What's new
What's new

Gear cutting machines with pitch error correction mechanisms

I've been curious about the gear grinding machine with the grinding wheel shaped like like a worm, or hob. Just off the top, the difference between optimum surface feet/minute between high speed steel cutters and grinding wheels is hanging me up.

Now I'll admit lack of any experience with gear grinding. (Examined a "gear shaving" machine years ago) My first expectations would be either a wheel dressed to the tooth space form and grinding one space at a time accurately indexed, OR a rack shaped grinding wheel in which rotation of the work and translation of wheel would generate the correct shape, but again it's one space at a time.

I'm imagining a grinding wheel in the shape of a hob to continuously generate tooth shapes would be really singing.
 
I've been curious about the gear grinding machine with the grinding wheel shaped like like a worm, or hob.
There's three basic methods ... one is the formed wheel grinds space method, that one is easy to figure out, it's just like the cutters people use on horizontal mills. That method was always kind of clunky before but it's made a giant leap with cnc dressers and indexing. Nowadays it's one of the fastest and most accurate methods. Youboob; hofler and niles would be where to look. The method has been around for ages tho, in the US Red Ring and Detroit were machine tool builders up until the sixties. .

Another is the threaded wheel type, was the fastest for a long time, if you look up reishauer on youboob should find quite a few videos demonstrating it. Works just the same as hobbing. I don't think anyone in the US has ever made these.

The last has been mostly retired but it was the most accurate when machines were mechanical. There's either one or two wheels with straight sides and the workpiece is rotated as it traverses past the wheel. Then the part indexes, repeat, repeat. In the highest accuracy versions of that, the wheel only touches the workpiece in a very tiny area. Maag (Switzerland) was a major builder worldwide. In the US Pratt & Whitney built two different versions of this. The gears in US-built radial engines were ground on P&W machines (duh :)).

(Examined a "gear shaving" machine years ago)
Totally different idea and purpose, only done on soft or pre-heat-treated gears. It's not grinding, it's a finishing process which removes a tiny bit of metal off the flanks. That's kind of gone away for a few reasons, which we can skip unless you are especially curious.
 
I owned a transmission shop for a few years when starting out. There are more people grinding gears out there than you think.
 
I looked at the spare parts documentation for several gear hobbing machines and compared the claimed accuracy of these machines (based on accumulated tooth pitch error) with the accuracy of their main worm wheel. On average, the difference is about 2.5 times! To make a wheel with an error of 1 minute, we need a machine that will have a wheel inside with an error of 20 seconds! How to make this wheel?
One "cheat" may be the size of the worm wheel in relation to the workpiece. For example, when I shape a 4" diameter spur gear, the only gear in the machine that is directly connected to the workpiece is about 24" in dia, all the rest turn several times faster than the workpiece. The other worm wheel is about 12" dia connected to a 3" dia cutter. So the cutter and workpiece errors are reduced by the size difference.
 
My first expectations would be either a wheel dressed to the tooth space form and grinding one space at a time accurately indexed, OR a rack shaped grinding wheel in which rotation of the work and translation of wheel would generate the correct shape, but again it's one space at a time.
A friend and supplier bought a CNC gear grinder about 20 years ago, and it worked sort of like a high speed hobber. It was $1.2M back then. The wheel was dressed by a rotating wheel with profile diamonds mounted in it, they had an involute shape i think, they at least were curved. The control fed the dressing wheel across the spinning grinding wheel to create the rack-like profile in the wheel, then the wheel was transversed along the length of the blank as both turned, much in the same manner as a hob is transversed along the length of a blank in a hobbing machine. The dresser did not have the tooth shape, it was a generic curve that the control created a toolpath for to generate the needed shape on the grinding wheel.
I got to see it running but I didn't get to see parts changed or the dressing done. I wanted to see how parts were indexed in the work holding, how it dressed a wheel etc, it even changed wheels like a VMC changes tools. The owner sold the shop and moved to ski country before I did.
 
I saw a manual gear grinder that used a flat sided wheel to grind gear teeth, much the way you can make an involute gear with a slitting saw. It turned the blank as it fed the wheel like a rack cutter, then indexed the gear and ground one side of the next tooth. was a largish machine, for 10"+ dia workpieces. was 30+ years ago at an auction in one the the former gear houses in Phila. May have been Reishauer. What i was most interested in was that it used tensioned sheetmetal strips attached to a wheel to push and pull the wheel which turned the workpiece, absolutely no backlash or gear error between that and the workpiece arbor. Wish I had a photo.
 
Thanks. I watched a Reishauer video of what was probably the most recent development. It shows the gear being automatically loaded and clamped. Then a non contact probe comes in as the gear rotates, pauses and retracts. Then the gear is ground as you describe. I suspect the probe magnetically sensed tooth locations so it can be synchronized with the grinding wheel. Then it's all electronic gear box.
 
  • Like
Reactions: Mud
One "cheat" may be the size of the worm wheel in relation to the workpiece. For example, when I shape a 4" diameter spur gear, the only gear in the machine that is directly connected to the workpiece is about 24" in dia
Yes, but if the machine’s characteristics indicate a kinematic error, for example, 100 seconds, then it will always be 100 seconds. Well, or less - this is the maximum permissible value. As a rule, the diameter of the main worm wheel on hobbing machines is slightly less than the maximum workpiece diameter for this machine. So the difference in diameters doesn't help much here.
 
Then a non contact probe comes in as the gear rotates, pauses and retracts. Then the gear is ground as you describe. I suspect the probe magnetically sensed tooth locations so it can be synchronized with the grinding wheel.
On machines before CNC it was simply possible to independently tighten the workpiece a little without rotating the grinding wheel - so that their relative position became correct.
 
I slowly continue my search for historical records. Alas, all the hopes associated with specific specialists lead to the fact that the people who worked with suitable machines have already died :(
A little information.
The Craven Brothers company, which also produced gear hobbing machines, registered a patent in 1937 for a device for correcting pitch errors of the vertical movement screw of the spindle:
This is not exactly what I'm looking for, but I think it will be interesting.
In addition, I finally found a non-Soviet gear hobbing machine without CNC and with a kinematic error correction system! This is a UPH 30 from David Brown.
Thanks to an incredible miracle, I found what seems like complete documentation for it, if anyone is interested, read it here:
There are documents in Russian and English, the name makes it clear what is what. The correction system in this machine is not mechanical, but electronic - resolvers, inductosyn and vacuum tubes. And this was in 1962!
I also read in one of the books that among the machines of the Schiess (or Schiess Defries) company from Germany there were models with correction devices. But surprisingly, I could not find a single catalog/brochure to see a list of their machines. If anyone has a catalog, I will be very grateful for it.
 
What i was most interested in was that it used tensioned sheetmetal strips attached to a wheel to push and pull the wheel which turned the workpiece, absolutely no backlash or gear error between that and the workpiece arbor. Wish I had a photo.
I actually used that technique for a job about 10 years ago to generate a large involute cam surface on manual machines. "Large" meaning about DP 0.25 with a pressure angle of about 40 degrees. It turned a tedious job of precise handle cranking to match precomputed profile positions into a simpler job of turning an accurate base circle disk, with compensation for the thickness of the metal strips. I roughed the involute surface out on the horizontal mill, then used the same fixture/mechanism to finish it on the surface grinder.
Caution: If you need more than one turn of the base circle disk (shouldn't!), you should wrap the strips helically around the base circle disk (cylinder) so they don't overlap, changing the effective base circle diameter as a step function of distance traveled. But a helical wrap also changes the effective base circle diameter, although it's a constant change, not a step change, so you can compute a compensated base circle disk diameter.
 
I actually used that technique for a job about 10 years ago to generate a large involute cam surface on manual machines.
Maag HSS grinders are all that way. I haven't looked at the SHS ones too much, they may be different. The very last Maags even had a mechanical differential for the bands so that you could get by with fewer pitch blocks, the mechanism adjusted for the size differential.

There's a small Kolb that may use pitch blocks and bands also.

Reishauers are all hobber-style worm-type grinders, they are proud of the fact that they invented that.

You're probably going to find more unusual mechanisms in testing equipment. People did much more unusual things with one-off measuring equipment than they did with production cutting machines - and there's a lot more places with five guys in white suits who are interested in that in the measuring tool world than in a place making 40,000 lb production gear machines.
 
In addition, I finally found a non-Soviet gear hobbing machine without CNC and with a kinematic error correction system! This is a UPH 30 from David Brown.
Milling man,

I have been looking through a couple of books I have on the history of the Marine Steam Turbine and read about David Brown making a 4.2m bull wheel (double helical second reduction gear wheel) for the de Laval Steam Turbine Company, Nacka, Sweden. This was for a then large oil tanker Josefina Thorden launched around 1955. De Laval's gear shop was unable to machine a gear wheel this large.

The books author (Ingvar Jung) was the chief engineer of de Laval from 1950 and writes "those who were there at her creation will never forget her". There follows a harrowing story of gearing problems, not caused by de Laval but by the ship builders.

Although the book is about marine turbine history, it became clear from the early days that reduction gearing was necessary, first single reduction, later double reduction. Making accurate gears was a problem from the start and most of the turbine builders had to develop their own machinery. For example, de Laval, Parsons, Westinghouse and General Electric all developed their own cutting or hobbing machinery including various methods of improving accuracy.

For example, Parsons invented the "creep method" for cutting his reduction gearing (presumably some time after 1909):

The Marine Turbine vol 1 pg 41a.jpg The Marine Turbine vol 1 pg 42a.jpg

If you like I can post some pages which mention gear cutting, however I can start another thread if you prefer.
 
Last edited:
I saw a manual gear grinder that used a flat sided wheel to grind gear teeth, much the way you can make an involute gear with a slitting saw. It turned the blank as it fed the wheel like a rack cutter, then indexed the gear and ground one side of the next tooth. was a largish machine, for 10"+ dia workpieces. was 30+ years ago at an auction in one the the former gear houses in Phila. May have been Reishauer. What i was most interested in was that it used tensioned sheetmetal strips attached to a wheel to push and pull the wheel which turned the workpiece, absolutely no backlash or gear error between that and the workpiece arbor. Wish I had a photo.
I have been wanting to build a similar device for grinding gear teeth on a bench grinder. But rather than grinding an involute, rather a simple radius chosen by drawing a circle on the gear in a cad program, to minimize the error from the involute. The error on a 24 tooth 20 degree pa gear is about 0.1 degrees with the gear too fat in the center of the tooth, which is easy to lap out. The origin of the radius is a little outside the base circle.

One of the amature gear cutters in a book form 50 years ago has some tables you can use for making gear cutters from a double sided button cutter. My method should be more accurate than his.

Anyhow, the issue is the grinding wheel is going to cut deeper into the root of the tooth in the middle of the gear. So for an 8" wheel the practical maximum gear thickness is about half an inch.
 
rather than grinding an involute, rather a simple radius chosen by drawing a circle on the gear in a cad program, to minimize the error from the involute.
Research 'cycloidal gears'. They have advantages and disadvantages.

@Peter S : If your book is correct, Parsons and deLaval were idiots. All those problems were solved long before 1955. Loooong before. If they were too cheap to take advantage of what was already available, what can you say ? And his creep thing is stupid. It does nothing that setting up for an extra tooth then helixing backward to the number you want doesn't already do.

All of that described in the attachments is stupid. Maag built shapers to 15 meters and they are way accurate over some cockamamie bullshit built by a wacko.

Gears and cams both seem to attract wackos.

Wait, let me rephrase that :D
 
Last edited:
@Peter S : If your book is correct, Parsons and deLaval were idiots. All those problems were solved long before 1955. Loooong before. If they were too cheap to take advantage of what was already available, what can you say ? And his creep thing is stupid. It does nothing that setting up for an extra tooth then helixing backward to the number you want doesn't already do.
EmGo,

The 1955 reference was only to do with David Brown, that company being mentioned by Milling man and I thought maybe not known by some PM members (except maybe for their tractors and Aston Martin).

The author, being the chief engineer at de Laval was probably not an idiot. The turbine company's I mentioned developed their own gear cutting methods in the early years of steam turbines, so early 1900's or earlier in de Laval's case.

De Laval was building steam turbines from 1893 and cutting accurate high speed gearing from 1895 for land-based geared turbines.

Parsons built their first geared turbine before 1909 and I guess their creep feed method shortly after that. It was no doubt superseded when better methods were invented.

Keep in mind the high input speed, high powers and yet the need to keep everything as light as possible while being attached to a possibly flexible foundation (ship hull).

The 1955 gearing mentioned above failed because the prop shaft thrust bearing was mounted to a weak foundation. This caused "resonant axial vibrations of the bull wheel, excited by the propellor". Result was fatigue failure of the gear teeth.
 

Attachments

  • The Marine Turbine vol 3 pg 87a.jpg
    The Marine Turbine vol 3 pg 87a.jpg
    746 KB · Views: 7
I'll add a few pages from The Marine Turbine, A Historical Review by a Swedish Engineer Part 1, The Days of Coal and Steam 1897-1927 by Ingvar Jung which gives a brief overview of the early days of turbine gearing. This book is not trying to give the history of gearing, so no doubt more could be added.

Milling man,
If this is disrupting your thread, please let me know and I will delete these posts.

The Marine Turbine vol 1 pg 37.jpg The Marine Turbine vol 1 pg 38.jpg The Marine Turbine vol 1 pg 39.jpg The Marine Turbine vol 1 pg 40.jpg The Marine Turbine vol 1 pg 41.jpg The Marine Turbine vol 1 pg 42.jpg
 
Last edited:








 
Back
Top