Linuxcnc silently stops responding

Dear all,

Linuxcnc randomly & silently stops responding, all outputs getting low (which indeed is good!), the E-Stop icon being activated (and cannot be deactivated), but without any (noticeable) warning or information window, leaving me no choice but to close linuxcnc and restart from scratch. And as you can guess, this always happens when almost at the end of a cutting job ....

My setup:

Linux 3.4-9-rtai-686-pae SMP PREEMPT

Debian 3.4.55-4

LinuxCNC 2.7.7-52 (NOT RIP-environment)

Librairie HostMot2 0.8.21

MESA 7i43-U-4 board (+7i42 + 7i37)

The debug level being set to 0x000002da, I got the following feedback from the command line after linuxcnc just got frozen:
Anyone an idea how the statement "Motion id 0..." should be understood ?

Another suspect thing is the "Status Window" from axis after freeze, which indicates (among others):andas in the ini file, accelerations as set (for each axes) as follows:

Could this be a possible indice on where to search from ?

I also suspected MESA breadboard to be exposed to excessive EMF originated from the spindle frequency converter, but I assume in this later case, the 7i43 watchdog would have spoken, which is not the case.

Anyone to advise me on the direction I should investigate, please ? I'm lost so far..... Thanks !

If the outputs (7I37 outputs especially) get disabled when this happens
it does suggest that communication to the 7I43 has been lost and the
watchdog has bitten.

(if communication is lost, you might get no warning that the watchdog had bitten)

In case this is a hardware issue, you could try re-routing the EPP cable between the PC and the 7I43
or adding line filter to the VFD (or doing a test run with the VFD turned off to see if its related )

Thank you Peter for this valuable information - I shall try another cut tomorrow and check accordingly (GMT+1 here), provided it fails again (I shall cut gold to help it happen!). In case it happens, I assume the INIT or DONE LEDs on the 7i43 shall toggle accordingly ?

One more question btw: is there any way of cross-checking the communication on the PC side by, say, reading a counter at the 7i43 in a python routine ? FPGA is currently loaded with SVST2_4_7I43B.BIT. Can you please suggest something on the subject, I haven't found that topic be addressed in the forum (but I may have missed something....)

Thanks for your advises !

Have you looked at the kernel log (with dmesg) for error messages?

OK Peter, I found out the io_error parameter toggled after the problem again arose (why did I neglected that, there is no word to my stupidness!). After checking all EPP connections again, found one loose internal connection that supposedly may have caused the comm error. Re-wired the EPP port, changed to a double shielded EPP cable, and up to now, everything works great again !

Thanks you so much Peter !