hey why arent configs in /etc?
* fenn decides to shut up before he gets saddled with writing an xml preferences dialog
so does anyone have any input on how they would like python bindings for canonical commands to look?
or any other neat ideas along that path to keep in mind
So, how do I upgrade?
[00:07:27] <SWPadnos> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?UpdatingTo2.1
you should be able to unmark all changes (in case you've marked anything), then click "mark all upgrades"
good night all
new gcam is posted
oh that is awesome.
crepincdotcom: no, that's way too deep
cradek: i just pulled the board off the mill... yes, i concur that that is way too deep
* crepincdotcom throws away another pcb
so i have a 1mm endmill and a 60 degree V
what do you suggest.
connical is fine for cutting MCB's
thats what i hear
what software are you using to generate toolpaths?
crepincdotcom: cut as shallow as you can reliably cut. looks like I cut .0045 deep: http://unpy.net/cgi-bin/viewcvs.cgi/eagle/ulp/singlesided.gpar?rev=1.3
cradek and jepler's script for eagle
I've been looking at kicad
cradek: then just tell the scripts that my width is whatever i calculate for that depth?
my main issue is that i cannot reliably zero my z axis
crepincdotcom, why not?
got a multimeter?
crepincdotcom: feeler gauge?
if you have a multimeter you have a reliable zero
i have a multimeter, but no feeler gauge. iv been using a peice of paper that is 0.003" thick
any dial caliper?
Twingy: what praytell do i do with the multimeter?
crepincdotcom, set it to continuity, tada
one lead on copper clad, other on endmill
that's a good idea for PCBs
i cant beleive i didnt think of that
silly kids :)
^^ i admit
i am a complete milling noob
in the more general case, tinfoil works better than paper, and you can measure it with the calipers and get a pretty constant thickness
I was talking to some one about that a few months ago and the idea just popped into my head
cradek: ah ok
how's this paper or tinfoil used??
a-l-p-h-a, tear test
a-l-p-h-a: move it around until its stuck under the spindle
you lower until it bites
I wonder.... could you connect one wire to the endmill, another to the copper, and wire them up as a home switch?
then hit homeZ and stand back
if you are dealing with metals then continuity meter is much better
ooooooooooohhhhhhhhhhhhhhhhhhhhhhh.... so you don't mill too much fibreglass??
jmkasunich_: only if you have plenty of extra cutters and PCBs
jmkasunich_: i havent read through the docs to figure out how to read in home switches yet
cradek: why you say that? if the homing velocity is very slow its no worse than any other touch off method
* crepincdotcom goes to try
anyone know of a reason that the *.cmp files would not work with emc2?
It is a leadscrew compensation file. X.cmp for that axis ...
I think alex got that working
There was also a temperature compensation file way back in the old days. I don't even remember it's name.
rayh: page 33 of the users manual talks about compensation
there is no requirement that the file end in .cmp, that must have just been a convention that was used
rayh: what does "gang lathe" mean?
multiple spindles all making the same part at the same time?
you don't want to know
well, copies of the same part
seemed like it was something about tool changing?
a set of tools along the x axis slideway.
Each tool gets set to a different zero for rotation and length.
rather than an indexer that puts different tools into the same cutting position.
that really doesn't sound any different from using a QCTP like I have
cradek: are you the one that has the emco?
cradek: when Axis is in Pause mode after a tool change, it doesnt let me jog to reset the z home
is there a way around that?
crepincdotcom: either stop and use "run from line" to restart, or more foolproof is to put each tool's part of the program in a separate gcode file
also, is a 1mm endmill ok for the roughing? is still seems to be making traces uber small as it goes by on both sides
I'd try it without roughing until you get the separations part right
how can i tell it no roughing?
0 rouging passes
3 outline passes still sounds ok?
here goes nothing....
sure 3 is fine
.0045 is the depth not the width
but I thought depth and width were the same for V
* crepincdotcom solves the other direction
if the V is 45 degrees then yes
mine is 60
so the slope is tan(60)
unless im wrong, in which case im screwed
hello good people.
does anyone know about being able to put images into pyvcp.xml files?
03jepler 07v2_1_branch * 10emc2/debian/rules.in: pluto_servo was no longer getting copied to docs/html/ after other changes
I know I heard someone mention but I forgot who it was.
he has several changes to pyvcp that someone should evaluate and maybe put in CVS -- he's asked me but I haven't gotten around to it
they're in pastebin somewhere I Think
I can hang around and see if he shows up.
jepler: I got oneshot working, the change you suggested for width as a pin is working with pyvcp sliders.
cutting in progress
look better this time?
the multimeter touch off is awsome :-)
owhite: if youre milling a PCB, put your multimeter on continuity
then put one wire to the top of the board, and one to the mill
you know you are z=0 right when it beeps :-)
have you heard of anyone succeeding in making PCBs with a laser?
i havent personally, but im not really in the PCB making scene
PCB stock burns at a lower temp than copper melts
yeah I dont think you could cut it directly. I'm wondering if I could maybe cut an overlay serving as a mask. CO2 lasers dont do well with copper.
probably be much better off resurecting my 3-axis router.
cradek: it still seems to be cutting too close on both sides, there are a couple traces that are Vs pointing up but i dont think theres copper left on them
is it consistent all over the board?
maybe the board is bulging up in places
even if there are traces there, theyre definatly not 50-mil like Eagle had them
could it be a tool diameter vs tool radius thing?
you need to tell either eagle or emc what size tool you are using, right?
if they as for dia and you give radius, or vice versa, it would cut wrong
well, the scripts jepler and cradek wrote, yes
if they ask
ok here's what you do
cut one pass in a new board at whatever minimal depth you can reliably use
measure the width
thats damn tiny to measure
width of trace, or width of cut?
but ill try
width of cut
yeah it is a pain to measure, I use a microscope
that i lack
that number goes in that screen when you generate the gcode
write a test program (plain g-code), two cuts 0.050 apart, then 0.040, 0.030, etc
and see when the copper in between disappears
I can point you to a very cheap microscope (<$100) with optical marks to measure in the microns, if you need it.
ill check it out, sure
I use it to measure kerf cuts of laser beams.
let me look around.
"Outline Width (mm)" is the number you want to set with this width
owhite: point away
03jepler 07v2_1_branch * 10emc2/debian/control.in: emc2-dev is useless without the main package, so make it a dependency
03jepler 07TRUNK * 10emc2/debian/control.in: merge rev 184.108.40.206: make -dev depend on the main package
I bet you could also cut a long narrow "V" shape and measure where along it the middle point disappears
heck you could do that on every board after it's mounted, and generate the gcode to perfectly match your depth settings for that board
ejholmegren_ and crep, I'm looking. I checked package and its unmarked.
[02:04:13] <owhite> http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=170081900177
cheap, I know. but I've even used it to take pictures. http://www.nilno.com/laser_dir/alignment/650_hole2.jpg
that's done by just holding a camera over the thing.
cradek: i thought of a potential issue
does the script take care of offsetting for the tool, or does it look it up in the tool table for axis? i havent put any of my tools in the axis table
it does not use emc's tool table or compensation
how big is your router bit?
60 degree V
:-( well thats an issue
it was hitting the end of the y axis
but not knowing
so everything was offset like 0.2"
sigh.... why did i use such a tiny mill
cuz a big one won't fit on your bed?
you have a point there
i have my own lab space now
thats where i am at the moment
they gave me a whole section of the cnc lab
so i retrospect i could have gone bigger.... but i didnt know i'd get this
well, perhaps this summer.
oh thats helpful
jmkasunich_: on the 14th, you had a conversation with tomp which ended with ..."GoSloJymbo can pick it up and take it home" . What can I pick up and where do I get it?
seems to have broken the motor as well
I don't recall that conversation
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2007-02-16.txt
I have a hard time following it. It's about the DRO display and waitusr.
tomp said that, not me
what he meant was we got the bugs out, now you should be able to use it
are all emc irc sessions logged?
I understand tomp said it. I just thought this meant I could find a listing somewhere which would spell the solution out.
maybe, but I dunno where
tomp had the code, I was just helping him debug it
[02:25:23] <crepincdotcom> http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2007-02-16.txt
Dont know if you guys watch american chopper, they're using a 5 axis water jet on the show.
crepincdotcom: you can measure things pretty easily with a flatbed scanner and a ruler
or even a digital camera
i'm cheap tho and never had one that can focus for shit
ok, who actually has a machine running steppers?
my steppers have 12" pigtails, and I'm trying to come up with ideas for connectors. any suggestions? I need to feed the long cable thru energychain
how many amps
card edge connectors?
fenn: I think he wants wire to wire
i know db9 is common
I've been using crimp on .25" vlades, but they're a bit bulky.
but that's probably too bulky
Jymmmm: I found 16AWG x 4 conductor speaker wire at home depot
4 conductors in a jacket
jymmm mini din-9 like old keyboards
sherline uses midi/din connectors
dunno how flexible
jmkasunich_ they have shielded too, which is what I'm using, but I think it's 18ga
i just had to rummage around in cradek's brain for a sec to find the answer :)
I need 6.1A, so I was looking for the bigger stuff - they had 18 (and maybe 20) too
fenn mini din doens't have any locks though.
I just like the ability to swap motors around w/o havign to re-run the cabling.
anyone use the barrel crimp connectors?
the "quick disconnects"?
[02:40:59] <jmkasunich_> http://rocky.digikey.com/WebLib/Switchcraft/Web%20Photos/EN3C4M.jpg
jmkasunich_ OD ?
I got those - waterproof (dunno if oil proof), rated 7A or so, about 1/2" OD
jmkasunich_ cost per pair?
I got chassis mount receptacles and cord mount plugs, about $12 for a set
ah, not too bad
not cheap, unless you compare it to MS style connectors
they have cord mount receptacles too, so you can do a middle-of-the-cable connection if you want
for 3A, you probably have a lot more choices though
I've been using these: http://www.maplin.co.uk/images/300/42728i0.jpg
ah - fastons
never been fond of those except for maybe inside a cabined
but the kind that is insulated on both sides
not fond at all when polarity matters
much rather have a 4 pin connector
4 wires all colored coded. you really cant screw it up =)
crepincdotcom: yeah, I tried that wiki, and 1) it didn't update, or 2) waitusr as I applied it is not valid. I am trying to find out which one. I am still getting error 15:command not found.
jmkasunich_ yeah, but this is to connect to the 12" pigtail leads from the steppers. The other end of the cable I'm using a 4c locking connector... sec I'll get a pic.
what about mate-n-lock?
jmkasunich_: cable end http://www.bills2way.com/equip/images/4pin.jpg
and the associated bulkhead mount
housings are under a buck, pins are $0.35 each
do you know current rating?
those are cb mic connectors
sure ain't 6 ;-)
check out the white plastic connectors on this: http://nilno.com/laser_dir/laser_controller/108_0830lg.JPG
those are mate-n-lock
that's what I went with. came from digikey, I could dig up a part number.
those are molex connectors
hey, if you don't need environmental resistance, you can use hard disk power connectors
jmkasunich_ most dont' have locks on them though
I like them for panel mount, but I dont think they are that great in-line on cables.
they would fit through the eChain though
oh hell.... no female
those are molex too, in my opinion. I
great.... they have 2 6 9 and 12 female, but not 4
if you go to digikey and put "mate-n-lock" in their search box you'll find them
I have the housings, just really need the pins, and I guess I could buy the cheap crimper too (the one I have is for REALLY small connectors)
I just need to find my extrator tools.... haven't seen them in a while =)
what I do is solder the wire _and_ crimp to those pins, by the way.
jmkasunich_ nice idea on the molex connextors, will work on locking them together "somehow"
if you won't be unmating them often, a small tie wrap would do the trick
jymmm: they have in - line versions for cable, and also panel mount.
owhite I want the crimper so it'll strain relief the insulated part as well
oh yeah crimping is essential
iirc the cheap crimper is $40
how many connectors do you need?
tie wrap or shrink wrap might do it
if it was only a couple, those waterproof circular ones are nicer, and at $12 bucks per mated pair would be cheaper than the crimp tool
but for 4, might as well buy the tool and have it for the future
That's what I was thinking as I've always used molex connectors in one fashion or another.
another trick (okay this is really minor) is to tin the wire with solder, crimp it, keep it in the crimper, and heat with a solder to join to the pin. keeping the crimper in place is helpful because it sinks the heat away and you dont melt the wire insulation.
and the pwr connector will fit in/out the echain w/o cutting it off too
er, heat with a solder _iron_...
what is echain?
waste of money
its a conduit that only flexes in one direction
like a tank tread sorta
oh got it.
drat: Note that even though the 5I20s FPGA can tolerate 5V signal inputs, its outputs will not swing to 5V. The outputs are push pull CMOS outputs that will drive to the output supply rail of 3.3V."
any gecko users around?
jmkasunich_, sort of, but not stepper drives
these are steppers
I'm seeing flickering of the power LED under load
when the stepper is energized but not moving (or moving slowly) the light is bright, and the supply is 55.25V
hmmm. do you have a charge pump or other monitor connected to the disable input?
when I run it as 12 revs/sec, the light gets dimmer, even tho the supply only drops by 0.5V
these drives don't have a disable input
I thought they did
that must be the 212
which is strange, since on 2/12 Mariss said on the geckodrive channel that they can be disabled
I'm writing a post to ask him
channel = the yahoo group
yeah - there was a nearly identical question inthe last few days - a guy with a PMDX-<something> with "noise problems"
pin 7 is labeled NC on these drives, and the manual doesn't mention a disable
though I could be confusing two recent issues
there was a thread about disableing drives for manual cranking
I remember that one also
the motor is a nice handwarmer
dang - I'm losing some steps too
about 3 microsteps per rev it seems
jmkasunich_ mechanical or electrical?
is there something other then keystick that will work on a 320x200 screen?
you could hack tkemc
320x200 is damn small
or pyvcp maybe
might be a bit lacking
could Ubuntu 6.06 be installed from the iso image residing on a hard drive in say a redhat machine? (old laptop w/o floppy, and cant boot from cd)
320x200 fits on a cheap LCD
tomp: mount the iso with -o loop
actually the best way would be to use pxe
[08:29:44] <fenn> http://myy.helia.fi/~karte/ubuntu_pxe.html
fenn: thanks, ah, the laptop is brain dead right now, no network... before i asked this Q i pulled the hd out and put it into a system that could boot cd. but soon (too late) realized it was probing all the wrong hdwr :(
fenn: nice trick tho, i read it
Does EMC2.1.0 have pyvcp installed in it? I have had to use head to run pyvcp and hal.
Goslowjimbo: pyvcp should be available in 2.1
it might not be 100% complete as in HEAD, but it should work
Well, I have tried to update head according to the wiki, and I still couldn't run "waitusr". It gave me a command not foud, error 15
command not found sorry
[09:54:15] <Goslowjimbo> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?UpdatingTo2.1
You doing any good today?
im making a nametag holder for the frontdoor at work..
to try out the dnc function of the big mill..
Did some testing of backlash comp yesterday.
Looks like it would work well enough for a machine I'm working on.
cool.. that's always good news :)
as usual we implement these things without big iron to test on ..
I asked about the leadscrew comp stuff and someone said you had it working?
well.. theoretically working
I worked on the userlevel part, and getting the stuff down to motion controller
Okay. I don't have any scales fine enough to set one up but someone asked.
and map each centimeter :)
How accurate do you figure we could get with a pot.
I'd say something around .1 ohm :P
well, pretty damn accurate :9
Got a question about a file from long ago.
But can't seem to find the cvs stuff at sourceforge.
all the CVS history, including from before the move to our own server, should be at http://cvs.linuxcnc.org/cvs/
(you can also check it out from that server, of course)
I thought so but I seem to remember a temperature compensation file.
rayh: it wasn't a file afaik
there was some stuff inside the motion controller called emcmotalter
or something like that
and that alter has been used for temp. compensation
" The alter interface is intended to support dynamic compensations of small
amounts based on a sensor-based machine model, for example from temperature.
alter does ring a bell
but you need an outside process to send the changes
alter by itself doesn't really do anything, except applying the offset
so you probably want a simulated model of the axis, and send the changes to the motion controller, once every few msec/secs
This stuff was way back when the repository was on WillS's sun workstation.
we don't have alter in emc2, but there is a different motor offset which can do just the same thing
What would that look like.
jepler: do you remember without looking? I think you used that for the stored position (from POSITION_FILE)
alex_joni: no, I don't remember.
* alex_joni looks
Back then we burried a thermocouple in the machine base, read it with some sort of process controller that sent a equal interval variable to alter.
right.. we probably want the same thing for emc2
only the NML call will be a different one
EMCMOT_SET_MOTOR_OFFSET <- that one goes to the motion controller
looking for a userlevel interface
You mean the temp was passed using NML???
I suspect so
this was during the run (if I understand it correctly)
ok, I just checked .. there is a function called emcAxisSetMotorOffset()
that one gets the data down to the motion controller, right now there is no NML hooks to do that from a GUI or another process, but I could add that easily
right now it can only load the compensation table a single time, there's no provision on the realtime side to do anything but append a new entry to the (fixed-size, ascending-order-only) compensation table
The use we made of it back then was to compensate for temperature stretch of leadscrews.
jepler: right, but if there's a process running every few seconds that affects the motor offset, then I think you have the same behaviour one had with alter
we got the K&T to set the pallet off onto the changer. Now we can start digging into it.
That's great skunkworks. Someday I've got to drive down there and see it.
I've a question
:) I should be interesting - it spins the z axis nut instead of the lead screw. Kinkda makes it a bit harder.
anyone knows what is an "XclientMessage"?
Function XClientMessage sends a message to a window with XSendEvent. The message is initialized with a particular protocol type and atom.
The format of the XClientMessage function is:
alex_joni: yes if you need single-entry compensation you can probably use EMCMOT_SET_BACKLASH without modifying the realtime code
but I can't imagine that thermal expansion is corrected well with a single backlash-like value
alex, can I ask a question? I would like to know if it is possible to do a thing
xemet: don't ask whether you may ask a question. just ask the question.
I've got this at university:
[13:54:12] <xemet> http://www.mindflux.com.au/products/spacetec/4000flx.html
it is an old spaceball, serial, not USB
I would like to use it to move the CNC
there is a driver for linux,, 3dxware
I've installed it and it works
there is a demoprogram that opens a little window and shows the axes coords
there seems to be a linux joystick driver for "spaceball" in the kernel, and the comments indicate it includes 4000flx.
if that is true, then configure the spaceball to work with the linux joystick driver, then use hal_joystick
I've contacted 3dconnexion asking how this program work
How can I configure the spaceball in order to use the linux joystick driver?
I don't know
I bet the Internet knows
[13:57:37] <jepler> http://www.gelato.unsw.edu.au/lxr/source/drivers/input/joystick/spaceball.c
how to use it?
I don't know
I bet the Internet nkows
[14:01:01] <jepler> http://www.charmed.com/txt/joystick.txt
scroll down to "inputattach"
so...If I succeed in configuring the spaceball in order to work with this driver...it will be seen as a joystick?
that's my impression from what I've just learned by searching the internet
I will search a lot!
It seems to me that temperature compensation should in effect change the INPUT_SCALE.
you can easily change the HAL parameters stepgen.X.position-scale or encoder.X.position-scale while emc is running but I don't know how the machine will behave if you actually do this
'inputattach' seems to come from the 'joystick' package on ubuntu
I've done it and it causes an instant following error.
Unless that number happens to be zero.
I think the only safe time to change that would be while sitting at joint=0
jepler: I've not understood a thing...I've to download and install this driver or it is already available in ubuntu?
yeah -- say you were at 1000 counts and you change position-scale from 1000 to 1001. Your position should change from 1000/1000 = 1.00 to 1000/1001 = 0.999
(if you're homed, I think that's at the home position)
xemet: the kernel driver (spaceball.ko) seems to be in the standard ubuntu kernel package and in our realtime kernel package
xemet: the "inputattach" program is in the "joystick" package which does not seem to be installed by default on ubuntu.
I will see if it is in the synaptick packages list
...ok, I've installed the joystick package...but If I type inputattach it says No such file or directory
ah, and there is a package named inputattach, already installed...
well...just realized, inputattach is present and works
the problem is the path of the serial port...
ok, found. /dev/ttyS0
uhm...well, if I run input attach, the spaceball make two bips.
but...if I use hal_joystick /dev/ttyS0 it says, device is not a joystick
and, if I try to calibrate it, it is not shown...
joysticks don't plug into serial ports
cradek: this device does
ok ignore me then
cradek: the docs I read lead me to believe inputattach hooks the serial port to the joystick driver inside the kernel
it is an old spaceball
after that /dev/input/<mumble> is supposed to work
look in dmesg after running inputattach in the background
+ r = c->args_func(get_cldat(interp), interp, argc, argv);
what is dmesg?
dmesg is a command that shows the kernel log
what have I to look for?
running dmesg before and after the inputattach
I don't know exactly, I hoped the last few lines would say whether the joystick was attached or not
but if it beeps, cleraly linux is communicating with it
the only thing that seems changed is at the ende serio: Serial Port ttyS0
the documentation seems to indicate you should be able to 'jstest /dev/input/js0' after doing the 'inputattach ... &'
if that does *not* work then your problem is still with the driver itself, not with emc's hal_joystick
no...js0 is not present
is "spaceball" in the list of modules shown by the "lsmod" command?
and it says Used by 0
are you running inputattach in the background (with a & at the end)?
no joydev is not present
try "sudo modprobe joydev"
ah...wait, I've typed inputattach --spaceball /dev/ttyS0
after that, tha spaceball beeps
but in the terminal
there is no prompt anymore
in unix, putting "&" at the end of a shell command runs it 'in the background', returning you to the shell prompt while the program is still running
I opened another terminal windows...
but I think the program is still running...
It is probably important that 'inputattach' keeps running during the whole time you want to use the spaceball
did you try the 'modprobe joydev' yet?
I've done it now
did that make /dev/input/js0 exist?
try restarting inputattach now that you've loaded joydev
I've to stop the one started before...
how to do that?
pressing ctrl-c will probably stop it
restarted but no js0
look at the files listed in /dev/input before and after you run 'inputattach'. see if anything is being created.
inputattach *is* communicating with your spaceball -- on my system without one, it prints after a few seconds: inputattach: device initialization failed
he is communicating...
but I don't know where is the device...I should be able to calibrate it for example using jscal
[15:09:39] <xemet> http://www.gelato.unsw.edu.au/lxr/source/Documentation/input/joystick.txt
read the point 2.2
do you think I have to do something like that?
you can try that, but ubuntu uses udev which is like devfs, and usually makes that unnecessary
did it make any difference?
but...I've to create nodes 0 1 2 3 or just 0?
probably only 0 until you have more than one joystick device you want to use
js0 is present now, but if I run input attach, and after jstest
it says No such device
while inputattach is running, does this command show any entries for your device? grep "N:" /proc/bus/input/devices
keyboard mouse and speaker
I'm out of guesses on this one
ok...I think it doesn't work...
maybe I will study it better later...
thank you very much for you supprt
I've to learn a lot
awallin, are you getting alot of spam lately on WP?
I've got akismet, and Spam Karma 2 running.
hi a-l-p-h-a, yes, two spam komments got through today. normally I don't get anything. something wrong with akismet lately?
dunno what... but yeah. something's wonky... I put spam karma 2 running now again.
I couldn't DELETE the damn spam either... something was buggy
[17:50:53] <a-l-p-h-a> http://unknowngenius.com/blog/wordpress/spam-karma/
FYI.... Molex 4pin (hdd power connector) ratings: 250V 22ga=6a, 18ga=8a, 16ga=10a, 14ga=10a
there's a way to switch root devices, and there's a way to mount iso images ( i have scripts for this ). is there a way to mount a 'live-cd' and install from it?
i have a laptop that cant boot cd and has no floppy
what os is currently on it?
but i can pull the hd out and load an iso image as a file, or as a partition
right now , nada, had damnsmalllinux last nite
Jymmmm: i can put damnsmalllinux on it, or dos... what might you suggest?
not sure, sounds reasonable, but if it's mounted, not sure how it can install
is there a way to issue g code to emc from the command line?
can you boot from usb?
Jymmmm: no usb, but might be able to put image on a cf in pcmcia slot when using damnsmalllinux
how big is hdd?
2.1 g :(
what I had in mind will take almost a gig
got 1g cf
eholmgren: dont know from cmd line. will mdi help?
eholmgren: yes, using mdi (a python script)
mdi -ini /path/to/ini 'command'
tomp: Just a thought (it worked really nice for me on a usb stick) http://www.knoppix.net/wiki/Bootable_USB_Key
tomp maybe alex_joni has an idea
though, I'm no longer allowed to root boxes at work with it now =)
wait, maybe that's not the one I used, hold on
alex_joni: is mdi in the wiki?
not sure, but there should be a manpage
tomp Ah, here ya go.... you can't get easier than this.... http://pendrivelinux.com/2007/01/01/usb-knoppix-510/
tomp THAT's the one I used.... large DL, but straight forward setup
tomp I did that from XP - works great
Jymmm: and i thought the 1st info was great ;)
didn't tomp say that he has no USB?
or is that "no USB stick", not "no USB ports" ?
SWPadnos: one can work around both )
harder to work around the "no USB port" thing for OS installation :)
well.. there are PCI ones
although I suspect the BIOS can't boot from those
PCI doesn't work so well on a laptop
SWPadnos: right, no usb, thinking maybe i can use pcmcia cf chip... the trick seems to be... i cant boot to the image, how to chgroot or something, to use it after booted on sacrificial system
tomp, can you boot from CF at all?
no, only from hd, no net not cd , not cf
then you have a problem.
you can get a 2.5" to 3.5" HD adapter, and stick a bootable image on the HD in a desktop machine
oh, i can move hd to another system to install stuff ( how i did dsl damnsmalllinux )
yes using that adapter
how did you install something on that laptop?
moved hd to another box, removed all other hd, booted cd on that box, into the laptop hd
moved hd back to laptop
does the laptop have a CD-ROM that it can't boot from?
btw: thanks guys
SWPadnos, right, laptop toshiba 730cdt, cannot boot from cd, cannot boot from floppy ( no got hdwr)
ok, but it has a CD-ROM - it just won't boot from it???
tomp does it have a PCMCIA slot?
ah. then stick the HD in another machine, install grub to it, and manually edit the boot line to read from HDD (or whatever the CD-ROM comes up as)
it'll boot grub from the HD, but you can tell GRUB to boot the OS on the CD
Jymmm: yes and i hacve 1g cf chip w adapter
if you want a somewhat easier (at boot time) way, then you can use syslinux instead
SWPadnos: that sounds great!.... lemme think
tomp it SHOULD be able to boot from the PCMCIA slot.... format the CF card with DOS as a test
Jymmmm: it doesnt know cf exists w/o loading a driver ( i think ) but thats an easy test
CF is the same electrically as IDE
CF is the same electrically as IDE HDD
CF has all of IDE built in, but the interface bus is often not the same (PCMCIA vs. PCI), so the machine may not boot from a PCMCIA HD (or CF)
SWPadnos If it's old, it might need card and socket driver
there are no drivers at boot time ...
but I do know that some pcmcia hdd's would boot up on the older stuff.
SWPadnos: so i just edit menu.lst, adding a paragraph/stanza for the cd ?
Jymmmm: I can try
tomp, I'm not sure what to add, but grub also provides a "command line" at boot time - you can get device lists, etc (it even has tab completion and some help)
so you can just get into grub, then poke around until it works :)
Jymmm: the bios screen is the ancient box text type, not the type where 2 colums of choices
tomp: format z /u/s/v:CFCARD
tomp: format z: /u/s/v:CFCARD replacing Z with the real drive letter =)
SWPadnos: will try something b4 I pester again: thanks all ! ( i have used the grub cmd line to fix stuff )
have fun :)
Jymmm: what format should the cf be? ext2/3 fat/fat32 ?
tomp you might buy an ide2cf adapter, then you can just buy a 4gb cf card =)
tomp, you'll probably have some issues with such a small HD. I think 4G (or more, waaaaaay more :) ) is highly recommended
swpadnos also thinks 30" dual monitors are recommended
SWPadnos: Works nicely on a 1GB usb stick... http://pendrivelinux.com/2007/01/01/usb-knoppix-510/
dsl works on a 64MB i think
but you already have dsl so nevermind
yeah - I guess the real question is which OS tomp wants to end up with
RH3 should fit as well ;)
SWPadnos you should slap yourself for that one
anyone speaks python in here?
SSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSS SSSSS SSSSSS
SWPadnos got the eChain mounted (did inside on the X axis), gotta buy some split look tubing and some molex pins (and maybe crimper)
alex_joni: wellcome back
Jymmm: because its so easy to try, i got the cf chip empty and mount says its vfat, is that same as fat32? i can mount the .iso image and copy the contents to the cf. I see teh bios allows booting the 2nd hd, whatever/wherever that might be
ok, vfat not same as fat32, but the format of the cf can be changed ;)
vfat encompasses fat32 and long file names
but can't vfat also indicate fat16 + long file names
you can use vfat for FAT12, FAT16, and FAT32 filesystems, I think
I'm sure vfat can do any combination, regardless of whether Windows may have ever used it :)
since FAT12 is floppies, I assume it did
k, thanks, was just about to format the cf ! will just copy contents of the mounted iso to it ( ubunutu 6.06 alternate for mem < 192 meg )
tomp: what they said.
Jymmmm is now known as Jymmm
hi, the HAL docs suggest using 'dot' in pin names. The utility Comp doesn't seem to let you use dots
comp adds dots where appropriate, I think
it used underscores, not dots
the name you use in comp has to be a valid C identifier
well, you should get soemthing like comp.0.pin_name
for ex if I say device.something, it gives an error
right - a dot is a structure member reference in C
if I say device_somthing it converts to device-somthing
So I guess there is no way woth comp to follow HAL docs guidelines
DavidMTL, that may be so. what are you trying to do?
Jymmm is now known as Jymmmmmm
use comp to create a .ko
for my device
I think the dot suggestion is for separating instances or functions from each other, not something to be used inside a name
I can write it in C but it defeats the purpose of comp
for example, parport.0.pin-08.out
there are several things that comp doesn't support - I think arrays are one of them
I will write in C then
note the dots in that parport pin: parport (module name) 0 (instance) pin-o8 (sub-device) out (function)
sort of :)
Isn't it called: parport.0.pin-08-out
in comp you'd create this with: pin out bit pin_08_out;
could be - I'm running from memory (and possibly imagination :) ) )here
I ws looking at various other hal files and saw dots all over that place in some of them
The usual structure is: <driver>.<instance>.<name>
in the .comp file 'pin' declrations, you're declaring <name> only
pier: removing acpi worked?
or <driver>.<instance>.<function>.<instance>.<name>.<instance>, though I'm not sure there are any that have all 3 instance numbers
ppmc does have the other 5 though: ppmc.0.encoder.0.phase-a (or similar)
yeah -- that you can't do in comp
hi eholmgren, yes indeen
good to hear
so in the driver for pluto-servo I used <driver>.<function>-<instance>-<name>
can't even remember where I read that, but I figured it'd be worth a shot :)
DavidMTL, which document (and roughly where) is the comment about using dots in HAL names?
then I should use stepgen_00_enable and it gets converted to arcnc100.0.stepgen-00-enable
eholmgren: emc runs beautifully now
IMHO this is perfectly acceptable
it's acceptable, but may not save too much time for something that has dozens of functions (since you have to explicitly name each, rather than using a loop to create pins that point to an array, for example)
SWPadnos: I will try to find it. But I may be confused from what I saw in existing hal devices like ppmc
you should look at the section on canonical device interfaces - those are what you should provide in your driver
unfortunately comp hurts as much as it helps when you want to have N identical functions in one instance (e.g., 4 step generators)
what you really want is some kind of array syntax, but it's not currently possible
eholmgren: now I am trying to learn a bit more about RT before daring to read emc source code
swp mentioned that earlier
so you'll have something like arcnc100.0.encoder.0.<various pins and params>
jepler: still having no luck with the spaceball, but something incredible happens...if I use inputattch --spaceorb instead of --spaceball...then my spaceball beeps and js0 is created in /dev/input/...but of course i do not have a spaceorb so if I try the calibration every axis remain always to 0...
and then arcnc100.0.pwm.0.<PWM interface pins>
At this point I think that therei is a bug in the spaceball module or my spaceball is not supported...
(actually, a PWM is treated as a DAC, so you'd implement the DAC interface for that)
eholmgren: I am playing around with my first c..p RT module doing a ramp with the speaker
xemet: that's too bad
eholmgren: but there is always a bit of an error with the frequency I get in the steady part of the ramp... 1506Hz, 1530Hz, etc, never 1500Hz straight :(
it's incredible...js0 is created also if I try all the other devices...mouse, joysticks etc...SpaceBall is the only that doesn't create js0
I'm not lucky...
so create js0 and be done with it
fenn: doesn't help, you get ENODEV
no it doesn't work...
ah, when I use the --spaceorb mode instead of --spaceball, a beautifull spaceorb fully functional is shown by my device manager...
of course when I use --spaceball the device manager shows only a new serial but no specific device
a-l-p-h-a/jepler here's your new wordpress theme http://themes.wordpress.net/testrun/?wptheme=1418
I've contacted the author of joystick driver Vojtech Pavlik...maybe he will answer my question
and I've contacted also the 3Dconnexion team...maybe they know the problem...
good night all
haha, that theme is great
03jmkasunich 07TRUNK * 10emc2/src/hal/drivers/hal_m5i20.c: hal bits are considered true for any non-zero value, not just 1
i've got a question about stepper motors
i've figured out how to determine which wires are for which coil
but i don't know how to determine the polarity
there is no polarity inherent in the motor
03jmkasunich 07v2_1_branch * 10emc2/src/hal/drivers/hal_m5i20.c: backport: hal bits are considered true for any non-zero value, not just 1
I think the person I'm looking for is tomp, but does anyone here know about inserting images into pyvcp.xml files? :-)
Bo^Dick it doesnt really matter ... if it goes backwartds, just reverse one coil
the only time it does matter is when trying to series/ // an 8 wire motor
then its easy ... try it from a PSU. if it locks then you have them in phase, if it doesnt lock the motor, its outta pahse
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2007-02-16.txt