#linuxcnc Logs

Apr 03 2020

12:15 AM Centurion-Dan2 is now known as Centurion_Dan
12:31 AM genoober: lucky duck
12:31 AM genoober: where ya at captain my captain
12:32 AM CaptHindsight: Chicago area
12:32 AM genoober: damn how is it colder over here in oregon
12:33 AM CaptHindsight: some dip in the jetstream
12:34 AM genoober: true...
12:35 AM CaptHindsight: https://www.wunderground.com/maps/wind/jet-stream
12:37 AM CaptHindsight: http://www.stormsurfing.com/cgi/display_alt.cgi?a=nam_250
12:45 AM genoober: very interesting
12:50 AM genoober: stupid question.... do i need to copy all of the configureation files from ~/linuxcnc-dev/configs/sim/axis/xhc-hb04 to my machine config folder at ~/linuxcnc/configs/MiniMill ?
12:52 AM genoober: i cant seem to find out what goes where and what to edit... everything just seems to say edit your hal files... but which ones? there are quite a few
12:52 AM XXCoder: pink_vampire: what did I mean on what?
01:09 AM Centurion-Dan2 is now known as Centurion_Dan
01:21 AM genoober: argk
02:00 AM Centurion-Dan2 is now known as Centurion_Dan
02:46 AM genoober: i think i may have found what has been killing me lol "libusb: error [_get_usbfs_fd] libusb couldn't open USB device /dev/bus/usb/001/013: Permission denied" for my whb04 pendant
02:46 AM genoober: im googling around but none of the solutions im finding are working
03:01 AM Centurion-Dan2 is now known as Centurion_Dan
03:23 AM Deejay: moin
03:50 AM rmu|w: genoober: you need to install a rules file in /etc/udev/rules.d to set up permissions for your pendant.
03:51 AM rmu|w: genoober: http://wiki.linuxcnc.org/cgi-bin/wiki.pl?Using_A_XHC-HB04_Wireless_MPG_Pendant
04:08 AM Tom_L: 29°F
04:08 AM Tom_L: morning
04:32 AM mr_boo: hi. is this a smart idea for a rail? https://imagebin.ca/v/5HpsJ3KX8jFw
04:36 AM Loetmichel: mr_boo: nope
04:37 AM mr_boo: what would be the obvious disadvantage?
04:37 AM Loetmichel: do three bearings and a massive ground smooth rod, no tube
04:37 AM XXCoder: mr can you move it around or twist etc
04:37 AM Loetmichel: imprecise surface on the square tube AND it will be twisted over the length
04:38 AM XXCoder: rails and such is pretty cheap lately
04:38 AM Loetmichel: http://www.cyrom.org/palbum/main.php?g2_itemId=8884 <- works better
04:38 AM mr_boo: i couldn't get the holes to line up so its already twisted
04:38 AM Loetmichel: but considering the price of SBRs these days i wouldnt bother at all
04:38 AM XXCoder: even MGNs is pretty cheap
04:39 AM XXCoder: in least compared to when I bought my sbr set lol
04:39 AM mr_boo: Loetmichel : how big are those rods?
04:39 AM Loetmichel: 16mm
04:39 AM Loetmichel: and standard 608 bearings
04:39 AM XXCoder: 20mm to 8mm various sizes
04:39 AM mr_boo: Loetmichel diameter?
04:39 AM Loetmichel: eh, 604
04:39 AM Loetmichel: diameter, yes
04:40 AM mr_boo: holy...
04:40 AM mr_boo: i think i'm gonna give up the idea of building me a cnc
04:40 AM XXCoder: why?
04:41 AM XXCoder: whats your usercase anyway?
04:41 AM mr_boo: too small workshop and not good enough equipment
04:41 AM XXCoder: no no
04:41 AM XXCoder: wood? plastics? hard steel?
04:41 AM XXCoder: project type in general
04:44 AM mr_boo: i would have liked to build a cnc from steel only
04:44 AM Loetmichel: mr_boo: buy a 6040/3020 with watercooled 800W spindle.
04:45 AM Loetmichel: you simply cant beat the bang for the buck on those
04:45 AM mr_boo: any particular brand?
04:45 AM Loetmichel: i have given up to try to make a better machine (they sure leave a lot of things to be desired) for a lesser price
04:45 AM Loetmichel: its all china brand anyways
04:46 AM mr_boo: what price range should i expect?
04:46 AM Loetmichel: look out for SBRs on y at least, ballscrews no acme and a watercooled spindle, not the 400W dc motor, that thing is crap
04:46 AM Loetmichel: about $1000 for a 6040, less for a 3040 and even less for a 3020
04:47 AM Loetmichel: never seen a 3020 with proper spindle though
04:47 AM Loetmichel: which country?
04:47 AM mr_boo: i'm in sweden
04:50 AM mr_boo: my dream in life is to have a cnc workshop
04:50 AM Loetmichel: mr_boo: someething like this: https://www.ebay.com/itm/USB-1500w-3Axis-Router-CNC-6040-Engraver-Carver-Machine-Water-cooling-VFD/274055757613
04:51 AM Loetmichel: just fromk your country or at least europe because customs
04:51 AM XXCoder: ac motor?
04:51 AM Loetmichel: ac spindle with VFD, yes
04:51 AM Loetmichel: and watercooled
04:51 AM Loetmichel: air cooled would also be ok
04:51 AM XXCoder: cool
04:52 AM Loetmichel: but the DC motors are crap because they have no bearings on the ER11 shaft at all
04:52 AM Loetmichel: they use the motor bearings
04:52 AM XXCoder: ac have torque across entire range? or in least more wide range
04:52 AM Loetmichel: torque is irrelevant for those small mill bits
04:52 AM Loetmichel: runout isnt though
04:53 AM Loetmichel: and those three phase spindles have a LOT less runout and a lot more rigidity against side forces than the "300/400W engraving motors"
04:54 AM Loetmichel: because they have proper 7xxx bearings on the rotor shaft
04:54 AM Loetmichel: i havent yet stalled mine. and those are only the 800W spindles
04:55 AM Loetmichel: not even with the 35mm facing mill ;)
04:55 AM Loetmichel: also you want to have all the RPM you can get for small mill bits
04:56 AM Loetmichel: and 24k is while not optimal a lot better than the ~10k the DC motors do
04:59 AM mr_boo: it's fortunate i haven't spent a lot of money for my failing cnc project
05:00 AM mr_boo: some $80 on stepper motors only
05:01 AM Loetmichel: mr_boo: to be perfectly clear: those CNC 6040 are not ideal machines. they have a bunch of flaws
05:01 AM Loetmichel: but they are certainly better than anything you could build for that kind of money yourself
05:01 AM jthornton: morning
05:01 AM mr_boo: still better results than what would have been obtainable from anything made by me
05:01 AM Loetmichel: indeed
05:01 AM XXCoder: jthornton: https://imgur.com/a/Mo4n3gV heh
05:02 AM XXCoder: mr_boo: what do you plan to cut with cnc? not build with :)
05:02 AM mr_boo: considering my small workshop where the vacuum cleaner takes up the only space to stand
05:02 AM mr_boo: XXCoder i thought of cutting plastic wood and circuit boards
05:03 AM XXCoder: that have very small sideload
05:03 AM XXCoder: you can build machine out of mdf and it'd be fine
05:03 AM XXCoder: (box style) Loetmichel show him your box cnc
05:04 AM XXCoder: whats important is decent rails movement system and design that allows you to dial those in and have reasonable precision machine
05:05 AM mr_boo: the nail in my coffin is my extremely small workshop https://imagebin.ca/v/5Hq2BJQNplR4
05:05 AM XXCoder: honestly thats more room that I have ;)
05:05 AM jthornton: build a small machine
05:05 AM mr_boo: the vacuum cleaner takes 50% of the space
05:06 AM mr_boo: also if i would build/buy a cnc it would have to be in my apartment
05:07 AM XXCoder: could build a cnc booth
05:08 AM XXCoder: keeps chips more managable
05:08 AM mr_boo: i have to reconsider a lot of things in my life
05:08 AM Loetmichel: XXCoder: that? -> http://www.cyrom.org/palbum/main.php?g2_itemId=11205&g2_imageViewsIndex=1
05:08 AM XXCoder: yep
05:09 AM XXCoder: mr_boo: check that link out. its huge but you can make small version
05:09 AM XXCoder: boxed cnc is very rigid for cheao
05:09 AM mr_boo: that machine looks neat and the workshop too
05:09 AM Loetmichel: better not use MDF though, use multiplex or waterproof ("marine") plywood
05:10 AM XXCoder: or protect all mdf wood some way. sealants before you assemble, then seal it again
05:10 AM Loetmichel: mr_boo: its not my first rodeo
05:10 AM Loetmichel: thats the eight machine i built over the years
05:10 AM Loetmichel: the one i built first had all the errors one could make ;)
05:11 AM Loetmichel: http://www.cyrom.org/palbum/main.php?g2_itemId=2859&g2_imageViewsIndex=1 <- that was my first try ;)
05:11 AM XXCoder: me, i have cnc router, 3d printer and tiny laser machine lol
05:12 AM mr_boo: whatever ones career might be it would now from this viewpoint appear that DIY on the CNC is the hard way to to anything
05:12 AM Loetmichel: 25 years of evolution ;)
05:12 AM XXCoder: loet i guess that predates cnc spam in aliexpress and such now?
05:12 AM XXCoder: knew it
05:12 AM mr_boo: just to drill a hole in a piece and get it at a certain location is really hard
05:13 AM jthornton: not really
05:13 AM Loetmichel: mr_boo: depends on how your education is
05:13 AM mr_boo: well if 0.5mm off is ok then its easy
05:13 AM XXCoder: jt I REALLY hate running hard metal jobs lol
05:13 AM XXCoder: baked steel is nasty material
05:13 AM Loetmichel: i had the advantage to have learnt metalworking from the ground up
05:14 AM mr_boo: but to drill a hole in a square rod with rounded corners and get all the holes in the center how do you do that?
05:14 AM jthornton: you can drill a hole very accurately if you know how or you can drill a hole in random locations if you don't know how
05:14 AM Loetmichel: you can drill a hole with a hand drill and a center punch to about 0,1mm, if you are good even 0.05mm precision easily
05:14 AM mr_boo: that was crazily hard and the rig that i made shifted eventually and all holes got off
05:14 AM Loetmichel: IF you had the right education
05:14 AM jthornton: yup
05:15 AM mr_boo: Loetmichel yeh but i had to repeat that for sixteen locations
05:15 AM mr_boo: and make them line up
05:15 AM mr_boo: one can see they don't line up well https://imagebin.ca/v/5HpsJ3KX8jFw
05:16 AM Loetmichel: yeah, thats what you learn in a metalworking apprenticeship.
05:16 AM Loetmichel: how to make thinks line up
05:17 AM mr_boo: the drill is 8mm and regardless of where the punch is there is a drift when drilling
05:17 AM Loetmichel: then you are doing it wrong
05:17 AM Loetmichel: also: predrill!
05:17 AM XXCoder: yeah should step up on sizes
05:18 AM mr_boo: i use to do that normally but i had so many holes here
05:18 AM XXCoder: i would make drill jig
05:18 AM XXCoder: if thats the issue
05:18 AM Loetmichel: mr_boo: experience tells you where you can omit predrilling and where you shouldnt
05:18 AM Loetmichel: if you dont have that, predrill everything
05:19 AM Loetmichel: but as i said at the beginning: making linear rails yourself is obsolete considering the prices for SBRs from china
05:20 AM mr_boo: still even whith these rails that i've made there are so much more to be done and i don't have access to welding anyway
05:21 AM mr_boo: my 5 squaremeter workshop are already occupied by a small milling machine, lathe, tablesaw and two grinders
05:21 AM mr_boo: the chair and vacuumcleaner takes up the remaining space
05:22 AM mr_boo: there is literally no space at all to move around
05:22 AM XXCoder: mount vacuum on wall
05:22 AM jthornton: sounds like you need to clean up and come up with a design that you can build
05:22 AM XXCoder: get longer hoses
05:22 AM XXCoder: make chair easy fold type (not same type as lawn ones but strong one
05:23 AM XXCoder: wall mount that too
05:23 AM mr_boo: jthornton i still would need to buy ballscrews or at least acme
05:23 AM Loetmichel: mr_boo: do a central vacuum. saves space
05:23 AM XXCoder: mr_boo: aliexpress have so many screws
05:23 AM XXCoder: I think ACME is fine for your needs
05:24 AM mr_boo: do the typical chinese 6040 machine have ballscrews?
05:24 AM Loetmichel: http://www.cyrom.org/palbum/main.php?g2_itemId=136 <- green are selfmade valves, red is the 50mm drainage tube, the shopvac is under the mill on the right
05:25 AM Loetmichel: both versions aviable, with ballscrews and with acme
05:25 AM XXCoder: mr make sure to buy better bearing stuff
05:25 AM Loetmichel: if you see an ad that doesent talk about ballscrews assume acme
05:25 AM XXCoder: ones in china machines is crunchy sometimes
05:25 AM XXCoder: or maybe just unpack and pack balls yourself
05:25 AM Loetmichel: XXCoder: it isnt
05:25 AM Loetmichel: acme is garbage for CNC use
05:25 AM XXCoder: as bonus you can sort balls and replace bad ones with better ones
05:28 AM Loetmichel: XXCoder: had no upper limit on my big cnc once... cant tell how often i had to search for 32 2mm balls and re-insert them into the nut... ;)
05:28 AM XXCoder: lol fun
05:28 AM XXCoder: dont lc have sft limit
05:28 AM XXCoder: *soft
05:28 AM Loetmichel: http://www.cyrom.org/palbum/main.php?g2_itemId=4669
05:29 AM Loetmichel: http://www.cyrom.org/palbum/main.php?g2_itemId=4663
05:29 AM Loetmichel: uf you forgot to lift the finger on pageup... you had "fun"
05:29 AM XXCoder: erm why is that on z assembly?
05:29 AM Loetmichel: hmm?
05:30 AM XXCoder: power stuff for spindle
05:30 AM XXCoder: vfd
05:30 AM Loetmichel: because i had probelms with the motor cables generating so much EMI that the Estop triggered
05:30 AM XXCoder: interesting
05:31 AM Loetmichel: so i made the cable from VFD to motor as short as possible
05:31 AM XXCoder: cant have heavy emf sheild?
05:31 AM Loetmichel: was less work to move the VFD ;)
05:31 AM XXCoder: wonder if can do ceiling mount but cable must be able to reach every spot. longer. hm
05:32 AM Loetmichel: same on my 6040 though
05:32 AM Loetmichel: had to shield and ferrite clip the motor cable or it would have "ghost steps" when the spindle was running
05:32 AM Loetmichel: luckily i work at a shielding company now ;)
05:33 AM XXCoder: :)
05:33 AM Loetmichel: before: http://www.cyrom.org/palbum/main.php?g2_itemId=12506 after: http://www.cyrom.org/palbum/main.php?g2_itemId=12560
05:33 AM SpeedEvil: Can you pick up the g-code from your transient emissions at 20m with a large dish and a supercomputer?
05:34 AM Loetmichel: you dont even need a dish there i would imagine
05:34 AM Loetmichel: those cheap machines are not shielded at all
05:34 AM SpeedEvil: yeah - I meant yours.
05:34 AM SpeedEvil: :)
05:35 AM XXCoder: i miss my cnc bah
05:35 AM XXCoder: but i plan to work on my laser engraver tomorrow.
05:35 AM Loetmichel: point for shielding the motor cable was that the PWM of the VFD produced enough EMI that the stepper driver inputs got triggered every now and then, even with the PC running
05:35 AM XXCoder: new 5.5v laser!
05:35 AM SpeedEvil: XXCoder: 300A?
05:36 AM XXCoder: erm I meant 5.5w
05:36 AM XXCoder: 5.5v 300a would be nasty laser indeed
05:36 AM mr_boo: is it a good idea to dry to buy a second hand cnc?
05:36 AM Loetmichel: SpeedEvil: hehe, been there done that... but with a 32*640 3mm green led pixel marquee
05:37 AM Loetmichel: needs quite a bit of power ;)
05:37 AM * SpeedEvil ponders 8W IR fiber coupled LASER.
05:37 AM Loetmichel: mr_boo: for a professional: good idea
05:37 AM Loetmichel: for someone that doesent know where to look at: nah, better not.
05:37 AM Loetmichel: imagine sending your wife to buy a cheap used car...
05:38 AM mr_boo: the StepMores 4040 is on my want list
05:38 AM SpeedEvil: 'professional' has almost no meaning.
05:38 AM SpeedEvil: If you know how to rebuild anything, then it can be a real bargain.
05:38 AM Loetmichel: SpeedEvil: prfessional as in "someone with experience with machines so that he knows where to look for hidden flaws"
05:38 AM SpeedEvil: yeah
05:38 AM syyl: i know a buisness who runs exclusively on used machines
05:39 AM syyl: 4 dmg vmcs now, all used
05:39 AM XXCoder: SpeedEvil: ir is fanastic for cutting clear stuff. clear to us anyway
05:39 AM SpeedEvil: XXCoder: nm?
05:39 AM XXCoder: not too sure. half of green?
05:40 AM SpeedEvil: twice, probably.
05:40 AM XXCoder: if I remember right green tend to use ir and double freq
05:40 AM SpeedEvil: Also, no
05:40 AM SpeedEvil: Well - 'no'
05:40 AM SpeedEvil: DPSS LASERs typically use 808nm or so pump diodes to pump a 1064nm crystal LASER, then doubled to 560nm or so.
05:41 AM XXCoder: ahh ok
05:41 AM XXCoder: watch out for china green ones, they tend to not always filter ir completely
05:41 AM SpeedEvil: Remove the doubling crystal, and you'd get 1um out with really decent focus.
05:41 AM SpeedEvil: remove the crystal LASER, and the source dimensions of the 808nm LASER may cause real issues.
05:41 AM XXCoder: would that also increase some more?
05:42 AM XXCoder: *lumen since theres less loss
05:44 AM SpeedEvil: lumens for 1064 are less (zero) than 808nm. 808nm is a little brighter than the moon, when you have the same watts per square meter as the sun hitting a surface.
05:44 AM SpeedEvil: Watts out at 808 are more, but the focussing is annoying as the emitter shape/number tends to be really annoying.
05:45 AM * jthornton watches 6 deer on the driveway camera
05:46 AM Loetmichel: syyl: sure, if they have the knowledge to asses the used machines before buying thats perfectly viable
05:46 AM Loetmichel: for a newb: not so much
05:46 AM SpeedEvil: A guarantee and a hand-holding support line can be a great thing.
05:46 AM SpeedEvil: You're gonna pay for it though.
05:53 AM * SpeedEvil sighs in article not covered by sci-hub.
05:53 AM Loetmichel: SpeedEvil: i am surprised every time that you can see which end is the "hot" end on an 850nm Fiberoptic NIC
05:53 AM * SpeedEvil is trying to find spectrometry of solder paste in the 0.5-2um range.
05:53 AM SpeedEvil: yeah
05:53 AM SpeedEvil: I think around there is where you start getting into 'if you can see it, and it's not pitch black, you're already at risk of eye damage'
05:53 AM SpeedEvil: 808nm is already falling off damn fast.
05:55 AM jymmmm: solder paste?
05:55 AM jymmmm: good morning
06:04 AM SpeedEvil: jymmmm: yes.
06:05 AM SpeedEvil: (wondering about LASER soldering again)
06:06 AM jymmmm: Oh, I thougt you meant oprical cememnt for securing fiber cable to connector
06:09 AM jymmmm: If you use a YaG laser, shouldn't be an issue
06:09 AM SpeedEvil: ^cheap
06:10 AM jymmmm: Titanium ring engraved with YaG... https://i.imgur.com/sqcqBG2.jpg
06:50 AM miss0r: meh. another mouse suffers from double click of death
06:53 AM Loetmichel: miss0r: i have a dozen or so dell mouses from the company at home
06:54 AM miss0r: I should get a stack as well... At the moment I have no spares.
06:54 AM Loetmichel: whenever they fail i just open the drawer on my desk, get a new one out, replug, throw the old on the "electrnics for the scrapheap" pile ;)
07:00 AM miss0r: yeah.. I know the owner of a pretty large computer hardware store.. I will ask him for a bucket full of generic USB mouses
07:00 AM miss0r: is it also called mice when you are speaking computers? :D
07:09 AM Loetmichel: mice, right
07:09 AM Loetmichel: sorry for not being a native english speaker
07:11 AM miss0r: hey, me neither. I was asking because I didn't know :D
07:11 AM miss0r: No reason to apologise to me anyway :]
07:12 AM sensille: axis axes :)
07:12 AM miss0r: hehe
07:21 AM Loetmichel: miss0r: we had them spare from a project where we bought a few 100 dell computers and bundled them with new Cherry keyboards and mouses
07:21 AM miss0r: Indeed
07:21 AM Loetmichel: so they are spare here and you find them in every corner ;)
07:22 AM miss0r: Its good to have
07:23 AM Loetmichel: running low though... still have a few 100 without cables because we used those in another project, but not many with the cable still attached ;)
07:23 AM Loetmichel: its funny. you can actually buy a mouse cheaper than a 3m usb cable... so we used two mouses, bite one cable off, affixed it to the other one
07:24 AM sensille: (mice)
07:24 AM Loetmichel: clipped a ferrite bead over the heatshrinked connection, perfect ;)
07:24 AM Loetmichel: sensille: YES, i KNOW now.
07:24 AM Loetmichel: keep forgetting though ;)
07:25 AM sensille: i know you know, that's the fun in it :P
07:25 AM Loetmichel: be careful when you meet me in a dark corner ;)
07:25 AM sensille: now even more
07:26 AM Loetmichel: indeed ;)
07:26 AM Loetmichel: hmm, our internel powergrid seems to be lacking some
07:27 AM Loetmichel: just heard the air compressor start... and the two shopvacs on the same circuit that supply the vacuum tables of the CNC mills with air toured down about 5%
07:27 AM Loetmichel: tour down= less RPM
07:27 AM Loetmichel: sorry if that wasnt clear
07:29 AM miss0r: I'm having a hard time motivating myself today
07:30 AM Loetmichel: *booarps* that was a nice brokkoli+sauce hollondaise + chicken pizza
07:30 AM Loetmichel: lunch break: done
07:31 AM miss0r: I feel dirty. I have mcdonalds for lunch
07:31 AM miss0r: and I feel it...
07:31 AM miss0r: I have a serious case of the mcTummy(r)
07:32 AM Loetmichel: for whatever reason that pizza is called "elvis" ,)
07:32 AM miss0r: lol
07:33 AM miss0r: if you eat it too much you will end up like he did... (And here I am not referring to being famous) :)
09:28 AM Loetmichel: maaan... two 800W shopvacs, two CNC 6040 routers, a small air compressor. All in a 20m^2 workshop without any ventilation. Running continously for 8 hours a day... SAUNAAAAA!!!111
09:29 AM pink_vampire: 10% ears left.
09:30 AM Loetmichel: hearing protection
09:33 AM pink_vampire: your shop is about the side of my home
09:34 AM Loetmichel: pink_vampire: its not big at all
09:34 AM Loetmichel: https://www.youtube.com/watch?v=lm-kpV2Msn0
09:42 AM pink_vampire: I know.
09:43 AM pink_vampire: I wish my house was bigger
09:44 AM pink_vampire: I like the enclosure that you made.
09:45 AM Loetmichel: yeah, its very useful to not have all that swarf distributed across the workplace
10:24 AM JesseG: lol yeah machines are good at distributing swarf. They think it's their primary purpose. Any machine work you actually get out of it is just a biproduct :P
10:27 AM Tom_L: Loetmichel, does it shake the cart when it's running?
10:47 AM Loetmichel: Tom_L: only when the brakes are not on
10:54 AM Loetmichel: hmpf. re at home from work. Now i feel naked... forgot the smartphone at the company :-(
10:55 AM Loetmichel: now i have to drive to the company tomorrow and get it back so i can login to any services. dumb me. :(
10:56 AM Tom_L: you know you miss work already...
10:57 AM Loetmichel: BS. i just need that thing to login anywhere
11:05 AM Tom_L: _unreal_, https://www.youtube.com/watch?v=_atw3e0nIrg
11:06 AM rmu|w: if you put the smarts into the phone...
11:43 AM jymmmm: pcw_home: This is updated at least daily... https://www.latimes.com/projects/california-coronavirus-cases-tracking-outbreak/
11:55 AM pcw_home: It looks like California has a handle on it though the peak of hospitalizations/deaths is maybe 2/3 weeks away
11:58 AM genoober: good morning all
11:59 AM jymmmm: pcw_home: but infections as of april 1 are 1250, that curve is getting steeper daily
12:00 PM genoober: Hey Tom_L I figured out the biggest part of my problem with my mpg i think :)
12:00 PM jymmmm: pcw_home: California is now twice as many as Italy.
12:02 PM Tom_L: good news
12:04 PM Tom_L: what was the issue?
12:05 PM pcw_home: New cases are flat in CA (Italy has 4X new cases/day)
12:05 PM genoober: indeed. unfortunatly none of the fixes seem to work. i can only access the device as root. no matter how i twiddle the udev rules
12:06 PM pcw_home: CA has about 1/10 the cases (though cases are probably undercounted by a factor of 10 or more)
12:12 PM Tom_L: genoober, did you also try setting (checking) permissions on the files?
12:13 PM genoober: yup i did
12:13 PM Tom_L: sounds like you're on the right track. i bet you figure it out
12:14 PM genoober: lsusb -l reports the device crw-rw-r-- 1 root root 189, 6 Apr 3 10:11 007
12:18 PM jymmmm: pcw_home: I have no clue what you are looking at, but new cases for April 1 are: Italy 4K and USA 41K,
12:19 PM genoober: i can manually change permission and the test tool ~/linuxcnc/bin/xhc-hb04 works but any reboot or removal of device and it reverts to root access only :(
12:21 PM jymmmm: genoober: are you logged in as root?
12:23 PM jymmmm: pcw_home: And California has 90 new cases today as of 9am this morning. doens't sound too flat to me
12:24 PM Tom_L: damn, regular heatwave goin on now... 30°F
12:24 PM pcw_home: I was only talking about CA, which does seem to be flattening out
12:26 PM pcw_home: (about 1200 new cases a day for the last 3 or so days)
12:26 PM jymmmm: pcw_home: California has had 1200+ new cases daily for the last three days, so maybe that's flattening IDK
12:27 PM pcw_home: I think mainly by luck we started the social distancing early in the progression
12:28 PM pcw_home: NY had widespread dissemination before they started (Italy also)
12:29 PM jymmmm: CV can live on metal 2 to 10 hours, longer on plastics. I saw some kid gathering shopping carts bare handed. He's more likely to get infected 20x greater, touch each cart, and infect 20x more people.
12:29 PM pcw_home: Lousiana will beat CA in a couple of days
12:30 PM pcw_home: maybe just a day
12:32 PM jymmmm: pcw_home: where are you seeing LA numbers?
12:33 PM pcw_home: Local stores here have workers wiping down carts
12:33 PM pcw_home: worldometer
12:33 PM Tom_L: they were doing that here before the stay home order
12:35 PM pcw_home: 10297 cases in LA 370 deaths
12:35 PM jymmmm: Join the blue hand group, have a barrier between and everything else. Hand sanitizer isn't gonna help if you are already infected.
12:35 PM Tom_L: Lousiana or LosAngeles?
12:36 PM pcw_home: Louisiana
12:36 PM Tom_L: wonder how NY is fairing
12:37 PM pcw_home: Looks pretty bad
12:40 PM genoober: nope im not logged in as root
12:42 PM Tom_L: thar's the problem
12:42 PM jymmmm: pcw_home: https://public.flourish.studio/visualisation/1725950/embed
12:43 PM genoober: linuxcnc wont run as root
12:43 PM Tom_L: you can make your changes though then log back in
12:44 PM genoober: i have and they always revert after a reboot or unplugging the device
12:46 PM jymmmm: pcw_home: that came from https://www.nola.com/news/coronavirus/article_7cb2af1c-6414-11ea-b729-93612370dd94.html
12:47 PM genoober: i do have a terminal logged in as root when i need to make root changes then close it immediatly... LOL ever since i fat fingered a a very stupid rm -R ./* back in the 90s
12:47 PM jymmmm: genoober: why don't you just SU as needed?
12:48 PM genoober: i do then close it immediatly before i forget
01:05 PM Choscura: guys I'm hanging out on facebook.com/dadpodiot/live
01:05 PM Choscura: also have a discord up
01:05 PM Choscura: https://discord.gg/hh4Jc3 iirc
01:06 PM rmu|w: genoober: did you reboot after changing udev rules?
01:32 PM rmu|w: genoober: run "udevadm control --reload-rules && udevadm trigger" as root after changing udev rules
01:33 PM rmu|w: https://unix.stackexchange.com/questions/39370/how-to-reload-udev-rules-without-reboot
01:34 PM Papagno: Hi . i have a question. because in axis , lathe.ini, when move X axis move reverse motion respect a real machine ? On real machine the spindle is lefth position and X= 0 is centre the spindle. when move X + axis move turret UP , in machinne reference direction.
01:35 PM Papagno: In axis X plus move down and drawing the piece is revert about a real machine.
01:37 PM Tom_L: change the sign of
01:37 PM Tom_L: scale in your ini
01:54 PM genoober: yes i have done both after every udev change
01:56 PM genoober: ls -l /dev/usb is now reporting crw-rw-rw- 1 genoober users 189, 7 Apr 3 11:02 008
01:56 PM genoober: and still not working
01:56 PM Tom_L: do you have ownership of the files you're using?
01:58 PM genoober: yup i do...
01:58 PM genoober: ive also changed owner to root and the groups
01:59 PM genoober: didnt figure that would change much but i gave it a whirl
01:59 PM rmu|w: genoober: if you manually chown the usb device files, does it work then?
01:59 PM Tom_L: you probably don't want to do that
01:59 PM rmu|w: 19:19 < genoober> i can manually change permission and the test tool ~/linuxcnc/bin/xhc-hb04 works but any reboot or removal of device and it reverts to root access only :(
02:00 PM genoober: yup
02:00 PM rmu|w: so your udev rules don't work properly?
02:01 PM genoober: yeah i kinda put everything back to what seems to be the recomended concensus on the forums...
02:01 PM rmu|w: consensus on a forum may be worst of it all
02:01 PM genoober: ohh LOL
02:02 PM genoober: well the current owner is root in my udev and group is users
02:03 PM rmu|w: and nor the test tool is not working?
02:03 PM rmu|w: s/nor/now/
02:04 PM rmu|w: you could try strace
02:04 PM rmu|w: strace ~/linuxcnc/bin/xhc-hb04 and look for the failing syscall
02:05 PM genoober: the test tool only works as root unless i chmod the usb pendant then i can but it reverts on reboot/logout/or if its unplugged
02:05 PM rmu|w: if it reverts then your udev rules don't work
02:06 PM rmu|w: lsusb should show you the device id and vendor id of the pendant
02:06 PM rmu|w: check if it is the same you used in the udev rules file
02:07 PM rmu|w: and check out https://unix.stackexchange.com/questions/200194/how-to-debug-an-udev-rule-in-etc-udev-rules-d
02:07 PM rmu|w: udevadm monitor -u
02:07 PM genoober: thats kind of what i thought too but ls -l of the device shows proper owners and permissions but if i chmod and give permissions to everything it works ive recursivly changed all permissions in the top linuxcnc dir so everything should work... im just missing something and im totally lost on it
02:08 PM genoober: i dont have strace
02:08 PM rmu|w: 21:07 < genoober> i dont have strace
02:08 PM rmu|w: not possible ;)
02:08 PM rmu|w: apt-get install strace
02:09 PM genoober: already started
02:09 PM genoober: ls -l /dev/usb is now reporting crw-rw-rw- 1 genoober users 189, 7 Apr 3 11:02 008
02:09 PM rmu|w: /dev/usb is a directory
02:10 PM rmu|w: at least here it is
02:11 PM genoober: ok ill typoe it all out
02:11 PM genoober: ls -l /dev/bus/usb/001/008
02:11 PM genoober: crw-rw-rw- 1 root users 189, 7 Apr 3 11:57 /dev/bus/usb/001/008
02:14 PM genoober: wow strace spit out a ton of gibberish....
02:14 PM rmu|w: do you have both /dev/usb is now reporting crw-rw-rw- 1 genoober users 189, 7 and /dev/bus/usb/001/008 crw-rw-rw- 1 root users 189, 7?
02:14 PM rmu|w: sorry
02:15 PM rmu|w: both /dev/usb crw-rw-rw- 1 genoober users 189, 7 and /dev/bus/usb/001/008 crw-rw-rw- 1 root users 189, 7?
02:15 PM genoober: lol no root and users is current i was fiddling with different users and groups earlier
02:17 PM rmu|w: i would * check vendor and device id in udev rules file and compare with output of lsusb (that is lsusb, not ls /dev/bus/usb... or something). * udevadm monitor * unplug/replug
02:18 PM genoober: vendor and device id's from lsusb are in line with what is in the udev rules
02:20 PM genoober: https://pastebin.com/APMKZquS
02:23 PM genoober: https://pastebin.com/4PmX2Fug
02:24 PM genoober: second one is from strace
02:25 PM genoober: first is devadm monitor
02:27 PM rmu|w: can you show your rules file?
02:32 PM rmu|w: hmm. strace output doesn't really help. "xhc-hb04: waiting for XHC-HB04 device"
02:41 PM genoober: ATTRS{idVendor}=="10ce", ATTRS{idProduct}=="eb93", MODE="666", OWNER="root", GROUP="users"
02:47 PM rmu|w: genoober: rules file on my system look like this
02:47 PM rmu|w: SUBSYSTEMS=="usb", ATTRS{idVendor}=="0483", ATTRS{idProduct}=="374b", \
02:47 PM rmu|w: MODE:="0666", \
02:47 PM rmu|w: SYMLINK+="stlinkv2-1_%n"
02:47 PM rmu|w: (ubuntu 18.04)
02:48 PM rmu|w: (MODE:="0666", colon)
02:50 PM rmu|w: genoober: in theory, it is possible that another rules file with higher priority overrides the settings, so you could also try renaming it into 99-whatever.rules
02:50 PM genoober: if i use SUBSYSTEM=="usb"..... it doesnt matter what i do in the rest of the rules file all the permissions are stuck root root
02:51 PM rmu|w: what a mess
02:51 PM rmu|w: those rule files are a mess
02:52 PM genoober: indeed
02:52 PM genoober: Failing cmd:
02:53 PM genoober: loadusr -W xhc-hb04 -s 1 -I /home/genoober/linuxcnc-dev-tst/lib/hallib/xhc-hb04-layout2.cfg -H
02:53 PM genoober: Are permissions correct?
02:54 PM rmu|w: ?
02:54 PM genoober: error message from trying to run linuxcnc
02:55 PM genoober: somewhere my permissions are not right what permissions does linuxcnc want LOL
02:56 PM rmu|w: did you try adding the colon to mode?
02:56 PM rmu|w: MODE:="0666"
02:56 PM genoober: will do
02:58 PM genoober: same error
03:00 PM rmu|w: https://askubuntu.com/questions/15570/configure-udev-to-change-permissions-on-usb-hid-device says to use SUBSYSTEM=="usb", ATTRS{idVendor}=="04b8", ATTR{idProduct}=="014a", MODE="0666" on ubuntu 18.04 (proper device/vendor id)
03:00 PM rmu|w: seems there is a difference between ATTRS and ATTR
03:01 PM rmu|w: https://mirrors.edge.kernel.org/pub/linux/utils/kernel/hotplug/udev/udev.html
03:02 PM rmu|w: udevadm info -q all -a /path/to/device should show you attributes of the device
03:06 PM rmu|w: hmm. colon shouldn't make a difference
03:06 PM rmu|w: something else is going on
03:07 PM genoober: https://pastebin.com/6LK0MXjA
03:08 PM genoober: should DRIVER=="usb" be usbhid?
03:11 PM unterhaus_: you know the conversation isn't going well when someone asks for a free body diagram
03:11 PM rmu|w: SUBSYSTEMS="usb", ATTR{idVendor}=="10ce", ATTR{idProduct}=="eb93", MODE:="0666", OWNER:="root", GROUP:="users"
03:11 PM rmu|w: should work
03:13 PM XXCoder: lol https://youtu.be/aFRG7nI5ZVY
03:13 PM XXCoder: dial test indictor clock
03:13 PM XXCoder: big one on wall
03:14 PM genoober: nice
03:15 PM genoober: changed my udev to what "should work" nogo
03:16 PM skunkworks: http://electronicsam.com/images/greenmachine/IMG_20200403_151042.jpg
03:17 PM rmu|w: ;-)
03:17 PM rmu|w: genoober: i'm out of ideas. you could look at /var/log/* for errors
03:18 PM Tom_L: :)
03:19 PM genoober: any idea which log i should be looking for?
03:19 PM rmu|w: syslog
03:22 PM genoober: Apr 2 23:53:39 debian systemd-udevd[313]: Invalid rule /etc/udev/rules.d/99-xhc-hb04.rules:2: error parsing ATTR attribute
03:23 PM rmu|w: is your clock off?
03:24 PM genoober: ohh crap thats yesterdays log
03:24 PM genoober: LOL
03:25 PM XXCoder: skunkworks: lol thats cool. if you could make matching twisted hex rod
03:25 PM rmu|w: maybe this helps
03:25 PM genoober: 13:24:49: The file "/var/log/syslog" does not look like a text file or the file encoding is not supported.
03:25 PM rmu|w: https://wiki.ubuntu.com/DebuggingUdev
03:26 PM rmu|w: sudo pkill udevd
03:26 PM pcw_home: skunkworks: looks like you have invented a new security fastener
03:26 PM rmu|w: sudo udevd --debug-trace --verbose --suppress-syslog
03:30 PM genoober: udevd not found
03:31 PM rmu|w: hmm. that was in pre-systemd-times...
03:34 PM rmu|w: /lib/systemd/systemd-udevd -D
03:38 PM genoober: there is a rules file in /lib/udev/rules.d/99-xhc.rules that conflicts with the rules in /etc/udev/rules.d/99-xhc.rules
03:41 PM genoober: made them the same
03:42 PM genoober: the only difference was productid
03:46 PM genoober: some progress... i can run test tool as non root
03:47 PM genoober: let me reboot and see if everything works post reboot
03:56 PM genoober: i can still run the test tool post reboot wahoooo.... next hurdle
04:37 PM Deejay: gn8
05:41 PM _unreal_: pcw_home, you live?
05:41 PM _unreal_: pcw_mesa, you live?
05:42 PM Tom_L: _unreal_, you see that guy's cnc build i posted for you?
05:46 PM pcw_mesa: I'm sort of here but a bit busy ATM
05:50 PM _unreal_: Tom_L, not sure?
05:50 PM Tom_L: https://www.youtube.com/watch?v=_atw3e0nIrg
05:50 PM _unreal_: pcw_mesa, hi um just wondering is there an off line method for driving the mesa's?
05:51 PM Tom_L: you can plug the ethernet in and power it from the usb if you want
05:52 PM Tom_L: or some 5v source
05:52 PM _unreal_: Tom_L, interesting
05:52 PM Tom_L: might give you an idea what to expect from yours
05:53 PM Tom_L: i skipped thru it
06:04 PM XXCoder: went to office depot, waited in line for 40 minutes
06:04 PM XXCoder: bought pfte tape. couldnt find good foldable table
06:05 PM XXCoder: went to lowes, entered fast but they have fewer people
06:05 PM XXCoder: i now have hobby table yay
06:14 PM genoober: silly question any way to test the xhc-04b driver besides running the binary at the command line? i cant figure out why its not loading
06:14 PM jthornton: we just got a state wide stay at home order, not sure what that means exactly
06:14 PM XXCoder: stay at home basically.
06:14 PM jdh: means stay at hoem
06:14 PM XXCoder: many companies will close unless essental
06:14 PM jthornton: I do that normally
06:14 PM jdh: me too
06:14 PM XXCoder: it shouldnt affect you too much.
06:15 PM jdh: but, I will go biking tomorrow
06:15 PM andypugh: genoober: Well, still command-line, but more linuxcnc-ish:
06:15 PM andypugh: halrun
06:15 PM andypugh: loadusr xhc_…..
06:15 PM XXCoder: jdh: thats also allowed in stay at home order
06:15 PM genoober: so something like halrun loadusr -W xhc-hb04 -s 1 -I /home/genoober/linuxcnc-dev-tst/lib/hallib/xhc-hb04-layout2.cfg -H
06:16 PM andypugh: You can then use halmeter / show pin etc to see what is going on.
06:16 PM XXCoder: one a day, stay away from people
06:16 PM jdh: I will be riding with someone else, but not close
06:16 PM andypugh: genoober: Aye, something like that.
06:16 PM andypugh: Start simple and work up :-)
06:17 PM andypugh: I can imagine the tcl not working without the rest of LinuxCNC to connect to.
06:19 PM genoober: hmmmm ok
06:19 PM jthornton: the biggest impact to normal is the limit of 25% of the buildings authorized fire or building code occupancy for buildings < 10k sq feet and 10% for larger buildings
06:20 PM genoober: well damn the device never seems to become ready
06:20 PM jthornton: I guess I better get my 2x4's tomorrow... need two straight ones
06:20 PM XXCoder: bring something to read
06:20 PM XXCoder: it took me 40 minutes to enter
06:20 PM jthornton: I go early in the morning before the lazy people get moving
06:21 PM genoober: order a palatek 25d4 the crate comes with 6 very strait 2x4s every time
06:22 PM * jthornton goes to prep for the shrimp and grits
07:18 PM XXCoder: lol new MPCNC https://www.thingiverse.com/thing:3813481
07:18 PM XXCoder: honestly better than pipes one.
07:36 PM _unreal_: genoober, I'll be ding the same thing I have one of those pendents as well
07:47 PM genoober: imstill stuck in permission hell and trying to get the driver to load
07:50 PM adamj12b: Does anyone have any experience dealign with the following error? hm2/hm2_5i25.0: Smart Serial port 0: DoIt not cleared from previous servo thread.
07:52 PM adamj12b: My system is running a servo_period of 400000 for a 2.5khz servo loop and latency tests out about 30uS.
07:55 PM _unreal_: genoober, https://forum.linuxcnc.org/38-general-linuxcnc-questions/35913-xhc-whb04b-6-linuxcnc-not-working?start=30
07:58 PM pcw_home: adamj12b: try looking at the various tmaxes
07:59 PM adamj12b: I have them, they seem ok but need an experts opinion. I was just about to post on the forum.
07:59 PM adamj12b: pcw_home Here are the Tmax;s https://photos.app.goo.gl/H1KoWkqkeLfKj5b49
08:00 PM adamj12b: clock speed of the system is 1.99GHz as reported by latency-histogram
08:03 PM pcw_home: I just see a blank page there
08:04 PM pcw_home: what do you get for servo-thread.tmax?
08:05 PM genoober: _unreal_ im already in dialout group got the udev rules done but still no luck
08:05 PM adamj12b: servo-thread.tmax is 743256
08:05 PM adamj12b: hm2 read is 651744 and write is 48816
08:07 PM adamj12b: New link https://www.dropbox.com/s/wb8ve2ximcr17q6/IMG_20200403_203433.jpg?dl=0
08:08 PM pcw_home: well 743256 is 371 usec so you are out of time with a 400 usec servo thread
08:09 PM pcw_home: not sure why read is so slow, the average value might be instructive
08:09 PM adamj12b: andy was not sure the other night, but would increasing the 6i25 to 66mhz help anything?
08:09 PM genoober: i seem to have the correct permissions according to ls -l /dev/bus/001/010 being readable and writable by "users" and owner is "root"
08:11 PM pcw_home: adamj12b: No, its not a PCI access issue
08:11 PM pcw_home: more likely cache related
08:12 PM pcw_home: Unfortunately the latency test does not use much memory or any hardware I/O so is pretty optimistic
08:13 PM pcw_home: you might try isolcpus 2,3
08:13 PM adamj12b: I am already isolcpus 3
08:14 PM pcw_home: depending on cache structure, isolcpus 2,3 might be better
08:15 PM adamj12b: OK, I can try that. If the read time drops I would know if it is successful?
08:16 PM pcw_home: yeah that and all the other tmax's
08:22 PM adamj12b: well they are lower now but the machine has only been up for a few minutes. It was up for a few hours while I was at dinner and came back to that error.
08:23 PM adamj12b: servo thread is at 258uS currently
08:26 PM pcw_home: is the average read time relatively short?
08:29 PM adamj12b: read time is now 236uS
08:29 PM adamj12b: write is 7.8 uS...
08:30 PM pcw_home: but whats the average read time?
08:31 PM adamj12b: Where would I find that? in halshow?
08:31 PM pcw_home: good way to check this is plot the time in halscope
08:32 PM pcw_home: while you move windows around, copy files, watch videos etc
08:32 PM adamj12b: ok I will do that. In halshow read time shows roughly 87000
08:32 PM adamj12b: I will plot it though.
08:33 PM pcw_home: that will give you an idea of what affects the latency
08:36 PM adamj12b: Its averaging 86k with spikes at 200ish and occasionally 400ish
08:36 PM skunkworks: pcw_home: http://electronicsam.com/images/greenmachine/IMG_20200403_155600.jpg
08:36 PM _unreal_: pcw_home, Soon as my freind gets my @#$%@#$$3 box built
08:36 PM _unreal_: I can start mounting all of the controller parts
08:36 PM skunkworks: 60Deg in .5"
08:36 PM _unreal_: the motor controller parts
08:37 PM _unreal_: skunkworks, interesting but aside from art is that useful?
08:38 PM skunkworks: not sure. Andy says a pasta maker.. ;)
08:38 PM _unreal_: skunkworks, maybe
08:38 PM _unreal_: skunkworks, if you could setup a block in the middle you could make hex noddles :)
08:38 PM skunkworks: I want to make the 'screw' taht fits that...
08:38 PM skunkworks: that
08:40 PM _unreal_: royal caribbean stocks down to $24 from $130
08:43 PM pcw_home: skunkworks: That's really neat, if you make a matching hex key you could have a hex one way security screw
08:44 PM _unreal_: pcw_home, when I get things setup
08:44 PM _unreal_: err let me re-write that
08:46 PM genoober: hal loadusr xhc-hb04 stalls waiting for device. the device seems to run when i use the xhc-hb04 binary... permissions seem correct as well as device ids in the udev rules... ive spent nearly 2 weeks in quarentine working on this and since last haloween off and on trying to get this going... does anybody live near portlan oregon? im thinking about in person help... im just about willing to pay cash moneys if someone can get this fixed
08:46 PM _unreal_: pcw_home, My goals, get all the basics working, brushed servos, quadrature encoders. quadrature movement control, VFD 1.5kw spindle, position tracking in lcnc of the servos if moved when motors are powered off
08:47 PM pcw_home: Dont your drives take step/dir?
08:47 PM _unreal_: genoober, I may get to the point as well ;)
08:47 PM _unreal_: pcw_home, no quadrature we talked about it the other day... the mesa cards support it
08:47 PM _unreal_: http://ve7it.cowlug.org/dspic-servo.html
08:48 PM pcw_home: I tend to lump step/dir/CW/CCW/Quadrature-out together as flavors of step/dir
08:48 PM _unreal_: ahh
08:48 PM _unreal_: pcw_home, is there a way to do offline motion control? with the mesa
08:48 PM _unreal_: ?
08:49 PM skunkworks: pcw_home: interesting - that would be neat :) bet it would have to be abit finer maybe..
08:49 PM pcw_home: _unreal_ you can do things like set step rates by hand
08:50 PM _unreal_: hum
08:50 PM _unreal_: not sure how useful it would be though?
08:50 PM _unreal_: and I have to still ensure a lock out
08:50 PM pcw_home: You can do it very awkwardly by hand or more easily with halcmd
08:53 PM adamj12b: pcw_home I will continue to dig into this. Just so Im going to correct direction, You said move windows, copy files. It seems to me that these 2 tests would show if its video releated(which I think it is) or maybe disk io related. There seems to be a way to disable the hardware graphics accelerator but im unsure how to yet.The latency is back to
08:53 PM adamj12b: 700+ on the servo thread after idling for the past 20 minutes.
08:55 PM pcw_home: adamj12b that (video causing latency spikes) has been my impression especially on slower machines
08:55 PM adamj12b: I dont need 60+fps. I need a faster servo loop...LOL
08:56 PM adamj12b: my drives run torque mode so I want faster then 1khz.
08:56 PM adamj12b: is there a way to reset tmax without restarting?
08:57 PM skunkworks: use the latency-histogram
08:57 PM pcw_home: skunkworks: what would you need to do the motion to make the twisted hex key (say from a rod) on the mill?
08:57 PM adamj12b: latency-histogram does not show any issues...its beautiful
08:58 PM pcw_home: adamj12b, yes just setp them to 0
08:59 PM adamj12b: skunkworks https://drive.google.com/file/d/1nBBHsOThisCVE-ic_Cd6aEBB5vfEGArM/view?usp=sharing
08:59 PM CaptHindsight: RTAI or preempt_rt and what CPU is this? adamj12b
08:59 PM CaptHindsight: haven't read the backlog
09:00 PM adamj12b: J1900 @ 1.99GHz Preempt 4.19.106-rt44-lcnc
09:00 PM adamj12b: that link I just shared is a histogram at 4khz.
09:01 PM CaptHindsight: looks very nice!
09:01 PM adamj12b: Earlier it was discussed that it doesn't load the system enough. Which I have always wondered
09:01 PM adamj12b: CaptHindsight I thought so too, But actually running im having errors on the mesa SSerial timing.
09:02 PM adamj12b: Ive slowed down to 2.5khz but dont have an up to date histogram.
09:03 PM adamj12b: my tmax on the servo thread creeps up to ~700000 with a 1.99ghz clock thats 351uS and im set up right now for a 400uS thread
09:07 PM pcw_home: too bad you cant do what the Ethernet driver does and just give up and wait for the next cycle if the read time gets too long
09:09 PM adamj12b: So is there any major issue to one cycle not finishing?
09:10 PM pcw_home: in general no, (just lower control loop bandwidth that cycle) but there are some specific cases where lost write cycles can be an issue
09:12 PM pcw_home: adamj12b: you can run the histogram on read times for example
09:13 PM adamj12b: So this vs high loop frequency for torque mode drives, would loop speed be more important then an occasional error?
09:13 PM pcw_home: (that is the latency histogram has a hal equivalent so you can get a histogram of any hal pin)
09:13 PM adamj12b: How would I do that? I would love to see it because its hard to see on the plot
09:14 PM adamj12b: I think I found it.
09:14 PM adamj12b: hal-histogram then.....
09:14 PM pcw_home: yeah I forgot the mane
09:14 PM adamj12b: servo.thread.time....thats obvious
09:14 PM pcw_home: name
09:15 PM adamj12b: hal-histogram hm2_5i25.0.read.tmax
09:15 PM adamj12b: OK! i just need these nudges in the right direction.
09:16 PM adamj12b: time, not tmax...
09:18 PM adamj12b: What binsize do you recomend?
09:19 PM pcw_home: enough that the max values dont get squashed to the ends
09:20 PM CaptHindsight: just found these http://yurtaev.com/
09:20 PM CaptHindsight: Network Servo Interface Adapters for LinuxCNC
09:20 PM CaptHindsight: A family of PCI interface adapters compatible with Mitsubishi network servo drives.
09:21 PM adamj12b: yea I think thats whats happening now. there is just one bar...
09:21 PM adamj12b: bin size is 100
09:22 PM CaptHindsight: https://www.youtube.com/watch?v=LTfe2ljmRpU pick-n-place using them
09:22 PM CaptHindsight: Linuxcnc and opencv
09:23 PM XXCoder: CaptHindsight: cool videp
09:23 PM adamj12b: https://photos.app.goo.gl/aAGJ5gS4a8bLDmo6A
09:23 PM CaptHindsight: https://github.com/yur7aev/linuxpnp A set of scripts for LinuxCNC doing SM components pick and place
09:25 PM pcw_home: hal-histogram --binsize 10000 --nbins 200 servo-thread.time
09:25 PM XXCoder: CaptHindsight: was it soldered in? or just put there?
09:28 PM adamj12b: ok I have 10000 but let me change nbins
09:31 PM pcw_home: yeah the default 50 bins is a bit crude
09:31 PM adamj12b: So, this is going to show me the frequency of the latency isuess?
09:32 PM pcw_home: great tool from dgarrett
09:32 PM adamj12b: I dont think I need 200, but I will let this run awhile to get a summary and then shrink it up to get a better image.
09:35 PM adamj12b: For sure. This is an awesome tool. I would think with is and a special sim running, you may be able to get a better idea of actual latency vs the latency test.
09:38 PM CaptHindsight: XXCoder: after the pick-n-place the board is sent through a reflow oven where the solder paste melts and holds the parts in place
09:41 PM CaptHindsight: https://www.youtube.com/watch?v=1G7HH3icagE the process up close
09:42 PM _unreal_: Ok just got my power adapter finished
09:42 PM XXCoder: ahh
09:43 PM _unreal_: so I can now power my 12v monitor from the computer
09:43 PM CaptHindsight: https://www.youtube.com/watch?v=saOHrw4ezGw Soldering overview: reflow, wave and hand soldering
09:43 PM _unreal_: I have a very expensive marine lcd monitor :)
09:43 PM _unreal_: got to love the shit they throw away
09:44 PM _unreal_: the monitor I have was $3k new now they sell for around $500 on ebay
09:44 PM _unreal_: I'm going to use it for my lcnc
09:44 PM _unreal_: wish it was touch screen though :(
09:53 PM adamj12b: CaptHindsight Did you see any pricing about those servo interfaces?
09:57 PM adamj12b: General question, Would creating a slower thread from my Servo_thread and run tasks in it that do not need to be run at servo thread speed be more
09:57 PM adamj12b: efficient
10:00 PM pcw_home: adamj12b My guess would be no since the would be run at the end of the servo thread every N cycles
10:02 PM adamj12b: Just a thought as I have things like calssic ladder for drawbar, operator buttons and oil pump functions running in the servo thread when they could easily run at 250-500hz with no ill effects.
10:02 PM pcw_home: Yeah too bad they cannot be distributed
10:03 PM pcw_home: (do a little every servo thread)
10:03 PM adamj12b: yes for sure.
10:03 PM adamj12b: sounds like something to talk to andy about. :)
10:05 PM adamj12b: the histograms have been running for about 2000 seconds.there are some small blips around the 400-600 mark
10:06 PM adamj12b: There is 3 clear intervals above 400k clock cycles
10:06 PM adamj12b: https://photos.app.goo.gl/9rLPhzXy6D2BFmFS8
10:07 PM pcw_home: I wonder if a do (almost) nothing base thread would help
10:07 PM adamj12b: I will try, what do you suggest?
10:10 PM pcw_home: just a hail Mary pass, institute a base thread at say 10 KHz and link it to something minimal like a software PWMgen
10:12 PM Tom_L: https://www.ebay.com/itm/2-3Axis-DRO-Digital-Readout-for-Milling-Lathe-Machine-Linear-Glass-Scales/392619866552?epid=13022805422&hash=item5b69f7b9b8:m:m1msCDC61Ngw2pcioUvYfBQ
10:12 PM Tom_L: what's the catch?
10:13 PM adamj12b: ok. I have to look into how to do that. I have only ever used your products. :P
10:21 PM adamj12b: pcw_home OK I have added a base thread with a period of 100000 and have restarted the histograms
10:29 PM adamj12b: crap, I forgot to add something to the base thread. Is that going to make a difference? The histogram already looks like crap...
10:30 PM XXCoder: oh did you disable power saving stuff?
10:30 PM XXCoder: go to bios and disable em all
10:31 PM adamj12b: O yes. My latency is great, im fighting a strange latency issue with SSerial of the mesa
10:31 PM XXCoder: Tom_L: weird
10:31 PM XXCoder: so cheap too
10:32 PM adamj12b: XXCoder Here is the latency histogram I shared earlier.
10:32 PM adamj12b: https://drive.google.com/file/d/1nBBHsOThisCVE-ic_Cd6aEBB5vfEGArM/view?usp=sharing
10:33 PM adamj12b: and the histogram of the servo thread and 5i25 read time
10:33 PM adamj12b: https://photos.app.goo.gl/9rLPhzXy6D2BFmFS8
10:33 PM XXCoder: im not exactly expert but disabling power saving stuff is common way to improve some stuff
10:33 PM XXCoder: it did for me heh
10:34 PM adamj12b: yea It helped me. isolcpus=3 made the biggest difference.
10:34 PM XXCoder: nice'
10:34 PM XXCoder: Tom_L: ah its different prices for lengths etc
10:34 PM XXCoder: 38 usd is for 2" one lol
10:35 PM Tom_L: yeah i'm not sure what 3 axis with dro would be
10:39 PM XXCoder: prices seem not bad
10:39 PM XXCoder: would be cool to have one for drill press
10:39 PM XXCoder: for say pattern holes
10:40 PM XXCoder: combined with xy table for drill oress
10:42 PM Tom_L: you still workin or on leave?
10:42 PM XXCoder: working
10:43 PM XXCoder: boss told us that some of us will be cut hours
10:43 PM XXCoder: almost 100% i will be
10:43 PM Tom_L: yeah they're cutting stuff here too
10:43 PM XXCoder: 2 machines and all lathes are "immortal" though. infinite parts
10:43 PM XXCoder: but if they fix robo I probably will keep working
10:43 PM XXCoder: 700 parts 15 a day
10:43 PM Tom_L: you do alot of lathe work there?
10:44 PM XXCoder: oh dunno is 20 lathes normal for shops? ;)
10:44 PM Tom_L: cnc?
10:44 PM XXCoder: yeah
10:44 PM Tom_L: my bud had one
10:44 PM XXCoder: well and one very large manual
10:44 PM Tom_L: the shop my kids are at don't have any
10:49 PM XXCoder: from what i understand one or maybe 2 lathes is normal
10:49 PM XXCoder: but we run many many lathe products
10:51 PM pcw_home: adamj12b: the latency issue is not really related to sserial except that linuxcnc needs some small amount of time from the last write of the servo thread cycle to the read on the next cycle
10:52 PM pcw_home: so if the whole servo thread period is used up you lose that required ~50 usec time
10:54 PM adamj12b: pcw_home I added the pwmgen and its running now. It appears to have made some difference but its still early. I was getting a realtime error when it was ust the base thread added, but that hasn't happened since I added the pwm.
10:54 PM adamj12b: What does it mean if adding the base thread at 10khz helps?
10:57 PM pcw_home: I think it means that the latency is caching related
10:58 PM adamj12b: caching of the processor or memory or??
10:59 PM pcw_home: keeping the critical bits of LinuxCNC in the memory cache
11:03 PM adamj12b: um... ok that doesnt sound fixable. lol
11:08 PM pcw_home: Depends on the cache size and LRU algorithm
11:08 PM pcw_home: CPUs with bigger caches seem to do better with LinuxCNC latency
11:11 PM adamj12b: this has a 2MB L2 cache, I cant seem to find a L3 cache listed...
11:12 PM adamj12b: The histograms seem to be updating much slower then they were before the base thread. but I have not gotten any realtime errors for sserial or servo thread.
11:14 PM pcw_home: You can see the caching issue by running different speed servo threads, slower threads will have higher peak latencies
11:15 PM adamj12b: I did see that in my bench testing.
11:15 PM adamj12b: I got better numbers at 4khz then I did at 1khz
11:15 PM pcw_home: yep
11:16 PM adamj12b: So is there a fix for that other then different motherboard?
11:16 PM adamj12b: well processor, but this one is built into the mobo...
11:17 PM pcw_home: I think its mainly speed and cache size related
11:19 PM pcw_home: you could also try isolcpus=1,2,3 (thats pretty much required on the RPI4)
11:19 PM adamj12b: I was thinking about trying that.
11:19 PM pcw_home: or be satisfied with 2 KHz or so
11:20 PM adamj12b: 2khz > 1khz
11:20 PM adamj12b: so...
11:20 PM pcw_home: curious if the base thread actually helps
11:21 PM adamj12b: would switching the 6i25 for the 7i92 thats on my desk help me at all? I know you mentioned something about ethernet earlier.
11:21 PM pcw_home: did you wire both PWM functions to the base thread?
11:21 PM adamj12b: Honestly, I think its bogging the system and the histograms are missing updates.
11:22 PM adamj12b: I wired make pulses to base and update to servo
11:22 PM pcw_home: 5 KHz might be enough for the base thread
11:22 PM pcw_home: I would wire both to the base thread (so the FP stuff is pulled it)
11:22 PM adamj12b: ok i can change that.
11:24 PM pcw_home: you can see how much time its wasting with its .time parameter
11:25 PM adamj12b: pwmgen.update needs FP so it has to be on servo
11:25 PM adamj12b: I changed to 5khz
11:26 PM adamj12b: I actually dont have update wired. only make pulses
11:26 PM adamj12b: do you have to assign the linuxcnc processes to the isolated processors?
11:33 PM pcw_home: supposedly isolcpus does that
11:35 PM pcw_home: My understanding is that linuxCNC threads are bound to the isolated CPUs but I'm not really sure of this
11:37 PM adamj12b: ok I will add it to the list of research.
11:38 PM adamj12b: It looks as though isolcpus=1,2,3 caused more harm then good on the histogram.
11:39 PM adamj12b: yes, it definitely made things worse.
11:39 PM pcw_home: yeah it really depends on how the cache is structured/shared between processors
11:40 PM adamj12b: so the best I can tell, I can try and knock the intergrated GPU down a few pegs somehow or slow the loop. seems like my options at this point.
11:41 PM CaptHindsight: Linuxcnc looks at the cpu core count and runs on the highest core
11:42 PM adamj12b: ok so isolcpus=3 is all thats really needed.
11:42 PM CaptHindsight: that is why running isolcpus=(highest core #) most often works best
11:43 PM adamj12b: makes sense.
11:43 PM CaptHindsight: you don't have to bother with isolcpus= if you use current RTAI
11:43 PM adamj12b: yea everything was fantastic on RTAI till I tried the 6i25 which turns out 64bit rtai + 6i25 causes a hard crash.
11:44 PM CaptHindsight: with some cache designs isolcpus = 3,2,1 has better numbers
11:44 PM adamj12b: really? flip it around....ok
11:44 PM CaptHindsight: this only leaves core 0 for non-real time
11:44 PM CaptHindsight: and realtime on core 3
11:45 PM CaptHindsight: and nothing runs on core 1 and 2
11:45 PM CaptHindsight: this is for a 4 core cpu
11:45 PM adamj12b: any suggestion on the integrated graphics? I dont need 60fps on 4x glx gears....
11:46 PM adamj12b: Ive heard you can disable hardware acceleration but I cant find any good info how to.
11:46 PM CaptHindsight: why would you want to disable it?
11:47 PM CaptHindsight: the Intel and the nvidia drivers can cause poor latency
11:48 PM CaptHindsight: they can be left out of the kernel when compiled
11:48 PM adamj12b: you saw the plot earlier. it tests good there but peter thinks its a cache issue. no point in wasting resources just to have a high fps graphics.
11:49 PM adamj12b: Is there any way to disable it for testing without recompiling the kernel?
11:49 PM _unreal_: yawn
11:51 PM _unreal_: adamj12b, are you running proprietary drivers for your gpu or mesa graphics drivers?
11:51 PM _unreal_: I dont see why you couldnt change your boot option to boot in svga compatible mode?
11:52 PM XXCoder: sure
11:52 PM XXCoder: hexediting
11:52 PM XXCoder: ;)
11:52 PM _unreal_: to disable 3d gpu
11:52 PM adamj12b: _unreal_ Honestly dont know. I am using a mint 19.2 image that BeagleBrainz compiled on the forums
11:52 PM CaptHindsight: adamj12b: you blacklist the Intel kernel module and use software rendering
11:53 PM _unreal_: adamj12b, omg... do you know who to see what drivers are being loaded?
11:53 PM CaptHindsight: adamj12b: check the instructions for the distro you are using for how to blacklist modules
11:54 PM adamj12b: CaptHindsight I will do that. Blacklisting modules is the nudge I needed.
11:54 PM CaptHindsight: adamj12b: if you do it wrong it will just use the Intel kernel module
11:54 PM _unreal_: lsmod
11:55 PM _unreal_: will list all of the modules that are being loaded
11:55 PM adamj12b: _unreal_ I assume you could find it using lspci but not positive on that
11:55 PM CaptHindsight: https://forums.linuxmint.com/viewtopic.php?t=85344 might be of interest
11:55 PM _unreal_: ? adamj12b lspci list's PCI devices
11:56 PM _unreal_: lsmod list's your modules DRIVERS that are being loaded
11:56 PM CaptHindsight: adamj12b: you're probably thinking of modprobe
11:56 PM _unreal_: lsmod |grep video
11:56 PM adamj12b: i915 is listed for video
11:57 PM _unreal_: thats intel
11:57 PM _unreal_: i915 here ;)
11:57 PM adamj12b: i know theres some kernel arguments to modify that. I came across them the other day
11:57 PM _unreal_: this laptop is a p6100 2ghz
11:58 PM adamj12b: wouldn't happen to know how to disable the 3d gpu would you? :P
11:58 PM _unreal_: adamj12b, are you running mesa?
11:58 PM adamj12b: Yes
11:58 PM adamj12b: 6i25 + 7i77 + other SSerial
11:58 PM _unreal_: then you need to disable mesa AND MAKE SURE YOU BACKUP before making changes
11:58 PM _unreal_: NOOOOO
11:58 PM adamj12b: srry
11:59 PM adamj12b: its 1am...
11:59 PM _unreal_: MESA gpu driers
11:59 PM adamj12b: that i dont know
11:59 PM _unreal_: Thats why I asked
11:59 PM _unreal_: are you running proprietary gpu drivers or the open source mesa?
11:59 PM _unreal_: you need to figure that out. then you can work on shutting down GPU