#linuxcnc Logs

Feb 25 2025

01:08 AM Deejay: moin
01:23 AM memleak: hey hey!
01:23 AM Deejay: hey :)
02:40 AM lcnc-relay: <oddname_skane@> top of the mornin to ya
04:47 AM Tom_L: morning
05:23 AM xxcoder: good very early morning
05:35 AM lcnc-relay: <oddname_skane@> how do I launch linuxcnc with debugger?
05:35 AM lcnc-relay: <oddname_skane@> im pretty certain i have editer the launch link before
05:35 AM lcnc-relay: <oddname_skane@> * edited
06:07 AM lcnc-relay: <t4kuy4@> Question, the tool Diameter compensation, how does that work? Do i manually have to set g41/42 or is that something thats done in cam?
07:04 AM jpa-: t4kuy4@: for mills, you usually do tool diameter setting in CAM completely and G41/G42 is not used
07:05 AM jpa-: in production settings G41/G42 is sometimes used with tool table, so that small differences between replacement tools can be compensated for without regenerating CAM files
08:05 AM lcnc-relay: <oddname_skane@> figured out how to edit the launcher 🙂 thanks anyway
08:05 AM lcnc-relay: <oddname_skane@> "CRITICAL - no preference file found, enable preferences in screenoptions widget"
08:05 AM lcnc-relay: what can I do with this error? 🙂
08:16 AM lcnc-relay: <oddname_skane@> qt-related
08:18 AM lcnc-relay: <oddname_skane@> hmmm do you have the res.qrc file btw?
08:18 AM lcnc-relay: <oddname_skane@> oh wrong tab 😄
08:25 AM lcnc-relay: <oddname_skane@> better question can I make so that my build of linuxcnc is "updated" to buildbot version?
08:28 AM lcnc-relay: <zincboy_ca_on@> If this is for an actual machine, you really don't want to do that. For a test system, the way I do it is to clone the git repository and do a local build and run-in-place.
08:33 AM lcnc-relay: <oddname_skane@> why wouldnt I wanna do it?
08:34 AM lcnc-relay: <skunkworks8841@> because there could be bugs.. You update and something doesn't work.
08:38 AM lcnc-relay: <oddname_skane@> trying to setup a different config but the config errors out regardingh QT, and its maker says it need the buildbot version
08:38 AM lcnc-relay: <oddname_skane@> https://paste.debian.net/1355229/ alot of qt-errors 🙂
08:39 AM lcnc-relay: <oddname_skane@> https://paste.debian.net/1355334/ plus this when trying to push a button in the added interface
08:42 AM lcnc-relay: <oddname_skane@> not really sure why it would matter though
09:41 AM Rab: big_kevin420, I know people in the Bay involved with Sudoroom, but I'm in Texas. No idea who checks their email, if anybody. I can get you in touch with an actual person if you want.
10:07 AM lcnc-relay: <voiditswarranty@> Hey, I have a weird thing going on on one axis on my surface grinder. The head up/down is run through quite a large gear reduction, so the scale factors for that axis are quite high. The weird thing is I can only command as little as 0.0004" movement. any movement smaller than that and the DRO position changes, but it does not actually send any movement commands to the motor (checked with an oscilloscope). Any ideas where...
10:07 AM lcnc-relay: ... to start looking at that? Using riocore with a traditional step/dir stepper motor, not sure if that affects this. I was thinking MIN_FERROR but playing with that value didn't change the behavior.
10:07 AM lcnc-relay: [AXIS_Y]
10:07 AM lcnc-relay: MAX_VELOCITY = 0.4
10:07 AM lcnc-relay: ... long message truncated: https://jauriarts.org/_heisenbridge/media/jauriarts.org/ouaWrrrwsioxrTeBLUXgxYpg/7qecjBMlBrU (28 lines)
10:22 AM lcnc-relay: <zincboy_ca_on@> Deadband?
10:28 AM lcnc-relay: <voiditswarranty@> moving it smaller didn't help any, going smaller than that just made hunting happen
10:29 AM lcnc-relay: <voiditswarranty@> I wonder if I found a floating point rounding error
10:39 AM lcnc-relay: <voiditswarranty@> huh, I think I got my scale math wrong for one thing. 2mm per rev leadscrew, 10:1 reduction, 200 steps per rotation stepper, 25,400 steps per inch. But that shouldn't change commanded movement resulting in no movement....
10:51 AM lcnc-relay: <voiditswarranty@> oh wait a sec I think I just solved it. it's very close to the minimum step size. riocore may not be accumulating step error for movements commanded that are smaller than the minimum step size.
10:51 AM lcnc-relay: <voiditswarranty@> guess I'll throw a 4x microstepping on it
11:05 AM lcnc-relay: <voiditswarranty@> ah nope, I was right when I wrote hte config. 5mm leadscrew, 400 steps per motor rotation, 10:1 reduction. I wonder what the G64 defaults are, would that be involved in a very small movement from a stop in a single axis?
11:56 AM lcnc-relay: <oddname_skane@> I believe I have some oddities with QT, what can I do to check if QT is fully functional etc?
12:23 PM roycroft: there is an open source handheld cnc router-ish thing now: https://www.youtube.com/watch?v=Tdpeqkf2NNs
12:24 PM roycroft: it uses relative positioning, so it won't ever be a serious competetor to the shaper origin, but it might be useful for some applications, and certainly costs a fraction of the cost of an origin to build
12:27 PM Rab: "This method [optical mouse sensor] resulted in an error of less than 3mm over a distance of 300mm." I wonder if that's cumulative, though, so if you move 0-300-0mm you're now 6mm off, and so on.
12:29 PM lcnc-relay: <oddname_skane@> its a start atleast for an open source solution 🙂
12:30 PM lcnc-relay: <oddname_skane@> hopefully it will improve 😄
12:33 PM roycroft: since it's relative positioning, i would assume cumulative
12:33 PM roycroft: for a while it will likely only improve so much
12:33 PM roycroft: i'm sure the developers are dancing around shaper's patents
12:34 PM roycroft: being very careful not to step on any of them
12:35 PM roycroft: it's an interesting project, though
12:35 PM roycroft: and if i ever have some spare time i might build one, even though i have an origin
12:36 PM roycroft: it's kind of big and clunky looking right now, too, but i'm sure they'll refine that
01:02 PM jpa-: Rab: i'd say that it has to be cumulative, otherwise it could be just calibrated out
01:03 PM lcnc-relay: <unterhausen@> with shaper, do you have to buy the fiducials occasionally?
01:04 PM jpa-: unterhausen@: yeah, and the shaper origin tape costs like 0.50 EUR per meter
01:04 PM xxcoder: roy pretty cool, though with base that wide, I wonder if they can simply have muliple sensors to lessen error
01:06 PM Rab: xxcoder, they say they use four optical pickups and average between them.
01:07 PM xxcoder: ahh missed that. thanks
01:08 PM Rab: https://www.cameronchaney.com/handheld-cnc-router
01:08 PM Unterhausen: locating that thing reminds me of working with mobile robots. I learned all about how to read barcodes
01:09 PM Unterhausen: there was a software package someone had used to get their masters degree, but it only worked if your robot never rotated at all
01:10 PM roycroft: this project uses multiple optical sensors to help detect rotation
01:10 PM roycroft: i just can't see a device that uses relative positioning every working really well
01:10 PM Unterhausen: yeah, but it made me think of having fixed overhead cameras
01:10 PM roycroft: folks complain about the expense of the fiduciary tape for the shaper origin
01:11 PM xxcoder: unter might not be bad idea, to combo those.
01:11 PM roycroft: but that cost is actually fairly minimal, and folks (including myself) make permanent workspaces all the time
01:11 PM Unterhausen: we got the code reading rotation in an afternoon
01:12 PM roycroft: if you're going to do a $3500 floor inlay with an origin you can afford to burn $50 worth of fiduciary tape for the job
01:12 PM Unterhausen: I thought about snitching to their advisor
01:12 PM roycroft: if you can't, then up the price to $3550 and you'll still get the job
01:13 PM Unterhausen: some of the things Eric Curtis does with his shaper look like maybe the tape wouldn't be a great investment. Like cutting mortises
01:13 PM roycroft: the big positioning issue with that compass thing is that it's trying to detect movement relative to the workpiece
01:13 PM roycroft: and some of the workpiece material will be hard to detect movement on
01:13 PM roycroft: you use a shaper plate with a fence for mortising
01:14 PM roycroft: the plate is a permanent tape field
01:14 PM roycroft: using shaper tape for a mortise is dumb in most cases
01:14 PM roycroft: you can use a shaper workstation for end mortises
01:14 PM Unterhausen: okay, I tune him out when he gets the shaper out, so I didn't notice that
01:14 PM Rab: String pot-style encoders could obviate the need for the tape. Like a Maslow CNC in reverse. A little finicky to set up, but so's the tape I guess.
01:15 PM Unterhausen: mostly just afraid he will talk me into buying one
01:15 PM Unterhausen: yeah, at some point you have a cnc router table though
01:15 PM roycroft: there's a project that's getting a lot of discussion on the shaper forum now about a big workstation that has slides like a slab flattener
01:15 PM roycroft: so you can pocket out large areas
01:16 PM lcnc-relay: <voiditswarranty@> roycroft: many DRO scales are relative, super super common. as usual, "it depends" 😄
01:16 PM roycroft: it uses a large tape field that folks get printed on an aluminium plate
01:16 PM Unterhausen: roycroft, is there a link to look at it?
01:16 PM roycroft: voiditswarranty: but those dro scales have a constant baseline to index against
01:17 PM lcnc-relay: <voiditswarranty@> true
01:17 PM lcnc-relay: <voiditswarranty@> optically it'll vary with the contrast of the background
01:17 PM roycroft: with a cnc router you're going to be trying to detect motion on constantly varying surfaces
01:17 PM lcnc-relay: <voiditswarranty@> yeah, b ut with high DPI sensors these days, the picture isn't changing very quickly at computer speeds
01:17 PM roycroft: i'm looking for that thread now
01:18 PM Unterhausen: thanks
01:19 PM roycroft: https://community.shapertools.com/t/zsws-zedsled-for-workstation/13719
01:19 PM roycroft: it's a fairly long thread
01:19 PM roycroft: be warned
01:20 PM roycroft: it started out as a sled for shaper workstation
01:20 PM roycroft: and evolved into a giant fiduciary grid for really big parts
01:21 PM roycroft: oh, that thread does not have the big grid part
01:21 PM roycroft: it must be another z-sled thread that branched off that one
01:22 PM roycroft: https://community.shapertools.com/t/a-whole-new-way-to-use-shaper/13500
01:22 PM roycroft: that's the one
01:22 PM roycroft: sorry
01:22 PM roycroft: the big fiduciary grid is printed on dibond
01:22 PM Unterhausen: I saw the printed fiducial panel
01:23 PM roycroft: there's another thread somewhere about a guy who built a boat using a shaper origin to jig up most of the curves
01:23 PM roycroft: his tape field was several meters wide
01:23 PM Unterhausen: the open source project could use encoders with a bit of work
01:23 PM roycroft: and still maintained really high accuracy from one end to the other
01:24 PM roycroft: yes, but that might be stepping on shaper patents
01:24 PM roycroft: i don't know where that line is, but it's something i'm sure they are considering
01:24 PM Unterhausen: it seems strange they would try to patent a cnc router
01:24 PM roycroft: festool bought shaper a couple years ago because they wanted to use the auto-correction techlogy in their tracksaws
01:24 PM xxcoder: not royter, the tape thingy
01:25 PM roycroft: i guess it's cheaper to buy a company than to license its patents
01:25 PM Unterhausen: I guess if you make your own encoder it would start being a problem
01:26 PM roycroft: and again, i don't know that to be the case re encoders being covered under patent, but i know that shaper hold several patents for their device
01:26 PM memleak: finally a break. 53F and sunny, perfect day for a swim.
01:26 PM roycroft: btw, the main domino patents expire this year
01:27 PM roycroft: so there will be domino clones before long
01:27 PM roycroft: probably none will work as well as the real thing, and those that work reasonably well will cost almost as much as the original
01:27 PM Unterhausen: youtube keeps trying to get me to watch domino killer videos. The problem is the thumbnails show that it's never a domino killer
01:27 PM roycroft: but there will be alibaba dominoish things for cheap soon
01:28 PM roycroft: there are no "domino killers"
01:28 PM roycroft: but also, dominos are not the solution for every bit of joinery
01:28 PM xxcoder: lol
01:28 PM Unterhausen: Yeah, I don't think something that allows you to wobble a drill around is going to hurt domino sales
01:28 PM roycroft: and btw
01:28 PM xxcoder: looked at some of thumbnails. funny
01:28 PM roycroft: as much as i like laura kampf and the work she does, her "through domino" joinery idea is just hideous
01:29 PM roycroft: dominos are very useful, and game changers in many ways, but they should NEVER be seen in a finished product
01:29 PM bjork1intosh: post structuralist avant garde joinery.
01:29 PM roycroft: something like that
01:30 PM Unterhausen: the fiber company is out digging up our street finally
01:30 PM roycroft: i actually watch her videos a lot less since she moved to la
01:30 PM Unterhausen: neighbor put her trash in their way
01:30 PM roycroft: when she's back in köln working on lise lotte i always watch
01:32 PM roycroft: so the deluge is finally over
01:32 PM roycroft: and the sun is almost shining
01:32 PM roycroft: i think i'll head out at lunch time, uncover the belt grinder, finish setting it up, and powering it up for the first time in a year
01:32 PM xxcoder: very windy here still. my sleep was prety crap because of it lol
01:32 PM roycroft: speaking of game changers, having that back online will be so awesome
01:33 PM roycroft: it's a bit windy here, but not too bad
01:33 PM roycroft: eugene got a lot of it - there were several trees downed on the u of o campus because of the wind, and scattered power outages in the southern valley
01:33 PM roycroft: for once we did not have a power outage, but we got all the rain and virtually none of the wind
01:35 PM Unterhausen: some people are trying to talk me into building jer schmidt's belt grinder
01:36 PM Unterhausen: those people are enablers
01:37 PM xxcoder: https://c.tenor.com/7RKKfJYJ22oAAAAd/tenor.gif
01:39 PM Unterhausen: exactly
01:40 PM Unterhausen: the cool thing about that belt grinder is the xy table
01:53 PM roycroft: i built jer smith's v1 grinder
01:53 PM roycroft: it is awesome
01:54 PM roycroft: i'm thinking about building a 2gen one
01:54 PM roycroft: and i have to decide soon, because some of the accessories do not work with both
01:54 PM roycroft: and i have a large contact wheel and a set of small contact wheels that i've not assembled on to arms yer
01:54 PM roycroft: yet
01:55 PM roycroft: i need to either make them to fit the first gen grinder or build the second gen grinder and build them to fit that
01:55 PM roycroft: and btw, i emailed jer about the 2nd gen plans that i bought from him, asking a couple questions about mods i was considering making
01:56 PM roycroft: mainly because the mount is different, and i had already built a custom stand for the first gen grinder
01:56 PM roycroft: instead of answering my questions, he sent me a set of modified plans to my specfications
01:56 PM roycroft: so kudos to him for that
01:57 PM Unterhausen: Didn't realize I was going to find another enabler
01:57 PM roycroft: haha
01:57 PM roycroft: honestly, it's the best 2x72 belt grinder design i've ever seen
01:58 PM roycroft: the one feature of the table not being attached to the belt drive puts it in a different class
01:58 PM Unterhausen: did you cut the parts out yourself?
01:58 PM roycroft: i sold my big belt sander - 6x89 - a few years ago, and have lamented selling that, because it was so nice for edge sanding long boards
01:59 PM roycroft: yes, i built it from scratch
01:59 PM roycroft: i even made my own tensioner spring
02:01 PM Unterhausen: I wonder if I have enough steel to make it, that would change things.
02:02 PM Unterhausen: some people in seattle had a group buy of the parts from osh cut
02:02 PM Unterhausen: If I did that, it would cost me 3x
02:05 PM Unterhaus_: fiber company is out there digging by hand
02:05 PM Unterhaus_: probably afraid they are going to cut the cable company's lines
02:18 PM Unterhaus_ is now known as Unterhausen
02:19 PM Unterhausen: distortion is the weakness of that design
02:22 PM roycroft: yes
02:22 PM roycroft: but it's better than using square tubing
02:23 PM roycroft: i think with a more powerful welder the heat would be more concentrated and you wouldn't get as much warping
02:23 PM roycroft: i have a more powerful welder now, so if i build a 2nd gen version i can test that theory
02:24 PM roycroft: jer just did a second go of a poor person's surface grinder attachment for his belt grinder, btw
02:25 PM roycroft: i'm thinking that would be good for the knifemaking crowd
02:45 PM Unterhausen: I need to learn more about distortion and mig welding. I mostly have it nailed for torch welding
02:46 PM Unterhausen: there are companies that use heat bending for all kinds of things
02:46 PM roycroft: i try to be good about it
02:47 PM Unterhausen: I would weld at the makerspace, I don't think they have oxy/acetylene there and I'm not sure if map gets things hot enough
02:47 PM roycroft: i.e. on something like that receiver tube i tacked it together then did small amounts of infill welding, and jumped around the part so i wasn't concentrating too much heat in one place at a time
02:47 PM roycroft: but it still warped too much
02:48 PM roycroft: i am not a particularly competent welder, i readily admit
02:48 PM roycroft: and much of the problem may have been lack of skill on my part, but i think using an underpowered welder was the big contributor
02:48 PM Unterhausen: I have to think about controlling distortion in this case. I am actually planning out a youtube on how it works on bike frames
02:48 PM roycroft: it seems unintuitive that too little power would result in too much heat
02:49 PM roycroft: but i'm fairly convinced that was the case
02:49 PM Unterhausen: weirdly, people that try to build bike frames with mapp often overheat things because of not enough heat. Very common problem
02:49 PM roycroft: yes
02:50 PM roycroft: my experinces brazing copper demonstrate that
02:50 PM roycroft: get in hot and fast and get out
02:50 PM roycroft: if you go in cool it takes forever and you end up overheating things by the time you get your joint hot enough
02:51 PM Unterhausen: oxy/acetylene lets you heat a wide area to avoid distortion.
02:51 PM roycroft: yes
02:51 PM roycroft: rosebuds are great for pre-heating cast iron, for example
02:51 PM roycroft: but you still want a fairly small flame right at the point of the braze
02:51 PM roycroft: fairly small but very hot
02:52 PM roycroft: anyway, have fun building jer's grinder
02:52 PM roycroft: i know you're going to do it :)
02:52 PM roycroft: oh, also, i upgraded the motor on mine from 2hp to 3hp when my original motor burned up
02:53 PM roycroft: i highly recommend 3hp
03:45 PM * JT-Shop opened the doors to the shops and garage
03:46 PM JT-Shop: 72°F
04:08 PM roycroft: it's alive!
04:08 PM * roycroft has belt grinder again
04:09 PM roycroft: funny, though, how the motor spun backwards when i wired it up according to the wiring diagram i made when i built it
04:09 PM roycroft: i wonder if we have australian power here
04:09 PM roycroft: anyway, as it's 3 phase, that was a quick fix
04:09 PM roycroft: it's getting pretty warm here as well, but not that warm
04:09 PM roycroft: i'm working outside in a t-shirt today
04:17 PM JT-Shop: same
04:30 PM roycroft: https://roycroft.us/BeltGrinder
04:31 PM roycroft: there's obviously some more cleanup, rust removal, and repainting needed
04:31 PM roycroft: but it works
04:32 PM roycroft: i think the bench top cleaned up nicely, considering it was covered with rust
04:33 PM JT-Shop: https://www.trick-tools.com/Multitool_2_x_48_inch_Belt_Grinder_Attachment_MT482_307
04:34 PM JT-Shop: I have that one on a jet grinder
04:34 PM Unterhausen: roycroft, nice build except for the welds. Which I wouldn't have noticed without you mentioning it
04:36 PM roycroft: the welds are strong but ugly as sin
04:36 PM roycroft: they are one of the reasons i'm considering building a gen 2
04:37 PM roycroft: that's a pretty nice attachment, jt-shop, and gets you a lot of the functionality of a belt grinder
04:38 PM Unterhausen: plans start to add up if you get all three sets
04:38 PM JT-Shop: yup it works very good, I did consider building one of those like yours
04:38 PM roycroft: i cannot emphasize enough, though, that the ability to rotate the drivetrain independently to the table is a game changer
04:39 PM Unterhausen: not sure what the most powerful motor I have sitting around is
04:39 PM roycroft: and if you're looking for interesting surface finishes, try rotating the belt 30 or 40 degrees, not 90 degrees, and you get some neat diagonal grinding patterns on your parts
04:39 PM roycroft: really, honestly, make sure it's 3hp :)
04:39 PM Unterhausen: probably have to buy one
04:39 PM roycroft: the 2hp was ok
04:40 PM roycroft: i got rid of it not because it did not have enough power, although i could slow it down if i was pushing hard
04:40 PM Unterhausen: I got an extra motor with my lathe because it was converted to single phase
04:40 PM roycroft: it was an open frame motor (it was something cheap i picked up locally), and got some grinding dust in it and released all the magic smoke
04:40 PM roycroft: i kind of anticipated that might happen
04:41 PM roycroft: the new motor is 3hp and is tefc
04:41 PM roycroft: i never knew how underpowered 2hp was until i installed the 3hp motor
04:41 PM roycroft: but keep in mind that before i built that belt grinder i was using a small 1x24 or so grinder with a 1/2hp motor
04:42 PM roycroft: the 2x72 was a huge upgrade
04:44 PM roycroft: i have the materials to make a glass platen, which will run a lot cooler, but i haven't gotten around to making it yet
04:45 PM roycroft: and as i said earlier, i might up grade to a 2gen grinder, in which case i'd need a new platen anyway - i'm not going to make any more attachments until i decide on whether to upgrade the grinder or not
04:45 PM roycroft: and i would really like to start using my contact wheels, so i should make that decision soon
05:04 PM Unterhausen: I think the most powerful motor I have is a servomotor.
05:04 PM Unterhausen: I have 1hp motors otherwise
05:07 PM lcnc-relay: <skunkworks8841@> Lol.. I think the biggest 3 phase motor we have is 20hp
05:10 PM roycroft: and now the workday is done - i think i'll go clean up the belt grinder tables and put some fresh boeshield t-9 on them
05:11 PM roycroft: i have the horizontal table that is in the pics, a long version of that table for edge sanding, and a 45 degree fixed table for chamfering and beveling
05:12 PM roycroft: i never built an adjustable table, and again, will hold off until i decide if i'm going to upgrade the grinder or not
05:12 PM Unterhausen: I have a weakness for collecting motors, but I do have my limits. No place to put them
05:15 PM roycroft: one other thing - jer schmidt designed his machine so it can be built with common machines like small welders and drill presses, and not require a mill and/or lathe or anything like that
05:15 PM roycroft: he also designed it using common hardware store fasteners
05:15 PM * JT-Shop is on a mission to get rid of crap I'll never use
05:16 PM lcnc-relay: <skunkworks8841@> I have already taken a 2.5 tons of electric motors to recyling
05:16 PM roycroft: i substituted shoulder bolts for most of the standard threaded bolts in my build, and i think that was a good decision
05:16 PM lcnc-relay: <skunkworks8841@> Recycling
05:16 PM roycroft: besides, going online to mcmaster is easier than going to the hardware store
05:18 PM * roycroft also substitutes socket head cap screws for hex head machine screws in most builds, because it's the way civilized people build things
05:19 PM roycroft: or to quote a well-know u-tuber, it's what separates us from the animals
05:20 PM Unterhausen: I have a portaband in a stand, so I can cut parts out. I just haven't gone through my steel collection to see if I have to track down material
05:20 PM roycroft: you need a little dom tubing
05:20 PM Unterhausen: Never have figured out how to buy steel around here in small quantity at a decent price. Might have to go to pittsburgh
05:20 PM roycroft: and some 1-1/2" square bar
05:21 PM roycroft: those things might not be in your collection
05:22 PM roycroft: i've found that onlinemetals.com is cheaper than my local (used to be local) steel yards, but after shipping, are a bit more than local, but not outrageously more
05:22 PM roycroft: the problem with them, which is not a problem for this build, is that you can't buy 20' sticks from them
05:22 PM roycroft: after 5' shipping gets really expensive
05:23 PM roycroft: abom swears by mcmaster for metal, but he must get the 4 finger discount from them that nobody else does
05:24 PM xxcoder: what not 5 finger? ;)
05:24 PM roycroft: no, i think he pays a little
05:24 PM xxcoder: but yeah some shapes is hard to ship. 20 feet?
05:24 PM roycroft: he doesn't mention them often enough for full sponsorship
05:24 PM roycroft: 20' is a standard length
05:25 PM roycroft: i like buying 10' lengths when i can unless that will cause a lot of waste, then i wrangle a 20' length
05:27 PM roycroft: ok, time to make grinder accesories shiny and clean
06:34 PM lcnc-relay: <lolthekidison@> https://cdn.discordapp.com/attachments/970073761770512384/1344102905170628659/20250225_181828.jpg?ex=67bfb0e0&is=67be5f60&hm=442c60e7e4cac6f87a3a67842ddf211aada402a1ad72e370bccd1d29543309e3&
06:35 PM lcnc-relay: <lolthekidison@> Anyone know what this is about? I thought I gave this error because my part was to big but even with a small test part is still says this.
07:19 PM Tom_L: X axis has a following error
07:21 PM Tom_L: max acceleration or max velocity may be too high
07:21 PM Tom_L: could be other reasons..
07:51 PM lcnc-relay: <grandixximo@> lolthekidison@: It depends on how you move your axis, the following error is the difference between pos-cmd and ps-fb, this means it's the distance between where the axis is expected to be as defined by your axis parameters, and where it actually is as defined by the feedback from wherever you have that hooked to.
07:51 PM lcnc-relay: you can modify how much you want to allow in the .ini file you have MIN_FERROR for very low speeds and FERROR for max speed.
07:51 PM lcnc-relay: Following errors can also come from improper setup of joints and axis parameters, if your joints can't run as fast as your axis can, you are going to have following errors.
07:52 PM lcnc-relay: <grandixximo@> * joint pos-cmd and ps-fb, this means it's the distance between where the joint is expected to be as defined by your joint
07:57 PM lcnc-relay: <grandixximo@> It could be an inherent problem with your settings, or it could be a problem with the hardware, depending on where the pos-fb is coming from
08:03 PM lcnc-relay: <lolthekidison@> Sorry trying to understand what you exactly mean. Some stuff is just a bit consing for me.
08:04 PM lcnc-relay: <lolthekidison@> I don't think it's a speed issues. I just reset my step conf Incase I messed something up somehow. I max speed was 100mm/s and my cutting speeds are around 10mm/s
08:08 PM lcnc-relay: <grandixximo@> what hardware are you running?
08:08 PM lcnc-relay: <lolthekidison@> Oh wow it works again
08:08 PM lcnc-relay: <lolthekidison@> Haha this is all a lot to get a hold of 😅
08:09 PM lcnc-relay: <grandixximo@> you have a computer running linuxcnc
08:09 PM lcnc-relay: <grandixximo@> that's all I know
08:09 PM lcnc-relay: <grandixximo@> from the picture
08:09 PM lcnc-relay: <lolthekidison@> Yes an i7 9500k or something. And office PC I got from work
08:10 PM lcnc-relay: <grandixximo@> how are you controlling the motors?
08:10 PM lcnc-relay: <lolthekidison@> I think the bios was originally messed up because after a few reboots it would delete Linux cnc from my drive. That seems to be fixed now
08:10 PM lcnc-relay: <lolthekidison@> Umm just like a generic parallel port breakout board
08:10 PM lcnc-relay: <grandixximo@> Yeah it's never gonna work
08:11 PM lcnc-relay: <lolthekidison@> https://cdn.discordapp.com/attachments/970073761770512384/1344127297061392384/rn_image_picker_lib_temp_c83fe5fc-5d6a-4183-a1d5-d86b993994a9.jpg?ex=67bfc798&is=67be7618&hm=ef7365bba7eadf10a3a03a206d00580482ecbd89426d0988ab9fdab41b3f995a&
08:11 PM lcnc-relay: <lolthekidison@> Why will it not?
08:11 PM lcnc-relay: <lolthekidison@> It has been working for some cuts
08:13 PM lcnc-relay: <grandixximo@> you have a general PCI LPT card in the computer?
08:14 PM lcnc-relay: <lolthekidison@> It's like one of the old school ones I guess yeah
08:14 PM lcnc-relay: <grandixximo@> close linuxcnc
08:14 PM lcnc-relay: <grandixximo@> run this command
08:15 PM lcnc-relay: <grandixximo@> latency-test
08:16 PM lcnc-relay: <lolthekidison@> Yea I've done that a bunch. Since resetting the bios I haven't done any optimizations. Max jitter 4294 base thread 44599
08:16 PM lcnc-relay: <grandixximo@> open youtube
08:16 PM lcnc-relay: <lolthekidison@> Opening chrome and files explorer
08:16 PM lcnc-relay: <lolthekidison@> Yea I have done all that at one point and gotten lower numbers than this
08:17 PM lcnc-relay: <lolthekidison@> Once every like 5 times I open Linux cnc now it will give a latency error. Before it didn't when I limited the cpu to 4 instead of like 8 cores
08:18 PM lcnc-relay: <grandixximo@> You can'
08:18 PM lcnc-relay: <grandixximo@> * can't run step generation with 44599 on base thread
08:18 PM lcnc-relay: <grandixximo@> what's your base period?
08:19 PM rdtsc: wonder if the CPU has "efficiency" cores - BIOS might have a way to turn those off
08:20 PM lcnc-relay: <grandixximo@> has anyone got a stable i7 step generation?
08:20 PM lcnc-relay: <lolthekidison@> https://cdn.discordapp.com/attachments/970073761770512384/1344129561570578516/rn_image_picker_lib_temp_43b6db72-83eb-4cdb-bfb9-23f6b41f2ceb.jpg?ex=67bfc9b4&is=67be7834&hm=66a09da766ba5cbbf85bb85b6efc17740cb3f583ccfa10000d27c8c32c7a0353&
08:20 PM lcnc-relay: <lolthekidison@> I've been able to keep the middle to below 10k before while many things running. Maybe I'm missing something though
08:22 PM lcnc-relay: <grandixximo@> can you post your ini file?
08:24 PM lcnc-relay: <grandixximo@> BASE_PERIOD = ....
08:24 PM lcnc-relay: SERVO_PERIOD = ....
08:24 PM lcnc-relay: what do you have these set at?
08:24 PM xxcoder: ah yes vitamin B6
08:25 PM xxcoder: she starts to talk about https://youtu.be/7W4E43MXPto?t=555
08:26 PM xxcoder: usa 100mg, and uk obly 12 mg
08:26 PM lcnc-relay: <lolthekidison@> https://cdn.discordapp.com/attachments/970073761770512384/1344131103237341225/20250225_201642.jpg?ex=67bfcb23&is=67be79a3&hm=6d0bf3886739ab32ddea87be706723682629c08e91784f57b767ff2ff288917c&
08:26 PM xxcoder: oops wrong channel
08:26 PM lcnc-relay: <lolthekidison@> This is the max 50,000 you can do in the stepconf wizard
08:28 PM lcnc-relay: <grandixximo@> run this
08:28 PM lcnc-relay: latency-histogram --base 65000
08:28 PM lcnc-relay: <skunkworks8841@> Lol.. making a block diagram for the hal that runs the monarch... It is more involved than I thought
08:28 PM lcnc-relay: <grandixximo@> runa a ferw glxgears
08:29 PM lcnc-relay: <lolthekidison@> Ok second I'm going to isolate the cores again
08:32 PM lcnc-relay: <grandixximo@> https://store.mesanet.com/index.php?route=product/category&path=83_84
08:32 PM lcnc-relay: <grandixximo@> my suggestion is looking into one of these
08:33 PM lcnc-relay: <lolthekidison@> grandixximo@: Is this what I should try to eventually get?
08:33 PM lcnc-relay: <grandixximo@> if isolating the cores doesn't get you good performance
08:33 PM lcnc-relay: <grandixximo@> when you use FPGA the steps are created by the FPGA you don't care about latency so much anymore
08:33 PM lcnc-relay: <grandixximo@> 1ms is good enough
08:34 PM lcnc-relay: <lolthekidison@> My friend was just saying not to trust preloaded or slow software. He said Linux cnc was the best because it was live movement. This is where I end up 😅
08:34 PM lcnc-relay: <grandixximo@> it's not linuxcnc fault
08:35 PM lcnc-relay: <lolthekidison@> Yea my PC is just not consistent enough to generate the steps?
08:35 PM lcnc-relay: <grandixximo@> Your CPU choice is not appropriate to have low latency enough to generate the steps by itself
08:35 PM lcnc-relay: <grandixximo@> Unfortunatly most beefy modern CPU don't have good latency
08:36 PM lcnc-relay: <lolthekidison@> Alright I'll give that test a go in a second. Luckily I got this computer for free and the CNC has been cheap
08:36 PM lcnc-relay: <grandixximo@> Because they try and do so much stuff every cycle, they don't really care if they get the cycle consistently in time
08:38 PM lcnc-relay: <grandixximo@> And with the power saving outlook of today, they rather sleep and keep it the CPU cool and saving power, than hitting every cycle to the ns
08:38 PM lcnc-relay: <grandixximo@> * rather then
08:40 PM lcnc-relay: <grandixximo@> https://imgur.com/a/Z3XCu56
08:40 PM lcnc-relay: <grandixximo@> This is what I'm running, and I don't make steps, but in theory I could
08:43 PM lcnc-relay: <lolthekidison@> I just disabled all muliticore power boost/ savings USB ports, most stuff in bios. Only using core 1/2 now
08:44 PM lcnc-relay: <grandixximo@> you have glxgears?
08:44 PM lcnc-relay: <lolthekidison@> Yea I just forget what panel it's in the top left
08:45 PM lcnc-relay: <grandixximo@> you can run from terminal
08:45 PM lcnc-relay: <grandixximo@> glxgears &
08:46 PM lcnc-relay: <grandixximo@> maximize and minize glxgears, that normally makes the worst spike you will ever see, or about
08:47 PM lcnc-relay: <grandixximo@> I normally open 9 glxgears instances, and maximize and minimize one of them
08:49 PM lcnc-relay: <grandixximo@> if you run histogram there might be a glxgears button at the bottom
08:49 PM lcnc-relay: <lolthekidison@> https://cdn.discordapp.com/attachments/970073761770512384/1344136957483155608/20250225_204007.jpg?ex=67bfd097&is=67be7f17&hm=ea6be761bed46e14538b72064a0c9cfedfb95b0d3a7d1e9b5b87e1b57311e4cb&
08:49 PM lcnc-relay: <lolthekidison@> Have yt and gears open, spammed a few screens and tabs
08:50 PM lcnc-relay: <grandixximo@> https://imgur.com/undefined
08:51 PM lcnc-relay: <grandixximo@> https://imgur.com/mVX4eoN
08:51 PM lcnc-relay: <lolthekidison@> Oh I missed you can open 9
08:52 PM lcnc-relay: <grandixximo@> You can't run steps with latency like that
08:54 PM lcnc-relay: <grandixximo@> it's not as bad as I thought, but it ain't super good
08:54 PM lcnc-relay: <grandixximo@> You maybe able to change your base period to 100000 or 120000, and run the machine at half of what you are running now
08:54 PM lcnc-relay: <lolthekidison@> Half speeds?
08:56 PM lcnc-relay: <grandixximo@> yes making half the steps
08:56 PM lcnc-relay: <grandixximo@> doubling the time between steps
08:57 PM lcnc-relay: <grandixximo@> time itself is fixed and running only in one direction unfortunately 😅
08:58 PM lcnc-relay: <lolthekidison@> Will it help if I'm not using so many micro steps?
08:59 PM lcnc-relay: <lolthekidison@> I'm using 8 i believe. I think for CNC that's sortve high? For 3d printers it's not alot
09:00 PM lcnc-relay: <lolthekidison@> grandixximo@: Also wow an n100 really does that good!? Isn't that like the most basic chip you can get in a bunch of laptop and mini PCs?
09:05 PM lcnc-relay: <grandixximo@> Beefy CPU is not how you get good latency, simple CPU is better, yes the N100 is pretty decent
09:08 PM lcnc-relay: <grandixximo@> if you make your steppers require half the step, you will go same speed but loose accuracy
09:11 PM Tom_L: a good load test is unigine heaven... way tougher than glxgears & yt
09:11 PM lcnc-relay: <lolthekidison@> I changed the base period in ini and you both were correct I was moving to fast
09:12 PM lcnc-relay: <lolthekidison@> I can do 65000 with the new CPU settings but if I change to say 100k it gives that error again
09:13 PM lcnc-relay: <lolthekidison@> grandixximo@: However that requires that ethernet card to work
09:14 PM lcnc-relay: <grandixximo@> There is an infinite way to setup linuxcnc
09:14 PM Tom_L: are you also running isolcpus?
09:14 PM lcnc-relay: <grandixximo@> I'm running ethercat
09:14 PM Tom_L: setting in grub
09:14 PM lcnc-relay: <lolthekidison@> Hehehe just what I need 😅 😂
09:14 PM lcnc-relay: <grandixximo@> the intel ese kernel does isolcpu itself
09:16 PM lcnc-relay: <grandixximo@> the MESA cards have ethernet and PCI
09:16 PM Tom_L: i was doing some testing for RTAI and found i was getting 15us running 4-5 glxgears with 2-3 yt but when i ran unigine heaven it jumped to 23.4
09:16 PM Tom_L: http://tom-itx.no-ip.biz:443/~webpage/cnc/JT-SHOP/flexgui/RTAI5.4.290.png
09:17 PM lcnc-relay: <grandixximo@> You have to consider if you are going to run something that heavy when you run linuxcnc
09:17 PM Tom_L: but i'm not using a base thread, just servo
09:17 PM Tom_L: i won't but it is a good test
09:17 PM lcnc-relay: <lolthekidison@> grandixximo@: Wait there's a way I can find the accuracy by taking the 1.8° steppers with the 4mm pitch lead screw right?
09:17 PM lcnc-relay: <grandixximo@> I also use only servo, base was just for showing
09:18 PM lcnc-relay: <grandixximo@> Tom_L: it's very worst case scenario
09:19 PM Tom_L: i use RTAI on mine however you can't if you're running ethernet
09:19 PM lcnc-relay: <grandixximo@> i5 gets better letancy than i7 in my experience
09:19 PM Tom_L: i'm running i5 here
09:20 PM Tom_L: i do have a spare i7 but i'm not using it
09:20 PM Tom_L: might be interesting to put it in the same board/setup and run some tests
09:21 PM lcnc-relay: <grandixximo@> I had a 6700K as my office PC, could not get good latency on it
09:21 PM lcnc-relay: <grandixximo@> but kernels have improved lots since
09:22 PM Tom_L: memleak has been doing alot with RTAI and i think preempt-rt as well
09:22 PM Tom_L: this new rtai he just came out with is quite a bit better than the previous one
09:23 PM Tom_L: and it's not been out that long
09:26 PM lcnc-relay: <grandixximo@> Yeah, and the newer linuxcnc kernels will have preempt patched pre applied, 6.2x i think
09:26 PM Tom_L: i've been installing those from the package manager
09:26 PM Tom_L: which is kinda nice
09:27 PM Tom_L: i think 6.1x was the last one i've installed
09:29 PM lcnc-relay: <grandixximo@> As of Linux 6.6, the PREEMPT_RT patches have been fully merged into the mainline kernel.
09:29 PM Tom_L: that's good news
09:29 PM lcnc-relay: <grandixximo@> It will trickle down to the repos eventually
09:29 PM Tom_L: 6.6?
09:29 PM lcnc-relay: <grandixximo@> yes
09:29 PM Tom_L: that's pre wheezy
09:29 PM lcnc-relay: <grandixximo@> I remembered 6.2 but i searched now and it's 6.6
09:30 PM Tom_L: bookworm is 12
09:30 PM lcnc-relay: <grandixximo@> no
09:30 PM lcnc-relay: <grandixximo@> linux
09:30 PM lcnc-relay: <grandixximo@> not debian
09:30 PM Tom_L: ahh ok
09:30 PM lcnc-relay: <grandixximo@> https://www.kernel.org/
09:30 PM Tom_L: yeah
09:31 PM lcnc-relay: <grandixximo@> https://github.com/torvalds/linux
09:31 PM Tom_L: i'm familiar with kernel org
09:31 PM lcnc-relay: <grandixximo@> yeah, it's cool that you can go look at the source anytime
09:32 PM lcnc-relay: <grandixximo@> Open source is cool
09:32 PM Tom_L: until it's trying to play catch up
09:32 PM Tom_L: then you get to play dependency hell
09:33 PM lcnc-relay: <grandixximo@> Just one more thing to nerd about
09:34 PM lcnc-relay: <grandixximo@> lolthekidison@: yes, easy math
09:35 PM lcnc-relay: <grandixximo@> it's direct drive no other reduction?
09:36 PM lcnc-relay: <grandixximo@> With a 1.8° stepper motor, which has 200 steps per revolution, and a 4 mm pitch lead screw, the resolution per full step is 4 mm divided by 200, resulting in 0.02 mm (20 microns) per step
09:36 PM lcnc-relay: <grandixximo@> 360/1.8=200 steps per revolution
09:40 PM lcnc-relay: <lolthekidison@> Yea I had to look it up how. I think I will maybe try to lower the micro steps and then see if I can keep up the speed a bit. The machine can't really cut fast either way but hopefully it helps
09:41 PM lcnc-relay: <lolthekidison@> I really appreciate everyone's help, thank you
09:44 PM Tom_L: it's always good to have some linuxcnc discussion in the linuxcnc channel :)
09:55 PM lcnc-relay: <grandixximo@> Tom_L: is the RTAI 5.4 pretty stable? are you running 24/7?