Gear Cutting

[chuckknigh]

It seems to me that a common use for CNC setups is cutting gears. Seems like a simple way to do this would be with a 2 axis machine...but not a traditional X-Y.

I'm picturing a dedicated machine, with a single linear axis, and a single rotary axis. I saw something similar used for making router bits, in an old "This Old House" episode.

Basically it tkes a blank and spins it along the rotary axis. A cutter head moves in and out, radially, cutting the teeth and gullets of the gears.

Since it's so simple, it should be cheap and easy to build. And, it would be a boon for the aspiring clockmakers. "Huge" gears could conceivably be made, easily, on a benchtop sized machine...a 12" gear, for example, would take up 6" side to side, plus the length of the linear axis.

Now...the real question. Would any existing software be able to drive it? I guess if I write a simple script to generate the G code directly, it wouldn't be too much of a problem.

Comments? Suggestions?

-- Chuck Knight

[sol]

I spent the better part of today playing with your idea after reading your post in the other thread...even made a mock up with a slow rpm motor. The problem I became tangled in was trying to get the curve of the flank/face's involute correct. If that is not that important then maybe simple x,y,z software can be used where X is the linear, Z is up and down (if it is even needed) and Y is the rotary table with the distance actually in degrees though read and written as inches... just so that one rotation of the table equals some multiple of 360 (like 720 +, to match stepper's gearing and resolution)

In short, you draw a gear, define points with X=distance from center/origin, Y= angle from origin, straighten it to a rack, g-code it, but then cut it with a rotary table.

Again my problem with this cobble was the gear faces were not pretty, one side more convex than the other..then I ran out of time and gumption because I realized I was literally reinventing the wheel. There is bound to be a straight forward code generator for this but a cursory Google-ing found nothing.

Would love to hear other ideas!

[HuFlungDung]

Chuck,

I believe that the machine you describe would be called a gear shaper. It uses a hardened master "gear" to function as a cutter on the reciprocating cutterhead. The side of the gear tooth profile is the cutting edge.

The slow rotation of this cutter head at the proper ratio to the blank is most likely set up by a mechanical gear train. Anyways, this allows for one cutter to cut all gears in one Diametral pitch category.

Although you could "cnc it", there would not be a whole lot of advantage, since the rest of the machinery needs to be built just to hold the work properly to do the cutting anyways.

In the hobbiest category, I have thought about using a so called "slotting attachment" on a cnc mill, but never actually got around to doing anything. The slotting attachment is simply a small reciprocating "shaper head drive" for use on a milling machine.

The toolbits for this type of attachement are simple "lathe bits" which you can grind yourself. Most guys could figure out a way to grind a simple angled toolnose, similar to an Acme thread tool. This would serve as a single "rack gear tooth" which you would use as a cutter. In the involute gear systems, any gear of a given Diametral pitch (Dp) will mesh with all others. This means that, given the correct motions, a gear rack tooth (straight gear in other words) will generate the correct profile for any number of teeth on a correctly sized blank. The trick here, is that you would need a 4th axis rotary table to rotate the blank, whilst you program one axis of the table movement to simulate the rolling motion of the gear blank past the cutter, which would be reciprocating up and down in the Z plane.

Clear?

[chuckknigh]

I've actually played with gear hobbing, which is essentially a variation of what you've described with the "rolling gear." I'm talking about a simple "straight router bit" type of cutter head that would move in and out, with respect to the center of the blank.

It could theoretically cut any DP and any tooth form, given the right rotational data, and that includes undercutting leaves on pinions.

-- Chuck Knight

[ESjaavik]

Of course this is possible if you use a bit not larger than the root radius of the tooth. That may amount to a problem. As for the path generation software it does exist. I visited a friend of my friend who have a wire EDM. (In his hobby garage! I'm still green of envy.) He showed me some gears he made. They were small, down to around 20mm. And some were not even round. They had strange shapes, but still could be put on shafts and merge perfectly as far as I could see and feel, although of course with constantly varying output speed. I don't know what software he used to generate the cutting paths, but could ask him.

But I'm going for the hobs. That may even be a good "first CNC" project. You need only one stepper/servo driven axis and that is the rotary table. Then you need an encoder on the spindle. A horizontal mill is the best, but a sturdy vertical would probably do for smaller size gear modules (pitches). Then the software will have to do the job of keeping the gear blank in a constant relation to the spindle. In other words a software gearset connecting the spindle with the rotary table. Then you can generate any number of teeth without having a large number of changewheels. Not to speak of the contraptions necessary to connect these between the spindle and the rotary table.

Feeding the cutter into the blank can be done by hand, auto feed or CNC, whatever you have. The reason I think this is a good first project is that you need only one drive and stepper or servo. No expensive ball screws and linear rails. Still there will be little or no compromize on the quality of the end product. It will familiarize you with the electronic parts and connecting them up. It can also be used for a number of other rotary table tasks. And if you go on to a full CNC mill, it can be integrated as a rotary axis. If you find out this electronic stuff is not for you, the rotary table is still a nice tool even when hand cranked, and almost mandatory for any gearmaking.

[foamcutter]

Ok guys,
Lets see some drawings of this contraption! Sounds interesting and useful.

[balsaman]

Any cnc with a 4th axis can do this. Any good CAM software can do this too.

Eric

[HuFlungDung]

The method I was thinking of bypassed the necessity of purchasing an expensive hob, because I would be using a simple flycutter ground from a $5 piece of HSS. The advantage would be that the cutter can be very accurately made, compared to handgrinding an involute form toolbit to match a particular gear tooth, which always requires a good sample gear (made by someone else) to use as a guage. I have used this method frequently to flycut gears and involute splines when a regular gear cutter was not readily available.

Hobbing requires the addition of an inclinable table axis, since the helix angle of the hob itself is seldom desirable in the finished gear. It is true that the hob can also cut helical gears, which the slotter-type setup I described above would not do.

Thin gears (perhaps up to 1" thick) can be very nicely roughed with a laser. A light finishing pass with a machine tool would be all that is required.

[chuckknigh]

Yes, but not everyone (and especially first timers) wants to go to the time, the trouble, and quite bluntly the expense to build a 4 axis machine...or even a 3 axis one. Especially if all they need is some gears. I mentioned this concept in the "Open Source" thread, and decided to mention it here, for debate.

I'll draw up a quick picture, but I can't do CAD very well. It'll probably be a hand drawn sketch, since that's truly quick and easy for me. Don't look for real details, in version 1. ;-)

So, let's go with some "off the shelf" pictures from the web.

Picture a stepper motor driven mount for holding a gear blank, probably patterned after this:



Just for reference, this mount is from Steve Bedair's site, and is used for hobbing telescope worm gears on his lathe. Works quite well from the looks of it.

Now, axis 2. We'd have a single linear axis, which would move a router or Dremel (with a straight cutting bit) in and out from the central point of rotation. As the gear blank rotates, it would cut the teeth. Reversing the rotation would allow the undercutting of leaves on pinions, if desired.

Shouldn't be too hard to build, really...and the cost would likely be minimal. Even a short linear axis, offering 6" of travel, would allow a 12" diameter gear to be produced...pretty big, for being made in a home shop. There are few gears in a clock bigger than 12", and that includes tower clocks!

Now, granted, this'd be a limited use machine, but aren't most "dedicated power tools" designed primarily for a single use? I know one of the strengths of CNC is its flexibility, but not everyone wants or needs that flexibility, and quite frankly, that cost.

-- Chuck Knight

[HuFlungDung]

As Einar pointed out, I don't really know what radius of tool you would be able to get away with, since there is not much room at the root of a small tooth. This is why lasar or wire edm is a good choice, because it can form extremely small fillets, and is non-contact to boot.

The main issue that I see, is generating the correct tooth shape. By generate, I mean using a simple cutter moving in a logical path that generates the involute itself.

Even in the simpler mechanism that you describe, does the rotation of the gear blank not have to be electronically cammed to the tool bit position? I think so, and this is a fairly complex cnc issue. It would be far simpler to make one tooth at a time, on a simple manual rotary table which you index for every tooth. A small XY cnc program takes care of the motion, and a simple cadcam toolpath takes care of the tool offset radius. This would enable quite a small cnc machine to cut large gears, since it only handles a single tooth.

[foamcutter]

Chuck,
Hand drawn sketches are the best kind. Besides I'm with you on the hand drawn being faster than CAD for some of us. Ron

[chuckknigh]

Of course the rotary axis would be controlled electronically, by stepper motor. I just posted that pic to show what I was talking about, with a gear blank mounted on a rotary axis.

You're right, of course, about the tool diameter being smaller than the root circle for the gear. Of course, you can get Dremel compatible carbide endmills in sizes of .015" and smaller, from any clockmaker's catalog, so the tool size should not be much of an issue.

Really, this is just a thought experiment, trying to deviate from the traditional X-Y-Z arrangement. I think it would work.

-- Chuck Knight