LinuxCNC and Autoleveller

[Country_Bubba]

metric is no problem. Actually it is developed in metric and conversions are done internally for imperial.
You should have no problems.

[daedelus]

I ran some experiments yesterday as promised and thought I would share the results here.

Test setup / plan:

Run a simple test pattern with some bits clamped underneath to exaggerate any warping etc. Etch the same test pattern using v0.8.3 and v0.7.7. Both times the units were set to inches and LinuxCNC was used.

This is the initial blank board setup:

Blank board setup

testsetup.jpg (137.48 KiB) Viewed 6770 times

Here you can see the rubber band sticking out the top of the board which I have clamped underneath the board and is clamped down tightly in all 4 corners.

The band gives a general bowed effect to the surface, so if AutoLeveller works, this bow should be compensated for during etching.

So, next I probe and etch the test pattern using v0.8.3 Here is the result:

Test zig zag pattern etch using AL 0.8.3

Test AL083.jpg (127.96 KiB) Viewed 6770 times

The depth of the etch remains constant throughout the test, i.e. the bow is successfully compensated for apart from a little bit at the end which etched a bit too deep. This was because the board was a touch too small for the pattern (miscalculation on the operators part) and as a result the final row of probing did not quite touch the surface and probed a bit deeper. So, given this, I expected the etching to be too deep here. I am very happy with the results and this is perfect as far as this test goes.

Next, I flipped the board over and etched the same pattern on the other side but this time using v0.7.7. The rest of the setup was pretty much exactly the same:

Test zig zag pattern etch using AL 0.7.7

Test AL077.jpg (146.48 KiB) Viewed 6770 times

This time, the board is mostly good but went a bit too deep in the middle. I must have changed the algorithm between these 2 versions at some point.

In conclusion, the later member versions give a better cut, and whilst 0.7.7 is not perfect it still performs well.

I am currently working on another new improved leveling algorithm, which, once complete and tested thoroughly, may be applied to both the member version and free version.

I hope this helps a bit and feel free to ask any questions about this or anything else here.

[putz182]

Excelent...

Can't wait the new version.

Don't know if this can help, but I found out that lower the feed rate when probing, better are the results.

Milling with 0.07mm.