Skip to main content

Best Practices to Achieve the Lowest Uncertainty in Measuring with Respect to Standard Test Conditions

Reference Cell / Module Calibration

Spectral error is always present because we are measuring with respect to a tabular reference spectrum. The possibility of reflected light artifacts may also exist in the researcher's test bed. These errors are minimized if the researcher sends the research cell to the calibration lab for calibration and then transfers that calibration to multiple stable reference devices. The process is as follows:

  1. Identify the reference device

    • 1.1. Packaged reference cell

      • 1.1.1. A packaged reference cell is stable but has a spectral error whose magnitude is sample-specific and possible reflection artifacts because it does not look like what is being tested. Only encapsulated reference cells are suitable for outdoor use. If used over long periods of time, a baseline infrared or electroluminescent image is useful to identify cracks or regions of nonuniform generation. A cell size of 2 cm × 2 cm allows the lowest area-related calibration uncertainty because that is the size of most primary reference cells.

  2. 1.2. Research cell

    • 1.2.1. This minimizes spectral error and possible reflection artifacts in setting the light level. The researcher must verify that the sample did not change during the calibration process, and the calibration must be transferred to stable packaged working standards.

  3. 1.3. Production module

    • 1.3.1. This is typically used to set the light level of similar modules being tested on a production line. A reference module minimizes the spectral error if relative quantum efficiency is the same as what is being manufactured. A reference module minimizes the spatial nonuniformity error compared to a smaller reference cell with lower calibration uncertainty. A reference module has a larger calibration uncertainty than a small-area cell because of errors in the correction for spatial nonuniformity of the light source. If used over long periods of time, a baseline infrared or electroluminescent image is useful to identify cracks or regions of nonuniform generation.

  4. Before sending the sample out for calibration, perform a light current-voltage (I-V) on the cell under the standard setup. A dark I-V is very sensitive to any changes in the diode properties, shunting, and series resistance, and it is useful to track changes before they impact the light I-V.

  5. Send the sample to NREL or another photovoltaic calibration lab for calibration.

  6. After the sample is returned with a calibration, verify that the sample has not changed.

  7. Set the light level of your simulator using your typical setup using their newly calibrated photovoltaic cell or module.

  8. Transfer this calibration to multiple working stable packaged working standards. This step is key because it now allows the user to set the light level with a packaged stable cell that corrects for simulator and other testbed stray-light artifacts if they are present, in addition to correcting for spectral error. This assumes that the test cell/reference cell geometry is fixed.

  9. Use the working standard to set or measure the light level on a routine basis.

  10. Compare working standards with calibrated reference samples that have been sitting in a cabinet from time to time. If control charts are used, then this interval could be 6 months to a year. If not, then working standards should be checked frequently against standards calibrated externally.

Calibration Traceability and Maintenance

  • Have a minimum of three calibrated devices (reference cells or modules) on hand at all times so that they can be compared against each other and the working standards to give confidence that the calibration has not drifted.
  • Identify one or more stable packaged cells or module for use in control charts to monitor the test bed and any potential drift in the reference device's calibration.
    • Measure the control sample at least once a week.
    • Plot percentage deviation from the average or calibration value with 3 standard deviation control limits.
    • Put together a basic decision tree for what to do if open-circuit voltage, short-circuit current, fill factor, and maximum power point are outside control limits.
  • When sending samples out for calibration, it is best if a sample that the lab has calibrated already is included so that a control chart of the lab's values for your sample can be maintained. Failure to do this will prevent you from being able to track any random error or drift in the calibration lab's values.
  • Have a set of working standards (reference cells or reference modules) for routine use that are well-calibrated against the lab's primary standards.


  • Wires or ribbons are the most repeatable and durable.
  • Placing an antireflection coating on a contact pad increases the chance of damage because the coating must be scratched through to reach the metal every time the cell is contacted.
  • Transparent conducting oxides (TCOs) such as tin oxide are resistive and can not be readily contacted with metal probes. Researchers often place indium on the TCO to provide a low-resistance contact surface. Test labs such as NREL use metalized rubber and sometimes indium or gold wire if the sample is very small.
  • In general, the lab will make a separate current and voltage connection on the end of a wire, lug in a junction box, or mate to the module connector.