Apparently this has happened a few times this week. We got it back on the rails with two people, using the page up to jog it up and push it back onto the rails. The controller wont let you go high enough.
Also the z probe seems to be dead, not sure if its just missing battery.
This is a bit perplexing. The soft limit can be “broken” if the motor stalls due to something jamming the axis. A stall would most likely occur on the lift, where LinuxCNC tells it to move up but the motor jams, fails to lift, and thus is not as high as LinuxCNC thinks it is.
But that would leave the spindle lower, not higher. FYI the primary cause of this would be crud on the rails, in particular epoxy tables being done with less-than-100% cure on the epoxy. The flakes are sticky.
I don’t think this can be from someone setting an offset in the tool table, as that should not affect the Safe-Z since that command operates in Machine Coord system which does not use offsets. Safe Z does not jog up looking for the homing switch, it relies on the machine home automatically established with Home All when it’s powered up.
If the homing switch was off, then it would likely dismount on Home All, and you’d see it sooner.
I’m not sure. I do mean to upgrade this to closed-loop servos soon which is a whole new performance game.
So the wireless MPG pendant won’t let you dial it back down? But the keyboard will? That would be weird and probably key if so. Are we sure that happened?
FYI if you power down the controller box (white power strip), you can hand-turn the leadscrew to raise and lower it
When it’s off the rails like this, it won’t move in z at all without help. PageUp with someone lifting the spindle assembly will get it to raise so you can get it back on the rails. The pendant isn’t sufficient for that, though maybe it’s just that spinning the dial doesn’t give enough power at 1x or 10x step size; I wonder if it might work if you went to 100x step size.
What happens if it stalls going down, if for instance the bit hits the table or piece? Then it will think it’s lower than it is, which seems like it could lead to overshooting going up. And since Safe-Z doesn’t look for the homing switch, there’s nothing to prevent it.
OK, so it’s not a weird software prob.
It would indeed act like this if it stalled going down. But that would generally require a great deal of back force, like you crashed a large bit into the stock or table without having the spindle in. If that had just happened, I presume that would have been mentioned.
The motor has it easy on the way down. And I’ve done a lot of performance tests on it for a lot of repeat high accelerations.
Are we sure this isn’t with the “3D” profile selected?
You’re right, given how often this has happened (I’ve witnessed it twice in the last week; Charlie has seen it once or twice more before that.), it seems unlikely that a head crash preceded each one.
That said, I did stall the z motor that way myself once last week. As you’ve probably seen, the plastic over the step-size button has worn through, making that somewhat flaky. I hadn’t realized that then, so I tried to press it twice to go from 10x to 1x, but it only acknowledged one press, putting it at 100x. So instead of sneaking down, I slammed the bit into the table. But I reset and rehomed everything afterward and didn’t have problems. And now I check the step size display every time. That button is a real crapshoot now; you can get 0, 1, or 2 steps through the cycle from a single attempted press.
But again, that seems unlikely to be a common cause for repeated incidents.
I don’t know what mode (regular/3D) they were in for certain, but since it will boot into the regular mode by default, it seems unlikely they all picked the wrong mode.
I did do performance testing on it last night and it performed correctly the whole time.
I did notice the Z travel limits do let the spindle crash into the table right now, with or without the nut. In the past I did try to keep the limits up with the current spoilboard thickness so it won’t be able to do that. It’s not all that meaningful though because the bit extends a variable length from the nut so there’s no protection against crashing the bit, just the nut itself.
We should replace the V-wheels at this point. They’re the best high-accuracy low-maintenance solution available (they actually perform way better than the linear rails the common mfgs use), but they have run for a crazy number of hrs on the original ones. I don’t see any apparent performance problem going on with them but sure we can replace them, it’s pretty easy.
In a couple of the cases I saw last week and this, it was when the user pressed SafeZ that it overran and jumped the V wheels. I don’t know what level the bit / head was at before that, but it seems to be a common thread.
I did have a head crash too so thats probably the cause
I think we need to emphasize that if the spindle hits anything hard enough to cause a stall, then you need to rehome the machine. And if you’re not sure, go ahead and rehome; it shouldn’t cause you any problems if you didn’t stall but may save the machine or your piece if you did.
