CamBam
News:
 
*
Welcome, Guest. Please login or register.
Did you miss your activation email?
January 25, 2020, 11:36:14 am


Login with username, password and session length


Pages: [1]
  Print  
Author Topic: Pocketing. Again.  (Read 484 times)
Dragonfly
CNC Jedi
*****
Online Online

Posts: 2247



View Profile
« on: January 10, 2020, 12:17:18 pm »

Some may get annoyed for which my apologies. But I continue to insist that current 'Pocket' MOP need some tweaking.
From my own experience I came to the conclusion that current algorithm tries to do the final pass at exactly the set step over. At any cost. even if that means doing extra pass(es) and long cut at 180° tool engagement.
And I think this is not right. There should be some margin of step over variation to adjust for best tool trajectory.
If you take a look at the pictures - there is a simple shape for pocketing.
- At step over = 0.4 the generated tool path is not what one expects and is worst approach for milling - doing a full load pass near the outer edge. This should not happen IMHO.
- At step over = 0.35 the tool paths are fine.

Also, I don't think the "optimization" with multi depth level pockets where every other level starts at where the previous had finished is useful. Sometimes I (and I suppose others too) spare extra time to play with step over and start point to make the milling more effective and predictable only to find that the next level starts at another point.


* pocket-040.jpg (217.25 KB, 753x662 - viewed 22 times.)

* pocket-035.jpg (221.99 KB, 833x670 - viewed 21 times.)
* WrongPocket.cb (4.49 KB - downloaded 7 times.)
Logged
dave benson
CNC Jedi
*****
Offline Offline

Posts: 1239


View Profile
« Reply #1 on: January 10, 2020, 14:07:13 pm »

Hi Fly
If I understand correctly.
That file did funky things to CB.
What you can do is use the Finish Stepover value to tune your toolpath.
Here I've left 0.25 mm to clean up.
Dave


* Finish stepover feature to tune toolpaths.PNG (57.69 KB, 848x464 - viewed 14 times.)

* pockets.PNG (89.47 KB, 1202x584 - viewed 18 times.)
Logged
Dragonfly
CNC Jedi
*****
Online Online

Posts: 2247



View Profile
« Reply #2 on: January 10, 2020, 15:47:52 pm »

Adjusting the step over does the trick too, Dave.
But as the second picture shows, I have the feeling that some internal calculations go wrong. This is not the first case I bump into this problem. Even simple shapes like the one in my file or a circle can produce very weird results. Producing sporadic cuts, double cuts or changing the milling direction of a pass.
I suspect it's the floating point aritmetics but am not able to isolate an example. Since some time I've been using Profile MOP with cut width where possible. Same moves but much more stable and predictable.
Logged
Dragonfly
CNC Jedi
*****
Online Online

Posts: 2247



View Profile
« Reply #3 on: January 10, 2020, 16:17:20 pm »

P.S.
I should mention that the offending shape was obtained by using positive 'Offset' command.
Looking at the points collection table I see tiny bulge values where they should not be present.
« Last Edit: January 10, 2020, 16:20:40 pm by Dragonfly » Logged
dave benson
CNC Jedi
*****
Offline Offline

Posts: 1239


View Profile
« Reply #4 on: January 11, 2020, 03:28:13 am »

Hi Fly

Yes this is one of the very issues I ran into making the collision detector.

In a file posted recently where the geometry was from an unknown source.
Pic 1 is of the original Layer geometry as the original tool made it.
pic 2  polylines were not joined.
pic 3 I joined the polylines.

At this point I generated a profile mop and a pocket mop which seemed to work
fine, At this stage you might think that the polyline is ok but this is far from the case.

In pic four I used the collision detector with a cleaning value of zero.
Using it like this unveils all of the dodgy geometry which only shows up in the
CD geometry layers not the original layer geometry.

Pic 5  Reveals that there are an embedded Arc and a microscopic polyline.
Pic 6 Zoomed in very far shows the offending geometry.

I simply deleted them and added a new polyline.

I run into this problem with imported geometry a fair bit, that's why I made the file cleaner as
it's operation is informed from what I learned from the CD and is not the same as the
inbuilt polyline clean CB function.

This is the first thing I do when CB gives me funky toolpaths which can happen with complex
geometry and using nearly all of manipulative functions like "Break at Intersections" and
chopping up arcs.

Mostly when I've had trouble with getting a job to run. it's not been the CB toolpathing that
caused me trouble, but the parent geometry and I didn't have an easy way to check using the
normal CB tools. The collision detector is very sensitive to this however and shows up all the
anomalies in it's layers.

A thing to remember is that the CD mops are "one shot only" that is to say, if you have generated
a tool path and it needs some work on it, then fix what needs to be fixed in the polyline
and delete all the CD layers and mops else they won't be altered on the next round of
mop generation.

As a test I ran the CD over the repaired polyline with a tool that is too small for the job
What you should get is a continuous line tool path. The complete opposite use of  what the tool is
supposed to do. Roll Eyes

The last pic is of the repaired geometry and using a clean up mop with a tool Dia of 0.5 mm.
This continuous line indicates that the line is perfect.
If you use this polyline with a mop and it's not working then it's a mop setting or a bug maybe.
Dave


* 1 Original Layer Geometry as per tool that made it.PNG (113.04 KB, 1148x605 - viewed 14 times.)

* 2 Poly lines were not joined.PNG (55.89 KB, 935x564 - viewed 13 times.)

* 3 Now CNTRL P for Single poly line.PNG (49.75 KB, 890x503 - viewed 12 times.)

* 4 Cleaning value set to Zero shows up the hidden polylines.PNG (20.62 KB, 294x162 - viewed 15 times.)
Logged
dave benson
CNC Jedi
*****
Offline Offline

Posts: 1239


View Profile
« Reply #5 on: January 11, 2020, 03:29:20 am »

more pics

Logged
dh42
Administrator
CNC Jedi
*****
Offline Offline

Posts: 5728



View Profile WWW
« Reply #6 on: January 11, 2020, 06:42:08 am »

Hello

@Dragon: About the extra points in your square (very tiny arcs about 0.005mm), you can remove them with this script

http://www.cambam.co.uk/forum/index.php?topic=5137.0

This is the same as the "polyline/clean" function except that the clean function remove only points that perfectly overlaps ; with the script you can set a distance value. (in this case, I set 0.01)

Note, that after cleaning (8 segments instead 16) the full load pass near the outer edge with the 0.4 step over go away.

Also when you generate the toolpath with 0.4 stepover (and the original 16 segments square), you get a warning "open polygon" that is not present with 0.35 step over, which means that CB is able to detect that something goes wrong, but the warning is not very explicit.

Quote
From my own experience I came to the conclusion that current algorithm tries to do the final pass at exactly the set step over.

Yes it seem it's the case ; on the picture the purple toolpaths are with 0.4 steopover, the yellow with 0.35 ; the last external toolpath exactly overlaps for both stepover.

note2: a polyline arc fit (0.02) do the trick too for removing extra small arcs ; the resulting polyline end up with 8 segments but .... the 4 straight lines are also converted to arcs (with a bulge of: -0.000483)

++
David


* Sans titre-1.png (37.55 KB, 1175x877 - viewed 13 times.)
« Last Edit: January 11, 2020, 07:23:42 am by dh42 » Logged
Dragonfly
CNC Jedi
*****
Online Online

Posts: 2247



View Profile
« Reply #7 on: January 11, 2020, 09:03:52 am »

David,
and Dave,
I admit I've not used your CD plugin. Will install it to see how it works. I did a polyline clean with CB but, as David points out, it removes only perfect overlaps. I must have been quite tired at the end of the day to not pay attention that there 16 segments. I even counted them but it did not ring any bells in my head. Smiley

I've had quite a number of such cases. Even with a perfect circle drawn in CB when the step over and diameter produce a remainder during calculation. Sometimes converting the circle to a poly does the trick and my guess is it's due to rounding.
This is the reason I raised the issue with pocketing again. I still think it needs some more flexibility. In more complex projects one can easily miss that some tool paths are not quite as expected.
There are other issues too when going to next depth level from clearance at plunge speed. It seems it is more frequent and connected to using spiral lead-in.

Thank you for the insight.
Ivan
Logged
Pages: [1]
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.21 | SMF © 2015, Simple Machines

Valid XHTML 1.0! Valid CSS! Dilber MC Theme by HarzeM
Page created in 0.122 seconds with 18 queries.

Copyright © 2018 HexRay Ltd. | Sitemap