 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 25th T Tuesday update. Let's get into it. There's a lot to talk about, so this one may go a little long. I'm going to let it go a little bit long, like maybe 25 minutes. We'll see. So last week we've got updates on the 3D printer, on the bill of materials, and on the PCB assembling. Coming up I want to deal with something, well, I'll explain it when I get to it. And also I did a bunch of stuff with new software for the movable feast machine for the T2TILE that I want to show you some a little bit of it as well, and that's why it'll take a little extra time. So let's go. ROHS stands for Reduction of Hazardous Substances. It's pronounced ROHAS as far as I can tell. It's a directive that comes out of the EU that says basically when we're building products, especially tech products, maybe stuff in general, I'm not sure, it would be better if we didn't have a lot of poison in them because we have to live with them and then put them in landfills or put them in rivers or wherever they end up. So it's a good thing. ROHAS is, and one of the big problems with ROHAS is lead in solder. And we talked about this earlier that, you know, with all of the prototypes that I've been hand soldering myself, these guys that we've been looking at, so forth, these are all using leaded solder because leaded solder is easy to handle. But that's why we go to the pros to get lead free. And I had had some discussions back and forth with ETS that I guess led to a misunderstanding or I misremembered or something. There was a little bit of difficulty because I was always kind of trying to get to an actual contract with language and ETS really didn't seem to want that. So they don't work with the contracts or the handshake and so on. So I said, well, okay, if that's the way it works. But now when I was actually getting serious about actually trying to get the stuff scheduled because the bill of materials is going to be there. And I was getting in touch with ETS and saying, you know, when can we schedule to do this? We're looking for 150 to 200 tiles with ROHAS processing and parts supplied, boards supplied by me and so forth. But the response was, you know, do the assemblies have to be lead free? They'd like to wave solder those and there was a typo saying the wave is lead free. They meant the wave is leaded or leaded in. And that they quoted based on assuming they could put lead, they could be using lead based solder. And so the implication is it will take longer and the price will go up yet again. And, you know, this is exactly what I was sort of worried about when I was trying in my nerd way to get everybody to agree on what the language was going to be and get it written writing and so forth. And now at the last minute this is coming up, so I didn't really feel that good about that. But really, the fault, you know, to the degree that there's any fault, it's mine for not either pushing the issue or at least getting some comparative quotes other than just the online guys that were so ridiculously expensive. Try to find some kind of apples to apples comparison to see whether, you know, this is reasonable or whether the price should go up or not. So in fact, this past week I did reach out to other PC board assemblers. There's more than one in Albuquerque, Albuquerque's big city by Albuquerque standards. And now I am talking with other folks about getting a quote and, you know, be very explicit Rojas right from the get go. And we'll see what they say. I mean, if their quote is wildly higher, if it's in line with the revised quote that we would get from ETS, well then, you know, so be it. But I just sort of feel like I need to do due diligence before just saying, okay. So that was a little bit of a bummer, but it make it hopefully it'll come out fine. However it comes out and we'll find out more information in the next week. So that's the scheduling issue. Are we going to have the boards being manufactured in process before the 26th, before the midseason half a year, according to a real season? Probably not. But we're making serious blows down in the last thing, steps of the thing. So that's the manufacturing story. On the 3D printing story. Last week, if you recall, I had been having a chat with the Prusa service people, well, just the Prusa customer support people and they couldn't tell me how to do warranties, get a warranty service or how much it would cost or how to get an RMA. And where we ended up was the customer service person had said, I left a message for the whoever it is when they come in in the morning. And in fact, that did arrive. So the guys at the department forwarded this to me when it came in the next day. And without going through it all too much, basically it's a quote for an hour of work for 34 bucks, which seems like a pretty good price for an hour of work based on the United States standards. But also two one-way DHL shipments, $110 each and $100 was what it cost to get the thing sent to me to begin with. But here there's absolutely no implication that if it is a warranty claim that Prusa would pay at least one direction of shipping, nothing like that. I can play a roundtrip shipping and I can pay an hour on top of it. And so that's $254 for them to take a look at it, see if it's a warranty claim and maybe fix it. There was an email that went with it that said it's just don't know how many hours it would really take. So just the price of one. So $254, that's an absolute minimum. Don't feel great about that. I got back with the folks in the chat room. Andrew Walpole was saying, well maybe in fact what you need to try and do is find somebody local in Albuquerque that could repair these things or understand it and so forth, these 3D proven systems things. Guys that do 3D stuff. In fact I reached out to 3D proven systems using their web based contact thing. My name is Dave Ackley. I'd like to introduce you. Now they had a list of manufacturers and Prusa was not on their list of manufacturers. So that didn't seem like it was too encouraging. In fact I got mail back from Lance Detmerce, a very nice guy, some technical part of the company. And we had a little chat in email over last week that sort of boiled down to yeah they don't really do Prusa stuff that could probably look at it because printers are all fairly similar but then they're going to $125 an hour for repairs in the US. So that didn't seem like something I wanted to pursue immediately although I really felt good about Lance Detmerce and 3D proven systems and he's spinning off a Route 9 company that's actually going to be focusing on contract 3D printing which in fact could use at some point. So that was not a solution although it ended up feeling good. So the solution, the next step was well let's just drag the printer in to put it on the bench and see what we can see about it. This is what it looks like when you take the steel spring steel bed off it that's held down by magnets. I was really suspecting this area in the back where the cable connects to the print bed. I mean this print bed slides back and forth to change the Z component of where it's printing and so this cable which is wrapped in textile is flexing all the time and so I took a look at it. I also was flipping it over and taking a look up underneath at the bottom. So this is looking up at the build plate from the bottom. This is the same old capped on tape that we saw when I was doing the reflow soldering myself. So this in fact right up underneath here is the thermistor, the heat sensor connected to a cable that's going out through that cable in the back that we're getting urban temp bed about that for some reason either this thing isn't signaling or the guy at the other end isn't hearing the signal. This looked fine to me and it's not like I was using any kind of weird plastic that's extremely high temperature that might be degrading the adhesive on the capped on tape PLA not that hot. So this really looked all not terrible which once again reinforced my notion that it was probably something to do with that cable connection. I looked at it, I found the screw holes at it, I went and I got my socket wrench and I unscrewed it, I took the bottom thing off and just took a look at it. So what you end up seeing you see these two lugs those are connecting the actual power lines that are going to the heater that takes a lot of current to just heat up the bed and then this thing that looks like basically a regular zip cord is presumably the two wires for the temperature sensor because it's the only two other wires that were there. Now when I opened this up it did seem to me like the cable was kind of tight and I couldn't really find out what was going on up this cable up the textile thing because it was held down or hot glued or something I wasn't entirely sure but when I decided to just close this thing back up I tried to ease the cable a little bit, you know, cram a little bit more of the textile and underneath the 3D printed clamps that I was screwing up so that maybe it wasn't pulling so much on the thing. And I screwed it back in with my socket wrench and set it back up. So now this was a reel of the prusiment that was almost out anyway so I thought well what the heck let's just give it a try see what happens and in fact it started out printing fine which proves nothing. When this thing gets erm in tent bed it could be two hours into the print it could be six hours into the print. So the fact that it got through the first ten minutes was nice but it didn't prove anything. But we were running out of the filament so I was going to have to change it so I got this roll of old silver that I had lying around from before because I really didn't want to blow more of the stuff that was matched to what was going to be in the cases if I didn't believe this was actually really going to work. So I paused the printer when it would get to the end because once again I still wasn't going to trust the automatic film changing system. I put in the new stuff, I did the load filament thing, I got it pouring out, half you know Mr. Twizzle, black color ice cream, half black, half silver and I thought everything was fine. I thought I was going to be able to go back and continue with these guys that now had a front black plate and would have gray behind it. It would be cool. No, for some reason it decided even though we had already loaded the filament and said it was fine it decided to poop a bunch more silver on top of the thing when it went back to fix it. I tried to pull that stuff off to see if I could keep going but no way and in fact I had already knocked one of Pinocchio's shoes out of position anyway so this print was ruined but not because of mint temp error. So I said well screw it let's just get the real stuff and put it in and see what happens. It's only money if I have to go buy more of this stuff and I started up a regular print and lo and behold eight hours later we had three more cases. So that was good. I did a few more goes. I got another few more sets of three cases. I felt like you know hey I have the healing hands. You know put it on the thing it's fixed. Yeah no after a while though it started to die and this one died early. The one that broke my heart was this one which was you know easily six of six hours into the print over halfway certainly through the thing before it died and so forth but nonetheless in this whole process because I didn't really have anything else to do I sort of have developed this whole superstitious procedure that I do every time I'm about to start a print I lift up the base. I push on where the capped on tape where the thermistor is in case it could have gotten early back down. I untrap the textile cable from the heat bed which gets caught but between that the control unit on the side there and the extruder. I loosen that up and I spread it out so it's more evenly back instead of being up bent so vertically at the joint and in fact that in that process I've learned that you know a couple of times when I did that I created an ER Mintemp bed which just served to convince me that yes that's where the problem is. It's in the cable connection to the bed but since I developed this superstitious process eventually I got really lifting up the thing and touching that thermistor because it doesn't seem to matter but easing the cable at the beginning of each thing I have not had a failure again an ER Mintemp bed failure since then and in fact we are now getting to the bottom of that reel that I started in this was a picture just from an hour or so you might occasionally hear the printer going in the background it's finishing off that roll as we speak. The net result is we got a whole another box of cases out of this so now we have three complete cases that's 36 in a box that's 108 cases we have another 24 of the older style that takes the socket cap countersink that puts us up at 108, 128, 130, something if we can get another box through or even better two boxes out of this thing we will have enough cases almost surely so that's the 3D printer story I hate not having a proper diagnosis for these things but there it is moving on the bill of materials yes so where we left off there was we had gone through this whole thing to get the P8 and P9 headers Beagle Bone Green we'll connect to the circuit board with had decided to go to AliExpress $43 for $200 with free shipping which I hate because it's not guaranteed any kind of fast could have been 23 days but in fact last week it arrived in LA and was expected to be delivered this past Thursday it arrived in Albuquerque it was not on Thursday but magically got rescheduled for Friday and it showed up Friday here it was once again $85 to me but if you look on the cover what this says is $5 value I hate this stuff and there it is there was not a lot of packing in along with the bags of connectors two bags of 200 connectors and then box and I think some of the pins actually that were crammed a little bit in the edge I'm happy about that but I think they're all usable so we'll see and there they are and again 200 pieces 200 pieces, 400 total takes two per tile that means we have capabilities for 200 tiles right there and with that the bill of materials for what needs to be soldered on to the board is complete we can go whenever we get an assembler that will work with us so that's quite milestone even though there's really nothing to show for as far as what works soon I hope in addition to the tiles we also need the intertile connectors we need a whole lot of them previously I had modified it to put a little the series resistors in the middle all these problems with oscilloscope noise and analog noise that turned out to be mostly just my brain damage I went back and I cleaned up the design of the intertile connector this is the 3D rendering of it this is the circuit layout circuit of it I like this layout much better I cleaned it up in several ways really I'm getting a little bit more confidence in this now that it's almost the end of the road for needing it at least in the T2 and I ordered 9 of them from Osh Park the idea is we'll get these quickly we'll check and make sure they're okay and then we're going to try to do an order for hundreds of these things from somebody probably in China we shall see to simultaneously make the circuit boards acquire the parts that go on it which is just two female connectors solder them all up and send it back to us boom in one go that goes and the boards in fact have been sent off to the FAB and we'll see them probably before I talk to you again that's the logistics division there's yet more May 9th and 10th I'm going to be in Vienna giving this invited talk at this thing called the the pioneer pioneers 2019 the pioneers festival something like that it's mostly about startups European based startups and entrepreneurs and so forth there's lots of speakers of which I am now one of them that I was invited and I'm hoping that there'll be a good excited crowd and we can spread the word about computation 2.0 about living computation about there is another way coming down the road literally for most startups but hopefully they'll be excited to hear about it my friend connected me with these guys and he's going to be there as well and he's been there previously and says you know they're good crowds they're excited they're big crowds and so forth so we shall see Vienna May 9th and 10th alright yeah but there's also new software engineering to talk about and I'm calling them data worms you'll see why in a minute I don't really not totally thrilled with the name I would turn to our community out here for suggestions on names I mean worms isn't terrible but well you'll see alright so to introduce this story there is a video on my original channel on the Dave Ackley YouTube channel that's now going to be 9 years old I'm sorry 8 years old going to be 8 years old this year from 2011 asymmetric diffusion of bonded structures in the movable fuse machine what the heck is that well so you'll watch it and what it is is these bunch of atoms that are connected by little lines and they wiggle and they move and they go around and when they get over here the yellow thing the yellow thing tells them to go down and the white thing tells them to go west and the green thing tells them to go north and in general it's kind of like a worm that that wiggles around and it can be directed to do things now one of the things that comes up this is a action shot from the middle of that video where so you can see these these the balls are these atoms the lines are these bonds between the atoms and this particular molecule has now gotten stuck it's gotten hooked on to the edge of a pointy little bit of wall or something like that now eventually this thing popped free and it kept on going so it was alright but this is one of the fundamental problems of the whole idea of bonds and the early versions of the movable fuse machine had this idea of bonds that you could have atoms that were aware of each other and wouldn't get too far away built into the low level engine but when we went to more recent versions of the movable fuse machine although more recent is like five years ago now we ditched the idea of bonds because it seemed they were too specific and we didn't really know what we wanted out of them so you don't see any of these lines in the movable feast machine simulator today and that's because bonds are not built into the underlying architecture you know you could the bonds are implemented with software they're basically a doubly linked list each atom in here remembers the site where it saw the guy who's downstream of it and the site where he saw the upstream of it and everybody agrees that when anybody moves you update the site numbers of the guys that are pointing at you and the guys that you're pointing at to keep the consistency of the list going and that's how bonds work in this past week well actually it goes back a couple of weeks into the wall of science period one of the things that I was working on was how can we get big things to move asymmetric diffusion of bonded structures has been a concern for the movable feast for years and so we had these big cell membranes that they kind of moved and so forth and that we had the 2D printer that had lines moving through it swap lines that would move through it that would kind of move a whole big 2D structure step by step one of the things I realized in the last week was that if you took the idea of a swap line where you waited until everybody caught up with you and then you moved one step and you boiled that down to a single dimension swap line so now it's just one atom swapping along a line it becomes very simple and so let's look at it hello alright here we are so now here let me get a look at this guy alright well what happened here so this thing is an s-worm he's got a bunch of types green means he's the head red means he's the tail this light green means he's the body and these blue things means that these are temporary structures that are being passed through from head to tail so if we do an individual event here like this guy can I do him oh I guess we're too close to the edge we'll just let it go automatic it'll be quick but you'll get the idea there so you see what happens every so often the head moves to a new site leaves a blue behind it and the blue gets swapped all the way down from the top to the bottom and when the blue swaps with the tail with the red guy it just gets consumed so the net effect is is that we get these little worms that are moving around at this point it's just moving randomly I mean did everybody still have that snake game or was that just from computers when I was a kid where you drive this worm around and you eat the little stars and every time you ate a star so you know this guy wound himself into a circle and the old game of snake right that means you lost that guy and that's what happens here and that's just inevitable in the sense that the whole idea of this thing is rather than using a bond that fills empty space you use these temp atoms that keep track of where the guy is by keeping track of where he had been and so if we have a bunch of these guys they will all go running around and the S worms they kind of stand for simple worms or stupid worms or something like that given that they're just wandering around randomly they wrap themselves in a knot where they can't get out rather quickly but we also have the B worm and alright let's just take this guy specifically we're going over 25 minutes but I'm almost done I'll show you what I wanted to show you oh that's a simple worm so well so alright so he's not gonna go anywhere let's get a B worm alright there's a B worm alright now if we put a wall around that guy alright now there's no place for the head to go because he only looks north south east west the simple worm just looks you see he doesn't even look diagonally because his structure isn't smart enough to do that because his next and previous pointers as a doubly linked list are two bits long it's just one two three or four which adjacent guy is upstream of you and which is downstream of you that's it but look what happens to the B worm now that when the head figures out it's trapped it builds a bridge yellow is the downstream edge of the bridge heading toward the tail the dark blue or black is the upstream abutment of the bridge heading from the tail towards the head and he escapes and so now he can get on out and go about his business and so on and you know the bridging worms the B worms they can get trapped too but they can only get trapped in a smaller number of circumstances if they're completely surround they can only bridge a single site now with with more programming they might actually be able to bridge two sites but it's a lot a lot more code and the code here is not quite as simple as I would like it to be but it's pretty simple and think about this once you know each of these things has seven bytes available one of these worms has 112 bytes of unused space that they could be carrying a hello world a string a number these things could meet up with each other and do arithmetic as they pass along each other you could compute with the data worms and once you have the bridging worms you can do things like these don't have to be 16 segments long this is just the particular demo that I organized you could make circuits you could have that circle circuit lines that hop over each other and go on about their business you could use this as a basis for much more understandable evident computation going on intracellular intracellular sending queries out and so on I had previously had a routing demo where we just sent around single atoms well what good is that we could redo that routing demo I would like to I haven't got the time that you would use data worms as passing through there so again we could be trapped we could be transmitting non-trivial payloads alright so this I'm very excited about it's you know really incredibly simple and that's kind of the point it's meant to be a basis and it's really a spatial doubly linked list and that's it simple I mean if you blow hole in the middle of one of these guys you know they're not designed to be amazingly robust they know more or less how to clean themselves up although it can take a while that guy's still been rolling around out there and so on and so I built a little track for these guys to run on imitating the 2011 video where you know green means go north and yellow means go south and so on and so forth you know it takes a while to go but I have made a time travel video that shows it happening much quicker where is it here it is like that this is going to start going faster and faster and faster I don't know if we can hear anything on the sound the pitch of the sound is reflecting the air the number of events per second which is kind of a reasonable thing to do okay at a time lots of developments happening hardware, software, all of this stuff the next update will be out in a week that will be the 26th T Tuesday update sorry this went long was it too long I think it had some good stuff in it I'm kind of getting excited about the whole thing coming together a little bit more I hope to see you next week