Homing procedure for axis with multiple slaves

Clive wrote:

You are wrong about the keeping active part 100%. If there is an isolated home switch on the axis and you travel left to right when homing everything works fine while you start homing on the left hand side of the switch. If you start homing positioned on the right side of the switch, Linuxcnc will still move left to right and crash into the limits. If the home switch remains active on the right hand side, Linuxcnc knows which way to go to get back to the home switch triggering edge before the homing sequence commences.. There are a number of posts about this on the forum where people have got into grief with this.

We may be both correct with this. I don't use travelling sensors mine are fixed at each end of travel so the homing routine always knows which direction to go.

I have a feeling you do.

Yes, your machine still complies with my rule so I win

I made the decision to implement every sensor available on all axes and only use a shared home/limit switch on the Z axis due to lack of room. I'm certain you could not use a HOME_SEARCH_VEL of 6000 mm/min with shared home and limits. Something would break. I wanted to make it faster but the tradeoff with coasting to a stop came into play. I probably coast about 50mm after the home edge is first triggered. It took me a lot of playing to come up with this as when I built my machine 2.8 had just been released and hardly anyone was using it for joint axes at the time so I was on my own.

Ok, fixed all the side issues.

I've modified the ini file as follows:
[KINS] joints=4, kinematics = zzzz
[AXIS_Z]
[JOINT_0]
[JOINT_1]
[JOINT_2]
[JOINT_3]

and assigned the same homing procedure to all the joints.

Now, for the HAL file:
- each axis section starts with a list of setp (pid P, I, D, bias, FF0, ...): I'm looking at a sample configuration done with the pncconf wizard, where the only slave axis has pid.z2.PGain, .... -> how should it look like in my case?

Posting your hal and ini files would help

Think I solved it:
-> the naming should match with the function call "loadrt pid names=..." at the beginning of the file.

Now fighting with the HALUI signals ("halui pin ... does not exist")
I'll keep you updated

HAL


INI

piter wrote:

Think I solved it:
-> the naming should match with the function call "loadrt pid names=..." at the beginning of the file.

Now fighting with the HALUI signals ("halui pin ... does not exist")
I'll keep you updated

Yes you will have to duplicate the Z in the section of the hal file stuff and changing where necessary Z 0,1,2,3 etc

piter wrote:

Now fighting with the HALUI signals ("halui pin ... does not exist")
I'll keep you updated

Easiest way to resolve this is to comment out the bad pin so Linuxcnc will load and then you can review in halshow to find the correct pin name.

Done! As rodw suggested, I removed all the halui related pins in the hal file, and everything loaded correctly.
I still get an error related to /usr/share/axis/images/axis.ngc, (X axis command, which does not exist here), but just acknowledge it.

As always, thank you all!

Pietro