Use feed hold only for a controlled motion pause when the machine has room to decelerate safely and no live tool, fire, electrical fault, loose fixture, or person in danger remains hazardous. Use a soft reset to abandon controller execution, not to prove the machine is safe. If there is immediate physical danger, use the exact machine’s documented emergency procedure immediately—do not wait for software to respond or spend time collecting evidence.
The distinction is not the button label. A sender may call several controls “pause,” “stop,” or “reset,” while they affect different parts of the system: motion, controller execution, spindle or coolant outputs, and physical energy isolation. In standard GRBL v1.1, feed hold pauses motion; a soft reset halts and reinitializes the controller; neither is, by itself, a complete-machine emergency function or proof that all energy has been removed.
Match the Response to the Hazard
A wrong artwork path moving through open air is a different situation from a clamp collision, a loose workpiece, smoke or fire, an electrical fault, or a person entering the work area. In the first case, a controlled pause may be appropriate if the machine can stop before contact and the tool will not remain dangerous afterward. In the second group, the immediate priority is the machine’s documented emergency procedure—not a software command.
Hypothetical router example: a router is cutting a sign when the programmed path begins moving toward an uncovered clamp. If safe deceleration clearance is already established and there is no imminent collision, feed hold can stop motion in a controlled way; the spindle may continue running, so the operator must still keep clear of the cutter. If the clamp has already been struck, the workpiece is loose, or the cutter is creating a hazard, use the machine’s emergency procedure rather than trying to manage the event from the sender.
Hypothetical laser example: a laser job begins producing unexpected smoke or flame. The immediate action is the machine’s documented emergency procedure and fire-response procedure—not a feed hold, reset, or screenshot. A software pause does not by itself establish that laser energy, electrical power, smoke, or fire risk has ended; exact laser-output behavior depends on the machine’s firmware, configuration, and hardware.
Feed Hold Is a Controlled Motion Pause
In documented GRBL v1.1, the real-time feed-hold command places the controller in a suspend or Hold state. If the machine is moving, it decelerates to a stop and then suspends; if it is jogging, feed hold cancels the current jog and flushes remaining jog motions from the planner buffer.
Feed hold can be appropriate when the problem is the programmed motion itself, there is sufficient safe deceleration space, and stopping motion leaves no continuing hazard. After the machine is fully held, GRBL’s cycle-start/resume command can resume a feed hold, subject to the controller’s documented state and configuration.
By machine-control definition, a standard GRBL feed hold does not disable the spindle or coolant; it affects motion only. That matters for both routers and lasers: a paused toolpath does not establish that a spinning cutter, coolant flow, or laser output is off. Actual laser behavior requires verification against the exact firmware build, machine configuration, and manufacturer documentation—not an assumption transferred from GRBL’s CNC-oriented documentation.
Soft Reset Abandons Execution
A soft reset—Ctrl-X in documented GRBL v1.1—immediately halts and resets the controller without a power cycle. It is an abort/reinitialization action, not a resume command.
The position consequence depends on what the machine was doing. If reset occurs during motion, GRBL raises an alarm because position may be lost; if it occurs while the machine is not moving, the documented firmware retains position and re-homing is not required. That documented position retention does not certify that the workholding, tool, material, or setup is still trustworthy after a physical incident. A controller can report a plausible coordinate while a clamp has shifted or a workpiece has moved.
A soft reset is therefore a choice to abandon the current job and return the controller to a fresh execution context. It is not a substitute for an emergency stop, and it does not establish energy isolation or verify that every machine subsystem is safe.
Emergency Stop Is a Machine-Level Decision
Immediate danger calls for the exact machine’s documented emergency procedure. That procedure may involve a physical emergency-stop control, power isolation, fire response, or another machine-specific action; the correct choice depends on the machine’s actual design and the hazard. A red sender button, USB disconnect, reset pin, enclosure opening, or software label does not by itself establish what electrical or mechanical action a control performs.
A software reset or disconnect does not by itself prove that every source of machine energy has been removed. The controller’s documented response and the installed machine’s physical response are separate questions. Follow the manufacturer’s emergency instructions and the applicable maintenance-isolation procedure; do not assume that resetting a controller makes energized equipment ready to service.
Do not bypass interlocks, limits, guards, grounding, ventilation, or manufacturer stop rules. Do not reach into a live tool area, service energized components, or perform a live test to see whether a control is safe. If the machine’s emergency behavior is unknown, treat the situation conservatively and follow the manufacturer’s documentation.
After the Hazard, Preserve Context
Once the immediate hazard is controlled, preserve what the controller can still report before changing its state. A fresh status report provides the state, current position, and speed data; alarm messages identify an emergency condition and an alarm code. Record the reported state, positions, alarm code or message, and the last known job context when available.
Then verify the physical setup before choosing recovery. Check that the workpiece, fixture, tool, and surrounding area are secure; confirm that no fire, smoke, electrical fault, or live-tool hazard remains; and follow the machine manufacturer’s recovery procedure. GRBL’s documentation warns that an alarm can mean position is not guaranteed, and that a reset during motion can cause lost steps, making re-homing highly recommended in those documented cases.
Do not restart simply because the sender shows a familiar state. A controller state describes the controller; it does not independently confirm that the physical setup is unchanged or that every hazard has been removed. The decision after an incident is therefore physical first: establish that the machine, workpiece, fixture, and surrounding area are safe, then decide whether the controller’s retained position is trustworthy enough for the manufacturer’s recovery procedure.
What if the sender freezes but the controller is still running?
A frozen sender does not prove the controller has stopped. GRBL real-time commands are handled separately from the normal streamed G-code interface, while queued motions can remain in the controller’s planner buffer; use the machine’s documented emergency procedure if the physical situation requires it, rather than assuming a frozen screen means motion has ended.
Does an error response prove that the controller’s queue is empty?
No. In character-counted streaming, previously sent blocks may remain in the serial receive buffer after a rejected line. An error response is not the same event as a feed hold or reset. Follow the exact sender’s interruption and streaming behavior rather than inferring a cleared queue from one message.
Can stored startup blocks run after a reset?
Yes, subject to the documented startup conditions. Motion in a startup block can create a new hazard during recovery. Standard GRBL does not run startup blocks when initializing in Alarm or immediately after an $X unlock; with homing enabled, it runs them after a successful homing cycle rather than at startup. Review stored blocks and the exact machine’s recovery procedure before reinitializing.
