Yes, it works!

Well, not that much to say, only that it’s pretty hard to capture this in its real beauty — for some reason unknown to me, photos will always be taken in the wrong moment. But, then again, who cares:



For now, just a bad short video with the mobile phone, but enough to upload it to YouTube for now:

Another one is printing right now — the kids have been asking!


Maybe it works!

This is so exciting! I put everything together today and, while there is still a way to go, now I definitely have a feeling like this is going to work. My first machine!

Assembling it I soon noticed small things which I could quickly address by printing new parts — mainly the bubble ring needing sufficient clearance on both sides, so I printed small spacers.

I decided to go for rubber bands instead of GT2 tooth belt, because I had no idea of how to glue a tooth belt to a continuous loop of a certain length, and, well, I want to publish this and a GT2 tooth belt isn’t exactly someone people have at home normally. Common broad rubber bands like we use them at office for large acts have a too small diameter. So I used elastic ribbon, the kind you have in your underwear. I used needle and thread to make a continuous loop; you have to take care that it’s just tight enough to have enough friction on the pulleys, but loose enough so you don’t have too much friction by skewing the axis. Well, it does work and it looks fantastic — when I have one on either side and the wood painted black as well, this will surely remind people of a jet engine! Well, maybe, at least:


Well, it works sort of. Problems remaining:

  • I need black elastic ribbon!
  • the large pulley behind the bubble wheel is so large that it pops most of the bubbles; I’m currently printing a smaller one, but if it get’s too small, my mechanical advantage might get too small, too. We’ll see in, let me see, 13 minutes!
  • the belt retainers on the pulley are too small for my rubber band, which keeps getting of after a short time, but I already accounted for that with the pulley printing just now;
  • my tank is not completely water-proof; it’s okay for now, but it get’s soapy on the outside. I’ll try making the walls a little stronger and with more infill for the next tank.

Aaaah. This feels very good.

The bubble fluid tank

Ok, contest deadline pressing I did the tank today. While the code is certainly chaos by now, I really love OpenSCAD: if you’re reasonably fast with typing, you can model quite fast, you have a very natural, easy way of accounting for certain constraints, and it’s fun. While of course a good OpenSCAD design should be fully parametric, allowing to easily scale my design to any wheel diameter, this is in reality not always easy to achieve, however. The tank I came up with still takes some modifications if you scale parameters by more than a bit, but hey, it’s my first machine!

Ok, so here’s the tank, fit with a mount to screw it onto a piece of wood which is supposed to be mounted on the back of your bike:


This takes some important things into consideration: the largest part of the tank should be only above the lowest bubble ring, to save on bubble fluid. And, to reduce spilling, I inserted bulkheads (which also serve as internal supports during printing), as well as a small wall in the upper part that also tries to reduce spilling. Here’s a cut to show how it looks inside:

tank-cutAbove the bulkhead walls you can see small holes supposed to let air circulate, so you don’t get problems with enclosed air when filling this with bubble fluid. Of course, a cap to close that refilling hole still has to be printed.

This has to be printed reversed, i.e. with the top down. It will still need support under that mount and probably in the middle of the half circle. There should be no support in the tank itself (except from above that refilling hole), as you won’t be able to remove it easily, and it won’t be needed because of the bulkhead walls — so I let the slicer generate support only from printbed. Although the first layer has a very large surface, I could get the first layers to print correctly until I turned on a raft; this is probably due to poor calibration of my printer.

This thing has now been printing for more than 4 hours, which is longer than my printer has done ever — it hasn’t even printed half of it, though. We’ll see. It’ll be a pain after this, when I’ll find out this doesn’t work …

Slowly getting somewhere …

Back from a long summer vacation I finally got back into working on this, with the deadline for the thingiverse windcraft tech challenge pressing me. The good news: I’ve finally found the design fault that made the wind wheel so problematic while slicing — the rendering quality of the linear_extrude command with tilt=50 greatly increases by not only increasing the number of facets ($fn=50) but also using the „slice“ parameter. While I have absolutely no idea what this parameter does exactly, neither do the authors of the OpenSCAD Manual, where it simply states: „The slices parameter can be used to improve the output.“ Well, setting slices=100 did it for me, so while not precise, the manual is obviously right. Because theres only a thin surface touching the bed, a raft is in order. Also I increased the brim surrounding the wings, to give it more stability during printing. The result is this:


On another note, I found someone aiming for something similar as I do — a wind bubble machine, published on thingiverse on July 1st, shortly before my vacation. It looks cool and the idea with the gear design is charming; I’ll try it out when I have time, but won’t bother now as it will be quite some printing and I did not find yet any fotos or videos of it working under real conditions. As of now, I will not go for such an, admittedly beautiful, design, for the following reasons:

  • I found out by now, that the resistance of the fluid actually is a problem, and this will probably increase with the teeth of the gear going through the liquid; this may be neglectable because of the mechanical advantage of the gears, but I would like to go without
  • with the bubble rings being inside the gear, you have less room where the bubble fluid is actually effective, so you have to refill more often
  • while having the fan mostly in front of the machine gives a compact overall footprint, it shields a lot of the travelling way of the bubble wheels, probably reducing the amount of bubbles produced during one turn of the wheel
  • the tank is not done with bicycle usage in mind and will definitely spill its liquid after a few meters on the street; so it may be fine for the garden, but not for my purposes

So I’ll continue with my tooth belt approach. As most people don’t have tooth belts at hand, I will also evaluate a rubber band or similar things.

Separating Wind and Bubble Wheel

I decided to try to separate wind and bubble wheel, connected with a belt, making it possible to

  • have a larger surface area for the wind wheel
  • have the bubble wands dive deeper into the tank, thus hopefully reducing the amount of spilling, while having a larger reservoir of bubble fluid
  • gain mechanical advantage by using different sized pulleys

second_ideaAn obvious problem is the friction of the belt transmission. A gear train would be even worse and would add more wind turbulences. At first I wanted to use a large rubber band for the transmission, but a friend suggested the tension would add too much friction, so I will go for some GT2 tooth belt I still have left in the drawer. I found a collection of parametric pulleys here, creating these two pulleys:

small_pulley large_pulleyThey’re supposed to be fixed on the shaft with an M3 screw and a captured nut. I did the same on my new bubble wheel and wind wheel:

bubble_wheel_2wind_wheel_2Trying to print that larger wind wheel (diameter of 160mm) unfortunately turns out tricky — I don’t know what’s wrong with my design, but both Cura and Slic3r decide to slice incomplete layers, the wings oscillating between their full width and getting to small in the middle, leaving a gap:

slicing-problemI have no idea what causes this, but for sure it is a problem of my design, which I actually liked for the simplicity of OpenSCAD:

linear_extrude(height=wing_height, twist=50) {
  for(rot = [0, 60, 120, 180, 240, 300]) {
      square([(ring_dia-inner_dia)/2+2, spoke_width]);

Well, for now I managed to print one by simply using a higher spoke_width of 2.8mm, but I still wonder what caused this … and even that had a sad ending midprint.

I think I’m frustrated for now, just in time to spend two weeks of vacation in Austria, thoroughly clean that nozzle and then try again …

First steps: the too simple solution

Designing the first version of the wind wheel, with bubble wands directly attached, given my lack of experience with OpenSCAD, was surprisingly simple and took 30-45 minutes. While there are plenty of wind wheels out there to download, nothing seemed to really it and I decided to do a simple linear_extrude, with a twist setting. The result looks nice:


For the shaft I used 2mm welding wire, one that comes in rods rather than as a roll, which worked fine. I printed a few small rings that could be pressed on the wire, to hold things in place. The whole thing is supposed to be attached to this bubble liquid tank:


And so, the same day, my first prototype:


Let’s be honest: it does make (a few) bubbles facing a fierce wind but we certainly can get better. Mostly this thing spills a lot of liquid with every little bump on the street. I learnt, that while having the wands directly attached to the wind wheel makes things very simple, my first design has some obvious and serious disadvantages to it:

  1. My printer’s build space will only allow for a certain size (in the case of my Kossel mini up to 170mm diameter, damn I’ll have to win some professional machine in a contest!). This means that the size of my bubble wands has to be substracted from the size that still is available for our wind wheel. If you take a look at the wheel above, it’s easy to estimate that getting rid of the bubble wands would allow me to print a wind wheel with more than double surface.
  2. As a consequence of 1., it’s difficult to reach a wind speed that will steadily overcome the friction the bubble wands meet in the liquid.
  3. As the wind wheel should not dip into the liquid, the height of the tank is limited by the size of the wind wheel.
  4. As a consequence of 1. (short bubble wands!) and 3. we have a very small tank.
  5. The size of the tank is even more problematic because we won’t get any bubbles anymore when the liquid level falls below the point where the bubble ring starts — at that point the soap won’t coat the ring anymore.

Because of 5., the tank stores much too much liquid which will never be used, which is a huge waste: because we can only make bubbles until the liquid level falls under a certain point, the following to tanks would get you the same amount of bubbles, but the first one will store a lot of wasted liquid:


But not only this is a lot of wasted liquid — it will also contribute to the problem of spilling liquid, because the amount of liquid in a simple tank will add to sloshing. That’s why in my second design of this tank I decided to go for the idea on the right. In addition, I decided to put bulkheads into the upper part of the tank to reduce sloshing effects:


On the photo you can also see a simple mount for the luggage rack of my bicycle. With this setup I was able to produce a few bubbles, but I had to get pretty fast and it still spilled much.

I now decided to drop this simple design and go for a more complex solution which should solve the problem of overcoming the friction in the bubble liquid as well as spilling and tank size. The idea is to separate the wind wheel from the bubble wand wheel.