Field Troubleshooting
First, collect this
- The software version:
|
cd /home/reuben/linuxcnc/configs/RoseEngineButler |
- The generator's output (what LinuxCNC will load):
|
cd /home/reuben/linuxcnc/configs/RoseEngineButler |
- The network check (which cards answer, and the Raspberry Pi's wired address):
|
bash /home/reuben/linuxcnc/configs/RoseEngineButler/REB_Change_Network.sh check |
- A copy of /home/reuben/Documents/REBset_v1.ini.
- Whether the machine has an expansion box, and whether Expansion Box - Connected is ticked in the Settings program (Axis Selection tab).
- If LinuxCNC starts: the HAL state, from a terminal while it is running:
|
halcmd show all > ~/all.txt |
- Any error text from starting LinuxCNC in diagnostic mode (see Diagnostics).
LinuxCNC won't start
- Run the generator by hand (above). If it reports a problem, it refuses to write the configuration.
- Check the Mesa cards are reachable: run the network check above, or ping 10.10.10.10 (base card) and ping 10.10.10.11 (expansion card), then read a card with mesaflash --device 7i92t --addr 10.10.10.10 --readhmid (see Mesa 7i92).
- The Raspberry Pi's wired address should be 10.10.10.100 (ip -4 addr show eth0). If it isn't, run REB_Change_Network.sh pi.
- A machine not yet moved to the 10.10.10.x network (Raspberry Pi on 192.168.1.120, card at its factory 192.168.1.121) still runs the base box, but the generator refuses to start with Expansion Box - Connected ticked. Move it with REB_Change_Network.sh (see Installation and Setup for Debian 13 Trixie, Part 3), or untick the expansion box.
- Expansion Box - Connected is ticked but the expansion box is off or unplugged: switch it on and plug it into the switch in the base box, or untick it.
- A card that doesn't answer at its 10.10.10.x address: check its cable to the switch, its power, and that its W6 jumper is up (W5 down). Jumpers are read only at power-up.
- Nothing else may be plugged into the switch in the base box - no router or other computers.
- Look for hm2 / gpio / num_inms errors in the diagnostic output.
"Base box lost contact" or "Expansion box lost contact"
If either Mesa card stops answering while LinuxCNC runs (a network cable, the switch, or a box's power), LinuxCNC drops into E-stop, lights the signal tower's red light and sounds the fault siren (if used), and shows which box it was. Check that box's network cable, the switch and its power, then press F1. While in E-stop, Rose Engine Butler tries to reconnect the card every few seconds; F1 resets the E-stop once the card answers again.
An axis doesn't move
- Is it assigned to a channel, and is the motor on the plug for that channel? See Axis Mapping and the generator output.
- Is that channel on the expansion box (06 - 09)? Then Expansion Box - Connected must be ticked, or the role is left out.
- On an earlier 8-channel box, channels 06 and 07 now belong to the expansion box; move a role on them to a base channel (00 - 05).
- Is its ENA on (green) on the main panel, and the machine powered on (F2)?
- Is the DM542T's 5V/24V switch set to 5V? At 24V (as new drivers come) the driver doesn't respond to the Mesa card.
- Is its Scale plausible? A value off by 1000x moves too little to see.
Limit switch, E-stop, or signal tower misbehaving
See Panel Accessories. Limit Jacks 4 - 6 (on the expansion box) only work while the expansion box is connected and ticked.
The customer's tool table or custom HAL changes disappeared after an update
Update Rose Engine Butler sets local changes to tracked files (such as REB_Custom/REB_Tool.tbl and REB_Custom/REB_Custom.hal) aside before pulling and puts them back afterwards. If the update changed the same lines, or the machine was updated before 30 September 2026 (when the update didn't put them back), they are still in the stash - see Updating Rose Engine Butler.
Going back to the previous version
See Updating Rose Engine Butler. A machine already moved to the 10.10.10.x network needs its network put back before it runs a version from before the two-box design.
|
eMail comments to |
