Hi,
Such dramatic steps in the level strongly suggest an electrical/mechanical fault of some type. Have you considered/measured how hot it might be getting inside your enclosure? The first test I would do is to monitor the voltage directly across the PV panel with a multimeter, to check that there are no faults with the panel.
Personally, I avoid using "preset" pots whenever possible (because they are basically a "mechanical" component), and certainly for any "precision" measurement. It should be possible to calculate a suitable fixed value, or simply substitute a fixed resistor after test. Then "trim" or calibrate the final value in software (i.e. within Cumulus, or I guess the WeatherDuino itself ?).
Probably, not directly related to your "steps", but putting any "window" over the solar sensor potentially upsets the sensor's "Cosine Response". This is because the amount of light reflected off the surfaces of the window (and thus not measured by the sensor) increases as the angle of incidence reduces (i.e. as the sun gets closer to the horizon).
The PV panel itself is presumably designed for direct use outdoors, but if it is necessary to protect the rest of the electronics (and electrical connections, etc.) then I recommend a thin (< 1mm) sheet of PTFE ("Teflon") for the "window". PTFE acts as an excellent "diffuser" to achieve a good Cosine response, is "transparent" to UV, visible and near-IR and has good physical and chemical (unreactive) properties. Note the little white dome in the FO solar pod, although I've no idea whether that works properly, or is even made from Teflon.
Cheers, Alan.
Such dramatic steps in the level strongly suggest an electrical/mechanical fault of some type. Have you considered/measured how hot it might be getting inside your enclosure? The first test I would do is to monitor the voltage directly across the PV panel with a multimeter, to check that there are no faults with the panel.
Personally, I avoid using "preset" pots whenever possible (because they are basically a "mechanical" component), and certainly for any "precision" measurement. It should be possible to calculate a suitable fixed value, or simply substitute a fixed resistor after test. Then "trim" or calibrate the final value in software (i.e. within Cumulus, or I guess the WeatherDuino itself ?).
Probably, not directly related to your "steps", but putting any "window" over the solar sensor potentially upsets the sensor's "Cosine Response". This is because the amount of light reflected off the surfaces of the window (and thus not measured by the sensor) increases as the angle of incidence reduces (i.e. as the sun gets closer to the horizon).
The PV panel itself is presumably designed for direct use outdoors, but if it is necessary to protect the rest of the electronics (and electrical connections, etc.) then I recommend a thin (< 1mm) sheet of PTFE ("Teflon") for the "window". PTFE acts as an excellent "diffuser" to achieve a good Cosine response, is "transparent" to UV, visible and near-IR and has good physical and chemical (unreactive) properties. Note the little white dome in the FO solar pod, although I've no idea whether that works properly, or is even made from Teflon.
Cheers, Alan.