The Ship in a Bottle: Printing a Hermetically Sealed Sea Scooter in One Go - EMF 2026

hi everybody. Um [applause] yeah, so um today absolutely I'm I'm if you're here for this talk which is shipping in a bottle printing sea hemetically um uh sealed sea street in one go um then you're in the right place. So, um, so let's just briefly go over what I'm going to kind of cover hopefully in this, uh, in this session. Um, first of all, I want us to all agree on what a sea scooter is. It's not something I was familiar with when I thought about this project, which is funny. Um, then talk about some of the interesting design challenges I've uh, added into the mix um, to make this an interesting build. um and then kind of walk through some of the main subsystems and how we kind of solve those uh design uh constraints uh for those subsystems which in a non-typical way that you do for something like a sea scooter. And finally, hopefully bring it all together to show you uh building a ship in a bottle and printing a sea scooter in a single piece using a 3D printer. And lastly, it's speedo time. Uh you can see hopefully seeing whether sing or swim and whether the thing actually works. So if we get through all that we're doing well. Okay. So first of all um what is a sea scooter? So if this is what you think of a sea scooter if first thought comes to mind then uh it's not this cuz this is not real but it is often when I uh talk to people they will think of something like this. Um a sea scooter is more like this or this. So they come in different shapes and sizes. Uh they're basically handheld devices which you can uh propel you under the water for like scuba diving or snorkeling. Um or uh just yeah recreational say different sizes and shapes. Um and it it it really sort of drives you along a few kilometers an hour some some a bit faster. Um so that was the challenge and I'm probably going to be trying to build something which looks a bit like this which this has obviously been done. There's hundreds of different types of sea scooters that exist in the world. Um so part of the the the drive for this project was um I really passionate about um making things and pushing different techniques like additive manufacturing to the limits. And I was thinking well how can I kind of showcase and and and test what you can do with like a consumer grade 3D printer. Um, and I was thinking, well, what kind of interesting constraints we put in place that only really 3D printing could or might be able to solve as a challenge. So, I'm going to try and build something this something that looks a bit like this with some constraints. So, there's only two real constraints I'm I'm adding into this. So, firstly, the hole must be um printed in a single piece. So, if you're f familiar with concepts like printing in place or um basically as a single print, which looks a bit like um like like this. So on a bill plate printing up, that's a hole. And secondly, the hole must be hemetically sealed to get some value out of doing that single print. It's going to have no holes or seals in in the hole at all. So all the points you've got, you've got propeller, you've got a master onoff switch, you've got throttle and data connections and power, we're going to cut all those links. So where you'd normally have cables or drive shafts coming out, we have to work out, can we do this in a way that uh doesn't require any any holes in the hole. And that is is probably the the trickiest part of the um uh the the design um challenge. So first subsystem, this thing's going to go in the water. Now 3D printing, it's not not not great. Typically, if you try and put something in a 3D print, it normally will leak to some degree, but I need to have some real reassurance. I was going to be putting a lot of electronics into this um that even if I went under underwater and not just below the water but you know meters deep that this thing wasn't going to leak on me. So I did a whole lo of testing uh built um all these at the top pictures is a lot of little mini um mini holes I um scaled down versions and I built myself uh with all different materials. Each color is a different material and different way of printing. And at the bottom we've got you can just see lined up testing them next to each other. I gave myself the target of at least 40 or well uh 40 m water depth which means it needs to be able to withstand uh four bar of water pressure which is is quite a lot of water pressure um equivalent I mean thing on the top. So, I built this test setup which you can sort of see the steel rods and it's a polycarbonate uh cylinder which I pressurized to to up to four atmospheres going up incrementally with all the different test holes testing how much water was taken up um during uh during a soap test. uh loads of really interesting results, but to to cut it short, um the the best and the the process that I ended up going for which was reliable was um ASA using a 3mm shell which was epoxy coated and using a a small nozzle like a regular nozzle that seemed to give me the most um reliable ability to go down to four bar and get no water ingress at all. So I I was happy I had a system which was uh watertight at the depths I needed to be watertight to. So we could then start thinking about what was going to go inside of this hole. So the next thing most things first thing you think about normally is the drivetrain. So I've got a propeller I need to drive but I can't make a hole for a shaft. So we have to look at what kind of other systems exist and I can't make any holes really for electrical wires. So um how do I drive this system? So the solution is using uh magnets uh a magnetic gear and that magnetic a magnetic gear um I won't go into too much detail. I'm actually doing a workshop on it which is almost directly after this I think in uh workshop 4 if you're interested building building these and a bit more information about them. But simplistically we've got um the great things about magnetic gear is they're non-cont. I'll explain what that means in reality in a second, but uh there's basically no need for holes and you get a reduction gearing for free. So you can take a motor and it gears it down as part of this non-cont drive. So I kind of simplify my overall um design. Um so the main parts of any of most magnetic gears is you've got inner rotor which is a magnet with it's magnetized or the north and south is uh red and blue. Uh we've got a middle layer which is the field concentrator layer which is all about focusing and channeling that magnetic um the magnetic uh flux through to the outer layer which is what interacts and drives the uh the outer rotor around. So this kind of a little animation to try and uh help visualize this as it's really counterintuitive. Um I think I'm going Yeah. Yeah. So as you can see kind of interaction between the um so red red and blue attracting each other wanting to move a little bit. So they're not quite in in the most uh optimum. So you can see there's a bit going to be a force exerted particularly on these three points. Uh and what happens as we rotate it that moves around. So it actually goes in the opposite direction which is what actually ends up driving the outer rotor in the opposite direction to the inner rotor. a really cool feature of a magnetic gear which is not just a direct magnetic coupling that you think of but actually ends up driving your outer the outer rotor in the opposite direction. So it's kind of like messes with your mind a bit. Um so for this system we we end up with a 4:1 uh gearbox ratio and actually this is the the ratio I ended up going with um for the C scooter build. Um and this is the cross-section. So this this is hopefully going to explain why this thing works. So, uh, on the outer we kind of, you can hopefully see there's magnets at the top and the bottom. That's the outer rotor. Then we've got the the inner rotor in here, which is another magnet. This bit's the inner chamber of the, uh, C scooter. And here we've got the the bit that's doing all the all the interesting stuff, which is the um, uh, field concentrator layer. So, embedded inside of this, which is the hole. So this is basically the hull of the sea scooter are the uh iron pins which is focusing the the uh uh magnetic flux lines through to the outer rotor. And because they're completely separate, we can seal it off with a cap at the end which means that the inside can be dry and the outside can be wet. So we can drive drive this system without any any worry for getting it wet. And I can design this in a way that it should be 3D printable in place which was also tricky. So this that's what this this design should achieve. So, next thing, right? Okay. So, we've got our drivetrain. Well, I'm kind of a bit fixed with that now because I've got the geometry and the the gearing ratio. Uh, if you want to get into sort of propeller design, it's like a whole whole thing in itself. So, I just went and looked at some competitions and and found the um the top five or six propellers and and scaled them up or down. Um, this is these are some of them. I also some interesting shaped ones. So, these all apparently did well in a competition. So, we've got sort of standard propellers of uh much sort of with with a root pitch with a ve sort of a a slender very slender propeller and lastly a tooidal propeller which I just really wanted to try because um I've I've never seen one work and I've heard interesting things about them. Uh one of the things is that they're supposed to be quite quiet. So, just a little uh little video clip. So I I created this test rig um which has got some uh nicely bodgeged together uh luggage scales which is measuring the uh the force that the uh the um the propeller is is putting out inside my bathtub. Uh and the current's going up. So I'm just gradually increasing the the the current on the power supply to see when it starts uh the torque slips and the gear slips. And that's the maximum uh power output I can get from each propeller design. Um, what you can't hear cuz I think I turned off the sound for this one was the this this was the the blue the blue propeller which is the um uh tooidal propeller. And it was just it was amazing. So so much quieter than all the other propellers. And I and it was printed badly. I mean it was the surfaces were not smooth. Um it was probably not exactly the right size scaling. I did a few different scaling sizes but it was just so much quieter and developed so much more uh power at like lower um RPM. which would work quite well for a magnetic gear because if you make it spin faster then the the sort of the peaks in uh in the in the torque spikes you get uh through resistance in the water make it more like to slip. So really good. It wasn't the best but propeller was was a really interesting uh uh thing to test. Great. So I have so I found the best propeller to use that couples with my my drivetrain. Next subsystem, right? Power system. So, okay, standard power system. We've got batteries and we've got need to get power in and we need to store the power so we can power our sea scooter without it being plugged in. So, first challenge is that this is quite a constrained system. I need to build this on a commercial uh uh not yeah at home uh 3D printer. So, I was kind of my my um my size constraints were were pretty strict. So, I need to make sure I've got a decent sized battery which I can fit inside my hole. um without it interfering with everything else and giving space for sort of a bit of uh flowing of air and uh trying not to let things overheat. The second thing is that as you can see with the power one, we we we cut that power. So normally to have obviously conductive cabling or something going in allowing you to be able to charge a battery pack. So we we're going to need to do this wirelessly. Um which I'll get on to in a minute. First of all, the uh for the battery system, I ended up going with a lithium-ion phosphate battery chemistry cuz it's as far as the lithiumion uh class, it's it's very stable. Uh these actually were supposed to go in a car. They were surplus being tested, I think. Um and I came up with uh a good voltage that won't do me any harm, but is high enough to be able to get as much power as possible, is about 26 volts, which is what what these run on. Uh to charge them, I think you need about 30, which will be important in a second. Um, and this system put eight of them together get 65 W hours. Uh, with the amount of power that's consumed, we get about 40 minutes of run time, which I thought that's that's all right for for a first go. If I can get anywhere near that, I'm doing well. Um, so great. So, I've got a system and and and then I had to develop my own uh battery housing. So, this is, you know, four four cells, then some uh uh collars, then another four cells, and the end cap just to try and make it as compact as possible. So, it really fits in amongst everything else that I'm going to need to fit in there. There's still a lot more to go in, but that's where where we got up to. Um, but hopefully still a good half of it free to be able to fill with electronicy things. So, the next step is great. So, I've got a battery. Now, I need to use something to get the power in. So, we're going to rely on something which I'm sure you're all familiar with, which is uh wireless charging, which there's lots of types of wireless charging from a toothbrush through to like sort of iPhone and more advanced things. Um, but I'm going to use uh inductive uh resonance coupled uh charging which is like uh Q chargers or what you get in your phone cuz that's relatively efficient and relatively widely available. Um you can see here we we kind of got my test setup. We we're measuring current and voltage and we're trying to put a load on it. Uh running at 24 volts cuz that's that's the um the voltage that we're going to be using for charging. Um, and I ended up having to create two of them uh in uh in in parallel and because I needed to get enough uh current to go through and also give a bit redundancy. We've also got a a boost converter to get the right voltage cuz I think it drops down to 12 on the outside and there's actually 5 mm of damp. So here you've got two plates that are 5 mm separated. It's this is not the plate. This is about this is a secondary coil but they're they're separated by 5 mm. We got about two amps going through it at the moment. Great. So it can put through a load. How about let's test it on the actual thing I want to use it for rather than lots of complicated calculations. So here we are. We are we now actually got the setup connected directly to the battery and we were able to get an amp of charging at nearly 30 volts into this battery pack which means I should be able to charge it in a couple hours which to be honest it was was was better than I thought I was going to end up with. So um I think we we are on to winner. All I have to do is is create some kind of docking station for this which um you you'll see in a minute to uh to be able to house the other side of the wireless charging. And then we we've kind of got our charging system sorted. So great. Right, next system down. This should work. So we can get power in, we can store the power, and we can drive the power out. But um at the moment, it's only going to do either all the power all the time or none of the power all the time. Oh, sorry. This is just a quick quick animation of of it all kind of like fitting together. So that the wireless charging coils we can we could just stack them on top of that battery pack uh with the the the boost converters behind them and also yeah the redundancy as well just in case cuz they were they weren't great quality ones and I had to I was a bit worried about overheating. So I uh I thought well you know if one breaks then I can still charge uh something and I've still got a system that works. Um so next subsystem wireless control. So we we can get power in we can do stuff. Now we should focus on making go faster and slower. So the throttle, the master switch and uh you know collecting data uh as well. So these challenges kind of constraints we've got again we've got no holes in the hole obviously uh for human interface devices to control the sea scooter and um electromatic fields don't really transmit well through water particularly for things like wireless and Bluetooth lower frequencies they do to some extent and the lower you go the better you get but then you get other complexities with antennas and uh data rates and stuff like that so for the sake of this and also the fact that if it's controlling the throttle I I really want this to be reliable cuz I don't want it to sort of just jam on or uh or get stuck off or or anything in between. Um so I wanted to be kind of like yeah pretty reliable. So I actually ended up going with um is this playing? I will play it. Okay, this is it all uh a combination of hall sensors read magnetic sensors for control and using some Wi-Fi to send stuff back telemetry and other stuff back. So here we're connecting up the master onoff switch onto it. There is sound to this one as well, but um it doesn't seem to be happening. Um and uh now this is the hall sensor. So this is the throttle. So as we go closer and further away, you can see it's spinning at different speeds. It also sounds different. Um next to it further down, that's going to be the switch that we can use in a sec. Uh which is another hall sensor which tells it when it's in the docking station and therefore outside of the water. And so it can start trying to uh use wireless going up to the cloud and coming back down to my base station. and see it. I've just put the magnet on the base station. So, it thinks it's in the base station. Now, we've got uh all kinds of telemetry. Can you read it? Okay, just about this one says pressure over here. So, we measure the internal pressure see because I you know I don't want to be doing this and it's it's it's slowly taking on water and building up pressure. Same with temperature. I don't want to overheat things start to melt. And uh also humidity as well. another sign for whether things getting wet inside and lots of other uh metrics as well. Uh that sort of just get downloaded from the last time the sea scooter ran. So I I think I think we got a system. I mean those are all the parts that that we needed to put together. Now is the trick. Can we bring it all together and put it into a single print? Now this is the whole point of this uh did that one. Yeah. So the print. So this is kind of bringing it all together. This is this is what one of my my 3D printers. Um so we know we need to print a hole in one of our main constraints. We need to print this hole in a single piece. Now this is the start of the print here. Um let's just go through it together. So I've kind of obviously sped it up. I think this print was about 13 hours. Um so first of all I'm dropping in that magnetic um iron conducting um concentrating layer. So these magnetic rods poking them in during the build. So we paused it. Pop a bit of glue in just to make sure that they're not going to rattle around at all. They were fairly snug, but um I needed to make sure there's a little bit of clearance so I didn't have to hammer them in. Right, we're off again. So, carry on printing. So, we've got magnetic gear as all the bits are in there now, ready to go. We've got a a threaded piece that we're just putting into the um the hole there. So, when we go on to our first major widget insertion. Okay, here we go. So, this this is what we just looked at. This is the main widget we're putting in. We're putting two in. This one weighs over three kilos with all the magnets in it. And we're just putting that straight into the build. This is all part of the same print. Dropping it down. It's really important that the height and the orientation because all the systems that are magnetically uh controlled need to know where the sensors are. So, we've got little notches that it's clicking into and it's got a a thread to be able to screw into the right height. Great. We started printing. It looks like it's going to fail. It it doesn't fail because I realized that 3 kilos pushes your base plate uh down. So, I had to move it up by.3 mm to save it. But I think I only lost like one uh one one lap, one uh layer. And we keep printing nearly all the way up to the top until we get to the top where top where we want to insert a second piece, which is the wireless charging units. They just connect in. This one's uh on skeleton thing. Still quite big, but not not as impressive as the last one. Um another upshot of this is the big flat surface we've got. makes this a lot easier to um to print over the top of and and there we go. So that that was one print everything inserted. I then uh covered it in in epoxy and um and let let's see if the thing will actually work. It's the maiden voyage [music] of the sea scooter. Let's take a swim and see how fast it's going to go. [music] >> Well, hey. [music] >> Okay. [music] >> Right. Let's give this a go. So, it's actually quite quite hard to find a [music] place that you can do this. Someone will let you test an experimental device in. >> Very cold. Oh my god, this >> someone let me use [music] their pond. >> Oh god, this is cold. >> And it was quite cold. >> So, it's gone in. And the first thing I do is I try and max out the throttle and it slips. So, we have to start again. So, we back up the throttle and we're off again. >> [applause] >> Thank you. So, the thing works which which for a first time and that that's the final uh printed 3D scooter. I didn't mention the handles, but that's that's how we get the the proportional control with the magnet in there as well. There's a bit of work there, but yes. So, so it worked. So, um, so you can print a ship in a bottle, uh, using a 3D printer, uh, and quite a lot of design. Um, I guess there were some things we maybe saw in there which you you can, uh, do. So, so it worked, which is fantastic, and I was very happy. What next? So, can we do better at the the waterproofing of it? So, the next thing I I kind of looked at, look more recently, is can we change the way we do 3D printing? So, there's some other techniques that you can do. You might be able to make out. It's maybe a bit difficult, but we've got uh normally you you do 3D printing in layers. Um this try a new technique which is where you stagger the layers. So it looks like a brick wall instead, which means it's harder for water to get through. It's no there's no planes that can separate and it's uh you can get a higher fill factor. And so I did a whole of testing to show uh how much better this worked. In fact, I don't even need to do the uh epoxy coating on it. I'm almost confident enough I can definitely make holes that I can go up to full bar and they don't leak at all without any uh uh without any um uh coatings or post-processing which really surprised me. Um and then the next thing was can we give it more power? So we've definitely the system we've designed I designed can definitely accommodate more power. Um the weakest link is the magnetic gear at the moment cuz that's that's maxing out its torque and then it starts slipping. So how about we uh we just make a bigger one. So, so this one is designed to instead of being like small compared to it's actually the exact same diameter as the hole. Uh, so the idea is it kind of is just more like a tube. This has got half a kilo of neodyinium magnets in it. Um, it puts out about uh between 5 and 10 times more torque from what I've measured than the the little uh the MIDI gear which I I've built. Um, so if if you like um this kind of stuff, I've got lots of videos on my YouTube channel if if people would like to have a look. a lot more detail, a lot of other things I've built. Um, and yeah, all this is documented there. Um, and uh, as I said, I've also got a magnetic gear talk which is going to happen shortly after here where we're actually building miniature magnetic gears. Um, so things you can take home uh, which work exactly like this. In fact, they're exactly the same uh, gearing and kind of like design but really scaled down to get the max power out of something which is as big as a large coin. Um, you can you can take home. So that is it for me and from design to make. Um thank you very much. If anybody's got any questions I will be uh in the Q&A tent. [applause]