Transcript Слайд 1 - Indico

Status Update from
the DC Spark Lab
Anders Korsbäck 17.3.2015
Two-Rate Model
Most of the time since CLIC workshop, the Large Electrode System has
been occupied collecting data for testing the Two-Rate Model under
different input voltages, pulse lengths and overall breakdown rates
-3
5M pulse window
25M pulse window
10
Breakdown rate
The latest run was done
at 2.8 kV, 3.5 us pulse
length, gap size 60 um,
with a total of 5356
breakdowns over 840
million pulses, overall
BDR of 6.37e-6
10
10
10
10
10
-4
Drift in BDR over time, long measurement
run at Large Electrode System, 2800 V, 60
um gap, 3.5 us pulse length. BDR
calculated through sliding window of 5 or 25
million pulses, minimum of 10 BDs per
window.
-5
-6
-7
-8
0
1
2
3
4
5
Cumulative nr of pulses
6
7
8
x 10
8
Two-Rate Model
And the statistics of number of pulses before BD...
-3
10
10
Probability density
Probability density
Two-exponential fit A + B
Short-term component A (BDR = 2.17e-3)
Long-term component B (BDR = 9.58e-5)
-4
New: 2.8 kV, 3.5 us
10
10
10
-3
Probability density
Two-exponential fit A + B
Short-term component A (BDR = 2.51e-3)
Long-term component (BDR = 1.64e-4)
-4
Old: 2.3 kV, 6 us
-5
-5
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
Nr of pulses before breakdown
0.2
0.4
0.6
0.8
1
1.2
1.4
Nr of pulses before breakdown
1.6
1.8
2
x 10
4
2
x 10
The Two-Rate Model has been
shown to fit under another set of
conditions! It continues its
victorious march towards
scientific relevance!
4
Probability density
Probability density
10
10
Probability density
Two-exponential fit A + B
Short-term component A (BDR = 1.64e-3)
Long-term component B (BDR = 9.38e-5)
-4
Old: 2.5 kV, 5.83 us
10
-5
0.2
0.4
0.6
0.8
1
1.2
1.4
Nr of pulses before breakdown
1.6
1.8
2
x 10
4
Time to breakdown at BDR = 1
Just a little thing I did on the side of the Two-Rate Model data
collection...
I had observed that, when running at BDR << 1, breakdowns
would almost all happen at the very end of the pulse:
18
x 10
5
16
Time to BD for
a run with pulse
length 6 us
Probability density
14
12
10
8
6
4
2
0
2.5
3
3.5
4
4.5
5
Time to breakdown (s)
5.5
6
6.5
x 10
-6
What’s the distribution of time to breakdown like if the pulse is
long and high enough that BDR = 1? Might the distribution give
some hints at the nature of the microstructural process behind
breakdown?
Time to breakdown at BDR = 1
Here are the obtained distributions, for measurements running a
few days at constant conditions and BDR near 1:
16
x 10
5
3750 V
3500 V
14
Probability densities (histograms
normalized to 1) of time to breakdown in
the Large Electrode System, 8 us pulse
length, BDR ~ 1, 60 um gap. The voltage
pulse has a dip at 2.0 and 4.1 us due to a
reflection along the PFL, explaining the
sudden drop in probability density in the
3750 V case.
Probability density
12
10
8
6
4
2
0
0
1
2
3
4
Time to breakdown
5
6
7
x 10
-6
Data is compromised by the ripple in the PFL due to the
reflection dip at 2 and 4 us, would need a flatter pulse.
Still, the apparent two-Gaussian distribution looks interesting,
might be consistent with Two-Rate Model: later Gaussian
primary BD, earlier Gaussian follow-up BD
Time to breakdown at BDR = 1
Drift in mean time to breakdown was also seen in the longer,
3750.V run, which lasted about 4 days:
Drift almost
unrealistically
sinusoidal, though
existence of longterm drift consistent
with BDR drift over
time at BDR << 1
Mean time to breakdown
4
x 10
3.5
-6
Drift of mean time to breakdown over a 4-day measurement
run in the Large Electrode System, pulse amplitude 3750 V,
length 8 us, gap size 60 us, BDR ~ 1. Mean time to
breakdown over the course of the run shown as a moving
average with different window sizes.
MA 11
MA 101
MA 1001
3
2.5
2
1.5
0
1000
2000
3000
4000
5000
6000
7000
Total nr of breakdowns
Model: The breakdown process can be described by microstructural ”mean time to breakdown”, BDR << 1 happens when
pulse length is significantly shorter than it, distribution of time to
BD is then the left tail-end of the distribution seen for BDR = 1