Delay Times in Breakdown Triggering
Antoine Descoeudres,
Sergio Calatroni, Mauro Taborelli
CERN, TS-MME
CLIC Breakdown Workshop
May 2008
CLIC Breakdown Workshop – CERN, May 2008
1 / 13
Time matters
•
t0 : high voltage is applied between the electrodes
ho
Q:
wl
ong
?
•
t > t0 :
no spark…
spark !
Study of delays can give information about the breakdown mechanism
CLIC Breakdown Workshop – CERN, May 2008
2 / 13
DC spark : experimental set-up
1)
UHV
2)
28 nF
power supply
(up to 12 kV)
V
V
HV probe
current probe
to scope
•
Charged capacitor connected to the electrodes during 2 sec, through a
high current mechanical relay
•
Voltage and current measured with probes, connected to a scope
¾ HV probe : 20kV, 75MHz, 1:1000, 100MΩ
¾ current transformer : 500MHz, 1kA peak
¾ scope : 1GHz
CLIC Breakdown Workshop – CERN, May 2008
3 / 13
DC spark : experimental set-up
NB: slow discharge of the capacitor during the voltage ‘pulse’ due to
finite resistance of HV probe (RC circuit)
28 nF
V
•
i
100 MΩ
circuit time constant :
τ = RC = 2.8 sec
resulting ‘pulse’ :
CLIC Breakdown Workshop – CERN, May 2008
4 / 13
DC spark : experimental set-up
•
The moving part of the mechanical relay
is bouncing sometimes (loss of contact)
…but voltage pulse is ‘flat’ over at least 5 ms
(and 5 ms = max. delay observed)
CLIC Breakdown Workshop – CERN, May 2008
5 / 13
Timing diagram of a spark
•
Schematic evolution of the voltage and current across the electrode gap
???
V
I
t
delay
HV relay
is closed
(< 5 ms)
spark
~ 2 μs
rising time
~ 100 ns
CLIC Breakdown Workshop – CERN, May 2008
6 / 13
Example of a spark
delay
CLIC Breakdown Workshop – CERN, May 2008
7 / 13
Delay times with Mo electrodes
•
Histogram of delays
population #2
‘delayed’ brkds
population #1
‘immediate’ brkds
average : 1.17 ms
(σ = 0.33 ms)
average : 129 ns
(σ = 16 ns)
•
voltage rising time: ~ 100 ns
2 populations, separated by a gap (1 – 100 μs)
Two different breakdown mechanisms ?
CLIC Breakdown Workshop – CERN, May 2008
8 / 13
Delay times with Mo electrodes
• Delays after conditioning
• Delays during conditioning
percentage of
delays < 200 ns :
82%
immediate brkds dominate
during conditioning
gap
24%
delayed brkds dominate
after conditioning
CLIC Breakdown Workshop – CERN, May 2008
9 / 13
Delay times with stainless steel electrodes
•
•
Conditioning
•
both populations observed at every field, but
Same tendency
???
¾ immediate breakdowns dominate during conditioning
¾ delayed breakdowns dominate after conditioning
•
pop. #2: delays increase linearly with breakdown field ???
CLIC Breakdown Workshop – CERN, May 2008
10 / 13
Delay times with copper electrodes
•
Cu : ‘immediate’ conditioning, lower breakdown field
•
Large majority of immediate breakdowns
average : 163 ns
(σ = 44 ns)
CLIC Breakdown Workshop – CERN, May 2008
11 / 13
Delay times with different materials
Cu
Ta
Mo
SS
Eb = 170 MV/m
Eb = 300 MV/m
Eb = 430 MV/m
Eb = 900 MV/m
R = 0.07
R = 0.29
R = 0.76
R = 0.83
R = fraction of delayed breakdowns (excluding conditioning phase)
R increases with average breakdown field
CLIC Breakdown Workshop – CERN, May 2008
12 / 13
Summary
•
Two populations of delays observed :
¾ immediate breakdowns (average ~ 120 ns)
¾ delayed breakdowns (from 0.1 to 5 ms, average ~ 1.3 ms)
•
Both populations observed at every field
¾ during conditioning mostly immediate brkds
¾ after conditioning mostly delayed brkds
¾ delays ~ breakdown field? (SS)
•
Two different breakdown mechanisms?
¾ immediate: associated with surface contamination? gas desorption?
¾ delayed: once the surface is cleaned?
•
Repartition of delayed / immediate breakdowns
¾ depends on the material
¾ the ratio of delayed brkds R increases with the average breakdown field of
the material
•
ideas are welcome!
Next : in BDR mode
CLIC Breakdown Workshop – CERN, May 2008
13 / 13
Thank you !
CLIC Breakdown Workshop – CERN, May 2008
14 / 13