#linuxcnc Logs
Sep 25 2023
#linuxcnc Calendar
01:41 AM Deejay: moin
03:38 AM rigid: i suppose no one in here knows details about configs/common/client.nml?
03:39 AM rigid: commit history suggests that it just has been moved around since git is used
03:40 AM rigid: forum search also doesn't bring up anything
03:59 AM rigid: client.nml and server.nml to me looks like the key for running the UI remotely
04:01 AM rigid: ahh, the forum search is limited to posts since last year. There's one post from 2020 mentioning it.
04:01 AM rigid: Seems rodw knows about it
04:08 AM travis_farmer: Morning :-)
04:26 AM Tom_L: morning
04:29 AM NoSpark: Good Morning
04:29 AM NoSpark: About to start converting a large CNC router to Linuxcnc
04:30 AM NoSpark: I have a question about how to handle vacuum tables with a vacuum sensor
04:30 AM Tom_L: single spindle?
04:30 AM Tom_L: most cnc's i've enountered to that manually
04:30 AM Tom_L: but i'm not a woodworker
04:31 AM Tom_L: custom M code maybe
04:31 AM NoSpark: Tom_L: It has a single spindle, but also has an 11 position drill head, and a separate probe with independent Z-axis
04:31 AM Tom_L: you should have fun
04:32 AM NoSpark: idk if you would call that 13 head or not *shrug*
04:32 AM NoSpark: Idk about the drilling head, cause I have a 17 pos tool rack
04:33 AM NoSpark: my question about the vacuum is, is there a way to "pause" the job if the vacuum fails?
04:33 AM Tom_L: anything is possible..
04:33 AM NoSpark: I'm looking to use QT Dragon (it is very pretty), and it seems to have a pause pin, so should I just connect it to that?
04:34 AM travis_farmer: there is if you know how to program a HAL comp program
04:34 AM Tom_L: read a pin on the vacuum sensor into lcnc
04:34 AM Tom_L: is the sensor analog?
04:34 AM NoSpark: Hal comp is pretty easy, and I can program custom hal components if I have to
04:34 AM Tom_L: you may have to do some coverting there if it is
04:34 AM rmu: NoSpark: there is a "feed hold" pin in motion component that stops movement
04:35 AM NoSpark: Tom_L: Digital, single setting screw with a High/low output
04:35 AM Tom_L: even easier
04:35 AM NoSpark: rmu: feed hold might set fire to the work, I really need to retract
04:35 AM rmu: but that maybe not enough. running tool without vacuum may grab the workpiece
04:35 AM rmu: in reality you only can do an e-stop
04:36 AM NoSpark: rmu: correct, I just had a vacuum failure machining a piece of perspex, which shattered
04:36 AM travis_farmer: i like an analog sensor, so you can detect low vacuum before it is an issue
04:36 AM NoSpark: the problem with estop is you can't resume the program
04:37 AM rmu: how do you get a "vacuum failure"?
04:37 AM NoSpark: rmu: usually the tool cuts all the way through the part, and breaks the vacuum
04:37 AM NoSpark: like when you are separating out pieces
04:38 AM rmu: the only thing that will fix that is a different setup and bigger vacuum pump
04:38 AM NoSpark: say for example you are cutting circles out of some 10mm plastic, if a bit of swarf gets jammed under one edge, it can lift the work, and break the vacuum
04:39 AM NoSpark: rum, I have 2x 13kw vacuums on this machine, a bigger vacuum pump is not really an option
04:39 AM NoSpark: *rmu
04:41 AM NoSpark: the machine is mainly used for machining corian, which runs about US$100 to $600 a square meter
04:41 AM rmu: maybe the plasma guys have a solution for the case the torch collides to pause/restart the program
04:54 AM NoSpark: maybe...
04:54 AM NoSpark: I don't suppose there is a "rewind" feature in linuxcnc
04:58 AM jpa-: you mean running toolpath in reverse? IIRC there is something for that
05:00 AM jpa-: https://forum.linuxcnc.org/24-hal-components/41812-reverse-run
05:03 AM JT-Cave: morning
05:50 AM NoSpark: what is the recommended method for gantry style routers? this is an YYXZ
05:51 AM NoSpark: Should I use the "gantry" rt component?
06:03 AM JT-Cave: use mesact
06:04 AM srk| is now known as srk
06:17 AM JT-Cave: or if all else fails rtfm http://linuxcnc.org/docs/stable/html/config/ini-config.html#_joint__lt_num_gt_section HOME_SEQUENCE
06:17 AM JT-Cave: http://linuxcnc.org/docs/stable/html/config/ini-homing.html#sec:homing-section
06:20 AM Tom_L: i'll build later today
06:21 AM JT-Cave: ok
06:28 AM NoSpark: I'm curious as to why the gantry component exists if trivkins can do everything it can?
06:29 AM NoSpark: is it just a legacy component?
06:38 AM JT-Cave: https://github.com/LinuxCNC/linuxcnc/commits/master/src/hal/components/gantry.comp
06:51 AM JT-Cave: rooster just crowed
09:41 AM JT-Shop: hmm the bp disk drive is full again and that's why I can't login
09:50 AM JT-Shop: hmm I'm stumped now
10:40 AM JT-Shop: the bp is debian 9.13...
10:53 AM roycroft: i'd boot off a cd, mount the root filesystem, clean up some disk space, then boot off the primary device
10:53 AM roycroft: and put a bigger disk in the machine
10:54 AM JT-Shop: I've tried the first one but really didn't find much to delete
10:55 AM roycroft: i don't know how you set up your filesystems, but if /var is part of the root filesystem then you might be able to zero out some logs in /var/log
10:55 AM JT-Shop: I deleted any .old from /var/log
10:56 AM JT-Shop: lunch
10:56 AM roycroft: look for something big in /var/log
10:56 AM roycroft: like /var/log/messages
10:56 AM roycroft: and it's really big you can zero it out
10:56 AM roycroft: that's where i typically see the problem on small systems with a single filesystem
10:57 AM * roycroft tries to always have a separate /var/log filesystem to avoid such problems
11:02 AM JT-Cave: how do you have a separate /var/log
11:02 AM rmu: jt-shop: do you know ncdu? have it installed?
11:02 AM rmu: jt-cave ofc
11:03 AM JT-Cave: rmu, not heard of ncdu
11:03 AM roycroft: i partition the disk at install time
11:03 AM rmu: https://en.wikipedia.org/wiki/Ncdu
11:03 AM roycroft: i am of the splitter camp, which seems to have all but died out
11:03 AM rmu: can help you find what's the culprit when you manage to login
11:03 AM roycroft: but i still find it useful to have several filesystems
11:04 AM JT-Cave: can't login or switch to terminal with ctl alt F1-6
11:05 AM rmu: then you probably have to reboot into recovery mode
11:05 AM rmu: how do you know disk is full if you can't login?
11:05 AM JT-Cave: I booted from a live cd
11:05 AM roycroft: i'll typically install with a 7.5GB /, but with a separate /usr, /home, /tmp, /var, /usr/local, /var/log, and whatever other filesystems are useful for the install (a separate /var/www and /var/lib/mysql for a web server, for example)
11:06 AM JT-Cave: booting to any option and it just goes to the login screen even though lightdm.conf has auto login enabled for me
11:06 AM roycroft: since debian now has mostly merged / and /usr, i've started doing installs with a larger / and no separate /usr
11:06 AM rmu: roycroft: separate /usr isn't really supported afaik
11:07 AM roycroft: rmu: a separate /usr works fine, but see my comment above
11:07 AM rmu: JT-Cave: can you boot into recovery mode? i.e. directly into a root shell?
11:07 AM JT-Cave: how?
11:08 AM rmu: in the grub menu
11:08 AM JT-Cave: I did get into the grub menu once and changed to terminal but ls didn't show much
11:09 AM rmu: take advanced options and there should be recovery mode
11:09 AM JT-Cave: it's debian 9.13 my luck number
11:09 AM JT-Cave: I'll look in a bit when lunch minute is up
11:10 AM roycroft: when you're able to log in, you can look for big files without installing a new utility
11:10 AM roycroft: find / -type d -size 50M -ls , for example, will list all files >50MB in size
11:10 AM roycroft: er
11:10 AM roycroft: find / -type f -size 50M -ls , for example, will list all files >50MB in size
11:10 AM rmu: i usually have a 2g-sized (real, not sparse) file named "spaceclaim" in /root/ that i can delete in case of space emergencies ;)
11:10 AM roycroft: that will list regular files >50MB
11:11 AM roycroft: the first would be directories >50MB
11:11 AM * roycroft just tries to ensure that / never fills up
11:12 AM roycroft: a 10GB /tmp and /var, at the very least, and a separate /home, really helps
11:12 AM roycroft: none of this helps solve the current problem, of course
11:24 AM JT-Shop: rt recovery press enter to continue takes me to the login screen
11:26 AM rmu: it should take you to a text menu where you can go to shell
11:26 AM JT-Shop: cannot open access to console the root account is locked
11:26 AM JT-Shop: it takes me to the login screen
11:27 AM JT-Shop: can't type anything only press enter
11:27 AM rmu: you can try editing command line arguments in grub and add "init=/bin/sh", that will drop you into a root shell, but nearly nothing in the system will be setup
11:28 AM rmu: but the idea with live-cd, mounting of root fs and exploring from there is probably easier / friendlier
11:29 AM roycroft: that's what i would do
11:29 AM rmu: use something like baobab, filelight or ncdu to find out where the large stuff hides
11:29 AM roycroft: and you should have the find command on the livecd
11:30 AM JT-Shop: I'll try that again after I get back from Foxy
11:34 AM JT-Cave: what's a pia is I have the old version of qtpyvcp and emc pre python3 and I'd really hate to recreate that from scratch
11:46 AM rmu: https://www.youtube.com/watch?v=XxbJw8PrIkc s/Developers/Backups/
11:53 AM roycroft: i used to see steve ballmer at mariners games
11:53 AM roycroft: he was all sweaty and crazy there too
12:08 PM bjork1intosh: roycroft, did he by chance remind you of the famous Billy Mays?
12:09 PM roycroft: kind of
12:09 PM roycroft: i could see him doing infomercials
12:10 PM bjork1intosh: but he did informercials!
12:10 PM bjork1intosh: for crappy versions of windows!
12:10 PM roycroft: that last statement is redundant
01:05 PM travis_farmer: i was impressed with the power of my machine... pushed a 1/8in (disposable) bit into plywood... had my pendant on 1.0in rather than 0.01in :-( then i bumped the MPG grabbing the pendant. that one is on me for not turning off the jogging when not in use...
01:07 PM travis_farmer: i have like 25 of those cheap 1/8in carbide bits, but i was glad it was one of those, rather than my amana bits
01:09 PM travis_farmer: in other news, my tool probe is getting bounce (electrical), think i should add a small value (0.1uf) capacitor? debounce the signal a tiny bit?
01:10 PM travis_farmer: i also have some 22pf but thought they would be too small
01:14 PM Tom_L: 1" jog sounds dangerous
01:14 PM travis_farmer: it was
01:15 PM Tom_L: JT-Cave, clone the drive to a bigger one
01:42 PM Tom_L: 2.0.1 must not have saved the feedback in the [SPINDLE] section
01:45 PM roycroft: sometimes i really dislike "smart" mail clients
01:47 PM roycroft: i have multiple email accounts that i use, and when i compose an email from a particular account, both mail.app and postbox silently change the sender address from what i use to the address of the last correspondence i had with the recipient
01:47 PM roycroft: i have to remember to go back and change it to the one i want
01:47 PM roycroft: and sometimes i don't remember
01:47 PM Tom_L: JT-Cave, pr #29 up
01:48 PM roycroft: it is really quite annoying, and i haven't found a setting to turn that crap off
03:08 PM JT-Shop: seems like because I didn't set a root password during the install root is locked... now to figure out how to get around that
03:11 PM travis_farmer: do you have access to the shadow file? you coud generate a shadow hash on one machine and inject it into the file for root on the other machine.
03:12 PM travis_farmer: s/coud/could/
03:13 PM JT-Shop: so clearly the disk is not full
03:14 PM JT-Shop: I have no clue what a shadow file is
03:14 PM rmu: JT-Shop: start from livecd, mount the disk, chroot to the disk, and set root passwd with "passwd" cmd
03:14 PM JT-Shop: ok I'm started from a live cd
03:16 PM rmu: /etc/passwd, the traditional unix user database, needed to be world readable, and back then password hashes were included there
03:16 PM roycroft: you can edit /etc/shadow and insert a password hash there
03:16 PM rmu: so the /etc/shadow file was invented that contains the password hashes and that file was only readable by root
03:16 PM travis_farmer: they are in etc/shadow now i think
03:16 PM roycroft: i can generate one for you if you wish
03:18 PM roycroft: $y$j9T$uOLu83JWP28v.m5nmLmF8.$5o9c.4CasTzv4UzmiuJouSMa2VhsKmBQm4qbMZrnK
03:18 PM roycroft: yD
03:18 PM roycroft: if you add that hash you can use the password "linuxcnc"
03:19 PM roycroft: look at /etc/shadow and you'll see what to do
03:21 PM JT-Shop: there's nothing like that in /etc/shadow just a few numbers after each item
03:21 PM roycroft: probably something like
03:21 PM roycroft: root:*:199180: .....
03:22 PM roycroft: if so, change the * to the hash i pasted above
03:22 PM roycroft: which starts with $ and ends with D
03:22 PM travis_farmer: or you could just give it split to two lines...
03:22 PM JT-Shop: root:*:19625:0:99999:7:::
03:22 PM * roycroft would not recommend that
03:22 PM roycroft: yeah
03:22 PM roycroft: the * means "there is no password"
03:23 PM travis_farmer: dang words were in my text already, i meant that the hash split to two lines
03:23 PM travis_farmer: in IRC
03:23 PM JT-Shop: user is the only one that is different
03:23 PM roycroft: yes, that is why i made it clear how the hash starts and ends
03:24 PM roycroft: do you know the password for "user"?
03:24 PM travis_farmer: right
03:24 PM JT-Shop: yes
03:24 PM roycroft: so you can copy that hash field to root
03:24 PM roycroft: and log in with that password
03:24 PM roycroft: i would make a copy of /etc/shadow before doing this
03:25 PM JT-Shop: I'm editing it with sudo nano
03:25 PM roycroft: i'm not a nano person, but if it were me i'd duplicate the "user" entry, delete the "root" line, and change the username in the duplicate from "user" to "root"
03:25 PM roycroft: however you do that with nano
03:26 PM JT-Shop: $y$j9T$uOLu83JWP28v.m5nmLmF8.$5o9c.4CasTzv4UzmiuJouSMa2VhsKmBQm4qbMZrnK
03:26 PM JT-Shop: I don't see a D in that hash
03:26 PM roycroft: it's two lines what i posted
03:26 PM roycroft: ends with rnKyD
03:26 PM JT-Shop: oh I see now :)
03:26 PM roycroft: so just add yD to what you pasted
03:28 PM travis_farmer: and just so you are aware, if it screws up, just liveCD it again, and re-edit it. no biggie
03:32 PM * roycroft recommends that jt change the password to something not that once he's able to log in as root
03:32 PM * travis_farmer agrees with roycroft ;-)
03:33 PM roycroft: that's why i set it to linuxcnc, and not chickenz
03:33 PM roycroft: because we all know that jt wants to use chickenz as the rood password
03:33 PM travis_farmer: lol
03:33 PM roycroft: so now he can set it to that safely, knowing that the hash was not posted
03:34 PM JT-Shop: nice unable to mount any usb drive I try :(
03:34 PM travis_farmer: hash is a one-way hash anyway, IIRC
03:35 PM travis_farmer: huh?
03:35 PM * Tom_L pulls up a chair with some popcorn to watch
03:36 PM * roycroft is not entirely sure he understands the fundamental problem
03:36 PM JT-Shop: hand me the .44
03:36 PM JT-Shop: the fundamental problem is I can't login to the BP mill at boot up
03:36 PM roycroft: if it's not letting you log in as root because there is no root password, why can't you log in as "user", and sudo from that?
03:36 PM roycroft: well that's the symptom
03:37 PM JT-Shop: no clue, can't use recovery mode either
03:37 PM roycroft: i thought the problem was that there is no root password, and it's not allowing logins as root without a password
03:37 PM JT-Shop: that's for recovery mode
03:37 PM roycroft: oh
03:37 PM JT-Shop: I did copy everything from the drive "I think"
03:38 PM JT-Shop: so reinstall could be the answer
03:38 PM travis_farmer: that does not sound very exciting...
03:40 PM travis_farmer: that was an inoppertune IRC disconnect... (had to review the logs)
03:40 PM JT-Shop: debian 9 live cd seems to have a problem with usb drives so stepping up a notch
03:41 PM JT-Shop: time to spoil most of the chickens... one is in the clink
03:48 PM rigid: does microstepping add variance? I calibrate 10mm movement down to 0.001mm accuratly. But when I move in 1mm steps, every step is off
03:49 PM rigid: when I move 1mm ten times, it's accurate again
03:50 PM rigid: i'm using 1/32 microstepping currently
03:50 PM rigid: cheapo DRV8825 drivers
04:01 PM roycroft: i'm no expert on this, but i've read that some drivers do not do microstepping properly
04:01 PM roycroft: so it may be an issue with the drivers
04:04 PM XXCoder: keep in mind that microstepping dont add any precision past certain point
04:04 PM Tom_L: JT-Shop, no way to ssh in?
04:04 PM Tom_L: probably not
04:05 PM XXCoder: stepper starts drafting a little, even skip steps at small enough microstep. it averages out but yeah there can be error
04:05 PM roycroft: "check out our high-yield savings accounts - we pay up to 0.50% apr"
04:05 PM Tom_L: in my mind, the steppers still cog and if the micro steps aren't enough to keep it where it should be, it will cog where it want's to go
04:05 PM roycroft: this in an email i just got from a credit union where i keep some of my accounts (not any savings accounts, other than the required shares account with the required $5)
04:06 PM roycroft: xxcoder: microstepping should not lose precision, though
04:06 PM Tom_L: roycroft, you should invest the $5 in that..
04:07 PM roycroft: my savings account at marcus, with no restrictions, is paying 4.40% currently
04:07 PM roycroft: that's where most of my savings is held
04:07 PM roycroft: that's better than most cds
04:08 PM rmu: rigid: sounds like variation in the leadscrew. how did you calibrate to 1µm accuracy? i doubt it.
04:08 PM JT-Shop: Tom_L, ssh is not turned on until you login as far as I can tell... also does not broadcast on the lan
04:08 PM roycroft: sshd is a system service that is enabled at boot time generally
04:09 PM roycroft: but it is not automatically enabled on installatin
04:09 PM roycroft: you have to enable the service when you install the os
04:09 PM rigid: i'm no expert either but if I understand microstepping correctly, it doesn't add accuracy at all, just granularity. This driver steps a sine wave so each step must have different height by definition
04:09 PM roycroft: correct, rigid
04:09 PM roycroft: unless the driver is broken
04:09 PM XXCoder: roy one of issues is much weaker holding torque for microstepping. it goes worse as you go 1:1 to 1:32
04:09 PM roycroft: in which case it adds granualarity but reduces accuracy
04:09 PM roycroft: from some reports i've read
04:10 PM rmu: rigid: the motor does take positions between full step positions, but linearity can be so-so and according to gecko drives it can't really resolve more than 10 positions between full steps
04:10 PM roycroft: gecko drives are supposed to be especially good at microstepping
04:10 PM rigid: rmu: the leascrew is the least I suspect. my 0.01 measure dial thingy matches exactly. might not really be 0.001 but 0.005 easily
04:10 PM roycroft: and xxcoder is correct - microstepping reduces holding torque
04:10 PM rigid: rmu: that makes sense
04:11 PM rigid: yeah, torque is not an issue until now
04:11 PM rigid: just crank up the current :-P
04:11 PM rmu: holding torque at intermediate positions
04:11 PM rigid: rmu: there's no force applied. so holding uses zero energy
04:12 PM rigid: *should use
04:12 PM rigid: i guess i'll try 1/4 microstepping tomorrow just to be sure
04:12 PM XXCoder: hold force is highest when its not moving
04:12 PM XXCoder: and highest current and heat
04:12 PM rigid: yeah, but there is no stepping
04:13 PM rmu: rigid: if it is systematic error that somehow magically disappears every 10mm i guess something else is wrong, like some scale factor
04:13 PM rigid: why should current when holding without force use more current than moving with force present?
04:13 PM XXCoder: any backlash?
04:14 PM rigid: rmu: right. that measure dial thingy just does 10mm. otherwise that'd be totally the next thing to try.
04:14 PM rmu: stepper motor is open loop so motor current doesn't depend much on applied load
04:14 PM rigid: XXCoder: backlash from what? it's the X axis moving back and forth
04:14 PM XXCoder: theres tons ponental sources
04:14 PM XXCoder: like nut can move a little on screw without screw rotating
04:14 PM rmu: there's a reason stepper motors get very hot even if not moving
04:15 PM rigid: rmu: you mean the motor uses the same current when I hold the tool while moving than when i don't hold it?
04:15 PM rigid: my current metere says otherwise
04:15 PM rmu: rigid: it depends on the driver
04:16 PM rigid: the reason is that the magnetic coil is basically a shortcut and non-stepper motors don't draw current when standing still
04:16 PM rigid: well, they do but you can't really hold them like with steppers
04:17 PM rigid: iirc BLDC can also brake
04:18 PM rigid: s/shortcut/short circuit/g
04:19 PM acer: rigid what kind of a machine are you micro stepping?
04:20 PM acer is now known as _unreal_
04:20 PM _unreal_: rigid, ?
04:20 PM rigid: _unreal_: atm simple XYZ china mill
04:21 PM rigid: but I know those NEMA 17s can do better
04:21 PM _unreal_: ok and what is the issue? also
04:21 PM _unreal_: "non-stepper motors don't draw current when standing still " depends on in general not true
04:21 PM rigid: I calibrate 10mm movement down to 0.001mm accuratly. But when I move in 1mm steps, every step is off. when I move 1mm ten times, it's accurate again
04:22 PM Tom_L: JT-Shop, clink being death row?
04:22 PM _unreal_: rigid, you want to avoid micro micro stepping as much as possible
04:22 PM JT-Shop: no, just isolated from the rest
04:22 PM JT-Shop: freezer camp is death row
04:22 PM _unreal_: micro stepping will get rid of vibrations and slow speeds
04:22 PM Tom_L: JT-Shop, you see the pr?
04:23 PM JT-Shop: yes, thanks
04:23 PM rigid: really? how does adding more steps increase speed?
04:23 PM _unreal_: it will "add" resolution. but at the extrema expense of holding torque.
04:23 PM roycroft: we've already been over this :)
04:23 PM rigid: torque is not an issue
04:24 PM _unreal_: <rigid> really? how does adding more steps increase speed? where did I say more steps increases speed
04:24 PM rigid: and I suspect that, since it's a sine wave, microstepping also takes a toll on accuracy.
04:24 PM _unreal_: adding more steps adds resolution. I.E. PIXILATED VERSE 4k pictured quality
04:24 PM rmu: rigid: what is the error you see
04:24 PM _unreal_: when micro stepping
04:25 PM rigid: (2023-09-25 23:19:42) _unreal_: micro stepping will get rid of ... slow speeds
04:25 PM roycroft: i think you've empirically determined that with your setup, rigid
04:25 PM roycroft: that is what you were complaining about in the first place
04:25 PM rigid: rmu: different error every 1mm ... sometimes just 0.1, sometimes 0.3
04:25 PM _unreal_: oh I see its AT slow speeds
04:25 PM _unreal_: not and slow speed
04:25 PM Tom_L: ..this should take the rest of the day..
04:25 PM _unreal_: higher micro stepping levels will remove vibration
04:26 PM rmu: rigid: can't really be the same if it cancels out after 10mm, can it
04:26 PM _unreal_: I can under stand the confusion rigid :)
04:26 PM rigid: roycroft: yeah, but i'm not sure about the cause. with this cheap stuff, everything is possible :)
04:26 PM Tom_L: don't get the cheap stuff :)
04:26 PM rigid: and i'm not sure if microstepping is the most likely cause... but almost
04:26 PM rmu: how many steps per mm?
04:26 PM _unreal_: rigid, what is the main issue you have?
04:27 PM roycroft: based on what you've described, i would not use microstepping until you can get some decent drives like geckos
04:27 PM roycroft: and then have another go at it
04:27 PM roycroft: maybe get one gecko drive to start with to see if that helps
04:27 PM rigid: Tom_L: sure, but you see those NEMA-17 steppers and leadscrews in printers/scanners doing quite high resolutions
04:27 PM roycroft: and if it does, replace the rest
04:27 PM _unreal_: micro stepping I'd in general go no higher then 1/16th stick to 1/4
04:28 PM rigid: roycroft: yeah, i'll try without and with 1/4 microstepping. if that's the cause, I should see a difference
04:28 PM rmu: that may sound harsh, bu IME, anybody that claims to have 1µm accuracy with a self-cooked stepper system doesn't know what they are talking about
04:28 PM _unreal_: rigid, what is the issue. I am zipping though the chat but have not found the main issue.
04:29 PM _unreal_: 1µm LOL
04:29 PM _unreal_: rigid, most printers scanners use DC motors with an encoder strip
04:29 PM rigid: rmu: well, obviously it's not 1um accuracy :) but the needle hits the dial mark rock solid at +/- 10mm ... i'd say it's _at least_ 0.05 (i can clearly spot half, 1/4th or 1/8th off dial mark)
04:30 PM rigid: and that dial gauge was not exactly cheap
04:31 PM _unreal_: I think I'm finally happy with my code https://pastebin.com/LFwpLny2
04:32 PM _unreal_: for anyone wondering, my code is for an ATTINY85. to drive 2 PWM outputs in series operation from one POT.
04:32 PM _unreal_: https://pastebin.com/LFwpLny2
04:32 PM rmu: how many steps per mm do you have
04:32 PM rmu: (full steps)
04:33 PM _unreal_: rmu, I think my little machine is 92 steps per mm.
04:33 PM _unreal_: :)
04:35 PM rmu: the motor also may be inaccurate, if you have 5mm leadscrew, your 10mm that are exact are 2 full rotations so won't reveal motor unlinearity
04:36 PM rmu: 1mm would be 40 full steps on a typical 200steps/rot hybrid stepper, so may be all over the place depending on accuracy of pole spacing in the motor
04:50 PM rigid: no difference without microstepping
04:50 PM rigid: the accuracy is even worse
04:51 PM rigid: did that go through the disconnect? -> no difference without microstepping
04:53 PM rigid: rmu: 1.01mm is exactly 50 full steps. 2.03mm is 100 and 3.08mm is 200
04:53 PM rigid: could even the cheapest stepper be that non-linear?
04:54 PM Tom_L: are you measuring at the stepper or somewhere else
04:54 PM Tom_L: the somewhere else could be the fault
04:56 PM rmu: 200 steps is one full rotation? then your lead screw is off
04:56 PM rigid: hm, scope & logic analyzer might be the logical next step. i never debugged CAM stuff before
04:56 PM * rigid would rather patch linuxcnc code :)
05:13 PM rigid: rmu: doesn't seem so. one rotation of the spindle is exactly 4mm for a lot of rotations
05:13 PM rigid: 32mm to be exact
05:14 PM rigid: 8 rotations actually
05:22 PM rmu: rigid: have a link to the stepper motor?
05:23 PM rmu: 50 full steps for 1mm, 100 full steps for 2 amd 200 full steps for 3mm can't be
05:25 PM rigid: not exactly 50 steps. 50 is 0.99mm, 51 is ~1.01mm
05:25 PM rigid: it's a brandless one. It's this CNC https://www.amazon.de/dp/B07G55SNZW
05:26 PM rmu: so when 50 steps are 1mm, 150steps should get you 3mm, not 200 steps
05:26 PM rigid: lemme check again
05:30 PM rigid: yeah, i meant to write 150 not 200
05:30 PM JT-Shop: I see two identical files shadow and shadow- and root:!:
05:32 PM Tom_L: what is vipw?
05:32 PM Tom_L: The best way to edit /etc/passwd, or shadow or group file is to use vipw command. Traditionally (under UNIX and Linux) if you use vi to edit /etc/passwd file and same time a user try to change a password while root editing file, then the user's change will not entered into file.
05:33 PM rigid: Tom_L: sorry, missed that. i measure the tool holder against the x axis
05:34 PM Tom_L: JT-Shop, https://www.cyberciti.biz/tips/what-is-the-best-way-to-edit-etcpasswd-shadow-and-group-files.html
05:34 PM rigid: it really looks like rmu is right and the motor is too nonlinear
05:34 PM Tom_L: what to put in them is between you and roycroft :)
05:35 PM rigid: traditionally "usermod" or "passwd" are used to edit shadow
05:35 PM Tom_L: if you're logged in
05:35 PM Tom_L: which is the catch 22
05:35 PM rigid: how would you run anything without being logged in?
05:35 PM Tom_L: live cd
05:36 PM Tom_L: if he were logged in the problem would be solved
05:36 PM rigid: ah... just copy/paste the line from one system to another
05:36 PM Tom_L: this is on a mill
05:36 PM rmu: chroot
05:36 PM rigid: or chroot
05:36 PM rmu: then use passwd
05:37 PM rigid: Tom_L: you want to change the password?
05:37 PM Tom_L: i don't want to do anything. it's jt's issue
05:37 PM rigid: ah
05:37 PM Tom_L: but he's unable to login
05:38 PM rigid: JT-Shop: you can just change the password of any user on any linux, copy the hash and replace the hash of the user on your mill. do you have another linux available?
05:38 PM Tom_L: that may be what he's working on
05:38 PM Tom_L: or enabling root
05:38 PM rigid: easiest would be to enable passwordless login and then use passwd
05:40 PM rigid: which is :: in passwd and shadow
05:41 PM rigid: (or :x: in passwd and :: in shadow, depending on the *nix flavor)
05:42 PM JT-Shop: hmm I have a terminal now...
05:45 PM rigid: great, run "passwd" to set a new password
05:46 PM JT-Shop: failed to change the password auth error
05:47 PM rigid: what kind of terminal do you have?
05:47 PM rigid: is "whoami" saying root?
05:47 PM JT-Shop: my prompt is root@(none):/#
05:47 PM JT-Shop: yup I'm root
05:48 PM rigid: does it prompt for an old password?
05:48 PM JT-Shop: no
05:48 PM JT-Shop: I added "init=/bin/bash" to the end of the parameter list
05:49 PM rigid: yeah that works. and if you run passwd, you enter your new password two times and it says exactly "failed to change the password auth error"?
05:49 PM JT-Shop: no not exactly
05:50 PM JT-Shop: the mill is not in my view from here
05:50 PM rigid: first thing you should do after this is setup ssh + public key :)
05:50 PM JT-Shop: I have ssh setup on all my pc's on the lan
05:51 PM rigid: i never had passwd say something like this. might be some additional security mechanism. can you "nano /etc/passwd" to edit the file?
05:51 PM rigid: yeah, but not on the mill obviously
05:52 PM JT-Shop: and the bp mill does not show up on the lan with angry ip scanner
05:52 PM rigid: when the system is fully booted, "ifconfig" will tell the IP
05:53 PM rigid: *all IPs
05:53 PM JT-Shop: I can view the file but can't edit it
05:53 PM JT-Shop: ifconfig command not found
05:54 PM rigid: your system isn't booted, yet
05:54 PM rigid: maybe the partition is mounted read-only. you could try "mount / -o remount,rw" and then try passwd again
05:55 PM rigid: wait, are you booting from a CD?
05:55 PM rigid: those usually come read-only :-P
05:56 PM JT-Shop: atm I'm booted into the hard drive
05:56 PM JT-Shop: time for dinner here putz with it tomorrow
06:32 PM roycroft: ifconfig is in /usr/sbin on debian 9, but it's not installed by default (stupid debian)