS Linac Beam Performance
Measures (through the entire linac)
Energy (GeV)
Rep Rate (Hz)
Pulse Length
(mSec)
Beam Current *
(mA)
Beam Power
(MW)
Design
Highest Ever
(Individual)
Highest Beam
Power
(Simultaneous)
1.0
60
1
1.01
60
1
0.88
30
0.55
26
20
13
1.5
0.18
0.18
* Time average including ~ 30% chopping
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
24
Energy Jitter – Pulse to Pulse
160
140
120
100
80
60
40
20
0
•Average phase diff =
112.63 deg : average
energy = 866.02 MeV
•RMS phase diff = 0.278 :
energy jitter = 0.15 MeV
11
3
11
3.
3
11
3.
6
11
3.
9
~ 0.4 MeV RMS
jitter,
•Max
phase diff = 1.2 deg
out of 886 MeV: energy jitter = 0.66 MeV
11
1.
5
11
1.
8
11
2.
1
11
2.
4
11
2.
7
Frequency
BPM 27-25
Delta phi (deg)
 Energy control pulse-to-pulse nor within a pulse has not been
a concern
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
25
Power Ramp-up Progress
ISIS Power Record
 We are starting to get to real beam power levels
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
26
Beam Loss / Activation
Contact / 30 cm dose
in mRem/hr
 SCL has a large aperture and should easily transport beam
 This past summer we observed higher than expected activation
levels in some warm sections (with quadrupoles) between the
cryomodules – not expected based on loss monitor levels
 Not well understood, possibly longitudinal loss
 Purposeful detuning of the warm linac results in loss patterns with
similar shape as the activation patterns
 “Dark current” from the ion source ?
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
27
SRF Test Facility
RF/Coupler
processing
Test Cave
Cryomodule
Assembly
VTA
Cryomodule
Assembly
Chemistry
Class 10
Class 10,000 Class 100
Mezzanine
(lab space)
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
28
Summary
 We have been operating the SNS SCL for ~ 2 years with
beam
 Generally it is quite forgiving
 Run with many cavities off / entire cryo-module removed /
gradients far from design
 Need tools to adapt to rapidly changing conditions
 Cavities are like individuals – each has it’s own set of
difficulties / strengths
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
29
Normal Conducting Linac
 CCL Systems designed and built
by Los Alamos
 402.5 MHz DTL was designed and
built by Los Alamos
 805 MHz CCL accelerates beam to
186 MeV
 Six tanks accelerate beam to 87
MeV
 System consists of 48 accelerating
segments, 48 quadrupoles, 32
steering magnets and diagnostics
 System includes 210 drift tubes,
transverse focusing via PM quads,
24 dipole correctors, and
associated beam diagnostics
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
30
The Beam Power Ramp Up Goal
 We need to ramp to full design power, at full final reliability with
decreasing beam study time by Oct. 2009
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
31
Cavity Fault Impact on Beam Arrival Times
for a Proton Linac
Arrival
Time:
Cavity:
Arrival
Time:
Cavity:
 Proton beams for high power applications (< 10 GeV) are not fully relativistic and the
velocity is energy dependent
 If a cavity fails, the beam arrives at downstream cavities later
 For SNS if an upstream cavity fails, the arrival time at downstream cavities can be
delayed up to 5 nsec
 This is over 1000 degrees phase setting of an 805 MHz RF cavity
 Our goal is to set the cavity to within ~ 1 degree
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
32
50
30
10
-10
-30
-50
-70
-90
… …
-110
0.52
Beam-RF Phase (deg)
Longitudinal Acceleration Modeling (Application
Programs – Online Model)
Medium-Beta
(g=0.61)
0.57
0.62
Parmila
0.67
0.72
OLM
 Drift-kick-drift method
 Assume design field profiles throughout the cavity
 Transit Time Factor is calculated at each gap, based on a fit of Superfish calculations
 The beam sees a large phase slip from gap to gap as it traverses the cavity
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
33
Test of the Cavity Recovery Method – Single
Cavity “Failure”
Turn off cavity 7
 Turned off cavity 7, rescaled the downstream cavity phase setpoints
 Downstream cavity phase setpoints changed > 1000 degrees
 A beam measurement check with the last cavity showed it was within 1
degree of the scaled prediction
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
34
Phase Scaling Error (deg)
Expected Errors from the Scaling Method (I)
Beam Energy (MeV)
 Uncertainty in the cavity positions leads to errors in the predicted
change in phase
 Relative cavity positions are known to a few mm, so < 1 degree error
is expected from this uncertainty
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
35
Phase Scaling Error (deg)
Expected Errors from the Scaling Method (II)
Beam Energy (MeV)
 Uncertainty in the energy gain/cavity results in errors in the predicted
change in cavity phase
 Energy gain is known to within a few hundred keV, so the error from
this uncertainty is 1-2 degrees
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
36
Cavity Fault Recovery Scheme at SNS
 Additional applications of the cavity recovery scheme
 Missing cryo-module tests to evaluate the impact on beam loss from
removing entire cryo-modules from service for repairs.
 Recovery from a control system failure that resulted in 3 broken cavity
tuners.
 While intended for use in recovering from a single cavity failure,
the scheme has been used more often to recover from more severe
situations
 Usually takes days to assess the situation, minutes to apply the
recovery scheme
 Previously took days to setup the cavities (now ~ 1 shift) with beam
based measurement techniques
 This technique is considered a “standard practice” by now at SNS
 Future improvements may include a more automated invocation
OAK RIDGE NATIONAL LABORATORY
U. S. DEPARTMENT OF ENERGY
37