 Computers keep changing the world, but their power and safety is limited by their rigid design. The T2 Tile project works for bigger and safer computing using living systems principles. Follow our progress here on T Tuesday Updates. This is the 35th T Tuesday Update. Let's get into it. Last time we set a six week goal for episode 40 to actually have a ring lotus of tiles, 133 tiles, mounted on frames, mounted on a wall, and running the MFM code sufficiently so that they could be benchmarked with drug physics. This is the first week. There's been a tremendous amount of action and I have a lot to talk about, so let's go. In the past week, a lot of progress on how we're going to take these tiles, collect them together into clumps, and be able to manage them, put them on the wall, talk about that. And then I want to do a demo of some of the improvements in the software. My goal for next week is that we take the MFMS, the Mobile Feast Machine Simulator, and get the code forked to get MFM T2, which is going to be the engine that's going to run the events on the tiles. And what we see now when we see the green dots just jumping around, that's actually MFMS. That's the simulator running on the tiles. So is it a simulation? Well, it is kind of because the tiles aren't talking to each other yet. So the goal for this week is to get the new version going and get it looking different enough that we can tell it's MFM T2 and not MFMS. All right, let's go. In the news today are top stories. While all this other stuff was going on with the manufacturing of Vienna and so forth, I was also starting to deal with trying to get the other intertile connectors, not the data and power ones, but the data only ones that are supposed to go in between the power zones. And I had an order going in with PCB way needed to finish with them because we had run out of stock on the connectors that we needed. So we had to wait for stock to come in to finish the previous order. So that was going to get folded into this current one. There was a problem that the components were a lot more expensive in the second order than they were in the first one. So I had to kind of go back and forth a little bit with Yolanda, my contact, and they reviewed it to see if the price could be cheaper. And eventually the price for the components, I got to be almost the same as it was on the first order, which is obviously what it should have been because the same component in fact in the same quantity after we had some negotiations. So that order went in and this past week, we got the sample confirmation where they make one and send you pictures of it to make sure it looks good. And here it is. These are the circuit boards for the data only intertile connectors. We only need like 200 of these or 160 of these or something to complete the ring lotus rather than the 400 plus of the data plus power ones. But you know, it looked fine. So I said, let's go for it. And in fact, the order shipped Sunday. And so the current status as of right now is it's supposed to arrive Tuesday, June 4th. That would be today as you're watching it. In fact, it is today now. But there's currently a clearance event which always strikes terror in my heart, especially with the political and economic and international climate that we have today. So the things apparently is still in Shenzhen. And I do not believe it's going to actually be here by Tuesday, by the end of day today. We shall see. So that's that. Also, in the news, ETS has been manufacturing these boards. I picked up 60 a while ago, plus an original set. And I made the Lotus from that 60. I have another 121 ready to go. They needed a few more little parts that were run short on certain things. And so this was the box of the 60 that I got way back when that I took a Lotus full out of. There's still tons of untouched boards in this box of 60. However, now we, whoops, where is it? There we go. We also have another box of 120. That means in my possession, we have over 180 stuffed circuit boards ready to go. Maybe for me or us. I don't know. I think we're going to try to, in a week or two, maybe we'll try to arrange a T2 tile assembly party of some sort. Maybe at UNM. I'll spring for a free lunch for folks that can come in and help assemble all these darn things, like we saw in the video a couple of weeks ago. So hopefully a couple of folks are in the Albuquerque area. They might be able to help out. It could even be fun. Who knows? So that's happening. And that's it for the news. So moving on. Oh yeah. So with the new goal of having a Ring Lotus on the wall and one of the goals, design and prototype rigid frame to hold one tile, inner frame connectors, external frame anchors, and so forth. No sooner than had to raise this goal than in the T2 tile lobby on Gitter. Andrew was all over this. He's a 3D printing guy with experience much more than I. And it was a very, to me, I thought, a wonderful discussion of, well, how can we attach to the wall? Could we use extrusion and make frames and print things or do custom things or would acrylic or all these various possibilities going by very rapidly and saying, well, that's going to run into an issue with this. Can we fix it and so forth? It was great. And so, and Andrew said, you know, could you make a diagram as me? Can I make a diagram showing the relative things? And so I showed, well, I showed the picture from the Gerber files from the KiCAD board layout with a one millimeter grid around it so that it would actually be possible to measure the exact positions of the mounting holes and so forth. That doesn't answer the other key question, which is how far apart are the tiles from each other? I took a couple of them and I took a picture with the ruler on it and it looked like it was about 12 millimeters from one tile to the next with the gap all in. Is that exact? Who knows? It's not going to be perfect because there's natural slop and everything. But that was enough that Andrew was going off and thinking about all these wild things, originally thinking that all these like three print things could attach to a aluminum extrude standard aluminum extrusion frame, but then maybe who needs the frame? Maybe just can make these things that would go together. So the rings, these rings that he's got that are connected, each of them would stretch over several tiles and allow bringing them together and anchoring them. He had a more refined design once the idea was maybe didn't need the extrusion at all and then he's starting to print. And here's the result. It's sort of a funny looking thing, kind of a little bit of lips, especially in the printed in red, but unbelievable. It's day one of the new goal. Right around the same time, Joby said, you know, okay, the assumption that I had been working with and that I'd said was that the idea was the case screws, these four socket screws that we went through trouble to buy a month or so ago, would screw into standoffs and the standoffs would screw into the frame on the wall somehow. And so that if you needed to remove a tile, you'd remove the four screws and that would allow you to take it off. The PCB off of the tile cover. But okay, you figure you take enough because you want to do it. And Joby says, well, you know, you could have a little snap fit connector so that instead of unscrewing the screw to remove a tile from the wall, you just pull it and it goes snap and comes off. And then in fact, you would have cute little ball feet for the things to stand on and it would be a separate step if you actually needed to get into the case to change something or whatever, then you'd unscrew the little ball foot from the case screw and open it up that way. And that sounded very cool. That wasn't enough just to make a suggestion. He prototyped at least two drafts of it, showed a cross section of one where, you know, you got the, this is the foot part where, and the idea was that you just take the screw, the M3 socket cap screw and you just whang it right into the plastic and it'll sort of self tap and it'll be tight enough to hold in particular to snap and pull out. And so he sent the STL files from, you know, he's using Fusion. I'm using auto open scads so I needed to model it myself. And so I tried his thing. I print them up and it kind of worked and it was pretty cool. And so I started making a mock-up of it in open scad. This is one of my early versions. I had a lot of problems with, especially the little lips, the little part here that are meant to hold the snap in, ending up snapping off and so forth. But, you know, I screwed around with it for a while. I mean, just to see. And the key that I liked about it was that I could print two of these that were pre-spaced at about 12 millimeters about the exact gap that was supposed to be between one and the next. And so that would space the tiles appropriately. So, at the same time, Andrew said, oh yeah, the snap fit connectors are a great idea, but especially if you're having problems with strength, that in order to make 3D printing work strong, what you want is wherever something is going to be under tension, you want the printing to run along that direction so that you have a line of plastic in the direction that's being pulled rather than multiple layers of plastic that are easily pulled apart. So if you make the thing square so that you could print it on its side, and he went up and drafted it and came up with a scheme where you'd take a little square and put a captive nut in it and so forth, and printed some up and put up the STL files for that as well. So I pulled those down. So this is his actual STL files in my slicing program. I have not yet tried to model these in OpenSCAD. I tried to print them. I had a little problem the first time, but I tried a little harder and the second time it was looking good and I printed up two sets of these things. So really it's the female connector, the male connector, and this is just a little demo anchor to show how the male anchor would fit into the frame. And so here they are assembled and so forth. And so here it is. And it does have a nice little like that. You put it in, nice snap. And it comes out pretty good. Now unfortunately the trapped nut here, that part is a problem because there's really not enough bolt coming through with the bolts that I have now. It could get longer bolts, but at the moment there's about 2.7, 2.8 if you're lucky millimeters. And that's not actually enough to reliably grab the nut this way. So at the moment I've been playing around with the Joby. Just go ahead and screw straight into the plastic and making it all tight enough. And so I took the idea of having these pairs, having got a little set of two, and make it into having a little spacer. So this gets two sides of one tile onto the next tile and also onto the alternate brick tile down below. And with these, and I added some various improvements, eventually I got to these. So these are the little feet. They have a little hex nut at the bottom for holding them to screw on. I've got a little washer built into them which also makes them print better because it gives them a slightly bigger base. And I printed up a bunch of the straps and the pairs and the nuts. And, oh yeah, at first I was using pliers but now you see I made a wrench for the thing. Here's the improved wrench with the two sides of it. And this thing is great. It's super strong. And if I take a second, so here's a tile. This tile actually I think has a problem with its Southwest connector. So I'm going to have to see if I can diagnose that better. But that's the reason it's being taken out of the step as it goes. So here's a new foot for the thing. It goes in the wrench like this. If you're weirded out by like screeching nails on the chalkboard, watch out for this because when we get down to the bottom the thing is going to scream. But here it is. So now we've got our wrench that we can use to hold it down. And there, see it actually bottoms out. And then we have another foot on it and it goes on to the connectors pretty well. It holds, it pops off. So, very nice. Take all that, put it all together and you can now do this. Look at this. I can pick it up. We've got all of these three buys and actually all kinds of different ways that you can arrange these straps. I don't completely understand it. This is the key master over here, the special one and so on. And I want to see if I don't know if I'm running out of time here. But let's see if I can get anywhere with this. If I can boot this thing up and show a demo of that as well. We're going to run too long. Okay. And so we've now gotten, we'll just leave that. So Wallace is booting up. So almost surely, a couple of things. Almost surely not all of these will come up. And that in fact is the the non-deterministic system D problem that is still being a problem. I still do not have a solution on knowing exactly how these things are going to work out. And of course, it's taking it's sweet time to boot. Oh, there we go. Well, encourage and got fairly far. Wow. Looks like maybe we got a complete boot. I miss it. I missed it. So now I'm talking to the key master here and what I want to do is see if I can demonstrate doing an actual change of code. And if I'm going to skip over it because we're real late on time. But if you look at the testing the Lotus T2 tile Lotus where I tried to push code from one to the other with the CDM copy.msc file. It's very weird. It looks like the thing moves in one direction and then all of a sudden other things start coming in to the thing where it should be going out. Why did that happen? It was a bug. It was a bug in the stats displayed. Now we've got a whole new stats display. Isn't this cool? Look at all the information that we have down the side. We have the grid voltage, the center temperature, the edge temperature, the amount of light we're getting. We got the frequency that the CPU is running at. By default, I set these things to also 120 megahertz. They can go a thousand megahertz. They can go a gigahertz, but it generates more heat. By the way, the fact that we now have these things up on little feet to get clear underneath it makes the operating temperature quite a bit lower and tipping it up so that you can just get convection going through it makes it a little bit better as well. So I'm having a little bit less of a worry about the whole thing melting down because of the heat. And in fact, I'm hoping we will be able to benchmark it at a one gigahertz when it comes time to do it. So now they all automatically display the debugging output when they first start up and then they switch to the MFM. But I'd rather see the debugging thing permanently if I could. And let's see. So this is the Perl script that's actually running this thing, the display, the debug display and so on. So it's TNV. Here it is. Okay. So there it is. See, I already did it. Well, that's okay. Well, all right. That's fine. So look, this guy says demo here on the last line. None of these guys do. That's the demo I'm preparing to do. And oh, right. There it is. So the idea is, is the way we make this move is the make file has a special target called CDM distribution. So we'll build that and that's going to produce a new MFZ file, which we can only do on the and there it is. TGG. Yes, that's it. And we just copy that to the common directory. And now the CDM, the common data manager should recognize this and tell you what, I'm going to, there it is. So it saw that it changed. It's now going to load it and install it on itself. Step one, that's also a new thing that it can do. And after it's installed it on itself, it will start announcing it to its neighbors. And hopefully the neighbors will then pull it down, install it themselves, start running. And all right. So there it is. It installed it itself and it should go. It should remember that it was displaying the, okay, so now we're back to this. But now in particular, we should be announcing it to these guys and they should be saying, pretty quickly, they should be saying, yes, we would like to see this. So we're expecting this guy to be pulling from Northwest and this guy to be pulling from West. This, in fact, is going to work. And so for nothing yet, and this one is running long, I don't know. You know, so Alan had a comment in the YouTube comment a couple of weeks ago about I should do more live demos and I should show more failures. Do people agree? Do you think that I'm doing this sort of thing? I mean, the problem, of course, being that it's hard to control how long it takes. And, of course, if I cut it afterwards, if I tighten it up with little, you know, YouTube chop-a-da-chop-a editing, then how, in what sense is it live? So I still don't see it moving up there. There we go. Okay, great. This guy is sending it out. This guy is pulling it in and doesn't look like this guy has picked it up yet. Another thing that you might see is, can you see it? Now there's little red lines along all the edges that are not connected. And now let's come on. Let's see this guy say demo. Oh, all right. Could this actually be working? How it got to him? How it went right around him? I don't know. I mean, I suppose it could have gone, I don't know. Look at that. Okay. That's the code push that we wanted to see a couple of weeks ago. Not sure why this guy didn't get it. I mean, probably if we stop him, maybe he just had a race. So if we go back and now go back to the thing, presumably maybe he'll have the new guy. And there it is. Okay. Demo. All right. Episode 40, July 9th. Ring Lotus benchmark ready on a wall. Sorry this ran long. Have a good week. I'm glad it worked. I'll see you next time.