Mitsubishi SSCNET

it looks like for some reason the card hasn't be configured by the BIOS. though if has a base address assigned... no idea why. try
setpci ... COMMAND=0x6
instead of 0x2 as the card is also a bus master...

nyx not found -
make sure the nyx-pci-dkms_2.4.1_amd64.deb has been installed. if it is - check
sudo dkms status
nyx should be listed as 'installed' for the currently running kernel.
if not, try
sudo dkms install nyx-pci/2.4.1
if successful
sudo modprobe nyx

We are moving forwards! I had to
And then reinstalled the packages from yurtaev.com/nyx24.html using dpkg -i

Now the driver seems to be working:

And servo info gives some more stuff:

Linuxcnc starts with the nyx-ssc3h file! All LED's on the card are looking good!

Note that I do not need to issue the setpci command after a reboot. So maybe I messed something up during the initial install..... sorry about that.

I am getting nothing from the axis, AC->Ab->AC status on the drive and following errors now. I am pretty sure it is because I need to home my servo drive (it has the absolute position batteries installed, so I need to define a home position). I will dig in the servo's manual and keep you guys posted.

Thanks a lot for the help!!!

Cheers,

flawa

drives should get initialized: Ab -> AC -> b##
if they got caught in a loop - it is either transmission errors thru the fiber or, most probably, a latency issue. what does the latency-histogram show?
make sure all power-saving stuff in the BIOS setup is turned off, also
/etc/default/grub:

GRUB_CMDLINE_LINUX_DEFAULT="quiet intel_idle.max_cstate=0 processor.max_cstate=1 isolcpus=1,2,3 idle=poll"

usually helps somewhat...

All BIOS options were already turned off, latency was quite good (<25 us on a 15min run, and did longer runs with similar performance).  




Surprisingly, after modifying /etc/default/grub, the isolcpus does not show in the second run:

 

But overall this latency seems to be within what you recommend, no?
I am still getting "sync glitches" though, and the D6 LED is behaving erratically.

[EDIT] I have linuxcnc running without sync glitches and D6 long on->short off, but still the same status.

I see that you have set the period to 222242 in the ini file. Is there a reason?

Also note that I only have 1 drive connected, is that is an issue? The drive's number is set to "0" (which corresponds to drive number 1 according to the manual) on the rotary switch. DIP switches are all off. I also tried to change the optical fibre to no avail.

I have to go sleep now, I will work on this on Sunday and try to figure out what's happening!

Thank you very much for your help!

cheers,

flawa


 

yes, the 222242ns period is there for a reason. the card uses a digital pll to keep sync with the linuxcnc by varying its 0.22ms cycle a bit according to the phase difference with the linuxcnc's one. MR-J4-B drives are rather strict on the cycle time - they tolerate max -0/+5 bytes of cycle length at 150Mbps. so that timing is about at the middle of that tolerance range

yet another retrofit in progress - traub tnc 65. originally mitsubishi meldas 500 controller with some customizations from traub i guess. mds-b/c1-v/sp servo drives. the oem controller was alive, so i got servo/spindle params from it. the machine has 3 spindles: main, counter and live tooling. main and counter have hi-res secondary encoders for C1,C2 axes plus hydraulic brakes. two carriages were mapped to Z,X and W,U axes and counterspindle to V, but a secondary independent control channel is badly needed for adequate utilization of the machine potential... also spindle-counnterspindle synchronization is in research...
the machine has no limit switches relying on absolute servo encoders. it has Z1,Z2,V overload sensors. it has two Z1-V proximity switches. linuxcnc homing was configured with rigid stops and servo motor torque limiting.