good job alex - emc up and runniing the tort.ncg :)
SkunkworksD: how about the booting?
was it still black?
or did you get a splash ?
sorry - there is stll no splash
can you pastebin a dmesg ?
(could it be my issue?
not sure, but I want to make sure
it's probably not you
but I need to know what hardware you have, etc (most is in dmesg)
SkunkworksD: run cat /proc/cmdline
paste in here (one line)
root=/dev/hda1 ro quiet splash
ok, so 'splash' is in there.. so far so good ;)
[12:34:23] <SkunkworksD> http://pastebin.com/765049
this is a ecs p4vmm2 motherboard with 1gb
CPU: Intel(R) Pentium(R) 4 CPU 2.40GHz stepping 09
with the latency test it gets tons of overruns - (right now it is using the onboard video)
was that from the dmesg?
so far so good - nice
skunkworks: I'll try to ask smarter people about the splash
will get back to asking you if I need more info.. ok?
I will be around. :)
skunkworks: neat, how's stability/performance/general behavior?
no issues yet - this is just a test machine so I am not actually controlling anything.
other than no splash screen on bootup - but we don't know if that is my issue or not.
skunkworks: does it output signals correctly to the parport/IO-card?
I am just running the sims right nwo
plus this machine has a onboard video card which doesn't seem to play with rt well.
skunkworks: I think it's not your issue, as you say the default kernel does produce a boot splash..
oh, onboard video tends to mess up with RT?
Lerneaen_Hydra: big time
that could explain the strange behavior I've been having
SWPadnos says mainly it is the shared memory issue
pauses (0.1sec) during a g1 move for instance
Lerneaen_Hydra: have you run the rt latency test yet?
yes, but it didn't give very strange values
yep. my panel PC has "integrated video", but the frambuffer is actually on the video controller chip
and I have no problems with that machine
26k as the OVL max
no overruns? where you opening and closing windows (like a web browser)
opening gimp, oo.org, moving windows, etc etc
try an opengl app
like one of the gl screensavers
my nvidia is fine until I run something with gl
how much of the integrated graphics would you need to disable?
it pauses for .1 second in the middle of a move?
oh, I'll test that
skunkworks: yes, something like 1/10th
Lerneaen_Hydra: typically you just plug in a pci/agp card and it disables
oh, that's enough? How do I get breezy to update and find the new card?
you tell apt to reconfigure X, I'd have to search the ubuntu wiki for the exact command
ok, I'll play with it a bit. What type of card would be suitable for an EMC machine? A simple agp card?
pretty much anything, even nvidia works with the open "nv" driver
would an old ati be ok with teh default driver?
not the closed-source one
unless you run into the problems that A-L-P-H-A and I have been hitting
I have a Radeon 8500DV, and X only works sometimes, and it locks while the screensaver is running (maybe when it tries to put the monitor into power-save mode)
it was a rage turbo or something, an agp card from 1998
would an old matrox card be better?
this machine worked forthe most part with breezy, but would still lock sometimes after the screensaver ran
Matrox would probably be better - something in the G100-G500 range
we put a matrox in the Mazak, and I have a matrox in my mill's machine
so no hardware need for opengl?
if only thay had released programming info on the new cards - I'd probably have a Parhelia in my Linux machines
AFAICT it only has support for basic 2d stuff
HW openGL tends to wreak havoc on RT performance
at least, the NVidia driver does
SWPadnos: the hardware GL ATI in my laptop, by good luck, works fine
cool. what chip is that?
lspci -vv | grep -i ati ;)
it's an oldish laptop
you mean go out to the car, bring in the laptop, plug it in and wait for it to boot, then lspci
it's probably a rage mobility something-or-other then
can't you just remotely power it up and ssh into it? ;)
are you using the ati driver ot fglrx?
probably fglrx, with HW openGL
I'm using the open one whatever it's called
ah -ati or radeon
or did you install gatos drivers?
I'll bring the thing in later and we can examine it
ok. any time is fine
I don't remember what I did, but it was a bit hairy and involved building some kernel modules
that shouldn't be aq problem ;)
yeah it wasn't bad but it was finicky
* alex_joni heads home
the install used .a instead of .so (or vice versa) so a lot of the files it installed weren't picked up by the x server
so I dinked with it until it worked
uh - oh
I hate that
"let's fight the system, instead of the problem"
now I will try not to change it for a long time.
reminds me of LabView ;)
oh, you havn't seen weeble & bob have you (cradeks favorite ;) )
[19:27:37] <Lerneaen_Hydra> http://www.weebl.jolt.co.uk/pie.htm
Lerneaen_Hydra: I haven't seen that in ages
jepler: just as... interesting... every time?
it never changes
[19:50:32] <jepler> http://www.weebl.jolt.co.uk/100.htm
as strange as usual
poor donkey :(
* Lerneaen_Hydra finds it rather humerous that a completely OT discussion is here in #emc-devel, while a very emc-orientated discussion is going on in #emc
SWPadnos: yay, Shon says his axis drift with arcs was just operator error
SWPadnos: you said earlier you figured it was being caused by misunderstanding about g92, and I bet you were right
cool. I noticed that email as well
cradek: of all the strnage "bugs" such as the axis drift, and my "issues" with G92 and touch off, how many are just operator error, rather than b0rken code?
Lerneaen_Hydra: so far, arguably all of them
Lerneaen_Hydra: there was some borked code in emc, but that was long ago :D
I suspect that G92.x was working according to spec, though the spec may only be understandable by the spec writer
cradek: looks like I (and many others) need to read through the documentation more in depth
that's what I think about G92 too.
SWPadnos: it's all in the docs
hmm, that's another way to look at it
the evidence mounts that maybe the spec for g92 should be revised
the spec DOES spell out the operation pretty carefully
I saw that, that's the only reason I know anything about G92.x ;)
that doesn't mean the way it works isn't stupid.
and in that argument I have no opinion :-)
it made sense to me, though I think one or two things weren't well thought out, like what happens when emc is restarted and parameters 5211-5216 are nonzero
or not well documented at least
I didn't find that documented (but it might be)
I find the "set g92.2 in axis, exit emc, start axis again, and g92 is suddenly applied again" to be the main thing that surprised me
Lerneaen_Hydra: I agree
that behavior seems odd compared to m2 which unapplies it
right - the surprise is that the offsets (in the parameter file) cause G92 to be turned on, and that's not explicitly stated in the docs
cradek: random idea; maybe the values for G92 should be shown somewhere onscreen, as well as whether they're applied or not?
* alex_joni ponders about shutdown
it would make more sense for those to get reset to 0 on shutdown or startup
I disagree about that
lol, crazy idea: how about having another tooltip (grayed out, hardly visible) that is in the original place (without G92 applied)
I agree that is crazy
I meant emc startup/shutdown, not hte GUI
cradek: it's late ;)
alex_joni: I find that rather cumbersome, it would be hard to see which value it is, you would just see A value
but it would make a lot of sence after enough beers
oh, another user's random musings. What does the home button in AXIS really do? (I should really read through the docs someday. soon.)
Lerneaen_Hydra: it homes the machine
sets coordinates to 0
at least on the selected axis
moves the axes to a home position or changes the value of the axis?
how that workes depends on the setup
on simple machines it just changes the value to 0
on complicated ones it moves to a switch, backs off, moves on the switch again, watched for an encoder signal, etc
but with home switches it's suitable for it to move to the physical switch?
the precise answer is it initiates a homing procedure on the selected axis: that can mean many things depending on your emc setup.
and the offset function is like touch off, except it changes it's value to 0 and not any value you want to enter=
cradek: so you could have a complicated script with g03, m8, m3, and so on in the homing procedure?
homing is a machine state in the motion controller
no, homing means certain things
configured by certain ini file references
it's nicely documented but in a strange place - I think alex is working on moving it
what are you limited to be able to do? set axis to a value, move said axis to a certain position, and move said axis untill IOpin X is high/low?
cradek: that sounds good
right now it's there.. I plan to include it in the emc2_user_manual soon
but not tonight
g'night all (it's late here too)
* Lerneaen_Hydra notices that the backlash compensation page is very empty
[20:36:28] <alex_joni> http://sourceforge.net/tracker/index.php?func=detail&aid=1460779&group_id=6744&atid=106744
that sounds like the problem I had too
anyway, gnight all
I suppose we could put G92.2 or G92.3 in our startup codes as desired
or maybe even m2
cradek: noticed that bug report?
yes I've seen it before
ok.. just seen it
I still think the BC moves should be added as segments before TP, then TP will blend/smooth them and honor all constraints
JMK disagrees wholeheartedly
the problem with my scheme is we would have to split arcs at quadrants
I see, but fail to comprehend now ;)
ok goodnight :-)
isn't the answer "set min_ferror larger than backlash, and let stepgen's constraints make the backlash compensation happen at the fastest acceptable rate?
cradek: does this look ok? http://sourceforge.net/forum/forum.php?forum_id=577599
[20:43:48] <alex_joni> http://sourceforge.net/forum/forum.php?forum_id=577600
yes looks great, thanks
fenn: the other day you suggested i start a list of suppliers and whatnots. what format does something like that need to be in?
how about: name in bold, followed by relative price, selection, and speed
and a representative list of the kind of products they sell
with a link to their page when you click on the name
a simple text doc?
i was thinking on the wiki
i'll make a template for you to copy
i have no idea what format that requires. but given an example, i'm sure i can match most any format.
wiki formatting is quite easy to learn
usemod (the wiki software) is somewhat annoying, and i cant get rid of the brackets around names, so you're going to have to live with it i think
[22:41:00] <fenn> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Suppliers
ok, so dumb question. how do i use this?
I wish this part were more obvious
[22:42:22] <fenn> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?BasicSteps
CamelCase is turned off so you have to use [[ ]]'s to make links
thanks, i'll look at it, and put together a good list before mucking with the actual wiki. i've bookmarked those 2 pages.
any special permissions or logins to edit the wiki?
well, i'm an advocate of just getting stuff out there so people can see it, rather than perfecting something that may not be useful
you have to enter "emc" in the administrator password box to be able to edit pages
oh, it will be a long way from perfect.....
if anyone else is watching... or, perhaps i should ask the board
is it really useful to have EMC1 stuff on the front of the wiki like that?
with "Versions" right at the top
kinda weird intro if you ask me
obviously we are shoving EMC1 into a dark corner, so why pretend to support it?
my stuff will be generic machine design components, just things i've found to be useful. no junk.
like sdp-si and msc?
or more specific, like brand recommendations?
perhaps it would be good to have a sort of "vote for this brand if you've used and are satisfied" kind of thing
then people will also know someone else who has used it and can ask them questiosn
let me get it going over the next couple of weeks and see how hard it is, and how it looks.
so i guess "parts and raw materials suppliers" doesnt quite fit your vision
or, does it?
right, well, you can edit any page so change whatever you feel needs changing
i will stick to the suppliers page. i really dont want to mess up anything else.
boy this wiki needs some work