OLPC How To: Solar Powered XO Laptop


I was gently reminded by a caring volunteer from the laptop.org wiki that OLPC never purposefully misled people by stating officially that a solar alternative was available for the XO.

However, it is easy to note by googling through blogs, forums and watching the enthusiastic Nirj Deva, the British congressman, that his belief that these are "sun-powered computers" was, if mistaken, quite widely shared.

The hope and somehow built-up expectation through the media and coffee-shop development experts was that a lot of juice would be provided either by yo-yos, hand cranks, solar panels, or gang charger. So many original supporters did complain or start dreaming too big. People felt misled, though staunch advocates of the official party line will tell them they tripped themselves.

So much so that one good day I got tired of it, and volunteered to get solar on the XO, if for no other reason than to stop the whining and accusations of vaporware. If they wanted solar and the deities didn't give it to them, they could get it from this here Prometheus.

olpc xo power connector
Solar powered XO laptop

Why wasn't solar developed by OLPC Central?

My theory is that it simply fell under the radar, and then got lost everywhere except press commentaries. Fact is, any self-respecting hacker can fix to feed 12 volts to an XO with half an hour of effort, using some odd parts from the garage 'tronics cadavers. This very same kind of hacker tend to feel that such simple pursuits are not worthy of their Matter of Britain time, and maybe they are right.

Not being blessed with that much self respect, and needing to clear the garage too, this was just up my alley. It also helped that I wanted to achieve enduring fame in OLPC's hallowed halls, now that the xoctoplug had been published by someone else. You know, they would say, there goes Yama, the sun god! (actually I've been called a sun god - wiracocha - when I was a kid, but that's another story).

How to grok solar

Such a simple problem has somewhat simple solutions. I needed an XO, which I got after some begging around in Austin (thanks, Anne Gentle!). I needed a plug, since I did not intend to cut the plug off the brick, this being a loaner, and I needed a solar panel.

This last part was easy, I had been running a little research on the side for year and a half with a cheapo Harbor Freight 1.5 W panel, originally intended to help me run a desktop motherboard on solar, but I was stumped at the time by the problem of pulling both 12 V and also 5 V.

olpc xo power connector
OLPC DC power input port

This was no longer a problem, XOs thrive in pretty much whatever you give 'em. So it was written, and it is the true truth. The fact that the panel had been all that time on my roof and still works just fine makes me trust these to recommend them to others.

The plug was the worst. It has been said that the power plugs for the prototypes of the XO came from a bag of components that had been given to Mary Lou Jepsen for free, from a discontinued North Korean-Contra project MIT had been developing in the 90's, and she used these because at the time the nascent OLPC could not afford to go to
Radio Shack for a few standard parts.

Well, I made that up, but it is possible that the real story, when finally gets told, will be almost as weird as that. I won't push further, because finally I went from friend to friend to find Richard Smith as friendly and helpful as they come. He describes himself as the "EC, battery, and power guy" of OLPC, and so I know that the issue is now in good hands, whatever happened before he took over the role.

So then one good day I actually had an hour available before sunset, I made myself a plug off some copper, a Bic pen and some glue, checked everything with my tester one last time, and went outside for the last 5 minutes of sun with a thoroughly discharged XO. I got the pretty orange LED coming on, but no time for more.

xo solar
Yamandu working hard


The XO would not power at all after the sun set, but that was par for what I expected from just a few minutes of solar rays.
Next day during my noon break I asked a colleague to come with me to the parking lot, hooked all together, and powered the machine.

He took a few pictures. First ever XO on solar sighted in the wild! Austin, Texas, 4/23/08 11:58. The shirt and yerbamate gourd are a salute to the valiant OLPC people in Uruguay, ¡Vamo pa'rriba La Celeste que no ni no! Sugar on Solar, April 23 11:58 am

So you did your stunt. What next?

Actually running an XO on solar in a parking lot of Austin is thoroughly meaningless if people cannot use this solution themselves. It's like if someone would dare to present the XO to the UN when supply is sparse, there is no ready instructional content and the mesh looks like a young lady's nylons after a rave. Nobody would do this, right? Except for the right reasons, and that's for another article. We all do owe a lot to Nicholas.

Anyway, I plan to offer some pre-assembled units and all data so other people can build their own or just buy from me. Reinvest the income in research and if any profit, 100% to solar for developing countries projects. I hope to be able to get some donations too, for even though I already have had 3 inquiries, it is unlikely this will become a real business, but who knows, this might be the XO killer peripheral.

Oh, and I also hope I will now be taken seriously for a couple developer's XOs.

The current (partial) solar solution would retail for under $30, and available as soon I get the good plugs, that apparently do exist for purchase. Maybe enough solar kits will be sold so I will be able to afford to buy the XOs I need on eBay :-)

Problem solved...

Well, no. The 1.5 W panel powers the computer fine but does not charge the battery. So I need to keep working. Updates in the XO Solar page of the OLPC wiki.

Related Entries


Thanks so much for your work! (And for the links to the existing over-the-counter stuff you're using). It's good that Richard Smith is communicative with you.

Perhaps one day the Potenco Yo-Yo will not be vaporware as well. But I'm not holding my breath.

Go, Yama! I'm renaming you Yamacandu. This is the spirit of Open Source at its finest. When you see that something is needed, just do it.

Nice post. A couple weeks ago at the OLPC country workshop at 1CC in Cambridge, Richard Smith gave a great presentation on power issues with the laptop. The video of his presentation isn't online yet, but it will probably be put here sometime:


The slides are available here:


He did talk quite a bit about solar power, and showed us a few demonstration panels that they were working on.

Not *quite* the first techy who hacked together a solar panel. I know that the incomparable Mike Lee, who runs the DC learning club (XO group), has been using his solar charger since at least december of last year.


I don't know the history of the solar charger at OLPC, but a trickle charger that is such a large fraction of the cost of the machine is a little bit of a hard sell.

this is really nice. I don't think I have use for a solar battery pack for the XO, but it would be fun to play around with. It would be great to get a few people in central park one day and just experiment with it. I'm sure that reading something on the XO in direct sunlight and having it being charged at the same time would just be the perfect bonus.

Keep up the good work!

2h 40 at 10 W, hey, I got it pretty close, I tested full charge at 174 minutes on a semi-cloudy day. (compare with page 5 of the Slides link Cris points at).
Independent confirmation of results is basic to the advancement of science and knowledge...

Now, just a silly question, why aren't more of these reports public?
Anyone has any idea, suggestion?

Thank you, Cris, for pointing to this info. A community kudos to you.

As to the Potenco, the problem is that you cannot expect the level of power output needed for more than 10 minutes before fatigue shuts down the wetware. Richard indicates it took 2 hours to charge using the crank?

It hurts a bit that the community has to re-invent things because (reliable, usable) info is so hard to find.

Thank you all for your kind comments, I love that Yamacandu moniker, my parents will feel honored as well, on what kind of people I has become my privilege to hang out with.

I bow my head in the direction of DC, and Curiouslee. I will retain, if you don't mind, the privilege of the first to publish a reproducible full charge? :-)

I did the exact same thing with the exact same panel a few weeks after getting my XO. I use that solar panel to top off the charge in my 1985 Oldsmobile b/c I have a bad alternator. I only tried it for a few minutes and it worked very well. I scavenged the plug off an old AC transformer from something I knew I had thrown away(even if you throw away the product, always save the transformer, they can come in handy for other things). Congrats on the success. BTW, those Harbor Freight panels are only 11.99 on sale now.

uh uh uh, John here has a point that requires a quick correction. The $11.99 panels *seem* to work.
*But* to recharge in less that 3 hours you need the full 10W setup, which :-( costs one hundred bucks. Details in the wiki page

Oh darn, it takes two hours to charge and the Smith slides say you get only four hours from a charge. What happened to the original ten-for-one promise? But Smith does say, "Not really viable until Gen 2" which implies it will work for the XO-2. I guess it was a harder problem than anticipated, but we are still going to get there.

I seem to recall posting a response to a post on olpcnews some months ago about using the harbor freight cheapo solar panels for 12volts for the OLPC glad someone else thought of it too.


"which implies it will work for the XO-2. I guess it was a harder problem than anticipated, but we are still going to get there."

Only if we are incorrigible optimists :-)

these are the numbers:
Currently you need 10 W output for 2h 40 (Smith) or 178 minutes (Yamacandu)

This is what the Potenco people found out and reported (nobody has arrived to make public further test data beyond the "preliminary" here. Notice this is no longer in their website)

So, an *adult* can produce an output of 10W for 10 minutes, and then he is "fatigued". Fatigue is a French word that means "Honey, where's the fresh batteries? My back hurts"

So you need, hmmm, 24 fatigued *adults* to achieve a full charge, or maybe only 6 considering that they might rest in half an hour if they are fit. Thus, for just one *adult* to achieve a full charge of an XO-2, it would have to have 1/6th of the power requirements of a current XO, that, let me point out, has already received kudos the world over for being at the very bleeding edge of energy efficiency.

Now, this project is supposed to be for children. I do not know how much energy a 9-year old outputs, 1/4 of an adult? And that kid will do 8 10-minute routines to the point of fatigue per day? (assuming a 300% improvement on present needs, 24 times 10 minutes for a current XO as Richard reports, 1/6 of the kid's time awake). Boy, we can go out later and feel proud of gold medals in boxing and baseball, thanks to the OLPC...

Sorry, won't happen. And if anyone tells us that it will, either they have the wrong information, or they have information they are not disclosing that somehow changes everything that we know of bio-efficiency. Or who knows why... Potenco still indicates their PCGs are meant for OLPC

LOL, I did a Mars Climate Orbiter on those numbers. 2 h 40 is 18 times 10, not 24 times 10... <:-)

As for the weird connector, the power brick for my IBM T-42p laptop charges my OLPC just fine. It's a little loose but it works great I do like only lugging one charger everywhere.

The comment involving 10W/getting fatigued in ten minutes is just silly, since it really depends on how the human body is producing that energy. As OLPC has already stated, the legs (quads) definitely the best source for efficient human power, and cranking a handle is pretty bad (forearm/wrist). Assuming ideal (100% efficient? laziness on my part..) conditions --
10W * 10 minutes = 4000 Joules
4000 Joules ~1.4 Calories

If you get tired from burning 1.4 Calories (out of a daily 1000-3000 or however people eat)..

Sure, the human body isn't that efficient, and neither is the electrical generation. Assuming 10% total efficiency from food->XO-1 bumps that number to 14 Calories. That sounds a lot more reasonable, but keep in mind that most people would be at a much slower rate.. the XO draws ~5-7W under steady use.

7W * 4 hours * 0.10 = 24 calories, and you're left with a fully charged laptop. I'm thinking my efficiency is still far too high, though.

You know, all of these wonderful things that we can do with these laptops makes me wonder just where will the limit be. I mean, I know there are obvious limitations given the hardware and whatnot, but even that does not stop everyone from finding ways around them. There seems to always be something new and interesting to do with these little guys and I for one am very happy to be watching our overall progress with them.

I'm wondering, just how far do you guys think we can take the XO?

Thanks for the update on solar power! I'm extremely interested in being able to charge the XO via off-grid methods.

I keep wishing that there was a crank charger that looked like the little exercise-bike pedal setups that I've seen. There are a few of these items that are JUST the pedals, for people to use while seated in a regular chair. If a full-sized bike can be rigged to charge a battery, why not a little item like this?

For that matter, I've seen a few mini-stair-stepper exercise things that could probably be developed into a charger, also.

I have a feeling that if these items are designed properly, they could become chargers for a wide range of devices, not just XOs. How's this for an advertising slogan: "Exercise your way to a lower electric bill! :)"


Why doesn't someone just open up the XO and resolder another power input plug to one already found easy, like a CD player type plug for example?

I've been called a sun god - wiracocha - when I was a kid, but that's another story).

Sorry, forgot to include the ??

From the last time we discussed this issue it was noted that the diameter of the connectors for the OLPC power and source vendors for it are listed on the laptop.org wiki

"exercise-bike pedal setups" are an intriguing concept, Donna, that definitely make more sense than anything arm-driven, as dax rightly points to. One problem with an "as OLPC has already stated" approach is that so far a lot of what is stated has not been proven... :-(((

The whole point of this note was (the clincher was in a paragraph that somehow got cut out, but anyway is in the wiki), lookit, here we actually did get a full charge of an XO battery using solar. Solar on the XO has been "sold" all over, but had NEVER been proven until this point (except of course for Mike's test that Seth kindly reminded us all of).

Now, human powered full charge of an XO has NEVER been proven, period, especially with a crank or yo-yo. Anytime you hear anybody say that the crank or yo-yo is anything else but well-wishers dreams, please, stand up, and ask them to show proof. There ain't any so far, sorry. My take is that it is impossible that way, and actually I would be happy to be proven wrong. Wish this were solved, for many kids need it in places solar is not the best (cloudy for weeks). Donna and Dax and the quads folks might have a chance with some bicycle contraption, but a full charge has never been proven that way either (your chance for some OLPC fame!).

which brings us to,
refer to the corresponding wikipedia article for the sordid details, but it appears I was quite cute :-) as a little kid, with clear brown hair. There is a tradition among Andean peoples that the sun gods were fair-skinned, with light colored hair. My mom made a point of taking me along to the markets, and thus I was "jiska wiracocha" to the salesladies there (little-sun-god in Aymara). Just if anyone was wondering, I put that for local flavor only. I do not condone passing oneself for a deity, even if one is very cute. I uploaded a picture to my User page in the wiki for a few days.

Ned: wiki don't work. Richard looking to solve this for good.

The XO-2 is supposed to use less power. The question is if it will be low enough to make human power practical.

Looks like its back:

The XO DC jack specs:

* 1.67mm Center pin diameter
* 5.5mm Outer barrel diameter
* 11mm contact length

Mechanical Drawing Image:DCJack.pdf

Note that a mating plug is shown on that drawing.

Some off-the-shelf parts that work:

* I finally found a 5.5x1.75mm connector as part of a laptop plug set made by Velleman. A distributor in New Jersey, Tequipment, sells the plug set (Model# PLUGSP3) for only USD $3.80. You'll have five extra plugs in other sizes to add to your parts box for other projects. Ordering from Tequipment was easy and delivery was prompt. I have photos posted on Flickr showing the plug and the slight modification necessary to use it with the Radio Shack 2-pin adaptaplug socket or you can solder directly to the pins on the back of the plug and seal appropriately. --Mike Lee 21:37, 4 June 2008 (EDT)
* I've been told by Someone Who Should Know that Digikey CP-014-ND is the real thing. I ordered a few, will report on it later. Yamaplos 16:12, 24 April 2008 (EDT)

Caution! These are not the right plugs! I had got my information from the person in charge of power at OLPC, so take any information about this carefully, whatever origin it has. $18 bucks down the drain. Yamaplos 16:42, 27 April 2008 (EDT)

(courtesy Richard Smith and Arjun Sarwal)

* Digikey part #CP-2199-ND (right angle) Current recommended part, tested at the OLPC office in Boston
* Digikey part #CP-2195-ND


sorry. when I said "don't work", I meant "no good". I'm waiting for Richard's results on some over-the-counter plugs he is due to receive Friday.

You may have inspired me to put a section in the http://wiki.laptop.org/go/Simplified_user_guide about power requirements - solar and human-energy power. :) I swear, the first question I get asked when I bring out my XO is still, where's the hand crank?

Nice job, Yama! Thanks for letting me feel I've contributed, however small.

Yama, thanks for sharing. This is more of a journal post than a complete How To. I'm looking forward to your ongoing adventures.

I laughed out load at your tongue-in-cheek comment:

"The plug was the worst. It has been said that the power plugs for the prototypes of the XO came from a bag of components that had been given to Mary Lou Jepsen for free, from a discontinued North Korean-Contra project MIT had been developing in the 90's, and she used these because at the time the nascent OLPC could not afford to go to
Radio Shack for a few standard parts."

For my own experiments, I gave up trying to do full charge time logging until I get peak sun in my area (Washington, D.C.) this summer. When I tried a solar panel with the XO last December in Ohio, and at various times through the winter, I quickly realized I only had about two hours of usable sunlight (but low intensity) during sunny days--not enough to do a full charge test in one sitting. I think when charging time is mentioned, the data should be qualified by location and season information. You have pretty good solar radiation being in Austin, TX, but much of the world is not so lucky. In addition to local/seasonal solar radiation, factors such as panel performance, orientation, efficiency and weather come into play.

Here are some references I am using:


This summer, I plan to tap into the charge circuit of one of my XOs to attach a USB power meter so I can graph the charge curves.

I really like that wiki page, Anne. There was someone on the Spanish side of things that was looking for something like that. And with Chris Leonard hack for translation, it should work for many languages at least at a basic level right now.
I especially like the tag marking that it is Community work. I will be adding that to my pages.

Mike, I will be following those power studies, and might even figure out a way to have someone pick up the tab for my solar-izing (hint hint everyone out there)

The results are in! Richard and Mike concur on the Digikey part #CP-2195-ND as working mostly though not quite. I missed my chance to grab one of them, Richard packaged and sent them to Cambodia after testing, so by the time I got around to ask for a sample (Digikey prices are a bit out of my league for small quantities, they charge a lot for "extra" handling, but that might have to do with them sending you a phone-book size catalog even for a minimum order. If you're part tree-hugger, as I am, you may want to tell them you don't want their catalog, or stay away from them altogether)

AND I got my developer's XO approved! yeeha! It's not here yet, but it is hopefully on its way eventually. It's been a good week. Have a great weekend, everyone!

Yama, just so you know from someone who has ordered from Digikey many times, they use minimal packaging. The order comes in a bubble padded plastic envelope or a small box. There's no catalog. Just a thermal printed scrap of paper with the part numbers.

I imagine the handling fee goes to American labor who have our typical high standard/cost of living. Also imagine the carbon gas trail the little connector leaves behind while in transit.

Congrats on your XO!

I'm down here in wintry (but sunny) New Zealand,& after being roped in for OLPC handouts in the Solomon Islands/New Guinea,have been exploring off grid XO-1 charging.Although I prefer a 20W polyX PV for such field work,considerable mileage has been made with a new 10W CIS (Copper Indium Selenide)PV- these are cheaper than classic Si based panels & indeed perform better in overcast & off angle sun. Even in low angle (but very clear air) conditions the XO 10W solar charging "sweet spot" was matched nicely by this CIS PV.

As a working XO takes ~6W,& with the inbuilt 6.7V 3 Ah battery storing ~20W.h, ~3 hours run time between charges can be expected.This indeed has been the norm in diverse kids XO use here. Allowing for conversion losses,a 10W PV pushing in 600mA will take ~twice this time to recharge.
Smaller panels,such as the 1½W car charger shown above,will take MUCH longer(days)to recharge! These will NOT directly run an out of the box XO drawing ~6Watts, although they can be a useful extender.

My preferred approach has been to use a buffering 12V 7Ah SLA (Sealed Lead Acid) battery -these are a cheap(& often free!) global standard- as a "Portable Power Pak",since such an approach allows energy gathering AWAY from the XO. Let's not forget charging setups left out on the sun can be damaged,stolen or ruined by the wind/rain!

I've been increasingly concerned that OLPC handouts in off grid regions are failing to appreciate the total lack of mains charging power! It's akin to donating a 4WD Mercedes to a village but not providing fuel...
Such remote regions often enjoy abundant sunlight,but unless further funding is provided schools running on the smell of an oily rag will not be available to afford even the simplest solar charging... Stan

Check pictures =>

EXTRA: I've also been exploring use of the NZ made "Smart Drive" washing machine motor-these are widely used in Australia/NZ for microhydro & wind. When exy-cycle connected some 50W can be ergonomically generated, but shipping of such a setup to remote regions is likely to mean they become a $$$ item. As larger PVs are ~US$5-10 a Watt (& much more readily shipped & installed) I see such human powered approaches as being really only an educational curiosity. Stan



We had a stranger walk up to us at our outdoor Philly OLPC XO user's group meeting in a park by the Schuylkill River. He asked us about the crank power and the yoyo generator. I stated that neither were available. Unfortunately, we decided at the last minute to go there, since the 30th Street Train station Wifi were not working for us! Maybe next time, I could bring my 15 Watt solar panel?