EOMA68 Computing Devices

An Earth-friendly way to easily upgrade and fix your own computer

Oct 25, 2017

Project update 54 of 75

Shenzhen Maker Faire 10-12 November 2017

by Luke Kenneth Casson Leighton

This update is primarily about the imminent Shenzhen Maker Faire 2017. The HDMI review continues (leaving time for SZMF). A particularly insightful and revealing conversation takes place on the Reprap Forum.

Revision 2.7.5 EOMA68-A20 HDMI Layout

So covering this first: the review continues, bear in mind that one mistake and it will be yet another $2,000 USD gone of the available budget, and no working HDMI. Also bear in mind that Richard is volunteering his time and expertise, for which all of us should be extremely grateful, not least because of the extremely valuable information that he’s providing. You can see the progress being made here and also help out.

The current issue to be resolved is that the impedance of each HDMI TX/RX line is, according to the CAD software, 89 Ohms when they each need to be 50 Ohms. The differential impedance is reasonably close. Richard is doing some calculations on the surrounding GND keep-out area so as to be able to get the right numbers. According to Analog Devices engineers, the differential-line impedance needs to be twice the single-line impedance, otherwise it defeats the point of the exercise.

All of this is way, way too complicated for me to research: I understand it but there’s absolutely no way I could do it on my own. So, we are relying on Richard… who, once again to reiterate and remind people, is volunteering his time and expertise. Without his help it would be necessary to simply risk another $2,000 USD and potentially end up with an extremely low probability of people getting boards with a working HDMI interface.

So, for those people who are asking, "when will the board be ready, when will it ship?" the answer remains and will always remain, "when it’s ready." Without a firm 100% proven pre-production prototype there is absolutely no point in trying to create any kind of schedule for the factory.

What that does is leave me free to deal with other aspects of this extremely comprehensive project, such as Shenzhen Maker Faire. It’s coming up again for another trip out of the country so as to continue the 90-day-visitor visa stream. The one after this will be FOSDEM 2017.

Shenzhen Maker Faire 2017

If you recall, last year I was riding (or lying down on) the back seat of a coach from Zhuhai to Shenzhen which leaked wonderfully warm exhaust fumes into the passenger compartment. On waking up alive, I passed a huge billboard for Shenzhen Maker Faire 2016: it was the first time I’d heard of it, and it was already the afternoon of the last day. I figured it would be silly to miss the opportunity to present the EOMA68 project there the following year, as well as the 3D printing and ecological and cost-saving aspects of what I’m doing. By giving them a huge list of over seven (and increasing) libre hardware/software projects I received approval for a free booth this year at their 300% over-subscribed website. The page went up this weekend.

I’ve basically been working very, very quickly, in some cases overnight, on a range of GPLv3+ licensed completely open designs, with assistance from various communities, and publishing all modifications as they happen, and all of which are targeted at both cost savings by using much lower-cost MCUs and components, or are based on boards that are extremely common and low-cost so that the components are easy to get hold of, and so on. Here’s a list of the various forums and sub-projects:

isn’t available in Shenzhen and has to be imported!) so it should hugely undercut the troubled Atmel / Microchip. The Due is incredibly (relatively) expensive in Shenzhen

RD3D. Takes six stepper modules, can take an extra two steppers on an expansion header, has an on-board microSD card slot, has five MOSFETs, and yet is still a two-layer board and uses pretty much exactly the same components as RAMPS. However, it also supports both the Arduino and the Due so people can transition from Marlin on slower 16-bit MCUs to the much better RepRapFirmware on a Due. Later (much later) I will work on porting RepRapFirmware to the STL47o.

two-layer and is designed to be hand-assembled. It’s also deliberately relatively large (65 mm x 78 mm) so as to have good thermal dissipation. Also there is far less pressure on the layout when the board is larger. It’s also designed to be usable on other 3D controller boards.

board](http://reprap.org/wiki/A4982-Dual_Stepper_Board), again using a large PCB size and the fact that the A4982 has an exposed pad for much better thermal dissipation. Again, it has the advantage of being hand-assemblable. The one thing, however, about the A4982 is that it doesn’t have the same sort of SPI-based controller interface as the Trinamic steppers…. so I added a ridiculously-low-cost STM8 MCU ($0.24) to control the VREF and MS1/MS2 micro-stepping over I2C instead of using a $1.20 digipot and still not being able to control MS1/MS2. It also might actually be possible to turn it into its own independent board with a G-code control interface over UART.

sensor](http://reprap.org/wiki/FilamentWidthSensorBoard) board which measures a shadow from three different angles in real time, so that changes in filament diameter can be compensated for. After buying some Chinese filament that had a whopping 10% variation in diameter, which due to a square law is an absolutely enormous 19% change in volume. It’s basically a re-implementation of a proprietary design: CC Non-Commercial is unfortunately a proprietary license, which prevents and prohibits anybody but its creator from profiting from it, stymying development.

re-implementation of dc42’s IR sensor board, using the much lower-cost STM8 MCU instead of the ATTiny45.

It’s worth mentioning that during the development of RD3D, a major design flaw was discovered in RAMPS 1.4: anyone using RAMPS 1.4 in their 3D printer should not push the total amount of current it uses beyond 8 Amps (total). That’s only 96 Watts (total) at 12 V, where all of that budget is required for the heated bed alone. A review is here. Basically, be prepared for burn-out, and if you are considering running the heated bed for prolonged periods of time to get up to 100

The status on all of these is that the PCBs have been ordered: half of them the components arrived this morning (Monday 24th at time of writing) and will be shipped to me later today; the other three have just literally been ordered this morning, but are unlikely to be manufactured before SZMF, so I will assemble them actually at the booth, alongside the Riki200 and the Cheap-and-Cheerful Taobao Clone.

The Two 3D Printers Going to SZMF2017

Just as an aside: the Riki200 is turning out to be the absolute best 3D printer I’ve ever made. After six assembly and four design attempts, it’s about time. On the Cheap-and-Cheerful Taobao clone, whilst mine has been okay - some of the 3D-printed parts snapped because the PLA quality is too low but the actual metal components were fine - my friend’s 3D printer has turned out to be a total nightmare. The ridiculous saga involves:

Chinese National so they actually had leverage to get the rest sent along)

actually had ripple in the metal

with the same brand name that I ordered has been fine)

it since I ordered mine… poorly)

the shipper used

We also blew up three sets of RAMPS 1.4 MOSFETs until we noticed that the replacement MK3 Alu heated bed we picked from a China clone has 2 oz copper… meaning that its resistance was half what it should be (0.7 Ohms). We "solved" this by simply… cutting out one of the heated bed tracks! We also destroyed two TMC2100 Polulu steppers because they were reversed, and one A4988 stepper. I’m sure there are a few other things we destroyed as well, but honestly there are so many I’m actually beginning to forget.

Regardless, I’m just absolutely amazed at both my friend’s patience, but also he is being challenged to solve each of the problems, and succeeding, which is great to witness, particularly as this is literally his first ever 3D printer.

The Reprap Forum Discussion

The relevant part starts here. If you recall, I returned to the Reprap forum after a couple of years break, and got the distinct and unnerving feeling that it had turned into a "ghost town." Neil, in a previous update, explained some of the background, but it turns out that there is a lot more than it first seemed.

As you can see from that first post referenced above in the thread, it turns out that a number of long-standing open hardware contributors are getting abusive messages from people who haven’t even bought from them, claiming "how can they possibly have designed and published and sold such utterly dangerous and criminally fraudulent s**t?" It turns out that there are Chinese criminals doing the trick of putting 10 uF capacitors into 100 uF cases, or doing shoddy manufacturing simply to be able to defraud people, and it’s getting dangerous. The designers - whose web site of course the criminals put loud and clear on the fraudulent product - have become so distressed by the torrent of abuse they receive from people deceived by the criminals that they’ve made a conscious decision to cease and desist publishing their work.

However, it turns out that there’s more: many of the open hardware engineers, those that don’t accept the offers of enslavement by proprietary Western companies, instead end up being exploited in some way. The worst and most public case was Makerbot (the people who took published prior art from forum online discussions, patented it, and used it to raise VC funding, but there’s a lot more to it than that). This seems to be a consistent theme, as it’s not just one person: it’s happened to several people, several times, and over the past few decades they’re getting fed up with it. So… they’re going underground.

This is basically a major blow for Open Hardware and for the 3D printing community, particularly in combination with the fact that many people are publishing their work in secret and/or publishing it - late - under proprietary licenses such as the CC "Non-Commercial" license. Development is now fragmented into separate focused communities: if you try to communicate with them and help them out, their reluctance to respond stems from their negative experiences of their work having been exploited. That includes successful, well-known, and well-liked 3D printing companies as well as individuals trying to make money from Open Hardware development.

As a libre hardware and software advocate, this isn’t something I can tolerate. So, I’m asking for ideas and would like to invite people who make a living from Open Hardware, and those people who would also like to support it, to think of some possible solutions. We cannot let criminals and spongers "win": there has to be a way. I’ll be using the designs that I’m making as an experiment to see what works, with one notable exception on what is not up for "negotiation": the 100% commitment to GPLv3+ licensed designs and firmware, published as it is being developed, not "after the fact" when of course it is far too late for anyone to help - is absolute.

As always, you can join the mailing list to get in touch, or, in this particular instance, perhaps join in the Reprap forum thread.

Sign up to receive future updates for EOMA68 Computing Devices.

Subscribe to the Crowd Supply newsletter, highlighting the latest creators and projects