Advanced Search

OK, in that case you will have to monitor both commanded position and reported actual position to measure the time it takes, then try some tuning if the result is something reasonable.
Perfect would be no delay at all, but that never is the case.
In any case, you should absolutely remove the fedback from the drives, tune the drives to work properly with feedback from stepgens, only after you get that working, then switch to drive feedback and tune PID and FF1 values.

Hi Tommylight

I am only using EtherCAT drives.
No Mesa or parallel port.

yes the encoders are reporting back, but to the drive, and the drive is sending the following info back to LinuxCNC
"actual-position" and "actual-velocity"

LinuxCNC is then sending out to the drive
"target-position" and "target-velocity".

fixed the logo problem. Can you show your calculation?

If you set the 'isOnSensitive' property, then the button will be disabled until the machine is on.

The mist button uses the mist_action property.

Thanks for the tip. Unfortunately I got the same result, and with the values looking good.

Finn

Are you sure you are running a real time kernel?

What does:

uname -a

report?
 

How much "problem" is there? What type of machine? Even in CNC boring heads with a single point cutter are still "a thing".

Maybe your axis are not exactly perpendicular?

Can you measure it simehow?
I had huge backlash on my 2nd hand machine, the rubbers of the couplers between ball screw and coupler were "disappeared".
Another thing I found is a damaged bearing that keeps the ball screw in place. Then the complete screw can move a little.

Happy trouble shooting

I installed debian 12 and then linuxcnc using the sudo apt install method on a computer that used to run linuxcnc on debian 10.

The jitter numbers are huge, around 50,000,000 ns.  On debian 10 they were more like 50,000 ns.

What happened?

Found this just now, the datacenter featured in Batman vs Superman is made by the EMC2 company, they are the reason we had to change the name of this project to LinuxCNC.
At least they did not sue us, so thank you...maybe?!?!?!

Might have better luck with this in the Debian forums, it is Linux related.

Yet another Flexgui Op. This time we tackled single point thread milling.

What settings are you using for the G251X? I have been trying to tune them with no success. The motors are very loud and produce a lot of vibration, especially at low speeds. Using generic Chinese drives with the same hal (minor tweak to the stepper scale) produces significantly less noise and vibration.

First make sure the scaling is correct for all axis, then check for backlash or loose screws or mounts.
Also check for mount deflection while machine is working.

Reading hal pins in gcode is tricky as you need to sync the interpreter to update the values:
linuxcnc.org/docs/stable/html/gcode/over...gcode:ini-hal-params

Try this:

Hi Collins,

It's far from linuxcnc compatible.
The hal works different and the rtapi works different.

The code is entirely different. The realtime approach is the same.