Development of a String Level I-V Curve Tracer

Yunus Erkaya, Isaac Flory, Sylvain X. Marsillac

Virginia Institute of Photovoltaics, Old Dominion University, Norfolk, VA, USA

Abstract — Monitoring systems with high precision and high load. When the switch is closed, the capacitor is connected to
reliability are of great importance in the interest of harnessing the PV array and as it charges in time, the voltage and current
solar power with ever increasing power outputs and efficiencies.
Typical monitoring systems do not allow for the tracing and change will allow tracking of the I-V curve. Different methods
analysis of I-V curves. A portable and light-weight method for can be utilized to measure voltage and current such as
tracing I-V curves of photovoltaic strings has been developed transducers, voltage dividers, and current measurement
using a capacitive load. resistors. PV string properties, such as short-circuit current
Index Terms — curve tracing, I-V measurement, fault and open-circuit voltage, determine the type of switch and the
detection, monitoring, photovoltaics. methods to measure current and voltage.
Selecting a proper value for the capacitor is important and
I. INTRODUCTION poses a certain challenge. The capacitor charge time can be
approximated with the following formula,
I-V curves of photovoltaic strings provide relevant 
t   1.1 C  (1)
information that ordinary monitoring systems at the inverter 
level and fault detection systems at combiner level cannot where, tcharge corresponds to charge time, Cload is the size of the
measure. Information about partial shading, photovoltaic panel capacitor, Voc is the open-circuit voltage, and Isc is the short-
mismatch, effects of solar insolation and temperature, shunt circuit current.
resistance, and series resistance can be gathered from the
analysis of a typical I-V curve. Many methods are described in
the literature regarding curve tracing, and some papers even
go as far as to compare every technology [1]. In our case, we
wanted to develop a system that would allow for the
measurements of I-V curves for a full string of panels (instead
of just one), which required some specific features.

II. INITIAL DESIGN

Fig. 1. Schematic of a capacitive load [2].
Originally a variable resistive load and three different
electronic load topologies were explored: buck, cascaded For instance, a photovoltaic array with Voc= 300V and Isc=
buck-boost, and boost converter; however, they were proven 8A at AM1.5 with Cload= 4700 µF will charge in ~194 ms. The
to be non-robust and quite unsatisfactory. It also took quite a same array under half sun will take twice as long to charge
long time (10 seconds) to trace the full curve, leading to large and if the short-circuit current drops to 1A, charge time will be
power sinking issues. Therefore, resistive load and electronic 1.55 s.
load topologies were abandoned in favor of a capacitive load. If a single photovoltaic panel at Voc= 37V and Isc= 8A is
tested with Cload= 4700 µF, the charge time will be as little as
TABLE I 24 ms. Under illumination that yields 1A short-circuit current,
SUMMARY OF I-V CURVE TRACING METHODS the charge time will increase to 191 ms. More values are
presented in Table 2.
Resistive Load Electronic Load Capacitive Load
Slowest Faster Fastest
TABLE II
Expensive Less Expensive Least Expensive
Heavy Light (heavy resistor) Lightest
SUMMARY OF CHARGE TIMES
Not scalable Semi-scalable Scalable (capacitor 1A 5A 10A
(resistor limitation) limitation) PV Array (300V) 1550 ms 310 ms 155 ms
Low Noise High Noise Low noise Single Panel (37V) 190 ms 38 ms 19 ms

Time variation depending on the source (i.e. single panel vs.

III. DESIGN AND FABRICATION string), along with fluctuations of solar insolation, makes
A simplified schematic of the capacitive load operation is charge times hard to predict with a degree of certainty. It must
described in Fig. 1. A mechanical or electrical switch is also be kept in mind that charge times should be high enough
utilized to connect the photovoltaic array with the capacitive to overcome the internal capacitance of large panels and

978-1-4799-4398-2/14/$31.00 ©2014 IEEE 3104

Authorized licensed use limited to: King Mongkuts Institute of Technology Ladkrabang provided by UniNet. Downloaded on June 05,2023 at 05:37:54 UTC from IEEE Xplore. Restrictions apply.
 strings, but not too high to prevent operating conditions from When current drops to under 2% of normal operation for
changing [3]. Therefore, it would be advisable to have a AM1.5, the curves exhibit significantly more noise as seen in
smaller capacitor value for large strings and a higher capacitor Fig. 3. This is due to the low precision used in taking
value for single panel measurements. This poses challenges in measurements that yield undesirable results in low light
creating a single device that allows for tracing all possible conditions. The capacitors took a considerably longer time to
operating scenarios as long as it is below the voltage and charge at 6561 ms, 4393 ms, and 2425 ms.
current handling capabilities.
The proposed solution to this problem is to have two 0.6 120
different capacitor values for the two different applications.
0.5 100 1 String IV
Since there is more concern with string level monitoring the

Current [A]
capacitor value has been optimized to allow for reliable string 0.4 80

Power [W]
2 String IV
level I-V curve tracing.
The I-V curve tracer, constructed on a PC board measures 0.3 60 3 String IV
3.90” x 3.60” (99 x 92 mm2) with a height of 1.25” (30 mm),
0.2 40 1 String
excluding capacitors, which makes it extremely portable for
Power
field applications. The capacitors are in the process of being 0.1 20 2 String
optimized to lower weight and volume. The maximum string Power
values supported are 450V and 10A, or 4.5 kW. 0 0
3 String
0 100 200 300 Power
Voltage [V]
IV. EXPERIMENTAL RESULTS
Fig. 3. I-V and power curves for 1-3 strings under 1.75%
A. Initial Design illumination.
Experimental results for the initial design were done in
January and can be seen in Fig. 2. At AM1.5 (1000 W/m2) the 10 2000
strings are rated for an open-circuit voltage of 300V and a
short-circuit current of 8.61A each; however, due to lower 8
1500
insolation levels in winter, measurements of 1, 2, and 3 strings
Current [A]

Power [W]
together were possible through a combiner box built in-house 6
that allows for string monitoring and testing. 1000
4
6 1400
500
5 1200 1 String IV 2

1000 2 String IV
4 0 0
Power [W]
Current [A]

800 0 100 200 300

3 String IV Voltage [V]
3
600
1 String
2 Power Fig. 4. I-V and power curves for 2 strings under 50% illumination.
400
2 String
1 200 Power With 50% illumination, the 2-string measurement exhibits
3 String the characteristics of a single-string measurement at one sun.
0 0 Power Therefore, it can be safe to assume that the I-V curve tracer
0 100 200 300 works remarkably well for the photovoltaic strings currently
Voltage [V]
located at the Virginia Institute of Photovoltaics.
Fig. 2. I-V and power curves for 1-3 strings under 25% B. Enhanced Design
illumination.
Experimental results from the initial design indicate that
As seen above, the curves are very smooth and exhibit very there needed to be improvements in accuracy, curve
smoothness and number of data points. Accuracy was
little noise when current is at the levels indicated. No
improved by better utilizing the full range of the analog-to-
smoothing algorithms or electronics were in effect. The
digital converter, which consecutively improved curve
capacitor charge times were 786 ms, 400 ms, and 264 ms for
smoothness. Finally, the enhanced design allows for 5 times
1-string, 2-string and 3-string measurements, respectively. the original data points.

978-1-4799-4398-2/14/$31.00 ©2014 IEEE 3105

Authorized licensed use limited to: King Mongkuts Institute of Technology Ladkrabang provided by UniNet. Downloaded on June 05,2023 at 05:37:54 UTC from IEEE Xplore. Restrictions apply.
 9 1600 8 1400
8 1400 7 1200
7 IV 5mF
1200 6
1000

Current [A]
Power [W]

Power [W]
6 IV 1mF
Current [A]

1000 5
5 800
800 IV 0.68mF 4
4 600
600 Power 5mF 3
3
400
400 Power 1 mF 2
2
200 1 200
1 Power
0 0 0.68mF 0 0
0 100 200 300 0 50 100 150 200 250 300
Voltage [V] Voltage [V]
Fig. 8. I-V and power curves of a single string with one shaded
Fig. 5. Effect of different capacitor values on I-V and power panel.
curves 8 1200

7
The effect of tcharge on the resulting I-V curve of a single 1000
photovoltaic string is seen in Fig. 5. In this case, the smallest 6

Current [A]

Power [W]
capacitance value performed equally well compared to others. 800
5
Shading tests were performed to analyze the performance in
non-ideal conditions. No shading, single cell, single panel and 4 600
dual panel shading results are shown Fig. 6 to 10.
3
9 1600 400
2
8 1400 200
1
7 1200
Current [A]

0 0
Power [W]

6
1000 0 50 100 150 200 250 300
5 Voltage [V]
800
4 Fig. 9. I-V and power curves of a single string with two shaded
600 panels.
3
2 400
Eq. 1 does not take shading effects into account when used
1 200 to predict charging times. Results show that Voc remained
0 0 nearly the same, ~250V, in all cases (single data points can be
0 50 100 150 200 250 300 seen in Fig. 8 and 9), except the shaded panels that did not
Voltage [V] provide any current to charge the capacitors beyond a certain
Fig. 6. I-V and power curves of a single string without shading. voltage.
9 1400 9 1400
8 8
1200 1200
7 7
1000
Current [A]

1000
Power [W]

6 6
Current [A]

Power [W]
5 800 5 800
4 600 4 600
3 3
400 400
2 2
200 200
1 1
0 0
0 0
0 50 100 150 200 250 300
0 50 100 150 200 250 300
Voltage [V] Voltage [V]
Fig. 7. I-V and power curves of a single string with a single cell Fig. 10. I-V and power curves of a single string charged to Voc
shaded. reading of voltmeter for a single shaded panel.

978-1-4799-4398-2/14/$31.00 ©2014 IEEE 3106

Authorized licensed use limited to: King Mongkuts Institute of Technology Ladkrabang provided by UniNet. Downloaded on June 05,2023 at 05:37:54 UTC from IEEE Xplore. Restrictions apply.
 Waiting for the panels to actually charge the capacitors to
the Voc voltage read with a voltmeter lengthened the charge
process by 50-80 times for a single panel depending on the
test. Data points with equal time spacing in Fig. 10 illustrate
that the charge time significantly increased beyond a certain
voltage to where the curve became nearly useless (only 15
points).
Therefore, it was concluded that even though Eq. 1 doesn’t
take shading effects into account, it can still be used in a
reasonable matter to trace I-V curves of photovoltaic strings
up to a point where the panels deliver 1.5% of their maximum
power.

V. CONCLUSION
In conclusion, an effective and portable solution for
measuring I-V curves of photovoltaic strings has been
developed. Currently limited to 450V and 10A, we are
working on scaling it to 1000V, making it compatible with UL
1703 and UL 1741 standards. Work is continuing to improve
the curve tracer to work in all conditions, including current
levels below 1A.

VI. ACKNOWLEDGEMENT
This work has been supported by Dominion Virginia Power.

REFERENCES
[1] E. Duran, M. Piliougine, M. Sidrach-de-Cardona, J. Galán,
“Different methods to obtain the I–V curve of PV modules: A
review,” in 33rd IEEE Photovoltaic Specialists Conference,
2008, pp. 1-6.
[2] T. H. Warner and C. H. Cox, “A high power current-voltage
curve tracer employing a capacitive load,” Solar Cells, vol. 7,
pp. 175-181, 1982.
[3] J. Muñoz and E. Lorenzo, “Capacitive load based on IGBTs for
on-site characterization of PV arrays,” Solar Energy, vol. 80,
(11) pp. 1489-1497, 2006.

978-1-4799-4398-2/14/$31.00 ©2014 IEEE 3107

Authorized licensed use limited to: King Mongkuts Institute of Technology Ladkrabang provided by UniNet. Downloaded on June 05,2023 at 05:37:54 UTC from IEEE Xplore. Restrictions apply.