Professional Documents
Culture Documents
Abstract— PWM inverters are widely used in different power integrated transformer and inductor based on leakage
electronics applications. PWM waveforms have a harmonic inductance is presented. In this solution, only a gapped leg is
content which should be filtered, typically based on LC needed (less radiated EMI), and no winding is placed around
topologies. If a transformer is required, the leakage inductance
its (less copper losses due to proximity effect caused by
can be used as a series inductance for the LC filter.
This work proposes an easy procedure for the integration of fringing flux).
the series inductance for harmonic filtering inside the
transformer, while the value of the leakage inductance can be II. BASIC APPROACH
determined by design. The filter inductor L is series connected with the
Index Terms—integrated magnetics, inverters
transformer, which means that L and the leakage inductance of
the transformer are series connected. The natural leakage
I. INTRODUCTION
inductance of the transformer usually has a very low value
Inverters in power electronics usually work with PWM respect to that filter inductor and can be neglected for filtering
waveforms which have a high harmonic content. This purposes. However, if this value could be increased, the use of
harmonic content should be filtered due its annoying effects an additional inductor could be avoided. There are 3 major
(losses increasing, load malfunction, EMI noise, etc). In single questions when using the leakage inductance as the filter
stage solutions, low frequency approach is required. An LC inductance:
filter implementation is shown in figure 1a as an example of
this application. • The inductance value must be high enough.
Figure 1a solution can be improved if a magnetic • The inductance value must be easy to design. The leakage
integration is carried out (figure 1b), in order to decrease both inductance can be obtained using advanced simulation
the overall volume and weight. techniques such as Finite Element Analysis, but these
techniques are not available for every designer.
Additional series inductor
Transformer • The inductance value should be repetitive in a
C
manufacturing process, so it should be controlled by design.
INVERTER + + +
VPWM VSIN VOUT
- - -
The proposed integrated magnetic component overcomes
PWM waveform Sinusoidal waveform these requirements. The presented work is focused on a 3-leg
(a) core shape. Figure 2 shows an example of the proposed
Integrated transformer and series inductor
integrated magnetic component, while figure 3 represents its
equivalent reluctance model.
C
The key point is that the magnetic flux Ɏ3 implies the
INVERTER + +
VPWM VOUT existence of a leakage inductance which value can be
- -
determined by design, because its magnetic path is closed
PWM waveform Sinusoidal waveform trough the magnetic core. Primary winding is divided in two
(b) sub-windings (N11 and N12): one of them (N12) is in the same
Figure 1 Inverter coupled with a transformer through an LC filter
leg that the whole secondary winding (N2) and the other
subwinding (N11) is placed in an ungapped independent leg.
Magnetic integration has been widely used in the last years Considering that the whole flux is confined within the core,
in power electronic applications, principally DC/DC there is no significant leakage inductance in the secondary
converters. The size, weight and cost of ferromagnetic cores winding N2, because the whole secondary flux Φ 2 is shared
are considerable. The number of these elements, needed in the with the N12 primary winding. However, there is a part of the
converter, can be reduced using this technique. Thus, it is primary flux (Φ 3) that is not shared with the secondary
useful to design coupled inductors, integrated transformers, winding, and therefore there is a significant leakage
and integrated inductor and transformer [1-4]. inductance in the primary winding. The magnitude of Φ 3 can
In this work, an easy and accurate design process of the be controlled by means of the air gap thickness.
1202
TABLE 1. COMPARISON BETWEEN TRANSFORMATION RATIO OBTAINED FROM
FEA SIMULATIONS AND CALCULATION RESULTS
wc Gap Transformation ratio
(µm) Calculated FEA Error %
la 0 0,667 0,667 0,09
200 0,532 0,532 0,05
400 0,519 0,519 0,08
cc 600 0,514 0,514 0,10
Figure 5 Air gap parameters TABLE 2. COMPARISON BETWEEN LEAKAGE INDUCTANCE VALUES OBTAINED
FROM SIMULATIONS AND CALCULATION RESULTS
If the permeability of the magnetic core is high enough, the Primary leakage Secondary leakage Ll 2
expressions of the electrical parameters of the integrated inductance, Ll1 (mH) inductance, Ll 2 (mH) Ll1 %
magnetic (figure 4) component became simpler. If Gap Calcu- FEA Error FEA
(leff 3 +leff 1 ) µr <<lgap , expressions (7), (8) and (9) can be written
(µm) lated %
0 0,902 0,920 1,09 0,002 0,03
200 0,217 0,223 2,55 0,002 1,02
as follows:
400 0,131 0,138 5,16 0,002 1,42
N2 Aeff ·µ0 ·µr Aeff ·µ0 600 0,097 0,104 7,59 0,002 1,71
tr = ; Lm2 = N22 ⋅ ; Ll = N112 ⋅ (12)
N11 + N12 leff 2 + leff 3 lgap
TABLE 3. COMPARISON BETWEEN MAGNETIZANTION INDUCTANCE VALUES
OBTAINED FROM SIMULATIONS AND CALCULATION RESULTS
Those equations allows to design the proposed magnetic Gap Magnetization inductance (referred to secondary winding) (mH)
(µm) Calculated FEA Error %
component following and easy procedure. 0 2,252 2,467 8,72
200 1,528 1,625 6,00
IV. SIMULATION RESULTS 400 1,468 1,559 5,82
600 1,534 1,446 5,75
Simulations using FEA (finite element analysis) have been
carried out in order to check the proposed approach to design
Ll1 simulated Ll1 calculated
integrated magnetic components. An E42/21/15 Ferroxcube Lm2 simulated Lm2 calculated
core with a relative permeability µr=2300 has been simulated, 3
according to the winding arrangement described in the 2,5
previous section with N11=20, N12=20, N2=20. Equation (13) 2
L(mH)
has been used in order to obtain the transformation ratio, once 1,5
Ɏ1/Ɏ2 relation was obtained from the FEA simulation. 1
N2
Tr = (13) 0,5
Φ
N 12 + N 11 ⋅ 1 0
Φ2 0 100 200 300 400 500 600
processed to determine the magnetizing inductance, the Figure 6. Leakage inductance and magnetizing inductance.
leakage inductance of the primary and secondary windings,
and the transformation ratio. The method followed is In table 1, is possible to see that the difference between
presented in [5], [6] and [7]. In all cases, simulations were analytical and simulation results is less than 0,1%. The
made under no saturation conditions. transformation ratio can be obtained through an analytical way
On the other hand, simple calculations according to the with great accuracy.
proposed design method ((7) and (8)) have been carried out in
order to obtain the primary winding leakage inductance and
0,8
the transformation ratio. Tables 1 and 2 and Figure 6 show all
the data obtained for different air gap thicknesses, comparing 0,6
the FEA results with the simple calculation values obtained 0,4
tr
the maximum leakage inductance and the minimum Figure 7. Transformation ratio.
transformation ratio are obtained for the minimum air gap
thickness. From table 2, different conclusions can be observed:
• Secondary leakage inductance can be neglected respect
the Primary leakage inductance.
• The error is reasonably low. In this case, an air gap length
of 1,4% of the total column length corresponds with a
1203
7,6% error. This difference is mainly due to “fringing core material is M0H oriented grain with a plate thickness of
flux” effect. 0,23mm.
• The bigger the gap length, the grater the error. 300
180
300
Two plots are displayed in Figure 8. The first one shows the
difference between primary leakage inductance values
calculated analytically and the simulation values. The second
60 60
plot shows the same results for magnetization inductance. Dimensions in mm.
Depth = 80mm.
50
In this work, no size optimization was developed. In the
next future, this final step will be carried out in order to obtain
40
an optimized model which will improve the overall size of the
30
system
%
1204
• Permeability of the material. 4
10
• Effective length of the magnetization flux path.
• Effective surface of the magnetization flux path. 3
10
Amplitude (Vp)
However, magnetization inductance value is stronger 2
10
enough and this error is not critical for our purposes.
On the other hand, the measured series inductance is bigger 1
10
than the analytically calculated. That difference is mainly due
to the fringing flux and the natural leakage flux of the 0
10
windings, and was perfectly expected.
-1
10 0
In figure 11 it is presented the transfer function Vout/Vin in a 10 10
1
10
2
10
3 4
10
frequency sweep, measured with an impedance analyzer under Frequency (hz)
(a)
no load working conditions, and with an external capacior of 5
10
10uF parallel connected in the secondary winding (figure 1b).
4
10
Amplitude (Vp)
3
10
2
10
1
10
0
10
-1
10 0 1 2 3 4
10 10 10 10 10
Frequency (hz)
(b)
1205
[4] Njiende, H.; Froehleke, N.; Cronje, W.A. “Modeling and analysis of
integrated magnetic components”; PESC '03. Page(s):283 - 288 vol.1
[5] Prieto, R.; Asensi, R.; Cobos, J.A.; Uceda, J. ”A full procedure to model
integrated magnetics based on FEA”; APEC '04. Page(s):952 - 957 vol.2
[6] Pleite, J; Prieto, R; Asensi, R; Cobos, J.A; Olías, E. “Modeling of
magnetic components based on finite elements techniques”. Power
Electronics Congress, 1996. Technical Proceedings. CIEP’96. V IEEE
Internacional. Page(s): 170 – 175.
[7] Valdivia Guerrero, Virgilio. “Técnicas de integración magnética.
Análisis, diseño y verificación experimental”. Master Thesis. Carlos III
University of Madrid.
[8] Pleite, J; Prieto, R; Asensi, R; Cobos, J.A; Olías, E. “Obtaining a
frequency dependent and effect distributed model of magnetic
components from actual measurements”. IEEE – PELS INTELEC’96.
Boston.
[9] Arun Balakrishnan, William T. Joines, Thomas G. Wilson. “Air-gap
reluctance and inductance calculations for magnetic circuits using a
Schwarz – Chirstoffel transformation”. IEEE Transactions on Power
Electronics, Volume 12, Issue 4, July 1997 Page(s):654 – 663.
1206