#linuxcnc-devel Logs

Feb 13 2018

#linuxcnc-devel Calendar

07:50 AM JT-Shop: pcw_home: hows the 7i76E's coming along?
07:52 AM jepler: unfortunately, "the GPL, with the additional restriction that you must obey the intent of the GPL" is not a license compatible with the GPL.
07:52 AM jepler:  
07:53 AM jepler: fg
07:53 AM jepler: whoops
07:53 AM jepler: https://lwn.net/SubscriberLink/747124/e2426dfab33c2cf8/
09:47 AM jepler: huh I thought 'git' was on the livecd
09:58 AM skunkworks: I think I noticed that. Sorry - didn't say anything. Easy to get git.
09:58 AM skunkworks: and could you install the current dependencies to compile linuxcnc?
09:59 AM skunkworks: or is that pushing it :)
10:02 AM jepler: once you smash right past the CD size limit, the only thing you have to guide you is feelings about whether people mind downloading 4GB instead of 2GB, so that just a few people can have development packages already available
10:02 AM jepler: weird, I thought seb merged that patch to get rid of yapps2 as a build dep
10:03 AM jepler: oh wait I have it backwards, we removed yapps2 from the source tree so that we actually use the system version we'd been depending on all along
10:51 AM jepler: this USB stick is seriously slow while doing apt updates. sigh.
11:00 AM skunkworks: there are definitely speed differences in sticks
11:00 AM skunkworks: (especially when I buy the cheapest ones I can find)
11:22 AM jepler: I'm not surprised at that last bit
11:29 AM jepler: this chromebook (HP cr3114) seems to get good RT performance when booted with the live image on USB, but of course there are no I/O connectors that could actually be used to run hardware.
11:29 AM jepler: (it's a couple of years old 14" chromebook, Celeron 2955U processor, 4GB RAM)
11:31 AM jepler: maybe somebody super crafty could turn an SD card slot into an I/O connector for Mesa
11:40 AM skunkworks: jepler, no Ethernet connection?
11:40 AM jepler: skunkworks: no, only wifi
11:40 AM skunkworks: darn
11:40 AM jepler: wifi, USB, HDMI, SD, GSM SIM
11:40 AM jepler: power
11:42 AM jepler: supersonic signals via the audio codec, that's the ticket.
11:42 AM skunkworks: I am very happy with this w530. I use it a lot to test Ethernet configs
11:42 AM jepler: otherwise .. QR codes?
11:42 AM skunkworks: the cpu fan gets a workout :)
11:43 AM skunkworks: you could have a sensor that senses 16 pixels on the screen?
11:43 AM skunkworks: or even 32~
11:44 AM jepler: you'd need a lot more than that, since you can only update the pixels once per 17ms or so
11:45 AM skunkworks: darn ;)
11:45 AM skunkworks: https://forum.linuxcnc.org/38-general-linuxcnc-questions/34014-faceted-surface-on-g2-lathe-command
11:46 AM skunkworks: only other thing I can think of is stiction
11:46 AM jepler: I'm not in the "biz" of developing hardware drivers at the moment anyway
12:02 PM jepler: .. once I finally got build deps installed, it was just 7 minutes to "make -j2" linuxcnc
12:30 PM pcw_mesa: JT-Shop being built now but we have a large number of kits (13 kits and more than 2K boards) at the assy house so theres a fair backlog
12:50 PM JT-Shopp: pcw_mesa: thanks
12:50 PM JT-Shopp is now known as JT-Shop
12:52 PM pcw_mesa: they ( 7I76E's ) are about 3rd on our priority list so maybe in a week or two
12:53 PM pcw_mesa: also shoud have the sinking output variants or 7I76E, 7I76, 7I77, 7I84
12:54 PM pcw_mesa: "of" instead of "or"
12:57 PM JT-Shop: cool
01:44 PM jepler: skunkworks: could each rotation of the spindle be seen as a "facet", I wonder?
01:44 PM skunkworks: hmmm
01:45 PM skunkworks: I don't know - my first thought was he set the naive cam too high that it bisected the arcs (but that would have cause large facets..- but his gcode program has strait g65
01:45 PM skunkworks: *g64
01:51 PM jepler: like http://www.mini-lathe.com/Mini_lathe/Reviews/a2z_qctp/pt01.jpg but on the fillet instead of on the end..
01:52 PM jepler: I wasn't entirely sure what I was seeing in the OP's photo
01:55 PM jepler: oh are the facets we're talking about the ones around the circumference on the left hand portion of the image? that can't be due to segmenting an arc...!
01:56 PM jepler: seems more like some kind of tool chatter
01:59 PM jepler: .. I ordered a couple of "sandisk ultra fit" flash drives. They're marketed as high performance and don't protrude much, which hopefully makes 'em nice for semipermanent use in this chromebook
02:01 PM skunkworks: no - I am assuming it is the arcs on the right side as that is what the gcode shows
02:17 PM pcw_mesa: I vaguely remember some discussion about the fact that this was ignored but does
02:17 PM pcw_mesa: [TRAJ]
02:17 PM pcw_mesa: CYCLE_TIME = 0.010
02:17 PM pcw_mesa: do anything anymore?
02:20 PM pcw_mesa: also in his .ini file the ferror limits are really wide (26 mm!) , it might help debugging if they were narrowed to something reasonable just in case its a stepgen issue of some kind
02:28 PM jepler: [TRAJ]CYCLE_TIME? I don't think so.
02:29 PM jepler: [EMCIO]CYCLE_TIME [TASK]CYCLE_TIME and (dependent on UI) [DISPLAY]CYCLE_TIME do I think
02:30 PM pcw_mesa: Yeah I thought [TRAJ] CYCLE_TIME was a no-op now
02:32 PM pcw_mesa: if it did have an influence, 10 ms linear interp updates migh make for noticeble faceting (but maybe not on a slow machine like the Sherline )
02:34 PM jepler: I'm not sure I trust my search 100%, but I don't think [TRAJ]CYCLE_TIME is in any inifiles in master branch
02:34 PM skunkworks: his feed rate is about 2 inches a minute
02:34 PM jepler: it's probably an heirloom inifile passed down from grandmother to granddaughter
02:34 PM skunkworks: for the given part
02:35 PM skunkworks: So I don't think that is it. could be the following error - but I don't see anyting in the ini file that would be causing that.
02:35 PM skunkworks: unless there was a realtime delay enough to cause issues - but I would think the part would be way off.
02:35 PM skunkworks: sticksion
02:36 PM pcw_mesa: yeah 2 IPM is only ~1/3 mil every 10 ms
02:36 PM pcw_mesa: so yeah, sounds like mechanics
02:46 PM jepler: sigh, there's a new ubuntu LTS coming up in April. Feels like I just updated the forum to the LTS it's on. :-/ oh well, that was disruptive because of the php5 -> php7 transition, hopefully 18.04 is easier and I don't put it off for 18 months after LTS comes out.
02:47 PM pcw_mesa: I would still reduce the ferror limits ( because 26 mm is crazy ) maybe to 0.02,0.1 or some such
02:51 PM cradek: if he's not using CSS that's maybe just what happens when your surface speed changes a lot
02:52 PM jepler: I should learn to "lscpu" instead of "cat /proc/cpuinfo"!
02:52 PM cradek: are they obviously facets, or just different looking finish?
02:54 PM jepler: a lot of v2.7 stock configs have [TRAJ]CYCLE_TIME
02:54 PM jepler: so I shouldn't blame a user so hard for having it
02:55 PM jepler: and FERROR = 26 is also in our shipped files
02:55 PM jepler: geez
02:55 PM * jepler goes back to being semiretired from linuxcnc
02:56 PM cradek: I bet if he does the same path with css+fpr the problem goes away
02:56 PM * cradek goes with jepler
02:57 PM jepler: (and that's still in master, in case somebody wants to fix it)
02:59 PM jepler: once is happenstance, twice is coincidence, three times is stock configuration
03:21 PM skunkworks: I kinda mentioned the following error
08:31 PM jepler: make[1]: Entering directory '/home/jepler/src/zopfli-1.0.0'
08:31 PM jepler: gcc src/zopfli/*.c -O2 -W -Wall -Wextra -ansi -pedantic -lm -D_FORTIFY_SOURCE=2 -Wl,-z,relro -o zopfli
08:31 PM jepler: make[1]: Leaving directory '/home/jepler/src/zopfli-1.0.0'
08:31 PM jepler: now that's a buildsystem