PDA

View Full Version : Gears, gears and more gears



John Stevenson
09-25-2010, 05:39 AM
Been working with Art Fenerty of Mach3 fame on his new gear-cutting program.
Basically brought about because he wanted to do clock gears and then it got all out of hand with eccentric gears and all that artistic crap.

Link to the site at www.gearotic.com (http://www.gearotic.com/) there is a demo copy but the tutorials are worth watching.

Anyway I wanted to get the program back to making mechanically perfect gears and bring in some ideas. Most of the cutting is for routers making thin gears with a vertical spindle but there is an option for doing spur and helicals travelling across the face.

Now for the good bit even though CNC is needed the gears are cut with a conventional off the shelf $3 end mill and that end mill will cut any gear provided it can reach the root, don't believe it, watch the video.......

http://www.youtube.com/v/qZJ95I3ZWro&

Not a great picture, the white nylon doesn't help but I'll redo it later on brass or similar. Main thing is it shows the operation, the cutter blocks a slot out and then by using all 4 axis at the same time it moves away, raises the cutter and rotates the 4th axis so each pass is at a different angle and part of the involute. You have the choice of selecting how many passes per tooth.

But because specials cutters are not needed it's now very easy to modify gears to fit an application. I cut two 14 tooth gears to text book specs, mounted then on two pins the required distance apart and they ran perfectly, I then made two more identical blanks but cut 13 on one and 15 on the other.

http://www.stevenson-engineers.co.uk/files/gear_cluster2.jpg

You can swap these in any order and they run fine. Helicals are still a work in progress, they work but the tooth shape is not correct yet.

http://www.stevenson-engineers.co.uk/files/helical_2.jpg

Again simple tooling, in most of these cases the tool was a throw away FC3 3mm end mill costing 2.99

[Edit]

Forgot to add but the program will be a work in progress, if there is a demand for an application then it may be added if possible. Bevels are definitely on the cards but the program is for general release first.

motorworks
09-25-2010, 07:04 AM
John
Great work!!
I will be sure and pick that program up....
and it great to see a GOOD post here!! LoL
eddie

jimmstruk
09-25-2010, 01:00 PM
Thanks John, its good to see some chips and a real worthwhile project and some fresh ideas! JIM

Smokedaddy
09-25-2010, 03:31 PM
Hi John,

Looks great to me. The video is fine, just a touch out of focus or a touch to close in filming it. I would love to be able to make something like this.

As always, thanks for sharing,
-SD:

small.planes
09-25-2010, 04:37 PM
Your back from the pub then.

Dave

motorworks
09-25-2010, 05:01 PM
"Your back from the pub then."

He never left the pub. His wife dropped off the laptop and some extra money....got to love the gal :)

eddie

John Stevenson
09-25-2010, 05:47 PM
Been back 10 minutes and playing with this program again.

Because everything is done in software you can play "what if ?" to your little hearts content.
Not quite what you want to fit in? How about a 18.5DP cutter or better still to get bang on a 18.67564 DP ?

Sooooooo.

I drew a 40 tooth XL series timing pulley up in CAD [ with the help of Mr SPI Industries :rolleyes: ] and got some measurements off it.

They went back to Gearotic [TM] and juggled some figures.

http://www.stevenson-engineers.co.uk/files/XL%20timing%20pulley%20gearotic.jpg

So if we use 16.25 DP we get an OD of 2.5240, we need 2.5240, OK so far

Then play with shift, stubbing and tooth width we get a total depth of 0.0571, we need 0.0550, 2 thou too deep, think I can live with this.

Tooth width is 0.1242 and we need 0.1240, even closer.

Now bung the dxf onto the original drawing and it's within a gnats left bollock.

Only fly in the ointment is it says it wants a-0079 thou cutter but I reckon it can get away with a 40 thou cutter. I think this is all to do with going off standards on the shift and stubbing etc and I have asked Art to check this.

It's still early days on this program but so was Mach3 at one stage and look where it is now.

Once saved all that is needed to make any XL pulley is to change the number of teeth.

.

Tom S
09-25-2010, 07:33 PM
That is one fantastic program. Imagine combining that with a 3D modeling program that uses parameters to punch out a model of any gear configuration that you could think of!

Then again, maybe not. We could end up with some crazy engineer designing geartrains using completely non-standard gears. Bah.

John Stevenson
09-25-2010, 07:49 PM
Like this ?

http://www.stevenson-engineers.co.uk/files/eccentric%20gear.jpg

:D

gaston
09-25-2010, 09:03 PM
Just got home from the Gears model engineering show in Portland and a person there had a gear set with no "round gears" there must have been 8 or more gears. Interesting to watch a bunch of odd shaped gears running together.

John Stevenson
09-26-2010, 09:54 AM
Got the new update from Art this morning and it's cured the screen problems and cutter size so a 47 thou cutter will fit.

Heres the comparison CAD file.

http://www.stevenson-engineers.co.uk/files/xl_compare.jpg

Purple is the true pulley, green is Arts program, by the time all the corners have been knocked off on a scotchbrite wheel it will be indistinguishable from original.

.

Tom S
09-26-2010, 01:59 PM
Like this ?

http://www.stevenson-engineers.co.uk/files/eccentric%20gear.jpg

:D

You, sir, can stay away from me.

John Stevenson
09-27-2010, 03:32 PM
Setup and did a 2.5Mod 11 tooth 45 degree helical today.

http://www.stevenson-engineers.co.uk/files/helical_new.jpg

Came out perfect, checked it on the gear verniers and it was within two thou of where it should be.

51 minutes with a 3mm end mill running at 500mm/min - proof.

http://www.stevenson-engineers.co.uk/files/helical%20screen.jpg

Elapsed time is next to the EStop button

Video here http://www.youtube.com/watch?v=OI8f6Lrwx00

Watch from 2.08 until 3.01 and you will see how it steps over in the Y and the 4th axis follows it to form the involute in a series of discrete steps.

Shout up if I'm boring you playing to an audience of one. :D

mochinist
09-27-2010, 03:43 PM
Was playing with the demo a little yesterday and am looking forward to the non demo release being available for purchase.


I bet that e-stop button is hard to hit with the mouse cursor when that o-sh!t moment happens.

Enraged
09-27-2010, 03:45 PM
how about intermittent gearing? I'm working on a project that requires intermittent rotary motion linked to a scotch yoke.

Erik Brewster
09-27-2010, 03:45 PM
That is some seriously impressive software. That gear looks great!

willmac
09-27-2010, 03:54 PM
John -

That is showing real progress - a tribute to thinking slightly outside of the box. Of course this will not challenge the traditional production methods for gears, but that is not the point. What you have is a accessible method that can be implemented in the garage workshop, allowing us to make things that would have been impractical before. Many thanks for that.

Just a thought - I think that you are using a parallel end mill. If you were able to use a tapered end mill (such as used for mould cavities etc) you could end up with a cut that was much closer to tangency with the required involute form, extend the length of cut and perhaps achieve an improved finish with fewer steps. I don't know if Gearotic is capable of generating the Gcode for this - the calculations would be a lot more complex - but it might be worth a look.

Bill

Mcgyver
09-27-2010, 04:04 PM
What you have is a accessible method that can be implemented in the garage workshop, allowing us to make things that would have been impractical before. Many thanks for that.
l

Exactly. when i bought the controllers and motors for my cnc retrofit recently I bought i fourth set....now i now why

John Stevenson
09-27-2010, 04:22 PM
Was playing with the demo a little yesterday and am looking forward to the non demo release being available for purchase.


I bet that e-stop button is hard to hit with the mouse cursor when that o-sh!t moment happens.
Touch screen, that's why the buttons have been re-designed so you can press them easily. However it's plain stupid to rely on a software e stop, and illegal here in the UK, so all the machines have big red buttons fitted at forehead hight, the idea is that as you get dragged in your forehead hits the E Stop :D

Bill,
We looked at this but passed on it because it then made the cutters special. J&L do some tapered end mills but only in limited sizes and they are about $50 each.
This way it uses off the shelf cutters costing pennies and you chose what you want or what fits, providing it is smaller than the root space, and it does display this information so you can use any smaller cutter.

Later the software is going to be able to generate and cut any involute cutter needed, ideal when you are onto very small gears where the only end mills will be very tiny.
Not new I have been doing this for ages, as has Evan but the software will automate to in that you won't have to draw and program the correct shape.

This will be very handy for doing timing belt pulleys where there are only three to a set unlike the 8 B&S cutters.

Like Mach it will be a work in hand with free updates, new today was the provision for cutting gears with either a right hand or left hand mounted 4th axis.

Art is trying to get this out first week in October before he goes away for a month, when he gets back it will be bug fixes and new operations.

Top of the list is bevel gears.

These will be done with the 4th axis laid down at the pitch angle [ or use a BS zero dividing head ] and the gear will be rolled just as in the video.

This will make a perfect bevel gear with no hand filing needed as is required now and will basically mimic a Gleason.

rdfeil
09-27-2010, 04:54 PM
Sir John,

Glad to have you back :D Always interesting to see your work and that of the people you work with. You are definitely NOT playing to an audience of one. Please keep posting the interesting ( or even the not so interesting ) things you do.

Looking forward to many more posts.

.RC.
09-27-2010, 05:09 PM
So essentially it is gear shaping (like the sunderland method) with an end mill..

Next up we want true bevel gears....

j king
09-27-2010, 08:20 PM
Glad your back John.. Gears look great. Fantastic job getting it figured out. Jim

snowman
09-27-2010, 11:04 PM
I have a complaint. In one of your videos there is an ugly gent on screen.

Why did you let Alistair become a movie star? I know that isn't you, my mental image of you is more of a british film star.

Erik Brewster
09-27-2010, 11:33 PM
I just saw your raster scan cut of a spiral bevel gear. Very interesting! All this gear work you are doing is four stars out of four!

John Stevenson
09-28-2010, 03:55 AM
I have a complaint. In one of your videos there is an ugly gent on screen.

Why did you let Alistair become a movie star? I know that isn't you, my mental image of you is more of a british film star.

I didn't stand a chance, I was just going to do the video when he pushed me aside and said " here hold this "

One good thing though is the camera has no audio other wise instead of a soundtrack talking about the table you would have heard begging posts in a Scotch accent that's hardly understandable.
Billy Connerly speaks the Queens English compares to this guy.
"ahh see you Jimmie "

Evan
09-28-2010, 08:55 AM
That look very interesting John. I will be looking forward to playing with it when I have some time.

lazlo
09-28-2010, 09:08 AM
Looks great John!

How do you sign up for the beta? Just drop Art an email?

RTPBurnsville
09-28-2010, 09:12 AM
Wow! I have been wanting to get a 4th for the Tormach and now there is a valid reason other than just because it's cool.

Thanks,
Robert

Michael Hall
09-28-2010, 10:41 AM
http://www.gearotic.com/downloads.html

Michael




Looks great John!

How do you sign up for the beta? Just drop Art an email?

photomankc
09-28-2010, 11:32 AM
Well crap!!!! I was just going to get all excited about getting my 3 axis mill all completed and now I'm going to have to get a forth!

That is way cool. I have seen a number of ways to make a gear out there but they were all a tad cumbersome or insanely expensive. This looks perfect for what I would need to do with an occasional gear for this or that project.

John Stevenson
09-28-2010, 07:05 PM
Managed to get the mate done to the helical today before bloody customers stopped play.

http://www.stevenson-engineers.co.uk/files/helical%20pair.jpg

Our local gear cutter popped in on his way to the hardening shop across the road, saw them and had an orgasm in the middle of the shop.

Will Simply Green fetch this off paintwork ?

Need to fit them to spindles at the correct centre distance to check mesh but they roll together very smoothly.

Might try doing some of the eccentric gears in plastic on the router just to see how they run and either give them to the grand kids or use them as throwing stars at next doors mangy moggie.

Need to go raid the scrap plastic bin at the local sign makers..........later.

Circlip
09-29-2010, 04:34 AM
Can't see what all the fuss is about. Bench grinder, bit of drill rod and whammo, profiled milling cutter.

Regards Ian.

dp
09-29-2010, 10:34 AM
All seems a bit overdone. Now here's a clever solution. Her other tool is a dremel. :D

http://www.gears.com/mill_pinion.jpg

snowman
09-29-2010, 11:18 AM
All seems a bit overdone. Now here's a clever solution. Her other tool is a dremel. :D


I've got one of those. The upkeep costs are high, especially if you prefer to ride them hard and put them away wet. I like to think my stuff is disposable, and as many times as I've drug this one out to the curb, it keeps finding its way back in.

lazlo
09-29-2010, 03:24 PM
All seems a bit overdone. Now here's a clever solution. Her other tool is a dremel. :D

Holy crap! That's a big gear :) Is that from a ship?

John: that helical set looks fantastic! Surface finish looks good.

.RC.
09-29-2010, 04:06 PM
Could be from a steel rolling mill or for the mining industry..

DickDastardly40
09-29-2010, 04:19 PM
I've got one of those. The upkeep costs are high, especially if you prefer to ride them hard and put them away wet. I like to think my stuff is disposable, and as many times as I've drug this one out to the curb, it keeps finding its way back in.

Snow,

Are you referring to the tool or its operator?

small.planes
09-29-2010, 04:21 PM
Nice helicals.
Can it do herringbone gears ( >>>> <<<< bad ascii art time) yet?

Dave

dp
09-29-2010, 04:23 PM
The picture is from http://gears.com/ which is a gear shop in New Mexico. I don't know anything else about it but suspect it is used in heavy dirt hauling equipment. Not many ships there, but you never know.

mochinist
09-29-2010, 04:31 PM
Holy crap! That's a big gear :) Is that from a ship?

There is a place here in Tempe AZ that makes gears so big, that they make them in either two or four pieces for shipping reasons I believe, I seen one that was at least 20' in diameter. Per my understanding most of their work is for big earth hauling/moving equipment.

rmuell01
09-29-2010, 04:57 PM
I've got one of those. The upkeep costs are high, especially if you prefer to ride them hard and put them away wet. I like to think my stuff is disposable, and as many times as I've drug this one out to the curb, it keeps finding its way back in.


are we still talking about the gear? :eek:

.RC.
09-29-2010, 06:04 PM
There is a place here in Tempe AZ that makes gears so big, that they make them in either two or four pieces for shipping reasons I believe, I seen one that was at least 20' in diameter. Per my understanding most of their work is for big earth hauling/moving equipment.

I think quite a few industries make use of large gears

http://www.hofmann.net.au/products_6_4_largemillskilngears.php

.RC.
09-29-2010, 08:02 PM
BTW do we call this new method of gear generation the "Stevenson" method or the "Sir John" method?

John Stevenson
09-29-2010, 08:07 PM
BTW do we call this new method of gear generation the "Stevenson" method or the "Sir John" method?

The Art Fenerty method.

After all it's his program that doing all the work, I'm just wasting brass bar. :D

dp
09-29-2010, 08:19 PM
BTW do we call this new method of gear generation the "Stevenson" method or the "Sir John" method?

It's pretty much how you cut gears on a shaper except the cutter is spinning, and the orientation of the cutter to the work is far more versatile.

http://www.eurospares.com/graphics/Tools/shapers/shaper%20gear%20cut.pdf

Evan
09-29-2010, 08:44 PM
I'm working on how to make CamBam do it without extra software or at most a simple plugin. I have already done profile cuts that are similar using CamBam to produce an involute profile cutter. It's just a matter of rinse and repeat.

lazlo
09-29-2010, 09:04 PM
It's pretty much how you cut gears on a shaper except the cutter is spinning, and the orientation of the cutter to the work is far more versatile.

http://www.eurospares.com/graphics/Tools/shapers/shaper%20gear%20cut.pdf

Dennis, the Shaper-cut gear is actually rolling the cutter through the enveloping cut. It's a form of gear hobbing -- essentially a Poor Man's Sundstrand.

What John is doing is milling the actual helical involute directly. Very cool.

dp
09-29-2010, 09:21 PM
Dennis, the Shaper-cut gear is actually rolling the cutter through the enveloping cut. It's a form of gear hobbing -- essentially a Poor Man's Sundstrand.

What John is doing is milling the actual helical involute directly. Very cool.

It's the same thing. The blank is rotating, the cutter is moving across the blank. Because of the greater versatility of the CNC, the cutter need not have a rack form (it has Z axis to simulate that) although it would use less g-code if it were a rack form. That same versatility also allows the needed relative rates to create a helix cut - not easily done with a shaper, but also not impossible. It is the equivalent of coupling the ram with the table to turn the blank with two influences. Entirely possible, entirely impractical.

.RC.
09-30-2010, 01:23 AM
Actually it has just dawned on me the large impact this new gear cutting method could have on the wider home hobby or jobbing shop...

Gear cutting for the most part has been a job reserved for specialised equipment..

Equipment that firstly is large in size and secondly uses expensive cutters.. For the small jobbing shop it is not feasible to own hobbers and fellow gear cutters just to cut a few dozen gears a year not to mention that every job would be a different module or diametral pitch so you would need to have umpteen number of cutters on hand that get limited use...

The size of the gear teeth to be cut using this methos is only limited to what will fit ont he rotary table... Hell even a CNC'd POS bridgeport could cut a big gear using this method... If the program gets to the point it can cut true bevel gears it would be a very handy thing to have...

DickDastardly40
09-30-2010, 03:22 AM
A further question: If I wanted to produce 2 or more mating gears of the exact same size or to put them back to back to create a double helical train, would it take less time to make one long gear like a worm wheel and split it into however many gears I wanted or to make a separate piece for each one?

John Stevenson
09-30-2010, 04:25 AM
Just what Ringer said,
Where it is going to score for me is those specials like short splines on the end of DC motors that drive hydraulic pumps of fork trucks.

I get quite a few of these, probably 20 per year, they seem to come in batches, all different.

At the moment I have to take the armature into town to the gear cutters and they make me a short stub up. I then have to stub this into the armature but not fit it, then have it hardened and fitted.

What with the running about and the delay at the gear cutters who is usually very busy turn round can take a week, this way I can get it down to a couple of days, less if it doesn't want hardening.

Standard gears it's usually quicker to buy off the shelf, this program will free you up from the specials.

John Stevenson
09-30-2010, 08:07 AM
Why don't I keep my big mouth shut [ no need to answer this ] just posted the spline post above and what comes thru the door?

Splined rotor all crapped up.

BUT THE DAMN THINGS INTERNAL.................

Evan
09-30-2010, 09:07 AM
That's a shaper job.... :D

lazlo
09-30-2010, 09:46 AM
Where it is going to score for me is those specials like short splines on the end of DC motors that drive hydraulic pumps of fork trucks.

John, why don't you use your electronic gear hobber?

John Stevenson
09-30-2010, 10:06 AM
John, why don't you use your electronic gear hobber?

Need to buy a series of 8/16DP, 10/20DP, 12/24DP, 16/32DP. 20/30 and 20/40DP gear hobs.
Not easy to find at a reasonable cost

becksmachine
09-30-2010, 11:41 AM
Why don't I keep my big mouth shut [ no need to answer this ] just posted the spline post above and what comes thru the door?

Splined rotor all crapped up.

BUT THE DAMN THINGS INTERNAL.................

Murphy Strikes Again!! ;)

Don'tcha just love this repair business, cool program John.

Dave

leesr
09-30-2010, 02:18 PM
John

I like the program. I would like to buy it and try it out.

leesr:)

John Stevenson
09-30-2010, 06:28 PM
Just been released today.

www.gearotic.com

$75, still got bugs in it but like Mach it's a one off payment, Art has about another week to work on bugs then he's off for 3 weeks to China and when he gets back, early November it's back to bug fixing and adding features.
Future updates are free.

Evan
10-03-2010, 11:21 PM
Well, I have figured out the math for cutting spur gears by the "new" method. This is a graphical representation of what must be calculated to determine the offsets for the Y axis and the amount of rotation of the A axis in order to cut a tooth wall. The other side is just the same thing in mirror image and the rest of the teeth are simply lather, rinse and repeat. The main thing is that you need to calculate the tooth shape via whatever approximation suits and then transform it to cartesian coordinates. After that it's all basic trig.

http://ixian.ca/pics7/gearcalc.jpg

When I find time I will write a plugin for CamBam to generate the code for simple spur gears. That may be awhile and it won't compete with Gearotic as it won't have the range of capability that Gearotic does. The gearotic program is very cool and I have only found one way to crash it so far. That is probably already fixed. $75 is a good price.

dp
10-03-2010, 11:29 PM
One thing about the process that had me curious is if it matters to cutter efficiency to first create coarse steps with the vertex of the cutter and then go back over it with the shank, only, to smooth the steps - reducing them to facets. Depending on the depth of the steps it may remove metal faster as the cutter moves to toward the tip of the tooth.

John Stevenson
10-06-2010, 05:00 PM
So instead of playing came to do the first paying job today and very home shop related.

http://www.stevenson-engineers.co.uk/files/pinion%20shaft1.jpg

This is the pinion shaft out of a lathe apron, early import that will probably prove very hard to get spares. Jammed up in a crash, bent it and broke a tooth off and bent the rest.

13 teeth 1.5MOD, about 17 DP so put the data into Gearotic and it spits all the details out, main one being because it's so small, only 7/8" diameter it needs a 1.5mm cutter or 1/16".

Give the program very sensible depths of cut, 0.25mm or 10 thou and slow feeds as I only have one cutter.

Load up and let rip, 59 minutes later I have this.

http://www.stevenson-engineers.co.uk/files/pinion%20shaft3.jpg

Removed and wiped down and this is the final finished gear.

http://www.stevenson-engineers.co.uk/files/pinion%20shaft2.jpg

This was done with 8 passes per side, default is 10 but you can see from the pic the quality is very good, in fact probably better than if I'd hobbed it which always leaves a snail track round the gear.

I may have been able to go a bit quicker but i was worried about the cutter and if it had broken the time to make a new blank would have been more that hanging on.

Very pleased with the way this came out, mikes up OK on the gear verniers and seeing as I don't have a 1.5MOD hob a lot quicker than farming it out.

It now means that the odd one off is now possible in the home shop.

.

camdigger
10-06-2010, 05:24 PM
Cue applause!

Well done John!

oldtiffie
10-09-2010, 03:11 AM
John.

I've just down-loaded and run the "demo"version of the Gearotic application.

http://www.gearotic.com/downloads.html

Still - as you say - a couple of bugs to iron out - but very impressive.

I should be able to buy the "real" version shortly as at US$70 it is very good value.

Glenn Wegman
10-09-2010, 10:21 AM
It may just be the pic, but the root appears to be double the width of the crest. Looks WAY out of proportion to mesh with another gear.

Evan
10-09-2010, 10:46 AM
I am sure the gear is correct. I now have the general case of the calculations completely worked out. In this diagram the angles and offsets were precaclulated using the arbitrary cartesian coordinates of the line segment ends on the curve of the gear face. The results of those calculations were then used to inform the CAD program as to placements of the guide lines. Then the CAD program was used to reconstruct the calculated face as a full check on the math. Next step is to write the program to do it automatically.

http://ixian.ca/pics7/gearmath.gif

BTW, Gearotic is a Canadian program. I haven't reverse engineered it. I have determined the algorithm myself. In Canada you can't patent software, algorithms, math or science and you can't patent a method of doing something if the method is only a mathematical algorithm.

John Stevenson
10-09-2010, 10:51 AM
It may just be the pic, but the root appears to be double the width of the crest. Looks WAY out of proportion to mesh with another gear.

Optical delusion......

Got this email this morning.

Hi John

The shaft arrived this morning, it fitted beautifully. The lathe is back together and running well.

Thank you so much. I'll square with you at the show.

RKW
10-09-2010, 11:39 AM
That's incredible. As you know, here in the USA everything is patented. I decided quite a while back that I was happy with my ideas and making them just for me. Anything patentable will just have to die with me. Firstly, it is ridiculously expensive to go through the process, and secondly, it will only be a very short period of time before it is stolen and made and sold by another country. No thanks.


In Canada you can't patent software, algorithms, math or science and you can't patent a method of doing something if the method is only a mathematical algorithm.

Evan
10-09-2010, 11:57 AM
The US is perhaps the only country where software and algorithms may be patented.

John Stevenson
10-09-2010, 12:09 PM
Just got a heads up email this morning.

Hi John
If you get Machinery Classified, have a look at page 4 of the 6th October issue.

It looks like the high end people are doing the same as Art cutting teeth with standard cutters.

**********

Went to have a look.

http://www.machineryclassified.co.uk/digital-magazine/Default.aspx

Page 4.

......Via an on line HPG licensed server.........

Bet that cost a couple of quid more than $75..............:confused: :D

lazlo
10-09-2010, 01:24 PM
It looks like the high end people are doing the same as Art cutting teeth with standard cutters.

**********

Went to have a look.

http://www.machineryclassified.co.uk/digital-magazine/Default.aspx

Page 4.

Bit of confusion in that text:


"Mathematical calculation, as described, is the best route for high-quality gears -- classes 2 or 3; while for classes 5 to 7, producing the gear machining cycle conventionally from the CAD model is better, it is said."

They're making a quality distinction between mathematically calculating the involute curve directly (their method), versus just cutting a DXF of the involute form (the existing method).

I think he's referring to AGMA classes, which are basically ABEC ratings for gears. In other words, how close to the correct tooth form you're getting. Class 1 and 2 gears are very low quality. Class 12 gears are used in ultra-precision equipment and are hobbed, hardened, then shaved or ground.

So I don't know if he's saying that cutting the involute form directly is "better" than hobbing for classes 5 - 7 (standard maintenance grade gears)? Or that mathematically calculating the involute curve is better than cutting a CAD model for AGMA Class 5 - 7 gears?

I think they're saying that you can't use either process (of directly cutting the involute tooth profile) for precision gears (AGMA Class 8 - 12)...


http://i164.photobucket.com/albums/u15/rtgeorge_album/AGMAStandards.png

John Stevenson
10-09-2010, 01:39 PM
Lazlo

Dunno ? I think I'll stick to Art's cheaper version, by the time I have bought a Hermes 5 axis machine plus paid to join the HPG Licensed Server I think either Hermes, HPG or myself is going to be financially embarrassed


:D :D

lazlo
10-09-2010, 01:54 PM
Dunno ? I think I'll stick to Art's cheaper version

I would imagine that Gearotic actually does the mathematical model of the involute tooth (which your linked advert claims is a superior method), since it models the dynamics of the gear train, and can generate G-Code.

http://www.gearotic.com/videos/ProjectTutorial2/ProjectTutorial2_controller.swf

In reality, the difference between modelling the tooth mathematically and generating G-Code directly, versus downloading a generic CAD model of a gear and cutting it as an ordinary 3D solid is the resolution of the steps (what we've often called "facets" here in the past).

When Gearotic generates the G-code, does it ask you for number of steps per tooth?

John Stevenson
10-09-2010, 04:34 PM
Yes the default is 10 but I have been cutting at 8 to trade speed against quality whilst beta cutting but to be honest on small teeth such as I have been cutting the quality looks fine.

I don't know what the maximum steps are and Art is in China at the moment until the beginning of November

philbur
10-09-2010, 08:00 PM
Here's one I made earlier. I think the tooth spacing looks wider than it actually is because of the flat bottomed root.

Phil:)

http://i186.photobucket.com/albums/x36/philbur/TEETH_013.jpg

Phil:)





http://www.stevenson-engineers.co.uk/files/pinion%20shaft2.jpg

oldtiffie
10-09-2010, 08:58 PM
Yes the default is 10 but I have been cutting at 8 to trade speed against quality whilst beta cutting but to be honest on small teeth such as I have been cutting the quality looks fine.

I don't know what the maximum steps are and Art is in China at the moment until the beginning of November

Good "back to basics" and practical post John.

All hobbing cutters - gear shapers included - cut "flats" or "facets" due either (both?) the cutter or the job indexing, stopping and cutting and then moving on etc.

The number of flats/facets cut by a hobbing cutter is fixed and is equal the to number of teeth that passes and cuts the tooth for each revolution of the cutter and for each tooth of the gear. For each revolution of the hobbing cutter the gear is indexed one tooth.

There is a practical limit of number of flats/facets that the hobbing cutter can actually cut without rubbing on the job and "losing" the cutter edges and/or work-hardening the job. It is all to do with minimum practical depth of cut.

As discussed in previous related threads, the larger the pressure angle and the larger the DP and number of teeth the nearer the involute curve is to an arc. The larger the arc the nearer it is to being a "flat" - ie the chord of an arc or circle.

Similarly, the larger the number of "flats" the closer the series of them along an curve the closer those small flats will be to the curve in shape.

Again, there is no point in having more flats than is necessary for a practical satisfactory gear. This is or should be decided on the job.

I much prefer the Gearotic system if for no other reason that the number of flats is optional and the cutting is done by a very sharp end-mill by climb milling and mounted for as close to the spindle bearings as possible for optimum stiffness. That cutter will do a much better job than any hobbing cutter as it can easily cut less than 0.001" if required.

The "accuracy" of gear teeth is more than slightly over-done at times here as all that is needed to measure them is either or both a "gear caliper" or a common shop caliper (either or both can be vernier or digital) as their accuracy of about +/- 0.001" is quite satisfactory. Any more reduction in tooth width can be achieved by setting more on the depth ("Z") of cut and effectively reducing the Pitch Circle Diameter - but it will have no real practical adverse effect in most cases.

The Gearotic system is or should be what the HSM gear-maker needs - if he has a 4-axis CNC mill set-up.

So far neither John Stevenson nor anyone else has mentioned the mill and the rotary table and the cutter source/s. My guess is that the mill is a "Sieg" (or similar), the rotary table a "Vertex" (or similar) - all of which are from China - as I would also guess the cutters are (also from China).

Let's not get too "exotic" or "high end" here as we are all HSM-ers after all with typical HSM tools and needs.

I hope we stick with gear-cutting at that level.

Glenn Wegman
10-09-2010, 09:05 PM
Thanks John and Phil,

Looks like the reflection from the light made the root look exceptionally wide. At first I thought it was possibly cut using cutter comp that didn't comp!

willmac
10-10-2010, 05:36 AM
OldTiffie-

I'm not sure I understand what you are saying here:

"All hobbing cutters - gear shapers included - cut "flats" or "facets" due either (both?) the cutter or the job indexing, stopping and cutting and then moving on etc.

The number of flats/facets cut by a hobbing cutter is fixed and is equal the to number of teeth that passes and cuts the tooth for each revolution of the cutter and for each tooth of the gear. For each revolution of the hobbing cutter the gear is indexed one tooth."

---

A hob cuts a gear in a continuous process. Neither the hob nor the work stop or index. The tooth space generation is carried out by the rotation of several cutter 'teeth' successively through the same tooth space. In the case of a Fellows type gear shaper, the tool reciprocates, so it does stop at the bottom and top of its stroke, but neither the work or the cutter indexes; the rotation of both is continuous in cut.

With perhaps a bit more development, the Gearotic method could be able to cut gears which are good enough for all home workshop and a lot of commercial work. I suspect that they will already be better than can be acheived with B&S type gear cutters, unless the number of teeth you want is optimum for that cutter, and perhaps even then. It would be interesting to do a comparison. The really big advantage is the fact that no expensive tooling is required. The only drawback is perhaps speed, which is no problem at all for home use or for one-off repairs etc.

John Stevenson
10-10-2010, 06:13 AM
Technically one of the most accurate methods of cutting is the Sunderland planer method in that as the cutter, in this case a rack, is passed thru the gear blank everything except the cutter motion is stationary so the facet it is machining is square to the blank and in the same plane.

A Fellows / Maxicut disk type planer is constantly revolving the cutter and blank as it cuts so the facet is actually a slow spiral.

Same applies to hobbing, in fact if you take a hobbed gear and hold it up to the light you can see a snail track around the gear where it catches up per rev. depending on how fast the hob was passed thru the work makes this more or less visual.

However we are talking minute measurements here but never the less the Sunderland method is far superior.

This is basically what Art's program does, a series of user defined facets, in fact it's the same process that is done with the shaper cut gears but with a standard cutter and not a form tool.

http://www.stevenson-engineers.co.uk/files/pinion%20shaft2.jpg

This took one hour to cut the gear, just the gear teeth, no blank machining. Slow granted to hob this would have taken 5 minutes, again just cutting the teeth, however in this case I don't have a hob and not many people have hobbing machines.

Remember the CNC isn't limited to just doing gears.

These helicals took 45 minutes, less teeth but larger and done in brass I was able to increase speeds and feeds and it was a stronger cutter, 3mm against 1.5mm

http://www.stevenson-engineers.co.uk/files/helical%20pair.jpg

To make it clear that's 45 minutes per gear.
To be honest I have done helicals before by the 4 axis method but I used a hand ground cutter and the gear was in phenolic for a vintage car dynamo drive, however counting programming, and grinding a cutter [ 3 attempts ] it took a lot longer.

Whilst it's very useful to cut gears this way it's just another tool to be used in the best application it will fit.

In conversation with Art and with what has been flying back and forward the program isn't going to say at this level. Bevel gears with no hand finishing are on the cards as it the making of form cutters where the program will not allow off the shelf cutters.

Stay tuned.

Evan
10-10-2010, 06:18 AM
The Gearotic system relies on CNC to make gears and I would say that that qualifies as "expensive tooling". For regular straight cut spur gears there are many methods using a CNC mill that will produce very satisfactory results including the spur gear on the shaft that John recently cut. The big plus of the gearotic program is the ability to cut other types of gears such as helicals.

John Stevenson
10-10-2010, 06:41 AM
I think you are missing the point Evan.
You need a gear hobber to hob gears and that's all it can do and is expensive.

Whilst a CNC mill is also expensive it is not confined to just cutting gears. In fact one of the main advantages of CNC mills over manual is the lack of special tooling and jigs that can soon payback the initial outlay.


Take this.

http://homepage.ntlworld.com/stevenson.engineers/lsteve/files/4th%20axis%20cam.jpg

Gearbox cam, produced with a 4th axis, standard cutters and a big box of one's and zero's that can be recycled.

In the past you would have needed probably 3 master cams, produced at great cost plus a complicated follower drive system, totally out of context for a one off.

However after this one off the machine can go on to do something totally different as regards the job, tooling and materials.

Evan
10-10-2010, 06:46 AM
You are going to wear out that picture one of these days John. :D

John Stevenson
10-10-2010, 07:08 AM
At least it's not posed <g> :D

I'll have to photoshop it to give it a new lease of life :rolleyes:

lazlo
10-10-2010, 10:32 AM
A hob cuts a gear in a continuous process. Neither the hob nor the work stop or index. The tooth space generation is carried out by the rotation of several cutter 'teeth' successively through the same tooth space.

We've collectively been trying to explain the enveloping cut on a gear hobber to Tiffie for years. I think that's one of the reasons John has him on Ignore :)

The facets he's imagining are from this picture I posted from a gear manufacturer, showing how each gash on the hob cuts tangent to the involute tooth form:

http://i164.photobucket.com/albums/u15/rtgeorge_album/hobbing.png

But the helix angle of the hob, and the continuous synchronized rotation of both the hob and the gear blank, and the steady increase in depth of cut averages out the tangent points, and the gear tooth comes out mirror smooth.

lazlo
10-10-2010, 10:43 AM
This is the How It's Made: Gear Hobbing video I posted a couple of months ago. The gear hobbing starts at 3:37 (gear shaping is immediately before that):

http://www.youtube.com/v/UHFT4zUkmMw&hl=en_US&fs=1&

No facets. Look at the reflection of the CMM probe:

http://i164.photobucket.com/albums/u15/rtgeorge_album/hobbedgear.png

John Stevenson
10-10-2010, 10:57 AM
Sunderland at work.

http://www.youtube.com/watch?v=pGG-C3FSPeM





.

lazlo
10-10-2010, 10:59 AM
Sunderland at work.

http://www.youtube.com/watch?v=pGG-C3FSPeM

Holy Cow that's a big shaper! They're apparently a used machine tool dealer. Did you notice this even bigger shaper?

http://www.youtube.com/v/pVzKwm6jQJw

John Stevenson
10-10-2010, 11:01 AM
Our local gear cutter has one 94" diameter and 26" face width.

.RC.
10-11-2010, 03:27 AM
A bit off topic but is there anything that is a mathematically perfect curve??

Any line between two points is always straight.... So can a perfect curve even exist? Or is a curve always made up of a group of points..

The Artful Bodger
10-11-2010, 03:44 AM
A bit off topic but is there anything that is a mathematically perfect curve??

Any line between two points is always straight.... So can a perfect curve even exist? Or is a curve always made up of a group of points..

Hmmm, I think there would be, or very close, amateur telescope makers etc manage to make curves perfect down to light wave lengths (cue Evan) but anyway, for a gear there must be a point where any imperfections would soon by rounded off by wear, I think.

jugs
10-11-2010, 04:15 AM
A bit off topic but is there anything that is a mathematically perfect curve??

Any line between two points is always straight.... So can a perfect curve even exist? Or is a curve always made up of a group of points..

A radius is a perfect curve, an ellipse (depending how it is generated) can also be perfect
BUT how do we measure it - by a group of straight line points..Hmmm...:confused: :D

Back on topic, this system looks like a huge advance for the small shop guys ( no not dwarfs), well done to all involved.

john
:)

Evan
10-11-2010, 04:18 AM
Any line between two points is always straight.... So can a perfect curve even exist? Or is a curve always made up of a group of points..

Your starting assumption is closely related to Xeno's Paradox. It states that to move from A to B you must move half of the distance, then half of the remaining distance, then half of that ad infinitum. Since it will take an infinite number of such steps you can never arrive at B.

Clearly that is not the case. The reason is that to move any distance at all requires a particular velocity. If you measure the distance moved in a series of steps that decrease to an infinitely small size the time required to cover that distance also decreases to an infinitely small amount of time.

As it happens there are actually classes of infinities. Not all infinite classes are the same size. The set of all rational numbers is infinitely larger than the set of all integers and there are an infinite number of irrational numbers between any pair of rational numbers.

That last declaration is the reason that a curve is a curve rather than a series of straight line segments. The length of straight line segments can only be represented accurately by rational numbers. The coordinates of a curve must be represented by both rational and irrational numbers since pi is irrational. Because of this the set of numbers used to represent a curve is infinitely larger than the set of numbers that may be used to represent a polygon. The curve cannot be accurately represented by only the set used to describe a polygon.

In other words, you cannot square the circle.

philbur
10-11-2010, 04:45 AM
The is of course a simply made cutter with an easy to generate rack tooth profile (straight sided) that will produce a spur gear by a similar process without the need for CNC.

Phil:)

willmac
10-11-2010, 05:47 AM
Perfect Curves?

You need to define what you mean by a "mathematically perfect curve". If you mean a curve that can be defined by a mathematical function, no matter how complex then you are casting the net very wide indeed. NURBS (Non Uniform Rational Basis Spline) can be used to define free form curves and surfaces and are widely used in CAD systems. Whether you would regard NURBS as a mathematically perfect curve depends on your viewpoint of what that means.

Give us a a bit more detail about what you want and we can give you a better answer.

Mind you, there are other types of perfect curves!

.RC.
10-11-2010, 06:05 AM
Well I was thinking about the perfect involute curve since the discussion of hobbing and shaping came up and how accurate their curves are and was wondering if a perfect curve is ever possible or if every curve is always made up of segments..

willmac
10-11-2010, 06:28 AM
An involute is a fairly simple mathematical function, so it can be calculated to arbitrary precision, just like a circle or an ellipse. So the answer to your question is that it is not necessary mathematically to approximate an involute by a series of segments. Actually the type of involute that we are interested in for gears is based on a 'base circle'. You can have involutes that are based on other types of geometry, but as far as I know they have no use in gearing.

Saying that there is a simple function for an involute does not mean that all manufacturing methods will generate an arbitrarily accurate curve. Hobbing, for example, generates something close to such a curve, but it is a segment by segment generated approximation.

CAM and CNC controllers likewise may approximate a mathematically well defined curve like an involute with a series of straight line moves or arcs, or in higher end controllers more sophisticated curves. In each case the accuracy of the machined curve depends on the number of steps/segments as well as all the other cutting parameters. Of course the amount of data needed for a curve depends on the degree of accuracy required. If you must approximate with straight lines and need a very accurate curve then you will generate a vast number of G01 moves - perhaps requiring DNC.

Evan
10-11-2010, 09:54 AM
Saying that there is a simple function for an involute does not mean that all manufacturing methods will generate an arbitrarily accurate curve. Hobbing, for example, generates something close to such a curve, but it is a segment by segment generated approximation.


The term "hobbing" is a general term and is not specific to a particular exact method. There are various ways to hob a gear and some of them use continuous rotary motion of the hob and the gear. There are no straight line segments generated.

Further, this is getting deeply into angels on the head of a pin territory. Using just ten facets to generate a tooth face brings the deviation from a true involute to less than .001" at the most at all points. Increasing that to just 12 brings it to less than .0005" and 15 takes it down to the normal variation caused by the surface finish marks.

John Stevenson
10-11-2010, 11:37 AM
So 16 will be micron level ?

Evan
10-11-2010, 11:47 AM
Getting close.

edit: Note also that the number of teeth on the gear determines how many facets are required to approximate it with any particular degree of accuracy. The more teeth the fewer facets required until you get to a rack which requires only one to "approximate" it perfectly. That is even better than micron level. :D

philbur
10-11-2010, 11:47 AM
I would have thought that the precision of the produced gear would have more to do woth the CNC machine performance than the precision of the mathamatics. Any one ever heard of lost motion, tool flex, tool runout, part runout etc etc.

Phil:)

Evan
10-11-2010, 11:54 AM
That is an entirely different topic Phil.

philbur
10-11-2010, 12:04 PM
But it is a lot more relevant to the HSMer Evan. I think a relative comparison of machine and mathematical precision would be of much more general interest.

Phil:)


That is an entirely different topic Phil.

precisionmetal
10-11-2010, 12:12 PM
Mathematically, I believe an involute is typically defined "instantaneously" (which is all that's required for doing gear calculations). If an infinite number of points was used, then the entire profile could be defined perfectly. If I'm not mistaken, I believe that the involute functions used in gear software are iterative -- i.e. it's a routine that continually loops until the error falls below a threshold.

A hob or gear shaper still does not generate a perfect involute because the cutting edge on the hob (or shaper cutter) is not continually in contact with the same "plane" of the profile. As the part/cutter roll/translate, the cutting edge is moving.

The only machine/method I know that actually generates a perfect involute is a Reishauer gear grinder, and of course it still has the issue like a hobbing machine where the feedrate through the part actually creates "scallops" (though the grinding wheels is typically so large that it's a complete non-issue).

PM

Evan
10-11-2010, 02:01 PM
A hob or gear shaper still does not generate a perfect involute because the cutting edge on the hob (or shaper cutter) is not continually in contact with the same "plane" of the profile. As the part/cutter roll/translate, the cutting edge is moving.


As I wrote, there is more than one way to hob a part. For example there is hobbing of worm wheels. A the hob rotates it cause the wheel to rotate. The wheel is gradually fed into the hob as both rotate until full depth is reached. There is no synchronism between the feed rate and the rotations of the hob and wheel. The feed rate isn't incremental or stepwise, it is continuous so there is no possibility of a repeating pattern being formed. The tooth profile of the hob is a straight line. This method generates a true involute mechanically.

While it isn't possible to describe a circular function to an infinite degree of accuracy mathematically Nature has no such limitation unless you want to invoke Planck Time.

http://metalshopborealis.ca/pics/wormwheel1.jpg

willmac
10-11-2010, 02:38 PM
PrecisionMetal-

The involute function doesn't need to be defined iteratively, it is really just a simple function. In parametric form using polar coordinates r,a, angles in radians with base circle radius R and parameter p it looks like:

r = R sec p
a = tan p - p

(would have used more conventional Greek symbols, but not sure how)

So you can evaluate it just like the equation for a circle or a parabola.

The question was about mathematically defined curves. Of course when actually cutting metal it is a different and infinitely messier world.

lazlo
10-11-2010, 03:22 PM
I believe that the involute functions used in gear software are iterative -- i.e. it's a routine that continually loops until the error falls below a threshold.

There's a closed form solution for the involute curve, so you don't want/need to iterate in the software. That's the competitive advantage the advertisement John posted earlier was claiming: that their gears were generated directly from a mathematical equation, and not just cut from directly from a CAD model, which is built from discrete line segments.

I'm pretty sure that Gearotic is doing the same thing -- they have a closed-form solution for a perfect involute curve, and will generate as many discrete points in the G-Code as you can handle.

As was mentioned in another thread, you could get even better accuracy if you used a tapered end mill, since it's pretty close to a rack form, and therefore much closer to the tangent of the involute tooth.

In any event, I don't think we're counting angels. Evan was suggesting that getting within a thou on a gear tooth was plenty of precision. But consider that the center-to-center tolerance on standard shop-grade (AGMA 8) gears is +/-0.0005" (that's not a typo :)). If you're + a thou on one tooth, and - a thou on the mating tooth, you're going to have a noisy gear that's going to wear pretty fast. I'm quite sure you get much better than a thou with Old School involute gear cutters and a dividing head.

By the way, people throw out "Gear Hobbing", "Gear Shaping", ... pretty loosely here, but this is a comparison of the accuracy of a variety of gear generating methods. The X-axis is the AGMA/DIN gear accuracy:

Gear Hobbing, Shaping, and Shaving
(http://books.google.com/books?id=_cH57g5T7acC&pg=PA184&lpg=PA184&#v=onepage&q&f=false)
http://i164.photobucket.com/albums/u15/rtgeorge_album/geargeneration.png

Evan
10-11-2010, 04:06 PM
But consider that the center-to-center tolerance on standard shop-grade (AGMA 8) gears is +/-0.0005" ...

Why would that matter? The entire advantage of involute gears is that centre to centre distance isn't at all critical.



I'm quite sure you get much better than a thou with Old School involute gear cutters and a dividing head.


Only for the particular gear that the cutter is made for. The rest will have a slightly incorrect tooth form.

lazlo
10-11-2010, 04:17 PM
I was searching for an animation of the involute contact surface to answer Evan's question, and just noticed this slick animation showing the enveloping cut of a hob.

This is an Ogg Video file, but it's fantastic. The green wave-front is the contact area of the hob as it rotates past the gear blank. In a real hobber, both are rotating, but it's a lot easier to visual when one reference frame is fixed:

http://upload.wikimedia.org/wikipedia/commons/f/f6/Skupaj_ogv_q10ifps2fr6.ogv

http://i164.photobucket.com/albums/u15/rtgeorge_album/gearhob.png

willmac
10-11-2010, 04:44 PM
Where the pitch circle is not concentric with the centre of rotation of the gear a lot of noise and wear can be created. The run out tolerances for gears when hobbing are extremely tight and having a gear blank not properly centred on the mandrel is one of the easiest ways of getting a poor gear. Varaiations such as this cause the gear to lose a bit of its conjugate action.

willmac
10-11-2010, 04:45 PM
Lazlo-

That is a beautiful demonstration. I also like the analogy with a wavefront. OldTiffie - are you convinced?

willmac
10-11-2010, 04:52 PM
I suggested the taper end mill and I still think that it would be a big improvement, both in quality and speed with Gearotic. In fact there is a trade-off between quality and speed using John's approach and a taper end mill would shift this quite considerably. As John S pointed out, tapered end mills cost more, but I think the bigger difficulty would be developing the code needed for working with a tapered cutter. It is not a new problem, but a fair bit more difficult that assuming a cylindrical cutter.

Evan
10-11-2010, 04:57 PM
A tapered cutter would be a major complication to the math. As it stands currently the height of the cutter doesn't need to be considered at all except for being in the correct place for the bottom cuts on round gears. After that all the rest of the cuts up the face may be made with no change in the height as the cutter will always clear the opposite face and will always be deep enough for the facet it is cutting.

John Stevenson
10-11-2010, 05:24 PM
Biggest problem with a tapered cutter is getting everyone to sing off the same hymn sheet.
You can get end mills anywhere in the world, I have some tapered cutters but the smallest ones often have a small diameter exceeding 1/8" so they will only do gears below 10 DP or 2.5 MOD

precisionmetal
10-11-2010, 06:13 PM
The involute function doesn't need to be defined iteratively, it is really just a simple function. In parametric form using polar coordinates r,a, angles in radians with base circle radius R and parameter p it looks like:

r = R sec p
a = tan p - p

(would have used more conventional Greek symbols, but not sure how)


Willmac,

Doesn't that function define points on the curve, rather than defining the curve?

Can you explain? thx

PM

dp
10-12-2010, 02:11 AM
There's a closed form solution for the involute curve, so you don't want/need to iterate in the software.

Unless your CNC is using form cutters, you are going to iterate. That's the nature of raster reproduction.

whitis
10-12-2010, 02:16 AM
A tapered endmill doesn't improve the geometry; the side of the taper is still a straight line. A regular end mill tilted at an angle (or offset from the axis of the rotary table) also cuts a taper. A taper has a potential mechanical advantage in that the whole end mill doesn't need to be as wimpy as the narrow root. It also can rough out slightly faster.



Doesn't that function define points on the curve, rather than defining the curve?

same thing. The parametric equations are continuous. It describes an infinite number of points along the curve, though in practice you will use a more manageable subset.


"There are various ways to hob a gear and some of them use continuous rotary motion of the hob and the gear. There are no straight line segments generated." - actually, the teeth on the hob make that a discontinuous process resulting in small facets and as the hob traverses along the length of the gear blank you produce scallops.

Spur gears are going to have facets, scallops, or both, depending on which way you do the cuts, whether you use this process or a hob. You can trade cutting time for accuracy and as you pointed out, Evan, it doesn't take that many passes to get the error down to very small levels.

It turns out that the finite number of rack teeth normally used to show how a gear hob generates the curve are actually a fair reflection of real world hobbing. Cutting your hob from a grinding wheel would be pretty close to the theoretical ideal but the grinding wheel wears (and generates dust that would wear the ways of your machine) and you still have minor scallops from the feed as you progress along the length of the gear. Because the ideal model of hobbing is a 2 dimensional one, not a three dimensional one.

Worms have an advantage for hobbing in that there is no feed along the length of a gear and this is probably one of the reasons they are used for high accuracy applications like telescope drives and rotary tables.

On a good gear hobbing system, the rotation of the hob and the gear blank is coupled by gears or motors. The imperfections in these gears or motors can have an effect on the quality of the results but a self running hob does not track perfectly either.

While the facets and scallops produced by the cutter can easily be reduced below 0.0001" or whatever reasonable limit we want to put on the accuracy of our machining, they can introduce some vibration and wear. Polishing can improve the smoothness and may slightly improve dimensional accuracy or make it worse.

A tooth hob might start from first principles to generate the involute, but when the rubber meets the road, the real world issues reduce it to an approximation.

Minor errors in the setup and alignment of the machine can introduce some errors, as well as backlash, and all the other usual sources of error.

Real world gear cutting processes produce imperfections.

The hob (or a circular rack) has a performance advantage since more than one tooth is cutting at a time. It also has a bigger radius which reduces the depth of the scallops allowing you to move faster (an advantage shared by a disk style cutter or a circular rack).


"Mathematically, I believe an involute is typically defined "instantaneously" (which is all that's required for doing gear calculations). If an infinite number of points was used, then the entire profile could be defined perfectly. If I'm not mistaken, I believe that the involute functions used in gear software are iterative -- i.e. it's a routine that continually loops until the error falls below a threshold."

Rather than "instantaneously", the appropriate word is "continously". Computers iterate along curves. Gear cutting software or not. Few minor exceptions are symbolic math and using recursion vs iteration but those really aren't relevant to the discussion. The size of the steps can be adjusted to produce an error below any specified value and can be adjusted dynamically, if need be. The practical limit is the limit of the stepper or servo motors. Or they solve for intersections of curves. They don't compute an infinite number of points along a curve and they don't need to.

lazlo, the gear in your video (post #84) does indeed have both facets and scallops; whats more, they are clearly visible (if you know what to look for) at 4:29 as they start to show the CMM probe and are even visible but less clearly in your lower magnification still picture (the angle is best on the second tooth from the left). Close to the level of surface finish produced by metal cutting process itself, but they are there. Nice vid, though. The number of facets is about 16. Width of the facets and width of the scallops is about the same, which is probably no accident.

The facets on sir john's gears are more dramatic because he isn't willing to wait around to make them smaller. If he took about twice as many passes (i.e. take 16) , the results would probably look somewhat similar to the gear in the video, cutting a gear with the same diametral pitch (or module), pressure angle, size, and type. But due to the smaller cutter diameter, he needs more scallops per inch/mm to get the same error.

Note that 16 number is consistent with Evans calculation.

It is quite possible, though, that what looks like grinding marks on the tool transfering to work are the real facets (call them micro facets) and the larger macro facets are actually irregularities between the smaller facets with irregularities between individual teeth on the hob and the pitch error on the hob or errors in the drive system making some facets stand out more than others. However, there aren't enough threads on the hob to account for the number of macro facets but the micro facets could be the effect of the different threads on the hob engaging the work at a slightly different angle than the theeth on the previous one subdividing the facets produced by indivdual teeth into smaller subfacets.

There are 12 teeth per revolution on the hob by 5 threads/rows so there could be as many as 60 facets, though the teeth on the edge might not engage of the full length of the involute.

You can also see while the reflection of the CMM is pretty clear in the tooth at point blank range, the reflection is very dull in the next tooth over. It isn't a mirror finish though it is approaching one.

Either way, the physical geometry of the hob is creating visible facets on the work which are a significant factor in the surface finish, as expected.

Philbur:
"I would have thought that the precision of the produced gear would have more to do woth the CNC machine performance than the precision of the mathamatics. Any one ever heard of lost motion, tool flex, tool runout, part runout etc etc."

Bingo. Once you take enough passes.

Evan:
"Note also that the number of teeth on the gear determines how many facets are required to approximate it with any particular degree of accuracy. The more teeth the fewer facets required until you get to a rack which requires only one to "approximate" it perfectly. That is even better than micron level."

First part is correct. Second is correct with regard to facets, but neglects that scallops exist even on the rack, though no different than machining any other flat surface with an endmill.

Evan
10-12-2010, 02:43 AM
Whitis,

With continuous rotation of the gear and the hob the cutting edge describes a curve as it passes over the tooth face. It isn't cutting a "facet" or a scallop. It is generating a small portion of the involute curve. The next cutting edge to encounter that tooth does the same at a slightly different part of the curve. Where the two portions of the curve overlap there is a region where no cut is made since the material is already removed in what is theoretically a perfect involute.


Rather than "instantaneously", the appropriate word is "continously".

Instantaneous is correct.


7.3.2 Properties of Involute Curves
The distance BK is equal to the arc AB, because link MN rolls without slipping on the circle.
For any instant, the instantaneous center of the motion of the line is its point of tangent with the circle.
Note: We have not defined the term instantaneous center previously. The instantaneous center or instant center is defined in two ways (Bradford & Guillet 43):
When two bodies have planar relative motion, the instant center is a point on one body about which the other rotates at the instant considered.
When two bodies have planar relative motion, the instant center is the point at which the bodies are relatively at rest at the instant considered.
The normal at any point of an involute is tangent to the base circle. Because of the property (2) of the involute curve, the motion of the point that is tracing the involute is perpendicular to the line at any instant, and hence the curve traced will also be perpendicular to the line at any instant.
There is no involute curve within the base circle.

http://www.scs.cmu.edu/~rapidproto/mechanisms/chpt7.html

dp
10-12-2010, 03:20 AM
When a circular cutter makes a linear pass it cuts in scallops. It can't be helped. I'm not understanding something.

John Stevenson
10-12-2010, 03:59 AM
Well I was going to cut some sample gears for the show today but now I'm not sure if whatever I cut can pass muster. ????????


I might just cut them and take them anyway. :(:(

.RC.
10-12-2010, 04:42 AM
Well I was going to cut some sample gears for the show today but now I'm not sure if whatever I cut can pass muster. ????????


I might just cut them and take them anyway. :(:(

Of course they will pass muster... This method is certainly good enough for 99% of uses..

We are just talking theoretical BS to keep the thread alive.... I am sure most of us know that in 99% of engineering work, working within a few thou is often good enough... We may like to brag how we work to nanometres, but it is all BS in the end and not needed...

Evan
10-12-2010, 06:02 AM
When a circular cutter makes a linear pass it cuts in scallops. It can't be helped. I'm not understanding something.


The work is rotating too. The orientation of the cutting edge to the work surface is changing in a smoothly varying manner during the passage of the cutting edge through the work. The cutting edge is a straight line. The presentation of the work to that edge changes angle during the pass to produce a convex cut.

The cutting edge is producing a "facet" but the facet isn't flat. It is curved and therefor is not a facet.

I will have to think about making an animation.

Richard Wilson
10-12-2010, 06:15 AM
I was impressed with John's method of cutting gears, and even I understood it. Now, having read all the learned discourses, I don't understand it anymore, but I'm still impressed with John's method. The only question i have for John is:- do you find it cheaper to make a pair of helical gears by your method than having them cut commercially. I recently paid 100 to have a pair of skew helical gears (10DP, 10T and 20T) cut on my steel blanks, would your method have been cheaper)

I am reminded of Ernest Hives (of Rolls Royce) reaction when shown the original, very simple, Whittle jet engine 'Don't worry, we will soon design the simplicity out of it!'

Richard

Evan
10-12-2010, 08:32 AM
It is Art Fenerty's method. John is testing it. Art Fenerty is the author of the Mach III CNC controller program.

http://gearotic.com/

vincemulhollon
10-12-2010, 08:42 AM
The only question i have for John is:- do you find it cheaper to make a pair of helical gears by your method than having them cut commercially. I recently paid 100 to have a pair of skew helical gears (10DP, 10T and 20T) cut on my steel blanks, would your method have been cheaper)

I'm guessing you can buy a large number of very small diameter endmills for 100 pounds, so it costs practically nothing for consumables and tooling. On the other hand, the "new method", or whatever it will be called, burns an astounding amount of machine time compared to more traditional methods... Which means it costs somewhere between nothing and a zillion pounds for labor (err, labour?)

lazlo
10-12-2010, 09:56 AM
When a circular cutter makes a linear pass it cuts in scallops. It can't be helped. I'm not understanding something.

The contact point is tangent, but both the cutter and the gear blank are continuously rotating.

If the hob had concentric discs (like a stack of involute cutters), it would cut facets. But the hob has a continuous helix, so as it rolls in and out of the cut (the "enveloping cut" watch the animation), it's cutting a smooth arc.

Watch the How It's Made video I posted, or any of the industrial gear hobbing videos on Youtube -- they come off the hobber perfectly smooth.

lazlo
10-12-2010, 10:00 AM
I was impressed with John's method of cutting gears, and even I understood it. Now, having read all the learned discourses, I don't understand it anymore

It's really simple: the Gearotic program simply cuts the outline of the gear tooth like any solid model.

In other words, download a CAD model of a gear to any CNC machine, hobby or industrial, and it will cut the gear from the solid. It's not hobbing, but produces a very nice gear.

Gearotic is effectively a library of gear models that you can cut. It's much more than that (it actually model gear kinematics and mathematically generates the gears), but for the purposes of the discussion...

philbur
10-12-2010, 10:57 AM
While we are on the subject of helical gears, and with all the experts assembled. How critical is the concentricity of the outside diameter of a gear to smooth/quiet running. The main gear in the headstock of my lathe is a 57 tooth module 2 precision ground helical gear. It makes a clack, clack noise (in sync with the gear rotation) which to me indicates either the teeth running out of true or some bent teeth. The outside diameter of the gear has a indicated run-out of 0.12mm.

I don't have the equipment to measure the teeth. I was however considering using my CNC mill (Tormach) and a 4th axis to do a comparative measurement of each tooth, say both ends and middle. The mill positioning is repeatable to better than 0.005mm when moving in the same direction to the same point.

Any comments would be appreciated.

Phil:)

dp
10-12-2010, 11:54 AM
The contact point is tangent, but both the cutter and the gear blank are continuously rotating.

If the hob had concentric discs (like a stack of involute cutters), it would cut facets. But the hob has a continuous helix, so as it rolls in and out of the cut (the "enveloping cut" watch the animation), it's cutting a smooth arc.

Watch the How It's Made video I posted, or any of the industrial gear hobbing videos on Youtube -- they come off the hobber perfectly smooth.

An extreme case is a power planer for wood - you see the cutter marks very clearly. There's not much functional difference between the planer cutter and a hob, assuming you're talking about this kind of hob:

http://chestofbooks.com/home-improvement/workshop/Machine-Shop-Work/images/Fig-278-Hob-for-Forming-Teeth-of-Worm-Geara.jpg

The quality of the end finish is the result of keeping the cuts close to each other so cutting imperfections are disguised. It doesn't mean the tool isn't cutting scallops. The wood planer is such a good example because there are relatively few cutters for the feed rate. Changing the feed rate/number of cutters doesn't change now they cut, though.

willmac
10-12-2010, 02:07 PM
DP -

Sorry but I disagree, there is an important difference between the wood planer and a hobbing cutter. The difference is that both the work and cutter rotate continuously and the cutter meets the work at a multiplicity of different orientations. That makes a big difference to the finish.

dp
10-12-2010, 03:37 PM
I guess a better example would be to describe the cutter as a spiral grinding thread (probably a thoroughly impractical method) which is truely continuous vs a toothed cutter which is not.

whitis
10-12-2010, 03:46 PM
Whitis,

With continuous rotation of the gear and the hob the cutting edge describes a curve as it passes over the tooth face. It isn't cutting a "facet" or a scallop. It is generating a small portion of the involute curve. The next cutting edge to encounter that tooth does the same at a slightly different part of the curve. Where the two portions of the curve overlap there is a region where no cut is made since the material is already removed in what is theoretically a perfect involute. .

That would be the case if the cutter were a continuous uninterrupted shape that magically cut every where it touched the work. It isn't. It has a finite number of cutting edges that are moving almost perpendicular to your involute cross section at a speed much faster than they are sliding across the involute. Thus, it does cut facets. There may be a very slight curvature but it is closer to flat than the involute you are trying to cut(it may be involute, but it is the wrong involute). It is not small sections of the involute that will merge together when you cut enough of them and ultimately has little effect on the number of facets you need to get within a certain error tolerance on the tooth profile.

That is the dominant effect; there are a number of minor effects that add or subtract. The rake angle of the cutting teeth counteracts this to a very slight extent. Continuous feed along the length of the work, however, adds to the angular velocity of the cutter working against the slight improvement offered by the rake angle. A further complication is that the hob is not of infinite diameter so it produces some outward motion which also adds or subtracts to the sliding motion. The relative direction of these various contributions relative to the sliding motion changes depending on whether you are looking at the front or back edge of the tooth.

To throw another kink in the work, the cutter doesn't cut perpendicular to the work but cuts along the edge of the trench. This would appear to add or subtract from the sliding motion (depending on which tooth edge you look at), but the portion of the cut where this is significant is in metal that will be removed later (except for worm gears).

Sir John: by all means show the gears.

John Stevenson
10-12-2010, 04:43 PM
Well I cut 5 today in between getting pestered by customers.

Two were just copies of the original helicals as the first pair have gone to China to the Sieg factory,
Two 12 teeth same MOD as the helicals, 2.5, to use two blanks up I already had.

Last gear was the same helical but done in steel as a test because we are going to be cutting this gear, or copies of it at the show.

Reasons:-

[1] Brass is bloody expensive and
[2] If you cut in brass, wax or aluminium they always ask "Can it cut steel" so we try to cut the hardest job at the shows.

Had an email tonight to ask can we cut a big Delrin gear of 12 DP as they guy has been quoted 270 for one.

270 is a lot of pints so we'll go for it if we can, limited to 12" diameter by virtue of having a 6" 4th axis but emailed back for more detail, just got enough time to get a sheet of Delrin in.

lazlo
10-12-2010, 06:58 PM
270 is a lot of pints so we'll go for it if we can

John, you're paid in beer? :)

gmatov
10-13-2010, 12:33 AM
I can only comment on this. You guys are way ahead of me on the gear generation.

"An extreme case is a power planer for wood - you see the cutter marks very clearly..

My 18 inch Rockwell Delta has a 4 inch head with 3 blades, about 4 1/2 inch arc of cut, 3450 RPM, and at the slowest feed for a finish cut, you will still see scallops in the finished surface.

I don't have the math to do the arc that the knives make, even with the slowest feed.

The same almost has to hold true with an end mill 2, 3 or 4 flute, running through the tooth gullets. You have 4 cuts per revolution, but you have a finite distance between each cutting edge taking its cut. It has to make a scallop, no matter how close they are to each other.

I don't find that to be objectionable. Most of my machinery, at work, had well worn in gearing. EVERYTHING was mirror smooth, though it had lots of backlash.

Way back then, with the chick and the air grinder and the big shaft with a helical gear, guess was a steel mill rolling mill.

In my mill, we had a few helicals. One, the first stand, met a slab imported from Brazil. Nobody realized their mill poured 10 inch slabs. Our mill was designed for 8 inch slabs. When it hit the rolls, it drove the shaft through the thrust bearing, out of the gear case, and shattered the runner and the bearings.

Helicals are not good in any interrupted high load operation.

All our other gear boxes were herring bone, neutral thrust.

The other post about BIG gears? Yes, we had gears over 20 feet in diameter, about 4 feet across the face, split. As you say, the only way to transport them, make them in 2 pieces, all keyed.

I like the system Sir John is testing. I can't use it, but hope many of you can, even if it does take an hour or more, when you could call up and order a new one for delivery in a week. Did I need one, this evening, and I had paid US 70 bucks for the program, I would surely make a gear to get back in operation, even if I had a replacement on order.

That would go into stock till the shop made replacement gave out. BUT, if it NEVER gave out, would you think you wasted money on the 70 buck program, and the pittance in material, and the hour, that you could have gotten 65 bucks for?

Cheers,

George

.RC.
10-13-2010, 02:45 AM
John, what do you use for a 4th axis?

Is it a converted rotary table or dividing head, or a factory built item?

John Stevenson
10-13-2010, 03:47 AM
John, what do you use for a 4th axis?

Is it a converted rotary table or dividing head, or a factory built item?

Converted RT, in my case a 6" Vertex that was converted when we were messing about with the testing of the Divisionmaster when it first came out.

I do have a dividing head converted as well

If you go to the old Divisionmaster site at http://www.divisionmaster.co.uk (http://www.divisionmaster.co.uk/)

Under Examples of use a lot of the tables etc are mine.

At the show we will be using a Sieg unit because we are tied to Sieg for the show gear, no difference, just another make.

Evan
10-13-2010, 06:00 AM
The accuracy of shop made gears in large sizes is a very important issue to me. I have made half a dozen so far up to 12 inches in diameter for the hour angle tracking mechanism of the equatorial mount of my telescope. Accuracy is of the utmost importance as even the slightest irregularities will show clearly as tracking errors in the photographs.

By slightest I mean that the maximum deviation that can be tolerated is only a few arc seconds per hour. It is especially critical if the error is not evenly distributed as that much error from tooth to tooth will make a gear useless even if it averages to zero error over a longer interval. Tooth to tooth variation must be kept below one part in 10,000 or better. In general with a telescope clock drive the error should be less than a few seconds per day. As there are 23 hours, 56 minutes and 4 seconds per day that is 86164 seconds so one part in ten thousand is no more than 8 seconds error total.

To produce gears of such accuracy requires extreme attention to detail and consistency. I am not quite there yet but am working on it.

.RC.
10-13-2010, 06:44 AM
That is the big thing and what a lot of us probably do not understand enough...

With industry, tolerances are usually set out in either the part plan or a standard, and a lot of the time those tolerances are there as the part is expected to last a certain amount of time

In the home shop where you are making something for your own use, there is no need to follow industry standards when you know the application does not warrant it.... For example making a small hobby engine... Industry standards may dictate x amount of tolerance for a bearing fit, while the home shop owner firstly may not have the capabilities to even be able to measure such dimensions to the accuracy required, so they can go to other methods like a more loose fit with loctite to make the fit correct... And while the experts might scorn at such a practice, who cares so long as it works and you are happy...

Much like this method of cutting gears.. yea it may not be as good as a hobber and they may not be as accurate as a gear ground on a reinsheusher (sp?) grinder but who cares... We have a few windmills where the gears in it are as cast..And they have lasted over 50 years of continuous use...

Machtool
10-13-2010, 08:24 AM
....... as a gear ground on a reinsheusher (sp?) grinder .
Gday Richard.

Im thinking, youre thinking of Reishauer.

http://www.reishauer.com/index_de,345.html (http://www.reishauer.com/index_de,345.html)
Regards Phil.

dp
10-13-2010, 11:02 AM
I can appreciate the "nice to know" bit - but there must be a line drawn between the "nice" and "need" to know as regards practical needs and acceptability as regards outcomes on HSM shop equipment.

I think I don't agree with this. I have a need to know quite a lot about the engine on my Harley as my life depends on it and keeping it in the best possible condition and tuning. (I demonstrated this over the weekend when a fine citizen of Oregon decided to pull into my lane without looking to see if there was room for both of us (there wasn't). The lane to my right was occupied as well so my only option was to swing over to the lane divider [I had less than 6" clearance between the handlebars and the intruder's truck] and hit the throttle hard. I did, the noise blast from the exhaust got everyone's attention, and I got out in front of the idiots very quickly. What my wife did next is legendary - don't pi$$ off a redhead!)

However - I really enjoy reading Norm and JS's stories about the WWII warbird engine work they've seen and done and I'll never have a need to know any of that but it is still immensely interesting to read. Probably because I'm in this hobby for the education many of the posts that are possibly topic creep but as likely expansive background on the original idea are interesting, especially when told by persons with first-hand experience. The quenching process for the crown gear being an example - I'll never use it but it is endlessly fascinating to learn about it and wonder about how the process was discovered.

Evan
10-13-2010, 01:03 PM
Everyone should have knowledge of the first principles that govern how something works. That is what makes us different from chimps. Knowing not just how to do something but why to do something is what makes it possible to apply that knowledge to other situations.

lazlo
10-13-2010, 02:12 PM
Or... you simply use a Reishauer gear grinder -- arguably the most accurate (and fastest) method there is for finishing/producing production gears.

The Reishauer looks like a gear hobber with an abrasive hob. I'm pretty sure the grinding wheel would have to be helical for this setup to work, if both spindles are rotating continuously:

http://www.niagaragear.com/what-s-new/there-s-even-more-with-our-new-reishauerae-rz300e-precision-gear-grinder/image

precisionmetal
10-13-2010, 02:27 PM
Yes... the grinding wheel on a Reishauer is like a thread, and the grinding wheel spindle is inclined at the helix angle of that "thread" (plus/minus whatever the helix angle of the gear being ground is... assuming it's not a spur gear).

Evan
10-13-2010, 03:03 PM
That wheel (or stack of wheels) must be insanely expensive. I can't believe that it is cost effective. It must also have a continuous dresser in the back to maintain the profile.

precisionmetal
10-13-2010, 03:20 PM
The grinding wheel is not expensive, but the crush roll dressing wheel that's impregnated with diamond is.

Not sure how the latest gen of Reishauers handle dressing... it may all be done CNC these days.

The grinding wheel is big on a Reishauer (under most circumstances) so wheel life is very good. Of course it's all flooded in oil during production grinding as well, so that helps too.

lazlo
10-13-2010, 04:10 PM
Yes... the grinding wheel on a Reishauer is like a thread, and the grinding wheel spindle is inclined at the helix angle of that "thread"

It's basically gear hobbing, where the hob has an infinite number of gashes. This shows the automated dresser -- it's a pair of diamond wheels that tilt to dress the abrasive hob to whatever pressure angle you're cutting. Slick!

http://www.youtube.com/v/m6gRprNkTkY

In reference to Evan's question -- in this demo they gash the gear first, and then finish-grind it in the Reishauer (Kapp).

John Stevenson
10-13-2010, 05:33 PM
Well got the samples done, because of all the ho haw about precision I decided to ignore all that and just cut them :)

http://www.stevenson-engineers.co.uk/files/gearcluster1.jpg

Another pair of helicals, two 12 teeth, same pitch, 2.5MOD spur gears and one steel one to prove we can cut these at the show.

Then for something different I did a few for the kids :D [ me included ]

http://www.stevenson-engineers.co.uk/files/eccentric%20gears.jpg

Mounted them on a lump of scrap melamine board with a small geared motor at the back, does about 6 rpm. Can't show it working as got the batteries on charge in the house but here's the simulation on Gearotic.

http://www.youtube.com/watch?v=zOmSrePm_UM

Quite interesting to watch, should keep me ammused for hours.

TGTool
10-13-2010, 06:02 PM
It's basically gear hobbing, where the hob has an infinite number of gashes. This shows the automated dresser -- it's a pair of diamond wheels that tilt to dress the abrasive hob to whatever pressure angle you're cutting. Slick!



Nice video and nice machines but it does raise a couple questions. Of course the wheel has to turn much slower than the normal surface speed for a grinding wheel to keep rotation speed within reasonable limits so the action might be closer to what would happen in hand stoning a surface.

The other thing is that the wheel itself doesn't look quite like a normal vitrified wheel. The surface texture has too many large voids making me wonder if it isn't some kind of polymer abrasive like a Cratex wheel on steroids. The spinning diamond wheel dressers still should be able to form a precise enough surface for what they need. If the wheel actually has some resilience to it that might also mean that you can't just go around one, dial in to the number you want and be finished. The gear shaver I saw I think had the same action. That is, so long as you continued to run the gear and cutter together you continued to remove material. Not like a surface or cylindrical grinder where you could come up to some number on your infeed dial and spark out. Hmm, lots to wonder about.

small.planes
10-13-2010, 06:12 PM
http://www.stevenson-engineers.co.uk/files/gearcluster1.jpg

Them spur gears are never going to run like that :p
and your crossed helicals are all backwards

(running away)
Dave

philbur
10-13-2010, 06:22 PM
Very nice John, I want the software. To save me reading all 16 pages of this thread, when will the finished sofware be available and where will I be able to buy it.

Phil:)

jkilroy
10-13-2010, 06:28 PM
Actually I'd expect that demo is greatly slowed down. The few gear grinders I have seen in action FLY, they are a blur of action.

KiloBravo
10-13-2010, 08:25 PM
Very nice John, I want the software. To save me reading all 16 pages of this thread, when will the finished sofware be available and where will I be able to buy it.

Phil:)

Just read the first post ;)

www.gearotic.com

precisionmetal
10-14-2010, 11:49 AM
Nice video and nice machines but it does raise a couple questions. Of course the wheel has to turn much slower than the normal surface speed for a grinding wheel to keep rotation speed within reasonable limits so the action might be closer to what would happen in hand stoning a surface.

No... not at all. A Reishauer runs "normal" surface feet for grinding as jkilroy mentions. The wheels is big, so it's not going to run at 3600 like a bench grinder. The surface feet at the cutting edge is high though.


The other thing is that the wheel itself doesn't look quite like a normal vitrified wheel. The surface texture has too many large voids making me wonder if it isn't some kind of polymer abrasive like a Cratex wheel on steroids.

Definitely a grinding wheel, and once set up it's definitely a "set to size" and go situation.

A machine in action: http://www.youtube.com/watch?v=RLmywAAHHeI

lazlo
10-14-2010, 12:51 PM
A Reishauer runs "normal" surface feet for grinding as jkilroy mentions. The wheels is big, so it's not going to run at 3600 like a bench grinder. The surface feet at the cutting edge is high though.

A machine in action: http://www.youtube.com/watch?v=RLmywAAHHeI

Video is a PITA because of the coolant -- you can only see a couple of seconds of the hob and the gear blank in the last couple of seconds. But the gear blank is rotating around 60 RPM, so for a 20 tooth gear (again, just eyeballing it), the hob would be spinning at 1200 RPM. That's surprisingly slow.

Evan
10-14-2010, 01:11 PM
According to my wife that will almost certainly be a stack of wheels which only makes sense. Much cheaper to replace one in the stack than the entire thing if something bad happens.

JCHannum
10-14-2010, 01:25 PM
This photo that Robert posted and the first twenty seconds of the KAPP demo video clearly show the spiral grind of the wheel.


http://www.niagaragear.com/what-s-new/there-s-even-more-with-our-new-reishauerae-rz300e-precision-gear-grinder/image

Evan
10-14-2010, 02:01 PM
My connection is too slow to watch videos online except at 3:00 am and even then they sputter.

2ManyHobbies
10-14-2010, 02:07 PM
http://www.stevenson-engineers.co.uk/files/eccentric%20gears.jpg

Mounted them on a lump of scrap melamine board with a small geared motor at the back, does about 6 rpm. Can't show it working as got the batteries on charge in the house but here's the simulation on Gearotic.

http://www.youtube.com/watch?v=zOmSrePm_UM

Quite interesting to watch, should keep me ammused for hours.
You need to replace a geartrain in an old clock with that and donate it to the pub.

Either that or put color wheels on the shafts and walk down the street turning a hand-crank and giving everybody migraines.

precisionmetal
10-14-2010, 04:34 PM
According to my wife that will almost certainly be a stack of wheels which only makes sense. Much cheaper to replace one in the stack than the entire thing if something bad happens.


One grinding wheel, one piece.

They are available in widths of 62mm, 84mm, and 104mm.

MCS
10-16-2010, 04:43 PM
What I like to know about the "Gearotic" gears is the noise (whine) after initial wear-in, compared to a hobbed gear. Not at automotive rpm's but in the region of 1000-3000 rpm, my assumed target range. After all, the product is the result of the total system resolution.

The next question would be a fatigue run, because of the, in my opinion weakened tooth bottom. I would instantly start grinding little radii on the endmill, but that's me.

Toolguy
10-16-2010, 04:56 PM
You can easily buy endmills with a corner radius already on them. If I were gearing up for this, I would get or make corner rad. endmills. That said, probably for a small model engine it may not matter.

philbur
10-16-2010, 05:00 PM
I'm not sure the procedure is going to be good for small gears. I think it will require some very small endmills and big rpm's.

Phil.)


You can easily buy endmills with a corner radius already on them. If I were gearing up for this, I would get or make corner rad. endmills. That said, probably for a small model engine it may not matter.

John Stevenson
10-16-2010, 06:41 PM
I'm not sure the procedure is going to be good for small gears. I think it will require some very small endmills and big rpm's.

Phil.)

There is work in the pipe line to allow small gears to be cut but for more information you will need to join the Gearotic forum for announcements.

I don't propose to give further details on here, anyone interested enough now has the information to keep abreast.

philbur
10-17-2010, 08:19 AM
I think the issue is not with parallel development, it is more to do with the intend of that development.

Continued poking with a sharp stick only exposes that intent more clearly.

Phil:)

JCHannum
10-17-2010, 08:27 AM
There is an enormous difference between a program that can produce a simple gear profile and one that can generate gears such as this one;

http://www.youtube.com/watch?v=zOmSrePm_UM

That can only occur by development of an idea and building on it over a period of time. It is the result of several people being involved and exchanging thoughts and ideas, and sharing successes and failures.

John is one of those people. He very generously spends a large amount of his time and energy sharing his experience and ideas on many boards. He is genuinely interested in advancing the knowledge and application of CNC machining and by his involvement on so many fronts and bringing together so many others with similar knowledge, he has facilitated some very significant advances in making CNC machining available to the HSM.

I have never seen an instance where John does not give credit to the originator of an idea or application. He more often as not downplays his own involvement in the development of the item, but it is usually quite obvious he has played a major role in bringing it to fruition.

Rather than knock him, refer to him as a thug, or try to one up him, I think we all should thank him for his contributions and hope that he continues to share his experiences and knowledge.