Back
[01:49:17] <CaptHindsight> yikes! $89 for BBB IO board with an FPGA and 256MB of SDRAM
[02:10:58] <KGB-linuxcnc> 03Chris Morley 05master dba8981 06linuxcnc 10lib/python/gladevcp/hal_actions.py 10lib/python/gladevcp/hal_mdihistory.py gladevcp -teach hal_mdihistory and hal_action mdi about NO_FORCE_HOMING * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=dba8981
[07:14:18] <KGB-linuxcnc> 03John Thornton 05v2.5_branch af63901 06linuxcnc 10docs/src/ladder/classic_ladder.txt docs: add a note about loading only one classicladder clp file * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=af63901
[10:20:06] <KGB-linuxcnc> 03Sebastian Kuzminsky 05joints_axes4 34306e7 06linuxcnc 10configs/puma/puma560.ini 10configs/puma/puma560_sim_6.hal puma560: make the config load under ja * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=34306e7
[10:45:09] <KGB-linuxcnc> 03Sebastian Kuzminsky 05scorbot-er-3 0c09231 06linuxcnc 03src/hal/user_comps/scorbot-er-3 scorbot-er-3: start adding a driver * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=0c09231
[10:45:09] <KGB-linuxcnc> 03Sebastian Kuzminsky 05scorbot-er-3 5afc5c7 06linuxcnc 10(6 files) scorbot-er-3: start making a config * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=5afc5c7
[11:04:05] <skunkworks__> scorebot-er?
[11:26:53] <seb_kuzminsky> skunkworks__: scorbot-er-3:
http://highlab.com/~seb/linuxcnc/scorbot-er-3/
[11:27:26] <skunkworks__> Nice!
[11:34:06] <skunkworks__> how does it interface now?
[11:35:55] <pcw_home> I _am_ working on the 6 axis Hbridge card but it will still be a while
[11:37:42] <pcw_home> Do you have the motor specs?
[11:39:14] <seb_kuzminsky> heh
[11:39:18] <skunkworks__> ooh - I have a puma....
[11:39:46] <seb_kuzminsky> i'll look up the motor specs next time i'm in the hack space - i'm pretty sure they're little 12V pittmans
[11:39:57] <seb_kuzminsky> skunkworks__: the interface right now is awful
[11:40:26] <seb_kuzminsky> there's a control box (visible under the monitor in the first picture)
[11:40:40] <seb_kuzminsky> it speaks rs-232 at 9600 bps to the PC
[11:41:10] <seb_kuzminsky> there's a goofy little ascii protocol for controlling it, implemented in the python hal component in that branch i just pushed
[11:41:36] <seb_kuzminsky> you basically tell the control box what encoder count you want each servo to go to, and it tries to do it for you
[11:41:44] <seb_kuzminsky> you can also query the limit switches
[11:42:35] <seb_kuzminsky> i hooked it up and got it running, and the other people in the hackspace were like "i thought linuxcnc didnt support serial" and i had to hem and haw about how it's just a prototype for testing ja and nontrivkins
[11:43:06] <jepler> how many positions per second can you command at 9600 bps?
[11:43:12] <jepler> 10 or 20?
[11:43:22] <jepler> less?
[11:43:43] <seb_kuzminsky> more
[11:44:17] <jepler> bbl
[11:45:10] <seb_kuzminsky> 9600 bps is ~100 bytes/second (they use TWO stop bits), and each servo position command is 4 bytes + the position you want, maybe 3-4 bytes, so call it 8 bytes per servo update
[11:45:17] <seb_kuzminsky> so i guess you're right, 10-20 ;-)
[11:45:52] <seb_kuzminsky> plus you need to waste time asking about the limit switches
[11:46:00] <seb_kuzminsky> it's not awesome
[11:46:21] <skunkworks__> I thought linuxcnc couldn't use serial....
[11:46:35] <seb_kuzminsky> and eventually i want to replace that control box with a little bit of hostmot2 stuff, and close the servo loop in linuxcnc the way it should be done
[11:46:41] <seb_kuzminsky> heh
[11:46:45] <skunkworks__> :)
[11:47:04] <seb_kuzminsky> yeah, and that robot's position refresh frequency demonstrates why ;-)
[11:47:19] <seb_kuzminsky> it's the kind of "can't" that means "shouldn't"
[11:47:29] <seb_kuzminsky> bbl
[12:38:59] <KGB-linuxcnc> 03Dewey Garrett 05dgarr/ngcgui-gcmc c4af98d 06linuxcnc 10(23 files in 6 dirs) ngcgui,pyngcgui: support gcmc files * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=c4af98d
[12:39:54] <dgarr> any ideas on this message (didn't appear with --dry-run):
[12:39:55] <dgarr> remote: fatal: bad object 0000000000000000000000000000000000000000
[12:39:56] <dgarr> To ssh://dgarr@git.linuxcnc.org/git/linuxcnc.git
[12:39:56] <dgarr> * [new branch] dgarr/ngcgui-gcmc -> dgarr/ngcgui-gcmc
[12:40:29] <cradek> that happens when creating new branches and can be ignored
[12:40:39] <dgarr> ok -- thanks
[12:40:41] <cradek> it's a bug in one of the hooks that trigger emails/buildbot/kgb/etc
[12:56:47] <skunkworks__> $ git clone --branch unified-build-candidate-3 --origin github-mah
https://github.com/mhaberler/linuxcnc.git [<directory>]
[12:56:49] <skunkworks__> heh
[12:56:58] <skunkworks__> pcw_home,
http://imagebin.org/283038
[12:58:32] <skunkworks__> mhaberler, the latencyhistogram doesn't seem to work with rt_preempt.. (or I am doing something wrong)
[12:58:49] <skunkworks__> it has gone up a bit but under 50us
[12:59:40] <mhaberler> "doesn't seem to work".. forensics?
[12:59:54] <skunkworks__> sorry - look at the screen shot..
[13:00:09] <skunkworks__> says rtapi module must be loaded..
[13:00:58] <pcw_home> Also the latency test under Preempt-RT seems a bit buggy (max times go down sometimes)
[13:01:02] <skunkworks__> latency-test seems to run as expedtec
[13:01:07] <skunkworks__> expected
[13:02:16] <skunkworks__> pcw_home, are you running 12.04?
[13:02:37] <pcw_home> Yes
[13:02:43] <skunkworks__> 32 or 64?
[13:02:49] <pcw_home> 32
[13:03:05] <skunkworks__> This is 64 - should be interesting to see if the 7i80 branch works
[13:03:48] <pcw_home> works for me but Ethernet latency depends a lot on hardware
[13:04:05] <mhaberler> skunkworks: do a 'realtime start' and then try to run it
[13:05:30] <skunkworks__> This has the RTL 8168 card - so I have a intel pro/100 plugged in
[13:06:36] <skunkworks__> mhaberler, that worked! thanks
[13:07:09] <mhaberler> it would be great if authors of such utilities gave ubc3 a minimal try - some code has rather narrow environmental assumptions, and that is one of them
[13:10:20] <mhaberler> this might be worth a try for the 8168 problem:
https://enc.com.au/2013/10/16/damn-you-unworking-r8168/#comment-2380 - if it works, it means no extra driver build needed
[13:13:02] <skunkworks__> mhaberler, in the emails about 7i80 - am I understanding that if I just grab and build the ub3 - I have the ethernet mesa stuff?
[13:21:58] <mhaberler> I think micges' latest (rt-preempt) is
http://git.linuxcnc.org/gitweb?p=linuxcnc.git;a=shortlog;h=refs/heads/ubc3-7i80
[13:22:11] <mhaberler> I havent tried this
[13:22:46] <mhaberler> this
https://github.com/mhaberler/linuxcnc/commits/ubc3-7i80-rtnet is his previous xenomai/rtnet work
[13:24:33] <skunkworks__> mhaberler, stupid question - how do I switch to that branch (rt-preemt)
[13:25:21] <mhaberler> there is no 'preempt' branch - this is the unified build for all kernels
[13:25:49] <mhaberler> all you need is an rt-preempt kernel running, then build and start it
[13:25:54] <mhaberler> the right thing will happen
[13:26:26] <mhaberler> was that what you meant to ask?
[13:29:10] <mhaberler> what I meant is - micges code will work with rt-preempt kernel only; the other stuff should work on any kernel unchanged
[13:30:21] <andypugh> Yesterday I pushed a script that changes the HAL pin names in config files to reflect the JA3 changes. I just remebered that there are also smart-serial pin name chabges (for the 7i64) between 2.5.3 and master. Would it be sensible to roll those changes up in the same script, or should I push a different version of the script for the 2.5-to-2.6 changes?
[13:31:23] <cradek> I don't think we know yet whether we'll merge ja4 before 2.6 but it's looking unlikely to me.
[13:31:34] <skunkworks__> right - I understand that - I have preempt running. I just pulled your master ubc3 and made it. it seems to work - but the the hm2_eth isn't there - so I need to switch to that branch. (Which I think I figured out..) building now.
[13:31:38] <cradek> previously we've just given instructions for changes like you're talking about
[13:32:07] <cradek> so I'm not too sure whether a script is appropriate -- but if it is, it should be separate from the ja4 config-updating script
[13:32:42] <andypugh> Are you also saying that you are unconvinced about the advisability of the JA4 script?
[13:33:19] <cradek> not really. that's a much more invasive type of change
[13:33:28] <cradek> I don't know enough about the sserial change to know how big it is
[13:33:40] <andypugh> Bah! I was just marshalling my robust arguments :-)
[13:34:03] <skunkworks__> andypugh, how much torque does that washing machine motor seem to have with the 2.2kw driver?
[13:34:12] <cradek> if we start having versioning (like you've done) it makes auto-updating, even for little things, a lot more realistic
[13:34:46] <andypugh> the sserial change is just that the current driver uses the canonical pin naming convention, and the driver in master uses the same (generic) code as the other sserial devices that came later, and copies their naming convention.
[13:35:32] <andypugh> (FWIW, nothing else uses the "canonical pin naming convention" anyway, as it states that digital inputs should be called "digin" and outputs "digout"
[13:36:31] <cradek> heh, whose canon is this? I don't think I've seen it
[13:37:42] <andypugh> http://linuxcnc.org/docs/html/hal/canonical-devices.html
[13:38:40] <andypugh> "By version 2.1, the HAL drivers should have all been updated to match these specs. Send an email if you spot any problems."
[13:38:48] <cradek> ha
[13:39:21] <andypugh> <sends email>
[13:54:04] <skunkworks__> mhaberler, this is my issue.. I have done this git clone --branch unified-build-candidate-3 --origin github-mah
https://github.com/mhaberler/linuxcnc.git [<directory>]
[13:54:25] <skunkworks__> but don't know how to switch and build the preempt 7i80 branch
[13:54:46] <mhaberler> which one do you want, the old xenomai or the newer which works with rt-preempt
[13:54:58] <skunkworks__> new with rt-preemt
[13:55:21] <mhaberler> ok. then use the branch from g.l.o.
http://git.linuxcnc.org/gitweb?p=linuxcnc.git;a=shortlog;h=refs/heads/ubc3-7i80
[13:55:35] <mhaberler> git fetch origin
[13:55:57] <mhaberler> git checkout -b ubc3-7i80 origin/ubc3-7i80
[13:56:26] <mhaberler> cd src; sh autogen.sh; ./configure;make ; sudo make log setuid
[13:57:00] <skunkworks__> trying.
[13:59:42] <skunkworks__> git fetch origin does not seem to be a git repository.. (
[14:00:06] <mhaberler> I assumed you are in a working linuxcnc source directory
[14:00:32] <mhaberler> if you have one, just cd into it
[14:00:40] <skunkworks__> well - I am in the one I did with this git clone --branch unified-build-candidate-3 --origin github-mah
https://github.com/mhaberler/linuxcnc.git [<directory>]
[14:00:59] <skunkworks__> hold on - I my have totally confused git
[14:01:02] <mhaberler> this one is from github; micges branch lives on g.l.o
[14:02:15] <mhaberler> you might try ' git clone --branch ubc3-7i80 git://git.linuxcnc.org/git/linuxcnc.git
[14:02:51] <mhaberler> that r8168 trick seems to work; the firmware-realtek package is needed
[14:04:01] <mhaberler> got a 3.12.5-rt kernel to run on an amd brazos board with r8168 and _without_ hand-build r8168 module
[14:04:24] <andypugh> skunkworks__: The motor consumes enough current to destroy my bench power supply, so the answer to that question will have to await me bolting together the big toroid I got from eBay and the reamins of said bench PSU.
[14:06:04] <skunkworks__> andypugh, heh - cool
[14:18:49] <skunkworks__> andypugh, wonder if it will need cooling
[14:19:08] <andypugh> It has built-in air stirrers
[14:20:45] <skunkworks__> neat - so what lathe are you putting that one?
[14:35:52] <skunkworks__> mhaberler, close..
http://pastebin.ca/2514987
[14:36:07] <skunkworks__> pcw_home, did you get this error ^
[14:36:18] <mhaberler> just do this:
[14:36:50] <mhaberler> ./configure --with-rt-preempt --with-posix
[14:37:13] <mhaberler> this will supress building for xenomai, which you have installed and micges didnt consider
[14:37:38] <skunkworks__> oh - heh
[14:38:10] <andypugh> skunkworks__: I don't know what on, or even if, I will ever use it.
[14:59:29] <skunkworks__> andypugh, did you have to do anything special - I get it to see the card - but the watchdog bites instantly
[15:04:12] <andypugh> It was days ago, I can't remember :-)
[15:05:12] <skunkworks__> heh
[15:05:45] <andypugh> The entirelty of my HAL:
http://pastebin.com/ipXrKN6a
[15:08:44] <skunkworks__> that is similar to what I did. Do you have to isolate the nic some how?
[15:08:49] <skunkworks__> or should it just work?
[15:09:07] <skunkworks__> the preempt latency seems to stay under 50us
[15:10:46] <alex_joni> http://www.youtube.com/watch?v=ZuE1JWFbbOU
[15:11:50] <andypugh> skunkworks__: I was directly connected from the motherboard to the 7i80. I may have turned off the wired network.
[15:14:40] <skunkworks__> hmm - same here.
[15:14:41] <skunkworks__> odd
[15:15:02] <skunkworks__> alex_joni, cool
[15:15:04] <skunkworks__> !
[15:15:08] <andypugh> You set the PC IP?
[15:15:35] <skunkworks__> the nic - yes. It sees the card it seems
[15:20:11] <andypugh> I am not sure what you mean by "nic". The IP address of the PC needs to be 192.168.1.1 (apparently)
[15:20:56] <skunkworks__> oh
[15:21:02] <skunkworks__> how do you set that?
[15:21:20] <alex_joni> ifconfig eth0 192.168.1.1
[15:21:27] <alex_joni> or whatever eth you have
[15:21:58] <alex_joni> (that's the quick change, the right way is to edit /etc/network/interfaces)
[15:25:33] <andypugh> I did it in the network control panel.
[15:27:00] <skunkworks__> do you remember what you put for the gateway?
[15:27:17] <skunkworks__> bbl
[15:36:31] <jepler> as I understand it, there would not be a gateway, because the network does not have a machine that will route packets to the internet at large
[15:41:54] <alex_joni> what jepler said
[15:49:56] <alex_joni> anyone has a longer g-code program handy?
[15:50:07] <alex_joni> (longer than the ones in nc_files)
[15:50:41] <micges> what kind of program?
[15:51:59] <alex_joni> any kind
[15:52:08] <alex_joni> just to see how 3d performs on the bbb
[15:53:17] <alex_joni> hmm.. found one which is 3500 lines, works fluidly
[15:59:24] <micges> alex_joni:
http://filebin.ca/15u1P9AatzEP
[16:00:00] <alex_joni> thanks
[16:00:56] <alex_joni> 500k lines .. :)
[16:01:20] <micges> :)
[16:01:41] <micges> if you load it entirely on bbb it will be big success
[16:03:33] <alex_joni> started loading
[16:05:10] <alex_joni> 50% done
[16:11:34] <alex_joni> http://imagebin.org/283055
[16:11:50] <alex_joni> micges: definately a bit sluggish.. but it still can be used
[16:12:24] <skunkworks> micges: did you see my issues with rt_preempt?
[16:12:31] <alex_joni> there is about 1s delay in the display handling
[16:12:33] <micges> that's great, few ywars ago I couldn't load it on few pc
[16:12:51] <skunkworks> logger[mah]:
[16:12:51] <logger[mah]> skunkworks: Log stored at
http://linuxcnc.mah.priv.at/irc/%23linuxcnc-devel/2013-12-19.html
[16:14:36] <alex_joni> night all
[16:18:07] <skunkworks> I am getting watchdog bites instantly
[16:18:28] <skunkworks> although - this is 64bit version... do you think that would matter?
[16:20:05] <skunkworks> for the hm2_eth
[16:26:50] <memleak> ifconfig eth0 192.168.1.1 netmask 255.255.255.0 broadcast 192.168.1.255 && route add default gateway <your gateway address>
[16:27:31] <memleak> just to clarify the discussion earlier about IP addresses and gateways
[16:28:17] <memleak> if you dont have an eth0 network adapter then you can run ifconfig -a to show a complete list of network adapters available
[16:30:28] <KGB-linuxcnc> 03Sebastian Kuzminsky 05scorbot-er-3 03a1c19 06linuxcnc 10src/Makefile 10src/hal/user_comps/Submakefile 04src/hal/user_comps/scorbot-er-3 03src/hal/user_comps/scorbot-er-3.py install the scorbot driver * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=03a1c19
[16:49:38] <Tom_itx> i'm not sure if here is the place to ask but i was wondering if a feature was gonna make it to 2.6. It has to do with a script in hal_manualtoolchange that allows an external button to work along side the manual toolchange in axis.
[16:50:09] <Tom_itx> at the time several thought it would be handy to have
[16:54:25] <dgarr> Ton_itx i'll put it in master if you will test the deb
[16:54:46] <andypugh> I am using it on my machine.
[16:55:06] <Tom_itx> i have the file and have tried it
[16:55:21] <Tom_itx> i'm waiting to mod my pendant before it becomes permanent here
[16:55:35] <dgarr> ok later today
[16:55:44] <andypugh> It helps a lot with Touchy, I can just mash the big green "run" button rather than try to hit the little fiddly dialog button.
[16:59:34] <micges> skunkworks: I have no idea, need test it
[17:15:11] <skunkworks> well - I can easilly install 32 and try it (see if it is my hardware or not)
[17:15:57] <skunkworks> this is what I pulled git clone --branch ubc3-7i80 git://git.linuxcnc.org/git/linuxcnc.git
[17:38:26] <PCW> skunkworks: you might try:
[17:38:27] <PCW> sudo ping -i .001 192.168.1.121 >elatency
[17:38:29] <PCW> for a minute or so to see what kind of latency you have
[17:38:30] <PCW> to send/recv packets
[17:40:34] <PCW> (maybe theres a ping latency histogram util)
[18:20:26] <KGB-linuxcnc> 03Andy Pugh 05joints_axes4 7e4fbb5 06linuxcnc 10scripts/update_ini Add dialog and force options to the update_ini script to make it more suited * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=7e4fbb5
[18:46:09] <skunkworks> pcw will try tomorrow
[18:47:54] <PCW> I had luck with some MBs with Preemt_RT other had good latency on the latency test but occasional 5ms packet delays
[18:48:42] <skunkworks> ok - good to know. The latency was good on the motherboard - this is with a intel pro/100
[18:48:59] <PCW> i suspect there's a fair amount of learning needed about Preempt _RT behaviour
[18:49:46] <PCW> the MB that just worked was some old core duo (E5200?) test MB we had around
[19:03:39] <andypugh> <happy> I can now launch any config from the picker in joints_axes and it will auto-convert and run :-)
[19:19:44] <KGB-linuxcnc> 03Dewey Garrett 05dgarr/ngcgui-gcmc 39447c8 06linuxcnc 10tcl/ngcgui.tcl ngcgui.tcl: fix msg for unexpected gcmc version * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=39447c8
[21:15:14] <KGB-linuxcnc> 03Dewey Garrett 05master 8c1630c 06linuxcnc 10src/emc/usr_intf/axis/scripts/hal_manualtoolchange.py hal_manualtoolchange: new pin for external button * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=8c1630c