Tangential Saw shaving 4th axis

[David]

Hi Art

A good day today. :)
Reversed A axis rotation as suggested and proceeded to cut a trial gear of 20 teeth 2 module using the previous parameters.
I used just rooting and toothing and had no problems producing a respectable gear with no root walls.
Without touching anything (same blank) I ran again but this time with rooting and root wall no toothing, which ran correctly and finished the gear. I have attached a couple of pictures.
Considering the blanks are large washers from the local hardware store, the saw at least half past its best, the mandrel scrap brass sticking too far out of the chuck and unsupported (the last two to aid vision and photography) the magnified tooth view is surprisingly good.
So far no code produced giving wild swings of A and deep Y cuts. Next job to try all three: rooting, root wall and toothing in one file.
Now it is beginning to work as it should I think this method has true potential with fine saws for cutting clock gears in brass.
Great work Art
Regards
David

[ArtF]

David:

  Excellent!!! . Very happy to see that result, and even happier to see the quality of the gears.
Its always good to see visual verification of such things. The code is produced very differently
that Gearotics old methods, is fairly new ans little tested, so Im always looking for verification.

Great work!

Thanks for working with me on it,

Art

[SkyMoBot]

Those do look quite excellent.  I'm going to have to get a set and try that out.  How long did it take to cut those?

[John S]

I have cut some small ones using the older Gearotic method and they came out well but one thing I found is that with thin saws they want to deflect so I used two heavy steel cheek plates either side of the saw with just enough tooth sticking out to get full depth and then a bit.

That really speeded things up and improved the quality.

[David]

Hi SkyMoBot

Glad you like the gear, even if it is just an old washer the cuts do seem pretty good.
I ran the program twice on the same blank with different things checked  as Art and I were looking at a particular issue
Feeds speeds were deliberately slow, X clearance generous; guessing if ran all three parameters together (rooting, root wall and toothing) it would have been about 2 hours.
I have produced the code for a 96 tooth mod 0.6 gear which I will try soon. Because of the smaller teeth each one will take a lot less time (about 90% lines of code per tooth) to cut, but a finer depth of cut might be better for small teeth so I guess around 3 hours. According to the software and my calculations a 0.6 mod can still be cut with a 0.015" saw but I will purchase some 0.2mm and some 0.3mm on Ebay and try those in the future.
Hope that helps
David

[David]

Hi JohnS

I had wondered about using supporting washers for the finer saws, thanks for the information and confirming the improvement it gives.
I will give that a try before I settle on actually producing mandrel and blanks for real the clock gears
Regards
David

[ArtF]

Hi:

  I use the supporting washers trick on my table saw for its blades on hardwood, it is a good way to
stop deflection. A a word about time... if you select a higher depth per pass you may ind it much faster,
in theory the saw should be able to take a fair amount due to its sinusoidal width in the direction of cut.
Secondly, I default to 10 segments , JohnS proved quite well that 8 seems optimal for time vs quality
and can save you two passes per tooth.

  Hopefully, this year we can have our simulator, Im still reading and researching the topic..

Art

 

[JustinO]

Nice gears considering the frugal cutters. You might improve the finish by lightly sliding the saws on a diamond hone to remove the most extremely "set" teeth.

Art, do you have a shaving option for cycloidal toothed gears?

--Justin

[ArtF]

Justin:


  You can do cycloidals, no special code is needed, G2 uses the tooth shape from the dxf as the shape to tangentally shave.
How well it works may vary, but the program wont complain if you do a cycloidal. The new algorithm is a generlize one, meant to work
as well as it can on any tooth shape...

Art

[David]

Hi Art

Tried a 96 tooth 0.6 module gear today with less success.
Saw width I appreciate is a little wide at 0.016" with a maximum suggested of 0.0175" so it should be OK as a trial, no too large warnings in gearotic.
Two problems occurred, on the first tooth shaving cut the tooth is all but machined away, the last 2 shaves cut air.
The dreaded 90 degree A rotation is back along with the full depth Y plunge just about clears but fouling the mill spindle is worryingly close.
Not sure if these smaller gears are the area you felt might need more work, I am happy to wait if that is the case,.
I still have a traction engine to finish and a clock to start  ;D
Attached a first part of the code
Regards
David

[ArtF]

David:

I see what you mean, The lines at line 92 take a terrible jump.

Can you'll post a snapshot of the screen that created it.

Im thinking it may be an overflow. The problem is the math that determines
the max diameter isnt exact, its a bit heuristic, so if its wrong, the saw may be a touch too large,
which may cause a nasty math error. But thats really good, if I can figure out exactly how to reproduce this
code, I think I can fin ally nip it, its an obvious overreaction to something in there..

  It seems after the 90 things go back to normal for the next tooth. This does seem reminisent of a bit
being too large. I often enter 0 as the diameter to see the toolpath, ( which usually looks good, ) and then
increase slowly, as you get larger youll see the toothing passes move away from the flank, when the path
on each side crosses over the other side, the bit is too large.

  Its a hard one to explain, but when you see it it gets more obvious. SO if you can , post a snap, and
Ill dig into it, and if your playing, set a zero and slowly increase to saw size to see if it looks as if its too
large  a bit.. I highly suspect its got something to do with small numbers of some sort..

Art



Art

[David]

Hi Art

First screen shot is the 96 tooth 0.6 module  with 0.015" saw as requested. I regenerated the code and still got the A 90 degree and deep Y plunge.
Second screen shot is same but with 0.001" saw, code generated also has the A 90 Degree and deep Y plunge.
Hope this information is helpful
David

[ArtF]

David:

Nope, all I get is good paths, no 90's.. Can you post the project file, maybe theres something weird in it..

Art

[David]

Hi Art

Just had a thought, as a Mac user I design either on the Mac or on a virtual PC, I wonder if that is the problem.
Not sure why you would get a different result on a mathematical process.
Before we do anything else I will take the project file into the workshop and generate the code on the PC that drives Mach 3 and the mill
Will post results later

David

[ArtF]

David:

  Thx, Im sure we can narrow it down. Im thinking its the root wall routines failing.. so turning them off by unchecking RootWall may make the 90's go away.. but I find it strange I cant duplicate it.. Ill bet its one  of those "Whoda think that" moments when I find it.. :)

Art