Project update 69 of 75
When the Intel Compute Card first came out, it was quite amusing to hear their marketing department claim a world-first on the sixth known worldwide introduction of modular computing. BugLabs, ICE Computers, kangaroo.cc, an expired patent from the 1990s, and of course EOMA68 all beat them to it. Note to Intel: sixth is not first, at all.
It was also quite nice to hear the enthusiasm of their partners, citing phrases and use cases that could have come directly out of the eco-computing white paper that I wrote three years before Intel became all enthusiastic about modular computing.
The shuttering of another Intel initiative, so soon after they abandoned the Compute Stick, did not come as a surprise. Whilst this project is small in comparison to the billion dollars available to Intel, ultimately they are beholden to profit maximisation, and that means it was guaranteed that they would terminate it.
NexDock, sadly, was lulled into believing Intel hype. What particularly shocked me, although it should not have, was hearing that encryption featured heavily in the Compute Card design. What was Intel thinking! Chinese ODMs have a hard enough time with stepping outside the cookie cutter model as it is!
EOMA68 is simple, uses other open standards, is properly open, is a certification mark, and is independent of corporate control, for very good reasons.
So, I just wanted to say thank you to Intel, for doing the expensive job of introducing modular computing to a wider audience, and for screwing it up by trying to add DRM and making it a closed standard where Intel is the sole exclusive source.
By being literally the polar opposite on every single aspect of EOMA68 in every way, and then FAILING (just like the "open" Project Ara from Google), Intel helped make my case and I didn’t have to spend any of the sponsorship money or backers’ pledged funds to do so.
So, Intel, when you have woken up a little bit, do get in touch, maybe you could spend say 10% of the $250,000 USD budget blown on developing custom Intel Compute Card casework and connectors alone, on helping make an Intel EOMA68 Card. Better yet, make one yourselves, and contact me to get it certified as compliant with the EOMA68 standard.
Maybe your team could continue the conversations that you stopped having with me, coincidentally a year before the Compute Card first came out.
I heard from Mike last week. He started on the cards and found a couple of components missing, so had to order those. In the meantime, he had some very practical questions.
In assembling the Micro Desktop, he wanted to know what to do about the arms that hold the card. It turns out that I have put the reset switch in the way. Sigh. So it needs desoldering then reattaching. Times 500. Sigh.
We also discussed making sure the socket is secure, so that pressure on the push button does not rip the socket off the PCB. He is going to track down some flat head bolts that will sit flush on the Micro Desktop wooden case floor.
Also, and this turns out to be really important, the card samples that were hand soldered have the PCMCIA header at a very slight angle, around 2 degrees. Doesn’t sound like much, does it? However, looking at the bare PCB card in the socket of the Micro Desktop shows that the far end of the Card is sitting 2 to 3 mm above the line. Pushing it down level, as it would be in its case and with the rails installed, would twist the PCMCIA header and cause the socket pins to bend slightly. That, in turn, results in so much friction that the amount of pressure needed to eject the card is so high it is in danger of damaging the metal ejector mechanism.
The solution is quite simple: Mike will have one of his engineers design an assembly jig, where the PCMCIA header will be held in place, dead level, whilst it is being soldered onto the Card with a heat gun. It is something of a pain to have to do it this way, as it means quite literally 1,000 hand-assembled cards, as all other components can be SMT machine mounted using pick and place.
These are practical manufacturing details that we should have been encountering two years ago, rather than having to do redesigns due to components going end of life (EOL), and so on. This is, however, just how it is. Accepting that and dealing with the issues as they come up, and moving on step by step is how success is achieved. There is no magic pixie dust here.
The NLnet Foundation approved the Libre RISC-V SoC, and the independent EU review confirmed. This will be discussed on that page in its own update. So, we have access to funding that is initially limited to €50,000 EUR. The limit is a rule that applies per person, per first project.
Therefore, there are a couple of workarounds: 1) find an extra person, and 2) split into smaller projects and complete one, quickly. Over the next few weeks I will apply both strategies, and that should pave the way for EOMA68 to receive funding as well.
Initially, the EOMA68 application was rejected as it was believed impossible to handle two full time projects. I pointed out that much of the time, EOMA68 is waiting often for weeks or months for suppliers, components, or other aspects of manufacturing.
The conversation with Michiel was a delight, I had no idea that NLnet, a registered charity, had access to such resources. They have kept a low profile and would like it to stay that way, because they want to minimise overhead and ensure that the projects they donate to receive as much of the funds as possible. However, they do take a very responsible attitude and make sure that both they and the projects they fund are properly and independently audited, and that’s an unavoidable fixed cost.
Overall I am just deeply impressed by their commitment to ethical funding and to their goal, and I am extremely puzzled as to why I have not heard of them before now, particularly as they funded Haralde Welte’s OsmoconBB research and many other software libre projects.
Still so much to do. It is just one step at a time…