Weather Station Update – Part 1

So over the past few months I have been experimenting with the weather station and have made some major changes.

See my first post here for a general description of what I did.

After a period of trial and error in the field I understood that the Raspberry Pi 2 that I was using with the Pimoroni weatherhat addon was using too much power for my solar panel and battery setup. After several modifications to attempt to reduce the power consumption of the Pi, I was only able to achieve 4 days on a full charge, with the added frustration that if there was no sun this would be reduced even more. One of my major annoyances due to this was that I was not able to collect any rain data!

The problem with the Pi is that it isn’t an embedded system rather a micro-computer. What I mean by this is that there is no inherent means to place the system into a hardware sleep or reduced power mode. I even exchanged the Pi board with a Pi Zero with high hopes, but to no avail. In the end I had to revert to the engineering route to solve the problem.

I approached this project in the lazy way: buy a bunch of off-the-shelf products, stick them all together and hope that it works well enough for my use-case. Well, unsurprisingly with this method it sort of half works!

What I needed was some “engineering glue” and a rethink, or to at least engage my brain!

The Pi Zero I managed to reduce to 150-200mA of power consumption. The battery I had was rated at 10Ah and the solar panel at 5W – so for 5V supply 1A in current. With the efficiency of the solar cell this reduced to 100-200mA of usable current.

For this system to work effectively, the solar setup would have to produce over twice the operating power of the hardware during the daylight hours to compensate for nighttime drain. Essentially I would need to target the 50mA range to guarantee the battery would last long enough to survive sunless periods and allow it to be recharged quickly during sun-full periods.

The calculation without a solar panel would be:

10000mAh battery capacity

200mA average usage

10000/200 = 50 hours or 2 days

Including a solar panel, let’s say with an average supply of 150mA, operating in sunlight for 10hours a day would mean this:

daylight hours: (150-200)*10 = -500mAh / day of net drain

"sunless" hours: 14 * 200 = -2800mAh / night of net drain

therefore in a 24hour period I’m consuming 3300mAh of battery capacity. 10000/3300 = 3 full days.

*sunless hours I define as the solar power not providing charge.

Like I said with some modifications I was able to boost this to 4 days, which wasn’t good enough. I would need a big net gain during daylight hours to compensate for the night and also charge the system. Targeting 50mA would get me this:

daylight hours: (150-50)*10 = 1000mAh / day of net gain

"sunless" hours: 14 * 50 = -700mAh / night of net drain

Therefore in a 24hour period I’m recovering 300mAh of battery capacity.

If the sun disappeared the system would be able to operate for (10000 / 24*50) 8 days. After which the net gain would allow the system the refill and operate normally.

As you can see the system engineering approach starts off with the use-case and derives the requirements. From there I can produce an expectation of what is needed to achieve the requirement and then select the appropriate solution and architectural design.

In my next post I will discuss the system tradeoff and go into more technical detail.