Back in 2014 I had an idea to add WiFi to a pet rock. https://hackaday.io/project/2058-wifi-rocks .
It turned out that it wasn't going to be feasible with the tiny solar panel that the "pet rock" had on its back. I always liked the idea of having wifi-enabled devices in my garden doing something useful, like telling me the temperature outdoors or how wet or dry the ground is. After inspiration from watching a video from Andreas Spiess (https://youtu.be/37kGva3NW8w) , I thought I'd revisit the project and see what I can do with some of the cheap modules available today.
Now most of the solar powered devices from the local hardware store are garden lights, so I'm starting with some of these as a base. This project will be just a bit of a blog to keep track of things and to share it with the community.
Just a quick update on the Arlec Garden Spotlight ( SKU: LVS020). After removing the solar panel from the spotlight, I disconnected the wires from the main YX8018 board ( with wire cutters ;-) ) . This spotlight also uses the basic driver circuit that many of the cheaper garden lights use. Compared to them, the battery is a bit bigger, and so is the LED, but the general charging method and LED powering is the same.
The sun was out, so some quick tests. The 75x60mm solar cell stats are:
Open Circuit voltage: 6.8V
Short circuit current: 94mA
With 11 cells in series, that breaks down around .62V per cell open circuit. According to Wikipedia, the maximum power point is approx 75-80% of the open circuit voltage. The panel will be around 5.44v based on this approximation.
A nice variety!! Some are linear and some are switch-mode. They all should work with the solar panel in the spotlight, although some have preset MPPT points, so some mods are required if 5.4V is the max power point. The CN3971 boards come default set to 6V, or selectable 6/9/12/18v, but the chip itself is rated from 4.5V and up to 28V.
One problem with the battery in the Solar light is that it doesn't have any protection. If it's discharged too much, the battery will stop working and not be able to be charged any more. Only the TP4056-based charger has built-in protection. Fortunately, one of my previous purchases from Aliexpress was a number of small pack-protection PCBs, so I can easily wire one of these in and utilise one of the other charger modules.
That's it for now. I'll update more once I've done some wiring!!
As I mentioned in the project description, this project is an evolution from my earlier project "WiFi Rocks". I though I'd give it another go and blog my progress.
Back then, I had purchased a little WiFi Linux board to put in the solar-powered rock. The power consumption of this device would have flattened the battery in the rock in no time at all. Given the prevalence of ESP8266 and ESP32 boards around now and their ability to use protocols like MQTT to upload data to the internet, I thought I'd base this new project on these devices.
This time round, I'm not limiting myself to a solar powered rock! I'm going to base it on something I could buy at my local Bunnings store here in Australia. It's a big hardware chain. Kind of like Home Depot in USA. Most of the solar powered things are garden lights of some description. Path lights. Motion sensor floodlights. Regular floodlights. Spotlights. You get the idea. Now many have pretty small solar panels, so I'm looking at something that's a bit bigger. I haven't really planned the high-level requirements yet. Basically I want to be able to use a standard size Li-ion battery for power storage. Could be 18650. Could be 14500. Could be 16340. I've then got the option then to buy a larger capacity battery than the light comes with.
I went there today and ended up with a couple of lights. First one is this one. The "Arlec 30 Lumens LED Garden Spot Solar Light". It's only IP44, so not waterproof. But it should withstand showers. Plus, I can always add some "waterproofing" later (silicon sealant). The reason I got this one is because it has a few nice features:
There's a 18650 battery under the solar panel. Granted it's a puny 1000mAh one, but it'll get an upgrade later.
The LED housing is made of aluminium. Might be good for transferring heat out (and in).
It has a flat back that can be used to mount other accessories like an antenna (more on this later)
The solar panel itself looks like its higher voltage than many of the other solar lamps.
So what do I mean by "looks" like it's higher voltage? Lets go back to some solar basics.
Most of these solar lights use polycrystalline silicon solar cells. A single silicon solar cell will output approximately 0.5V. Some higher efficiency cells output a little more, but 0.5v is a close enough approximation. Now, to charge a battery, we need a voltage slightly above the voltage of the fully charged battery. This can be done a couple of ways. We could boost the voltage of the single solar cell using a boost converter. There are some energy harvesting ICs that will startup at this low voltage. But there are losses involved with boosting from such a low starting point and the system could be more efficient by having a higher input voltage. This can be done by putting the solar cells in series. So depending on the output voltage you need, you just put more cells in the string. Now alot of the smaller garden lights use a single 1.25v NiCad AAA battery for energy storage. To reach or exceed that voltage, multiple silicon cells in series are required. These garden lights often use a single chip "Joule Thief" chip, like YX8018 to drive the single LED once the sun goes down and charge the battery during the day. There's plenty of articles online to describe this process. HERE's one from Analog Devices. It describes a mini 3x3cm solar panel that outputs 2.7V in full sunlight. This would be around 5 cells in series. The little epoxy encased solar panels are actually made up of small pieces of silicon cells soldered together to make the higher voltage string. Embedding the cells in epoxy keeps them together and protected from the elements.
Back to the solar panel on the Arlec Spotlight. I noticed that this panel had 11 cells joined together. So without any real-world testing, I figured that this panel would output...