Professional Documents
Culture Documents
Dynamic Modeling of The Dual-Active Bridge
Dynamic Modeling of The Dual-Active Bridge
net/publication/4359117
Conference Paper in PESC Record - IEEE Annual Power Electronics Specialists Conference · July 2008
DOI: 10.1109/PESC.2008.4591971 · Source: IEEE Xplore
CITATIONS READS
34 776
2 authors:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by H.-P. Nee on 04 June 2014.
Abstract—In the present paper the staedy-state and the ZVS conditions. Simultaneously, the snubber capacitor,
dynamic behavior of the Dual-Active Bridge topology have which is connected across the transistor TA − , will be
been studied. The small-signal model of the converter has discharged and will eventually force the diode D A − to be
been derived and theoretical and experimental results are forward-biased. Therefore, the inductor current starts
presented.
flowing through the diode D A − and diodes Ds1 and Ds 4 ,
The Dual-Active Bridge is an attractive solution for high-
power applications where a bi-directional operation is as shown in Figure 1(d).
needed. Similarly, the transistor TA − is turned on at ZVS as
shown in Figure 1(e) and Figure 1(f), and the
I. INTRODUCTION commutation sequence will be repeated as above.
The Dual-Active Bridge (DAB) topology which was TA + DA + Cs Ts1 Ds1 Ts 3 Ds 3
+Vdc
first presented by Kheraluwala et.al in [1], [2], and [3] Ideal T / F
consists of two switch-mode active bridges, one operating Lσ C R
in the inversion mode and the other in the rectification
mode. The ac-terminals of the bridges are interconnected −Vdc
TA − DA − Cs Ts 2 Ds 2 Ts 4 Ds 4
by means of a high-frequency transformer, enabling
power flow in both directions as shown in Figure 1. Each (a)
bridge can be controlled to generate a high-frequency TA + DA + Cs Ts1 Ds1 Ts 3 Ds 3
square-wave voltage at its transformer terminals +Vdc
Ideal T / F
( ±Vdc ,±V0 ). By incorporating a controlled amount of Lσ C R
M=1
In Figure 2 the input voltage, the output voltage and, 0.7 M=0,8
0.6
Vdc + Vdc
0.4
0
ωt
− Vdc 0.3
V0 + V0 0.2
0.1
0 ωt
0
− V0 0 0.5 1 1.5 2 2.5 3 3.5
Phase shift angle in rad
458
2 Vdc0
the Single-Active Bridge topology [4] similar dynamic
A11 = − characteristics should be expected. This implies that the
Ts ( Vdc 0 + x 20 ) dynamics of the DAB topology are predictable and non-
oscillatory. The phase-margin of the system is 35 degrees
§ Vdc 0 Ts ( ( φ0 − 3 π ) Vdc 0 − 2 x 20 π )+ · and the cross-over frequency is f c = 4.8 k Hz .
¨ ¸
¨ π Ts Lσ ( Vdc0 + x 20 )2 ¸
¨ ¸ B. Small-signal control-to-state transfer function
¨ ¸
¨ ¸ (11) The small-signal control-to-state transfer function for
1¨ 4 π Lσ x10 Vdc 0 − π Ts x 20 2 ¸
A12 = + the DAB converter is defined in Eq. (18).
2¨ π Ts Lσ ( Vdc 0 + x 20 )2 ¸
¨ ¸ ~
¨ ¸ ) x1 ( s
¨ ¸ Gsd ( s ) = = [ s I − A0 ]−1 [ E ]
~
(18)
¨ ¸ φ ( s ) U (s )= 0 ~
¨ ¸
© ¹
N (s )
Gsd ( s )= G sd
(19)
x 20 ( Ts φ 0 x 20 − 2 π Lσ x10 ) + π Ts Vdc0 ( Vdc0 + 2 x 20 ) (12) Δ(s )
B11 =
π Ts Lσ ( Vdc 0 + x 20 )2 where
x20 x20 § ·
E11 =
π Lσ ( M + 1 )
(13) N G sd ( s )= ¨ s+
1
¸ (20)
π Lσ ( M + 1 ) ¨© R C ¸
¹
~ ~
where x0 , Vdc 0 , and D are the dc terms and x , V dc , and
Bode Diagram
~ 30
d are the signal-frequency ac terms.
20
Magnitude (dB)
A. Small-signal control to output transfer function
10
The control-to-output transfer function of the DAB
topology can be defined as 0
~
x2 ( s
) -10
= [ s I − A0 ]−1 [ E ] 45
Gvd ( s ) = ~
(14)
φ(s ) ~
U (s )= 0 0
Phase (deg)
N Gvd (s)
Gvd ( s )= (15)
Δ(s ) -90
10
1
10
2 3
10
4
10
5
10
6
10
= [ s I − A0 ]−1 [ B0 ]
40
Gvg ( s ) = ~
(21)
U(s )
20
~
φ =0
Magnitude (dB)
0
Thus,
(s)
-20
N Gvg
-40 Gvg ( s )= (22)
Δ s)
(
-60
0
The numerator of the small-signal transfer function
-45 corresponds to
Phase (deg)
-90
N Gvg ( s ) = B11 (23)
-135
C
-180
1 2 3 4 5 6
10 10 10 10 10 10
Frequency (rad/sec)
459
function, a LHP zero is present which is forcing the phase
20
Bode Diagram
angle to overshoot the frequency axis and increase the
0
magnitude of the source-to-state transfer function. Similar
characteristics where observed in the SAB topology.
Magnitude (dB)
-20
-40
IV. CONTROL AND REGULATION: SMALL-SIGNAL
-60
CONTROLLER
-80
-90
is introduced. As stated previously the maximum power
can be achieved for a phase-shift angle equal to 90
-135
degrees. Minimum power can be achieved at phase-shift
-180 angle which equals zero. Minimum power or zero output
1 2 3 4 5 6
10 10 10 10 10 10 power does not imply zero current. The current is still
Frequency (rad/sec)
circulating through the devices causing losses.
Figure 6: Small-signal source-to-output transfer function. Bode Consequently, zero active power does not correspond to
diagram. zero reactive power. If the duty ratio is kept constant, it is
As shown in Figure 6, the phase margin of the source- not possible to achieve zero reactive power if the output
to-output transfer function is adequate, 145 degrees, power is zero.
implying that the stability criteria are fulfilled. On the
other hand, and at low frequencies, small variations In order to study and design a controller for the DAB
around the steady-state value of the input voltage are topology the small-signal transfer functions are employed.
slightly amplified in the output of the converter. In the The control configuration is identical to the case of the
SAB topology the small-signal variations where damped SAB converter. Thus, two PI regulators are connected in a
resulting in lower filter capacitance. As predicted by cascade scheme with two loops, a voltage loop and an
Figure 6, the 300 Hz ripple should be slightly amplified. inner current loop as shown in Figure 8. The states of
interest are the inductor current and the output voltage.
D. Small-signal source-to-state transfer function ~
V dc Gvg
~ +
x1 ( s
)
Control-to-output +
= [ s I − A0 ]−1 [ B0 ]
Σ
Gsg ( s ) = (24)
~
V0
~
U(s)
~
V0
~ −
φ =0
~
Vref I ref φ
+ Σ PI Σ PI
+
−
N G sg(s) Control-to-state +
Σ
Gsg ( s )= (25) +
Δ(s ) G sd
Control-to-state
Where, ~
V dc Gsg
§ ·
N G sg ( s ) = B11 ¨¨ s + 1 ¸
¸ (26) Source-to-state
© RC ¹
Power Stage
Bode Diagram
-10
460
V. MEASUREMENTS
Small-signal variations around the reference voltage
600 A 10 kW prototype has been built and tested in the
Voltage in volts 400 laboratory. The specifications of the converter are
200
summarized in Table 1.
0
60
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Input voltage (pole-to-pole) Vd 400[Volts]
Current in A
40
Output Power P0 >8 [kW]
20
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Switching frequency fs 10 [kHz]
30
Voltage in volts
In Figure 10 a 300 Hz small-signal ripple is added to At nominal power the phase-shift angle should be 90
the dc-link voltage. As predicted by the small-signal degrees. Due to load power handling limitations the
source-to-output transfer functions, small-signal phase-shift angle was limited to values less than 90
perturbations added to the line voltage are amplified and degrees during measurements, and the nominal power
are added to the output voltage of the converter. Despite delivered to the load was 10 kW. Consequently, the
the inherent property of the converter to amplify the converter can deliver considerably higher power than 10
perturbations, the controller should attenuate the small- kW at a phase-shift angle of 90 degrees. Additionally, the
signal variations to acceptable levels. As shown in the converter was operated as a step-down converter and at
figure the output voltage is oscillating with 300 Hz but the certain loads as a step-up converter for demonstration
ripple is only 1% of the nominal voltage. reasons.
Small-signal line variations In Figure 12 the inductor current and the secondary
snubber capacitor voltage are shown. Thus, at i Lσ ( φ ) the
Voltage in volts
384
382
secondary placed controllable switches are turned-off and
380
0.8 0.805 0.81 0.815 0.82 0.825 0.83 0.835 0.84 0.845 0.85 the snubber capacitors are charging up, forcing the
15
antiparallel diode to commutate the current.
Current in A
5 20
0.8 0.805 0.81 0.815 0.82 0.825 0.83 0.835 0.84 0.845 0.85
Current in A
40
Voltage in volts
20 0
0
-20 -20
-40
0.8 0.805 0.81 0.815 0.82 0.825 0.83 0.835 0.84 0.845 0.85
Time in seconds 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
-4
x 10
integral part of the voltage controller is forcing it to follow Figure 12: Measured waveforms. Primary current and snubber
the reference value. The current instantly increases in capacitor voltage
order to cope with the phase-shift variation but returns In order to fulfil the soft-switching constraints, the
back to its steady-state value with a certain derivative. controllable switches must turn off a finite current.
Phase-shift angle variations can neither cause instabilities Additionally, the phase-shift angle must be limited within
nor oscillations to the system assuring that the system is a certain range. By examining Figure 12 it is obvious that
robust and well-defined. the secondary placed controllable switches are turned-off
Small-signal phase-shift angle step
under ZVS conditions resulting to almost negligible
400
switching losses. At turn-off, and when the capacitors are
Voltage in volts
10
5 winding capacitance.
0
0 0.1 0.2 0.3 0.4 0.5 0.6 The inductor current and voltage at a conversion ratio
5
0,86 are shown in Figure 13. As discussed in [4]
Angle
461
the oscillating current is flowing through the snubber flowing through the magnetising inductance and the
capacitors implying that are part of the resonance circuit. winding capacitance of the transformer.
Nevertheless, the different frequencies appears in the Measured Secondary Current at M=0.96 and R=16
40
current, and as a result two different resonance modes are
present. 30
Current in A
10
0 0
Current in A
-20 0
-50
0.5 1 1.5 2 1.25 1.3 1.35 1.4
-4 -4
x 10 x 10
30 -10
20 20
10
Current in A
Current in A
10
-20
0 0
-10 -10
Voltage in volts
200
-100 Time in Seconds -4
x 10
0 -200
-300
-200
-400
-400
-500
Figure 15: Measured secondary current at nominal load
0.5 1 1.5 2 1.25 1.3 1.35 1.4
Time in seconds -4 Time in seconds -4
x 10 x 10
Measured Primary and Secondry Current and the difference (ip-is) at M=0.96 and R=16
40
Current in A
20
Figure 13: Measured inductor current, snubber current, and 0
inductor voltage -20
-40
In addition very high frequency oscillations are taking 40
0 1
-4
2
x 10
place when the charging current is reaching the peak
Current in A
20
0
Measurements have also been performed at a higher
conversion ratio as shown in Figure 14. As shown in the -10
0 1 2
figure the oscillations are not affected by the conversion Time in seconds
x 10
-4
Current in A
0 0
-20
The DAB converter has the ability to operate in the
0.5 1 1.5 2
-50
1.25 1.3 1.35 1.4
step-up mode. In this mode of operation the conduction
time of the primary placed diodes corresponds to the
-4 -4
x 10 x 10
40
20
20
conduction time of the primary placed controllable
Current in A
Current in A
0
0
Voltage in volts
200
-100
0 -200
-200
-300
-400
the secondary placed controllable switches is considerably
-400
0.5 1
Time in seconds
1.5
-4
2
-500
1.25 1.3
Time in seconds
1.35 1.4
-4
increased in the step-up mode. Specifically, the
x 10 x 10
conduction time of the secondary placed controllable
switches corresponds to the conduction time of the
Figure 14: Measured inductor current, snubber current, and
inductor voltage at M=0,96 secondary placed diodes when the converter operates in
the step-down mode. This implies that the conduction
When the secondary placed controllable switches are losses of the secondary bridge are higher in the step-up
turned off the current will not commutate to the diode. mode than in the step-down mode.
The current is clamped at zero. This due to the reversal of For clarification reasons, when the converter operates in
the voltage across the winding capacitance, thus, the the step-down mode the conduction time of the primary
winding capacitance is charged and/or discharged. This is placed controllable switches is considerably higher than
clearly shown in Figure 15. Thus, oscillations are present the conduction time of the diodes. For the secondary
only when the secondary placed controllable switches are bridge, the conduction time of the diodes is considerably
turned off. higher than the conduction time of the controllable
In Figure 16 the primary and the secondary currents are switches at step-down. The measurements in the step-up
shown. The difference between the primary and the mode were carried out just to demonstrate the ability of
secondary current is shown in the last plot. As shown in the DAB converter to boost the output voltage within
the figure the oscillations occur during the commutation reasonable limits with maintained high efficiency.
of the secondary placed controllable switches. Assuming In both Figure 17 and Figure 18 the inductor current
that the primary and the secondary current are in phase the and voltage and, the charging/discharging snubber
difference of the two currents should represent the current capacitor current are shown. The converter has the ability
to boost the output voltage corresponding to 1.4 times the
462
input dc link voltage. As in the step-down operation the therefore, the conduction losses are considerably higher
oscillations are present in the step-up mode as well. The than in the step-down mode.
oscillations are expected to be damped at light loads and Output power versus efficiency
96
amplified at heavy loads as already discussed.
95
Measured Inductor Current and Voltage at M=1.4 and R=25
M=0.96
40
94
20
Current in A
93
M=1
Efficiency in %
0
92
-20
91
-40
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
-4
90
x 10
89
M=1.4
Voltage in volts
200
88
0
87
3000 4000 5000 6000 7000 8000 9000 10000 11000
-200 Output power in watts
Current in A
0 0
Current in A
0 20
-50
0
-20
rapidly decreased by 10%. In both cases the system is
0.5 1 1.5 2
-40
1.4 1.42 1.44 1.46 1.48 1.5
neither unstable nor oscillatory. The voltage is increased
and/or decreased monotonically as governed by the
-4 -4
x 10 x 10
200
200
controller, in order to reach the new reference value.
Voltage in volts
Voltage in volts
100
0
0 + 10% Step
-200
-100
0.5 1 1.5 2 1.4 1.42 1.44 1.46 1.48 1.5
Time in seconds -4 Time in seconds -4 20
Current in A
x 10 x 10
VI. EFFICIENCY MEASUREMENTS 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045 0.05
300
0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045 0.05
The efficiency of the converter at different conversion Time in seconds
efficiency. This is also true for low power levels where the
soft-switching conditions are fulfilled. For high powers 350
Voltage in volts
300
the peak current is considerably higher implying increased 250
contributing to a decrease in the efficiency of the 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045 0.05
Time in seconds
converter. When the converter is operating as a step-up
converter the peak value of the inductor current is (b)
considerably higher, Figure 18, than in the step-down Figure 20: Dynamic response at heavy load, (a) + 10% and (b) -
mode. At the specified load the peak inductor current is 10%
1,8 times higher in the step-up mode than in the step-down
mode. Consequently, the averaged switch and diode Similarly, in Figure 21 the dynamic behaviour of the
current is 1,8 higher than in the step-down operation, and converter operating at light load is presented. The
converter behaves in an identical way as in the heavy load.
463
-10% Step 3. Third window in the figures: Measured and reference
20
value for the phase-shift angle
10
Current in A
0
4. Last window in the figures: Alive signal, indicating
-10
that both bridges are synchronised and are operating
-20
0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045 0.05 VIII. CONCLUSIONS
The Dual-Active Bridge employing a conventional
350
phase-shift control strategy has been analysed and studied.
Voltage in volts
300
250 Both the steady-state and the dynamic behaviour of the
200
150
converter has been analysed and presented. Converter
100 dynamics have been presented by using a state-space
0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045 0.05 averaging technique based on the symmetry assumption.
Time in seconds
Additionally, the small-signal model of the converter has
Figure 21: Dynamic response at light load for -10% step been derived. As a result the dynamic behavior of the
converter and the system in general has been studied. It
In Figure 22 and in Figure 23 the performance of the has been shown that both the power plants and the whole
controller during voltage reference steps is presented. In system including the controller are stable.
both cases the controller is neither unstable nor Measurements have shown that the oscillations are
oscillatory. present even in the DAB topology. However, these
450
vin (r); vouRef (b); vout (g)
oscillations are a result of a resonance between the
400 inductor and the winding capacitance of the transformer.
350
300
The measured efficiency of the converter is varying
0.05 0.1 0.15
iDes (r); idc (b)
0.2 0.25 0.3
between 93% and 95%.
60
0
steps was studied in the laboratory and the results are
x 10
-5 0.05 0.1 0.15 0.2
phiTDes (r); phiTDesLim (b); phiTMeas (g); t2Per (c)
0.25 0.3
presented in this chapter. The system was proven to be
2
1.5 stable and that it can deal successfully with steps in the
0.5
1
reference voltage without causing any instabilities.
0.05 0.1 0.15 0.2 0.25 0.3
6
REFERENCES
opmode (r)
300
power applications”, IEEE Transc. on industry applications, Vol.
0.05 0.1 0.15
iDes (r); idc (b)
0.2 0.25 0.3 27, No. 1, January/February 1991
60
40
[4] G. D. Demetriades, “ On small-signal analysis and control of the
20
0
single- and the dual-active bridge topologies”, Doctoral
-20
-40
Dissertation, Royal Institute of Technology, Sweden, ISBN-91-
20
x 10
-6 0.05 0.1 0.15 0.2
phiTDes (r); phiTDesLim (b); phiTMeas (g); t2Per (c)
0.25 0.3
7283-966-x, 2005.
10
[5] G. D. Demetriades, Hans-Peter Nee, “Evaluation of different soft-
switching topologies for high-power applications”, NORPIE 2003,
0
0.05 0.1 0.15 0.2 0.25 0.3
Stockholm, Sweden
6
[6] D. M. Mitchell, “DC-DC switching regulator analysis”, Published
opmode (r)
4
by DMMitchel Consultants, ISBN 0-07-042597-3
2
0
0 0.05 0.1 0.15 0.2 0.25 0.3
Time (s)
464