Advanced Search

That can't be right? Maybe it works, but you normally don't want nor need an IP-address on the ethercat interface.
Here is my config for I226-V in /etc/network/interfaces.
The i226 uses the Intel igc driver, the energy-efficient mode is off by default, so the igb.EEE=0 is not needed and also is for the igb driver, not the igc.
You can rename the nic with a .link file, google it the best tip there.

 

rodw wrote:

iface eno1 inet dhcp
    hardware-irq-coalesce-rx-usecs 0

Ehm.... it worked! Thanks! The networkinterface for EtherCAT is no longer in the list of networks and now the internet connection defaults to the proper interface, and EtherCAT works like a charm now...but....why did that work? :-D I don't understand since nothing in that configuration seems indicate "oh, by the way, this will hide the interface from UI and prevent InternetConnection from trying to use it"...i've looked up the hardware-irq-coalesce-rx-usecs parameter and that only seems to be some sort of tuning, not something that should do what it just did....

Sorry, i'm just trying to understand it... the worst part is when something works, and i don't understand why it works

PS: I also disabled the power-stuff in GRUB as you suggested. Thanks!

/Thomas

minimal fpga project, just edit the pins file to match your pinout:
github.com/multigcs/FPGA-blinky/tree/main/Colorlight5A-75E

i can't remember but i think it was already for the v8.2 board

if it works, you can take a look here:
github.com/multigcs/riocore/tree/dev/rio...olorlight5A-75B-v8.0

That was it, thought I must have missed something.
Thanks
Creating the spindle-at-speed for a stopped spindle seems to have fixed the issue I was having. I added a delay because I prefer to have the spindle actually stopped (or close to) before the movement happens.
I don't see a way to make the delay time proportional to the starting speed of the spindle, so for now I've set it to five seconds.

not sure but you could try adding this to /etc/network/interfacesauto eno1
iface eno1 inet dhcp
    hardware-irq-coalesce-rx-usecs 0
replace eno1 with the NIC name reported by ip a for ethercat
Restart the network

Also add this to your boot parameters using grub-customizer to disable energy efficient internet igb.EEE=0

So far it hasn’t complained yet.
Old HDD drives don’t like having their power cut off like that, but modern SSDs don’t really care.

besriworld wrote:

I don't know if it's correct but I directly turn off the main power switch and everything goes off.

hehe! That is one way to do it! But I don't think Linux likes getting the power cut like that? Or is that assumption wrong?

/Thomas

Hi,

I use a "Odroid H4" as master and it has two network interfaces. Perfect for LinuxCNC with EtherCAT and cabled networking. However, I'm having the issue that both network interfaces shows up as if they both can connect to the internet: 


Two things annoy me:
1. Is there a way to rename these devices? The naming is kinda useless as it is and it bothers my OCD, preferably i'd call one Internet and the other EtherCAT or something like that, but i've seem to have tried everything google suggests me to rename them, without any luck...apparently not the correct one yet, ideas?
2. The top network card is the one connected to the internet...but every time the machine starts, it attempts to connect to "Internet connection" via the bottom one. These attempts causes the EtherCAT bus to reset every minute or so...to solve this i have to press "Disconnect" at the lower card and press "Internet connection" at the upper card...at every start of the system Can i somehow mark the EtherCAT card as "hands off!" for internet-stuff? And the internet is automagically connected via the upper one? First i thought "easy fix", i'll just swap cables and use it the other way around....BUT lo and behold, the problem was the exact same, but now the interfaces was just switched around...ideas?

I know this isn't "LinuxCNC and ethercat specifics", but i assume others here have had the same issues, so i hope it is ok...

/Thomas

I don't know if it's correct but I directly turn off the main power switch and everything goes off.

I've been wanting this to, so thanks for the input!....but.... how do you guys handle the shutting down process? Currently i'm quiting LinuxCNC with ctrl+q, going to Application-->Logoff in Debian and clicks Shutdown....waits untill the screen turns off and then flicking the main-switch on the machine.... is there a better way?

Hi,

I'm trying to setup the  Vers WLR v9 probe  and i got it working fine together with the Vers toolsetter. However, there is a pin out for errors. This triggers if there is issues with the communication for the probe(e.g. if it runs out of battery, is not turned on, communication fails and so), but how to configure this in HAL to halt the probe-cycle? First i thought about making some logic to check if motion-type is "probing" and combine that with the error pin. But the issue is the famous "one pin for all probing", so i don't think i "know" if i'm using the 3D probe and should monitor the error pin or i'm using the toolsetter and the 3D probe is intentional not active(causing the error-pin to trigger!). I've seen several posts about using M-codes to determine which probing cycle to run, but i've also seen people warn about this approach since it requires you to remember to use them each time....

Am i missing something obvious here or have others found a clever solution to this?

/Thomas

I believe that I have figured out my problem, however I don't quite know how to fix it. I'm using subroutines to perform different types of feature probing operations. An example program would look something like this:

preamble
o<probe_z> call [...]
o<probe_xy_boss> call []
end

When I load one of these gcode programs, I immediately get "unknown word starting with c" errors from the 3D plotter. It seems like the plotter isn't recognizing the syntax for named subroutines. If you ignore the error, then the first subroutine will run without issue. After the first probe is completed, the error will appear again for the second o code and terminate the program. I attempted to fix this by wrapping each call in a conditional like this:

o100 if [#<_task> EQ 1]
o<probe_z> call [...]
o100 endif
o101 if [#<_task> EQ 1]
o<probe_xy_boss> call []
o101 endif
end

This stopped any errors from appearing prior to cycle start. The first subroutine will execute without issue, and then the error will get triggered again immediately after the first probe. Limiting the file to a single subroutine solves the issue. The #<_task> wrapping method doesn't seem to work more than once in a file in this situation. I suspect that this has something to do with the fact that the first subroutine contains a probe operation which limits the lookahead of the controller's planner. Perhaps the backplotter is getting re-triggered after the first probe operation, but the machine is still in automatic mode since it's mid-file? Any ideas would be appreciated. I'm using probe-basic.

I am updating a grinding machine which only has the y and z axes controlled by a motor, the x axis is hydraulically driven with limit sensors, I already have the machine working with an old 5i23, I configured the x axis to not have errors in pncconf by configuring a linear encoder mounted on the x axis with the intention of being able to visualize them but when running a program it shows me a tracking error obviously given that there is no x movement command and the encoder starts to count, is it possible to disable the hili of the x axis only keeping the encoder count to be able to visualize the x axis on the screen? 
 

Danke tommy,

werde es nachher gleich probieren.

karlhe

Habe ich brobiert, leider nur freecad 0.2, nich 1.0

 

How did you create the config?
If using the wizard, there is "invert" on the right side of every pin, so check that for direction of the axis you need.
You can also edit the .ini file and add " - " (minus sign) in front of the SCALE value for the joints/axis you need to change direction.