Bevel Gear Set-up 2

[artbuc]

Power is transmitted to my Ariens mower through a gearbox mounted on the mower deck. Ariens has been using this design for at least 25 years and before that it was used by New Holland. The mower is no longer in production but Ariens will build one for you from parts which is what I just did. Unfortunately, the new gearbox had excessive endplay on the output shaft (0.028" vs spec of 0.003/0.008")which could not be fixed. Usually there are shims which can be removed to reduce endplay but this unit had excessive endplay with no shims. The replacement gearbox was the same. My dealer did a non-Ariens approved modification to reduce endplay (something he has done many times over the years) but now the box has a high pitch whine at full speed and an intermittent scraping noise at half-speed.

Ariens Engineering Department has set-up another gearbox and is sending it to the dealer. They are not forthcoming about what caused the original problem other than saying it was due to an accumulation of tolerances. This doesn't strike me as plausible.

The box design is somewhat crude. It is aluminum diecast with the bearing races for the vertical output shaft located by non-machined shoulders in the diecast. The horizontal input shaft bearing races on either side of the input gear are adjustable to set input shaft endplay and gear backlash. The gears are straight bevel and have no markings to set mounting distance or to identify mating teeth. The service manual does not specify gear mounting distance but says backlash should be 0.005/0.015".

Unfortunately, I think all of the new Holland/Ariens engineers who know about this box have retired and Ariens says they will obsolete the part instead of investing much effort to find the problem. I am expecting the third box will not be much better than the first two.

Can someone give me some suggestions on how to set-up this box? Or, where can I find a person or shop that can set this box up for me? Thanks for your help.

PS I will try to post a drawing of the gearbox although I think you all can probably visualize it.

 

[MikeHalloran]

Thanks for the update.
Do let us know how well it works under load, too.




Mike Halloran
Pembroke Pines, FL, USA

 

[artbuc]

Terry said

"As I mentioned previously, the .030 inch shaft end play seems way to large for your gearbox's tapered roller bearing sets. If your gearbox housing is diecast aluminum and your shafts are steel, then that shaft's axial clearance should be set close to zero at room temp, since the clearance will only increase at operating temps. The tapered roller bearings in your drawing look to have a contact angle around 20deg, which would equate to a radial runout of about .010 inch. And that amount of radial runout would not normally be acceptable at the gear mesh point."

Ok. Al expands 2x steel but the Al case will be air cooling while the steel shafts will not. It may be that this relative cooling will largely offset Al's higher thermal expansion. Or, maybe the steel shafts will actually expand more than the Al case. I am not sure how to allow for thermal expansion. Also, on the input shaft, differential shaft expansion will decrease endplay where as on the output shaft, differential shaft expansion will increase endplay. I want to get this box right but I don't want to outsmart myself. Because I don't know actual operating temperatures of the shafts and case, I am inclined to set both shafts at close to zero even though the spec is 0.003-0.008". Does this seem reasonable? Thanks.

 

[mfgenggear]

"Ok. Al expands 2x steel but the Al case will be air cooling while the steel shafts will not. It may be that this relative cooling will largely offset Al's higher thermal expansion. Or, maybe the steel shafts will actually expand more than the Al case. I am not sure how to allow for thermal expansion. Also, on the input shaft, differential shaft expansion will decrease endplay where as on the output shaft, differential shaft expansion will increase endplay. I want to get this box right but I don't want to outsmart myself. Because I don't know actual operating temperatures of the shafts and case, I am inclined to set both shafts at close to zero even though the spec is 0.003-0.008". Does this seem reasonable?"

No the bearings or gears will seize, must have the mfg end play
or there will be to much friction & heat.

Mfgenggear

 

[tbuelna]

mfgenggear,

"No the bearings or gears will seize, must have the mfg end play or there will be to much friction & heat."

I still disagree. Based on what the sketch shows, .028" end play in a gear shaft mounted in tapered roller bearings would produce excessive runout at the gear pitch line. And apparently, that's what seems to be occurring. Tapered roller bearing pairs are designed to run with little or no axial play. If there is excessive endplay/radial clearance, the tapered roller ends will lose the aligning contact with the inner race shoulders and will skew and skid. This will result in excessive bearing heat generation and scoring failure.

As for clearance changes due to CTE mismatch between shaft and housing, there's no way it will be anywhere near .030 inch. Looking at the sketch, I'd guess that the bearing spacing is about 6 inches max. With a CTE mismatch between the aluminum housing and steel shaft of around 6ppm/degF, that gearshaft would need to get almost 850degF hotter than the housing to take up .028 axial clearance.

Finally, if one shaft is OK with the specified .003 to .008 inch endplay, using a similar bearing arrangement and subjected to similar operating loads and speeds, then why wouldn't the other shaft also be OK?

Terry

A quick test can be done by assembling the gearbox with your desired clearances and heat soaking it in an oven to whatever temperature you think it will see. Take it out and check to see if it still turns freely.

 

[artbuc]

Terry, the factory spec for endplay is 0.003-0.008" for both shafts. However, the box came from the factory with 0.028" endplay on the vertical output shaft. When this was corrected, the box began to howl. Apparently, the excessive output shaft endplay allowed the misaligned gears to run relatively quietly. Of course the box would have self-destructed in short order operating like this. Even after just a few hours the bearing races started to spin in the diecast Al case and I now have to attempt a repair using a chemical retaining compound. I have brought the gears into alignment (see earlier post) and want to be sure I set the box up right before I rerun it under load.

I can do the temp test as you suggested, but it will not account for temperature gradients that will exist in normal operation. Calculating relative thermal expansion is straightforward if you assume all components are at the same temperature. My concern is that the Al box will stay cool relative to the internal steel shafts. This temperature may or may not be enough to offset Al's 2x thermal expansion.

I was wondering if there are rules of thumb or other general experience which would indicate the relative operating temperature of a gearbox case compared to its internal moving parts?

 

[MikeHalloran]

It appears that your assembly process is now superior to whatever the factory is doing.

Maybe they will subcontract the assembly operation to you.

It might bring in a little beer money.


Mike Halloran
Pembroke Pines, FL, USA

 

[artbuc]

Mike, I assume you are making fun of me? Are you saying to set the endplay to factory spec and stop worrying about differential thermal expansion? After all we are talking about a mower, not a space ship! Terry got me thinking that the factory endplay spec was too loose. Guess I am getting a little too fine tuned.

 

[MikeHalloran]

I am absolutely not making fun of you. I think you have got the right idea, getting the contact pattern right first.
From there, you need to set the bearings so that the backlash cannot be forced to zero, and then to provide just enough axial play to not jam the bearings, for which the 3 to 8 mils sounds just fine.

The only remaining question is, how many can you do in an hour? That will help you set a price for outsourcing. I am absolutely not kidding about that, either. Ariens has clearly lost whatever expertise they once had, and whatever (s)he was doing was not documented well enough so the survivors could do it.



Mike Halloran
Pembroke Pines, FL, USA

 

[artbuc]

Hey, I am a retired chemical engineer who just entered the fascinating world of gearbox design. I temporarily lost control of myself. You have to be a little off to enjoy watching Arrow Gear videos on YouTube!

The Ariens story is sad. I am sure that when this box was first designed in the early 70's, component parts were speced to achieve good gear mesh on a reproducible basis. Over time, suppliers changed and specs drifted to the point where a couple shims were required here and there to make things right but bo one is assigned to perform the checks and install the shims. Gearbox assembly has been subbed out to people who just put it together. Heck, Ariens only QC test is to run it and listen for noise. If it is sufficiently quiet (whatever that means) they let it go. No one checks or cares if it has 30 thou BL or 30 thou endplay.

My old box lasted 18 years. And, if I knew then what I know now, I could have given it a tune-up every 4 or 5 years and it probably would have lasted forever.

I have now officially buried the differential thermal expansion bogeyman!

 

[mfgenggear]

Terry

since this is not a precision assembly
I mean it is not precision machined to the spec's that we are used too. there fore the backlash & end play is better to be more loose then tighter.

Probably very loose tolerance gears. guess !
& the housing according to the OP is not machined.
so more end play & back lash is required.
I would say that the assembly of the bevels & shafts
are not square & there is probably misalignment between the
the mounting surfaces.

To make this assembly better, bore out the holes (mounting surfaces) & press a steel or bronze bushing into the holes
& re machine to a better precision.

but lots cost to do all this work, so i recommend live with a little more end play.

I agree .028 end play is to much .
.012 end play should be fine.

my two cents

Take Care

Mfgenggear

 

[tbuelna]

mfgenggear,

I agree that there's likely geometry errors. There's only one reason gears howl- the gear teeth moving in/out of mesh are fighting each other. The numbers of teeth (19/22) on these gears make the problem worse with a low contact ratio.

Thoroughly enjoyed the discussion though.

Terry

 

[artbuc]

Purely qualitative, but as I posted above, I am expecting to say bye-bye to the howl. Before adjusting output gear MD, it has hard to turn the input shaft by hand because of binding on 10 of the 19 output gear teeth. Now, they turn as smoothly as you can imagine with a very good contact pattern. The shafts may not be exactly perpendicular and/or in the same plane, but any misalignment is too small to see in the contact pattern. I am waiting for my Loctite 640 & primer so I can secure the output shaft bearing races. After I do this I will confirm the shim requirement to set output shaft endplay at 0.003" and order them from Ariens. I thought it would be easy to buy shims but aparently these are a custom size (25/32 x 1.00).

The only remaining question I have is how to prevent anymore bearing race spinout on the input shaft. I don't want to Loctite them because I may need to move them someday to adjust BL or endplay. I may drill/tap the case and use a setscrew. Of course, the races may not spin now that the gears are properly aligned.

 

[MikeHalloran]

Use Loctite PST on the bearing races. It will bond well enough to hold them, but remain separable.



Mike Halloran
Pembroke Pines, FL, USA

 

[artbuc]

I talked with Ariens about the misaligned gears and how I was able to get dramatic improvement (at least under no load conditions) by decreasing the MD of the output gear. I asked them what they are doing to make sure their boxes are set up right. They told me they run them on a dyno for 15 minutes. I assume they are checking for noise and vibration.

Is a 15 minute dyno run sufficient to determine if the gears have been properly set-up? The latest box they sent me was relatively quiet initially but was howling badly after 3 hours.

 

[MikeHalloran]

I would NOT assume that they check anything on the dyno.
They are apparently assuming that a 'bad' box will frag itself within 15 minutes, no instrumentation needed.
Sounds like an MBA is running things there.






Mike Halloran
Pembroke Pines, FL, USA

 

[artbuc]

Mike, please be more specific. Are you saying a 15 minute run on a dyno is meaningless? Can any dynamic test be a substitute for properly setting up the gears? Can any 15 minute dynamic test determine if gears are properly aligned or not? Thanks.

PS Hope to have my old box, which I believe is set-up properly, running by the end of the week.

 

[MikeHalloran]

A 15 minute dyno run on gears that are not set up correctly can only harm the gears, and the pieces that fall off will harm the bearings.

If the gears _are_ set up correctly, a minute or so with a varying load and an accelerometer or a microphone should be sufficient to verify it. Lacking a computer controlled dyno, it might take a bit longer; 15 minutes should be more than sufficient even with primitive tools.






Mike Halloran
Pembroke Pines, FL, USA

 

[tbuelna]

artbuc,

If you got a smoother operating bevel gear set by reducing the mounting distance of one of the gears, then what you likely had was an issue with contact ratio and index errors. This would be especially true with the low numbers of teeth and low precision your gears have. They likely need some tip mods or face profile mods.

A 15 minute dyno run is more than sufficient for acceptance testing of your gearbox. If it is going to fail in scoring, it will happen in a minute or two.

Don't be too concerned about backlash, it's not usually an issue with involute geometry gears.

Regards,
Terry

 

[artbuc]

Thanks Terry. Tip and/or face profile mods doesn't sound like anything I can do. I am beginning to think the howling I hear may be unavoidable with the poor quality gears Ariens is using these days. My old box didn't howl for 18 years but maybe the gears were much better? Still waiting for my Loctite...

 

[artbuc]

Hello and Happy Thanksgiving to all. Thought I would give you an update. I have operated the box for about 10 hours after securing the output shaft bearing races with Loctite and decreasing the mounting distance of the output gear by 0.015". The box was much quieter but still had a high pitch howl at full throttle (3500rpm). The biggest improvement came when I de-throttled. The box wound down smoothly and quietly compared to before when it sounded like it wanted to self-destruct.

I don't know what to make of the contact pattern. The output (overhung) gear had a nice smooth (no lines or scoring) pattern in the center of the tooth. However, the input (straddle-mounted) gear had a high pattern concentrated at the toe. There are definite lines of contact which seem to be the same for all teeth. Also, the leading edge (drive side of tooth) of the top land is rounded over to the point where you can catch your fingernail on the ridge when you drag it across the top land perpendicular to the face width. It looks like the input gear is taking a beating whereas the output gear looks good.

Can you conclude anything from this rough verbal description?