Meynard PDF

Design Methods and Tools
for Power Electronics

T. Meynard G. Fontes
Laplace N7 CNRS
Toulouse
Outline of the presentation
Simulation vs Design
Semi Analytic Design
Designing with objects and optimization

SIMULATION VS DESIGN
Standard simulation tools
Simulation is an injection :
one circuit gives a single set of waveforms.
Design tools
Design is NOT an injection :
there can be an infinity of solution for a given set of specifications
(and sometimes NO solution!...)
Hidden quantities in
standard simulation tools
Which of the components in this circuit is the biggest? => L? C? the heatsink?
Which of the components in this circuit is the most expensive?
If L is halved and C doubled to get the same voltage ripple, will the filter be smaller?
Or less expensive?
SEMI ANALYTIC DESIGN
Commutation Cell Design
Start
User defined
Specifications
Measure
Efficiency &Weight
f
sw
SC
Filter Design
HV & LV
standards
Simulate
Build
Test
loss estimate
more accurate
loss estimate
select
heatsink
A standard design process
COMMUTATION CELL DESIGN
9

Rules of the game

a) Define a global switch requirement (Voltage current, Frequency, Duty Cycle,
Case Temperature,..)

b) Evaluate the limit of operation of a switch to determine haw many must be
connected in series and parallel to fulfill requirements

c) Evaluate losses and other characteristics of the design

d) Repeat a) to c) for each component and compare results and make a choice

Find the MacroSwitch
with the best efficiency
10
I
sw
0

V
sw
D/f
sw
1/f
sw
O
case

1-Define MacroSwitch requirements
Voltage to be switched
+
Maximum collector-emitter voltage in switching mode

Number of series connected switches

2-Find number of series connected switches
V
sw
V
ce
1 )
.
int(
max margin
+ =
CE
sw
Series
V V
V
n
11
=> P
sw
= f
sw
E
on,off
= f
sw
( A
on,off
+ B
on,off
.I + C
on,off
. I )

=> P
cond
= V
CE
I = (V
T
+ R
T
. I) I = V
T
I+R
T
. I

3-Find Maximum Current per switch for this profile :
evaluate variation of losses as a function of the current
12
Conduction losses
+
switching losses

I
max
(f
sw
)

O
jMax

R
Ojc

3-Find Maximum Current per switch for this profile :
solve thermal equation
Conduction losses + Switching losses = Maximum Power extracted
.
. + .
= 0
= .
2
4.
() =
.
2 .
. +
. +
. +
. =

13
0 A
100 A
200 A
300 A
400 A
1E+3 Hz 1E+4 Hz 1E+5 Hz
C
u
r
r
e
n
t

I
s
w
Switching Frequency f
sw
SEMIX 302GB12T4s
Vsw=900; D=0.5; dTetha_J_C= 60C
Maximum Power
Current /frequency operating area of a switch
Maximum Power
() =
.
2 .

14
4-Find number of parallel connected switches
5-Find losses and efficency
V
sw
I
sw
= .
2
+
2

= 1

18
6-Build a MacroSwitch
with each component of the database
I
sw
0

V
sw
D/f
sw
1/f
sw
O
case

E
x
p
l
o
r
e

I
G
B
T

d
a
t
a
b
a
s
e

19
0.0%
0.5%
1.0%
1.5%
2.0%
2.5%
3.0%
0 50 100 150 200 250
Coefficients de pertes des diffrents modules
Vcommut=3000V; Icommut=100A; DutyCycle=50%; fsw=0Hz; DTjc=50C
600V (7 en srie)
1200V (4 en srie)
1700V (3 en srie)
3300V (2 en srie)
6500V (1 en srie)
Find the IGBT-based MacroSwitch with the best efficiency
6500Vdevices=>
E
x
p
l
o
r
e

I
G
B
T

d
a
t
a
b
a
s
e

(
*
)
LossCoeff =Loss in one switch / Ouput Power (Buck converter configuration)
L
o
s
s
C
o
e
f
f

(
*
)

RefNr (in Voltage Class)
20
0.0%
0.5%
1.0%
1.5%
2.0%
2.5%
3.0%
0 50 100 150 200 250
600V (7 en srie)
1200V (4 en srie)
1700V (3 en srie)
3300V (2 en srie)
6500V (1 en srie)
1700Vdevices
=>
(
*
)
L
o
s
s
C
o
e
f
f

(
*
)

E
x
p
l
o
r
e

I
G
B
T

d
a
t
a
b
a
s
e

21
0.0%
0.5%
1.0%
1.5%
2.0%
2.5%
3.0%
0 50 100 150 200 250
600V (7 en srie)
1200V (4 en srie)
1700V (3 en srie)
3300V (2 en srie)
6500V (1 en srie)
1200Vdevices=>
(
*
)
L
o
s
s
C
o
e
f
f

(
*
)

E
x
p
l
o
r
e

I
G
B
T

d
a
t
a
b
a
s
e

22
0.0%
0.5%
1.0%
1.5%
2.0%
2.5%
3.0%
0 50 100 150 200 250
600V (7 en srie)
1200V (4 en srie)
1700V (3 en srie)
3300V (2 en srie)
6500V (1 en srie)
600Vdevices=>
(
*
)
L
o
s
s
C
o
e
f
f

(
*
)

E
x
p
l
o
r
e

I
G
B
T

d
a
t
a
b
a
s
e

23
EXTENSION OF THE METHOD FOR
VOLTAGE SOURCE INVERTERS WITH
COMPLEX CONTROL PATTERNS
25
Rules of the game
Split the system in :
a modulation/topology dependant subsystem,
and a device specific subsystem.

( )
2
0
2
0
. . .
1
. .
1
. .
1
. .
1
RMS T avg T
T
T
T
T
T
T
T T
T
T
cond
i R i V dt i
T
R dt i
T
V dt i i R V
T
dt i v
T
P + =
|
|
.
|
\
|
+
|
|
.
|
\
|
= + = =
} } } }
( ) ( )
|
|
.
|
\
|
+ + + + + =

=> => OFF ON
cell off cell off off
def
cell
ON OFF
cell on cell on on
def
cell
switching
I C I B A
V
V
I C I B A
V
V
f P
2 2
mod
. . . . .
|
|
.
|
\
|
+ + + + + =

=> => => => => => OFF ON
cell cell off
OFF ON
cell cell off
OFF ON
cell off
ON OFF
cell cell on cell
ON OFF
cell on
ON OFF
on cell on
def
switching
off off off off off on on on on
I V C I V B V A I V C I V B V A
V
f
P
2 2 mod
. . . . .
26
PASSIVE COMPONENTS FOR
MULTILEVEL CONVERTERS
31
Rules of the game

Establish design criteria for filters and specific/internal components
Apply them to all configuration
Compare stored energy
Evaluate converter size or cost based on a combined lost/stored energy
criterion
32
FILTER DESIGN FOR
MULTILEVEL CONVERTERS
33
Commutation Cell with Filters

LV LV
LV
C L
f
t 2
1
0
=
HV HV
HV
C L
f
t 2
1
0
=
HV
C
V
HV
C
HV
V
LV
L
I
LV
I
LV
C
V
LV
C
HV
L
I
chop
I
chop
V
sw
f
D 1
D
HV
L
LV
L
34
One (of the many) approach of filter design

Different functions of passive components
Limit the impact of the converter on the external world :
=> limit the ripples of current on HV side and voltage on the LV side (Steady-state)
Limit the impact of the external world on the converter :
=> limit HV and LV variations induced by load steps (Transient response)
Limit the impact of the converter on itself :
=> limit the ripples of curent ripple and and HV voltage ripple (Steady-state)
High Voltage filter Low Voltage filter Commutation Cell
High Voltage filter LowVoltage filter Commutation Cell
35
AC Equivalent circuit of a Two-Level Cell
HV
C
V
HV
C
HV
V
LV
L
I
LV
I
HV
L
I
chop
I
HV
L
LV
I
sw
f
HV
V
sw
f
LV
C
V
LV
C
HV HV
HV
C L
f
t 2
1
0
=
LV
L
LV LV
LV
C L
f
t 2
1
0
=
36
HV HV
HV
C L
f
t 2
1
0
=
HV
C
V
HV
C
HV
V
2 /
LV
L
I
LV
I
LV
C
V
LV
C
HV
L
I
chop
I
LV
L
sw
f
HV
L
LV
L
D 1
D
D 1
D
D 1
D
D 1
D
Series-Parallel MultiLevel Cell with Filters

37
AC-equivalent circuit of Series-Parallel MultiLevel Cell
LV p LV
LV
C n L
f
. 2
1
0
t
=
HV
C
V
HV
C
HV
V
LV
L
I
LV
I
HV
L
I
HV
L
p
LV
n
I
sw p
f n
p s
HV
n n
V
.
sw p s
f n n
LV
C
V
LV
C
p LV
n L /
HV HV
HV
C L
f
t 2
1
0
=
38
LV
ripple sw s p
LV p LV
LV p LV
LV
sw s p
LV
s p HV
LV
ripple pk LV
ripple
V f n n
C n L
C n L
f
f n n
f
n n V
V
V
%
5 . 1
0
2
0
%
2
) ( 2
1
. 2
1
1 2
t
t
t
t
=
=
|
|
.
|
\
|
= =

Steady state, time domain : worst case ripples
Pulsation on the Low Voltage side (2
nd
order filter)
HV
ripple sw p
HV HV
HV HV
HV
sw p
HV
p LV
HV
ripple pk HV
ripple
I f n
C L
C L
f
f n
f
n I
I
I
%
5 . 1
0
2
0
%
2
. 2
1
. 2
1
1 2
max
t
t
t
t
=
=
|
|
.
|
\
|
= =

Pulsation on the High Voltage side (2
nd
order filter)
39
1kHz 10kHz 100kHz 1MHz 10MHz
t
sw Cell
HV
f n
f
V
envelope
.
2
= =>
2
0
2
) (
|
|
.
|
\
|
=
salient salient
sw HV
salient
f
f
f
f V
f gab
t
Case 1 : n
Cell
. f
sw
< f
salient
=>

) . . sin(
2
D n
n
V
A
HV
n
t
t
=
Ripples : from time domain to frequency domain
-20bB/dcd
f
switching
+ -40dB/dcd = -60dB/dcd
B
1
ripple

f
0
=>Conclusion : the ripple requirement allows increasing f
0
when increasing f
switching

t
HV
V 2
Ripple
requirement
100dB
200dB
0dB
41
EMC standards : frequency domain formulation
-20bB/dcd
f
switching
+ -40dB/dcd = -60dB/dcd
B
1
ripple

f
0
=>Conclusion : salient point of EMC standards imposes decreasing f
0
when increasing f
switching

EN55022A
1kHz 10kHz 100kHz 1MHz 10MHz
100dB
200dB
0dB
Ripple
requirement
42
4
1
2
3
4
Required cut-off frequency vs switching frequency

Steady state : ripples and standards combined
1
2
3
44
Steady state : ripples and standards combined
Required cut-off frequency vs switching frequency

for MultiCell converters

10
3
10
4
10
5
10
6
10
2
10
3
10
4
10
5
switching frequency [Hz]
c
u
t
o
f
f

f
r
e
q
u
e
n
c
y

[
H
z
]

nCell =1
10
3
10
4
10
5
10
6
10
2
10
3
10
4
10
5
Simplified EN55022A Filter for multiCell Chopper
c
u
t
o
f
f

f
r
e
q
u
e
n
c
y

[
H
z
]

nCell =2
nCell =1
10
3
10
4
10
5
10
6
10
2
10
3
10
4
10
5
c
u
t
o
f
f

f
r
e
q
u
e
n
c
y

[
H
z
]

nCell =5
nCell =2
nCell =1
10
3
10
4
10
5
10
6
10
2
10
3
10
4
10
5
c
u
t
o
f
f

f
r
e
q
u
e
n
c
y

[
H
z
]

nCell =10
nCell = 5
nCell = 2
nCell = 1
2
0
d
B
@
1
0
0
k
H
z

45
Step response, average model :
full load => no load
HV
C
V
HV
C
HV
V
HV
L
LV
C
V
LV
C
p LV
n L /
Voltage overshoots
Best response of the control to limit
overshoot on LV side : impose D=0
max
) 0 (
LV
t
L
I I
LV
=
=
Worst Case :
max
) 0 (
%; 100
LV
t
L
I I D
HV
= =
=
0
max
LV
I
0.V
HV
0.I
LV
49
LV
LV
L
C
L
I
LV
LV
C
V
HV
V
LV
LV
LV
C
L
I
max
0
HV
HV
L
C
L
I
HV
HV
C
V
HV
V
HV
HV
LV
C
L
I
max
0
Worst Case
Low Voltage Side High Voltage Side
Step response, state plane analysis
full load => no load
50
max
%
%
.
..... .......... . ..........
.
LV
HV
ovrsht HV
HV
HV
HV
HV
HV
LV
HV
ovrsht
I
V V
C
L
V
C
L
I
V = =
Limit the voltage overshoot on the High Voltage Side
Full load => no load : dynamic requirement HV side
51
HV
C
V
HV
C
HV
V
HV
L
LV
C
V
LV
C
p LV
n L /
Voltage dips
Best response of the control to limit
voltage dip on LV side : maintain D=100%
0
) 0 (
=
= t
L
LV
I
Worst Case : 0 %; 100
) 0 (
= =
= t
L
HV
I D
0
max
LV
I
Step response, average model :
no load => full load
52
Low Voltage Side High Voltage Side
LV
p LV
L
C
n L
I
LV
LV
C
V
HV
V
LV
p LV
LV
C
n L
I
max
0
HV
HV
L
C
L
I
HV
HV
C
V
HV
V
HV
HV
LV
C
L
I
max
0
Worst Case
Step response, state plane analysis
no load => full load
53
max
.
..... .......... . ..........
.
LV
LV
dip HV
LV
p LV
HV
LV p
LV
LV
LV
dip
I
V V
C
n L
V
C n
L
I
V = =
Limit the voltage dip on the Low Voltage Side
No load => Full load : dynamic requirement LV side
54
Calculation of the components

=
= =
=
HV HV
LV
HV
ripple pk HV
ripple
p Cell
C L
f
I
I
I Rip
n n
. 2
1
0
% %
max
t
=
= =
=
LV p LV
HV
LV
ripple pk LV
ripple
s p Cell
C n L
f
V
V
V Rip
n n n
. 2
1
.
0
% %
t
( ) ( )
( )
|
|
.
|
\
|
=
sw HV
sw Cell salient
sw Cell salient sw Cell
f V
f n f
f n f gab Rip f n f
. 2
. , max .
. , max ;
2
min
3
%
5 . 1
0
t t
High Voltage side Low Voltage side
55
Valid for uncoupled AND coupled magnetic components
=
=
LV
dip HV
LV
LV
LV
LV
LV
LV
dip HV
p LV
V V f
I
C
I f
V V
n L
% 0
0
%
. . 2
. 2
.
max
max
t
t
=
=
HV
ovrsht HV
HV
LV
HV
LV
HV
HV
ovrsht HV
HV
V V f
I
C
I f
V V
L
% 0
0
%
. . 2
. 2
.
max
max
t
t
=
=
LV
LV p LV
LV
LV
dip HV
LV
p LV
f
C n L
I
V V
C
n L
0
%
. 2
1
.
.
max
t
=
=
HV
HV HV
LV
HV
ovrsht HV
HV
HV
f
C L
I
V V
C
L
0
%
. 2
1
.
.
max
t

56
Valid for uncoupled AND coupled magnetic components

Example #1 : 2-level converter
=> from 10 to 150kHz, the tendancy is an increase of passive components
10
3
10
4
10
5
10
6
10
-4
10
-3
10
-2
L

a
n
d

C

v
a
l
u
e
s

[
H

o
r

F
]
MultiCell Chopper with: nS=1; VHV=800; ILVmax=250; IHVmax=125; relativeOutRipple=0.01; VHVovershoot=0.1; VLVdip=0.05;standardHV=HVDCA;standardLV=EN55022A

LHV
CHV
LLV
CLV
57

Example #2 : series 2-cell converter
=> the HV filter is unchanged, L
LV
and C
LV
are reduced if f
sw
> 80kHz
10
3
10
4
10
5
10
6
10
-4
10
-3
10
-2
L

a
n
d

C

v
a
l
u
e
s

[
H

o
r

F
]

LHV @nS=1
CHV @nS=1
LLV/nP@nS=1
CLV @nS=1
LHV @nS=2
CHV @nS=2
LLV/nP@nS=2
CLV @nS=2
58

Example #3 : parallel 2-cell converter
=> all passive components are reduced if f
sw
> 80kHz
10
3
10
4
10
5
10
6
10
-4
10
-3
10
-2
L

a
n
d

C

v
a
l
u
e
s

[
H

o
r

F
]

LHV @nP=1
CHV @nP=1
LLV/nP@nP=1
CLV @nP=1
LHV @nP=2
CHV @nP=2
LLV/nP@nP=2
CLV @nP=2
59

Example #4 : parallel multiCell converter
=> with 10 // cells, all passive components start decreasing at f
sw
> 15kHz
10
3
10
4
10
5
10
6
10
-4
10
-3
10
-2
L

a
n
d

C

v
a
l
u
e
s

[
H

o
r

F
]

LHV @nP=1
CHV @nP=1
LLV/nP@nP=1
CLV @nP=1
LHV @nP=2
CHV @nP=2
LLV/nP@nP=2
CLV @nP=2
60
Combined requirements

For the same amount of energy, magnetic components are (2 to 10 times??) bigger, heavier and more
expensive than capacitor => Reducing the inductances and increasing the capacitance leaves room for
optimization
C
HV
L
HV
0

L
LV
C
LV
0

(and increasing the inductance must not be rejected a priori!)
61
Candidates for
volume reduction
Current design
Candidates for
volume reduction
Current design
Commutation Cell Design
Start
User defined
Specifications
Measure
Efficiency &Weight
f
sw
SC
Filter Design
HV & LV
standards
Simulate
Build
Test
loss estimate
more accurate
loss estimate
select
heatsink
A standard design process
weight estimate ?
VOLUME OF PASSIVE COMPONENTS AND
FILTERS FOR MULTILEVEL CONVERTERS
65
Area Product of Magnetic Components : Inductors

Basic formulation
w w
eff t
eff
A k
I n
j
.
.
=
c t
A n
I L
B
.

.
=
eff w
eff
c w
j k B
I
I L A A
. .
. . =
Drawback : L, I
^
and I
eff
are not independent variables
so the influence of L for example is not obvious
Advantage : allows selecting the core
( )
eff w DC
DC DC
DC
c w
j k B f I I
I
I
I
I
I V
A A
. .
. . 4
12
1
1 .
2
1 . .
.
2
A
|
|
.
|
\
|
A
+
|
|
.
|
\
|
A
+
=
2
12
1
1
|
|
.
|
\
|
A
+ =
DC
DC eff
I
I
I I
|
|
.
|
\
|
A
+ =
DC
DC
I
I
I I
2
1 .
f I
V
L
. 4A
=
Taking into account : ; and

we get :
( )
DC
DC DC
DC
eff w c w
I I
I
I
I
I
I V
j k B f A A
NAP
A
|
|
.
|
\
|
A
+
|
|
.
|
\
|
A
+
= =
4
12
1
1 .
2
1
.
. .
. . .
2
0
0,2
0,4
0,6
0,8
1
1,2
1,4
1,6
0% 50% 100% 150% 200%
DC
I
I A
DC
eff w c w
I V
j k B f A A
.
. .
. . .
U
s
u
a
l

d
e
s
i
g
n

Z
V
S

A
w
A
c
V
AI
66

Improved formulation #3 : combining copper losses and core losses
Limits on Core Loss Density and Copper Loss Density can be combined to form an
Improved Normalized Area Product :

Though elegant, this formulation could be misleading :
B
ACmax
and R
AC
/R
DC
are very difficult to determine a priori :
B
ACmax
should be chosen to limit core temperature rise which in practice depends on
f and core material (loss), size (volume/surface ratio), shape, cooling conditions
R
AC
/R
DC
depends on f, shape, number of turns, conductor material
2
max
12
1
1 . ;
2
1 max
8
1
.
. . . . .
|
|
.
|
\
|
A
+
|
|
.
|
\
|
A
+ = =
DC DC
AC
AC
sat
DC DC
eff w sat w c copper
core
I
I
R
R
B
B
I I I V
j k f B A A
NAP
Improved
73

Interleaved converters with uncoupled inductors
4 / 3
2
2
2
4 3
.
12
1
1
.
2
1
8
1
.
.
. . .
.
|
|
|
.
|
\
|
|
|
.
|
\
| A
+
|
|
.
|
\
|
A
+ =
|
|
.
|
\
|
=
DCtot
tot p
tot p
DCtot
p
p
DCtot
eff w
shape
tot
I
I n
I n
I
n
n
I V
j k f B
K
Vol
NVol
DCtot
tot
I
I A
NVol
Basic formulation
75

Interleaved converters with uncoupled inductors
4 / 3
2
2
max
2
4 3
12
1
1 ;
2
1 max
8
1
.
.
. . .
.
|
|
|
.
|
\
|
|
|
.
|
\
| A
+
|
|
.
|
\
|
A
+ =
|
|
.
|
\
|
=
DCtot
tot p
DC
AC
AC
sat
DCtot tot p p
p
DC
eff w sat
shape
tot
I
I n
R
R
B
B
I I n n
n
I V
j k f B
K
Vol
NVol
Improved formulation
DCtot
tot
I
I A
NVol
76
p
n /
Area Product of Magnetic Components : ICTs

Interleaved converters with coupled inductors (or InterCell Transformers = ICTs)
p
ind
tot
n
I
I
A
= A
DCind p DCtot
I n I . =
DCtot
tot p
p DCtot
tot p
DCind
ind
I
I n
n I
I n
I
I
A
=
A
=
A
2
4 / 3
. . .
8
.
12
1
1
.
2
1 . .
.
2
2
2
|
|
|
|
|
|
.
|
\
|
|
|
.
|
\
| A
+
|
|
.
|
\
|
A
+
=
eff w
DCtot
tot p
tot p
DCtot
p
DCtot
shape p tot
j k f B
I
I n
I n
I
n
I
V
K n Vol
=> Total
volume
of the n
p

coupled
inductors :
p
n .
Compared with uncoupled inductors:
Fluxes unchanged ,
Current ripples reduced
Basic formulation
77

DCtot
tot
I
I A
4 / 3
2
2
4 3
12
1
1
.
2
1
8
1
.
.
. . .
.
|
|
|
.
|
\
|
|
|
.
|
\
|
A
+
|
|
.
|
\
|
A
+ =
|
|
.
|
\
|
=
DCtot
tot
tot p
DCtot
p
p
DC
eff w
shape
tot
I
I
I n
I
n
n
I V
j k f B
K
Vol
NVol
NVol
Basic formulation
78

4 / 3
2
max
2
4 3
12
1
1 ;
.
2
1 max
8
1
.
.
. . .
.
|
|
|
.
|
\
|
|
|
.
|
\
| A
+
|
|
.
|
\
|
A
+ =
|
|
.
|
\
|
=
DCtot
tot
DC
AC
AC
sat
DCtot tot p p
p
DC
eff w sat
shape
tot
I
I
R
R
B
B
I I n n
n
I V
j k f B
K
Vol
NVol
DCtot
tot
I
I A
NVol
Improved formulation
Excel
Link
79
sat
AC
sat
AC
p
DCtot tot
AC
sat
DCtot tot p
B
B
B
B
n
I I
B
B
I I n
max
max
2
max
2
1
2 2
1
= A =
A
+
|
|
|
|
.
|
\
|
=
sat
AC
sat
AC
P DC AC P
B
B
B
B
n R R n
uncoupled
max
max
2
3
2
1
2
;
24 1
min _
|
|
|
|
.
|
\
|
=
sat
AC
sat
AC
P DC AC P
B
B
B
B
n R R n
coupled
max
max
2
3
3 / 2
1
2
;
24 1
min _
Ripple giving the minimum volume of Magnetic Component :
85
Crossing(f
1
=f
2
)
Minimum volume of LV-side uncoupled inductors and InterCell Transformers
Cf ProduitAire Selfs&ICTs.xls
Ripple giving the minimum volume (as a function of B
ACmax
/B
sat
and AI/I
DC
)
86
Maximum value of B
AC
for a loss density of 250mW/cm
3
vs frequency
Technology-related data : Core loss limitation
|
|
.
|
\
|
|
.
|
\
|
~
72 . 0
/ 250
max
13600
2 . 1 min
3
f
B
cm mW
AC
Maximum value of B
AC
3
vs frequency
87
|
|
.
|
\
|
|
.
|
\
|
~
76 . 0
/ 500
max
21400
2 . 1 min
3
f
B
cm mW
AC
Maximum value of B
AC
3
vs frequency
|
|
.
|
\
|
|
.
|
\
|
~
76 . 0
/ 750
max
26200
2 . 1 min
3
f
B
cm mW
AC
Filters with minimum volumes

V
o
l
t
a
g
e

r
i
p
p
l
e

(
#
4
)

Base point
C
HV
L
HV
Min. volume(#4)
Base point
L
LV
C
LV
92
Candidates for
volume reduction
VolL
HV

Vol C
HV

VolL
HV

Vol C
HV

M
i
n
i
m
u
m

Candidates for
volume reduction
Feasible points all use a capacitance greater than that of the base point which means capacitors bigger than basepoint.
Smaller filters can only be found by reducing the volume of the magnetic part.
For HV filter this is only possible by decreasing L
HV
, but for LV filter, the magnetic component with the smaller volume can be
obtained for a smaller or higher inductance. The minimum volume of the whole filter will be found somewhere between
these two values of inductances (base point inductance and inductance with the minimum volume), by following either the
constant LC (L_MinVol < L_BasePoint) or the constant L/C (L_MinVol > L_BasePoint) line.
Filters with minimum volumes

V
o
l
t
a
g
e

r
i
p
p
l
e

(
#
4
)

Base point
C
HV
L
HV
Base point
L
LV
C
LV
93
Candidates for
volume reduction
VolL
HV

Vol C
HV

VolL
HV

Vol C
HV

M
i
n
i
m
u
m

Candidates for
volume reduction
Min. volume(#4)
Feasible points all use a capacitance greater than that of the base point which means capacitors bigger than basepoint.
Smaller filters can only be found by reducing the volume of the magnetic part.
For HV filter this is only possible by decreasing L
HV
, but for LV filter, the magnetic component with the smaller volume can be
obtained for a smaller or higher inductance. The minimum volume of the whole filter will be found somewhere between
these two values of inductances (base point inductance and inductance with the minimum volume), by following either the
constant LC (L_MinVol < L_BasePoint) or the constant L/C (L_MinVol > L_BasePoint) line.
Commutation Cell
Design
Start
User defined
Specifications
SemiCon. & Materials
Database
End SC list ?
Efficiency vs
Specific Power
Losses
HeatSink Weight
[f
min
,f
max
] SC list
Losses
Weight
Filter
Design
HV & LV
standards
End f
sw
list ?
Design process

n
P
n
S
Start
User defined
Specifications
SemiCon. & Materials
Database
End SC list ?
Efficiency vs
Specific Power
Losses
HeatSink Weight
[f
min
,f
max
] SC list
Losses
Weight
&

0
,
0

HV & LV
standards
End f
sw
list ?
Design process
0.9 0.91 0.92 0.93 0.94 0.95 0.96 0.97 0.98 0.99 1
0
1
2
3
4
5
6
7
8
9
Efficiency [%]
S
p
e
c
i
f
i
c

P
o
w
e
r

[
k
W
/
k
g
]
V
HV
=600V ; D =50%; Power=6kW; Coupled

Combining SC data and passive component data :
feasible points in the (SpecificPower,Efficiency) plane
S
p
e
c
i
f
i
c

P
o
w
e
r

[

k
W

/

k
g

]

Efficiency [ n.u.]
Increasing f
sw
IPW90R120C3
(Si 900V)
96
0.9 0.91 0.92 0.93 0.94 0.95 0.96 0.97 0.98 0.99 1
0
1
2
3
4
5
6
7
8
9
Efficiency [%]
S
p
e
c
i
f
i
c

P
o
w
e
r

[
k
W
/
k
g
]
V
HV

IPW90R120C3
SPP11N80C3
IPP65R074C6
SPP20N60C3
STY139N65M5
APT60N60BCS
IPB60R099CP
STP20NM60FD
IPD50R399CP
IPI50R350CP
IPB200N25N3
IRFP4332PbF
IRFP4768PbF
IPB107N20N3
BSB280N15NZ3
BSZ240N12NS3
BSC060N10NS3 G
BSC109N10NS3 G
IPB025N10N3
IPB025N10N3 G
SiJ482DP
BSC042NE7NS3
BSC028N06NS
BSC014N04LSI
IRF7946PbF
IRLH5034PbF
SiR640DP
BSC0925ND
BSB013NE2LXI
S
p
e
c
i
f
i
c

P
o
w
e
r

[

k
W

/

k
g

]

Efficiency [ n.u.]
APT60N60BCS
(Si 600V)
IPW90R120C3
(Si 900V)
97
0.9 0.91 0.92 0.93 0.94 0.95 0.96 0.97 0.98 0.99 1
0
1
2
3
4
5
6
7
8
9
Efficiency [%]
S
p
e
c
i
f
i
c

P
o
w
e
r

[
k
W
/
k
g
]
V
HV

IPW90R120C3
APT60N60BCS
EPC2010
EPC1011
EPC2001
EPC1005
EPC2014
CMF10120D
CMF20120D
S
p
e
c
i
f
i
c

P
o
w
e
r

[

k
W

/

k
g

]

Efficiency [ n.u.]
APT60N60BCS
(Si 600V)
EPC1005
(GaN 60V)
CMF20120D
(SiC 1200V)
IPW90R120C3
(Si 900V)
98
Main blocks are nS, nP compatible
Design check
Graphic Vectorization in PLECS
Dedicated blocks to help vectorization
Vectorized filters
Vectorized filters with pre-design
MacroCell is composed of Flying Cap Legs (nS commutation cells in series) that can typically be used in parallel connection (nP legs in parallel).
Ideally, all commutation cells should all be controlled with the same duty cycle, but those in series with a phase-shift of 2pi/nS, and those in
parallel with a phase-shift of 2.pi/nP . When nP and nS are coprime, the input and output ripples are periodic at nP.nS.fSw.

Author : TM
vHV = 100, fSw = 20e3 % Input Voltage, switching frequency
rNom = 5, R = rNom % Nominal load, test load
nP = 3 % number of cells in parallel
nS = 12 % number of cells in series
Vectorized commutation cell
Vectorized Cyclic Cascade
InterCell Transformer
Vectorized Monolithic
InterCell Transformer
Vectorized and configurable magnetic
components for interleaved converters
Vectorized phase-shift makes
the whole block vectorized
Vectorized regulator and
control signal generator
D
i
f
f

1
-
2

>

0

D
i
f
f

2
-
3

>

0

A

>

0

A

>

0

A

<

0

A

>

0

A

<

0

Sampling must
be synchronized
with carriers to
avoid multiple
switching in the
same period
Delayed
sampling causes
errors on the
integral of the
difference that
are never
compensated for
Each ref step
must be handled
to cancel the
integral of the
difference of any
pair of control
signals
Vectorized equalizing sampler
A simple circuit
allows open-loop
compensation of
these unbalances
without
increasing the
number of
switchings
Vectorized equalizing sampler
x1e-5 integral of the differences x1e-6 integral of the differences
sampled duties and ref sampled duties and ref
f
sample
=sampRate .f
sw
phaseDelay =
sampRate .(0:n
Cell
1)
n
Cell
. f
sample

[n
Cell
] [1]
[1] [1]
[n
Cell
]
[1] [1]
[n
Cell
] [n
Cell
]
[n
Cell
] [n
Cell
]
Design check
Link to
MAcroCellw
RegulatedL
VFilter
DESIGNING WITH OBJECTS
AND OPTIMIZATION
A real-world object
Shape
Dimensions
Material
Weight
Cost
Model
Real World
Object
Shape
Dimensions
Material
Weight
Cost
Model
Real World
Object
The only part of a real-world object
that is known to a standard simulator
A real-world object
Designing a real-world object
Apply stimuli
according to specifications :
Voltage,
Current,
Switching Pattern,
Ambiant temperature

Check compatibility
with maximum ratings :
Peak voltage
Peak & RMS current
Peak induction
Losses => Temperature

Real World
Object
Constants
Converter specs
Project:
Directory Filename
ICT dimensions
ICT:
Materials Geometry
Example : InterCell Transformer
Perfs
at a blink
Main
Characteristics
Constraints
Constants
Converter specs
Project:
Directory Filename
ICT dimensions
ICT:
Materials Geometry
Perfs
at a blink
ICT dimensions
after optimization
Constrained
quantities
Main
Characteristics
1
st
Optimization
Check and refine opt.
Example : InterCell Transformer
Designing a full system
HV
C
V
HV
C
HV
V
2 /
LV
L
I
LV
I
LV
C
V
LV
C
LV
L
HV
L
LV
L
D 1
D
D 1
D
D 1
D
D 1
D
Collect objects
shape
dimensions
materials

Weight

Losses??
shape
dimensions
materials

Weight

Losses??
shape
dimensions
materials

Weight

Losses??
shape
dimensions
materials

Weight

Losses??
shape
dimensions
materials

Weight

Losses??
shape
dimensions
materials

Weight

Losses??
shape
dimensions
materials

Weight

Losses??
S
o
u
r
c
e

L
o
a
d

Air speed
COOLING

Losses??
FILTER

Losses??
FILTER

Losses??
Compose objects
shape
dimensions
materials

Weight

Losses??
shape
dimensions
materials

Weight

Losses??
shape
dimensions
materials

Weight

Losses??
shape
dimensions
materials

Weight

Losses??
shape
dimensions
materials

Weight

Losses??
shape
dimensions
materials

Weight

Losses??
shape
dimensions
materials

Weight

Losses??
S
o
u
r
c
e

L
o
a
d

FULL SYSTEM

=> Simulate full system, find waveforms and evaluate losses at last!
COOLING

Losses??
FILTER

Losses??
FILTER

Losses??
Build full system model and solve
shape
dimensions
materials

Weight

Losses!
shape
dimensions
materials

Weight

Losses!
shape
dimensions
materials

Weight

Losses!
shape
dimensions
materials

Weight

Losses!
shape
dimensions
materials

Weight

Losses!
shape
dimensions
materials

Weight

Losses!
shape
dimensions
materials

Weight

Losses!
S
o
u
r
c
e

L
o
a
d

Inputs
Outputs
Outputs
Outputs
Need for a fast solver
Steady-state waveforms are needed

Accelerated determination of steady-state waveforms
with a standard (time-domain) simulator is not the best choice.

In most cases simplifications can be made to allow frequency domain analysis
which inherently is a direct determination of steady state waveforms.

Main assumptions to allow standard frequency analysis (linear system) :

Intrinsic non-linearities of components (saturation of permeability of magnetic
materials, exponential V(I) characteritics of diodes, etc) can be neglected :

Voltage/current ripple applied to commutation cells can be neglected
to decouple the HV and LV sides,

Influence of spontaneous commutations can be neglected,
Direct determination of
the operating point
Specified
Specified
?
?
duty cycle
??
Principle used for approximate
determination of the operating point
Apply 1V@fRef to the LV side and solve LV circuit
Find amplitude v
load
per Volt and delay
Scale v
AC
and select phase to match v
load
specifications
Find power delivered by v
AC
and scale for specs (P
LV
#v
AC
)
Lossless commutation cell => P
HV
=P
LV

Constant v
HV
=> only i
HV
DC
gives P
HV

Solve HV circuit with I
dc
=1A and find internal resistance
Find i
HV
DC
and v
HV
such that P
HV
=P
LV
Find duty cycle so that v
AC
= D(t).v
HV
Assumptions :
=> linear systems, lossless commutation cell, v
HV
is constant
Equations used for approximate
Assumptions :
=> linear systems, lossless commutation cell, v
HV
is constant
=
2
+
2
4.
2

Approximate time domain
opPoint known here
Approximate frequency domain
Solve separate normalized circuit (I
DC
=1A ; V
AC
=1V 0) using a single frequency! (DC and f
Ref
)
=

v
H
V

2
+
2
4.
2

I
LV
Full frequency domain analysis
using the operating point
HV
C
V
HV
C
HV
V
LV
L
I
LV
I
HV
L
I
HV
L
p
LV
n
I
sw p
f n
p s
HV
n n
V
.
sw p s
f n n
LV
C
V
LV
C
p LV
n L /
139
The control pattern duty(t) determined previously allows direct calculation of the
steady state waveforms at a point that is very close to the specified point:
The circuit is split in independant linear subcircuits
The spectra of the souvrces are derived from duty(t),
and

(time domain multiplication by duty(t) followed by FFT, or direct convolution of spectra)

The circuit is solved in the frequency domain
If necessary time waveforms regenerated using iFFT.
FULL SYSTEM
Optimize at last
shape, materials, dimensions
Voltages, currents, temperatures, weight, losses
Objective Constraints
Optimization routine

Meynard PDF

Caricato da

Informazioni sul documento

Descrizione originale:

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

Meynard PDF

Caricato da

Copyright:

Formati disponibili

Design Methods and Tools

for Power Electronics

Potrebbero piacerti anche