#linuxcnc-devel Logs

Jan 21 2018

#linuxcnc-devel Calendar

06:16 AM linuxcnc-build: build #3491 of 1405.rip-wheezy-armhf is complete: Failure [4failed compile] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1405.rip-wheezy-armhf/builds/3491 blamelist: John Thornton <bjt128@gmail.com>
06:45 AM linuxcnc-build: build #5340 of 0000.checkin is complete: Failure [4failed] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/0000.checkin/builds/5340 blamelist: John Thornton <bjt128@gmail.com>
08:20 AM jthornton: hmmm
12:21 PM seb_kuzminsky: jthornton: i'm working on the 2.7.12 release and i came accross your commit 444c31afc, "Docs: the linux-image-rt-amd64 has been removed or is not longer available"
12:21 PM seb_kuzminsky: that does not seem to be accurate, to me:
12:21 PM seb_kuzminsky: https://packages.debian.org/search?keywords=linux-image-rt-amd64
12:23 PM seb_kuzminsky: i guess it's missing from "mainline" jessie, though it is in jessie-backports
12:23 PM seb_kuzminsky: is that what you ran into?
01:52 PM seb_kuzminsky: JT-Shop: did you see my comment about linux-image-rt-amd64 above? any clues for me?
02:17 PM speedy__: when im MDI mode, can linuxcnc be switched to manual mode by gcode? example: execute MDI code via jog dial -> mdi program ends -> linuxcnc remains in MDI mode but manual mode is desired.
02:28 PM JT-Shop: seb_kuzminsky: no, I missed it
02:28 PM JT-Shop: reading back that might have been the problem, as I recall the 32bit was fine but the 64bit was not found
02:35 PM seb_kuzminsky: i don't want to de-emphasize jessie-amd64 as a valid platform. maybe we should say "look in backports if you can't find it" instead of taking it out
02:49 PM JT-Shop: might be better if we told them how to install it, look in backports is pretty hard to understand if you know nothing about linux
02:49 PM JT-Shop: I was trying to install following the manual to the letter when I had that problem
02:50 PM JT-Shop: while I have your ear, if you install from build bot will you get notifications of updates somehow?
02:56 PM JT-Shop: last part for the day... I'll be glad for some quiet in the shop
03:05 PM seb_kuzminsky: JT-Shop: if you have the buildbot deb archive in your sources.list, then you'll get notification of new packages from there, just like you do for new packages from the normal deb archives
03:06 PM seb_kuzminsky: i understand why you removed the option for linux-image-rt-amd64 now
03:06 PM seb_kuzminsky: i'll see about making the install instructions better for that option, and put it back in
03:11 PM JT-Shop: I'll try it again in the morning and see if I can enable backports and make it work
03:29 PM seb_kuzminsky: JT-Shop: was it jessie you were trying?
03:45 PM seb_kuzminsky: speedy: gcode doesn't have access to the mdi/manual mode (there's also an 'auto' mode for running programs)
03:47 PM seb_kuzminsky: however, you could maybe make a 'user-defined command m-code' in the form of a shell script that changes the mode
03:47 PM seb_kuzminsky: http://linuxcnc.org/docs/devel/html/gcode/m-code.html#mcode:m100-m199
03:47 PM seb_kuzminsky: then your mdi code can call that m-code at the end
03:48 PM seb_kuzminsky: there'll probably be a problem trying to switch out of mdi mode while the m-command is still running
03:48 PM seb_kuzminsky: so you might have to fork and do the mode-switch in a background process
04:54 PM rene-dev: seb_kuzminsky boolean2 is the -not pin for inputs
05:12 PM seb_kuzminsky: do you think 2.7 has the same bug?
05:16 PM rene-dev: no
05:16 PM rene-dev: it depends
05:16 PM rene-dev: it was introduced with the commit I linked in my commit
05:16 PM seb_kuzminsky: i see
05:17 PM rene-dev: I dont know if that ended up in 2.7
05:19 PM rene-dev: that commti doesnt create the -not and -invert pins for bidirectional IO
05:19 PM rene-dev: but the pins are still used in the read and write function, so the driver will crash
05:19 PM seb_kuzminsky: i just looked, nothing like 31029ef is in 2.7
05:19 PM seb_kuzminsky: ok cool, thanks for the fix
05:24 PM rene-dev: seb_kuzminsky Im looking at abs encoder support... I dont see how that differs form inidiate homing? http://linuxcnc.org/docs/devel/html/config/ini-homing.html#_home_absolute_encoder
05:25 PM rene-dev: really abs encoder should add a pin, because abs encoder still need to be homes, just not as often.
05:25 PM rene-dev: so linuxcnc should be aware when they loose position, due to battery failure or something
05:30 PM rene-dev: hmm, it doesnt seem to be all that easy to get a pin all the way into the homing state machine :D
05:35 PM seb_kuzminsky: rene-dev: i've never used absolute encoders, only regular relative ones
05:35 PM rene-dev: they just have a battery, and count multiturns, and tell you if the value is still valid
05:35 PM seb_kuzminsky: do abs encoders generally have a "i forgot where i was" signal, for things like battery replacement?
05:36 PM seb_kuzminsky: and you want that homing should pay attention to the "value is valid" signal and do different things if it's true or false?
05:37 PM seb_kuzminsky: on true we just use the valid value (like now, i guess?)
05:37 PM seb_kuzminsky: on false, do the normal home to home switches thing
05:38 PM seb_kuzminsky: and somehow reset the abs encoder position there? or record it in linuxcnc and compensate somehow?
05:47 PM rene-dev: yes, you usually have to reset them after a battery failure
05:48 PM rene-dev: so the joint would need a is-abs pin
05:48 PM rene-dev: on true, use it with some offset
05:48 PM rene-dev: on false, home and store that offset
05:49 PM seb_kuzminsky: where is that offset currently stored? or is it? i have no idea
05:49 PM rene-dev: the reset depends on how its done... some just reset, some you can program the oiffset. linuxcnc shoudl not care about that
05:49 PM rene-dev: in [JOINT_n]HOME
05:49 PM rene-dev: but linuxcnc cant write to that...
05:50 PM rene-dev: what about using the positions file?
05:50 PM seb_kuzminsky: ah, i see
05:50 PM rene-dev: to let the hardware know that homing has performed, the index enable pin could be used. it already does whats needed on reset
05:51 PM rene-dev: so currently you would need 2 configs, one to home, and one to run. and there is no buildin way for error detection.
05:52 PM rene-dev: POSITION_FILE does something else... but maybe it can still be used?
05:52 PM rene-dev: I think writing to the ini is not a good idea
05:55 PM rene-dev: seb_kuzminsky https://pastebin.com/h3hzmEUD
05:56 PM rene-dev: thats what this guy uses http://www.yurtaev.com/
05:57 PM rene-dev: http://www.yurtaev.com/files/linuxcnc_2.7.11ssc10/
05:57 PM rene-dev: scroll down to absolute encoder support, it explains how it works
05:57 PM rene-dev: http://www.yurtaev.com/yssc2p-software.html
06:14 PM pcw_home: some recent absolute encoders use the Weigand effect so need no batteries
06:15 PM pcw_home: ( and many older designs one have gears )
06:17 PM pcw_home: The Fanuc encoders have a "I lost power" bit in the serial data
07:24 PM skunkworks: anyone have an idea? I am trying to get the latest dxf2gcode working and after lots and lots of flailing around I am down to this error
07:24 PM skunkworks: Reading https://pypi.python.org/simple/pyqt5/
07:24 PM skunkworks: No local packages or working download links found for pyqt5
07:24 PM skunkworks: error: Could not find suitable distribution for Requirement.parse('pyqt5')
07:25 PM skunkworks: but pyqt5 is installed
07:28 PM skunkworks: https://sourceforge.net/p/dxf2gcode/sourcecode/ci/master/tree/source/README.txt
07:28 PM rene-dev: pcw_home yes, in the end it is all the same problem :)
07:28 PM rene-dev: pcw_home do you have any documentation on the fanuc serial encoders?
07:28 PM skunkworks: google has failed me
07:29 PM rene-dev: thats one of the few protocols missing in stmbl...
08:00 PM pcw_home: rene-dev theres a little description in the hm2 regmap file (77 data bits 1.024 Mbaud 5 bit CRC)
08:01 PM rene-dev: ah, ok. I currently dont have one.
08:01 PM rene-dev: how do I get it to talk?
08:01 PM pcw_home: ( thats type I, (RS-422) type II is RS-485 and 2.048 Mbaud )
08:03 PM pcw_home: Type I needs a request pulse, Type II I'm not sure yet, I have a 16M count/turn type II encoder but have not had time to play with it
08:08 PM pcw_home: 65536 and 1M count/turn are the most common Type I, Type II go up to 32M counts/turn AFAIK
08:11 PM pcw_home: they all include a low res (1024) absolute track for commutation (does not require battery)
09:30 PM seb_kuzminsky: skunkworks: funny, i've been meaning to try to package dxf2gcode to make it easier to install
09:30 PM seb_kuzminsky: skunkworks: what distro are you trying it on?
09:41 PM jepler: skunkworks: python3-pyqt5 make sure you have the version for python3 installed?
09:46 PM skunksleep: seb_kuzminsky: this was on stretch
09:46 PM skunksleep: jepler: thanks I will check
09:48 PM skunksleep: I remember the last time I tried it (a long time ago) it just worked