I have BB's so I may try this with this gearbox, though I probably I won't be able to repost the design since the original is patreon exclusive.

I am afraid that airsoft BBs are too small to make a solid bearing. Because of small diameter they may deform plastic raceway. And also here you would need to make deep grove bearing or angular contact bearing (for the trust loads (axial), which would not be easy with the small BBs.

BTW, Let’s Print youtube channel have tested different 3D printed bearings, and Airsoft BBs one was not durable at all.

@@Skyentific Great, thanks for saving me the time of attempting that

I have tried a small 3D printed NEMA 17 7:1 cycloidal gearbox using BB's as bearings on double angular contact races. Did not work out well. Too much play in radial and axial.

Thank you. I think this kind of actuator can be used for many applications. And this is exciting :)

Interesting idea...thanks for sharing 😀

I've been using drill + sandpaper

That might require pva between the two rings though but maybe pausing and adding kapton tape on the seam might work.

That would add quite a bit of backlash in the gear though, and that's not what you want for robotic acuators.

This is a great idea!!!

I don't randomise the seam. Instead I just file it with sandpaper or similar.

@@eugene-d I tried this. It takes too much time to send 38 rollers per bearing, so it is 76 per joint, times 6 or 7 joints. It is too much :)))

@@Skyentific , that's true. Then fine-tuning the slicer settings seems very reasonable.

@@Skyentific You can try using the hand drill and some 3d printed fixtures to vastly accelerate the sanding process

Thank you!

Absolutely, I like the video to push towards high load tests!

Using a cheap brushed motor with a printed plastic ring and a photointerrupter like paper printers use should make it at the same price but including an esp32/100+mhz arm board (stm32) and the driver. Using a cheap magnetic position sensor on the output might work pretty well too. Maybe using both given the price of an optointerruptor and printable transparent paper. I think making a cheap stm32 or esp32 board with an h bridge and connections for an optointerruptor and/or magnetic position sensor would be a great thing. It should be doable for

@@satibel why people are bent on using brushless servo motors when TM motors and other brushed DC motors are very cheap. Are we putting a million miles on them? if economics are the problem then use the cheap stuff to demonstrate potential value of machinery. The only real cost is the servo driver and PSU.

@@togowack with 12V brushed motors, you can use a free or very cheap second hand pc atx psu, they can usually handle 20A or more, so it's not even a problem, just use one per motor and run them at ~50% load, so they last and have better efficiency. windshield wiper motors are also usually fairly cheap (10-20 bucks in a junkyard, especially if you buy like 5) and have plenty of power for most applications (30 to 120W with over 10Nm of torque) also if using that with like a 1:2-1:3 pulley or even direct drive you can probably have a good enough actuator (15-45 rpm would be decent)

@@togowack You guys are spot on the money, BLs are overkill for a general robotics application. I've been testing with some large brushed worm and 'box' type gear motors and they give huge power for the money. With a potentiometer or encoder you can get a reasonable accuracy. Lowering costs is key especially once a robot needs ten or twenty joints.

Awesome tip. Gonna experiment with this.

I do this but go one step more. If two surfaces meet and are fastened with screws there is no need for the surfaces to be flat. They are flat when made of metal because flat is an easy shape to machine. But with 3D plastic why not make is wavy or dimpled like a golf ball. I use truncated pyramids with long rectangular bases between the screws. Or if the surfaces are round like pipe flanges the bases are made of concentric arcs. I use about 1/4 of the surface area with these. You might call this "intermittent tongue and groove." The idea is that 100% of the shear force is taken by the interlocking geometry and the screws only see tension force In Fusion360, I fist draw the base, extrude up with 15 degree taper ange. Then use this as a tool to subtract from the mating flange. Finally, I fillet ONLY the make features. It is self-aligning but, better is is 4x stronger in shear

Спасибо!

I think the problem is that the rollers have a cone shape. He mentions it at 5:42

@@brad3378 You're right I must have missed that. I just don't like the idea of using 3d printed plastic for bearings. Might only work in the short term.

@@onurmemis3618 I am not at my PC right now so I can't look it up right now but look up the metal bearing used by the video featuring the 3D printed Halbach array motor. They're less than $10 each on Amazon