A Challenge – How Would You Machine This Part?

[JasonBModerator @jasonb]

So will you be doing a detailed write up for MEW

 

Is it teh LSM one as I thought they did the castings? or are they just the blanks.

 

Jason

 

Edited By JasonB on 04/06/2011 13:24:23

[Anonymous]

Hi Jason,

 

Yes, I will be writing this up for submission to MEW. And before there’s any whinging about yet more CNC, the majority of the article will be about the design of bevel gears, as well as a summary of the machining methods, manual and CNC. The actual CNC machining part will be quite short, as it’s not that interesting.

 

I’m building my SCC Burrells from the Filby drawings, as modified by LSM; but I bought my drawings, and all the castings, from John Rex in Pontefract. I’ve had a quick look on the LSM website and they don’t mention castings for the differential pinions, just the crown wheels. Similarly John Rex sells the crown wheel castings, but not the pinions. The John Rex crown wheels are just blank castings, I assume the LSM ones are similar.

 

Regards,

 

Andrew

[Steve WithnellParticipant @stevewithnell34426]

Hi Andrew,

 

Questions questions…

 

Which solid modeller are you using?

 

Is 10 teeth pushing it a bit? I thought 12 was considered the lower limit?

 

The design book I was given is “Gear Design Simplified” by Franklin Jones and its the best laid out /easiest to follow I’ve seen so far.

 

I’ve only cut parallel depth bevel gears so far, but they do work and don’t whinge much…

 

 

Regards

 

Steve

 

[Anonymous]

Hi Steve,

 

There’s nowt wrong with questions!

 

I’m using Alibre for solid modelling. I have the ‘Expert’ version, so including sheet metal and all the bells and whistles. I’ve been using it for about 6 years now. I bought it primarily for my work, but it is also very useful for model engineering. For anything other than a simple arbor, or similar, I tend to design in Alibre to ensure that parts fit together, will move together without interfering and then be able to produce 2D drawings. It has a few foibles but overall I think it is good value for money. Solidworks was outside my budget. In the distance past I’ve used ProEngineer, but didn’t get on with it, very difficult to use.

 

That’s the book! First time I used it, mumble, mumble, years ago I borrowed it from the library at RAE Farnborough, but eventually had to buy my own copy from Amazon.

 

As I understand it 12 teeth for involute gears is about the limit without introducing undercut, which weakens the teeth, I think. Certainly the standard B&S style involute cutters only go down to 12 teeth. But there’s no theoretical limit, just practical ones. Well may be a one tooth gear would be a bit difficult!

 

Regards,

 

Andrew

Edited By Andrew Johnston on 04/06/2011 16:01:32

Edited By Andrew Johnston on 04/06/2011 16:02:27

[JasonBModerator @jasonb]

Andrew, there is a bit about machining the diff for a Burrell (singel cyl) but quite a bit is the same and they are John Rex parts on Conrads site though he went for parallel depth.

[Steve WithnellParticipant @stevewithnell34426]

 

This is the set of bevel gears I made for my Stuart Victoria. I’d like to claim I made these with a set of false teeth, so I don’t need CNC, but it wouldn’t be true. These are 40DP parallel depth, so not as manly as Andrews pucker 6DP gears, but they did the job. I set is 1:1 and the other set is 2:1. I ignored the advice regarding odd numbers of teeth because the engine was never likely to run more than an hour or two any way.

 

 

[Anonymous]

JasonB: Thanks for the link to Conrad’s site; an interesting read. At least I didn’t fall into the trap of not having the number of teeth on the crown wheel divisible by 3 when I re-designed the differential using parallel tooth bevels.

 

Steve: Wow, nice looking gears! I may be asking you for advice in the future; there are two bevel gears needed for the governor on my traction engines. I think they’re about 20DP so milling by the CNC route will need a small cutter. So I may use the parallel depth method for them. Just one comment; the space between the teeth on the two gears to the left seems quite deep in relation to the tooth width. Is this a consequence of the parallel depth method?

 

Regards,

 

Andrew

[Anonymous]

Now for a brief discussion on the various methods of producing bevel gears.

 

First, thanks to Nick for the drawings, very good! I’m sure the method would work, as would the use of a hacksaw and file. But my word it would be a bit tedious. I wonder what sort of templates would be needed to ensure accuracy, and how would you make them?

 

As an aside Richard mentioned the Hiraoka books; I’ve heard good reports about them. I have a friend who has an obsession with Shay locomotives; I’ll have to ask hin if he has the books.

 

Now onto the suggestion by ady of using a shaper. I considered this too, essentially I guess one is making a crude Gleason bevel gear planer. It seems a lot of work though, and secondly my shaper is sadly out of action at the moment, awaiting repair of a broken casting.

So finally we come to the production of the bevel gears on a milling machine. First, a statement; it is not possible to machine a true bevel gear on a conventional milling machine, with one caveat.

 

JohnS has succinctly described the two approximation methods, so I won’t repeat them here, but just make some comments.

 

On the first method (that requires ‘adjustment’ afterwards) as I understand it the gear is designed using the DP at the outer diameter of the gear. An involute cutter is then selected that has the correct tooth form for this DP and the appropriate number of teeth. But, here’s the difficulty, this cutter has to pass through the tooth gap at the small end of the gear. So you cannot use a standard involute cutter. Instead you need a special involute cutter that has the correct form but is thinner than standard. Can anyone confirm this? As far as I’m aware these cutters are no longer available; at least I couldn’t find any, not even from China or India. I have made bevel gears in the past using this method, but I had the facilities of the main workshop at RAE Farnborough behind me. They lent me the special cutter required. Here are the gears prototyped in aluminium, the final gears were made in EN32 and were case hardened:

 

If you look closely at the lower gear you might be able to spot the ‘deliberate’ mistake.

 

The parallel depth method avoids the need for a special cutter as the design is based around the DP at the smaller end, and hence a standard involute cutter is fine.

 

Now onto the post by JohnS, really interesting. I downloaded the Gearotic demo last night and had a play. It looks very good, but I cannot output data from the demo so cannot see what the G-code looks like. Are the bevels generated true bevel gears, or do they use the parallel depth method? The other thing I couldn’t see how to control was the stepover of the tool, which determines the surface finish, trading this off against run time. The cutting time JohnS got is much quicker than mine. I wonder if this is due to a coarser stepover, or a more sophisticated toolpath. My CAM system will only interpolate X and Z, The Y axis never moves and A is only used as an indexer. It would be more efficient to interpolate 3 or even 4 axes. This would also allow one to generate a ‘Sturz’ style cutting strategy which would be more efficient and give a better finish for a coarser stepover. Further I don’t see why a tilting dividing head is needed? I didn’t use one, but just used the 4th axis as an indexer, no cuts took place while the rotary table was moving. It would have been possible to cut the pinion using a 3 axis mill, but this would have required a long series end mill. A 4mm ball end cutter with a flute length of at least 22mm would be a bit delicate! I will be cutting the crown bevel gears using 3 axes, as the teeth are much flatter, so we don’t need the 4th axis.

 

Now, back to the caveat. It would be possible to machine a true bevel gear on a manual mill by essentially replicating what the CNC mill does. In other words a long list of X, Y, Z and A co-ordinates that are stepped through by twiddling the handles. It wouldn’t half be tedious though.

 

Regards,

 

Andrew

Edited By Andrew Johnston on 05/06/2011 11:32:20

[John Stevenson 1Participant @johnstevenson1]

Andrew,

Replies to questions from the Gearotic questions.

 

Bevels are true bevels and not parallel depth bevels. Not sure what you see in the demo program but when you go to the output manager where you enter tool size, feeds etc you have a choice on depth of cut per pass, step over and how many ‘slices’ per pass you want to do, default is 10 but can be altered to suit. I tend to use 8 to get a decent finish but reasonable cutting times.

 

Gearotic interpolates in 3 axis X, Y and A as it roll the gear, it basically mimics the Gleason but uses a rotation cutter in stead of a planing action.

 

Gearotic uses a tilting dividing head as it doesn’t move Z when cutting, head is tilted at root angle which is parallel to the bed. This is done to make the maths easier otherwise it would be interpolating in 4 axis.

 

Now, back to your caveat.

 

Somewhere I have a program based on a spread sheet, not done by me but by a guy who’s name I can’t think of at the moment to give recognition.

 

In it he uses the No 1 cutter which is from 135 teeth to a rack and takes a pass then moves over in Y, rotates the dividing head and takes another pass etc on both sides.

Basically what Gearotic does but all manual moves.

 

To be perfectly correct you would need a straight sided rack cutter and not the 135 tooth cutter which has some involute on it but I’m assuming this was the best idea using readily available off the shaft cutters.

One number 1 cutter will do all gears in a range.

 

Perhaps the program could be automated for doing your small bevel gears ?

 

John S.

[Nicholas FarrParticipant @nicholasfarr14254]

Hi Andrew, I,m pleased you liked my drawings, they were done while polishing off a large glass of Cótes Du Rhóne, and moving onto a Sauvignon Blanc. They were (if you hadn’t guessed) a little tongue-in-cheek in their own way, (I thought the use of a mini grinder would have said it all).

I had gathered that they were done by CNC, but as I’ve had no experience with CNC or cutting bevel gears for that matter, and the fact that others had already suggested CNC, I thought I’d submit a some what comical alternative. Yes I agree with you, it would be a pretty tedious method, especially to get accuracy, and any templates would take as much, if not more work.

They look a fine pair of gears BTW, and hope your cast iron ones turn out just as good.

Regards Nick.

Edited By Nicholas Farr on 05/06/2011 17:25:13

[mgjParticipant @mgj]

Well its good thing they had CNC when they made traction engines!!

 

Andrew – you really have to ask why? Or I do, because it seems to be a very wonderful solution to a problem which never existed in the first place. Unless one simply wants to do it this way for fun – which is as good a reason as any, if one has the time and the pennies..

 

One simply doesn’t need a true bevel for this application – and with the clearances in a TE diff, it wouldn’t matter if you used lantern pinions.

 

A parallel depth bevel is, accepted, an approximation, though it does offer certain advantages in running in certain circumstances. Despite the (very small) approximation,  at 20 RPM you ain’t going to notice, nor even when winching, and certainly not at the 1/2RPM differential velocity when the bevels are doing their differential bit. –

 

And we go to all this trouble for something that can be chopped out of a standard bit of cast iron or steel? The calculations take about 5 minutes on a scientific calculator, order the cutter – it will be smaller than nominal but as you say, a standard off hte shelf DP cutter – , three evenings, a vertical mill, a simple dividing head and for addition, a DTI to set the blank roll. It will take longer to make the arbors to hold these blanks than it will do do the gear cutting, and you have saved the trouble and cost of getting everything cast anyway.

 

And why do we need 4 axis CNC for something that was originally made in a blacksmiths shop?

 

Forgive me, but it seems to me, this is a wonderful way of making a very simple bit of machining very complicated and expensive, for nil gain in performance.

 

Did you hear BTW of the bid BAe put in in the 1980s for some NBC fan housings (sort of aircon units) to go in armoured vehicles. Most people used pressed steel housings. The BAe bid was 4 times the price of anyone elses- but then they intended to start with a huge chunk of steel and a milling machine?

Edited By mgj on 05/06/2011 19:05:32

Edited By mgj on 05/06/2011 19:07:59

[JasonBModerator @jasonb]

As mgj says the full size were just as cast. If you have a copy of Gilberts Suppliment there are a couple of very good pics of the diff wheels on a Fowler and Marshal and you can easily see the sand cast texture.

 

I think they were run in with a mix if sand & grease

 

You don’t need teh accuracy, it snot like a car diff going a 110s of rpm and finely adjustable preload. You can’t adjust the preload on a TE and when the axle or 2nd shaft warms up and gets longer all those fine fits will all go to pot anyway.

 

You could always machine them from test wax which would be quick and use that as a pattern for lost wax casting, the finish would be more than usable, these are the as cast teeth on my fowler

 

Jason

 

Edited By JasonB on 05/06/2011 19:56:47

[Anonymous]

Hi Nick,

 

Yes, I rather gathered that the drawings were intended to be tongue-in-cheek, but nonetheless amusing. They certainly lightened my day; excellent choice of accompanying beverages by the way!

 

I shall be interested to see how I get on cutting the cast iron. I’m nowhere near the speeds recommended by the cutter manufacturer for aluminium, in fact I’m running at the bottom end of the speeds they recommend for cast iron. So as a first pass I’ll stick with the same speeds and feeds and see how I get on.

 

Regards,

 

Andrew

 

[Anonymous]

Hi JohnS,

 

Thanks for the explanations of Gearotic. Now that I look at it with fresh eyes I see that the cutting information is available on the output page. I also now understand why the 5 axis is needed. Tormach do a 5 axis add-on, driven rotary table with a manual tilt, but it is pretty expensive and would be expensive to ship, plus VAT, plus import duty plus….well I expect you get the picture. I could machine a fixed block to tilt my rotary table, but I think I’ll stick with my existing method for the moment.

 

However, I am interested in Gearotic. I did look at it more than a year ago but decided against buying it. But now, after consulting the keeper of the purse strings, I think I will make a purchase. If nothing else I will be interested to look at the generated G-code and plot out the consequent toolpath. At some point I will need to machine a helical gear for the steering on the traction engines and Gearotic will give me another option, ie, CNC versus manual machining. Even if I decide to machine the helical gears manually Gearotic may help to generate a 3D model.

 

Thanks again.

 

Regards,

 

Andrew

[Colin Jacobs 1Participant @colinjacobs1]

Posted by Donald Mitchell on 02/06/2011 16:52:53:

I’d use my grandpa’s old hacksaw and finish the teeth off with a nice rat tailed file I bought in Woolworths about 30 years ago; I can work to about 2/10 of a thou with that file, who needs CNC

Donald

In Bonnie Scotland

And much more rewarding I bet

[John Stevenson 1Participant @johnstevenson1]

Andrew,

Sorry have I misled you ?

5 axis isn’t needed just a tilting dividing head in place of the 4th axis which will still be A.

Then the normal X, Y, and Z axis.

 

John S.

[Anonymous]

And so to the delicate task of replying to JasonB and mgj. I hope you don’t mind me making it a combined reply.

 

I am fully aware that the differential gears in full size are ‘as cast’ and that they are slow moving and lightly loaded. I suspect that at the time it would have been pretty difficult, and very expensive, to get bevel gears machined. I believe that the Gleason bevel planer was invented in 1874, but I assume it took quite a while to cross the Atlantic, and become established. I hadn’t thought of having the gears cast, although it would have been an interesting exercise. Of course in order to make the patterns I would have had to go through the same design process, which is what took most of the time!

 

mjg – Well why not?!

 

I could have made the bevels by the parallel depth method and as mentioned I did do all the calculations for a range of pinion and crown wheel sizes to get ODs that were a close match to the existing drawings. However it would have still involved quite a lot of work in re-designing other parts and ensuring that they still fitted together. I must be slow, because I’m sure I couldn’t do the necessary calculations in 5 minutes, and it would take me longer that three nights to do the machining. As an aside I am actually building two engines, and this design has three pinions in the differential, not two. So that’s six pinions and four crown wheels.

 

There are some things in mjg’s post I don’t follow. What are the advantages of parallel tooth bevels? Errr, I thought that was exactly what my method did do, it makes the gears from a standard length of bar? I don’t follow the reasoning for using a vertical mill, surely a horizontal mill would be easier? Lastly I don’t understand why more than one arbor would be needed. I only used one arbor to make the pinions, took me about half an hour to make. It doubles up for use in the lathe to make the initial blanks, on the mill to cut the teeth, and back on the lathe to finish the front and back chamfers.

 

I have a vertical mill, a horizontal mill, a CNC mill and a universal dividing head, so I could have chosen any of the methods mentioned. I’ve even got a hacksaw and round file, but not the patience! After a lot of thinking and designing, and false starts, I did it the way I did because it was a challenge, I’m interested in the design and manufacturing process. I feel that I now know a lot more about the way bevel gears are designed, the mathematics of involute curves and the tooth shapes and how to calculate them, and the means of making the gears, including the approximate methods.

 

At the end of the day, each to his own, and good luck to them.

 

Regards,

 

Andrew

 

PS: Did you know that the use of the involute curve of the circle was first suggested for use in gears by Euler?

[Steve WithnellParticipant @stevewithnell34426]

Posted by Andrew Johnston on 05/06/2011 10:45:21:

 

Steve: Wow, nice looking gears! I may be asking you for advice in the future; there are two bevel gears needed for the governor on my traction engines. I think they’re about 20DP so milling by the CNC route will need a small cutter. So I may use the parallel depth method for them. Just one comment; the space between the teeth on the two gears to the left seems quite deep in relation to the tooth width. Is this a consequence of the parallel depth method?

 

Regards,

 

Andrew

I don’t know! (might be I cut a bit too deep!) Here is the spreadsheet I used to design them:

http://www.btinternet.com/~steve.withnell/lathe/gears.xls

 

I built it originally from the Ivan Laws book, then made some mods based on Franklin Jones.

[Anonymous]

JohnS,

 

Presumably the dividing head still needs to be driven by a stepper motor? I do have a universal dividing head, but it is not motorised.

 

Regards,

 

Andrew

[mgjParticipant @mgj]

Andrew- of course, if one wants to, and I did make that point. But turning these things out in ali first to make a pattern for a set of true bevels seems a very long and relatively expensive way around the houses.

 

Further, the earlier part of this thread created an impression (to me) that one needs to go to such lengths to make a set of diff gears for a TE, when actually one does not, and a very satisfactory set of gears can be knocked out with the much more common vertical mill and some kind of indexing head.

 

I just had the impression that perhaps the approach was more how to make a set of true bevels, when possibly a more holistic view – how to make a satisfactory TE diff might have advantages.

 

[JasonBModerator @jasonb]

I don’t think Andrews intension is to use the alloy ones as a pattern, I feel he was just testing the programme on alloy and will machine the final ones from CI bar.

 

A horizontal miller is obviously the better machine for the job as it will be a lot more rigid when you get upto 6DP but not many of us have one which is likely why mgj suggested a vertical setup. You cab also use a rotary table and index plates with a vertical mill as the TR can just be skewed round to the required angle.

 

J

Edited By JasonB on 06/06/2011 18:27:03

[Richard ParsonsParticipant @richardparsons61721]

Anderw In the Hiraoka Climax locomotive the author scratch builds everything (even the wheels). From his writings I understand he has only a lathe, a vertical mill (perhaps) and a pillar drill. His bevel gears have an offset axis. That is the two shafts by pass each other at right angles. He describes not only the tooling, the main parts of which are the ‘wedge’ which sets the bevel angle, the work piece carrier which doubles as dividing head, but also the cutters and how to make them (the way he does this needs no backing off!). His gauges are very simple.

As you have a Cleaver Number Contraption (CNC) then OK use it. He does it all by ‘Mandraulic methods and has FUN, but you take half the time (and can drink tea whilst the machine works for you. When I am cutting a slice oh a bit of real railway rail (the only source of good steel I have) I enjoy listening to a strongman sawing (my power saw) whilst I contemplate a large mug of Rosie and a Zemle es Lekver – a jam roll-.

Good luck

Dick

Edited By Richard Parsons on 06/06/2011 18:27:16

[mgjParticipant @mgj]

Jason – I thought they were to be cast, but it could be a mix up in terminology.

 

I agree about the milling machine – I did my 8DP gears on a Dore Westbury, the largest being about 9″ in dia. 6DP will be a bit of a shaker – which is one reason why on my vertical mill I got one that would do a sensible bottom speed (54 RPM IIRC) and not the 100+ rpm that many machines will do. A horizontal would be better, but if one gets the head down and keeps the quill extension to a minimum it will be fine (but I’m not using a DW for the 4″ engine!)

 

To finish really. Were I building a gunsight with zero backlash and all then ‘d specify true bevels and all. When I built the DW, Quorn and GHT dividing head, then setting to .0002″ was normal (and boring and painful) for certain bits. But when one is building a TE, and one is struggling to keep the hornplates more or less flat after all that riveting, and they are what carries crank and 2nd motion shafts, the gears will fit where they touch, like it or not!

 

And especially if it is a road locomotive with springs and the axle is moving around!

 

Edited By mgj on 06/06/2011 19:11:11

[John Stevenson 1Participant @johnstevenson1]

These were traction engine bevels true but it’s the learning curve that matters, the next job could be a set of camshaft gears to a DOHC engine or perhaps cast gears are good enough here ?

 

Remove the blinkers.

 

[EDIT]

 

Giles Parkes ran me up tonight, now Giles is 138 years old and dead set on cutting gears the old fashioned way, no CNC for Giles.

 

However the reason for the phone call was he’s been thinking  [ good pastime that  ] and wanted to know if with the correct programming the CNC could put spiral gashes into a hob blank and then go on the generate the hob teeth WITH relief, all at one go.

 

I must admit I had never thought of this and yes it can be done which would make a very accurate hob without all the banging that occurs with the backing off process.

Helical gashed hobs will also cut smoother .

 

Nice one Giles, proves at even at 138 he can still see the curvature of the earth.

 

John S.

Edited By John Stevenson on 06/06/2011 19:37:29

[JasonBModerator @jasonb]

Pic of a set of embrio pinions from a full size Fowler resteration may be of interest, From the machining marks I would assume they were done by CNC.

 

 

However the holes for the pinion pins were very definatly done using manual methods

 

Photos taken from Carl’s Fowler A4 rebuild on TT forum

Edited By JasonB on 06/06/2011 19:49:01

[Author]

Posts