So, I’m trying to figure out why no 2 passes are ever the same.
At first, I thought I was skipping steps, but I’m running 85oz/in 2A steppers, at 2.2A (1.1V, I’ve tested at 0.7 and it was like they had no torque at all, I could stop them with a finger. Now I can’t stop them without a lot of effort) with DRV8825, so they should have much more than the required torque. At the same time, it’s not overheating, the drivers are kept cool by a 120mm fan, and I’ve checked all of the steppers and drivers multiple times during the cut, and they don’t get hot, slightly warm at most.
So, ok, maybe it is belt stretch, I’m using a no name belt after all, it could be defective or something, but as far as I know it should be a constant, not something that happens only sometimes on multiple passes.
Then I thought: maybe the bearing are shit/too tight/not tight enough, but they roll easily when unpowered, and I left the machine running some cycles for an hour and the end it was perfectly accurate.
Then I started thinking about the weirder issues. Maybe the import spindle gets too hot, and it softens the toolholder, causing it to flex. That would explain why the cut starts off being pretty good, but then drifts. But I left the spindle running for 30 minutes, and at the end there was no noticeable flex on the holder, no matter how much force I applied (I removed it from the middle assembly, to avoid damaging anything).
I thought about my CAM, but the issue presents itself with really light workloads:
I’ve tested two cuts on plywood, both 2mm DOC, 10k rpm spindle, 3mm endmill (the second cut was made with a fresh one, just in case), 900mm/min feeds. So, everything pretty normal and slow.
[attachment file=69369]
This is a 2d contour of a circle. At the points where it is more squished, the movement slowed a lot, then accelerated, and after a while came back to normal speed.
This repeated for both passes. At the end it came back to zero, but the shape was completely wrong.
So, ok, maybe I’m asking too much of my machine, even if I’ve seen people pushing it much, much harder.
So, here comes test 2: [attachment file=69370]
as you can see, the second pass was larger in a spot.
This is a 2d adactive cut, same doc, and feeds, optimal load 1.2mm, just with a fresh endmill. So, even if the spindle was not pushing hard, it still wasn’t correct. There was no slowdown, and I kept the area clean with a vacuum.
The kicker is: it came home perfectly to X0Y0, I changed the bit to an engraving vbit, and cut the drawing, and it did it perfectly, and again came back to zero. Just to clear up that it wasn’t the spindle wire causing an interference on something.
I don’t know, I’ve run out of ideas.
I’ve changed the board from my previous tests, now I’m using a fresh ramps and mega 2560, with new drv8825 drivers, and I’ve made sure to remove any play from all carriages (in my previous thread there was some, and I’m sure it was causing issues). There are no cracked pieces.
The machine has a 60x60cm usable area, 7cm Z axis, everything was self sourced.
