Tips/Techniques Printed worm drive rotary table

[Arbutus]

The job was to build a modular rotary mount as part of a synchronized camera positioning system. This stage is a horizontal base mount for large (2Kg +) cameras costing more than my car, so it has to be sturdy and do the job reliably.

Previous iterations were made from CNC milled 6061with Boston Gear components which is an expensive option, so I wondered if the durability and long term performance of 3D printed parts could be an option. Quite frankly, until this latest generation of printers and engineering filaments came along, the quality of a 3D printed gear was so awful I wouldn't have tried this.

Now I can consistently print plastic gears with a reasonable surface finish, within tolerance. So this is a bit of an experiment. Can I replace the expensive gears and casing with a 3D printed equivalent? What are the required tolerances for a weathertight assembly? Are these plastics up to the job?

This stage uses PETG for the casing, since it is durable, tough, weather resistant and dimensionally stable. The worm is a module 1 thread with a trapezoidal tooth form. It is made from PA-GF, a glass reinforced high temperature nylon, chosen because it is tough, wear resistant and self-lubricating. The worm wheel is 60T M1, with the profile relieved to mate fully with the worm. I used Nylon10 for this since it has a smooth, wear resistant finish.

The mechanics include a thrust bearing and a ball bearing for the input shaft, and two large bearings to support the worm wheel in the casing.

A close-up of the gear teeth shows no visible damage after running 12 hours at 650 RPM. The lubricant is vaseline, since many oils and greases will destroy plastic.

So far, I am impressed. I think the PA-GF is too rough and abrasive for this application, but the Nylon10 looks like the better option for gears.

All the parts printed within 0.1mm of the designed dimensions.

I'll run this again, next time with the gearbox under load.

 

[David]

That is impressive!

 

[Susquatch]

Ya, I'm impressed! Holy Crap that's cool!

 

[PeterT]

Nice. How are you retaining the screws & studs & such within 3DP?

 

[Arbutus]

M3 embedded inserts from the back side. They are 5.4mm OD and melt perfectly into a 5.2mm hole.

 

[PeterT]

Good, that's why I was asking. The M3 inserts I bought suggest 5.0mm dia x 4.0mm depth counterbore. Pretty close. They have melted in nicely too & so far doing the job. I saw some other insert variety packs on Amazon I was tempted. They also had different lengths for same thread (shorties & longies). I wonder if the knurl/counterbores are slightly different between them

https://www.amazon.ca/dp/B0D475LRH3?ref=ppx_yo2ov_dt_b_fed_asin_title&th=1

 

[Susquatch]

They also had different lengths for same thread (shorties & longies).

Not sure it matters, or even that it's relevant, but it might interest you to know (if you didn't already) that a thread length of any more than 1.5 times the diameter is just wasted. The mating threads don't take up any tension after that.

 

[PeterT]

Yup, thanks I was aware of that. And any typical steel fastener is going to pull the insert out of the plastic like like a carrot from loam long before the brass threads give way. (Edit well not that bad if insert can be located on the other side of object like example). The only reason I can think of is they (the plastic industry where these probably originate) might desire the added length for manufacturing reasons. Maybe tooling related or better insertion alignment. Or longer surface/knurl gives them a bit more holding power in injection molded plastic? I tried pressing in an insert with CA in the hole (heatless). It needs a different diameter & even so I think its crushing the counterbore wall a bit & weakening?

Interesting side note. I buzzed a 3DP part through my scroll saw. It was like a magic trick. Blade cuts through, shows a cut seam, but the 2 parts are very much stuck together. I've seen this movie before. The blade is providing enough heat via friction to melt the plastic right after it made the kerf & sew itself back together again just as fast.

 

[djberta]

Very nice. Functional print.

 

[Arbutus]

And after another long run, there is some noticeable wear and the gearing now spins more freely.
The input shaft backlash is now about 2 degrees, whereas before it was friction locked. The output shaft backlash (back driven free play) is extremely small and probably better than my mill's RT.

Some more images after about 18 hours total run time. The last run was at 1300 RPM.

Just to put things in perspective, this camera stage might rotate 180 degrees in an hour when its working on set, so spinning at 1300 is like a century of use!

The wear is on one side of the worm wheel teeth, which are now very smooth. The worm is also visibly smoother over a small length. There was initially some contact at the tip and root of the teeth which wore away within the first few minutes and was probably the source of the fine plastic debris.

I have also tried Nylon for stepper motor mounts and some other machine parts. Printing large pieces is difficult due to warping. I cheat with that because I use a chamber heater on the P1S so all my ABS, Nylon and other high temperature filaments behave properly. YMMV.

Nylon is horribly tenacious stuff. It glues itself together very easily. Great for inter-laminar strength, but really bad if the part needs support and you don't have an AMS.

The parts are flexible and energy absorbing, almost indestructible. Not at all stiff or brittle like PLA or PETG. I really had a hard time breaking apart a nylon bracket with a hammer.

I printed the holes in the worm wheel at 2.8mm. An M3 screw happily self-taps into the nylon (thread forming, not thread cutting) and it is very secure. The plastic is very soft though, so overtightening these small screws can easily strip the hole.

Unlike metal gearing after running in, the lube is clean.

Total cost for the filament - around $3.00. Print time 1:35 at 0.15 layer height, 0.4mm nozzle.

IIRC... Boston Gear equivalent parts: SS Worm+shaft $78 US Acetyl worm wheel $38 Bronze worm wheel $93.

Jury is still out, but this is looking good.

 

[Dan Dubeau]

Wow, that's pretty impressive. I wouldn't have thought printed parts would survive very long in this application, but TBH my printing knowledge is a few years behind what's current.

 

[1018Machine]

Cool stuff, you mention using Vaseline as a lubricant but being a petroleum product would it not start to deteriorate the plastic too? Would silicone grease be better?

 

[PeterT]

@Arbutus did you generate the gear profiles using your own CAD or found a (suitably accurate) solid model?

 

[djberta]

If you need more strength there is the carbon fiber filaments. They have been out for some time now and I think my last roll of carbon fiber pla was 20 bucks.

 

[Arbutus]

Cool stuff, you mention using Vaseline as a lubricant but being a petroleum product would it not start to deteriorate the plastic too? Would silicone grease be better?

Could be. In previous tests one of the concerns I had was oil creeping between laminations over time. Exposure to oil and stress is not a good environment for most 3DP parts. That doesnt seem to happen with vaseline. I'll try other greases on the next round of tests.

The Nylon10 and PA-GF i'm using have almost perfect interlaminar bonding, which may be why it's not a problem in this case.

 

[Arbutus]

@Arbutus did you generate the gear profiles using your own CAD or found a (suitably accurate) solid model?

I do both. There are some very good gear generator add-ins for Fusion which can generate tooth profiles correctly, or sometimes I will design from scratch - for example creating a cycloidal drive epitrochoid tooth profile.

McMaster-Carr, KHK Metric, SPI and many other retailers have good online models. If you have a known, manufactured gear - try this: find it's mate from an online catalog - download the .STEP file and print it with your favourite filament and see how well it fits. A little blue on the surfaces helps you see where the contact points are. Then tune the printer to smooth things out as much as possible. 0.12 to 0.18mm layer heights and a slow print speed seem to do the trick eventually.

Tolerances are the gotcha. 3DP plastic gears are an order of magnitude less precise than metal gears and this needs to be taken into account when designing. In a metal gear train I aim for a 0.001" clearance. i.e. if the theoretical centre-centre distance is 1.000, I would increase that to 1.001. For a plastic gear train that would probably be (in this example) 1.005 for a smooth running system. Larger gaps cause unnecessary backlash and noise.

 

[Arbutus]

If you need more strength there is the carbon fiber filaments. They have been out for some time now and I think my last roll of carbon fiber pla was 20 bucks.

In this job, I used a glass reinforced nylon, but it is quite rough and abrasive. The pure Nylon10 seemed to work much better for the worm wheel. There's also Kevlar reinforced ABS in my stash so that is another filament to play with. The Nylon-GF is $75 CAD for 1kg - so three times the cost of basic filaments. Not to be wasted as a support material!

 

[Mcgyver]

great stuff! Thanks for posting it

 

[djberta]

In this job, I used a glass reinforced nylon, but it is quite rough and abrasive. The pure Nylon10 seemed to work much better for the worm wheel. There's also Kevlar reinforced ABS in my stash so that is another filament to play with. The Nylon-GF is $75 CAD for 1kg - so three times the cost of basic filaments. Not to be wasted as a support material!

Good to know. I did not think about it being abrasive.

 

[jcdammeyer]

A different option for a rotary table drive is the harmonic drive concept.

Here's what JoeMac said when he sent this to me:

I found enough stuff to give it a run. Runs and sound perfect after all these years. No crunching or growling. The PLA seems to have stood up but disassembly and inspection will have to wait a bit. Maybe I’ll try it on the FluidNC setup if I ever get around to getting it up and running.

We both build similar versions.
Mine along with the bigger version of a different style.

His with the clear plate shows what is happening.