View Full Version : Shaper cut gears

John Stevenson
10-13-2003, 04:53 PM

[This message has been edited by John Stevenson (edited 01-08-2006).]

10-13-2003, 06:16 PM
Keep us informed of your progress!

Forrest Addy
10-13-2003, 07:38 PM
John, I did some thinking on this and I've concluded to a certainly one could make true involute spur and bevel gears on a shaper if he were dedicated. He could also make helical gears.

The problem isn't theory but working out the details and appling them.

I try to find some links.

10-13-2003, 08:09 PM
Here is one Gear Cutting Group that might offer some insite:
Also, this Metal Shapers Group has had postings on cutting gears:
Good luck,
Jim Larsen

10-13-2003, 08:17 PM
Here are a few other URL's that are helpful in gear cutting:
This Meshing with Gears Discussion Forum is part of the above and is full of great info:
Good luck,
Jim Larsen

John Stevenson
10-13-2003, 08:30 PM
Many thanks for the links. I did know about the last three and visit them but the first one is new to me.

John S.

10-13-2003, 08:33 PM
Okay lets get it over---I am green with envy--after having a 7inch shaper for all summer, I would like a bigger one. The one you bought looks like a real gem.
So if 1) you ground a tool shaped like the tooth of a rack 2) you decided what the pitch distance of the rack was 3) you traversed the gear at the correct pitch spacing, taking several depth cuts at each space 4) indexed the gear one tooth and repeated step 3. Would you create the correct shaped gear tooth for the particuliar tooth count?
I guess I will try to use my CAD system to model this idea.

Peter S
10-13-2003, 10:08 PM
Any chance you could scan the ME article? I would like to read that one if possible.
It would be a good article to put in the YG files mentioned above.

BTW, if you don't know, there is a list on the following home page


of about 10 YG which are dedicated to running old ME articles. See each groups 'Files'. Note also, there are also "extra" groups not mentioned, eg mwhints2 and mwhints3

10-13-2003, 10:45 PM
Ever see a cutter for a Fellows gear shaper?Its a disc with many involute teeth,but as I remeber its not tooth count specific like a common milling cutter,and they go for cheap on ebay,anyway my only experience with gear cutting on a shaper has been large rack gears,but form what I have heard and read in books it is possible to cut many different and perverted gears on a shaper.

Forrest Addy
10-13-2003, 11:09 PM
I found the thread on shaper cut gears I started on Practical Machinist 5/05/03. John made substantial contributions made the time.


The idea has been around since the turn of the century. My take on the subject of cutting gears on a shaper is it's a good, practical, and accurate method but it's inherently slow for production.

If electronic indexing was coupled to a mechanical work piece roll/translation mechanism shaper cut true involute gears would be well within the capabilities of a home shop owner who's had to shy away too many times from the cost of soft steel gears from Martin or Boston Gear.

[This message has been edited by Forrest Addy (edited 10-13-2003).]

10-13-2003, 11:43 PM
Don't mean to steal the thread but I just cut two 48 DP , 20 deg P.A. gears for a model engine I'm building. I ground a 40 degree included angle lathe tool and made my own straight tooth (not helical) gear hob out of 01 and hardened it. I then set up the gear blank in one of those cheap indexing fixtures and the hob in the vertical mill. The hob has 5 teeth and the middle tooth must be lined up with the center of the gear blank. I then proceeded to cut the teeth one at a time. As one whole depth is being cut by the middle tooth, the cutting teeth above and below are also forming the gear teeth above and below. This continues all the way around the gear. It probably doesn't form a perfectly smooth involute tooth, but it is darned close and the resulting gears run pretty smooth. I first learned of this method in an article written in Strictly I.C. several years back.


10-14-2003, 12:34 AM
Sunderland made BIG gear shapers.

They used a section of rack as a cutter, and rotated the gear as they reciprocated and moved the rack. The rack generated the profile as the tooth rotated into and out of mesh.

While fairly slow, I understand they made very accurate gears, and were not really that much slower than other methods for large size gears.

The rack was working on several teeth at a time. So, it was considerably faster than most any single cutter methods, and way cheaper than a hob method.

John Stevenson
10-14-2003, 05:50 AM

[This message has been edited by John Stevenson (edited 01-08-2006).]

Forrest Addy
10-14-2003, 10:42 AM
Pretty much as I thunk it out but I don't have the electronic smarts to coordinate the step motor index with the roll.

Inherent resolution is important too. "Tenths" (microns) matter to gear tooth profiles. The index and the roll are simple math proportioning steps to index and translation to generate a true involute. I'd suggest a step rate and reduction gearing to give 100,000 steps per inch of roll and transation.

In these days of low cost servo motors (I have six three hundred watt motors with rotary encoders I bought off ebay) and Rutec or Gecko step and direction servo motor drives, US Digital feedback components, low cost step and direction drivers, only a bit of software development stands in the way of quality home brew gears.

metal mite
10-14-2003, 11:03 AM
I think they're using the rack to rotate the blank past the cutter, not cut the gear.

You have to already have a gear the same number of teeth to go this way.

How about a plate with a slot(on the indexing head) and stationary pin (mounted on the machine) to rotate the gear blank past the cutter.

Mount this plate on the back of a freely rotating head and put index holes on the blank somehow.

Have to set the pin at the proper pitch diameter in relation to the gear.


Benjamin Borowsky
10-14-2003, 12:19 PM
Forgive the really basic question, but couldn't a mill do these gears fairly easily if it were CNC'd?

Please enlighten me, in simple words of less than three sylables...


metal mite
10-14-2003, 12:50 PM
Sure, a mill (gear tooth cutter) will do just as well. It muy be better than some rube goldberg approach.

Don't see how a cnc would improve things much.

I do mill sprockets on the cnc, but a gear has too small a radius at the root.

Some of us just like to do it in different ways, like s--.

Not me, I have a selection of gear tooth cutters from ebay and flea markets.

I saw (Generating gear shape) it in a book with two bands around each side of a disk as a tensioner.

One tied off each side of the indexer.
Course you need a disk the radius of the pitch dia minus half the band thickness.

Need two bands to eliminate the backlash.

Heck, we spend days (weeks) argueing bout the error cutting a taper with the tailstock .02 below center.

Some people got nothing better to do.


10-14-2003, 02:08 PM

The Sunderland cut with the rack cutter, but yes, to do that sort of approach you have to either have the same size gear and another rack section, OR a gearbox to set the rotation vs the rack movement.

The point of the rack cutter is to cut multiple teeth at once. The same exact thing could be done with a single tooth cutter, with more back-tracking.

The Sunderland as I understand had a gearbox setup....John S can likely clear up details better than I.

Forrest Addy
10-14-2003, 03:37 PM

Yes, a CNC mill could cut excellent milled tooth gears but they would still be milled.

Milled gear teeth are a compromise. The formed cutters used cut a good involute for only one number (N) of teeth and that is in the center of the range of tooth numbers the cutter is intended for. The gears having N on either side of the best profile have error in proportion to their departure from ideal N and they consequently run rougher. There are "half" gear cutters whose numbers over-lap the "integer" cutters but consider: you need 7 to 13 cutters to cut one particular diametral pitch from 11 teeth to a rack. At $35 to $80 each that adds up into real money.

Milled gear teeth are OK for low speed (ie hand cranked or under 200 RPM) operation but above that the inevitable errors in the profile start weaving their evil spell. Because of their errors they will inevitably cog. With some careful and highly skilled handwork a milled gear could be stoned and coaxed a couple of classes higher in the accuracy scale but they're still a compromise.

A gear whose tooth flanks are a generated involute can be run to any practical RPM. The method John is discussing by its very nature will generate true involute gear teeth and the gearing thus shaped will be perhaps two or five classes higher right off the machine than milled gearing.

The other big attraction is the tooling for a "single point gear shaping" is very low in cost and easily maintained. The single cutting tool may be hand ground and it amounts to an Acme threading tool with a tip width to suit the diametral pitch and half angles to suit the pressure angle.

Besides there's a hellova challenge here for a home shop machinist. Who can resist that?

[This message has been edited by Forrest Addy (edited 10-14-2003).]

Benjamin Borowsky
10-14-2003, 04:16 PM
Thanks, Forrest. Always like to learn.


10-14-2003, 08:05 PM

Here is a link to some photos of the CNC gear shaping machines built in upstate NY and in germany by Gleason.


I work for Gleason and while I must apologize for the poor quality of the web site, the machine are pretty impressive. If you want more specific info or photos, let me know.


10-14-2003, 09:18 PM

I'll also be interested in how this turns out.

I rigged up a crude system to generate gears on my shaper but I never followed through with it after making one or two gears.


John Stevenson
10-14-2003, 09:19 PM

[This message has been edited by John Stevenson (edited 01-08-2006).]

Forrest Addy
10-15-2003, 12:16 AM
Yeah, John,

You're right on the numbering system.As soon as I read your words I recalled the lecture in cutting gears Ray McBride gave us that day so many years ago.

I feel you do need a bit more accuracy in the theoretical resolution than you would think for mechanical accuracy. 0.00025 unblended error is significant for a tooth flank. This error is insignificant in low speed gears but enough to be heard or felt as the errors come into the mesh.

Machine mechanical inputs that come in a series of stiff little steps in lieu of continuous motions as from gear trains defeats this "elastic integration". Smooth motions become bumpy and the way systems subject to stick-slip and overshoot. At least that's what I fear would happen if the gear class envelope were pushed and indicentally that's what pioneers in numerical control had to cope with when step motor drives were in vogue.

The dynamics of gear cutting integrate many errors resulting in a "fairing" of the critical curves resulting in surprising accuracy. Not every gear is destined for a helicopter transmission or astronomical telescope motion control but they should be made to a standard acceptable for the intended service and the gear maker should keep the subtleties in mind as he sets up for gear cutting.

I've had gears on the involute tester and there's only a tenths or so difference between high speed gears that run smooth and quiet and gears at are noisy as hell.

Case in point: several pairs of gears were cut with an 8 dP 20 PA stub tooth hob. The hob failed via breakage in the roughing cut with the last gear but not enough for the blemish to affect the final finishing.

The problem is there was no hob available in time to finish the gear. An expedient was devised where we used a full depth hob holding the short to stub depth and side shifting a bit to get the tooth thickness and size over wires.

That gear when run with the pinion cut with a stub hob was noisy as hell. They screamed at 1200 RPM (+/- ?) whereas the others were smooth and quiet.

Those "tenths" will haunt you: There's what's called a cycloidal tip and root relief in the AGMA standard rack (and I assume also in the ISO and relevant Brtish standards). This is a slight ramp that permits the tooth to take up and ease its load gradually and each are only about 15% of the flank face (I can't give a good figure. My handbook is a wet walk out to the shop.)

A gear cut stub tooth with a full depth hob will naturally have an extended root relief, a slightly wider bearing face, and a shortened tip relief. This displaces the bearing area up the flank. The gears showed good bearing when set-up in the testing machine. And their involutes were on the button except for the tip relief of course.

The gears were removed and their wear pattern checked. The normal polish was present but it was half the height on the flank as in the gears which were both were cut with a stub hob. The line of action consisted of a series of micro-slams instead of overlapping load sharing.

This error was enough to reject the gearing for noise and I was the guy who made a new blank in a hurry and shepherded it through QA, heat treating, and the rest while the only other suitable hob in the Navy system was flown up from San Diego over-night. THAT cost more than the $600 toilet seat.

I do ramble on.

[This message has been edited by Forrest Addy (edited 10-14-2003).]

10-15-2003, 12:40 AM
Hows about the gear shaving machines that used to be used for automotive type gears....Red Circle or some such name which escapes me.

That whole process was a process involving a skew "gear" that was really a cutter. It meshed and shaved along the tooth like a shaper due to the skew.

If I recall, it was a finishing operation that gave a very accurate gear. Avoided grinding.

John S....that Sunderland operator must have been a busy dude........I wouldn't even want to have to stack all those gears, to horse them in and out of the machine and get them mounted would be a pain in the back..

I assume they were cut in "sticks" instead of one at a time........

John Stevenson
10-15-2003, 05:20 AM

[This message has been edited by John Stevenson (edited 01-08-2006).]

metal mite
10-15-2003, 08:49 AM

What you propose should be entirely possible.

Probably have to retrofit the shaper table with ball screws, or at least antibacklash nuts.

Have to make a cnc rotary head with an indexing mechanism built in.

None of it is too complicated.

Even the programing isn't too complicated.
Just ratios between the feeds.

If you want a fully automatic cycle, that complicates things some.

I like the idea with a pin in a slot better
anyway but I'm an old timer.

Just can't be tripping over knat tards or you'll never be done.

How many guys on here can measure a tenth reliably anyway.

A little lapping compound works wonders.

I've seen submarine drive shaft couplings lapped in a huge home built lapping lathe to remove the hob marks.


John Stevenson
10-15-2003, 11:04 AM

[This message has been edited by John Stevenson (edited 01-08-2006).]

Forrest Addy
10-15-2003, 11:13 AM
John, being my one and only irrepresible devil's desciple self I was pointing out the limitations compromises in accuracy impose on gearing.

There's bound to be interest whenever a revelation and simplification of gear manufacture combined with an order of magnitude reduction in tooling cost is made. About about half of the world's home shop machinists will spring to quivering attention and ask: could I make new gear for my boat's out-drive, race car car transmission, motor cycle? My answer is yes - but...

Hence my fussing about limitations.

Moving a 1800 Lb shaper? Just enough to induce a pleasent sweat to a man of your talents. I moved my 24" shaper in my 3/4 ton pick-up - with 90 PSI air in the rear tires at speeds under 25 MPH (ever hear of really scary roll coupled oscillation?).

Incidentally: most "herringbone" marine reduction gears were technically double helical gears cut on a hobbing machine. There had to be a space between the two rows of teeth for the hob to run out. True herringbone gears where the teeth intersect in a sharp apex were cut in this country on Farrel-Sykes horizontal gear shapers. The shaper cutters looked like fellows cutters but they were equipped with sophisticated helical guide bushings. The shaper cutters went back and forth alternately as the gear cutters and the blank were slowly rotated in mesh by the usual timed grear train. Each cutter cut to the center and reversed back so precisely there was no burr or over-cut at the center.

A factoid sticks in my mind: a matched pair of F/S cutters for a 3 dP 20 degree Pa 30 degree helix herringbone gear having a 6" face costs the same as three Cadillacs.

The cutters had to be purchased in matched opposed pairs and their guides were made on a Farrel-Sykes horizontal gear machine.

Which leads us to: how did the first set of F/S guides ever get made?

[This message has been edited by Forrest Addy (edited 10-15-2003).]

Herb W
10-15-2003, 12:40 PM
I'm one of the home shop guys who's following this topic with much interest, though not quite at the quivering stage http://bbs.homeshopmachinist.net//wink.gif

Being able to cut the full range of tooth numbers with a simple cutter is certainly an attractive proposition...particularly for those of us who only need to cut a few change gears etc now and then and wouldn't get good value from purchased form cutters.

I've just assumed that gears for car transmissions and the like were beyond what could be done in a home shop setting. One could make gears that would 'look' right, but aside from the issue of precision, since were talking about making unhardened gears, I presume they wouldn't last very long.

So, just what are the limitations for these proposed shaper cut gears? Just how demanding an application can we expect them to be able to handle?

For applications where hardened gears are required, what's the procedure? Are they roughed out, then heat treated, then finished by a grinding operation?

As for shaper hauling, my work truck is an ex-military chev c30 'CUCV'. Dealers that get them like to advertise them as 1&1/4 ton pickups...where they get that from, I have no idea. I just consider it a good sturdy pickup. Anyway, when I hauled the shaper home (24" Alba) it was the first time I had enough weight in the old beast to make it ride decent...but then at 3500 lbs, it's a lightweight for a 24.

10-15-2003, 03:06 PM
In reply to your question; "How did the first set of F/S guides ever get made?'
Never having seen these animals I can only guess but I feel confident saying that the first ones were made the way the very first one of almost machine toolwas made; With a file...and a lot more skill and patience than I'll ever have.

10-15-2003, 03:43 PM
Same old story. All new machines are made with used machines having wear.

10-15-2003, 07:05 PM
BTW, I have shaper-cut gears, on an Atlas, and it worked quite well. I didn't use the generation method, I mathced the profile to an existing meshing gear of similar size and went on my merry way.

Worked fine.

I would sure like to do the generation method.

Oh, I would suggest doing the rack cutter, even at the cost of an extra step. it takes a LONG FRIGGIN TIME to cut each tooth in steel. The rack would be cutting on more than one. If you can keep "outline chatter" down it will be a lot nicer

10-15-2003, 07:33 PM
One company in California CNC mills large DP gears for downhole oil pumps. They have spent millions developing and perfecting the programs to do this on their Haas mills - which was cheaper than shipping the huge parts to Canada to be cut on a shaper.

So it is possible to produce high quality gears that do not need to be lapped afterwords on a cnc mill, but it is unlikely that it will ever be done in a home shop with out the involute form cutter unless someone is willing to put in years of development time.

John Stevenson
10-15-2003, 07:54 PM

[This message has been edited by John Stevenson (edited 01-08-2006).]