• aditya mehrotra.

one stabilizer updates: designing and printing a genuinely cool carrier [updates]

So part of what I'm trying to do is learn how to design and model interesting shapes in cad programs like fusion 360. So the camera carrier on the rig broke yesterday (because it wasn't strong enough and the layer adhesion was off) so we decided to design another one and see how this one faired.

So this was the initial design as well as the initial carrier as you can see we had to put duct tape around it since it had broken. The stabilizer is coming along well, we decided that instead of wasting a print on a huge surrounding frame we'd just build a better carrier and focus on a nicer hand-held, smaller system. Hopefully by tomorrow when this thing is done printing we'll be able to rig this whole thing up and go test it. We might be able to get most of the electronics done today using the NAZA and the phantom battery and other components as the brains and power.

But first, I want to talk about the stabilizer's new carrier design and here it is!

Basically the new design used the sculpt workspace a much nicer workspace for creating more complex shapes. The reason for this shape was to maximize the internal volume, shift the mounting location of the camera backwards, and finally to make the whole thing stronger. We're now going to FEA this design using ABS as the material. Notice how the design has a lighter more skeletal feel to it, and the design is actually designed to be printed very easily.

WE should note this one will be printed out of wood because that's the filament we have the most of and for design purposes. It'll look good.

So first we uploaded the stl and then reduced its mesh density: https://toglefritz.com/convert-a-complex-stl-into-a-body-in-fusion-360/

Then we went into the design workspace and selected "convert mesh to b-rep" which should cover the message to a solid body and it did!

Now for some reason it's still a surface which means we need to convert it to a solid before we FEA it at all.

So there's some edges we need to stitch - I think the modeling we did was less than ideal. Here's some information on fusion surface stitching: https://knowledge.autodesk.com/support/fusion-360/learn-explore/caas/video/youtube/lesson/146730-courseId-101856.html

What we're going to try to do is scale down, simplify and re-stitch, and then scale-up the model. Scaling down might help us get all the free edges that won't be stitched together. There's a lot of them probably because we didn't so such a hot job of creating this model but that's why we're doing it so we can learn. Most of the error seem to be in the corners of random boxes. The corners aren't going so hot for some reason. When we scale them down this might be able to fix the issue, it'll lower the gap enough where 10mm gaps should not exist. The geometry could also be really weird because of the number of faces and triangles that are happening. It also might be easier to stitch not in the STL so maybe we'll try that.

So we went back to the regular design area and here it is.

See what it is, is there's two or three places where there is a hole that we need to fix with PATCH before we FEA this creation. Let's do that.

So we're generating patches for different parts of the structure every time we see a whole. Then when we are done we're going to stitch everything together to create one solid body.

There we go :) that's everything 104 free edges, 104 stitched and now we have a solid body that we can FEA! Let's first determine the loads of simulation. We're going to say the camera and the lens together are like 2kgs which is about 20N of force. We're also going to say that the servo can produce 20Nm of torque, and that one side of this thing will be fixed. Let's see what the FEA does with it.

(1) Let's change the material to ABS plastic.

(2) We're going to make both the 20Nm torque and 20N force act in the SAME direction to put the most stress on the plastic. This should never happen but regardless we're designing for it at the moment.

(3) We now FIX the opposite face that will be attached to the lower arms.

Now we're ready to generate the mesh and run the FEA! Note that this FEA is slightly inaccurate, it's always more accurate to do Assembly FEA for parts rather than part FEA because that accounts for the transfer of loads between devices but no-matter. This is a worst-case test. Let's see how it performs.

Now when we analyze these results we should be careful because first of all, we're 3D printing this which means we need to account for things like layer direction, and second, we need to realize that this body is SOLID and the 3D print is not solid it only has 25% infill so we need to try our best to determine if the part will be fine with infill before printing it. Finally, we're using WOOD PLA and not ABS which will change the results slightly as well.

So at first if we look at this, the analysis kind of seems unhappy since the minimum safety factor is 1.437. But if we look closer, that's in a very small tiny area point. As we move the arrow up, we notice most of the part has a safety factor of > 3. Now where does that leave us if we have 25% infill. If we move the arrow up more, most of the interior of the part has a safety factor of 8+ which means this piece should be fine even when 3D printed. Only some minor outer edges have a lower safety factor. And it should be noted that this part will not be exposed to the level of fixation that we provided here. It's over-constrained compared to the application. All of these combined tells us that this part should pass FEA.

Regardless, at some point in the future I want to look at this link: https://my3dmatter.com/influence-infill-layer-height-pattern/ which described how we should print things and also how we could FEA 3D prints. Maybe there's a way to export the file that Cura generates and FEA that I don't know we have to see. For now, I think this is ready for printing. If we don't do anything too crazy, the layers should not come unbounded.

So here we go printing! I think this is 0.28mm layer height, 25% infill, it should take 36h and it's wood PLA filament. We can't expect the same performance as the FEA but we can expect for a camera rig it'll be fine if we don't abuse it.

We'll check back later as the print progresses. But I do want to say both the CF filament and the wood filament when printed at 220C and 70C have amazing build plate adhesion.

Sad face - we need to clean out our 3D printer nozzle because printing with all these fibers has gotten it very upset. I think we'll probably get some hardened nozzles. That's a waste of good filament :(((((( now I know why they say use a hardened nozzle or something.

We'll come back to this :(

#updates #omgrobots

1 view0 comments
© copyright 2019 | aditya mehrotra