Influence of solar cell capacitance on the measurement of I-V-curves of PV-modules

S. Mau and T. Krametz

arsenal research, Faradaygasse 3, 1030 Vienna, Austria Phone: +43 - 50 550 - 66 52, - 66 71, Fax: +43 - 50 550 - 63 90 E-Mail: stefan.mau@arsenal.ac.at, Internet: www.arsenal.ac.at

ABSTRACT: Longer periods of warranty for PV-modules and increased competition among PV-module manufacturers made the quality testing and exact measurement of the maximum power output of modules a very important issue. To guarantee correct and reproducible measurements, I-V-curves are normally traced under simulated irradiance. Two types of solar simulators are available for the measurement of PV-modules: Steady state and flash simulators. The latter achieve the necessary uniformity and spectral match without a noticeable increase in the modules temperature. The reason for this is the shortness of the light flash in the range of 0.5 - 20 ms. The whole I-V-curve is traced in one (single-flash) or more flashes (multi-flash) [1]. Using shorter measurement intervals, even for crystalline solar cells, cell capacitance may affect the accuracy of the measurement. The focus of this paper is the change in curve shape and thus the error for the assessment of the maximum power output caused by this phenomenon as well as it’s correlation to open circuit voltage decay (OCVD).

Keywords: Qualification and testing, solar cell capacitance, open circuit voltage decay

2 EXPERIMENTS 1 INTRODUCTION

For the following investigations a class A flash Nowadays flash simulators are widely used for the I-simulator (Berger-Lichttechnik) with a xenon-lamp, that V-characterisation of PV-modules. The advantages of

generates a 10 ms long light flash, was used. During the these simulators are in the better uniformity of the

tracing of the I-V-curve at STC the modules were irradiance in the measurement plane and the negligible

connected to an adjustable ohmic load. The regular increase of temperature of the module during the

ohmic step was 40 mOhm. However several jumps up to measurement. However, special precautions must be

260 mOhm appeared. taken to ensure the accuracy of the transient

measurement results if the I-V-curve of PV-modules with

In a first part of this work the maximum a high capacitance is traced. Significant shifts of the

measurement duration according to IEC 60904-9 was curve at higher voltages near the maximum power point

determined. This standard defines the temporary leading to the wrong estimation of the maximum power

inhomogeneity of the flash to be less than ± 2 %. output have been reported [2, 3, 4].

Therefore the light intensity was traced during the flash

at different starting voltages of the light source. The This shift may be related to the capacitance of PV-signal of a Si-sensor was measured with a clamp-on modules which, at higher voltages near Vmpp mainly

ammeter and recorded with a digital scope. On the base affected by the diffusion capacitance [5, 6]. The diffusion

of the traced characteristics, the maximum duration for capacitance is correlated to the lifetime of the minority

the subsequent I-V-measurements was determined. carriers in the quasi-neutral bulk of the solar cell [7]. If

increasing lifetime leads to a higher diffusion capacity, a

The second part of this work looks at the I-V-capacitor has to be included in the equivalent circuit of

characteristics of 13 PV-modules, traced at different the module. For correct simulation of the electrical

measurement periods to detect a possible shift of the behavior of the PV-module it is placed parallel to the

curve, using shorter measurement periods. To simulate diode and the shunt resistance. Charging or discharging

the same load characteristics for all modules, the ohmic the capacitor may influence the measured current and

load was parameterized as shown in table 1. voltage of the PV-module by using shorter measurement

periods. In this work the I-V-characteristics of 13

modules were traced for different measurement periods Load curve 1 2 3 to find out whether PV-module capacitance may have an

Measurement period [ms] 2.5 8.5 17 influence on the curve shape at shorter measurement

Meas. period / point [µs] 45 45 90 periods.

Number of meas. points 55 190 190

The open circuit voltage decay (OCVD) Ohmic steps [* RMPP] ~0,052 ~0,013 ~0,013 measurement is often used to investigate the minority Flash single single multi carrier lifetime of a solar cell. However, if the diffusion capacitance of a PV-module is increased, this will Table 1: Ohmic load characteristic during I-V-strongly affect the shape of the OCVD of the module [8]. measurement In this work, electrically excited OCVD was used to estimate the response time of the module and these The ohmic load increased linearly up to where the results were compared to the measured I-V-curves. RMPP (=UMPP/IMPP) doubled. Above that value the load

curve was defined according to the exponential behavior of the I-V-characteristic. The error in the estimation of the maximum power output, resulting from the low number of measurement points for load curve 1, was not

more than 0.5 %. For the visualisation of possible transient effects, the module voltage and current were traced with a digital scope with a sampling rate of 500 kB/s that was directly connected to the modules junctions. Table 2 lists the modules according to their solar cell manufacturers, cell dimensions and materials.

Subsequently the OCVD behavior of the modules was investigated. Therefore the tempered PV-module (25 °C) was serially connected to a power transistor and a power supply. During the 5 seconds on-state of the transistor, an adjustable voltage was applied to the PV-module. The turn-off time of the transistor was 5 µs. In the subsequent off-state, the PV-module voltage was monitored with a digital scope. In discrete steps the voltage was reduced from VOC to ~0.7*VOC.

3 RESULTS

The intensity of the light flash against the time is displayed in figure 1. 1.06 flash period: 8,5 ms 1.05

2,5 ms][% 1.04

ytis U = 90 %netn

1.03U = 88 %U = 86 %I evit 1.02aleR 1.01 1.00 0.99

0123456710111213

Time [ms]

Figure 1: Relative intensity of the light flash against the time at different starting voltages of the light source

About 1.5 ms after activating the flasher the displayed plateau of nearly 11 ms appeared. The light intensity is stabilized by controlling the discharging current of the light source inducing a 3.5 kHz oscillation. The oscillation of the measured PV-module current is automatically corrected by simulator software. Increasing the starting voltage of the light source resulted in a higher mean intensity but also in a sharper decrease at the beginning and the end of the flash. To comply with the requirements according the temporary inhomogeneity of the IEC 60904-9 a 2.5 ms time delay and an 86 % starting voltage of the lamps were chosen to trace the I-V-curves.

The results of the I-V-measurement of the 13 PV-modules are presented in table 2. The I-V-curves of all PV-modules were measured at 2.5 ms, 8.5 ms and 17 ms and each of them were traced from SC to OC and vice versa. For 9 modules there was no shift of the MPP observed when using shorter measurement periods. Also the digital scope showed no transient effects which could have influenced the measurement accuracy. So the I-V-curves were assumed to be correct.

However, 4 PV-modules revealed a shift in curve shape using shorter measurement periods. At higher voltages curve shape depended heavily on the

measurement time and direction. For most of the curves the shift of MPP was higher when measuring from OC to SC than in the reverse direction. This effect is in accordance with literature that reports charging transients to be faster than discharging transients. This phenomenon isn’t however yet fully explained. As expected the simultaneous current and voltage tracing using a digital scope revealed the ongoing transient effects of current and voltage.

Manufacturer Dim. Material Shift of MPP

2.5 ms1

8.5 m1 17 ms1 BP 5” m-Si ok ok ok Central Elect. 5” m-Si ok ok ok Chaori 4” m-Si ok ok ok Ersol 5” p-Si ok ok ok ISFH 4” m-Si ok ok ok Kyocera 6.2” p-Si ok ok ok

Microsol 6” m-Si -5.2 % 2 -3.1

% 2

ok +9.6 % 3 +2.1 % 3

ok Mitsubishi 6” p-Si ok ok ok

Sanyo 4” m-Si3 -6.8 % 2 -2.9 % 2

ok +19.7 %

3 +4.8 % 3 ok Sharp 5” m-Si ok ok ok

+4.1 % 3

ok ok Webel 6.2” m-Si -2.3 % 2

+6.6 %

3 ok ok +1.8 % 3 ok Yingli 5” p-Si ok ok ok Yuhuan

6” p-Si ok ok ok

Table 2: Solar cell data and shift of maximum power point at different measurement times, m-Si: mono-Si, p-Si: poly-Si, 1) Measurement time, 2) SC-OC, 3) OC-SC

When starting the I-V-measurement at the OC, the capacitor was charged during the determination of the OC-voltage. During the tracing of the curve the capacitor continuously discharged, resulting in higher volt0ages and therefore an overestimation of the maximum power output. When starting the measurement at the SC, the capacitor charged during the measurement which led to an underestimation of the maximum power output.

4.5 MPP-area:4.0 Voltage 3.5 a 3.0 ] Current A[ 2.5

2,5 ms / OC-SC a Timetn8,5 ms / OC-SCerru2.0C 17 msb b Voltage1.5 8,5 ms / SC-OC2,5 ms / SC-OCc 1.0 CurrentTime

0.5

Voltage

0.0

c 010203040506070Current

Voltage [V]Time

Figure 2: I-V-curves, i(t) and u(t) for different measurement times and directions for one module

This effect is shown in Figure 2 for one module. The large diagram shows the I-V-curves measured with all three load curves. Tracing the I-V-curve at 8.5 ms or faster led to a shift in curve shape. The only

measurement period that gave the correct I-V-characteristic of the module was at 17 ms. In this case identical curves were traced in both measurement directions. The top and bottom diagram on the right side show the current and voltage for the 2.5 ms measurement from OC to SC and SC to OC respectively. They clearly exhibit the decreasing or increasing voltage and current of the module, during the measurement of the single points, due to the discharging or charging of the capacitor. The middle diagram shows the current and voltage for the 17 ms measurement. Due to doubled measurement time per point and the increased number of points used, which led to smaller steps, there were no transient effects visible.

1.1 1 drop due to series resistance0.9 0.8 ]co0.7 V /[0.6 e gat0.5 loV0.4 0.3 0.2 0.1 0 00.10.20.30.40.50.60.70.81.1 Time [ms]1 decay:0.9 -22 mV/µs0.8 shift due to increased]co0.7 diffusion capacitanceV /[0.6 decay: eg -190 mV/µsat0.5l oV0.4 0.3 0.2 0.1 0 00.10.20.30.40.50.60.70.8Time [ms]

Figure 3: The “normal” OCVD (top) and the OCVD influenced by increased diffusion capacitance (bottom)

Figure 3 shows the characteristic diagrams of two modules from the OCVD-measurement. The two diagrams show the time before and after the turn-off of the power transistor for three different voltage levels. The upper diagram demonstrates the typical OCVD for calculating the lifetime of the minority carriers. After a first drop due to the series resistance, voltage decreased in a typical non-linear way. This behavior was measured for all modules which exhabited no shift in the curve shape using shorter measurement periods. The lower diagram shows this same characteristic at lower voltages but at higher voltages there was a noticeable shift in the curve shape. This shift was caused by an increase in diffusion capacitance. This particular behavior was found in all modules which exhibited a shift in the I-V-curve

using shorter measurement periods. The reduced voltage decay of -22 mV/µs (Voc) compared to -190 mV/µs (0.7*Voc) is thought to be the reason for the strong curve shift at higher voltages because the module wasn’t able to follow the voltage steps given by the load.

4 CONCLUSIONS

In this work 13 PV-modules were used to study the influence of solar cell capacitance on the measurement of the I-V-curves, using a flash simulator. For 4 modules, shorter measurement times led to a shift in the shape of the I-V-curve and therefore to the wrong estimation of the maximum power output of the PV-module. Measurement of the open circuit voltage decay showed that this effect was related to an increased diffusion capacitance at higher voltages. However, an appropriate simultaneous voltage and current tracing as well as the measurement from OC to SC and vice versa are useful tools in detecting problems resulting from increased solar cell capacitance.

REFERENCES

[1] W. Herrmann, Leistungsmessung von PV-Modulen mit Sonnensimulatoren – Bewertung der Lichtqualität mit neuartiger Messtechnik, 19. Symposium Photovoltaische Solarenergie, Staffelstein (2004)

[2] H. A. Ossenbrink, Do multi-flash solar simulators measure the wrong fillfactor?,

[3] K. Bücher, Accurate production line testing of high capacity modules, high efficiency modules and very large area building integrated PV modules, 17. EUPVSEC, Munich (2001)

[4] D. Chenvidhya, On dynamic and static I-V characteristics of solar cell modules having low and high fill factors, 3rd WCPEC (2003) Osaka, Japan

[5] G. Friesen, Capacitance effects in high-efficiency cells, Solar Energy Materials and Solar Cell, Volume 48 (1997) pp. 77-83

[6] R. Salach-Bielecki, Influence of junction parameters on the poen circuit voltage decay in solar cells, Opto-Electronics Review, Volume 12 (1) (2004) pp. 79-83

[7] M.P. Deshmukh, Measurement of silicon and GaAs/Ge solar cells ac parameters, Solar Energy, Volume 78 (2005) pp. 1-4

[8] T. Pisarkiewicz, Photodecay method in investigation of materials and photovoltaic structures, Opto-Electronics Review, Volume 12 (1) (2004) pp. 33-40