Solar Shed mega hobbyist. Finally a HN topic I can contribute to.<p>I've spent last few years trying to build out my solar shed mainly to have auxilary power to charge weed whacker, leaf blower, power lights, power a camera, etc. Just basic 12v power.<p>The number one lesson is to 4x your power needs. 4x your watts and 4x your battery amp hours. Don't cheap out on charge controllers.<p>There's never enough sun light to fully charge your batteries and your panels are never at the right angle to maximize sunlight. Especially in less solar friendly areas like Michigan.<p>Lead acid is fine if you over compensate with panels. Lead acid will last longer in the cold than LiFEP04. Die hards will say you cant charge LiFEP04 in winter but I do. I accept the fact that my batteries will degrade faster. When you can build your own packs, its a lot easier to find which single set of cells went out and cheaper to replace. Its a risk I accept.<p>If you are starting out, get a 200 Ah/hr LiFEP04 battery and 400 watts solar. Its a very good starting point for 12v and gives room for expandability.<p>Avoid foldables. Avoid Harbor Freight. Avoid Jackery Solar Panels. You need Quality Panels.<p>If you wanna learn more, check out Will Prowse <a href="https://www.youtube.com/@WillProwse">https://www.youtube.com/@WillProwse</a>
I have recently gone down the solar panel rabbit hole too but I never tried the "fancy portable mini panels" cause they seem to weak and overpriced.<p>Instead, I acquired two large 140W panels that had been removed from an old installation for the princely sum of $20 (USD).<p>After struggling with trying to find connectors that I could pair with the MC3 connectors (remember, these panels came off an old installation), I bought a dozen MC4 and just cut off the old MC3 connectors...MC4 is the way to go these days.<p>Next came the surprise at finding out about "MPPT" - in all my years looking at solar panels on roofs, I never knew they were so finicky about voltage vs current curves...real MPPT charge controllers start at $70 for a Victron 15Amp and I decided to try my luck with Chinese sellers on AliExpress...turns out the charger was a PWM fake sold as a MPPT charge controller.<p>Anyway, I also needed a battery since the charger doesn't work without a battery source. That's another saga but short story - I built my own 3S battery pack out of 18650s.
1. Lead Acid is far cheaper, though its bigger and heavier than LiFePo. 12V @12Ah is only $35 from a brand-name (<a href="https://batteryinthecloud.com/products/ps-12120" rel="nofollow noreferrer">https://batteryinthecloud.com/products/ps-12120</a>), and closer to $25 from no-name brands on Amazon.<p>2. Modules to charge LeadAcid are so cheap, they don't even make them. This specified PS-battery hsa a 13.5V to 13.8V "standby" voltage, meaning that 13.65V from the Solar Panel is all you need to have a UPS. Connect the 13.65V source from the Solar-panels to the + and - leads of the 12V battery, and volia. You get 13.65V when solar is available, and 12V from the battery when the solar cuts off. The end. Isn't that easy?<p>3. Grid-tie should be similarly easy, though I don't have too much experience with it myself. IMO, buy a professional AC->DC converter, probably at 19V or some other suitable voltage and then get a DC-DC buck converter to go from 19V to 14.35V, and then a diode (0.7V dropoff) to hook up to your batteries in parallel. You'll also need a diode in your Solar-panels cause you don't want your Grid-tie system "charging your solar panels" (that'd probably create a fire and/or damage them...).<p>Hobbyists should NOT deal with main-power themselves, but there's a gross-many number of AC-to-DC converters available from $10 to $40.<p>4. Don't battery balance. Just buy bigger batteries. If 12Ah isn't enough, buy a 20Ah battery. If 20Ah isn't enough, buy a 33Ah battery, etc. etc. The limit is whatever you're comfortable with (the bigger batteries give more current which can be more dangerous)
There are some really handy databases with detailed insolation data available for sizing your solar projects. <a href="https://nsrdb.nrel.gov/" rel="nofollow noreferrer">https://nsrdb.nrel.gov/</a> is one place to find this stuff.<p>Years ago I worked on environmental monitoring systems for snow and ice management. One of the things I did was help spec out solar systems for remote installations in Alaska and Canada.<p>The basic issue was "How much sun will there be in a typical summer?" combined with "How much power can we store in our batteries?" and "How much power can we draw from the lead acid batteries without them freezing?" (who knew that a discharged battery freezes at a lower temperature than a charged one? I sure didn't).<p>It's a pretty straightforward process, actually. The company I worked for had a pretty detailed spreadsheet to help with bids, but essentially you:
1. look up the insolation data for the install site in one of the databases with that info (like <a href="https://nsrdb.nrel.gov/" rel="nofollow noreferrer">https://nsrdb.nrel.gov/</a>)
2. compute rate of loss over time for your installation (power consumption +_self-discharge)
3. look at the battery datasheets to determine how much you can discharge the battery at any given time based on anticipated temperature ranges
4. work out what your tolerances are for temperature and insolation variability<p>Usually there would be a couple of options that were in the cost sweet spot where you had a bit more solar and bit less battery or vice versa. Some of these sites were up around the arctic circle, so battery was pretty important in those cases, since you're looking at months without appreciable sun combined with low temperatures.<p>This wasn't even the major part of my job, but it was a small company and I wore a lot of hats.
I started out like the author of the blog post with a 60Watt panel and a cheap non-mppt solar controller. My balcony orientation was shit and that didn’t work out. Since then I upgraded “quite a bit”…<p>The challenge is more difficult than it seems. A raspberry pi 4b does ~3.4 watt idle so that’s 84 watt-hour per day. This may not sound like a lot, but for battery-powered devices it quite a bit. Microcontrollers can run months on that amount of energy, but they aren’t as convenient as the Pi.<p>The 12v 12ah battery used, contains 144 watt hour. Only enough to cover less than two full days with little to no sunlight.<p>The 50 watt panel is already small, but the non-mppt controller makes it even more inefficient. With good sunlight, the panel can easily run the pi and charge the battery during daytime.<p>The “real” challenge is to keep the Pi online during days of overcast weather. A 150 watt panel may only do 4-8 watt under those circumstances and that’s not enough.<p>But I always love these projects none the less.
I did a similar project a while back with a Pi Zero W, which has the perk of maxing at like 1w. Still, my takeaway was "whatever you think you need for a panel, triple it, and whatever you think you need for a battery, multiply it by like 10." Cloudy days tend to come together, the power output for your panel will drop off way more than you think, and your battery won't charge nearly as fast as you're hoping.
"the Raspberry is actually powered from the USB output of the controller"<p>I don't know about his controller but these cheap controllers will power the USBs even on low battery. They will kill the battery (he even mentions this bellow).<p>The right way to have USB output is to connect a car charger to the 12V output of the controller and then connect your load to that charger. That 12V output is cut when the battery is low and reconnected after it is charged. These voltages can be set in the controller. This way the battery will never discharge so much that the controller doesn't even recognize the type of battery (6V, 12V)
Nice to see solar projects like this. We have 4 solar setups on the property currently.<p>One for main house. Another for the pump house. A temporary garage setup and one for a studio.<p>A good way to learn to manage your own power generation and setting up systems.
I've been looking for the same sort of mini charge controller with grid backup that is mentioned in the problems and challenges also with no luck. Is this really not a product that exists?<p>I've long wanted to do a sort of small scale migration to solar. Offset the load from my home server with a few solar panels and source any excess power needed from the grid while also charging the batteries/providing power from solar when the sun is up. I could probably engineer something myself but doing things right on high power circuits is not worth the effort/risk. Any suggestions?
It's down right now. Archived version:
<a href="https://web.archive.org/web/20230822000517/http://blog.rfox.eu/en/Hardware/SolarPi_experiment_2_Finally_something_that_works.html" rel="nofollow noreferrer">https://web.archive.org/web/20230822000517/http://blog.rfox....</a>
I remember tuning down the clock speed of the raspberry pi cpu when playing with flashrom, maybe that could also be used to reduce its power consumption?<p>Not sure what rPi version the author is using, i think i remember i was using the original one ? It was 700mhz by default i think i clocked down to 200mhz and 50mhz. Granted, it was mostly unusable (bye bye openssh) at 50 mhz.
IME, a 50w panel is not enough. You'll never get full output, even in perfect alignment and a sunny day. I probably get an average of 70% or less from my 4x250w panels.
> This means that the solar power cannot provide enough power to charge the battery, although I am using a 60W solar panel for ~5W Raspberry Pi.<p>Just because there's "light" doesn't mean the panel will produce power. The sun has to shine BRIGHTLY.<p>This is why solar power in Berlin or London is a waste of resources, because the sun doesn't shine BRIGHTLY there.