How to make a weather station really work.

So I got all the pieces together since my last post about weather satellite images. The hardware update is I got a very nice antenna:

Click the photo to see on Amazon.com

That alone improved reception, by eliminating a lot of noise. Also added an LNA:

Click the photo to see on Amazon.com

That’s placed right at the feed point of the antenna, it amplifies the raw signal, filters it for just the 137mhz band, and amplifies it again. (note, if you intend it for all weather operation, you’ll have to do something to weatherize this unit)

Those two things gave me the most clear satellite images I’ve gotten yet, and more importantly, it pulls the signal in from lower satellite elevations, so I get more of the pass, meaning I get more of an image in the north/south direction. You can see with this photo, I get almost from the arctic down to a good chunk of Florida on a good pass:

Now it was time to make this work in the garden for real. So I have a back corner garden, it has the biggest view of the sky in my yard, and gets a good amount of sun for the solar panel… So I’ve had a setup of a 25 watt solar panel, a 21 amp hour gel battery, and a charge controller that I’ve used to put other experiments out in the garden… however those experiments were done with a small circuit board with a relatively cheap ESP32 on it… That draws maybe 400ma… and I was able to run that experiment (A magnetic field logger) pretty far into the winter before it quit.

Now the the Raspberry PI 4 can draw up to 3 amps, which is significantly more than the panel/battery combo was designed for… but I gave it a try. I had it attached in such a way that the charge controller would cut power to the PI if the battery dropped below 10.5 volts — to save the battery… It’s a deep cycle battery, but it should not be drawn down to zero.

For waterproofing… the battery and charge controller are in a plastic battery box that has 2 downward facing vents on it. But there’s a problem… on a night when the dew is strong, it gets moisture condensing in the box… this is not good. It really needs a fan for airflow.. but I’m not sure if I can afford the power…

So since I don’t want my pie to get damp, I bought a waterproof electrical box… put the PI and the SDR radio in it and all looked happy. Then a sunny day came, and wow great for the solar panel, but not so great that the waterproof box has no where to dissipate heat… running compute intensive stuff with the radio and LNA running caused the PI to detect an overheat situation. Also at the same time it drained the battery down and the charge controller shut the whole show down…

So I tried again… Now I have a task that runs once a minute on the PI. it uses “vcgencmd” to get the throttle state of the PI… if it detects a persistent under-voltage, or over heat it kills all the satellite tasks and hopes for things to cool down before the next pass.

Ok so another very hot day came along, and at least the task shutdown worked, and the pi began to cool off. But not fast enough for me… So I moved the whole thing into the battery box again, and now the task is…figure out how to get heat out of a sealed plastic box.

I’ve done calculations and the panel should be able to more than keep up with the PI, meaning it should result in a net charging up of the battery… The problem is.. when a processor runs hot it draws more current than normal. And it’s probably why when the first heat situation hit, and was runaway, it drained the battery down below the threshold.

One more thing… I have an RTL_SDR dongle… if you turn the bias tee on to power the LNA.. well it stays on even when you’re not using the radio.. It stays on until you explicitly turn it off… That never happens in Raspberry-NOAA-V2.. so I inserted a command to shut the bias tee down between passes. This saves 300ma which is significant.

So now also on order is a “Witty PI” — which is a daughter board you attach to the PI.. it does nifty power management, like,, shut the pi down on low voltage, hight temperature, and or on a schedule… That’s exactly what I need… I can ditch the charge controller low voltage system and use the Witty PI’s version… if voltage goes to low it will do a graceful shutdown of the pi, and wait till voltage charges back up again, if the temp goes too high, it will do the same,… and for extra power saving, I can shut the pi down on a schedule, since the satellites have a consistent morning and evening schedule…there’s a good 8 hours of no passes, that I can basically shut down and draw zero power during.

By the way.. the pi itself has no real way to shut down and come back on again on it’s own because it has no real time clock. also when you halt a PI 4, it still consumes 300ma on it’s own just to sit and do nothing. The Witty PI will bring that down to an absolute zero for the pi because it controls the PI power source. The Witty PI itself draws less than 10ma when idle.

Leave a comment