XO Laptop Electrical Considerations at Scale


When working on the TCO analysis of some large scale deployments of ICTs in developing countries I was really concerned with a main impediment to the use in most schools in many rural areas: the availability of electricity.

Fancy XO laptop gang charger

In many countries low cost devices could make a significant difference in education in rural and isolated areas. But of course, most of these areas in developing countries don't have grid electricity, and won't have for many years to come as the infrastructure needed is just too expensive.

For a long time I have been worried about the total energy requirements for a school setup where all the students have laptops, even low-cost low-energy ones.

To start with, the costs of setting up adequate classrooms for the students to work with personal devices are not few, since you normally have to install flooring, windows, window shades, adequate roof, have adequate benches and chairs, and above all that, electricity!

Schools need some ability to charge the devices if they are to be used all day at school as well as at home (where there may or may not be electricity available).

You will be surprised to know that for many African schools that are labeled as having "grid electricity" this might mean 1 light bulb with a cable running 200metres from another building! And just imagine a classroom with 70 to 100 devices all connected and being charged!

The cabling, the cable layout, the good floor and ceilings and the equipment needed to produce a stable and safe electrical installation (and I highlight safe because we are talking about children that might die without the safe part of the electrical setup). And you also have to consider that in most rural environments kids wont have electricity to recharge the device at home.

In many schools I've seen, however, the issue has been solved not by placing plugs available for every child at every bench of the classroom, but by placing a large charging-station at one corner of the classroom, or even in another room.

Homebrew XO laptop gang charger

In these cases you will need also a charging policy, since kids will normally arrive home in the mornings with their computers with no charge left. So you'll have to define charging times, or have extra batteries, so that you ensue that every kid has a charged computer when it is time to work with them in class, and then enough time to recharge them to go home.

But what if the school doesn't have grid electricity? what are the options with solar, wind, water or hybrid energy generation? or with fossil fuel generators?

The Computer Aid report shows OLPC making the lowest energy consumption of all devices tested. The XO is supposed to consume "about 10% of typical laptop" but I found it difficult to find out exactly what the figures were, since my own estimation showed it consumed 11 to 7 watts, depending on if it was being used while charging, charging only or being used for different activities.

This doesn't look like much right? But when you have 1000 XO's, 8 servers and 50 access points it adds up to 30 Kw a day (about the same for a computer lab with 20 PCs), and the investment to provide off-grid energy can be half a million US in solar panels and batteries, then you starting thinking about it in more detail.

Depending on how the device is used, batteries can last 2 to 3 hours, and take about the same time to charge. But performance will be worse as batteries get older (BTW, here I found a very nice articles on XO batteries in Spanish)

I sincerely hope the energy consumption can be reduced for version 2 as they promised. Until then I believe it is almost impossible to provide alternative off-grid energy sources for large ICT school setups, not even if they have XOs. I'm writing a manual on available options, so if you have any examples, devices or ideas please post as a comment.

Update: As Martin Langhoff, OLPC's School Server Architect, points out in the comments OLPC's deployment guide contains quite a bit of information on deployment power requirements. Most interesting of all the (somewhat oddly named) Workbook contains detailed instructions and a spreadsheet for calculating the expected power cost of a large-scale OLPC deployment.

Related Entries

1 TrackBack

When working on the TCO analysis of some large scale deployments of ICTs in developing countries I was really concerned with a main impediment to the use in most ... [more]


Why so much handwaving and inaccurate data? Are there no editors on OLPC News doing basic fact checking... or asking for facts?

OLPC publishes a good guide to power draw of the XOs as part of the deployment guide. In fact, the spreadsheet outlines a lot of additional costs.

It is a bit of a laborious process to work through the spreadsheet for each school, but a deployment team needs to do it at some stage. And we hope it will help shape the deployment plans.

Please do look at the wikipage here http://wiki.laptop.org/go/Deployment_Guide - specially the 'Workbook' and 'Power' sections.

The information is easily available. A search through our wiki would have found it.

And in terms of lowering power consumption,

- the latest 8.2 and 8.2.1 offer an "aggressive power savings" mode

- we're working on 1.5 which is an opportunity to apply some lessons learned -- this isn't too far away

OLPC News team: we put a ton of work in each of these things. Publishing articles that fail at the "did you google that?" test doesn't help anyone.

Huh. So OLPC now has a deployment cost estimator. I scan the wiki on a regular basis for new content and I never found this tool.

Maybe its the term "workbook" which doesn't say much, or the lack of any publicity that this tool existed, but getting huffy that we didn't know about it when Google has only 10 results, total, for olpc "deployment workbook" (which isn't a phrase we'd look for anyway) is a little over the top, Martin.

Hi Wayan,

First: Please do the right thing -- edit this damn article and put the links I provided somewhere. This will avoid misinforming your readers, and is an honest thing to do. Most readers won't get to the comments section.


As you say, Pperhaps you haven't found that page, but you aren't writing an article that claims that it doesn't exist :-)

If someone is writing an article and claims that OLPC doesn't provide guidance on topic X, is it over the top to expect that they googled the keywords? The author didn't spend 10 minutes trying obvious things like


the search for 'power' has the deployment guide as the 3rd hit. Anyone working on deployment planning would perhaps forget to search for 'deployment planning' or 'deployment guide', but surely see that?

Oops! Does she know about google? See:

Deployment guide

Deployment planning

Furthermore, I'd advise that in their research they scan the mailing lists and seriously consider posting some questions. They might get valuable information that sometimes is hard to find or not quite up to date on the Wiki.

People are putting a ton of work into this, doing the best they can. To publish "headline articles" that lack the most basic fact checking is not helpful.

I don't mind fair criticism, but you will find that good quality fair and balanced writing takes a ton of work.

If the OLPC News team does the hard work of providing quality critique, then it's hugely helpful. OTOH, if you just publish shoddy articles, and can't recognise when the article is plain wrong, then you can do a lot of damage.

Please please please, put some pressure on authors to research their articles. Fact check them a bit.

It's honest.

It's the right thing to do.

@both of you and the author...

The magic word is "power" then "school" or "deployment" gets you up there...
Maybe a better wiki tagging could reveal these pages in "electricity" and "energy" searches.

However the I think the whole issue is just a letter difference!
The article talks about "35Kw a day", meaning obviously 35 KWh a day! which is actually the ~6KW for 1000 XO (as estimated by the work sheet) for 6-7 hours of work!

So some refreshing of highschool physics (by everybody)is what is really due here! And the addition of the "h" of course...

As a mental exercise and based on Martin's reference, a school system of 100 XO's could be supplied by a well conditioned adult on a bicycle charger for 15 to 30 minutes (500 watt output). If there were three chargers, 6 hour school day and 8 hour charging day, then only about 1/5 of the adult power and 15 minutes of riding (pys. ed.?) by each XO user would be needed (not that I would have small children riding one of those things). This is about a 25 / 1 ratio of use to work time. As the ratio goes to 100 (less power used) the riding time goes to less than 4 minutes. Just a perspective on human power and the low power requirement of the XO.

Hello everyone and thanks for the comments!
I’m glad that the article served to allow more people to get to know the deployment guide and the cost estimator. I did review OLPC deployment Guide several times, but I always found it lacking so much info that I didn’t check it for these last months. Mea culpa mea culpa! (Ive done the corresponding self-flagellation, don’t worry :)). I’m also very glad that OLPC has created some kind of TCO for deployment, maybe inspired in the one we launched with Jon Camfield in October? Countries always welcome this kind of tools and I’m glad they can have it now.

My only concern here, and this is not without appreciation of the volunteers who invest their time in developing material for OLPC site, is that sometimes the content is a little too overly positive, too "optimized". I know, they want the devices to be used all over the world, but most of the time they don’t really know how REAL the world over there is. We must not forget that we are talking about countries that are making decisions (budgetary decisions) that would affect their future, and I’ve seen too may ICT projects become just a big business for some and a very bad business for the governments and the citizens, that it is my main concern to avoid over-positive estimations of costs over the lifetime of the installations. In this sense I suggest you review the Deployment guide in general in search of this potential “overly positive” information.

So, knowing that we have such a good audience here may I ask the OLPC team to publish more specific data on the XO consumption, for example what it consumes in the average class use, while accessing internet, using the camera, charging only, charging when in use, etc.? I believe this would be a very helpful data and not difficult to obtain from the several deployments, don’t you think? It would also be interesting to know how temperature is affecting battery life and charging times, for example.
Thank you all again for the input and for the good info

PS: and thanks for the corrections about the capital letter in kW.

"PS: and thanks for the corrections about the capital letter in kW"

Wow! You ARE talking for 30KW of power! Is not physics. Math is the problem. :-)
In the most demanding scenario, if all 8 servers (8x200)are streaming video through all 50 access points (50x5) and this is watched by all 1000 XO (1000x10) you are still ~12KW.
But how many movies are they going to watch every day?...

Seriously though, I'm using the XO daily for all kind of things without power-saving, radio on and the screen in full brightness. 5-7W is all it uses, and is overclocked too. So 30KW is a more likely number for 4000 XOs (plus infrastructure) in pretty heavy use.

precisely, I dont see what the problem is. I was refering to "30 kW a day", you are talking about (max) 9 kW an hour. Lets say that the equipment is on 5-6 hours a day in the average school? am I missing something?
Thanks for you data.

No problem!
In my 1st post above I exactly pointed out that you probably mean 30KWh (energy) a day and not 30KW (power) a day and thus was no need for Martin Langhoff's "infuriation"
In your response however, you PS "thanks for the corrections about the capital letter in kW". So I could only assume that you stand by the KW and not KWh. This triggered the second post. :-)

I cannot open the worksheet! It is designed for the proprietary Microsoft Excel program. This project should be using all open formats like ODF.

The worksheet opens and operates perfectly well in OpenOffice.org. Perhaps you could argue that it should be provided in both ODF *and* Excel, but if you have to pick a single format for broadest usability, Excel is the clear winner.

Open source/open standards purism is a good thing and has an important role in a project like this one, but when it comes to getting real-world deployments done, it has to take a back seat to pragmatism.

This is a vital topic. In Nepal, we have concluded there is no viable option to power from the grid. The schools are required to connect to the grid to provide power to the charging stations. They must also provide power for the UPS supporting the server and wireless access points.

We have developed a second generation charging station (see http://redmine.olenepal.org/wiki/deployment/Power for details). The charging station is a wood design which can be fabricated by local woodworkers.

Students use the laptops in school on battery and then return them to the charging station. Students take the laptops home but are discouraged in some schools from taking the AC adapters with them.

More research is needed to determine the usage practice. For example, how many hours per day do the students use the laptops in class? If electricity is not available during parts of the school day, how is the classroom usage of XOs changed? Do the students have access to power at home? How can the usage schedule be adapted at school to allow the children to take a charged laptop home?

The third generation charging station probably should have a single power supply delivering DC to the laptops via permanently attached power leads. This would allow more efficient energy conversion and free the AC adapters to go to the homes.

We probably need to explore having two batteries per laptop as an option. In this case the charging station should be able to charge batteries separate from the laptop.

Naturally, having the students in a school propose off-grid alternatives for their school would be a very productive collaborative learning experience. Imagine several teams - solar, wind, human power, geothermal, .....


So Martin, we have the OLPCorps teams leaving for Rwanda. Do you have advice on how they can power up 100 XO's twice each day?

On electric grid, that's probably 1.5-2x the battery needs of 100 XOs, but for solar its, 2-2.5x as the solar panels would not have the opportunity to recharge the batteries between draw-downs. Or 3x to have a recharge in the middle of the day?