argh, I just fixed most of those same things
but of course jepler's looks better anyway
except I'm not so sure about that dark background
cradek: you mean the background of the "description" area?
change it back if you feel strongly
here's my logic, such as it was: that text is descriptive and not editable, just like the labels at the top of the page
yeah, I can sure see that
how about the same color, instead of darker?
oh, they're not the same color for you?
no, not even close
can you tell me the rgb values you are seeing?
for me, it's all #d9d9d9
d9d9 d9d9 d9d9 in the text box
eaea e8e8 e3e3 elsewhere
that's not even grey
something stuck a bunch of crap in my X resources
03jepler * 10emc2/tcl/bin/pickconfig.tcl: don't hand-code the color of text widget.
now they're still the same color on my system; on yours, maybe they match now
lerman_ is now known as lerman
I see the text widget is -relief flat, but it's definitely sunken on my screen
unless I don't know what flat means
cradek: the -flat on the text widget itself is there so that the sunken relief (of the scrolledwindow) goes around the text *and* the scrollbar
cradek: Maybe it should be flat, though
right, I see
when I change it to flat, I get an otherworldly white line around it
so forget it, it's fine sunken
03cradek * 10emc2/tcl/bin/pickconfig.tcl: remove debug output
03jepler * 10emc2/tcl/bin/pickconfig.tcl: show OK button as default
so, whats new?
oh I dunno
a hundred things in emc
a dozen things in axis
skunkworks: any chance to try axis 1.1.1 again?
I keep wondering whats the best way to automate this sheet loader
its a PLC and motion sort of task I guess
I have not had a chance yet - within the next few days
03cradek * 10emc2/src/emc/motion/usrmotintf.cc: remove debug output
new emc2, emc2-dev ubuntu packages in the repository
[02:35:40] <cradek> http://timeguy.com/cradek/emc/ubuntu
new emc2, emc2-dev ubuntu packages in the repository
I'm tempted to move the TESTING tag, since I tested and made packages
I didn't get the little "update" ballon when I booted just now
what do you think?
I think it only polls once in a while
run the update manager and hit refresh
go ahead - I thought about doing it to catch the version stuff that was added yesterday
also my halcmd changes
stand by one
want to test that
its important to me that halcmd loadrt work outside the emc environment
to date it has relied on emc.run to set HAL_RTMOD_DIR
I'm hoping module_helper gets around that
and if not, maybe we can figure out how to do it
no I don't think it changes that
the module helper is only a replacement for insmod/rmmod; it has no idea where files are
the update ballon just popped up ;-)
but it does - the whitelist
the whitelist is not specific enough for that today, but SMOP
I hate to add complexity to it though; it's setuid, has been audited, and is fully functional
well somehow or another it just worked
the complexity for knowing where everything is and shuffling paths around belongs outside the setuid code
did a run-in-place build, scripts/realtime start, bin/halcmd loadrt threads
and it worked fine
or maybe thats the existing "if the env var isn't set, try to figure it out" code working
looks like it does some crazyass things with /proc/self if the env is not there
I'm guilty of those things
that must be what saved you
but I thought they were busted
certainly are busted for installed
you could now compile in the path to rtlib. you have all the information you need.
ok, we have a configure script and makefile that are aware of whether we're building RIP or install
yes, thats where I was going
but how? I sure don't want halcmd.c.in
just follow my example for EMC2_BIN_DIR
but before you do this, do we move TESTING?
-DEMC2_BIN_DIR in the makefile?
if this is a simple fix, I'd like to have it part of the new TESTING
also, seems like -D might be the way to get EMC2_VERSION into C code as well
or should we "just shoot the engineer and ship the product"
alex said he knows how to do that correctly (config.h) and we should wait for him
he said today he will do it when he gets home
go ahead and move TESTING
and set VERSION to "Testing 2006-02-14" or something like that
also: set another tag, TESTING-2006-02-14
that way when TESTING moves, we'll still be able to checkout this particular one if needed
heh, we need a release checklist
I dunno, these tags are going to accumulate fast if we do that
but tags are free, as far as I can tell
is there anyplace were a proliferation of them will get messy?
ok, forget what I said about a special tag
after all, the version contains the date
you're gonna change VERSION, then tag, then change it back, right?
so the tag actually applies to a window a few minutes wide, not a date
if VERSION is the only thing that changes, who cares... but suppose there are real commits either before or after the tag, but on the same day
but you can sure look at the date of that VERSION revision to get the tree
OK, I'll shut up now
I don't understand what you just said
(more of me trying to figure out how it can break)
".. on the same day"
9pm, somebody commits something important (not a change to version)
10pm, version is changed, tag is applied, version is changed back
11pm, work continues on head, with another commit
2 months later, I want to check out that TESTING to help someone
I need date and time if I want to get the right thing
you'd get the 10pm version
sure, you have that
it's in VERSION's log
that;s why I shut up ;-)
but you must admit it can be error prone if you're not aware of the details
everything is like that!
a tag would make it bullet proof
at the expense of having lots of them build up over time
it depends how much we expect to need/use these versions
lesser of two evils I guess - I suspect that 99% of the time if somebody says "I'm running Testing-<2monthsago> and I need help" our first response will be "get the current one", so the case I was worried about will be rare
* jmkasunich shuts up
ok I'll get on with it, let's see if I can get this right
I'm tempted to put a checklist on a wiki page
I'll make notes
make notes on the wiki ;-)
03cradek * 10emc2/VERSION: moving TESTING
03cradek * 10emc2/VERSION: Done tagging
maybe next time I'll be able to do it without reading the info pages
anything beyond checkout or commit I read the book
too many options and too many ways to get confused
maybe I need a book
but info is like a book with hyperlinks
I wasn't referring to a paper book, don't have the CVS one
but there is http://cvsbook.red-bean.com/
just remembered something I should have done
updated the history file
np, just do it and move the tag on that file up
I'll just do the update and leave the tag alone ;-)
I'll move it for you
NEWS is the file, right?
yes, but I don't see it
I've never touched it
I seem to recall alex saying something about info from that winding up in the debs somewhere?
yes that kind of thing goes in /usr/share/doc/emc2
also there's a (debian-mandated) changelog with a particular format, so we're duplicating effort somewhat.
maybe thats what alex was talking about
where is that located?
or you mean same dir
both derived from CVS: /docs/NEWS tho?
no, both written separately
or does the info have to be entered somewhere else
the debian one is debian/changelog, and I write an entry every time I make a new deb
seems kinda dumb
we should only have one
or do you prefer it this way, so you can decide what gets noted in the deb log?
it's tough because major changes (that go in NEWS) don't exactly coincide with deb-making (which requires an entry in debian/changelog)
when you see the new emc2 deb in the update manager, the latest debian/changelog entry shows to tell you what changes you're going to get
hmm, I just clicked on the update manager
emc2 is new (as expected)
but both "changes" and "description" under Details are blank
is there an update to update-manager available?
I think I got one yesterday
emc2 is the only update today
it shows ver alpha21?
dpkg -l update-manager
do you have 0.37.99.cjr1?
* Jymmm waves
it worked for me...
cjr? is this from ubuntu or you?
whats all this talk about ubuntu?
why are you releasing versions of ubuntu's install-manager?
jmkasunich: to get it to read changelogs for packages that aren't theirs
jmkasunich: they have it hardcoded to contact only THEIR webserver for changelogs
I think that feature is important for emc releases, so I fixed it
jmkasunich: I'm building debian packages that make installing/running emc2 on ubuntu easy
gonna submit a patch to them for inclusion?
or are they likely to consider the hardcoded server a feature
cradek whu ubuntu instead of sarge?
w h y
I don't know, I'm pretty disappointed in how they had it hardcoded
Jymmm: easier to install, more polished
Jymmm: excellent i18n support
Ok, fancy wrapper, but the contents are a grab bag
Jymmm: it's way too late to debate the choice for me, feel free to build sarge packages if you like
cradek It's nto really a debate as much as understanding the reasoning behind it.
Jymmm: it's a choice I made, nothing more
and these pkgs will work under Sarge too?
I'd be rather nervous putting the kernel packages into a sarge box
untested as the version, or untested as in nfc
Jymmm: I have never run sarge
jmkasunich why's that?
as ubuntu is based on sarge.
cause kernels are scary ;-)
well, sarge unstable
jmkasunich: I suspect it would work ok...
thats good, but I still ain't gonna try it
my goal is a turnkey setup, which means I can concentrate on one OS release.
If others can use the packages, great, but I don't need the added complexity of many-platform testing.
remember I said the details/changes and details/description didn't work
gotta click on the package to make them appear
* Jymmm hands jmkasunich a P
I would probably have realized that if there was more than one
I agree it's NOT obvious
ok, the CVS tag is Testing-2006-02-14
the deb is 2.0.0-alpha21
and the deb installed Help/About says Prerelease HEAD
should we attempt to coordinate those?
seems to me the release process should be:
unfortunately I think 2.0.0 is NOT "newer" than 2.0.0-alpha21
so we've already got a mess regarding the debs
that will at least make the displayed (About) version match
from looking around, it seems to me that there's no jood linux software to generate NC code from a model of any sort. is that true?
the actual deb version needs to comply with debian rules, right?
I think debian is not strict about the exact format of the version string
seems to be cool with both .'s ans -'s
orpheus: I'm not familiar with that area (more of a coder than a user), but I think you are right, the pickings are kind of slim
just not _
I'm not sure exactly what algorithm they use for "newer than"
that's something I could probably find out.
* orpheus goes to try writing a simple implementation of such
worst we'd have to do is have the testers dpkg --purge emc2; apt-get install emc2
cradek: you can always use apt-preferences if you need to force some version rule
yeah our early adopters will be able to handle whatever small thing is necessary if we change numbering schemes.
jmkasunich: I hadn't thought hard about synchronizing these things, but you're right they should be.
how useable is that dow days?
orpheus: yes, there are debs for ubuntu
orpheus: it's still in testing, as there is not an offical emc2 release, but it works great.
any clue if things explode with debian? and weird dependancies that come from ubuntu
[03:58:26] <cradek> http://timeguy.com/cradek/emc/ubuntu
orpheus: as far as I know, nobody has tested yet on debian.
to be clear: nobody has tested the debs on debian
oh, well, you may have youself a first
the source compiles and runs fine on BDI-4.several, which is debian
right, I'm talking in particular about the debs
the source works on plenty of platforms
I wouldn't say plenty
orpheus: what kernel version is your debian system?
BDI-Live and 4.xx and ubuntu
jmkasunich: any modern linux distribution.
assuming it has a RT patched kernel and the right dependencies
which is a non-trivial assumption
ah, well, i'm not quite that far yet
orpheus: the realtime kernel/rtai debs are based on 2.6.12 which is the version in ubuntu 5.10
orpheus: oh you're not installed yet? then install ubuntu!
i was going to use this laptop i had lying around
but it's only a 300 mhz
that'll work ok, I installed on a 266 last week
i've been hearing that's too slow
oh it's slow allright, but not "too slow"
the default - full gnome
it's not bad if you get rid of gnome-terminal
no, i've heard that emc's realtime stuff will say "no"
that's not true, it works fine. lots of people run it on machines that speed.
even AXIS runs acceptably on my laptop.
you aren't gonna have screaming rapids with a stepper machine, but it will work
(there has been some recent talk about the RT having problems, that was a bug and its fixed now)
i already have debian installed, and i'm not reinstalling (darned maching won't boot from cd, and my floppy drive died)
jmkasunich define 'not screaming'
orpheus: sounds like a bad machine to experiment with then...
200MHz CPU can "probalbly" run steppers at the default 50uS BASE_PERIOD
oh, always blame the machine!
well, i'm also just now getting my mill set up too
which means a max of 10K steps/sec
jmkasunich what would that be about in IPM?
probalby best to stay under 5K steps/sec
orpheus: if you're going to run a real mill, you do not want to use a laptop. They're notorious troublemakers.
Jymm: silly question, that depends on the machine
full stepping with a 1" pitch screw? or 10x microstepping, 2:1 gear reduction, and 20TPI?
it makes a huge difference
orpheus: mine causes the realtime to stutter whenever the CPU fan turns on or off
steppers aren't here yet, then i have to mount them... point is, i was just going to see what the software looked like
ah, for that, it's fine
.5" 5tpi 8micro
Jymmm: can you do the math yourself please?
* Jymmm needs to write a calculator sometime
Jymm: .5" pitch is 2 tpi
well, it was my hope to use that mahine, but oh well
orpheus: some people report success with some laptops, some (me) report insurmountable problems
orpheus: all depends on luck (your laptop)
.5" diameter, 5tpi
ok, diameter doesn't matter
cradek I could if I knew.
it's just that i don't have tons of space
so a laptop would be good
200 steps/rev * 8 microsteps = 1600 steps/rev
times 5 tpi = 8000 steps/inch
so at 5000 steps/sec, you'd get 5000/8000 = 0.625 inches per second
here's some good questions:
times 60 = 37.5 ipm
jmkasunich Seriously, you think a lil js calculator would be nice?
Jymm: there will be a quiz next week, so remember that math
if i'm going to run 3 steppers from a parallel port, all i really need is a parallel cable to gut, and 3 stepper drivers, right?
Jymm: no need, its just basic math
orpheus: more or less - the parallel port lines usually go directly to the stepper drivers
orpheus: what size motors?
jmkasunich basic yes, but lots of items =)
sometimes you have an intermediate board of some kind that allows you to hook up things like estop or limit switches too.
Xylotex has a nice little 3 channel microstepping board, that already has the DB-25 on it
no cable gutting needed
about $130 I think
it's cheap too
whic brings me to another question, where can i get drivers for not-too-much
but only handles 24V I think
goes up to 2.5A, 24V
as in, up to 24V?
or up to 2.5A at 3 V?
orpheus: I think the cheapest solution would be the xylotex
up to 24V
you'll want to run your motors on 24V
NINE POUND HAMBURGER!!!
max supply voltage is actually 35V, recommended max is 30, most folks find 24V to be most convenient
preferably 30-40, but you'd need better drivers
I have a xylotex, I like it
but I don't run a machine with it, just some steppers on a testbench
dude, they're 3V steppers, isn't 30-40V a bad idea?
no, the driver limits the current to a safe value
no - the 3V spec is the zero speed (DC) voltage rating
the drives are choppers, and step down the supply voltage to limit current
higher voltage is needed to get good torque at high speed
general recommendation for top stepper performance with choppers is about 10x to 25x the motor voltage
so basically, it's 24V spikes that average to less than 3v
yes, at low speed
as the motor speeds up, the voltage rises, but the current remains fixed
its current that makes torque (and also makes the motor hot)
orpheus: motors are huge inductors, you need the high voltage to lower the time constant
so you don't want too much current
you run steppers about 20 to 25 times it's rated voltage
what kind of machine do you want to run with these motors?
it's a harbor freight, one of those crazy micromills
micro or mini?
30 lbs or 150 lbs ;-)
UPS said "no" when it was packed in a single box
the only person I know with one of those (Jon Elson) used NEMA34 steppers
* orpheus crosses fingers
I dunno if it _needs_ NEMA34
the 23s might be a little underpowered... might be ok if you keep your speed down (and depending on the leadscrews)
ballscrews or acme? (probably acme)
do you know the pitch?
any other differences?
those motors will probably be ok with acme 20 pitch
but it's not going to jog real fast.
'cause if there's a leadscrew that doesn't involve me putting push berings into my motor mounts, it'd be cool
i have time
nnot push bearings, what do they call them...
the machine was built for manual use, right?
so the screws should already be able to handle the thrust
are they 1/4"-20?
8mm i think
that's a funny acme, usually 1/4-16
funny to have metric diameter and inch pitch
well, whatever the inside of a 608 bearing is...
I don't know much about leadscrews :-/
I would expect the end of the screw that goes in the bearing is turned down to fit
the threaded part shouldn't be in the bearing
might be a 1/2 inch screw, turned to 8mm at the bearing seat, or something like that
well, dang.... my z looks different than the y
I would hope a 150lb class mill (minimill as opposed to micromill) would have bigger than 1/4-20 screws
gimmie 2 mins with a caliper...
jmkasunich no way a screw would be 1/4" on a minimill, would it? I mena I know they're cheap, but not THAT cheap.
1/4" is more like micromill/Sherline size I think
jmkasunich sherline I would expect.
I'm sure sherline uses 1/4" acme
I'm not even sure its acme
I know it's 20 tpi
wait, I'm thinking soemthign else
course I haven't seen that many, and my eyes don't focus on something as fine as 20tpi ;-)
jmkasunich: I think my metric sherline is acme
jmkasunich: it's 1mm pitch
yeah but easy to count
unlike 16tpi leadscrews :-P
my shoptask has 10tpi, 5/8 dia acme, my Van Norman has 5tpi about 1" acme
the shoptask is the one I intend to CNC
16tpi is easy to count too (if you hold a rule with 1/16 grads up to it)
but with a drawing in thousandths, seems like it'd be hell
z is 3/8ish, y is 7/16ish
what the heck
different sizes.. go figure
diameter isn't all that important
also, that's from the outside of the threads
pitch is what counts
that matters more
I need to sleep too
still recovering from 3am this weekend
ah, should i leave you alone till after my steppers get here?
you can always show up here and ask questions
they're in the mail
somebody will answer
quick question tho
xylotech says $145, is there any other place i should be looking?
xylotex is the only one I have personal experience with
I'm sure there are others, nothing springs to mind right now
orpheus: Just be careful when wiring it up
you're not looking to sell your or anything, right?
yeah, 24v hurts
geckos are a step up, about $120 per axis (instead of $150 for three, or $50 per axis)
no, not selling ;-)
24V doesn't hurt people
but shorted motor windings hurts drives
jmkasunich heh heh heh
it can at 9 amps...
that too tho
including shorts to ground
I burned mine out
* Jymmm hands jmkasunich a wire... "touch it"
words from the wise
Jeff (the guy at xylotex) repaired it for $27 bucks
24V doesn't kill people, people kill people
you can get knocked on your ass from a D battery
* jmkasunich hands Jymm a piece of 500MCM connected to a 480V 800A feeder
will most any ltp-port-based driver work with emc/
ebay has some variety
jmkasunich hey now, mine was only connected to 24v
hard to answer that
"almost any" is a pretty wide range
google "emc + device name here"
Jymm: I work with 480-690V at 600+ amps all the time
orpheus have you got your motors already?
I don't worry about 24C
in the mail
24V that is
24C isn't that hot either
orpheus Xylotex is a bipolar driver
I dunno where you guys are, but its 11:38 here and I gotta work in the morning
i got that 8-wire means wire it any way you neeed to
thanks for answering newbie questions
orpheus pretty much, I just ordered the motors from Jeff too.
ebay was just too cheap for that part
but he includes a power supply...
oh well, too late now
he doesn't INCLUDE a PS/ but you can get one from him.
@$VDC switching PS
24VDC switching PS
there's n list of "known" working hardware, is there/
you dont want a switching PS supplying motors
fenn be fair now
linear would be best, but switching will work.
a non-regulated supply would be best
transformer, bridge, capacitor
i don't even know what mine is...
it's just 'big' , and 'hot' and times
does it have a transformer
a big one
otherwise it wouldnt be big, just hot
it's some 15-year-old 15-amp 25-volt variable supply. 2 knobs, 2 meters, one switch.
now i just have to borrow it back
sounds like it'll work
oh, what's the difference between unipolar and bipolar anyways?
unipolar gets pushed on one end, bipolar gets pushed and pulled
they're both basically 2-phase motors, no?
bipolar has two wires coming out for each coil.. unipolar has one wire for each and a common
point beingthat bipolar is better, as then my motors are 282oz/in , not 200
and microstepping is the ability to make a less-than-full step
are there any devices emc will work with that are done with a serial interface
looks like no, at least for these drivers
dang, they were cheap too
most serial interfaces are too slow
er, i guess
dunno why actually..
most serial motion controllers are fed Gcode, and then create the motion in hardware
anyway, no, there's no serial port interface in HAL
$75 would have been good for 3
3 single axis
[05:14:20] <orpheus> http://cgi.ebay.com/LOT-3-API-DM-224I-Microstepping-Motor-Control_W0QQitemZ7589998370QQcategoryZ71394QQrdZ1QQcmdZViewItem
see if that goes
expect those to go up quite a bit more
but.. what are they?
aside from a good reason to have 3 comm ports
those arent even stepper drivers
er, i think
no wonder that didn't make sense
excuse me while i fail to read correcty
or type correctly
i dunno what they are smoking
its a stepper drive that takes encoder inputs
you probably feed them gcode or similar and they do all the acceleration and such
not really what you want for cnc
but it's per-axis...
right that's the problem
point is the xyloctex is looking better
no, ebay search, i don't want a solution to my bipolar personality...
bill20r3 is getting these http://cgi.ebay.com/CENTENT-CN0121-REM-CN0121_W0QQitemZ7577664420QQcategoryZ97184QQssPageNameZWDVWQQrdZ1QQcmdZViewItem
wait that doesn't look right
take a gander at these.. http://tinyurl.com/9okjf
that's a monster heatsink
but.. 3 amps.. matches the motors, but i'd need to find a voltage rating
[05:26:27] <fenn> http://cgi.ebay.com/Bi-polar-Stepper-Motor-Control-Board-10A-CNC_W0QQitemZ7588545113QQcategoryZ58294QQrdZ1QQcmdZViewItem
noheatsink up to 6A
wow it's $17 on their website
ah its backordered though
but... what's the catch?
* fenn has been waiting for a month for his stupid electronics to arrive
from a different company
i'm not totally sure it regulates the power going to the motors.. is that possible?
like, it may let an arbitrary amperage through, whatever you give it
do you see any catches?
wish i could zoom in more to read the labels on the chips
well i can make a pretty good guess - neat how you can see the traces on both sides of the board
they *might* respond to emails
it just doesn't seem to have enough on it...
i think the brown-green-brown-(brown?) resistors are the current sense resistors
at the top and bottom
the little 6-pin chips are the opto's for step/dir
that means something...
i'm CS not EE
what's an opto do/
little LED/photodiode pair that protects against static discharge and the like
there's no electrical connection
ah a gold band is 0.1 multiplier..
blah i give up
thanks for trying
think i may try for them, since i don't really need microstepping
i'm a little interested in those drivers myself
i'll let you know how long the backorder is/ if they work when i get mine
i'll probably be in here then, so be sure tot ask
hmm the other controller doesn't have a current control
the other one they sell
bah i should just make my own
* fenn sacrifices 20 virgin goats to the UPS god
come on packages!
orpheus and why dont you need microstepping?
well, one step is... hurmm
microstepping isn't used for precision movement as much as overcoming stepper motor issues.
microstepping is for wuuses
like not skipping steps
fenn go take your geritol or prunes or sotheing
orpheus smoother motion for one
they don't list that t
the driver does microstepping\
so i was assuming not
[05:59:54] <orpheus> http://www.electronicutopia.com/contents/en-us/p289.html?lmd=38747.972917
if you wanna see
ah, well, i gtg
my, it is late
* fenn dreams of field-oriented controlled stepper motors..
hope they're microstepping
the one's i may order
New game started, http://sre.servegame.com
: This is a webgame based on SOLAR REALMS ELITE, a famous BBS game back in 1992 (Earth2025/BRE fans are welcome)
cradek: looks like your message to the bdi list went through
hope I got everything right
cradek - I am trying to read info on the ubuntu page. I can't with ie 6
are they trying to keep windows users away? ;)
do you mean the timeguy site?
no - you have a link to ubuntu from your site
never mind - I need to scroll down ;)
they have a large white space before the text - so you have to scroll down a page to see anything
cradek: huh, your (timeguy.com) front page is very wide on IE
cradek: "internal" pages seem to be OK
jepler: strange - it used to work...
skunkworks: that's IE's bogus rendering of CSS
skunkworks: I recommend an upgrade to firefox
sounds good. I have it installed on one of my computers here at work. going to have to play with it again.
why do you like ubuntu?
Does your emc2-install.sh - make it a rt kernel? Or am I confused?
skunkworks: emc2-install.sh adds a new repository for debian packages, and the installs the needed ones
the sh installs a few packages, yes one of them is a rt kernel package
Cool - I might have to try this.
nice work - looks like it would be much easier than the bdi + emc2
well it doesn't require compiling, and you get the advantage of a community-supported platform to run on
Is axis also installed?
so in my mind, there are a lot of advantages
yes, axis is installed
hi chris, skunk
the ubuntu update manager is nice - when it finds updates, it prompts you, shows the changes you can expect, and lets you install them by clicking ok
sounds nice - they have a 64 bit version for the amd64 proccessors. Do you thinks this would work rt with emc2? Speed improvement?
skunkworks: I don't think anybody's yet used emc2 on a machine with a 64-bit kernel.
jepler: is my front page better on IE now?
skunkworks: I've read a few messages on the rtai mailing list that indicate a 64-bit realtime kernel is under development, but I'm not sure where the kernel patches are.
cradek: I'll check.
no biggie - don't have a amd64 processor anyway ;)
I do (but it's not for emc development)
cradek - does it get emc2 from sourceforge - or do you make a package every so often?
I make packages every so often
I plan to make a new one every time we move the TESTING tag in cvs
lately there has been a new package every day or so - development has been thundering along at a pretty good pace
I hope we can keep it up right 'til the release
neet - so I would just redownload the emc2-istall.sh and then follow your install directions
no, you get a prompt on the screen that says an update is available
then you click OK
just like the OS's updates
I think I am going to have to try it.
downloading the iso as we where talking
you can also get ubuntu with bittorrent, sometimes that is faster
it is going as fast as I would expect from our connection here at work. around 50kbs
ooh, that's not very fast
still you'll have it today I guess
no - It would be faster at home but I don't have a way to burn iso's there yet
cheap burning software
skunkworks: you can also call ubuntu and have them send you free cd's
I see that
hi samco ;)
websys - are you supposed to get 7+ inches of snow also?
03cradek * 10emc2/src/ (configure configure.in): get the version number into config.h, thanks alex
alex_joni: Do you know what was broken to cause emc2 to run slow. (in laymans terms) ;) I see John Kasunich mentioned in the list that there was an issue.
having to raise the base_period
skunkworks: it's fixed now
and I know too, but I'll let chris have the pleasure to explain :D
I forgot to mention that John Kasunich said it was also fixed
there is an RTAI call that tells the time in nanoseconds
that was all over the realtime code, in order to tell how long things were taking
however that RTAI call is so slow it was taking much longer than the code it was meant to measure
so jmk replaced it with a scheme that measures in CPU clock cycles, not nanoseconds
this is a very quick measurement, but obviously the measurements are not as simple to interpret
that makes sense - thanks
skunkworks: was that layman anough?
enough even.. can't type ;)
there is an awfull complexity if you start to look at these problems in detail
is there then something that then tests to see how many clock cycles there is in a certain amount of time? at startup maybe - to see what each clock cycle represents?
no, but you can get that information from /proc/cpuinfo
so now you just have to do a little math to get from cycles to seconds
there I go again - making things more complicatecx
skunkworks: but the problems gets worse, because the CPU freq might change on some processors
some turn the freq down on idle time, some stop it completely, etc
right - power saving and such - but I would think the rt kernal would disable that
it's a big mess that needs sorting out
some is HW in the CPU you can't control
but it easily grows in complexity
you start with a simple task (measuring time), and end up with dependency and versioning hell
because you need to use a RTAI function that's consistent from 24.1.x to 3.0,3.1,3.2 and now 3.3 RTAI
I can imagine actually. Been in similar situations
again - you guys are doing great work.
skunkworks: don't mention it ;)
or better yet, mention it :P
we also appreciate people using it and reporting their results, it helps a lot
On a side note - I think I finally got my car straitend out. Running 16psi to redline so far. no knock.
ended up being a tuning issue
I needed to tune to what ran right not what the numbers where telling me.
should be able to run 20-24psi with higher octane fuel or propane/alky injection.
oh - and bigger injectors.
03jepler * 10emc2/ (4 files in 3 dirs): generate rtapi.conf from configure, not from a makefile
03jepler * 10emc2/bin/.cvsignore: new generated file
03jepler * 10emc2/src/module_helper/ (Submakefile module_helper.c): make emc_module_helper a bit more restrictive by taking the module extension from Makefile.inc instead of having two extensions
les_w: toc toc!
hi giacus and k4ts
bleh...I could not find the formula for zero axial stress at the glue line for a two component composite beam.
So I had to derive it
the blackboard is full
hi les_w :)
just to prove the beam element thicknesss need to be the 1/square root of the young's modulus
I know I had that in scholl...
don't worry, I get paid even to relearn what I forgot!
could be nice 1 gb ram bank in the head ..
My memory is very short :(
funny - I just spent an hour yesterday calculating a markup for pricing that we thought would have to be a formula. we had 2 equations - substiuted and simplified. It simplified to a constant ;)
for the first time I feel I am forgetting some school stuff
we where like "that can't be right" - but it was
I spent two or three hours on this
I smoke.. :( I think smoking damage the memory
thats a very bad thing
I want a glue line on a component that is two disks of differing materials.
what thickness ratio results in no tensile or compressive stress at the glue line?
that was the problem
thought it would pop right up in google
but it didn't
already tried wikipedia ?
well, it's solved now...on the blackboard.
oh, ok. Company is bugging me to travel and I still have not found a car.
some have problems ...
I wanted to ask you something.. what's a normal feedrate on your emc machine?
what can it do?
oh about 600 ipm
say without actual carving/machining
much more if I changed out the ballscrews
positioning speed, not rapids
It is a combination wood/metal cutting machine
so 600 is enough
les_w: right, was just wondering as a reference
spindle hp limits cutting speed
I know of an emc2 user that is doing F9000 (mm) and not complaining
I usually don't run that high though because emc is pretty rough due to the tp
before the #ifdef in tp.c I could not run more than 120 or so without possible damage
well, maybe someday there will be a new TP
Hi - I installed
how'd it go?
I installed Ubuntu
smooth - but
emc2 installed but when I run it I get the spash screen and that is it
the splash screen goes away and nothing
start a terminal and run emc at the prompt
where is it now - location
just type emc
so you get the config picker? what do you pick?
insmod error inserting ....
yes I get the picker
which config are you picking? sim? sim-AXIS?
let me log into here from there - then I can copy and paste
did you reboot?
I mean, after the install?
ok let's have a look at the exact error message
skunkworks: can you open a terminal and run : 'uname -a'
wow that was easy
insmod: error inserting '/lib/modules/2.6.12-magma/kernel/adeos/adeos.ko': -1 File exists
insmod: error inserting '/usr/realtime-2.6.12-magma/modules/rtai_hal.ko': -1 File exists
sam__: seems like there is an emc already running, or not cleanly shutdown
get about 10 lines similar to that
easy or hard fix?
let me reboot again and see
ok, that's the easy fix :D
I had run emc2 2 or 3 times with no results so it may have been runing somewhere in the back ground
yes, could be, let's try it at the shell first so you can see any error message
skunkworks: you probably chose sim, and didn't wait for mini to come up
RTAPI: ERROR: version mismatch 0 vs 529
ugh, that's a new one
do you want more or does that tell you? looks like rt wasn't installed correctly?
alex_joni: any idea?
sam__: can you get the exact first few lines?
RTAPI: ERROR: version mismatch 0 vs 529
HAL: ERROR: rtapi init failed
IOCONTROL: ERROR: hal_init() failed
can't initialize the HAL
RTAPI: ERROR: version mismatch 0 vs 529
something I did again?
it's an rtapi internal problem
the only thing I could say that I did was there was one update that I installed between when emc2 installed and I rebooted
SHMEM is not init'ed corectly
or it's inited somehow unexpected
I'm trying to figure out what's happening right now
cradek: the rtapi code uses some static revision code inside rtapi_common.h
but this works on several other machines!
the only way I'm seeing right now for this to happen is if you mixed a newer rtai_rtapi.c with an older rtapi_common.h
ok, if it does work on other machines, I'll keep looking :)
sam__: dpkg -l "emc2*"
sam__: what version number do you have?
ok, the second value 529 looks ok, the 0 looks like shm isn't working
or not properly..how about udev & co?
|/ Err?=(none)/Hold/Reinst-required/X=both-problems (Status,Err: uppercase=bad)
||/ Name Version Description
ii emc2 2.0.0-alpha21 PC based motion controller for real-time Lin
ii emc2-axis 1.2a2-0.2 AXIS front-end for emc2
does that make sense?
sam__: can you open a console and type what I'm asking you?
sam__: as in be our eyes and fingers ;)
sure there right now - shoot away
lsmod | grep rtapi
ok, same as me
cradek: you mean the version
rtapi 25664 2 motmod,hal_lib
rtai_sem 14976 1 rtapi
rtai_shm 8192 1 rtapi
rtai_fifos 23500 1 rtapi
rtai_up 69400 4 rtapi,rtai_sem,rtai_shm,rtai_fifos
rtai_hal 20888 5 rtapi,rtai_sem,rtai_shm,rtai_fifos,rtai_up
sam__: cd /proc/rtapi/
dr-xr-xr-x 2 root root 0 2006-02-15 16:18 .
dr-xr-xr-x 95 root root 0 2006-02-15 16:05 ..
-rw-rw-rw- 1 root root 0 2006-02-15 16:19 debug
-r--r--r-- 1 root root 0 2006-02-15 16:19 fifos
-r--r--r-- 1 root root 0 2006-02-15 16:19 modules
-r--r--r-- 1 root root 0 2006-02-15 16:19 sems
-r--r--r-- 1 root root 0 2006-02-15 16:19 shmem
-r--r--r-- 1 root root 0 2006-02-15 16:19 status
-r--r--r-- 1 root root 0 2006-02-15 16:19 tasks
sam__: cat shmem
**** RTAPI SHARED MEMORY ****
ID Users Key Size
01 2/0 1212238881 65500
02 1/0 111 172520
echo "5" > ./debug
I didn't ;)
sam__: no worries
Feb 15 16:07:12 ubuntu kernel: [ 136.079551] HAL: creating thread base-thread, 50000 nsec
Feb 15 16:07:12 ubuntu kernel: [ 136.079615] HAL: thread created
Feb 15 16:07:12 ubuntu kernel: [ 136.079620] HAL: creating thread servo-thread, 1000000 nsec
Feb 15 16:07:12 ubuntu kernel: [ 136.079641] HAL: thread created
Feb 15 16:07:12 ubuntu kernel: [ 136.079646] HAL: creating thread traj-thread, 10000000 nsec
Feb 15 16:07:12 ubuntu kernel: [ 136.079656] HAL: thread created
Feb 15 16:07:12 ubuntu kernel: [ 136.079698] MOTION: setting Traj cycle time to 10000000 nsecs
Feb 15 16:07:12 ubuntu kernel: [ 136.079705] MOTION: setting Servo cycle time to 1000000 nsecs
Feb 15 16:07:12 ubuntu kernel: [ 136.079709] MOTION: init_threads() complete
Feb 15 16:07:12 ubuntu kernel: [ 136.079713] MOTION: init_module() complete
ok, RT part is working as it should
but the user-part is somehow borked
ls -al /dev/shm*
drwxrwxrwt 2 root root 40 2006-02-15 16:05 .
drwxr-xr-x 12 root root 0 2006-02-15 16:06 ..
yes, what chris said
there's a symlink iirc
I've never seen that before
but never mind..
a udev thing
sam__: still there?
ls -al /dev/RTAI_SHM
crw-rw-rw- 1 root root 10, 254 2006-02-15 16:07 /dev/RTAI_SHM
cradek: so far I'm getting this:
RT part of shm & co is working, the user part does access SHM , can alloc space for new data, but the allocation is somehow borked
why would this be different on just sam's machine?
we know it's the same binaries
I can only imagine some bad checking
the initializing code is looking for a magic number
if it's there, it assumes it's alreadyinitialized, and does nothing more
but I can't imagine that he's getting that value repeatedly (even after reboot) by chance
sam__: any way you could set up a ssh on that box?
is it connected directly to the internet?
though a router
so not directly..
I could make this as the dmz
yes,for me to ssh in and look around
it might be helpfull, I have not many ideas, to ask of you, left...
If you tell me how to setup the ssh - I could set this internal ip as the DMZ
is it a web-based router? you could set up port forwarding for ssh, it usually is pretty simple
ssh should already be there
that to - what ever you want
whatever you think is simpler
can you see my /msg message?
to set up ssh is simply 'sudo apt-get install ssh'
I think it's already there,I didn't have to do that
sam__: is there anything unusual about your machine? is it extremely old or extremely new?
he must be working on his router
actually on ssh
run from the top menu: system/administration/services
turn on "remote shell server"
it's working, router is still icky
hi! you must be a EE?
cradek: ssh is ok now
Just tried to build TESTING, run fails
I should add, this is Gene H.
Well, let me see if a copy paste works brb
gene@shop:~/emc2$ scripts/emc &
gene@shop:~/emc2$ EMC2 - TESTING-2006-02-14
Machine configuration directory is '/home/gene/emc2/configs/stepper/'
Machine configuration file is 'stepper_inch.ini'
insmod: error inserting '/lib/modules/184.108.40.206-magma/kernel/adeos/adeos.ko': -1 Operation not permitted
insmod: error inserting '/lib/modules/220.127.116.11-magma/rtai/rtai_hal.ko': -1 Operation not permitted
insmod: error inserting '/lib/modules/18.104.22.168-magma/rtai/rtai_up.ko': -1 Operation not permitted
insmod: error inserting '/lib/modules/22.214.171.124-magma/rtai/rtai_fifos.ko': -1 Operation not permitted
insmod: error inserting '/lib/modules/126.96.36.199-magma/rtai/rtai_shm.ko': -1 Operation not permitted
ok, you can stop
insmod: error inserting '/lib/modules/188.8.131.52-magma/rtai/rtai_sem.ko': -1 Operation not permitted
insmod: error inserting '/lib/modules/184.108.40.206-magma/rtai/rtai_math.ko': -1 Operation not permitted
insmod: error inserting '/home/gene/emc2/rtlib/rtapi.ko': -1 Operation not permitted
insmod: error inserting '/home/gene/emc2/rtlib/hal_lib.ko': -1 Operation not permitted
ERROR: Could not load 'rtapi'
ERROR: Could not load 'hal_lib'
hi gene, I just answered your list email
Realtime system did not load
Shutting down and cleaning up EMC...
+ Exit 255 scripts/emc
I think you skipped the "sudo make setuid" setup
sorry - the machine is an ecs motherboard p4vmm2 -I don't remember the proccesor - is there a way to tell in here
Ahh, my mailer takes about 3 minutes inhouse.
Sparky_: you didn't configure --enable-run-in-place
sam__: I can get that info now
processor : 0
vendor_id : GenuineIntel
cpu family : 15
model : 2
model name : Intel(R) Pentium(R) 4 CPU 2.40GHz
stepping : 9
cpu MHz : 1800.365
cache size : 512 KB
alex - that is a little scary
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 2
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe cid xtpr
bogomips : 3555.32
scarier than moving your mouse around?
I think sow
No, I did a sudo make setuid too. I even copied all that to a script and ran it the first time, adding the cvs up -rTESTING to the top of the script
Sparky_: you didn't ./configure --enable-run-in-place
sam__: still working?
Yes I have, 3 times now
Sparky_: paste src/Makefile.inc
did I just lock up?
Ok, but first, heres the script
gene@shop:~/emc2$ cat switch2TEST
cvs up -rTESTING
sudo make setuid
skunkworks: yup, sorry
I just ran a RTAI testsuite test
and it crashed
Humm, sorry, /home/gene/emc2/src has no such file
do you want me to reboot?
Sparky_: that is NOT possible
skunkworks: please do
Sparky_: ls -l /home/gene/emc2/bin/emc_module_helper
-rwxr-xr-x 1 gene gene 32127 2006-02-15 17:17 /home/gene/emc2/bin/emc_module_helper
Sparky_: ls -l /home/gene/emc2/src/Makefile*
Whats not possible Alex?
that Makefile.inc is not there
go to /home/gene/emc2/src again, do the sudo make setuid
those permissions are wrong
alex_joni - back up
do you still want to look around?
Thats crazy Chris, its there and I can't 'see' them, but
gene@shop:~/emc2/src$ ls -l /home/gene/emc2/src/Makefile*
-rw-r--r-- 1 gene gene 16283 2006-02-15 16:54 /home/gene/emc2/src/Makefile
skunkworks: I'm going for that test again
-rw-r--r-- 1 gene gene 3947 2006-02-15 17:15 /home/gene/emc2/src/Makefile.inc
-rw-r--r-- 1 gene gene 3284 2006-02-14 22:13 /home/gene/emc2/src/Makefile.inc.in
skunkworks: if it crashes again, please reboot :/
ok, so Makefile.inc is there
but as chris said the permissions are not right
issue the 'sudo make setuid' manually
Damned if I know guys, heres an ls of that dir I'm sitting in the pwd output
blocks.ko encoder.mod.o hal_m5i20.mod.o hal_tiro.o scope_rt.o
blocks.mod.c encoder.o hal_m5i20.o libnml siggen.ko
blocks.mod.o encoder_ratio.ko hal_motenc.ko Makefile siggen.mod.c
blocks.o encoder_ratio.mod.c hal_motenc.mod.c Makefile.inc siggen.mod.o
classicladder_rt.ko encoder_ratio.mod.o hal_motenc.mod.o Makefile.inc.in siggen.o
classicladder_rt.mod.c encoder_ratio.o hal_motenc.o Make.rules stepgen.ko
classicladder_rt.mod.o freqgen.ko hal_parport.ko module_helper stepgen.mod.c
classicladder_rt.o freqgen.mod.c hal_parport.mod.c motmod.ko stepgen.mod.o
CodingStyle freqgen.mod.o hal_parport.mod.o motmod.mod.c stepgen.o
config.h freqgen.o hal_parport.o motmod.mod.o Submakefile.skel
config.h.in hal hal_ppmc.ko motmod.o supply.ko
config.log hal_ax5214h.ko hal_ppmc.mod.c objects supply.mod.c
config.status hal_ax5214h.mod.c hal_ppmc.mod.o pid.ko supply.mod.o
configure hal_ax5214h.mod.o hal_ppmc.o pid.mod.c supply.o
configure.in hal_ax5214h.o hal_skeleton.ko pid.mod.o tests
CVS hal_evoreg.ko hal_skeleton.mod.c pid.o threads.ko
debounce.ko hal_evoreg.mod.c hal_skeleton.mod.o po threads.mod.c
debounce.mod.c hal_evoreg.mod.o hal_skeleton.o rtapi threads.mod.o
debounce.mod.o hal_evoreg.o hal_stg.ko rtapi.ko threads.o
debounce.o hal_lib.ko hal_stg.mod.c rtapi.mod.c timedelay.ko
depends hal_lib.mod.c hal_stg.mod.o rtapi.mod.o timedelay.mod.c
doxconfig hal_lib.mod.o hal_stg.o rtapi.o timedelay.mod.o
emc hal_lib.o hal_tiro.ko scope_rt.ko timedelay.o
encoder.ko hal_m5i20.ko hal_tiro.mod.c scope_rt.mod.c
encoder.mod.c hal_m5i20.mod.c hal_tiro.mod.o scope_rt.mod.o
So where is Makefile.inc
a few lines above
01:10 < Sparky_> blocks.o encoder_ratio.mod.c hal_motenc.mod.c
And even an sudo ls doesn't show them to me???????
what do you mean doesn't show? it's right there in front of you
can you see my chat alex?
no, not sure about dcc chat
Damn I'm going blind this ls doesn't sort the same way this machines does.
but let's use /msg
Ok got a cat of it, it big, what do you want to see
type:sudo make setuid
it should work then
gene@shop:~/emc2/src$ sudo make setuid
Makefile:5: /Makefile.inc: No such file or directory
Makefile:67: depends/rtapi/_ulapi.d: No such file or directory
make: *** No rule to make target `rtapi/_ulapi.c', needed by `depends/rtapi/_ulapi.d'. Stop.
And I didn't get that error any of the previous times I did it???
cd'd to /home/gene/emc2/src...
The diff is that I'm ssh'd into it now, before I was on its keyboard
Sparky_: echo $PWD
gene@shop:~/emc2/src$ echo $PWD
Should I save my stepper_inich.ini and just blow this away and get a new checkout?
Sparky_: edit Makefile, change on line 5 $(PWD) to `pwd`
argh, that won't work
wonder why bdi loses $PWD with sudo
Douh why not
Sparky_: give me a minute.
Or do we need to put that in a pair of ``?
That didn't fly either
Sparky_: change line 5 to say: include /home/gene/emc2/src/Makefile.inc
les_w: I got 23 hot chicks hangin' out in my living room: http://fenn.dyndns.org/pub/camera/DCP_0600.JPG
gene@shop:~/emc2/src$ vim Makefile
make -C /usr/src/kernel-source-220.127.116.11-magma SUBDIRS=/home/gene/emc2/src CC=gcc-2.95 V=0 -o /usr/src/kernel-source-18.104.22.168-magma/Module.symvers modules
make: Entering directory `/usr/src/kernel-source-22.214.171.124-magma'
Building modules, stage 2.
make: Leaving directory `/usr/src/kernel-source-126.96.36.199-magma'
cp *.ko ../rtlib/
You now need to run 'sudo make setuid' in order to run in place.
gene@shop:~/emc2/src$ sudo make setuid
chown root ../bin/emc_module_helper
chmod 4750 ../bin/emc_module_helper
So that might have worked, test time, watch for smoke...
gene@shop:~/emc2/src$ cd ..
EMC2 - TESTING-2006-02-14
Machine configuration directory is '/home/gene/emc2/configs/stepper/'
Machine configuration file is 'stepper_inch.ini'
insmod: error inserting '/lib/modules/188.8.131.52-magma/rtai/rtai_math.ko': -1 File exists
iniFind is depreciated
But its running!!!!
you'll only get that once
the next time it will be cleaned ok
the rtai_math is still in there because of the last attempts
Now lets see if it can run 3D_Chips
ok the (one?) problem is that $PWD is lost on bdi's sudo, so sudo make setuid fails.
$PWD comes from @PWD@ from Makefile.inc, right?
no, it comes from the shell
joint 1 following error
emc/task/taskintf.cc 787: Error on axis 1, command number 342
your current directory is always $PWD in the environment
BUT... This was at a feed override of 250%!!
cradek: then maybe just use `pwd`?
I've sen syntax like $`pwd/rest_of_name before I think
I've seen syntax like $`pwd`/rest_of_name before I think
alex_joni: no, it needs to not change for the kbuild part
Chris: Have you looked at tp.c recently?
Sparky_: I sense this is a loaded question
One could say that I guess :)
cradek: it's a trick question, don't answer
Well, I was asking to see if I could take those N095G4P.0001's back out of 3d-chips.ngc
Sparky_: I think the direction-reversing bug is fixed.
Ok, lemme check, brb
Yippers, its gone. I'd forgotten that I had commented that line before. I just ran it up to 234% feed override before it got a tummy ache, and thats faster than my motors can go.
if you have your ini set right, feed override honors the constraints
But ATM, any joint error detected is forever as it comes back with every stroke of the F2 key
yeah, that's a reported bug - it will eventually go away if you keep hitting F2
That is I think, something in my .ini file thats way off level...
yeah, you should not get following errors, no matter what the feed override.
It can also be reduced to a one time reset with the right settings in the .ini file. I had it that way a week ago but wanted more speed.
Ahh, Chris, but I still am. My PERIOD is at 50u-s, can I now lower that, it feels peppier than befoire thats for sure
yes, you can now lower that, that bug is fixed too
Ok, trying 45u-s now
running at 200%
running at 220%
running at 230%
Going to try 40u-s now
And thats ok so far at 240%, faster than my motors
running at 280%, the backplot is flying compared to thee max speed of these motors
And at BASE_PERIOD=40u-s, I can just barely feel the lag, it WILL go lower yet I'm sure
Joint 2 error at end of first downstroke, feed override=300%, BASE_PERIOD=35u-s
And it still feels good, so 30u-s is next
Joint 2 error at end of first downstroke, feed override=350%, BASE_PERIOD=30u-s, machine still feels good