SSI absolute encoders and analog servos

Adding "pcie_aspm=off" to grub may take care of this. I will try it tomorrow. I will try to see what I can get out of halscope. Any detailed info of what I should check with scope?

Tero

Large latencies or problems with the time of day clock can cause loss of sync (that why I worry about the NMI)
I'll setup a copy of your config here and run it couple of days to see if I see any issues since I have not tested the SSI stuff
more recently than around 2.6

Ok, I will run Memory tests and if not I will replace PC with another one to see if it is hardware based.

Tero

Hmmm after modifying config to :

loadrt hm2_pci config="ssi_chan_1=yencoder%32e

#SSI stuff
setp hm2_5i25.0.ssi.01.frequency-khz 400
setp hm2_5i25.0.ssi.01.timer-number 2
setp hm2_5i25.0.dpll.02.timer-us -150
setp hm2_5i25.0.ssi.01.yencoder.scale 10000

This warning is gone... I moved this to ssi_channel 1 and also used timer no 2.

Tero

Tero

Its extremely unlikely that the SSI warning is related to SSI or timer channels
Its probably more likely a random hardware or PC latency issue of some kind

(I have been running ssi channel 0 and timer 1 on my test machine for about a day now -- no issues)

Might it be an issue with 5i25 card itself? I am asking because it just hit me that this is the card you sent me free of charge(test card of some sort) Because you were temporary out of stock. I forgot to replace it with brand new one....


Tero

Its unlikely its related to the 5I25 (a bad card is much more likely to have a bad I/O pin or not be recognized, this is more likely a system issue)

Have you tried looking at the DPLL phase error? Sometimes the PCI clock is far enough off that the DPLL base frequency
needs to be tweaked

Ok,

Here is phase error:

Seems weird to me... Same pattern than I have with axis. f-error

forum.linuxcnc.org/38-general-linuxcnc-q...id-tuning-once-again

Tero

Hmm... No lock at all so cannot possible work, can you post your hal and ini files here.

Also what is the scale of the horizontal in that halscope plot? it looks like the
DPLL base frequency has been set completely wrong somehow

I will in few moments. Could this be related:

forum.linuxcnc.org/49-basic-configuratio...wsum-switch-bouncing

Tero