A weather station is a classic backyard-scientist device. My wife and I currently garden based on folklore or regional statistics rather than solid data for our specific lot.
What location in the yard gets the most light? Is it too early to plant tomatoes because of frost (what was it like last year at this time)?
For this reason, I decided to make a weather recorder. The weather station needs to:
Weather station in clear project case.
The prototype is installed in a clear plastic pencil box, rather than an element-proof project enclosure. If the initial experiment proves successful, I'll machine up something nice.
The weather station consists of:
(There is another article dedicated to the circuit for solar power and battery recharging).
The weather station uses an arrangement of thermistors similar to the PCB thermal analysis project. There are six temperature sensors in total:
Soil temperature probe under a rock.
Over the winter, I’m going to dig a deep hole and measure actual frost depth.
Here are the temperature, solar panel voltage, and battery pack measurements logged for several days.
Graph of weather data for June 29 to July 2, 2010 in Chicago, Illinois USA.
The light blue and salmon-colored lines at the top are voltage measurements that correspond to the secondary axis (right side numbers). You can see that the solar panel voltage goes way up during the day, oscillates a bit, and declines. The blue battery pack voltage declines steadily at night as it supplies power, and then rises steadily as it is recharged during the day.
The red line shows the temperature inside the project case goes up to around 110 °F during the day, with a cool period in the afternoon when the case is in shadow.
The blue line shows the soil temperature and the other lines show the air temperature. The soil is insulated and has a larger thermal mass than the air, so it changes values less rapidly and less radically than the air.
Notice something interesting? The air is coldest at dawn.
Zooming in a bit, here are two line graphs of overnight temperatures. (The battery and solar panel voltages do not appear on these graphs.)
Graph of overnight temperatures for June 28 to June 29 2010 in Chicago.
As expected, the air temperatures change faster than the soil temperature. Again, the soil retains heat because of its large mass.
There is a bump in the middle due to a change in weather patterns (warmer weather was blowing in). Some nights will be like that.
If we can believe that the temperature sensors have high enough resolution and are correct relative to each other, then the fact that the air gets slightly colder the farther away it is from the warm summer soil is as expected.
But, something does not look right about the box temperature. It mirrors the other air temperatures, but is one to two degrees colder. That doesn’t make sense. If anything, the project box should retain heat and the circuitry should produce a slight warming.
The day was very humid, and water condensed inside the project box at night. The project box thermistor was pressed up against the top lid, where the water was condensing. So, perhaps the condensing water was the source of the temperature drop.
I thought that heat was released when a gas condenses. But, perhaps the surrounding area is heated slightly but the condensed liquid cools?
Graph of overnight temperatures for June 29 to June 30 2010 in Chicago.
The next evening is more like I expected.
Even so, I expect the box temperature to always be a few degrees warmer than the air temperature at the same height (53 cm), because of the heat coming from the electronics.
Also, why is the highest-height air temperature warmer than the next two lower-height air temperatures? Well, maybe there really was a layer of warmer air at that height that evening.
On the next pages, we'll take a closer look at the weather station circuitry and the mistakes I’ve discovered so far...