Calibration passed. I created a 12"x12" test square in CAD and saved the .dxf. I imported the .dxf in ESTCAM and it aligned to a 1 inch grid showing 12x12". I defined my bit as a 1/4" Up Spiral and created the GCode and loaded it and zeroed the home (about 56" from the lower left post). The Maslow followed the tool path and completed the job with no errors, but the square measures 1/16" too large in both the x and y dimensions. I measured the width of the cut made by the tool path and it is 1/4". (I was thinking I may have miss-measured the bit diameter.)
Any ideas what I should double check or repeat? Thanks!
I will change the home and cut another test box. Thanks! The line width is 0 in CAD, and I’ve sent .DXF files to cabinet shops to use their CNC before with good results. I’m going to sleep up this today and do some more PD in the short term future.
How about this idea:. If someone can post a 12" test square for 15/64" material cut down say 3/16th with tabs and a 1/4" up Spiral bit
when importing a drawing (svg/dxf) into a CAM program, it just has the lines so
you have to tell it which to do.
if you are doing this in a CAD/CAM program (such as fusion 360) it knows the
3d result you are going for and makes these decisions automatically.
if the error was the diameter of the bit, it would be obvious that this was your
problem, since you are off by much less than that, something else is happening
(possibly happening as well as this sort of problem)
It is also possible that you are just hitting the limits of accuracy of your
maslow. When you did your calibration, do you know what fitness score you got?
The maslow is aiming for an accuracy of 0.5-1mm and 1/32" is in that ballpark.
Thanks for your comments. I re-ran calibration with a fitness score of 0.5193282821041084 (is this good?) I am using a 1/4" and creating the part on the outside of the imported DXF file created in CAD. It shows up in the CAM grid as a 12"x12". with the outside (and touching) the line shown. I cut three squares and they measured 12 x 12 1/16", 12 x 12 3/32", and 12 1/16 x 12 1/8". This accuracy is OK for organic shapes, but I would like to get things to 1/32" or better for assemblies. Any ideas? The test squares were cut about 1 foot from each other in the center of the calibration area.
Thanks for your comments. I re-ran calibration with a fitness score of
0.5193282821041084 (is this good?)
That is estimating that the anchor position in known with an error of ~2mm
(1/fitness)
I am using a 1/4" and creating the part on
the outside of the imported DXF file created in CAD. It shows up in the CAM
grid as a 12"x12". with the outside (and touching) the line shown. I cut
three squares and they measured 12 x 12 1/16", 12 x 12 3/32", and 12 1/16 x 12
1/8". This accuracy is OK for organic shapes, but I would like to get things
to 1/32" or better for assemblies. Any ideas? The test squares were cut about
1 foot from each other in the center of the calibration area.
you are on the leading edge of testing accuracy here, so we don’t know for sure.
One thing is that the maslow is very repeatable, so if you can map out the
errors, you can compensate with them in software (similar to bed leveling on 3d
printers where something takes gcode that assumes everything is flat and square
and distorts it to match reality). right now there is no software that will do
this, but it can be done.
As for the accuracy of your machine, I would look closely at how rigid it is, is
there any chance that the frame is flexing (recording a video of the anchors and
then viewing it fast forward helps spot such things.
I’ve been pretty distracted with other bugs, but getting back to focusing on accuracy is top of my list once I get through those. That does seem worse than I would expect, but I will need some time to dig into testing what the accuracy on my machine is now and then focus there before I can give exact numbers
Thanks guys for the input and attention. I will be making some more tests when I get some more scrap sheet stock off the rack. Perhaps a grid of drilled holes over a larger area. I’ve had the M4 disconnect a couple of times during my testing (but my laptop remained connected to the internet) so I am also going to shoot this issue with a network heat map and maybe moving/adding a node on my wifi. Baby steps.
