The sheet is already lost. The last panel is offset, the profile is cutting air, and the operator’s first instinct is to blame the CAM file. That instinct is usually wrong, and chasing it wastes the only thing a postmortem is good for: narrowing the failure class before the next sheet goes on the table.
Here is the uncomfortable fact that reframes the whole investigation. ShopBot’s own specification sheet describes the PRSalpha as using closed-loop steppers with 7.2:1 gearboxes and constant positional feedback. If an obstruction is hit or the tool is cutting too fast, the drivers attempt to correct position before activating an alarm, stopping the machine, and displaying an alarm on the monitor. Once reset and homed, cutting should be resumable (ShopBot Buddy/PRS spec sheet, JUN2023).
That changes the diagnosis. A PRSalpha that ends a job in the wrong place either alarmed and you missed it, or it was never moving the commanded distance in the first place. The open-loop “the stepper just gave up silently” story belongs to the PRSstandard, which ShopBot documents as open-loop steppers with 3.6:1 gearboxes and no positional feedback — steps can be lost and not noticed until the file completes or the operator stops. If you are running an alpha and blaming torque rolloff at 100% rapid, you are debugging the wrong machine.
First fork: wrong distance, or lost position mid-cut?
These are different failure classes with different evidence, and treating them as one problem is why postmortems stall.
Class A — the machine is not moving the commanded distance at all. Command a 10-inch move and measure it. If it does not travel 10 inches, the problem predates the cut. ShopBot’s troubleshooting guidance is blunt: this is usually a unit-values problem, and the documented fix is to reset default settings and load the settings file that matches the tool (ShopBot FAQ: My Tool Isn’t Moving the Correct Distance). The same document warns that a settings file for the wrong tool type will not run the machine correctly — a Desktop setting file on a PRSalpha is not a subtle error, it is a different machine.
Class B — the machine moves correctly in a test but loses position during the file. Now the suspects are obstruction, hold-down failure, feed rate, or a mechanical fault the loop could not correct before the alarm threshold. The closed-loop driver will try to recover; what it cannot do is recover from a pinion that is not fully engaged in the rack.
The unit-value check comes before anything mechanical
Before you touch a wrench, read back the unit values. For a PRSalpha, ShopBot documents defaults of 2482.8171 steps per inch on X and Y and 2979.3805 on Z, with a resolution multiple of 5 on all axes (SBG 00171, Settings Unit Values). Pull them with the VU command and compare. If they do not match, you have found your failure class in under two minutes, and the CAM file is innocent.
The same document is explicit about the order of operations: check the rack and pinion first. A worn rack or a pinion not completely engaged in the rack should be checked before proceeding with unit-value adjustments. Adjusting unit values to compensate for a mechanical mesh problem hides the fault and guarantees it recurs.
Where in the file did it happen?
The rapid/cutting speed split is the most useful piece of evidence in the file log. ShopBot documents PRSalpha rapid transit speeds up to 1800 inches per minute and cutting speeds up to 720 inches per minute (spec sheet). Those are different regimes with different loads.
A skip that occurs during a rapid move and one that occurs in a full-depth cut point at different causes. If the offset appears after a long traverse between nested parts, look at acceleration, clearance height, and whether the gantry clipped a clamp or a hold-down. If it appears mid-cut, look at chip load, material movement, and whether the sheet lifted. The move log tells you which happened; guessing does not.
This is also where the CAM side earns a fair hearing. A post-processor that emits rapids at 100% into a region the machine cannot traverse cleanly — a clamp in the path, a low clearance plane — will produce a collision the closed loop reports as an alarm, not a silent step loss. Check the clearance plane and the rapid moves before re-posting the whole file.
Mechanical suspects and their signatures
Each of these leaves a different trace, which is why the postmortem record matters more than any single test:
- Rack-and-pinion mesh: wrong distance on a commanded move, often axis-specific. Check engagement before touching unit values.
- Worn pinion: progressive drift, worse on one axis, visible wear on the gear teeth.
- Obstruction: sudden offset at a specific file location, often near a clamp or the edge of the sheet. On an alpha, expect an alarm.
- Hold-down failure: the sheet lifts, the cut wanders, and the offset follows the part geometry rather than the axis.
ShopBot’s user guide is direct about why the operator is part of the detection system: never leave the tool unattended while it is running, because a cutting error or workpiece slippage the tool cannot detect might occur (SBG 00142 User Guide). On a closed-loop alpha, the alarm is the backup; the operator watching the cut is the primary detector.
The postmortem record: five fields, every time
Without a repeatable record, every recurrence restarts from zero. Keep it short enough that it actually gets filled in:
- Axis and measured offset. Which axis, how far off, measured — not estimated.
- Unit values read back. VU output for X, Y, Z, compared against the documented PRSalpha defaults.
- Mesh condition. Rack and pinion engagement checked, wear noted.
- File position. Where in the file the offset appeared, and whether it was a rapid or a cut.
- Hold-down method. Vacuum, screws, tabs — and whether the sheet moved.
Five fields, filled in at the machine, turn a mystery into a trend. After three occurrences you can see whether the fault is axis-specific, speed-specific, or material-specific.
Re-home and resume, or scrap the sheet?
The decision rule is about risk, not optimism. ShopBot’s spec sheet states that after a closed-loop alarm, once reset and homed, cutting should be resumable. That is the documented path for an alpha that alarmed cleanly.
But resumability is not the same as salvage. If the offset is small and the remaining features are non-critical — a pocket that will be hidden, a profile with a sanding allowance — re-home, re-zero, and resume. If the offset crosses a joinery face, a visible edge, or a tolerance the assembly depends on, the sheet is scrap. Re-cutting one sheet costs material and an hour; shipping a part that is 3/8 inch off costs a callback.
The postmortem exists so that decision is made from evidence rather than hope. Verify the unit values, check the mesh, read the move log, and write down what you found. The next sheet depends on it.
FAQ
My PRSalpha stopped mid-job with an alarm. Did it skip a step?
Not in the open-loop sense. The spec sheet describes the alpha as closed-loop with positional feedback; the drivers attempt to correct position before alarming. The alarm is the failure signal, not a silent skip. Reset, home, and check the postmortem fields before resuming.
How do I know if my machine is an alpha or a standard?
ShopBot’s settings document identifies alpha control boxes as larger, with serial numbers beginning with PRSA; standard boxes are smaller, with serial numbers beginning with PRS-S. The drive architecture differs, and so does the failure mode.
Should I adjust unit values to fix a wrong-distance problem?
Only after checking rack-and-pinion mesh. ShopBot’s documentation says to check the mesh first, then reset to the correct settings file. Adjusting unit values around a mechanical fault hides it.
What are the correct unit values for a PRSalpha?
2482.8171 steps per inch on X and Y, 2979.3805 on Z, resolution multiple of 5 on all axes, per SBG 00171. Read them back with VU and compare.