final project -- weekly assignments -- about me -- fab academy
I redid this assignment in 2018, as I wasn't able to finish it in 2016, and didn't really want to go on with the old results. So, look below for this year's attempt. The old stuff stays in here for future reference, and as a reminder of ye good olde times.
Due to work troubles, I couldn't go to Kamp-Lintfort to mill a mold, and Daniel couldn't, either. So, we built a mill. From a 3D printer. And tried to get it to run with fabmodules-generated G-code.
It works surprisingly well...
Lessons learned, so far:
The results are quite nice for the parts without lost Z-steps (Daniel's test part, as mine isn't done yet):
Sadly, there is a lot of work left to do in here...
final project -- weekly assignments -- about me -- fab academy
final project -- weekly assignments -- about me -- fab academy
It's 2018, and while the Fablab still has no working mill to use, I do have access to one. Not on my own, as it's a real machine and not a toy, and I'm not a regular user. But I can get access to it, big thanks to Prof. Seabra and Daniel from the Technical Centre for Thermal and Mechanical Process Engineering:
Given the circumstances, I will not be milling anything from wax, as it could react with the coolant in the machine and lead to a very expensive and very hard to clean out problem I don't want to be responsible for. I will also not mill anything from plastic - It would be interesting to try, and we will try at some point, but I couldn't get any piece of millable plastic large enough in time. So, I will mill a mould from aluminium, which is a bit easier to handle (the piece of stock I bought is still almost 3kg, a piece of steel would be a lot heavier) than steel.
For this assignment, I want to make something pretty mundane - New grips for a small joystick. It's loosely related to my original final project (which I will still build, just not during fabacademy), as it's the kind of joystick I want to use for the placer, and a few other machine ideas:
Old remote controls are extremely cheap to get, as nobody really wants them any more since the hostile takeover of the 2.4GHz systems. I don't want one to use it, either, but the joystick modules in there are usually really good (as opposed to a lot of newer systems), not only electrically but also mechanically, so they are a really good base for my own input devices.
Fun fact: You can buy the exact same modules as industrial joysticks, and they are still being made, for around 200€. A used remote is around 1/10th of that, and has two of them...
So, I really like those little sticks, but what I really don't like is their tips - They end in rather sharp-cut "crowns" of metal teeth, which is great of you're standing outside in the rain with thick gloves on and still want to have precise control. It is not great, however, when you have the joystick integrated into a panel and want to blindly grab there with your bare paws... So, I want to have a rubber grip for them. There is a lot of soft casting silicone left at the lab, so that's a match. The test cast with it looked good, too (those drops are water from cleaning it):
I would have liked pink silicone better, but that is actually hard to get if you don't want to mix the colouring yourself.
Designing the part started, in this case, with an idea and callipers to get the measurements. The form is really simple, too: A half dome on top, a slightly tapered body below and a flat bottom with a hole for the stick in it. I measured the original grip of the stick at 8mm, estimated that a 30mm diameter grip would be really chubby, but nice to grip (when you have your palm on a panel and want to just carefully move the stick with your fingers). That set the dome radius, and the rest was drawn by how it felt right:
I also rounded off the edge of the bottom side, since we only have a ball end mill with a radius of 3mm for finishing:
I used Autodesk Inventor to model it, as that is what I'm most used to.
Putting a mould around that design would be possible in a number of ways, and, depending on the method of casting used, most of them are not very sensible. I mainly considered two versions:
The first version would have had the distinct advantage of having less parts, and the inflow channel could have been hidden in a spot low at the side of the grip so its remains are less visible. It is a nice, simple concept. I didn't do it like that, though, as the bottom part would have been really deep and hard to machine - down to being impossible with the tools we have. I didn't want to buy a 100€ tool just for that.
What I did draw was the second version. The inflow channel is in about the most visible spot possible - right at the top of the grip. A good, sharp knife should solve that problem, though. The mould gets simple that way, though, and a lot better to machine. I drew one symmetrical mould half, which has, of course, to be milled twice:
The third part of the mould is a simple cylindrical dowel pin of the same type I use to align the mould parts:
Two of those eight holes in each half are used for alignment, four for pressing together the parts. Having M8 screws in there should be way, way more than enough to hold the parts together tightly.
At what will end up the top end of the mould, I also added a rather large funnel for the silicone. I hope to be able to just cast all the silicone I need into there, then have it slowly drip into the actual mould from there - the channel is rather tight (to be easy to cut off).
From having a model of the mould, the CAM system will bring you to an actual set of instructions for the machine. We use Autodesk HSM Ultimate here at the moment, and it seems to work, albeit a bit finicky at times.
The very first step of the CAM process actually was not done in the CAM system, but back in Inventor: The machine has no ball head tool (for finishing) long enough to finish out the 20mm deep funnel, so I had to make it slightly slimmer. That funnel is an absolute "don't care" thing, I just want a large volume in there so I don't have to refill it, and I gladly just reduced its diameter to 15mm:
Next, the mould half had to be fit into an available piece of stock. In my case, that is an aluminium profile with a 50x50mm cross section and the length I saw it off to be. Having a stock of something like 50x30mm would, of course, have been more efficient, but that profile wasn't available to buy.
Then, a lot of milling phases have to be defined. The process is iterative, and each step first has to be simulated, the results checked, settings tweaked, simulated again, ...
What I ended up with is a series of different strategies applied to a piece of metal:
It was clear, at this point, that the opening for the centre pin can not be cleared out perfectly - it's just not possible with a realistic tool geometry and three-axis milling. So, the pin will stick out a bit more, which is perfectly acceptable for now - All aspects of the grip design are estimates, anyway. If it turns out the grip will end up too low like that, I can still grind down the pin to match in what the machine can do there.
While the actual CNC milling is a rather simple affair (the machine does the work, you "just" have to check it's doing the right thing), setting everything up is not. The whole production process for a mould half is a lot more work:
I couldn't take that many photos of the milling process, as it's quite hard to photograph - most of what you see is an inferno of brown water splashing everywhere, through windows that are covered in brown water, too. A few are there, though:
Drilling a hole. The process is is surprisingly unspectacular and quiet.
During the Adaptive Clearing Cycle. Lots and lots of brown water, and a slowly emerging shape.
The tool change allows for a somewhat decent photo opportunity. From the distance and through the water covered windows, it already looks like the mould it's set to become.
Finishing cycle of the funnel. You can see the edge of the finished area when it's not behind the tool, and the difference in surface when it's momentarily not submerged in water.
After the last finishing pass, the mould looks shiny, but is still mostly full of water.
As the mould halves come out of the mill, they are shiny, but their edges are still sharp.
After very careful deburring, the mould halves are ready to use. If the grips made in them turn out to match what I want to have, I might still polish them some more by hand, but right now, that is not necessary.
Casting is, again, a multi-step process: First, since silicone can stick to aluminium like glue, a separating agent has to be applied. I used the "wax" from a very cheap candle for that, which is a bit hard to apply, but can be smeared on and then made into a thin, nice surface using a paper towel and patience:
Here, the left mould half is covered in wax, the right one is not.
The centre pin, as a third part of the mould, has to be covered, too, of course. As a small advantage, it sticks in place that way:
Also visible here is that the pin is not at the bottom of its opening, as that geometry was impossible to mill. I'm still ignoring that problem for now, if I want to have a shorter hole in the grips I can still grind off from the pin or buy a shorter one.
Then, the mould halves are screwed together, using two additional pins to have them aligned:
Now, it is time to prepare the silicone, which is made from two components that have to be weighed in and mixed:
The silicone is then poured into the funnel of the mould, and the mould placed into a small vacuum chamber. Evacuating the vacuum chamber for a few times, then letting air back in transports the silicone from the funnel into the mould.
After letting the silicone cure for a few hours (and getting some desperately needed sleep), I opened the mould to find... bubbles. Only a few of them, but more than I would have liked. It should be possible to get them out in future casts (by letting the silicone "cook" for a few minutes at a strong vacuum, so the silicone bubbles take the air with them), but for now, it is usable and looks good enough for me to use. I will decide after some testing if the form is good, or if I need to change it before casting more of them.
A sharp knife easily takes care of the pouring channel, leaving a knob-sized joystick grip:
The source files for the grip are available for download, of course.
final project -- weekly assignments -- about me -- fab academy
