 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 28th T Tuesday Update. Let's get into it. Last week we finally got the parts delivered to ETS, the manufacturer of the circuit board manufacturing company in Albuquerque that's going to be assembling the board's forest but getting all the solder on parts onto the circuit boards. Since then, I did some reviewing for this upcoming artificial life conference that I submitted a paper to a few weeks ago that we haven't really talked about yet. In addition this week, the tiles themselves, the T2TILEs that we're going to have 150, 100, I don't know, somewhere between, somewhere less than 200 because that's the number of boards we have, but it might be close to 200. We'll see how it goes. But the tiles themselves are not enough. We also need the ITCs, the intertile connectors. So I've been trying to find ways to get those manufactured. Got an update on that. And coming up in the coming week, the manufacturing seems like it's likely to start. So that's very exciting. So let's bunch of stuff I want to go through quickly. So yes, so the, you know, I spent 24 years as a professor, but I hate grading. Well, in particular, I hate giving people bad grades because, you know, nobody likes it. It's not fun. I don't like it. So, you know, I sort of feel sort of offended when I have to give bad grades, but you know, I do what I do. So I make up, you know, rules for myself to say that, you know, well, you know, you need this, this, this, you make a rubric that says, you know, you need to go, and then you sort of take it out of your own hands and say, well, sorry, you know, the rubrics is whatever. So I make up these things when I'm reviewing as well and sort of check all the stuff off. And so I did all these papers. I kind of like one of them. The other ones, not so much. So I've been dragging my feet. The deadline is tomorrow, so I'll put in the notes and that'll be done. In addition, there's this, I don't know exactly what to call it, a startup conference, a trade show or something that I agreed to go talk to in Vienna in less than a month now that, you know, they've got me on the schedule. So, you know, right. So, which of these doesn't belong here. But, you know, there it is at the afternoon, first afternoon of the conference. I come right after a talk on pioneering cultured meat. And they have me down for talking about artificial life. More recently they filed up with an email and said, what title do I want? But maybe I'll just leave it as artificial life. I'm not really sure. In addition, there's the next day, a future of AI thing, which was originally structured to be as a debate of some sort. But I'm not sure how that's actually going to come out. So we shall see. But at the very least, I've started to be thinking about what's my talk going to be. It's only 20 minutes. I'm thinking of staying in the framework of artificial life. Start with the meaning of life, like the A-Life video that I assume most folks have seen here. But then end with the T2 tile as an architecture for computing for the future that's going to use artificial life as its fundamental model and do that in 20 minutes with time for questions it could have. So we shall see. But that's starting to develop as well. All right. So the manufacturing, the getting the tiles done. So last week we packaged up all the parts that we've been buying, stuck it in the trunk, drove it over to ETS, dumped it all off on them, took a look at the machine that was going to be putting them together. Since then, I've had a bunch of email back and forth with Robert Evans, who's I think the owner of the place as well, seems to be sort of supervising the T2 manufacturing for me, which is great. And so, you know, the data that I sent him was not in the format that he wanted, which it doesn't actually surprise me. So the KiCAD puts out this, for all the surface mount parts, you need to know exactly where they are down to the fractional millimeter and how they're oriented. So the pick and place machine, once it picks it up off of the tape and spins it around to the correct orientation, knows where to put it down. So this .paz file that KiCAD produces that has this information, position X, position Y, rotation and so forth. But it's printed out, you know, all padded with spaces. It's not a machine readable form. So he was looking for a CSV file, so I made a little purl script. It makes me happy to be able to make a little purl script and just actually get something done without a lot of design fussing and reduce the file down to this. So the designator, the C1s, particularly important ones are these FIDS that stand for Heduchils. They're just actual little target markers. There's three of them so that you get, you know, collinear ones and then an orthogonal one so that the pick and place machine which uses cameras can actually look at the board, find the Heduchils and know where it is relative to all the other things. So the position of the Heduchils in the .paz file or in this case a CSV file sort of defines the coordinate system for all of the other stuff. So I sent it off to him, you know, he was also asking about why the numbers were all like, you know, minus 100 and so forth. And I don't know. That's just what Kaik had produced for me. I did some Googling about it and it seemed like, you know, man, I do not want to mess with this. You know, I could just take the CSV file and just do spreadsheet stuff to it, like move everything around to any given offset. But I did not dare want to go back to the Kaik had thing and try to regenerate the Paz file with some different origin. So we went back a couple of times and Robert, you know, bless him, said, you know, I can figure it out. That's what I was hoping he would say. So we'll see. So that was early last week. So then yesterday, Monday, I just sent him a note saying, is there anything else I can do about, you know, his task is to get this stencil made, the official heavy duty stencil, and also figure out the pick and place stuff from the Paz file thing and then actually, you know, say when is it going to happen? So I got a mail back from him yesterday as well. I was out of town for a couple of days. The stencil is here and I'll be programming the pick and place tomorrow. Probably run them Wednesday. That's tomorrow from now. If you want to come video. Yeah, I want to come video. So if all goes well or if, you know, barring, I guess at this point, we're sort of past all going well and we're at the point of, you know, unless there's a surprising bad thing, it might start to exist or at least the surface mount parts may get mounted on the circuit boards. That's not the whole story this week. So that's very exciting. All right, that's the manufacturing, the boards front, the 3D printing, the cases. Each of these cases that I've been accumulating has 36. Each of these boxes has 36 cases in it. So if we have five boxes, that's 180 cases. We have one, two, three, four. We have five. This fifth box up here is not quite full. We really need just a few more goes on the four ups on the 3D printer and the cases will be ready to go. We can start moving on to the ITC handles. So progress there as well. And that's been a long road. Okay, right. And then it's the intertile connectors. This is the PD, the power and data connector. We also have the PO's, the power onlys, and then a few very special ones, one per lotus, the power injector ones that we use to feed power into them. I haven't talked about them really. But it's the idea that it's 16 pins going to 16 pins. They're just directly mapped straight over with these two by eight with this polarizing rib down one side to make it so that it only goes in the orientation that it wants to go. Here's another picture at it from another angle. And now this particular little header piece, the thing with the two by eight, the female header and the polarizing rib is not that easy to get. I've eventually found three different sources of it. There was the Y pay less solence connector, which is, you know, lovely. And you go to the website and it's all in English, which helps me, but, you know, very expensive, whatever it is, you know, 60 cents each. If you're, 57 cents each, if you're getting a thousand of them, then there was the much cheaper, like 16 cents each from 4UCon. That was a whole production that you remember after you've been here for a while that had a minimum order quantity of 1,000 and a very long lead time. And then I found this third one at this YXCon company that, and of course I didn't really appreciate being a, you know, white American, well, somewhat white. So that, you know, 4UCon and YXCon, they look the same to me, but, you know, one of them is in Taiwan and one of them is in the People's Republic of China. So they don't think each other is all that similar. PCB Way and C2Us are both in the People's Republic. So a company that was in the People's Republic, which is YXCon, seemed like it would be more likely to work. But unlike the 4UCon lovely, lovely data sheet that had no decisions on it, except how many pins did you want, everything else was all set, the XYCon one had to sort of fill in the blank in order to figure out what the part number is with all this different information, which terrifies me, of course, because most of the, you know, you presume all the really important stuff is in Chinese and the stuff that's in English is, you know, as right as it happens to be. So I've been talking, I had sent an order off to Seed Studios to say, can you make the circuit board and get the parts and assemble it? I had first given them the Solon's connector, the SFH-11, blah, blah, blah, and they had specified it and it was incredibly expensive. It was more than the 64 cents each. It was 68 cents each and so that wasn't great. So net result was going to be not quite, you know, $1,400 and plus in a month of working time and so I wanted to see if I could do better than that. So I looked at that at a YXCon data sheet and I made up a part number and that's where we were last week. They had accepted it, they hadn't quoted on it because they said it was pending. A couple of days later after last week's update, indeed they actually came through and they said that it's confirmed the part cost would be 183 bucks which was a lot, lot less than the Solon's connector and in fact you divide this out, it's 20 cents each, which is not bad. I mean the best we ever saw was the 16 cents from 4Uconn but couldn't really buy it and everywhere else it was 60 cents and up and so the total price for 450 from Seed Studios $654 and change if that part number is right and that's the problem. I don't really have a lot of confidence that if I pick the wrong part number that Seed Studios is going to push back and say oh by the way this is not going to fit. I'm just going to go ahead and order them and figure out that it doesn't fit and it'll be on me. The problem is, you know, so let's take a look at this. So this is the XYcon. It's a F185-12 and then you pick number of pins per row. Okay, well so I need 16 pins total arranged in two rows of 8 so you'd think you'd put 08 there. Then it's contact plating full of gold, semi-gold. What do I want? I put a 0 which is the teeniest that it's 0.8 micro inches of gold. Anything less than some small number of micro inches they call gold flash and there are people that say that it doesn't really do any good. The whole point of having gold connectors is that it makes very low friction and very high conductivity so you don't get corrosion problems. You have very solid low resistance connections and so forth. I'll unmate them and mate them again and when you have this gold flash, this flash gold, very thin stuff, apparently it gets worn off fairly quickly. I'm not going to sweat that. I use the no gold at all connectors and I very rarely had any actual trouble. So even flash gold I feel pretty good about and when I was looking for other parts to Google just to compare, would you have any F185-12 blah blah blah A0, the all gold in the thinnest possible plating was the most popular thing. And then what kind of insulator material? I don't even know what kind of insulator material I want. I mean I do know a little bit but again I just picked the one that seemed most popular out there. SY, why SY? It's like Simon says. What packing do you want? A polyethylene bag blah blah blah. I put AY because Googling other things seemed that most popular and then three. The three is totally important because the three is the one that says you actually need the polarizing rib. You can get the exact same part. F185-12 blah blah blah blah one and it doesn't have these plugged ends or it doesn't have the polarizing rib or both or whatever. So this is why I'm so terrified that I guessed wrong on this. In particular, I did not go with F185-12-08. I said F185-12-16. Why did I do that? Because it seemed to me Googling around that in most part numbers that from YXCon, even though it says number of pins per row, it's actually the number of pins on the entire connector. And they go different ways, different companies. For example, the Sullins connector, this DO8 inside the part number there, the D stands for dual or double. It means two row. And the 08 does in fact mean pins per row. So it's a 16 pin connector in two rows of 8. Other folks, they do the number of pins on the entire thing. There's no consistent markings. Oh yeah, and then another wonderful thing, of course. So the 4Ucon guys, who I certainly wish I could do it, could work with them. When they start numbering the pins, see all of these connectors, the ones that are meant with the polarizing rib, they have four pin, the two end rows are not real pins. They're just plugged plastic that allows you to line it up and it allows the header that goes, the shrouded header that goes around the male pins to be sort of uniform all the way around. And so like very reasonable thing, the 4Ucon data sheet does not count the plugged holes one, two, three, four when it starts labeling what the holes are. But the XYCon claims it does one, two, three, four even though one and two are plugged. So what are we even talking about as a row? Is a row with four open pins plus two plugged? Is that supposed to be a six? It's my fault for not reading Chinese. I Googled all kinds of related part numbers and tried to figure it out and really it seems like that's very close to what I was going for. 1216 is what I went for. A0, the thinnest gold flash. The B, whatever that was the type of plastic. The A, coming up polyethylene bag. Type one, I don't even know what it is but it's not the one that I want. And the 1218 is listed as a two times nine position and all these other ones that are in two, 1236 is two times 18 positions. So I went with 16 but I was really hoping that I'd get some pushback from SEED or I went on to PCBWay as well as I talked about last week and for them since I had a customer service person who was dedicated to me that I had interacted with before I said well how about trying to do the PDs and the DOs at the same time since you could click on put two designs on one board you could click on tell PCBWay to make the panel that contains both of these designs and so I tried it. It did not go well my person said thanks for assembly inquire I didn't exactly we had several go rounds my problem was I wanted to send them two sets of Gerber files one for design A, one for design B and have them put them together the long and short of it is with a lot of confusion and this kind of undercut my feeling that it was very nice to have actually interact with because it took a long time to get to the bottom line eventually it couldn't be please make the two order separately but then how did they get panelized together and so forth. I finally made a picture saying here's your interface I click panel by PCBWay I type two into the number of designs that's all accepted how do we get that to happen and the answer is well no we only can panel one same PCB so that was not the greatest interaction. On the other hand I was asking more of PCBWay than I had asked of Seed Studios so I went back and I did just the same thing let's just do the PD's see what their quote is so I took their bomb format and I put the 4Ucon part number on there because that would be the one that I was most confident of what it was but I did list the XYcon and the Solons specifically saying the Solons is not preferred it's too expensive they accepted it they gave me a quote the $287 did not include the part cost that was just the assembly cost and then quickly enough they gave me a price for making the circuit boards and they were reviewing for doing the assembly parts and eventually it came through for $800 in component cost $300 total it turned out it was because they in fact were quoting the digit key part as well so it was $300 less than Seed Studios so that was competitive but it still wasn't really what I wanted to do so I got back with them saying it could be possible to do the XYcon part could that be sourced or is there a reason? What I would really love them to say is that's the wrong part it's upside down it has 16 pieces it has 32 something they'll check again that's where that stands now we'll see what happens the next update will be in a week I think we're going to have manufactured tiles in progress if not in hand hope to see you next week