Transcript breakdown statistics and conditioningx

Statistics of breakdown and
conditioning in pulsed dc and rf
systems
Anders Korsback, Jorge Giner Navaro, Robin
Rajamaki and Walter Wuensch
mini-MeVArc, 21 March 2016
Walter Wuensch, CERN
Motivation
Statistics
Physics – The statistical properties of breakdown may give us insight into
the evolution of the surface under pulses, the underlying trigger
mechanism and what happens to the surface after breakdown.
Practical – We try to operate the structures ever closer the gradient limit
but we demand reliability and life-time. The statistical properties may
give the essential life functions.
Conditioning
Physics – Determining what exactly gets better as a structure conditions
could give us insight into etc.
Practical – Conditioning is long, months at 50 Hz, and consequently
expensive. How can we shorten this process or replace it with another
one?
mini-MeVArc, 21 March 2016
Walter Wuensch, CERN
Breakdown statistics: RF and DC
Anders Korsbäck
CERN / University of Helsinki
Robin Rajamäki
CERN / Aalto University
Jorge Giner Navarro
CERN / University of Valencia
Walter Wuensch
CERN
WW note: The first
time I saw a BD
interval plot was by
W. Farabolini
What is breakdown statistics?
•The operational history of an accelerating structure tested in Xbox-1
is shown. Instead of accumulating breakdowns at a constant rate, it
shows a “staircase” structure on many scales in a self-similar way.
•Hence, a single (overall) breakdown rate, i.e. nbreakdowns/npulses, is
clearly insufficient to describe what’s going on. It doesn’t say
anything about when breakdowns happen…
•…in relation to the overall history
•…in relation to each other
Two-Rate Statistics
•Or, to visualize what was just explained, let’s return to the operational
history vector and to number of pulses to breakdown:
□: non-breakdown pulse, ■: primary BD, ■: follow-up BD
□□□□□□■□■□□■□□□□□□□□□□□□□□□□□■■□■■□□□□□□□□■■□■□□□□
7
2 3
18
1 21
9
1 2
•Red numbers are
values for nr of
pulses to BD for
primary BDs, blue for
follow/up BDs. Red
and blue are
individually
Poissonian, giving a
two-exponential
probability density
when put together
2, 3, 1, 2, 1, 1, 2
7, 18, 9
Comparison of rf and dc
700
Data
Long-term BDR= 2.59e-005
Short-term BDR= 2.07e-003
Two-exponential fit
-3
10
rf
600
400
300
KEK
Probability density
Cumulative BDs
500
-4
10
-5
10
200
100
-6
0
10
0
dc
Cumulative nr of breakdowns
3.5
1
x 10
2
3
Cumulative pulses
4
5
6
7
x 10
0
1
2
3
4
5
6
Number of pulses before breakdown
7
8
4
x 10
4
3
2.5
2
1.5
1
0.5
0
0
2
4
mini-MeVArc, 21Cumulative
Marchnr2016
of pulses
6
8
x 10
8
Walter Wuensch, CERN
Breakdown positioning in
CLIC prototype RF
accelerating structures
CLIC workshop 2016
R. Rajamäki*, W. Farabolini, J. Giner Navarro, T. Argyropoulos, B.
Woolley, W. Wuensch
19.01.2016
*Aalto university / CERN
Introduction
• What?
• Localize BDs in RF accelerating structures
• Why?
• Structure diagnostics
• Breakdown studies
• How?
• RF power and phase
• Directional coupler
• Structure vibrations
• Accelerometer
• Electron emission
• Faraday cup
• Spectrometer
• Photons
• PMT/ camera
• X-ray
Structure diagnostics (1/2)
TD26CC
y-axis projection
BD cell map
x-axis projection
Method comparison (3/3)
τd,edge
1.
3.
2.
Observations:
1. Methods are generally in agreement
2. Non-symmetric spread
3. Peak of correlation method
What about possible expanations?
τd,corr2d
Breakdown migration?
ii.
i.
Mainly upstream migration
Downstream migration
iii.
Upstream migration
Possible migration scenario
Courtesy of W. Farabolini
Spatio-temporal correlations
i.
iii.
ii.
(𝑛)
Δτd
=
(𝑛)
τd
Vertical lines = artefacts of conditioning algorithm
(𝑛−1)
− τd
Breakdowns arriving shortly after each other occur close to each other.
Newest pulsed dc data
Long pulsed dc run with electrodes prepared with same procedure as rf structures.
mini-MeVArc, 21 March 2016
Walter Wuensch, CERN
Heat treatment and joining
KEK/SLAC
Tsinghua U.
SINAP 21 March 2016
mini-MeVArc,
CERN
Walter Wuensch, CERN
BDR as a function of field
mini-MeVArc, 21 March 2016
Walter Wuensch, CERN
And corresponding distributions
mini-MeVArc, 21 March 2016
Walter Wuensch, CERN
Performance summary at CLIC specifications
30 5
𝐵𝐷𝑅 ∝ 𝐸 𝜏
mini-MeVArc, 21 March 2016
Walter Wuensch, CERN
Conditioning
Accelerating structures do not run right away at full specification – pulse length and
gradient need to be gradually increased while pulsing. Typical behaviour looks like this:
Pulse length steps
BDR falls
during flat E
run
4 million pulses per day at 50 Hz
mini-MeVArc, 21 March 2016
Walter Wuensch, CERN
Comparing conditioning
Scaled gradient vs cumulative number of PULSES
Pulses
𝐵𝐷𝑅 ∝ 𝐸 30 𝜏 5
Scaled gradient vs cumulative number of BREAKDOWNS
Breakdowns
mini-MeVArc, 21 March 2016
Walter Wuensch, CERN
Newest pulsed dc data
Long pulsed dc run with electrodes prepared with same procedure as rf structures.
mini-MeVArc, 21 March 2016
Walter Wuensch, CERN
Longer term operation
7 months @ 50 Hz
rf
pulsed dc
30 days @ 1 kHz
mini-MeVArc, 21 March 2016
Walter Wuensch, CERN
Long term evolution of BDR
pulsed dc
rf
Pulses
mini-MeVArc, 21 March 2016
Walter Wuensch, CERN
Effect of venting system
-82
10
Normalized BDR ( (MV/m)
-74
10
Normalized BDR ( (MV/m) -30 ns-6 )
-83
10
-30
-6
ns )
Test vent of dc
system, 3 days
Before venting
After venting
gradient -3.146
gradient -28.14
-84
10
-85
10
-76
10
-86
10
-78
10
9.8
1
9.9
10
10
Cumulative nr of pulses
-80
10
4
-82
10
2
1. Feedback phase, gradient -7.87
2. Constant voltages phase, gradient -2.59
3. Stepped voltage phase, gradient -3.15
4. Reconditioning after 3 day vent, gradient -28.12
-84
10
8
10
mini-MeVArc,
21 March 2016
9
10
Cumulative nr of pulses
3
10
10
Walter Wuensch, CERN
Dynamic Vacuum Meter is
at CERN
Antti Meriläinen1,2, Robin Rajamäki3, Ivan Kassamakov1,2, Walter
Wuensch3, Kenneth Österberg1,2 and Edward Hæggström1
1) Department of Physics, University of Helsinki
2) Helsinki Institute of Physics
3) CERN
www.helsinki.fi/yliopisto
19.1.2016
Dynamic Vacuum Meter
Vacuum tube or AS-element
Breakdown
p(t)
RQCM = 6.5 mm
50 cm
x = 5 ± 0.5 cm
RBeam= 5 mm
Quartz crystal
microbalance (QCM)
as reference
Dr. Walter Wuensch, Introduction to
CLIC, Collaboration meeting at HIP
19.10.2010
www.helsinki.fi/yliopisto
19.1.2016
Electrode Design
RDesign
For 0.1 mm gap, r ≥ 0.3 mm
RDesign = 0.5 mm ≥ 0.3 www.helsinki.fi/yliopisto
mm
19.1.2016
Cad Design
www.helsinki.fi/yliopisto
19.1.2016
Optics and Electrodes
www.helsinki.fi/yliopisto
19.1.2016
DVM & DC Spark
Estimated signal for conditioning
process
Cu atoms
release
Estimated signal for Breakdown
Cu atoms
release
www.helsinki.fi/yliopisto
Breakdown
Spark
19.1.2016
New HV pulser
15 kV, 1 kHz Marx
generator from ISEL,
Lisbon.
Fast rise and fall time!
mini-MeVArc, 21 March 2016
Walter Wuensch, CERN
Electrode pipeline
Running
• Hard copper, with acoustic sensors - running
Available
• Cu, CuAg, stainless steel - SLAC
• Nb
• 3-D printed Ti
• Small ridge for optical access
• About 10 pairs of diamond machined Cu
• Ceramic spacers
Under preparation
• Voids - Helsinki
• Diamond like coating - PSI
mini-MeVArc, 21 March 2016
Walter Wuensch, CERN
Nb electrode made in central
workshop.
We propose to build this
cool-able to cryogenic
temperatures.
mini-MeVArc, 21 March 2016
Walter Wuensch, CERN