OLPC XO Repair Data from Costa Rica


Resumen en español al final del artículo

In my recent interview with zunia.org I mentioned that I'd like for OLPC and other ICT4E projects to be more open and share informations such as detailed cost breakdowns, concepts for teacher training, details about Internet connectivity and maintenance systems, processes for community inclusion, and similar artifacts. I strongly believe that such data, information, and knowledge being more widely available would be beneficial to other countries and organizations.

As such I was very happy when I received an e-mail with the subject "Repairs data" from Daniel Castro (Executive Director of Fundación Quirós Tanzi which distributed 1,500 XOs in Costa Rica earlier this year) a couple of days ago:

Thought you'd be interested in some of these numbers. The first column is our parts usage so far, for repair operations. The second is our projection for end of 2012 (we have 1550 deployed XOs), and the third is our projection for 2013, with 5000 XOs.
A lot has been said, or speculated, about repair costs, and this is where we are. When we started the program, we had estimated a 5%/year investment in parts vs initial XO investments. If you see the total for 2013, with 5000 deployed laptos, is $48,584.40, which is about 4,8% of the cost of those 5000 laptops, assuming $200 per XO (prices fluctuate between $190-$210). As you see, we've had to repair keyboards on ~13% of our machines, but they are relatively cheap.

I've seen here and there percentages for discarded computers in other programs (20%, 30%), but this data suggests that the costs of putting those computers back in service is not 20 or 30% of the initial investment, but closer to 5-10%, which is of course a more manageable number. Indeed, if you take a look at our projection for 2012 and assumed each repair case is from a different computer and that we didn't perform the repairs, we would have a total of 289 XO's out of stock, which would be 18,6% of the original deployment.

Upon following up with Daniel he shared some more details about these figures:

All costs are in dollars, the cost per part comes from OLPC, except for the 'tarjetas madre' (motheboards) which is a conservative estimate we made up for this analysis. The cost of actually performing the repairs (transportation, personnel) are not included, as these will vary a lot according to the model used in the particular deployment. We have one full time tech working with us right now and his workload is probably at around 80% of his capacity.

Aside of this in itself being very interesting data Daniel also notes that it's also been the basis for developing a payment policy for students' families:

These numbers are calculated based on our current replacement policy. Seeing that a lot of the damages were due to child's abuse of the machine, we are instituting a payment policy. The student's family will have to pay a symbolic amount when the student damages his computer due to misuse, which goes up for every repeat abuse. We expect this to significantly lower the damages. I'd be more than happy to share the results in a some months if you're interested.

Needless to say that I can't wait to see how things work out over the coming months and years and whether Fundación QT's projections in terms of hardware issues and associated costs hold up.

But for now these data points are certainly a great resource and starting point for other new OLPC projects. I'd really like to thank Daniel and Fundación QT for sharing this information and hope that other projects with follow suit so we can get a more comprehensive overview of maintenance issues and their impacts on ICT4E initiatives.

Resumen en español: Hace poco que mencioné que a mi me gustaría si los proyectos de OLPC o otros tipos de ICT4E fueron más abiertos en cuanto a compartir datos y información sobre aspectos como el mantemiento, la capacitación de docentes y sus costos. Por esto estuve muy feliz cuando me llegó un mensaje de Daniel Castro (Director Ejecutivo de Fundación Quirós Tanzi que distribuyó 1,500 XOs en Costa Rica al inicio del año) llamado "datos de reparaciones". La tabla arriba contiene sus datos y es un buen recurso y punto de partida para otros proyectos de OLPC. Espero que más iniciativas seguirán este ejemplo para que podamos obtener una visión más amplia de los problemas de mantenimiento y sus impactos sobre las iniciativas ICT4E.


Get OLPC News daily - enter your email address:

Related Entries

3 TrackBacks

Resumen en español al final del artículo

When I saw all the interest in the OLPC XO Repair Data from Costa Rica which we pu... [more]

Resumen en español al final del artículo [more]

English summary at the end of the article (We will publish a full English translation next week)

En nuestras series mensuales de entrevistas con personas claves en las implementaciones de OLPC en todo el mundo estamos hablando co... [more]


The amount of repairs will depend on whether the XO's stay in school or travel home.
Base on experience with 100 XO's in U.S., the keyboard failures appear to be about the same for either case. The LCD replacement rate is higher when XO's go home (limited data). The number is 6 keyboards and 3 displays per past school year. I have access to only higher cost keyboard assemblies ($40) and U.S. costs are higher (display at $ 85).

Thanks a lot for sharing this information, much appreciated.


Our computers do go to the children's homes every day. We have no data as to what would happen if they didn't in terms of damages, but our anecdotal evidence does suggest that most damages happen outside the schools. However, the decision to keep the computer's in the school should not be taken lightly, as that directly impacts the possibility for children's empowerment.

Thanks to Daniel Castro for this excellent data and update, and to Christoph for the writeup!

If you have seen the keyboard selection for the 1.75s -- the keyboard is the item that received the most revision and attention on the hardware refresh.

SJ, thanks a lot for the reminder about the new membrane keyboard design (http://www.olpcnews.com/hardware/keyboard/olpc_improves_the_xos_membrane_keyboard.html). Daniel had also referred to it in our e-mail exchange but I cut that part for the sake of brevity.

Is the new keyboard (base upper assembly) backwardly compatible, as an assembly, for the XO-1 ?

If I remember correctly Paul Fox and Martin Langhoff made comments about that question on our original article about the new keyboard design.

In modeling new deployments we use $7.32 per laptop in repair parts excluding manual labor. Adjusting the LCD price and motherboard costs downward to reflect more accurate costs, these Costa Rica numbers are very close to the estimate. With the new keyboard design in XO 1.75, this cost should decline some what.

This repair estimated cost assumes the laptop is taken home. Better training in the care of the laptop can make a significant improvement in repairs required.

Bob, thanks a lot for this information.

As for the "better training" part of your comment: Some of the volunteers in Uruguay have also told me that organizing information-sessions for parents upon the laptop distribution in 1st grade and telling them about the educational value of the XOs have gone a long way in reducing repairs.

Nicaragua also has statistics that show a reduction in repair costs after more extensive training.

Thanks for sharing this information.

I just got back from my hand-holding sessions with Champion Teachers in Occidental Mindoro Philippines. I found out that 15 chargers from one school are already busted. These came about when students started bringing home the XOs. Of course students don't have surge protectors at home so if there are power outages in their areas, the chargers get affected.

I will try to gather data from the tech lead support of each school so we could also share some relevant information regarding repairs.

Tessa, thanks, it would be great to get repair data from the Philippines!


We had an issue with the chargers that resulted in OLPC replacing our chargers. We use the green 'brick' charger. I would suggest raising the issue with OLPC if you can.