#linuxcnc Logs

Sep 10 2023

01:24 AM Deejay: moin
01:50 AM travis_farmer: Morning :-)
01:55 AM pere: hi
01:56 AM travis_farmer: :-)
04:58 AM JT-Cave: morning
05:21 AM rigid: can anyone recommend a (cheap) "mini lathe" with a stepper motor? i even lack the terms to properly search for it :)
05:22 AM rigid: actually, I just need the clamping thing. I found those that attach to a drill, but I probably need to do a lot of machining to fit it to a stepper motor shaft
05:38 AM Tom_L: morning
05:42 AM -!- #linuxcnc mode set to +v by ChanServ
05:42 AM -!- #linuxcnc mode set to +v by ChanServ
06:01 AM travis_farmer: JT, did you see my convorsation yesterday about MesaCT on windows? https://www.tjfhome.net/community/index.php?resources/running-mesact-on-windows-10.2/ (my install notes)
06:29 AM JT-Cave: yes I did and thanks for testing it
06:29 AM travis_farmer: not a problem :-)
06:38 AM JT-Cave: rooster just crowed
07:05 AM JT-Cave: and the buck just broke his food bowl...
07:06 AM travis_farmer: well heck, no dessert for him ;-)
07:17 AM JT-Cave: time to ride
07:39 AM rigid: ah, "lathe chuck" was the term. and they're available with varying shaft diameters https://www.aliexpress.com/item/1005005513833700.html yay
10:13 AM mrec: pcw--home: can you give me the part number of p1 of the 7i96s? I'd like to buy a suitable counterpart for it
10:13 AM mrec: it's the expansion port connector
10:14 AM pcw--home: You want the flat cable IDC connector?
10:15 AM mrec: yes
10:15 AM mrec: should I still add a buffer or can I directly connect encoders to it?
10:16 AM pcw--home: we normally use AMP 1658621-6
10:16 AM pcw--home: If you connect an encoder directly I would all say 100 Ohm series resistors
10:16 AM pcw--home: s/all/add/
10:18 AM mrec: I think our local electronic store has something similar I'll check tomorrow
10:19 AM pcw--home: the GPIO (as inputs) will take up to +7V but negative spikes (-0.7V or less) that are not current limited will damage the input clamp diode on the bus switches
10:21 AM pcw--home: How do you get a negative input spike? Its easy, just ground a GPIO pin to frame ground that has ~2V of HF unlimited current noise (from a VFD etc) relative to the 7I96 ground
10:22 AM mrec: I will have a closer look at that tomorrow, I only wired up the glass scales to the default encoder onboard
10:23 AM mrec: the board was lying around for nearly a year but now I want to finish moving my machine away from parallel port
10:24 AM mrec: so far I only modified the efinix project to support the encoders via expansion port
10:42 AM pcw--home: The best encoder interface is differential if your hardware supports it
10:47 AM mrec: the glass scales are single ended.
11:05 AM Tom_L: from 2.7 to 2.8 ilopass lowpass.n.scale was dropped however lowpass.n.load was added. is this the same functionality?
11:06 AM Tom_L: in the documentation at least...
11:07 AM Tom_L: or is scale still a part of ilowpass but not documented?
11:09 AM JT-Shop: http://linuxcnc.org/docs/stable/html/man/man9/ilowpass.9.html
11:11 AM -!- #linuxcnc mode set to +v by ChanServ
11:11 AM Tom_L: i see now... there's an ilowpass and a lowpass :)
11:12 AM Tom_L: wonder why there's 2 variants
11:17 AM JT-Shop: no clue here
11:26 AM JT-Shop: nap time
11:26 AM Tom_L: JT-Shop, every now and then my pendant scale for the mpg gets off a bit
11:26 AM Tom_L: any ideas?
11:27 AM Tom_L: like it doesn't see one of the scales
11:27 AM Tom_L: i checked the wiring
11:27 AM Tom_L: i looked at it in halshow and the mux4 out doesn't switch to the low setting
11:28 AM Tom_L: the 2 inputs are a binary switch: 00 01 10 11
11:28 AM Tom_L: afik
11:32 AM travis_farmer: mux4 needs 4 inputs: http://linuxcnc.org/docs/2.9/html/man/man9/mux4.9.html from what i see
11:32 AM travis_farmer: nope, i read wrong...
11:58 AM travis_farmer: i was running through the calculations for charge_pump ( http://linuxcnc.org/docs/2.9/html/man/man9/charge_pump.9.html ), and man my machine runs slow... says it is running with a base frequency of around 178.57Hz (SERVO_PERIOD = 2,800,000) :-( and i thought the machine was running good-enough with this computer...
12:15 PM travis_farmer: i mean, should i yet again be on the lookout for better hardware? or is 2.8 million ns fine for Mesa hardware? it just seems like it would be dirt-slow
12:18 PM travis_farmer: it seems to perform OK for the runs i have done... am i just paranoid?
12:19 PM CaptHindsight: base thread with Mesa?
12:19 PM CaptHindsight: Mesa uses a servo thread
12:20 PM travis_farmer: i am using servo thread
12:20 PM CaptHindsight: ok
12:20 PM CaptHindsight: and when you run a latency test what are your results?
12:21 PM CaptHindsight: on the servo thread
12:22 PM travis_farmer: i don't remember the numbers, but they are a lot less than 2.8 million ns. but i run that high just to keep from getting realtime errors
12:22 PM travis_farmer: i can run out and re-test...
12:22 PM CaptHindsight: then you weren't getting accurate test results or ....
12:23 PM CaptHindsight: we have discover something else run amok on your system
12:23 PM * travis_farmer heads out to shop
12:30 PM roycroft: it's amok time
12:31 PM roycroft: watch out for spok - he's not being very logical
12:31 PM tjf_shop: travis_farmer is out in shop
12:32 PM * roycroft just ruined a part because of an error on the drawing
12:32 PM roycroft: now i'm fixing up all the errors
12:32 PM roycroft: the assembly was fine, because the same error was on all the parts
12:33 PM roycroft: fortunately there are only three distinct parts, so the fixes should not take long
12:33 PM tjf_shop: running latency-histogram --nobase
12:46 PM tjf_shop: set isolcpus=3 and max servothread is 49.9us with youtube running
12:49 PM tjf_shop: running hexchat, if it's better...
12:49 PM tjf_shop: servoperiod is 58.2us now
12:51 PM tjf_shop: not getting any IRC chat, has hexchat failed me???
12:55 PM CaptHindsight: failed?
12:56 PM tjf_shop: nope, came through
01:01 PM tjf_shop: held steady at 58.2us, so i tried 100000ns servo period, no go (realtime error). bumped to 1000000, and it's running tort.ngs right now, and much faster!
01:03 PM tjf_shop: i didn't have isolcpus=3 before, so i think that was the issue
01:11 PM CaptHindsight: mebe
01:11 PM tjf_shop: ok, i have raised my mm/min up to 5340, from 4000-something, so it is an improvement, for a crap computer
01:12 PM tjf_shop: no realtime errors
01:12 PM CaptHindsight: so not hardware error, just user error
01:12 PM tjf_shop: apparently...
01:13 PM CaptHindsight: it's ok this time
01:14 PM tjf_shop: i am happy with the rapids, i will leave it as is for now, and check with a real machine run when i get the software (Vcarve Pro)
01:15 PM CaptHindsight: intel or amd cpu?
01:15 PM tjf_shop: intel
01:16 PM CaptHindsight: the way AMD does their caches tends to give higher latency than Intel unless you cripple the cpu down to just two cores
01:17 PM CaptHindsight: maybe it's a firmware issue but we won't know until we get more open source firmware from AMD
01:18 PM CaptHindsight: https://www.phoronix.com/news/AMD-Genoa-Onyx-Coreboot
01:40 PM travis_farmer: back inside the house... i don't know why i am having such IRC issues...
01:41 PM travis_farmer: would not let me connect for several minutes...
02:15 PM travis_farmer: just hypothetical... i have a Dell PowerEdge 1950 with dual 4-core (i think) Xeon CPUs, and 32GB RAM. i heard the Xeon CPUs have better latency results... i wonder how it would fair
02:17 PM travis_farmer: or am i just better off to leave well-enough alone?
02:21 PM travis_farmer: actualy, it is dual 2-core Xeon CPUs...
02:32 PM travis_farmer: gonna try it tomorrow, and see. if i use isolcpus with a dual CPU, how best to set that up? i currently have it set to 3 on my running machine
02:34 PM CaptHindsight: it's a non issue with Mesa and servo threads
02:34 PM travis_farmer: what do you mean?
02:36 PM CaptHindsight: just about anything other than some laptops and some boards with broken BIOSes are fine with Mesa and servo threads
02:37 PM travis_farmer: Hmm, ok. guess i can take it out of my running machine then...
02:39 PM CaptHindsight: IF your system gets 58.2uS latency during the test and somehow has issues at 100uS then you're not running the test long enough or with enough memory allocating apps running
02:39 PM CaptHindsight: test again with youtubes for at least a couple hours
02:40 PM travis_farmer: will do tomorrow :-)
02:40 PM CaptHindsight: " so i tried 100000ns servo period, no go (realtime error)." does this happen each and every time or just the first time you start LCNC?
02:40 PM travis_farmer: as soon as it starts
02:41 PM CaptHindsight: each and every time or just the first time you start LCNC? is the question
02:41 PM travis_farmer: every time
02:41 PM CaptHindsight: which latency test are you running?
02:42 PM travis_farmer: none at the moment, but i was running "latency-histogram --nobase" with youtube running and i had 4 glx gears
02:43 PM CaptHindsight: post a pic of the results next you get a chance
02:43 PM CaptHindsight: please
02:43 PM travis_farmer: certainly :-)
02:44 PM CaptHindsight: glx gears doesn't have much of an effect as it used to many years ago
02:44 PM travis_farmer: ok, but the youtube is fine?
02:44 PM CaptHindsight: FF playing youtubes is still one of the best tests
02:44 PM travis_farmer: ok
02:45 PM CaptHindsight: haing good latency on the test for a few hours and then having immediate LCNC latency issues ususally is some config problem with LCNC
02:45 PM CaptHindsight: haing/having
02:46 PM travis_farmer: ok
02:46 PM travis_farmer: https://github.com/travis-farmer/linuxcnc are all my configs
03:46 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/pendant/new/Pendant_face2.jpg
03:46 PM Tom_L: facelift for my pendant
03:47 PM travis_farmer: nice
03:47 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/pendant/new/Pendant_old1.jpg
03:47 PM Tom_L: engraving was a bit light on the bottom
03:48 PM travis_farmer: yeah, easier to read the new one
03:48 PM Tom_L: especially once i add black to the letters
03:48 PM travis_farmer: yup :-)
04:04 PM skunkworks: Tom_L: very nice!
04:05 PM skunkworks: OMG - those push button switches.. That was my childhood. For some reason dad had a ton of those
04:11 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/pendant/new/Pendant_face4.jpg
04:16 PM JT-Shop: nice
04:16 PM travis_farmer: seems to be pouring rain right now..
04:16 PM roycroft: it's hot and sunny here
04:17 PM roycroft: and i have to cut the grass today still
04:17 PM roycroft: well, cut the weeds
04:17 PM roycroft: the grass hasn't grown since may
04:19 PM skunkworks: Tom_L: what does tool and spindle buttons do?
05:20 PM Tom_L: spindle turns it on and off, tool is for manual tool change
05:21 PM skunkworks: Huh.. twice now I have seen someone showing an error to the effect - exceeded negative soft limit (0.000000) on joint 0
05:21 PM Tom_L: i use the pendant more than the screen
05:21 PM skunkworks: joint must be unhomed, jogged into limits, rehomed?
05:22 PM Tom_L: i assume so
05:22 PM skunkworks: never run into that
05:22 PM skunkworks: is that where you are homing to a hard limit?
05:22 PM Tom_L: no idea
05:22 PM skunkworks: weird
05:22 PM Tom_L: headin back out to swap the new face out
05:23 PM skunkworks: going to need more pictures :)
05:36 PM markgonz: hello everyone
05:37 PM markgonz: I am looking for some help from a linuxcnc pro
05:37 PM markgonz: I need to wire my spindle vfd to mesa card
05:37 PM JT-Shop: just ask your question
05:38 PM markgonz: I have omron mx2 vfd , this is the manual https://assets.omron.eu/downloads/manual/en/v3/i570_mx2_users_manual_en.pdf
05:38 PM markgonz: and mesa 7i96s
05:38 PM markgonz: I want to do 0-10v to control the spindle
05:39 PM markgonz: but I get lost with how to wire this thing
05:40 PM markgonz: page 201 on the manual explains how to use inteligent input terminals and I thing there is the important info
05:40 PM JT-Shop: a 7i96S spindle control is a pot type so you need 10v to one side 0v to the other side and the wiper is going to the spindle
05:43 PM markgonz: this means I need and external supply?
05:43 PM JT-Shop: yup
05:43 PM markgonz: I read that the inverter had a internal supply that can be used
05:43 PM JT-Shop: quite possible if it has 10v
05:43 PM markgonz: is 24v I think
05:44 PM markgonz: my problem is that I am a total noob at electronics
05:45 PM markgonz: and I wish to learn but right now this spindle wiring is driving me nuts
05:46 PM JT-Shop: you might consider just having spindle on/off and change the freq manually
05:47 PM markgonz: ok that would be fine
05:47 PM markgonz: is all I need
05:47 PM JT-Shop: are you using the Mesa Configuration Tool?
05:48 PM markgonz: I am not sure, is it the same as the pncconfig wizard?
05:48 PM JT-Shop: nope
05:49 PM JT-Shop: https://gnipsel.com/linuxcnc/mesact/index.html
05:50 PM markgonz: ok I will look into it
05:52 PM JT-Shop: I'm the author so if you have a problem reach out here
05:52 PM markgonz: ok thanks,thanks for your work
05:52 PM * JT-Shop goes to put a tomahawk on the grill
05:53 PM markgonz: https://www.youtube.com/watch?v=bt_uiCL2v3E just found your video of 2 weeks ago, that will help me understand it more
05:53 PM JT-Shop: I need to make more...
05:56 PM markgonz: but my idea was to figure out the damn wiring first because the vfd manual seems hard to understand
05:58 PM markgonz: I think my option is to learn enough electronics to understand it or hire someone at this point
05:58 PM markgonz: because almost everything on it seems a new language to me..
05:59 PM markgonz: and I understand reading the damn manual is not a job for the linuxcnc people that already do so much for this project
06:40 PM Tom_L: skunkworks, http://tom-itx.no-ip.biz:81/~webpage/cnc/pendant/new/Pendant1.jpg
06:40 PM Tom_L: works better now too
06:41 PM Tom_L: must have had a wire that wiggled loose or something
07:04 PM -!- #linuxcnc mode set to +v by ChanServ
07:04 PM -!- #linuxcnc mode set to +v by ChanServ
07:27 PM solarwind: CaptHindsight The chinese profiled linear rails, VERY pleased with all of them
07:28 PM solarwind: testing with dial indicator and absolute position sensors and stuff shows they work well within tolerances of the genuine versions
07:28 PM solarwind: The cheap chinese ballscrews on the other hand, very questionable and variable depending on the source
07:28 PM solarwind: the leadscrews are a lot better in terms of quality and consistency
07:29 PM solarwind: the HGR rails especially, damn, they are REALLY rigid and smooth
07:29 PM solarwind: even the cheap MGN rails which cost about as much as a pizza
07:29 PM solarwind: they are so good, no reason to buy anything else like the cylindrical linear guides
07:31 PM solarwind: I am seeing that the grease is getting slightly gray after 24h of constant load testing at 1m/s back and forth on a 1m rail
07:35 PM solarwind: The HGR blocks have 2 grease zerk ports, so easy to hook up a continuous oiler/grease pump to them. The MGN blocks don't seem to have that, but those are more for 3D printers and such anyway
07:44 PM skunkworks: Tom_L: nice!
07:55 PM Tom_L: skunkworks, cut anything on the lathe yet?
08:21 PM CaptHindsight: https://www.mobil.com/en/lubricants/for-personal-vehicles/our-products/products/mobil-1-synthetic-grease
08:23 PM CaptHindsight: https://www.linearmotiontips.com/how-to-choose-a-lubricant-for-recirculating-linear-bearings/
08:24 PM CaptHindsight: solarwind: ^^
09:51 PM CaptHindsight: https://www.youtube.com/watch?v=cYpYGbC2Drg
10:05 PM roycroft: when i first was exposed to gtaw welding in a welding class the instructor had me butt weld some copper without filler
10:05 PM roycroft: i won't say i was good at it but i got the parts welded together
10:07 PM roycroft: i remember having to really cook it and move really fast
10:07 PM perry_j1987: CaptHindsight you going to fabtech
10:07 PM CaptHindsight: not this time
10:07 PM perry_j1987: got some better place to be eh
10:09 PM CaptHindsight: maybe the one next year up in Canadia
10:09 PM CaptHindsight: not much really new at fabtech
10:10 PM perry_j1987: what makes you think that
10:10 PM perry_j1987: all the people debuting stuff they worked on through the pandemic haha
10:12 PM CaptHindsight: been in the industry for over 40 years
10:12 PM CaptHindsight: some yeah they have their 2023 models on display
10:12 PM CaptHindsight: but not really new tech
10:14 PM CaptHindsight: and "look what we got patented that was easy and obvious"
10:14 PM perry_j1987: *hates patents
10:14 PM CaptHindsight: especially in the additive manufacturing section
10:16 PM perry_j1987: that's held the world back for decades
10:16 PM CaptHindsight: https://www.fabtechexpo.com/exhibit/additive
10:19 PM roycroft: patents foster innovation
10:19 PM roycroft: patent abuse is what holds things back
10:20 PM CaptHindsight: https://www.fabtechexpo.com/special-events heh
10:21 PM CaptHindsight: a who is not who of fab stars
11:50 PM solarwind: CaptHindsight thanks, I'll take a look at those tomorrow
11:51 PM solarwind: roycroft yeah, for a task like that, I use flat tungsten - that is - no point, just ground to have a flat face in order to focus the arc into where I point it
11:51 PM solarwind: and I use 3.2mm lanthanted and crank it up to 300A
11:51 PM solarwind: light it up and away you to
11:51 PM solarwind: go*
11:54 PM solarwind: chill bars can be a good thing depending on what you're doing since the goal is to not get the rest of the piece super hot, just the area you're welding
11:55 PM solarwind: I made huge welding chill pads for this purpose, out of solid aluminum slabs, with water channels milled inside them
11:55 PM solarwind: kind of like a huge welding mat (a pair of welding mats so I can adjust the gap between them) that is actively water cooled
11:56 PM solarwind: I can pump so much heat into it if I strike an arc directly on it, without forming a puddle, but the water definitely boils inside
11:57 PM solarwind: can hear the popping and steam coming out of the water outlet port