Ok, I created a video of the PCB engraving run from the autoleveling job I posted earlier. The system state was unchanged from the job. Before running, I confirmed the zero point was unchaged (using a test probe).
As you can see in the video, some areas are too shallow and some are too deep.
https://drive.google.com/file/d/0B8J3Ge-s_kbOZXBobTJ1RHB3N2M/view?usp=sharing
Here’s what I’m seeing in that video. 1) You have a lot of runout on that spindle to the point where it’s sort of painful to watch how inaccurate these milling lines are due to the spindle. 2) Your PCB flexes as the bit touches it, so your mounting of the PCB is not great. 3) I am seeing that your endmill is actually dragging over the copper which means you have a dull endmill.
The auto-leveller, from what I deduce, did an amazing job for you and the results are way better than i would expect given the runout, dull endmill and pcb flex.
I think you saw a cleaner milling job in your non-autoleveled picture from yesterday because the pcb was flexing against the endmill way more intensely than the auto-leveled run, so it overcame the dull endmill sliding over the copper.
Well, you saw previous photos, where the milling looked great without autoleveling. I have dozens of examples like that. Also, its clear that some lines, right next to each other, were assigned vastly different z levels. For instance, one line starts and ends at z=.09 and it intersects with a line with start/end z=.221. Notice the curve on the lower left. There’s a hard edge where it goes from barely touching to deep into the PCB.
The other run you saw, the milling was clean in both cases. The only difference was that the autoleveled one didn’t engrave deep enough in some places.
Also, the milling bit may be dull, but its a brand new PCB milling bit that has milled only 4 PCBs up to this point (I changed to a new one, its a 10 degree .2mm bit; I’ll change to a .1mm when I can get autoleveling to work properly.)
Hmm, if there is a z=.221 intersecting a z=0.09 then that it as problem. I do know that really long lines do have a problem that I don’t break them into smaller pieces, but that has tended to not be an issue with people’s gcode and I can’t see in yours where that’s the issue. So, can you point out or show a screenshot of the lines where that intersect is happening?
Have you tried slowing down your spindle speed to control the runout? You may find reducing it 20 or 30% gets you a beautiful fine point at the tip rather than the blurriness I see in the video.