#linuxcnc-devel Logs

Mar 13 2019

12:12 PM -!- #linuxcnc-devel mode set to +v by ChanServ
12:16 PM -!- #linuxcnc-devel mode set to +v by ChanServ
12:47 PM -!- #linuxcnc-devel mode set to +v by ChanServ
12:51 PM -!- #linuxcnc-devel mode set to +v by ChanServ
01:01 PM -!- #linuxcnc-devel mode set to +v by ChanServ
01:12 PM -!- #linuxcnc-devel mode set to +v by ChanServ
01:15 PM -!- #linuxcnc-devel mode set to +v by ChanServ
01:28 PM -!- #linuxcnc-devel mode set to +v by ChanServ
01:33 PM -!- #linuxcnc-devel mode set to +v by ChanServ
01:34 PM -!- #linuxcnc-devel mode set to +v by ChanServ
01:42 PM -!- #linuxcnc-devel mode set to +v by ChanServ
01:46 PM -!- #linuxcnc-devel mode set to +v by ChanServ
09:49 PM Roguish: pcw_home: having serious problems with my setup. the watch_dog is biting me all over. no clue why. had a 5i22, then tried my 5i24. same thing. is this OS related?
09:50 PM Roguish: have never had this problem before.
09:56 PM pcw_home: WD bite is usually latency related
09:57 PM pcw_home: could be hal file related also (wrong WD timeout setting vs servo thread)
10:11 PM Roguish: there's a WD timeout setting?
10:12 PM pcw_home: yes, in the hal file
10:13 PM Roguish: OK, found it. not sure how to set it. it's in the hostmod2 driver section. timeout_ns
10:25 PM Roguish: pcw_home: so like this? loadrt hostmod2 timeout_ns=10000000 that's 2x the default of 5000000
10:27 PM Roguish: pcw_home: I have SERVO_PERIOD=1000000
10:30 PM pcw_home: Yes thats OK
10:30 PM pcw_home: what latency do you get on your host?
10:31 PM Roguish: not sure right now. haven't checked for a while. it's an asrock q1900m board
10:31 PM Roguish: most everything extra turned off in the bios
10:32 PM Roguish: running master pre-empt
10:34 PM Roguish: i've also been getting spurious other errors like limit switches tripped when nothing is near them. and lots of ferrors, even though i've retuned for the 7i40. a fairly soft tuning. not too aggressive.
10:36 PM pcw_home: maybe a noise issue, does the problem go away if motor power is off?
10:36 PM Roguish: no idea.
10:37 PM Roguish: been running from a launcher. i'll run from a command prompt. can see more errors.
10:37 PM pcw_home: If you can try that, it would determine the source of the problem
10:37 PM Roguish: are the 7i40's noisy?
10:39 PM pcw_home: the probably have faster output switching than the LM18200
10:39 PM Roguish: i have pwm_frequency = 40000
10:39 PM pcw_home: and much better ground connections to the host
10:40 PM pcw_home: so if you have a ground loop of some kind it will cause more issues
10:40 PM pcw_home: (better ground connection because all 24 flat cable grounds are used)
10:40 PM Roguish: motor power lines are not shielded.
10:42 PM pcw_home: you might need to shield them + use a common mode choke (giant ferrite bead) on the motor
10:42 PM pcw_home: and perhaps power supply lines
10:42 PM Roguish: what do you think the servo period and pwm_freq ought to be
10:42 PM pcw_home: servo period maybe 2 KHz PWM 20 to 40 KHz
10:44 PM pcw_home: you can get a ground loop if your motor power ground is tied to ground somewhere in the power supply (like a PC power supply)
10:44 PM Roguish: 2 KHz == 500,000 ns
10:44 PM pcw_home: yes
10:45 PM Roguish: I disconnected the PS grounds. I had a wire from the neg term to a common terminal for all the power supplies. I disconnected those.
10:47 PM Roguish: Peter you're juggling conversations......
10:47 PM pcw_home: Does sound like a ground loop (if the problem goes away with motor power off its a noise issue)
10:47 PM pcw_home: I do have some common mode chokes about the size of a walnut
10:49 PM Roguish: https://www.amazon.com/Filter-ferrite-Choke-Digital-Cable/dp/B07CVHXS6P/ref=sr_1_16?keywords=electronic+choke&qid=1552535116&s=gateway&sr=8-16
10:52 PM pcw_home: Yeah those are worth a try (mine are not split so you need the thread the wire through)
10:54 PM Roguish: thanks. thats' ok. in the middle of machining for an important client. all the problems have cost me way too many hours. i'll get some chokes or find some shielded power cables.
11:33 PM Tom_itx is now known as Tom_L