PELSDenverSlides PDF

Resonant Power Conversion
Prof. Bob Erickson

Colorado Power Electronics Center
Department of Electrical, Computer, and Energy Engineering
University of Colorado, Boulder
Outline
1.  Introduction to resonant power conversion

2.  Simple frequency-domain modeling of resonant converters with the
fundamental approximation
3.  The series and parallel resonant converters, and zero-voltage switching
4.  Design techniques: shaping the tank characteristics to achieve desired
output I-V characteristics, achieve zero-voltage switching, and improve
light-load efficiency
5.  A resonant converter that has found substantial recent commercial
application: the Ls-Lp-C converter
Introduction to Resonant Conversion
Resonant power converters contain resonant L-C networks whose

voltage and current waveforms vary sinusoidally during one or more
subintervals of each switching period. These sinusoidal variations are
large in magnitude, and the small ripple approximation does not apply.
Some types of resonant converters:
•  Dc-to-high-frequency-ac inverters
•  Resonant dc-dc converters
•  Resonant inverters or rectifiers producing line-frequency ac
Fundamentals of Power Electronics 2 Chapter 19: Resonant Conversion

A basic class of resonant inverters
NS NT
is(t) i(t)
Basic circuit +
L Cs +
dc Resistive
source + vs(t) Cp v(t) load
–
vg(t) R
– –
Switch network Resonant tank network
Several resonant tank networks

L Cs L L Cs
Cp Cp
Series tank network Parallel tank network LCC tank network

Tank network responds only to fundamental
component of switched waveforms
Switch
output
voltage Tank current and output
spectrum voltage are essentially
fs 3fs 5fs
sinusoids at the switching
f
frequency fs.
Resonant Output can be controlled
tank
response by variation of switching
frequency, closer to or
away from the tank
fs 3fs 5fs f resonant frequency
Tank
current
spectrum
fs 3fs 5fs f

Derivation of a resonant dc-dc converter
Rectify and filter the output of a dc-high-frequency-ac inverter

Transfer function
H(s)
is(t) iR(t) i(t)
+
+ L Cs +
dc
source +– vs(t) vR(t) v(t) R
vg(t)
–
–
–
NS NT NR NF
Switch network Resonant tank network Rectifier network Low-pass dc
filter load
network
The series resonant dc-dc converter

Resonant conversion: advantages
The chief advantage of resonant converters: reduced switching loss

Zero-current switching
Zero-voltage switching
Turn-on or turn-off transitions of semiconductor devices can occur at
zero crossings of tank voltage or current waveforms, thereby reducing
or eliminating some of the switching loss mechanisms. Hence
resonant converters can operate at higher switching frequencies than
comparable PWM converters
Zero-voltage switching also reduces converter-generated EMI
Zero-current switching can be used to commutate SCRs
In specialized applications, resonant networks may be unavoidable
High voltage converters: significant transformer leakage
inductance and winding capacitance leads to resonant network
Resonant conversion: disadvantages
Can optimize performance at one operating point, but not with wide
range of input voltage and load power variations
Significant currents may circulate through the tank elements, even
when the load is disconnected, leading to poor efficiency at light load
Quasi-sinusoidal waveforms exhibit higher peak values than
equivalent rectangular waveforms
These considerations lead to increased conduction losses, which can
offset the reduction in switching loss
Resonant converters are usually controlled by variation of switching
frequency. In some schemes, the range of switching frequencies can
be very large
Complexity of analysis

19.1 Sinusoidal analysis of resonant converters
A resonant dc-dc converter: Transfer function

H(s)
is(t) iR(t) i(t)
+
+ L Cs +
dc
vg(t)
–
–
–
NS NT NR NF
Switch network Resonant tank network Rectifier network Low-pass dc
filter load
network
If tank responds primarily to fundamental component of switch

network output voltage waveform, then harmonics can be neglected.
Let us model all ac waveforms by their fundamental components.

The sinusoidal approximation
Switch
output
voltage Tank current and output
spectrum voltage are essentially
fs 3fs 5fs
sinusoids at the switching
f
frequency fs.
Resonant Neglect harmonics of
tank
response switch output voltage
waveform, and model only
the fundamental
fs 3fs 5fs f component.
Tank
current Remaining ac waveforms
spectrum can be found via phasor
analysis.
fs 3fs 5fs f

19.1.1 Controlled switch network model
4
NS Vg
Fundamental component
is(t) Vg vs1(t)
1
+ vs(t)
t
vg + 2 vs(t)
– 2
–
1
– Vg
Switch network
If the switch network produces a The fundamental component is

square wave, then its output 4Vg
voltage has the following Fourier vs1(t) = sin (st) = Vs1 sin (st)
series:
4Vg So model switch network output port
vs(t) = 1n sin (nst) with voltage source of value vs1(t)
n = 1, 3, 5,...

Model of switch network input port
NS Is1
is(t)
1
+ ig(t)
vg + 2 vs(t)
– 2
– s t
1 is(t)
Switch network s
Assume that switch network T s/2

output current is i g(t) T = 2 i g()d
i s(t) I s1 sin (st – s) s Ts 0
s T /2
It is desired to model the dc 2 I s1 sin (s – s)d

Ts 0
component (average value) 2 I cos ( )
of the switch network input = s1 s
current.
Switch network: equivalent circuit
+ is1(t) =
Is1 sin (st – s)
vs1(t) =
vg 2I s1 +
cos ( s) 4Vg –
sin (st)
–
•  Switch network converts dc to ac

•  Dc components of input port waveforms are modeled
•  Fundamental ac components of output port waveforms are modeled
•  Model is power conservative: predicted average input and output
powers are equal

19.1.2 Modeling the rectifier and capacitive
filter networks
iR(t) | iR(t) | i(t) V vR(t)

+ +
vR(t) v(t) R iR(t) s t
– –
NR NF
Rectifier network Low-pass dc –V
filter load
network
R
Assume large output filter If iR(t) is a sinusoid:

capacitor, having small ripple. i R(t) = I R1 sin (st – R)
vR(t) is a square wave, having Then vR(t) has the following
zero crossings in phase with tank Fourier series:

4V
output current iR(t). vR(t) = 1n sin (nst – R)
n = 1, 3, 5,

Sinusoidal approximation: rectifier
Again, since tank responds only to fundamental components of applied

waveforms, harmonics in vR(t) can be neglected. vR(t) becomes
vR1(t) = 4V
sin (st – R) = V R1 sin (st – R)
Actual waveforms with harmonics ignored
V 4 vR1(t)
vR(t) V
fundamental
iR(t) s t iR1(t) s t
vR1(t)
i R1(t) =
Re
–V
Re = 82 R

R R

Rectifier dc output port model
iR(t) | iR(t) | i(t)

Output capacitor charge balance: dc
+ +
load current is equal to average
vR(t) v(t)
rectified tank output current
R
i R(t) =I
– – Ts
NR NF
Rectifier network Low-pass dc Hence
filter load T s/2
network
I= 2 I R1 sin (st – R) dt
TS 0
V vR(t) =2I
R1
iR(t) s t
–V
R

Equivalent circuit of rectifier
iR1(t) I
Rectifier input port:
+ +
Fundamental components of
current and voltage are 2
vR1(t) Re I R1 V R
sinusoids that are in phase
Hence rectifier presents a – –
resistive load to tank network
Re = 82 R
Effective resistance Re is
vR1(t) 8 V Rectifier equivalent circuit

Re = =
i R(t) 2 I
With a resistive load R, this becomes
Re = 82 R = 0.8106R


19.1.3 Resonant tank network
Transfer function
H(s)
is1(t) iR1(t)
+
+ Zi Resonant
vs1(t) network vR1(t) Re
–
Model of ac waveforms is now reduced to a linear circuit. Tank

network is excited by effective sinusoidal voltage (switch network
output port), and is load by effective resistive load (rectifier input port).
Can solve for transfer function via conventional linear circuit analysis.

Solution of tank network waveforms
Transfer function: Transfer function

vR1(s) H(s)
= H(s)
vs1(s) is1(t) iR1(t)
+
Ratio of peak values of input and
Zi Resonant
output voltages: vs1(t) +
network vR1(t) Re
–
VR1
= H(s) –
Vs1 s = j s
Solution for tank output current:

vR1(s) H(s)
i R(s) = = v (s)
Re Re s1
which has peak magnitude

H(s) s = j
s
I R1 = Vs1
Re

19.1.4 Solution of converter
voltage conversion ratio M = V/Vg
Transfer function
H(s)
is1(t) iR1(t) I
+ +
+ + Zi Resonant 2
Vg network vR1(t) Re I R1 V R
– –
– –
vs1(t) =
2I s1 Re = 82 R
cos ( s) 4Vg
sin (st)
Eliminate Re:
M= V = R 2

1 H(s) 4

Vg Re s = j s
V = H(s)
Vg s = j s
V I I R1 VR1 Vs1
I I R1 VR1 Vs1 Vg

Conversion ratio M
V = H(s)
Vg s = j s
So we have shown that the conversion ratio of a resonant converter,

having switch and rectifier networks as in previous slides, is equal to
the magnitude of the tank network transfer function. This transfer
function is evaluated with the tank loaded by the effective rectifier
input resistance Re.

19.2 Examples
19.2.1 Series resonant converter
transfer function
H(s)
is(t) iR(t) i(t)
+
+ L Cs +
dc
vg(t)
–
–
–
NS NT NR NF
switch network resonant tank network rectifier network low-pass dc
filter load
network

Model: series resonant converter
transfer function H(s)
L C
is1(t) iR1(t) I
+ +
+ + Zi 2
Vg vR1(t) Re I R1 V R
– –
– –
2I s1
vs1(t) = Re = 82 R
series tank network
cos ( s) 4Vg
sin (st)
Re Re 1 = 2 f
H(s) = =
Z i(s) R + sL + 1
0 = 0
e
LC
sC L
s R0 = M = H( js) = 1
Q e 0 C 2
= R 1+Q 2 1 –F
1+ s + s
2
Qe = 0 e
F
Q e 0 0 Re
Construction of Zi
|| Zi ||
1
C
L
f0
R0
Qe = R0 / Re
Re

Construction of H
|| H ||
1
Qe = Re / R0
Re / R0
f0
R /
e
C L
R e

19.2.2 Subharmonic modes of the SRC
switch
output Example: excitation of
voltage tank by third harmonic of
spectrum switching frequency
fs 3fs 5fs f Can now approximate vs(t)
by its third harmonic:
resonant
tank 4Vg
response vs(t) vsn(t) = n sin (nst)
fs 3fs 5fs f
Result of analysis:
tank
current V H( jns)
spectrum M= = n
Vg
fs 3fs 5fs f

Subharmonic modes of SRC
M
1
1
3
1
5
etc.
1 f 1 f
5 0 3 0 f0 fs

19.2.3 Parallel resonant dc-dc converter
is(t) iR(t) i(t)

+
+ L
+
dc
source +– vs(t) Cp vR(t) v(t) R
vg(t)
–
–
–
NS NT NR NF
switch network resonant tank network rectifier network low-pass filter dc
network load
Differs from series resonant converter as follows:

Different tank network
Rectifier is driven by sinusoidal voltage, and is connected to
inductive-input low-pass filter
Need a new model for rectifier and filter networks

Model of uncontrolled rectifier
with inductive filter network
I iR(t) iR(t) i(t)

+
+
vR(t) s t
vR(t) v(t) R
–
–
–I NR NF
k rectifier network low-pass filter dc
R network load
4
I iR1(t)
fundamental
Fundamental component of iR(t):
vR1(t) s t i R1(t) = 4I
sin (st – R)
vR1(t)
i R1(t) =
Re
2
Re = R
8
R
Effective resistance Re
Again define
vR1(t) VR1
Re = =
i R1(t) 4I
In steady state, the dc output voltage V is equal to the average value
of | vR |:
T s/2
V= 2 2V
VR1 sin (st – R) dt = R1
TS 0
For a resistive load, V = IR. The effective resistance Re can then be

expressed
2
Re = R = 1.2337R
8

Equivalent circuit model of uncontrolled rectifier
with inductive filter network
iR1(t) I
+ +
vR1(t) Re 2V + V
R1 – R
– –
2
Re = R
8

Equivalent circuit model
Parallel resonant dc-dc converter
transfer function H(s)
is1(t) iR1(t) I
+ +
L
+ + Zi 2 +
Vg C vR1(t) Re V R1 V R
– – –
– –
2
vs1(t) = Re = R
2I s1 8
parallel tank network
cos ( s) 4Vg
sin (st)
M = V = 82 H(s) H(s) =
Z o(s)
Vg s = j s sL
Z o(s) = sL || 1 || Re
sC
Construction of Zo
|| Zo ||
Re
Qe = Re / R0
R0
f0
L 1
C

Construction of H
|| H ||
Re / R0
Qe = Re / R0
1
f0
1
2LC

Dc conversion ratio of the PRC
8 Z o(s)
M= 2 = 82 1
sL s = j s 1+ s + s 2
Q e 0 0
s = j s
= 82 1
2
1–F 2 2
+ F
Qe
8 Re R
At resonance, this becomes M= 2 =
R0 R0
•  PRC can step up the voltage, provided R > R0
•  PRC can produce M approaching infinity, provided output current is
limited to value less than Vg / R0

19.4.1 Operation of the full bridge below
resonance: Zero-current switching
Series resonant converter example

+
Q1 Q3 L C
vds1(t) D1 D3
+
iQ1(t) –
Vg +
– vs(t)
– is(t)
Q2 Q4
D2 D4
Operation below resonance: input tank current leads voltage

Zero-current switching (ZCS) occurs

Tank input impedance
Operation below
resonance: tank input || Zi ||
1
impedance Zi is C
L
dominated by tank
capacitor.
∠Zi is positive, and f0
R0
tank input current
Qe = R0 /Re
leads tank input Re
voltage.
Zero crossing of the
tank input current
waveform is(t) occurs
before the zero
crossing of the voltage
vs(t).

Switch network waveforms, below resonance
Zero-current switching
vs1(t)
Vg
vs(t)
+
t L C
Q1 Q3
vds1(t) D1 D3
+
iQ1(t) –
– Vg
vs(t)
is(t)
– is(t)
Q2 Q4
D2 D4
Ts
+ t
2
t Ts t
2
Conduction sequence: Q1–D1–Q2–D2

Conducting Q1 D1 Q2 D2
devices: Q4 D4 Q3 D3
Q1 is turned off during D1 conduction
interval, without loss
“Hard” “Soft” “Hard” “Soft”
turn-on of turn-off of turn-on of turn-off of
Q1, Q4 Q1, Q4 Q 2, Q 3 Q 2, Q 3

ZCS turn-on transition: hard switching
vds1(t) Vg
+
Q1 Q3 L C
vds1(t) D1 D3
+
t iQ1(t) –
ids(t) vs(t)
– is(t)
Ts Q2 Q4
+ t D2 D4
2
t Ts t
Conducting Q1 D1 2 Q2 D2
devices: Q4 D4 Q3 D3
Q1 turns on while D2 is conducting. Stored
“Hard” “Soft” charge of D2 and of semiconductor output
turn-on of turn-off of capacitances must be removed. Transistor
Q1, Q4 Q1, Q4
turn-on transition is identical to hard-
switched PWM, and switching loss occurs.

19.4.2 Operation of the full bridge above
resonance: Zero-voltage switching
Series resonant converter example

+
Q1 Q3 L C
vds1(t) D1 D3
+
iQ1(t) –
Vg +
– vs(t)
– is(t)
Q2 Q4
D2 D4
Operation above resonance: input tank current lags voltage

Zero-voltage switching (ZVS) occurs

Tank input impedance
Operation above
resonance: tank input || Zi ||
1
impedance Zi is C
L
dominated by tank
inductor.
∠Zi is negative, and f0
R0
tank input current lags
Qe = R0 /Re
tank input voltage. Re
Zero crossing of the
tank input current
waveform is(t) occurs
after the zero crossing
of the voltage vs(t).

Switch network waveforms, above resonance
Zero-voltage switching
+
Q1 Q3 L C
vds1(t) D1 D3
+
iQ1(t) –
vs(t)
– is(t)
Q2 Q4
D2 D4
Conduction sequence: D1–Q1–D2–Q2

Q1 is turned on during D1 conduction
interval, without loss

ZVS turn-off transition: hard switching?
+
Q1 Q3 L C
vds1(t) D1 D3
+
iQ1(t) –
vs(t)
– is(t)
Q2 Q4
D2 D4
When Q1 turns off, D2 must begin

conducting. Voltage across Q1 must
increase to Vg. Transistor turn-off transition
is identical to hard-switched PWM.
Switching loss may occur (but see next
slide).
Soft switching at the ZVS turn-off transition
•  Introduce small
capacitors Cleg across
each device (or use
device output
capacitances).
•  Introduce delay
between turn-off of Q1
and turn-on of Q2.
Tank current is(t) charges and
discharges Cleg. Turn-off transition
becomes lossless. During commutation
interval, no devices conduct.
So zero-voltage switching exhibits low
switching loss: losses due to diode
stored charge and device output
capacitances are eliminated.
19.4 Load-dependent properties
of resonant converters
Resonant inverter design objectives:
1. Operate with a specified load characteristic and range of operating

points
•  With a nonlinear load, must properly match inverter output
characteristic to load characteristic
2. Obtain zero-voltage switching or zero-current switching
•  Preferably, obtain these properties at all loads
•  Could allow ZVS property to be lost at light load, if necessary
3. Minimize transistor currents and conduction losses
•  To obtain good efficiency at light load, the transistor current should
scale proportionally to load current (in resonant converters, it often
doesnʼt!)

Inverter output characteristics
Let H∞ be the open-circuit (R∞)

transfer function:
v(s)
H (s) =
vs1(s) R
and let Zo0 be the output impedance

(with vi short-circuit). Then,
v(s) = H (s)vs1(s) R This result can be rearranged to obtain
R + Z o0(s)
2 2 2 2 2
v( js) + i( js) Z o0( js) = H ( js) vs( js)
The output voltage magnitude is:
2 2
2 H ( js) vs( js) Hence, at a given frequency, the
v( js) =
Z o0( js)
2 output characteristic (i.e., the
1+ relationship between || v || and || i ||) of
R2
any resonant inverter of this class is
v( js)
with R= elliptical.
i( js)
Inverter output characteristics
General resonant inverter i

Inverter output
output characteristics are characteristic
H vs
elliptical, of the form I sc =
d
Z o0 oa
e d l
2 2 a tch Z o0
v( js) i( js) I sc M =
+ =1 2 R
V 2oc I 2sc
with
Voc = H ( js) vs( js)
Voc Voc = H vs v
H ( js) vs( js) Voc 2
I sc = =
Z o0( js) Z o0( js)
This result is valid provided that (i) the resonant network is purely reactive,
and (ii) the load is purely resistive.
Matching ellipse
to application requirements
Electronic ballast Electrosurgical generator
|| io || || io ||
50
inverter characteristic
inverter characteristic 2A
40
0W
d
d loa
lamp characteristic t che
ma
|| vo || 2kV || vo ||

Input impedance of the resonant tank network
Transfer function
H(s)
is(t) i(t)
Z (s) Effective
1+ R 1 + o0 sinusoidal Resonant
Z o0(s) R source network + Effective
Z i(s) = Z i0(s) = Z i(s) + resistive
1+ R Z o(s) Zi Zo v(t)
vs1(t) – load
Z o(s) 1 + R
R Purely reactive –
where
v vi vo vo
Z i0 = i Z i = Z o0 = Z o =
ii R 0 ii R
– io vi short circuit
– io vi open circuit

Zi0 and Zi∞ for 3 common inverters
Series 1
L Cs C
Z i0(s) = sL + 1
s || Zi ||
L sC s
Zi Zo || Zi0 ||
Z i(s) =
f
Parallel 1
C
L p
Z i0(s) = sL
L
Zi Zo || Zi ||
Cp
Z i(s) = sL + 1
|| Zi0 || sC p
f
LCC 1
C +
1
Cs Z i0(s) = sL + 1
s
L C
p
1
C L sC s
s
|| Zi ||
Zi Zo
Cp || Zi0 || Z i(s) = sL + 1 + 1
sC p sC s
f

Other relations
Reciprocity If the tank network is purely reactive,

Z i0 Z o0
then each of its impedances and
= transfer functions have zero real
Z i Z o
parts: Z = – Z*
i0 i0
*
Z i = – Zi
Tank transfer function Z o0 = – Z*

o0
*
H (s) Z o = – Zo
H(s) = *
1+ R H = – H
Z o0
Hence, the input impedance
vo(s) magnitude is
where H = R 2
vi(s) 1+ 2
R
2 2
Z o0
*
Z i = Z iZ i = Z i0
H
2
= Z o0 1 – 1 R 2
Z i0 Z i 1+ 2
Z o
A Theorem relating transistor current variations to
load resistance R
Theorem 1: If the tank network is purely reactive, then its input impedance
|| Zi || is a monotonic function of the load resistance R.
  So as the load resistance R varies from 0 to ∞, the resonant network

input impedance || Zi || varies monotonically from the short-circuit value 
|| Zi0 || to the open-circuit value || Zi∞ ||.
  The impedances || Zi∞ || and || Zi0 || are easy to construct.
  If you want to minimize the circulating tank currents at light load,
maximize || Zi∞ ||.
  Note: for many inverters, || Zi∞ || < || Zi0 || ! The no-load transistor current
is therefore greater than the short-circuit transistor current.

Proof of Theorem 1
Previously shown:   Differentiate:

1+ R 1 – 1 R
2 2 2
Z o0 d Zi
2 Z o0 Z o
2 2 2
Zi = Z i0 = 2 Z i0
dR 2
1+ R
2
2 1+ R
2
Z o Z o
  Derivative has roots at: So the resonant network input

impedance is a monotonic function
(i) R = 0 of R, over the range 0 < R < ∞.
(ii) R =
In the special case || Zi0 || = || Zi∞ ||, 
(iii) Z o0 = Z o , or Z i0 = Z i
|| Zi || is independent of R.

Example: || Zi || of LCC
•  for f < f m, || Zi || increases 1

C +
with increasing R . || Zi || s 1
C f f
p
•  for f > f m, || Zi || decreases L
R
ing
with increasing R .
s
rea
inc
1
•  at a given frequency f, || Zi ||
inc
C
rea
s
sin
is a monotonic function of R.
gR
•  Itʼs not necessary to draw fm
the entire plot: just construct
|| Zi0 || and || Zi∞ ||.
f

Discussion: LCC
1
C + LCC example
|| Zi0 || and || Zi∞ || both represent series 1 f0 f
s
resonant impedances, whose Bode || Zi || C
p
diagrams are easily constructed. 1 L
|| Zi0 || and || Zi∞ || intersect at frequency fm. C
s
For f < fm || Zi ||

then || Zi0 || < || Zi∞ || ; hence transistor || Zi0 ||
current decreases as load current 1
f0 =
decreases fm 2 LC s
For f > fm f = 1
2 LC s||C p
then || Zi0 || > || Zi∞ || ; hence transistor
fm = 1
current increases as load current 2 LC s||2C p
decreases, and transistor current is
greater than or equal to short-circuit f
current for all R
L Cs L Cs
Zi Zi0
Cp Cp

Discussion -series and parallel
Series 1 •  No-load transistor current = 0, both above

L Cs C
|| Zi ||
s
L
and below resonance.
Zi Zo || Zi0 ||
•  ZCS below resonance, ZVS above
f
resonance
Parallel 1 •  Above resonance: no-load transistor current

C
L p
L
is greater than short-circuit transistor
Zi Zo || Zi || current. ZVS.
Cp
|| Zi0 ||
f
•  Below resonance: no-load transistor current
is less than short-circuit current (for f <fm),
LCC 1
C + but determined by || Zi∞ ||. ZCS.
1
L Cs s
C
p
1 L
C
s
|| Zi ||
Zi Cp Zo
|| Zi0 ||

A Theorem relating the ZVS/ZCS boundary to load
resistance R
Theorem 2: If the tank network is purely reactive, then the boundary between
zero-current switching and zero-voltage switching occurs when the load
resistance R is equal to the critical value Rcrit, given by
– Z i
Rcrit = Z o0
Z i0
It is assumed that zero-current switching (ZCS) occurs when the tank input
impedance is capacitive in nature, while zero-voltage switching (ZVS) occurs when
the tank is inductive in nature. This assumption gives a necessary but not sufficient
condition for ZVS when significant semiconductor output capacitance is present.

Proof of Theorem 2
Previously shown:
Note that Zi∞, Zo0, and Zo∞ have zero
Z o0 real parts. Hence,
1+
Z i = Z i R Z
Z 1 + o0
1 + o Rcrit
R Im Z i(Rcrit) = Im Z i Re
Z
If ZCS occurs when Zi is capacitive, 1 + o
Rcrit
while ZVS occurs when Zi is inductive,
then the boundary is determined by Z o0Z o
1–
∠Zi = 0. Hence, the critical load Rcrit is R 2crit
= Im Z i Re 2
the resistance which causes the Z o
imaginary part of Zi to be zero: 1+
R 2crit
Im Z i(Rcrit) = 0 Solution for Rcrit yields

– Z i
Rcrit = Z o0
Z i0

Discussion —Theorem 2
– Z i
Rcrit = Z o0
Z i0
  Again, Zi∞, Zi0, and Zo0 are pure imaginary quantities.
  If Zi∞ and Zi0 have the same phase (both inductive or both capacitive),
then there is no real solution for Rcrit.
  Hence, if at a given frequency Zi∞ and Zi0 are both capacitive, then ZCS
occurs for all loads. If Zi∞ and Zi0 are both inductive, then ZVS occurs for
all loads.
  If Zi∞ and Zi0 have opposite phase (one is capacitive and the other is
inductive), then there is a real solution for Rcrit. The boundary between
ZVS and ZCS operation is then given by R = Rcrit.
  Note that R = || Zo0 || corresponds to operation at matched load with
maximum output power. The boundary is expressed in terms of this
matched load impedance, and the ratio Zi∞ / Zi0.

LCC example
  For f > f∞, ZVS occurs for all R. f0 f
  For f < f0, ZCS occurs for all R. || Zi || 1

ZCS ZCS: R>Rcrit ZVS
C + for all R ZVS: R<Rcrit for all R
  For f0 < f < f∞, ZVS occurs for  s 1
C L
R< Rcrit, and ZCS occurs for  1
p
C
R> Rcrit. Z i || Zi ||
s
|| Zi0 || Z i0 {
  Note that R = || Zo0 || corresponds
to operation at matched load with
maximum output power. The f1 fm
boundary is expressed in terms of
this matched load impedance,
f
and the ratio Zi∞ / Zi0.
– Z i
Rcrit = Z o0
Z i0

LCC example, continued
Zi
R 90˚
R=0
60˚
ZCS
incr
30˚
easi
ng R
R crit 0˚
ZVS -30˚
||
Z o0
||
-60˚
R=
-90˚ f
f0 fm f f0 f
Typical dependence of Rcrit and matched-load Typical dependence of tank input impedance phase
impedance || Zo0 || on frequency f, LCC example. vs. load R and frequency, LCC example.

A popular subclass of tank circuits
Tank network Tank Series branch reactance Xs Shunt branch reactance X p

jXs Series L – 1
C
Parallel L – 1
C
jXp LCC L – 1 – 1
C s C p
LLC L s – 1 L p
C
2 2
Key equations M
a + J =1 (output characteristic)
b
M= 1 (control characteristic)
Z i0 = jXs, Z i = j Xs + X p 2
1 + Qe
Xp a2 b
H () =
X p + Xs Xp
jXs X p a = H () =
Z o0() = = jXsH () X p + Xs
X p + Xs
H () Ro Ro
Z Xs b= =
Rcrit = Z o0 – i = X p – Z o0() Xs
Z i0 Xs + X p
V I R R
At f = fm: Xs = – 12 X p M = out , J = out o , Qe = o
Vin Vin Re
Important frequencies of popular tank circuits
Tank 0 m
Series 1
LC
Parallel 1 1
LC 2LC
LCC 1 1 1
LC s LC sC p LC s2C p
LLC 1 1 1
L sC Ls + L p C Lp
Ls + C
2
The LLC tank network
DC-DC converter circuit
Tank network input impedances
Zi0 = sLs + 1/sC Zi∞ = s(Ls + Lp) + 1/sC

LLC
Tank input impedance plots Zi0, Zi∞
Good
characteristics
obtained for fs > fm
•  Switch current
varies directly
with load current
•  ZVS obtained at
most operating
points including
matched load
LLC output characteristic
for fm < fs < f0
Rcrit < || Zo0 || so ZVS region includes matched load

LLC: ZVS boundary
in the region f0 < fs < f∞
Rcrit 1– F
= nF 1+n
n = 4
Ro0 1+n 2
F –1
Ls
Ro0 =
C
Lp
n=
Ls
f
F= s
f ZVS for R > Rcrit
ZCS for R < Rcrit

LLC
Conversion Ratio M vs. Switching Frequency F, Lp/Ls = 4
Qs = 0
||H∞||
fs Qs = 0.25
F=
f
0.35
f = 1
2 (L s + L p)C
R 0.5
Qs = o0
Re
Ls 0.75
Ro0 =
C 1
1.5
2
3
5
10
25
fs = f fs = fm fs = f0
LLC
Output plane characteristic, Lp/Ls = 4
fs < f∞ F= 1+n f∞ < fs < f0
F=1
Increasing F
F=1
Increasing F
F = 0.5
fs
F=
f f0 < fs
1 Increasing F
f =
2 (L s + L p)C F= 1+n
f0 = 1
2 L sC
f0 = 1 + n f
Lp
n=
Ls 1+n
fm = f
M= V 1+ n
Vg 2
IR F = 10
J = o0
Vg
Summary
1.  Simple models approximate the tank waveforms by their fundamental

sinusoidal components, and that facilitate resonant converter design
using conventional frequency response methods
2.  Several theorems show how to employ frequency response plots to
determine important properties:
•  The inverter or dc-dc converter elliptical output characteristic
•  How the transistor currents vary with load, and how to shape the
tank impedances to improve light-load efficiency
•  Dependence of the ZVS/ZCS boundary on load
3.  The LLC tank network exhibits the following advantages:
•  Buck-boost conversion ratio
•  Wide range of operating points exhibiting zero-voltage switching,
i.e., for f0 < fs or for f∞ < fs < f0 with Re > Rcrit
•  Transistor currents vary directly with load current for fm < fs
Selected References
1.  R. L. Steigerwald, “A Comparison of Half-Bridge Resonant Converter

Topologies,” IEEE APEC, 1987 Record, pp. 135-144.
2.  S. Johnson, “Steady-State Analysis and Design of the Parallel
Resonant Converter,” M.S. Thesis, Univ. of Colo., 1986.
3.  R. Erickson and D. Maksimovic, Fundamentals of Power Electronics,
2nd ed., Chapter 19, Springer, 2001.
4.  B. Yang, F. Lee, A. Zhang, and G. Huang, “LLC Resonant Converter
for Front-End DC/DC Conversion,” IEEE APEC, 2002 Record, pp.
1108-1112.

PELSDenverSlides PDF

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

PELSDenverSlides PDF

Uploaded by

Copyright:

Available Formats

Resonant Power Conversion

Prof. Bob Erickson

1. Introduction to resonant power conversion

Resonant power converters contain resonant L-C networks whose

Fundamentals of Power Electronics 2 Chapter 19: Resonant Conversion

Switch network Resonant tank network

Several resonant tank networks

Series tank network Parallel tank network LCC tank network

Fundamentals of Power Electronics 3 Chapter 19: Resonant Conversion

Fundamentals of Power Electronics 4 Chapter 19: Resonant Conversion

Rectify and filter the output of a dc-high-frequency-ac inverter

Fundamentals of Power Electronics 5 Chapter 19: Resonant Conversion

The chief advantage of resonant converters: reduced switching loss

Fundamentals of Power Electronics 9 Chapter 19: Resonant Conversion

A resonant dc-dc converter: Transfer function

If tank responds primarily to fundamental component of switch

Fundamentals of Power Electronics 11 Chapter 19: Resonant Conversion

Fundamentals of Power Electronics 12 Chapter 19: Resonant Conversion

If the switch network produces a The fundamental component is

Fundamentals of Power Electronics 13 Chapter 19: Resonant Conversion

Assume that switch network T s/2

It is desired to model the dc 2 I s1 sin (s –  s)d

• Switch network converts dc to ac

Fundamentals of Power Electronics 15 Chapter 19: Resonant Conversion

iR(t) | iR(t) | i(t) V vR(t)

vR(t) v(t) R iR(t) s t

Assume large output filter If iR(t) is a sinusoid:

Fundamentals of Power Electronics 16 Chapter 19: Resonant Conversion

Again, since tank responds only to fundamental components of applied

Actual waveforms with harmonics ignored

Fundamentals of Power Electronics 17 Chapter 19: Resonant Conversion

iR(t) | iR(t) | i(t)

Fundamentals of Power Electronics 18 Chapter 19: Resonant Conversion

vR1(t) 8 V Rectifier equivalent circuit

With a resistive load R, this becomes

Fundamentals of Power Electronics 19 Chapter 19: Resonant Conversion

Model of ac waveforms is now reduced to a linear circuit. Tank

Fundamentals of Power Electronics 20 Chapter 19: Resonant Conversion

Transfer function: Transfer function

Solution for tank output current:

which has peak magnitude

Fundamentals of Power Electronics 21 Chapter 19: Resonant Conversion

Fundamentals of Power Electronics 22 Chapter 19: Resonant Conversion

So we have shown that the conversion ratio of a resonant converter,

Fundamentals of Power Electronics 23 Chapter 19: Resonant Conversion

Fundamentals of Power Electronics 24 Chapter 19: Resonant Conversion

transfer function H(s)

Fundamentals of Power Electronics 26 Chapter 19: Resonant Conversion

Fundamentals of Power Electronics 27 Chapter 19: Resonant Conversion

Fundamentals of Power Electronics 28 Chapter 19: Resonant Conversion

Fundamentals of Power Electronics 29 Chapter 19: Resonant Conversion

is(t) iR(t) i(t)

Differs from series resonant converter as follows:

Fundamentals of Power Electronics 30 Chapter 19: Resonant Conversion

I iR(t) iR(t) i(t)

For a resistive load, V = IR. The effective resistance Re can then be

Fundamentals of Power Electronics 32 Chapter 19: Resonant Conversion

Fundamentals of Power Electronics 33 Chapter 19: Resonant Conversion

transfer function H(s)

Fundamentals of Power Electronics 35 Chapter 19: Resonant Conversion

Fundamentals of Power Electronics 36 Chapter 19: Resonant Conversion

Fundamentals of Power Electronics 37 Chapter 19: Resonant Conversion

1.  Introduction to resonant power conversion

It is desired to model the dc 2 I s1 sin (s – s)d

•  Switch network converts dc to ac

  So as the load resistance R varies from 0 to ∞, the resonant network

Previously shown:   Differentiate:

  Derivative has roots at: So the resonant network input

•  for f < f m, || Zi || increases 1

Series 1 •  No-load transistor current = 0, both above

Parallel 1 •  Above resonance: no-load transistor current

  For f > f∞, ZVS occurs for all R. f0 f

  For f < f0, ZCS occurs for all R. || Zi || 1

1.  Simple models approximate the tank waveforms by their fundamental

1.  R. L. Steigerwald, “A Comparison of Half-Bridge Resonant Converter