Advanced Search

halscope and record statusword and controlword.

well we sorta already did that. there is no setting that doesnt either flicker the drives or slam the axis. like it does't respect search or latch vel at all.

that has to be a clue. something ELSE is making it home. maybe it IS doing internal drive homing?

"Divide and conquer".
Comment out everything except one single axis and try until you find it.

Hakan wrote:

Yes it does have ethercat drives. And modbus nonsense as well.

 

bah!
haha. clearly i've botched something somewhere. i just cant see it.

Hakan wrote:

I don't think it is a good idea to combine homing on current position with HOME_OFFSET.
That means is set's zero exactly where it is and immediately after moves to HOME_OFFSET.

ive removed that now, now, no change. also that offset is + direction and it is slamming the thing full speed into the table.

Yes it does have ethercat drives. And modbus nonsense as well.
 

does your lathe use ethercat drives?

this is what i had before with the pulse drives. everything was fine. And i could have sworn in my first ethercat program *before* all this modbus nonsense it was fine as well.

I don't think it is a good idea to combine homing on current position with HOME_OFFSET.
That means is set's zero exactly where it is and immediately after moves to HOME_OFFSET.

That's odd. Here is what I have in my lathe where it homes on current position.

If I set like the table in your posts, it still moves and slams the axis into the end stops.
If I set homing sequence to -1 it flickers the drive.

I do not want any internal homing on the drive. I want no motion whatsoever.

What "normally" works is not here.

The docs at linuxcnc.org/docs/html/config/ini-homing.html says it
 

The cia402 component says

 For using the servo drives internal homing procedure configure your
joint homing to Home on Index Pulse only and connect the components
home Input to the index-enable Pin:

I don't see you have connected that pin.

To debug, use halscope and record controlword and statusword to and from the drive.

Later I will summarize and correct the first message of the topic.
First, it turned out that it was pointless to consider the topic without taking into account the electrical component and connection diagram.
Then it turned out that neither the Snnnnn parameter nor the M68 Qnnn control the laser power as it should (the hypothesis from the message www.forum.linuxcnc.org/10-advanced-confi...-laser?start=0#92752 is not correct in part 50%).

Colleagues, here we are trying to use the Snnnnn or M68 E0 Q 0.nn commands to turn the laser on and off.

What is known about the time of their execution and delays? What is the point of using M68?

Yeah the thread is long and full of failed attempts and iterations.
That's my development cycle, sadly.
The last post summarizes it. And this post show the performance forum.linuxcnc.org/show-your-stuff/56269...m-60?start=20#336805 It keeps performing like that even on water and rusty pieces.

Because the torch is the way it is, the ground leads in the torch, used during the arc start, needs to be cut and a small voltage applied which one measure to detect contact with the material. The rest is implementation and packaging in a nice format.
If you are willing to do surgery on the long cable to the torch you can cut it open and access the ground leads in there.
Instead, I made a short extension cable where all modifications were done.

I also used two separate voltage readers but I don't think that is necessary.
One uses the built-in voltage port, one measures directly between material ground and torch tip.
I will cut some parts later today so I can compare the voltage output of the two.
Of course, Mesa's THCAD cards can be used, the full voltage version.

Ok, so I sorta understand what is going on. searchvel set to 0 is supposed to work. It was in my file before, HOWEVER it causes my ethercat drives to reset repeatedly, apparently cause the home routine toggles enable. This is basically what was happing with the M3 pins, some manner of conflict.

So how do we get around this?