Solar Powered Mesh Node Project: Part 1

The goal of this project it to put together a self-contained solar powered node station that can operate indefinitely without external support.  The design strategy will be to shoot low with a minimal configuration to see how short we fall, rather than throwing money at an overkill system to start. This is my first solar power project so it should be a good learning experience.  

Let’s guess at some rough numbers and then pick some parts.  Let’s suppose we want to run an Ubiquity Nanostation which is spec’d at 8W max power consumption.  4W is probably more realistic for an average usage, so 4W x 24h = 96Wh per day.  We’ll use a nominal 12V battery so 96Wh / 12V = 8Ah.

Common sealed lead acid battery capacities include 7Ah which sells for about $20 but we can spend $5 more to almost double the capacity to 12Ah with a battery such as this one:

In the winter in my area we can have as little as 8 hours of daylight.  That means we have to average at least 96Wh / 8h = 12W of solar power over a day.  Accounting for less than peak power due to the angle of the sun during most of the day and less than clear skies, this 25W panel may be just barely adequate but offers the most bang for the buck:
Finally, we need a charge controller to prevent the battery from overcharging and to disconnect the load when the battery becomes too discharged.  This one is about 10x overkill at 20A but the price about the same as lower rated units and it has a number of nice features:
The display shows battery/charge voltage and can be cycled through a number of pages to show info such as solar input amperage and load amperage which is very helpful.  The various voltage parameters are factory configured for lead-acid batteries but can be adjusted to accommodate several alternate battery chemistries.  The 5V USB power ports would be handy for powering a small part 15 access point, microcontroller, etc.

The instructions stress that it is important to connect the battery before connecting the solar panel and to disconnect the solar panel before disconnecting the battery, but it doesn’t say exactly why.

Finally, we need a way to monitor and log how the system is doing over hours, days and weeks. At a minimum I’ll need to record the battery voltage. It would make sense to use the mesh node for connectivity. While a Raspberry Pi might be our first thought due to the built-in Ethernet port, it would have a significant current draw that can’t be hibernated between readings.

I happen to have an Arduino Ethernet shield I picked up at a hamfest that should be much more suitable for this task.  It uses little current and can sleep for most of the time, powering up once or twice an hour to send a data point to a small server somewhere else on the mesh.

In Part 2 we’ll start putting pieces together for some initial tests and eventually talk about mounting hardware and outdoor enclosures.