Olimex Announces Their Open Source Laptop

A few months ago at the Hackaday | Belgrade conference, [Tsvetan Usunov], the brains behind Olimex, gave a talk on a project he’s been working on. He’s creating an Open Source Hacker’s Laptop. The impetus for this project came to [Tsvetan] after looking at how many laptops he’s thrown away over the years. Battery capacity degrades, keyboards have a fight with coffee, and manufacturers seem to purposely make laptops hard to repair.

Now, this do it yourself, Open Source Hardware and hacker-friendly laptop is complete. The Olimex TERES I laptop has been built, plastic has been injected into molds, and all the mechanical and electronic CAD files are up on GitHub. This Open Source laptop is done, but you can’t buy it quite yet; for that, we’ll have to wait until Olimex comes back from FOSDEM.

The design of this laptop is completely Open Source. Usually when we hear this phrase, the Open Source part only means the electronics and firmware. Yes, there are exceptions, but the STL files for the PiTop, the ‘3D printable Raspberry Pi laptop’ are not available, rendering the ‘3D printable’ part of PiTop’s marketing splurge incongruent with reality. If you want to build a case for the Open Source laptop to date, [Bunnie]’s Novena, random GitHub repos are the best source. The Olimex TERES I is completely different; not only can you simply buy all the parts for the laptop, the hardware files are going up too. To be fair, this laptop is built with injection molded parts and will probably be extremely difficult to print on a standard desktop filament printer. The effort is there, though, and this laptop can truly be built from source.

As far as specs go, this should be a fairly capable laptop. The core PCB is built around an Allwinner ARM Cortex-A53, sporting 1GB of DDR3L RAM, 4GB of eMMC Flash, WiFi, Bluetooth, a camera, and an 11.6″ 1366×768 display. Compared to an off-the-shelf, bargain-basement consumer craptop, those aren’t great specs, but at least the price is consummate with performance: The TERES I will sell for only €225, or about $250 USD. That’s almost impulse buy territory, and we can’t wait to get our hands on one.

118 thoughts on “Olimex Announces Their Open Source Laptop

        1. It is a pitty that the economics of chipmaking haven’t panned out on this. Nobody will take ARM seriously as a primary computer until you can load one up with 16+ GB of RAM, and nobody will build an ARM chip with a decent memory controller until people take ARM seriously as a desktop/laptop CPU.

          1. @Fritzchen yeah, that you rent from them and never see. I want one I can hold in my hands and run at home. The last ARM server I opened up was crazy small inside, like smaller than mini-ITX and with what looked like a chipset heatsink covering the CPU. 25W of output isn’t hard to deal with when you have monster server fans I guess.

      1. Memory training is very complex, especially for DDR3 on up. Intel and AMD spend tons of time and money qualifying memory IC and module vendors and writing libraries to tune the memory timings at boot. They do this because PC manufacturers depend on removable 3rd party memory modules for built-to-order sales.

        Most high end consumer ARM SoCs end up in products that are low price point and/or tight form factor. Removable memory isn’t necessary since the devices are constrained by a fixed or limited function OS that won’t benefit greatly from future expansion. ARM SoC manufacturers unfortunately don’t have a good reason to support removable memory as they would lose money on that feature.

  1. Wow, good job! As soon as I get enough money, I’ll be looking into getting one. It looks just like the Samsung NP900X4D I’m using right now – I’m wondering if they took one of the Samsung laptops and got the molds/designed around the existing cases =)

    Didn’t see the LCD panel in the “replacement parts” section, hope it’ll be available sooner or later.

    1. Well consider that they won’t have the economies of scale that Dell, Acer, etc do. That makes a huge difference. Injection molding has a big setup cost.

      I’d pay it for a true open source laptop without hidden blobs.

        1. Look into the oema68 project. All the closely integrated ICs are encased in a pcmcia card enclosure, for insertion into a small desktop module cradle, a laptop (3D printable, really. Already done on a FDM.), tablet or phone. This lets you re-use/upgrade the screen, keyboard, power supply, and peripherals. The 1st gen module is expected to ship in about a month for a cost of $60 usd.

          Yes, the first gen is an Allwinner A20, but the oema68 spec is for the physical format and electrical interface. You could build any system you can fit inside. Right now they are focused on non-spyware SoC (not Intel).

      1. Economies of scale do indeed not apply, but neither does anything involving injection molding – they do nothing of that sort. This is a generic Shenzen case they are buying wholesale as-is…

        As for the “laptop” part, Olimex helpfully specified since then (on their blog) that this is meant to be a tinker toy for hackers which nobody should mistake for anything like an actual laptop (seeing as how a laptop very much should be comparable to a desktop except for gaming…):

        “We think of TERES1 as to become portable lab for hackers, to may program Arduino boards, sniff protocols at hardware level, capture analogue signals etc. it will be still good to browse internet, edit text files, code embedded software, but do not expect it to replace your desktop.”

    2. For a niche market piece of hardware it isn’t too bad but it seems they’re already sold out.
      It would be nice if it had a little more memory and a SATA connector to add a 2.5″ hard drive or SSD.

  2. I just love it when people make “OpenSource” products with parts that require binary blobs or signing NDAs to get components to work… To me, it isn’t OpenSource unless everyone can see every line of code.

      1. It’s not like everybody can whip up things on a scale of a whole x68 system…
        My guess is more like “very few people” – but I also guess that there is nothing in terms of hardware choices to realize a project like this 100% open source and open hardware.

    1. I just love it when people make “OpenSource” products with parts that require proprietary chip dies. To me, it isn’t OpenSource unless everyone can see every transistor interconnection.

      I think HaD had a pretty good article about it. It’s good when it’s more open then the status quo, as it advances towards a more open future. Don’t expect everything to happen at once.

      1. I hate it when people make “Open Source” chips with HDL/circuit diagrams, but without publishing the silicon doping details & schematics to build the lithography machine. To me, it isn’t Open Source unless I can build it from a collection of pure elements.

        1. I can’t speak for the grandparent post. Maybe he is just being a purist. I don’t know. For me the problem is support lifetime. Binary blobs mean that you will probably be unable to get software updates long before the hardware quits functioning or is too slow to be useful. I don’t really want to use decade old equipment either but I do think that the now expected cycle of replacing everything every year or two is way too wasteful both in money and in resources. Save some landfill space for our kids’ junk please!

          Unfortunately as far as I am aware anyway there just isn’t anything out there that will run decently without a binary blob. Chip manufacturers are evil bastards! So.. I wouldn’t be so critical of the not entirely open source open sources stuff that is out there either. They are just doing the best they can with what is available.

        1. Neither does this Laptop. EOMA68 is not designed to be a “Development board”. It’s a plugable computer. Where the SoC/CPU can be many types. Allwinner A10/A20(Currently build with 2GiB RAM) /A64/tec. Rockchip rk3288 (Already being developed with 4GiB RAM). Or even a Pi SoC: Broadcom BCM2835/BCM2836 (Pi 2)/BCM2837 (Pi 3) is possible.

          You could even make on with a Cortex M0 or Z80 processor if you like.

          Al those have different capabilities. So breaking out all the pins is impossible if you like interchangeability.

          The A20 Crowdfunding was met and on it’s way to complete.

          And far more open than Olimex. More open in software and more open in hardware and more open in communication.

          So where’s crap?

      1. I stopped checking it out a few years ago after waiting in vain for a few more years before that for anything to actually materialize. And before you mention their current campaign, right now they’re “offering” a frankenhorror only a mother could love – even this… toy is more advanced and usable than that.

    2. As long as the logic board has a standardised interface for the various outputs and inputs then the issue of the chosen logic board can be fixed down the line with a new logic board using a better combination of parts.

      Personally I think using the RPi Compute Module pinout might have been a good idea but when they started the latest board wasn’t available. However RPi is a lot more open than anything AllWinner related, and likely will get upgrades every so often.

      The next issue is the rather poor display, although at 11.6″ it’s not the worst situation.

  3. “but at least the price is consummate with performance: The TERES I will sell for only €225, or about $250”

    No, not even close. I picked up a new i3 Asus with 4G ram, Win 10, and a 1080 screen for $350.

    If you are throwing a laptop away for a bad battery or flakey keyboard you probably shouldnt be buying a hackable laptop. Third party batteries are often $10 shipped from the US on ebay and used keyboard can be swapped in two minutes. I have laptops as old as PIII that are running and still used (running old software for servo drive commissioning)

      1. Wow, that was a fun read. Sounds like the guys first pc build. He didn’t stop and think about why the manuals are more of a technical nature and not friendly “just plug A into B”. Building the computer is surprisingly easy after you know how. Knowing which parts to buy takes way more research time than putting it all together. Especially if you plan on running linux. The only thing you ever really need from a manual is the pinout for the motherboard interfacing with the case. But this guy had his build plan handed to him by another dedicated soul. He should at least feel accomplished! Pleased! Excited! or something positive.

        I’m sure there are PC Master Race sub-tiers and this fellow occupies a lower dungeon. He might have bribed his way through the gate but he will surely fail to maintain his rig and end up buying a console again. He will be cast out and back with the peasantry in a few years.

    1. From the article: “But that’s not really the point with a project like this. Open source hardware is still relatively rare, and Olimex’s project might not result in the computer of my dreams, but it does scratch a certain itch that off-the-shelf proprietary computers never will.”

  4. Waaaaay to expensive for those specs. I can do the same thing with a $60 Beaglebone. Lower the price to $100 and I’ll take another look. The last time I bought a laptop, I spent $250 on eBay and got a Lenovo business class in near-perfect condition with Core i5s, 4GB, etc.

      1. You bloody well should, if you have any intention of actually making any use of it is a practical fashion. You might even find a mint-condition Intel Transformer pad with 3G, GPS, IPS included, for about $120 like I did.

        1. what is your definition of open source hardware?
          diagram? board view? bios code? SMC firmware code?
          all of that is available for X220, and as a bonus you can run it 100% blob free unlike this funny Allwinner toy (1GB of ram aahahahhaha, thats 3 chrome tabs).

          1. Permissive licenses.
            Ware the X220 “open source” The manufacturing files, gerbers, schematics and all code along with the toolchains for compiling it would be freely available with permissive licneces that would allow people to modify and commercially exploit.
            Just because one can get the schematics for 10usd online does not make it opensource, gonkai at most.

            The X220 was also 1299usd when it was launched.
            So comparing a used surplus high volume machine to a new niche small volume device is a bit unfair apples to oranges thing.
            Sure you get more bang for the buck, if that’s your only goal.
            But can you get documentation for even the low level stuff that talks to the battery?
            Can you grab the pcb cad files and do your own modified version of the boards and have it manufactured and it’s legal?

            This is always the same on hackaday comments, some small volume niche opensource product comes along and everyone misses the point of it and complains how mass manufactured thing X is better than it coz they can get it on ebay for less, omitting the the real point like openness of the design, availability of sources and so on.

            I’d compare this thing to the Novena or EOMA68 laptop, where this one is only a fourth or a sixth of the price.
            Hell, as the pcb files are open only your own skill limits you from making a spin of it with RK3288 that has 4gb ram and more stuff.

            Or that’s my take on the subject anyway.

          2. >But can you get documentation for even the low level stuff that talks to the battery?


            >Can you grab the pcb cad files and do your own modified version of the boards and have it manufactured and it’s legal?

            like this https://www.reddit.com/r/thinkpad/comments/5gcszl/update_on_51nb_t70/ ?

            >availability of sources and so on

            Its you who misses Allwinner is a habitual GPL violator and you wont get any blob source code from them

    1. If that is something you claim: no, they don’t. The claim in the article is obviously wrong.

      If that is questioning the claim in the article: it clearly isn’t true, in fact gluing components together could make notebook chassis stiffer, reduce some construction problems and enable even slimmer designs. Instead they are made to be repairable however due to the current state of the art they generally aren’t for amateurs anymore…

    2. He did say “SEEM TO”, he didn’t say they actually do… It’s a suit if hyperbole. But I appreciate his point. I’ve taken apart my fair share of laptops, and one generation model may be easy to take apart and work on, but just one generation later and the RAM is under the keyboard or the hard drive is buried.

      I appreciate this opensource effort, because it starts commoditizing the design and production of a portable computer, or at the very least lowers the barriers to know where to start with your own design! We have this mentality with desktop computers, where you can either buy a generic model, or do it yourself and customize it to death…

      I’ve been waiting for this type of thing to happen for some time, and it’s finally rolling into view!

  5. I agree with the naysayers based purely on specs. Quad core A53 isn’t so bad (not great but it’ll do), it’s the 1GB of RAM and 4GB eMMC that make it uninteresting for me. Seriously I’ve got $35 boards kicking around with 2GB RAM and 16GB eMMC, those components can’t be that expensive! The other big issue is the display, I’m not sure what it is but it’s likely to be a TN panel running off LVDS instead of eDP, iirc there aren’t many models of those and they’re all the worst panels it’s possible to inflict upon human eyes. Low contrast ratio (150-200:1), low colour space coverage, screen door effect, generally horrible.

    Now if they’re using eDP making it relatively easy to find a good replacement panel (maybe the panels used in the XPS 13 would fit), and the board is replaceable with something beefier maybe even with a SODIMM slot on it then I’m in! Just won’t buy the kit with an included terrible display and mainboard.

    1. LVDS is an interface, eDP is an interface. LVDS consumes more power (not much but…) and doesn’t support advanced features like partial display refresh (not that the eDP panels I’ve looked at support that anyway – as it adds costs and complications). IPS is a type of TFT technology as is TN.

      There are plenty of IPS or good quality TN screens available with LVDS interfaces. TN screens can be very good with contrast ratios of 1000:1 and better, 30 bit color resolution etc. In fact some top-of-the-line gaming notebook computers are shipping today with TN screens as the higher quality option! The main problem with TN screens are a smaller view angle, something that isn’t a huge problem for small notebook computers in the first place and something that

      IPS panels tend to be better in general but that is because they are (or were) high-end alternatives. Again there are plenty of good TN panels even using an LVDS interface – the problem in this case being that the A64 only supports one LVDS lane limiting the resolution.

    2. The N116BGE-EA2 shown in the build manual is an eDP 1 lane TN screen with 500:1 contrast, 220 cd/m2, 6 bit colors (2^(6*3) physical colors before things like temporal dithering). It is non-glare and have good response time according to the specs.

    3. As the design will support a 1920×1080 screen the interface will reasonably be a 2 lane eDP which means you can install an IPS panel (though the shipped FHD kit is likely to use one already).

      Now if only this place had edit support for comments…

  6. This is neat, if I could hack it to fit a windows tablet motherboard, you can get those for $10 with 2gb of ram and a quad core Atom, or even get a cherry trail with 4gb for a little more.

  7. I didn’t read anywhere that this was the most powerful lightest cheapeast amazing laptop on the market.

    What I did read is that it is and Open Source laptop.

    Sure you can buy a cheaper faster lighter laptop but – they aint open source.

    Open source doesn’t mean it’s free…….

  8. My dad trod on his laptop, it’s a couple years old. Took about 10 minutes to find a replacement display on eBay and another 10 minutes, after waiting a couple of days for it to arrive, 10 minutes to fit it. Very easy. UK sourced.

    The keyboard died on my £200 chromebook I brought about 4 years ago. 10 minutes to find a replacement on ebay, 10 minutes to fit. Again, UK sourced.

    I agree things can be a pain to fix, but with google by your side, everything comes apart. :)

    Maybe one day, to help reduce landfill, companies will be forced to make their devices easier to repair.

    P.s. The lower end arm based chrome books are very hackable. £250 for quad core + 4gig ram. Check out the ASUS Chromebook Flip C100PA

    1. Acer C710 here, been may main home PC for a while now (on linux) 160gb ssd, 8gb of ram (and one slot still free…) dual core celeron 847. not a rocketship but will happily brows the web / stream hd to my telly and run a VM with windows 7 or xp and various car diagnostic stuff no problemo.. wish more chromebooks were upgradeable like this :(

    2. The problem with repairing notebook computers is that parts are custom made, the motherboard, the sub-boards for connectors, the fans, the heatsinks the keyboard etc. There have been attempts of making standardized designs (Intel did one) but even then not all parts were interchangeable and most manufacturers didn’t even ship such computers. Compal did for two generations IIRC however that meant that the keyboard could be swapped – even the MXM graphics cards (which was/is an Nvidia standard) couldn’t be changed freely due to BIOS compatibility problems.

      I can’t see any standardization in the future that would help consumers repair their machines – if anything the opposite will be true.

  9. Why not build the motherboard for a compute module, which could be switched later, have a receptacle for the ESP32 talking care of Wifi/Bt and with its ULP the battery management. Add another receptacle for an USB C, PD, switch, dp/edp board (learning about usb dev), AS alternative to a HDMI, micro power, mini usb and lan board.
    Then we need a csi bus connector for a camera BUT also a more rugged cable too.

    Make the raspy GPIO pins retractable, add a second PCMCIA bay were you got all the 46 raspi GPIO and some 22 pins from the ESP32 (ex: SD bus), let’s see what people can do with it (reuse of pcmcia/cardbus hardware/cases or low power interconnects..)
    This is would be a great upgrade / recycling of my Newton message pad, it’s also need a new 6 inch 4:3 display with mipi/dsi 2/4 lanes, ePaper could be possible…

      1. I know easier said than done, but at 4) and 6) we leave the DIY, open source idea.

        The core of 2) is to make it as “simple” as possible, staying really modular. Looking at the schematics of the cm expander board, it doesn’t look complicated at all. Or am I wrong?
        Build a very small 4 layer pcb, for lines crossings, keep the “motherboard” simple with 2 layer. (But won’t be big anyway)

        3) the ANX7688 from Analogix can convert the hdmi signal and act as switch and PD. It would be nice, if we could persuade these manufacturer to make an evaluation board that can be recycled for tutorial purpose with raspi compatibility… But this too, should be a module.
        But I also could begin with a simpler board/module, with Hdmi connector/protection, perhaps an usb2 switch and barrel connector…
        More troubles is the display, bc: 4:3 and dsi are out of time, but there are “essays” with rpi & epaper, as there are with esp8266 & rpi.

  10. Yeah, should be simple. All you need to do is:

    1) write a ESP32 firmware for using ESP32 as a WiFi/BT slave (Espressif hasn’t made one yet), also incorporate all the features you want into the firmware – and make it work stably and quickly at that! (quite a task, TBH)
    2) Design a board hosting everything – I have this feeling two layers are pushing it with all the high-speed stuff going on with peripherals, but 4-layer boards are cheap aren’t they? (tip: they aren’t)
    3) Make CM talk USB C and DisplayPort. Is the hardest task actually.
    4) Optimise it for production – only after that, you’re likely looking at 100$ just for the hardware, probably without the CM price included. Then, you need to recoup your software/hardware development costs – 10$ per unit should do, right? (won’t)
    5) Yeah, write software. Software usually is the thing making or breaking the project.
    6) Find enough people to buy it from you so you’re not left with massive debts.

    That said, wondering about what could be done is cool, However, reality sets in pretty quickly. I’m building a Pi Zero phone (ZeroPhone) and, with all the simple circuitry, it’s quite a task anyway, trust me, ZeroPhone project is at least for a year’s worth of work before it’s ready and the thing you describe will be much harder.

  11. I’ve been looking for a more open PC with zero binary blobs, and this is quite interesting. In my opinion this has potential because in addition to the openness, it also looks polished like a professional laptop.
    However the hardware specs seem a bit lacking. I can understand the non upgradable RAM, but then 1GByte seems really scarce for a modern Linux machine. 4G of FLASH is also small. Is there a way to attach an SSD? I guess no, at least a permanent SD card? I hope yes. But then I’d be worried of long term reliability especially since I’ll have to have a swap file with so little RAM. How many USB ports does it have? The spec doesn’t mention that. If I have to carry this around as a laptop, It must be capable to program microcontroller development boards, so at least two USB are needed (one for power/jtag, the other for an FT232) and no a hub is a big no. Also the screen at 1366×768 is just barely acceptable. Come on, smartphones have greater resolutions these days. Finally, can it function with zero binary blobs? Allwinner isn’t on the list of free software enthusiasts, if I recall.

    Summarizing, I believe this is quite close to cross the line between a toy and a real open laptop, but not there yet. People who design this must understand that there are some minimum requirements these days that people expect from a real laptop. It’s not a raspberry pi, it should be something that should withstand daily use without regretting not to have bought a chromebook. My wish list is: 4G RAM, SATA connection to real hard drive, 2 or 3 USB ports, and either a higher resolution screen, or a way to attach an external monitor (VGA, HDMI). I’m not saying it is easy, but to be more than a toy there is a line to cross.

    1. 2x USB, MicroSD, HDMI, audio and power are the external connectors.

      Internally I see a micro USB port, which is most likely an OTG port, and an unused ribbon connector.
      Not seeing any free expansion pins, which is a bit of a shame.

  12. This product is meant for people from the industry, not for the general public. It is NOT ASSEMBLED!!! Most of the people that commented here would never manage to assemble it, anyway. Of course, there are cheaper and faster laptops for watching movies. Can you hook wires to your fancy laptops and attach sensors or extra hardware over I2C, SPI, UART? No, you can’t blink a LED with Dell or HP or whatever.

  13. A lot of people here are missing the point. It’s not intended to compete on price and features but to be open. It should be a lot easier to develop a replacement mainboard for this machine because of the documentation. I for one can’t wait for the first Z80 or 6502 based laptop.

  14. Tell me something, why do i want an open source laptop “fully” replaceble, if it is more expensive than branded ones. And it doesnt have the same power other have.

    Second, 11″ a laptop? that is at the most, a netbook, that is selling for same price 5 years ago and maybe with same specs.

    Is a smartphone less capable than this laptop? cause the only difference i see is the screen size..

    plus laptops allow to change ram.. as keyboards, screens… So i dont see the point of making an expensive, low specs laptopt unless just to fill the hype of open source that it is not!

    I would prefer that someone create the case for currently branded available laptops, would be cheaper to develop, and would allow the same replacement they claim in here… In 4 or 5 laptops that i have had, i dont remember one that i could not change the screen, keyboard, ram, hard drive… I even have one without his down plate because of heating problems and it is working.

    1. Well, people are currently wasting 300$ to have a Raspberry laptop solution (pi-top) which make the Teres-1 relevant because it is more an integrated solution than the pi-top. It can be a good solution (= not hacky like ARM Chromebooks) for a portable native ARM development environment.

  15. Well done Cvetan, and good luck. I’m just thinking that the target audience is not chosen well, what we really need is something with lots of open ports, think arduino or old parallel/serial ports, and analog inputs, where you can just stick/connect wires in easily. None of the present computers allow this, AFAIK. It would be an extraordinary thing for hardware development. Presently you must have a PC with cable to Arduino/Rpi, wires hanging everywhere. Not the best setup for hardware labs.

  16. Is it possible to design something this open using x86?

    Arm is great when you have source code for everything you want to run but I don’t see a whole lot of Arm binaries out there.

    Mainly… can’t run Wine and can’t run the proper drivers for my printer. I bought the damn thing because it said it supported Linux on the box! Sure.. Specifically Ubuntu x86 and pretty much nothing else! I miss the days when openprinting.org had all the information one needed to chose a printer but the day I was shopping I couldn’t find decent information on a single printer in the whole store!

    1. No. Intel and friends are so far in the land of proprietary software and designs, that I suspect they even lost track of what comes from where and why it is there themselves. They probably just keep reusing blobs of binaries and silicon that nobody has any source code for anymore, and that they found on the laptops of their employees 5 years after they retired.

  17. If I bought this, I’d have to:
    Remove the touchpad,
    Replace the whole keyboard (with pok3r or HHKB or V60, DSA or similar keycaps),
    Replace the screen (with an ultra-widescreeen LCD),
    All of which would require – a totally new case.

    By that stage, how much of the original machine would be left?

    1. You’d better check out EOMA68. The laptop is 3d printed and all files are open and parameterized. Thus you can create an laptop with with MX keys keyboard and any screen to your liking.

      It would become rather thick but probably awesome!

  18. Might as well buy a chromebook and install Linux, I just bought a very nice used one for under 50.00 on ebay.ou can pick up a high quality chromebook that can we upgraded (in some cases). But this 1gb BS is starting to get old. I can’t operate on 1gb and haven’t been able to in almost 10 years unless i’m running straight command line and it still sucks,

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s