Correcting bias in measured module temperature coefficients

Temperature coefficients for PV modules describe the change with temperature of current, voltage and power. Coefficients are commonly determined by linear regression using measured module output at fixed irradiance and varying temperatures. We compare temperature coefficients determined for the same modules from both outdoor and indoor measurements. We find systematic bias in the temperature coefficients for voltage and power, with values derived from indoor measurements consistently smaller in absolute value than values derived from outdoor testing during which the module temperature is measured as specified in IEC 61853-1. Our work suggests that the bias results from a corresponding bias in the estimated module temperature. However we have not identified an alternative arrangement of a few thermocouples that would result in consistent values for temperature coefficients from either indoor or outdoor measurements.