No abstract

The measurement of the photovoltaic (PV) performance with respect to reference conditions requires measuring the performance with respect to a reference spectrum. Procedures were developed in the mid 1980s to correct measurements for errors relating to the spectral irradiance of the light source being different from the standard and the responsivity of the irradiance detector being different from the device under test. In principle, these procedures are exact, but require the measurement of the spectral irradiance of the light source and responsivity of the test device. This is problematic for most facilities that measure module performance. It has been suggested that a polynomial fit of the short-circuit current (iJ measured under natural sunlight divided by the total broadband irradiance as a function of air mass provides an accurate spectral correction factor. The polynomial correction factor is normalized to unity at an absolute air mass of 1.5. The polynomial correction factor is compared with the spectral correction factor for a variety of devices at two locations. INTRODUCTlONMeasurements of PV performance under natural sunlight with respect to standard reference conditions require translating the data to the reference irradiance, spectrum, and temperature. This paper compares two methods for correcting the data to the reference spectral irradiance. The spectral correction factor provides a means to exactly correct the measured current to an arbitrary tabular reference spectral irradiance distribution. Because all reference spectral irradiance distributions are only a function of wavelength, with angular information not included, the spectral correction factor can be written as [l] where E, , , , is the total irradiance, E&) is the spectral irradiance of the reference spectrum, E&) is the spectral irradiance of the solar spectrum, S,(h) is the spectral responsivity of the reference detector, and S,(h) is the spectral responsivity of the test device whose measured short-circuit current is I-. If the reference detector is a thermal detector then S,(h) is constant and drops out. The units of spectral irradiance are Wm-zwm-'. The units of spectral responsivity are AW' for a semiconductor-based reference detector and VW' for a thermal detector such as a pyranometer. If a pyranometer is used as the reference detector to measure the total irradiance, then S&) is constant and drops out. Commercial spectroradiometers cannot measure the spectral irradiance from 300 nm to 4000 nm, so an additional uncertainty in the spectral correction factor is introduced by limiting the range of spectral correction to a Si-based detector of 300 nm to 1100 nm 121.It has been suggested that a fourth-order polynomial fit of I= divided by the total broadband irradiance as a function of air mass provides an accurate spectral correction factor [3]. The polynomial correction factor is normalized to unity at an absolute air mass of 1.5. Corrections based on the absolute air mass are most accurate when the reference spectrum is close to ...