Gear hobbing process question, what causes difference on one flank vs the other

[SBAER]

Hi

We make the gears for our first prototype by wire EDM and that works pretty well. When we need to make the first run of 20 gearboxes we get the gears hobbed by a local shop that specializes in this type of work. Gear noise is a big issue for these gearboxes, we are trying to make the quietest gearboxes possible but still limiting ourselves to AGMA Class 9 tolerances. We don't fully understand how all the individual tolerances that make an AGMA 9 gear an AGMA 9 gear affect the noise of the gear train over its intended speed range.

There is one gear set in the current gearbox under development that is excessively noisy in one direction but not in the other direction. We can change the direction that the objectionable noise occurs in by flipping the gear over (pinion orientation does not affect noise). The whole lot of 20 gears has the same issue to some degree of the other, some are very quiet in one direction but really loud in the other, others are not as quiet in the good direction but also not as loud in the other. The gears are 4140HT, nitrided after hobbing. We have a shadowgraph that allows us to compare the generated gear profile to the CAD (theoretical) profile, but the 10:1 magnification is not high enough to conclusively highlight any anomaly (the gear has a module of 2.25). If we look at flanks of the gear that has run for a couple of hours under a microscope we see a different wear pattern on one flank vs the other. See pictures below



I have uploaded a video with sound if you are curious enough

Relating Sound and Wear Pattern_1.mp4

drive.google.com

My question is; what could have gone wrong in the hobbing process to cause this? Is this worn equipment, badely sharpened hob, sloppy setup or something else?

 

[Mark Rand]

First, my guess would be that the hob was worn.
Second, as a thought. If low noise is paramount, it it possible to redesign with helical gears? That would take bi-directional thrust bearings for the bi-directional rotation.

 

[EmGo]

First, my guess would be that the hob was worn.

No, it's a much worse fuckup than that. It actually looks like those were cut on a shaper and the parts moved on the arbor as the cutter fed in, then didn't clean on the next pass.

But that's just a guess, it would be hard to make that particular kind of a mess with a hobber.

Don't think it's necessary to go helical but again from the photo, those teeth would look terrible from a quietness aspect, even if they weren't butchered

It's not a little process thing, SBAER. This one was a total flockup of some kind. Maybe rude to say this but I'd definitely change suppliers. There's decent gear shops out there.

We don't fully understand how all the individual tolerances that make an AGMA 9 gear an AGMA 9 gear affect the noise of the gear train over its intended speed range.

The only part of that that will make much difference to noise is tooth-to-tooth spacing. Gear classes are more about strength and accuracy than noise ... for example, an AGMA 9 gearset of 20* PA teeth will be noisier than an AGMA 9 set of 14.5* PA teeth. Same accuracy but lower PA usually means quieter. Noise reduction will usually be dealt with in the design phase, or there are methods to improve it post-heat treat : shaving, honing, grinding, lapping (not with each other, with a crossed-axis lap) all reduce noise.

But again, them tooths look like they will be noisy no matter what you do ! They could be AGMA 15 and I bet they'd still be loud.

 

[10 fingers]

I can't tell for certain from the photos, but lack of feed scallops suggests these gears were shaped. Even so, is a HORRIBLE finish. Cutter had to be very dull to tear at the flank that badly. How did gear shop inspect these gears? Have a value for pitch line runout? What about involute form error? What is contact ratio? This looks like the work of some hack, hobbyist.

 

[SBAER]

Thanks everyone, I did not consider the fact that these gears were possibly shaped and not hobbed (I am a slow learner

I appreciate suggestions as to which parameters to optimize for if noise is a prime consideration. This is for a medium volume ( ramps from 500/year to 5000/years), medium duty application that is cost sensitive. Helical gears are not on the table, nor is grinding. We have have been making addendum modifications to optimizing the contact ratio (its about 1.5). As long as we are cutting these gears from 4140HT contact stress is always the lowest factor of safety and is the determinant of gear life. The reduction ratio is kind of high at 79/18. Looking to avoid a catastrophic failure mode.
We were hoping to come up with a design that at some point in the future could be done in powder metallurgy, hence the 20PA but maybe we should drop that idea and lower the pressure angle.

 

[triumph406]

Thanks everyone, I did not consider the fact that these gears were possibly shaped and not hobbed (I am a slow learner


.

.Ask the supplier

 

[10 fingers]

Thanks everyone, I did not consider the fact that these gears were possibly shaped and not hobbed (I am a slow learner

I appreciate suggestions as to which parameters to optimize for if noise is a prime consideration. This is for a medium volume ( ramps from 500/year to 5000/years), medium duty application that is cost sensitive. Helical gears are not on the table, nor is grinding. We have have been making addendum modifications to optimizing the contact ratio (its about 1.5). As long as we are cutting these gears from 4140HT contact stress is always the lowest factor of safety and is the determinant of gear life. The reduction ratio is kind of high at 79/18. Looking to avoid a catastrophic failure mode.
We were hoping to come up with a design that at some point in the future could be done in powder metallurgy, hence the 20PA but maybe we should drop that idea and lower the pressure angle.

I want to convey this in the most constructive way possible, You Are Way Over Your Head. If one of my employees tried to learn gear theory from asking strangers on the internet vs. studying a text book I wouldn't hesitate to fire him. If you are serious, need to buy, read, and study this, https://www.amazon.com/Handbook-Practical-Design-Darle-Dudley/dp/0070179514

 

[EmGo]

This is for a medium volume ( ramps from 500/year to 5000/years), medium duty application that is cost sensitive. Helical gears are not on the table, nor is grinding. We have have been making addendum modifications to optimizing the contact ratio (its about 1.5). As long as we are cutting these gears from 4140HT contact stress is always the lowest factor of safety and is the determinant of gear life. The reduction ratio is kind of high at 79/18. Looking to avoid a catastrophic failure mode.
We were hoping to come up with a design that at some point in the future could be done in powder metallurgy, hence the 20PA but maybe we should drop that idea and lower the pressure angle.

A few things ... I disagree that you need to pay someone, you can bang this out yourself fine but it will take longer and maybe require more testing. Might still be cheaper in the end ...

Powder metal doesn't care what the pressure angle is, so that's not an issue.

4:1 is not a terrible reduction but the low number of teeth on your pinion kinda might be.

1.5 contact ratio will NEVER be quiet. NEVER.

Helical doesn't cost any more than spur.

The teeth you have look terrible. I am surprised they don't drive you out of the building with the screaming rattling noise.

If sound is an issue, then you might scrounge around for an old Red Ring sound tester, or build something like it yourself. Shaving, honing and lapping will all reduce the noise. (Again, before someone gets bad ideas, NOT lapping the two parts together, that doesn't work.)

But back to the beginning, those teeth look really really really awful. I am kinda surprised they work at all and I don't mean just the surface finish. You'd be better off with something out of Boston Gear than those things.

Time to start over, I'm afraid. A contact ratio of 1.5 ? UGH ! ugh uggity-ugh ugh.

 

[SBAER]

I want to convey this in the most constructive way possible, You Are Way Over Your Head. If one of my employees tried to learn gear theory from asking strangers on the internet vs. studying a text book I wouldn't hesitate to fire him. If you are serious, need to buy, read, and study this, https://www.amazon.com/Handbook-Practical-Design-Darle-Dudley/dp/0070179514

The problem is I would have to fire myself. I am over my head in all aspects of my business, if you think my gear knowledge is wanting you would be shocked by gaping holes in my understanding of managing people.

 

[DDoug]

The problem is I would have to fire myself. I am over my head in all aspects of my business, if you think my gear knowledge is wanting you would be shocked by gaping holes in my understanding of managing people.

Hire a gear expert as a consultant, try a semi retired one.

You may well have to pay for a corrective grinding/lapping after HT, whether you want to or not.

 

[SBAER]

A few things ... I disagree that you need to pay someone, you can bang this out yourself fine but it will take longer and maybe require more testing. Might still be cheaper in the end ...

Powder metal doesn't care what the pressure angle is, so that's not an issue.

4:1 is not a terrible reduction but the low number of teeth on your pinion kinda might be.

1.5 contact ratio will NEVER be quiet. NEVER.

Helical doesn't cost any more than spur.

The teeth you have look terrible. I am surprised they don't drive you out of the building with the screaming rattling noise.

If sound is an issue, then you might scrounge around for an old Red Ring sound tester, or build something like it yourself. Shaving, honing and lapping will all reduce the noise. (Again, before someone gets bad ideas, NOT lapping the two parts together, that doesn't work.)

But back to the beginning, those teeth look really really really awful. I am kinda surprised they work at all and I don't mean just the surface finish. You'd be better off with something out of Boston Gear than those things.

Time to start over, I'm afraid. A contact ratio of 1.5 ? UGH ! ugh uggity-ugh ugh.

Hi Emanuel

The problem with the helical gears is not so much the gears themselves as the added expense of transmitting the axial forces. There are gear reducers for this application that are built to an overall higher quality level (helical gears) but that is not the part of the market we are targeting. We have a similar reducer in production, the noise level is deemed to be acceptable.

The powder metallurgy gears have with a lower root strength than wrought gears hence the higher PA to compensate.

We have not found it so easy to get gears made quickly (within 4 to 6 weeks), even when we specify gears that can be cut from standard tooling. Going from the wire EDM cut prototype gears which we make in-house to the batch of 20 hobbed/shaped samples required for pre-production verification has been a logistical challenge. If we send the hobbed gears back to the shop that made them because the surface finish sucks the project schedule gets obliterated.

Goggle does not seem to have much awareness of the Red Ring Checker but I did find this advertisement on Ebay.



National Broach and Machine company appears to still be in existence although the gear testing stuff seems to have been sold to Nachi and then to Involute Gear and Machine, I found a National Broach gear accuracy patent that looks like it is doing something similar to measuring over pins.


There is no doubt having more capability to measure gears in-house would be helpful. A CMM machine that can measure along a prescribed spline might be easier to find in this day and age.

 

[EmGo]

The problem with the helical gears is not so much the gears themselves as the added expense of transmitting the axial forces.

The axial forces are not as big as you think, just deep groove ball bearings would probably work fine but at this point, maybe that's cart before horse ...

The powder metallurgy gears

Powder metal is going to eat you alive on tooling cost, you haven't even got a working version yet, way too early to consider that.

We have not found it so easy to get gears made quickly (within 4 to 6 weeks), even when we specify gears that can be cut from standard tooling. Going from the wire EDM cut prototype gears which we make in-house to the batch of 20 hobbed/shaped samples required for pre-production verification has been a logistical challenge. If we send the hobbed gears back to the shop that made them because the surface finish sucks the project schedule gets obliterated.

Here's a root problem that you can attack. If you were on the west coast or wanted to send to shijiazhuang I could make a few suggestions (Ground teeth from china are probably cheaper than hobbed in the US ... )

Either find someone you can work with (sounds like you tried and there aren't any locally ?) or do it yourself. Your parts would go on a 7A Fellows shaper (or an 8 or a 4, those are less common.) Should be able to find one for five-ish. That's a bit of a chunk for a project that's not even off the ground yet but will pay for itself pretty fast. They aren't that hard to run. If you have your own you can turn up a couple blanks, walk into the other room, cut some teeth and have something to test in a day. Then when you get what you like, the cost is just the guy to load blanks and push the go button. Gear machines are just as automated as cnc, except they cost a lot less to repair.

So that is what I would do ....

Goggle does not seem to have much awareness of the Red Ring Checker

There's lots of "testers" of different types, I was thinking of the sound tester, which is really just a pair of shafts that you put your parts on, in an enclosure with an acoustic horn so you can listen to them run under load. You can find them used, haven't been made in decades, Red Ring is effectively gone, along with almost all the rest of the US gear machine industry.

There is no doubt having more capability to measure gears in-house would be helpful. A CMM machine that can measure along a prescribed spline might be easier to find in this day and age.

There's software for cmms to do that, there are dedicated cmm-type machines especially built for gears, and there's the older stuff like involute and lead testers. Lots of that isn't going to mean anything to you and won't really do what you want.

I would suggest a nice redliner. For your purposes those will tell you everything you need to know and they are faster and easier to use.

So ... if you are convinced you will never do helicals, a small gear shaper. Otherwise, a hobber. A setup of some sort for sound testing. And later on a redliner ... that'll give you the basics to take charge of your own destiny

 

[David_M]

"Some gear designers have used and advocated a 200% pinion addendum and a 0% gear addendum. This design has all recess action when the pinion drives and no approach action. If gears are somewhat accurate, they will run more smoothly and quietly as the arc of the recess is increased and the arc of approach decreased. There are special cases in which the design of a gear set justifies a very long pinion addendum and a very short gear addendum." - Dr. Stephen P. Radzevich - Dudley's Handbook of Practical Gear Design and Manufacture, Second Edition.

I'm not suggesting the above applies to your case; I posted it to show something. I have Kindle for pc on my computer, and I think books are much easier to search using keywords like 'quiet, quietness, quietly, etc.' this way: that and an extensive library of reference material in Kindle format.

P.S. follow Emanuel Goldstein's advice.

 

[MCritchley]

I want to convey this in the most constructive way possible, You Are Way Over Your Head. If one of my employees tried to learn gear theory from asking strangers on the internet vs. studying a text book I wouldn't hesitate to fire him. If you are serious, need to buy, read, and study this, https://www.amazon.com/Handbook-Practical-Design-Darle-Dudley/dp/0070179514

Thanks for the link, I bought the copy of the book you linked.
Every machinist or fine engineering mechanic can bone up on gearing.

 

[EmGo]

"Some gear designers have used and advocated a 200% pinion addendum and a 0% gear addendum. "

Back when we had rear-drive cars, the ring and pinion were designed along those lines. Remember when backing up made big howl, but going forward was quiet? But if OP's device needs to run in both directions, that trick won't work

 

[EPAIII]

I can not contribute anything to the specific gear problem but it seems that your back is against the wall with this first production run. No time to go to another gear company and no time to send them back and hope they can correct or replace them.

Have you discussed this with your gear shop? Frankly, if that was a product made by a company I owned or managed, I would be bending over backwards to make it good.

Assuming that does not work and with a lead time in weeks, which I have seen in gear orders before, perhaps doing replacements with your own EDM equipment for this first batch is something you should consider. Even if you can only do one a day, that's only 20 days instead of nearly double that getting them made by others. And you should have a good one on the second day with additional ones on each of the following days. Getting someone to work overnight may speed the process up to ten days or even less. If it was my company I would be there myself.

That should give you some breathing time to find a better gear shop. Or resolve the problem with the present one.

 

[DDoug]

What does your drawing look like ?
I worked with a gear engineer, detailing drawings, and he had me make a large table of specialized gear tolerances.

 

[SBAER]

The drawing had all the gear tolerancing in the table, it is a sizeable list most of which can't be measured without some pretty specialized equipment (CMM with gear software package). Determining what values to put in the boxes is what you pay the gear consultant to use his software for. EDMing the batch of 20 gears is not impossible but this would cost a fair bit of money. The schedule has been thrown out the window for this project, hopefully we will learn enough to avoid the problem on the next project

 

[DDoug]

The drawing had all the gear tolerancing in the table, it is a sizeable list most of which can't be measured without some pretty specialized equipment (CMM with gear software package). Determining what values to put in the boxes is what you pay the gear consultant to use his software for. EDMing the batch of 20 gears is not impossible but this would cost a fair bit of money. The schedule has been thrown out the window for this project, hopefully we will learn enough to avoid the problem on the next project

So you didn't hold the vendor to those tolerances ?

 

[EmGo]

So you didn't hold the vendor to those tolerances ?

How is he going to measure it ? Without very specialized equipment, you can't. CMM doesn't cut it.

It's like heat treating, only worse. Most people don't have Rockwell testers and have to go by what the heat treater says. If the heat treater messes up, how will you know ?

Gear measuring equipment is way more complex than that. I've only had one or two customers in my entire life who could measure the gears I've made.

Another problem is, all those numbers that software puts on a data sheet ? Pretty much meaningless. There's only a couple that actually have any effect. So that's also another layer of b.s. in the process.