#linuxcnc Logs
Sep 26 2020
#linuxcnc Calendar
12:53 AM Deejay: moin
01:02 AM pink_vampire: hi
01:05 AM XXCoder: hey
01:05 AM XXCoder: been a bit
01:05 AM XXCoder: hows ya
01:05 AM pink_vampire: yeah..
02:00 AM CaptHindsight: skunkworks: wutz weird aboudit?
02:01 AM CaptHindsight: https://forum.linuxcnc.org/media/kunena/attachments/1324/blacklist-gpu.txt
02:01 AM CaptHindsight: https://pastebin.pl/view/2f57585f
02:02 AM CaptHindsight: https://forum.linuxcnc.org/18-computer/36879-raspberry-pi-4?start=440#183363
04:12 AM Tom_L: morning
04:13 AM XXCoder: hey
04:27 AM Deejay: re
04:37 AM centreline: Morning all
05:43 AM JT-Cave: morning
05:43 AM XXCoder: heys
08:47 AM -!- #linuxcnc mode set to +v by weber.freenode.net
09:26 AM skunkworks: CaptHindsight: then I am doing something wrong. It didn't seem to effect the latency
09:55 AM jymmmm: morning
10:28 AM CaptHindsight: skunkworks: did video playback get choppy?
10:31 AM CaptHindsight: do you have isolcpus=1,2,3, idle=poll in cmdline.txt
10:31 AM CaptHindsight: arm_freq=1500 in the config.txt
10:31 AM CaptHindsight: force_turbo=1 goes into the config.txt file, not the cmdline.txt
11:14 AM Tom_L: how hot does that make it run?
12:06 PM CaptHindsight: Tom_L: Rpi4 needs a heatsink to run properly
12:07 PM CaptHindsight: force turbo is supposed to keep the cores from speed changing
12:07 PM CaptHindsight: arm_freq=1500 is the default speed for the Rpi4
12:08 PM CaptHindsight: i list it anyway so hopefully it won't change freq on me
12:08 PM CaptHindsight: if the firmware wasn't crap it would be helpful, but Boradcom so FU
12:16 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/Rpi4/Rpi4_top.jpg
12:16 PM Tom_L: seems to do rather well
12:16 PM Tom_L: would be nice if it was wider for the header
12:20 PM Loetmichel: Hmpf. Just cleaned my spam drawer out at gmx... those "girl wants to meet you" spams getting really good lately. Took me alsmost the whole mail to notice its spam and not geniue. Impeccable german, even some personal info strewn in (i wonder how that leaked)... wouldnt it be for the obvious promotion of some "partner service" website at the end i wouldnt have caught it. That said: why do
12:20 PM Loetmichel: they even bother with a happily married man in his 50s? Do they really think one would belive that a 20yo girl would fall for a grumpy old grandpa?
12:20 PM Tom_L: how much money do you have?
12:21 PM centreline: I get german spam like that since I moved our mail server to a german provider, & I dont speak german :D
12:23 PM Loetmichel: Tom_L: enough ;)
12:23 PM centreline: you can check if its a mail account thats leaked at https://haveibeenpwned.com/
12:24 PM centreline: I know it sounds scammy to put your mail in a site like that (yes I am paranoid) but Troy Hunt is the real thing, and its not to resell your email details.
12:25 PM centreline: my day job is dealing with stuff like this...
12:28 PM Tom_L: we get as many phone scams as email if not more
12:29 PM CaptHindsight: Tom_L: I use a similar heatsink, I can overclock to 2100MHz
12:31 PM Tom_L: i wonder if anyone but sam is using an rpi4 on real hardware
12:33 PM pcw_home: I've been running a test system for 6 months or so but no mechanical hardware
12:34 PM pcw_home: and a RPI3 for a year or so previous
12:34 PM Tom_L: i've had mine on the mill but just for tests
12:34 PM Tom_L: and not since the 2.8 release
12:35 PM Loetmichel: centreline: i know that site. and according to it my mail was breached by 5 different websites... none of which i have an account.
12:35 PM Loetmichel: so funny how they got my mail adress
12:36 PM centreline: you might not have had a account on the actual sites, but they had your user details in their databases and leaked it out, sometimes its not as simple as X mail address used for Y site is in Y's database
12:37 PM centreline: troy finds dumps on the internet, and adds them to the haveIbeenpwned database under the provider they leaked from
12:39 PM centreline: but if the site names aren't familiar to you, I dont know of a easy way to search down the link between them and one you do know sorry
12:44 PM Loetmichel: centreline: no problem. As i change my PWs often it is of no consequence either way.
12:44 PM Loetmichel: was just curious
12:44 PM centreline: I dont know my passwords for websites, I use keepass-xc and a yubikey to generate a new one for each place
12:45 PM Loetmichel: hehe, i tend to use brain 1.0. but i do change them often, and have them personalized for each website.
12:45 PM centreline: I had brain overflow a few years back and it cured it for me (I have a lot of creds for work etc too)
12:48 PM centreline: if someone wanted my creds enough, they'd just come round my place and beat me till I unlock it all though :D
12:48 PM centreline: "rubber hose decryption method"
12:49 PM Loetmichel: aka .> https://xkcd.com/538/
01:21 PM centreline: yes xkcd has it covered always. just pushed my max vel up and have familiar noises from machine when moving, welcome back old friend...
01:54 PM centreline: so I need to wire ena to the steppers via a relay to invert the enable logic, so if I have a hardware crash the 7i76 outputs fail low and disable everything... low
01:54 PM centreline: thinking out loud here, but correct me if its wrong, please
02:28 PM Tom_L: centreline, are you using a mesa card or parport?
02:28 PM Tom_L: nm. you can reverse the logic in the hal file
02:30 PM centreline: but as I understand it if the pc crashes, the 7i96 fails active low on all pins
02:30 PM centreline: so to disable drive in that case, it would have to be something pulling a relay to 5v high, unless it was told not to or a crash
02:31 PM Tom_L: test it out
02:31 PM Tom_L: i wired mine to a relay :)
02:32 PM Tom_L: and i ended up at least on mine reversing the logic so on power up my drivers were disabled until out of estop and power on states were both true
03:03 PM Loetmichel: Haaach, industrial grade Ultraonic cleaners are a boot to glasses people... tried to clean my glasses for 5 minute straigt: still fat in the corners of the lenses and stripes all over. 30 sec in the Sonic: squeaky clean! ;)
03:09 PM centreline: Tom_L, I will, currently experimenting with stepgen and accel parameters and ended up with maxaccel at 800, then hit jog 5mm and the table moved 0.5mm, so stepscale out by factor of 10 :)
03:10 PM centreline: this is vastly different to the config and timings I used with my parport bob setup...
03:20 PM robotustra: veegee, Toronto is not overcrouded and not overpriced :)
03:33 PM -!- #linuxcnc mode set to +v by ChanServ
03:34 PM skunkworks: CaptHindsight: all that.. (I migght be running 1800mhz or so
03:34 PM skunkworks: but latency shot up quicly with and went above 200us
04:06 PM Deejay: gn8
04:39 PM centreline: hmm just one last issue for the night, after homing the axis goes the wrong direction...
04:40 PM CaptHindsight: skunkworks: try arm_freq=1500 in the config.txt if it still runs at 1800 then their rpi configs are broken
04:42 PM centreline: dumb mistake I hadnt configured HOME_OFFSET on that one axis and it tried to go to 0 into a limit switch
04:42 PM Tom_L: centreline,
04:42 PM Tom_L: mmm
04:42 PM CaptHindsight: skunkworks: maybe our Rpi4's have different firmware
04:42 PM Tom_L: i was gonna say reverse scale in the ini
04:43 PM centreline: I had to do that on another, but put the - prefix on this one and wondered why it had no effect
04:44 PM centreline: its late, I guess its that time where its safer to do whats left fresh tomorrow
04:44 PM centreline: ok gnite all, and pcw_home thanks again for putting me right on the stepper wiring
04:44 PM roycroft: well, my bench is essentially done
04:44 PM roycroft: i need to let the finish cure for a week or so, then i can rub it out one more time and apply some wax
04:45 PM roycroft: i also need to flatten the bench top, which i'll start to do in a week or so
04:45 PM roycroft: but that will be something that happens over time - the bench top will move for quite a while before it settles down
04:45 PM roycroft: so i'll be flattening it slowly
04:46 PM roycroft: it's only out of plane by about 1/16" as it sits, so it's quite usable as it is
04:46 PM roycroft: but i want to get it dead flat, or as close to dead flat as i can
04:47 PM roycroft: and speaking of dead, i still have to finish shaping the deadman and drilling it for whatever i use in it - holdfast or bench dogs, but that's a decision i'm putting off for a while, and that's only a couple hours of work
04:47 PM roycroft: i'm at just under 245 hours total time working on it, so i finished it within my time budget
04:48 PM roycroft: tomorrow i will rearrange the shop, move the old bench out of the way, and install the new one in its place
04:50 PM skunkworks: CaptHindsight: I am also using an older image for the rpi4
04:50 PM skunkworks: I get <100us as long as I don't run any video
04:52 PM robotustra: roycroft, https://www.youtube.com/watch?v=-MwCJpEuC44
04:54 PM roycroft: i don't care if it's level, robotustra
04:54 PM roycroft: i just want it to be as flat as i can reasonably make it
04:55 PM robotustra: there is a way how to make ideally flat table top
04:55 PM robotustra: with the help of router
04:56 PM roycroft: yeah, i could spend a couple hours setting up a router sled and then get it fairly flat
04:56 PM roycroft: or i could spend a half hour with a jack plane, a jointer plane, and a pair of winding sticks and get it even more flat
04:56 PM roycroft: without all the noise and mess
04:57 PM robotustra: I think you know this techunique https://www.youtube.com/watch?v=E_s8WCMu0GQ
04:58 PM roycroft: yeah
04:58 PM roycroft: i
04:58 PM roycroft: am not interested in that at all
04:58 PM roycroft: it takes too long, is way too messy, and way way way too noisy
04:59 PM roycroft: and doesn't leave a very nice finish - i'd have to plane it after anyway
04:59 PM robotustra: yep. do you have planes?
04:59 PM roycroft: a router sled is for people who don't know how to use a plane
04:59 PM robotustra: true
05:00 PM roycroft: www.roycroft.us/Planes.jpeg
05:00 PM robotustra: I use japanese plane
05:00 PM roycroft: i've added a low angle jack plane since then, but i can't use that on the bench top
05:00 PM roycroft: i like the planes i have
05:01 PM roycroft: the wooden plane to the left of the center on the bottom is a scrub plane
05:01 PM roycroft: with a pretty severe radius
05:01 PM roycroft: if i have some rough timber that is seriously twisted i can take down the high corners really fast with that
05:02 PM robotustra: I would use a violin technique to make the big surface flat
05:02 PM roycroft: it's impressive how big a chip it will take
05:02 PM roycroft: but it leaves a very corrugated surface
05:02 PM roycroft: which does not matter, because that's what the jack plane, jointer plane, and smoothing plane are for
05:03 PM robotustra: how do you determine what places to shave?
05:03 PM roycroft: one can flatten any board with just a no. 5 jack plane
05:03 PM roycroft: i rock it, eyeball it, and use winding sticks
05:03 PM rs[m]: https://www.osadl.org/Profile-of-system-in-rack-7-slot-2.qa-profile-r7s2.0.html?shadow=1
05:04 PM roycroft: if i have a large board what i usually do is take a few passes across each end with the jack plane first
05:04 PM roycroft: to get them flat
05:04 PM roycroft: then i use winding sticks to determine how out of plane they are
05:04 PM roycroft: then i work on the high corners
05:04 PM roycroft: flatten the ends again
05:04 PM robotustra: I think I know better way
05:04 PM roycroft: repeat until they are coplanar
05:04 PM roycroft: then i plane the edges straight
05:04 PM robotustra: you can mark it once and make it dead flat
05:04 PM roycroft: without touching he ends
05:05 PM roycroft: once that's done i have the entire perimeter flat
05:05 PM roycroft: then i just plane the middle until it's flat
05:05 PM roycroft: it works really well
05:05 PM robotustra: if you want I can share the idea
05:06 PM roycroft: sure, as long as you don't expect me to agree that it's a better way :)
05:06 PM roycroft: i may or may not agree
05:06 PM roycroft: but if i don't i'll let you know why
05:09 PM robotustra: 1) make a jig like for routing table 2) use awl instead of router, set the proper length, the lowest spot on your surface 3) pick holes of the surface evenly 4) shave with planer till the marks goes off
05:09 PM roycroft: that would probably work
05:10 PM roycroft: but it would take a long time
05:10 PM roycroft: i think my technique would be a lot faster
05:10 PM robotustra: as you wish
05:10 PM roycroft: your method still requires all the planing
05:10 PM roycroft: and in addition you have to set up the sled
05:10 PM roycroft: and level it
05:11 PM roycroft: and then make all the depth marks
05:11 PM roycroft: so it would work pretty well, but would be a lot more complicated than how i do it
05:11 PM robotustra: you don't need them too much - I think 2x2 inch is enough
05:11 PM roycroft: and there's no need to do all of that
05:12 PM roycroft: my jointer plane is 22"/56cm long
05:12 PM roycroft: and it's very very straight
05:12 PM roycroft: so if i make the two long ends of the board coplaner, i use them as guides for the jointer plane
05:13 PM roycroft: it will stop cutting when the middle of the board is in the same plane as the edges
05:13 PM robotustra: if I had 1/16" on a wooden top - I would probably stoped
05:13 PM roycroft: as i said, it won't take very long to flatten it
05:14 PM roycroft: a lot less time than it would take to set up any kind of sled/jig
05:14 PM roycroft: which is why i'm just going to flatten it
05:14 PM roycroft: i just reinstalled the hold down bolts
05:14 PM roycroft: and i just applied finish to the undercarriage
05:14 PM roycroft: once i move the bench to its permanent home i'll let it sit for a week to settle
05:15 PM roycroft: then spend a half hour flattening it
05:15 PM roycroft: and then in a month or so check it again
05:15 PM roycroft: over the course of the first year or so it will probably warp a bit here and there
05:15 PM roycroft: but it should not take long to flatten it if i keep at it
05:16 PM roycroft: the thing is, if i want to flatten a board i need to have a flat surface on which to plane it
05:16 PM roycroft: so it is important for the bench top to be as flat as is practical to make it
05:22 PM robotustra: well, I have to make some square nuts to mount my part on a turn table
05:29 PM JT-Cave: roycroft, do you use a scrub plane to rough flatten the surface?
05:39 PM skunkworks: CaptHindsight: for me - I get the best latency with the fake kms driver (but still cannot run youtube - then it is >200us)
05:42 PM roycroft: yes
05:42 PM roycroft: it hogs off huge amounts of material
05:42 PM roycroft: so i use it to get the board fairly coplaner
05:42 PM roycroft: it's especially useful with twisted boards
06:03 PM CaptHindsight: skunkworks: I am using the Andy's img off the LCNC site https://www.linuxcnc.org/iso/linuxcnc-2.8.0-pi4.zip
06:35 PM veegee: robotustra downtown Toronto is horrible. Anywhere else is outside the core is not crowded or overpriced
06:35 PM veegee: but no french, so that alone should be a reason
06:36 PM veegee: Even just 7km linear distance away from downtown is fine
06:37 PM veegee: the greater toronto area is the whole point, it has the highest availability of anything you want
06:37 PM veegee: If it's in Canada, it'll be in the GTA
07:04 PM robotustra: veegee, imagine - I don't use French in Montreal
07:04 PM robotustra: so I don't care
07:06 PM robotustra: The island of Montreal actually split into 2 parts, East Island and West Island, which is English speaking mainly, where I live,
07:08 PM robotustra: I don't intersect with french people much, and if there is so - if they want something from me, like managers from work - they have to speak English anyway.
07:12 PM veegee: lol! the anti french comment is more light hearted humour
07:12 PM veegee: I have nothing against them
07:12 PM veegee: I can speak french anyway
07:13 PM CaptHindsight: skunkworks: looks like i should post the firmware version used as well
07:19 PM robotustra: veegee, I compare french to some local dialect, which is useless if you want to communicate with clever people, from outer world
07:20 PM veegee: yeah
07:21 PM CaptHindsight: I wish they would use French down here
07:22 PM CaptHindsight: they do in a small area in Louisiana
07:22 PM CaptHindsight: but it's mostly English and Spanish
07:25 PM robotustra: According to principle of minimal action people should intercommunicate the simplest language IMHO
07:29 PM XXCoder: well rounded people knows 2 languages at least
07:29 PM XXCoder: and not similiar language either
07:36 PM robotustra: I already know 4, and want to study 2 more, but I don't think I'll learn them to the extent I can freely talk to people with thin languages, and I repel this idea
07:37 PM robotustra: 4 is enough.
07:37 PM robotustra: I have to leave some space to study machining :)
07:38 PM XXCoder: my brother is learning swedish language
07:38 PM XXCoder: pretty cool
07:38 PM XXCoder: its 10 times harder for deaf lol
07:38 PM XXCoder: (yes my bro is deaf too)
07:39 PM robotustra: sign language is already a complex one
07:54 PM robotustra: nice https://www.youtube.com/watch?v=bXrnqcNgL00
07:56 PM XXCoder: ugh cant stand that lol
07:56 PM XXCoder: hes constantly under tesion
08:01 PM robotustra: but a bit funny
09:01 PM BitEvil is now known as SpeedEvil
10:11 PM roycroft: why does it bug me so much when abom alwasy says "michitoyuh"?
10:13 PM Tom_L: should he be saying starret instead?
10:14 PM roycroft: he says that a lot
10:14 PM roycroft: but he pronounces starrett fairly accurately
10:14 PM roycroft: mitutoyo, however, seems out of his grasp, so he says something completely different
10:15 PM Tom_L: so it appears that motion.spindle-brake is asserted on M5 but not the transition from M3 to M4
10:15 PM Tom_L: ie during a rigid tap cycle
10:17 PM cradek: that's sure what I'd expect
10:17 PM cradek: is it not what you expect?
10:17 PM Tom_L: my control says to assert brake before reversing direction
10:18 PM Tom_L: so i don't know how serious to take that
10:18 PM Tom_L: tap cycles aren't going to be high rpm
10:18 PM cradek: weird
10:19 PM cradek: yeah and reversing smoothly is more important than reversing quickly
10:19 PM Tom_L: i'll have to do some testing
10:19 PM Tom_L: the way i set up the sherline, it took an eternity to reverse but it did it just fine
10:21 PM Tom_L: so on every spindle stop the brake is going active
10:22 PM Tom_L: and if the vfd needs brake asserted before a direction change i could add a one shot pulse to it on either transition maybe
10:31 PM Tom_L: i'm actually doing that on the sherline controller now since it really doesn't have reverse
10:31 PM Tom_L: the pulse hits the inhibit line which resets the control to do it's ramp up cycle before jumping to full speed
10:32 PM Tom_L: it's ramp time is set with a pot
10:32 PM Tom_L: http://tom-itx.no-ip.biz:81/~webpage/cnc/PDF/KBIC_Data_Sheet.pdf
10:32 PM Tom_L: fwiw
10:33 PM Tom_L: P10