I gave it a go a few years ago. It seemed like all the solar battery chargers I could find were for lead acid. So I got one of those batteries you'd find in a UPS and a panel. Initially it was working alright but the battery very quickly degraded to the point the pi would run out of power every night. Realistically just a waste of resources unless you are doing it purely for fun.
Isn't fun the main point of artisanal technology projects like this? Did you really expect to save money by using solar at such a micro scale where you get precisely none of the economics of scale? Besides, lowtechmagazine also goes offline from time to time: https://solar.lowtechmagazine.com/about.html#often
It seems like your battery perhaps wasn't big enough, or you haven't taken enough measures to engineer your system. This certainly isn't plug-and-play without getting an enormously oversized solar and battery setup.
It's interesting that they're serving uncompressed text, they could be serving gzipped content-encoding which would save a bit of bandwidth which translates to power usage.
I love the concept, but I'm extremely skeptical that the dithered PNG images are smaller than a (much nicer looking, but lossy) JPEG. It's fine if that's the aesthetic they're going for, but it doesn't seem like it would reduce the server load...
For the record, one of the 640x480 PNGs is 41 KB, and I just saved an elaborate, full-color (and higher resolution) JPEG at a smaller file size (quality was around 50, I think). The JPEG looks completely fine on my (normal DPI) screen.
Edit: It does look like they scale up the PNGs (at least when I view their site), and lossless images definitely scale better. Maybe that was the motivation.
The part that bothers me a bit is that the website doesn’t actually do anything useful - there isn’t really any content. I guess one could argue that the proof of concept of a solar powered server is interesting enough to justify the site’s existence - but Low Tech Magazine has already had a solar-powered version for years AND they actually host interesting content.
If someone hosts a solar-powered website that doesn’t do anything useful, is that really helping the environment? Or is it just contributing to e-waste when the parts eventually fail? And does it really have a smaller environmental footprint than just hosting a site at Neocities or something?
That being said I recognize that this is just someone’s little hobby project, and they are more than welcome to build fun little projects like this and share them if they like. My comments are more directed at a general tendency of techies to engage consumerism under the guise of environmentalism.
As I said, “they are more than welcome to build fun little projects like this and share them if they like”.
But if they are going to imply that this is more environmentally friendly than hosting a site in a data center…
> Data centers and servers traditionally consume large amounts of data [I assume they mean energy?]. But this lightweight solar powered website relies only on energy from the sun.
… then I think it is fair game to examine those claims.
In all likelihood, I'd say their claims are right-ish. The caveats are:
* has their PV cell been made in such a way as to ever reclaim its embodied energy? Smaller panels tend not to achieve this over their lifetimes, larger ones do.
* data centres _where_? If we are talking data centres that feed heat into a district heating system as part of its cooling, then probably not. Free cooled or air conditioned systems probably lose, though.
There's also the matter of designing for a limited medium: over time there's a (very difficult to quantify) carbon saving from serving a site that is small and cacheable vs running some monster SPA with a RoR back end. Think computation-per-request
So.. maybe? It is an apples and apples comparison, but we're not specifying which kind clearly enough.
I’m assuming static content hosted on some serverless architecture (so you’re not reserving CPU cycles when nobody is actually using the site), where the host is able to optimize things pretty heavily.
I’m also assuming that the data center gets a significant amount of their energy from renewable sources. Google, for example, offsets 100% of their energy usage with renewable energy. [1]
Probably good, but again, heat reclamation vs free cooling vs AC also matters. In terms of environmental impact, where the heat goes is of enormous consequence. If Google are using AC, they're raising energy demand and therefore lengthening the road to decarbonisation. If they are using free cooling, that's sort of OK. If they're pumping heat to people's houses, that's fantastic as it displaces their heat demands.
No dispute here that economies of scale apply on the compute side.
Now only if they could demo a k8s cluster on a solar-powered RPI full rack with high-availability hosting a multiplayer game and show rolling software update to v2.
It would be much more energy efficient to just serve the page from some static hosting like GitHub pages where the server is already running and when no one is requesting your page, it uses 0w rather than having to run a full OS constantly.
That way you also avoid the creation of a battery, computer, and solar panels to host a few MB of files.
Given the off-grid mentality that often comes with solar power enthusiasts, hosting on a computer in a different city sort of puts you back on-grid. I like to imagine that the maintainer is also talking to their friends about mesh networking.
For a hobby, sure. Its fun and novel while a few documents sitting on S3 is boring. But from a practical and environmental standpoint, it's purely worse. The OP link is currently dead while I've never seen github pages down after years of using it.
From a practical standpoint Open-Source software is often also a hard sell, if you ignore the value of the gained freedom. This linked project isn't much different from someone running Gentoo with Libreboot and a custom kernel. Is it necessary or practical? Maybe not. Environmentally friendly to recompile everything? Nope. But that person probably learned a lot in the process and had some fun with it.
> I've never seen github pages down after years of using it.
Sure, though with those services the actual uptime doesn't exceed the user account's lifetime.
As hobbies go, I think it's practical. There are applications where handling things in the cloud just isn't an option. Many of them are also cases where you don't have reliable power.
Given the amount of useful information provided by this website, it would be most efficient to write “Battery: 95%” and “* battery level may be incorrect” on a sheet of paper and “host” it in a drawer somewhere.
Oh, that makes me think of another interesting angle: Designing pages to minimize energy consumption on the client device. Of course, that probably reduces server load at the same time for some things - fewer bytes moved over the network helps both sides - but also includes things like less complex CSS/JS.
There’s something charming about a website being tied to something physical like how this one requires sunny weather, reminds me of the blog post “I wish my web server were in the corner of my room” https://interconnected.org/home/2022/10/10/servers
although if this site is truly meant to be environmentally friendly I feel like that’s completely negated by the fact it uses lithium ion batteries, switching to lithium for such a low power project just moves the environmental damage somewhere else
I did a similar project a while back - my two takeaways were that you need a bigger solar panel than you think and a bigger battery than you think. It’s not even cloudy days - the difference between the amount of sun on a sunny day in October and December is significant (at least, in Northern California).
Technically not completely powered by solar power, since they decided to include remote analytics[1] served by a cdn which is probably hosted in one of those data centers they mention. :)
Also, the hardware likely has not been manufactured in a carbon neutral way, shipping of the components created its own footprint (likely via giant oil guzzling ship), and externals (e.g. DNS) are also not running on solar. It’s a cool project but not as environmentally friendly as it might appear.
(If we want to get really pedantic, the solar panels and battery need a few years to pay back their own CO2 debt for when they were produced. This server might need to run 3 years before it’s carbon negative. Maybe even longer because not all of the captured energy can be stored in the battery.)
Pretty cool for the hack factor here. I live in a neighborhood that is on the grid, but subscribed to 100% renewable energy plan.
Way less cooler than this project, but I can claim the same (well, renewable) just by running a website on my laptop! :)
I enjoy looking at dithered images like they are using. It's too bad the lovely dithering turns into periodic-looking patterns when scaled down per the 60vh-width directive in the css.
The extreme power consumption of a RaspberryPI 4 is about 7W [1]. The battery is 144Wh. If battery can discharge 100% that's about 20 hours and about 10 hours for 50% discharge. This all assumes the CPU is pegged at 100% and it's not an older model which consumes less power.
If you put any effort into power optimization you can get an r-pi under 2W at fairly high CPU load. r-pi like could also be one of the much more efficient single board computers that run under 1W
> Certainly would take a lot it seems to drain it to 0% before the sun came up.
It depends.
Different types of batteries have different recommended levels of discharge. Acid batteries should not be discharged more than 50%, while lithium batteries can be discharged up to 80-90%. The Blue Ion 2.0 lithium battery is the best one available, as it can be discharged up to 100%.
I have my router, wifi, and a mac mini server that are connected to a UPS, which is connected to a LiFePO4 battery, which is charged by solar panels. But you don’t see me bragging abo- oh wait.
I am curious about what kind of setup would be optimal to charge a LiOn array, instead of Lead-Acid. Taking into account the requirements of the project, every solar mppt charger for residential use is very over spec.
The typical "solar usb power banks" carry a very small solar cell, and probably don't bother implementing mppt. I am clueless on where to look for a mppt charger for sub 500W installations. Ideally, that supports LiOn to recycle old laptop batteries
I suppose this dithering would decrease PNG file size somewhat, because it would increase the number of repeated pixel subsequences, improving the efficacy of the Huffman coding. But has anyone actually done an analysis of how much?
Definitely Arizona. AFAIK nowhere else does MST without DST. And the photos both appear to be of places in Arizona. Fun fact, within AZ, the Navajo Nation does DST while the rest, including the Hopi Reservation enclosed by the Navajo Nation, does not.
The site owner appears to prefer privacy so I’m not gonna speculate any further.
It's pretty unusual for a party to fly an environmentally-friendly banner overhead, though it's not something I'd put past the Burning Man crowd.
I do like that you went there however, because partying and consumerism overlap substantially. And TFA amounts to green-washed consumerism.
This is coming from someone who has spent entire days working from beaches powered by a portable solar panel. It was certainly a celebration of how far solar panels had progressed that the thing could power my thinkpad all day. But it would have been asinine and dishonest for me to portray it as environmentally-friendly. I just added more e-waste to the existing plentiful powered outlets everywhere situation, because I wanted to be at the beach and play with a solar panel.
- Yes they mention it on their website. Love LowTechMagazine.
I've been meaning to try this out sometime. It'd be a fun side project.