About six months ago, I built and installed a basic homemade weather station in my backyard. The device measures temperature and sunlight over many months, recording the information on a flash chip. I analyzed the data to investigate:
The solar panel on the weather station is primarily for recharging the battery pack. The battery pack provides power for continued operation overnight. The history of the voltage of the solar panel is recorded in the flash chip, because the voltage level indicates how sunny the day is, and the time of sunrise and sunset.
I got a bright idea (sorry about the pun) that the solar panel voltage at night would be affected by moonlight. Therefore, it should indicate the phases of the moon from full moon (brightest) to new moon (darkest).
On the following graph, the lowest levels of yellow lines occur at night. I’ve cropped out the upper levels of the yellow lines, since we aren’t interested in the daytime voltages.
Solar panel voltage versus moon phases July 2010.
As you can clearly see, the brighter the moonlight, the higher the nighttime voltage. Wait! It doesn’t show that at all! If moonlight had the most significant influence on nighttime solar panel voltage, you should see a gradual increase in voltage from July 11 2010 to July 26 2010, as the moon became brighter.
A single set of data is not statistically reliable. It could have been a cloudy month (it wasn’t). The moon could have been hidden by trees, the horizon, or buildings (it wasn’t). Or, there may have been other issues that obscured the expected results.
I plotted other months and came to the same conclusion. This solar panel’s voltage is not primarily influenced by moonlight. So, what causes the variations?
Overnight air temperature versus solar panel voltage.
For the solar panel used in my weather stations, when connected to a load (battery and device), temperature is the most influential factor in determining the nighttime voltage. For example, notice that July 2 has the lowest temperature and lowest voltage, whereas July 4 has the highest temperature and voltage. The correlation is consistent throughout the graph.
These results aren’t completely surprising, since these solar panels are semiconductors and all semiconductors are affected by temperature. The surprise is that the solar panels are more influenced by temperature than moonlight.
To detect moon phases, I could use a photodiode or photocell that isn’t attempting to power the device. That would provide greater photosensitivity. Furthermore, the resulting reading could be mathematically compensated for temperature.
On the next page, we'll see the weather station fail due to the death of a battery. Was it caused by freezing temperatures?