Printing Small Gears - Wierd Tapering Issue

[peteGSX 1]

Hi all, firstly I realise I'm going to be stretching the limits of an FDM 3D printer with what I'm trying to achieve, but I reckon it's doable with some patience.

 

I have a model railway project underway requiring me to print some narrower versions of existing gears. I've used Fusion 360 with the GF Gear Generator add-in to model the new gears, and it seems like the dimensions will do the job nicely.

 

I'm using Cura 4.11 for slicing, and I have a Creality Ender 3 V2 printer.

 

The issue I'm having is when I attempt to print, and the top surface of the gear is a larger diameter than the bottom surface, resulting in a bevelled edge or taper if you prefer (I'm printing the gears on their side).

 

So far, I've tried the standard 0.4mm nozzle as well as 0.2mm, as well as varying layer thicknesses from 0.08mm to 0.2mm. I've changed infill from the default 20% to 100% (recommended for gears for strength), and reviewing the GCode for all of these options shows that the printed gears should be correct.

 

I'm generating supports with Cura, and have tried lifting them a couple of mm off the bed with a higher support in case the initial layers were having issues, but that's no better either.

 

Is anyone able to shed some light on why this might be so? I'm sure there's some setting magic in Cura that I'm missing.

 

I've attached the two STL files as well as various GCode files of the options I've tried.

 

Any help will be greatly appreciated!

Mid Gear v5.stl Driven Gear v5.stl Mid and Driven Gears - 0.2mm Slower Super Quality 0.2mm layers.gcode Mid and Driven Gears - 0.2mm Slower Ultra Quality with Support.gcode Mid and Driven Gears - 0.2mm Ultra Quality with Support.gcode Mid and Driven Gears - again.gcode

[peteGSX 1]

Quick update on this is that one of my team members from work has used Prusa slicer to print the mid gear on his Prusa Mini and it appears to have the same affliction. He managed to get a photo also that highlights the issue.

[GregValiant 1,342]

The teeth of the gear are hard to support because they are so small.  I think what is happening is that after the hub prints then the first layer of the teeth falls past the support interface.  Typical for any round feature the plastic then gets pulled into the middle.

There are a couple of things you can try.  (I would go with the .2 nozzle at .1 layer height)

Try printing it with the hub below the plate (set the Z of the part to -0.5) and see how it goes.  If you really need that hub then print them separately and glue them on.

A second option is to set the "Support Horizontal Expansion" to 1.0 and the "Support Roof Horizontal Expansion" to 1.0 and use a very high Support Interface Density of > 75% or so.  It looks like setting the Support Roof Pattern to concentric might help as well.

Essentially it looks like it's all about supporting the first layers of the teeth so they keep their shape even though they are .5 above the build plate.

[peteGSX 1]

Thanks very much Greg! I'd like to keep the hubs in place without needing to glue them, so will give the second option a go first before trying without the hub.

 

I'll be sure to post back, and your logic sounds more than reasonable to me!

[peteGSX 1]

Ok, tried the extra support settings with 0.2mm nozzle and 0.1mm layers. The quality is awesome! Unfortunately, the problem persists despite the obvious extra support around the base of the gears.

 

Off to work shortly but you've kicked my brain into gear here with some other potential ways to support those teeth, as well as trying without the hubs, flat on the bed. I'll keep experimenting and will report back.

[GregValiant 1,342]

Just for a test to isolate the problem - sink the gear into the build plate 0.5mm to bypass the bottom facing hub.  I'm too lazy to change nozzles so you will have to do all the heavy lifting.

[peteGSX 1]

Yep, cool, that is definitely on my experimenting list for the long weekend! I've also got one ready to go using Meshmixer supports as well which I'll try this afternoon after work.

 

I'm more than happy to do all the heavy lifting, after all this is my problem that needs solving 🙂

[peteGSX 1]

Ok, the Meshmixer supports didn't work. The issue was just as bad (if not worse), and given they're tree supports, one gear actually didn't make it through the print and got knocked off so I guess they were a bit too weak (I reduced post and tip size).

 

Anyway, I gave direct on the bed a try this evening after ditching the axle supports altogether and enlarging the centre hole so I could print them separately and glue them in.

 

Unfortunately now it seems that the bottom side of the teeth is now printing larger than the top side teeth.

 

This is with the 0.2mm nozzle, initial layer 1.12mm and layers set to 0.08mm. I set infill to 100% also. The bed was still hot when I took the photo so I haven't taken them off yet.

 

 

[GregValiant 1,342]

Nice job on the bevel gears.  Their tough to do.

 

I'll switch nozzles in a while and print one of those gears.  Maybe something will jump out at me.

I use up my wet PLA on these test prints and it may have an effect on small parts like that.  We'll see.  Maybe my mind is just stuck on the "support" thing, but it seemed to be the problem until I saw the "bevel" gears.

[peteGSX 1]

Hahaha if only my goal was to print bevel gears! 😁

 

I'll keep fiddling as well and see if anything makes an improvement. Your support comments did make sense to me, but I'm still pretty new to 3D printing and have been focused on larger items so far, so this finer printing is all new.

 

Keen to learn and improve though!

[peteGSX 1]

One more go tonight. I kept the slicer settings as above but just added a raft. Reverted to the original behaviour of the inverted bevel I had first up, so no improvement there.

[peteGSX 1]

Ok so today I grabbed a gear test print off Thingiverse, scaled it to 20% of it's original size, and no bevel/taper issues to be seen.

 

Looking at the fact that those gears are on a reasonably thick bed, I tried a brim instead of the skirt or raft, and much better! I did slow the print speed to 20mm/s as well.

 

I'm trying again now just straight on the bed with a skirt and no brim at 20mm/s, and I've also been able to make the gears a little wider which may or may not help.

[GregValiant 1,342]

They are so small that trimming is an issue.  I've lost my .2 nozzle so I can't do a fine print.  The print I did with the .4 stuck well but I also got the bevel/angle effect (it's also how I know that trimming is an issue).  Maybe there is some shrinkage going on as well but I would think that would be uniform all the way up.

A lot of science goes on in that .2 wide by .1 tall rectangle under the nozzle.  Maybe slowing down further for the first 8 or 10 layers (up into the gear teeth) would help.  It's a tiny and fairly complex model.

[peteGSX 1]

The straight on the bed print literally just finished, and while it's not as bad as it was, they're still bevelled.

 

Printing with the brim is by far the best result so far, and yeah I agree I reckon trimming is going to be an issue.

 

Sounds like a good idea on the more, slower layers. At the moment I've just set initial and print speed to 20mm/s, but I might do initial at 10 and do that for the first 10 layers and I might try a raft again instead of the brim.

 

I'm feeling a little more hopeful now though, so thanks a heap!

[GregValiant 1,342]

The gears I've printed have been larger and I didn't have any issues.  I think this whole problem is a question of the small size of the parts.  Trial and error is necessary as few people print functional detailed pieces like those.

[peteGSX 1]

Yeah ok, makes sense. All the Googling I've done for 3D printed gears is larger stuff, and I knew I'd be pushing the friendship trying to get small gears printed, but I'll keep trying and persevere.

 

The raft and slow print is a no-go, it just finished and still has the narrower base.

 

Looks like a brim is the way to go, so I'll try that again tomorrow with the slower speed initial layers and see how that goes. If it means a bit of post-processing trimming pain I'll cope. The brim has most definitely given the best results so far and the ones I did earlier today are almost useable.

[GregValiant 1,342]

There is a setting for "brim distance" from the part.  I set it to .1mm and it makes for easier removal and it seems to work just as well.  I just thought I'd throw that up here in case you started running out of things to think about.

[peteGSX 1]

Haha cheers! Scrolling through the settings in Cura makes my brain hurt, but it's always good to know another helpful one 😁

 

I'm doing the brim print now with the 10 layers at 10mm/s, concentric infill, etc. but will give brim distance a shot as well.

[peteGSX 1]

Success! Well, I believe at least good enough for what I need anyway but not perfect.

 

Using the Ultra Quality Cura profile (0.08mm layer height) and the 0.2mm nozzle I ended up making these modifications:

• Infill density 100%
• Infill pattern Concentric
• Print, initial layer, and skirt speed 20mm/s
• Minimum layer time 35 seconds
• Skirt for adhesion

I think they're good enough to use, so I'll print the axle hubs next to glue in and see how that goes.

 

The big things seemed to be a combination of the slow speed and extra cooling time. I suspected the initial layers were staying too hot, allowing the layers on top to compress them and spread them out instead of maintaining the correct shape.

 

I still have no clue why printing with supports and the axle hub in place causes the opposite issue but as long as I can print the hubs separately and glue them in it's a non-issue for me.

 

Thanks for the help!

[geert_2 558]

Yes, I could imagine that insufficient cooling causes issues. Try printing a dummy block next to the models, to move away the nozzle for some time and allow the gears to cool down. The goal is to keep the flow through the nozzle very constant, so the temperature and viscosity of the melt is also very constant. Thus do not "park" the head aside somewhere, but let it keep printing dummy blocks instead.

 

A hot nozzle that stays on top of a tiny print, will keep radiating heat and prevent the model from cooling and solidifying.

 

To make sure that the bed is at the correct height, and there are no Z-movement irregularities: if you print a test cube, does that have perfectly straight sides and corners, or do they also show some sort of beveling distortion, or elephant feet?

 

Picture: dummy blocks to improve the cones: this reduces heat deformation (insufficient cooling) but does not fully eliminate it:

 

[peteGSX 1]

Ah yeah that makes good sense!

 

I did end up getting a set of gears I'm very happy with after all that, and my final settings are what I posted above but with the minimum layer time extended to 40 seconds.

 

The other thing I ended up changing was the bed temperature, which I lowered from 60 degrees to 50 degrees.

 

This seems to have solved the problem for me and while I'm not at the point of testing the gears for use yet, they certainly seem to mesh well and appear to be the right dimensions.

 

To answer your elephant's foot query, no, I haven't seen that with any of my other prints. I revisit bed levelling often and am confident it's set correctly.

 

Thanks for the tip!

[geert_2 558]

Something else that just occured to me: if you want to print each gear in multiple parts and then glue them together, you might consider making the mating shapes square instead of round. That will prevent them from slipping. I remember that one of my HO-model trains had a slipping gear on the motor axis, when I was a kid. The gear was all metal back then, but a way too soft metal. I had to hammer the inner opening of the gear out of shape to make it clamp again with a tight fit.

 

I am still wondering if 3D-printed gears can withstand real use, since even real metal or injection moulded gears would wear out?

 

[peteGSX 1]

That's an interesting point actually. The larger idler gear will be fine as the teeth do all the work and the axle is just to spin on, but the smaller driven gears will be driving the wheels, so slippage would be a problem.

 

I've yet to see how they turn out for longevity, but the gears they're replacing are plastic, and I've seen many comments during Google searches that PLA lasts surprisingly well in low stress gear applications.

 

With what I'm doing, I'd consider it low stress, but time will tell.

[geert_2 558]

I have never printed gears, so no experience with that aspect. It is an interesting test, and I am curious about the long term results.

 

I did print quite a few clamps in PET and PLA. My PLA-clamps suffered from creep deformation, thus slow deformation when under continuous load, so they would soon lose their clamping force and start sliding around. Under higher loads, continuous or intermittent, microcracks did form. In my car all PLA items would soon deform due to heat, even in a very mild spring- or autumn sun.

 

So I would guess the gears might work for light use in a railway bus (or how do you call such things, not a regular tram, but a sort of diesel bus with manual gearbox, driving on train tracks?). But if you are going to pull heavy freight trains, or if the gears are close to the warm motor, I would doubt they would survive for very long.

 

My PET clamps would also snap suddenly when overloaded, but they are more flexible and can handle deformations better, with less permanent creep deformation. But the load they can withstand is not that strong.

 

A few pics:

 

Microcracks in PLA filament that was kept under continuous load (bending) for some time:

 

Idem, seen through a microscope:

 

Cracks and permanent deformation in carabiner hooks in PLA (left, cream), but not in PET (right, green):

 

The PET carabiners do also snap after some time, suddenly and without warning, without slow creep deformation, nor cracks. This is the cross-section of the fracture surface:

 

 

[peteGSX 1]

That's very interesting to see! I'm certainly not expecting these to be as long lasting as brass gears or anything like that, and given I'm modeling in N scale I expect the freight to be relatively light. These gears are actually for converting an N scale locomotive to run on Z scale track for a narrow gauge railway.

 

Interestingly enough I printed a second set on the weekend with my newly saved small gear profile containing all those settings I mentioned earlier, and while the large idler gear came out well, the two driven gears again showed some deformity with the side on the build plate being a slightly larger diameter.

 

I can't find what I did with my original tests using the 0.4mm nozzle, but I don't recall noticing this issue with those, so I'm thinking I might try with the larger nozzle again and see if that makes any difference.

 

As an aside, I've contemplated taking the easy path out and looking for somewhere to buy small gears, but there's nowhere locally here I can find in Australia that has them narrow enough, and the only alternative seems to be buying bulk lots of 50 to 200 from China, of which maybe 10% will be useful to me which is a bit wasteful (and shipping from China at the moment seems to be taking a couple of months!).