6
$\begingroup$

In a design that I need to prototype, I have a coaxial, rotational joint between 2 cylinders connected to another coaxial rotational joint between 2 cylinders by a hinge. I have attached the individual part files as well as their assembly where you can use your mouse cursor to check how they move. Click here for the SolidWorks parts and assembly files

Initially when I 3d printed them by splitting them in half, the outer tube did not close, even after filing the inner tube sufficiently. enter image description here

Could you tell me an alternative method to get this complicated joint working? I need to 3D print them. I plan to use ABS for the 3D printing.

Thanks for your patience :)

EDIT: Lubrication: I plan to use grease to lubricate the sliding surfaces.

Clearances: I printed them without clearances initially (this is my first ever 3D print and I wanted to know how accurate it would be). I then used Dremel and filing tools to file away about 0.5 mm of both the surfaces. It should have been enough for mating the parts. The gap between the edges of the outer shell while trying to close it around is 3-4 mm which doesn't suggest 3D printing inaccuracy. I suspect that the outer tube isn't circular enough in cross section. It meets the inner tube at the the two ends of the semicircle and not inside its belly (you can see it in the picture). I have already tried the obvious, which is removing material from the surfaces wherever the parts touch. The model seems to be too warped for any success. The parts were printed upwards from the bottom, so it is not the warping due to gravity.

Improve this question
$\endgroup$
5
  • 1
    $\begingroup$ Can you just attach photos to the question? The google drive links require permission to open. $\endgroup$
    – Ryan Carlyle
    Mar 7, 2016 at 19:18
  • $\begingroup$ Fixed it. Thanks. You would be able to see the picture and the other files now $\endgroup$
    – Shivanand
    Mar 7, 2016 at 20:51
  • $\begingroup$ Is your issue that the two halves of the outer part are not mating up? What kind of tolerances do you have in the model? It's normal for FDM parts to be oversized by ~0.1mm in X and Y. $\endgroup$
    – Ryan Carlyle
    Mar 8, 2016 at 4:19
  • $\begingroup$ Simple question, but I sometimes forget when I'm designing. Did you account for mating tolerances? As Ryan suggested, you may consider creating a "machining model" that accounts for the capabilities of the process. In the case of FDM, shrinking the part by about 0.007" might help as well as generally giving yourself about 0.005" between surfaces. I wouldn't expect much accuracy beyond that though $\endgroup$
    – tbm0115
    Mar 8, 2016 at 6:26
  • $\begingroup$ I have added some more details in the question (under 'EDIT'). Sorry if I wasn't clear. $\endgroup$
    – Shivanand
    Mar 8, 2016 at 14:08

3 Answers 3

Reset to default
3
$\begingroup$

Consider ball bearings

Depending on the requirements for your joint, I would consider making a design that utilizes ball bearings. With ball bearings, your join will not wear down as quickly, and also move smoothly.

Handling dimensional inaccuracy

The topic of how to achieve dimensional accuracy of FDM printers has already been discussed in this question. You might want have a look at Ryan Carlyle's answer for a specific routine on how to calibrate a Cartesian printer.

From the question linked, a summarized approach for handling dimensional inaccuracy could be:

  1. Add any desired tolerances into the joint in your CAD model, without regarding the inaccuracy of the actual printer you will use.
  2. Tune your printer and slicer to reproduce those dimensions set by your CAD program.

Generally speaking, the printer settings should handle filament independent dimensional accuracy, while the slicer should be used to account for different characteristics in each filament type. Correcting size through your CAD model is ill-advised, since you should be able to use the same model for multiple printers.

Improve this answer
$\endgroup$
2
  • 1
    $\begingroup$ Good points. However ... there is some latent inaccuracies in most FDM printers (especially desktop-styles) that are basically impossible to rectify using slicers / printer settings without upgrading hardware or (depending on the resolution required) changing processes as @fred_dot_u mentioned. $\endgroup$
    – SRDC
    Mar 9, 2016 at 4:04
  • $\begingroup$ @SRDC, are you thinking of the extrusion width's effect on dimensional accuracy? I agree with your comment, and see how this conflicts with the ideal scenario where you shouldn't "fit the model to the printer". Still, for most people, I believe a well calibrated FDM printer gets close enough. If that is the case for the OP, however, I cannot tell. :-) $\endgroup$
    – Tormod Haugene
    Mar 9, 2016 at 6:23
4
$\begingroup$

Have you considered to use a service to have the part printed? You can print almost anything in one "pass" using selective laser sintering, typically done in nylon. I am currently un-boxing an SLS printer, so I cannot currently offer to print it, but the capabilities of SLS are substantial for parts of this nature. I've read of entire planetary gearboxes being printed, fully assembled, with appropriate clearances for operation, once removed and "dusted," or cleaned of residual powder.

The printer I hope to be assembling soon has 80 micron powder, so the resolution is quite high, although I forget at the moment the x/y/z figures.

Do you think such a method would suit your purposes?

Edit (add):

because FDM printing by nature causes "oozing" especially in the x/y directions, I would not be surprised to learn that your outer piece has a smaller inside diameter than your design has specified. You may have to make adjustments to the design to compensate. Those adjustments would be specific to a particular printer/filament combination as well. Calibration prints would make the task slightly easier.

Improve this answer
$\endgroup$
2
  • $\begingroup$ SLS could work. I need to check if I have facilities nearby which can do it. It was not a problem of oozing, since I have removed about 0.5mm on both mating surfaces. I suspect the outer shell isn't circular. You can see in the picture that the outer tube meets the inner tube at the the two ends of the semicircle and not inside its belly $\endgroup$
    – Shivanand
    Mar 8, 2016 at 16:34
  • $\begingroup$ I can now see that the ends of each "arc" are contacting, while the centers of the arcs appear to have air gaps. If force is applied, do the ends join and do the centers contact the inner cylinder? If so, I'd suspect that the outer parts are distorted from a proper cylinder. I've done cross-section prints of a 3d model in order to determine tolerances and clearance without printing the entire project to find something went wrong. If you can slice the bottom of all three pieces and print them, you may find that you have distorted semi-cylinders that have to be compensated/adjusted. $\endgroup$
    – fred_dot_u
    Mar 8, 2016 at 22:29
2
$\begingroup$

Depending on your tolerances and how many you need - and how valuable your time is - you can 3D print in ABS with a regular FDM printer, then clear out support material and do an Acetone vapor treatment or vapor wash. It will be smooth enough for most prototypes.

Improve this answer
$\endgroup$
2
  • $\begingroup$ I wouldn't recommend a vapor treatment on something that is tightly toleranced. There are too many variables in the process that can throw to part out of tolerance. $\endgroup$
    – tbm0115
    Mar 8, 2016 at 6:22
  • 1
    $\begingroup$ I would agree - perhaps I should have been a bit more specific ... depending on your tolerances. If you need machine-type tolerances, don't bother with vapor treating, as tbm0115 said. However, if all you're needing is a first pass to try and see/feel/explore how everything works together and if it's really doing what you thought it would, then vapor treating could be great - especially if you have relatively large features compared to your filament size / resolution. $\endgroup$
    – SRDC
    Mar 9, 2016 at 4:02

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .