Clarification on Controller State (IDLE / HOLD / RUN) During M6 Tool Change Modes (0–4 and ATC)

[cimygy]

Hello,

I am currently studying the M6 tool change implementation in grblHAL, and I would like to clarify how the controller's reported state behaves during the different tool change modes.

From the documentation, I understand the functional differences between tool change modes 0–4 and the Automatic Tool Changer (ATC) mode. However, what I am specifically trying to confirm is the controller's internal state transition:

Core Questions on Controller State

Controller State: When an M6 command is executed, what is the controller's reported state? (i.e., Does it enter HOLD state, IDLE state, or remain in RUN state during the critical tool change procedure?)

Mode Differences: Does the state differ depending on the configured tool change mode (0–4 and ATC)?

Command Permissions in HOLD: If the state is HOLD, which G-codes / M-codes are allowed or restricted while the machine is paused? (For example: probing commands for tool length measurement, axis moves, spindle commands, coolant commands, etc.)

Reason for the Question (Workflow Integration)

My typical tool change workflow requires the following sequence:

Move the machine to a designated tool parking position.

Change the tool (manual/semi-automatic).

Perform tool length measurement (possibly using a probe - G38.2/G38.3).

Return to the previous machining position and resume cutting.

Since the validity of modal commands (like those needed for probing and movement) depends heavily on whether the machine is in HOLD or IDLE, knowing the exact state during M6 processing is critical for correct macro or host controller integration.

If there is a specific reference in the source code or documentation (e.g., a specific C file or Wiki section) that describes this state transition logic, I would appreciate being pointed to it.

Thank you very much for your time and support.

[terjeio]

Controller State: When an M6 command is executed, what is the controller's reported state? (i.e., Does it enter HOLD state, IDLE state, or remain in RUN state during the critical tool change procedure?)

Mode Differences: Does the state differ depending on the configured tool change mode (0–4 and ATC)?

Yes, whe the in-build tool change routines are running the state is reported as Tool even if/when the internal state changes, IIRC ATC implementations may differ; if macro based the state is reported according to the internal state, if C code based not neccessarily.
Common to all modes is that the 'ok' response to the M6 command should not arrive until the tool change is complete. The exception is is that commands issued by the sender when the Tool state is active will be 'ok'ed as usual.

Since the validity of modal commands (like those needed for probing and movement) depends heavily on whether the machine is in HOLD or IDLE, knowing the exact state during M6 processing is critical for correct macro or host controller integration.

Since the validity of modal commands (like those needed for probing and movement) depends heavily on whether the machine is in HOLD or IDLE, knowing the exact state during M6 processing is critical for correct macro or host controller integration.

By "host controller" do you mean the sender?
FYI macros running on the controller cannot query anything during Hold state since execution of gcode is suspended

If there is a specific reference in the source code or documentation (e.g., a specific C file or Wiki section) that describes this state transition logic, I would appreciate being pointed to it.

The only documentation is for the internal tool change protocol (mode 0 - 3), ATC implementations will likely differ and since they are made by 3rd parties any documentation must be provided by them? For ATC implementations the M6 command is just like any other, either wait for the 'ok' response or until there is space in the input buffer before sending further commands.

[cimygy]

Thank you for your quick response.

I was thinking about using the ATC Controller in manual mode to replace the sander communication and use modal commands to move the tool to the tool change position, change the tool, move it to the probing position, probe, correct the tool length, and then finish the job at the original position.

That's why I was wondering if it's possible to use the G command in manual mode.
I think I still need to learn more.
I'll prepare a test CNC soon and try it out while actually operating it.

I'm building a custom board based on the RP2350B.

I apologize for asking an awkward question without studying it thoroughly.
(Please understand my poor English.)

[skasti]

I have automatic tool change working on my cnc, but I am running a somewhat modified version of grblHAL to do so. But I think the core concepts should be possible without those mods; https://github.com/skasti/grblhal-macros

I am not using the probing routines, as I like to probe a few times and average out the reading, but this is probably not necessary. Just haven't gotten around to changing it 🙈

You can see roughly how it works here: https://youtu.be/AR_KXNpowi4?si=FsbZrAgfwEIhAbGr&t=97