tooth mesh when using Mechanism

: "Steve" wrote : When using Mechanism and setting up 2 gears, set the number of teeth etc. : done this. When drag rotation is done everything moves fine - HOWEVER - the : teeth don't mesh graphically - is there a trick to this to give a more : realistic appearance? : Steve, I'm not sure if I understand the problem. It sounds like you're saying that they start out meshed okay, but when you define them as gears, with the ratios and all, that somehow they don't stay meshed when you move them. Is that about it?

David Janes

Steve wrote:

you're not gonna T me for this, so it's just as well you handed them out IA :-)

I've never been able to understand the fascination people have for modelling gear teeth. Are you doing this just so your boss can whack his weewee rapidly ? If so, that's okay. Continue on as before. Otherwise, is this for Real Engineering (tm) ? If so, forget it. Think 'GIGO.'

a) How does the software create the involute curve ? Does it use a mathematical formula to derive an involute from the base circle ? if not, think GIGO. If you don't KNOW, how can you trust your end result ?

b) how does the software deal with the tooth thinning that's necessary in Real Life (tm) for gears to mesh without interference due to manufacturing tolerances ? GIGO.

c) Let's ASSume that you are trying to solve the most basic of gear tooth failure problems : pitting at the pitch line leading to spalling, loss of case, and eventual failure.

This is caused by loads exceeding the film strength of the lubricant, leading to micro-welding which then pulls little pieces out of the mating parts usually starting at the pitch line then continuing to catastrophe. You would think that if you had good tooth profile curves and smart software that takes into account the contact ratio, thus the varying stress on the teeth that occur as the number of teeth carrying the load varies from 1.5 to 2 (approximately, in a decent design) then you might get some fairly accurate results. IF the software takes these factors into account and IF you have the profiles down pat and IF the teeeth aren't modified with tip and root relief to overcome the bending loaded-tooth problem, then maybe you could do this in Pro/E. Even tho it'd take about fifty times as long as using an HP-41, you could probably do it. That's IF you knew that all these items were calculated correctly in the software. Otherwise, GIGO.

d) How about a more real-world (hate to keep bringing these ugly complications in here) situation ? There are two common modes for tooth failure. The first is the one we just discussed - overload leading to oil film failure. The second is failure due to fatigue starting in the root. If it is uncommon to have a correctly-defined involute in most software, it is almost unique to have the trochoids that comprise the root fillets done correctly in software. The only programs which I know of that do that are two-D specialized gear programs which create the fillet by an iterative generation creating a shape conjugate to the correctly- defined cutter geometry. If your fillet isn't the correct shape, how the hell can you trust the analysis ? Especially when it's the fillet shape which is the critical portion of the analysis ? I't common to gain up to 15% in fatigue strength just by modifying the root fillet shape. Does Pro/E do this ? CAN Pro/e do this ? Oh, by the way - hobbed roots are not the same shape at all as shaper-cut roots. So do you know which process will be used to make the parts ? Again, will the parts be finish-shaved with tip and root relief ? Or cut by a hob with tip and root relief ? Or crowned ? unless you know ALL these factors, AND how the software creates its little Virtual gear, your analysis will be useless. No, WORSE than useless because someone will probably believe it. After all, it looks so good on-screen and it came out of a computer, right ? Sigh.

There are many more factors besdies these involved; I've just touched on the most basic ones. Suffice to say, using solid modelled gear teeth for anything beyond amazing the masses is not yet ready for prime time.

I'm sure there are smart people playing with solid-modelled gear tooth analysis, but they already know the problems. I can think of one PhD. who is using APL/2 for a research project, but it's most certainly not commercial software. The people who ask "how can I model a gear tooth so I can test it in Mechanica ?" are the ones who should beware. GIGO. At the present time it is *much* faster and more accurate to analyze gear teeth using traditional methods - the good ol' calculator approach. Leave the tooth-meshing display for the times when you want to amaze your ignorant boss - and better hope he *is* ignorant enough to not start asking the obvious questions. Otherwise, don't waste your time playing with your pecker in Pro/E by modelling gear teeth. It's useless. Worse than useless - it gives false results.

wow hamei! Did this guy touch a nerve!? I was going to say that I've seen some pretty decent involutes get drawn, but you're right... it was a specialized 2D Gear application that knew about most of your points. The graphics wouldn't really help too much if you didn't heed the other warnings coming out of the software.

Actually, I think Steve was just trying to animate some gears that he tossed approximate teeth on (nothing really wrong with that) and he wanted to see them stay lined up. They didn't... because he didn't have any real way to line them up. Probably an initial condition on the joint would have gotten them lined up. I'll bet they were always the same amount of mis-allignment, right? Not incrementally getting worse.

-meld

hamei wrote:

Sorry all - been away... To answer a few questions.. Hamei... The teeth were generated within Pro/E for the sole purpose of motion representation only. I don't plan to do gear design or analysis in Pro/E. I don't have the formulas I used in front of me presently - but if you need to know then post and I'll post back. Your comment about modeling gear teeth - depending on the business etc. modeling of teeth is required for drawing purposes and thus will be in the assembly as well

David... Yes you are correct ... If I assemble in assembly mode - you can use a number of different items and configurations to simulate the gear teeth to mesh in a reasonably accurate fashion. However when you convert to Mechanism and create pin connections the orientation of the teeth from the 2 gears doesn't match up as if they were properly meshed, movement is OK - however if you are using this assembly as a presentation - it is noticeably not correct.

TIA

What are you using for the driving dimensions ? You should be using the pitch diameters. At the pitch diameter two gears MUST move at the same pitch-line speed (otherwise the teeth break and fall off :-)

i don't know Mechanism at all but what you basically have with a pair of gears in mesh is a pair of rolls friction-driving each other with no slip. Think of it that way rather than teeth-as-pegs pushing on each other, maybe you can find a way to simulate the motion more effectively. You may be getting screwed up because the sliding velocity everywhere *except* at the pitch point of the mesh is constantly changing. If Mechanism has a way to represent friction rolls you'll be home free.

I still say it's a waste of time tho :-)

i don't see what the problem with generating gear teeth is, we managed to do this years ago using proe using proper involute curves, compensating for correction factors, class, tooth root radius etc & all done in accordance with iso6336 parts 1 to 5. Unfortunately we are not going to be developing this into a commercial product, but it's fast and easy to customise any aspect of the gears. Just a little bit of hard work is all that is called for.

Right. Only problem is GIGO.

There is no such thing as a "tooth root radius." If you just want to make a cute animation, fine. If you want to do analysis, forget it. You are not including some of the most critical factors in an analysis. Gear roots are *not* all created equal. The manufacturing methods make a huge difference in the shape of the root. The roots are *not* a simple "radius," either.

btw, there's no way in hell you could be "in accordance with iso

6336 parts 1 thru 5" when ISO Method A (which you erroneously call part 1) "...requires precise measurements of actual test gearing ..."

GIGO.

iso6336 part 1 is "calculation of capacity of spur and helical gears, basic principles, introduction and general influence factors" iso6336 part 2 is "calculation of surface durablility (pitting)" iso6336 part 3 is "calculation of tooth bending strength" there is no part 4 iso6336 part 5 is "strength and quality of materilas" Method A that you quote is used throughout this spec and relies on physically measuring already produced gears. It is therefore pointless to use Method A. The "tooth root radius" is generated by the hob used to cut the gear. All of our hobs are to BS978 part 1 fig.1. The tooth root radius is accounted for in iso6336 part 3 section 5. Our proe software does not use mechanica to do the simulation. All the relevant data we require is generated using relations.

That's right. It is pointless unless you wish your results to match the real world (tm).

Or the shaper cutter. And it isn't a radius, it's a trochoid because, as you said, it's *generated.* The 'radius' is not a copy of the tool radius because during the cutting action the tool and part move relative to each other. That's what generated *means*. This is not like milling with a form tool.

I'm sure you accounted for that in your software, right ? the fact that the "radius" in the root is not in fact a copy of the radius at the tip of the cutting teeth ?

And all your hobs use 'standard' dimensions ? Which 'standard' ? DP ? Module ? Shaper depth ? Preshave depth ? Pregrind ? Split pitch ? Old AGMA ? new AGMA ? Composite tooth ? What do you do about tooth thinning for backlash, ignore it ? What about long and short addendum gearsets, you don't do that ? Never compensate for equal wear ? never run extra-depth teeth for smoother running and less noise ? Never use full radius tip hobs for extra fatigue strength ? You stick with a 20* pressure angle ALL the time ? you're aware (no, probably not) that there is no such thing as a "standard" tooth ? Nobody who is a professional gear person sticks with "standard" tooth relations or "standard" proportions which means that in gears there IS no "standard" except for the convenience of buying off-the-shelf cutting tools for commercial grade work. "Standard" cutting tools which one then uses at non- 'standard' depths to create non-'standard' teeth for the purpose required. In a real gear design 'standards' are *starting* points. For that matter,if you are doing production you don't even use "standard" pitches. Go look at an automotive gearbox some time. Or a set of plastic gears. Or timing gears. Or powder metal gears. Or anything intended to be quiet. Or carry large loads with tip and root relief. In fact, you can even buy off the shelf cutters set up with protuberance for tip and root relief ... I suppose you account for that, too ?

There is no such thing as "standard" in the gear world, except for beginners who don't know any better.

That's really really nice. And I'm sure you know that shaper-cut teeth are deeper than hobbed teeth ? And a totally different shape in the root ? I'm also sure you know that the roots of even hobbed teeth are NOT simple radii ? Did you know that a large percentage of gear tooth breakage occurs because of cracks starting in the corners of the roots ? And that often deeper roots with larger curves are effective in increasing fatigue strength ? Usually by around an 'insignificant' 15% ? Cut with 'standard' full-radius (on the tip) cutting tools ? You're aware that there are *several* 'standards' for different gear cutting tools and *several* 'standard' tooth depths so that just about every set of gears is *different* ?

You account for this in your software, do you ? unh-hunh.

As I said, GIGO.

Yeah, all the relevant information probably *could* be generated using relations. It *could* be, but dollars to donuts in your case it *isn't* because you aren't even aware of what the relations are.

And that was my initial point : this all *can* be done in software (in fact some people are doing it in software, but not off-the-shelf pro/e, that's for sure) but it's a lot more work than it sounds like and if you don't know what you are doing the results won't match the real world which is sort of the point of analysis, so in effect it's meaningless. In fact, worse than meaningless because it gives people who think they know everything just because they looked in a book and pulled some numbers off a chart the impression that they know what they're doing. Which, at the end of the day, could be very destructive. My point was not that gear tooth analysis *can't* be done, but that people asking the question "How can I model gear teeth in ProE so I can analyze them ?" really *shouldn't* be doing this because if they don't already know what they're doing they'll get bad results. Unless, of course, they just want to make a pretty picture to impress the boss, in which case it's fine (as long as no one mistakes the results for accurate.)