Troubleshooting
Symptom → fix. Learned on real boards.
Cutting
| Symptom | Cause → fix |
|---|
| Cut far deeper than the trace depth on a screwed-down board | The extra depth for a board held at points. It is a quarter of the arch the map shows, capped; if that is still too much, untick Work the extra depth out from the probe map and set Cut deeper by yourself. Make sure the map is applied — with the warp off the whole range is charged. |
| Head lifts to the top, cuts air or shallow | Depth outside the 60 mm Z stroke → raise the work surface. Zero Z here gives the check the real surface. |
| Cuts through in some areas, not others | The surface is not flat against a 0.15 mm cut → probe the bed. Missed spots afterwards → rework. |
| ✕ marks on the traces, "spots will be shorted" | Nets closer than the bit. Widen the clearance in KiCad or use a narrower bit; milling more carefully will not fix it. |
| VPanel refuses the file with an NC error | A coordinate outside the machine's travel, or a lift above the Z ceiling. The Off the bed check and the Z-reach check name the cause. |
| Wrong depth after a bit change | Re-zero Z after every bit change and after the flip. Never XY. |
Probing and the link
| Symptom | Cause → fix |
|---|
| Connect fails | The Arduino Serial Monitor holds the port, wrong port, or replug and Rescan the serial ports. |
| A whole column of probe points failed | They were outside the machine's travel. The grid is now clipped to the reachable copper, and the job itself is off the bed — see the checks. |
| no touch | Clip off, dirty copper, board not isolated from the bed, or the bit started too high. |
| unstable | A flaky contact: clean the copper, reseat the clip. |
| The probe will not start: "no live position" | The readout has to be alive first; wait a moment after connecting. |
| A lone spike in the 3D surface | A flaky touch. Clear that row and probe again. |
| Nothing jogs after a STOP | Make one Z step first (Page Up); the firmware holds until then. |
| No machine bar, just a sentence | Linux: the link is Windows-only. Export here, send from the CNC PC. |
Panels
| Symptom | Cause → fix |
|---|
| "Two boards overlap" and the export refuses | Move one; Lay them side by side or Butt them together. |
| A board's side is not in the cut-out | Its edge is on the sheet's edge, so there is nothing to cut there. Set the sheet's real size and corner if that is wrong. |
| The panel does not fit the reachable copper | The sheet ends before the travel does, or the boards plus a gap are wider than the sheet. Move the sheet, or cut the boards one after the other. |
Double-sided
| Symptom | Cause → fix |
|---|
| Top mirrored or offset after the flip | Physical flip ≠ expected direction. Fiducial fits cannot catch this. Jog to a distinctive board hole with the top step selected before cutting. |
| "Too far out to cut" | Re-seat the board and probe again; Find it for me is repeatable. The files are not written under a bad fit without asking. |
| NC job ignores my origin | NC obeys G54 only. XY at the machine origin, zero only Z. |
Setups
| Symptom | Cause → fix |
|---|
| "Setup loaded, but … could not be read" | The Gerber folder moved, or the file is from the first interface with something this one cannot read. It says what; check the job before you cut. |
| Can't find exports or setups | Documents\SRM-CAM\exports and …\sessions. Dialogs reopen the last folder used. |
| The window never opens | Read Documents\SRM-CAM\gui2.log. On Linux without a GPU driver, run with GERBER2RML_GL=software. |
Still stuck?
Open an issue with the symptom, the step you were on, and the text of the check or the toast.