P.Harmonic Gear Reducer

by LoboCNC Jun 7, 2018
Download All Files

Thing Apps Enabled

Please Login to Comment

Such a clever design! Well done!

I printed the flexible part in PP, polypropylene. It should in theory be a very good plastic for this application, it does not fatigue when bent. It should last forever.
I have had it running as fast as my stepper can go for 30 minutes now, still going strong.

Post a "make" if you get a chance!

Really clever and smart design!

Well, yes of course.
I expect printing precision will continue to improve.

You seem to have the skills to do a good job with what we have.
As you've noted your PLA surfaces are not wearing noticeably.
So I guess youve got your pressure angles pretty good
despite the current equipment limitations.

I only really bring it up to check my understanding.
I'm learning from a standing start as we speak.
I know very little except for what youve told me
or pointed me to.

I have two payload targets in mind.
At least one of them may be impossible.
I don't know.
The larger one would be in the 15-20kg range moving quite slowly.
The smaller one would be in the 3-7kg range, also moving slowly.

I should probably quantify slowly somewhat.
The slowest speed would be one revolution per day!
The fastest maybe 6RPM.
This could be achieved with the addition of a pulley and belt gearing
Before or after the harmonic drive.
Before would seem better to me.

And I'll be using a stepper motor with a silent stepper stick to achieve microstep control.
I only have PLA but the ABS I ordered should arrive shortly and I've found a local supplier of inexpensive PETG.
if I can work it out I want to mash this design up with Emmet's.
Herringbone sun and planets with helical fixed and flexible rings.
Should be a fun learning experience, whatever the outcome.
OK, that's all until I can start my own remix.

It took me awhile to bend my head around how this works. Finally internalised the visualisation. The planet gears serve double duty; they are the rotor and they are the teeth of the fixed wheel meshing with the flexible wheel. They ride in the actual fixed wheel, causing the rotation and always in synch so that they are effectively the same as the fixed wheel but reversing the direction the teeth face. This allows the flexible teeth to face inwards and gives us this clever design.

My only question is this: Won't using a planetary gear introduce backlash? One of the desirable qualities of a steainwave gear is zero backlash. It seems as though this would be lost.

You are right that the planetary section can induce some backlash. The only backlash, however, occurs between the planets and the fixed ring because all of the other teeth mesh points are pre-loaded by the flexing ring gear. In fact, you could also take out that backlash by having the sun be slightly barrel shaped, causing the planets to be pressed into the sun at the top and pressed outwards into the fixed ring at the bottom. Of course, all of this is in the noise when dealing with plastic gears that can simple be over-meshed slightly to eliminate backlash.

Not quite following the barrel shaped sun gear idea.

Won't over meshing lead to increased wear on plastic gears?

Especially under load. At which point backlash would return.

I'm not looking for a nice educational demo of a principle.

I want a functional gearbox that works under load and is durable.

I'm not expecting to do that in PLA.

I'd have to work up a diagram of the barrel shaped sun to explain it better. Any way you look at it though, any zero-backlash gearing (that doesn't involve some sort of parallel spring-loaded bias gearing) ends up with some over-meshing and increased wear. With metal gears which are very stiff, even tiny amounts of over meshing will create a ton of friction and wear. With acetal or nylon gears, though, a tiny amount of over meshing does not create huge contact forces, but what you gain in reduced backlash you lose in decreased drive train stiffness.

Decreased stiffness? Because of the barrel you mean? Reducing contact area.

Is that true of strain wave drives? One of their highlighted and desirable features is zero backlash. Another is high ratios and torque in a compact form factor. Ultimately I'm aiming at something in the region of 200-1. Maybe not on my first print ;-)

With plastic gears, the stiffness is less just because the plastic teeth deform more under load. And yes, strain wave/harmonic drive are, in fact over meshed, but the over meshing mechanism is that when the flexing cup deforms into an ellipse, the gear teeth on the end are no longer exactly parallel. They are canted outward slightly and forced into the teeth on the fixed ring by the springiness of the cup. There's another type of pancake harmonic drive that doesn't have a cup, just a flexible ring gear, but this type I believe is low-backlash but not completely zero backlash.

Thank you. I'll google the pancake drive. So if you were me which direction would you explore next? I agree that nylon and pom seem the most durable materials, although both are problematic to print and nylon is surprisingly hygroscopic. I'll probably cut my teeth (sorry) on abs and PETG before attempting nylon or pom. I will probably need them for anything practically durable under load. They are used for gears already so I have some hope I am not pursuing the impossible. Hopefully the abs and PETG will give me an easier learning curve and be somewhat durable to play with.

MechEngineerMike has a great overview of 3d printed gears:
He rates PLA over ABS and PETG for printing gears. I've been printing some compound planetary gearheads with PLA and found them to be almost irritatingly durable. (I keep hoping they will wear themselves in to run more smoothly, but there is surprisingly little wear.) If you need high ratio zero backlash gearing, I think you'll have to do some sort of harmonic gearing that incorporates a flexible element to pre-load the teeth to take out backlash. There's no way to FDM print conventional gears accurate enough to run smoothly without backlash. I'm actually looking into getting some gears SLS printed in nylon to see if I can get close to zero backlash.

Thanks for the link.

Are you running them under load to wear them in?

Laser sintering sounds exotic. Is it expensive?

Be interested in how that works out for you.

I've run the gears under load with lubrication and (mostly) unloaded at high speed without lubrication. (I'm a little afraid to run under load without lubrication for fear they'll grind themselves to bits. ) So far, the original surface and visible layer lines are very much intact. Regarding SLS, you can get online quotes from places like Shapeways. Definitely more expensive than printing yourself, but it you are working on an engineering prototype, way cheaper than any other options.

Interesting. What kind of load were you using?

It may seem a bit crazy but have you thought of using a polishing or even a fine grinding compound in the moving gears?
If you keep an eye on it, to avoid overshooting, it may give you the finish you're wanting.

I've been testing a compound planetary gearheads in a robot arm (which I may publish someday). In reality, having gears wear in against each other will improve the smoothness, but also increase the backlash, which is my bigger concern. (I ended up solving the smoothness problem by using helical gears instead.) The smallest gears I'm trying to print are 40 pitch, and realistically I don't think there's any way I can get the accuracy needed for near zero backlash with FDM.

Ah yes, I saw the herringbone design that Emmet did. Why does that work better?

Can you not get the backlash down enough by overmeshing or spring loading?

I read MechEMikes interesting info. His opinions on materials seem partly theoretical.
He apears to dismisses PETG based on it's qualities without actually trying it out.
One of his commenters reports positive results and it seems to work well enough in your design.

Helical or herringbone gear teeth have each tooth gradually engage and disengage rather than the entire tooth face hitting the opposing tooth all at once. It averages out any irregularities in the tooth profiles and is much smoother. Overmeshing can work of your tooth profiles are accurate enough, otherwise, you are just grinding the irregularities against each other. Spring loading can be a little cumbersone to incorporate, unless it comes for free like with a harmonic drive.

Regarding PETG gears, I'm guessing wear resistance might be his main issue. You should message him and see what he says.

It might. I will. Re. Wear. Doesn't a pressure angle cause the teeth to roll around each other without really rubbing?
Hence your persistent surface on the PLA gear.

With perfectly formed, perfectly mounted, and perfectly rigid gears, you get pure rolling contact and virtually no wear.. Reality, especially with limited 3D printing accuracy, is a little different.

SLA printed from universal material - failed after few minutes, but with decent lubrication worked flawlessly: https://www.youtube.com/watch?v=DJOF9DoujVE

That looks great! I don't know much about the different SLA materials available, but I wonder if there isn't something a little tougher that might make it viable.

Actually there is material that mimics PE, should be just right for this purpose.

Do you know what that's called?

What you have here is quite novel! Well done!

While this may break as a harmonic gear reducer in PLA, some of the concepts here would translate to a rotary series elastic actuator that would probably have a longer life and multiple applications.

Thanks for sharing this!

You are welcome! I did reprint the flexi-gear in PETG and so for it seems to be holding up nicely.

Did U print everything using PLA? even this flexible part? I really liked this model. Congrats for the excellent job.

Initially, everything was printed in PLA. But as you might expect, after 10 min. of high speed operation, the PLA on the flexible gear cracked. I then reprinted it in PETG, and it is still going strong.

Oh. I suspected haha. Thanks.

Excellent work, as always.

I would think this would be great for a panoramic time-lapse!

Just a comment to your description statement. with 0.4mm nozzle you can also try to print with 0.2mm line width. Especially if you will be running multiple perimeters. I've been experimenting with that for some time - and it gives really good results. All you have to do is cheat by providing 0.2mm into setup of your slicer. And it works. You can get really good results in terms of the vertical and horizontal resolution with such settings.

Thanks for the tip! I didn't think you could get away with a line width only 50% of the nozzle diameter, but I'll have to give it a try.

Very clever. I designed a rigid harmonic drive but you are limited in ratios. The flexible gear is genius.

The flex gear implementation is really original and very cleverly done, genious! Thanks for sharing

Thanks for the kind comments!