Anyone know how to check ram on a linux system )ubuntu)?
That's all I can think of I haven't checked trying to fix this erratic stalling...
reboot, hit escape to get the boot menu, pick memtest
cradek: That sounds simple enough even for me. Thanks,
davidf: did you decide to try a different machine?
bad ram will generally cause applications to crash
hey that reminds me
in cvs yesterday jmk and I put in some reporting of realtime anomolies
it would be nice if you could try it, but to do that you'll have to compile the cvs version, as it's not in 2.0
cradek, not yet. There is some dust in this box, and the stralling is so seldom and random, It just doesn't seem like something like incompatibility.
Sometimes it will go for 5 minutes perfectly, then a 2 second buzz, and continues (though offset, of course.)
I'd be glad to do it if I can figger it out.
I think there are instructions in the wiki - I'm looking for them now
Don't know if I have the requisite compiler / linker installed.
it's all very easy to install
[00:21:09] <cradek> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl/emcinfo.pl?Installing_EMC2#On_Ubuntu_5_10_and_6_06_from_source
ok I'll have a read.
on step 2.2 use 'cvs -z5 -d:ext:firstname.lastname@example.org:/cvs co -d emc2.head emc2' instead of what it says there
cradek, do you think its possible that there is something in 2.0.3 that might be causing this? I don't think it ever happened before I upgraded.
no I really don't ...
we carefully put only very safe bugfixes in those 2.0 versions
did you try downgrading to one of the earlier versions to see if the problem goes away? That's always an option
It's just SO weird. Perfect behavior in every way for a whole run, then later, Zap! out of the blue.
you could certainly try 2.0.1 or whatever you were using before
the configs are compatible back to before 2.0.0, you could even go back to 2.0.0
I even changed to a 7406 open collector off the parport, wewired, checked every connection, added an AC line filter, spent an hour on the phone with a Parker tech, still the same crap.
well I still think it's your computer.
this checking in cvs head *might* show it, if so.
Seems like it would be doing it more often.
what's that you said?
yeah I guess so.
4 lines up.
'this checking in cvs head *might* show it, if so.'?
that's the development version
the one I suggested you compile and try
oh, you mean if I compile & run the dev version?
I'm going to go do the same so I have a baseline
ok, I'll give that a try right now. Do I need to uninstall it first or anything?
jepler, sorry I didn't see your question till now. No I haven't but I was thinking of doing that.
no you don't have to uninstall
ok. I've started the process...
getting an error message:
0% [Connecting to us.archive.ubuntu.com (18.104.22.168)]
breezy/main libglib1.2-dev 1.2.10-10ubuntu1
Could not connect to us.archive.ubuntu.com:80 (22.214.171.124). - connect (113 No route to host) [IP: 126.96.36.199 80]
0% [Connecting to us.archive.ubuntu.com (188.8.131.52)]
try 'sudo apt-get update'
huh maybe their mirrors are down again
yeah its trying 2 different IP's same error on both- cant connect.
huh, I'm able to install packages from us.archive.ubuntu.com with apt-get
how do i make it stop trying to connect - cant type any other commands.
davidf: you can use the european ubuntu mirrors: in /etc/apt/sources.list change all the us.archive to uk.archive
ok got a bunch of stuff, now its back to the same IP it cant hit:
99% [Connecting to us.archive.ubuntu.com (184.108.40.206)]
then sudo apt-get update again
cradek, so just open that file in text editor, and change everything from us.arcgive to uk.archive?
yes, but make sure you use a text editor that doesn't add extra newlines where they don't belong
everything in that file is commented out. Should I remove the comments?
everything? can't be.
hmm. you're right.
ok. I'll just change the active stuff by hand.
hmm. cant edit it. guess it is read only.
I need to set permissions I guess?
geez all the hand holding you guys have to do.
sudoedit or sudo edit?
ok now how do i save it?
or do i?
look at the help on the bottom of the screen, I think it's control-w
Did that, but it is giving me a weird name to save it as.
that's fine, just hit return
afterward sudoedit will put it in the right place for you
now sudo apt-get update
and if that goes well, you can go back to the wiki instructions
do I want to build axis?
yes do both of the build-dep
03cradek 07HEAD * 10emc2/src/emc/motion/control.c:
tighten this test to 10%, as my reliable machine and jmk's both show
slop around 1% max.
on step 2.2 use 'cvs -z5 -d:ext:email@example.com:/cvs co -d emc2.head emc2' instead of what it says there
cradek, is there a way to copy that line from this chat? ctrl - something?
cant do it with the mouse.
just highlight it, then go to the terminal and middle-click
two button mouse.
pulkl its tail?
try both buttons together for middle-click
does it have a wheel?
wow, old mouse then
picked it up from a parking lot after an auction. :)
nobody wanted it. felt sorry for it & took it hoome.
you sound like me
"so what if this computer is the size of a school bus, it's PERFECTLY FINE"
bash: cvs: command not found
you skipped a step then
oh. ok... let me check
sudo apt-get install cvs
that's the one
why do I want to get an apartment anyway?
never noticed that
Media change: please insert the disc labeled
'Ubuntu 5.10 _Breezy Badger_ - Release i386 (20051012)'
in the drive '/cdrom/' and press enter
now the fun starts looking for that.
cancel that again, and comment out the first line (cdrom) of /etc/apt/sources.list
whatcha mean by cancel that agin?
just hit control-c
then sudoedit /etc/apt/sources.list again
smuckers is selling pre-made PBJ sandwiches with no crust
for when you just don't have time to "cook" your own sandwich I guess
When my sis was little, she went through a phase where she absolutely refused to eat anything but jelly sandwiches with the crust cut off. Maybe that's what happened to her brain...
my parents regularly buy pre-fried microwaveable bacon
seems not long ago we would have all had a good laugh at that
it's like pre-toasted microwaveable toast
(which I've also seen)
actually its not too bad, but extremely thin and expensive. Saves a paper towel though. That costs something.
you've got to be kidding.
crust, or not?
well forget that then.
believe it or not, it was "bimbo" brand
ok I think I got lost here.
I tried the cvs -z5 -d:ext:firstname.lastname@example.org:/cvs co -d emc2.head emc2
and I'm getting cmd not found again.
did you do apt-get install cvs?
did you do sudo apt-get install cvs?
do it again?
isn't that the one we aborted because it wanted the cd?
ok then do it again
here's the response after that:
dave@ubuntu:~$ cvs -z5 -d:ext:email@example.com:/cvs co -d emc2.head emc2
bash: cvs: command not found
dave@ubuntu:~$ apt-get install cvs
E: Could not open lock file /var/lib/apt/lists/lock - open (13 Permission denied)
E: Unable to lock the list directory
dave@ubuntu:~$ sudo apt-get install cvs
E: Type 'restricted' is not known on line 2 in source list /etc/apt/sources.list
E: The list of sources could not be read.
just a sec...
ok, jepler was right to warn of bad CR's
ok that step is done....
darn cant seem to get to the top dir.
did cd /
you want to get the cvs in your home directory
just "cd" by itself to get there
no I'm there, but cd src gives no such dir.
dave@ubuntu:~$ cd src
bash: cd: src: No such file or directory
it's inside the emc2.head directory you just checked out
ok. The instructions just say 'from the top directory, do cd src. Probably should idiot proof that.
please edit the wiki instructions however you see fit
Now you're getting dangerous.
dave@ubuntu:~$ cd emc2.head/src
bash: cd: emc2.head/src: No such file or directory
did you do the checkout?
I think so.
it would have printed lots of files that it was getting
dave@ubuntu:~$ cvs -z5 -d:ext:firstname.lastname@example.org:/cvs co -d emc2.head emc2
that should have created a directory emc2.head
dave@ubuntu:~$ cvs -z5 -d:ext:email@example.com:/cvs co -d emc2.head emc2
The authenticity of host 'cvs.linuxcnc.org (220.127.116.11)' can't be established.
DSA key fingerprint is 5b:b7:3d:04:9d:dd:86:ad:dc:c6:78:81:d4:3d:05:8e.
Are you sure you want to continue connecting (yes/no)?
ok thats really a bunch of files.
that's all the source for emc2 and its documentation
ok thats done.
ok you should be able to do step 2.3 now
the "top" directory is that emc2.head
I tweaked those instructions in 2.2 and 2.3
still doing the make
it might take a bit the first time
has to compile everthing the first time right?
some of this is coming back to me. Did some of this years ago but its pretty foggy
You now need to run 'sudo make setuid' in order to run in place.
chown root ../bin/emc_module_helper
chmod 4750 ../bin/emc_module_helper
bash: cd..: command not found
looks like a successful build
ok, did scripts/emc & emc came up.
is your custom config shown?
now try it?
yeah just run it
hmm, clicked ok & it disappeared.
what config did you choose?
custom one. Probably hit cancel by mistake. Ran it from the task bar & its up.
that's 2.0.3 then
you need to run the one in the emc2.head directory
you now have two versions, the released and the development, just like developers do
03jepler 07HEAD * 10emc2/docs/NEWS: this probably bears mentioning
03jepler 07HEAD * 10emc2/docs/NEWS: put items in chronological order
a 2.0 configuration won't work with the development version without changes
right. got errors.
make a copy of your config and let's fix it up
I forget what the differences are -- I know one is to 'loadrt trivkins' but I think there's another
the whole directory
or the hal files?
if your directory is 'david' copy it to a new directory 'david.head' or something like that
this way you can run both versions by picking the appropriate config
jepler: nml file I think
oh the spindle stuff?
what's a good name for a wiki page to document this? There should be one...
so what new?
it's long but descriptive
davidf: I think jepler's going to write a wiki page with instructions
well I'm starting one
you might not want to wait .. in fact I was hoping that you'd figure it out and I'd just write it in the wiki
well let's see
copy ~/emc2.head/configs/common/emc.nml ~/emc2/configs/myconfig
find the halfile that has loadrt EMCMOT in it
it's probably core_stepper.hal in your custom config
above that line put "loadrt trivkins"
if you have spindle control, find the halfile that hooks up the spindle signals, and change e.g. iocontrol.0.spindle-on to motion.spindle-on
this might be in standard_pinout.hal
I think this is it
when you said 'find the file that has loadrt EMCMOT in it... Did you mean find the line in myconfigs, of actually 'find the file'?
-of, + or
I did mean find the file
it could be in any hal file depending on how you organized your custom config
thinjk thats in the ini...
no it'll be a hal file
I'm guessing core_stepper.hal for you
OK, here's my first stab: http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?UpdatingConfigurationsForDevelopmentVersions
well now wait a sec...
jepler: nice showing the diffs
I did the dir copy thing, but instaed of getting a dir, I got a file called myconfigs
look above just above err cp.
that made a file in that dir. Is that what you wanted?
in that command you were supposed to replace "myconfig" with the name of your config
ok, so now its myconfig. Fine.
yes we're trying to replace the emc.nml file in your config with the new one distributed with the devel source
but I thought I was copying a whole dir.
yes you were supposed to do that first, I thought you had that done
let's back up and look at the big picture
the idea here is to make a new config (which is a directory with some files in it) that's like the old one, but with some things updated
think I know what you want
one of those things we're updating is to replace the .nml file with the one from the new source download
so copy the whole dir, make one right next to it with a new name
if you find any other items add them to that page...
and link it from somewhere
if I were as smart as jepler I'd go to bed too
so copy ~/emc2.head to ~emc2dev.head maybe?
If you want to do this later thats fine.
where's your custom config?
its in there.
ok what's it called
really? that's the one you've been running in 2.0.3?
I have an ini file called inchMill50K.ini in the standard place.
I didnt make any new dirs.
what directory is that one in?
Just new files.
ok and that's from your home directory right?
cp -R stepper-xyza stepper-xyza.dev
cp ~/emc2.head/configs/common/emc.nml ~/emc2/configs/stepper-xyza.dev
now pull up ~/emc2/configs/stepper-xyza.dev/stepper_xyza.hal in your editor
I see. That's something like I thought you wanted.
make SURE you're changing stuff in the .dev config now
we don't want to mess up your config for 2.0.3
at about line 5 there's loadrt [EMCMOT]EMCMOT ........
above that, add a line that says "loadrt trivkins" (with no quotes)
ok save and exit that file
pull up standard_pinout.hal in the same .dev directory
near the bottom find linksp spindle_on iocontrol.0.spindle-on
change "iocontrol.0" to "motion"
so it says .... motion.spindle-on
ok save and exit that one too
I think we might be done
go back to ~/emc2.head and run scripts/emc
pick your .dev config and see if it starts
EMC2 - pre-2.1 CVS HEAD
Machine configuration directory is '/home/dave/emc2/configs/stepper-xyza/'
Machine configuration file is 'inchMill50K.ini'
insmod: error inserting '/home/dave/emc2.head/rtlib/motmod.ko': -1 Unknown symbol in module
HAL:5: ERROR: insmod failed, returned 1
HAL config file /home/dave/emc2/configs/stepper-xyza//stepper_xyza.hal failed.
Shutting down and cleaning up EMC...
+ Terminated $EMC2_BIN_DIR/$EMCSERVER -ini $INIFILE
EMC terminated with an error. You can find more information in the log files
as well as in the output of the shell command 'dmesg'
now wait a second
you picked stepper-xyza
you need to pick stepper-xyza.dev
didnt see that. did we make that?
that's the one you were just editing
... I hope
stepper-xyza.dev is a dir...
emc configs are dirs with files in them
err, an emc config is a dir with some files in it
in there is inchmill50K.ini.
you should see that in the config chooser that comes up first
But I didnt see it in the list when emc2 ran. Let me try again. I picked the old one.
if it's not there we messed something up.
there was a file called stepper-xyza.
no it was a directory.
with your ini file in it right?
I cliked that & it ran. But I dont know if its a copy of inchMill50k or not.
well I double clicked it & emc2 ran...
no its not right. It looks like the default file.
your customs should be at the top of the list of configs
right got it. But it came up with a file already loaded.
emc2_axis in itallics. right?
yes that's normal
it's just for looks
but you can run it if you want
ok. Should I run that g code now?
so does your mill work now?
well we'll see.
I forgot to home the z axis
Let me stop & restart so I can see the drawing.
ok great, it runs
But that happens all the time.
Well obviously not ALL the time or we wouldnt be doing tyhis.
now the whole point of this is that there's a new error message when a particular problem with the realtime subsystem is detected
I am hoping you get that error at the same time your axis stalls
if so, that will confirm my suspicion about your machine still having problems.
so far its tracing back over the drawing perfectly.
the feed may be too slow in that file to test your machine well
runniong 3rd time...
you could run whatever gave you problems before
no its pretty fast actually. faster than I was running
but this is mostly straight lines, longer than the script lettering (lots of ver short line segments) so the accel has more time to speed up.
yeah your low accel is going to really slow you down on lots of short lines
looks like F 60 that right?
don't remember, possibly
Think my max vel in the ini is 30 right now.
ok, it's just going 30 then
looks about right.
ran well again.
you can definitely hear if it stalls right?
since the moves are short it may not be off much after a stall
see, this is what is such a pita. Wont screw up when you want it to, just whenever you think its fixed.
I don't think we could have fixed it ... you're just "lucky" right now
I'm crossing my fingers for that error
It's so cool to see it going back over the 4th time, & not be able to tell you drew it more than once.
Even after 10 or 20 homes on x & y I cant see even a hair's breadth of error with thiese home switches.
Geez, could it be the file I was cutting???!!!!
I swear I'll shoot myself.
Maybe I shold load that test file.
still cutting emc2 axis...
but no problem in the gcode can cause stalls on a properly working machine
I'm going to get to bed
oh crap! Just got that error message and I clicked it before I read it all!
look in your dmesg
Said unexpected real-time delay.. & something else...
that shows the system log
the error will also be in there
is that a halcmd?
nope, it's a regular linux command
I typed dmesg|tail -10 and nothing happened.
in a terminal?
it should show some log messages
try just dmesg
just a blinking cursor on a blank line.
there's no prompt?
is that the terminal running emc?
ok it's busy, use a new one
[10560.592454] 984331: ERROR: Unexpected realtime delay; check dmesg for details.
[10560.592479] In recent history there were
[10560.592485] 723239, 722550, 724707, 724601, and 720988
[10560.592491] elapsed clocks between calls to the motion controller.
[10560.592498] This time, there were 11132841 which is so anomolously
[10560.592504] large that it probably signifies a problem with your
[10560.592510] realtime configuration. For the rest of this run of
[10560.592524] EMC, this message will be suppressed.
well there we have it
well there YOU have it maybe.
your machine got lost doing something else (ignoring emc) for 10 million clock cycles
well what a goof-off.
that's .0125 seconds at 800MHz clock
did you get a stall when that happened, or were you between program runs?
q is, what?
this absolutely confirms that something about your computer is incompatible with realtime
but what can take that kind of processor time?
looks like a network check of on line or something
that's what you have to figure out :-)
would a BFH help?
help me, anyway.
we've checked the obvious things, onboard video, ACPI
Too bad I'm in the basement. I could throw it out the window but its at ground level.
it wouldn't be as satisfying then I suppose
There is a table saw down here though...
describe the computer to me
Well, lets see, its a piece of shit.
now at least you know your time would not be wasted trying a different computer
733 MHz PIII
network card, dsl modem
Those little bus slots
any other cards in it?
no ISA slots etc
integrated everthing, compaq
so the cards in it are just ethernet and video?
on board ethernet.
so the only card in it is video?
integrated video, not in use.
yes, video card.
does it have onboard sound?
think so. 3 sockets for phone jacks.
maybe you should disable that in the bios
and anything else onboard that you're not using
I think I tried to do that...
are you using any of the usb?
I bet your two button mouse isn't usb...
no. I disabled the usb & that didn't help either.
Then I went back to default config I think after I turned off everthing one by one to no avail.
once you've grasped at each of the straws a few times, you might just have to try another machine.
does it have ide disk or scsi?
regular ide or sata?
switch it over to another box?
I think regular.
But I'm not sure.
what model of video card is it?
I saw that awhile ago, cant remember the name, but I think Jepler was talking about one yesterday.
0000:02:04.0 VGA compatible controller: Matrox Graphics, Inc. MGA G200 (rev 01)
I have a G450 in mine and it's known good
I would suspect a G200 would also be ok then
I think I'm out of advice except 'try another box'
So you dont think any high level process can grab the puter's brain then huh?
if you take the video card and disc, everything will probably just work
just switch it to a new box, huh?
something is sure grabbing the CPU for a long time (.01 seconds), we just don't know what
I bet that's dozens of step pulses missing
no wonder it stalls
Is there some way to list all the processes currently running (like the windows task manager?
yes but none of those are the culprit
something's taking the CPU completely away from Linux
it's a piece of hardware or programming in the system bios
no, it's actually in the hardware somewhere beneath all the software.
like dma memory refresh
yes it could be something doing lots of dma
that shouldnt take that long though.
for mem refresh
I think onboard video does that (suspends the processor, moves some memory blocks around)
no but it's something at that level that's wrong.
ok it's bedtime
ok, well, Why dont I let you off the hook till tomorrow
now you know what you have to do next
or forever if you want of course.
I want to see this resolved too
right. where's that BFH?
actually I think this is great news
we have a nice test in emc that detects problems that were invisible before
Thanks Chris. Youre the best.
visible problems are MUCH easier to fix.
Yes that is really neat. Great job.
you now have a visible problem and you'll KNOW when it's fixed because the error will be gone.
ok goodnight, keep us posted, jmk will be happy to hear about this too
Still cant believe how downright NICE you guys all are.
oh did I forget to send you my consulting bill!?
goodnight & thankls loads.
No problem. I'll put it in the stack.
looking for cheap hardware to develop a home machine using linux and EMC any direction would be great... I have a mini mill I made using floppy steppers and parallel port now I am hungry for more...
how does this sound as a backup solution?
use 7zip, to compress files.
use par2, to create recovery archives as well, just incase the DVD gets scratched (25% recovery method).
so 25% of the total data can be fubared, but since I have the par2s, I could recover whatever was missing.
[11:54:32] <A-L-P-H-A> http://en.wikipedia.org/wiki/PAR2
[11:54:44] <A-L-P-H-A> http://en.wikipedia.org/wiki/7zip
To keep you guys informed: I have finally finished my rtnet driver for HAL. We are now going to run some tests on it to test its performance - if you are very interested and don't mind the pretty long installation procedure, you're welcome to test with us
one day, I'll want to test
I may want to look at it next week, though I won't be able to do any testing since the only time I'll have will be on a plane (and I'm not sure I can take a network as carry-on)
glarrrrgh - I can't believe this. my cell phone now says "Unauthorized Charger" when I plug it into the computer to charge
it used to work!
also, "glarrrrgh" sounds like a pretty funny screensaver
her I was, thinking it would say "Charge Complete". like it does every other time
rendered 3D arrrrrrgh
cradek_ is now known as cradek
evening (or morning) all
morning where I am...
I think there is a problem with the emc2-instal.sh script for breezy badger. It has www.linuxcnc.COM in it.
oops! that would be a mistake.
yeah - I changed it to .org and the "update" was still looking for .com.... so I looked in the sources.list file the GREP added something to. And it was inthere as COM from the first attempt to run.
SWPadnos: can you fix it?
Then I hadto argue with ubuntu that I could not work out how to become super user in - till I found an ubuntu forum that told me just to use some magic synaptic manager thing
and it is all good now
Who do I contact so someone can fix up the file that people will download. (save others having the same problem)
let me see if I can fix it
Up till tonight I was just playing with a live-CD. This is the first time I tried a permanent inst.
sorry - fix the script? where is it?
on linuxcnc.org somewhere
heh - ok
I think I have a login there - I'm looking for it
I can log in to fix the problem. if I know where it is
[13:57:05] <cradek> http://linuxcnc.org/emc2-install.sh
[13:57:11] <cradek> http://linuxcnc.org/dapper/emc2-install.sh
[13:57:23] <andrewm1973> http://www.linuxcnc.org/emc2-install.sh
oops - you beat me to it.
be sure to check both
I just rebooted the linux machine and am pretending to cut up a 3d-Chips in AXIS-Sim now - so all worked out good in the end :D
Now I just need to build my etch-servo circuit board and I can play in real life.
ok. fixed the one at root level. is there a reason the breezy script is in the root and the dapper script is in the dapper/ dir ?
(the one in dapper/ was OK already)
SWPadnos: history? once there was only one, now there are two
I suppose they could have descriptive names like emc2-dapper-install.sh and emc2-breezy-install.sh instead
or one script could detect the distribution
Could there be some legacy reason to not have a different name for them ?
hey - that might actually reduce workload when new Ubuntu releases come out ;)
(though we'd probably need 6 or 7 releases to make up for the extra effort for detection)
andrewm1973, I don't think so, except that the wiki instructions remain the same after the download step (ie, "run emc2-install.sh")
SWPadnos: then you have to add detecting whether the packages are available for that distribution
(. /etc/lsb-release; echo $DISTRIB_CODENAME)
sure. I guess we'd need an automated patch-and-release system as well - one that tracks changes to the kernel, RTAI, and emc, and makes new LiveCDs and debs whenever needed
well guys - thanks for gettign that fixed up.
or we can just have a second copy of emc2-install.sh in breezy/ ;)
sure. thanks for pointing it out
It's past midnight here and I think I should go have a kip.
a kip? hmmm
kip = sleep (kinda)
the internet says ... http://www.worldwidewords.org/qa/qa-kip1.htm
skunkworks: what's up?
trying to make it tru a friday without too may issue
03jepler 07HEAD * 10emc2/src/emc/task/emctaskmain.cc:
introduce a new internal flag, emcTaskEager, which is set whenever task is in
the process of adding new segments to the queue. Halscope debug traces show
that running 'spiral.ngc', the tcqLen() rises to 1000 in about 1 second (and
should be relatively independent of [TASK]CYCLE_TIME), while the old code might
enqueue as few as 30 per second.
sorry - yes
you have missed my spelling issues - haven't you?
jepler: how is the etch-o-sketch? have you got an itch for a mill or lathe?
lerneaen_hydra is now known as lerneaen_hydra_s
lerneaen_hydra_s is now known as hydra_school
skunkworks: I still haven't run it since the end of 'fest
still need to solder the wires back on the encoder?
skunkworks: I did that and tested that the quadrature inputs worked again, but never actually ran it
and I even had something interesting in mind to do: skip using g-code at all, and use 'siggen' to make it generate spirograph-type designs
jepler, cradek helped me a lot last nite. We found an RT error with dmesg while emc ran. .01 sec. lost uP time. Good job!
yeah I saw the end of that conversation still on my screen this morning
Its a real nice thing.
I noticed it also gives a lot of info on what's on the system... more than the bios setup
One thing I saw was Bluetooth. Could that be a problem, & could you please tell me what it is?
bluetooth is a short-range wireless communications system
that's what I thought. Didn't know I had that. Must be built in to the MB. Didn't see anything in the Bios about it to turn it off.
you probably don't have the hardware, just the driver loading (just in case)
unless it's a laptop (which it isn't, IIRC)
right its a tower.
there are a zillion things that are enabled by default, so they "just work"
some day, one of us should put up a wiki page on "disabling all that unnecessary and possibly harmful stuff"
That would be great. yeah, if I understood more, the dmesg output might give a clue to what might be the problem. It is really nice but I'm clueless about most of it.
That would be great. Yeah, if I understood more, the dmesg output might give a clue to what might be the problem. It is really nice but I'm clueless about most of it.
hey bill - how is the mill coming?
any new pictures?
I drilled mounting holes in the motor-mount plate and they were spaced wrong.
got pretty frustrated and took some time off from it.
measuring stuff is hard. :-|
do you know roughly how critical the tolerances for power mosfet's?
I dont know really, I let the smoke out of a IRLML6402 mosfet, and I'm trying to find a replacement
well... once the genie's out... it's out for good.
it's a SMD part, but I could use a non-smd replacement if I could find one.
it's not like the cartoons, where they can get it back in.
but I just dont know how much wiggle room I've got with the numbers.
depends on the whole board.
who made it?
it's an olimex AVR development board, AVR-P40-USB.
[16:56:25] <bill20r3> http://www.olimex.com/dev/avr-p40-usb-8535.html
and where's this smd part?
the little black thing between what looks to be a transistor, and resistor?
it's not in that picture, and not in the schematic.
(that schematic is out of date)
but I found the part number on another similar board that they make.
at the top-center of this: http://www.olimex.com/dev/images/pic-p40-usb-sch.gif
bill20r3: that FET is just for enabling the microcontroller once the USB negotiation has happened? If you don't care about obeying the USB spec, just unsolder it and short together the source and drain, or whatever they're called on a FET
except that if it blew up, there may be another problem (since USB can only source 500 mA, in theory)
the usb programming part still works, but the board doesnt work normally.
so I can burn code to a chip, but can't run it.
burn via USB, or via the 10-pin header?
powered from USB, or from the programming dongle/board?
it looks like 5V is supplied through the 10-pin header
the usb is just a serial port(connected to the avr), and power source.
the 5v comes from the usb.
the jtag dongle is optoisolated, so it gets power from the dev-board
pin 2 of ICSP is connected to +5V ...
ISP isnt' JTAG
if you plug it in to USB and measure the voltage at U1 pin 11 is it +5V? I suppose the FET could have failed as a short
it's a dual-purpose dongle, I referred tothe wrong purpose. :-)
but then you'd expect the whole thing to work...
I haven't put a meter to it yet.
the photo on the product page is wrong - it doesn't show the FT232 chip in the board (unless it's tiny and hdden by the USB connector)
yeah, it is wrong.
and the schematic for the avr version is wrong too, only the pic version of that board has FET1 on the schematic.
the ft232 chip is on the bottom
a Fet is *basicially* a transistor, right?
yes, it's a kind of transistor, but it has some special characteristics
ok. I'm just trying to work out it's purpose, in head.
its purpose is to turn on power to the rest of the circuit once USB negotiation has been performed and turn it off if the USB device is told to enter sleep mode
USB permits small current before negotiation (10mA?) and tiny current (in the range of nA) for most devices in sleep mode
ahh. so the avr wont start executing anything until the usb negotation is done?
that makes sense, especially if your code uses the rs232 port.
so the usb chip isn't switched, but the AVR is, and that 'switch' is broken.
well that doesn't explain why you'd still be able to reprogram the device
that's why I suggested you measure Vcc at the AVR, to determine if the FET failed open or shorted
yeah, I'll have to read up on what pins get used during the flashing.
* bill20r3 looks around for a usb cable.
about 3.7v on vcc, the datasheet says it wants 4.5-5.5
that could explain why it's kinda working
which avr is it?
the other way that you can accidentally power an AVR is through the I/O pins
they contain protection diodes that are internally connected to VCC,
so if you have a logic 1 on one of the pins of the ft232 that's connected to the AVR, that could be the source of this 3.7v you're measuring
in either case, I need to replace that fet
I think so
I do think you're likely to get away with replacing it with a jumper...
(see page 50 of http://atmel.com/dyn/resources/prod_documents/doc2466.pdf
if you didn't follow what I said about the diode to vcc)
as long as I'm not doing anything that uses the rs232 immediatly on boot?
it's true that you'd be sure to lose a power-on identification string or the like
I'm not using the rs232 at all anyway
then another option would be to power it from a wall-wart or somethingand not plug in the USB at all
thanks for your help w/ this, I gotta go eat.
hi guys.. finally home ;)
it's the start of a 3-day weekend here in the US
give me another 2 hours and it will be the start
well, true, it's 2 or more hours before the end of work .. but who's working?
not me - dave just woke me up emptying the trash
need more coffee so you're alive after work? ;)
err - after "work"
cradek: the secret is to not look too startled
and to rest on a napkin, so there's no drool when you sit up
I sure screwed it up then
though the napkin sticking to your cheek would be a dead giveaway
hmmm - should I get the $66 Athlon 64 FX3000, or the $669 Opteron 180?
Any of you guys know if its podssible to put a 20 Gig HD in a PII?
sure I think so
Cant get it to work. Tried 2 diff ones.
as long as you have the correct interface, which at the PII era was almost assuredly IDE (or SCSI)
what problem does it have?
I got the one to see the drive & start linux, but got some errors
It gets as far as loading modules, and stops with Alert - /dev/hdb1 does not exist, then drops to a shell.
you should configure it as a master drive
that should be hda1
then I get Cant access tty job control turned off.
what other IDE devices are in this machine?
pretty sure it is hdb1 but I can hook it back up & try it again.
none I think. Has a CD rom
that would be one ;)
and you're not connecting this in addition to another drive, right? (ie, you replaced the drive that was there with this one)
is it connected at the end of the cable, or at the middle connector?
ok. it should be at the end
also tried jumper setting for enable cable select as to master / slave.
the drive probably has a set of 3 jumper positions, labeled something like Master, Slave, Cable select (or MA SL CS)
only MA / master shuold be jumpered
And also jumpered as single drive, master, and master with non-ata slave (I think thats what it says)
Interesting thing, maybe this might help,
ok - single drive is proper then
unless the CD-ROM is connected to the same cable
no, different cable.
it should be slave, not master
I have to have a jumper on one setting to even see it on the PII, but I get that error. I get the same error on the PIII with that jumper setting, and I have to take all the jumpers off to work right on the PIII
not as the only drive on a chain, I think
Slave setting is no jumper, and the box doesnt see it at all that way.
The other P2 I have doesnt see it no matter what.
but module is looking for /dev/hdb1
it shuoldn't be - that's what made me think the drive wasa slave
Actually, SWPadnos I think that makes sense now.
Lets see, I installed linux with this drive in a PIII just as it came.
The drive had no jumpers. So that means it was set as slave at install.
So linux is looking for a slave on the P3, and the P2 needs a master. Hmm?
sort of, but if you put a master in there, it won't boot from the slave
the newer machine is probably a bit more forgiving of where the boot disk is
Can I change the linux sys to see it as a master and set the jumper for that? ie look for /dev/hda1?
davidf: several changes needed for that
erm - well. I suspect that's harder than reinstalling ;)
you need to update the bootmanager, then edit /etc/fstab and probably some other places too
the good news is that once you reinstall as hda, you'll be able to move the drive back and forth like we said ;)
initrd comes to mind as well
you can do it, but only with another running linux in place
SWPadnos: I don't think initrd needs to change
or boot it in the PIII, make the changes, then set it as master and hope nothing is screwed up
it only holds modules and such
Well if it does, then I reinstall anyway.
I was just hoping not to have to back up a lot of email, data, etc.
davidf: it's probably best if you do before fiddling with it
* alex_joni heads to bed early
what about the other drive you tried?
have a good weekend guys
see you Alex
Maybe I'll just buy another used drive & install a new linux / emc on that for the P2 & keep this as utility.
SWPadnos, not sure what drive you mean?
you said you tried two drives
Just the one drive back & forth between boxes. The orig P2 drive has win 98 I'd like to keep.
ah - ok. my misunderstanding
didn't you install ubuntu to some drive though?
yes, a 20 Gig Seagate. Thats the one we are discussing
so what needs to be backed up for you to do a reinstall on that drive (once it's jumpered as master)?
email is really the main thing I cant get over again.
and some G code files
The rest I can dl again.
I'd stick it in the PIII, and pull off whatever you need
it shouldn't matter that it stalls running emc ;)
&& then set it as master & reinstall?
yep. that's what I'd do (assuming I'm restricted to using hardware on hand ;) )
prety much the case right now.
Hate to have to go from a 733 P3 to a 466 P2 though.
Such is life.
Maybe I'll get brave and try partitioning the 6 gig on the windows box & put linux there.
Anyway thanks SWPadnos cradek et al
sure - good luck
if you use map command in grub config, maybe it'll work
it seems to start booting, but it's looking for /dev/hdb1 later in the boot (like at mount time or something)
with map command he can show master disk as /dev/hdb
see you Lerneaen_Hydra
I'll be here tomorrow
skunkworks: the userspace 'task' program does a little work and then sleeps for [TASK]CYCLE_TIME
acemi you still there?
so how hard would that be ?
I dont know my way around linux.
you have to boot with PIII and then you have to edit /boot/grub/menu.lst
Im on that box now.
did you open the file?
but if I mess up I better have stuff backed up I guess.
yes, menu.lst ?
back up your emc configs before you do much more
I have emails & other stuff to do too. That'l take awhile.
is it just a one liner, or what?
[20:28:57] <acemi> http://www.linuxquestions.org/questions/showthread.php?s=46d991790d933d189366259166a0382b&p=2391150#post2391150
there is an example here
my english is not good, i can explain easyly
-> i can't
So I really just need:
map (hd1) (hd0)
map (hd0) (hd1)
and that's all?
you must change the root() line too
(ahead of those 2.)
it'll be root(hd0,0)
because your disk will be master disk
but with map, you show it as slave
think I can do that. Better back up first. Worth a try. Thanks.
SWPadnos, Hate to bother you again, but I'm about to do something I better ask someone to look at first
# title Linux
# root (hd0,1)
# kernel /vmlinuz root=/dev/hda2 ro
Is that gonna break somthin?