Help designing a tiny solar system

walden, 5 months ago (edited 5 months ago)

30w might not be sufficient to run the pi with a camera, etc., especially taking into account cloudy days and whatnot. Do you live somewhere particularly sunny?

I’d start off with a single 100w panel at a minimum. That’s what I have, and in the winter (right now) it maxes out at around 45 watts, and only for a couple hours a day. The rest of the day it’s a trickle.

Does the Pi need to be on 24/7? Certain charge controllers allow you to implement a schedule for the “load” terminals. You’ll want something like this to run the Rpi.

Does it get below freezing where you are? LiFePO4 get damaged if you charge them below freezing. In that case a deep-discharge rated Lead Acid battery might be better.

I just built out a solar setup for my shed. It has different needs than yours, but I’ll be running a Pi 3B+ 24/7 and I think my setup should handle that ok, even in the winter months.

A 100w panel should be about $100-120USD. I recommend a nice solar charge controller – something like the Victron SmartSolar 100/20 at a minimum (100/30 would be better for adding another panel or 2 in the future). Watch some YouTube on why Victron is great, especially if you’re a nerd and want to check the status of stuff from your house via bluetooth.

For a panel, charger, and battery, you’re in the $400 range. Try and use some spare wire you have laying around – 12 gauge should be fine. If you go cheaper and get a 30w panel and a cheapo charge controller from AliExpress (nothing wrong with that, just fewer features), you might end up spending more in the future if it’s not sufficient.

MossyHabitat, 5 months ago

Thanks for the reply. Do you have stats on how many watt hours you collect from the 100w panel in the Dec/Jan (assuming you’re in the northern hemisphere), and how those stats vary on overcast days vs sunny? I’m around 37°n and relatively sunny in the winter, except for a gloomy week here or there - an adequate battery buffer will be needed for those days, and the capacity can be shifted to ventilation fans in the summer.

While it does freeze here (occasionally down to 0° F) the battery will be inside the chicken coop where the temp has always remained above freezing. Initially my concern with lead acid was the risk of gassing into the birds’ airspace, which is why I’m drawn to the LiFePO4 chemistry. Victron does seem great for heavier(medium)-duty scenarios like sheds or RVs.

The pi is a zero, which pulls about 25-33% the current of a 3b, and I’ve been able to run it off a 10,000 mah battery bank for over 24 hours. It is beneficial to have it running continuously during daylight hours only, aside from logging weather lows, but I can’t think of a practical method to shut it down & bring it online automatically - at that point it’d probably make more sense to switch to an ESP32 device feeding data to a primary server inside the house.

walden, 5 months ago (edited 5 months ago)

I’m around 41°N (southern New England). I installed a Victron SmartSolar 30A just within the past week, so the stats I have are limited. The best I’ve seen so far was yesterday where I got 50Wh. Most other days it has been 40Wh. Today is a bit cloudy so I’ll try and update you on what I get.

Edit: I made a spreadsheet to sort of figure out what I can accomplish with 50Wh – and it’s not much! If the Pi draws 2.5 watts, the solar will only cover 20 hours not including any other electricity I want to use.

This time of year has the shortest days, so if I can bump it to 60+ Wh a day that will be better. My property also has the downside of a hill to the west, so it makes direct sunlight even shorter. I’ve already ordered a 2nd 100w panel from Harbor Freight so that will help.

fratermus, 5 months ago (edited 5 months ago)

Do you have stats on how many watt hours you collect from the 100w panel in the Dec/Jan (assuming you’re in the northern hemisphere),

You can model average insolation (and use that to extrapolate average harvest) by month using tools like PVwatts. Here’s a walkthrough.

Using Bowling Green, KY as an example since it’s on the 37th. 100w of flatmounted panel on an MPPT solar charge controller would average:

<span style="color:#323232;">Solar wattage	100
</span><span style="color:#323232;">Month	Daily Wh Avg
</span><span style="color:#323232;">Jan	168
</span><span style="color:#323232;">Feb	249
</span><span style="color:#323232;">Mar	331
</span><span style="color:#323232;">Apr	426
</span><span style="color:#323232;">May	513
</span><span style="color:#323232;">Jun	598
</span><span style="color:#323232;">Jul	555
</span><span style="color:#323232;">Aug	506
</span><span style="color:#323232;">Sep	396
</span><span style="color:#323232;">Oct	305
</span><span style="color:#323232;">Nov	201
</span><span style="color:#323232;">Dec	156
</span><span style="color:#323232;">Average	367
</span><span style="color:#323232;">
</span>

Derate those yields by ~18% if using PWM. <-- rule of thumb, not gospel

if we are on the west coast instead, here’s Santa Cruz, California:

<span style="color:#323232;">Solar wattage	100
</span><span style="color:#323232;">Month	Daily Wh Avg
</span><span style="color:#323232;">Jan	206
</span><span style="color:#323232;">Feb	286
</span><span style="color:#323232;">Mar	386
</span><span style="color:#323232;">Apr	519
</span><span style="color:#323232;">May	582
</span><span style="color:#323232;">Jun	642
</span><span style="color:#323232;">Jul	605
</span><span style="color:#323232;">Aug	542
</span><span style="color:#323232;">Sep	447
</span><span style="color:#323232;">Oct	349
</span><span style="color:#323232;">Nov	245
</span><span style="color:#323232;">Dec	183
</span><span style="color:#323232;">Average	416
</span>

and how those stats vary on overcast days vs sunny?

The figures above are daily averages, including normal weather patterns and are how we size our systems. But for the sake of curiosity/understanding, my observations have been that if my clear-day harvest is X then overcast is 0.6X, bright overcast is 0.7X and dark/rainy is 0.05-0.10X. Cloud-edge effect and other reflective phenomena can result in harvest >1X.

While it does freeze here (occasionally down to 0° F) the battery will be inside the chicken coop where the temp has always remained above freezing.

A battery warming solution could be implemented for $20 (warming pads, 12v temp controller).

walden, 5 months ago (edited 5 months ago)

Not OP but this is really neat. Thanks.

Over the past few days my system has been severely under performing in relation to these calculations. For January it says 290 Wh per day, but I’m only seeing 50. I do have a lot of trees, so probably some shading… maybe even a lot of shading.

<span style="color:#323232;">Jan	290.3
</span><span style="color:#323232;">Feb	392.9
</span><span style="color:#323232;">Mar	387.1
</span><span style="color:#323232;">Apr	400.0
</span><span style="color:#323232;">May	387.1
</span><span style="color:#323232;">Jun	400.0
</span><span style="color:#323232;">Jul	419.4
</span><span style="color:#323232;">Aug	387.1
</span><span style="color:#323232;">Sep	400.0
</span><span style="color:#323232;">Oct	322.6
</span><span style="color:#323232;">Nov	300.0
</span><span style="color:#323232;">Dec	290.3
</span>