Advanced Search

Good job tracking that one down!

Did you previously increase the libusb log level somehow? If so, then maybe that's why it segfaulted for you, but not for me. regardless, looks like a libusb bug to me.

Perhaps, I found out what is happening to the drives. I read the PDO mapping part of the manual again and it seems that the PDO mapping objects - 1C12h(RxPDO) and 1C13h(TxPDO) are set to some default values. The PDO mapping cannot be saved in the EEPROM of the drive and therefore the master needs to set the desired mapping in the PREOP state of the drive every time when the drives are powered on.

The output in terminal that show "Gap" in PDO entry is because there is no object mapped to that free slot - PDO mapping is set to fixed.  Here  is the PDO mapping part of the manual. It can be seen that 1C12h should be set to 1600h and 1C13h should be set to 1A00h to activate the variable PDO mapping.

Attached is the PDO mapping objects description from the manual and it shows that they expect data type of ARR(perhaps array). Apparently, LinuxCNC EtherCAT master should define a variable PDO mapping and what objects to include in the mapping.

In my opinion, the PDO mapping should be defined in the .hal file or create a script that executes once when the EtherCAT master detects slaves, similar to the first scan bit of a PLC.

The compensation needs to allow for tilt of the saddle as it moves along the bed. There is typically more wear on the front way than the rear, and this is (I believe) what causes the machined diameter to vary.

You could measure a parallel test bar with a dial indicator at various points along the bed, and use the "lincurve" component to calculate the required offset (and then add that on between motor-pos-cmd and and the input to the PID)

Any that you do find will be second-hand. And old.

I would follow PCW's advice and change to a supported FPGA card. The 6i24 is in-stock at Mesa, but would need a PCIe motherboard rather than PCI.
It doesn't look like Mesa have stock of any full-size PCI cards at the moment.
(The 5i24 is the PCI version of the 6i24 but shows as out of stock)

The 6i24 will work with the same daughter cards as the 5i20.

It would be necessary to change "5i20" to "6i24" throughout the HAL file too.

No, there hasn't been a 2.10 release, and that was the development version when my email was sent.

This is the state of the .comp in the 2.9.8 release tag.
github.com/LinuxCNC/linuxcnc/blob/v2.9.8...nts/toggle2nist.comp

(no debounce pins)

So far so good. Next big thing is the wireless jog wheel. 

Okay I changed it and there was no improvement. 

Thank you in advance =) 

Looking more carefully at your hal file:

setp joint.0.jog-vel-mode 1

Means that the jog will continue as long as jog count <> 0

(this is normally intended for input devices like joysticks that return to 0 when not pressed)

for position mode jogging with a MPG, you should have:

setp joint.0.jog-vel-mode 0




 

I found it.

Ok, I didn't find the reason but I found a way around it.
It always crashed for me with "libusb: debug [libusb_unref_device] destroy device 3.109"
I ran it in the debugger and saw it stopped with a SIGSEGV.
After spending too much time figuring out how to turn on core dumps, I got a core dump.
Here is the stop point.  

 

The one wonders what ctx_log_handler is
I commented out this line in libusb_unref_device()
 
and then the xhc actually restarted.
Here is the log of that. 

were you able to get it set up and working?

how do i change the pref_edit.py from this one out of the forum with mine becaus when i am just copy paste than occurs an error . regards tom 

The links : NativeCAM-Python3-Debian-12-Installationguide2026-02-20.pdf
: forum.linuxcnc.org/media/kunena/attachments/40714/pref_edit.py

Amazing support Peter, but I really don't understand the expectation of the original post. 

Your Mesa cards are very resonably priced. You have obviously put years of work into designing and updating these cards, and provide outstanding support to us here in the forum.

Someone reverse engineers your work, floods the market with knock-offs, undercuts your livelyhood, and then expects support?

You are a better person than me Peter. In your place, I would be tweaking Mesaflash to brick any knock off which gets connected to it
Cheers,
Mark

something like this:

 

and it will generate the code for all FPGA's and MCU's

the interlinks are RS422 over RJ45 cables (like smart-serial)

Ok, got it, good to know - and I do not know that much about Linuxcnc and lastly, Pnnconf / MESA stuff. All new.

But, all limits working now. I just need to check the probe setup - think that I can do after the limit switch and HalShow stuff initiation.

But, my ESTOP is still beyond me. Saw threads where JT and I believe a few others talked about inserting lines that linked the ESTOP pin to other pins or feedback - tries several but beyond me once again. 

ESTOP - loaded all pins and all ESTOP, Abort, Charge pump references into HalShow, enabled [it's a NC switch] my external ESTOP button and nothing in Halshow changed color. Yes, my external button is wired into the mx3660, and yes it shuts down the motors on push.

I tried to move the ESTOP to different pins that were open and then Halshow - no change. The probe pins work in Halshow so I know that is at least on the right pin and gives feedback - I can figure that out.

So any suggestions on ESTOP? Else I will live with it the way it is. Still have a lot of changes to do with travel and positioning in the ini for axes but that's easy stuff, just takes me a little longer one step at a time.