vermicular engineering

Does anyone have something bad to say about worm gears? Or good actually, good is what I am looking for. I want to experiment with drive shafts this year and worm gears look to be a great way to do it, especially with multiple things running on a common drive shaft (say like 3 wheels).

My thoughts so far:

They can have poor efficiency, but you can do similar reductions with fewer steps (and weight and volume). Maybe one hardened and polished worm on a polished phosphor bronze gear is equally lossy as the 3 stage rolled steel on steel spur gear transmission it replaces. Ideally anyhow, this is the case.

Even in ideal situations, they do generate heat. I’d like to run them “open” which is to say with HD grease rather than in an enclosed oil filled housing. 1800rpm is the maximum Boston Gear lists them at, but that is for all day every day industrial work. The easiest thing to do would be to run them straight from a CIM at closer to 5KRPM. Extrapolating from the BG tables, I am still OK on power ratings, but I don’t want to weld them together.

At sufficient pitch angles they cannot be back driven, so you have built in position holding, which saves oodles of complexity.

Motion Industries seems to be the only e-commerce supplier for Boston Gear and have a 6WD suite of worm gears for ~$800; a touch pricey. Anybody have a cheaper supplier, or manufacturer? Mcmaster only has CI gears, which is not ideal, though their 6WD suite would be more like $600.

So who has used worm gears and what nightmares/wonderful dreams have they wrought? Oh, and anybody ever licorice a drive shaft a much lower than expected torque values?


Hi Travis,

I think you will find the following threads helpful:

The general consensus among teams that have tried using worm gears in their drivetrains is that mechanical alignment is of the utmost importance to maximize your (already reduced) efficiency and avoid breaking things. Also, make sure you think long and hard about what materials to use for your worm. Also, be cognizant of the effects of sudden slopping and collisions with robots and field elements on the teeth of your worm.

As far as running your worm at 5000+ RPM, while YMMV, I would doubt that an un-reduced CIM would have the torque to overcome the considerable friction in a worm drive.

Personally, I would reconsider any design using worms in the drive train - there are other (easier) ways to be non-backdriveable besides worm gears.

Hope this helps.

As you noted, these are expensive, in my mind their biggest drawback.

These are better suited for manipulator motion than for drivetrains, particularly because there are far better solutions for drivetrains where space is not such an issue. Think robot elbow instead, where they can be great.

That being said, a cheap source of worm gears for experimentation is window lifter motors. No, they are not strong enough for drivetrain or big manipulators, but you get some nice gear sets for cheap that you can play with and learn.

Carefully consider the axial/thrust loads on the worm shaft, and what types of supports and bearings will be used.

Our 2003 robot was 4 wheel worm drive. It was common knowledge that standing within 3 feet caused you to be covered in grease, almost magically. We had electronics covered in it as well. Definitely the messiest robot to date.

We used worm gears in our robot last year with great success. We didn’t experience a single failure in our drive system. Although, I suspect the low friction surface in last years game made it very easy on the drives :slight_smile:

The gears we used were a bit unusual as far as worm gears go, since the worm had 6 starts. This meant that the reduction was fairly low (5:1 I think), and the gears were back drivable. We obtained the gears by taking apart a pair of industrial gearhead motors that came from a surplus shop. We got several sets of gears for quite a low cost.

As others have mentioned, precisely aligning the gears is key. We machined a housing for the worm with bearings supporting the shaft on both ends. Then a shaft coupler connected to the CIM motor above. The coupler makes things less compact than directly mounting the worm to the motor, but it allows for misalignment, and prevents axial loads from being transferred to the motor bearings. It also simplifies changing the motor.

For lubrication we used grease, and the gears were enclosed with a housing we built. A small hole at the top of the housing allowed re-greasing with a syringe.

Anyway, I have attached a couple pictures of our design. If anyone is interested, I can send Inventor files.


In general teams have found that the thrust loads and gear alignment during run and direction change are formidable problems to overcome. In products that use them, the case/enclosure is substantial in order to hold everything in alignment. I just had part of one open a few minutes ago from our satellite dish azimuth drive. The right angle worm reduction weighs almost as much as the 3/4 hp motor and is fully lubed. All it is used for is to couple to a lead screw that is driving the dish.

All insightful points, well taken.

Many people bring up worm gear efficiency as a problem. I looked deep into the roymech (great online reference by the way) page on gear efficiency and for single start 14.5° worms at 5krpm the efficiency is 89% – pretty fair actually. Efficiency scales as v^-0.25 so you do worse while accelerating.

There was a lot said about reacting axial loads. Does anyone have experience using an E style retainer clip for this purpose? My friends over at Machinery’s told me that an ANSI standard 1/2" shaft E clip will resist 600lbs of static load. I know you have to react that eventually, but I have something sinister in mind.

Several folks said they broke teeth. This indicates to me that worms are not necessarily inherently crummy (though one is suggested to use bronze, not steel gears), but perhaps the gut instincts that work fine for spur gear sizing need a fudge factor to be applicable to worm designs.

The final concern was alignment, on which I am going to just have to bite the bullet.

An alternative at this point would be a largish single stage spur gear reduction – I just discovered Boston Gear’s 5T(!) stem pinions – followed by a smallish reduction and right angle transition using helical gears. Helical gears, being a somewhat special case of worm gears, should have a lot of the same challenges. Anyone have something good/bad to say about errr Crickian engineering?


E-clips would probably work fine for this. Snap rings would probably work better. Your best bet would be to have a structural/geometry constraint such as shoulders on the shaft. flanged or tapered bearings, etc to avoid using a clip at all.

In supporting the axial load, the concern in addition to the support/constraint is the bearing. It can be done (has been thousands of times over, but it’s just more work than the traditional spur gear setup. Typical single row radial bearings are really only intended for no more than about 25% of their radial rating applied in the axial direction. Angular contact bearings are fantastic, but can be expensive. Some teams get by by using bronze thrust washers/bushings and grease. Tapered roller bearings (similar to automotive wheel bearings, but smaller) could be another option.