Upload
others
View
16
Download
0
Embed Size (px)
Citation preview
Mauricio Lopes – FNAL
2
Multipoles, Conformal Mapping,
Pole tip design
Multipole Magnet Nomenclature
• The dipole has two poles and field index n=1.
• The quadrupole has four poles and field index n=2.
• The sextupole has six poles and field index n=3.
• In general, the N-pole magnet has N poles and fieldindex n=N/2.
US Particle Accelerator School – Austin, TX – Winter 2016 2
Even Number of Poles
• Rotational periodicity does not allow odd number of poles.Suppose we consider a magnet with an odd number of poles.
• One example is a magnet with three poles spaced at 120degrees. The first pole is positive, the second is negative, the
third is positive and we return to the first pole which would
need to be negative to maintain the periodicity but is positive.
US Particle Accelerator School – Austin, TX – Winter 2016 3
Characterization of Error Fields
• Since satisfies LaPlace’s equation,
must also satisfy LaPlace’s equation.
• Fields of specific magnet types are characterized by the function
where the first term (N) is the “fundamental” and the remainder of the terms (n) represent the “error” fields.
∑=n
n zCF
∑≠
+=Nn
nn
NN zCzCF
nCzF =
US Particle Accelerator School – Austin, TX – Winter 2016 4
∑≠
+=1
1n
nndipole zCzCF
∑≠
+=2
22
n
nnquadrupole zCzCF
∑≠
+=3
33
n
nnsextupole zCzCF
∑≠
+=Nn
nn
NNNpole zCzCF
22
∑≠
=Nn
nn zCsErrorField
US Particle Accelerator School – Austin, TX – Winter 2016 5
Allowed Multipole Errors
US Particle Accelerator School – Austin, TX – Winter 2016 6
• The error multipoles can be divided among allowed orsystematic and random errors.
• The systematic errors are those inherent in the design andsubject to symmetry and polarity constraints.
• Symmetry constraints require the errors to repeat andchange polarities at angles spaced at π/N, where N is theindex of the fundamental field.
• In the figure, the poles are not symmetricalabout their respective centerlines. This is to
illustrate rotational symmetry of the N poles.
US Particle Accelerator School – Austin, TX – Winter 2016 7
• Requiring the function to repeat and change signs according to the symmetry requirements:
• USing the “polar” form of the function of the
complex variable:
( ) θπθ NpoleNpole FNF ∆−=
+∆
∑
∑
++
+=
=
+∆
+
Nni
NnzC
ezCN
F
n
n
Ninn
n
πθπθ
πθπθ
sincos
( ) ( )∑∑∑ +===∆ θθθ θ ninzCezCzCF nninnnnn sincos
US Particle Accelerator School – Austin, TX – Winter 2016 8
• In order to have alternating signs for the poles, the following two conditions must be satisfied.
• Rewriting;
( )θπθ nN
n coscos −≡
+ ( )θπθ nN
n sinsin −≡
+
( ) ( ) ( )
( ) ( ) ( )θπθπθπθ
θπθπθπθ
nN
nn
N
nn
Nn
nN
nn
N
nn
Nn
sinsincoscossinsin
cossinsincoscoscos
−≡+=
+
−≡−=
+
US Particle Accelerator School – Austin, TX – Winter 2016 9
• Therefore;
integers. odd all , 7, 5, 3, ,1 1cos
integers. all , 4, 3, 2, ,1 0sin
L
L
=⇒−=
=⇒=
N
n
N
nN
n
N
n
π
π
• The more restrictive condition is;
integers. all , 4, 3, 2, ,1 ere wh
12integers odd all
L=
+==
m
mN
n
• Rewriting;
( ) integers. all , ,4 ,3 ,2 ,1 where12 L=+= mmNnallowedUS Particle Accelerator School – Austin, TX – Winter 2016 10
Examples
US Particle Accelerator School – Austin, TX – Winter 2016 11
– For the dipole, N=1, the allowed errormultipoles are n=3, 5, 7, 9, 11, 13, 15, …
– For the quadrupole, N=2, the allowed errormultipoles are n=6, 10, 14, 18, 22, …
– For the sextupole, N=3, the allowed errormultipoles are n=9, 15, 21, 27, 33, 39, …
Magnet Field Uniformity
• In general, the two dimensional magnetfield quality can be improved by the amountof excess pole beyond the boundary of thegood field region.
• The amount of excess pole can be reduced,for the same required field quality, if oneoptimizes the pole by adding features(bumps) to the edge of the pole.
US Particle Accelerator School – Austin, TX – Winter 2016 12
• The relation between the field quality and "pole overhang" are summarized by simple equations for a window frame dipole magnet with fields below saturation.
US Particle Accelerator School – Austin, TX – Winter 2016 13
( )[ ]75.077.2exp100
1 −−=
∆x
B
B
dunoptimize
( )[ ]39.017.7exp100
1 −−=
∆x
B
B
optimized
gap
overhang" pole"
halfh
ax ==
90.0ln 36.0 −∆−=B
Bx dunoptimize
25.0ln 14.0 −∆−=B
Bxoptimized
These expressions are very important since they give general rules for the design
of window frame dipole designs. It will be seen later that these expressions can
also be applied to quadrupole and gradient magnets.
US Particle Accelerator School – Austin, TX – Winter 2016 14
• Graphically:
2.52.01.51.00.510 -6
10 -5
10 -4
10 -3
10 -2
10 -1
Optimized
Unoptimized
Dipole Field Quality as a Function of Pole Overhang
x = ah
∆ BB
US Particle Accelerator School – Austin, TX – Winter 2016 15
Introduction to conformal mapping
• This section introduces conformal mapping.
– Conformal mapping is used to extend the techniques of ensuring dipole field quality to quadrupole field quality.
– Conformal mapping can be used to analyze and/or optimize the quadrupole or sextupole pole contours in by using methods applied to dipole magnets.
• Conformal mapping maps one magnet geometry into another.
• This tool can be used to extend knowledge regarding onemagnet geometry into another magnet geometry.
US Particle Accelerator School – Austin, TX – Winter 2016 16
Mapping a Quadrupole into a Dipole
• The quadrupole pole can be described by ahyperbola;
US Particle Accelerator School – Austin, TX – Winter 2016 17
constant a2
==C
Vxy
Where V is the scalar potential and C is the coefficient of the
function, F, of a complex variable.
The expression for the hyperbola can
be rewritten;
2
2hxy =
We introduce the complex
function; ( )
h
ivx
h
zivuw
22 +==+=
h
yxwu
22
Re−==
hh
xywv === 2Im
h
xyi
h
yxivuw
222 +−=+=Rewriting;
since
2
2hxy =
ihh
yxw +−=
22
Therefore;
the equation of a dipole since the imaginary (vertical) component is a
constant, h.
US Particle Accelerator School – Austin, TX – Winter 2016 18
Mapping a Dipole into a Quadrupole
• In order to map the Dipole into the Quadrupole, we use the polar forms of the functions;
US Particle Accelerator School – Austin, TX – Winter 2016 19
φieww = θiezz =and
Since was used to convert the quadrupole into the dipole,
φiewhhwz ==2h
zw
2
=
θφ
ii ezewhz == 2 whz =2
φθ =therefore; and
Finally; 2coscos
φθ whzx ==
2sinsin
φθ whzy ==
Mapping a Dipole into a Sextupole
• In order to map the Dipole into the Sextupole, we use the polar forms of the functions;
US Particle Accelerator School – Austin, TX – Winter 2016 20
φieww = θiezz =and
Since was used to convert the quadrupole into the dipole,
φiewhwhz 223 ==
2
3
h
zw =
θφ
ii
ezewhz == 33 2 32 whz =
3
φθ =therefore; and
Finally; 3coscos 3 2
φθ whzx ==
3sinsin 3 2
φθ whzy ==
Mapping a Dipole into 2N-pole
• General formula to map a Dipole into a 2N-pole
US Particle Accelerator School – Austin, TX – Winter 2016 21
Nwhx N N
φcos1−=
Nwhy N N
φsin1−=
=
+=
+=
u
v
vuw
ivuw
arctan
22
φ
• u and v are the dipole coordinates
• x and y are the coordinates for the 2N-pole
Example: Ideal Dipole
US Particle Accelerator School – Austin, TX – Winter 2016 22
Ideal Dipole mapped in a Quadrupole
US Particle Accelerator School – Austin, TX – Winter 2016 23
Ideal Dipole mapped in a Sextupole
US Particle Accelerator School – Austin, TX – Winter 2016 24
Optimized Dipole
US Particle Accelerator School – Austin, TX – Winter 2016 25
Optimized Dipole mapped in a Quadrupole
US Particle Accelerator School – Austin, TX – Winter 2016 26
Optimized Dipole mapped in a Sextupole
US Particle Accelerator School – Austin, TX – Winter 2016 27
The gradient magnet
US Particle Accelerator School – Austin, TX – Winter 2016 28
xBB
hBxy
o
o
')(
+=
'B
BxX o+=
XB
hB
B
BXBB
hBXy o
oo
o
''
')( =
−+=
'2
2
B
hBHXY o==
'
22B
hBH o=
( )H
XYi
H
YX
H
iYX
H
Zivuw
2222 ++=+==+=
H
YXu
+=2
H
XYv
2=
Gradient into a Dipole Magnet
US Particle Accelerator School – Austin, TX – Winter 2016 29
'2
'2
2
B
hB
yB
Bx
uo
o +
+=
'
22
B
hBH
H
XYv o===
Dipole into a Gradient Magnet
( )uwHB
BxX o +=+=
2'( )uwHyY −==
2
Gradient into a Dipole Magnet
US Particle Accelerator School – Austin, TX – Winter 2016 30
Dipole into a Gradient Magnet
(not optimized)
US Particle Accelerator School – Austin, TX – Winter 2016 31
Dipole into a Gradient Magnet
(optimized)
US Particle Accelerator School – Austin, TX – Winter 2016 32
The Septum Quadrupole
US Particle Accelerator School – Austin, TX – Winter 2016 33
• In order to maximize the number of collisions andinteractions in a collider, the two beams must be tightly
focused as close to the interaction region as possible. At
these close locations where the final focus quadrupoles are
located, the two crossing beams are very close to each
other. Therefore, for the septum quadrupoles, it is not
possible to take advantage of the potential field quality
improvements provided by a generous pole overhang. It is
necessary to design a quadrupole by using knowledge
acquired about the performance of a good field quality
dipole. This dipole is the window frame magnet.
• The conformal map of the window frame dipole aperture and the centers of the separate conductors is illustrated.
• The conductor shape does not have to be mapped since the current acts as a point source at the conductor center.
US Particle Accelerator School – Austin, TX – Winter 2016 34
Quadrupole Field Quality
• The figure shows the pole contour of a quadrupole and its required good field region.
The pole cutoff, the point at which theunoptimized or optimized quadrupolehyperbolic pole is truncated, alsodetermines the potential field quality forthe two dimensional unsaturatedquadrupole magnet.
US Particle Accelerator School – Austin, TX – Winter 2016 35
• The location of this pole cutoff has design implications. It affects the saturation characteristics of the magnet since the iron at the edge of the quadrupole pole is the first part of the pole area to exhibit saturation effects as magnet excitation is increased. Also, it determines the width of the gap between adjacent poles and thus the width of the coil that can be installed (for a two piece quadrupole). The field quality advantages of a two piece quadrupole over a four piece quadrupole will be discussed in a later section.
US Particle Accelerator School – Austin, TX – Winter 2016 36
• Given; (uc, v
c) satisfying dipole uniformity
requirements.
• Find; (xc, y
c) satisfying the same requirements for
quadrupoles.
h
rr
h
zw regionfieldgood
20
2
=⇒=
US Particle Accelerator School – Austin, TX – Winter 2016 37
[ ]factor 90.0ln 36.0 dunoptimizehB
Bha dunoptimize −=
+∆−=
[ ]factor 25.0ln 14.0 optimizedhB
Bhaoptimized −=
+∆−=
[ ] h= and 2
02
0cc vfactorhh
ra
h
ru −=+=
For the Dipole;
Therefore;
US Particle Accelerator School – Austin, TX – Winter 2016 38
ihh
yxivuw +−=+=
22
h
rr
h
zw regionfieldgood
20
2
=⇒=
h
r00 =ρSubstituting a unitless (normalized) good field region,
2sinsin
2coscos
φθ
φθ
whzy
whzx
==
==and using the conformal
mapping expressions,
2
cos1
2sin
2
cos1
2cos
φφ
φφ
−=
+=and the half angle formula,
US Particle Accelerator School – Austin, TX – Winter 2016 39
[ ]( ) [ ]( )
[ ]( ) [ ]( )factorfactorh
y
factorfactorh
x
c
c
−−+−=
−++−=
20
220
20
220
2
11
2
1
2
11
2
1
ρρ
ρρwe get,
[ ] [ ] 12
1
24
1
2222
2
2
20
22
2
20
2222
+
−=
+
−=
+
=+
=
factorh
r
h
hfactor
h
r
h
v
h
u
h
vu
h
wccccc
and substituting,
US Particle Accelerator School – Austin, TX – Winter 2016 40
+∆−−+
+∆−=
+∆−++
+∆−=
90.0ln 36.02
1190.0ln 36.0
2
1
90.0ln 36.02
1190.0ln 36.0
2
1
20
2
20
dunoptimize
20
2
20
dunoptimize
B
B
B
B
h
y
B
B
B
B
h
x
c
c
ρρ
ρρ
+∆−−+
+∆−=
+∆−++
+∆−=
25.0ln 14.02
1125.0ln 14.0
2
1
25.0ln 14.02
1125.0ln 14.0
2
1
20
2
20
optimized
20
2
20
optimized
B
B
B
B
h
y
B
B
B
B
h
x
c
c
ρρ
ρρ
Substituting the appropriate factors for the unoptimized and optimized dipole
cases, we get finally for the quadrupoles;
US Particle Accelerator School – Austin, TX – Winter 2016 41
• The equations are graphed in a variety of formatsto summarize the information available in theexpressions. The expressions are graphed for boththe optimized and unoptimized pole to illustratethe advantages of pole edge shaping in order toenhance the field. The quality at various goodfield radii are computed since the beamtypicallyoccupies only a fraction of the aperture due torestrictions of the beampipe.
US Particle Accelerator School – Austin, TX – Winter 2016 42
2.01.51.01.010 -6
10 -5
10 -4
10 -3
10 -2
as a Function of Pole Cutoff
∆ BB
xch
Optimized pole
Unoptimized pole
ρ0 = 0.9
ρ0 = 0.9
ρ0 = 0.8
ρ0 = 0.8
ρ0 = 0.7
ρ0 = 0.7
ρ0 = 0.6ρ0 = 0.6
US Particle Accelerator School – Austin, TX – Winter 2016 43
h
r00 =ρ
10 -210 -310 -410 -510 -60.2
0.3
0.4
0.5
∆ BB
ych
ρ0 = 0.9
ρ0 = 0.9
ρ0 = 0.8
ρ0 = 0.8
ρ0 = 0.7
ρ0 = 0.7
ρ0 = 0.6
ρ0 = 0.6
Uno
ptim
ize
dO
ptim
ize
d
US Particle Accelerator School – Austin, TX – Winter 2016 44
h
r00 =ρ
10 -210 -310 -410 -510 -61.0
1.2
1.4
1.6
1.8
2.0
2.2
2.4
Ratio of Peak Field to Poletip Field
∆ BB
BcutoffBpole
ρ0 = 0.9
ρ0 = 0.9
ρ0 = 0.8
ρ0 = 0.8
ρ0 = 0.7
ρ0 = 0.7
ρ0 = 0.6
ρ0 = 0.6 Unoptimized
Optimized
US Particle Accelerator School – Austin, TX – Winter 2016 45
h
r00 =ρ
Typically, 1max >h
z
12
quadrupole
>>
=
h
z
h
w
dipole
1 3
sextupole
>>
=
h
z
h
w
dipole
and
Therefore,
in the mapped space.
However, there is a problem in the mapping of the
quadrupole and sextupole to the dipole space.
US Particle Accelerator School – Austin, TX – Winter 2016 46
• When mapping fromthe quadrupole or sextupole geometriesto the dipole space, the FEMcomputation is initially made inthe original geometry and a vector potential map is obtainedat some reference radius which includes the pole contour.
US Particle Accelerator School – Austin, TX – Winter 2016 47
The vector potential values are then
mapped into the dipole (w) space and
used as boundary values for the
problem.
( ) ( )θφ ,, refref rawA =
quadrupole
refref h
rw
=
2
sextupole
refref
h
rw
=
2
3
quadrupoleθφ 2=
sextupoleθφ 3=
US Particle Accelerator School – Austin, TX – Winter 2016 48
General guidelines for
Quadrupole/Sextupole Pole Optimization
US Particle Accelerator School – Austin, TX – Winter 2016 49
• It is far easier to visualize the required shape of pole edge bumpson a dipole rather than the bumps on a quadrupole or sextupolepole.
• It is also easier to evaluate the uniformity of a constant fieldfor a dipole rather than the uniformity of the linear orquadratic field distribution for a quadrupole or sextupole.
• Therefore, the pole contour is optimized in the dipole spaceand mapped back into the quadrupole or sextupole space.
• The process of pole optimization is similar to that of analysisin the dipole space.
– Choose a quadrupole pole width which will provide therequired field uniformity at the required pole radius.
• The pole cutoff for the quadrupole can beobtained from the graphs developed earlier using the
dipole pole arguments.
• The sextupole cutoff can be computed by conformalmapping the pole overhang from the dipole space
using
( )cc yx ,
3 2whz =
US Particle Accelerator School – Austin, TX – Winter 2016 50
• Select the theoretical ideal pole contour.
2
2hxy =
3323 hyyx =−
for the quadrupole.
for the sextupole.
• Select a practical coil geometry.
– Expressions for the required excitation and practicalcurrent densities will be developed in a later lecture.
• Select a yoke geometry that will not saturate.
• Run a FEM code in the quadrupole or sextupole space.
• From the solution, edit the vector potential values at a fixedreference radius.
US Particle Accelerator School – Austin, TX – Winter 2016 51
• Map the vector potentials, the good field region and thepole contour.
• Design the pole bump such that the field in the mappedgood field region satisfies the required uniformity.
US Particle Accelerator School – Austin, TX – Winter 2016 52
• Map the optimized dipole pole contour back into thequadrupole (or sextupole) space.
• Reanalyze using the FEM code.
US Particle Accelerator School – Austin, TX – Winter 2016 53
Closure
US Particle Accelerator School – Austin, TX – Winter 2016 54
• The function, zn, is important since it represents differentfield shapes. Moreover, by simple mathematics, thisfunction can be manipulated by taking a root or by takingit to a higher power. The mathematics of manipulationallows for the mapping of one magnet type to another,extending the knowledge of one magnet type to anothermagnet type.
• One can make a significant design effort optimizing onesimple magnet type (the dipole) to the optimization of amuch more difficult magnet type (the quadrupole andsextupole).
• The tools available in FEM codes can be exploited to verifythat the performance of the simple dipole can bereproduced in a higher order field.
Next…
US Particle Accelerator School – Austin, TX – Winter 2016 55
• Perturbations
• Magnet excitation