Solar Powered Wireless Laptop Web Server

I got my laptop wired up to my solar panel this afternoon (Sat May 11th, 2002). (photos)

This web page is hosted on the laptop featured in the photos... so it might go off-line without notice. :)

The panel is a set of 42 Siemens/Arco commercial-grade cells probably running at about 14% efficiency. The cells are nominally 0.57 V open-circuit, 3.0 A short-circuit, so the power would be somewhere lower than the ideal 1.5 W/cell. Each square cell is just a tad over 10cm or 4" on a side.

The original wiring was one string in series, making it about 24 V open circuit, 3.0 A short circuit. I rewired it to have 2 parallel strings giving me 12 V open circuit, 6.0 A short circuit. With the laptop hooked up, it ran at about 10 V, 2 A, with the current rising to meet demand. This was not ideal and would have been operating in it's more efficient band if it were a 12 V nominal panel at maximum power.

As you can see, the system isn't quite for production yet... It could use some silicone to attempt to waterproof it and a sheet of plywood to stabilise it. The meters are interesting, and it would be interesting to get a analog-digital converter to get stats live on the laptop. :)

The laptop is an IBM A21p. It idles at about 20-25 W full screen brightness. Charging seems to make a difference of about 5 W.

The maximum draw I have seen from the solar panel was around 30 W, full brightness, charging the battery, wireless network, playing sound and disk access all at the same time. I didn't try adding more devices such as spinning up the CDROM or adding USB devices...

Shutting off the display reduces idle to 11 W, running at the dimmest screen setting is about 15 W.

I even took out the battery at one point to convince myself that it was working as assumed (and made sure I or the cats didn't walk in front of the panel, shading it).

The IBM AC adaptor is rated at 150 W in AC, 72 W out DC max. The IBM auto/air DC adaptor is rated 136 W in DC max, 72 W out DC max (57.6 W nominal). I ran the output of the panel through the IBM auto/air DC adaptor. The native input of the laptop is 16 V, max 3.36 A. I did not go directly in since I assumed that the AC and DC adaptors do regulation. The panel sits at 24 V open circuit, but dips lower than that... I could have tried a tap off at roughly 16 V, but then I could not use the full panel because I would end up with one string of 19 V open circuit (16 V under load) and a second useless one of 5 V. Dunno, might work! I would not be wasting any power in the adaptor then.

I did some crude tests with different kinds of light...

Conditions                                                Short Open
                                                          mA       V
====================================================================
15 W fluorescent, equivalent to a 60 W incandescent:      23     3.4
outside, overcast, 17:00 today:                          300    10.3
yesterday, full sun, 16:00:                             6000    12.0

Someone asked: How flexible is that panel? Could you make it into a sombrero?

Uh, no. It is flexible, but should not be. It needs to be mounted on a sheet of plywood to properly support and protect it. I am contemplating an amorphous panel from Arbour that could be rolled up, it is that flexible-friendly.

I like the idea of a hat though, since it would provide shade while you are sitting in the park hacking away so that you would avoid sunburn, sunstroke and could actually see your screen in its shade.

I would really like to get this sort of system powering my firewall and ADSL modem.

For a permanent system, you would need to consider insolation duty cycles, battery capacity and solar panel capacity.

The page is actually running off my main server now... To have it running full-time, you would have to have enough battery storage to run it for the longest period of time you anticipate between sunshine. You would also have to have enough solar panel to charge enough battery to run it in that sunshine duty cycle. Overcast conditions reduce the amount of power produced by the panel by a factor of 5 to 10.

I have not done all these calculations. My guesses would start at something like 5 days max between bouts of sunshine, and a duty cycle of 8 hours per day useful charging time and maybe 1 day in 4 of sun. For a 50W machine, that would require 50W * (24h/d) * 5d = 6000Wh of battery (1W = 1A * 1V). In 12V batteries, that would be 500Ah of battery. For the solar panel, if you have (50w * (24h/d) * 4d) / (8h) = 600w of panel, you should be able to keep the batteries topped up sufficiently.

Adjust these numbers for your local conditions. Winter conditions and snow cover of panels will affect this. In our latitude (45 degrees north) a good tilt on the panel is necessary for both keeping the panel clear and for maintaining the most optimal solar angle. Tracking systems do exist, but they are more complicated.

Note: I took these at different exposures to show the detail of the solar panel, wiring, meter displays and laptop screen. The diodes are there to bypass a shaded string. One is not connected because of the string re-wiring.


All photos (c) 2002 Richard Guy Briggs Email Web


Last modified by Richard Guy Briggs on Sat Oct 4 01:13:49 EDT 2003