[00:13:14] <cradek> http://cats.iamnear.net/?q=68502
I guess this is supposed to show you five photos of cats near you
... what a failure in my case
good night ;)
goodnight, thanks again
I have started writing an article targeted at Windows users that don't realize what they are missing it's not done, but thought I would solicit input as to if it's worthwhile or succestions or....critisism welcome http://pminmo.com/emc.pdf
heh .. I'll have a look, but my ability to relate to the windows crowd is so bad that I wouldn't trust anything I have to say about the article
looks good, needs a picture
needs lots of things......
this all reminds me (for no good reason) .. I still haven't tried out the linux sheetcam
I get the impression most of the people here are Linux savvy, spend little time in MS world. Probably just me, but I hit lots of frustrations making the transition
What kinds of frustrations?
I'm not sure anyone here has used MS Word in a decade or more (except maybe Alex and Jymmm)
I used to use WordPerfect, but I think the last time was about 5 years ago
hardware incompatabilities to start, dual boot issues
I moved quite a few wordperfect files over to oo not long ago.
yep - once OO got a mostly working WP import filter, I was able to dump WP completely
not know linux at all, its a whole different language
pick one hw incompatability or dual boot and describe briefly.
the one thing it still does better is envelopes - getting them right in OO is tricky
(and of course there's "reveal codes" :) )
hm this isn't a good sign. on the first run, sheetcam says "the plugin has thrown an error. do you want to edit the macro?" and the default seems to be "yes".
That one I can understand why why would you need to know Linux to run a ubuntu pc.
okay video. What about simply using a generic.
when the pc comes up and you fave 640 x 480 and can't scroll your toast
I guess I've not seen that low a resolution from an install in a while.
I'm not sure I would be running today if it wasn't for your classes at the workshop ray
Did 8.04 fail to set up your wireless.
yes it did
three different "compatable" cards
I do favor external boxes but have worked at least two cards.
One took a bunch of hunting and a wrapper compile.
yes I've given up , will order an access point
Do you use the live cd and run from it to see how the install might go.
yes I did
four different pc's
and that failed?
How old were they?
the one I first started with was the one I spent sooo much time one
the last three went smoother
Ah that is a good sign.
but I had to decide I REALLY wanted EMC to go through the hassle
Yep and apparently you wanted it bad enough.
I'm still struggling with stuff from time to time, but I'm getting there
One real problem with MS tm loaded boxes is that they tend to be much to much and much to new.
You mentioned grub issues?
decided to order an Atom board to dedicate
little Intel itx board
Okay? On board video?
If so what's the chipset?
[01:50:53] <pminmo> http://www.newegg.com/Product/Product.aspx?Item=N82E16813121342&Tpk=atom
[01:51:29] <SWPadnos> http://www.newegg.com/Product/Product.aspx?Item=N82E16813135091
try that one :)
I haven't ordered one yet, but the price is staggeringly low
ouch, the otherone is already on order
they're both good deals, but T80 (less $20) including MB/CPU/RAM... O_O
at $80 -20 ...
actually I wanted an itx though
ah, it is a little smaller
just not sure about how the onboard video and rtai, but it has one pci slot
yep - the GOAL3+ also has a PCI and a PCIe x16 slot
I should just order that and be done with it :)
(but I'm leaving for Germany in a week, and there's no use having it gather dust while I'm gone)
oh hmm - 2 PCI slots
I see some issues with that Intel GMA 950 chipset
well I'm painting my house, replacing some siding, landscaping while the weather is nice........same dust
somebody has an atom running emc2
Something about display and having to get a newer drive software for wide displays.
Not emc ubuntu.
I search there first.
yes, it's on the linuxcnc website
I just don't remember where
and I still can't navigate it well
Ah maybe the wiki.
need to run....stuff to prep for tomorrows tasks...I'm ready for winter and I don't like cold weather
darn, missed seb
hmmm - I wonder if you have a minute to talk about a lyx problem I'm having
I am here
hey, thanks for testing out the hostmot2 driver, sorry it's not going smoother ...
Did the dmesg output tell you anything?
not really, it ends without any output from the hostmot2 driver
hm, try this:
echo 4 | sudo dd of=/proc/rtapi/debug
that will turn on all debug output, that's how i normally run when debugging stuff
then re-load the hostmot2 driver with those debug flags and try to make the pwm move
then pastebin *that* dmesg output
k, give me a minute. It is on another computer.
pastebin has a 20k limit - it may just truncate
so maybe paste only the juicy parts :)
(or maybe it's >20k, but issy ran into that earlier today)
Checking for a typo, I am getting no such directory.
it appears when the realtime kernel is loaded
Oh, sorry. I thought I had to run that before emc.
after starting emc but before loading hostmot2
oh, ok, don't have it set up that way. Running a full configuration. Let me do it that way.
fwiw, here's the script i use to play with hostmot2 on my development machine:
[02:56:04] <seb_kuzminsky> http://highlab.com/~seb/bzr/fabrication/projects/mesa-electronics/test
that's using the hm2.hal that's in configs/hostmot2
is there a way to set the rtapi debug level in a full-up machine config? there's got to be
does DEBUG=0x7FFFFFFF in the ini file not do what you want?
actually, you should be able to change the realtime script to do that echo for you (probably doesn't need root either, just echo)
i have no idea what that DEBUG statement would do in an ini file, i've never used one ;-)
echo 4 >/proc/rtapi/debug should do it
I am doing something wrong. After I do the "loadrt hm2_5i20 ..." I am getting no pins.
I guess you could put that in a HAL file
DEBUG in an ini file sets a number of debugging flags for the user parts of EMC
no connection with rtapi debug level
ok, I wasn't sure if the RTAPI stuff was also connected to that
it probably should be, but isn;t
ehj: does the hm2_5i20 driver load at all?
if you get no pins then it didnt load and the problem is probably earlier...
is the hostmot2 driver successfully loaded, along with the rest of the realtime environment?
Yes, the driver loads, because if I try to load it again I get an error. I can then unload it.
I will have to pick this up later. It is getting too late. I will try again tomorrow.
ok, anytime you get to a place where it's not doing what it aught, pastebin the dmesg and we'll take a look
the tail end of the dmesg ;)
see you later
sure, we'll get it squared
hint: "dmesg -c" to clear the log, then run, then dmesg to see only stuff since you cleared it
man what a strange bug
pwm works fine on mine here
i'm suspecting cables & connectors right now
except that it works with the old 5i20 driver
heh -sorry ;)
*pop* went my thought-ballon
does it help if I said hm2 pwm didn't work for me either?
yes that does help...
didnt know you'd tried it
jepler and I were trying to get dac output from the 7i33 and there was none
same symptoms as ehj? /enable works, dir works, pwm does nothing
I did not check the pwm dir, I only had the 7i33 output to test
was this with a 5i20?
it stayed at 0v but the enable did come on
have i lost my head?
I wonder if you had the 7i33 jumpered for external power
hold on, let me go try this out in the garage, back in 5
or did it work with old 5i20 with no changes?
no it is jumpered for internal
yes it works fine with the old driver
seb_kuzminsky, one thing to check - make sure that the md5sum from your bitfile matches the one in CVS/the deb package
here are the sums from my CVS checkout: http://www.pastebin.ca/1200708
here are the sums from my CVS checkout: http://www.pastebin.ca/1200708
in case you want to compare to yours (probably not the problem, but might as well check)
that's what i get here too
cradek: which firmware did you try?
I think there's an error in hostmot2.vhd in the 5i20 dir
oh hmm - maybe not
well its still working here
do you have all the PWMs on?
& 3 encoders & 2 stepgens
i'll try it with just one pwmgen
still no problem
have you checked the md5sums? :)
i compared mine to the ones you pastebinned and they're the same
ok, so we know there's a problem :)
hm, cradek, are you on CVS TRUNK? or 2.2?
behavior of encoders has changed (current top-of-tree TRUNK differs from hm2-0.10)
looks like jeff's encoder index changes had unintended consequences
but i dont think that's related to the pwm problem people are having
both top-of-tree TRUNK and hm2-0.10 have working pwm for me here
jmkasunich: you're right that i dont need sudo to write to /proc/rtapi/debug, but i *do* need dd:
> echo 4 > /proc/rtapi/debug
bash: /proc/rtapi/debug: cannot overwrite existing file
i've got noclobber on, hold on...
you could try echo 4 >> /proc/rtapi/debug
ah yes without noclobber it works just fine:
echo 4 >| /proc/rtapi/debug
never mind me....
i'm an idiot
EMC: 03seb 07TRUNK * 10emc2/src/hal/drivers/mesa-hostmot2/hostmot2.h: oops, i've been forgetting to update the version number
pwmgen should work now
alex_joni: what meaning have feedrate_blackout variable in axis ?
in this code:
if time.time() > feedrate_blackout:
vupdate(vars.feedrate, int(100 * self.stat.feedrate + .5))
I have no idea
I think it updates the feedrate from the status buffer only ever so often
otherwise the feedrate would jump all over the place (very irregular display)
but this is just me speculating :D
micges: after the feedrate has been modified by axis, the slider is not updated from the stat buffer until after 'feedrate_blackout' seconds have passed. If this is not done, then the slider will seem to randomly return to older values when it is being modified by axis (by clicking, dragging, keypresses, or whatever)
change that to 'if 1' and drag the slider left and right; you'll easily see the effect
(contrary to what alex said, when axis hasn't altered the feedrate recently, it immediately displays new values when they appear in the status buffer, so if you haven't touched the axis feedrate slider in the last few seconds and then you use a halui based method to change FO, it will be reflected immediately)
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emcdevel/2008-09-12.txt
good morning skunkworks
Good morning jepler. It is friday!
micges: see how much I know :)
skunkworks: even better
it's friday after 4PM
alex_joni: just means monday is coming that much faster for you ;)
well.. actually I still got tons of work, till I go home
yeah.. fun :/
I have a dell blade server that is boots the livecd to busybox. I am having trouble finding info on that.
I get the impression that it is because it doesn't see any drives?
(it has 4 scsi drives in it)
maybe you pressed ctrl-alt-[1-6] on accident
na - rebooted a few times now. Searching the ubuntu forums again :)
maybe you can poke around in the shell to see what the prob is
I get to the initramfs prompt
skunkworks: check the log.. it probably failed on something
maybe try to pass some boot params, like allgenericide
ok - do you know who you are talking to? How do I check the log? ;)
I get a lot of sd* unknown patition table..
there is a driver sd needs updating entry
alex_joni: at a fresh boot to the busybox prompt - there isn't any errors above (shift-page-up)
dapper seems to boot on it :)
maybe try installing dapper, then upgrading :)
* skunkworks doesn't have a problem using dapper..
hi alex :-)
seb_kuzminsky: we had a user with a small 5i20 problem yesterday
seems linux hotplug was picking the 5i20 card up, and the driver ended up insmod-ed on boot time
little request_firmware side effect? :)
beats me :D
heh - 168 updates for dapper.
can somebody tell me if is possible to run the emc tru remote x server without runing X
hooray, looks like maybe my isp got their routing issues fixed
hopefully this stays up
can somebody tell me if is possible to run the emc tru x-server remote without runing X on the linux
I routinely run axis over ssh's X forwarding. X also happens to normally be running on that system, but just for grins I stopped X on that system with /etc/init.d/gdm stop and it still works fine
yep - I've made an embedded system that never runs X, and graphical X apps still work remotely
X is installed though
my idea is to use pc104+ geode board , stopping the video , and remotely pass it to x - server
I'm not sure exactly what can be removed on the "target" system
no need to remove eniting
ok, if there's enough space on the "hard disk" for an X install, then you're all set
I changed the runlevels so that runlevel 2 is a console multiuser boot, and runlevel 3 is graphical multiuser (like everyone other than Debian does :) )
SWPadnos: the emc .deb package has concentrated on having correct dependency information and a single monolithic package. it's almost certainly a larger set than you need to just run emc plus just one of the GUIs.
I got rid of a lot of junk in the startup scripts too
the grafic on the rt linux is something i dont like , generaly the shared memory is not for rt linux.
patches to configure+makefiles to make it easier to build emc subsets, and pacthes to debian/* to make non-monolithic packages with leaner dependencies would both be thoughtfully considered for inclusion
another issue is that to work correctly , you have to forget the acceleratin on RT systems
I also noticed that timing got worse when I was using networking, so YMMV
jepler, OK, I'll keep that in mind
I think we may want 3 (maybe more) packages. emc-core, emc-gui, emc-doc
yeah, maybe that too
actually, emc-sim-hal and emc-rt-hal
the video acceleration writes to the memory directly , and stopping all the rest. otherwice will not be an acceleration
but I'm not sure quite how to break it out yet
SWPadnos: yeah I'd leave that alone at first
actually, HAL is not really EMC, so hal-rt and hal-sim may be appropriate (except for the fact that there are other things called HAL in Linux)
sure, there might be a better package name
jepler, yep, sim can be a complete system anyway, since it isn't meant to run RT machines
and it can be ignored for now also ;)
ok. let's try , we need open source x-server for windows and compilation for the emc-axis
I can do the windows X part as open source ,
cygwin-X works fine with EMC programs
some features of AXIS will be missing if you want to actually execute it on a remote PC (versus running it on the target and displaying the UI on a remote PC)
jepler , can you do the compilation for the linux
the axis runs on the target , only the video , keyboard , mouse and sound will be remoted out
I dont know haw will be the RT linux with the net..
issy: I don't understand what you are asking me to compile
the intel chips work with hardware mac hub , but most of the rest are software
Well , I would like to try the emc without the x
runing it with the axis guy on the local computer
and transfering the video and keyboard to the
that doesn't require any change to axis; that is a basic capability of the X Windowing System.
boot the EMC computer, don't log in. using a different computer, do "ssh -X ip.of.EMC.PC". run EMC from that SSH terminal
yes , need only to remove the x
from the linux emc computer
ok, then on the EMC computer, switch to a console and login. do `sudo /etc/init.d/gdm stop'
switch back to the X screen and hit ctrl-alt-backspace to kill the X server (or use kill from the console)
this is the top google hit for 'ubuntu disable X server', it looks relevant: http://ubuntuforums.org/showthread.php?t=43516
then go to the remote PC and use ssh -X
thanks. I will try it now
with realtc net chips work worse!!!!!
the software hub
WAUUUUUUUU 6000 / interupts/second....
let see the intel
I have an embedded PC with a core2 duo 1.8 GHz, which often has latencies less than 1000ns. if I run video or network on that machine, the latencies spike to 14000 ns several times a second
mine is Intel
240 interupts/second with intel
if you want an Ubuntu installation without the X server, then you want to start with the "server" install CD from ubuntu instead of the emc2 live cd. then add emc2 by hand. this will pull in X-related libs but shouldn't pull in any X server
who cares how many interrupts there are? the important thing is how long they prevent the RT kernel from doing its work
too much for realtec...
with intel the latency is 18000
the number of interrupts from your network card will be proportional to the amount of network traffic on your ethernet segment
with realtec 36000
the packages and the size are the same
the difference is twice.
well if you intend to use mesa or pluto boards, the latency is not so important
I'm not sure I know what you're saying. as long as you do, I'm OK with that :)
but for steppers at hi speed ..... using the parport....
welll , not the ideal
heh - for my application, I needed 10000 interrupts/sec, and had to do floating point on a few dozen analog I/Os every time
what are you using
I used a Mesa card to interface to my analog card
mesa + 7i37 for digital, custom board for analog
also connected to mesa
with the mesa you can work with latency up to 40000 , try to connect at this latency steppers to the parport... with no mesa , and you will see
oh, of course. I know high latency is bad for steppers, and for servos controlled with parallel port PWM
in my case, 40000 would have been way too much latency anyway - that would have been a 40% phase shift in the PID output
it what i am sayng . with net using software hun , the latency junp twice
I had the same problem with an Intel chipset. it only went to 14000 or so, but that's a lot compared to 600 (yes, six hundred)
i see that the intel net is hardware , but many thers are software .
so , when using steppers forget the net if its not hardware mac
doesn't bother me - I have no intention of using steppers ;)
the test machine I am using is 800 mhz via c7 , and works fine with the steppers , if the net is disabled
i got speed at 14/m- min
if net and video are disabled ;)
(which makes it hard to see what's going on :) )
seems the net drivers and the rt are not intended to work together
mean net drivers
I've been tempted to search various net drivers for CLI calls, but haven't gotten around to it
unfortunately in the last editions of the net chips , all the manufacturers are using software mac and hub's as they are easy to use.
I have to ask Peter walice if can be do inthe mesa fpga ,
then the remote communication well be much bether
yes , ehernet
you would need external hardware to do that. you can put the ethernet MAC into the FPGA, but the PHY and magnetics have to be hardware
yes , this is siple to be done
simple , sorry
sort of. remember that the FPGA connects to the outside world over a ribbon cable. unless you make a daughteboard that plugs directly into one of the connectors, and you are careful with the termination resistors (you'd probably have to remove them), signal integrity will be a problem
you can buy pc104+ cpu and mesa 7I65 (pc104) and you have 8 axis machine , little bigger than cigarette box
you can put it on your machine and with single ethernet cable to connect it to the remote station
it's expensive, but you can do it :)
this is important for me , because I am manufacturer and the remote administratin wll be easy for me
PC104 processor cards are very expensive compared to normal PCs
not realy expencive.
I have machines all over the world ,can you imagine haw much we have to pay for fixing software with no remote administration?
considering that I can buy a complete motherboard+CPU+512M memory system for $80 (minus a $20 rebate for now), PC/104 is definitely "relatively" expensive
the mine are 163 USD
ok, that's not bad
if you are doing for inhouse , it is ok.
geode 500 Mhz
for industrial application must be fanless
look for board 6100 at www.aewin.co.tw
the Geode is a pretty slow CPU, that may be contributing to the latencies you're seeing
I guess those routing problems aren't quite completely fixed :)
no idea , but I will test them
The problem will be to find or to write the open source X-server for windows
so you don't like Cygwin/X ?
no a bunch of dll's
lot of dll's with target to do something that is the base of the linux
need to be open sources.
and not bigger than 1 mega
cygwin/X is open source, except that it allows linking with proprietary executables
[16:49:38] <SWPadnos> http://x.cygwin.com/
they're looking fora maintainer too :)
the dll\s are not open source.
tonight I have to finish the reference points of my gui , and during the wikend will work on mine idea to see haw will be with the geode and the remote
guess jepler's routing isn't fixed afterall
there's a lot of network trouble in lincoln this morning. it's getting better though.
similar things going on here, or between here and everywhere else
but not a lot of rain to "explain" it
cradek: can you tell me you ideas about run from line ? I read that you have thouths to fix it once for all
I think it should do exactly what it's asked: start executing at the selected line. Any modes that are required for that to make sense (spindle, G90, tool loaded, G43, coolant) would need to be set by the operator in MDI or manual mode before running.
I think this is the only way that the operator can always get the behavior he wants.
you could write "safe start" spots in the gcode program. a "safe start" would be a place where all modal commands are set. at every tool change would be a good place for this. then starting there is very easy.
I still don't understand what happens if you select run from line inside a function :)
neither does EMC
alex_joni: hopefully, all hell breaks loose
that's certainly what cradek is asking for
it would execute starting where you say, and then give an error when it gets to the O-endsub
obviously you would not want to do that.
I can see one good use of pause/start-from-line though: to change the number of iterations of a loop
or to change some endpoint for a while condition
dunno how to code it though :)
I think the reality is that looping/while/sub code is fairly incompatible with restarting in the middle of the program
I haven't fully convinced myself that it has to be that way, but I think you're right for the foreseeable future
also, I don't care about that, and doing what the user says is the best design that can be actually implemented (in contrast to doing what the user wants)
yes, it's much easier to parse what people say
the current run from line, basicly starts interpreting the file from the beginning
and compares the current line with the preset line
so you want to ditch that, and just start from the preset line?
sounds good :D
it seems like the current code will work nearly as well as cradek's idea when you write "restart blocks" into your gcode
but I think AXIS will need to truncate the preview too
one thing I commonly have in gcode files is some settings at the top using parameters; I think that these settings would be lost under cradek's scheme, but are retained under the current scheme
jepler: yes it does.
jepler: with the current code you have the advantage of setting interp internal things .. like G21, etc
exactly, otoh one can set those in MDI
jepler: you could step through them, and then run-from-line where you want.
or, set them in mdi
I am not sure if that's true or not. These are settings in named parameters or numbered parameters that aren't written to the var file that I'm talking about ..
jepler: maybe cradek is saying that his run from line can work after stepping a couple of times
without stopping the program :)
I could be wrong about how parameter saving works
it might be nice to have a wiki article or doc section about writing gcode *today* that is usable with *today's* run from line
safe start spots currently work fine. I write them at tool changes.
would you be willing to write it up? either for the wiki or for me to add to the docs?
I hesitate to document behavior we are going to ditch
a utility that scans a file up to a point and prints a G-code line to set modal codes to the state they'd be in had the program been run would be useful
I wonder if SAI could be coerced into doing that
get the codes, stick that in MDI, remove what you don't want, then restart
bbl, I forgot to eat today
oh. me too
hmm.. not me
cradek: those ideas are cool
this is actually the same steps to restart that I've programmed into axis
1. set params in MDI mode
2. goto safe start point (begining of selected line)
3. run program
cradek: SWPadnos: Are you still here?
Lerman, I'm back now