Joint 2 error only while cutting

Hello - I've had this machine running LinuxCNC (Version 2.7.1) for about 5 years now. It has worked super reliably. Without making any changes to the machine, when I run a program now it'll trigger a "joint 2 on limit switch" error and trigger the estop. The limit switches aren't close to the being touched by any part of the machine.

So far I've tried:

• Air cutting (this works just fine)
• Dropping the feed rate super, super low (still triggers joint 2 error)
• Replacing the Z axis limit switches (still triggers joint 2 error)
• Watching the Z limit switches / estop with the hal meter (the resolution I need may be too low to trigger? I wasn't able to see anything while triggering on the switches while the program was running; I was able to see it when manually pressing the switches)
• Starting LinuxCNC through the command line; the output that it logs when the limit is hit is "Error on axis 2, command number 484" (I'm not sure what command number means in this context).

Things I've noticed:

• When this happens, it shuts down the machine momentarily (E-stop is engaged). Whatever is triggering the limits is reset immediately, I don't need to check the 'override limits' checkbox to get the machine to work again.
• The Z-axis position doesn't change like it does when manually triggering the home / limit switches.
• It doesn't happen when cutting in the Z direction (which is a small percentage of the time)
• It hasn't happened with a rectangular / circular pocket cut (half stepover for the cutter)
• It happens when cutting contours (full bit width is engaged) - making me think it's either something with vibration, or electrical noise (motors working harder)
• When I manually trigger the limit switches, the Z axis position changes in LinuxCNC (pressing the top home switch makes the machine think it's at +0.25; pressing the Z max limit switch makes it think it's at -3.5). When this is happening while cutting, it doesn't change the position of the machine which makes me think it's something else that's happening.
• It is always my Z axis that shows the error; could be something with the way I have the axes wired that's causing it to trip?
• I usually use a 1/8" bit, but I'm using a 1/4" now - unlikely to be related as I've cut aluminum with 1/4" and didn't experience this (I'm cutting plywood right now)
• I am using Fusion 360 to output the nc files.

Attached are the hal / ini files I'm using for the machine.

Any guidance for where to go from here would be greatly appreciated. I'm confused what would cause this start unexpectedly and I'd like to understand what's happening before trying something like adding debounces to the limits before knowing what's triggering the estop.

My guess would be a loose connection, bad switch, broken
or shorted wire... I guess it could also be EMI but that isn't
something that typically changes a lot over time.

I would set up halshow to monitor motions limit in hal pin
and wiggle cables around to see if you can cause the limit signal
to go high.

My guess would be a loose connection, bad switch, broken
or shorted wire... I guess it could also be EMI but that isn't
something that typically changes a lot over time.

I would set up halshow to monitor motions limit in hal pin
and wiggle cables around to see if you can cause the limit signal
to go high.

Halshow is exactly the kind of thing I was looking for!

The wires passed the preliminary wiggle and jiggle test. Their resistance readings seem right too (~0.3Ω regularly, open loop when switches are pressed). Maybe while cutting something different is happening, but halshow isn't showing me visual differences while jiggling.

Did you change the wall socket where the machine was plugged in?
Was there a storm near you lately?
Did the control box get hit hard by something/someone?

Wall socket is the same, no storms since I used it last. Control box is hasn't been hit. The power switch is off when I'm not using it, but there is still a parallel port connection - maybe something could have happened from the computer to the controller. Weird it's only the one axis though.

The other thing I changed earlier this year (and have put a lot of miles on the machine since then) was to hook up a touch probe and swap the Z axis from two dedicated inputs to a shared home / max limit input. I'll try undoing that tomorrow and see if it behaves differently.

You might be onto something PCW - this might not be LinuxCNC related at all.

I jumped the switch inputs at the control box after homing and it worked great. Swapped the wiring around (I have two independent 2-conductor wires running to the Z-axis for switches) and it happened again. So both of the wires broke at the same time? They were installed together, and it has been cold in the shop; maybe the temperature was enough to fatigue the copper in both of them.

I'll swap the wiring out and see what happens.