Re: Threaded rod through ball bearings
 

Nice isometric view. There's a big stress concentrator where the shaft leaves the outer bearing and before it enters the coupling nut. The shaft will fatigue and break there.

Epoxy or most other fillers will not increase the yield strength of the shaft in bending, nor remove the stress concentrators due to thread geometry, since their yield strengths are orders of magnitude below that of steel.

I can't give you the forumlae, but the number of threads engaged is the key to calculating the force to strip threads - in other words, a function of cross-section of material engaged by the nut. If you think of the thread ridge as a straight ridge instead of a spiral, it'll make the calculations easier while not being too inaccurate. A 3/8" coupling nut with more than about 10-12 threads engaged will have a good chance of breaking the shaft before the threads strip. Hope that's what you're looking for.

Seems to me that a hex shaft, turned down to a rod for the bearings and outer wheel, then threaded on the outer portion, would be plenty strong and not difficult. Do you have a lathe? Can it cut threads? Good learning experience...

Don

Don

Re: Threaded rod through ball bearings
 

I was thinking about lathes....I have half a dozen friends who have lathes....they really aren't that hard to find!

Re: Threaded rod through ball bearings
 

Though the epoxy is an order of magnitude weaker than the steel (estimated 3 500 psi tensile yield), it ought to significantly reduce the effects of fatigue on the stress concentration. (Fatigue being a major reason reason why parts fail at stress concentrations.) It provides an alternate load path for small but repeated stresses to be transferred within the shaft. The effective net stress in the areas of stress concentration ought to be much lower, assuming that the fatigue-induced stresses do not cause the epoxy to yield, or the joint to split. I'm not sure if you're running it enough to warrant protection against fatigue, though—the back of the envelope says that you shouldn't see enough iterations of stresses anywhere near the fatigue strength for the few hours that a typical FIRST robot runs.

When it yields, then yes, all bets are off. But shouldn't the shaft be designed not to yield due to bending loads, under normal operation? If it isn't operating in the elastic region (of the steel), it's probably a bad design to begin with, because you can expect it to need replacing, once it deforms. And realistically, for a 130 lb robot, is it likely to see enough stress in the shaft to plastically deform it? Assuming all the robot's weight is on one wheel, and the stress concentration factor is 2 (that ought to be conservative), you're looking at around 3 600 psi stress in the shaft, which would just barely cause the epoxy to fail, if it were withstanding the load on its own. With robots potentially bouncing and falling off of things, maybe this isn't enough—I'm not sure about the strain rate sensitivity. But when you figure that the shaft is still mostly steel, that the epoxy would mostly be in compression (where it's at least twice as strong), and that the thread crests would be bearing some of the compressive load inside the bearing, it looks like it ought to work.

By way of rough analogy, consider that epoxy-coated bolts don't lose their epoxy coating when they are pretensioned. Further, they don't lose their corrosion resistance under these circumstances. That implies that the epoxy coating remains intact and unbroken, even under tensile stress. While I realize that the epoxy is not fulfilling a structural role, and that the stress isn't repeated, it does suggest that a high-strength epoxy won't merely crumble and fail when the bolt is loaded.

Also, in mild steel, the yield strength (tensile/shear/fatigue) of weld material should be of the same order of magnitude as the unhardened base metal (and in many cases, nearly identical). If the bolt is unhardened, it's very likely that this will make it a fair bit stronger.

On another note, looking closer at the design (like clicking on the ImageShack picture to make it bigger...), I realized that you weren't using Trantorque- or Fairloc-like couplings, but rather a standard AndyMark hex hub. That would make it a little trickier to attach the wheel and sprocket to a shoulder bolt. Maybe there's another way to simplify this, though: try a hex-bore bearing and use hex shaft throughout.

Re: Threaded rod through ball bearings
 

If you do not have a lathe or milling machine, you might try checking with a local community college that has machining courses. Our local college was kind enough to turn down a bunch of sprockets for us when we had a weight problem.

Re: Threaded rod through ball bearings
 

We have one each of a mill and lathe, but they're always a bottleneck when it comes to getting robots together -- hence my desire to avoid using a lot of parts that need to be extensively machined. The milling machine allows us to put external keyways into shaft, but it's more difficult to key things like sprocket hubs, so I try to avoid using them whenever I can get away with it.

I did some digging last night and it turns out that all of the adapters and spacers and bolts needed to avoid using a keyed, steel sprocket make the weight saved almost negligible, so I've been reworking the design to use 5/8" keyed shaft and keyed sprockets. Since the wheel is comprised of three aluminum plates stacked together, I can cut plasma cut a key into the hub of each.

This means abandoning the use of threaded rod and coupling nuts altogether at the cost of redesigning the gearbox a bit. It seems right now that the simplest solution that'll allow me to have a hex output to the cantilevered center wheel and a hex shaft inside the gearbox is a bushing with a 1/2" hex opening and a .625" outer diameter. I can cut two or three thin pieces of metal or plastic to make the bushing since I can't broach a hex hole the traditional way.

Re: Threaded rod through ball bearings
 

They make hex bore bearings? Cool.

The epoxy on bolt threads (when mated) is under compression, but this shaft has no such forces between the bearing and the outer shaft coupler. Material under compression will have better tensile survival since the absolute force never gets far into the other size of zero. I would expect the joint to split, but that might not be reasonable on my part.

However, to go any further, I'd have to do some experiments. I do agree, completely, that if the design exceeds (or even approaches) the edge of the elastic region, it's not a good design.

But, Madison seems to be moving towards a 5/8" shaft, so I guess it's moot.

Don

Re: Threaded rod through ball bearings
 

as used in conveyor systems, apparently...

http://www.google.com/search?hl=en&q...=Google+Search

Re: Threaded rod through ball bearings
 

They're sold by McMaster-Carr as 'Conveyor Bearings' and are mentioned in the Timken catalog on a page about farm equipment. I didn't know they existed until Tristan mentioned them.

It appears there are very few varieties commonly available, however, and unfortunately, 1/2" is not among them.

Re: Threaded rod through ball bearings
 

These guys say that they can make 1/2" bores by special order. I don't know what the cost premium or lead time on those would be, however.