#emc-devel | Logs for 2008-08-23

[02:58:37] <jepler> the cool news: cradek's X axis has homed to index + index mask
[02:58:57] <jepler> the bad news: it required recompiling the old hostmot firmware, because its pinout was wrong (the index mask was overlaid on an output pin)
[02:59:09] <cradek> whee!
[02:59:10] <jepler> the good news: I managed to wrestle xilinx ise into building it
[02:59:16] <jepler> the bad news: it says it fails several timing constraints
[02:59:28] <jepler> the good news: it seems to work OK nevertheless
[02:59:41] <jepler> the bad news: the watchdog of the old hostmot is terribly broken (all it does is set pwm outputs to around 0V)
[02:59:54] <jepler> .. I think that sums it up
[03:00:28] <cradek> you have some cool news and some bad news and some good news and some bad news and some good news and some bad news?
[03:00:38] <jepler> yes
[03:00:43] <jepler> which do you want to hear first?
[03:00:53] <cradek> too late to have a relevant opinion
[03:01:04] <jepler> oh
[03:01:08] <jepler> nevermind then
[03:01:14] <cradek> I'm thrilled about the index mask working though
[03:01:15] <jepler> what I'm surprised by is the fact that I made it home before ingrid did
[03:01:23] <cradek> it's a hard machine to home and that fixes it right up
[03:04:35] <jepler> the good news: it looks like hostmot2 applies the watchdog to all output pins, not just the "dac" pwm outputs
[03:04:44] <cradek> nice
[03:05:44] <jepler> it switches the direction register of the port to all-inputs
[03:06:00] <jepler> s/the port/all the ports/
[03:06:09] <cradek> so all the isolated outputs should be in the no-conduct mode
[03:06:28] <cradek> that's definitely what I want
[03:06:50] <jepler> the manpage says they're pulled up -- is that still "no conduct" mode?
[03:07:58] <cradek> well pulled up only means something at the m5i20 I think - what the 7i37 does is another issue, and I assume it does the sane thing
[03:14:05] <jepler> 7i37 outputs are active low, and so are 7i33 enable inputs
[03:32:07] <SWPadnos> jepler, cradek, there should have been a fixed version of the old hostmot in CVS and/or the package
[03:32:44] <SWPadnos> the ones that were made for Anders (or whoever it was that needed the 100MHz clock) had the fix
[03:33:31] <SWPadnos> also, PeteW had sent me a fixed version, but I never committed it because I thought someone else had (or maybe I thought we'd finish the 5i2x driver or something)
[03:45:57] <cradek> which fix specifically do you mean?
[03:46:19] <SWPadnos> for the index mask pinout problem
[03:46:31] <SWPadnos> but maybe I'm thinking of the secondary encoder pinout problem
[03:47:24] <cradek> our driver didn't support index mask, so it wouldn't have really mattered
[03:47:38] <cradek> but secondary encoders are there (or at least their pins are)
[03:47:45] <SWPadnos> hmmm - it's entirely possible that I'm confused
[03:47:55] <SWPadnos> I think the secondary encoders were assigned to the wrong pins
[03:48:16] <SWPadnos> the code had something like pin_num(i+3) instead of (i*4+3)
[03:48:43] <SWPadnos> so they'd get all screwed up as i iterated from 0 to 3
[03:49:03] <cradek> I think we're going to jump to hm2 and put in the features we need
[03:49:15] <cradek> he thinks he can build it (he rebuilt hm4 tonight)
[03:50:11] <SWPadnos> it shouldn't be hard. the hardest thing for me (other than getting the silly Xilinx tools installed) was getting the pin file
[03:50:16] <SWPadnos> err - UCF file
[03:50:24] <SWPadnos> if you have that, then it should be simple
[03:50:31] <cradek> I think we do?
[03:50:50] <SWPadnos> if he built hostmot, then you have one you can start with
[03:51:12] <SWPadnos> it shouldn't need many changes, since you're using the same card and the pin names should be the same
[03:52:56] <pminmo> ise web version has ucf and pin issues if that helps
[03:53:30] <SWPadnos> ?
[03:54:25] <pminmo> documented bugs
[03:54:37] <SWPadnos> oh, interesting. which version(s)?
[03:55:20] <pminmo> webpak version of ise for a while, pace and ucf files
[03:55:53] <SWPadnos> the webpack has gone from version 6.something to 10.somerthing in the last few years - there are a lot of versions :)
[03:56:33] <pminmo> the problem has been there for a while, current version back
[03:56:40] <pminmo> backwards
[03:56:46] <SWPadnos> ok, good to know
[03:57:10] <SWPadnos> I certainly had problems with incremental compile using 9.1 or 9.2 or thereabouts
[03:57:25] <SWPadnos> like, turn it off because I'm paranoid, and it's still enabled
[03:58:47] <SWPadnos> actually, I was having odd problems that seemed to have nothing to do with the code, so I turned that off, but it didn't help
[03:59:09] <SWPadnos> but doing a full build made fully functional code, so I assumed it was some incremental compile bug
[04:00:04] <pminmo> top level changes can get completely screwed up and it shows up when you go to pace with a know good config
[04:00:37] <SWPadnos> I was changing a functional module, not the top level
[04:00:58] <SWPadnos> at least I figured it out - it was slow but reliable to recompile the whole thing
[04:01:50] <pminmo> if pace starts spewing errors when you think it shouldn't look at the modeled code
[04:02:22] <SWPadnos> oh, there were no errors, the module just didn't work (as evidenced by an external scope)
[04:02:32] <SWPadnos> rebuild all, and it magically works, no code changes
[04:03:38] <pminmo> webpack = lots of "quirks"
[04:04:34] <SWPadnos> I should break down and buy the real one, but I was annoyed that the versions that come with the development kits are time-limited licenses
[04:04:45] <SWPadnos> (I could be thinking of Altera though, can't remember)
[04:05:17] <pminmo> I think quartus license model is similar
[04:05:42] <SWPadnos> yeah, come to think of it, it was a NIOS kit that expired on me
[04:05:56] <pminmo> as far as I'm concerned, quartus - free web version is as buggy if not more
[04:06:51] <SWPadnos> yeah, could be. I haven't generated any real designs with it so I can't say
[04:08:47] <pminmo> they both work, but both have issues. anyway I was looking for bigjohnt, talk about docs but he doesn't seem to be on
[04:09:15] <SWPadnos> no, he does seem to be missing
[04:09:21] <SWPadnos> it is a Friday night, after all :)
[04:10:37] <pminmo> Friday nights...........I remember those :-)
[04:10:46] <SWPadnos> had one just last week
[04:10:52] <SWPadnos> didn't do anything, but it was Friday :)
[04:11:16] <pminmo> :-) later
[04:11:24] <SWPadnos> see you
[04:11:53] <CIA-36> EMC: 03jepler 07TRUNK * 10emc2/src/hal/drivers/mesa-hostmot2/firmware/5i20/SOURCE/Makefile:
[04:11:53] <CIA-36> EMC: I haven't tested the resulting bit file (or even made it be generated in the
[04:11:53] <CIA-36> EMC: right location with the right name), but now I can generate one with 'make'
[04:11:53] <CIA-36> EMC: instead of in the ise gui. this makes me go "yay".
[04:12:03] <CIA-36> EMC: 03jepler 07TRUNK * 10emc2/src/hal/drivers/mesa-hostmot2/firmware/xilinx.mk:
[04:12:03] <CIA-36> EMC: I haven't tested the resulting bit file (or even made it be generated in the
[04:12:03] <CIA-36> EMC: right location with the right name), but now I can generate one with 'make'
[04:12:03] <CIA-36> EMC: instead of in the ise gui. this makes me go "yay".
[04:12:51] <jepler> (and thanks so much to jmkasunich_ for originally writing the makefile rules to drive xilinx)
[04:13:05] <SWPadnos> I was just about to ask about that :)
[14:01:07] <jepler> oh the pain of developing hardware drivers
[14:02:35] <cradek> howso?
[14:06:26] <cradek> hm, in new cl, you have a variable (%I0), symbol name (CLinB0), and signalname/comment (>>gui-estop)
[14:07:05] <cradek> I like the display that shows all three, but I would rather see the hal name in the display, like the old behavior
[14:07:23] <jepler> cradek: oh, I oopsed the kernel once and had to reboot
[14:08:15] <cradek> oh
[14:09:18] <SWPadnos> what's really fun is when you're doing lower level development, and writing some thing to some wrong PCI address space causes the computer to die or reboot spontaneously
[14:12:24] <jepler> OK -- I think I've succeeded in adding these to hm2's encoder: index-enable, index-mask, index-polarity, index-mask-polarity, counter-mode
[14:12:43] <jepler> too bad I don't have a servo connected to test any of this
[14:12:51] <SWPadnos> so connect one ;)
[14:12:59] <jepler> bah
[14:13:01] <jepler> too much work
[14:13:06] <SWPadnos> oh, that's too hard, right
[14:13:22] <SWPadnos> I have servos, but haven't got a current EMC install on a machine with a Mesa card
[14:13:23] <cradek> wow
[14:13:29] <cradek> counter-mode!?
[14:13:38] <jepler> yeah, up/down instead of quadrature, apparently
[14:13:58] <cradek> I think that saves me finding an encoder...
[14:14:01] <cradek> (if this one works)
[14:14:07] <cradek> it has only one channel
[14:14:13] <cradek> saved them cents and cents I'm sure
[14:15:29] <alex_joni> heh
[14:15:40] <jepler> oh and also filter (to select between the 3- or 15-clock filter of encoder inputs)
[14:16:01] <cradek> by 9:15 am - that's amazing
[14:16:25] <jepler> hopefully it doesn't take until 2010 to debug the index-enable
[14:16:35] <jepler> time to go find breakfast
[14:16:54] <SWPadnos> 2010 is only 16 months away
[14:17:54] <alex_joni> emc2 2.4 by then?
[14:18:35] <SWPadnos> 3.0!
[14:18:44] <alex_joni> you forgot ftw
[14:18:54] <SWPadnos> profit!
[15:17:45] <jepler> index-enable works except that it doesn't change the index-enable pin when index is seen...
[15:19:29] <CIA-36> EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/common/Document_Header.lyx: standard document header file
[15:20:42] <CIA-36> EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/common/Getting_EMC.lyx: Getting EMC section for Getting Started Guide
[15:25:42] <CIA-36> EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/gcode/main.lyx: add info about manual tool change
[15:27:36] <jepler> ok now that works
[15:27:36] <jepler> too
[15:27:44] <jepler> hm, counter-mode doesn't seem to work
[15:28:40] <CIA-36> EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/gui/axis.lyx: manual tool change
[15:30:56] <CIA-36> EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/config/ini_config.lyx: added servo calib program info
[15:32:55] <jepler> oh wait -- it does, there's just something surprising about it
[15:33:01] <CIA-36> EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/ (Getting_Started.lyx Submakefile): add getting started
[15:33:01] <CIA-36> EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/quickstart/stepper_quickstart.lyx: add getting started
[15:35:49] <cradek> jepler: head out anytime, breakfast is over
[15:35:56] <jepler> (hm2 "counter" mode is: count on rising edge of A in the direction indicated by B -- that's the same as the broken quadrature decoder that everyone writes first)
[15:36:13] <jepler> cradek: OK, half an hour or so probably
[15:36:31] <alex_joni> whee.. sounds like fun
[15:36:35] <cradek> cool
[15:37:14] <alex_joni> jepler: remember to take your camera
[15:37:30] <alex_joni> chris doesn't seem to take any pictures of his retrofit
[15:37:48] <cradek> bill has taken a few. I don't know if he's done anything with them.
[15:38:11] <CIA-36> EMC: 03jepler 07TRUNK * 10emc2/src/hal/drivers/mesa-hostmot2/ (encoder.c hostmot2.h):
[15:38:11] <CIA-36> EMC: expand the hm2 encoder interface, adding a new pin and some new parameters:
[15:38:11] <CIA-36> EMC: pin: index-enable
[15:38:11] <CIA-36> EMC: parameters: index-invert index-mask index-mask-invert filter counter-mode
[15:45:39] <alex_joni> jepler: what is stride?
[15:49:51] <jepler> alex_joni: the distance between the successive places your foot steps while walking or running.
[15:50:08] <alex_joni> * alex_joni tries to apply that to hostmot2
[15:50:11] <jepler> in the case of hm2 I think it's the distance in the address space from one register to another
[15:50:16] <alex_joni> aha
[15:50:31] <jepler> so if I want to write the register for the i'th encoder whatsis, I have to go i*stride bytes past the first register
[15:50:42] <alex_joni> I see... nice
[15:50:58] <jepler> cradek: anything else I need to bring out besides my laptop and its power supply?
[15:51:11] <cradek> a network cable if you have one
[15:51:19] <alex_joni> and good mood
[15:51:22] <jepler> ok
[15:59:24] <jmkasunich_> HNC work?
[16:02:28] <alex_joni> sounds like it
[16:13:33] <pminmo> help cvs joht said for me to do a .compile and make to . compile i get no command found is that a cvs command?
[16:13:42] <pminmo> or?
[16:14:43] <alex_joni> pminmo: not making a lot of sense
[16:15:43] <pminmo> thats normal for me... I'm trying to get some document updates John did from cvs
[16:16:05] <pminmo> I have a trunk version build a couple of days ago
[16:16:30] <pminmo> email said to do a cvs update which i did and worked
[16:16:53] <pminmo> then he said to do the .compile and make
[16:17:51] <jmkasunich_> not compile
[16:17:52] <jmkasunich_> configure
[16:17:57] <jmkasunich_> ./configure
[16:18:29] <alex_joni> pminmo: you need to cd to the src directory
[16:18:32] <jmkasunich_> you need to do that in the src/ directory, one level down from the top of your CVS checkout
[16:18:37] <alex_joni> and there you'll find a ./configure script
[16:20:03] <pminmo> ok, thanks didn't realize it was a script
[16:20:18] <alex_joni> well.. the idea is that you can run it
[16:20:28] <alex_joni> it doesn't matter if it's a script or a binary file :)
[16:22:23] <jepler> http://emergent.unpy.net/index.cgi-files/sandbox/hm2.hal
[16:23:44] <skunkworks> working?
[16:26:00] <pminmo> yea kind of, it did the compile and make but I didn't get the new doc but I think I needed to do the make in a different context
[16:26:21] <jepler> http://emergent.unpy.net/index.cgi-files/sandbox/hm2-pins
[16:27:22] <skunkworks> wow
[16:27:39] <SWPadnos> I wonder why seb chose to use 3 digits for the GPIO pin numbers
[16:27:50] <SWPadnos> taht is a lot of pins and parameters, I must say :)
[16:27:52] <SWPadnos> that
[16:29:01] <SWPadnos> oh, interesting - I misread the numbers. I thought those were ...Px.<the pin number>, but they're Px.<an ascending number across all connectors>
[16:29:59] <skunkworks> This is big
[16:30:08] <SWPadnos> jepler, did you specify that all GPIOs should be inputs, or is that the default?
[16:30:14] <SWPadnos> * SWPadnos didn't see any outputs
[16:31:16] <skunkworks> where else can you get hardware that can be used as either servo/stepper/both
[16:31:32] <SWPadnos> parallel port! :)
[16:32:09] <skunkworks> well... You know what I mean
[16:32:11] <skunkworks> ;)
[16:33:01] <SWPadnos> it could be interesting, but more or less usless, to make a very simple PCI I/O card that just has 32 or 64 ins and 32 or 64 outs - half a dozen instructions and you've done a full update
[16:33:25] <SWPadnos> so you'd have insanely low execution time, but still terrible jitter, for the fast thread
[16:33:50] <SWPadnos> gotta run - time for food with the wife :)
[17:09:19] <alex_joni> hmm.. is that the guy with the 297/298 drivers?
[17:14:01] <skunkworks> pmino? yes
[17:14:51] <skunkworks> pminmo
[17:15:55] <skunkworks> he has used to use mach.. says he gets much better performance from emc
[17:24:13] <alex_joni> performance without a proper metric is useless as a term
[17:26:08] <jmkasunich_> what's this metric stuff? don't you know performance is measured in inches!
[17:32:17] <fenn> hoss powwah
[17:40:21] <jmkasunich_> ok, thats weird - axis just locked up
[17:41:38] <jmkasunich_> ok, thats _really_ wierd: while I was sitting over here, it "unlocked" and executed all the commands that I had issued
[17:41:46] <jmkasunich_> which included starting and stopping the spindle
[17:42:53] <jmkasunich_> * jmkasunich_ just stopped trusting his machine
[17:46:53] <logger_dev> * logger_dev is already logging
[17:47:25] <AlanC> hola
[17:48:17] <jmkasunich_> hi
[17:48:26] <AlanC> hello... i get the next message when i try to install emc2 on an hardy 8.04 -64 bit... http://www.linuxcnc.org/emc2/dists/dapper/Release Unable to find expected entry emc2.2/binary-amd64/Pachages in Meta-Index file (malformed Release file?)
[17:48:38] <AlanC> what do you think i should do?
[17:49:20] <jmkasunich_> I'm not very familiar with the packaging
[17:49:29] <jmkasunich_> I'm not sure there are binary packages for 64 bit
[17:50:23] <AlanC> well i tried to find a package for 64 but the binary wasn't there... i mean there's only a binary for 32...
[17:50:44] <AlanC> is there some way to install it without formatting my lap...
[17:51:00] <jmkasunich_> if there is no binary for 64, then you cannot do a binary install on 64
[17:51:18] <cradek> the 64 bit kernel and rtai and emc2 are all pretty experimental
[17:51:32] <cradek> I wouldn't use them for running a machine
[17:53:12] <jmkasunich_> cradek: any idea why a running user space process (axis) would just "go away" for 30 seconds, then start running again?
[17:53:36] <jmkasunich_> it came back before I could run top to see if it was usy
[17:53:36] <cradek> not really
[17:53:37] <AlanC> mmm so i'm owned
[17:53:53] <cradek> I would also not use a laptop to run machinery
[17:53:58] <jmkasunich_> AlanC: do you need to run 64 bit for some other reason?
[17:54:00] <cradek> you can run the simulator mode on any of this of course
[17:54:13] <AlanC> yeah well i just want the simulator mode
[17:54:37] <jmkasunich_> well, if you just want simulator mode, things get a lot simpler
[17:54:42] <AlanC> you mean there is a simulator mode that can be installed on hardy 64?
[17:55:05] <jmkasunich_> I don't know if there is a binary package for it, but sim mode will compile on many flavors of linux, including 64 bit
[17:55:14] <jmkasunich_> no special kernel or rtai is required for sim
[17:55:20] <cradek> yeah you will have to build it
[17:55:27] <AlanC> how... can i get that
[17:55:53] <AlanC> sorry i'm new to linux
[17:56:10] <jmkasunich_> gimme a minute and I'll give you a url
[17:56:59] <AlanC> thanks
[17:57:18] <jmkasunich_> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Installing_EMC2#On_Ubuntu_6_06_or_8_04_from_source
[17:57:40] <jmkasunich_> you should skip step 2.3, and do 2.4 instead
[17:58:04] <fenn> btw this might be easier: http://www.linuxcnc.org/dapper/emc2-install-sim.sh
[17:58:16] <AlanC> excelent let me try that
[17:58:56] <cradek> that will give you the 32 bit packates
[17:59:11] <jmkasunich_> and they are compiled for dapper, not hardy
[17:59:32] <cradek> I think there are hardy 32 sim packages
[17:59:48] <cradek> I don't know whether those will run on your 64 machine.
[17:59:52] <jmkasunich_> well, that script fetches dapper ones
[17:59:57] <AlanC> it didn't work
[18:00:04] <fenn> sorry i thought he was on dapper
[18:00:17] <jmkasunich_> you should follow the instructions on the webpage I posted
[18:01:32] <AlanC> ok
[18:02:10] <alex_joni> there are no sim packages for hardy
[18:02:22] <alex_joni> at least none that I know of
[18:02:46] <alex_joni> so the instructions from jmkasunich_ is what you need to follow
[18:04:06] <AlanC> ok so i have to do all steps except 2.3
[18:05:10] <alex_joni> you don't need 2.5 either
[18:10:38] <CIA-36> EMC: 03cradek 07TRUNK * 10emc2/src/emc/usr_intf/axis/scripts/axis.py: remove debug prints
[18:13:17] <AlanC> defeated
[18:13:39] <jmkasunich_> ?
[18:14:40] <jmkasunich_> AlanC: if you tell us what actually happened, somebody might be able to help you
[18:14:48] <jmkasunich_> "defeated" doesn't mean anything
[18:16:28] <AlanC> i'm just doing the steps, but i get stuff like thi
[18:16:29] <AlanC> s
[18:16:47] <AlanC> cvs -z5 -d:ext:anon@cvs.linuxcnc.org:/cvs co -rRELEASE_2_1_6 -demc2.1.6 emc2
[18:16:47] <AlanC> The authenticity of host 'cvs.linuxcnc.org (' can't be established.
[18:16:47] <AlanC> DSA key fingerprint is 5b:b7:3d:04:9d:dd:86:ad:dc:c6:78:81:d4:3d:05:8e.
[18:16:47] <AlanC> Are you sure you want to continue connecting (yes/no)?
[18:16:47] <AlanC> Host key verification failed.
[18:16:48] <AlanC> cvs [checkout aborted]: end of file from server (consult above messages if any)
[18:17:12] <alex_joni> AlanC: when it says are you sure you want to continue connecting ..
[18:17:16] <alex_joni> you need to type 'yes'
[18:17:37] <AlanC> oh
[18:17:40] <AlanC> hehe
[18:18:03] <AlanC> ok i'll start everything again
[18:18:41] <jmkasunich_> do one step at at time - if it looks like there was a problem, don't waste time moving to the next step - ask for help
[18:19:35] <AlanC> what means this
[18:19:35] <AlanC> You need to add the linuxcnc repository to your /etc/apt/sources
[18:24:02] <CIA-36> EMC: 03alex_joni 07TRUNK * 10emc2/debian/emc2.files.in: add tormach sample configs
[18:25:30] <alex_joni> AlanC: you need to edit /etc/apt/sources.list
[18:25:35] <alex_joni> and add a couple lines
[18:25:49] <AlanC> W: Imposible obtener cdrom:[Ubuntu 8.04 _Hardy Heron_ - Release amd64 (20080423)]/dists/hardy/main/binary-amd64/Packages.gz Por favor utilice apt-cdrom para hacer que APT reconozca este CD. apt-get update no se puede usar para agregar nuevos CDs
[18:25:49] <AlanC> W: Imposible obtener cdrom:[Ubuntu 8.04 _Hardy Heron_ - Release amd64 (20080423)]/dists/hardy/restricted/binary-amd64/Packages.gz Por favor utilice apt-cdrom para hacer que APT reconozca este CD. apt-get update no se puede usar para agregar nuevos CDs
[18:25:49] <AlanC> W: Imposible obtener http://www.linuxcnc.org/emc2/dists/dapper/Release Unable to find expected entry emc2.2/binary-amd64/Packages in Meta-index file (malformed Release file?)
[18:25:51] <AlanC> E: Algunos archivos de índice no se han podido descargar, se han ignorado,
[18:25:53] <AlanC> o se ha utilizado unos antiguos en su lugar.
[18:25:55] <AlanC> Leyendo lista de paquetes... Hecho
[18:25:57] <AlanC> Creando árbol de dependencias
[18:25:59] <AlanC> Leyendo la información de estado... Hecho
[18:26:01] <AlanC> E: No se pudo encontrar el paquete emc2
[18:26:53] <AlanC> Ok i'm just doing the steps but i don't know that even thus i get all this error messages will it work...
[18:27:06] <alex_joni> nope, it won't
[18:27:12] <alex_joni> AlanC: you need to edit /etc/apt/sources.list
[18:27:14] <alex_joni> and add a couple lines
[18:27:17] <alex_joni> deb http://www.linuxcnc.org/hardy hardy base emc2.2
[18:27:17] <alex_joni> deb-src http://www.linuxcnc.org/hardy hardy base emc2.2
[18:28:07] <AlanC> haha
[18:28:19] <AlanC> so i paste what?
[18:28:50] <alex_joni> * alex_joni is busy doing other things
[18:30:14] <AlanC> well people thank you very much for your help i thing i'd better wait until there is a version for 64... probably it was an error to install hardy 64.
[18:32:17] <CIA-36> EMC: 03alex_joni 07TRUNK * 10emc2/debian/emc2.files.in: add 2 more files recently got added
[18:38:22] <SWPadnos> you know, on the "restart at line" subject, it might be nice to have a "stop when this line completes" feature
[18:38:47] <SWPadnos> that way, you don't have to deal with the problems of restarting in the middle of a move, especially an arc move
[18:39:03] <alex_joni> I think you can..
[18:39:14] <alex_joni> using step
[18:39:14] <SWPadnos> I know you can if you have optional stop codes
[18:39:18] <alex_joni> or that
[18:39:29] <SWPadnos> ok, so when it's running, you hit step and it'll stop at the end of the move?
[18:39:34] <alex_joni> hmm.. not sure
[18:39:38] <SWPadnos> yeah, dunno
[18:39:41] <alex_joni> maybe it was pause/step
[18:39:49] <alex_joni> working on packages atm, so I can't test
[18:40:34] <SWPadnos> yep - I noticed. thanks for adding the Tormach config - I totally spaced on the .deb stuff
[18:41:43] <alex_joni> np
[18:41:53] <alex_joni> I just noticed, cause I'm adding sim stuff for hardy
[18:43:19] <CIA-36> EMC: 03rayh 07TRUNK * 10emc2/src/emc/usr_intf/halui.cc: oops fix relative z typo.
[18:44:10] <SWPadnos> cool. I just put a new HD in my laptop and installed Hardy, so I'll finally be able to have a real sim install (with menu items for manuals and everything :)
[18:44:48] <alex_joni> coming up in a couple of minutes
[18:48:14] <CIA-36> EMC: 03alex_joni 07TRUNK * 10emc2/debian/extras-sim-Ubuntu-8.04/emc2.files: files for sim on 8.04
[18:48:14] <CIA-36> EMC: 03alex_joni 07TRUNK * 10emc2/debian/extras-sim-Ubuntu-8.04/etc/xdg/menus/applications-merged/cnc.menu: files for sim on 8.04
[18:48:17] <CIA-36> EMC: 03alex_joni 07TRUNK * 10emc2/debian/extras-sim-Ubuntu-8.04/usr/share/applications/ (5 files): files for sim on 8.04
[18:48:21] <CIA-36> EMC: 03alex_joni 07TRUNK * 10emc2/debian/extras-sim-Ubuntu-8.04/usr/share/desktop-directories/cnc.directory: files for sim on 8.04
[18:48:22] <CIA-36> EMC: 03alex_joni 07TRUNK * 10emc2/debian/extras-sim-Ubuntu-8.04/usr/share/pixmaps/emc2icon.png: files for sim on 8.04
[18:48:34] <SWPadnos> I really need to work on a reliable RT+SMP kernel
[18:48:48] <alex_joni> try the amd64 jeff did
[18:48:55] <SWPadnos> simce I have a dual-core laptop and several other dual or quad cores
[18:48:59] <SWPadnos> ok, I will
[18:49:10] <SWPadnos> hmmm - true. I did install 64-bit
[18:50:56] <SWPadnos> I've used the experimental SMP kernel that cradek made a long time ago, and it works nicel
[18:50:57] <SWPadnos> y
[18:51:40] <SWPadnos> argh. for some reason Windows decided that I shouldn't be able to ssh into dreamhost
[18:52:21] <alex_joni> on dinero?
[18:52:35] <SWPadnos> hmmm. or maybe running from the "run program" dialog is a problem or something
[18:52:45] <alex_joni> putty ftw
[18:52:46] <SWPadnos> it works when I run it from a Take Command prompt
[18:52:53] <SWPadnos> yeah, using putty - of course :)
[18:53:05] <SWPadnos> I did just upgrade to 0.60
[18:55:10] <cradek> http://timeguy.com/cradek-files/emc/encoder-index.fix.patch
[18:57:14] <SWPadnos> does that actually work correctly?
[18:57:14] <CIA-36> EMC: 03alex_joni 07v2_2_branch * 10emc2/debian/extras-sim-Ubuntu-8.04/etc/xdg/menus/applications-merged/cnc.menu: new sim files for 8.04
[18:57:15] <CIA-36> EMC: 03alex_joni 07v2_2_branch * 10emc2/debian/extras-sim-Ubuntu-8.04/usr/share/applications/ (5 files): new sim files for 8.04
[18:57:15] <CIA-36> EMC: 03alex_joni 07v2_2_branch * 10emc2/debian/extras-sim-Ubuntu-8.04/usr/share/desktop-directories/cnc.directory: new sim files for 8.04
[18:57:16] <CIA-36> EMC: 03alex_joni 07v2_2_branch * 10emc2/debian/extras-sim-Ubuntu-8.04/usr/share/pixmaps/emc2icon.png: new sim files for 8.04
[18:57:44] <SWPadnos> ah, right. prev_count needs to be set to 0 so the present count will become the actual count later
[18:59:25] <jmkasunich_> hmm, I think I talked you guys into a bad decision at the workshop last year
[18:59:36] <SWPadnos> heh
[18:59:39] <jmkasunich_> motion.spindle-speed-out is documented as being in revs/minute
[18:59:39] <rayh> How's that?
[18:59:57] <jmkasunich_> motion.spindle-speed-in is used for feed/rev, and seems to be in revs/second
[19:00:03] <SWPadnos> it should be RPS, like all the other HAL units are
[19:00:16] <jmkasunich_> well, that is debatable
[19:00:20] <SWPadnos> yes
[19:00:27] <SWPadnos> since everyone knows RPM and hot Hz
[19:00:28] <SWPadnos> not
[19:00:29] <rayh> Darn and RPM is so much more common.
[19:00:30] <jmkasunich_> but having the two pins using different units is not right
[19:00:44] <alex_joni> jmkasunich_: it was something I picked a while ago
[19:00:48] <alex_joni> and no one complained :(
[19:00:52] <SWPadnos> heh
[19:01:00] <alex_joni> I agree it's not consistent..
[19:01:01] <jmkasunich_> oh, I thought it was jeff at the workshop last year
[19:01:07] <SWPadnos> don't look a gift patch in the screen? :)
[19:01:15] <alex_joni> I think he did speed-in
[19:01:30] <jmkasunich_> he did feed per rev, which added feed in
[19:01:34] <alex_joni> right
[19:01:42] <alex_joni> ok.. so RPM is older :)
[19:01:46] <jmkasunich_> right
[19:01:47] <alex_joni> causing a precedent
[19:01:49] <alex_joni> :)
[19:01:54] <jmkasunich_> and rpm is far more familiar to users
[19:02:40] <jmkasunich_> I talked jeff into using RPS for speed-in, because that is what you get from "encoder.0.velocity"
[19:03:02] <jmkasunich_> in hindsight that was dumb, but we can't change it now (could change it for 2.3 I suppose)
[19:03:21] <jmkasunich_> in any case, we need to document it
[19:04:10] <jmkasunich_> imagine my astonishment when my lathe tried to cut at 60x the commanded feed/rev
[19:04:15] <jmkasunich_> (fortunately I was cutting air)
[19:22:56] <alex_joni> jepler: any idea why there's a /usr/lib/emc/tcl/hal.so ?
[19:23:09] <alex_joni> as opposed to most .so's that are in /usr/lib/emc2 ?
[19:43:39] <CIA-36> EMC: 03alex_joni 07v2_2_branch * 10emc2/debian/changelog: note sim package for 8.04
[20:05:43] <CIA-36> EMC: 03jepler 07TRUNK * 10emc2/src/hal/drivers/mesa-hostmot2/encoder.c: fix encoder -- the prior version didn't work right if the count ever got very big
[20:08:57] <skunkworks> * skunkworks invisions a mini cncfest at cradeks this weekend.
[21:48:51] <cradek> I'm seeing a nasty jump at index time when I home to index: http://timeguy.com/cradek-files/emc/index-jump.png
[21:52:18] <cradek> I just replaced the image with a better one
[21:54:10] <cradek> I only get this the first time I home after starting EMC, and the further from home the axis is when I start, the worse the jump is
[21:57:27] <CIA-36> EMC: 03cradek 07TRUNK * 10emc2/src/emc/motion/control.c: honor [TRAJ] ACCELERATION settings in free mode planner
[22:04:54] <cradek> this is trunk/hal_m5i20
[22:07:10] <cradek> in fact, if I have my ferrors set reasonably, it ferrors
[22:16:28] <jmkasunich_> cradek: that is "normal"
[22:17:14] <jmkasunich_> when you hit the index, position feedback jumps from wherever it was to zero
[22:17:24] <jmkasunich_> the first time, "whereever" can be very far away
[22:17:31] <jmkasunich_> subsequent times, it is very close
[22:17:43] <cradek> ouch
[22:17:53] <jmkasunich_> anyway, when EMC sees the jump, it resets its internal offsets so that the command makes the same jump
[22:17:56] <cradek> if it doesn't ferror, it's very violent
[22:18:06] <jmkasunich_> so the error as seen by PID should remains small
[22:18:39] <jmkasunich_> however, the step in command will be multiplied by your FF1 gain and appear on the PID output for one period
[22:19:04] <jmkasunich_> I'm surprised that it is "violent"
[22:19:17] <SWPadnos> high torque servos
[22:19:24] <cradek> it jerks the screw 1/4 turn and then back
[22:19:31] <jmkasunich_> something is fscked then
[22:19:59] <jmkasunich_> the huge jumps seen on your plot are "normal", we need to look closer
[22:20:07] <cradek> it's extremely high resolution. maybe it's just that the counts are very high
[22:20:08] <jmkasunich_> can you capture pid.error and pid.output
[22:20:14] <cradek> pid.error is there
[22:20:22] <cradek> err axis.ferror is there
[22:20:28] <jmkasunich_> oh, the purple one?
[22:20:37] <cradek> yes
[22:20:52] <jmkasunich_> it doesn't get large for many mS after the jump
[22:20:56] <jmkasunich_> that is NOT what I would expect
[22:21:13] <jmkasunich_> I'd expect a huge PID output due to the FF1 term, that dies away immediately
[22:21:21] <jmkasunich_> (1 servo period)
[22:21:30] <cradek> I see enc-count is low millions before index
[22:22:28] <jmkasunich_> what you are seeing is weird, and we need more info
[22:22:52] <cradek> ok, I will try
[22:22:57] <jmkasunich_> the large jump does NOT in itself cause large error, and cannot be directly causing something that happens 10s of mS later
[22:23:18] <jmkasunich_> it might be indirect - the FF1 for 1mS kicks the axis, and it takes many mS to stop moving
[22:23:50] <jmkasunich_> but the error seems to be increasing exponentially, which means the "kick" is still going on 10mS or more after the jump
[22:24:04] <cradek> yeah I wish you could scale/zoom :-)
[22:24:12] <jmkasunich_> turn off the count
[22:24:27] <jmkasunich_> keep pos cmd, and fb, add pid error, and pid output
[22:24:36] <jmkasunich_> well, you already have pid error
[22:24:48] <jmkasunich_> you might want to turn on the debug PID pins and look at them
[22:24:57] <jmkasunich_> man pid - I don't recall the details
[22:25:20] <jmkasunich_> I have two passes left on my bore, then I'll be back - don't want to get distracted this close to the end
[22:25:24] <cradek> ok
[22:30:12] <jmkasunich_> actually, one pass - cutting now
[22:35:31] <cradek> http://timeguy.com/cradek-files/emc/jump2.png
[22:36:11] <jmkasunich_> murphy lives
[22:36:28] <cradek> oh no
[22:36:44] <cradek> oversize?
[22:36:55] <jmkasunich_> prior three passes hit their targest by 0.0001, 0.0004, and 0.0001, this one came in 0.0012 over
[22:37:23] <jmkasunich_> I can live with it, but its annoying
[22:37:26] <cradek> fffffffff
[22:37:30] <cradek> that sucks
[22:37:46] <jmkasunich_> I'll scrape the surfaces where the top cap bolts onto the base - take a thou off there
[22:38:27] <cradek> are you sure you didn't get the wrong mark on the boring head?
[22:39:58] <cradek> wonder if I have my servo thread stuff misordered somehow
[22:40:04] <cradek> could that do this?
[22:40:10] <jmkasunich_> possibly
[22:40:27] <cradek> should pid calc before or after motion?
[22:40:33] <jmkasunich_> it should be: read encoders, run command, run control, do pid, write dacs
[22:41:13] <cradek> aha
[22:41:18] <cradek> I had pid before motion
[22:41:29] <cradek> that would make it uncorrected for a period I think
[22:41:40] <jmkasunich_> if you run PID before motion, then it will see the fb jump without a corresponding cmd jump, huge error, cranks up the integrator
[22:42:02] <cradek> where are we going and why are we in this handbasket?
[22:42:50] <cradek> that fixed it!
[22:42:57] <cradek> thanks, that was stupid
[22:43:08] <cradek> I wonder if the right order is documented somewhere.
[22:44:00] <jmkasunich_> probably not
[22:44:18] <jmkasunich_> I thought everybody knew the right order ;-)
[22:44:26] <cradek> yep now it can home from 3" away with a .001 ferror limit
[22:45:23] <cradek> 3" ~= 1 megaencodercount
[22:46:40] <cradek> on to Z wiring...
[22:54:20] <BigJohnT> why does this page http://linuxcnc.org/docview/html/ say it refers to 2.2.7~cvs?
[23:08:22] <cradek> it's generated from the cvs branch v2_2_branch, and that's what the version number there says
[23:08:57] <cradek> it's probably a feature, since then doc fixes are available to the users between the stabel releases
[23:09:00] <cradek> stable
[23:10:17] <BigJohnT> ok thanks
[23:40:02] <BigJohnT> I've added getting started to docs.xml and index.tmpl but I get this error when I run make make: *** No rule to make target `objects/Getting_Started.xml', needed by `objects/xref.xml'. Stop.
[23:40:15] <BigJohnT> what am I missing?
[23:43:15] <BigJohnT> I also added it to the submakefile
[23:50:45] <BigJohnT> http://pastebin.ca/1182808 the changes I made
[23:52:12] <BigJohnT> * BigJohnT goes to make some pizza