 So we get the chance to learn from someone else's mistakes today. Lots of different hours spent trying things out and I'm pleased to see what we can learn from Leaven today who's going to show us... Well, try to take us through 50 slides. This is a great challenge to set yourself. I hope you can get through many of them. It's a little bit like a Pecha Kucha, except 50 instead of six. So Leaven's standard is from Brussels where he works with the Frey Institute. He's been teaching robotics courses there. And so Leaven, thank you very much. Thank you. Good evening. This is where you can find me. My name is Leaven Stundar. I work at the Brussels University. Over here I'm at the Belgian Village if you want to meet me later or tell me all the stupid things I said were wrong. You're welcome for a beer and a discussion. These are some references from smarter people that taught me a lot. Functional design for 3D printing is an awesome book if you want to do structural efficient 3D prints. How to build your everything really, really fast is written by Charles Guarn. He works at MIT. He has water jets. His text is a lot more about aluminum, more high-end robots, stuff I can't afford. Gorilla Guide to CNC Machining is an awesome text on making really, really tiny gears really, really accurately for model robots. This is the place I work. It's FabLab Brussels. We have the general FabLab equipment. We have laser cutters. We have 3D printers. We make our own PCBs and we have milling machines. What I want you to know about this place is that we're an engineering department and that we built this place for our students. I'm not really sure how the situation is in Germany. I think it's a lot better, but in Belgium we have a very theoretically oriented engineering education. And we are a group of teachers, engineers who think that's not okay. We think it's not okay that you can get a degree in engineering without ever holding a screwdriver. And that's what this lab is about. Now, the context of this talk is that in order to do that, in order to have all my students actually build stuff, I have to build stuff really cheaply. The image on the left is a prosthetic foot. It's a beautifully engineered thing. It's made by a doctoral student at my university. It also costs a couple of thousands euros to manufacture. It's two weeks of CNC milling by a qualified technician. You can afford to do that for a doctor. You can't afford to do that for every student. The robot on the right is 300 euros of parts, some MDF dust laser cut and some plastic parts. So what we've been doing is trying to make cheap robots and try to get them working really well. The assignment for my students this year was to build a Jenga robot. We brought one to the Belgian village. The goal of the assignment was to build a Jenga playing robot and at the end have a competition. And if the robot does not work, you fail the course. Disclaimer, I'm an expert at nothing. I know a little bit of everything, which makes me a bit nervous before this crowd. I wrote a stock with my students in mind as an instructional discourse before they start building their robots. I'm also not affiliated with any brand of tools, but I can't avoid showing some tools. If I show brands it's because I like the tool. There. What I want to do today is show you 50 tricks, slides, techniques, stupid things, and hope these are instructional. So there we go. One. Now you can't just 3D print everything. And now this is how 3D printers are branded and sold. It's a stupid idea. You should design your parts for the machine. You're going to make them. That's true for CNC machining. That's true for laser cutting. That's also true for 3D printing. Point in case, this is a little casing for a motor with a coupling. If you turn it upside down, it's quite ugly. It's quite inefficient. It takes 15 minutes to redesign it and make it look like this. What happened here is we took out the material where it doesn't do anything, which is the holes in the side. And then we put 45 degree slopes on all the edges that stick out. But you end up with something that looks better, prints more than an hour faster and is 15 grams cheaper. Number three, chamfer all your corners. If the challenge is to build robots in plastic, as opposed to robots in aluminum, you need to deal with the fact that you're working with a weak material. Your sharp corners will be failure ports. That's where the plastic will break. Put 45 degree angles at all your inside corners and you will have a stronger part. And side note, the new carbon-filled PLA filament is amazing. It's stiffer than PLA. It looks gorgeous and then it completely destroys your print head because it erodes it, gets stainless steel print heads if you want to play with that. Four, ignore 3D printer reviews. They're mostly useless. I have never read a 3D printer review that talks about one thing I'm interested in. I am not interested in how many print heads are on a 3D printer. I am not interested in how big the print area is or how many materials or colors I can print with. I'm interested in how reliable it is and how easy it is to fix. They will all break down. If you want to buy your first 3D printer, buy one that a friend has. Buy one that a fab lab in your neighborhood has. Buy one that somebody who can help you with it has. That will make a lot of difference. This one is a terrible idea as a structural way to build a printer kit. Now I need to qualify that. This is my Huxley. It's very cute. I love it to death. I love Adrian who built it. I have my students built these because it's the original. It's still a pain to get this thing level, to get this thing square. And it's not the way you should design your machines. What you should do is use your CNC machine, in this case a laser cutter. Rely on the accuracy of your CNC machine and put that accuracy in the parts, not in the assembly. This 3D printer clicks together and will be square and level enough without needing any measurement tools. This thing is a lot easier to build than the previous version. I'm sorry. I lost a couple of slides. I'm sorry. One second please. There we go. So it was here. Six. Use structures for stiffness, not thicker sheets. You can make really strong parts by using your material intelligently. This is Z-axis for a little CNC machine I once built. It's entirely constructed from 3mm of MDF and it's rigid enough. It's rigid because you use the material in different orientations and get stiffness that way. Understand speed. Everything you will make on a laser cutter will be finished 20 times faster than anything you have to 3D print. 3D printers are great for small, complicated parts. You're a complete idiot if you try to 3D print your entire robot. This is a 3D printed, entirely 3D printed robot that I made. Now I have to explain this one. As I mentioned before, we wanted to have our students build Jenga playing robots. But we had a discussion among colleagues if this assignment was too difficult. We had a discussion on Thursday. So I called my colleague on Monday and said if I can build one over the weekend and it works, does that actually mean that it's easy enough for our students to build one in half a year? And he started discussing it while I thought it was totally obvious that if I call him on Monday that I actually built a thing over the weekend. So, and point being, I only had a 3D printer. I have a 3D printer in my kitchen. I don't have a laser cutter at home. So how do you pull this off? Well, you want to design modular and you want to use as little plastic as possible. You also want to start printing before the whole thing is designed because there's 40 hours of printing time in this robot. So when this part starts printing, the rest of the robot hasn't been designed yet. You have an hour then to draw the second part. Again, the next part hasn't been finished yet, but you can put in mounting holes and build from there. Your mounting holes are free. What you can do is just throw in a grid of holes and then figure out where the next part goes while this part is printing. This is not one part. This is four separate parts that fit together. You can do smaller, simpler parts faster. At a time when part one is printing, I still have to figure out where to mount motor holder for, but I have a grid of holes so you can figure it out. And if you use 10 by 10 grid holes, then you can actually fit your mechanical connection. Obviously, you can do the same thing on a laser cutter. And you can do the same thing on a really big laser cutter that cuts steel. This is the furniture for another lap I built. We did everything. We basically made our own giant McKenna. Talking about holes, let's talk about drills. Center drill bits are awesome. These drill bits are intended to use in a lathe to make a tiny little center hole and then switch out with a regular drill. You can use them in a drill to drill into brittle materials. They're great for drilling Plexi. They're great for drilling PCBs. If you want to do holes, you want Forstner bits. There are hole saws. You can buy them. They're crap. These are a lot better. Seventy. Step drill bits make huge hole with little power. This is a drill I use more than any other drill bit. You need to be careful with it. It's not cheap and you cannot sharpen them, but they allow you to make holes in steel with very little battery-powered drills. You need to take care of them so you need to keep them cold. If you have a drill press or a drill without any coolant fluid on it, you can get by with WD-40 in a pinch. It actually works really nice. If you use WD-40, you should realize it is a solvent, not lubrication. If you spray WD-40 on your chain, you just dissolved all the grease and removed all the grease. The next day, your chain will be dry. You do not need very large drills. The small 3-cell Lippo drills are actually more powerful than the high-end drills 10 years ago. They built 18-fold drill drills for people who build houses and stuff. It's a macho game. They will break your wrist. 21. Buy cheap drills. Put a better battery on it. I used this for years. This is a great way to get cheap tools. 22. If you want to mount small nuts, these are a great tool to have. They're also remarkably cheap. You can get these tools at your local pharmacist. I know you cannot get them in England, but in Belgium you can just ask for needles, syringes, scalpels and these forceps. They will just sell them to you. 24. Keep a magnet in your toolbox. Another way to mount small bolts and nuts is to magnetize your screwdriver. These are two types of machine screws. You can get them both in. One of them will not work with an electric screwdriver. The flathead will not center your bit and you will get annoyed and never use them again. 26. Don't buy cheap cutters. I'm a big fan of cheap tools. Buying cheap cutters is pointless. If you try to cut something that's harder than your cutter, you will destroy it. Get a decent one. It will last you a lifetime. These will not last you a lifetime if you have good ones. They're for copper, not for steel. If you try to cut copper with these, you will destroy them. 28. Make your housing 25 larger than you think it needs to be. It will still be too small and nobody draws wiring harnesses in CAD programs. 29. Lay out your electronics flat. Stack shields. I use Arduino's myself. I like them. But the whole shield idea if you start to make your own boards is a terrible idea because if you stack two shields, how are you going to get to the bottom one? Lay your electronics out flat so you can debug them. Don't put open battery holders at the bottom of your robot. Not even if you don't have any other place to put them. It's a stupid idea. 31. Duck tape is not the same as gaffer tape. You know both and you need to know the difference. Duck tape works on wet surfaces. Duck tape is super, but it will stain whatever you put the duct tape on. Duck tape is what stage technicians use to tape down cables. It does not mark the floor, but it doesn't work on wet surfaces. If you don't know which is which, you will get into trouble. Get a large cutting disc for your drum. I use this thing all the time. Don't get the little clicks discs. They're awful. These are great. You go through a lot, but they work a lot nicer than a large grinder. Obviously use goggles. Figure out that you need to clean your goggles. They're like regular glasses. Only they get dirty even faster. Put a little bottle of lens cleaning fluid in your toolbox. 34. Buy your work gloves one size too small. I have a lot of gloves. I still haven't found a perfect one. The perfect glove is one you can use with a desk grinder and also screw in a little knot. This helps. This is the best I got so far. Obviously don't buy flammable gloves. 35. 3D printing gears is a miserable idea. You should not do it. This is an assignment for a stair-climbing robot. Those gears break. Why do they break? Because you're printing in plastic, not in a lume. And if you make gears, one, they have to be positioned really carefully or they will not grip. Unless you build gears with really large teeth. That helps a bit. But the main point is that all your force is transmitted to a couple of teeth making contact. Timing belts are a lot more forgiving. This is a timing belt. They're a lot more forgiving because they use the entire contour of your little wheel to transmit a power. The idea is you do not put point loads on your plastic because it will break. You want to spread out the forces. They also need to be aligned a lot less carefully. There's a lot more room for error. There's a lot more room for misalignment. 36. Use two bearings on one axle. Use two, not three, not one. The reason is your bearing is not designed to handle rotation. If your axle rotates inside your bearing everything will lock up. They're meant to take actual loads as shown in the first diagram. Also don't put three on them. That will not work. You can't align them properly. So the whole thing will lock up or one of the bearings won't do anything. 37. You will be convinced your design is an exception to this rule. You will be wrong. 38. Preload your bearings. This is an important one and one which I think is under-emphasized. If you use the bearings as in the top graph your axle can move around. A bearing is constructed from an outer ring, an inner ring and then there are balls in between those that facilitate the whole rotation. There's slack on those bolts. They can jiggle a bit. You need to take out that slack by pushing the two bearings towards each other. That's the function of the little black part in the robot. It allows me to tighten the nuts on both ends and squeeze those bearings together without everything going wobbly. Talking about bearings. 608 bearings cost less than half than any other size bearings. That's just a random remark. The interesting part is why this is so. They cost half than any other bearing on the planet because they're more of them than any other bearing. 608 bearings is what's in every roller blade on the planet. That is an interesting idea to know. That's an interesting concept. 40. This is a very, very low cost motor. It's amazingly powerful. The idea is the same. This motor is a car park. This is actually a German shop. We bought stuff there for years. If anybody knows what this one is for, I would love to know it. The previous version were windscreen wiper motors. This one has actually a screw and a nut on it. So I'm lost right now and I haven't figured out yet what's it for. We use it anyway. I have to get back to this little robot because I lied about the timing belts. The timing belts are very strong, but now your system will fail at the indicated point. That's where you have the smallest lever so the largest force. Your motor will strip your little plastic wheel off its axle. This is a terrible way to fix them in place. It works in aluminum. You should not copy it. Even if you put a hidden nut in there, it still won't work. You're going to split your plastic. This is better. It's the same idea as before. You are using a weak material, so spread forces over a larger area. You're using all the plastic that's there to clamp on to that level. A motor axle. 42. Understand accuracy. 3D printers are not a magic solution. In engineering standards, they're relatively imprecise. You can get a tenth of a millimeter accuracy. A laser cutter will give you about the same. A CNC mill will give you two hundredths of a millimeter. A well-aligned lathe will give you a hundredths of a millimeter. Sometimes you need that. This is one of those examples. If you want to do really, really small gear wheels, you want to go to a CNC mill. You do not want a 3D printer. You will want to save for a 3D printer and buy a 3D printer you need a lathe. There are two kinds of students in my class. It's not something I become optimistic about, but the basic thing is some people have been building stuff before they went to university. The two kinds of people I have are the students that already have a lathe and students who don't have a lathe. It's not a sexy machine. It's not a hype machine. It's the most accurate tool you can get for the least amount of money. You don't need a big one. You don't need a heavy one. A small one will do just fine. I've been printing with Nylon as a game changer. I've been printing with the Talman Nylons last year. Everything I said before might be a lie about not printing gears. This thing is awesome. It's halfway between plastics and aluminum. You can do amazing stuff with this. I had very good results with it. Which brings me to the next point. Learn your plastics. Most people who have been printing or visited FabLab know these plastics. You have your PLA, your ABS, which are your standard 3D printing plastics. You have your PMMA. That's actually the real name for Plexi, which is a brand name, which most people use for laser cutting. These are also interesting and it's worthwhile to learn to know them. Polyethylene, polypropylene, polystyrene, polyvinyl, chloride, nylon. The problem is I have never found a really good book on plastics, because most of them tell you about a hundred different kinds of plastics. You need to know about ten. But it's really interesting to know those ten, if you will be working with 3D printers and laser cutters. The other problem I have with books, plastic, is that if I want to identify plastic, I do that by smell and I do that by touch. And that's really hard to capture in a book. Basically what I'm saying is put plastic on fire. If you do this with polypropylene, you will notice it smells like candle wax. It's a carbohydrate. It's about half the price of PMMA. It's less brittle, so you can do this kind of stuff. This is a connector for a dome. The image on the bottom is the back side. You can see it's cut less cleanly. So you have a clean side, a side you have to clean up. But it's bendy, so you can do stuff you can't do without Plexi. It costs half the size and it's a really... Well, it's a beautiful plastic. It's one of the least polluting plastics we have. One of the reasons I want you to know about plastics is because there's a lot of difference. PVC is disgusting. This thing is actually quite nice. Polystyrene is good for housings. You can get it with different service finishes. This is a little case for a remote controller. It's also cheaper than the Plexi. You can glue it very cleanly. You can get it in different colors. It's not that stiff, but it's really good, as I say, to do stuff like housings. PVC is stuff you know from those electrical tubes or from little swimming pools or from toys. This is the plastic that gave plastic a bad name in the 80s. You should avoid using it. You should not laser cut it. It will destroy your laser. Unfortunately, it's one of the tougher plastics. It's one of the cheaper plastics. Sometimes there's not a similar product that can do what you want. Still, it's a bad idea. If you want to weld plastics, this is not the reason you need to know which plastic is which. The best way to glue plastic is to weld it. What welding means is that you dissolve the two pieces of plastic that are in contact. The solvent dissolves and in the end you have one piece of plastic. You can do that with all plastics. You can do it with polypropylene and polyethylene. For other plastics, you need to use the right solvent. Figure it out. If you do it right, you have a weld that is as strong as the material. It's used a lot in Plexi, but you can also use it with polystyrene. Everybody uses it for ABS. 49. Super glue does not cure wet air. It glues with moisture. That's the reason why if you put it between two pieces of plastics, it will take forever to cure. That's also the reason if you put it on your fingers, it will cure instantly. And you won't get your fingers apart. 50. If you have a laser cutter, make custom boxes for your project. And there's one thing I want you to do is to treasure the stuff you have. I cannot say how many stuff of me got destroyed before I got smart enough to start making custom boxes for my robots. That's it. Thank you. I hope you enjoyed it. Thank you, Leven. Thank you for that.