In 2016, I wrote an article about Luke Leighton's [1] crowdfunding campaign to build a modular, recyclable computer (Figure 1). Three years, and dozens of updates later, the project is about to test its first printed circuit boards (PCBs), and production appears just around the corner (Figure 2). Behind this milestone is a complicated story of changing specifications, the challenges of production in China and Taiwan, personal hardship, and delays; all of which illustrates the challenges that new manufacturers face when bringing open hardware to release.

Figure 1: Luke Leighton has been working towards a module computer for seven years.

Figure 2: Three years after a successful crowdfunding campaign, the project is testing its PCBs.

Leighton's EOMA68 design [2] has several distinguishing features. Designed to be a completely free laptop, it is also designed to be environmentally responsible, with a bamboo frame and a case that can be repaired with a 3D printer or even wooden parts (Figure 3). Technically, its greatest innovation is the storing of computer cards housed in recycled PCMCIA cards that can be easily swapped in and out of the machine (Figure 4). The building of EOMA68 laptops is accompanied by the establishment of an open EOMA68 standard that can be used by anyone. Parts of the project are already certified as free hardware by the Free Software Foundation.

Figure 3: The EOMA68 laptop features a modular case that can be repaired with 3D printing.

Figure 4: Computer cards make swapping operating systems easy.

The project's idealism has attracted loyal supporters, many of whom have volunteered their services to make both the hardware and the standard a reality. Leighton specifically singles out Chris Waid of ThinkPenguin, who sponsored the design of the 15.6-inch laptop housing. Another long-term volunteer is Richard Wilbur, who spent eight months working with high-speed differential pairs necessary for HDMI support.

Leighton himself has spent seven years on the project, first developing the standard and then trying to manufacture the laptop on a shoestring budget, often at personal expense. In the course of his efforts, Leighton has seen his already modest income halved as well as his family's eviction from their home. The experience might have been eased by seeking investors, but Leighton declares that the EOMA68 standard "cannot be allowed to be compromised by profit maximizing. By logical implication, I am absolutely prevented from approaching or accepting all and any share-allocational funding, including standard venture capital investor funding." He is, however, applying for grants from various nonprofit foundations.

Shifting Specifications and Components

Seven years is a long time in computing specifications. It can mean several generations in accepted standards. The EOMA68 standard has seen at least two major revisions: the replacement of SATA with a second USB port, and the removal of gigabyte ethernet with optional USB support. According to Leighton, each of these changes cost around $10,000, and delayed progress by six months. "However," he adds, "without them, EOMA68 would not be a viable long-term standard, so it was costly but absolutely necessary." The alternative would be a standard and product that was obsolete before either was released.

A similar problem has slowed manufacturing. The project requires very specific components, such as PCMCIA connectors and mid-mount USB-OTG and Micro HDMI Type D. "We're currently on the third mid-mount USB-OTG connector, and the fourth Micro HDMI Type D connector," Leighton says. "Each time the chosen component went to end-of-life, we had to do a complete redesign of the PCB." Each change required a thorough review of components, a process that, because of the project's limited budget, could take "3-4 months including component ordering, PCB manufacturing, assembling, [and] then testing. It's amazing how much time is spent waiting for other people."

To further complicate matters, this updating can often not be done by outsiders. "It is the most frustrating and irritating thing in the world to find no contact details for a supposedly open project, and no source or design repositories. [And] on finally making contact, the designer says, 'Oh, I will release the design files when I have finished them. I do not want anyone to criticize or perhaps steal my work.' Such a lack of trust terminates any possibility for others to help you to avoid serious basic design flaws."

Instead, such work has to be done in-house. For instance, after finding that outsourcing CAD and PCB layout was often time-consuming and unsatisfactory in itself, Leighton found it easier to learn this work himself and design the laptop's case as he learned.

Manufacturing in Asia

Like many first-time open hardware manufacturers, Leighton chose to manufacture in China or Taiwan. However, as others have found before him, manufacturing in Asia can be challenging for those who live in Europe or North America.

To start with, parts that are sold in North America may not be available in Asia. Sometimes, the parts have reached their end-of-life in Asia. At other times, components have been sold to larger manufacturers. The only way to get the parts that would be common in North America would be to order them from an American source, which means a six to eight week delay, special paperwork, and additional expenses, including 40 percent import duty.

"It's just absolutely critical to make sure that the components you use are the ones that the factory can actually get," Leighton observes. Except, he adds, "even just finding out what's available is a near-impossible task" – even if you are actually on site. [Sourcing materials] is truly a bazaar, and you are expected to have word-of-mouth contacts. The bottom line is that if you want to design a product that is to be manufactured in volume, make absolutely sure that you scheduled at least 6-12 months for component sourcing. No, that's not a joke or a misprint."

Leighton goes on to advise: "Take the time to establish good relationships with Chinese sourcing agents – respect and pay them adequately. If you happen to have friends there, cherish them. Alternatively, if you can move to Taiwan or Hong Kong, do so, as the cost savings will be immense over time. If not, you have to take into consideration the fact that each PCB and every component sent to you will have a 3-6 weeks delay for arrival and will incur international courier-level costs, plus import duty. Suddenly, the cost of moving seems like a sensible financial decision."

With the Goal in Sight

Leighton admits that some early backers have become discouraged by the delays. However, he adds that "there's nothing I can do about that. If they want to receive a failed product that is non-upgradeable and has no long-term future, I could deliver that to them, but I won't. We have one shot at getting this right. A few years' delay is worth it to create a stable decades-long standard that can be relied on."

Still, with release in sight, Leighton is already looking ahead to what comes next (Figure 5). First priority is a complete free reference design. Although as trademark holder, Leighton cannot compete with licensees of the design, he hopes to establish an EOMA68 certified program to maintain standards. He hopes to establish a foundation to sub-license the EOMA68 certification mark to ensure that everyone will "properly conform to the safety and interoperability aspects." There will be no charge or licensing fee, but certification will be mandatory before any product can be certified as compliant. "This is just how it has to be," Leighton says. "It's primarily down to end-user safety."

Figure 5: An early prototype of the laptop: Note the do-it-yourself housing for the computer card.

In addition, he plans to work on a free, RISC-V processor. It "will be fully libre right to the bedrock: CPU, GPU, VPU, everything. Even the hardware design source code is libre and is already being developed." Ultimately, he hopes to see an EOMA68 prototype and a design that can be used generally in the development of open hardware devices of every form factor.

Meanwhile, Leighton gives this advice to would-be open hardware developers:

• "Do the research. Find similar projects and study them; I learned from projects such as Openmoko and OpenPandora. Hardly anyone aged 25 has even heard of those, despite them being incredibly important and containing extremely valuable lessons.
• "Have a clear goal as to the scope and scale of what you want to achieve. If you only intend to make 50 units, the approach is radically different from wanting to do 1,000 or 10,000 or 100,000 or a million units. Each level of ambition requires a totally different strategy. 50 units, you can easily buy components off of Digi-Key and can use a European or USA-based PCB manufacturer. 5,000 units, you'd best find a Chinese factory.
• "Work out a strategy in advance, which allows you to fund the ongoing development for several years.
• "Don't just set up a blog. Set up a mailing list, because a mailing list lets people talk amongst themselves, whereas a blog restricts them to talking only to you. People who go to the trouble of talking to you and helping you are your front-leaders and your indirect word-of-mouth marketeers. Respect, appreciate, and empower them.
• "Write a blog anyway. Demonstrate to people that you have something rational and useful to say, and they are much more likely to offer assistance and advice. If they learned something from you because you presented it clearly, it should come as no surprise that they want to help you, and you should accept that help and make it easy to find you.
• "Be prepared to write talks and do presentations at conferences. Accept and appreciate that, at the end of the talk, the people who ask questions and also those who come up afterwards to thank and to talk to you would like an opportunity to communicate with you. Listen attentively to what they have to say, and with good grace.
• "Do not make the mistake of ceding control of the business to others. A long-term project such as designing hardware and properly seeing it through to product is something that requires huge sustained year-on-year effort. Plan accordingly and retain 100% control of the business whilst at the same time respecting the value and worth of contributors."

How successful the EOMA68 laptop will be remains to be seen. However, one thing is clear: By being open about the challenges of producing open hardware, Leighton and other first-time entrepreneurs like him are already creating a pool of knowledge that will increase the chance for others to succeed.