A solar simulator for measuring performance of large area concentrator photovoltaic (CPV) modules is presented. Its illumination system is based on a Xenon flash light and a large area collimator mirror, which simulates natural sun light. Quality requirements imposed by the CPV systems have been characterized: irradiance level and uniformity at the receiver, light collimation and spectral distribution. The simulator allows indoor fast and cost-effective performance characterization and classification of CPV systems at the production line as well as module rating carried out by laboratories.
© 2008 Optical Society of America
Concentrator Photovoltaics (CPV) is re-emerging as a result of recent advances in high efficiency solar cells . A CPV system consists of an optical system which concentrates solar radiation onto a photovoltaic receiver with a smaller area. This reduction in semiconductor area and the use of more efficient solar cells can lead to a significant cut in the system cost in dollars per watt. However, the lack of appropriate characterization tools for these systems is an obstacle in following a fast learning curve in improving the technology and reducing costs. Measurements under real sun have no practical use in a production line because of weather variability and high time consumption. On the contrary, an indoor solar simulator would allow fast and cost-effective factory testing. Unfortunately, CPV systems impose some strict constraints on the simulator design: besides the conditions needed for conventional flat modules related to irradiance level, spatial irradiance uniformity and light spectrum, the light has to be collimated for the illumination of CPV modules. In a concentrator, only the light impinging the optical aperture within a small angle is transmitted to the cell, therefore the light source must have an angular size similar to that of the sun, i.e. ±0.275°. Conventional simulators do not fulfil this requirement for module-size systems, since typical PV modules (flat plates) do not have this angular requirement.
IES-UPM has developed a solar simulator for CPV systems with the required light collimation by means of a large-area parabolic mirror, which collimates the divergent light beam coming from a small Xenon flash bulb. There are some other basic requirements for a solar simulator, i.e. appropriate irradiance level, spatial uniformity and spectral distribution, which have been also characterized. They are presented here as a demonstration of the validity of the simulator.
2. Light collimation requirement for CPV solar simulators
Typical solar simulators for modules employ a point light source, usually a flash bulb. Its angular size seen by the receiver depends on its section area and the distance between them. If a lamp of 6 cm in diameter is used, it should be placed at more than 100 metres away to assure an angular size lower than that of the sun across a 1-meter-long receiver (Fig. 1). With artificial light it is not possible to achieve real sun irradiance (850 W/m2 is the standard level assumed for direct normal irradiance) for such long distances. Therefore the light source of a CPV solar simulator has to be optically adapted to convert its divergent light beam into a collimated one.
The IES-UPM concept for a CPV solar simulator is based on collimation through a parabolic reflector. The parabola is the geometrical figure in which every ray coming from its focus is reflected back parallel to its axis. Thus, if this mirror is illuminated with a small lamp from its focus, it will produce a collimated light spot in reflection, with the same size as the mirror. A CPV module placed normal to its axis will receive light within its acceptance angle. The angular size of this light will be that of the lamp seen by the mirror (Fig. 2). For a bulb of 6 cm in diameter, at least a 6 m focal distance is required in order to produce the angular size of the sun.
There are two main difficulties involved in manufacturing such a collimator mirror. First, the large area that has to be illuminated, due to the aperture areas used in current CPV modules, requires a mirror of hundreds of centimeters in diameter. This is completely out of range for most optics manufacturers. Secondly, the long focal distance geometry required means extremely accurate optical quality, which is only available in the prohibitively expensive astronomical telescope industry. The IES-UPM has developed a mirror in collaboration with the Spanish company JUPASA with high optical quality but at a much lower cost, thus overcoming the aforementioned technological problems. It is a bulk aluminium piece with a diameter of 2 m and a focal distance of 6 m, mounted on a stable metallic structure (see Fig. 3). Metallic mirroring is obtained through an aluminium-based reflective coating. This allows large CPV modules with any area falling within this circle of 2 m in diameter to be measured.
3. System description
The CPV solar simulator designed at IES-UPM is based on a multi-flash testing system  . A Xenon flash lamp is triggered to illuminate the receiver, and this is biased at a different voltage at each flash pulse, recording different pairs of current-voltage (I, V) points during the fall of the flash pulse. Every current level corresponds to a different irradiance level (through a reference cell), so at the end of the measurement a family of I-V curves at several concentrations is obtained. A PC controls the whole process: triggering the lamp, biasing the device under test and acquiring data. Figure 4 depicts all of the system elements.
Since flash generator discharge is limited by the maximum energy of the lamp, the faster the discharge the higher the radiance of the lamp. A flash lamp without a plateau has been chosen in order to maximize radiance. The flash repetition rate is around 4 seconds. The number of points to be measured for the IV-curve is a trade off between accuracy and speed (measured modules throughput). Five points per IV-curve may be enough for in-line module sorting, allowing one measurement every 20 seconds.
4. Validity assessment: characterization of the simulator
The purpose of a solar simulator is to predict and compare the real sun behavior of PV systems, usually by rating their performance at standard and significant conditions. This establishes four basic requirements on a CPV solar simulator: appropriate irradiance level and spatial uniformity, collimation of the light and spectral distribution. These relate to the illumination system, which in our case corresponds to the flash lamp together with the collimator mirror as a whole.
4.1 Irradiance level
Irradiance level at the receiver is adjustable between the standard levels of 800 to 1000 W/m2 (direct normal irradiance). A commercial Xenon flash lamp and generator is used, which allows the triggering voltage to be varied. The higher the triggering voltage, the higher the peak power. Irradiance decay over time is also used to vary the power level, since a multi-flash system is used.
4.2 Spatial uniformity
For measuring the light uniformity on the entire receiver surface a sensor with narrow angular aperture has been used: a single lens-cell concentrator with around +/-1° acceptance angle, i.e. it accepts light in a similar way to CPV systems. The uniformity measured depends on the size of the aperture lens, since the non-uniformities of the irradiance are smoothed as the aperture area increases. For a 3 cm × 3 cm lens a +/-5% uniformity has been measured. Figure 6 shows the uniformity map of half a mirror, with measurement points equally spaced at 10 cm.
4.3 Collimation of the light
Regarding the angular size of the source, we can consider at least three different features in the mirror that can have an influence on it: overall profile, local defects, and scattering. The mirror profile was measured by mechanical contact, demonstrating it as being within 10 microns of the desired shape. This is too rough for the highly strict telescope optical requirements (e.g. maximum peak to valley error of ¼ wave), but accurate enough for the collimation of the extensive light source in the solar simulator. Local defects found are small and their effects are smoothed by the aperture area of the CPV system, since it averages functionally the non-uniformities within the area of its smallest optical assembly. Scattering has the overall and uniform effect of widening the angular size of the light source, which is caused by the mirror surface finishing.
The uniformity map shown above is a measure of the collimation of the light, since it is given for a narrow angular aperture. Nevertheless, it has been also quantitatively evaluated by means of a CCD camera. The light profile of our Xenon flash lamp has been photographed as seen from the receiver plane, after reflection in the mirror. As it can be seen in Fig. 7, the angular profile has the same geometry as the lamp, i.e. a toroidal flash bulb. This is actually a photograph of the artificial ‘sun’ incoming the CPV module.
The angular size of the profile can be defined as the incidence angle for which 90% of the power is enclosed, which has been measured to be ±0.43°, similar to the angular size of the sun plus some circumsolar radiation. 50% of the power is within ±0.23°. This demonstrates the simulator to be a valid indoor characterization tool for high concentrator PV modules.
4.4 Spectral distribution
The so-called multijunction (MJ) cells made of III-V materials (e.g. GaInP/GaInAs/Ge) are the most typically used as receivers in high-concentration PV systems. This makes system performance very sensitive to spectrum, since MJ cells are a series-connected stack of two or more sub-cells, each one generating current in response to the light of different spectral bands. The overall current is limited by the subcell with the minimum current. Current matching is defined as the ratio between subcell currents, which in optimal performance equals one.
In order to rate the actual performance of a CPV system, the solar simulator has to provide similar current matching in MJ cells as under reference conditions, therefore its spectral distribution has to be studied. In a Xenon flash lamp this is a function of the triggering voltage and the time from the peak in the pulse decay . These dependences have been measured for our flash lamp with a HR4000 spectrometer from Ocean Optics, which allows fast measurements with integration times down to 10 µs to be taken. Increasing the triggering voltage leads to higher equivalent colour temperatures. On the other hand the color temperature is reduced throughout the flash pulse decay. One can take advantage of these dependences to accurately adjust the spectral distribution in order to get similarity with that of the sun. Figure 8 shows a measured spectrum from the simulator together with the reference distribution AM1.5D - Low AOD .
The question arising is how to quantify similarity between spectra. The standard for flat plate solar simulators establishes fixed integration spectral bands for comparison with reference spectrum . But in a CPV system equipping MJ cells, it makes more sense to study the zones defined by the spectral response of the sub-cells, since spectral likeness should lead to the same current matching. In GaInP/GaInAs/Ge triple-junction cells (most often used in CPV), the Germanium cell is designed to generate about 20% current in excess in operating conditions. If this last condition is roughly assured with the flash simulator, only the spectral range covering GaInP and GaInAs subcells has to be studied carefully (see Fig. 8).
The current of each subcell for a given spectrum can be estimated through its spectral response, in accordance with Eq. (1). Thus, the ratio of photogenerated currents for the two cells being studied should be the same as under the reference spectrum to have a valid solar simulator spectrum.
I L,subcell photogenerated current
SR(λ) spectral response [A/W]
E(λ) spectral irradiance [W/m2/nm]
However, it may be difficult to take accurate spectral measurements for such short integration times (tens to hundreds of microseconds), or spectral response curves of subcells are simply not available. A more direct measurement of this spectral similarity is to measure directly these subcell currents isolated. In monolithic MJ cells this is not feasible, but their component cells (also known as ‘isotype’ cells) can be used. A component cell has the same spectral response as one of the subcells of the MJ stack, usually by growing an optically equivalent semiconductor structure but keeping the other subcells electrically inactive (or isotype). If the photogenerated current of component cells under reference spectrum AM1.5D is known, their ratio can be compared to that under the solar simulator. Typically this value will be given for top and middle junctions in a triple junction cell, and the photogenerated current will be approximated by the short-circuit current (ISC), as in Eq. (2). ‘Spectral similarity’ equal to 1 means the perfect matching of the spectra.
In the equation, AM1.5D superscript stands for the current measured under these reference conditions , and ‘Top’ and ‘Middle’ subscripts refer to the corresponding subcells. This value is not only related to the spectrum itself, but also obviously to the solar cell behavior, so component cells with same technology as those under the concentrator should be used in order to find the best figure for spectral similarity.
Component cells of triple junction cells from Spectrolab have been used in order to measure this spectral similarity between our flash simulator and the reference AM1.5D under which they were calibrated. Matching point (‘Spectral similarity’ equals 1) is achieved for a wide range of triggering voltage levels if different delays from peak power are considered. Fig. 9 plots this measurement for the solar simulator illumination system (lamp plus collimator mirror) as a function of time, and at a given triggering voltage. In this case spectral matching is achieved at 850 W/m2, which is usually taken as the reference irradiance level for CPV systems rating .
The first solar simulator able to measure large area CPV modules under reference conditions indoors has been developed. Its illumination system is based on a Xenon flash light and a large area collimator mirror, which simulates natural sun light. It has been thoroughly characterized, demonstrating appropriate irradiance level, spatial uniformity and collimation of the light.
The spectral similarity with standard spectrum has been also characterized in two different ways: through fast spectrometry and by measurement of the current of 3J component (‘isotype’) cells, demonstrating its validity.
The simulator is already being commercialized, allowing indoor fast and cost-effective performance characterization of CPV systems at the production line.
Authors are grateful to R. Herrero for the CCD photos of the light source, to I. García for the spectral response measurements of the component cells from Spectrolab, and to G. Jüngst from INTA Spasolab for the calibration work on the fast spectrometer. This work has been mainly supported by the Spanish Ministry of Education and Science through Project DISCO-FV (ENE2004-04669/ALT) and under Consolider Ingenio 2010 Program, through the project GENESIS-FV (CSD2006-0004). Solfocus and Isofotón have partially supported this project at different stages of its development. Also the European Commission within the project FULLSPECTRUM of the VI Framework Program and by the Comunidad de Madrid within the NUMANCIA programme (S-05050/ENE/0310). César Domínguez is thankful to the Spanish Ministerio de Educación y Ciencia for his FPI grant.
References and links
1. A. Bett et al., “High-concentration PV using III–V solar cells,” in Proceedings of IEEE 4th World Conference on Photovoltaic Energy Conversion (Institute of Electrical and Electronics Engineers, 2006)
2. W. Keogh and A. Cuevas, “Simple Flashlamp I–V Testing of Solar Cells,” in Proceedings of IEEE 26th Photovoltaic Specialists Conference (Institute of Electrical and Electronics Engineers, 1997).
3. I. Antón, R. Solar, G. Sala, and D. Pachón, “IV Testing of Concentration Modules and Cells with Non-Uniform Light Patterns,” Proceedings of the 17th European Photovoltaic Solar Energy Conference and Exhibition (2001), pp. 611–614.
4. J. Kiehl, K. Emery, and A. Andreas, “Testing Concentrator Cells: Spectral Considerations of a Flash Lamp System,” Proceedings of the 19th European Photovoltaic Solar Energy Conference and Exhibition (2004).
5. C. A. Gueymard, D. Myers, and K. Emery, “Proposed reference irradiance spectra for solar energy systems testing”, Solar Energy , 73, 443–467 (2002). [CrossRef]
6. ISO 15387:2005 Space systems -- Single-junction solar cells -- Measurements and calibration procedures (International Organization for Standardization, Geneva, Switzerland, 2005).
7. ASTM E 2527 - 06 Standard Test Method for Rating Electrical Performance of Concentrator Terrestrial Photovoltaic Modules and Systems Under Natural Sunlight (ASTM International, West Conshohocken, United States, 2006)