Transcript Nieizotermiczne charakterystyki impulsowych układów zasilających

Modelling the Half-bridge Dc-Dc
Converter with Selfheating Taken
into Account
Małgorzata Godlewska, Krzysztof Górecki
Department of Marine Electronics, Gdynia Maritime
University, Poland
Outline




Introduction
Models of Components of the Converter
Investigations Results
Conclusions
Introduction






In the switch-mode power supply systems requiring galvanic isolation
between the input and the output half-bridge dc-dc converters are often
used.
An important component of such a converter is a pulse transformer with
a split secondary winding and semiconductor devices.
In the analysis and design of electronic circuits a computer simulations
that require computer models of all components of the analysed system
are commonly used.
These models should take into account all relevant phenomena in each
of these components, and at the same time should be simple enough to
enable obtaining the results of calculations rapidly.
Therefore, many models of electronic components dedicated to
different applications and characterised by varying accuracy are
presented in the literature.
This paper presents the results of simulations and measurements of the
half-bridge dc-dc converter taking into account self-heating phenomena
in semiconductor devices and in the transformer.
Models of Components of the
Converter


The most important element of the converter is a transformer.
In its model, three blocks are distinguished:



Core Model,
Winding Model
Thermal Model.
H
B
C1
 The model of the core is
based on the modified JilesAtherton model.
 In this model the magnetic
force corresponds to the
voltage on the terminal H.
 The magnetic flux density
corresponds to the voltage at
the output terminal B.
 Power losses in the core are
equal to the voltage at
terminal Ploss.
EH
Ma
G1
CR
C
m
A
Em
EA1
Ealf EC
D1
C3
E4
RC
R2
E5
R1
a
M
C2
Ploss
C4
R1
C5 E
DB1
E11
R3
EP
Core Model
1a
Winding Model
Cthun
1b
RS1
ERS1
EV1
Vl1
Vl11
Rthun
GR
pthur1 pthu
2a
3a
ERS3
Vl2
Vl3
EV2
EV3 3b
Cthr1
....
2b
RS3
Rthu1
Cthrn
GL1
ERS2
TU
RR
ERMS1
RS2
Cthu1
....
Rthrn
TR
Rthr1
Ta
Thermal Model
pthru1 pthr
Models of Components of the
Converter (2)







The winding model includes three circuits representing respectively: the
primary winding and two secondary windings.
In the model of the primary winding the resistor RS1 represents the series
resistance of the winding at the reference temperature, the controlled voltage
source ERS1 describes resistance changes with the change of temperature TU.
The controlled voltage source EV models the voltage induced in the primary
winding; the controlled current source GL1 represents the magnetising current,
the controlled current source GR - energy losses in the core.
Voltage VRMS1 is used to designate peak magnetisation and is calculated by
means of the controlled voltage source ERMS1.
The secondary winding model contains only elements used for modelling the
voltage winding (EV) and the series resistance of that winding (RS, and ERS)
The thermal model allows the calculation of the core temperature TR and the
winding temperature TU taking into account self-heating and mutual thermal
coupling between the core and windings.
The MOS transistor and the diode are described with the use of linear hybrid
electrothermal models.
Investigations results
half-bridge dc-dc converter

Vin
• The examined converters
containing the sequence
C
toroidal :
• powder core RTP26.9x14.5x11 made of the
material T106-26,
• ferrite core RTF-25x15x10
made of the material F-867
C
• the nanocrystalline core
RTN-26x16x10 made of the
material M-070.
• Each transformer includes three
windings of 20 coils made of
enamel copper wire of the
diameter of 0.8 mm
1
L1
R1
T1
L2
Ut1
R2
L3
D1
2
D2
T2
L4
Vt2
C0
R0
Uout
Investigations results (2)



the converter containing the transformer with the RTP core f = 200
kHz, d = 0.3 Vin = 25 V.
 The measured watt-hour
efficiency reaches the
maximum 0.72 for
R0 = 20 Ω, and then
decreases to 0.03 with
resistance R0 = 10 kΩ.
Good agreement between the results of measurements and
calculations for the authors’ model of the transformer.
For the model of an ideal transformer large discrepancies (even
more than 25%) with the results of measurements and calculations.
Investigations results (3)




the converter containing the transformer with the RTF or RTN core.
the measured maximum watt-hour efficiency was 0.83 for R0 =50Ω.
The RTF core is characterised by the lowest value of the saturation
flux density and the large surface magnetisation curve.
Hence, the results are largely in line with the measurements - both
for simulation of the authors’ transformer model and the model
describing the linear transformer.
Investigations results (4)




the converter with the transformer containing the core RTP and the
RTN at f = 200 kHz, and Vin = 25 V.
The maximum value of the measured Uwy = 12.46 V is for the powder core
at R0 = 10 kΩ and 15.3 V at R0 = 10 kΩ - for the nanocrystalline core.
Big discrepancies between the results of measurements and simulations for
small values of R0 (up to 100% for load resistance R0 = 5 Ω).
This reflects the fact that in the analysis of component of power losses in
the considered system was neglected, which is important at high output
currents.
Investigations results (5)
the temperature dependence of the transformer core and windings
for each of the three considered cores at:
f = 200 kHz and Vin = 25 V
f = 100 kHz and Vin = 60 V.



For f = 200 kHz the highest value of temperaturę was obtained for
the powder core. It is equal to 36.4 °C at R0 = 10 kΩ.
For f = 100 kHz the powder core reaches the highest temperature up to 100°C for load resistance of 200Ω
Conclusions



The paper proposes a method of modelling the half-bridge dc-dc
converter with thermal phenomena in the semiconductor and
magnetic devices taken into account.
The form of the used electrothermal models of the considered
components of the converter is presented and their usefulness was
verified for the system containing transformers with powder, ferrite
and nanocrystalline cores, the power MOS transistors and Schottky
diode.
The simulations and experiments results show that the use of the
authors’ electrothermal models is more accurate for modelling the
considered converter than the classical models of the system
components and that also allows determining the internal
temperatures of that converter components.
Conclusions (2)


A particularly significant advantage of the electrothermal model of
the transformer over the linear model of this element are observed
for dc-dc converter containing the transformer with powder core.
The presented model of components of the considered converter
can be successfully used to simulate other switch-mode power
supplies.