Worm Gear revisited

Some might remember I did a Worm Gear Setup a while back and it “kinda” worked. Well it failed due to it jamming and destroying bearings and stuff. I overlooked the fact that if you use a worm gear the way I was and hang 150lb from it - well the 150 lb transfer from the Worm gear to the worm and even though the rotational torque needed to turn it is little - according to ratio and friction but the Worm gear still works as a lever and the Worm holds and transfers the torque- caused by the 150 lb (in that case about 1/2 of it due to the way levers work) to the case and the only thing holding it was a 688 bearing which died under the thrust of 75lb vertical to its rotation plane

So a thing that should be able to be fixed with a thrust bearing protecting the 688 in that direction.

So Ali Express delivered some thrust bearings. In the meanwhile My filament tester prototype died - as I pasted previously - due to stripping out 5/16 bolts and nuts I used to break the plastic test pieces. Before it broke it showed a 50 mm2 cross section PETG piece broke at about 230lb average and I don’t want to go lighter on the test pieces as 5x10 or 7x7 are dimensions often used by themselves or in multiples in my designs so I want to test close to real world use conditions. And I want to do stronger stuff too so I am shooting for 450-500 lb (or 3 robots or a whole alliance buddy climb - lol)

The core will be a 40 tooth Worm gear (yeah I could go other ways like planetary etc too) but I did planetaries successfully and I Worm gear is still waiting for success. So I just printed the Worm Gear with 2 HTD 5 30 tooth pulleys attached


The reason for 2 pulleys are 2 fold. First to keep it centered and 2nd HTD-5 15mm Belts according to Gates can be trusted up to 275 lb
The Gear has 40 teeth with a 2.8mm Modulus. I arrived at that from Hints of previous tests. Based on the area of this print I would expect a tooth to pull at a load of somewhere between 120-150lb. At 40 teeth at least 4 teeth are engaged to some point between the worm and the gear and I am using 2 worms also the bigger the gear the bigger the lever in comparison to the .94 in lever of the 30 tooth pulley reducing the load on the thrust bearings. The pulley has a 1/2 in square axle hole to allow a steel 1/2in square stock or tubing to be inserted the axle will have some adapters to 6805 bearings (4) Which should easily hold the load as they are statically rated at about 650 lbf and 4 bearings as I intend to extend the axle to connect external loads like the rack and pinion prototype I posted about earlier.

For the last 2 days I have been working on the bottom half of the case
W450lb_Case (Master).pdf (1.6 MB)
(3d Pdf viewable in Adobe Acrobat)
Its about 90 % done. RN I am working on the input gears . A 19 tooth Mounted to the CIM Motor, A 2nd 19 Tooth as a reversing gear and 2 59 tooth gears that will hold the worms. So overall a (59 * 40 * 1.06)/19 = 131 : 1 reduction. The Cim will be controlled by an Arduino and a BTS7960 so limited to <40ish amps at 12 Volts which translates to about 5 in/lb at the shaft max or theoretically about 650 lb lifting at the output. But I expect a loss of up to 30 % along the way especially due to friction at the Worm/gear area - even with lubrication. Hence this ratio to get close to the target 450 lb.

The max speed at max load will be probably 30 rpm translating into something a little shore of 3 in/sec. Which is good for testing tensile strength and deflection and such. If one would want to go faster - well add some motors and drop the reduction.

If it is at least somewhat successful - I will make a folder on grabcad for it and share the files. Either way I will publish the results.

So its time to comment below to put down what part will fail or not and at what weight load it will fail. The test intended is to hang my scale from a static place and connect the other end to the 2 belts and then give the motor as many Volts and Amps necessary to get the weight desired up to a max of about 36 Amps and then hold it for a little while.

Things still to do: Design and print rest of gears, 2nd and 3rd part of the enclosure, mount steel and aluminum tubing, connect arduino to BTS 7960, write code, and most likely add a touch screen to control the whole thing.

4 Likes

I keep hoping someone in FRC will come up with an easy to integrate worm gear with enough ratio to actually lock… Here’s hoping you are getting closer to it!

What are you planning for the worm itself? Threaded rod of some sort, or printed?
If you are printing you should look at the geometry of a double enveloping worm/wheel. That will radically push up your tooth engagement at the cost of MUCH more complicated geometry.
There is going to be a significant separating force between the worm and wheel, so don’t forget that when you are planning your bearings. The axial load on the worm is likely to be much higher than the separating force…

Yeah I am going to 3DP worm / Herringbone assembly. And I will use 2 one on each side


This is from the prior version - this one is from a 19:1 worm this one will be a 40:1 and I think the one in this pic was a 1.2 mm modulus herring bone gear. I had to up it to 1.5 to accomodate the bigger Worm gear plus it saves me a nozzle change as I can print a 1.5 modulus gear with a .8 nozzle

I went into a Worm Gear assembly to get one that does not backdrive or not enough to move a motor that is not under power. All my prior designs - production or prototype - were of the backdriving kind (regular gears, planetary and cycloids)

1 Like

If your CAD will do it, making the worm hour glass shaped will increase tooth contact. You have a single enveloping worm wheel already, so you are already doing quite well!

Inventor doesnt do that to my knowledge

Globoid (double enveloping) worms are a pain to model, and I’m not even sure this one is accurate. :sweat_smile:

This was done as a modeling self-challenge based on the industrial gearboxes we used at my former job.

5 Likes

Looks really neat - maybe for version 2. 0 its definitely printable

The worm wheel teeth look too shallow…
It looks like you are mapping the regular straight worm threads onto curve…
image
Patent drawings!
https://patents.google.com/patent/US2935886
Fancy picture with cross section:
image

I just might try that one day

They meet at the central plane in a standard acme worm profile with a complete depth.

It is somewhat in the nature of the globoid worm for the teeth on the gear to be ‘football shaped’ as the leading and lagging parts of the worm pass by, the sweep of the worm teeth in either direction ends up leaving a washout, or clearance in their wake.

This image is representative of what I am used to seeing in the industry, if I made the hub on my gear narrower it would look a lot closer to this example. IIRC I was adding extra material to have a bearing race baked into the outer lip on each side of the gear to later be designed into a housing.

If you have Solidworks the model can be downloaded here.

Otherwise, apologies for co-opting this thread.

4 Likes

Wow! Sweet model and some primo brass!
It’s a worn gear, so its on topic, right?

1 Like

Its on topic as far as I am concerned. Anything cool and educational is always welcome

22 hours later the first part of the housing is printed


The Plastic mainly functions to mount the other parts to. All is to be reinforced with 1/2 in (in this case steel) tubing The plastic positions it and makes sure the holes are easy to drill where they go and also to enclose it to keep the grease from flying all over the place. There are a few more enclosure pieces coming. But first some gears

4 Likes

Small progress today


The tester/gearbox got some “feet”. As you can see like in most of my designs the mechanical burden is carried by the aluminum or this case Steel tubing. Its a poor mans way to built strong and accurate and keep 3DP which is slow to a minimum. So 3DP is used to accurately position cheep steel/aluminum like thos tubes that I pick up locally between $.5 to $1 a foot depening on if its steel or aluminum - slightly more if its solid

First 2 gears printed too. HIPS which usually works well in this case and both HIPS and Nylon will hold the CIM shaft well enough with the key to stall the motor and not let loose and we are going to run this way below stall torque as we got to keep it under 40A

Also the “ugliness” above the gear was to allow me to mount the motor I neglected proper access for a screw driver to mount the motor so I called my Dremel for help. Next iteration will have that issue fixed

4 Likes

Back to the drawing board


No big deal to install 1 gear/worm assembly and the good news is everything fits. Yet it is nearly impossible to install and properly line up the 2nd gear/worm assembly

So I will split it in 2 and see how that goes. Plus the print was not that great as I should probably print them 1 at a time and relocate the seam on the inside somewhere else then in the “back” of the outside layer where the slicer (prusaslicer) put it by default. I feel at some point there might be a V2.0. Well its quite a learning experience.

A single herringbone drive for both gears and both worms? It sounds like the clocking/indexing doesn’t quite work for parallel power paths, since both of the worms need to engage while both of the herringbones are also engaged? Not really sure if the seam locations are the problem?

My instinct is a 1:1 belt transfer somewhere in there might add resilience against clocking issues? But as I puzzle through it I can’t prove my instinct is right.

First IMO a belt might not solve the problem on paper as its still a straight indexing Similar to a gear train with ideally even less “slop/backlash” That is why belts are used on printers and CNCs

Great care was given to make it symmetrical to make indexing easier. The seam was not a problem with indexing but it introduced a rough spot slightly binding every revolution. At least on paper the 2nd worm would index properly rotated 180 deg which the herringbone allows.

In real life this is a low cost operation (due to self financing and a host of other reasons) The teeth on the gears are roughly spaced 6 deg. A 6 deg turn translates in moving the wheel roughly up .1mm So If I multiply this with both sides. The axle of the wormgear is a 1/2in 16 gauge steel tube. The hole to take it needs a clearance of .3mm on each side to allow insertion without brute force hammering as the tube is extruded not machined. Even with that there is less slop than on any of the Andymark shafts/hubs. On the first prototype that felt tight there was a backlash on reversing torque of about 1mm which translated into having the choice of picking any one of 4 teeth. So that due to the reduction allowed quite a bit of freedom in indexing. Under torque the axle of the Worm Gear wheel aligned itself to “make up” for it. So indexing is not a real problem in this case (within reason) there are also other areas of inaccuracies. So there is a built in backlash in the Herring Bone gears to account for variances in printing (printer tolerances ar +/- .05mm in theory probably more) so there is a .1mm backlash built in then there are the cheap bearings using bolts as axle or even the shaft of the CIM motor (which on most has a slight wobble too) And the thrust bearings which were less than $1 each also have some tolerances. So that all adds up

All in all I am not that big of a fan of worm gears. They have their use for providing high reductions easily and they don’t back drive. But they are very inefficient due to the high friction and as they dont backdrive and jolt can do some serious damage to them.

I have embarqued on that project as Some on CD said things along the way “would be nice to do a worm drive” and some even said “bet it cant be done in 3dp” So I give it a shot and report. I need something for my material tester so for now the worm gear is it. It most likely even if it works will be replace with a Cycloidal or a multi stage/compound planetary. In best case the Wormgear setup will get another version that lifts 1 robot and will be on my website for all of CD/FIRS to download and use if so desired.

For now between the initial prototyp and this I am in it with about 5 kg of HIPS so $50 in filament and about $3 in destroyed bearings and even retired I can do that as my hobby.

1 Like

One idea for load sharing on the two worms is to switch to straight gears or helical, then add a rubber washer in the thrust direction on each worm. That will give you a stiff spring. Stiffer = thin, wide rubber washer between two steel washers. Soft end of the scale is a o-ring.
You can also add shim washers to adjust Axial position.

yeah that is a “plan B” ATM. There also is some “spring” in HIPS (not as much as most Nylons) . Well I will keep you guys up to date on that. I am currently exploring ways to easier mount this - which probably will enclose both the thrust bearing and regular bearing in the gear which will require a 5ish mm washer which I was leaning towards printing in TPU.

Another thing will be mounting the worm to the gear. The current idea is to have both bearings enclosed in the gear and maybe tape a piece of paper over and slide it in sideways until it mates with one of the small gears. The paper is to hold the bearings as there is no place for fingers. Later the paper can be ripped out. Then there will be the need of either use captive nuts (that means >= M4) in the gear spokes or inserts (need to figure a way to insert them reliably straight with a solder iron) Then after the first worm is in place and something on top is installed to hold the axle in straight then the 2nd worm can be slid in from the side to align with the Worm Gear and then the Herring bonde gear train can be turned (and the worm slightly to keep good contact with both gears) until the mount holes line up so the worm and herringbone gear can be connected. Then Lube and some final enclosure parts

Then up to the top to mount the scale, some limit switches, some mounts for the arduinos and BTS7960s Making a PC board to allow both speed and current control and the proper setup of limit switches and a quadrature encoder. IOW a “bit” of software and some UI to control it.

AND THEN… See if it works as desired - if not there might be the need for some iterations. So not a trivial task.

Can you post a cross section through the worm?