I just got my roughed out solar powered wrt.  Working it through the paces to see if it is viable before building and implementing  an enclosure for it.  Consists of a 9ah agm battery (similar to a ups battery in size)  a battery life solar panel 2.5w version and the router.   Has anyone else put there mesh on solar power or any hints to things to avoid?

Attachments
 tmp_IMAG0542-294614660.jpg [976 KB]

First find out what the power draw is of the device while it is being used.

I would suggest sizing the battery such that it can handle the power draw for 5 days, without a charge, and without discharging itself beyond 50% discharged.

I would also suggest sizing the solar array so that it can recharge the discharged battery bank within one days worth of sun.

Use midwinter for figuring out the amount of time that would be in your area.

Install a low voltage drop out device. Your battery is going to be the most expensive part, so protecting it from being over drawn is worth investing into.

Consider one of the charge controllers that can take higher voltage panels and drop it down to what your battery will use.

Lots of 240W panels with 30+ V for sale now with very attractive pricing, under $1 per watt.

Sizing the battery and solar array in this fashion will keep you on the air reliably. Configure for worse case scenario, 5 days of clouds, midwinter, etc.

For 24/7 operations off of solar power, you end up building something that might at first blush look way overbuilt.

Power consumption varies greatly with the version number of the WRT-54G, with the later models using much less power.

06:20:01 up 1 day, 1:07, load average: 0.14, 0.04, 0.01 with the battery reading ~3.65v

I ran a cron job to output the uptime every 5 minutes to a text file to monitor it. So all I had to do was "tail" it to get the last one when I re powered it on house current.

This is a v3 WRT54G router I have a few others here along with some motorola ones that are support.

I suspect that the battery wasn't fully charged when I started this ... my fault on that thinking that the battery would be shipped fully charged.

So off to charge the other one I received and making sure it is fully charged before starting the next test

This node was active on the mesh in house (5 nodes with a gateway) test but had no Lan connected clients to this one only moving data via wifi.

I see there is a snmpd daemon in the package list might have to add those to the nodes and see the throughput and other over time to see if that affects the battery also.

more to follow.

23:28 was the uptime this go round.

Started with a charged battery at 13.33v and ran it till it quit.  The router did handle traffic (Wifi) no lan clients,  during the time it was running I did check voltage during withn a DMM but didn't leave it connected. 

Running dual batteries should double the runtime but would also add to the weight of the box hanging from the pole.

Since it sunny outside I am going to hook it up and let the cell charge it with the sun and watch the voltage with the DMM.

I agree with KB7SVP's methodology, though it will specify a big system.  Figuring on 500 mA continuously, that's 12 Ah a day, or 60 Ah for 5 days.  To discharge 50%, that means a 120 Ah battery.  At an average of 4 hours of sun a day (the equivalent of the panels putting out full power for 4 hours), you'd need about 15 amps of panel to replenish the charge lost to the router (assuming just the 60 Ah).

Take a look at what the use case is.  If you want continuous operation from solar, you're looking at a system like that.  It should be pretty reliable, and might very well be cheaper/easier than bringing the grid into a remote location.  How many days do you want to be able to go without sun?  (I'd probably pick 3 myself.) 

On the other hand, if you want to be able to run the node on a temporary basis, you can get by with something a lot smaller, especially if you'll be using it in a sunny location during the day.  Remember that lead-acid batteries don't like to sit around at a lower state of charge - it would be preferable to bring it to full at the end of the day.

The charge controllers KB7SVP refers to are called maximum power-point tracking (MPPT) controllers, and are basically switching buck converters.  They use a microprocessor to search for the duty cycle that puts the most current into the battery, and as a side effect can take a higher voltage array which is nice.  They are of course a little more expensive, but for a system around the size of what I described they're not  a bad investment.  In general, a lot of the smaller controllers also have a built-in low-voltage disconnect (LVD) option for shutting loads off when the battery gets too low.

It would be great to have a mesh-wide power monitoring application using a derivative of the cron job listed below.  Is there any way that DC supply voltage to a node can be read and stored?  This data would then be remotely monitored for node 'health'.  This would be HUGE for long-term emcomm applications where system up-time is critical.  

73 from St. Croix de NP2X

Setup here is a WRT54GL on the tower in a large PVC junction box with four VRLA batteries charged by a 30 watt panel through a Steca Solsum 6.6c charge controller. Battery capacity is ~30 Ah. It's been up continuously for the past month, but we've had plenty of sun so it hasn't had much of a capacity test. Skies have been heavily overcast this weekend. I'll be surprised if it survives until tomorrow.

The Solsum has a low-voltage disconnect. I'm keeping a close eye on that, though, since a single battery I had out there froze last winter. Remote voltage monitoring would be useful.

Next step here is to add two Nanostation Loco M2 devices on top of the tower. They'll provide some longer distance, sectorized coverage while the WRT54GL will be more local - perhaps with one 24 dBi dish for a particular point-to-point need. I'm still thinking through the added power draw if I LAN connect them to the WRT54GL (another 2x200 mA?). The Loco M2's will be powered over the single Cat5e that I have running to the tower, but I'm using the data side of the cable for the WRT54GL.

Everything is working on the bench right now. Waiting for better tower work weather before putting it up.