Comason Bipand Uni

5
1 Abstract— An inverter is essential for the interfacing of photovoltaic panels with the AC network. There are many possible inverter topologies and inverter switching schemes and each one will have its own relative advantages and disadvantages. Efficiency and output current distortion are two important factors governing the choice of inverter system. In this paper, it is argued that current controlled inverters offer significant advantages from the point of view of minimisation of current distortion. Two inverter switching strategies are explored in detail. These are the unipolar current controlled inverter and the bipolar current controlled inverter. With respect to low frequency distortion, previously published works provide theoretical arguments in favour of bipolar switching. On the other hand it has also been argued that the unipolar switched inverter offers reduced switching losses and generates less EMI. On efficiency grounds, it appears that the unipolar switched inverter has an advantage. However, experimental results presented in this paper show that the level of low frequency current distortion in the unipolar switched inverter is such that it can only comply with Australian Standard 4777.2 above a minimum output current. On the other hand it is shown that at the same current levels bipolar switching results in reduced low frequency harmonics. I. INTRODUCTION rid connected single phase photovoltaic systems may be unipolar switched or bipolar switched. They can be current controlled or voltage controlled. In this paper the focus is on current controlled systems. These have advantages such as active current wave shaping, inherent current limitation and automatic synchronisation with the AC network [1]. For simplicity and excellent dynamic performance characteristics, hysteretic control [2] has been adopted for the current loop. The purpose of this paper is to compare low frequency output current distortion of unipolar switched inverters and bipolar switched inverters. Unipolar switched inverters have the advantage of higher efficiency due to reduced switching loss [3], but it has been shown theoretically [4] that distortion of their output current can be significant, specially at low Manuscript received September 20, 2007. L. Bowtell and T. Ahfock are with Faculty of Engineering Surveying, University of Southern Queensland, Australia ([email protected]) power levels. Grid-connected inverters have to operate within distortion limits specified in Australian Standards 4777.2. II. SYSTEM DESCRIPTION Figure 1 is a representation of the overall system. Fig. 1: Overall System The output terminals of the solar panels are directly connected to the DC input bus of the inverter. The voltage control loop is digitally implemented. It incorporates a maximum power tracker. The tracker routine is invoked once every few seconds. It involves incrementing or decrementing the voltage reference signal for the voltage control loop and monitoring of the resulting change in PV power output a few seconds later. If the output power rises the next increment of the reference voltage is made in the same direction as the previous one otherwise it is made in the opposite direction. If there is no change reference voltage is left unchanged. COMPARISON BETWEEN UNIPOLAR AND BIPOLAR SINGLE PHASE GRID- CONNECTED INVERTERS FOR PV APPLICATIONS Les Bowtell and Tony Ahfock G

description

hvgbvh

Transcript of Comason Bipand Uni

Page 1: Comason Bipand Uni

1

Abstract— An inverter is essential for the interfacing of

photovoltaic panels with the AC network. There are many possible inverter topologies and inverter switching schemes and each one will have its own relative advantages and disadvantages. Efficiency and output current distortion are two important factors governing the choice of inverter system. In this paper, it is argued that current controlled inverters offer significant advantages from the point of view of minimisation of current distortion. Two inverter switching strategies are explored in detail. These are the unipolar current controlled inverter and the bipolar current controlled inverter. With respect to low frequency distortion, previously published works provide theoretical arguments in favour of bipolar switching. On the other hand it has also been argued that the unipolar switched inverter offers reduced switching losses and generates less EMI. On efficiency grounds, it appears that the unipolar switched inverter has an advantage. However, experimental results presented in this paper show that the level of low frequency current distortion in the unipolar switched inverter is such that it can only comply with Australian Standard 4777.2 above a minimum output current. On the other hand it is shown that at the same current levels bipolar switching results in reduced low frequency harmonics.

I. INTRODUCTION

rid connected single phase photovoltaic systems may be unipolar switched or bipolar switched. They can be

current controlled or voltage controlled. In this paper the focus is on current controlled systems. These have advantages such as active current wave shaping, inherent current limitation and automatic synchronisation with the AC network [1]. For simplicity and excellent dynamic performance characteristics, hysteretic control [2] has been adopted for the current loop.

The purpose of this paper is to compare low frequency output current distortion of unipolar switched inverters and bipolar switched inverters. Unipolar switched inverters have the advantage of higher efficiency due to reduced switching loss [3], but it has been shown theoretically [4] that distortion of their output current can be significant, specially at low

Manuscript received September 20, 2007. L. Bowtell and T. Ahfock are with Faculty of Engineering Surveying, University of Southern Queensland, Australia ([email protected])

power levels. Grid-connected inverters have to operate within distortion limits specified in Australian Standards 4777.2.

II. SYSTEM DESCRIPTION

Figure 1 is a representation of the overall system.

Fig. 1: Overall System

The output terminals of the solar panels are directly

connected to the DC input bus of the inverter. The voltage control loop is digitally implemented. It incorporates a maximum power tracker. The tracker routine is invoked once every few seconds. It involves incrementing or decrementing the voltage reference signal for the voltage control loop and monitoring of the resulting change in PV power output a few seconds later. If the output power rises the next increment of the reference voltage is made in the same direction as the previous one otherwise it is made in the opposite direction. If there is no change reference voltage is left unchanged.

COMPARISON BETWEEN UNIPOLAR AND BIPOLAR SINGLE PHASE GRID-

CONNECTED INVERTERS FOR PV APPLICATIONS

Les Bowtell and Tony Ahfock

G

Page 2: Comason Bipand Uni

2

Output power from the PV array tends to raise the DC bus voltage by charging the storage capacitor. The voltage controller operates to keep the voltage constant at the level determined by the maximum power tracker. It is essentially a digitally implemented PI controller. The controller output, which at steady state is a constant value, is raised or lowered until the DC bus voltage signal fed back to the controller matches the voltage reference signal. The voltage reference signal is multiplied by the AC supply voltage signal to produce the reference current for the hysteretic current control loop. The multiplier is digitally implemented.

A. Unipolar Mode Current Control

During the entire positive half cycle of the source voltage

( sv ), insulated gate bipolar transistor T3 is kept off and T4 is kept on. Transistor T1 is switched on when the inverter output

current ( si ) goes below the bottom limit of the hysteretic

band. This causes si to rise while it flows through T1 and T4.

When current si goes above the upper limit of the band T1 is

switched off. This causes si to fall while it flows through D2 and T4.

During the entire negative half cycle of the voltage ( sv ), transistor T4 is kept off and T3 is kept on. Transistor T2 is

switched on when the inverter output current ( si ) goes above

the top limit of the hysteretic band. This causes si to rise

negatively while it flows through T3 and T2. When current si goes outside the lower limit of the band T2 is switched off.

This causes si to fall towards zero while it flows through D1 and T3

B. Bipolar Mode Current Control

During the positive half cycle of the source voltage ( sv ),

when current si falls below the bottom limit of the hysteretic band, T1 and T4 are switched on. As a result the current rises through T1 and T4. If the current rises above the top limit of

the hysteretic band, T1 and T4 are switched off and current si falls through D2 and D3.

During the negative half cycle of the source voltage ( sv ),

when current si goes above the top limit of the hysteretic band, T2 and T3 are switched on. As a result the current rises negatively through T2 and T3. When the current goes below the bottom limit of the hysteretic band, T2 and T3 are switched

off and si falls towards zero through D1 and D4. Note that bipolar switching as described here is subtly

different from classical bipolar switching. The difference being that in the classical case, during freewheeling through a diode pair, the transistor pair across the conducting diode pair will be turned on so that a reverse path is always available for the current. However, for simplicity the experimental inverter was operated as described above rather than as per classical bipolar switching. This has no consequence for most of the AC cycle

because except near the zero crossing of sv , there is no attempt by the current to reverse. The only observable effect is discontinuity of the inverter output current for a short time interval near the supply voltage zero crossing.

III. CAUSES OF LOW FREQUENCY HARMONICS

The following causes of low frequency distortion in the

inverter output current have been identified:

1) harmonic content in the signal voltage refv 2) harmonic content in the voltage controller output signal

[5]; 3) switching delay; and 4) inductor non-linearity;

A. Multiplier Input Signal Voltage

Without any signal conditioning, multiplier signal refv is merely an attenuated version of the AC mains voltage. In that

case total harmonic distortion in refv would be equal to total harmonic distortion of the supply voltage which was measured as 3.4 %. Assuming there were no other cause of distortion,

the distortion in refv gets replicated in si . The purpose of the zero phase shift filter in figure 1 is to minimise distortion in

si that is directly caused by the harmonics in the mains voltage. Total harmonic distortion at the output of the filter was 1.5 %.

B. Harmonic Content in Voltage Controller Output Signal

The feedback signal to the voltage controller contains harmonics because of distortion in the DC bus voltage. The main cause of that distortion is components at 100Hz and at higher multiples of 100Hz which are part of the storage capacitor current. With an analogue PI controller, the harmonics propagate to the controller output. These get

modulated by refv and produce a dominant third harmonic component in the reference current [5]. Since a digital controller has been adopted, it has been possible to completely eliminate the 100 Hz harmonics from the PI controller output with the result that it is purely DC at steady state. Thus the

multiplier output is not distorted if refv is purely sinusoidal.

Page 3: Comason Bipand Uni

3

C. Switching Delay

Detailed analysis presented by Sharma and Ahfock [4] shows that switching delay can cause significant total harmonic distortion in the output current of unipolar switched current controlled single-phase photovoltaic systems. This is particularly the case at low power output. If switching frequency harmonics are ignored, the deviation between

inverter output current is and reference current refi is given by equation 1 which has been derived using the mathematical model given in reference [4].

( )sgn

2s d c d s

s ref

v t V t vi i

L L− = − (1)

where dt is the switching delay

If there was no switching delay, as expected, the right hand side of equation (1) is zero. The second term on the right hand side is proportional to the source voltage and does not contribute to distortion since the source voltage has been assumed to be sinusoidal. The first term on the right hand side is a square wave component and therefore represents odd

harmonics in si . For the case of bipolar switching,

d ss ref

t vi i

L

−− = (2)

Equation 2 was deduced from the mathematical model

presented in reference [4]. It predicts that switching delay

causes current si to deviate from the reference, but no distortion results because the deviation is proportional to the source voltage which has been assumed sinusoidal.

D. Inductor Saturation

In the absence of switching delay, the mathematical model

in reference [4] predicts zero low frequency harmonic distortion for both unipolar switching and bipolar switching. If there is saturation and switching delay, then inductance L in

the above equations becomes a function of current si . When

supply voltage sv is low, L assumes higher values and vice-versa. It can be deduced from equations (1) and (2) that the simultaneous presence of switching delay and saturation will

cause distortion in current si essentially because inductance

L is not a constant. In this paper inductor saturation is not a cause of distortion since the highest current considered is within the linear range of the inductor.

IV. EXPERIMENTAL RESULTS

Experimental results are displayed in Figures 2 to 5 and in

table 1. All tests were performed with inductance L equal to

10 mH, DC bus voltage cV at 400V and AC supply voltage sV at 230 V. Both at high output current and at low output current, low frequency distortion is higher for unipolar switching.

Low frequency distortion is present in the bipolar case

because refv is not purely sinusoidal, because of imposed

blanking times near the zero crossing of supply voltage sv and because of discontinuity in the current waveform just before the zero crossings of the supply voltage. As mentioned before, this discontinuity results from the fact that T3 and T4 are kept

off during the positive half cycle of sv and T1 and T2 are

kept off during the negative half cycle of sv . This last reason, it is believed, is the most important cause of low frequency harmonics in the bipolar case.

Low frequency harmonics is present in the unipolar case

because refv is not purely sinusoidal, because of imposed

blanking times near the zero crossings of supply voltage sv and because of switching delay (equation 1). The last reason, it is believed is the most important cause of low frequency harmonics in the unipolar case.

Fig. 2: Inverter Output Current

(Bipolar switching; 2refi A= )

Fig. 3 Inverter Output Current

(Unipolar switching; 2refi A= )

Page 4: Comason Bipand Uni

4

TABLE I TOTAL HARMONIC DISTORTION (EXCLUDES SWITCHING FREQUENCY

HARMONICS)

Reference Current A (rms)

% THD Unipolar Switching

%THD Bipolar

Switching 0.5 11.25% 8.37%

0.75 8.18% 5.02%

1.0 6.34% 3.8%

2.0 3.42% 2.06%

The harmonic spectrum at low output current for the case of

unipolar switching is given in figure 5. The harmonic distribution has strong resemblance to that of a square wave as predicted by equation (1).

The harmonic spectrum at low output current for the case of bipolar switching is given in figure 6. It is postulated that the observed low frequency harmonics can be significantly reduced by preventing discontinuity of current at the zero crossing near the zero crossing of the supply voltage.

Fig. 5: Inverter Output Current

(Bipolar switching; 0.5refi A= )

Fig. 4 Inverter Output Current

(Unipolar switching; 0.5refi A= )

Fig. 5 Harmonic Spectrum of Inverter Output Current (Unipolar 1A, THD =

6.34 %)

Fig. 6: Harmonic Spectrum of Inverter

Output Current (Bipolar 0.5A, THD = 8.37%)

V. CONCLUSION

One of measures of quality of power from a grid connected photovoltaic system is the level of low frequency harmonic content in its output current. Several causes of low frequency harmonics have been identified for single phase grid connected systems operating in current controlled unipolar or bipolar mode. Techniques have been proposed to reduce low frequency harmonic content for both unipolar switching and bipolar switching. It is also demonstrated, both theoretically and experimentally, that compared to unipolar switching, bipolar switching has reduced levels of low frequency harmonics especially when power output is low. If compliance to Australian Standard 4777 is a requirement, then unipolar switching may not be an option below a certain power level. Bipolar switching is more likely to meet AS4777 harmonic requirements at low power levels provided current diacontinuity near the supply voltage zero crossing is avoided. However unipolar switching may offer higher efficiency because of reduced switching loss. An optimum approach may be to operate the inverter in unipolar mode for most of the AC cycle and to swap to bipolar mode near the supply voltage zero

Page 5: Comason Bipand Uni

5

crossing. This is currently being investigated.

REFERENCES

[1] Borle L, Dymond M S, Nayar C V, Philips S J, “Current Controlled Grid Connected Inverter”, Proceedings of the Australian and New Zealand Solar Energy Society Conference, pp 374-379, December 1993.

[2] Harashima F, Inaba H, Kondo S, Takashima N, “Microprocessor-Controlled SIT Inverter for Solar

[3] Energy Systems”, IEEE Transactions on Industrial Electronics”, pp 50-55, Vol 34, No. 1, February 1987.

[4] Liaw C M, Chen T H, Wang T C, Cho G J, Lee C M and Wang C T, “Design and Implementation of a Single Phase Current-Forced Switching Mode Bilateral Converter”, IEE Proceedings PtB, No.3, pp 129-136, May 1991.

[5] Sharma R , Ahfock A, “Distortion in Single Phase Current Controlled PV Inverters For Grid Connection”, Proceedings of the Australasian Universities Power Engineering Conference , AUPEC04, Brisbane, Australia, September 2004

[6] Sharma R and Ahfock A , “Performance Analysis of Utility Connected Photovoltaic Generation ” Proceedings of the Australasian Universities Power Engineering Conference, Brisbane, Australia, September 1992