 Computers keep changing the world, but their power and safety is limited by their rigid design. The T2TILE project works for bigger and safer computing using living systems principles. Follow our progress here on T Tuesday Updates. This is the 27th T Tuesday Update. Let's get into it. Last week we were confirming that we were going to be able to do ROHOS production hazardous substances when we make the boards with ETS and we fixed the damn 3D printer after all this time. This week we have kind of a big milestone I'll tell you about it. In the coming week I need to finish the reviewing of the other papers in the ALF conference that I signed up for and then there's also a bunch of other stuff which is in the process as well that I'll talk about as we go along. So let's get into it. On the manufacturing front we had finally at long last gotten to all green on the parts that we needed to assemble the tiles themselves. Everything that was going to be soldered to the circuit board plus the circuit board itself. We had and I had spoken to ETS and they had said yes we're going to do lead free and so forth. So I was like you know is it possible to might it be possible to get started next week and ETS says yes we still have availability. So it meant it was time to actually get the parts together and take them over there. So I made up a little packing sheet with all of the stuff that was going to be needed the circuit boards plus everything that was going to be soldered to the circuit board which means not the beagle bones greens themselves not the LCD not the displace but everything else was all going to get packed up and we started packing it up. We filled a box in the end we filled five boxes and here it is sitting in the house. The connectors the surface mount parts the through hole parts and this square box down here is the circuit boards themselves. The 200 circuit boards that we got from PCB way several weeks ago. We put it all in the trunk of the car and we drove over to ETS. Now all previous times when I visited ETS I had gone through this door which leads to the offices. But this time it was time well yeah and there I am looking cranky as usual but actually I'm pretty excited. For the first time we went through this door which leads to the manufacturing part of the building. And here's our T2 tile five boxes of parts inside ETS along with the packing list I've filled out and so forth now. I had never been in this side it was very interesting to look around. Robert the owner took us around a little bit and showed us some of the stuff. I took just a couple of pictures. I'm going to get back there when we get to the point of actually manufacturing some of these and try to get some video or just for the record. But that will be coming up well we'll see pretty soon now. So this is one of his pick and place machines. It's apparently an older one I wouldn't even know how to evaluate it. But again the whole point of the design of the T2 tile was no heroic engineering. So an old machine is perfect. I really don't think the tolerances are all that incredibly tight. So if you haven't seen one of these things before they're really quite amazing. There's a track that goes along here that's essentially like a conveyor belt. So the board comes in from a previous machine or gets fed in manually by the operator. So the previous step with the stencil and the squeegeeing of the paste is done before here. There's a machine sitting right next to this that I didn't get a picture of that does the squeegeeing automatically. They also have a manual one that we don't know how it's going to work for the T2 tiles. In any event then the board slides in here comes down these rails sits up wherever it is. And this thing has got three pick and place units. All these racks down here there's another picture here it is you can see better connected to each one of these racks is one of these reels. And on each of these reels is the tape containing a teeny teeny tiny service mount part. So the idea is these things come over robotically they come down they pick up the next one from the whatever the appropriate row is for what part they need. They spin around to get it to the position they're going to do take it to the correct position on the board and place it. They pick and place according to the instructions that are programmed into the thing. And if it's all been super optimized with putting the reels in the correct order for what they do the thing can actually pick up multiple parts simultaneously off of multiple different reels. Again whether it'll get that optimized for pitiful little hundred and fifty two hundred tiles who knows. But it's probably this is the machine that it's going to go. Since we were over there and dropped it off. Before we can actually do this I'm having ETS actually make the final stencil according to their own specifications because you know the stencils I was getting with these little things like that. The stencils they use are these big things that have cast aluminum frames and they go into a you know seriously rigid mountain sheen that needs the frames to be a certain size. So I said look you know just have them do it for whatever it is two hundred dollars three hundred dollars something like that depending on how exacting the requirements that teach you to I don't think they're that bad who knows. But so they're now working on that and in fact I've got I got some email later after we had dropped the stuff off saying you know asking questions about the pods file that Kaik had produces they would rather have a CSV spreadsheet. So I converted that and we'll see if that's all going well I'm terribly afraid that somehow some numbers will get dropped and that I mean I guess the real point is that you know when I'm thinking as a as a computer scientist. I imagine everything has to be absolutely perfect but of course this is a much more robust process because it's human in the loop. You know they're going to try to program this machine but then they're going to try it and run one through and say you know did the parts land where they're supposed to land and if something's gone wrong. Hopefully you know it'll be obvious if it misses the board supposed to go if something got rotated or whatever hopefully you know I'll be there something that we can test the first few of these things it will find out that it doesn't work. So it's not like it's ballistic from the moment that I change a spreadsheet number to yes your $15,000 is gone and there's no way back so we shall see but the tile manufacturing has begun. It's possible if all goes well that the assembly could actually start next week again probably not until after T Tuesday update but who knows. We shall see it depends how long it takes to get to the point of being able to send out the stencil apparently his stencil people take a couple of days. We shall see but it was really exciting just to see all the machinery you know I've seen pick and place machines before but never ones that were about to be doing our stuff. So that's the manufacturing front. On the 3D printing scene last week we got the thing fixed as far as the min temp bed error goes. Haven't seen a sign of it yet. We're cranking out the cases again so much so that we were essentially running out of the plastic that we're using for the Prusament Galaxy black. So put in order for a couple more rolls with the Prusa folks but that was going to take a week or so so needed to do something else. So I was kind of excited now to have the printer working well so I tried some other stuff. Check this out this is just one of these things that you download from Thingiverse. Guess what this is? It's one of these Serpinski pyramids. It's a mathematical thing and it's really wild if you watch it printing because essentially it's a single line winding in and out and in and out and up and over and down and around the whole thing. As it's one of those things, you know, as if the entire universe was a single electron zipping all around acting like all of the rest of the material. That's kind of what this thing is like. And it came out great and here it is. Let's take a good look at it. It's, you know, fractal and recursive and, you know, it's incredibly light but it has no supports even though it's got all these big empty overhangs. It's very cool and, you know, now we're never going to see it here with all this light but also it's glowing the dark. So it's obviously the concentration of pure mid power in the universe right here. So that was very fun. But, you know, there was actual stuff that needed to be printed that might not need to be in the galaxy, the Prusman galaxy black. And in particular that was these DO connectors. In addition to the tiles, we need the intertile connectors. We talked about them several times. We've seen this one. This is the PD connector, the power plus data. Those are the ones that are yellow in that picture that go in the within a power within a lotus within a power zone. We also need these DOs, the data only that are going to connect between power zones. And my original idea, I'm not 100% sure if I'm still sold on this was to use a different color plastic for the DO connectors to sort of visually set off the power zones. Now the reason I hesitate about it is I'm not sure how important the power zones are in the grand scheme of the T2 tiles, the grid, what's it going on on the displays and so forth. So I'm not sure. But I was thinking about it. So originally I was imagining that the data only connectors would be gray. So I designed this one. This is a similar thing to the other one. So I did print up a bunch of them. So here's a sheet of the DO connectors getting started up. It's 18 of them, three rows of six. It takes about 12 hours for this to build up. Oh yeah, I didn't mention that, you know, this thing, this is a record for me for a amount of time on a single print. It was about a day and a half. So these took just 12 hours. Then they had to be assembled afterwards. Again, the strap that goes underneath the tab for pushing down on and then the cap that locks the two of them together. I made a box full of them. I did another run. And this will hold us for a while, no problem. And eventually, oh yeah, actually this led to, there was a comment on last week's video from Sam Gentel. One option for the connectors would be to have a specialized tool that hooks around and under them. So like a plastic forceps so that you could just have a single remover rather than having all of these zillions and zillions of these guys. Especially these ones that I was talking about last week that kind of protrude up above the surface of the tile. Thank you for the comment, Sam. It served to remind me how long I've been working on the T2 tile. So I went back in my old directories and dug up this picture from the summer of 2017 when I was doing just that. So I had designed these little plastic little pincer things with the little notches in them that would grab onto the sides of the intertile connectors to pull them out and so forth. And eventually I moved away from this approach to the individual connectors that you see now. Not for any fundamental reason, but just sort of a death of a thousand little design cuts the way these things kind of happen. Part of the problem is that when the tiles are actually hooked up they're quite close together so there isn't that much room for that much width of these forceps to go in between. So they have to be fairly narrow and yet they have to be able not just to pull to yank them out but to exert pushing force on them. And these designs would not push very well because they didn't have enough of a foot. One of the things that this additional tab does is it gives you a flat plate so that you've got the strap on the bottom when you're pulling and you've got this flat plate, this kind of shoe that gets you a nice firm footing when you're pushing. These things didn't have that. You could imagine taking the forceps and sort of building a bit of a shoe up above them. You have to be careful to get it all to fit in. But then it's hard to 3D print because you don't have one side down and so forth. All of these problems could probably be solved and then you end up in the situation where you never have an ITC wrench when you've got it. So in the end I said let's have it travel together. We'll see. But it really did just remind me how long this road has been just to get to where we are here. I dug up a couple of other even older pictures. This is one of the early board designs. The board started out with mounting holes in the corner before it turned into those crazy little notch things and then turned back into mounting holes. But in particular the key here is you can see if you look close that in this one some of the intertile connectors on the board like this one are female and this one and some of them are male like these guys over here. Because the original idea was that it was going to be three male headers and three female headers and the intertile connector was going to be interchange your one male and one female. So you just lined it up and put it in whichever way you did. That turned out to have, it was a cool idea and you couldn't go wrong with it. But it turned out that it made it really difficult if you wanted to also support using ribbon cable because in the ribbon cable world when you have male and female connectors of the male and female they switch the pairs relative to each other. So if you put a male on one end and a female on the other end you do not get a straight through wiring. You get all of the pins, each pair of pins being swapped. I killed a beagle bone green or maybe two. You know, many botans died to bring us this information. So in the end I said okay we're going to make all of the things that are in the position of a ribbon cable be all the same gender so they'll all be straight through. And instead we'll use these polarized rib parts that are so hard to get to. I'm going to move on. Alright, and what's this? Oh yeah, this is an even earlier version of the intertile connector where the idea was you're going to just be able to reach in with your fingernails and grab it around a nice rounded edge and pull it out. That was before I understood the actual height of the LCD and the surface of the tile compared to where the intertile connector was going to be. But eventually it shipped. It arrived yesterday. We've got it. Now we've got two more kilograms of the galaxy black. This will be plenty to finish the tiles. It's possible we may even finish them by next week. We shall see, but we're back in business. Alright, now I'm not going to have time to really go all through this, but as far as the ITC's go, we need the circuit board plus these polarized female headers that I was just talking about. We need hundreds of these things. And I do not want to solder them all up. The idea was to try to get someone, one of these, not just PCB manufacturer, but PCB assembly. And so I have now tried to get a quote from Seed Studios. First, you put in the information for the PCB and you can look at it with the Gerber file. So this is the PD one. I actually, I did make this gap a little bit wider in the design. So there'll be a little bit more clearance. We'll see how that goes. But now you go on to the second step and you add a spreadsheet with the parts that you want to them to put on. So this part number, SFH 11, blah, blah, blah. That's the Digi-Key, Y-Payless, Sullins connector that they want 68 cents a piece for. And we're talking 900 of them. So this total quote was $1,400. An operation fee of $434. Assembly cost, I don't even know where all this stuff came from. And the 60 cents, they were charging me more than I expected because I was expecting it to be like 57 cents each because we're going to need more than 1,000 in the end, but I was only specifying 450 times 2 is 900. So it should have been like 64 cents. Seed Studios was quoting me like 68 cents, not sure what was going on there. When I updated it to, let's do 500 of them then to get the price break. No, it changed. Now instead of actually giving me a quote at all, it said it was unable to match it. Why? Because Seed Studios has only said they had 956 available to ship. They didn't have 1,000. I mean, it's very impressive that the Seed Studios can fear this all out in real time, but it doesn't actually solve my problem. I'm going to skip this stuff, but I ended up coming up with a part number for another company. I don't know if this part number is correct, but I gave it to Seed Studios. I also gave it to PCBway. PCBway, these guys are sort of their website is much more humble in the sense that they say, you know, if you can upload it, upload it, or you can skip this step and just email stuff to me. And everybody who attaches to PCBway gets a special consultant that, you know, for some reason always appears to be female. Who knows what, but I've interacted with Yohanda Chen a couple of times before, like when we had the solder mask issue between the black and the blue and so forth. So I went ahead and I sent a thing in. I'm hoping to get a quote, and I said, well, let's try to do the PDs and the DOs at the same time. That has now gone out. Don't know what's going to happen yet. We'll find out next week. Well, hopefully. And that's the story there. At a time we're building the tiles. Thanks for being here. Hope to see you next week.