Problem with 4th axis toolpath

[ArtF]

Greg:

  Not many ways around this one. Ill have to add a clearing routine for this first pass, It'll take me a couple days, I had to fix up a problem that showed up in the helical passes. Version 1.64.5 has the helical fix in it, but doesnt yet fix this trouble yet.

Art

[Greolt]

Art

......... . Ill have to add a clearing routine for this first pass..... ......

Perhaps some of those superfluo us passes prior to meeting tangency may need to reintrodu ced. 

They were in fact cutting something before they started on the finish of the involute.

While you are at it I will give you another thing to think about. 

Moving the tool to the Z depth of the base diameter may work out mathemati cally but in practice it is not low enough in all cases.

As an example I have drawn a mod 2.5, 30 tooth, 25 degree pressure angle spur gear.

Below are two pics (zoomed in to show detail) of the gear and tool at the second last flank cut and then the last cut.

As shown on the last flank cut, the material needed to be cleared is missed altogethe r.

Looks to me like we need a progressi ve lowering of the Z axis beyond the depth of the base diameter.

Greg

[ArtF]

Hi Greg:

  This one I think can be left to resolutio n. If the steps/flank is set to 8, then those last two pics woudl do just as you say, but if set to 10, less would be lost, if set higher..e ven less.
  I'll give this some thought but I dont see it as a problem as yet, its normal that with fewer flank steps the tooth gets less accurate, we found in testing that 8 is usually enough for a good tooth as I couldnt see the differenc e between 12 and 8 for example.

  The way I see it, it shouldnt be necessary to go lower than base, the involute curve generatio n starts at the base, so the bottom of tool is at the start of the involute. Its then possible to reach all parts of the involute from that depth at one angle or another. It'd be more a matter of rotating back a bit and moving Y back a touch to get the missed portion in your example for instance. But selecting more flank steps does that in essence.

   I now have a clearing routine developed but not yet released till I run some more tests on it to be sure. ( I hate screwing up peoples blanks..). What I did was make the tool go to the depth it cleared at , only corrected for angular displacem ent, then go pass by pass down to the base circle level before starting the shaving process. Kinda of a double clearance routine. The first pass of the flanking is basically repeated going deeper each pass till we're at base depth, then the rest of the steps are done.Hope fully out soon.

Art

[ArtF]

Greg:

  Those pics did show me though that the last pass shouldnt exist, its the mathmatic al point where the base circle touches the top of the curve.. shouldnt ever be necessary to cut that step.. Ill see its removed..

Art

[Greolt]

  This one I think can be left to resolutio n. If the steps/flank is set to 8, then those last two pics woudl do just as you say, but if set to 10, less would be lost, if set higher..e ven less.
  I'll give this some thought but I dont see it as a problem as yet, its normal that with fewer flank steps the tooth gets less accurate, we found in testing that 8 is usually enough for a good tooth as I couldnt see the differenc e between 12 and 8 for example.

That example above was in fact set at 10. 

But I see what you mean that it all comes down to resolutio n. 

We can always zoom further in and see material left, no matter how high the resolutio n.

What we need is a serviceab le gear, not absolute perfectio n. 

However having said that, experienc e so far has shown me that the tip of the involute is the most important part as far as a smooth mesh is concerned .

It is always the tip that catches if it is a bit sticky.

Greg

[ArtF]

Greg:

 True, that was my point, the stepping will never be perfect .. but perhaps its best to ensure a minimum amount is left at the tip, perhaps a forced cut just a certain distance down as the final iteration? I guess experienc e will tell us, I dont mind fudging about as we see fit when users report what causes binding and what doesnt.

  The full fix will be out soon in terms of the 4th axis work, then we'll just stockpile any experienc es from then on to see what should be modified to make it better..

   The bevel routines will be separate to themselev es I think so that I dont have to screw
around with what works well. I hate it when a fix breaks somethign that works, so Ill try
to be very very carefull from this point.

   Thanks for the analysis though, its great to have multiple minds looking at this, it gets
hard on the head at times.

Art

[ArtF]

Hi Greg:

  Version 1.65.0 is onmline, and I just cut  agear with it to make sure the clearing worked.. Its as many teeth as I could cut, and all went well including the flank clearance .. should help..

Art