What's new
What's new

Making Q10 gears with a 4/5 axis machining center

Mark Winsor

Feb 4, 2013
Hartford, CT, USA
Is it possible to make Q10 spur gears (say 2mod, 1.8 inch pitch diameter, carburized 8620) with an accurate 4 axis machining center? The tolerances work out to about +/- .0002" all over the tooth profile, and this would involve hard milling after the carburizing. It's definitely a challenge, but I thought I'd check if anyone has tried.

More details on the motivation: Gear specialists take somewhere between 10 and 18 weeks to make 3 gears, too long for me. My understanding is that the lead times are long because

1) They're busy/focusing on larger customers
2) I use unusual pressure angles and root geometries, requiring special tooling, which also has a lead time
3) Many don't do their own heat treatment/ID grinding/etc, and those extra processes have lead time

Another frustration is that with the exception of EDM'd gears, which don't work for other reasons, it's near impossible to make a CAD model which matches the finished part +/- .0005" on the entire tooth, and I have a need for that. The reason for this is because the finish shaping/grinding machines do not follow a CAD model, they follow a bunch of numerical values from a gear design table and it's difficult to figure out how these values are interpreted to produce the final part. It also has to do with the design of the tooling, and I have not had good luck trying to figure all this out.

I'd love to make gears just like any other part, where I get something back in 4 weeks which matches the CAD model provided, and meets the tolerances on the print. If anyone thinks this is possible and is up for trying, please PM me or reply here :)
Wow- where to start? Gear teeth of that quality are generated and not formed by a cutting tool following a driven path from a programmed profile. Essentially the process of generating the involute profile on the tooth flank is an inherent part of the gear hobbing, shaping or grinding machine. Some grinders do form a tooth space on the wheel that simply reproduce the wheel shape on the tooth flank but they are not that common. ALL type of gear machines pretty much rely on creating the involute profile by rolling out a rack tooth in one form or another. You might Google "involute curve generation" to get a better idea of how this works.

Are making Q10 gears for a specific application or so you just think you need them? Having the analytical testing equipment to confirm they are actually Q10 gears is another matter. Another issue- you may well have an AGMA 10 gear after hobbing, but if its carburized 8620, it may not be AGMA 10 after HT without grinding.

If ya just need close to a Agma 10 there are other materials that are much more stable than 8620 where you won't get much HT distortion. Do you need full Agma 10 or just tooth spacing accuracy or involute accuracy- lots of questions I know.

There are a few of us on here that can cut a pretty accurate gear, and have been doing it for a long time, but could not meet the exact requirements of a full blown AGMA 10. personally if I have that kind of job, we cut the teeth, do the heat treat and send it to another shop that specializes in gear grinding to do the step

Actually you will not find a properly specified gear that requires a profile tolerance to a cad model but rather to AGMA specs with regard to involute profile,lead tolerance, tooth thickness, tooth spacing and runout.

Hope this helps a little. Dan
Thanks for the input- to answer some points- Q10 gears are required for the application according to a gear consultant hired to analyze the application. I'm the engineer integrating them into the rest of the system and dealing with the manufacturing. There are plenty of gear inspection shops, so measurement isn't a problem, that only takes a few days if the work is planned in advance. The 8620 gears are currently ground after heat treat for the reasons you mentioned. I'm vaguely aware of how hobbing and other gear manufacturing processes work, but none of the above really answers my question.

Why can't gears be 3D milled on a 4 axis machine and a ball endmill? I'm used to working with Haas equipment, which isn't accurate enough for Q10, but I'm guessing some of the higher end machines could be. The reason for my question is because there are probably 20X more shops out there that can 3D mill vs. those that make gears, and it would really expand my options.

As for a "properly specified gear", I could care less if I'm properly specifying or not as long as I get a profile that's accurate to +/- .0002", which is a summary of what Q10 means in the size range in question.
Last edited:
Hard milling using small diameter, relatively flexible endmills, is going to have you chasing your tail trying to hold that tight tolerance, IMO. The scallop height restriction itself would require an ungodly amount of time on the machine. You'd likely have to probe frequently to figure out if the tool wear is properly compensated.

While I'd also like to have the ability to 3d mill gear teeth, I don't think that this would be a time saver. I've cut a few worm gears with this sort of approach, and regardless of the fact that I wasn't obtaining those tight tolerances, it took many hours to machine a single gear. A hobbing machine would probably cut it in 10 minutes, and it would be better quality.

I have a fair amount of experience, here. Specifically. Short answer - no. Not going to happen. I am happy to entertain an extended conversation and all the factors of why, right here. But I also know what you are going for and the things you are encountering with other places. Based on your call out, I expect that you are using the old AGMA standards, not the new ones. How do you expect to measure these?

EDIT - Disregard, I saw your other post now about sending them out.
Last edited:
Why can't gears be 3D milled on a 4 axis machine and a ball endmill?

For the exact same reasons that I have written about many times with regard to why they usually cannot be made well via WEDM. Because quite simply, the machines are not capable of the motion used to create them properly, nor are the softwares used for programming them to do so.

Because of the Involute Shape.

Do a search on that and my name and you should find your answers, directly.

As for a "properly specified gear", I could care less if I'm properly specifying or not as long as I get a profile that's accurate to +/- .0002", which is a summary of what Q10 means in the size range in question.

In a simplified explanation, yes that is true. In a micro examination of the conditions, that does not hold up, linearly.

The rotaries have errors. The linear axiis have errors. The spindle has errors. And all of these combine to create LARGER errors.

That does not even address the concern over linking the various motions. Most commonly done with gearing, but also done with servos in modern equipment. Point is, they are purpose built machines, that DO link them properly. Unlike normally common manufacturing machines.

Do you really think for a minute that all the bigger machine company's are not all over this? :)

I spent a week down at Mazak and they were all abuzz about their newest offering for Gear Cutting, claiming exactly what you are looking for. I know just a little bit about this, and actually bumped into the head application engineer for the program while there. We had a fairly lengthy discussion of the details and capabilities and when presented with real questions even he acquiesced that they cannot ACTUALLY hold those tolerances. They are a best case, perfect world scenario. And they cannot do it upon demand. Period.

That said, there ARE ways around it. Two that I am aware of. It is a subject that I have been working on for years, and have recently made good progress in. And, no. It is not subject to disclosure at this time. :)
Thanks guys, that makes sense about the scallop height and tool wear, I have not tried programming a toolpath so I don't know what it would take. That's why I was asking if anyone had tried. Could this problem be solved by using a custom tool that has the profile of the tooth? It would look something like a hob, but still allow an accurate milling machine and rotary device to do the job. I feel like the tolerances could be held this way, at least for softer materials.

As for EDM, there are plenty of places cutting Q10+ gears using EDM, the issue with that is that EDM lacks the ability to crown teeth, and leaves a couple layers of bad surface finish, decarb, and hardness differences that may need to be dealt with before running, depending on who you believe.

WJR- I will be in touch at some point with my prints, to see if you're interested in helping. I wish you luck on your new methods, as the gear industry leaves something to be desired from the perspective of a generic engineer. Maybe a good opportunity for change