You know the old saying. When all you have open is KiCad, everything looks like a PCB. That was certainly true for [Evan], who needed to replace a small part recently and turned to PCBs to get the job done.
The part in question was a sheered apart detent cam from a retractable cord reel. Glue and epoxy might have worked, and [Evan] was worried about how a 3D printed PLA part would have held up. The part is an extruded 2D shape, making PCBs a non-traditional but viable choice. Using the old scanner trick, he traced the outline in KiCad 7 (which adds image references). Then with the five boards stacked up, solid core wire, solder, and a propane torch worth of heat fused it. Ultimately, this machine’s tolerances are generous, so it worked wonderfully.
Was it the “right” tool for the job? Right or wrong, it is hard to argue that in terms of durability and ease per dollar, this doesn’t come out on top. PCB files are on GitHub if you have a 5020TF-4c retractable cord reel that needs a new cam. PCBs have a fun way of adopting different use cases like enclosures, but perhaps the idea of PCBs as a mechanical part could be applied elsewhere.
Good job getting the job done. I would have just stuck with 3D printed PLA or PETG, though. If a straight printed part didn’t hold up, I would try printing it at 100% infill and then remelt it in packed salt.
I made a force feedback steering wheel from scratch, and the two stage 40:1 gearbox is all PLA and skateboard bearings.
A chunk of brass and a needle file is probably the best (and imho easiest) way. If it’s not a part with close tolerances like this.
I’m very interested in WHY would you need 40:1 gearbox for this and how do you actually allow that wheel to turn. Seems pretty interesting.
The wheel has two DC motors in it, of the type in a cordless drill. I probably only needed one, but two was more fun. If they’re say 150 revolutions per second at max voltage, the gearbox drops the wheel down to about 4. That’s pretty fast steady-state, but about right for sudden steering movements. Unpowered it freewheels pretty easily, but when powered up the control algorithm matches duty cycle to measured velocity so that the wheel moves effortlessly for lack of any force effects being actively applied.
I originally started with an 80:1 gear reduction, but it wasn’t necessary and did have slightly worse feel. When I switched to 40:1, I also wired the two motors in series rather than parallel, and it’s pretty good.
3D printing – PA (nylon). CNC – aluminum. By hand – aluminum or brass (hacksaw and files).
Or he could have used their 3D printing service since he was already on their website and made his life a little easier
I don’t think it would have made his life easier. He says himself that he’s not confident in his solid-modeling CAD skills, but makes PCBs extremely regularly. I doubt it would have taken him less time to sharpen dull solid-modeling skills while making the part than to use one of his daily tools.
It also probably would have cost a little more ($30 shipping included for a simple PCB, I mocked his part up and it’s ~$20 before shipping), and been weaker (PLA vs fiberglass). For him, though, it probably would have been free either way as this was a sponsored deal.
The profile is exactly the same for 2D or 3D (really 2.5D), send PCBWay the vector (DXF etc) from KiCad and tell them how thick you want it.
Laser cut, milled or 3D printed in metal and you get to skip the soldering part.
Yeah, textbook “hammer / nail” syndrome.
Or he could have them make it from nylon or even steel. Probably also more expensive though. Still, a piece of steel and an angle grinder would have been my choice.
If he was able to trace the part from an image in KiCad, he could easily have done it in fusion 360 or similar. It really isn’t that different a process.
Woven fiberglass in resin, excellent strength I presume, probably cheaper than an OEM replacement part, if available.
Brilliant!
The glass in the FR4 that is exposed on the edges is very an abrasive, so this isn’t an application I would typically recommend (that pawl is going to wear out fairly quickly). But it is certainly going to be strong enough to hold up!
Yeah good point. Nylon makes a very nice low-friction contact surface. Edges of fiberglass are basically sandpaper. I don’t know if this is some kind of taboo around here, but one can just cut a shape out of nylon with simple hand tools in a couple minutes… No need to involve robots or chemical processes. Insane, I know!
For a one-off design, a cheap nylon chopping board + cheap hand tools is hard to beat. They are also very resistant to hot water and caustic solutions (like the inside of a dishwasher). Nylon parts are easy to tap for screw threads. You can 3D print nylon (Some people even use “line trimmer” cord) but you need a hot nozzle and it can warp a bit, or spit if the filament has been exposed to too much humidity.
Take the aluminum foil commonly wrapped around chocolate bars, and line the cog faces with that.
Most home kitchens have foil already, but with my method you get a chocolate bar.
In this very specific use case, phenolic would’ve been a very good choice. Phenolic gears are common in automotive applications, where they’re exposed to heat and oil.
But i don’t know if the board house offers it. For electronics it’s much worse than FR4.
Then he’ll just have to order another PCB to replace the pawl
Not a hack. It was carefully designed. Since it was carefully designed it could have been properly implemented. Tough crowd, I know.
The PCB idea is very cool. A 3D print with nylon would also have been a great choice here.
I faced a similar issue on an automatic garage control hybrid gear/sprocket. No replacements were available so I loaded up OpenSCAD, generated the part and had it printed in nylon. Nylon part lasted for about 5 years until the rest of the system wore down.
I made a similar project as an experiment. I used woodgears.ca gear generator and ordered 5 gears made from pcb material. the idea was to use two to three roller microswitches to make a rotary encoder. It sorta worked, but there was too much hysteresis in the switches to roll properly.
That pawl doesn’t look right. Installed upside down?
That’s what I thought at first as well, but it looks like the pawl reverses when it passes across the portion of the gear that doesn’t have teeth. That section is cut deeper and allows the pawl to flip.
Now that I know what’s inside of these I will take better care pulling and letting go of the cord. Gently. Because there is a little plastic tooth inside ready to break. Don’t trip over the cord and send it flying back in.
Am I the only one who notices that PCBway logo on the PCB, it looks more like an add than it does like a fix. Why is that logo on top and not the link to the website of the one who actually did the project. This smells like product placement…
Yeah it was sponsored