Vous êtes sur la page 1sur 6

THE APPLICATION OF THE HALF BRIDGE DUAL CONVERTER TO PHOTOVOLTAIC APPLICATIONS Quan Li, Peter Wolfs and Steven

Senini Central Queensland University Abstract The half bridge dual converter is a current sourced converter topology that has been previously developed from the half bridge converter. This paper examines some of the unique properties of this converter which make it suitable for lower voltage higher current applications. These conditions often arise in photovoltaic applications. An example of the use of this topology as an 85W, low voltage photovoltaic input, 360Vdc output converter for use in an inverter application is presented. 1. INTRODUCTION effective inverter design. To achieve a compact high efficiency design, many designers opt to use a DC-DC converter to translate the output voltage of a module, which is typically between 12Vdc and perhaps 48Vdc, to a level that allows direct inversion to produce AC, typically 360Vdc. The DC-DC converter represents a significant design challenge, especially in regard to obtaining efficiencies in the mid to high 90% range. This paper will examine a DC-DC converter topology that is expected to offer some advantages in this application. Encouraging experimental results have been achieved with a hard-switched 100kHz version of the converter. Several opportunities for further development are highlighted. 2. THE HALF BRIDGE DUAL CONVERTER

Photovoltaics (PVs) are currently being produced at approximately 200 MWp annually, as impressive as the figure may sound, momentary reflection shows that this is still a small part of the global energy need. Fortunately photovoltaic usage continues to grow rapidly. If the current growth rates of 20% and better annually continue, photovoltaics will be a significant energy source within a few decades, [1]. While the traditional markets of remote area power (RAPS) and telecommunications remain important, a new area of growth is grid interactive PV, [2]. While large centralised systems have been constructed, smaller systems appear to be growing in importance. These tend to be modestly sized arrays, typically a few kilowatts that generate directly into the power system. A recent local example is the solar school program, promoted by Stanwell Corporation that aims to place 5kW grid interactive arrays into 200 Queensland schools. Smaller grid interactive systems can make use of at least two technologies: Arrays with medium string voltages and grid interactive inverters in the low kilowatt range Panel sized inverters, or module integrated converters/inverters (MICs).

The half bridge dual converter is a current sourced converter topology which was developed from the more conventional voltage sourced half bridge converter using duality theory, [4]. Figures 1 and 2 respectively show the half bridge converter and its dual topology.

E/2 E

Q1 R V

The panel and module integrated inverters have sub kilowatt ratings, typically 50-200W, and convert the output of a module directly into AC. A grid interactive system of any size is developed by simply connecting the integrated modules and inverters to the AC system, [3]. While this approach brings the benefits of a simplified modular design, the designer of the module integrated inverter faces significant challenges in developing a small rating, but high efficiency, high reliability, cost

E/2

Q2

Figure 1. The Half Bridge Converter As the half bridge dual converter has been developed using duality theory, every attribute of the half bridge converter exists, but in a dual form in its descendent, [4]. Some key points are: The half bridge converter uses two capacitors to divide the supply voltage into two. Alternate

closures of the switches apply either +E/2 or E/2 across the transformer primary. The dual converter uses two inductors to divide the available supply current into two. Alternate switch openings direct either +I/2 or I/2 to flow through the primary. At least one switch is open at all times in the half bridge coverter, in its dual at least one switch is closed. The closure of both switches is destructive for the half bridge, causing a short circuit of the voltage source input, in the dual converter opening both switches is destructive causing a open circuit of the input current source. The half bridge is buck derived, the dual is boost derived. The output rectifier stage in the half bridge converter uses a current stiff inductive filter, the dual filter uses a voltage stiff capacitive filter. While the half bridge converter is suited to higher DC bus voltages, the dual converter is suited to higher DC bus currents.

Q1

Q2

Figure 3. The Push-Pull Converter The relative advantages offered by the dual converter include lower primary side average and RMS switch currents and less stress for the output rectifiers. For the dual converter, the primary switches carry either the input current I or I/2 when both devices overlap. The transformer turns ratio is selected to give 50% duty cycle at the highest input voltage. The boost nature of the dual converter allows lower voltage inputs to be covered by increasing the Mosfet overlap time. The peak device voltage stress is twice the maximum input voltage. In the push-pull circuit the transformer turns ratio needs to be selected to cover the lowest allowable input voltage. Under normal conditions the Mosfets must work at less than 50% duty cycle, this raises the peak current further increasing the switch current ratings. A two to one input voltage range, for example results in peak stresses that are twice the average input current. The voltage stresses are equivalent to the dual converter case, twice the maximum input voltage. The output rectifiers of the push-pull converter are stressed by two factors, the high turns ratio needed if a wide input voltage range must be designed for, and reverse recovery. The output rectifiers for the dual converter work into a capacitive filter, this clamps the reverse voltage to 360Vp, under all conditions. No additional snubbers are required to cope with the reverse recovery of the rectifier diodes. A bonus is that the dual design typically allows, faster recovery 600V devices to be used while a push-pull design typically requires slower 1000V devices, once allowances for voltage regulation and reverse recovery snubbers are made. The 1000V devices bring additional penalties in switching loss and forward conduction drop. A disadvantage suffered by the dual converter is a higher sensitivity to transformer primary to secondary leakage inductance. The energy stored in this inductance is dissipated in hard-switched versions of the converter. Push-pull converters are not completely immune however. Successful operation does require tightly coupled primary windings.

I/2

I/2

I Q1 Q2

Figure 2. The Half Bridge Dual Converter The last point is of significance in photovoltaic applications where the voltages tend to be low. An important implementation point is that it is generally possible to replace the input current source with a voltage source in a practical application. The two inductors act to make a voltage source appear current stiff. 3. COMPETING TECHNOLOGIES

Until now, the reflexive design choice for a 50-100W DC-DC converter, operating from a low voltage photovoltaic source would have been the centre tapped push-pull converter as shown in Figure 3.

4.

HARD SWITCHED CONVERTER SIMULATION RESULTS

In order to provide a starting point, this paper considers the design optimisation of a 100kHz hard switched converter intended to be used with a customised solar module using 44 BP solar cells and having a nominal 85W, 17Vdc, 5Adc rating. The intention is to progressively examine soft switching techniques for this converter and raise both efficiency and switching frequency. Figure 4 shows the actual experimental circuit. The simulation is performed with SIMULINK and is based on an idealised version of the experimental circuit. The device models do not include parasitic components and the primary side voltage limiting snubber, consisting of two diodes, a capacitor and zener diode is omitted from the simulation model.
176H 176H

inductance energy. A relatively low level of transformer leakage inductance, 0.5H referred to the primary side, was obtained by interleaving the primary and secondary windings, [5]. The losses in the snubber are a concern and some lossless solutions will be examined later. However for this topology the losses could be reduced by raising the clamping voltage but this would require a higher Mosfet drain rating. For the same die area this would cause some additional conduction loss due to higher Rdson. Temperature rise based estimates place the snubber losses at1.5W Figure 6 part(c) shows the primary current, during the overlap times, when both Mosfets conduct, some oscillations related to the device capacitances and transformer leakage inductance are observed. Ultimately we hope to develop lossless snubber network, or resonant switching methods that will make better use of both of these components. Figure 6 part (e) confirms that the output diodes are free from any over voltages due to reverse recovery. As the hard switched converter was intended to be a forerunner for higher frequency, low loss converters, significant effort was put into identifying the significant sources of loss. The converter was operated at slightly above rated load, with an input power of 97W and an output power of 90W. The measured loss, 7W, is in good agreement with the total thermal loss estimate of 6.5W. The measured efficiency was 93% under these conditions. Table 1 contains a list of the losses estimated by allowing the converter to reach thermal equilibrium and observing temperature rises. Ultimately, however the target efficiency for this application must be in the high ninety percent range. Further increases in efficiency can be gained in the snubber network. Elimination of the loss would raise the converter efficiency to above 94%. The other major loss areas are the magnetic components. Some improvement can be gained by selecting a better ferrite grade. The transformer used Philips 3C80, [6], significantly better grades exist. Component Transformer Inductor Mosfet Zener Diode Total Temperature Rise (K) 17 13 11 45 5 N/A Estimated Power Loss(W) 2.1 1.1 0.2 1.5 0.08 6.5

17V

20V Q1 Q2

1F

R V

Figure 4. Experimental Circuit The results of the simulation are shown in Figure 5. The leading edges of the Mosfet current waveforms shown in Figure 5 part (a) sometimes show a transient current spike. This is an artefact of the simulation models used in the SIMULINK package. An interesting result is that the Mosfet peak currents show very little ripple, in comparison to the primary currents. It can be shown that the 50% duty ratio case results in zero input current ripple for this converter. The output diode voltage waveforms are exceptionally clean, no reverse recovery voltage is seen. 5. EXPERIMENTAL RESULTS

Figure 6 contains the experimental waveforms that match the simulated results. Good agreement is obtained at the cycle by cycle time scale. As expected small variations exist at the switching time scale as the simulation model does not yet include detailed switching models. The Mosfet current, part (a) confirms that the leading edge spike noted in the simulation is indeed an artefact. The current edge is complicated, however by oscillatory currents related to the device capacitances. Figure 6 part (b) shows the Mosfet turn off voltage. This is limited by the primary side snubber network that absorbs the leakage

Table 1. Converter Losses-Room Temperature: 23C

10

10

Transformer Primary Current (A)

Mosfet Q1 Current (A)

-5

-5

-10

20 0.2

40 0.4

60 0.6

80 0.8

100 1
x 10
-4

-10

20 0.2

40 0.4 Time (s) (c)

60 0.6

80 0.8

100 1
x 10
-4

Time (s) (a)


60 50

80

Transformer Primary Voltage (V)

60 40 20 0 -20 -40 -60 -80

Mosfet Q1 Voltage (V)

40 30 20 10 0 -10 -20

20 0.2

40 0.4

60 0.6

80 0.8

100 1
x 10
-4

20 0.2

40 0.4 Time (s) (d)

60 0.6

80 0.8

100 1
x 10
-4

Time (s) (b)

Figure 5. Simulated Waveforms (a) Mosfet Q1 current (b)Mosfet Q1 Voltage


Diode Voltage (V)

0 -50 -100 -150 -200 -250 -300 -350 -400

6.

SOFT SWITCHING OPTIONS

The experimental results have highlighted the effects produced by the leakage inductance and other parasitic effects. Our future goals are to raise both switching frequency and efficiency. Three possible methods may bring improvements. These are: The use of improved construction techniques such as surface mount technology or co-axial transformer designs, [7], to reduce leakage inductance. Resonant switching, these techniques may be able to reduce switching losses and make better use of the parasitic elements.

20 0.2

40 0.4 Time (s) (e)

60 0.6

80 0.8

100 1
x 10
-4

Figure 5. Simulated Waveforms (continued) (c) Primary Current (d) Primary Voltage (e) Diode Voltage

2>

2>

2) Ch 2:

2.5Amp 10 us

2) Ch 2:

2.5Amp 10 us

(a)

(c)

1>

1>

1) Ch 1:

10 Volt 10 us

1) Math:

20 Volt 10 us

(b)

(d)

Figure 6. Experimental Waveforms (a) Mosfet Q1 current (b) Mosfet Q1 Voltage Lossless snubbers, these can recover energy stored in the leakage inductance and reduce switching losses, [4,8].

2>

The resonant switching approach can help in reducing switching losses, [9], but is problematic in that the switching device current and/or voltage stress may significantly increase. It is often difficult to maintain the resonance conditions across a wide load range. In the medium term some lossless snubber solutions may be better. Figure 7 shows a variation of the converter that includes lossless or non-dissipative snubbers that recover the leakage inductance and reduce the turn off switching losses. The operation sequence is as follows. Assume Cs is initially charged to E:

2) Math:

50 Volt 10 us (e)

Figure 6. Experimental Waveforms (continued) (c) Primary Current (d) Primary Voltage (e) Diode Voltage

L R V

E Cs Ds

Q1

+ Vcs Dr Lr

controlled by construction, to a degree, further development is required to develop energy recovery methods, such as non-dissipative snubbers. While reasonable performance has been obtained with a hard switched converter, the topology is under active study with the aim of developing variations that are suited to higher frequency, higher density and low loss applications. 8. REFERENCES

Q2

Figure 7. Converter with non-dissipative snubber Q1 turns off, and its drain current transfers to the loop, Cs, Ds and the supply E, given the initial charge on Cs, Q1 turns off with zero voltage. Cs linearly charges until the normal off state voltage is reached, normally this is above +E. The stored energy in the leakage inductance continues to drive a reducing current into Cs, the current falls to zero once the leakage inductance energy is recovered, the capacitor, and the drain voltage, is driven above the normal off state voltage. Ds blocks in the reverse, the capacitor remains in this state until Q1 conducts once more. At Q1s next conduction, a resonant current is established in Lr, Dr and Cs, this current reverses the polarity of Cs, once a voltage of E is reached, Ds conducts allow Lr to discharge into the input source E.

[1] Photovoltaic Energy for the New Millennium: the US National Photovoltaics Program Plan, SunWorld, Vol. 24, No. 1, pp. 4-5, March/April 2000. [2] D. Butler, Domestic Rooftop Systems A Cost Effective Answer to Greenhouse Emissions or a Political Sop to the Green Lobby, Energy in Central Queensland Facing the environmental Challenge, IEAust. Central Region Conference, Central Queensland University, Jun. 23-25, 2000. [3] M. Meinhardt, T. ODonnell, H. Schneider, J. Flannery, C. Mathuna, P. Zacharias and T. Krieger, Miniaturised Low Profile module Integrated Converter for Photovoltaic Applications with Integrated Magnetic Components, Proc. 1999 IEEE 14th Applied Power Electronics Conference, 1999. [4] P. J. Wolfs, A Current-Sourced Dc-Dc Converter Derived via the Duality Principle from the HalfBridge Converter, IEEE Trans. Industrial Electronics, Vol. 40, No. 1, pp. 139-144, Feb. 1993. [5] E. C. Snelling, Soft Ferrites, Properties and Applications. London: Butterworths , 1988, pp 335-338. [6] Soft Ferrites, Philips 1996 Data Handbook MA01, Netherlands, Aug. 30, 1995, pp. 94-95. [7] M. H. Kheraluwala, D. W. Novotny and D. M. Divan, Coaxially Wound Transformers for HighPower High-Frequency Applications, IEEE Trans. Power Electronics, Vol. 7, No. 1, pp. 5462, Jan. 1992. [8] J. D. Van Wyk and J. A. Ferreira, Transistor Inverter Design Optimization in the Frequency Range above 5 kHz up to 50 kVA, IEEE Trans. Industry Applications, Vol. IA-19, No. 2, pp. 296302, March/April 1983. [9] S. Freeland and R. D. Middlebrook, A Unified Analysis of Converters with Resonant Switches, Proc. IEEE PESC Conf. Rec. 1987, pp. 20-30.

While this approach is theoretically lossless it has some disadvantages that need attention. The circuit is relatively complex and additional components are required. It can be practically difficult to construct the loop Q1, Cs, Ds and the supply E. This loop needs to have very low inductance if adequate voltage clamping is to be maintained. The optimisation of this and other snubber networks is an ongoing project. 7. CONCLUSION

This paper has presented the dual converter an attractive candidate for low voltage input, high voltage output DC-DC conversion in PV systems. It offers a natural advantages in terms of conduction loss in low voltage systems and is much easier on the high voltage rectifier diodes that can be so problematic in the conventional designs. One area of relative disadvantage is a higher level of sensitivity to transformer leakage inductance. While this can be

Vous aimerez peut-être aussi