Project update 53 of 75
So, good news: the low-cost test PCB with the JAE DC3 Micro-HDMI connector came back successful; a trip to HK yielded some parts for the new printer (which is awesome); some planned trips; With the test PCB ok, the 2.7.5 pre-production test can go ahead.
So, I thought it best to do a quick recap, after having been contacted by a couple of people after the last update. I must apologise to one of them: I went completely ballistic after having very carefully over the past few months outlined exactly why and what decisions have been made, and making it clear that I am expecting people to get in touch on the mailing list (this is a libre project after all) to make their opinions and advice known, not least so that you have the opportunity to take direct responsibility for making this project a success.
I received a message which can be basically summarised as, "what do you think you’re doing wasting time designing a 3D printer, you’re a business, you’re supposed to making us our orders, give me the CAD files and I’ll go get the PCBs made up for myself". Having been up most nights for the past three months, working extremely hard and in a completely open and transparent manner for over six years now, I have to admit I wasn’t really mentally in a position to do anything other than get totally upset and angry on reading that message, which, whilst you may think I might be over-exaggerating as to the way I’ve framed its summary, I obviously can’t show you the original but I really am not laying it on.
Now… I’m trying to think why someone would write something like that, to understand where they’re coming from. What communications failure could have occurred over such a long time, such that they are completely unaware of what it means that this is a libre project, that as the Guardian of the EOMA68 Standard, I’m a Certifier (a type of Trademark, specifically a "Certification Mark" which applies to standards, not to trade) who is not permitted to sell "product" that could possibly be construed as competing with the businesses to which the Certification applies (think "HDMI" or "BLE" if you are not familiar with the concept), that I’ve made it clear many times that there’s not enough funds to complete everything and that I have carefully outlined strategies to bring in extra funds to cover that, in ways that help directly with the existing campaign - all of this done completely transparently. Not to mention that right on the very front page of the Crowd Supply campaign it specifically says, "all CAD files are already available with the exception of the Card which is still GPLv3+ licensed and will be released when the boards ship in order not to disrupt supply chains, contracts with manufacturers and suppliers." Basically, the absolute last thing we need is for someone to come barging in with absolutely no experience in dealing with some of the single-source suppliers, ordering only 20 or 50 parts and then getting angry because, after looking at the incoming email and wondering why this person expects them to sell them only \$2 worth of parts @ \$0.10 per part when their normal MOQ is 3,000, they tell the unknown person basically to get knotted. Word somehow gets back to them that this person is somehow related to me, they won’t speak to me because they’re fed up of having their time wasted: the A20 or other board design which critically depends on that part now has to be redesigned, and the project’s screwed for at least another $2,000 USD round of PCB prototyping and assembly plus many months of searching for a new part and redesigning the PCB to use it. If one even exists.
All of this I’ve described - at least once - in updates over the past year.
So whilst some (or maybe quite a lot, if I am honest) irritation may be evident in the above, it comes attached with a genuine and rather important question that I really could do with some help answering: what the heck am I doing wrong? How could communication and access to information about this project be improved?
Back when the first attempt to use this extremely small connector failed, a hard re-think had to be done. Bear in mind that - as repeated in previous updates - this is not an "avoidable" cost: you cannot know what you do not know until you encounter the problem (often sequentially). Continuing to spend $2,000 USD to get pre-production samples done when the rest of the PCB was fine, that would be simply fiscally irresponsible. So instead, trading time for budget, the decision was made to create a small, two-layer, very low cost PCB onto which a new test footprint could be done, and that put through the production re-flow ovens to see if all the pins successfully connected to the pads.
The good news: the test PCB was a success: Mike kindly sent one of the photos, yesterday, from his factory in Shenzhen. So, as long as that exact same pattern is used, we have a high probability of good yields on the production PCBs. The next phase will be to complete the new HDMI layout review and go to pre-production prototyping for the 2.7.5 PCBs. Once again: only when those are absolutely one hundred percent as best as possible to determine as being suitable to go to production will production actually take place. Not before.
So, as mentioned in previous updates - a recap for those people who may not have been following along - the 3D printer(s) plural which are being designed along a "maximising units per hour per dollar" strategy so as to be able to reduce the three YEARS worth of 3D printing needed down to reasonable and sane levels - are going well. A quick rewind as to why I went to Hong Kong: I’m a guest here in Taiwan as a visitor. Every 90 days it is necessary to leave this beautiful country and to return in order to reset the clock. This apparently can be done indefinitely. However it requires a certain amount of planning: where can I usefully go every three months?
The planned schedule includes a trip to Shenzhen for the Maker Faire (10th to 12th November) as well as FOSDEM 2017. There’s approximately three months between each, they’re useful, so why not. However it didn’t quite sync up with the last trip out of Taiwan (four months not three) so I had to make an extra stop in Hong Kong. Turns out that a friend of mine wanted to make a 3D printer (his first), so in the spare time between waiting for Richard to help with the HDMI Review, and in the extra spare time waiting for that DC3 Test PCB to come back positive, I’ve been helping my friend out. This video clip of the printer in action shows the progress to date.
We’ve yet to get the extruder commissioned, we burned out one of the RAMPS 1.4 MOSFETS on the heated bed after my friend didn’t quite tighten down the screws on the heated bed wires properly; he’s off to the markets to buy a new N-Channel MOSFET (all of $0.50 in Shenzhen, wow!), you can follow along the hilarity in the forum at this test of the "make it as low-cost as possible, pile ‘em’on like ants” way to achieve the maximum units per hour per dollar production target.
Whilst I was there he picked up some 5mm OD metal rods from the HK markets, sawn off literally from 4 m lengths - burrs to cut your fingers thrown in for free - and, after filing them down so that they could go in my backpack through the airport metal detectors and not freak out the security people too much, hopped back on a plane two days later and began assembly. The result: absolutely awesome.
When it’s run at only 50 mm/sec so as to do overhanging parts, they come out absolutely beautifully. Best quality from a 3D printer that I’ve ever made. Pushing speeds up to 150 mm/sec is fine, too. By the time it gets to 200 mm/sec outer edge speeds, 300 mm/sec infill and 350 mm/sec travel speeds things begin to get a little hairy, and I am having to hold the frame down so that it doesn’t shake itself off the table. The result is significant pulling at corners due to the previous layer not having cooled.
The next stage is therefore to try adding enough cooling to make sure that the filament - which is being laid down at a rate so fast that the object is still a blob of jelly for several layers - goes solid before adding the next layer. I naively thought that this could be achieved by monster 120 mm PC fans. This turns out to be completely and utterly wrong: axial fans are good at delivering high speed, low pressure airflow. Try to push that through a small hole, you get back-pressure: airflow just goes round the sides of the blades and it ends up with so much air going backwards that you actually think you got the fan direction wrong. If you’ve seen the Mythbusters "Personal Jetpack" episode you will be familiar with why you don’t design axial fans with a 3 mm gap to the edge of the cowlling like there is on these cheap-and-nasty Shenzhen Market 120 mm knock-offs that don’t even have metal bearings.
Apparently this is another of those things that is supposed to be well-known about 3D printing: you use "blowers" (a.k.a., impellers) which are lower speed, but much higher pressure. Air comes in the centre, goes out the side. I will be using four of them: one is simply not enough for the planned extrusion rate. My friend is going to the markets in Shenzhen to pick some up. Anyway, here’s a current picture where you can see that I’ve created air guides out of masking tape (yes really - extremely light-weight), used coat-hanger mild steel to create a frame to hold the fans above the frame (why waste money on extrusion when absolute rigidity is completely unnecessary), and overall come up with an alien-looking design which is definitely unique.
I’m not entirely sure how I’m going to sell the idea of replacing those two useless 120 mm fans with a pair of round 50 mm impellers each per flexible 12"… air-tube. Perhaps people familiar with the background as to why Angus of AC/DC’s stage costume is a schoolboy outfit will understand when I say that this 3D printer might end up being loved and cherished by a very, very special, rich, successful, creative and highly intelligent (and married) sub-community of the 3D printing world…
Overall though, I am extremely pleased to be able to say that after five attempts at designing (and six or seven at assembling) a 3D printer over the past four years, finally something seems to be actually going right. Can’t do it right until you know what’s wrong, eh?
This is finally coming to a close, there’s only so much space in which to try things so many ways. It’s a very tight space, so a lot of compromises have had to be made. Overall though, the original design which worked perfectly well (paid $10,000 USD to have it done) was absolutely dreadful, but was never actually tested for EMI. It could have failed horribly: we will never know. With Richard’s help, this design has been analysed in excruciating detail over the past couple of months so it is anticipated that, at least for the HDMI part, its EM interference will be about as low as it can possibly get. There needs to be one final round of review, after which this PCB will go to pre-production testing just like the others. When those come back - which would be in about 6-8 weeks time - we’ll know if they can finally go to production. Let me put it another way: with or without the HDMI connector working, they’re going to production.
So, just a reminder: if anyone would like to help this project out by paying for (and preferably using as a developer board to help with software) one of the ten planned 2.7.5 pre-production prototypes, please make yourself known on the mailing list. There’s enough money to cover the Micro-Desktop production (450 units) and the EOMA68-A20 production (800 units) but the less money that comes out of the available budget, the better.
Finally, please, please, people, please understand: this is your project. You helped back it financially but that doesn’t make me the expert: it just makes me the default person who, in the absence of any other input when it’s most needed, will make the decision anyway because I have to, and I know that I have a decision-making ability which is not totally suited to this type of role. When I said that I need your help to make it a success, and that you can follow along and help out at any time, I really meant it. That’s what being a proper libre project is truly about, unlike the ones in recent years that you keep seeing try the "slap the word ‘open’ on it and rake in the money" tactic. The idea is to fully document the techniques, ideas, successes, and failures on a constant transparent and ongoing basis, so that people can learn just as much as I have from the other truly open projects such as the OpenMoko, OpenPandora, and many others.