compiling grblHAL RP2040 from source on Linux

[chris2560]

I'm struggling to compile grblHAL from source on Linux.

The web builder will not allow me to select the generic 4 axis pinout, which I need because I'm building the circuit board and can't/unwilling to use extra io expanders when there's plenty of pins on the Pico for my simple 4 axis setup. A PicoCNC board is of no use because I'm controlling a Seig X1, running DM556 stepper drivers with 3NM steppers. The generic4axis pinout looks good, very similar to the pinout I've been using on a breadboard for the last year, but I'm now adding the 4th axis and need the newer firmware features like hard limits enabled on XYZ but disabled on A.

There is no documentation / I cannot find the documentation for how to compile grblHAL RP2040 from source using cmake and make commands or using Arduino IDE. I can easily obtain the source code with the git clone command, but cmake wants an SDK AND a tool path chain. Trying to give it the filepaths to the sdk and tool path chains within my RP2040 module install in Arduino IDE has worked before, but won't work now. I keep getting __retarget_lock_acquire_recursive errors.

[terjeio]

VSCode is the default build environment.

I am using cmake + make for the web builder.
The commands I use are:

cd <project folder>
cmake .
make

The build process locates the SDK via the PICO_SDK_PATH environment variable which has to be set first.
The toolchain should already be on PATH?
I installed it with this command:
apt-get -y install cmake gcc-arm-none-eabi libnewlib-arm-none-eabi

For the time beeing I am not going to spend time on adding support for the Arduino IDE.

[chris2560]

Thank you,

I have had some chance to do some more testing, on 2 different picos (though not a pico W because I don't own one)

The 3 axis map seems to compile and run fine, and all direction and limit pins are functioning correctly, I haven't yet tried running the machine from it yet, my machine doesn't yet have automatic control of the spindle, the 3 spindle control pins do seem to function correctly when checked with a voltmeter. I've checked the step, direction and limit pins, and wired my A axis motor into x, y and z and checked they run in forwards and reverse correctly.

The step and limit pins all function correctly on the generic 4 axis map.
But the generic 4 axis map's direction pins seem problematic, on the current firmware version (20231005) and older firmware version from a year ago. The Y direction controls pin D6 and the z direction outputs to pin D7, the x seems to not have a direction pin and D5 works as the A axis step pin, the source code clearly sets the direction pins as X=D6, Y=D7, Z=D8, A=Z+1=D9.

When the 3 axis pinout's direction pins are set to x=D7, y=D5, z=D6 it doesn't work at all, nor when the 4 axis map's direction pins are set to x=D7, y=D8, z=D9.

Is there some set of rules which define what pins can be used for what and in what order? (similar to the strict rules on GRBL when I was setting up GRBL on an Atmega1284 2 years ago)?

[terjeio]

There are strict rules depending on how pins are written to.
E.g. #define DIRECTION_OUTMODE GPIO_SHIFT5 uses a shift and writes all direction pins in one go. BTW this should be GPIO_SHIFT6 if I am not mistaken, it would explain the mismatch described above.

Other options for writing to pins or a set of pins are defined here, with variyng degree of freedom to chose pin numbers.

Note that step pins are written by PIO code in the 4 axis map - which requires the pins to be consecutive.

[chris2560]

That's solved it.
Changing line 40 of generic_4axis_map.h
from

#define DIRECTION_OUTMODE GPIO_SHIFT5

to

#define DIRECTION_OUTMODE GPIO_SHIFT6

and once compiled, the step, direction and limit pins all function correctly.

I'd seen the shifts defined in drive.h but not understood what they do, but basically the

#define X/Y/Z_DIRECTION_PIN 6/7/8

lines have no effect?

Please can line 40 of generic_4axis_map.h be changed on the repo accordingly?

Many thanks

[terjeio]

The pin definitions are only for reference and for generationg the $pins command output when a shift is used.
The repo and the Web Builder has now been updated.

[chris2560]

Many thanks,
The web builder now compiles the generic 4th axis map really well - (The version I'd compiled which used native USB seemed to weirdly disconnect from serial every time I started all 3 XYZ axis moving at once or maybe when I spun the spindle up to 5000rpm, but this was fixed with re-compiling for UART and using my FTDI UART adaptor - I suspect it may have been my system overloading some IO pins or a dodgy wire disconnecting from the laptop when the workbench started vibrating as the spindle spun up)

[chris2560]

Should my 4th axis home just like the XYZ axis?
I've got an infrared beam break for a limit switch and 2 drive transistors, connecting via 50ohms and 1microFarad capacitance to the IO pin, just like the other axis.
When the A axis motor and limit switch is plugged into the Z axis pins, it homes just fine, however when plugged into the A axis pins and $HA is called, it simply zero's the machine coordinate A immediately and doesn't seem to move the A axis either direction as it does.

[MSG:]

is printed to the console when $HA is sent.

I can post the parameters here if necessary.

[terjeio]

Have you enabled A-axis homing? $44-$47 are the settings for the different passes, and one of the passes should have the A-Axis bit set. E.g. $47=8 for homing A separately.

[chris2560]

You've solved my trouble yet again, it now tries to home the A axis when I send $Ha. (just a shame my infrared limit switch is not working reliably now, but that's my problem, not any problem with the software)
Many Thanks