j. marcos alonso€¦ · next step is to model the b-h curve and to obtain a valid expression for...
TRANSCRIPT
10/12/2018
1
J. Marcos Alonso
Electrical Eng. Dept., Campus Viesques, Ed. 3, Room 3.2.20
33204-Gijon, Asturias, Spain
SPAIN
Introduction to VIs: Why? What? How?
Analytical Modeling of VIs
SPICE-based Modeling of VIs
Examples of Applications
Conclusions
10/12/2018
2
Why?
• Variable inductors can provide additional control
parameters to improve converter operation:
4
BUCK DC-DC CONVERTER
Duty Cycle
Frequency
10/12/2018
3
• Variable inductors can provide additional control
parameters to improve converter operation:
5
BUCK DC-DC CONVERTER
Inductance
What?
10/12/2018
4
Air Core Inductors are non-variable inductors:
B(T)
H(A/m)
L (H)
I (A)
0=B/H is constant
The inductance does not depend on the
current or any other electrical parameter
I
I
7
A
r
Tapped inductors can be used as variable inductors:
8
Discrete variation of the inductance
Difficult implementation of selector switch (bidirectional), especially
at high frequency
Difficult implementation of closed-loop control
Not practical in many applications
10/12/2018
5
Variable inductors based on moving cores:
9
Continuous variation of the inductance
Mechanical control of the inductance
Difficult implementation of closed-loop control
Not practical in many applications
Magnetic Core Inductors are inherent variable inductors:
B(T)
H(A/m)
L (H)
I (A)
is variable, it depends on the
operating point (B, H)
The inductance depends on the
current circulating through the
winding
It has no independent controlConstant Variable
Constant Variable
4
10/12/2018
6
Magnetic Core Inductors are inherent variable inductors:
B(T)
H(A/m)
L (H)
I (A)
is variable, it depends on the
operating point (B, H)
The inductance depends on the
current circulating through the
winding
It has no independent controlConstant Variable
Constant Variable
4
Saturation
≠Short Circuit
Magnetic Core Inductors are variable inductors:
L (H)
I (A)
For variable inductor operating in the saturation region the effective
inductance must be considered:
Constant Variable
5
10/12/2018
7
Magnetic Core Inductors are variable inductors:
L (H)
I (A)
This type of variable inductor can be used in different applications:
Power factor correction, Maximum Power Point Tracking in PV applications,
voltage regulation, etc.[1]
However, application is not flexible due to its inherent dependence of the inductance
versus current
A controllable variable inductor would be much more interesting for general
applications
Constant Variable
[1] W.G. Hurley, W.H. Wölfle; Transformers and Inductors for Power Electronics. Wiley. 2013.6
A controllable variable inductor is a device whose inductance can be varied
by using an electrical magnitude, typically a current:
Icontrol
L (H)
VariableIcontrol
Now, the inductance depends on the control current
The effect of the current circulating through the inductance can be
neglected
The equation that rules the VI operation will be simpler:
7
(A)
10/12/2018
8
How?
The trick is to superpose a DC flux to the AC flux created by the inductor
main winding
By shifting the DC operating point along the B-H curve the permeability is
modulated
This modifies the inductance seen from the AC winding
DC FluxAC Flux
g
Modulated by DC flux
DC current
DC current
Variable Inductance:
Depends on DC current
8
10/12/2018
9
The AC flux generates a voltage across the DC winding
This voltage can be high and could damage the DC current source used to
supply the DC winding.
DC FluxAC Flux
g
DC current
DC current
PROBLEM:
9
Adding a series choke to the DC winding can filter the reflected AC voltage
DC FluxAC Flux
g
DC current
DC current
A POSSIBLE SOLUTION:
Disadvantages: complex, bulky and inefficient
10
10/12/2018
10
Adding two auxiliary windings in reverse polarity on the outer legs
Both DC windings generate DC flux in the same direction
The AC flux enters each DC winding with a different polarity
As a consequence, the AC voltages reflected across the DC windings
compensate each other
The DC current source sees almost no AC voltage across its terminals
A BETTER SOLUTION:
11
Triple E structure Quadruple U structure
12
10/12/2018
11
Introduction to VIs
Analytical Modeling of VIs
SPICE-based Modeling of VIs
Examples of Applications
Conclusions
Main ac winding on the central arm
DC bias windings on the lateral arms
DC bias current modulates material permeability
Goal: Obtain the variation of the inductance as a function of the dc bias
current. It is also important to determine the total inductance seen from the
auxiliary windings because this inductance will strongly affect to the dynamic
behavior of the VI.
22
CONTROLIdc
L
dc
Ndc Ndc
Np
lg
A1 A2
M1
M2
ac
10/12/2018
12
23
It is assumed that the ac component is small compared to the dc level
Superposition law can be applied to calculate dc and ac components
24
Once solved, the dc winding inductance can be calculated:
Analyzing the circuit:
The following equation must be solved to obtain
the magnetic flux density dc level:
10/12/2018
13
25
AC reluctances must be considered:
Calculation of AC flux and AC inductance is straightforward:
26
10/12/2018
14
27
DC magnetic flux density:
DC winding inductance:
28
AC magnetic flux:
AC winding inductance:
10/12/2018
15
Introduction to VIs
Analytical Modeling of VIs
SPICE-based Modeling of VIs
Examples of Applications
Conclusions
Importance of modeling of magnetic devices prior to manufacturing
Provide a better understanding of complex magnetic structures
Two typical modeling methods:
Finite Element Analysis (FEA): high accuracy, time consuming.
SPICE based behavioral models: quicker implementation and
simulation.
Previous works:
Ben-Yaakov 2005
10/12/2018
16
Relatively complex structures
Several windings in different arms
AC and DC excitation
Different applications
31
CONTROLIdc
L
dc
Ndc Ndc
Np
lg
A1 A2
M1
M2
ac
L ( H)
0 0.1 0.2 0.3 0.4 0.51
2
3
4
5
IDC (A)
Required elements:
Constant reluctance elements: air gap, non magnetic materials
Variable reluctance elements: depend on magnetic flux density (B)
Windings: magnetic – electrical interaction
32
CONTROLIdc
L
dc
Ndc Ndc
Np
lg
A1 A2
M1
M2
ac
+- +
-Ndc Idc Ndc IdcNpIp
Rl RrRc
Rg
() () ()
fcfl fr
Fdc Fp Fdc+-
10/12/2018
17
33
In SPICE simulator:
Voltage Magnetomotive Force
Current Magnetic Flux
Resistance Reluctance
34
ℜ𝑚 =𝑙𝑚
𝜇 𝐴𝑚
But can only be considered as a
constant when the material operates in
linear region!
10/12/2018
18
In a general case the material has to be
modeled in the whole range:
35
Material
B-H Curve
𝜇(𝐵) =𝑑𝐵
𝑑𝐻
ℜ𝑚(𝐵) =𝑙𝑚
𝜇(𝐵) 𝐴𝑚
Reluctance depends on
Magnetic Flux density B:
Next step is to model the B-H Curve and to obtain a valid expression for 𝜇(𝐵)
Neglecting hysteresis effects
Simple relationship to avoid SPICE convergence problems
36
Hyperbolic model:
𝐻 𝐵 =𝑘1
𝑘2𝐵+ 1
Brauer’s model:
𝐻 𝐵 = 𝑘1𝑒𝑘2𝐵
2+ 𝑘3 𝐵
𝜇 𝐵 = 𝑘1 1 + 2𝑘2𝐵2 𝑒𝑘2𝐵
2+ 𝑘3
−1 Direct expression of non-linear
material permeability:
𝑑
𝑑𝐵
10/12/2018
19
TDK-EPCOS N87 MATERIAl
37
N87 Datasheet
TDK-EPCOS N87 MATERIAl
38
𝜇 𝐵 = 𝑘1 1 + 2𝑘2𝐵2 𝑒𝑘2𝐵
2+ 𝑘3
−1
Maximum relative error 26% @ 0.3 T
0 0.1 0.2 0.3 0.4 0.5
B (T)
H (
A/m
)
0
500
1000
1500
Datasheet
Brauer’s Model
TDK-EPCOS N87 @ 25 ºC
𝐻 𝐵 = 𝑘1𝑒𝑘2𝐵
2+ 𝑘3 𝐵
10/12/2018
20
39
+-
value={(lm/(V(um)*Am))*I(Em)}
Em Gmb
value={I(Em)/Am}
Bm
Gmu1 1
value={ud(V(Bm))}
um
.func ud(B) { 1/( (1+2*k2*B*B)*exp(k2*B*B) + k3 ) } ; differential permeability
+
-
FR
fR
Resistance emulation Bm calculation d calculation
Can be implemented as an isolated component
40
Electric Variables
Magnetic Variables
Geometric
Parameters
10/12/2018
21
41
+-EFw
value={ Nw*I(EVw) }value={ -I(EFw) }
GVw
1m
{Nw}
Lw
Magnetic Part
(MMF)
Electric Part
(Winding)
+-
+-
EVw
1
+
-
vw
iw
+
-
Fw
fw+
-
fwddt
Nw
Implementation of M.M.F.Implementation electric-magnetic interaction
Airgap Linear Reluctor
Non Linear Reluctor Winding
Available for evaluation at: http://www.unioviedo.net/ate/marcosaa
B
OutputL
Output
10/12/2018
22
43
OUTER ARMS
21.74
(*) Dimensions in mm. Not to scale.
9.10
10.92
3.15
CENTRAL ARM
5.20
0.60
12.20 gap
EFD25 CORE*
CONTROLIdc
L
dc
Ndc Ndc
Np
lg
A1 A2
M1
M2
ac
44
AirgapMain Winding
Non-linear
Reluctors
Auxiliary
Windings
Inductance
Calculation
Parameters
10/12/2018
23
45
LTspice circuit for the simulation of the VI, including the calculation of the ac
winding inductance:
46
DC magnetic flux density in the outer
arms: theoretical results (blue solid line)
and simulation results (blue squares).
AC winding inductance: theoretical
results (blue solid line) and
simulation results (blue squares).
theoretical
simulation
theoretical
simulation
10/12/2018
24
47
LTspice circuit for the simulation of the VI including the calculation of the dc bias
winding inductance.
48
DC bias winding inductance
DC bias winding inductance theoretical results (blue solid) and
simulation results (red triangles).
DC bias winding effective inductance (L_(dc_eff)): theoretical results
(green solid) and simulation results (blue squares)
theoretical
simulation
theoretical
simulation
10/12/2018
25
49
AC winding inductance
AC winding inductance: theoretical results (blue solid line) and
experimental measurements (red triangles).
theoretical
experimental
50
VI Current
VI Voltage
Right Arm Induction
Left Arm Induction
Center Arm InductionVI Inherent Time
Response
0.5 V
1.0 V
0.0 V 0.0 mA
50 mA
100 mA
350 mV
-350 mV
0.0 mV
0.2 us 0.6 us 1.0 us 1.4 us
V(out) I(R5)
V(bc) V(bd) V(bi)
10/12/2018
26
51
VI Model
52
-10V
-5V
0.0us 0.5us 1.0us 1.5us 2.0us 2.5us 3.0us 3.5us 4.0us 4.5us
0V
5V
-15V
10V
-10V
-5V
0V
5V
10V
15V
5.0us
V(lt1,mid)
V(lt2,mid)
SIMULATION EXPERIMENTAL
Input and Output
Voltage
10/12/2018
27
53
-5V
0.5us 1.0us 1.5us 2.0us 2.5us 3.0us 3.5us 4.0us 4.5us
0V
5V
-20V
10V
-10V
0V
10V
20V
5.0us
-10V
0.0us
V(lt1,lt2)
V(lt2,mid)
SIMULATION EXPERIMENTAL
Inductor and
Output Voltage
54
SIMULATION EXPERIMENTAL
-50V
0.5us 1.0us 1.5us 2.0us 2.5us 3.0us 3.5us 4.0us 4.5us
0V
50V
-150V
100V
5.0us
100V
-100V
-50V
0V
50V
-100V
100V
-50V
0V
50V
100V
0.0us-100V
V(fc)
-V(auxl-)
V(fc, auxl-)
Voltage across
bias windings
10/12/2018
28
55
2uV
0.5us 1.0us 1.5us 2.0us 2.5us 3.0us 3.5us 4.0us 4.5us
-400mV
3uV
5.0us
-200mV
0.0us
1uV
0mV
200mV
400mV V(bc) V(br) V(bl)
V(l_spice)
Magnetic Flux Density B (mT)
Middle arm Right arm
Left arm
Inductance (uH)
56
RMS Output Voltage (V)
DC Bias Current (A)
0
0
0.1 0.2 0.3 0.4
RM
S O
utp
ut V
olta
ge
(V
)
5
6
7
8
9
RL
=0.5
W
RL
=1.0W
0 0.1 0.2 0.3 0.4
5
6
7
8
9
Vo_exp
Vo_sim
Ibiask
Experimental
Simulation (LTspice model)
0.5
10/12/2018
29
Introduction to VIs
Analytical Modeling of VIs
SPICE-based Modeling of VIs
Examples of Applications
Conclusions
Use of Magnetic Regulators to Improve Power Converter Performance
2222222
22
441 LfLCfR
RV
R
VP
LA
LAE
LA
LALA
10/12/2018
30
Universal Ballast with Variable Inductance and Digital Control
Variable
Inductance
Resonant TankRectifier
Low Pass
Filter
R
LoadVCC
+
-
Vo
VCC L CS
CO
R
+
-
VOS
+
-
+
-
VCC /2
VCC /2
VCC L
CP CO R
LO
+
-
VOP
+
-
VCC /2
+
-
VCC /2
Resonant converter
LC series resonant converter
LC parallel resonant converter
10/12/2018
31
L
CP
REP
8
2
RV1
IREP
+
-
VREP
VCC L
CP CO R
LO
+
-
VOP
+
-
VCC /2
+
-
VCC /2
INDUCTANCE
R=
DC
OU
TP
UT
VO
LT
AG
E
R1
R2
R3
R4
Inductance
excursion
Lmin Lmax
R1< R2 R3< <R4
Target value
VCC
L
CP CF R
LF
+
-
Vo
Current
Source Current
Reference
Voltage
Reference
ZF
Zi
Variable
Inductor
Idc
Error
Amplifier
e
1
n
n
Magnetic Control of PRCEquivalent circuit
Can be used to control other resonant converters at constant frequency
Forward converter
with demag.
winding.
One output for each
LED branch.
Variable inductor
w/ auxiliary
windings. Output current sensing and regulation
10/12/2018
32
VDC
M1
M2
CB2
D12
D22
Co2
Vg
Rg
LED Lamp #2
1
1
D11
D21
Co1
Vg
Rg
LED Lamp #1
1
1
CB1
L1
L2
n1
n2
To other arrangements
To other arrangements
Application to several LED arrays
Each branch can be controlled independently
Multi-array LED lamp
380 V
VDC
M1
M2
CB
L
D1
D2
Co
+-
VDC
2Vg
Rg
LED
Lamp
n
1
1
i
+
-
vg
+
-
v1
is
+
-
Vo
Io
L
VDC+- 2
nVo+-
i
jvg v1
Equivalent Circuit
t
-
nVo
j
VDC
2
VDC
2
-nVo
t
2
TS TS
vg
v1
t
(t)
(t)
i(t)Ip
-Ip
t
nIp
IO
iS (t)
10/12/2018
33
Variable inductors can be used to provide additional control
parameters in power electronics converters.
Analytical modeling of VI is useful for the first design and
evaluation of the variable inductor.
SPICE based models can be used to simulate the complete converter
under VI control.
VI have been tested successfully to perform control of power
converters in different applications
New ideas and applications are expected in the near future. There
are possibilities to develop new ideas using VI for the control of
power converters.
18
Gijón, Asturias, Spain