#linuxcnc Logs

Apr 04 2023

01:03 AM Deejay: moin
01:30 AM -!- #linuxcnc mode set to +v by ChanServ
01:54 AM travis_farmer[m]: Morning ☕️
02:38 AM -!- #linuxcnc mode set to +v by ChanServ
04:45 AM Tom_L: morning
04:47 AM JT-Cave: morning
06:28 AM JT-Cave: the c5 has a P0492 and it's only on the late 2003 and 2004 c5's and I can't figure out if it's bank 1 (drivers side) or bank 2 (passengers side)
06:30 AM JT-Cave: crap I found it DTC P0491 for bank 1 or DTC P0492 for bank 2
06:30 AM JT-Cave: the bank 2 check valve is between the intake manifold and the firewall
06:55 AM JT-Cave: https://www.motortrend.com/how-to/1602-how-to-fix-secondary-air-injection-code-on-a-c5-corvette/
06:55 AM JT-Cave: that looks easier to do
07:02 AM * spray[m] uploaded a video: (4146KiB) < https://libera.ems.host/_matrix/media/v3/download/jauriarts.org/WLbsjVFIofQKxQxKdpEwohUh/trim.A81ACA8E-F0D7-416A-B069-0480BF8DD093.mov >
07:02 AM spray[m]: Hello, can anybody point out, why the jogging acts strange? When it comes close to desired position it gradually slows down and mostly does not even reach precise position..
07:17 AM skunkworks[m]: tuning problem? Do you have a PID loop in there somewhere? Do you have your ini/hal files posted somewhere?
07:18 AM skunkworks[m]: go up to machine - halmeter - select the joint0 folloing error. see what it does when you jog it.
08:17 AM CaptHindsight[m]: I think he is using Remora
08:19 AM CaptHindsight[m]: he was a couple of days ago
08:40 AM spray[m]: Yes remora ☝️👍. Good memory!!
08:41 AM spray[m]: No servos with feedback. Just JMC integrated servo, which is basically the same as simple stepper motor
08:41 AM spray[m]: I can share ini and hal laters
09:20 AM JT[m]: Just fingered out what the EV button is for on the Insight
09:21 AM travis_farmer[m]: extreme velocity mode? ;-)
09:21 AM JT[m]: Electric Vehicle mode
09:22 AM travis_farmer[m]: i know, i was making a joke ;-)
09:41 AM roycroft: the electric vehicle mode is the extreme velocity mode
09:48 AM skunkworks[m]: spray linuxcnc uses feedback internally - so even using steppers there is some form of it.
09:58 AM spray[m]: Right
09:58 AM spray[m]: How does it cope with motors with no feedback?
09:59 AM skunkworks[m]: the loop is closed internally between the stepgen and motion.
10:00 AM skunkworks[m]: spray (@_discord_397707718582075420:jauriarts.org) https://www.youtube.com/watch?v=f7dI0JYC5q4
10:17 AM spray[m]: So It looks to me, that I need PID tuning to be done somehow to optimize the software internal feedback
10:24 AM skunkworks[m]: really need to see your ini/hal
12:10 PM mrec: does anyone use those ihsv57 jmc motors?
12:10 PM mrec: I wonder the provided torque curve doesn't seem to be right
12:11 PM mrec: a 200w delta motor delivers a better performance, while the ihsv57 will pass out relatively quick
12:15 PM JT-Shop: nap time
12:48 PM mrec: yes but there's nothing that can be done it seems to improve the performance
12:48 PM mrec: I'm using it with a 350w powersupply, I noticed that a lower voltage seems to be better rather than a higher one
12:49 PM mrec: with 40V it passes out quicker than with 30V
12:49 PM mrec: my application needs relatively high speed 42m/min is the maximum I'm able to get with it
12:50 PM mrec: I'm also using one on my CNC on the Z axis, but I prepared it already to replace it with a stepper motor it also passes out on low speed/medium torqure
12:51 PM mrec: Z is okay if you can maintain a minimum speed, very low speed is torturing it
12:52 PM mrec: I don't believe their torque curve, that's something copied from somewhere there's no such thing as a linear torque curve with that motor.
01:15 PM JT-Shop: yea my simple woodturning inserts have shipped
01:16 PM Tom_L: lunch first, then nap
01:27 PM JT-Shop: good plan
01:28 PM * JT-Shop parts off the medallion @80 rpm and very slow speed...
01:45 PM skunkworks[m]: spray - only if your setup is using pid.. It could be just a position loop to the stepgens.
01:55 PM * spray[m] posted a file: (2KiB) < https://libera.ems.host/_matrix/media/v3/download/jauriarts.org/uxRhkxkIXtNYjpELzYfnBfdw/remora-nvem.hal >
01:55 PM * spray[m] posted a file: (3KiB) < https://libera.ems.host/_matrix/media/v3/download/jauriarts.org/IBvnZxbeJEcrSKwmqRfgIrWr/remora-nvem.ini >
01:55 PM spray[m]: here you go
01:57 PM CaptHindsight[m]: how do attachments sent from Matrix or Deluge appear on IRC?
01:57 PM XXCoder: as url
01:57 PM CaptHindsight[m]: thanks
02:07 PM -!- #linuxcnc mode set to +v by ChanServ
02:26 PM Loetmichel: FINALLY! After 2 days, 1 hour and 7 minutes i finished the Mech. 400% of the usual tabletop model size. First try broke apart while removing Support material :( -> http://www.cyrom.org/palbum/main.php?g2_itemId=18107&g2_imageViewsIndex=1
02:36 PM TurBoss: Nice! reminds me TA robots
02:38 PM Loetmichel: Battletech (the tabletop)
02:43 PM skunkworks[m]: position mode.
02:44 PM spray[m]: Is that for me?
02:54 PM skunkworks[m]: yes.
02:56 PM skunkworks[m]: so no pid. So - check your following errror - something doesn't seem right. You have headroom.
02:57 PM skunkworks[m]: how does the computer communicate with remora?
02:58 PM skunkworks[m]: 833 hz servo thread?
03:12 PM -!- #linuxcnc mode set to +v by ChanServ
03:44 PM roycroft: i went to jerry's at lunch today to check price and availability of some plastic laminate
03:44 PM roycroft: first, they can't get anything smaller than a 4'x8' sheet any more
03:44 PM roycroft: they used to be able to order cut sizes much smaller
03:44 PM roycroft: so fine, i'll get a 4x8
03:44 PM roycroft: $91 for the color i wanted in formica brand
03:44 PM roycroft: but they wanted $255 shipping for it
03:45 PM roycroft: they have a wilsonart color that's not what i want, but it's $96 with no shipping charge
03:45 PM roycroft: i asked where the formica comes from and they said it comes from a warehouse in portland
03:46 PM XXCoder: 255? its cheaper to drive there yourself. even with truck. wow
03:46 PM roycroft: they also said they could not add it to an order and save the shipping charge
03:46 PM roycroft: it can roll up and fit in the back seat of my prius
03:46 PM roycroft: it should be like $20 for shipping, if anything
03:47 PM XXCoder: cool :) yeah wasnt sure if it could be bent thats why I menioned truck
03:47 PM roycroft: yes, without heating it can be rolled up loosely
03:47 PM roycroft: with heating it can be molded
03:47 PM roycroft: think of a molded countertop
03:48 PM roycroft: but without heating i can roll up an 8' long sheet to a diameter of 18" or so
03:48 PM XXCoder: 18" thats plenty small enough
03:48 PM roycroft: yes
03:48 PM roycroft: i might just take a whole different approach to this
03:48 PM roycroft: this would be for the extension table on my table saw, which will also house my router lift
03:49 PM roycroft: and i'm thinking i might laminate some whiteboard material to the top of it
03:49 PM roycroft: it would be nice to be able to scribble notes on it and erase them easily when i don't need them any more
03:49 PM XXCoder: isnt there whitewall paint?
03:49 PM XXCoder: I recall something like that
03:50 PM roycroft: i don't want to paint it - it needs to be more durable than paint
03:50 PM XXCoder: if I remember right its similiar as it hardens
03:50 PM XXCoder: could be misrememebring lemme do quick look
03:50 PM roycroft: but yes, there is white board paint
03:51 PM XXCoder: https://www.remarkablecoating.com/remarkable-now-leed-compliance-certified-whiteboard-paint/
03:52 PM XXCoder: interesting. well yeah probably not as good as regular whitewall dunno
03:53 PM roycroft: plastic laminate would be best
03:53 PM roycroft: gloss white formica would be fine
03:53 PM roycroft: or wilsonart
03:53 PM roycroft: since i don't want to pay the extra $255 :)
03:53 PM XXCoder: heck even driving there is cheaper by quite a lot
03:54 PM roycroft: yes, it is, and i could combine it with another trip
03:54 PM XXCoder: lets assume ricious cost of $100 of gas one way, its still cheaper
03:55 PM roycroft: but i'm not sure i even want to deal with an outfit that charge that much for shipping
03:57 PM XXCoder: thats always up to you
03:58 PM XXCoder: in many years, I only bought from walmart 2 times
03:58 PM XXCoder: both times being literally cant buy elsewhere
03:58 PM roycroft: i went to a walmart once, back in 1994 or 1995
03:59 PM roycroft: the store was full of cheap, low quality chinese junk
03:59 PM XXCoder: yeah and last time I went it was really messy
03:59 PM roycroft: i saw no reason to ever go back
03:59 PM XXCoder: not enough workers
04:00 PM XXCoder: second one I just ordered online. didnt care to enter again
04:04 PM spray[m]: I have ec500 controller connected to pc with ethernet cable. Other than that, I have no idea. From what I read I understand that I dont need the base thread, since the stepping is done on ec500 controller. It should be capable of 460kHz per axis
04:05 PM spray[m]: I was getting following error on joint 0 as well
04:05 PM rmu: spray[m]: did you fix the following error?
04:06 PM spray[m]: I did experiment with slowing down and it seemed to help
04:06 PM spray[m]: But Im not entirely sure its fixed for 100%
04:07 PM rmu: what did you slow down
04:07 PM spray[m]: Max feed rate
04:07 PM spray[m]: Does servo thread period impact my max feed rate?
04:08 PM spray[m]: And does my computer latency/max jitter impact servo thread period?
04:08 PM rmu: no (for sane feedrates)
04:10 PM spray[m]: So i would like to have 5000 steps per rotation on 5mm ball screw. JMC integrated servo is 3000rpm which is 50r per second. So I should be able to reach 250mm/s
04:11 PM spray[m]: 5000 steps at 50r per seconds is 250kHz stepping rate, right?
04:12 PM rmu: did you find halscope?
04:12 PM spray[m]: So what computer latency and what servo thread period do I need, so that I dont get following joint 0 error?
04:13 PM rmu: following error has nothing to do with servo thread period or latency / jitter
04:13 PM spray[m]: I found it, but I didnt succeed with plotting cmd position and real position
04:13 PM spray[m]: Alright. Good to know
04:14 PM rmu: imagine remore outputs max. 100kHz and you command it to output 120kHz, then you will get following error. (numbers made up)
04:15 PM rmu: remora
04:15 PM * spray[m] uploaded an image: (7044KiB) < https://libera.ems.host/_matrix/media/v3/download/jauriarts.org/GhFABNTwZXVrLJsKGTOwcQwF/7AB418F7-4FCC-49E8-9AB0-45FD4DAD057B.jpg >
04:15 PM rmu: you will have to look at commanded vs. "feed back" position with halscope
04:16 PM spray[m]: Im firing up linuxcnc
04:16 PM rmu: that should be OK, but that kernel isn't without problems IME
04:17 PM spray[m]: Should I switch to different one? Which one?
04:18 PM * spray[m] uploaded an image: (8893KiB) < https://libera.ems.host/_matrix/media/v3/download/jauriarts.org/wYPSSeGOSMvtPeXRolALBJAG/AAB915FF-BA93-425B-817A-36356FC47867.jpg >
04:18 PM spray[m]: BTW i always get this one at start☝️
04:18 PM spray[m]: Does it matter?
04:18 PM rmu: keep it for now
04:19 PM rmu: thats not good
04:19 PM rmu: your linuxcnc, did you compile it yourself?
04:21 PM spray[m]: I followed latest guide for installing probebasic
04:21 PM spray[m]: Its bookworm debian and there is an apt repository with linuxcnc debs
04:22 PM spray[m]: So I just installed linuxcnc via apt
04:28 PM rmu: do you know the "make" of your network interface?
04:31 PM rmu: (realtek, intel, ...)
04:33 PM rmu: if it's possible on your hardware, you could try this https://packages.debian.org/de/buster/linux-image-rt-amd64 kernel, it seems to work better than newer debian rt preempt kernels.
04:35 PM spray[m]: Realtek
04:36 PM * spray[m] uploaded an image: (7721KiB) < https://libera.ems.host/_matrix/media/v3/download/jauriarts.org/uHpOdwiqrnoZfkCtBsGCfPHq/8420C3BF-EFBF-4F25-92D8-07C2A5F1C3D1.jpg >
04:36 PM spray[m]: Last line in terminal
04:41 PM rmu: how many cores are there? did you isolate one or more?
04:41 PM rmu: (isolcpus=...)
04:42 PM spray[m]: Should be 4cores 4 threads i5 6500
04:42 PM * spray[m] uploaded an image: (7482KiB) < https://libera.ems.host/_matrix/media/v3/download/jauriarts.org/NqLCqDFHZDQAyWSEmVEPdHKY/4BDD5F1F-1C64-4F06-8DE6-2A1A812B15D7.jpg >
04:42 PM spray[m]: Im not aware of that
04:43 PM rmu: type "cat /proc/cmdline" and look for isolcpus=
04:45 PM spray[m]: Let me install the kernel you suggested first
04:45 PM skunkworks[m]: or run latency-histogram it will show if you have isolcups.
04:46 PM rmu: it says no_isolcpus here https://libera.ems.host/_matrix/media/v3/download/jauriarts.org/GhFABNTwZXVrLJsKGTOwcQwF/7AB418F7-4FCC-49E8-9AB0-45FD4DAD057B.jpg
04:46 PM spray[m]: Right
04:47 PM rmu: spray[m]: next step before doing anything else, add isolcpus to kernel cmdline, reboot, and try again
04:47 PM rmu: isolcpus=3
04:48 PM rmu: https://askubuntu.com/questions/19486/how-do-i-add-a-kernel-boot-parameter
05:00 PM spray[m]: Got it. But I did not change the kernel yet
05:00 PM * spray[m] uploaded an image: (9845KiB) < https://libera.ems.host/_matrix/media/v3/download/jauriarts.org/xImpPUEijvjdWNvtgNKwCxqg/DA1A2564-F915-466A-924A-0D56BC1FEB4E.jpg >
05:04 PM spray[m]: Ok. No initial error now
05:07 PM spray[m]: Thank you very much for now. I dont get the first error. I get back when I change the kernel for the one you suggested.
05:09 PM rmu: it's past midnight in my timezone, off to bed
05:11 PM rmu: i would only change kernel if you experience further latency problems. next thing i would check if udp packet flow is working as expected
05:12 PM rmu: i guess remora will be similar to mesa eth protocol, i.e. send one UDP packet to request current values, wait for answer, and at last write new values in third UDP packet, then wait for next servo period
05:13 PM rmu: use direct ethernet connection without switch, make sure linuxcnc is the only thing using that interface
05:51 PM spray[m]: We are in the same timezone then 👍
05:53 PM * JT3196[m] uploaded an image: (1082KiB) < https://libera.ems.host/_matrix/media/v3/download/jauriarts.org/XuHaWnONaQAOQqjxpUaEmEQE/image.png >
05:53 PM JT3196[m]: I'm confused what's causing this
05:54 PM JT3196[m]: This is using ethercat and corexy kinematics
05:59 PM * JT[m] uploaded an image: (283KiB) < https://libera.ems.host/_matrix/media/v3/download/matrix.org/LXvocJKsOmFZuQGDHTDKLXNT/20230404_174426.jpg >
06:00 PM XXCoder: start of another bowl nice
06:00 PM JT-Shop: the wedding bowl a bit late...
06:01 PM * JT[m] uploaded an image: (278KiB) < https://libera.ems.host/_matrix/media/v3/download/matrix.org/XWVOWlBjHKGaiokWLMAZYPNN/20230404_143233.jpg >
06:06 PM XXCoder: bottom, or covered?
06:06 PM JT-Shop: bottom
06:06 PM XXCoder: nice
06:07 PM JT-Shop: thanks
06:09 PM roycroft: that turned out really nice, jt
06:09 PM roycroft: i'm glad you went with the brass badge
06:09 PM JT-Shop: thanks
06:09 PM Tom_L: almost looks like a pro did it :)
06:09 PM JT-Shop: lol
06:10 PM * JT[m] uploaded an image: (264KiB) < https://libera.ems.host/_matrix/media/v3/download/matrix.org/ARSyltvodPuYBPaQwDcdfdsK/20230404_175520.jpg >
06:10 PM JT-Shop: the pattern is set in stone now
06:10 PM Tom_L: must be a tight fit if you used string
06:11 PM JT-Shop: pretty close fit and it's flossing
06:11 PM XXCoder: dots is your alignment aid?
06:12 PM JT-Shop: yup, the sides are not exactly identical so the two dots indicate the same end on all rings
06:20 PM roycroft: tomorrow is pay day, but money is tight this month
06:21 PM roycroft: why did i just get sale catalogs/flyers from woodcraft, rockler, travers, lee valley, and grizzly in the mail today?
06:21 PM XXCoder: they know
06:21 PM XXCoder: ;)
06:21 PM Tom_L: because you ordered from them?
06:22 PM roycroft: they think they know
06:22 PM roycroft: tomorrow my bank account will be fat
06:22 PM roycroft: thursday i pay bills
06:22 PM roycroft: friday my bank account will be very lean
06:22 PM JT-Shop: lol
06:23 PM ILoveApplePi: ping
06:24 PM JT3196[m]: pong
06:24 PM roycroft: boingy boingy boingy
06:25 PM ZincBoy[CAON][m]: I don't think very many people have experience with what you are doing (ethercat and corexy). So your pool of people that can respond to your questions is very low. From what you posted there is very little to say. The following errors are due to servo loop tuning or position feedback issues. Then you have the soft limit which means that joint when out of its allowed range (possibly due to the above). You would need to post INI
06:25 PM ZincBoy[CAON][m]: and HAL files to hope to get more info.
06:26 PM JT3196[m]: But it's not anywhere near any of the soft limits, there must be a bug in translation as we are near the center of the control area. I'll post the configs shortly
06:27 PM Tom_L: are units correct?
06:27 PM ZincBoy[CAON][m]: Note that it is a joint soft error limit. Not an axis soft error. The joint limits are pre kinimatics.
06:27 PM JT3196[m]: https://pastebin.com/Q8u9jZLe ini
06:27 PM JT3196[m]: https://pastebin.com/jy3909QH hal
06:29 PM ZincBoy[CAON][m]: Yup. Your min/max limits are the same for your axis and joints. I don't believe this is correct for a corexy kinematic setup. Axis X != Joint_0 in this config.
06:29 PM JT3196[m]: Can I just not set them? because the joint is unrelated it's free to spin
06:30 PM JT3196[m]: Errr axis rather
06:31 PM ZincBoy[CAON][m]: I don't think they can be unset. I have the limits set to +/-1e89 on my rotary axis.
06:32 PM ZincBoy[CAON][m]: You want the limits on the Axis values. The limits on the Joints need to be larger/different.
06:33 PM ZincBoy[CAON][m]: If you don't have limits on your XYZ linuxcnc will happily run into the hard limits.
06:33 PM JT3196[m]: My understanding would be that the Joint is the Cartesian motion where as the axis is the raw motor position used for corexy
06:33 PM JT3196[m]: Is that wrong?
06:33 PM ZincBoy[CAON][m]: Inverse.
06:33 PM JT3196[m]: Ah okay
06:34 PM ZincBoy[CAON][m]: Axis is the cartesian location. Joint is the physical actuator.
06:35 PM ZincBoy[CAON][m]: Hence the homing being located in the joint section.
06:35 PM JT3196[m]: The problem for me is that with corexy home position is based on both motors current state
06:40 PM ZincBoy[CAON][m]: Looks like the corexy comp is sitting between linuxcnc and the joints. So you are not using a corexy kinematics but a corexy translation layer. Not ideal but it will work. The above limit discussion assumed you were using a corexy kinematic.
06:41 PM ZincBoy[CAON][m]: Your HAL file is not using PID for the axis. I have found that things don't work as expected if you delete the PID. Don't know why. Just my experience.
06:42 PM JT3196[m]: I'm just not sure what to do about PID because all of the feedback control is done on the drivers
06:42 PM ZincBoy[CAON][m]: Also, you need to check that the corexy.jX-motor-pos-fb scale matches the joint.X.motor-pos-cmd
06:42 PM ZincBoy[CAON][m]: The PID is use the same way as in stepper configs.
06:43 PM ZincBoy[CAON][m]: I highly suggest starting with one of the generated stepper configs and modify from there.
06:43 PM ZincBoy[CAON][m]: Going from scratch is possible but you really need to understand everything that is going on.
06:44 PM ZincBoy[CAON][m]: Check the scaling using halscope and see what is happening when you jog.
06:56 PM JT3196[m]: Yeah i was trying to go through some of the anemic ethercat examples for these drivers that I had found on github, I'll try and start with a stepper example and work forwards. I couldn't find any examples using the corexy kins so I had to use the HAL comp.
06:58 PM JT3196[m]: Thanks for your help I really appreciate it
07:33 PM -!- #linuxcnc mode set to +v by ChanServ
07:48 PM * dudul[m] uploaded a video: (5819KiB) < https://libera.ems.host/_matrix/media/v3/download/jauriarts.org/BkULjoGiJDfjUCqqnzwqMEyu/VID_20230316_134813_HSR_12014925968.mp4 >
07:48 PM dudul[m]: Is it safe to do this?! Servo is ethercat delta asd-a2L, ballscrew is tbi, double nut. Sini ka300 linear encoder is connected to drive
07:48 PM dudul[m]: I want this axis to make a reciprocating move of a length of 0.04mm in 5ms . Its total mass is 25kg
07:48 PM dudul[m]: And it should be synchronized with the c and z axis too.
07:49 PM CloudEvil: hmm
07:49 PM dudul[m]: * Is it safe to do this?! Servo is ethercat delta asd-a2L, ballscrew is tbi, double nut. Sini ka300 linear encoder is connected to drive
07:49 PM dudul[m]: I want this axis to make a reciprocating move of a length of 0.04mm in 5ms . Its total mass is 25kg
07:49 PM dudul[m]: And it should be synchronized with the c and z axis too.
07:49 PM dudul[m]: Video starts with normal speed and then slow motion begins
07:56 PM ZincBoy[CAON][m]: Define "safe" You would need to do the calculations to determine critical speeds and loads.
07:59 PM CloudEvil: I have not done the computations, but for most stuff that's well within the elastic range anyway
08:01 PM ZincBoy[CAON][m]: The bigger factor is going to be how long it will work. Big difference between 5 seconds and 5 years.
08:01 PM CloudEvil: Indeed
08:02 PM dudul[m]: Thats what i want to know!
08:02 PM dudul[m]: It should work for commercial tasks. So maybe 24/7 on 1 year
08:04 PM CloudEvil: If I've not blown off the units, that's of order 1G
08:04 PM CloudEvil: which is not ridiculous
08:04 PM ZincBoy[CAON][m]: You will need to go through the load/velocity/life charts for all the components and work it out. If you don't have those charts for a component then I would suggest replacing with a component that does. Otherwise the best answer is "try it and see". For something like this you would be much better off with a linear servo.
08:05 PM CloudEvil: But resonances will make it unlikely your motion system will actually meaningfully respond to a 200hz signal even if your motors have any actual response at that speed.
08:05 PM CloudEvil: ^and you have no backlash
08:05 PM CloudEvil: An additional actuator designed to directly make the final motion might work better
08:06 PM ZincBoy[CAON][m]: Yes a voice coil style actuator might work.
08:13 PM dudul[m]: The problem is that i am in iran and i dont have acces to vcm
08:13 PM dudul[m]: Is it logical to diy it ?
08:21 PM dudul[m]: I tested this on a test axis and the whole table and system was shaking badly.