#linuxcnc-devel | Logs for 2014-07-25

[08:01:49] <KGB-linuxcnc> 03John Thornton 05v2.5_branch 24b0b3d 06linuxcnc 10docs/src/config/ini_config.txt Docs: fix startup code example. * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=24b0b3d
[09:35:53] <cradek> JT-Shop: another INCAM spam...
[09:37:45] <skunkworks_> sounds like a monty python skit...
[09:39:06] <pcw_home> spamincan
[09:52:57] <cradek> people should stop responding to steve...
[09:53:11] <cradek> (I should talk)
[09:53:16] <skunkworks__> heh]
[10:02:33] <pcw_home> years ago I volunteered to make Steve a encoder resolution divider (completer with index widener)
[10:02:33] <Roguish> anyone out there familiar with the bldc comp? andy pugh? pcw?
[10:02:35] <pcw_home> in a ~$1.50 CPLD. I made one up on one of our test CPLD boards, tested it thoroughly and sent it off to him no charge
[10:02:36] <pcw_home> he was unable to make it work and proceeded to tel me that his Fanuc field engineer friend had also verified
[10:02:38] <pcw_home> that the logic was faulty. That was my last dealing with Steve
[10:03:18] <pcw_home> I know nothing
[10:04:12] <Roguish> what the heck is this 'personality' thing in bldc comp?
[10:04:14] <cradek> ooh that sounds useful. I made a similar thing in an atmega8 back before I had the (ugh) pluto on my sherline lathe
[10:06:22] <cradek> I actually had a usable dumb servo system that way (parport pwm driving L298 dual H bridge ran both servos)
[10:07:07] <cradek> the divider did work right, but it didn't do index
[10:07:07] <skunkworks__> That was cool - it also had only like 10 pwm levels iirc. worked just fine.
[10:07:20] <cradek> I'm sure yours was better (and obviously faster)
[10:07:57] <cradek> yeah, the pluto is better mostly because it can read the full encoder resolution, but also it tunes up a bit nicer with more pwm levels.
[10:10:07] <cradek> Roguish: in http://www.linuxcnc.org/docs/html/man/man9/bldc.9.html it says "ignore the auto-generated SYNOPSIS above", so maybe you're supposed to ignore the personality thing?
[10:10:28] <pcw_home> I think I had jumpers to set the divide ratio (index was a bit of a kludge, N counts wide after the real index signal)
[10:11:22] <pcw_home> so the leading edge was correct but the trailing index edge depended on direction
[10:12:28] <Roguish> cradek: yeah. saw that. in the doc there is continual reference to the personality for the pins and parameters. how does the personality correlate?
[10:12:50] <cradek> I don't know. what happens if you just ignore personality?
[10:13:11] <Roguish> ex: bldc.N.init bit in [if ( personality & 0x05 ) == 4]
[10:13:19] <memfrob> when using the 2.6 branch does sim axis work for anyone else out of the box? i get errors while trying to run the example gcode files without modifying anything.
[10:13:26] <pcw_home> I thin ive only used the cfg- stuff
[10:13:32] <cradek> memfrob: uh what error?
[10:14:46] <cradek> oh move exceeds limit on Z? select Z, poke touch off, enter 1, ok
[10:14:53] <memfrob> yes.
[10:14:55] * cradek rubs his crystal ball
[10:14:59] <Roguish> i have put in all the pin and prameters and slowly eliminated them thru watching the errors. the old texas shotgun approach.
[10:16:01] <memfrob> ok thanks. why is that step not necessary for 2.5 branch?
[10:16:05] <pcw_home> bldc is kind of a swiss army knife comp
[10:16:06] <cradek> it is
[10:16:12] <cradek> you must've already done it
[10:16:25] <memfrob> hmm.. ok
[10:16:27] <Roguish> simple motor, cfg=q i think.
[10:17:23] <pcw_home> no Halls ?
[10:17:24] <Roguish> pcw_home: i'm trying to run a simple bldc with a 7i39
[10:18:08] <Roguish> has halls, and external encoder. encoder not for commutation.
[10:19:29] <pcw_home> normally better to use the encoder + Halls (so bldc can use sine wave commutation)
[10:19:53] <Roguish> should be cfg=h. maybe deleted the wrong letter late last night.
[10:21:17] <pcw_home> if you have the halls wired up, you need to check the commutation angle (should step from 0 to 1 monotonically in 0.1666666 steps)
[10:22:12] <pcw_home> (per electrical rotation, a 4 pole motor has 2 electrical rotations per turn)
[10:22:15] <Roguish> i am pretty sure it is a 60 degree motor
[10:23:05] <pcw_home> apply a little current to one winding and turn the shaft by hand
[10:23:52] <pcw_home> 1 detent = 2 pole
[10:23:54] <pcw_home> 2 detents=4 pole etc
[10:24:37] <cradek> proposed splash screen related fix: http://timeguy.com/gitweb?p=linuxcnc.git;a=commitdiff;h=49eee1e10e079db6026453f5c77722c751e6dbbe
[10:25:09] <cradek> :-P
[10:26:34] <pcw_home> friendlier if demos work without complaint
[10:27:08] <cradek> yeah, although I'm sympathetic to the "you oughta learn touch off right away" argument
[10:29:28] <cradek> and demo configs shouldn't be weird (Z0 should always be top of travel)
[10:29:43] <cradek> but like I said, it may be the least bad solution
[10:29:54] <cradek> now seb has to decide...
[10:30:05] * cradek is happy to not be the RM
[10:30:59] <cradek> hm, it only needs 3mm, not an inch
[10:52:29] <skunkworks__> well - as some point out.. linuxcnc is not friendly... I say make them enter a 64 digit activation code before you can even start configuring..
[10:55:28] <cradek> that's why nobody will ever use it
[10:55:31] <pcw_home> Is linuxCNC right for you? First Solve this simple puzzle...
[10:57:09] <skunkworks__> heh
[10:57:12] <pcw_home> Or the guy with the Traub on the forum ("I connected everything, it doesnt work but be a firmware problem")
[10:57:30] <pcw_home> s/burt/must/
[11:05:44] <archivist> bet he has never done a days work fault finding
[11:07:18] <cradek> well I failed to help someone on the forum who doesn't realize that an internal thread cut on the front side is just like an external thread cut on the back side
[11:07:30] <cradek> sigh
[11:07:41] <jepler> I've got a patch almost ready to implement this: http://pastebin.com/713cXn94
[11:07:44] <jepler> cradek: OK for 2.5?
[11:08:36] <skunkworks__> heh
[11:08:37] <cradek> jepler: better try it on a branch so buildbot tests it
[11:13:07] <skunkworks__> http://linuxcnc.org/index.php/english/forum/38-general-linuxcnc-questions/28138-new-scrypt-image-to-gcodepy-how-to-put-on-git
[11:14:44] <jepler> OK, patch is ready: http://emergent.unpythonic.net/files/sandbox/0001-Restrict-use-of-LinuxCNC-to-paying-customers.patch
[11:14:52] <jepler> you can pay me $1 for a trial license or $1000 for a full license.
[11:15:12] <cradek> jepler: go home, you're drunk
[11:17:47] <jepler> cradek: because I'm such a nice guy, here's a key for tetrode-cradek: 7ffa2c10-101e-4f96-8fb6-bc3934e21862
[11:18:16] <JT-Shop> cradek, on the forum?
[11:18:49] <cradek> I'm having a hard time knowing whether your nonsense patch was an accident
[11:19:11] <jepler> oh that's the wrong patch
[11:19:14] <jepler> but that's funny too
[11:19:22] <cradek> differently funny
[11:19:44] <jepler> http://emergent.unpythonic.net/files/sandbox/0001-V2Restrict-use-of-LinuxCNC-to-paying-customers.patch
[11:20:11] <cradek> JT-Shop: yeah http://linuxcnc.org/index.php/english/forum/31-cad-cam/28142-incam-m25-is-now-of
[11:20:14] <cradek> fering-a-free-trial#49113
[11:20:20] <cradek> oops
[11:20:37] <cradek> http://linuxcnc.org/index.php/english/forum/31-cad-cam/28142-incam-m25-is-now-offering-a-free-trial#49113
[11:21:28] <cradek> jepler: funny! looks about like code I've found in the wiki and bbs.
[11:23:54] <jepler> I am sad to say, I've given some thought to how software activation via serial number might work (though not because I would like to add it to linuxcnc, promise!)
[11:24:26] <skunkworks__> jepler is going to sell some software!
[11:24:36] <jepler> this is not a very good scheme, I'm worried some people would be able to bypass it
[11:24:44] <cradek> hah
[11:25:14] <cradek> jepler: be sure to add base64 as a build dep
[11:25:45] <jepler> cradek: coreutils is Essential: yes
[11:26:02] <cradek> ah ok
[11:27:40] <seb_kuzminsky> cradek: is moving Z+ limit to 1.001 less wrong than moving the program's top-of-work to Z=-1?
[11:27:57] <jepler> seb_kuzminsky: yes, did you read cradek's rationale in the patch?
[11:28:11] <jepler> leaving top-of-work at Z0 still encourages correct practice (touch off with 0 = top of work)
[11:28:23] <seb_kuzminsky> i didnt read that part :-/
[11:28:37] <seb_kuzminsky> oh yeah, it says right there
[11:28:47] * seb_kuzminsky goes back to lurk mode in shame
[11:28:55] <cradek> it would be bad to have it surprisingly plunge an inch
[11:30:19] <JT-Shop> cradek, squashed again
[11:30:24] <jepler> hm, do you think it would be more user-centric to have activation keys like filamentous-handicapper-stimulus-globe than 2eb692f0-b6f3-49fb-afe7-5ebbbc53a6a7 ?
[11:30:25] <cradek> thanks!
[11:30:58] <jepler> cradek: OK, install with this key instead: pugnacious-blemished-beneficial-sapience
[11:31:13] <jepler> no code change needed, because I'm that awesome
[11:31:21] <cradek> wow
[11:31:28] <cradek> I'm impressed, and you're still drunk
[11:31:35] <jepler> must be
[11:39:17] <Roguish> pcw_home: i think i got it. had to swap encoder wires in connector and (from notes of long ago) set motor to pattern 43 and cfg to h
[11:40:22] <Roguish> now doing simple PID tuning and scale settings, etc. to get correct speed.
[11:42:01] <pcw_home> qh mode will be smoother but requires a bit more setup
[11:43:06] <Roguish> i'll get the h mode correct, and then attempt the qh. at least i know the encoder is correct and the halls work,and the pattern.
[11:51:56] <pcw_home> the encoder direction needs to match the Hall direction or funny things happen
[11:59:33] <seb_kuzminsky> Roguish: could you try out the 0.8.13 hostmot2 firmware packages? they fix the permissions problem you reported
[12:01:02] <seb_kuzminsky> pcw_home: shall we have an Nth attempt at updating the hostmot2 source in git (& the compiled firmwares on wlo)?
[12:03:51] <pcw_home> Sure, what can I do to help?
[12:06:25] <seb_kuzminsky> check your new vhdl into the master branch of the hostmot2-firmware.git repo on glo, the buildbot will try it, and we'll see where i need to patch up the build system & debian packaging
[12:08:58] <KGB-linuxcnc> 03Sebastian Kuzminsky 05v0 6692770 06hostmot2-firmware fast forward * 14http://git.linuxcnc.org/?p=hostmot2-firmware.git;a=commitdiff;h=6692770
[12:09:12] <seb_kuzminsky> i'll help with any part of that or any questions you might have
[12:20:24] <cradek> ooooh, I will also help any way I can
[12:40:59] <pcw_home> Let me first add the @Card@ and @Pin@ to all the top level files if not there already so they are ready for scripting
[12:41:00] <pcw_home> without needing to make a copy
[12:41:05] <seb_kuzminsky> there's now two branches in hostmot2-firmware.git: master and v0
[12:41:18] <seb_kuzminsky> this mirrors the development and stable branches in linuxcnc.git
[12:41:53] <seb_kuzminsky> they currently point to the same commit, but they'll diverge as we/pcw adds the new vhdl to master
[12:42:54] <pcw_home> Yeah its kind of a forklift upgrade at this point so should not mess with V0
[12:43:08] <seb_kuzminsky> yep
[12:45:14] <seb_kuzminsky> there's still some buildbot configuration to do to support the multiple branches, but that can happen over time
[12:49:28] <cradek> ooh, I booted my test machine and hostmot2 updates popped up
[12:50:08] <seb_kuzminsky> 0.8.13, has the fix for the permissions problem that Roguish pointed out
[12:50:47] <cradek> yep it says 0.8.10 to 0.8.13
[12:50:48] <seb_kuzminsky> err, or 0.8.17, if you're getting them form the hm2 buildbot
[12:50:50] <seb_kuzminsky> ah
[12:51:09] <cradek> nope the latest image points at only the real release places
[12:51:15] <seb_kuzminsky> yay
[12:51:43] <seb_kuzminsky> i'm gonna look at your pending 2.6 patches (and jepler's) and try to do 2.6.0 this weekend
[12:51:52] <seb_kuzminsky> anyone know of any blockers?
[12:52:20] <cradek> I don't, but I haven't looked at the bugtracker for a while
[12:52:44] <seb_kuzminsky> none of those should hold up the release any longer, imo
[12:53:04] <cradek> ok
[12:53:06] <seb_kuzminsky> i'll make point releases as the known bugs get fixed
[12:53:07] <cradek> I'm sure it's ready
[12:53:12] <seb_kuzminsky> i think so too
[12:53:22] <seb_kuzminsky> it's sure been working well for me
[12:53:29] <cradek> yep, just do it, me too.
[12:53:43] <seb_kuzminsky> the last-second kernel upgrade was a bit scary, but i think it's working
[12:53:56] <cradek> I can have an iso a half hour after you do the deed
[12:54:01] <Roguish> thank you seb
[12:54:22] <seb_kuzminsky> Roguish: did it fix the problem for you?
[12:55:50] <Roguish> well, i changed the permissions manually. so i can see the directory.
[12:56:28] <seb_kuzminsky> oh, right
[12:56:50] <seb_kuzminsky> well, i bet if you "dpkg --purge" all your hostmot2-firmware packages, and reinstall them from www.linuxcnc.org, they'll still work right
[12:56:59] <Roguish> i am using the 'hybrid' iso, so i'm not sure how to use a new build. probably the git thing, right?
[12:57:17] <seb_kuzminsky> you have two options
[12:57:29] <seb_kuzminsky> 1: "sudo apt-get update && sudo apt-get -uV dist-upgrade"
[12:57:46] <seb_kuzminsky> 2: read the wiki & figure out git & build it yourself
[12:58:27] <Roguish> i can probably build it my self, but the update looks far easier (not to screw up....)
[12:58:53] <seb_kuzminsky> it's the easy way for sure, it's how almost all of our users get their upgrades
[13:00:18] <Roguish> i've used the git and build and don't mind it. i'm just trying to get a new setup with bldc running and don't need other things to do.
[13:00:46] <Roguish> i'll copy the command line you gave above and save it for a little later.
[13:01:55] <Roguish> hey, what is the command to create a new thread at a particular frequency?
[13:02:21] <cradek> there's a "thread" hal component that you load
[13:02:25] <cradek> er, threads
[13:02:37] <cradek> http://www.linuxcnc.org/docs/html/man/man9/threads.9.html
[13:03:14] <cradek> be aware you can only create threads that are slower than the existing fastest thread
[13:03:30] <cradek> usually motion creates your "first" one or two threads for you
[13:03:35] <Roguish> ok. it shows in the html doc. been looking in the pdf
[13:03:53] <cradek> I think the pdf should have the manpages too...?
[13:06:20] <cradek> hm, maybe not
[13:06:34] <Roguish> HAL configuration shows 1 thread, servo thread. set in the .ini file.
[13:06:58] <Roguish> are there any hidden threads?
[13:07:00] <cradek> ok cool, you can make any new thread slower than that
[13:07:04] <cradek> no
[13:08:05] <Roguish> ok. i'll figure it out. last question (maybe) where's doc on halscope?
[13:08:15] <cradek> the hal manual
[13:08:36] <Roguish> again, pdf's give very little. have not checked html
[13:09:14] <cradek> the "hal manual" pdf has 10 pages on halscope
[13:09:15] <Roguish> i would like a faster sampling rate, therefor the question about threads.
[13:09:37] <cradek> ok, how about you ask the real question at the root of all your questions
[13:10:31] <cradek> what are you trying to see in halscope?
[13:12:26] <Roguish> following error over commanded position over actual position, and over/under shoot. both in position and velocity. for tuning.
[13:13:04] <Roguish> give a step impulse and see the response.
[13:13:40] <Roguish> would also like to do the same in freq domain in a Bode plot.
[13:14:31] * skunkworks__ just fidles with the pid numbers until it looks good...
[13:14:34] <skunkworks__> ;)
[13:14:41] <pcw_home> I think the bode plot will need some comp help
[13:16:12] <cradek> Roguish: the servo cycle (the only thing you can plot with your one thread) is all the information you have to work with - the position is sampled and the command is generated at that rate
[13:16:26] <pcw_home> Bare HBridge tuning is sort of 1/2 way between torque mode and velocity mode
[13:16:42] <cradek> I have not had good luck tuning a step response - it just seems to overwhelm everything - but you could easily simulate that by turning up your acceleration and using gcode
[13:16:50] <pcw_home> (needs FF1 FF2 and significant D term)
[13:17:56] <cradek> I like to program a loop in gcode: rapid right, small pause, rapid left, long pause - repeating forever. then get halscope to trigger right and show you the whole thing
[13:18:05] <pcw_home> also faster than 1 KHz thread rate is probably needed unless the motor is driving significant (relative to its torque) mass
[13:18:54] <Roguish> cradek: i'll speed up the servo cycle and create a new thread.. i also use a gcode to shuttle it back an forth. works fine.
[13:19:25] <Roguish> pcw_home: are u refering to the 7i39 as a bare hbridge?
[13:19:28] <cradek> I don't think I understand why you want a new thread
[13:20:20] <pcw_home> Yeah normally 1 thread is enough
[13:20:34] <Roguish> cradek: maybe i don't need one if i just speed up the servo cycle.
[13:20:50] <Roguish> you are right. keep it simple.
[13:21:11] <cradek> you can't make a new thread that's faster than motion's fastest (in your case only) thread
[13:21:24] <cradek> even if you could, there'd be nothing more to plot
[13:22:13] <pcw_home> also for the 7I39, set the 3pwmgens dead time to 0
[13:22:24] <Roguish> I don't know exactly what it would take to create a Bode plot, but it might be worth it. my controls friends all say it's the better tuning method.
[13:23:35] <Roguish> is there a 'feature request form' ?
[13:23:54] <cradek> yes on sourceforge there's a feature tracker
[13:25:53] <Roguish> gentlemen. thank you all. you've been very helpful. gotta go swing a hammer. building the wife a shed/art studio. oh joy.
[13:26:16] <cradek> cool
[13:34:33] <Tom_itx> so is it really true? 2.6 is about to go out the door?
[13:34:48] <cradek> oh yes
[13:34:57] <Tom_itx> hurray!
[13:35:09] <Tom_itx> is there a new feature list somewhere?
[13:35:19] <mozmck> how can I compare two git branches? just put them in two directories and use meld?
[13:35:33] <cradek> mozmck: git diff branch1..branch2
[13:35:51] <mozmck> ah - thanks!
[13:35:56] <cradek> mozmck: or git log branch1..branch2 if you want to see the commits, not the code changes
[13:36:24] <Tom_itx> finally got my ethernet back up after a reinstall...
[13:36:28] <Tom_itx> not sure what caused that
[13:36:29] <cradek> Tom_itx: I don't think the changelog is written yet. that's one of the last (manualish) tasks.
[13:36:53] <Tom_itx> is the latest on the buildbot site?
[13:37:05] <cradek> do you mean latest changelog?
[13:37:12] <Tom_itx> latest code
[13:37:16] <Tom_itx> and / or changelog
[13:37:27] <cradek> yes always. but 2.6.0~pre5 was just released and that's a real release on the main site
[13:37:45] <cradek> buildbot constantly makes packages on all active branches
[13:38:08] <cradek> there's also a new test os-install image
[13:38:25] <Tom_itx> which flavor is it?
[13:38:57] <Tom_itx> i may do an install on a ssd and see how it flies
[13:39:07] <cradek> debian stable 7.x with rtai
[13:39:23] <cradek> yeah do. we would love to hear the feedback.
[13:39:40] <Tom_itx> i'd have done a bit of testing but my interweb quit
[13:39:55] <Tom_itx> got tired of jacking with it so i just reinstalled
[13:39:56] <cradek> Tom_itx: http://article.gmane.org/gmane.linux.distributions.emc.user/52401
[13:40:47] <Tom_itx> it's not on the main page yet is it?
[13:40:59] <cradek> nope
[13:41:05] <cradek> still testing
[13:42:00] <Tom_itx> is there an iso for it yet or just build it from the buildbot page?
[13:42:11] <cradek> see that url a couple lines up
[13:42:30] <cradek> that's the announcement from the users list
[13:42:39] <Tom_itx> ok
[13:43:23] <Tom_itx> hybrid.iso
[13:44:28] <Tom_itx> do i need unetbootin for a usb stick?
[13:44:44] <cradek> READ THE MESSAGE
[13:45:14] * Tom_itx tucks his tail and runs
[13:45:32] <cradek> haha
[13:45:33] <cradek> sorry
[13:45:54] <Tom_itx> i'm too used to irc :)
[13:45:57] <cradek> but the words of the message, they are for a purpose
[13:45:58] <cradek> :-)
[13:46:34] <Tom_itx> i'll check it out when i get back... got a couple errands to run
[13:47:32] <skunkworks__> Can I use zsync to get it?
[13:47:42] * cradek shakes his fist
[13:48:39] <atom1> hahaha
[13:49:04] <Tom_itx> I recommend using zsync to download it:
[13:49:16] <Tom_itx> 3rd to last line
[13:52:08] <cradek> Will it fit on a CD?
[13:52:25] <skunkworks__> awww - that was my next question!
[13:52:26] <cradek> (I can play this too)
[13:52:30] <cradek> haha
[13:53:14] <mozmck> can I install it on my iphone? wait, I don't have one, how about walmart net10 phone?
[13:53:26] <cradek> > The common thread in all of your problems is you.
[13:53:30] <Tom_itx> honest question: will a SD with usb adapter be the same as a USB stick?
[13:53:42] <cradek> Tom_itx: yes I think so
[13:54:17] <skunkworks__> yes. I have done this.
[13:55:21] <cradek> think I'm going to drag out my old sherline lathe this weekend and get it going again
[13:56:11] <skunkworks__> yay!
[13:56:29] <skunkworks__> are you going to switch to a 5i25? or keep using the pluto?
[13:56:42] <cradek> if the pluto still works I will use it with 2.6
[13:57:27] <skunkworks__> How can I get 2.6?
[13:57:37] <cradek> I may be the new maintainer, if I want it to keep working
[13:58:20] <skunkworks__> I still use the pluto for odd testing.. (testing encoders and such...) I figure if I kill it - it has paid for its self
[13:59:14] <skunkworks__> itself
[13:59:57] <cradek> my lathe driver stuff is all in a pretty box with connectors, and the screws are all in it and everything
[14:00:03] <cradek> it's weirdly finished
[14:01:32] <skunkworks__> not odd for you.. The jr was finished also.. and the hardige..
[14:01:48] <cradek> the hardinge is awful - the cabinet door doesn't even close
[14:01:54] <cradek> but yeah the jr is pretty nicely done
[14:02:11] <skunkworks__> why doesn't the door close?
[14:02:25] <cradek> mostly because of the wires going in and out, and the heat problems
[14:02:25] <skunkworks__> stuff doesn't fit or wires running out the door?
[14:02:34] <skunkworks__> ah.
[14:02:47] <skunkworks__> we have found that the control box on the k&t probably needs fans..
[14:08:00] <skunkworks__> jepler, what branch should I test for the 7i80/5i25?
[14:08:07] <skunkworks__> (uspace)
[14:10:20] <skunkworks__> does hm2-eth-v3 have support for pci also?
[14:12:18] <jepler> skunkworks__: yes, -v2 and -v3 are supposed to work with all hm2 hardware
[14:12:26] <skunkworks__> ok
[14:12:29] <skunkworks__> thanks
[14:12:29] <jepler> skunkworks__: let me know if your testing says otherwise
[14:12:39] <skunkworks__> will do
[14:14:29] <jepler> you're the best
[14:14:43] <skunkworks__> sorry it has taken so long...
[14:15:55] <jepler> what has?
[14:16:03] <skunkworks__> getting back to testing.
[14:25:48] <skunkworks__> 7i80 works!
[14:27:00] <jepler> skunkworks__: you do a ton of testing, no hard feelings from me if you take a few days off for life
[14:27:07] <jepler> heck, I too most of the last two or three years off
[14:30:01] <skunkworks__> yes - (and I worried about that... ) but I am glad you are back in force!
[14:32:29] <atom1> SD on an adapter looks like SCSI
[14:36:45] <atom1> ok the light's blinkin so something is being written :)
[14:43:14] <skunkworks__> jepler, http://pastebin.com/y1LFhacj
[14:43:42] <jepler> skunkworks__: something specific I'm looking for in there?
[14:43:59] <skunkworks__> ethernet and pci seem to work together...
[14:44:05] <jepler> ah, both are loaded at once?
[14:44:09] <jepler> I actually didn't try that..
[14:44:11] <skunkworks__> heh
[14:45:24] <skunkworks__> I think that was a double paste.. it is a little confusing at the bottom
[14:46:41] <cradek> anyone know anything about the glade stuff? http://linuxcnc.org/index.php/english/forum/48-gladevcp/28147-no-linuxcnc-widgets-with-wheezy-iso-260pre5#49140
[14:49:43] <Tom_itx> usb stick not bootable
[14:50:13] <Tom_itx> i get an icon of debian wheezy on the screen
[14:51:03] <jepler> Tom_itx: did you check bios boot order?
[14:51:20] <Tom_itx> yep, gettin ready to double check it...
[14:54:21] <kwallace> I think GaldeVCP needs Glade 2.something and not Glade 3?
[14:54:56] <kwallace> GladeVCP
[14:55:20] <cradek> how did you write the image?
[14:55:38] <cradek> oh hmm, if you get an icon, it's probably bootable...
[14:56:21] <Tom_itx> i used dd to write it
[14:56:40] <jepler> what /dev/xxx did you write it to?
[14:56:57] <Tom_itx> i didn't write it down
[14:57:13] <Tom_itx> it was similar to the instruction page
[14:57:51] <Tom_itx> the files are on the stick
[14:58:08] <cradek> did you use dd?
[14:58:15] <Tom_itx> yes
[14:58:32] <cradek> then it's bootable, it must be a bios thing
[14:58:35] <Tom_itx> checked the BIOS: boot USB first, hdd order: sandisk then SSD
[14:58:48] <Tom_itx> i'll burn a dvd
[14:59:34] <Tom_itx> it's a intel atom 525
[15:05:06] <jepler> cradek: as you probably knew, the sample configs gladevcp_panel and gladevcp_tab both work in binary.hybrid.iso with md5sum 89ce14617d0c1af19fba95e9403af681
[15:05:15] <jepler> cradek: I think the user needs to supply additional information..
[15:05:50] <cradek> it sounds like editing or creating a panel doesn't work as expected, and I don't know what expected is
[15:06:12] <jepler> doop doop: gscreen refers to sound files that don't exist on debian
[15:06:31] <jepler> src/emc/usr_intf/gmoccapy/gmoccapy.py: self.alert_sound = "/usr/share/sounds/ubuntu/stereo/bell.ogg"
[15:06:33] <cradek> it probably shouldn't make sounds anyway
[15:06:40] <cradek> BEEP
[15:10:53] <Tom_itx> those windows directions are just for a USB iso image aren't they?
[15:11:06] <skunkworks__> cradek, http://www.youtube.com/watch?v=8Gv0H-vPoDc
[15:11:43] <skunkworks__> Tom_itx, correct - you can burn a dvd in the normal way.
[15:32:47] <Tom_itx> dvd seems to work
[15:33:18] <cradek> cool, I don't have what it takes to try that, thanks
[15:33:44] <Tom_itx> must be a boot issue with my USB
[15:33:52] <Tom_itx> it _should_ work
[15:34:15] <cradek> this machine next to me would only boot from usb successfully every third try or so. I don't know what the pattern is
[15:34:23] <cradek> also only the first time after power up
[15:34:39] <cradek> something's obviously flaky
[15:34:42] <cradek> it's years old though
[15:35:23] <Tom_itx> i've got an old clunker that would only work with usb with 8.xx
[15:36:20] <Tom_itx> i'm pretty sure i've booted with it with unetbootin
[15:36:50] <Tom_itx> the D525 that is..
[15:38:50] <CaptHindsight> https://www.osadl.org/Latency-plot-of-system-in-rack-4-slot.qa-latencyplot-r4s7.0.html does this latency plot resemble anything seen with preempt_rt and linuxcnc latency test?
[15:39:08] <CaptHindsight> on their D525 boards
[15:50:49] <pcw_d525> sebastian: OK so i have the latest hm2 vhdl source on a linux machine
[15:50:51] <pcw_d525> what next?
[15:52:28] <jepler> pcw_d525: have you created a git clone of the "hostmot2-firmware" repository before?
[15:53:48] <jepler> pcw_d525: I think that what seb_kuzminsky had in mind was for you to start by cloning hostmot2-firmware
[15:53:51] <jepler> git clone git://git.linuxcnc.org/git/hostmot2-firmware.git
[15:54:29] <cradek> no, use ssh://pcw@git.linuxcnc.org/git/hostmot2-firmware.git
[15:54:42] <jepler> oh yes, what cradek said, if you've got git push access set up
[15:55:19] <cradek> (you'll need to have the ssh key we configured previously)
[15:55:20] <Tom_itx> ok i'll copy my configs over this evening and give it a try.
[15:55:27] <Tom_itx> looks like it installed ok
[15:55:30] <cradek> yay
[15:55:47] <Tom_itx> installed a default 3axis sherline
[15:56:00] <Tom_itx> gotta run for now
[15:56:12] <jepler> pcw_d525: and then copy or move the .vhd files into hostmot2-firmware
[15:56:24] <jepler> add them all: git add *.vhd
[15:56:37] <CaptHindsight> ouch the 3.4.9 rtai pae on this new A10 APU is getting >1,000,000 on the 1mS and 200K on the 25uS
[15:56:50] <jepler> create a commit: git commit -m"update formware files"
[15:56:59] <Tom_itx> do all the files still go in the default places?
[15:57:12] <jepler> (if your git is not set up, this may force you to set some git configuration items before it'll let you commit)
[15:57:12] <Tom_itx> mesa bit files etc
[15:57:37] <jepler> pcw_d525: and finally, "git push". the buildbit will try to build the result, and seb says he'll help work out the kinks
[15:58:02] <CaptHindsight> it must be the video driver since ran the same board/APU with gentoo and 3.4.55 rtai and it was ~12K with KMS and software rendering, no HW accel
[15:58:04] <jepler> Tom_itx: I think so .. /lib/firmware/
[15:58:06] <cradek> Tom_itx: that image already has the firmwares and lots of other good stuff installed
[15:58:14] <Tom_itx> k
[15:58:22] <Tom_itx> there's a couple files i'm concerned about
[15:58:25] <Tom_itx> for my install
[15:58:31] <Tom_itx> i'll look it over
[15:58:33] <cradek> Tom_itx: the bad old days of installing firmwares manually are almost over
[15:58:42] <Tom_itx> aww
[15:59:07] <Tom_itx> k. bbl
[16:05:51] <jepler> bbl
[16:07:35] <CaptHindsight> so on really new AMD boards and APU's you have to run off KMS and use software rendering, the gpu drivers kill latency
[16:07:50] <CaptHindsight> run/turn off
[16:09:52] <CaptHindsight> the same board/APU above runs about the same latency with the Livecd since it doesn't use HW accell
[16:42:43] <CaptHindsight> after turning off KMS and removing the hardware drivers 193K on 1mS thread and 23K on the 25uS
[16:43:17] <CaptHindsight> I've never seen the 1mS thread have worse numbers than the 25uS
[16:44:50] <CaptHindsight> using Gentoo and 3.4.55 RTAI with KMS off and using software rendering we get ~13K on both the 1mS and 25uS
[17:10:43] <skunkworks> the little bit of testing - radeon video is even worse with the latest rtai. nvidia seems great
[17:11:47] <pcw_d525> I dont want to overwrite anything in hostmot2-firmware V0, only master
[17:11:49] <pcw_d525> (not sure if i have the right keys on this machine either)
[17:12:46] <cradek> if your ssh:// clone worked, the key is right
[17:21:53] <jepler> pcw_d525: If "git branch" says "* master" then you are on master branch, and any commit you make will not change "v0".
[17:22:09] <jepler> pcw_d525: after you commit, you can verify what will be pushed with: git push --dry-run
[17:22:20] <jepler> you should see that it does not say it will modify v0
[17:49:51] <pcw_d525> must not have the right git stuff setup
[19:29:25] <Tom_itx> anybody happen to know how to set autologin on the new debian ver?
[21:18:58] <atom1> Error: could not insert module /usr/realtime-3.4.9-rtai-686-pae/modules/linuxcnc/hm2.ko: Unknown symbol in module
[21:19:22] <atom1> running my default config from 2.5
[21:22:15] <atom1> sherline.hal:41: insmod failed, returned -1
[21:22:40] <atom1> line 41: loadrt [HOSTMOT2](DRIVER) config=[HOSTMOT2](CONFIG)
[21:23:07] <andypugh> Interesting
[21:23:52] <atom1> i just thought i'd try my config straight from 2.5
[21:24:10] <atom1> i did add a bit file to the firmware dir
[21:24:22] <andypugh> http://www.linuxcnc.org/index.php/english/forum/9-installing-linuxcnc/28055-crash-when-start-emc2-error#49108
[21:24:34] <atom1> and checked to make sure hal_manualtoolchange was indeed updated as i had it
[21:24:41] <andypugh> Is similar, fail-to-launch out of the blue.
[21:25:18] <andypugh> Ah, maybe not so similar if you _changed_ somethig
[21:25:19] <atom1> yeah this was from the .iso i just burned
[21:25:33] <atom1> all i changed was adding a bit file i needed
[21:26:02] <atom1> and checked to make sure hal_manualtoolchange had the physical button code still in it which it does
[21:26:27] <andypugh> I guess that [HOSTMOT](CONFIG) is now not quite right then. Typo in bitfile name?
[21:26:46] <andypugh> Bitile not actually there?
[21:26:48] <atom1> lemme check
[21:27:40] <andypugh> Paste your [HOSTMOT2](CONFIG) line. This might be my fault. :-)
[21:29:27] <atom1> CONFIG="firmware=hm2/7i43-4/SVST2_4_7I47SS.BIT num_encoders=2 num_stepgens=4 sserial_port_0=20xxxxxx"
[21:29:29] <atom1> that?
[21:29:44] <andypugh> Yeah.
[21:30:10] <andypugh> I changed the sserial_port grammar, but I think that’s right.
[21:31:00] <andypugh> does /lib/firmware/hm2/ actually have a 7i43-4 folder?
[21:31:11] <atom1> yes
[21:31:25] <atom1> it just didn't have my custom file
[21:31:57] <andypugh> That won’t help
[21:32:03] <atom1> it's there and spelled correct
[21:32:09] <atom1> case sensitive as well
[21:33:10] <atom1> -rw-r--r-- 1 root root 212490 Jul 25 16:37 SVST2_4_7I47_SS.BIT
[21:33:14] <andypugh> is that _all_ that dmesg says? i think I have seen a clue as to _what_ symbol is not recognised
[21:33:26] <atom1> that wasn't dmesg
[21:33:35] <atom1> that was the error from linuxcnc startup
[21:34:26] <atom1> i'll try to post dmesg
[21:34:51] <andypugh> www.pastebin.org
[21:36:11] <atom1> it says the bitfile wasn't found
[21:37:03] <atom1> one sec, i'll post it
[21:38:39] <Tom_itx> http://paste.debian.net/111724/
[21:39:01] <atom1> the file is in the directory though, i copied it as root
[21:40:11] <andypugh> I wonder if there is an invisible character hiding in the INI file?
[21:40:40] <andypugh> Try without hm2/ as an experiment?
[21:41:03] <atom1> [HOSTMOT2]
[21:41:03] <atom1> DRIVER=hm2_7i43
[21:41:03] <atom1> BOARD=7i43
[21:41:12] <andypugh> copy/pasted from ls to INI ?
[21:41:14] <atom1> then the config line
[21:41:21] <atom1> which?
[21:41:48] <atom1> the pastebin was dmesg > txt file
[21:41:58] <andypugh> I generally hard-code board and config in the HAL. Nothing else needs it, so I don’t see why it belongs in the INI.
[21:42:18] <atom1> i got that from an early example file
[21:42:47] <andypugh> Yes, it’s that way in all the configs. I just don’t like it
[21:42:53] <atom1> heh
[21:43:35] <andypugh> Anyway, what I meant was to run “ls” on the firmware dir then copy/paste the filename into the INI
[21:43:56] <atom1> i'll try it
[21:46:19] <atom1> ahh i see it
[21:46:25] <jepler> hm2.ko sounds weird. hostmot2.ko and hm2_7i43.ko are the typical module names for the hostmot2 main driver and hardware-dependent driver.
[21:46:43] <atom1> there's an _SS at the end instead of just SS
[21:46:59] <atom1> duh
[21:47:41] <atom1> ok 2.6.0-pre5 on Sheline Mill up and running
[21:48:30] <atom1> spindle widget isn't working but that could be something else
[21:49:14] <jepler> atom1: I think the login manager is called lightdm. If so, this documentation will be relevant: https://wiki.debian.org/LightDM#Enable_autologin
[21:49:23] <andypugh> Gah! We both read that so many times and missed the _ !
[21:49:26] <jepler> .. if that's accurate, you have to edit a text file, not use a gui
[21:49:42] <atom1> ok
[21:56:58] <atom1> let's find out...
[22:00:37] <atom1> autologin worked
[22:00:47] <atom1> but none of my default settings are saved
[22:00:53] <atom1> like screen background