High Voltage Engineering

Enrique Gaxiola
Many thanks to the Electrical Power Systems Group, Eindhoven University of Technology, The Netherlands
& CERN AB-BT Group colleagues

Introductory examples
Theoretical foundation and numerical field
simulation methods
Generation of high voltages
Insulation and Breakdown
Measurement techniques
CAS on Small Accelerators

Introduction E.Gaxiola:
Studied Power Engineering
Ph.D. on Dielectric Breakdown in Insulating Gases;
Non-Uniform Fields and Space Charge Effects
Industry R&D on Plasma Physics / Gas Discharges
CERN Accelerators & Beam, Beam Transfer,
Kicker Innovations:
Electromagnetism
Beam impedance reduction
Vacuum high voltage breakdown in traveling wave
structures.
Pulsed power semiconductor applications
CAS on Small Accelerators

CERN Septa and Kicker examples

4 x MKQAV

Large Hadron Collider

14 TeV
Super Proton Synchrotron
450 GeV
Proton Synchrotron
26 GeV

4 x KKQAH

LHC
30 x MKD
12 x MKBV
8 x MKBH

4 x MKI
4 x MKP
4 x MKE

SPS
3 x MKDH

MKQH
5 x MKE

2 x MKDV

PS
PSComplex

MKQV

2 x TK

4 x MKI

9 x KFA71
4 x KFA45

3 x KFA79

4 x KSW
4 x EK

5 x BI.DIS

Reference [1]

Septum: E 12 MV/m
T = d.c.
l= 0.8 15m
Kicker: V=80kV
B = 0.1-0.3 T T = 10 ns - 200s
l=0.2 16m
RF cavities: High gradients,
E 150MV/m
CAS on Small Accelerators

PS septa SEH23

SPS septa ZS

Voltage: 300 kV

CAS on Small Accelerators

spacers

SPS injection kicker magnets

ferrites

beam gap

30 kV

magnets
CAS on Small Accelerators

 SPS extraction kickers

Courtesy: E2V Technologies

Generators
Pulse Forming Network
60 kV

Thyratron gas discharge switches

Power Semiconductor Diode stack

Magnets
30 kV

72 kV

CAS on Small Accelerators

 Maxwell equations for calculating

Electromagnetic fields, voltages, currents
Analytical
Numerical

CAS on Small Accelerators

Breakdown
Electrical
Fields,
Geometry
High fields
Field enhancement
Field steering

Insulation and
Breakdown

Medium
Gas
Liquids
Solids
Vacuum
Charges in fields
Ionisation
Breakdown
CAS on Small Accelerators

NUMERICAL FIELD SIMULATION METHODS

-CSM (Charge Simulation Method):

(Coulomb)

Electrode configuration is replaced by a set of discrete charges

- FDM (Finite Difference Method):
Laplace equation is discretised on a rectangular grid
- FEM (Finite Element Method):

Vector Fields (Opera, Tosca), Ansys, Ansoft

Potential distribution corresponds with minimum electric field energy

(w=E2)

- BEM (Boundary Element Method):

IES (Electro, Oersted)

Potential and its derivative in normal direction on boundary are sufficient

CAS on Small Accelerators

Procedure FEM
1. Generate mesh of triangles:
2. Calculate matrix coefficients: [S ]ij = ( i j ) A
3. Solve matrix equation:

[S

kf

U f
S kp = 0
U p

4. Determine equipotential lines and/or field lines

Procedure BEM
1. Generate elements along interfaces
2. Generate matrix coefficients:

ln ri
H ij =
ds , Gij = ln ri ds

n
Sj n
Sj
n

(H

3. Solve matrix equation:

j =1

4. Determine potential on abritary position:

1
U ( x0 , y0 ) =
2

ij

n
n

ln
r

U j
ds Q j ln rds
j =1 S n

j =1
S

ij )U j = Gij Q j
j =1

CAS on Small Accelerators

Generation of High Voltages

AC Sources (50/60 Hz)
High voltage transformer
Resonance source

(one coil; divided coils; cascade)

(series; parallel)

DC Sources
Rectifier circuits
Electrostatic generator

(single stage; cascade)

(van de Graaff generator)

Pulse sources
Pulse circuits (single stage; cascade; pulse transformer)
Traveling wave generators (PFL; PFN; transmission line transformer)

CAS on Small Accelerators

Cascaded High voltage transformer

Courtesy: Delft Univ.of Techn.

900 kV
400 A

1: primary coil
2: secondary coil
3: tertiary coil

CAS on Small Accelerators

Resonance Source
L

cap.
deler

L
L

C test

Courtesy: Eindhoven Univ.of Techn.

Equivalent
Circuit:

+ Waveform: almost perfect sinusoidal

C

+ Power: 1/Q of normal transformer

+ Short circuit: Q0 results in V0

H ( ) =
0 =

1
LC

Q 0
1 + Q ( 0 0 )
2

and Q =

1 L
R C

- No resistive load
2

900 kV
100 mA
CAS on Small Accelerators

1`

D1
C1

(Greinacher; Cockcroft - Walton)

Courtesy: Delft Univ.of Techn.

C1` D `
1
2`

VDC

Cascaded Rectifier

D2

stage 1

C2
C2` D `
2

Dn-1

stage 2

Cn-1

Cn-1` D `
n-1
n`

Dn

stage n-1

Cn
Cn` D `
n
amplitude: Vmax

VDC = 2nVmax
stage n

Voltage: 2 MV

Reduce V (~n2) and V (~n3) by:

larger Cs (more energy in cascade)
higher f (up to tens of kilohertz)
CAS on Small Accelerators

Single-Stage Pulse Source

V0

G
R1
dU
dU
+
+ bU = 0
a
2
dt
dt

U(t)

C1 a =

1
1
1
1
+
+
;b =
R1C1 R2C2 R1C2
R1C1 R2C2

R2

C2

60 kV
1 kA

spark gaps

2 (discharge)
1 (front)
LC-oscillations
t

Courtesy: Eindhoven Univ.of Techn.

U (t ) = V0 e t / 2 e t / 1

if C1 >> C2 and R2 >> R1

rise time:
1=R1C2
discharge time: 2=R2C1

Standard lightning surge pulse: 1.2 / 50 s

CAS on Small Accelerators

Vpulse
R2`

Bn
R`

B2
C 1`

G1

C1`

R1`

C``

R2`

R`

C1`

(Marx Generator)

V pulse = n VDC

R`

C1`

VDC
R1: front resistor
R2: discharge resistor

R1`
R2`

stage 1 R`

VDC

R1`

900 kV
100 mA

R2`

stage 2

A1

C 1`

R2`

stage 3

R2`

B1
R`

R2`
A2

G2
C`

R`

R`

R1``

A3
C1`

B3

R2`

stage n

C 1`
G3

R`

Cascade Pulse Source

An

Gn

C1`

R1`

Total discharge capacity: 1/C1=1/C1

Front resistance: R1=R1+R1
Discharge resistance: R2=R2

Courtesy: Kema,
Netherlands
CAS onThe
Small
Accelerators

Voltage: 300 kV

Pulse Source with Transformer

resistivity neglected

d 2 I1
d 2I2
primary : L1C1 2 MC1 2 + I1 = 0 U1
dt
dt
d 2I2
d 2 I1
secondary : L2C2 2 MC2 2 + I 2 = 0 C1
dt
dt

R1

U1 (t ) =

U2

I2

I1

C2

R2

d1
dt

U 2 (t ) =

d 2
dt

Eigen frequencies from characteristic equation:

L1C1 MC1 i1 1 i1
1
1

= 2 L1C1 2 L2C2 2 M 2C1C2 = 0

MC2 L2C2 i2 i2
Approximation: transformer almost ideal: k=M/(L1L2)1
1

1
=
L1C1 + L2C2

1
L1 (C1 + C2' )

slow oscillation

1
1 k 2

1
1
+
=
L1C1 L2C2

1
Leq (C1 // C2' )

fast oscillation
CAS on Small Accelerators

Pulse Forming Line / Network

charge cable

load cable

amplitude: VDC
duration: 2L/v

VDC

80 kV, 10 kA, T=20ns - 10s

Transmission Line Transformer
S
VDC

cables
parallel

cables in
series

150 kV
1 kA

Courtesy: Eindhoven Univ.of Techn.

CAS on Small Accelerators

Insulation and Breakdown

In Gases

Ionisation and Avalanche Formation

Townsend and Streamer Breakdown
Paschen Law: Gas Type
Breakdown Along Insulator
Inhomogeneous Fields, Pulsed Voltages, Corona

Insulating Liquids
Solid Insulation

Breakdown types, Surface tracking, Partial discharges, Polarisation, tan

Vacuum Insulation
Applications, Breakdown, Cathode Triple-Point, Insulator Surface Charging, Conditioning

CAS on Small Accelerators

400kV

800kV South Africa

400kV Geertruidenberg,
The Netherlands

CAS on Small Accelerators

1st

free electron

Cosmic radiation
Shortwave UV
Radio active isotopes

In air:
2.5 x 1019 molecules/cm3
1000 ions/cm3
10 electrons/cm3

Free path, effective cross-section

Townsends 1st ionisation coefficient

One electron creates new electrons per unit length
A + e A+ + 2e

ne(x=d) = n0ed
/p = f (E/P)
CAS on Small Accelerators

- vs. E/p

Electro-negative gasses
Attachment of electrons to ions

1.) air
2.) SF6

electrons: ne(x=d) = n0e( -)d

negative ions:
no ( ) d
n ( x = d ) =
[e
1]

Avalanche Breakdown;
creation of secondaries
Townsends 2nd ionisation coefficient

Reference [3]

one ion or photon creates new electrons

at cathode
ne = n0(ed-1)

E
1st
electron

Breakdown if:

# secondary electrons n0
d ln(1/ + 1)

steep function of E/p ed very steep (E/p)critical and Vd

well defined of weak influence

New 1st
electron

+
Avalanche
+ +
ion
r
o
n
o
+ +

Phot

Secondary

Paschen law / breakdown field

Townsend breakdown criterion d = K:
Ed
B
=
p ln(Apd/K)

Vd =

with

Bpd
ln(Apd/K)

Ed and Vd depend only on p*d

1: SF6
2: air
3: H2
4: Ne

p: pressure

A=I/kT
B=ViI/kT

d: gap length

Typically practically
Ebd= 10 kV/cm
at 1 bar in air
Reference [2]

Vbd,Paschen min, air 300 V

Small p*d, d<< : few collisions, high field required for ionisation
Large p*d, d>> : collision dominated, small energy build-up, high Vd
CAS on Small Accelerators

E0

E0

E0 + E

Streamer breakdown
Space charge field E E0
Field enhancement
extra ionising collisions ()
High excitation UV photons
when 1 electron grows into ca. 108
then
E large enough for streamer
breakdown (ne 2108 in avalanche head)

Result:
Secondary avalanches, directional effect (channel formation)
Grows out into a breakdown within 1 gap crossing (anode and/or cathode directed)

Characteristic:
Very fast
Independent of electrodes (no need for electrode surface secondaries)
Important at large distances (lightning)

CAS on Small Accelerators

Townsend, unless:

Townsend: d ln(1/ + 1) 7...9

( 10-4...10-3)
Streamer: d 18...20

Strong non-uniform field

(small electrodes, few secondary electrons)
Pulsed voltages
Townsend
Streamer

slow, ion drift, subsequent gap transitions

fast, photons, 1 gap transition

High pressure
Less diffusion
E high
photons absorbed
in front of cathode
positive ions
slower

Courtesy: Eindhoven Univ.of Techn.

CAS on Small
Laser-induced streamer breakdown
inAccelerators
air.

Breakdown along insulator

Surface charge
(Non-regular) surface conduction
Particles / contaminations on surface
Non-regularities (scratches, ridges)

Field enhancement
Increased breakdown probability
Reference [2]

Courtesy: Eindhoven Univ.of Techn.

Prebreakdown along insulatior in air.

CAS on Small Accelerators

Breakdown at pulse voltages;

time-lag

Reference [2]

ts, wait for first elektron

tf, breakdown formation
Townsend or Streamer

Short pulses, high breakdown voltage

CAS on Small Accelerators

Non-uniform fields; Corona

Breakdown conditions:
Global Full breakdown
Local Streamer breakdown
Partial discharge

Reference [2]

Non-uniform field:
Discharge starts in high field
region
....and extinguishes in low
field region
CAS on Small Accelerators

Corona
10 ns

- Power loss; EM noise

- Chemical corrosion
+ Useful applications

Pulsed corona discharges:

20 ns

30 ns

Courtesy: Eindhoven Univ.of Techn.

Fast, short duration HV pulses

Many streamers, high density
Generation of electrons, radicals,
excited molecules, UV
E.g. Flue gas cleaning

40 ns

Transmission line transformer

CAS on Small Accelerators

Solid insulation
Breakdown field strength:
Very clean (lab): high
Practical: lower due to imperfections

Voids
Absorbed water
Contaminations
Structural deformations

Anorganic

Quartz,mica,glas

Natural

Porcelain

Synthetic

Paper

Al O
2 3

Polymerisation

HD,LD,XL PE
Teflon
Polystyrene, PVC,
polypropene,etc

Epoxy

Spacer

Capacitor

Polyethelene
Organic

Feedthrough

Cable

+ Oil

Synthetic

Disc insulator

Hardener
Filler

Spec. properties:

Requirements:

Mechanical strength
Contact with electrodes and
semiconducting layers
Resistant to high T, UV, dirt,
contamination, rain, ice,
desert sand

Problems:

Surface tracking
Partial discharges

In voids
(in material or at electrodes,
often created at production).

Moisture content
high T
losses
bonding

Moulding in mold

Types of solid
insulation materials

Surface tracking
Remedies:

Geometry
Creeping distance (IEC-norm)
Surface treatment (emaillate, coating)
Washing
CAS on Small Accelerators

Vacuum insulation
Applications:

Advantages:

What is vacuum?

Disadvantages:

Vacuum circuit breaker

Cathode Ray Tubes / accelerators
Elektron microscope
X-ray tube
Transceiver tube

Pressure at which no collisions

for Brownian temperature
movements of electrons
>> characteristic distances
E.g. p = 10-6 bar, = 400 m

Self healing
No dielectric losses
High breakdown fieldstrength
Non flammable
Non toxic, non contaminating

Requires hermetic containment

and mechanical support
Quality determind by:
electrodes and insulators
Material choice, machining
Contaminations, conditioning

Characteristics of vacuum breakdown

No 1st electron from gas
Cathode emission
primary: photoemission, thermic emission, field emission, Schottkyemission
secondary: e.g. e- bombarded anode +ion collides at cathode e-

No breakdown medium
No multiplication through collision ionisation
Medium in which the breakdown occurs has to be created
(evaporated from electrodes, insulators)

Important: prevent field emission

Keep field at cathode and cathode triple point as low as
possible
Insulator surface charging, conditioning
CAS on Small Accelerators

14

Breakdown vs.field cathode-triple-point

Courtesy: ESTEC / ESA

Insulator surface
charging

Conditioning
effect lost
when charge
compensated

01

Conditioning
# Conditioning
breakdowns needed
to reach 50kV hold
voltage
Ref [12]

CAS on Small Accelerators

Insulating liquids
Requirements:
Applications: Transformers
Cables
Capacitors
Bushings

Pure, dry and free of gases

r (high for Cs, low for trafo)
(demi water r,d.c. = 80)
Stable (T), non-flammable,
non toxic (pcbs), ageing,
viscosity

No interface
problems
Combined
cooling/insulation
Cheap (no mould)
Liquid tight housing
Courtesy: Sandia labs, U.S.A.

Breakdown fieldstrength:

Very clean (lab): high 1 - 4 MV/cm (In practice much lower)

Important at production:outgassing, filtering, drying
Mineral oil (old time application, cheap, flammable)
Synthetic oil (purer, specifically made, more expensive)
Silicon oil (very stable up to high T, non-toxic, expensive)

Liquid H2, N2, Ar, He (supra-conductors)

Demi-water (incidental applications, pulsed power)
Limitation Vbd:
Inclusions: Partial discharges Oil decomposition Breakdown
Growth (pressure increase)
extension in field direction

Particles drift to region with highest E bridge formation breakdown

+

+
+

CAS on Small Accelerators

Transformer:
Mineral oil: Insulation and cooling
Paper: Barrier for charge carriers and chain formation
Mechanical strength

Ageing

Thermical and electrical (partial discharges)

Lifetime: 30 years, strongly dependent on temperature, short-circuits, overloading , over-voltages
Breakage of oil moleculs, Creation of gasses, Concentration of various gas
components indication for exceeded temperature
(as specified in IEC599)

Lifetime
Time in which paper looses 50 % of its mechanical strenght
Strongly dependent on:
Moisture (from 0.2 % to 2 % accelerated ageing factor 20)
Oxygen (presence accelerates ageing by a factor 2)

Measurement techniques
Partial discharges
UV, fast electrons, ions, heat
Deterioration void:

Oxidation, degradation through ion-impact

Pitting, followed by treeing

Eventually breakdown

Acceptable lifetime? Preferably no partial

discharges.

High sensitivity measurements on often large

objects
Qapp Qreal, still useful, because measure for
dissipated energy, thereby for induced damage
relative measurement

AC voltage phase resolved

discharge pattern detection Type of defect
CAS on Small Accelerators

Partial discharges
Before

V
a

c
Qtot = Q

a >> b << c

So:

c V = a V + b (V - V) + c V
V = V b/(a+b) V b/a

V- V
b

Q = aV + b (V - V) + c V
Q - c V = (a + b)(V - V)

After

V
b

Before:
After:

V-V

a
V

c
Qtot = Q - cV

V-V

Apparent charge:
Ctot V = (a + bc/(b+c)) V
(a+b) V aV
Qapp
Qreal

Rdamp

it

Test object

aV bV b r .d void
=
=
cV cV c
d insul

High sensitivity measurement because b/c << 1

b
AC
source

a
Cc

Zm, often RLC

Measure with resonant RLC circuit:

Excitation by short pulse it

Qapp gives it (Qapp = itdt)
No 50 Hz problem
V = q/C exp(-t) { cost - / sint }
Cc gives it if Cc>>Cobject
=1/(2RC) =[1/(LC) - 2]-1/2
Calibration through injecting known charge

Loss angle, tan()

Sources:

CHV

Conduction (for DC or LF)

Partial discharges
Polarisation
jCV

Schering bridge:
i=0, RC=R4C4
Gives: tan()
parallel:
serie:

1/RC
RC

test
object

i
R4

1/R.V

Tan :

Shielding

Bulk parameter
No difference between phases

PD:
Detection of weakest spot
Largest activity and asymmetry
in blue phase (ridge discharges)

C4

R3

Summary
Seen many basic high voltage engineering technology
aspects here:
High voltage generation
Field calculations
Discharge phenomena

The above to be applied in your practical accelerator

environments as needed:
Vacuum feed through: Triple points
Breakdown field strength in air 10kV/cm
Challenging calculations for real practical geometries.
CAS on Small Accelerators

References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]

M.Benedikt, P.Collier, V.Mertens, J.Poole, K.Schindl (Eds.), LHC Design Report, Vol. III, The LHC
Injector Chain CERN-2004-003, 15 December 2004.
E. Kuffel, W.S. Zaengl, J. Kuffel: High Voltage Engineering: Fundamentals, second edition, ButterworthHeinemann, 2000.
A.J. Schwab, Hochspannungsmesstechnik, Zweite Auflage, Springer, 1981.
L.L. Alston: High-Voltage Technology, Oxford University Press, 1968.
K.J. Binns, P.J. Lawrenson, C.W. Trowbridge: The Analytical and Numerical Solution of Electric and
Magnetic Fields, Wiley, 1992.
R.P. Feynman, R.B. Leighton, M. Sands: The Feynman Lectures on Physics, Addison-Wesley Publishing
Company, 1977.
E. Kreyszig: Advanced Engineering Mathematics, Wiley, 1979.
L.V. Bewley: Two-dimensional Fields in Electrical Engineering, Dover, 1963.
Energy Information Administration: http://www.eia.doe.gov.
R.F. Harrington, Field Computation by Moment Methods, The Macmillan Company, New York, 1968,
pp.1 -35.
P.P. Silvester, R.L. Ferrari: Finite Elements for Electrical Engineers, Cambridge University Press, 1983.
J. Wetzer et al., Final Report of the Study on Optimization of Insulators for Bridged Vacuum Gaps,
EHC/PW/PW/RAP93027, Rider to ESTEC Contract 7186/87/NL/JG(SC), 1993.

CAS on Small Accelerators

Appendix I

Maxwell equations in integral form

D dA = dV = Q

Electrostatic
D dA = Qomsl.

(1)

E dl =

E dl = 0

(2)

Magnetostatic
B dA = 0

(3)

H dl = I

(4)

omsl .

d omsl.
d

B
dA
dt
dt

B dA = 0
H dl = ( J +

D
) dA
t

omsl .

Maxwell equations in differential form

D =
E =

B
t

B = 0

H = J +

D
t

Electrostatic
D =

No space charge
D = 0

(5)

E = 0

E = 0

(6)

Magnetostatic
B = 0

In area without source

B = 0

(7)

H = J

H = 0

(8)
CAS on Small Accelerators

Appendix III Finite

Element Method (FEM)

Field energy minimal inside each closed region G:

2

W = E dV = U dV
1
2

1
2

Assume U satisfies Laplace equation, but U' does not, then

WU' - WU 0:

WU ' WU = 12 U ' U dV = L = 12 U 'U dV 0

2

Field energy for one element (2-dim)

Potential is linear inside element: U = a + b x + c y = (1 x

on corners:

U1 1 x1

U 2 = 1 x2
U 1 x
3
3

y1 a

y2 b
y3 c

Potential can be written as:

U = (1 x

1 x1

y ) 1 x 2
1 x

y1

y2
y3

U1
3
U 2 = U i i ( x, y )
i =1
U 3

y
a

y ) b
c

1
2

e
3
x

Field energy in element (e):

[ ]

W ( e ) = 12 U T S(e) U with Sij( e ) = ( i j )dxdy

CAS on Small Accelerators
s are linear in x and y is constant: Sij=(ij) A(e)

All elements together:

Total field energy of n elements

U T = (U1 L U m U m +1 L U n ) (U f U p )
free

prescribed

free: potential values to be determined

prescribed: potential according to boundary conditions

W = U [S ]U = U
1
2

1
2

Partial derivatives of W to Uk are zero for 1km (m equations):

W
=0
U k

[S

kf

S kp

T
f'

T
p'

S f ' f
S
p' f

S f ' p U f
S p ' p U p

U f
U = 0
p

Boundary Element Method (BEM)

Boundaries uniquely prescribe potential distribution

Laplace equation : u = 0
border 1 : u ( x , y )

border 2 : q ( x , y )

(Dirichlet )
u
n

r = ( x x0 ) 2 + ( y y0 ) 2
6

P (x0,y0)
) c

P
r

(Neumann)

(x,y)

ln r
1

P0=(x0,y0) inside : u ( x0 , y0 ) =

q
x
y
r
u
(
x
,
y
)
(
,
)
ln

ds

n
2

(0,0)

Problem: u(x,y) and q(x,y) not both known at the same time
CAS on Small Accelerators

 (uv vu ) nds = (uv vu )dxdy

Green II (2-dim):

v(x,y)=0 for PP0(x0,y0)

Choose v(x,y)=ln(1/r)

Exclude region around P0 by means of circle c

ln r 1
1 u
1
1
+c (u n ln r n )ds = (u ln r ln r u )dxdy = 0
ln r 1
u
u
ln r
u
ln r ds + u
ln r 1 ds = 0
n
n
n
n
c

lim 2 1 u

+
u
ln

d = 2 u ( x0 , y0 )
0 0 n

u ( xi , yi ) =

Pi=(xi,yi) on border :
Discretisation:

(u ( x, y)

U ( xi , yi ) = U j
j =1

In matrix notation:

(H
n

j =1

ij )U j = Gij Q j
n

ij

j =1

ln r
q ( x, y ) ln r )ds
n

ln rij

Sj

ds Q j ln rij ds
j =1

Sj

ij

ij

2
1

n node

element

6
j+1
j

Generates missing information

CAS on Small Accelerators