Actuator foot weakness - stage Y-axis actuator broke :(

My kid got a bit too enthusiastic today while operating the microscope remotely, and the foot of the X-axis actuator detached from the stage linkage. Reprint time!

Is there a reason that linkage can’t be more robust? I would print this with ABS, but I don’t have very good ventilation in the room where my printer is, so I generally try and avoid it, especially for 8 hours. If I can make something less fragile and still use PLA that would be preferable.

I’m fine with modifying the STL myself, but I wondered if there’s some reason why it’s constructed the way it is (so small), and whether modifying it will impact the functionality.

Screen Shot 2020-06-16 at 8.19.31 PM

@r.w.bowman maybe that semicircular part at the foot of the actuator could be more of a rectangle?

Also, I see in the sangaboard firmware some code for end stops - I don’t see any STL files or instructions for end stop mounting, so I’m wondering - how do I add end stops? I guess I’d have to use very sensitive and well calibrated bump switches for that. Has anyone constructed a microscope with functional end stops? I’d love to avoid breaking my microscope.

sorry that broke - it is a bit of a weakpoint (and your cross-section shows one of my attempts to improve it, which is the very thin hole through the small semicircular part - it adds some perimeter, which is stronger than infill).

The reason it’s semicircular is for print bed adhesion - a rectangle would have sharp corners, which are more inclined to peel up from the print bed. We could just make it taller, i.e. extend it upwards as you’ve got it in that picture. That would probably help a little, at the expense of changing the shape of the microscope feet, and thus the footprint of the microscope (which is a little annoying, as various folk have now designed accessories that rely on the footprint). It probably is something we should improve in a future version though.

Just to check, how long is the screw in your actuator? The easiest way to break it in the way you describe is using a screw that’s a little too long, as that will exert a huge force on the bottom of the hole down the inside of the column.

Ah, I was wondering why that tiny channel was there. That makes sense now.

The print bed adhesion in my case is a non-issue - I’m using a glass plate with PLA that has unusually good adhesion. Of course, this doesn’t help the folks that do have adhesion troubles. I completely understand why a semicircle is preferable given the context.

I’m using a 25mm pan head screw for the actuator, with a hex nut glued flush with the bottom of the head, so the exposed length of the screw is actually a bit shorter than the design. The other axis is fine, so maybe this was a combination of a really narrow linkage and too-fast print settings.

3 possibilities:

  1. I originally printed with 2 perimeters. Printing the first 2mm of the part with 3 perimeters instead of 2 results in the base of the actuator being printed entirely with perimeters, which should in theory be stronger. I’m going to try this first.

  2. Since there are about .8mm of height from the z origin to where the linkage connects to the base of the actuator, maybe the footprint could be kept the same, but the part could loft out a bit through those first 4 or so layers? That would make the cross-sectional distance from the inside face of the actuator to the outside point 5mm instead of 4.2, giving that section a bit more rigidity perpendicular to the line where the most stress is occurring.

  3. There will still be an issue with repeated flex - PLA is not particularly ductile, so it’s likely that delamination will happen eventually, despite the larger cross-section at the base of the actuator. As long as the stress is occurring along those right-angle intersections, it will cause delamination. Shifting the stress to the midpoint of the linkage (as shown below) might slow that process down. Ideally the linkage should fracture due to repeated stress before the base of the actuator delaminates.


I just look at that and think ‘glue’ - both to repair it rather than reprint, and for making nice fillets. :zipper_mouth_face:

My woodworking mindset keeps overtaking the idea that it should be printed right so it does not need modifying.

Same here! I tried gluing it but since the part is nested inside the frame, and because the layers pulled apart unevenly, it’s pretty impossible :frowning:

As for fillets - yeah, maybe, but whatever glue I use would need to be as flexible as the filament itself, if not more flexible. And I don’t know a glue aside from rubber cement that would fit the bill…

@wschalle We are starting to do some FEA on the flexure to work out how best to limit the stress. The issue with the fillets is that by the time you slice it you have quite a different shape.

@j.stirling Yeah, I thought about that a bit too. The base of the fillet is 2x nozzle width, but depending on the xy resolution of the printer, it may be too subtle of a detail.

FEA on 3d printed parts seems incredibly complicated due to the wide variations in material properties and print settings. I’m really curious - how are you approaching it?

As a simple experiment, I could print out just this part of the linkage with fillets and without, on .2mm and .1mm settings. Then I could measure the moment of force at failure for each and compare.

Would that be useful?

I think regardless of what the stress analysis shows, because of the orientation of most FDM printers, joints on the xy plane will be weak unless they are designed with moment resisting features of some kind.

The part that breaks is not actually a part that is supposed to bend I think. In that case a multimaterial solution would be to print it with a small hole vertically and glue in something with the ‘grain’ the other way: a pin, or a small bolt or a matchstick (woodworking again)

The plan with the FEA is that if we are staying within the elastic limit then material properties are a scaling factor (Young’s modulus). Then we hope to be able to draw layered structures similar to the final sliced object, we can then look at shaping the layers to to smooth out the change in compliance. Not FEA experts here, at first there will be some delving into the literature.

I think it’s Z resolution that really hurts filleting here - even just the jump from 2-3 (or 3-4 depending on height) layers is quite a big one. Our thinking is that we might be able to be a bit cunning with the shape of the layers to get a stiffness gradient, and in fact one of our PhD students (who isn’t on this forum yet) is starting to model exactly that. As Julian says, though, we have quite a bit to learn about FEA first…

Have you thought about defining additional volumes which can be used as “modifiers” in PrusaSlicer to add more infill to certain, weak sections on the mainbody? Someone may look into if this could be done with 3mf rather than .stl output. But I suspect that the whole pipeline OpenScad -> PrusaSlicer + Infill Parameters + Modifier Volumes is just broken.

You’re right, the pipeline of CAD -> shape -> slicer doesn’t allow nearly as much information to be passed through as we’d like. To be honest, I think the weakpoints in the microscope are mostly not improvable by increasing the infill settings - depending on which parts you think are “weak”! The thin vertical legs will never be super strong, and the flexures will always break if dropped, but provided my printer is working reliably, I have usually found it to be surprisingly robust in normal operation.

I do have a very early prototype for v7 of the microscope that uses a totally redesigned (and much, much stronger) actuator - but that will be a while in the coming, I think!

1 Like