abstract - psma...abstract 2 simple equations of various loss components in inductive components...
TRANSCRIPT
A simple analytical method to calculate air gap induced eddy current losses in inductive components
Baumann, Grübl, Malcolm
Sumida-Europe Components, Germany
Abstract
2
Simple equations of various loss components in inductive components such as ohmic losses, core losses, proximity losses and skin effect losses are well known and are presented by literature. To calculate air gap induced losses in the winding, time-consuming computer simulation programs with the numerical finite element analysis are commonly used. On the other hand, the existing accurate analytical calculation methods of several authors are mostly not easy to handle for the user and have to be implemented in a computer design tool which doesn´t allow an understanding of what is really going on. We are proposing a simple analytical method which enables us to estimate the so-called gap losses which are proportional to the proximity losses. The model is based on a simple geometrical consideration which also takes into account the number of distributed air gaps. The derivated equations work in a closed form. First, the eddy current losses according to Butterworth are explained. Then the general stray field influence is modeled by using the fringing factor as a stray field cross section equivalent. Afterwards, the model is expanded to a location-dependent stray field which considers the influence of the distance of the winding to the air gap. The evaluation of the equations is done with the finite element analysis by using several examples. The comparison shows a satisfying correlation. In the end, the algorithm is simplified to a rule of thumb which enables first approximation of gap losses by hand. The rule can be used for a quick and simple double-check of the inductor layout, for double-checking computer calculation, or it can be used as a design rule.
Confidential © SUMIDA. All rights reserved.
In today‘s presentation
3
1. Current Situation 2. Fringing factor 3. Eddy current losses according to Butterworth 4. Model for gap losses 5. Determine gap losses with the help of proximity losses 6. Simplification 7. Evaluation via simulation 8. Limits and weakness 9. Rule of thump
Confidential © SUMIDA. All rights reserved.
Current Situation
4
• To calculate air gap induced losses in the winding, time-consuming numerical computer
simulation programs with the numerical finite element analysis are commonly used [1]
1. http://www.femm.info/wiki/HomePage 2. Albach M, Roßmanith H (2001 The influence of air gap size and winding position on the proximity losses in high frequency transformers. PESC, Vancouver, Canada, S 1485-1490 3. Jensen RA, Sullivan CR (2003) Optimal core dimensional ratios for minimizing winding loss in high-frequency gapped-inductor windings. APEC, Bd 2, S 1164–1169 4. Dowell PL (1966) Effects of eddy currents in transformer windings. PROC IEE, Bd 113, Nr 8, S 1387-1394 5. Albach M (2017) Induktivitäten in der Leistungselektronik. Springer Vieweg
)cos()cosh()sin()sinh(2)(
xxxxxxD
+−
⋅=
nhb
PP
w
w
prox
gap
43
≈
[5]
• A simple analytical equation which enables first approximation of gap losses by hand is to our knowledge still missing
• On the other hand existing accurate analytical calculation methods are not easy to handle for the user or have to be implemented in a computer design tool [2, 3]
• Simple equations of various loss components in inductive components such as Ohmic losses, core losses, proximity losses and skin effect losses are well known [4]
Confidential © SUMIDA. All rights reserved.
Definition of the fringing factor
5
The fringing factor F describes the increase of the effective magnetic cross section in the area of the air gap [1]:
L
et A
Aswith ⋅≈ 0: µ
ep AA /
Ae: effective cross section area AL: normalized inductivity Vp: fringing volume Vgap: volume of air gap
We have the integral information about the stray field volume Vw:
( )1−= FVV gapw
The fringing volume Vw is the key for the following model
1. Snelling EC (1988) Soft Ferrites – Properties and Applications.. Butterworths, second edition
gap
p
e
p
t
p
VV
AA
ss
F ===
wV
gapp VV /
tp ss /
pepw sAAV )( −=
Confidential © SUMIDA. All rights reserved.
Calculation of the fringing factor
6
Following is a simple equation based on ds=sp/2 from Mohan [1]:
Expanding to rectangular cross section with Uk as circumference of Ae [3] :
1. N. Mohan, T. M. Undeland, and W. P. Robbins -“Power Electronics –Converter, Applications, and Design”, John Wiley & Sons, Inc., 2003 2. McLyman WT (2004) Transformer and Inductor Design Handbook. Marcel Dekker, New York 3. Albach M (2017) Induktivitäten in der Leistungselektronik. Springer Vieweg
bw: length of winding space n: number of distributed air gaps
Ans
hbnshnsb
F p
kk
pkpk /5,21
)/)(/(+≈
++=
McLyman gives the most popular correlation with sp as air gap and Ae as magnetic cross section [2]:
bk: with of center leg hk: depth of center leg
1
7,0
3,07,0
3,01/
2ln2/
1−
−≈
+=
e
kwt
p
w
e
kp
AnUbs
nsb
AnUs
F
Ans
Anbsns
bA
nsF
t
e
wtp
w
e
p
5,21
1
2,11
1/
2ln/
1
7,0
3,07,0−
≈−
≈+= ps
Ae pA
Confidential © SUMIDA. All rights reserved.
ts
Eddy current losses according to Butterworth
7
The electron compensation due to the conductivity in longitudinal direction results in power losses [1,2] :
For a sinus similar stray field a voltage Ueff is induced [1,2]:
By integration dx we get the power losses [5,6]:
cu
d
cuprox
lbdBdxxlbBPρ
ωρ
ω24
ˆˆ 3222/
0
222
== ∫
The well-known equation for a round wire results with b=3π /16 ∙d. The flux density B² is perpendicular to the axis in longitudinal direction [1,2]:
cuprox
ldBPρ
πω128
ˆ 422
=
tB ωsinˆ
Flux density due to stray field of neighbored windings or air gap
b
l
AR I
AB
dx d
dtdNU ind
φ−=Excess of electrons
x
1. Snelling EC (1988) Soft Ferrites – Properties and Applications.. Butterworths, second edition 2. Butterworths, s.: 'Eddy-current losses in cylindrical conductors with special application to the alternating current resistance of short coils', Phil. Trans., [A], 1921, 222, p. 57
bdxlRwithdxlbxB
RU
dP cu
cu
effprox
ρρ
ω 2:ˆ 2222
===
22ˆ
2
ˆ xlBABNUdtdNU B
effindωωφ
==→−=
Confidential © SUMIDA. All rights reserved.
Proximity losses proportional to the winding volume
8
Fwcu
gprox ABfdl
pP ⋅⋅⋅⋅⋅= 222
0
2ˆ
8 ρπ
For a round wire diameter d0 we get the proximity losses specified as a function of B²AF:
π2
4:d
lAplwith wFG ⋅⋅⋅=
Fcu
gprox VBfd
pP ⋅⋅⋅⋅= 222
0
2ˆ
8 ρπ
cuprox
ldBPρ
πω128
ˆ 422
=
Substitution of the total length l of wire in the proximity losses with the help of the filling factor pG=Acu/AF :
Adapted to the volume of winding space VF we get the losses proportional to B²VF. This equation is only valid for a small resistance factor Fr<2 :
Confidential © SUMIDA. All rights reserved.
Induced losses in copper wire by the fringing field
Additional proximity loses due to the air gap
9
Additional proximity losses results from fringing flux density B(x,y)gap of the air gap:
( ) Fgapwirecu
gprox VyxByxBfd
pP
2220
2
),(),(8
+⋅=
ρπ
distribution of proximity losses Pprox=1,3mW; constant current [1]
1. Albach M (2017) Induktivitäten in der Leistungselektronik. Springer Vieweg
a) The classical proximity losses are high in outer winding space due to missing field compensation due to missing proximity of a wire
distribution of proximity losses Pprox=9,5mW including a air gap [1]
After solving the bracket we get an expression containing the scalar product:
b) The additional losses due to the air gap are high in the center of the component
Location of the wire field and gap field:
( ) Fgapwiregapwirecu
gprox VyxByxByxByxBfd
pP ⋅++⋅= α
ρπ cos),(ˆ),(ˆ2),(ˆ),(ˆ8
22220
2
α
2xB
2yB
Confidential © SUMIDA. All rights reserved.
Definition of the gap losses
10
( ) Fgapwirewcu
gprox VBBfdl
pP ⋅+⋅≈ 2222
0
2ˆˆ
8 ρπ
0),(ˆ),(ˆ: >>− gapwire yxByxBwith
As a result classical proximity losses not appear in the same position as the additional proximity losses due to air gap. This results that the scalar product can be neglected.
We get two field components which are independent from each other. Therefore the gap losses can be defined in first approximation as a separate loss component:
( )Fgapwcu
ggap VBfdl
pP 222
0
2ˆ
8 ρπ
≈
Confidential © SUMIDA. All rights reserved.
Location and direction of the wire field and gap field
Model for the gap losses
Generating a type of function for B(V) with the requirement B(V)V is constant according to a constant magnetic moment from a magnetic dipole [1].
Now we insert the initial condition B(0)=Bw and we get:
( )VV
VBVBw
ww
+=
ˆˆ
( )VB̂
wB̂
V
( ) wwVBVVB ˆˆ =( ) constVVB =⋅ˆ
In order to get simple equations, the basic idea is to solve in a bulk and not wire by wire. Now we take into account that the local B is not constant within the winding volume V.
( ) VVBfdp
Pcu
ggap
2220
2ˆ
8⋅⋅⋅=
ρπ
Fringing flux in the winding volume
Fringing flux volume equivalent
Das Bild kann zurzeit nicht angezeigt werden.
11 Confidential © SUMIDA. All rights reserved.
FB
AABBwith core
p
ecorew ==:
1. Kohlrausch (1996) Praktische Physik, Band2. B.G. Teubner Stuttgart. 24. Auflage
Fringing flux in dependence of winding volume
Model for the gap losses
12
Only regarding the volume of windings which are lying in the field B(V) we get via integration the product B²(V)V:
wFw
w
w
w
VVVVV
VVVk
σσ +≈
+
−+
=1
1 Vσ : volume of air gap upholstery VF: effective winding space σw: distance to the air gap Uk : circumference of Ae
We get a winding trans flux number 0<k<1:
( ) ( )
+
−+
⋅=
+==⋅ ∫∫
Fw
w
w
www
V
V w
wwV
V VVV
VVVVBdV
VVVBdVVBVVB
FF
σσσ
2
2
22 ˆˆˆˆ
( )1:
−=
≈
FVVUsVwith
gapw
kpwσσ
wσ
wV σV
σV
( )VB2ˆ
2ˆwB
FV
Confidential © SUMIDA. All rights reserved.
Now we get a simple equation for the gap losses including the volume of the air gap Vgap:
( ) kF
FVBfdp
P gapcorecu
ggap ⋅
−⋅⋅= 2
2220
2 1ˆ8 ρ
π
(only valid for low resistance factor Fr<2)
FBBwith core
w =:
Determine gap losses with the help of proximity losses
13
The middle flux density in the winding space is according [1] : As a result we get the B-Field in the winding space to:
;ˆˆ:
L
ecore
NAABIwith =
The fringing flux density Bw is the corresponding flux density equal to the cross section of fringing air gap Ap:
Simplification via setting both B-equations into the rescaled relation above: p
ecorew A
ABBˆˆ =
2
02ˆ
31ˆ
=
Lw
ecorewire Ab
ABB µ
22 2ˆ: BBwith =
The gap losses Pgap can be determined with the proximity losses Pprox by dividing both equations. This rescaling is useful because the accurarity of the Pgap improves according to the accuratiry of the Pprox.
1. Snelling EC (1988) Soft Ferrites – Properties and Applications.. Butterworths, second edition
kVAµbAV
PP
Fp
wLw
prox
gap22
0
223=
2
02
2
ˆ
31
=
wbINB µ
kVB
VBVBVB
PP
Fwire
ww
Fwire
Fgap
prox
gap2
2
2
2
ˆˆ
ˆˆ
== kF
FV
VBB
F
gapcoregap 2
22 )1(ˆ
ˆ −=
fringing flux volume equivalent
wwVB̂
the increase of effective cross section due to fringing
Ae pA
Confidential © SUMIDA. All rights reserved.
Simplification with substitution of some parameters
14
The effective air gap area is proportional to the fringing factor:
kVAµVbA
PP
Fp
wwL
prox
gap22
0
223=Regarding the volume ratio we get:
Fringing field volume Vw can be described by the effective air gap area Ap and the effective air gap sp : ( ) pepw sAAV −=
ep FAA =
The effective air gap sp can be described by fringing: tp Fss =
The air gap st without fringing is:
ksA
µA
Vb
PP
orkF
FVµ
bAPP
p
eL
F
w
prox
gap
F
wL
prox
gap
−=
−
=0
2
0
2 313
1
5,21−
−=
AnsF t
This results in:
With modificated fringing from McLyman we get the simplification:
Let AAs /0 ⋅= µ
11
)1(11
−−
−
⋅+=
+=
FAU
VVk
e
kw
w
σσ
klnh
AbPP
ww
ew
prox
gap 5,7≈
The winding trans flux number:
Confidential © SUMIDA. All rights reserved.
Simplification based on the area ratio via Mohan
15
Regarding the area relation we get:
Including the number of distributed air gap n we get : n
sds p
2=
The stray field window Aw with the extension ds and the length of gap sp is: pw sdsA ⋅=
The actual cross section of fringing air gap Ap: ep FAA =
The actual total air gap sp can be describe by the fringing factor: tp Fss =
This results in: knh
bPP
w
w
prox
gap
23
≈
The air gap st without fringing is: L
et A
As ⋅= 0µ
sp
Ap
AσAw
The stray field extension ds is according [1]: 2ps
ds =
pw sn
AAkwith σσ 21
1
1
1:+
=+
=
1. N. Mohan, T. M. Undeland, and W. P. Robbins -“Power Electronics –Converter, Applications, and Design”, John Wiley & Sons, Inc., 2003
kAAµbAA
PP
kVAµVbA
PP
Fp
wLw
prox
gap
Fp
wwL
prox
gap22
0
22
220
22 33→=
Confidential © SUMIDA. All rights reserved.
σwn=1
µi
bw
Ae
Af
hw
Simulation tools for evaluation at different conditions [1;2]
16
Different air gaps Different distance to air gap Different frequency
Number of air gaps Different depth Different window shapes 1. http://www.femm.info/wiki/HomePage 2. Martin Grübl, Sumdia Components & Modules GmbH Confidential © SUMIDA. All rights reserved.
2D simulation tools for evaluation wire losses at different air gaps [1;2]
17
Gap losses are in first approximation independent from the distance of the air gap (step function)
1. http://www.femm.info/wiki/HomePage 2. Martin Grübl, Sumdia Components & Modules GmbH
Component: Wire: Air gap: Curent: L-value:
JMAG: 0,0743 W Model: 0,0623 W FEMM : 0,0122 W
JMAG: 0,2489 W Model: 0,3169 W FEMM : 0,0769 W
E36 58Wdg, 1,25mm
1 mm 0,2 A, 100kHz
0,556 mH
E36 58Wdg, 1,25mm
2 mm 0,2 A, 100kHz
0,296 mH
E36 58Wdg, 1,25mm
4 mm 0,2 A, 100kHz
0.157 mH
JMAG: 0,3323 W Model: 0,3092 W FEMM : 0,1004 W
JMAG: 0,2996 W Model : 0,3091 W FEMM : 0,0936 W
E36 58Wdg, 1,25mm
0 mm 0,2 A, 100kHz
19,35mH
Confidential © SUMIDA. All rights reserved.
2D simulation for evaluation wire losses at different number of gaps [1;2]
18
Gap losses are inversely proportional to number of air gaps: Pgap~1/n
1. http://www.femm.info/wiki/HomePage 2. Martin Grübl, Sumdia Components & Modules GmbH
Component: Wire: Air gap: Curent: L-value:
JMAG: 0,2489 W Model: 0,3169 W FEMM : 0,0769 W
E36 58Wdg, 1,25mm
1x4 mm 0,2 A, 100kHz
0.157 mH
JMAG: 0,0757 W Model: 0,1168 W FEMM: 0,0211 W
E36 58Wdg, 1,25mm
4x1 mm 0,2 A, 100kHz
0.139 mH
JMAG: 0,1241 W Model: 0,1863 W FEMM : 0,0389 W
E36 58Wdg, 1,25mm
2x2 mm 0,2 A, 100kHz
0.147 mH
Confidential © SUMIDA. All rights reserved.
2D simulation tools for evaluation at different distance to gap [1;2]
19
This increasing distance the losses decrease with: Pgap~1/(1+2nσ/sp)
1. http://www.femm.info/wiki/HomePage 2. Martin Grübl, Sumdia Components & Modules GmbH
Component: Wire: Distance to gap: Air gap: Current: L-value:
JMAG: 0,2082 W Model: 0,2211 W FEMM: 0,0616 W
E36 58Wdg, 1,25mm
2mm 1 mm
0,2 A, 100kHz 0,556 mH
E36 58Wdg, 1,25mm
1mm 1 mm
0,2 A, 100kHz 0,556 mH
JMAG: 0,3323 W Model: 0,3092 W FEMM: 0,1004 W
Confidential © SUMIDA. All rights reserved.
2D simulation result of FEMM [1]
20 1. http://www.femm.info/wiki/HomePage 2. Dowell PL (1966) Effects of eddy currents in transformer windings. PROC IEE, Bd 113, Nr 8, S 1387-1394
y = 1,1948xR² = 0,4237
0
0,2
0,4
0,6
0,8
1
1,2
0 0,2 0,4 0,6 0,8 1
Mod
el (P
tota
l/W)
Simulation (Ptotal/W)
The proximity losses are calculated via Dowell [2]
y = 1,3188xR² = 0,507
1
1,1
1,2
1,3
1,4
1,5
1,6
1,7
1,8
1,9
2
1 1,05 1,1 1,15 1,2 1,25 1,3
F (M
cLym
an)
F (Simulation)
Worse correlation because 2D fringing does not correlate with reality (F-Lyman ≠ F-Simulation)
Confidential © SUMIDA. All rights reserved.
y = 1,0856xR² = 0,9467
0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1
0 0,2 0,4 0,6 0,8 1
Mod
el (P
tota
l/W)
Simulation (Ptotal/W)
2D simulation of “JMAG transformer module” with 3D correction [1]
21
y = 1,0661xR² = 0,904
1
1,1
1,2
1,3
1,4
1,5
1,6
1,7
1,8
1,9
2
1 1,1 1,2 1,3 1,4 1,5 1,6 1,7 1,8
F (M
cLym
an)
F (Simulation)
Better correlation because 3D correction improves fringing calculation (F-Lyman ≈ F-Simulation)
1. Manfred Wohlstreicher, Sumdia Components & Modules GmbH 2. Dowell PL (1966) Effects of eddy currents in transformer windings. PROC IEE, Bd 113, Nr 8, S 1387-1394
kF
FVµ
bAPP
F
wL
prox
gap
−
=13
0
2
−
⋅+=
)1(1/1
FAUk
e
kwσ
Confidential © SUMIDA. All rights reserved.
The proximity losses are calculated via Dowell [2]
Limits and weakness
22
• The overlay of wire field with the air gap fringing field is neglected • The different directions of the air gap field are not taken into account • The model was only tested on round wires
Confidential © SUMIDA. All rights reserved.
Rule of thumb
23
knh
bPP
w
w
prox
gap
23
≈
Pgap: gap losses Pgap: gap losses bw: length of winding bulk hw: height of winding bulk n: number of air gaps sp: distance of air gap σw: distance to the air gap
5,0/21
1: ≈+
=pw sn
kwithσ
y = 1,1046xR² = 0,8636
0
0,2
0,4
0,6
0,8
1
1,2
0 0,2 0,4 0,6 0,8 1
Mod
el (P
tota
l/W)
Simulation (Ptotal/W)
Confidential © SUMIDA. All rights reserved.
σwn=1
µi
bw
Ae
Af
hw
Thank you for your attention!