I finally figured out basic toolpaths in Fusion 360, and my first order of business was to check the accuracy of my machine setup. Here’s the simple shape I created and cut in redwood with a #201 bit:
As designed, the center square is 100mm on each side. I used a 3D adaptive contour, leaving 0.03mm for a 2D contouring finishing pass. Here’s a diagram with my Starrett digital caliper measurements:
So, my Y-axis results in dimensions that are too 0.15% too long, whereas the X-axis results are within 0.05%. Is there anything I can do about this? (I’ve read about entering GBL commands for scaling).
I’m wondering what you-all think of the cross-dimension difference of 0.14mm. I assume this comes from the squareness of my assembly. Should I loosen the Y rails, gantry platforms, and hybrid table bolts and try to square it up better, or is this about as good as I should expect?
I know it may seem I’m being overly critical, but what I’ve learned with new tools is that time spent at the beginning setting them up as accurately as possible pays benefits down the road.
Thanks, is the $100 variable retained within the Shapeoko controller through power cycling, or do I have to enter it each time, or somehow get it inserted into every file I sent to the unit?
It’s retained in the controller until you (or some rouge sender) explicitly change it again.
What you may find is that when you get everything calibrated to make that 100mm square perfectly, your 200mm square is off. Or the next 100mm square you make in the other corner is off. When you change end mills, it may be off. When you switch harder or softer materials, it may be off.
Don’t be discouraged by this. Even machines much more expensive have similar issues, and someone like @RichCournoyer could take you down a rabbit hole of how a truly accurate part needs to be designed/milled…he seems like he knows what he’s doing. We convinced @Julien to use optical mouse parts to accurately map his machine’s accuracy across the wasteboard, but I think his trivets would turn out just as excellent had he not.
The Pro should be REALLY accurate. I’d calibrate your belts for the size of stuff you may typically make and let it ride until you need to calibrate again.
Sorry if you thought I was discouraged - that isn’t my feeling and wasn’t my intent to convey at all.
As a traditional modern woodworker I’m all too familiar with the many variables that come into play in setups. And we haven’t even talked about the nature of wood movement - that redwood I milled will have different dimension in width (across the grain) tomorrow morning. I’ll probably cut a piece of MDF for my next run - at least that would move evenly. Also, many woodworking hand tools like try squares are only milled accurate to 0.001" per inch, which for the 4" (approx) square would be 0.1mm.
While I have a few projects in mind for my Shapeoko, the first one is a 12 piece hexagon. Built with traditional good woodworking tools, I was able to get angular accuracy of 1/10º, which, of course, when assembled meant I had a 1.2º gap! The traditional method would be to incrementally remove material from half of the 12 pieces until I got tight joints - I’m to only need light touch up sanding out of the Shapeoko for that.
One tip I picked up along the way when going for precision is to pull the gantry all the way forward to touch the front plates, then power up. In hopes they pickup the same starting step.
The OCD in me had me redo the X-Axis belt. As you know, the belt loops through and so there’s a “tail” of the belt that lies underneath the tensioned belt. That tail was hanging off to the side and wouldn’t lie straight. So, I moved the carriage to the left side and undid the belt attachment at the right. Then lined things up and re-tensioned.
And, then realized that I would need to recalibrate again!
So, I reset $100 and $101 to 40 and recut another test square. Guess what? The square came out very very close to square - so much so that the variation between X and Y is about the same as the comparing the two X or two Y measurements against each other. I do believe that this time I didn’t pull as hard on the belt as I had when I first assembled the machine.
Other people have pointed this out before, but just another confirmation that belt tension does matter.
