4★★★★모듈레이터의 종류 - Fukuda - RF Sourced and Modulator (mainly for linac)

RF Sourced and Modulator
(mainly for Linac)

S. Fukuda
KEK: High Energy accelerator Research
Organization

For
School for Accelerator Technology
And APplication

2012/2/1
S. Fukuda Schooi for Accelerator Technology and
Application
1
Introduction
RF Source Importance
Modern Accelerator is so called RF accelerator, and RF Is
the key technology.

Accelerator frequency is ruled by the existence of the rf
source of matched frequency.

RF system is expensive and large fraction of total cost is
shared in RF source. Therefore choice of RF is very important.

RF source is the source of failure (arcing etc.) and careful
availability consideration and constant effort of maintenance
are required.

2012/2/1
S. Fukuda Schooi for Accelerator
Technology and Application
2
Index
1.Introduction
2.Introduction : Accelerator and RF
3.RF Source suitable to accelerator
4.Klystron Theory
5.Klystron parameters
6.RF Source other than Klystron
7.Modulator General
8.Modulator- Line Type Modulator
9.Modulator- Other than Line Type Modulator

2012/2/1
3
Introduction(1)
Circular Accelerator
Lorentz Force Equation
F=eE+ev x B
F
v
B
Cyclotron
Electron has a circular orbit
In the magnetic field.
Revolution time is constant if B=const.
RF field is repeadedly used.
2012/2/1
4
History of the Linac(1)
Ising(1924)
Wiedroes Linac
(1928)
Alvarez Linac
(1946)
Acceleration
Acceleration
Drift (constant velocity)
Drift (constant velocity)
2012/2/1
5
Introduction(2)
Linear Accelerator (Linac)
Charged particles are accelerated linearly.

High frequency electromagnetic wave (micro-
Wave) is used for the acceleration.
Principle:
2012/2/1
6
History of the Linac(1)
Linac was developed after the World War II.
Due to the Radar Technology for military purpose

Short pulse high power microwave source.
Klystron was invented by Varian Brothers, in 1939.
(a few microsecond, 10 MW 10cm microwave)
Related high-power power supply
(300kV,300A pulsed source)
RF Technology is the key technology for linear accelerator
Reflection
2012/2/1
7
History of Linac(2)
Stanford University Electron Linac(1950-)
Linac=Very big electron microscope
R. Hofstadter studied nuclear structure by electron
scattering(1954-57) and was awarded a Nobel prize (1961).
Nuclear surface
(~10
-13
cm
Smooth Density
Change
Comparing with Dr. R. L. Moessbauer, Nobel
Winner In 1961, he was called the son of
Big Science.
2 mile accelerator was constructed in SLAC
(Stanford Linear Accelerator Center, CA, USA) (1966-)
SLC(SLAC linear Collider) was constructed (1983-)
Both cases, RF source (Klystron) was developed in SLAC
2012/2/1
8
Variant of Linac - Electron Linac
Since the electron mass is light, it is easy to be accelerated up
to the light velocity (velocity is constant).
overall structures are the same

Longest Linac is SLAC 2-mile Linac (3.2km and 20GeV)
Second longest linac is KEK linac (500m and 8 GeV)
Small linac was developed for medical use (2-3m and 30 MeV)
Short pulse linac of which pulse duration of a few to 20 microsec.
-----wavelength of 10cm (S-band), 5cm (C-band) and X-band (2.5cm)
Long pulse linac of which pulse duration of a millisecond
------wavelength of 20cm class (L-band, ex., 1.3GHz)
2012/2/1
9
Variant of Linac - Proton Linac
Since the proton mass is 2000 times heavier than electron, it is
hard to be accelerated up to the light velocity and depending on
the |(=v/c), several structures are used in proton linac and each
structure has a suitable frequency.
RFQ structures-----VHF band such as 201MHz-432MHz
Alvarez structures/DTL(Drift Tube Linac)-----same as RFQ
ACS(Anular Coupled Structure), SCS(Single Coupled-cell Structure) etc------
-Harmonics of low | section (0.8-1GHz)

2012/2/1
10
RFQ Structure DTL Structure Element of ACS
RF Sources suitable for accelerator
Factors required to accelerator rf source
1. Frequency range - higher than 0.3 GHz, klystron is best rf source. Less
than 0.3 GHz, Solid state amplifier, IOT and Tetrode are used.
2. Peak power capability-related with energy gain, short linac, and cost
benefit. How to minimize the discharge rate in tube and structures.
3. Average power capability related with duty cycle or repetition rate. Cw
accelerator.
4. Gain relate with driver amplifier. High gain is preferable; Generally
klystron is high gain such as 50 dB. IOT is around 20 dB. Power tetrode is
poor gain.
5. Phase stability klystron is voltage driven device and if applied voltage is
stable, phase property is excellent.
6. Simplicity, Availability and Long life klystron is much matured device
and satisfy these requirements.
2012/2/1
11
Available RF Sources states of art now
Average power vs frequency plot Peak power vs frequency plot
Frequency range
Suitable for accelerator
Most available rf source for
accelerator use, klystron is
a best device for
average and peak power
capability.
Efficiency (%) Bandwidth
(%)
Gain
(dB)
Relative
Operating
Voltage
Relative
Complexity of
Operation
Gridded Tube 10-50 1-10 6-15 Low 1
Klystron 30-70 1-5 40-60 High 2
Magnetron 40-80 High 3
Helix TWT 20-40 30-120 30-50 High 3
Coupled Cavity TWT 20-40 5-40 30-50 High 3
Gyrotron 10-40 1 30-40
High
5
IOT 1070 2025 High 3
Comparison of various RF
sources
2012/2/1
12
Klystron Theory
2012/2/1
13
Klystron Structure and Mechanism
Key Component of Klystron
Electron Gun
RF Cavity
Drift Tube
RF Window
Beam Collector
Focusing Magnet
Cooing System

Basic Mechanism of Klystron
Electron is velocity-modulated in an input cavity.
Then, electrons form bunch; density modulation
Bunches are de-accelerated in an output cavity
---> beam energy to rf power
See: Inverse process of accelerator
Accelerator ----> rf power to beam energy

2012/2/1
14
Important parameter; Perveance
Simon's formula
Definition of Perveance
Low perveance klystron has
high efficiency but applied
voltage is high.
Accelerator Laboratory, KEK
It characterize the space charge
force and deeply related with the
bunching formation and therefore
with efficiency.
2012/2/1
15
Beam Bunching Theory
Basic Mechanism of Klystron(Linear theory)
Electron is velocity-modulated in an input cavity.
Then, electrons form bunch; density modulation
For this bunching mechanism, a simple linear theory is educative. There are two
approaches:
- Ballistic theory: electron which has a modulated velocity propagate without any
interaction with other electrons.
- Space-charge wave theory: entire electrons behave like a wave which is
ruled by space charge force, and automatically it contains the space charge force
effect.

More Realistic Analysis (large Signal Analysis)
For the interaction region near to output cavity, linear theory is not applicable and non
linear large signal analysis are required.
One dimensional analysis: Disc model
2.5 dimensional analysis: Particle in cell Analysis including magnetic interaction
3 dimensional analysis: MAFIA / MAGIC code

2012/2/1
16
Klystron: Ballistic Theory (1)
Treatment of individual electrons without interaction

Inititial electron energy:

Electron Energy gain in the input cavity:

Assume V
1
<<V
0
: Linearization
The arrival time t
2
in the second cavity depends on
the departure time t
1
in the first cavity with
the assumption of an infinite thin gap:

or with

and called bunching parameter
0
2
2
1
eV mu =
t eV mu mu e sin
2
1
2
1
1
2
0
2
=
2
1
0
1
0
) sin 1 ( t
V
V
m u u e + =
) sin
2
1 (
0
1
0
t
V
V
m u u e + =
1
0 0
1
0
1
1
1
0
1
0
1 1 2
t sin
V 2u
mV
2u
1
t
) t sin
2V
mV
(1 u
l
t
u
l
t t e
e
=
+
+ = + =
0
0
u
e
u =
1
1 0
V 2
mV u
= X
1 0 1 2
t sin t t e u e e X + =
2012/2/1

17
Klystron: Ballistic Theory (2)
Because of charge conservation:
Charge in the input cavity between time t
1
and t
1
+dt
1
equals
the charge in the output cavity between time t
2
and t
2
+dt
2

then

0
1,57
3,14
4,71
6,28
0 1,57 3,14 4,71 6,28
Input Gap Departure Phase
O
u
t
p
u
t

G
a
p

A
r
r
i
v
a
l

P
h
a
s
e
X=0
X=0.5
X=1
X=1.5
1
2
1 2 1
1
2
2 2 1 1
t
t
t cos 1
t
t
t t
d
d
I I and X
d
d
with d I d I = = = e
1
1
2
t cos 1 e X
I
I
=
Fourier transformation of the current in the output gap I
2
)] sin( ) ( cos [
0 2 0 2
1
0 2 u
e
u
e + + =

=
t b t
n
a I I n n
n
) ( ) ( cos ) / 1 (
2 0 2 2
0
0
t
d
t
n
I an
e
u
e t
t u
t u
=
}
+
=
t
t
e e e t ) ( ) sin ( cos ) / (
1 1 1 0 t
d
t
X
t
n
I an
0 ) ( ) sin ( sin ) / (
1 1 1 0
= =
}
t
t
e e e t
t
d
t
X
t
n
I bn
with J
n
Besselfunction of the n- th

order
) ( ) ( sin ) / 1 (
2 0 2 2
0
0
t
d
t
n
I bn
e
u
e t
t u
t u
=
}
+
) ( cos ) ( 2
0 1
1
0 0 2 u
e + =

t
n nX
J I I I n
) cos( ) ( 2
0 1 0 u
e
e
= t X
J I I
P V I V I P Beam
58 . 0 ) 2 / )( 2 / ( 58 . 0 2
0 0
= = =
e e e
2012/2/1 S. Fukuda Schooi for Accelerator Technology and Application

18
Klystron: Space-Charge Waves Theory
Space charge forces counteract the bunching
Any perturbation in an electron beam excites an oscillation with the plasma
frequency

Therefore we have 2 waves with the Phase constants

And therefore
The group velocity is
The density modulations appear at a distance of

0
0
0
c
e
m
e
p
=
) / 1 (
1
e e | |
p e e
+ = ) / 1 (
2
e e | |
p e e
=
u
e
e
/ e | =
) / 1 /(
1
e e
p
e e
u u
+ =
) / 1 /(
2
e e
p e e u u
=
u
d d
u
e
e
g
= = | e/
p
e p
u
e t
/ 2 =
This means that the drift space or the distance between cavities is determined by
the plasma frequency (klystron current) and the electron velocity (klystron voltage)
and is given by .

4 /
p
More realistic approach: Space charge of electrons in the metallic tube is
reduced due to the mirror effect. Therefore plasma frequency reduced factor is
Introduced, and reduced plasma frequency (wavelength) determines optimum
length of klystron.
2012/2/1

19
Comparison of the beam trace among
two approach
Ballistic Analysis Approach
No space charge interaction
Space-charge wave approach
Space charge repulsion is included
2012/2/1
20
More Realistic Analysis of Klystron
One dimensional disc model
Constant diameter beam is expressed as series
of disc and moving as ballistic manner.
Space charge among discs is considered
Cavity-beam interaction is properly considered.

2.5 dimensional PIC program
Electron particle in 3D
Solving under axial symmetric condition
Space-charge and magnetic field effect are
included.
Realistic approach and fairly good
agreement between simulation and test
2012/2/1
21
Klystron Parfameters
2012/2/1
22
Klystron Efficiency Issues
Klystron efficiency is important for high duty machine
and cw accelerator use
There are two approches to raise the klystron
efficiency
- Low pervence beam klystron
High voltage, low current way and arcing in
the gun area is concerned for high power
application
Multi-beam Approch to decrease voltage
- Using Higher order harmonics cavity
From the basic fact that saw-tooth like voltage
modulation gives the high efficiency
Most popular way is using 2
nd
harmonics
cavity in the bunching cavity region
More complicated structure to introduce
multiple harmonics cavities
- By using computer simulation,a wide band tuning
klystron and a very high efficiency klystron such as
80% are realized.
Concept for high efficiency tube
Using the harmonics

Another way is using long distant
drift tube to mixing second
Harmonics component by split
Orbit due to space charge.
2012/2/1
23
Multibeam Klystron
Idea
Klystron with low perveance: => High efficiency but high voltage
Klystron with low perveance and low high voltage : =>low high voltage but low power
Solution
Klystron with many low perveance beams:
=> low perveance per beam thus high efficiency
low voltage compared to klystron with single low perveance beam
Measured performance
Operation Frequency: 1.3GHz
Cathode Voltage: 117kV
Beam Current: 131A
perveance: 3.27
Number of Beams: 7
Cathode loading: 5.5A/cm
2
Max. RF Peak Power: 10MW
RF Pulse Duration: 1.5ms
Repetition Rate: 10Hz
RF Average Power: 150kW
Efficiency: 65%
Gain: 48.2dB
Solenoid Power: 6kW
Length: 2.5m
Lifetime (goal): ~40000h
THALES TH1801
2012/2/1
S. Fukuda Schooi for Accelerator Technology and Application
24
Klystron Characteristics
Childs Law- In space charge limit current, emission current is as
follows.

where is perveance, and I is current, and V is applied voltage.
Then, power is

Powers variation

Klystrons Impedance

Therefore, klystrons impedance varies with applied voltage.

2 / 3
V P I =
2 / 5
V P V I P

= - =
V P V P
V
I
V
Z

1
2 / 3
= = =
V
V
V P
V V P
P
P o
o
o
2
5
5 . 2
2 / 5
2 / 3
= =
P
2012/2/1

25
Phase Variation of RF from Klystron
For linac, phase stability of RF is extreamly important
because variation dirctly reduces to energy variaion DE
Since klystron is voltage driven device and votage determine the electron
velocity, rf phase strongly depends on the voltage variation.

and

Usual S-band tube operating at 300kV range, this
phase variation Is roughly 6-8 deg./(dV/V%).
If you want to get AE /E=0.1%, then 0.025% of
voltage flatness is required.

Water cooling variation of tube body also causes
rf phase variation.
This happens since water temeperature changes
the gain cavitiys detuning frequency and bunch
center changes from original position.
this phase variation Is roughly
0.5-1 deg.phase/1 deg water temp.

-8
-6
-4
-2
0
2
4
6
8
10
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
Ratio of applied voltage @300kV (%)
P
h
a
s
e

(
d
e
g
r
e
e
)
simulation
measured
-25
-15
-5
5
15
25
-10 -5 0 5 10
Dwater temperature [degree]
D
o
u
t
p
u
t

p
h
a
s
e

[
d
e
g
r
e
e
]
free (ANSYS)
rigid (ANSYS)
Experiment
2
2
0
0
0
0
) 1 (
1
1
2 2
mc
eV
c
L
u
L
+
= =
t t
u A
)
%
deg
( 8 6
) (
V
V V
A
A
u A
=
2012/2/1

26
Other issues for Klystron
Other issues important to handle klystron are omitted here
ad only lising up below.
Beam focusing of klystron :Brillouin focussing or confined
flow:
Electron gun issues and arcing problem there
RF window and its failure or protection from break-downer.

Multipactoring in the window / drift-tube
Cathode emission property and observation of emission
history.
Instability by various reasons

2012/2/1

27
RF Source other than Klystron
Klystron comprises of several cavities and their sizes are proportional to
wavelength. Therefore, lower is the frequency, larger is the klystron size and
actual lower limit of klystron is around 300MHz.
Tetrode
Proton linac low beta section such as 200 MHz, Tetrode (or Triode) is used.
RF system is complicated because each grid requires rf circuit.
Final stage ( a few MW) has a poor gain such as 6dB, and multi-stage
amplifier system is employed.
IOT(Klystrode)
For 100-1300 MHz Range and up to 100kW (peak or cw), IOT is used, High
efficiency such as 60% is achievable. No saturation nature and even under
LLRF control, high efficiency is achievable. Low gain nature requires
relatively large driver amplifier.
Solid state amplifier
For the same frequency range as IOT, promising device
2012/2/1
28
RF Source other than Klystron (2)
Magnetron
Magnetron is an oscillator and its oscillation frequency varies. Feedback to
stabilize or tracking the frequency variation is inevitable. Crossfield amplifier
is not so popular. On the other hand since magnetron is cheap, this rf source
is used for small accelerator application
Gyroklystron
Since gyrotron is oscillator, gyroklystron was intensively studied for NLC (X-
band or more higher frequency ILC). Higher the frequency more suitable this
device is, while now intense study is not reported. Around 30-50MW output
level was reported. Expensive and difficult handling then klystron.
2012/2/1
29
30kW IOT 1.3GHz for ERL Use
30W IOTCPI)

Specification
2012/2/1
30
High Efficiency but low gain device
El
ec
tro
n
bu
nc
h
Electron bunch is modulated in cathode-grid
region, and no higher frequency device.
UHF band there are lots of broadcast transmitter
Application of around 100kW cw..
Progress of Recent Solid State Amplifier
2012/2/1
31
For CW use, solid-state amplifiers replace to
Klystron. High efficiency is achievable.

In KEK, 20-30 kW cw solid-sate amplifier
(1.3GHz)is more likely candidate than
IOT or Klystron for cERL.
Modulator
2012/2/1
32
Requirement to RF source from
accelerator specification
Energy Gain E
It depends on the accelerator final energy, and
it is the SQRT of P (proportional to electric field)
Energy width oE/E should be as small as possible
this corresponds to minimize P/P and /.

In order to achieve these requirements, good quality of
modulator (good stability of output pulse) is required.
2012/2/1
33
Various Modulator
Modulator for short pulse ( microsecond order)
Line type modulator: most popular, low cost, simple
Pulse forming
PFN (pulse forming network)
Blum Line
Hard tube pulser (Solid state amplifier)
Pulse amplifier
Pulse generation by IGBT
Magnetic compressor modulator
Marx Generator
Modulator for long pulse( a few hundred microseconds to a
few milliseconds)
IGBT modulator with bouncer circuits
Marx generator
~
~
~
2012/2/1

34
Modulator(1)
Line-type Modulator
2012/2/1
35
KEKB Klystron and Modulator
Max. Peak Output Power 108 MW
Max. Average Output Power 30 kW
Pulse Transformer Ratio 1:13.5
Primary Output Voltage 22.5 kV
Primary Output Current 4.8 kA
Total PFN Capacitance 0.6 F
Pulse Rise time(10-90%) 0.8 s
Pulse Flatness(P-P) 0.3 %
Pulse Width 5.6 s
Thyratron Anode Voltage 45 kV
Thyratron Anode Current 4.8 kA
Thyratron Average Anode Current 1.3 A
Repetition Rate 50 Hz
Modulator Specifications
Output Power 46 MW
RF Pulse Width 4.0 s
Efficiency 45 %
Perveance 2.1 A/V
3/2
Beam Voltage 298 kV
Repetition Rate 50Hz
Klystron Specifications
Pulse Transformer Tank
Klystron
Pulse Modulator
SLED
2012/2/1
36
Circuit Diagram of KEKB Modulator
Klystron voltage,
current
waveforms and rf
waveform
Accelerator Laboratory
37
PFN-type modulator
LC resonant charging
De-Qing
Single thyratorn switch

All components are unit type
2012/2/1
37
Principle of Line Type Modulator
(Most popular modulator)
Basic Circuit

Pulse width is determined by the traveling time of the line

Matching condition (Z0 is characteristic impedance of co-axial line)

line of length the is where
u

.
2
= t
R Z =
0
2012/2/1

38
Line Type Modulator
(Most popular modulator)
DC power
supply
Charging
Circuit
Discharging
Circuit
Pulse
Transformer
Klystron
Inverter P/S
2012/2/1
39
Discharging Circuit
Cable Equivalent circuit

Analysis of N stage pulse forming network (PFN)
1
st
stage

r-th stage

n-th stage

Solving this exactly is difficult and using Laplace transformation with suitable
approximation or computer simulation is frequently used.

E d i i
C
t i R
dt
t di
L
t
= + +
}
t t t )) ( ) ( (
1
) (
) (
2 1
0
1
1
Open end
0 )) ( ) ( (
1
)) ( ) ( (
1 ) (
1
0
1
0
= + +
+
} }
t t t t t t d i i
C
d i i
C dt
t di
L
r r
t
r r
t
r
0 ) (
1
)) ( ) ( (
1 ) (
0
1
0
= + +
} }

t t t t t d i
C
d i i
C dt
t di
L
t
n n n
t
n
2012/2/1

40
Analysis of P
Laplace transformation
r-stage

General solution

If R=0 (Open end condition)

If matching condition( )

0 ) (
1
) ( )
2
( ) (
1
1 1
= + +
+
p I
Cp
p I
Cp
Lp p I
Cp
r r r
2
2
2
2 2
2 2
2
1
1 cosh
1
2
) 1 sinh( sinh ) 2 (
) 1 cosh(
) 1 sinh( sinh ) 2 (
) 1 sinh(
sinh cosh ) (
e
u e
u e u e
u
u e u e
u
u u
u u
p
LC L
b
b
L
E
a
n n bp
n a
B
n n bp
n a
A
where
r B r A Be Ae p I
r r
r
+ = = =
+ +
+
=
+ +
+
=
+ ~ + =

u u
u
n n
n
CE p I
sinh ) 1 sinh(
sinh
) (
1
+
=

= =

(

+
= =

N
N
N
N
N N
N
N
n
N
C C L L where
C L p p
C
L
n
n
pC p pI
V
n p Z
1 1
1
) coth( 1
sinh
) 1 sinh( 1
) (
) , (
u
u
C L C L R
N N
= =
) 1 (
2
sinh ) 1
2
sinh 2 ( ) 1 sinh(
sinh
) (
2
1
N
N
C
L
p
N
N
N
n
e
C
L
p
V
n n
n
CE p I

+ +
=
u
u
u
u
2012/2/1
41
Characteristics of PFN
For n-stage PFN
pulse width

Characteristic impedance of PFN

Requirement of PFN-Parameter
Klystrons operation point V
s
,

I
s
then Z
s
= V
s
/ I
s
Step up ratio of pulse transformer n, then primary impedance is Z
S
= Z
S
/ n
2

From matching condition, PFN characteristic impedance Z
PFN
= Z
S

Pulse width t is come from Klystrons requirement or system design.
Total capacitance of PFN is derived
by energy equation: energy stored in
PFN=energy supplied to the load

stage number is determined by flat top condition

nC C C nL L L where
LC n C L
i N i N
N N
= = = =
~ =

,
2 2 t
C
L
C
L
Z
N
N
PFN
= =
2
0
2
2
2
1
C
p p
T
p p C T
V
I V
C or
dt I V V C
t
t
=
=
}
2012/2/1
42
PFN Simulation Example
L=1.3[H]
L2=0[H]
C=0.015[F]
Stage number of
PFN
1,3,5,10,20
2012/2/1
43
PFN Issues (Pulse flat top)
Since charging-discharging cycle is repeated in each PFN stage, flat top is folding
shape of n charging waveforms. Ringing of flat top of the pulse is strongly depend
on the stage number n of the PFN. For large n, smoother flat top, but cost is
expensive.
Usually, capacitors varies a few to ten % from nominal, and such variation should
be checked by analysis. C -> C+ affects the variation of Z and current i, and
then causes to V+ ,

On the other hand, effect of the inductance variation is,

Therefore, for the flat top adjustment, variation of capacitor is more serious (4
times larger effect). In order to compensate the droop come from pulse
transformer, larger capacitor should be set at the first stage, and positioning with
the capacitance value is recommended.
Since capacitance is hard to change, introduction of variable inductance is often
employed. Variable mechanism example: mechanical shorted structure, cylinder to
using eddy current effect.

V A
C A
0
4
1
)
L
L
V
V
L
A A
=
0 0 0
, ,
12
11
)
3
2
4
1
( ) ) )
C
C
C
C
C
C
V
V
V
V
V
V
I C Z C C
A A A A A A
~ = + = + =
2012/2/1
44
Issues (Pulse flat top)-II
First modulator for PF injector linac
n=16
C=0.018F
L=0.67H
Problems: Capacitors lead inductance and
connecting wire inductance is comparable with
nominal inductance of 0.67H.
Cylindrical capacitor to reduce lead inductance is
Introduced.

Second modulator for KEKB injector linac employs
Parallel connected PFN to overcome above problems.
n=16
C=0.015F
L=1.3H

N-parallel PFN is frequently used to achieve easy flat
top adjustment by variable inductance.
s ZC C L n
C L Z
T i i
i i
t
O
5 . 3 2 2
6
= = =
= =
s ZC C L n
Z Z Z C L Z
T i i
i i or
t
O O
5 . 5 2 2
7 . 4 // 4 . 9
2 1 2 1
= = =
= = = =
2012/2/1
45
Capacitor Bank Consideration

W
v
=
1
2
c
r
c
0
E
2
Energy Density of Capacitor
WStored Energy(J)
vArea of Dielectric Material(m
3
)
c
Dielectric constant
c
c of vacuum(8.85x10
-12
F/m)
EField Gradient(V/m)
No-Healing
Self-Healing(SH)
50~100V/m
~200V/m
Metal icon
Thin Film by Evaporation
A few of 100

Dielectric Film
Dielectric Film
Metal icon
Metal Thickness~7m
Dielectric Film
Dielectric MaterialCapacitor Thin Film(
r
~4.5), plastic Film(
r
=2.0~2.3)
2012/2/1
46
Capacitor
Down-sizing of capacitor
V =
2W
qcE
2
Case volume V of capacitor:
W : stored energy
c : dielectric constant
q : packing factor
E : field strength( ~ 60 V/m for no-healing capacitor)
Self-Healing Type Capacitor
Test Capacitor Breakdown Capacitor Element
ATF Modulator
0
50
100
150
200
250
300
10
5
10
6
10
7
10
8
10
9
10
10
SH
NH
Ref(n=20.8)
E
l
e
c
t
r
i
c

f
i
e
l
d

s
t
r
e
n
g
t
h
(
V
/
m
)
Lifetime(Shots)
L
L
0
=
V
V
0
|
\

|
.
|
|
n
Accelerated Life Test
Electric field strength vs Lifetime
50pps X 7,000H X 8 Years
Capacitor
Volume
1/4
Cross section of dielectric/electrode
Configulation(not to scale)
2012/2/1
47
Switching Device
Typical switching device for high power application
Spark-gap: Trigatron such as Ignitron and Crossatron
Thyratron: most popular high voltage switching device

2012/2/1
48
Switching Device
Important feature of thyratron
Thyratron is switching the high voltage pulse and current of
10kA peak is switched off. If it is not properly controlled, pulse
time jitter is seriously generated.
A keep-alive circuit or pre-trigger circuit is very important to
reduce the pulse jitter. Many thyratron has a hydrogen reservoir
in side and inner pressure of hydrogen is kept proper level.
Frequent watch and adjustment of reservoir voltage are very
important.
Ranging techniques to determine the proper operation voltage of
reservoir of the thyratron:
Change the reservoir voltage during the operation and observe
the upper limit in which continuous discharging start and the
lower limit in which thyratron quits operation, and then
operation voltage of reservoir is set in the middle.
2012/2/1
49
Modulator failure distribution
Total operation time : 6322 Hours
Machine failure time : 114 Hours

Operation Statistics in FY2007
Keep-alive current tuning
Thyratron failure
Air cooling fan

Modulator availability=0.997
Modulator failure time = 17.6 Hours
Linac failure distribution
53% of RF failure
Reliabilty of an RF system is directly linked to
the linac availability.
2012/2/1
50
Long term evaluation of the thyratron used
in KEK
Life distribution of thyratron : total=74
Period(September 1998 February 2008
Failure causes
Established using way
Acceptance test (all delivered are tested and only accepted ones are used
Replacement before life (at important section, every 2 years (-14,000hrs), new tubes are used
Reuse (replaced tubes mentioned above are stored as stand-by and set in not-so-important section
Prediction of the life (judged by the failure mode of keep-alive)
Maintenance in a yearreservoir voltage adjustmentcheck of jitter and pulse timing
Revised CX2411
Quality varies with year lots and company
Life is longer to 33.4 khrs year by year. Maintenance aims for min. cost and max. life operation.
0
2
4
6
8
10
0 10 20 30 40 50 60 70 80
CX2410K
F241
L4888B
N
u
m
b
e
r

o
f

T
u
b
e
s
Time(kHours)
Arcing
etc
Reservoir failure
G1 arcing
Failure of keep-
alive: 25%
2012/2/1
51
Status of Thyratron

Litton EEV ITT
L4888B CX2410K F-241
45 kV, 5 kA, 6 s, 50 Hz Switching

Operation period ( Sep. 1998 Feb.2008) : No. of Thyratrons=74
Lifetime Profile
Failure Modes Distribution
Thyratron Quality ?
0
2
4
6
8
10
0 10 20 30 40 50 60 70 80
CX2410K
F241
L4888B
N
u
m
b
e
r

o
f

T
u
b
e
s
Time(kHours)
Driver circuit
Keep alive has ~250 mA dc current at 100 V
Thyratron circuit
Keep alive Failure
G1 discharge
High voltage
Break Down
Reservoir Failure
Others
2012/2/1
52
Maintenance activity for Thyratrons

Acceptance Test
Break down rate < 0.05/ 1 hour at 100 hours operating
Switching Jitter < 10 ns
Check anode delay time

Exchange new thyratron in advance
Most important modulators ( such as Buncher section)
at intervals of two years(~14,000 hours) , Exchanged one is reused.

Checking and Tunning
Reservoir voltage and keep-alive current, switching jitter

Regular maintenance
Thyratron Ranging and exchange bad one at intervals of one year

Thyratron MTTF : 33,400 hours (as of Feb. 2008)
53
To maximum lifetime, and to minimize cost.
2012/2/1
53
Matching and mismatching
For lossless transmission line, current transformation is

And its inverse transformation,

Therefore, depending on , waveform is different after t=2o.
matching condition and no reflection
mismatch (positive mismatch)
extreme case :open end
lower operation point than normal
mismatch (negative mismatch)
extreme case :short end
(
+
+
=
+
+
=
+
=

o o
o
o
o
o
p p
p
g
p
p
g g
e
R Z
R Z
e
R Z
R Z
R Z p
e V
e
R Z
R Z
e
R Z p
V
p Z R p
V
p i
4 2
0
0 2
0
0
0
2
2
0
0
2
0 0
) ( 1
) (
) 1 (
1
1
) ( ) coth (
) (
| | | | | |
6 , 4 , 2 ), (
0 0 ) ( 0 1 ) (
) 6 ( ) 4 ( ) ( ) 4 ( ) 2 ( ) 4 ( ) 2 ( ) 2 ( 1 ) (
2
0
0
0
0
0
0
0
0
= =
< = > =
)
`
+
+
+
+

+
=
n n t t
t for t U and t for t U where
t U t U
R Z
R Z
t U t U
R Z
R Z
t U t U
R Z
R Z
t U
R Z
V
t i
o A
A A A A
o o o o o o o
R
0
Z R =
0
Z R <
0
Z R >
Protection Circuit for load discharge etc.

If discharge occurred in klystron or pulse transformer circuit, pulse
reflection occurs due to the mismatching effects. Serious case,
undesirable inverse voltage causes the failure of various devices.
Therefore, protection circuit is employed in the system and it force to
stop operation with a pulse or a few pulses later.
Fast protection: End of line clipper (within a pulse response)

Rather slow protection: reversed shunt
circuit
Average of inversed current due to the load
discharge is sensed and stop operation by
meter relay etc.
Charging Circuit is replaced by Inverter P/S
Charging Circuit can be replaced by
rather small inverter P/S.
It is possible to eliminate charging
reactor, and therefore modulator
becomes smaller.
Stability depends on the inverters
stability. De-Qing circuit is
eliminated.
Inverters failure rate is carefully
evaluated.
Charging is conducted linearly.
2012/2/1
56
Compact Modulator
(Charger is replace to Inverter P/S)
1.8
4.7
Compact Modulator
Present modulator
57
Start the development for C-band scheme of SuperKEKB

Decrease the modulator size to one-third that of the
existing modulator.

Switching power supply is essential to reduce modulator size.

Single unit for easy maintainability
Output voltage 50 kV(max.)
Output power 30 kJ/s
Voltage regulation 0.1%
Efficiency >80%
Power factor >85%
Input voltage 420 V, 3 Phase, 50 Hz, AC
Cooling Water 5 liters/min.
Size 19 rack mount
< 530mm(H), 480mm(W),
< 700mm(D)
Operation Single and Parallel operation
Specifications
2012/2/1
57
Discharging and Noise Problem
Discharging Circuit
True grounded position for
high power discharge current.
1 point earthing principle
Tri-axial cable is
desired
Multi-grounded with
filtering
2012/2/1
58
Discharging and Noise Problem (II)
Since pulse modulator produces a few microsecond high power pulse, it is a serious noise
generator. Especially the cabinet which contains thyratron, high voltage switching device,
generate large noises. Control signal (relay signal etc.) and analogue monitor cable from this
cabinet should be completely filtered from high
frequency noise.

For control signal for relay, signal is
come from discharge cabinet thru low
pass filter

Analogue signal is come from discharge
cabinet thru co-axial choke, which
reject common noise.

Discharging Cabinet
Which generate serious
noise
Charging Cabinet
DeQing cabinet
Control Cabinet
Filtering case
EMI mesh contactor
2012/2/1
59
Modulator other than Line-Type
2012/2/1
Application
60
Pulse Modulator Using Multi-Series Switch(1)
Series Switch
Series Switch Modularor
Storage Capacitor
PS
PS
Cell Modulator
Marx Type Modulator
Flexible
Active waveform control
Pulse waveform
Pulse flatness
2012/2/1
61
Bypass-Diodes(Freewheeling diodes)
Energy Storage Capacitors
17.3F
Cell-Modulator
IGBT
Mitsubishi CM1200-66H(1200A,3.3kV)
Resister Load
10
10-Stage Test Modulator
Max. 20 kV
Pulse Modulator Using Multi-Series
Switch(2)
2012/2/1
62
IGBT: CM1200HB-66H(1200A, 3.3kV)
Applied Voltage: 2 kV/stage, Repetition rate: 2 Hz
Storage Cap.=17.3 F(film), Resistor Load=10
20 kV, 2.0 kA Output Pulse Operation
2.8 s
-500
0
500
1000
1500
2000
0 1 2 3 4 5 6 7
O
u
t
p
u
t

C
u
r
r
e
n
t
(
A
)
Time(s)
With compensation
Without compensation
1st stage(for compensation)
3rd-10th stage(Main trigger)
2nd stage(for compensation)
Trigger Timing Chart
Active Waveform Control
IGBT: CM1200HA-66H x 5 + CM1200HB-66H x 5
Applied Voltage : 1 kV/stage, Repetition rate: 2 Hz
Storage Cap.=17.3 F, Resistor Load=4.1
Pulse Modulator Using Multi-Series Switch(3)
2012/2/1
63
Parameters K2-3 Unit
RF Peak Power 30- 60 [MW]
Pulse Voltage 280 - 450 [kV]
Pulse Current 230 - 450 [A]
Modulator Peak Power 160 [MW]
Modulator Average Power 0,5 - 100 [kW]
Mains: 1-phase / 3-phase 3
Cooling Water
RF POWER UP TO 60 MW MODULATOR PEAK
POWER UP TO 160 MW
Commercial Solid-state Amplifier
(Scandinova)
2012/2/1
64
Series Switch Modulator
(Diversified Technologies, Inc. )
IGBT Series Switch

140kV, 500A switch shown at left in
use at CPI

As a Phase II SBIR, DTI is building a
120 kV, 130 A version with a
bouncer to be delivered to SLAC at
the end of 2006
SNS High Voltage Converter
Modulator (Unit installed at SLAC)

RECTIFIER
TRANSFORMER
AND FILTERS
SCR
REGULATOR SWITCHING
BOOST
TRANS-
FORMER
HV RECTIFIER
AND FILTER
NETWORK
13.8KV
3
INPUT
LINE CHOKE
5
th
HARMONIC
TRAP
7
th
HARMONIC
TRAP
50mH
A
B C
3
(ON/OFF)
4mH
400A
4mH
400A
6 EACH
6 EACH
RTN
A B C
-HV -HV -HV 10ohm 20mH
.03uF
.03uF
.05uF
VMON
HV
OUTPUT
RECTIFIER
TRANSFORMER
AND FILTERS
SCR
REGULATOR
HVCM
EQUIPMENT
CONTROL RACK
ENERGY
STORAGE
Other
Alternative
Modulators
Longer Pulse Modulator with sag
compensation with RC circuit
2012/2/1
Application
67
Long Pulse Modulator with Bouncer Circuit(1)
CR
t
e V V

=
0
Klystron Impedance
R=120kV/140A
=857
C
t
Pulse width

C =
t
R D
r
t=1.7ms
Dr=1%
R=857
D
r
Droop
Secondary Capacitance
C=198F

E =
1
2
CV
k
2
E=1.4MJ
V
k
=120kV
If droop of 20% is allowed, then C
and E are 1/20
Save Capacitor bank and space
to1/20
24min run
of 1kW
heater
C=9.9F (Secondary)
C=1426F(Primary)
Switch
2012/2/1
68
10kV
-120kV
140A
1.7 ms
5Hz
1.7kA
2012/2/1
69
C
L

SCR(Need to Adjust
Protection SW for Safety
C Discharge SW for
Overvoltage

Characteristics
Simple and Compact
20% droop is compressed to1%
Easy Adjustment
Cheep Cost
Charging Diode
Specification
Voltage+-1kV Current
Invented by FNAL Quentine
Kerns
2012/2/1
70
Actual Waveform by Bouncer Circuit
Es=20kV, Pw=1.7ms, fr=5pps
Rise-time(10-90%)=33s
Es=17kV, Pw=1.7ms, fr=5pps
Flatness=0.8%(p-p)
With Bouncer Circuit
-100
-95
-90
-85
-2 -1.5 -1 -0.5 0 0.5
K
l
y
s
t
r
o
n

V
o
l
t
a
g
e
(
k
V
)
Time(ms)
Es=17kV
fr=5pps
0.3ms
0.7ms
0.52ms
0.4ms
0.6ms
-4000
-3000
-2000
-1000
0
1000
2000
3000
4000 -140
-120
-100
-80
-60
-40
-20
0
20
-3 -2 -1 0 1 2 3 4
B
o
u
n
c
e
r

V
o
l
t
a
g
e
(
V
)
/
C
u
r
r
e
n
t
(
A
)
K
l
y
s
t
r
o
n

V
o
l
t
a
g
e
(
k
V
)
Time(ms)
Klystron Voltage
Bouncer Voltage
Bouncer Current
2012/2/1
71
Protection of Klystron at Breakdown
Es=9.0 kV, Pw=1.7 ms, fr= 5 pps

W =100V I
k
}
dt
Energy deposit in klystron
from gun spark
W=2.0 J < design value
Arc voltage100 V
72 2012/2/1
72
Pulse Transformer Modulator Status
10 units have been built, 3 by FNAL
and 7 by industry (PPT with
components from ABB, FUG,
Poynting).
8 modulators are in operation.
10 years operation experience.
Working towards a more cost efficient
and compact design.
FNAL building two more, one each for
ILC and HINS programs SLAC has
bulit switching circuits.
HVPS and Pulse Forming Unit
IGCT Stack
Marx Modulator for ILC
2012/2/1
Application
74
2012/2/1
75
2012/2/1
Application
76
SLACs New Type Marx Modulator : P2
Vendor Design Marx Modulator for ILC
2012/2/1
77
Thomson Marx Modulator: Low voltage Marx and Pulse Transformers
DTIs Marx Modulator: 120kV Direct Pulse Generation
Marx Modulator's Feature
Good flat top pulse when pulse duration is
long such as a few millisecond.
( timing adjustment of individual Marx cell
enable to make a good flat top)
Even a direct output pulse case, acring
probability in the modulater is small due to
comprise of the low voltage cells assembly
Fast shutoff of the output when klystron fails.
2012/2/1
78
Thanks for listening my lecture

For RF source, PDS (power distribution system such as waveguide components)
are also important but this time completely omitted.

2012/2/1
79

4★★★★모듈레이터의 종류 - Fukuda - RF Sourced and Modulator (mainly for linac)

Загружено:

Сведения о документе

Исходное описание:

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

4★★★★모듈레이터의 종류 - Fukuda - RF Sourced and Modulator (mainly for linac)

Загружено:

Авторское право:

Доступные форматы

RF Sourced and Modulator

(mainly for Linac)

Protection Circuit for load discharge etc.

Вам также может понравиться