45 degree worm M2.5

[ArtF]

Yeah, Im still seeing things I dont like as well.. Still testing, Ill let you know..

Art

[ArtF]

Phil:

  lol..that display is really hard to analyse by eye, I now need thicker glasses. Anyway, try a thickness of .1mm and a tool of .1mm in diam.. does it look quite clean now?
I think the overlap your seeing is due to the correction pre and post slice for the tool diameter. Looked at straight in with a larger than 0 diam , it gets more and more overlapping at that scale. With a dia of .1 on the tool you can more easily see the shape of the tooth at center.

  The reason this is so hard to see is we're dealing with a 4mm  radius object being cut with a 1mm bit.. thats quite a ratio, try it with a
50mm radius and a 5mm bit ( 10:1 ) to see much better whats going on. Try a 12 tooth for example, it becomes much more obvious that the path is correct.. Even a 10 tooth of the same module shows a nice path..its just the smallness of the numbers on that 4 tooth that makes things very hard to grasp.

  There WAS definitely an error though, the roughing was badly broken in terms of the numbers and I can easily see it causing what happened
on that cut, so at the very least its probably worth a retest if you wish, Im pretty confident of the numbers now. (Not that thats  a definitive proof. :-) ), so we owe you a thank you for the bug, it was a bit complex to find, but your cut showed the proof.

Art

[philbur]

Hi Art, still seems to be the same problem. The lower tooth "space" has been run with rough and fine and on the upper tooth "space" I just allowed the fine to carry on and cut a second space with out any roughing.

I'll be away working until next weekend so no chance to try anything until then.

Phil :)

PS: using version referenced in post #27

PPS: the surface finish on the upper fine looks much worse in the photo than it actually is.

[philbur]

It's an M2.5 cut with a 3mm end mill.

here's the project file:

[ArtF]

Phil:

Thanks, that test was invaluable. I can see the  change from what I did, Im a bit stymied , but Ill analyse it and
see what needs to be done..

Art

[ArtF]

Phil:

  Is the zeroing on that exact? I ask because the software numbers seem to be perfect. I think Helical is becoming a smokescreen on this problem though and making it hard to see whats going on. Can you switch to a normal spur? Try the same cut just without helical, Im curious if the same effect shows. What the photo seems to show ( compared to what I'm seeing in the numbers..) is that the tool is zeroed about .2mm ( or whatever the depth of those gouge lines are) too close to center...

  A test done on a spur of the same size would tell me I think. Also , Id try a run say.. 0.3mm off the bar for a zero. Im thinking the
path is too close to zero, this would produce exactly what I'm seeing. You see, if the fine pass is too close, because it cuts on a tangent, no gouge
will occur, the tooth will simply be a bit thinner and very hard to notice, but the rooting, because its basically milling on a plane will cause a gouge. So I'm now curious if this is a situation where the zeroing distance is just slightly off..perhaps from software or from setup.. not hard to be out .3mm on my rig.
  It may be why I haven't seen this yet. Try a spur in the program, notice how close to profile the spur gear gets cut on its toolpath. ( Easy to see with no helical rotations involved in the display.) So if your system shows gouges on a spur, we than at least know its systemic to something ( either your center or my models numbers..), but if your spur is perfect I can then suspect its something in the rotational calculators.

  I appreciate your testing on this, No rush on this , we can discuss it when you return, dont work too hard. :)

Note:  In Theory as one increases the zeroed center distance the finish of a cut tooth smooths and gets thicker, this is because the tangental cuts skim a bit longer surface than called for and the rooting cuts less deeply in the tooth profile direction.

Art

[ArtF]

Phil:

  PS:

    To illustrate, look at the rooting ( on very fine passes ) on this version of your worm ( as a spur). You can see the rooting doesnt touch
the profile ( thought its very very close.). Any zero shift will gouge. In fact on reviewing all this Ive decided its way too close and will be changing it
by the time you get back to be far less close on roughing. That in itself may solve the issue, but I just wanted to show how the spur may better visualise whatever is going on.

Art

[philbur]

Art,

The z zero was within + or minus .03mm.

I notice in your spur gear screenshot that the line of the cut profile doesn't follow the line of the tooth profile but gradually converges on it, also it does still leave a small anomaly at the tip of the tooth, and both of these points are in common with the issues seen on the helical. Have you run that spur with a larger tool and coarse feed? Unfortunately I can't load gearotic to my employers' PC.
Also I have a comment on the roughing process. Firstly I noticed when I allowed the cutter to run the fine programme without first roughing it handles it very well. So this raises the question how to give the user more control over the aggressiveness of the operation. Watching a 2hr + 4 tooth cutting operation get boring really quickly after the first time.

Obviously with roughing you want to take as larger cuts as the tool/machine can reasonably handle however the current roughing also provides the finishing cut below the baseline, for which you would want to take small steps. My proposal for involutes would go something like this:

Step 1: use an end mill to rough slot to maximum depth and width, cut per step user defined (leaving material for a finishing cut - user defined). It would need to auto do some widening if the useable flute length is not sufficient for the full depth.

Step 2: Use the same end mill to rough the volute by tangential shaving cut per step user defined (leaving material for a finishing cut-user defined).

Step3: Use the same end mill run a finishing tangential shaving, cut step size user defined. Note: Why not nominate the tangential point on the cutter a short distance up the side (say 15% of cutter diameter) rather than the tip. This would avoid any risk of step cuts dues to inadequacies of the machine, cutter or the user.

Step 4: Change the cutter to a bullnose and finish roughing and do a finish pass on the root. Rough and finish cut size to be user defined.

Just a thought,

Phil :)

[ArtF]

Phil:

>>I notice in your spur gear screenshot that the line of the cut profile doesn't follow the line of the tooth profile but gradually converges on it, also it does still leave a small anomaly at the tip of the tooth, and both of these points are in common with the issues seen on the helical. Have you run that spur with a larger tool and coarse feed?

  No, unfortunatley I'm not setup at the moment to run such a test so Im working solely with numbers.  But lets discuss that photo and its convergance. The toolpath shown is exagerated of course as its very small step down, but in roughing there is no rotation. So lets just look at that spur gear. With no rotation the path above should generate a tooth shape just barely away from the profile at the top, but a bit further away at the bottom. The photo uses the actual numbers for position so in theory the shape above shows a very tight root for a tooth that will not rotate during the cut.
  I did this for a reason, in early GM I had many people tell me I left too much on the flanks during the root, it caused the bit to be stressed
(Primarily of course with small bits) during the tangental run.

>>Step 1: use an end mill to rough slot to maximum depth and width, cut per step user defined (leaving material for a finishing cut - user defined). It would need to auto do some widening if the useable flute length is not sufficien t for the full depth.

  This is basically what is being done. Look at the photo, this starts at the top, and works its way down clearing each depth from side to side until
it misses the profile by the setting of 10% of the bit diameter plus a clearance factor. THEN, two passes are made, they are made with rotation, first the straight wall flank of the root is rotated to be straight up and down and a pass is made ( with side offset ) so the bit side just cuts that wall.

  The photo doesnt show the tangental path, but in the tangental passes the flute is placed so it just rubs the profile edge, thus making a clean line up the involute.

  Now, your cuts shows something bad happening with the root cuts only, and it seemed to show the root was cutting too wide. So again, lets look to the
roughing pass in the photo, how could it cut wide and into the tooth? I see only 1 way, if the toolpath is moved downwards we get exactly as you got in
cutting. And I do mean exactly. Its what the numbers show. Now, does that mean you were off zero? No!, it could mean IM off zero. Im crunching the numbers
to see where I can be off, but logically, if the path shows proper, but the cut isnt, there must be an offset somewhere..the path , so long as the display
doesnt show a cutoff must be correct, but perhaps the output code isnt. Im still looking but cant find it yet.

  Now I do think cutting a spur is the secret,.. same machine, same operation, just 4 tooth spur instead of helical. This at least makes it pretty easy to tell at what stage things are getting hosed. Since there is no rotation during rooting you should be able to see if the initial rooting is gouging the sides as it roots, or if the tangental shave shaves off any gouge marks, which is what should happen.

  The reason GT shows a simulation picture was exactly for me to troubleshoot such things, it uses separate code which draws the output by the numbers on its blank. Any gouges should show up as profile crossover, and Im not seeing it at all here. BUT its possible my zero is slightly off, this WOULD show on such a small gear and likely not at all on others. A spur run should show us what it is..

    You idea's on more complexity in the machining as options, certainly for persons such as yourself cutting things such as worms is noted, I do
intend to completely re-code the GCoder at any rate, so I will put more options in there at that time, but by the numbers the current code should cut that worm just fine.. whatever is wrong is very close to being found. Its warming up here finally, so it wont be long before I can run some tests myself to see where any problem may lie..


  If you were zeroed within .01mm, then depth shouldn't be an issue, well..just image the toolpath in the photo, a .1mm drop in that path wouldnt kill you, though it would gouge the top of the profile very slightly... Whatever is off sees to be .5mm off or so going by your photos...which made me wonder what happens if the zero was raised above the stock as a test..

  Take note too though that the reason this is so critical is because it only has 4 teeth. Normally the rooting on its side to side journey never comes to be almost on top of a profile, this happens only below 6 teeth or so, the fewer teeth, the more critical the root becomes as the top pathes become parallel to the actual tooth profile. ( see the top of the rooting in the profile..very tight at the top..) Still..if I can find why its so critical on the 4 tooth I'm sure it improves all tooth counts..

  On a side note, all of this does point to a simple rule, the geometry shows me " The Zero distance in 4th axis becomes more critical to be exact as the tooth count drops due to the perpendicular nature of the tip curve to the tool in low tooth counts." . I always find such insights interesting.

Art
 




               

[philbur]

Art, is the fact that even for you spur screenshot the cutter does not correctly follow the involute profile but converges on it not an indicator of a systematic error in the gcode at least.

Phil :)

[ArtF]

Phil:

   Looks like it ...but not really. Consider if the bit is down in the root. With a 1mm bit the toolpath must remain .5mm
from the side. Else it gouges. Now consider each point up the involute slope. For the outside tip of the bit to miss
gouging the edge, it can come closer and closer to the profile to maintain a .5mm distance on the flat plane of
the bottom of the bit to the sloping involute profile. Measure at any point from the toolpath sideways to the
profile and it is .5mm. It just LOOKS like its getting closer..but again, it IS critical and only happens with
very few teeth because normally the outside tip of a tooth is never on such a 90 degree tangent to the tool..

  Hmm, hard to explain in words. Draw a circle. Starting at the top , draw a .25" line from the circle out .25" and put a dot. Move down 1/8" and repeat till you get 1/2 way down. Actually, thats easy to draw so here it is in the attachment. Its an illusion that the squiggly line is getting closer to the circle on the X plane. All the squigglies are the same distance out on that plane, and its the only plane that matters. The same effect is in the toolpath, only worse because the involute shape is going onto a flat plane ..faster...its as if your zooming into the very top of this illusion below.  Now it is possible thats where the numbers are bad..Ill check that to be sure, the trouble with tangents is that they approach infinity as they approach 90 degrees. Again, 4 teeth puts us very close to 90 degrees..good chance thats where the error is..
   Hmm, you know, Pulling up the bit zero to .1 or .2mm may be all it takes to fix it anyway, as the tangents shaving would still work the same and all thats affected is the rooting angles..which are so critical that a .1mm upwards motion of the rooting bit is amplified greatly in effect at those angles..Ill give it some thought..

Art


 

   

[philbur]

Art, yes I see now about the horizontal distance remaining constant.

Phil :)

[ArtF]

Phil:

  Now imagine the entire path is lifted just marginally. The side clearance stays the same on the root, but
the clearance from the profile grows exponentially as the zero is raised. ( It also shrinks exponentially as its lowered.),
Its why I asked about center, it wasnt I figured you were off by much, if any, but on 4 tooth gears in particular it becomes so critical is ain't funny.. ( Where on 15 tooth gears it isn't very critical at all..).
  Maybe I should build in something that raises the root by a clearance amount when the two planes get so
parallel.. :)

Art

[philbur]

Yes but if the issue was a case of off center then I don't think the artifact would appear equally on adjacent tooth flanks.

Phil :)

[ArtF]

Actually , it would be identical. Look at that example toolpath photo gain. Mentally lower the toolpath,
as you see both flanks now get equally hit.. identically so in fact. :)

Art