MakerBot Print is our newest print-prepration software, which supports native CAD files and STL assemblies,
allows you to interact with all your printers via the Cloud, and many other exciting new features.

Download Now


43.3:1 Compound Planetary Gearbox For Robotic Arm Project!

by Gear_Down_For_What Feb 11, 2017
Download All Files

Thing Apps Enabled

Please Login to Comment

Wrong number of teeth on the ring gear. Ring gear in photo has 28 teeth. I ordered this printed and received ring gears with 26 teeth. Can bearly fit 3 planetary gears. Is there a work around?

Amazed at how much work you have done and made available for the world. Well done
I am working on a robotic arm project for a disability application. Wanted to make your small robotic arm to prototype the project
Started by downloading and printing out 43.3:1 small gearbox with 4 planet gears and 4 ring gears.
Ring gears are too small in diameter, when installed have about 10mm gap in circumference, consequently gearbox does not work. Similarly the outer ring closure for testing is too small in diameter to go over the ring gear.
I thought my 3D printing might be too coarse to get gears to mesh properly and hence extend diameter across planet gears causing the problem. Re-printed at 0.1mm layer height but still a problem.
I have experimented with scaling up a ring gear inner & outer diameter by x 1.04 which reduces the circumferential gap to about 4mm but this may be distorting the gear teeth profile?
I have looked at your project in Fusion 360 but I am not experienced enough to see what might be the problem
Can you offer any suggestions?

Hi Thanks for the Good work.. and for your videos
But there is one problem to your files.. The file that has the planet Gears (3 module or 4 module) must be corrupted when you export them

i have download them many times (all of them or just only these ones ) and always the slicer program cant see them right.. it shows only 2 gears (2 & 4) and not all of them.. half )

all the other files are ok when i slice them.. no problem.

Also i try to open them in Blender and there was also a problem.. not display all of them

i think it would be better if you could upload them all the planetary Gears as units and not as a big only file

Thanks in advance

I test and other planetary boxes and still something wrong with some of the files . some have open faces .

But lucky me i find one that all are fine :) (66.46 : 1 Compound Planetary Gearbox for Robotic Arm)


The zip file with the separate planet gears made the magic ;)

the program can see them fine... no problem neither with other files of this package

thanks :)

time to print them and play with them to see how can i make use of them :)

66.46 : 1 Compound Planetary Gearbox for Robotic Arm

The gears in the "3 module planet gears" file don't fit the sun gear, don't bother printing them.

Edit: there is no issue

Others, including me, have printed the entire robot and it fit for us.

What issue are you having?

Looked at it again and I figured it out, I printed the wrong file from a different gear. There is no problem! Sorry!

Assembled and working now.

I’m glad you got it working :)

Great work.
I like to see gear-bearings in practical use.

I thought about designing conical gear-bearings with "soft sinusodial herringbone" design preventing roller slide-out (and with cycloid teeth).
I think with such a design it should be possible to reduced clearance down to zero by just tensioning the bearing axially.

I think the bad shaking seen visible in the video (maximal vibration amplitudes I guess ~5mm)
seems to come from some issues with the motor control plus amplification by mechanical resonance.
There's the reprap morgan 3D-printer. A scara robot with remotely similar design. It has parallel mechanics though.
It archives incredible accuracies I guess somewhere in the 10um range or even below.
That difference can't be just the different mechanics.
So I think if you focus on motor control improvement there should be orders of magnitude to gain in performance.

3D printer firmware (e.g. marlin) is already quite advanced.
I don't know if it could be adopted for controlling such a pick and place robot.

Regarding the super time consuming programming.
If I ever come about to making a pick and place robot (This one: http://reprap.org/wiki/RepRec_Pick_%26_Place_Robots)
I'd like to try making a library for it in a ridiculous high level language like Haskell.
Tying all the required software plumbing together sounds like a major nightmare though.

Sorry my first response was a little rushed.
I really like your ideas for the conical gear bearing, I'm wondering if it could be possible to accomplish that with a differential planetary. With a differential planetary being as complex and confusing as it is, I'm sure it would be a hard thing to model. If I understand you right, your basically talking about making into more of a tapered roller bearing style? It would be awesome to easily adjust backlash.

A big problem with my program is the amount of calculations that it has to do per second. I'm probably doing it in the worst way possible, especially because I only started learning Arduino code a year ago. Like you mentioned if you can do it in a better programming language or on faster hardware It would much more fluid movements.

I'm wondering if it could be possible to accomplish that with a differential planetary.<<

Yes, right after posting I realized that you've actually need planetary gearboxes.
That is what this "thing" is about after all.
I think I once started to ponder about this a bit but came to no conclusion.
I'm not sure if this is possible. (... parallel force path tensioning issue ... mumble mumble ...)
Conical gears are better suitable for differential gear-boxes which are
basically highly deformed planetary gearboxes.
But this geometry does not fit your robot.

If I understand you right, your basically talking about making into more of a tapered roller bearing style?<<

Exactly. Just with gear teeth.

In roller bearings some part of the force is always converted to in roller axis slide-out force.
Conventional roller bearings capture this force with cages or insets.
For insets see here: https://de.wikipedia.org/wiki/Datei:Schema_Roller_bearing,_tapered.svg
But when toothed rollers are used they can be made in a herringbone shape which can catch this load instead.
I think when gear-bearings that have sharp herringbone kinks in tooth-extrusion-direction are loaded
with forces running along the roller axes this might not be a too good idea.
The load concentrates too punctually near the kinks.
Therefore I think a soft (sine?) shaped extrusion curve should be much better.

Here's some info about conventional non-geared tapered roller bearings:
They come in X, O and thrust configuration:
X: https://commons.wikimedia.org/wiki/File:Montage_x_techno.png
O: https://commons.wikimedia.org/wiki/File:Montage_o_techno.png
thrust: http://www.deinabearing.com/Tapered-Roller-Thrust-Bearing.html
-- found no freely reusable images for this last one :(
The nomenclature refers to the lines of force not the
shape / roller orientation. (I Just figured that finally out right now. Yay!)
These kind of bearings can be found in motorcycle wheels.

It would be awesome to easily adjust backlash.<<

I think free movement backlash could literally be reduced down to zero.
Depending on the tension force the unavoidably remaining flex-backlash
could be reduced at cost of increased friction and wear. A trade-off.

About generating the needed conical gears:
The cone angles must meet certain geometry constrains
such that speeds match and there is pure rolling and no sliding.

I found that hyperbolic gears are the highest form gears could be abstracted to.
Given any two oblique angled axes and a gearing ratio a pair of matching gears can be generated.
I found some papers describing the math.
(This is made manageable with the very useful Plücker screw vectors -- https://en.wikipedia.org/wiki/Pl%C3%BCcker_coordinates)
But when modelling this in OpenSCAD IIRC I hit some roadblocks with non matching signs.
So I've not yet published this work yet.
There seems to be someone else trying something similar:
But I think this model doesn't draw from the theory of the papers I've read.

Turns out only cycloid gears are easily and straightforwardly generalizable to the hyperbolic case.
Evolvent teeth are an open research problem.

That's not a problem tough.
Cycloid gears are usually avoided since they
A) are hard to manufacture with conventional means.
But we do 3D-Printing for which soft curves are ideal. And
B) because they need zero lash (they have non adjustable inter-axis-distance).
But zero lash is what's targeted here anyway. And
C) because their varying attack angle can cause vibrations when high forces are transmitted.
But these are no steel gears so really high forces are not present.

Regarding the micro-controller programming:

I think I can give a bit of history of the 3D printer firmware marlin:
The developers really worked miracles there.
I have an Ultimaker Original 3D-printer from the very first generation.

The first firmware (I think it was sprinter back then) sounded like a machine gun.
The more "corners" a circle had (the better triangulated) the worse the issue.
As I understand this was an issue with missing look-ahead.
At every line segment endpoint there was a full stop.

With one really groundbreaking firmware update (look-ahead) this was solved.
It works somehow like that:
If the next (few) line segments have almost the same direction as previous one => keep speed.
If there's a sharp kink a few segments ahead => slowly reduce speed in advance.

The processor of the UM Original (early days) is not the strongest one:
(goes with this: http://reprap.org/wiki/Ultimaker%27s_v1.5.3_PCB)

Also there where problems printing directly from via USB cable connection.
since there is no dedicated USB decoding hardware this is done in software.
Excessive load frequently caused fatal hiccups.
Printing from SD solved that 100%.

I like your ideas, if you can figure out how to model them. haha. <<

I realize what I've gotten myself into there.
So I've identified and split of a more manageable sub-problem that would be useful standing on its own.
This one: http://reprap.org/wiki/ReChain_Frame_System
I already have some 3D-models revolving around that one.
But nothing fits together yet. (I'm still way below the percolation limit where the pieces star to fall in place naturally)
One aspect are friction-avoiding tensioning mechanisms.
I literally now have a dozen detailed out ways how they do not work.
I really need to publish all that stuff in a coherent way.

by dlajoie

I'm sorry I didn't see your latest post, but better late than never right?

on the roller bearings, it there any reason that the rotational axis of both sides cannot be parallel? otherwise the gear needs to be flexible haha

I like your ideas, if you can figure out how to model them. haha.

And yes the robot does have a resonance issue that I haven't managed to solve yet

how much to print me the Planet_Gears i have tried like 3 time get half and get move off by head

hello the best king about planetary gear i fiind this
perhaps you can design and mixt exactly the same with your planetary gears
that is soo nice to have that

When using One_Piece_Sun this replace Sun Gear 1,2 and 3. is this correct and do you forfeit torque.I don't see Sun Gear 2 in the one piece

Yes the one piece sun replaces Sun gears 1,2 & 3.

No the one piece Sun does not cause any additional friction, it actually helps keep the planet gears straight and probably reduces friction.

The One Piece sun does not contact the Planets in Module #2 because there isn't enough room to have a gear there, and i don't believe its necessary to have a gear there because the planets are supported from both ends, though it would be better to have a gear there, if there was room.

if your talking about the independent #2 Sun missing from the "thing" files, I re uploaded it yesterday, I don't know why it disappeared.

Thank you, Using the one piece sun do you get less torque.

I dont understand the question. Less output Torque? probably not.

Have you test the max load it can lift with Nema 17

No, not very scientifically.

I am trying to use the 4 module version, but Sun Gear 1 and Sun Gear 3 are both too big to fit in the 4th slot on the Optional 4 Module Planet Gear. It seems like slot 1 and slot 3 on the 4 module planet gears are narrower than slots 2 and 4?
Also, the instructions reference a Sun Gear 2 that should be used in slot 4, yet the files list does not include anything by that name.

Sun gear #2 is uploaded. I figured out how to do it from my phone :)

I'm really sorry about this. I will upload the #2 sun gear as soon as I get home, and reply to this so you know it's done. I don't know where it went, sorry!
I'm surprised that nobody else mentioned this.

I finally had a chance to print out the #2 Sun Gear, works perfectly, thank you!

Glad you like it! Haha I'm also going to release a one piece sun gear with slightly more clearance tonight too!

Great news on the sun gear with more clearance! My biggest issue has been the tight tolerances, It takes a lot of force to get the EZ Print ring holders to fit around the ring gears once all the gears are inserted. Managed to break an entire set of planet gears and a one piece sun gear by using too much force (hammer).... I would rather have just a little bit more slop in the system if it meant an easier time swapping parts around.

Great job, I'm also building a robot using the InMoov design, I'm not a mechanical engineer. Can you remix to be 50:1 for me leg joints

Are you asking for the exact same thing accept to change the gear ratio to 50:1?

I'm no engineer myself, to be honest. If it is something that lots of people request i could spend some time figuring out how to backwards calculate the gear ratio but for now I'm just going by "well that outta be a large enough gear ratio"

I'm only really able to guess what the gear ratio outcome is on these gearboxes at this point.

I found this 50:1 http://www.thingiverse.com/thing:2071318. which is also your.

Compound Planetary Gearbox (38.75:1 or 50.4) No Bearings, No Hardware!