DIY-hypocycloid gears

Hi all,

this is a very interesting high reduction gear for high torque applications what can be made on a small CNC-mill out from a sheet of POM.

Viktor

looks like it would be perfect for the stepper extruder, especially if one can be made to directly mount on the front of a nema17 motor. I'll have a go when my machine is printing if no-one beats me to it

---

-----------------------------------------------
Wooden Mendel
Teacup Firmware

I made an animation of a differential hypocycloid gearbox with a 100:1 reduction, designed to fit on the face of a nema17 stepper.

I've never seen these before, so the animation really helped me understand what was going on.

sure looks more like an epicycloid than a hypocycloid to me, and note that the python script in first post's link expects some values to be radius despite saying that they're all diameter.

I'll export the various pieces as STLs in the next few days in case someone wants to print one and tell me how it turns out

---

-----------------------------------------------
Wooden Mendel
Teacup Firmware

... excellent

I'm waiting too for feedback with fabbed surfaces ... with my CNC-mill and POM i can make really perfect surfaces with very low friction, but how is this with reprapped surfaces?

---

Viktor
--------
Aufruf zum Projekt "Müll-freie Meere" - [reprap.org] -- Deutsche Facebook-Gruppe - [www.facebook.com]

Call for the project "garbage-free seas" - [reprap.org]

my design incorporates a 608 bearing between the eccentric coupler and the hypocycloids, considering adding a bearing on the output shaft too. purple in the video is the bearing- 8mm ID, 22mm OD.

---

-----------------------------------------------
Wooden Mendel
Teacup Firmware

... the inner bearings are essential - i'm asking for the friction between teeth and columns and for the transmission to the out-axis ...

---

Viktor
--------
Aufruf zum Projekt "Müll-freie Meere" - [reprap.org] -- Deutsche Facebook-Gruppe - [www.facebook.com]

Call for the project "garbage-free seas" - [reprap.org]

Since there's so much rolling motion compared to sliding in those areas, I reckon squirting some grease in might be enough.. I thought there would be far more sliding than the animation shows.

For high load applications, using roller pins instead of just a shaped shell would indeed be important.

---

-----------------------------------------------
Wooden Mendel
Teacup Firmware

Well HDPE does not have much friction and gear teeth come out pretty smooth in the tangential direction when made by FFF.

---

[www.hydraraptor.blogspot.com]

... i'm planning to use this as DIY-replacement for harmonic drives with extreme high angular resolution, so i have to eliminate all backslash and similar ...

---

Viktor
--------
Aufruf zum Projekt "Müll-freie Meere" - [reprap.org] -- Deutsche Facebook-Gruppe - [www.facebook.com]

Call for the project "garbage-free seas" - [reprap.org]

Looking at the animation, it would appear that this design would have significant backlash unless you get very high tolerance on the gearing / pins. Is this my own misconception?

Interestingly, the topology is reminiscent of the the split annulus epicyclic gear, where the top hypocycloid gears maps to the top half of annulus and planetary gear, and the bottom hypocycloid gears map to the bottom half of annulus and planetary gear. The minor difference is that the hypocycloid gear transmits power by an off center rolling, whereas the split annulus epicyclic gear uses the sun gear to transmit power to one half of the planetary gear.

In terms of printability, it would seem the hypocycloid has the advantage of much courser 'gears', as the planetary requires conventional gear teeth.

BeagleFury, the hypocycloid design can be made to have no backlash at all, as all the teeth can be in contact with all the rollers simultaneously. This is its advantage over the split annulus epicyclic. However, this probably requires roller pins on bearings rather than a printed shell, so for reprap we must print it with a tiny amount of clearance.

How much backlash will this give us? I have no idea, print it and tell me. Maybe enough for a strip of teflon tape? Maybe we just need enough for some grease to slip in...

---

-----------------------------------------------
Wooden Mendel
Teacup Firmware

Triffid_Hunter Wrote:
-------------------------------------------------------
> BeagleFury, the hypocycloid design can be made to
> have no backlash at all, as all the teeth can be
> in contact with all the rollers simultaneously.
> This is its advantage over the split annulus
> epicyclic.

I believe the split annulus with 5 planetary gears has 5 points of contact at all times (every 72 degrees) -- even better if one adjusts the gears to use an angle (not herringbone because then you couldn't slide the planetary gears/sun gears into the assembly, but slight angle at least one gear pitch between top and base should allow you to twist the gears into place.)

However, epicycle too will require a fair amount of tolerance to avoid backlash; the advantage there would be the larger number of gearteeth that would limit the maximum amount of backlash at the output end.

I certainly intend to try out both designs once I get a printer working and the few parts needed to make it a better printer, as one of these gearboxes would certainly count as making a better printer -- one can gear down the extruder stepper motor to get better control and accuracy on extruded filament. If someone beats me to it, it would be nice so I can know which design to try first (the better one, presumably )

haven't had a chance to play with this lately, so I'm attaching my blender model for you all to play with

edit: file is too big to attach (13M), put it ifferential_hypocycloid.blend" target="_blank"  rel="nofollow">on the wiki: Differential_hypocycloid.blend

edit: had to wrap in a code block to stop stupid emoticon code from mangling the url
edit: didn't work, trying html character code

Edited 4 time(s). Last edit at 04/04/2010 11:06PM by Triffid_Hunter.

---

-----------------------------------------------
Wooden Mendel
Teacup Firmware

I thing there are some problems with this design.
This design is IMHO not dual-stage, but two separate stages with differential output. First stage has 1:10 ratio, second has 1:9 ratio. Both stages are driven from input shaft. The output is derived as difference between theese two stages.
Assuming a as first stage ration, b second stage, I input speed, O output :
a*I=b*(I-O)-O (second stage sees it's cage rotation substracted from input)
This could be simplified to:
O=I*(b-a)/(1+b)

It's consistent fro 10 + 9 lobes (1:100), 9+10 lobes (-1:99, consistent with simulation). I'll try to check this with simulation.

Someone (http://www.youtube.com/watch?v=SH46bpe1cNA&feature=mfu_in_order&list=UL comment) tried 1:15+1:10, the resulting ratio would be 1:33, he mentions 1:32 in the comment.

There may be major problem with efficiency, but this is only my guess: Both stages are loaded by output differential mechanism with torque similar to (or half of?) output torque. So the loses will be comparable with single-stage design, but the output speed will be only (aproximately) 1/10 of single stage - so the power efficiency will be only 1/10 of single stage (or the loses 10x higher? Not sure here ... )

Quick solidworks silulation (10+9 lobes) gave ~30% effiency when considering greased steel-stell contact (0.08 friction), about 10% with dry steel ( maybe .4 cof, not sure). But I have very little confidence in theese results.

(edited after some research)

Edited 1 time(s). Last edit at 02/03/2011 09:38AM by ledvinap.