The noise in a small planetary gear set 2

[spigor]

I've come accross a small size planetary gear set, module 1 mm, external dia. round 50 mm. The gear set uses spur gears. All the gears are made of tempered steel, 22-26 HRc. The set has been noisy, so grinding of the planet gears and the sun gear to the DIN 6 quality has been ordered. The ring gear has not been ground, as it has been impossible to do at the manufacturer's. The quality improvement of the gears did bring very little noise reduction, if any.
I remember same thing has happend years ago- a small gears being noisy, little or no change after grinding, even if a profile crowning has been applied. What did help a little, was reducing of the addendum on the external gears (don't remember which though, probably both the planets and the sun), but this is of course detrimental to their load capacity, and actually should not happen, so I thought back then it had been an accident.

Any ideas/comments on how to quiet this set?

 

[BrianE22]

Is the lubrication grease or oil? A heavier oil might help.

 

[gearcutter]

Can you post the gear data and describe the application/load conditions?

 

[spigor]

Here is the gear data, all dimensions metric:
1.0M 20PA
x=0
2.2WD
24 Axle distance
The sun gear: 33T 35-.1 OD -.036/-.110 tooth thickness tol.
The planet gears: 15T 16.8-.1 OD -.036/-.110 tooth thickness tol.
Ring gear: 63T 61.3+.1 OD +.036/0 gap thickness tol.

The gear is noisy even when turned idle by hand.
The input sun gear speed is supposed to be 2.500 min^-1.
It's lubricated with heavy oil.
The planet gears are on a carrier, there are bearings between the elements that keep them in right place, so the excessive amount of backlash should not have any influence on the noise. BTW, the noise is definitely comming from the meshing.

Can any obvious source of noise be identified here?

 

[Tmoose]

"Can any obvious source of noise be identified here??

I'd be interested in the frequency content of the noise ( sound ).
Octave band type measurements are helpful, but spectral analysis is way better.
This can be done with a microphone, PC with a sound card, and some ancient FFT freeware.
Probably with a 99 cent phone app these days too.

input rpm and ratios must be known.

The fact you say it is noisy even when turned by hand ( rpm much lower than service) suggests that resonant frequencies of the various loudspeakers, I mean gearbox housing surfaces, are not involved, but I would expect vibration measurements ( again, detailed frequency content) with flyweight acceleromenters would reveal the frequencies of interest, which is the first step to understanding what to try.

 

[gearcutter]

From your data it would appear that the set has had no corrections applied.
It is mandatory that the annulus have at least X = -0.20
Without this correction; the tips of the annulus will make non-involute contact with the mating sun gears.
In your case; the teeth of the annulus will extend past the root form diameter of the sun gears by approximately 0.20mm.
That right there is where your noise would be coming from.

 

[tbuelna]

Great point gearcutter. The small number of teeth on the planet gears immediately jumps out as a potential problem. A 15T spur planet, 33T sun, and 63T ring combination would require quite a bit of tweaking of all the gears to provide smooth meshing. If you experience mesh interference (or "noise") when just rotating the planetary by hand, then you have serious problems that a different lubricant or smoother gear flank surfaces likely won't resolve. A detailed geometry study is what is needed first. And then producing a modified set of gears based on the results of the study.

There are lots of available technical references for optimizing the geometry of simple epicyclic spur gear drives. Definitely worth taking a look at before you spend any more money reworking your gear sets.

 

[gearcutter]

Actually; I've made a mistake in reading the OP's data.

At the top of the table the data states that X = 0.
But based on the diameters shown; the tip diameters of the planets and annulus have been truncated. This is a very lazy way of solving the non-involute contact problem.
While it removes the interference; it significantly lowers the contact ratio, which will cause the mesh to feel rough when rotated and also increases the level of noise.

I've seen similar results like this before. It's usually data that's been spat out by software, where the software warns that there is contact inference and suggests the amount of truncation that needs to be applied. The user then applies the correction without realising that they've only created another problem.

As tbuelna suggests; it might be a good idea for someone to go over the design for you.

 

[tbuelna]

I can see where there would be interference between the 15T planet teeth and the 63T ring teeth with unmodified geometries. Profile shift on the planet and ring gear teeth will help, as well as some tip relief on the planet teeth. Contact ratio is usually less of an issue at the planet/ring mesh than at the sun/planet mesh. Tooth bending at the planet teeth can be a problem since they experience full reversing loads. The indexing errors in the planet carrier can present a problem if there are more than 3 planets used, or the bearing system creates an over-constrained condition between the sun, carrier, planets, and ring.

 

[spigor]

Thank you guys, it looks like you have pointed me in the right direction. The gear software has found interference between the planets and the ring gear.
There are 3 planet gears.
I will tell the customer, that this gear set has not been designed properly and needs some tweaking. I'll keep you informed if this would be continued.

Merry Christmas!

 

[tbuelna]

spigor-

You might also want to check the tolerance stack-up between the carrier journals and planet pins, ring gear, housing interfaces, and sun gear. Based on your description it appears you have a fixed ring gear and a 3-planet carrier mounted in a pair of bearings, but you did not mention how the input sun gear is mounted. Your gear drive may be over-constrained which can create mesh interferences if there is some amount of dimensional error stack-up in the gears teeth/bearing journals/housings. With this type of 3-planet epicyclic you could float the carrier to get better load sharing between the planets.

 

[loki3000]

It is mandatory that the annulus have at least X = -0.20
Without this correction; the tips of the annulus will make non-involute contact with the mating sun gears.

didn't you mean the K value? (km)
i didn't go and calculate it though.

 

[gearcutter]

No, it was X that I was referring to.
According to the AGMA standards:

X = Profile shift coefficient

K = Tip-shortening coefficient

Base on the data give by the OP; K has been applied to the gears but not X.........and, in my view, there-in lies the problem.

 

[spigor]

tbuelna,

I'm looking at the gear set and the input sun gear has been floated. The planet carrier and the annulus are fixed as there's a bearing between them. I suppose that the floating of the sun gear could also do the job to some extent, but in this design the sun gear has an involute spline inside, looks like flank centering, so the motor shaft will take some freedom away.

 

[tbuelna]

spigor-

If your 3-planet factorizing simple epicyclic has a radial constraint between the output carrier and fixed ring gear, and the input sun gear is floating, then there should still be good load sharing between the 3 sun/planet meshes. So the interference you describe is likely occurring at the planet/ring meshes. This might be due (at least in part) to the lack of tip relief and profile shift applied to the 15T planets and ring gear as described by gearcutter and loki3000. Or the interference at the ring/planet meshes could also be due to tolerance errors in the carrier and housing features.

You also mentioned some face crown was applied to the sun and planet teeth at some point. With spur gear epicyclics, a common problem is displacement of the planet pins due to torsional wind-up in the carrier structure. Some epicyclics use spherical roller bearings for the planets to address this problem. As a general rule the L/D of your spur gear planets should be <1.0, and if the L/D is greater then there may not be uniform load distribution along the tooth face. If the spur gear planet L/D is >1.0 then you would normally apply some amount of lead compensation or face profile modification to produce a uniform face contact under load.

What would be helpful is to first perform a methodical diagnosis of the deficient gear drive, starting with a check of location/clearance/runout conditions in the assembly and followed by a detailed dimensional inspection of the individual components. This will determine if you have a manufacturing problem or a design problem. You should not implement design changes until you are confident there are no manufacturing issues with your existing components.

Good luck to you and keep us informed of your efforts to resolve this problem. It's quite interesting and I might even learn something from this discussion.
Terry