Turning down the OD of a gear

I’m designing a gearbox that would work really nicely if a gear with an OD of 1.700 had an OD that was .020 less. How bad would it be to to turn down the OD of the gear .020? Also, does anybody know exactly the OD of a CIM motor? The website claims 2.54 max, but do they come in under this?

I am not sure about turning down a gear. The CIM motors are 2.52 max, they vary depending on manufacturing within a couple hundredths.

If you turn down a gear .020"…

… Your chance of the gear slipping will go up, albeit slightly.
… Your chance of messing up the tooth profile will go up–you really don’t want to mess that up, so make sure you clean up carefully after whatever machining op you do.
… If the gear is heat treated, you’re probably going to change the heat treating. This will probably lead to unintended consequences, good and bad (but most likely mostly bad).

But, you should be able to get away with it. For a little while, at any rate.

Now, I have to wonder… Why not just move the gear .020" farther away from the other gear, in some direction? And, what’s with asking if the CIM comes in under its stated diameter? If you’re building a really tight gearbox, I suggest that you also CAD in the assembly tools for that gearbox, just to make sure they will actually work…

The idea was to have three CIMs drive a large center gear, then have a smaller gear located underneath the CIMs, driven by this center shaft.

Ah, I see. Maybe move each CIM .010" out from the center, and move the other smaller gear .010" down? Or is there some other space constraint that would prevent that?

You can expect the Cims to be 2.536" OD. I think turning down the teeth of the gear could work, but you would be shortening the involute profile of the gear which could lead to problems depending on the power moving through the gear.

That would mess up the center to center distances.

That is true. On the other hand, part of design is designing for what is available. If the only available gear is .020 larger than needed, then that needs to be considered. Center-to-center shouldn’t affect it–matter of fact, I’m specifically suggesting changing the center-to-center BECAUSE the center-to-center is too small for that gear, and the only thing a center-to-center will affect is where on the gear contact is made (though, that little detail can introduce some backlash, among other things). That can also be mitigated by playing with a few other things–for example, if your center-center is exact distance, most FRC folks will add 2-3 thou to that to account for manufacturing tolerances; if you make it 10 thou instead, you’ve really only added 7-8 to what is normally done.

Personally, I wouldn’t attempt turning down a gear once it was cut, for the reasons listed above–there’s too many things that could go wrong, minor enough it would seem, but just too many items to tempt Murphy and his gremlins. That goes for both a mill and a lathe, BTW. If someone really knew what they were doing, then that could work, but in FRC, those types of folks are rarer then we’d like to think, unfortunately.

Is it really that common in FRC to add 2-3 thou? (I honestly don’t know) For every gearbox I’ve designed I’ve only added half a thou and haven’t had any problems with wear.

I do see the problems with the actually trimming the OD, depending on the team’s machining expertise and capabilities it could be quite difficult to do. correctly

It’s been pretty common in the past. Part of the wear thing is that our gears are only running for about an hour per event, give or take a bit (some teams will put more on practicing), which is only about 300,000 cycles for anything going at about CIM free speed, and most gears will be going slower than that. That’s not a lot for a typical gear. Even factoring in sudden reverses and the fact that not all gears will be going all the time, it’s not a lot of cycles. Once you start getting through a district schedule plus a full round of post-seasons plus demos plus driver practice… Then you might start seeing more wear.

Also, it’s a lot easier to add 2-3 thou than 0.5 thou, something about the latter being much harder to do reliably, and needing better measuring tools. (Back in my day, a lot fewer teams had CNC capability. Then again, there were a lot fewer teams.) I’d go into a discussion of tolerancing and its effects, but that would be pretty far off topic, so if you’re interested drop me a PM. Short version, X.000 vs X.0005 is probably going to be more expensive and time-consuming one way than the other. Take a wild guess which way is higher cost.

I think this practice is also becoming slightly less common because of the prevalence of VEXPro gears. Their hex is oversize, while the shafts are undersize; this leads to slop. Adding .003 to c-c can lead to very noticeable backlash.

Adding three thou is a psuedo-standard in FRC. It’s a good rule of thumb for custom gearboxes. It’s also one way to artificially shift your tolerance band.

Yeah… We had a 6(!) stage arm transmission this year and we had over seven degrees of slop in our original arm. We added the three thou but all of the loose hexes added way up. All of the shock loading of the arm flopping around in the straight up position led to multiple gear failures.

So yeah, I think we’re going to .001 next year. I’ve heard people run zero backlash with Vex gears with no problems, but I’d like a little bit more space for grease in there, and this avoids having to break in gearboxes for any extended period of time. (The kids made me promise I wouldn’t push for a geared arm again too)

My experience is similar to Eric’s: the addition of up to 0.003 in was necessitated by gearboxes being produced by hand by inexperienced machinists on milling machines of uncertain provenance. If you’re CNCing the gearbox plates, you can in all likelihood get away with less. But remember that the added centre distance isn’t just for position of the holes, it’s also for runout of the gear and perpendicularity of the axle. (For example, in 2006, 188 had a bad EDM job that screwed up some expensive hardened gears, and necessitated a lot of rubbing compound to “solve” the runout.)

I would be interested in hearing from some of the VexPro people about the design and tolerances of their gears. How far from standard tooth profiles can they be expected to deviate at perfect form (e.g. due to profile shifting, radiusing, etc.), and how much variance is introduced in manufacturing?

By the way, what are the dimensions and diametral pitch of this gear? It’s kind of important to know how much of the tooth is being lost. But maybe that’s part of the solution: you could go with the largest teeth that will work. In most cases, the limit will be set by the undercut of teeth on the pinion as you go down to small diameters. (To improve that, prefer a higher pressure angle given gears of a certain diameter and the same pitch.)

We did notice that too, but since we machine our own hex, we make it oversize to mate their bores nicely and haven’t had those issues.

Can you really hit a half thou accuracy? Even running finish and float passes and really getting to know the machine, program, and material, I can usually only hit around 1-2 thou accuracy on our CNC for stuff like hole to hole distances.

Yeah, 3 thou is the standard number. I’ve heard that it mostly came into being in a time when FRC gears were pretty low quality, VP gears nowadays are high enough quality that you don’t really need the adder. Another reason for the adder is that if you have a bit of runout on your shafts, or poor quality gear teeth, and you put the gears at the nominal pitch diameter, there’ll be places in the rotation where the gears get sort of jammed together, and you lose efficiency. Adding a few thou helps make sure that the gears aren’t getting jammed into each other in case of tolerance issues somewhere else.

Quick question to the OP:
Why does this matter so much? If it’s for clearance to spacers, just move the spacers. And if it’s for ground clearance, forget it. You will need more ground clearance than 0.02" due to the tread wear, compression, and random stuff on the ground.

CIMs are always 2.536" outer diameter unfortunately. You could look into RS-775, but if this is for drive CIMs would work better.

Turning the OD of a gear can have some negative issues as people have stated you are removing 10% of the active tooth depth. Depending on the tooth counts used this can have a much more dramatic effect than a 10% reduction in torque capacity. One of the factors that drive the rating of a spur gear is known as the contact ratio. It is essentially a measure of what percentage of time one tooth is in contact verses two teeth in contact. When you truncate the OD of one gear you shift that contact ratio closer towards one tooth in contact more often than two teeth. You could imagine when the load is shared between multiple teeth more more load can be transmitted.

Additionally many times gears have what is known as tip relief. Where the involute is modified slightly on the tips of gears to allow them to enter and exit mesh more smoothly especially when a load is applied. I do not know if FRC gears have tip relief machined in, but if they do you will remove it and possibly cause the gears wear more quickly and be noisy.

How many teeth are on the CIM pinion? Would it be possible to use a larger pinion? If you are trying to keep a specific ratio vexpro might make just what you need. They appear to have some CIM pinions with modified addendums or shifted centers. (The modified addendums will allow you to add .025" of center distance without the increase in backlash or shaving the OD of your gears basically lets you set the center distance up as if the gear arrangement had an extra tooth).

See page 9.

http://content.vexrobotics.com/vexpro/pdf/VEXpro_Gear_Reference_Guide.pdf

This is very case-dependent. Just remember that the teeth are designed to be in contact at the pitch-circle diameter, not at the outside diameter. So cutting the outside will have no effect on many systems.

So as long as you have ‘regular’ gears, not some kind of ten-tooth pinion or something, the outside two percent of the tooth should never be needed as the next tooth is engaged long before it disengages.

Also, technically, involute teeth aren’t sliding, they’re 100% rolling contact, but if you change the center-to-center distance, even by adding a measley 3 thousandths, you’ll have some sliding contact and less gear efficiency, more wear.

But as someone above mentioned, what is the gear pitch?–if the teeth are very small, 0.020 could be alot. if they’re big teeth, 0.020 is less concern.

Quoted for truth: It may not be obvious unless you have studied gears, but the teeth if an involute gear don’t just push on each other, they actually ROLL across each other. The sliding action (which leads to wear and wants a lubricant) is almost zero. That’s one of the reasons center-to-center distances are so very important for gearboxes.

And if anyone reading this thread hasn’t had a chance to study a little bit about how gears work… take a look. We generally tend to look at gears and say, “Oh, it’s just a gear.” and take them for granted, but the math and science that goes into gear design is truly fascinating. The modern gear is the product of centuries of engineering evolution.

I know I’ve still got a lot to learn about how gears work, wear and are manufactured.

Jason