Real life observations

Here's the place for discussion related to CAM/CNC and the development of the Path module.
Forum rules
Be nice to others! Respect the FreeCAD code of conduct!
Post Reply
cahlfors
Posts: 91
Joined: Tue Jan 24, 2012 8:27 pm
Location: Sweden

Real life observations

Post by cahlfors »

I know you like real life examples and pictures, so I an planning to provide some here. The objective is of course not to criticize, but to help developers improve the available tools.

The part is not centered in the stock. The reason is that I did a first test in a piece of wood, that happened to be slightly smaller than my aluminum stock. Resizing the stock is a lot of work (lots of dependencies), so I just skipped that part. The idea with wood is that it is a whole lot cheaper to scrap and start over again, should Path or I do something wrong. I am an electrical engineer, so do not have proper machinist schooling. This is just for my hobby projects.
20210406_160921_RESIZE.jpg
20210406_160921_RESIZE.jpg (132.76 KiB) Viewed 1271 times

One first, very positive, finding so far, is that what you see, is indeed what you get. The difficulty is in interpreting and scrutinizing every detail in the generated path, such as tool engagement and direction of cut. On top of that, you also have to consider part and cutting tool holding, such as vises, clamps, collets and chucks. All of these can interfere.

The part is milled with the bottom first, then turned over and a new origin is probed.
Skärmbild från 2021-04-03 20-26-56.png
Skärmbild från 2021-04-03 20-26-56.png (53.72 KiB) Viewed 1271 times
Face milling
The bottom (step 1) is milled with a 50mm face mill of this kind:
20210406_102035_RESIZE.jpg
20210406_102035_RESIZE.jpg (119.21 KiB) Viewed 1271 times

Generating a toolpath for a mill bigger than the part does not work reliably. I laborated with stepover, boundary shape and pattern. At 90% stepover and line, it was possible to get two passes with approximately 50% tool engagement, which is what I was after, but the path is not symmetrical. Very strange.

The line pattern does not make lines, which I was expecting. One line across, reposition, and make another line with the proper stepover and respecting the cut mode (climb/conventional), is what I expected. In this example, the first facing operation is made to full depth in just one pass, but if it is done in two or more passes, the tool will find its new depth at the end of the previous pass and follow the path backwards, disrespecting the cut mode. With a large face mill of this kind, this is a problem.

The lower shelf (MillFace lower) was done with the help of a helper object. I see that there is now a dressup that does essentially the same thing. I was unable to use line here, so tried other strategies and ended up with spiral. Of course, with spiral, it is only possible to respect the cut mode if you are going around a part. Ideally, I would have used line here, doing the outer part climb milling at 50% tool engagement, lift, go back and do the inner part of the shelf the same way.

For the top side, I gave up the large face mill. It did not provide a nice surface finish that I was hoping for anyway. Instead I used a 12mm end mill.
It was easy to select the three horizontal surfaces as region, although they are at different depths. Nice!
With "face region" and "offset" pattern, the cut mode was respected on the entire surface. Nice!
Skärmbild från 2021-04-06 14-10-46.png
Skärmbild från 2021-04-06 14-10-46.png (57.45 KiB) Viewed 1271 times

On MillFace003, the line pattern creates a zigzag pattern that does not respect the cut mode. I found no way around that. It also does not quite reach all the way to the "wall" of the part.
Disastrously, on the next cut depth, it starts the processing from the wrong end, creating a slot with 100% tool engagment, leaving a very ugly finish as result. I did not observe this in the path. The chips in the picture are welded to the part. This is often the result of "conventional" cutting, as each cut starts with the cutting edge rubbing against the wall before the pressure is enough to break through and start the cutting action.
20210406_102928_RESIZE.jpg
20210406_102928_RESIZE.jpg (150.74 KiB) Viewed 1271 times
MillFace004 shows the result with 50% tool engagement.
I ran out of picture space here. Pls see next post!

100% tool engagement can be ok, if it is planned for and the cutting depth and feed rates set accordingly, but not as a surprise when 50% has been set... But again, this information was there in the path. I failed to see it.

Profile
With lead in, it is a perfect tool - no issues, but I wish that we will get a tool that allows us to do multiple horizontal passes. If my stock had been thicker, one pass would not have been enough to reach the target shape. But I think that is for another tool. Profile is designed to cut things out of a slab, board or sheet metal - one cut only.

Pocket with spiral pattern
What a charm! It generates very efficient tool paths that allows you to cut holes better than drilled and slots without breaking the tools or overheating. One little issue, though. By default start depth, the pattern starts a millimeter or two inside the part. You have to trick it to start above the part to get it right.

I can't add more pictures here. They will be in the next post.

OS: Linux Mint 20.1 (X-Cinnamon/cinnamon2d)
Word size of OS: 64-bit
Word size of FreeCAD: 64-bit
Version: 0.19.
Build type: Release
Branch: unknown
Hash: e8566f22bbeb0b7204e3c45519d0963e8881100b
Python version: 3.8.5
Qt version: 5.12.8
Coin version: 4.0.0
OCC version: 7.5.1
Locale: Swedish/Sweden (sv_SE)
Last edited by cahlfors on Tue Apr 06, 2021 2:35 pm, edited 1 time in total.
electrical engineer
Ubuntu or Mint
Mendel 3d-printer/OctoPrint
LinuxCNC
Optimum BF20L mill
cahlfors
Posts: 91
Joined: Tue Jan 24, 2012 8:27 pm
Location: Sweden

Re: Real life observations

Post by cahlfors »

20210406_155642_RESIZE.jpg
20210406_155642_RESIZE.jpg (156.54 KiB) Viewed 1268 times
50% tool engagement conventional cut.
20210406_162205_RESIZE.jpg
20210406_162205_RESIZE.jpg (129.41 KiB) Viewed 1268 times
Wood prototype and actual part.

/Chris
electrical engineer
Ubuntu or Mint
Mendel 3d-printer/OctoPrint
LinuxCNC
Optimum BF20L mill
cahlfors
Posts: 91
Joined: Tue Jan 24, 2012 8:27 pm
Location: Sweden

Re: Real life observations

Post by cahlfors »

I meant to attach the file:
elevation.fcstd
(417.48 KiB) Downloaded 30 times
/Chris
electrical engineer
Ubuntu or Mint
Mendel 3d-printer/OctoPrint
LinuxCNC
Optimum BF20L mill
User avatar
sliptonic
Veteran
Posts: 3459
Joined: Tue Oct 25, 2011 10:46 pm
Location: Columbia, Missouri
Contact:

Re: Real life observations

Post by sliptonic »

Really interesting thread. Thanks for taking the time to write this up!
User avatar
onekk
Veteran
Posts: 6207
Joined: Sat Jan 17, 2015 7:48 am
Contact:

Re: Real life observations

Post by onekk »

From the images, it seem that you have had problem with XY zero.

And don't know your actual settings, but it seems that you have ad a X shift that is then propagated on the successive jobs.

Bad finishing could simply be caused by that "fact" that your CNC have tried to mill a "too big" piece of metal due to the X shift.

The threeshold in the aluminium piece is the proof of the worng zero.

chances are some offset where not applied if threeshold width is maybe equal to the "stepover" or half ot the "tool diameter", maybe simply some "tool diameter compensation" was skipped by Path and not inserted in the GCode.

To see what is gone wrong, the original Gcode files is needed to be analyzed by someone and some information about your mill and the postorocessor you have used.

I know that there were had some problems with "splitting output file" in which a "tool change" Gcode was missing, (but I'm using my bad memory).

Hope it helps.

Regards

Carlo D.
GitHub page: https://github.com/onekk/freecad-doc.
- In deep articles on FreeCAD.
- Learning how to model with scripting.
- Various other stuffs.

Blog: https://okkmkblog.wordpress.com/
cahlfors
Posts: 91
Joined: Tue Jan 24, 2012 8:27 pm
Location: Sweden

Re: Real life observations

Post by cahlfors »

Sorry, not able to upload the gcode files here. I have used the LinuxCNC postprocessor, tweaked for my setup.
There was no shift during machining. As I said, what came out, was indeed what went in. It is all available to see in the fcstd file. I should have spent more time doing that... :lol: :geek:

Anyway, the accidental "shelf" was easy to remove manually. The part is now cleaned up and installed in my project. It works beautifully. Thank you all developers for a fantastic job! It is SOO nice not having to export to HeeksCAM and doing the postprocessing there. It was a rather manual process and easy to make mistakes. Now I can do it all in FreeCAD, seamlessly, and with much greater functionality! AWESOME! :D
electrical engineer
Ubuntu or Mint
Mendel 3d-printer/OctoPrint
LinuxCNC
Optimum BF20L mill
User avatar
onekk
Veteran
Posts: 6207
Joined: Sat Jan 17, 2015 7:48 am
Contact:

Re: Real life observations

Post by onekk »

Wooden part is different from aluminium part, so I guessed the shift from the threshold in the aluminium part that is not present in the wooden part.

finish on aluminium part could depend on excessive tool engaging so no room is left for the chips to be cleared from the work done so some chips will be "remachined" and the finish is due to the chips catched between cutting edge and workpiece.

Many information is lacking to give some hints by some competent guy (I'm also not an expert). Maybe there is around some machinist in this forum, if not try to post the images on a machinist forum and see if my guess about feed and speed is right or wrong.

Regards

Carlo D.
GitHub page: https://github.com/onekk/freecad-doc.
- In deep articles on FreeCAD.
- Learning how to model with scripting.
- Various other stuffs.

Blog: https://okkmkblog.wordpress.com/
Post Reply