By Harvey Serreze

Solar simulators are used by the PV module manufacturing industry to determine the power rating of solar modules. Uncertainties or errors in power rating measurement can significantly impact a module manufacturer’s revenue. Consequently, module testing at the completion of manufacture under simulated outdoor operating conditions must give an accurate result; solar simulators can provide an accepted and convenient way to achieve this goal. Several equipment manufacturers offer a variety of solar simulators making the decision of what tool to buy not an easy task. International standards such as IEC 60904-091) define in detail the requirements needed to meet classes of performance. However, a simple examination of key simulator specifications may not always lead to the optimum choice. Often it is best to look beyond the basic specifications in selecting the proper tool for the job. Solar simulators are not perfect light sources and the quality of their optical output can strongly contribute to the uncertainty of the power measurement.

Spectrum

Although spectral classification of a simulator is a key factor to consider, its determination by the manufacturer may not be as easy as one might expect. Specifically, a Class A spectrum (the highest performance level defined by the standard) requires that the spectrum emitted by the simulator be within ±25% of the standard AM1.5G spectrum within each of six wavelength intervals from 400 to 1100 nm. Since most simulators utilize pulsed illumination to minimize module heating during test, this factor adds significant challenge to the ability to accurately measure the output spectrum of the simulator tool.

Measurement of the output spectrum over nearly a three-to-one wavelength range (400 to 1100 nm) from a pulsed optical source can be difficult to achieve. Few single-unit spectrometers can span this wavelength range without sacrifice. The presence of fairly intense peaks in the spectrum of commonly-used Xe flash lamps sets the maximum sensitivity above which the spectrometer might saturate. Since most spectrometers utilize silicon detectors, selection of proper spectrometer design specifications such as blaze wavelength to achieve adequate long wavelength sensitivity can result in inadequate short wavelength sensitivity. Other factors to consider during the measurement process are adequate calibration of the instrument and proper collection of an accurate light sample from the simulator under test.

Recognizing these difficulties, the National Renewable Energy Laboratory (NREL) in Golden, CO, the U.S.A., developed a measurement tool specifically to test pulsed solar simulators. Their Pulse Analysis Spectroradiometer System or PASS is able to very accurately measure the spectrum of a pulsed simulator2). Outside of NREL, there are very few such instruments in existence. Spire is fortunate enough to have a copy of their PASS─a photo of the unit is shown in Figure 1. Using such an instrument which is recalibrated by NREL annually, the company is able to check the calibration of its secondary instruments most of which are dual-spectrometer units3). This is very important for both accurate internal measurements of its simulators prior to shipment and for field use at customer sites throughout the world to check and/or certify the spectrum.

Irradiance Nonuniformity

Another often overlooked or misunderstood parameter is the spatial nonuniformity of irradiance at the test plane. While classification standards for nonuniformity have been set by agencies such as IEC in their aforementioned 60904-09 document, there remain a few issues that are vague or undefined. Specifically, a Class A nonuniformity designation for large area simulators requires that the spatial resolution of measurement utilize individual pixels or test points each no larger than 400 cm2. For example, this means that a simulator of 1.5 m x 2 m area must be measured at no less than 75 points. Furthermore, the test cell must cover at least 80% of the pixel area and, therefore, must be no smaller than 320 cm2 in size for this particular example. However, there are situations where higher nonuniformity measurement resolution is needed or utilized. For example, TUV Rheinland employs a uniformity test system having an equivalent resolution of approximately 248 pixels over 1.5 m x 2 m. Spire has developed a high resolution ‘micro-uniformity’ test measurement system capable of measuring the same 1.5 m x 2 m area with a resolution of over 1728 pixels with each pixel having a size of approximately 17 cm2. A photograph of this system appears in Figure 2 and a typical micro-uniformity map for its new advanced simulator appears in Figure 3. The excellent high-resolution uniformity of this new, compact design tool is apparent.

The irradiance nonuniformity value for a simulator is defined using the irradiance maximum and minimum values rather than from a statistical analysis of the irradiance pattern; consequently, there can be a wide variation in the reported nonuniformity value of a given tool depending on the size of the individual test point and the number of pixels, even though each determination is performed in accordance with the IEC standard. An example of such variation is shown in Figure 4. Because the entire region to the right of the dotted red line is within the boundaries set by the IEC standard, the uniformity can be legitimately reported as being under 1% at low resolution (well under the 2% maximum set by the Class A definition), while at higher resolution it may rise to nearly 1.5%. For applications such as single or multi-crystalline Si where the individual cell sizes are large and generally square in shape, there is relatively little need for high resolution uniformity. However, for thin-film applications where individual cells can be very narrow (1 cm or less) and quite wide (1 m or more), the need for high resolution uniformity can be much more critical.

Flash Duration

The length or duration of the light pulse is a parameter which is not addressed by any standard. As a minimum, the duration of the pulse must be sufficiently long to perform one complete sweep of the current-voltage (I-V) characteristic. Depending on the desired number of data points and the data sampling rate, the minimum pulse duration can typically be in the 10 ms range. Again, for applications such as traditional crystalline Si where the cells are able to reach steady-state equilibrium within this 10 ms time period, such short pulse durations are adequate. However, many of the newer, advanced thin-film structures and also some high-efficiency Si ones do not reach operating equilibrium during such short pulses; this can result in a significant underestimate of their power output capability. The causes for this behavior range from high capacitance to long diffusion length and slow trapping/detrapping times and are being studied by numerous research groups throughout the world. An example of this slow response behavior for an advanced CIGS production module is shown in Figure 5. It is apparent that steady-state power would not be reached during a typical 10-20 ms duration light flash. By not utilizing a simulator with adequate pulse duration (in this case, approximately 100 ms), the manufacturer would be underestimating the output power by nearly 2% or possibly more which could result in a significant loss of potential sales revenue.

Other Issues

Finally, there are other features that may not receive much attention, but can be important in selecting the best simulator to use. A good example of this is the ability to measure series resistance in a production environment. Outlined in IEC 608914) is the recommended procedure to determine the internal series resistance of a PV module. This procedure requires measurement of the I-V characteristic under three different irradiance levels. This can certainly be done by subjecting the module under test to three successive flashes, but in a production environment this will slow down the throughput considerably. Alternatively, if the simulator can provide three intensity levels within a single flash such as that shown in Figure 6, this measurement can be done with no impact on throughput. To achieve this ability, however, it is necessary that the simulator have the necessary performance to provide pulse durations in excess of 100 ms at one sun and even longer at lower intensity.

In summary, selection of the best solar simulator for a given application and test environment must be made by examining more than just the basic specifications. Several key quality parameters should be considered and evaluated such as spectral measurement accuracy, uniformity of irradiance, and pulse duration. Credibility in the manufacturer’s ability to accurately measure and report spectrum and uniformity of the measurement tool is important. Knowing and understanding the effects of any irradiance nonuniformity is also important, particularly for applications that utilize non-standard cell sizes and interconnect configurations. For many of the newer, more advanced thin-film and crystalline silicon technologies, the utilization of light pulses of adequate duration may be of great concern. Finally, features that allow determination of secondary parameters such as series resistance without slowing down production throughput may also be important issues.

Dr. Harvey Serreze is Product Development Manager for Solar Simulators at Spire Solar, Inc. (www.spirecorp.com). He received his Ph.D. in Electrical Engineering from Tufts University in 1974. Dr. Serreze was responsible for fabricating the first EFG silicon solar cells at Mobil-Tyco Solar Energy Corp. and for designing and building their first solar simulator test system. He joined Spire in 1993 where he has been developing and characterizing next-generation simulators.

The author is deeply indebted to Jason Burns and Nanditha Chandrasekhar for significant technical contributions.

REFERENCES

1) ‘International Standard--Photovoltaic Devices--Part 9. Solar simulator performance requirements’, IEC 60904-9, Edition 2.0, 2007-10, International Electrotechnical Commission (IEC), Geneva, Switzerland.

2) Afshin M. Andreas and Daryl R. Myers, ‘Pulse Analysis Spectroradiometer System for Measuring the Spectral Distribution of Flash Solar Simulators’, presented at the SPIE Optics and Photonics 2008 Conference: Optical Modeling and Measurements for Solar Energy Systems II, San Diego, CA (August 10?14, 2008).

3) H.B. Serreze and R.G. Little, ‘Large Area Solar Simulators--Critical Tools for Module Manufacturing’, Photovoltaics International, 1, 108-111 (August 2008).

4) ‘International Standard--Photovoltaic devices--Procedures for temperature and irradiance corrections to measured I-V characteristics’, IEC 60891, Edition 2.0, 2009-12, International Electrotechnical Commission (IEC), Geneva, Switzerland.