alex_joni: around ?
ok I see that when one makes that change t;he system must be rebooted before you can run apps
hey there. quick question: why is the gcode line incorrect: N106 G0 G28 G91 Z0.
error "cannot use 2 gcodes that both use axis values"
cannot g0 with g28
two g commands in one line
must be g0 OR g28
ok. it comes out of a fanuc post in Mastercam
can you post lines around the error one ?
N104 G0 G17 G40 G80 G90 G94 G98
N106 G28 G0 G91 Z0.
N108 G28 g0 X0. Y0.
N110 ( 3/8 DRILL TOOL - 1 DIA. OFF. - 0 LEN. - 0 DIA. - .375 )
actually, G0 comes in front of G28 (my editing is moving things around a bit)
can you pastebin.org whole file ?
sure, just a sec...
[00:27:40] <Roguish_> http://pastebin.com/d49836bca
G0 and G28 are both motion-producing codes, so I'm surprised MasterCam would output them both on one line
unless their software treats that as some special case (like G28 needs to know a speed and G0 is shorthand for "really fast")
SWPadnos: It is very possible
my solution is to remove G0 from line N106 and N108
it's also kind of silly to have G90 on N104, then G91 at N106 ...
sounds good to me. this is from a 'generic post. i'm doing a bit of editting in the post.
working on the 5axis setup.
ah - cool
SWPadnos: Bystronic lasers have G0 before everything too
long ago I wrote post to them :)
another silly (dumb) question. in the 5axis demo, is(are) the B and C axes 'nutating'?
I think :)
i think it would be described as a 'tilting head' but i'm not sure.
they aren't "nodal", as they'd say in the film industry
well, since you can change the angle, it's not like a nutation thruster
check this: http://numeryx.com/cnc/index.htm
i think it is an X Y Z C B setup.
with a virtual 'W'
anyone still interested of my DXF post to EMC ?
[00:38:46] <micges> http://www.emc-files.webpark.pl/import.tar.gz
I hope somebody looks at it. (I can't at the moment, and have no DXFs to try it with anyway :( )
my import generate files 0.ngc to (layers_count).ngc in /tmp
oh and its still alpha ;P
micges: is there a way to get english?
I got it to build and run :-)
Błąd: /home/chris/ARC-TEST.DXF: brak warstwy A
I get this (and nothing in the preview) with a dxf containing polylines made of lines+arcs
do you want the dxf file?
[01:00:35] <cradek> http://timeguy.com/cradek-files/emc/ARC-TEST.DXF
I have olny dxfs form artioscad, autocad and qcad
this is from autocad
and where is header ?
this is exactly how autocad wrote it
my wrote about 200 lines of initialisation ..
oh I selected Entities
this is autocad R12
must be because I selected Entities with dxfout
select all if you can
I will fix to allow such files too
that fixes it
but many of the arcs are wrong
wrong direction maybe
[01:05:24] <cradek> http://timeguy.com/cradek-files/emc/ARC-TEST.DXF
^ new file with header
can you read it with autocad? I can make a screenshot if not, so you can see the correct arcs
[01:08:20] <cradek> http://timeguy.com/cradek-files/emc/arc-test.png
Hi guys. I'm having problems with Emc2... it just won't start anymore. I rebooted and upgraded and I'm not having any luck. I get the following message: http://pastebin.ca/876745.
I'd be very happy to get pointed in the right direction. I didn't touch the system much, but I was playing with python scripts when I noticed the problem. Any help would be great.
try load to emc, maybe only preview is wrong
micges: I cannot figure out how to run it because of the language - can you say how
do you all know how I can remove this mod....
insmod: error inserting '/usr/realtime-2.6.15-magma/modules/rtai_hal.ko': -1 File exists
then push button "zaladuj"
window goes away
file will be in /tmp/*.ngc
owhite: try to run emc2 again
1.ngc starts with a line "0"
Hugomatic: did you create a file called "gcode.py" somewhere, like in the folder with your inifile? If so, remove or rename it and the files gcode.pyc and gcode.pyo if they exist
and ends with line G2 ...
jepler, its exactly what I did. thanks
acemi: still complains that rtai_hal.ko exists.
I added %. It is also missing feed rate
Hugomatic: instead of finding the "usual" module called gcode, axis found the file gcode.py when axis asked for the 'gcode' module
cradek: its pure gfile
owhite: you can see loaded module with this command: sudo lsmod | head
now it loads but the arcs are wrong
to remove a module: sudo rmmod module_name
acemi: I saw your recent change on the wiki. libpth-dev should only be necessary when compiling with --enable-simulator (userspace fake realtime), not when building for realtime and hardware control.
I added % ... F99 ... %
micges: I think you are using the wrong bulge for adjacent arcs
jepler, I'll whip up another name. I'm trying to make my own set of routines in Python. If I may, any preference between a "print myroutine()" over a "myreoutine()" that does the print inside?
owhite: try 'halrun -U' to try to force the realtime environment to unload.
I will check arcs and add some header
micges: this test file is one I made to test Realize :-)
Hugomatic: that's a matter of personal preference or style. I do some of both (and also a third option: pass a file into myroutine() and .write() or print>> to it)
how about > halrun unloadrt
micges: you are right the preview and export are different! both are wrong though
owhite: halrun probably won't start if it gets an error inserting rtai_hal .. (unless you give -U which is meant to run when the realtime system is already (partly) loaded)
halrun -U is like running the following sequence of commands: halcmd -R; halcmd stop; halcmd unload all; $REALTIME stop
cradek: I know that can happen
(break halcmd lock; stop rtapi threads; unload all components; unload realtime support)
ok Ill fix this and tomorrow second test :)
tell me if I can make more test files for you, I'm happy to help
I'm excited about your program
it is Free software, and people will find it useful
I hope so :)
lots of people ask about dxf-gcode conversion
anyone have any suggestions for testing my new M5I20 card, if I dont have anything connected to it yet? Like can I use pyiotest.hal to turn the LEDs on the M5i20 board?
ok can you make mm unit of arc-test ?
will be faster to me fixit
I don't know how to change units - I draw to the scale I want and disregard the units
can you tell me how?
[01:26:55] <cradek> http://timeguy.com/cradek-files/emc/units.png
this is all I have and I don't think it matters at all
cradek: what happened to your screen?
its all black.... did you burn it out or something?
cradek: have autocad opened ?
create new draw
type UNITS MM in command line
open test arc
and copy to the new mm unit drawing
the screen units.png is what I get when I tupe UNITS
I cannot type 'units mm'
cradek: has "Autocad BC"
I don't understand why it matters, the drawing's scale is not important to getting the arc directions right
yes but here is 2:30 night :D
pretend it is already in MM but the arcs are very small :-)
goodnight! tomorrow is soon enough.
especially soon for you :-)
I know but I usually stay long
best ideas :P
jepler: I tested the puma interface in simulator mode and I got no error. when I tested this interface in realtime mode in the same machine some time X crash. but I can't repeat this error. sometime it run without problem
I got EMC running again, and its 2.2.3... and it looks like the tool compensation bug was fixed. :-)
jmkasunich: I think I could make fun of your prehistoric cad software too!
Hugomatic: yep thanks for that report
I suspect that the unexcepted realtime delay causes this crash
cradek: you wanted my shipping address for you wwv clock?
you wanted my address for your large check?
cradek: you mean easycad?
its far from prehistoric
which clock do you like?
it has a GUI and everything
you guys have any idea if I can load up other .bit files on the M5i20? There are some precompiled files like hostmot5_4.bit, but I'd like to load up a config with more I/O.
fuck. just fuck.
jmkasunich: with my autocad the gui (menus, icons) are on the tablet - I have one, but no permanent "workstation" to use it - so I use the mouse tablet-emulation
I just thought it funny to see the old white monospace on black
yeah it looks a bit dated, but it still works as well as ever
qemu+freedos is really great for this
toast_school is now known as toastydeath
lesson from the weekend - a sharp carbide cutter + dawn/water mix cuts acrylic really nice
dawn the soap?
we used windex
applied by a dollar store marinade injector
probably more expensive ;)
both have detergent in them.
nice broken up chips of consistant size
windex is mostly alcohol and soap.
twice2: oh I like that one too :-)
probably my favorite clock - it has been running constantly since I made it
another lesson - strap clamps + blocks is extremely useful on tiny machines when you are operating near the machine's movement envelope
ds2: that's good advice for any machine
toastydeath: I have never needed that on a real bridge port. the 6inch kurt served all my needs
we do it very frequently
I think it says 2002 on it so that's 6 years - no sign of age (darkening) on the tubes yet
most of what I do fits well on a 6inch kurt
I probably clamp stuff down more often than I use the vize
jmkasunich: what made you use the clamps over the vise?
we held a bunch of plexiglass down with rubber cement.
flexibility and ability to hold large or odd-shaped parts more rigidly
iirc it was patterns for a pin router.
but doesn't the clamps get in the way?
you clamp according to the cuts you have in mind
skunkworks: how did you remove rubber cement w/o damaging the plexi?
yes they get in the way :-)
I think the plexi still had the protective paper on it.
so the clamps may have to move in mid work?
but, so does a vise
[01:52:22] <jmkasunich> http://www.metalworking.com/dropbox/_2001_retired_files/Facemil3.jpg
cutting from the side, clamps on top
cutting all around, bolt in center: http://jmkasunich.com/pics/spindle-encoder-mount-1832.jpg
that looks a lot more then 1/4HP ;)
aluminum and a sharp shell mill.... sex
I even use clamps for lathe tools: http://jmkasunich.com/pics/threading-tool-1833.jpg
owhite, at the moment there's no other 5i20 config to load
there's a lot of work a kurt juest doesn't hold.
that lathe tool mount cracks me up
so you are suppose to keep the stud closer to the work then the step blocks?
owhite, but you can test the card if you have the write-misc (or misc_write or whatever) function in a thread, and you change the LED signal(s)
its a lever
most of the force goes to whatever the stud is closer to
owhite, those are rhe 8 LEDs onthe card, so you have to be looking at the PCI 5i20 itself to see them change
hmmm and the studs are suppose to just long enough to accomodate the nut?
... are the ...
sometimes I use a vise and clamps: http://www.metalworking.com/dropbox/_2001_retired_files/Worm09.jpg
SWPadnos: got it.
(and a live center ;-)
no, they just need to be short enough for your tools not to hit
but the strap clamp kits come with a bunch of different sizes.
ah. i need to get some shorter studs
I have 4 each of 2", 3", 4", 6" and 8" studs
crashed into the studs once last night :(
jmkasunich: home made hobbin mill?
8" heh... that's funny 8" is almost my Z travel!
SWPadnos: would you know how many total I/O I get with the current 5i20 config?
that's a homemade facemill?
my "kit" came with 2 lengths of studs - too long and too short :(
owhite, 32 inputs and 16 outputs, I think
or the other way around
where I'm trimming the edges of a stack of plates?
jmkasunich: it is some import kit that was design for a rotary table (I have a puny mill, std sizes don't work for me)
jmkasunich: how did you time it all?
just need to find some 6mm all thread :/
I use 3/8-16, instead of the more common half-inch
(spindle to gear)
I use 6mm or 10-32
skunkworks: I gashed it first (rough cut the teeth on a dividing head)
ds2: can you just make some 1/4-20 T-nuts?
ah - so it already had some teeth
1/4 = 6.35mm
cradek: that's the other option... any reason you suggesting 1/4-20?
just because you will not find more common all-thread
and bigger is better if it fits
i bet 1/4 is much stronger than #10
the mill came with 10-32 accessories, the kit is 6mm so it would be nice not to have a 3rd size of stuff
3/8-16 is the 2nd most common, but probably too big for your machine
1/4 barely fits the tslots
maxnc is 1/4-20. you could get a kit made for that
yeah same with maxnc
really? interesting. taig says to use 10-32
[02:02:07] <cradek> http://www.positiveflow.com/maxaccess.htm
lol, what a rip
well it would take a day+ to make
heh, I can get 1/4-20 threaded rod in 10ft lengths for about $3
just need stock for the tnuts
much longer on a maxnc :-)
I just made a bunch - wish I had known you needed some
I think I made 6 - they are nice ones
those look like regular nuts, thought they were suppose to be spherical?
lol, thats funny.... "SLABBER/SURFACER" 1-1/4" diameter
you can get both
it is easy enough, just need to do it. trying to finish something by a deadline
flat nuts are probably good enough
"heavy duty milling vise"
1-1/4 is **big**
the mill with 10-32 aluminum t-nuts
cradek: did you mill off the tnuts in a strip or just band saw them off afterwards?
I'd saw them off
I used the bandsaw to separate them, but all parts are milled to size
you're better off with a 1-1/4" insert mill
cheaper to "resharpen"
by the time you mill off 3-4 nuts, you've turned at least one nut into chips
sure, but who cares
SWPadnos: are you there?
yes, but milling off is easier in a CNC mill
ds2 not neccessarily
I made them out of 1/4 x 1" Al flat bar
gets into clamping again
with softjaws, just come back and have it all done
cradek: aluminum is good enough?
you can clamp a long bar into the vise with enough top sticking up that you can mill the steps that make it a T in the entire length
for soft jaws?
sure, you want the T-nut to break before the (Al) table
but you can't mill it into individual nuts without hitting the vise
oh good point
jmkasunich: softjaws on the vise
4" of 1x1 bar will make 4 t-nuts for a machine the size of mine, unless you decide to mill them apart, then it only makes 3
SWPadnos: cool. can you tell me if it's practical to solder a tiny tiny resistor onto a pre-fabbed MCU board?
takes longer too - I don't even bother clean up the saw cut ends, other than deburring
maddash: people hand solder surface mount all the time
I've soldered 0603 resistors onto nvidia video cards :)
takes longer but a proper CNC machine would let me just walk away
jmkasunich: milling and bandsaw take about the same amount of time for me :-)
cradek: you don't understand -- the rabbit module I'm using (4100) is missing a resistor, and these buggers are TINY.
you'd still need to reclamp I bet
band saw here == small hand held saw from HF
that changes things a bit
SWPadnos: I'm asking you because you've probably seen a rabbit board before and have a rough idea of how small the components are
maddash: so what size is it?
SWPadnos: is write-misc a module like hal_ppmc or probe_parport?
yes. that'll be harder because it's conformal coated (at least mine is)
1206, 0804, 0603, 0402?
cradek: what kind of mill are you using?
owhite, it's a function in the m5i20 module
ds2: he has a max-nc, and a bport
someone doing a 0402 part by hand? is there going to be a youtube video?
if you use `halcmd show funct m5i20` it'll show you the real name
jmkasunich: I'm not sure. here are the schematics: http://www.rabbit.com/documentation/schemat/090-0228.pdf
look at page 3of4
(or is that show func?)
won't tell you the resistor size
the resistors on mine are 0603 and 0402
I don't think my ruler has small enough markings to measure
ok, what resistor is missing?
oh maybe you were thinking of misc-update.
jmkasunich: R7 - "NOT INSTALLED"
owhite, that sounds right
those MaxNC accessories are sure expensive... I made the drill arbor for lot less then their asking
what are these 4-digit numbers you guys are throwing around?
but you want it to be installed? why?
a 1206 surface mount resistor is 0.120 long x 0.060 wide
0603 is 0.060" long x 0.030" wide
maddash, just drop a small piece of wire over the solder blobs that are there already, then heat them individually
then make sure you didn't spread it all over the place ;)
SWPadnos: but the schematics say that a 0 ohm resistor belongs there, not a wire...
zero ohm resistors are a lot easier to find than other kinds
0ohm == wire!
on my board, it looks like there's one trace between the pads, but it's got soldermask on it, so it should be easy to not short to it
0ohm is nicer for automated assembly
zero-ohm resistor = screw-up cover-up
solder =/= 0 ohm
(or future-proofing, when I do it ;) )
or "use the same PCB for different things based on which zero ohm resistors you install"
I think the board I'm looking at is a 3720, so YMMV
it is just marketing appeasement so they can cripple products and dump them for less all in the name of 'market share'
seems like .1mm
it looks like 0603 pads to me, but who knows
you'll need a relatively fine soldering iron, ad non mine the conformal coat does extend to the R7 pads, so it'll stink when you heat it
if yours is the same
maddash: why exactly do you want to install a resistor (or jumper) there?
need to use the BUSY pin for something?
jmkasunich: because that's the BUSY pin of the A/D, and I want to hook it up to the CPU
using the A/D "closed loop"
what does that mean?
ie, you want to see when it's done, so you can read then - rather than waiting for a set amount of time then reading anyway
oh, like a servo, then
yeah, that's why I said "closed loop" ;)
sense when it's ready and read the data - makes sense to me
yay, so I can bring it to my dad's ex-client for soldering
should be fine
remember to use a soldering gun and acid core ;)
and a very very hot heat gun
I don't understand why they would leave out that resistor, particularly since they admit to using a 0ohm resistor
is it a configuration resistor?
pull up normall if it is not stuff; grounded if it is stuff
like a hard coded dipswitch
no, it's "don't waste a port pin unless you want the busy signal
the BUSY pin is actually connected to an external interrupt-capable pin of the rabbit
so for once, these guys have it almost-right
right - I bet that pin is also connected to the header so you can use it as an external interrupt instead
(external = from the carrier board)
so it is like a /OE signal controlled by the resistor?
the resistor either completes the wire from the A/D busy signal, or it doesn't
if the resistor is unpopulated, the A/D busy line floats, and the PE2 line either is unconnected or goes only to the external bus connector of the rabbit module
not output enable
so they left BUSY out so that pin that it would otherwise occupy could be used on the protoboard?
Oh like that
interrupt pins can be scarce commodities, so exposing as many as possible is a good thing
/OE == active low Ouput Enable
if your application needs the A/D to be interrupt-driven, you populate the resistor
at the very least, they should provide a jumper on the proto-board, but oh no, that'd be too costly
the proto board or on the core module?
I don't see much of a difference convenience wise btwn a jumper block, dipswitch, or a pads that need to be shorted
for them to put a jumper on the dev board, they'd need to run both wires out the main connector, which would be aPITA
ds2: i think the last one is permanent
maddash, no, just remove the resistor/wire if you want to change it back
maddash: not really... but I get influenced by RF folks who can't have a connector in the middle (impedence issues)
or you could try to find some 0603 switch so you can change it at will
as in, 'tear it off'?
nothing that 15seconds + a tweezer can't change
there's such thing?
or a fine-point wire cutter if you choose the wire route
we have a mobo with some tiny switches, but not that tiny
I always thought that soldering was like welding, on a smaller scale
you don't even need a fine-tip oron to remove 0603 parts
no, you can re-melt it
2 irons and lift it out like tongs or chopsticks
or one with a tip that's bigger than 0603 ;)
spoken like a true software weenie
maddash: I been around welders, welding is not permanent.. tack it, inspect, don't like it, angle grind it and repeat
soldering is a little easier :)
<-- not skilled enough to do 1 iron removal yet
heh - 1-iron removal is easy, depending on how nice you want to leave the board ;)
if that's the case, pop it in an oven and just wipe it off :P
with small parts it's easy, unless you want to save the part also
I work with a guy that is a master at removing chip resistors
just heat one pad and lift it with the iron
then melt the other side and the part comes off
then hit it with tiny chem-wick and it looks like there's never been a part there
he used to make resonators
assuming part doesn't snap in half...
if you don't want to save the part (a good assumption for $0.0001 resistors), then don't worry if you get two halves
it's not worth the time to think about it
it does if it drops on the board and you can't find it til it shorts out something else later ;)
heh - yeah, that's the hard part with such tiny parts - you really have to know where they are
at all times
i've had things short out underneath parts from either solder spray or bits of old parts
hmmm. the doctor says "don't do that"
I sent an e-mail to the emc-users list now. It's waiting for moderation because it's big (~400K). It contains my compile log which shows compile time warning. there are 2466 warnings
oh - that's a lot
on ubuntu or some other distro/kernel/compiler combination?
debian lenny, 2.6.22 kernel gcc-4.2 rtai-3.6
my router just gives up sometimes
keep thinking I should get a new one
most of them are the same: warning: deprecated conversion from string constant to ‘char*’
oh - odd
can you post that somewhere on the net?
don't you love the guys that are maintaining gcc nowadays
I don't think it's a good thing to send a 400k mail to all the list recipients
yes, 1 min
I guess the compile farm should probably use the verbose build if we want to see all this stuff (if that's not happening)
[02:51:06] <acemi> http://emrah.com/temp/emc_make.log
i noticed the deprecated conversion warning too
I see why they did that, a string constant doesn't allocate memory, so if you try to write to it, you will segfault
or that's the implication I read in the docs
hmmm - well, there's one that could be a problem:
/home/acemi/proje_usb/emc2/20080129/src/hal/drivers/mesa_5i2x/hal_5i2x.c: In function â¤?init_moduleâ¤?:
/home/acemi/proje_usb/emc2/20080129/src/hal/drivers/mesa_5i2x/hal_5i2x.c:185: warning: â¤?pci_find_deviceâ¤? is deprecated (declared at include/linux/pci.h:477)
I think we use pci_find_device in the new 5i2x driver as well
depricated doesn't mean it doesn't work
and probably in the ax5214 or whatever that other PCI I/O card was
it does mean that it won't work with a later kernel
deprecated: still works but we're gonna remove it ;)
linux developers are the worst enemy of linux sometimes
I hate it when my code gets depricated on....
nothing like some of the crap microsoft pulls, but still.....
keep it with things or take a snapshot and branch off on your own
probably doesn't work with selinux
one day you way want to come back to mainline but until then...
problem is the hackers
if you don't keep up
usually they get backported if in the kernel
ok - there are only pci_find_device warnings and a redefinition of max and min (in updown.c) - as far as C files go
in userspace, you are screwed anyways
everything else is .cc files, so the extra pedantry of the cpp compiler is getting us into trouble
userspace doesn't usually see this kind of nonsensical arbitrary change
i sent a copy of e-mail without attachement. I added the link
you can turn off the warning about char*
userspace is the biggest problem
you can fall behind on kernels and be mostly okay but space is the front line
ds2, userspace almost never changes, and when it does, it almost always still works with old binaries
the gcc guys have been making user space developers scramble for the last few years
oh - that's another issue ;)
SWPadnos: userspace is where the security exploits are (usually)
as far as kernel development goes, I kinda wish they'd lay off a little
acemi: emrah.com keeps timing out
I still use 2.4 kernels so... :P
that's a little painful
fenn: it works for me but I can install the file in another location too
<-- does this suppress the 'deprecated conversion' warning?
Use pci_get_device instead of pci_find_device as pci device returned from pci_find_device can disappear at any momemt in time.
SWPadnos_ is now known as SWPLinux
[03:06:28] <SWPadnos> http://emrah.com/temp/emc_make.log
jepler: I'll try
I can't seem to load that page
that's because SWP is special
works for me
[03:07:17] <jepler> http://emergent.unpy.net/index.cgi-files/sandbox/emc_make.log.gz
oh it's one warning over and over?
cradek: yeah, the compiler on acemi's system likes to complain about passing a string literal to a function that takes 'char *', even though this is pretty common style..
works for me, on Windows and Linux
WARNING: /home/acemi/proje_usb/emc2/20080129/src/trivkins: 'kinematicsType' exported twice. Previous export was in /home/acemi/proje_usb/emc2/20080129/src/5axiskins.ko
WARNING: "rt_sem_wait" [/home/acemi/proje_usb/emc2/20080129/src/rtapi.ko] undefined!
^^^ and these are kernel makefile bugs IMO
actually, the culprit is stuff like char *msg; msg = "THIS STRING";
is it still kosher to use the pci_* functions?
those should be const char * msg
(within a driver)
or is that char * const msg
instead of find
find works, but it returns something that is transitory
no - const char * - the pointer is to constant chars (I think)
i mean, thought it was switched over to have having board files list things and not have driver really know about how they are attached?
or is that a non-x86ism?
that would be a good thing
that may be an optional abstraction, for devices that could be connected via different buses
but I'm not sure why a pci device shouldn't know it's a pci device
but if you're writing a driver for a PCI card, you should be able to use some PCI function to find it
jepler: the diff don't work
sounds like a theoretical good thing in search of a problem
yes but by not knowing about it, it should be able to transparently handle other similarbusses like cardbus
you could have a PCIe to PCI adapter though (an external card cage), in which case you need the system to be able to find PCI devices at the end of another bus
or maybe even the serial version of PCI
acemi: still gives same warnings? or doesn't work at all?
exactly the same result
the driver writer is expected to know when they can share code across multiple versions and when they can't
acemi: OK .. well I don't have any more guesses in a vacuum
calling pci_* explicitly implies you only care about things really connected to the PCI bus even though there might be other compatable stuff
in that case, your driver is only for the PCI version ...
there's another consideration here, and that's that we at least pay lip service to compatability across a wide range of kernel versions -- in practice, we only compile on 2.6.15 and greater, though.
how do they handle identification?
or that part of the code is meant for PCI-specific device initialization
so the religious fevor of avoiding knowledge inside the driver is limited to non-x86 stuff
no, it's just got its place is all
your driver is suppose to register a name and the code in arch is suppose to handle finding it and telling you about it
dear OS, here is my pci vendor id and device id, please tell me where to find my hardware, please?
if you have a device that needs to know how it's connected (for example, for optimizations), then the driver needs to be able to get that information from the system so it can provide it to the card
Microsoft tells people not to name their drivers, it now takes pages of code to open a driver
as opposed to "CreateFile(\\\\my driver)
there are benefits to both ways but my point was more of what's the current trend
current trend is to follow foofoo theoretical nonsense
although making everything work the same is a good thing
there are times when not knowing how a device is connected is useful (like CD-RW drive), and other times when it's critical (like a 5i20) ;)
if you have a PCI card reduce to pure IP and resynth as part of a new x86 processor core... it would be nice not have ot redo the driver.just drop in a few structures in you arch/i386/board-XyZ and the driver finds it
I go to sleep, bye
I'll have to check that out
but I been swimming in arch/arm mostly
it would also hopefully reduce the number of changes you have to make to your driver
due to foo-foo theoretical nonsense wars
ds2: are there good discussions of current hardware driver development?
eric_U: nope, just indirectly from comments rejecting patch that are submitted upstream
just recently, there was a big effort to make drivers not know about i2c addresses but have it contained in the stuff in the board file... think it has something to do with sharing and being able to send the patches eventually to mainline
sad thing is, the way development has been going, they'll do some device independent architecture, realize its shortcomings, and do it again
that's a 2.7 feature ;)
I gotta get my milling machine back online
the new wizard is good idea, imo. might inspire next gen emc :)
I had some snarky comment about allowing people to design their own screens in VB, but I'll just leave it at that
I still wonder how long it's going to take before Ballmer kills Mach
SWPGot the button in.... NICE.... 24V LED AC/DC, one set each of NO and NC contacts
and finger guard
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2008-01-29.txt
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2008-01-29.txt
I saw you were looking for me earlier
(but it was 2am here, and I was happily asleep)
I wanted you to look my import
[09:18:31] <micges> http://www.emc-files.webpark.pl/import.tar.gz
post dxf files to ngc
check irc talk about 2 am with cris and test program if you want(have time)
micges: looks nice :)
(from what I read..)
especially the GPL v2 part
cradek: around ?
a bit early for him
or late :)
alex_joni, I saw your comment about axis filter this morning,
micges: cradek is usually around in 3-4 hours from now
archivist: axis filter?
"Jan 28 19:58:10 alex_joni ARCHIvist: hmm, then we need http filters for AXIS :)
I did a bit more gerber decoding and finding oddities in Kicad output last night/this morning
i'm on my xbox... i ws thinking it would be sweet to have emc running on this with a usb stepper driver
is it ffeasable?
emc uses a realtime kernel
and doesnt use usb because of latency issues with it
there is a debian distro for the xbox
i wonder if the realtime patches would compile with the xebian kerne?
does it have a parallel port
or pci bus
it does have some sort of pci bus
i bet i could solder tiny wires to it somehow
maybe a pci stepper driver
i guess a lot of work for something so slow
it's a 733 celly processor though
yes just use an old pc
733 is fast enough for steppers
oh i know i have ubuntu-emc running on a 233 p2!
and axis is usable
running a 3 axis mini mill
but the xbox is so small and tidy... i was jsut pondering
i couldnt get it running on a machine like light
er, that light
sry, need sleeps
i get up to ten ipm from 3 small steppers
the limiting factor being my driver
does linuxcnc support RTA Mind Series stepper motor controllers?
slarba_: got a link for those?
slarba_: like this? http://www.totalmotionsystems.co.uk/products/sdrives/mindb.html
if they are step & direction as the one I pasted, then sure.. I see no reason why they wouldn't work
slarba_: if it's the MIND T model, then it's not as simple
those seem to accept serial RS232 or RS485 connections
so you would have to write a bit of code to make them work with emc2 (and I can't say for sure they are suitable for emc2, without seeing a manual.. which I can't find on their site)
yeah, my controllers are MIND S
no, MIND T
they're programmable, with 128 memory locations for motion commands, simple repetition commands etc
so you have a manual describing the protocol?
if so, you might be able to adapt emc2 to work with them (I suspect it'll take a bit of work though..)
yes I'm familiar with the protocol
I actually have a working cnc machine. the software and motion command interpreter is written in python+pyserial :)
joint 3 following bugzez dey be gone, just tested 2.2.3
I managed to add one, which has been cured by 2.2.3
alex_joni: M66 bug-be-gone
BigJohnT: yay :)
* BigJohnT off to breakfast
can you fix my hardware bugz too (bug 1: machine needs finishing)
* alex_joni marks bug #1 as works for me & pending
high weight overloading on the table test, passes, "design" weight 5-10lb ish seems happy at 40lb+
[13:42:47] <skunkworks> http://www.cnczone.com/forums/showthread.php?p=401483#post401483
* BigJohnT off to my machine shop talk to you guys later
big john should have put a smiley face on his reply.. Those mach people are finatics
that's funny, I have bug #1 too
daisy.ngc come with the live cd?
does it work with a 3 axis machine?
needs 4 axis though - unless you modify it
I guess it needs 4 axes because of the harmony?
I've gotta say, that is the best sounding cnc music I've heard yet
some people would probably download the livecd just to play it on their machine
I know I would ;)
I am sure the video is only 3 axis... we should get cradek to make a video on his machine.
where is the code on a vanilla emc install?
sorry, locate worked
ok, so I have no clue what is going on in that file :)
no wonder it doesn't work on mach
I have a lot to learn
BigJohnT: you should have smiley faced your post.. those mach guys are fanatical..
LOL you must have read my post
* skunkworks is samco.
we're usually so nice to the mach fanatics, a few reminders that we have better music isn't going to hurt them
I guess if you paid for your software you would be a fanitic too LOL
it's too bad that nobody seems to make a step/dir to analog converter any more
$150 not that much in the greater scheme of things
I paid for this s** it must be betta!
if windows didn't scare the crap out of me, I'd think more seriously about it
I like windoz and make my living using it but I still would not play CNC with it...
* BigJohnT added :D to my post
good job :)
buy a chevy or get the cadillac for free?
main reason is that lots of people can barely use windows, linux scares them
eric_U: ther is the $200 linux pc from walmert now
the fanatic fan base probably convinces a lot of people too
it's not the money though JymmmEMC, I'd say some of those guys think that linux is communis'
in reallity only a few of us are communists
eric_U: No, I mean that an inexpensive PC that is sold in a mass distrobution chain is using linux.
that's for the people that don't know enough to be askeered
feelin sorry for Bill
from what I've heard, it's really not that bad.... comes with Open Office, mail, drawing, etc. For the mainstream they don't need much more.
linux is probably better for most people
OH btw... MacBook Pro + Leopard + Boot Camp... NICE
never understood why I should spend thousands for a home computer and software
eric_U: for the support?
just doesn't make economic sense to me
eric_U: Computers are the only thing that I know that cost thousands today, and are worth hundreds within six months.
eric_U: if a screw is loose on your chassi or one of the desktop icons looks bad you can always give the seller a call
one screw holding the case on is more than enough
whats a case
in fact, I didn't use cases until I spilled coke on my machine and ruined my ethernet card
* archivist has two caseless here
archivist: the thing your chinese food comes in
then I said "where's that darn case?"
I did lose one to coffee
my case isn't bolted together though
got a free mobo for it though
* Vq^ has amazingly never lost anything to coffee
inside of case==another shelf to store stuff
one of the guys at work got a computer out of the trash, he's been using it for almost 2 years now
that thing must be incredibly obsolete
my case has a stack of dirty dishes on top
My laserjet4 is ex trash
skunkworks: got a link for BigJohnT's post?
2 screws hold in harddrives. 1 screw holds the case cover - if exists.
[14:36:14] <ALS> http://www.cnczone.com/forums/showthread.php?p=401483#post401483
I've been trying to use more hardware to hold the internals together
[14:36:30] <skunkworks> http://www.cnczone.com/forums/showthread.php?t=51348
computer zen thing
the only thing I really put all the screws in - is the motherboard.
BigJohnT: ha :)
alex_joni: I thought it fit...
like a wet kipper
* archivist plays daisy, wee
oo that means I have emc
who needs a case.. http://www.electronicsam.com/images/pentIII2.JPG
I remember the technical guy from Sun with a motherboard bolted to the side of his desk
they couldn't make the machines fast enough, so if he had a full machine the sales guys would take it from him and ship it
archivist has emc cause if he had mach he could not play daisy
what happens if you try to run that on a machine that only has xyz?
* archivist waits to wind up the rest here when they return to musac
so you have a 4 axis machine?
and doesnt sound good
well building it, some bits "loose"
archivist: that's a dual CPU board?
I don't think my mill will play it too well, but I guess it can't hurt anything
alex_joni, er where
[14:53:47] <alex_joni> http://www.electronicsam.com/images/pentIII2.JPG
that was skunky's comp
or is that samco's ?
* alex_joni is tired.. time to go home actually
I like the fans
especially the case cooler on the PSU
alex_joni, I have mine visible somewhere its dual, cant remember the url atm
yah - they where just screwed onto the heatsinks.. the heatsinks where from slower slot1 processors.
my son's computer is kinda like that, I put a 120 mm case fan blowing on it
archivist: for duals there are special RT kernels.. but they aren't quite perfect
it's harder to make a kernel for SPM machines, and have it run on all possible hardware
alex_joni, emc is on a single
archivist: the default kernel .. yes
but if you're happy with the performance.. don't sweat it
the SMP rt kernels haven't worked for me on single processor machines
obviously not :D
but I haven't tried recently
I browse on a dual pentium II and have emc on an Athlon 850
eric_U: because your hardware is too old :)
skunkworks: those are PIII
* BigJohnT listening to Jefferson Airplane's Lather
BigJohnT: brown-noser ;) http://www.cnczone.com/forums/showthread.php?t=51436
You should be working : )
argh my dad's soldering buddy is in New Mexico
now I have to buy a soldering kit, and flip a coin to see which one of us (dad and me) will do the soldering
because I have absolutely no experience soldering, and my dad's a brutish fellow with no regard for the small and fine things (figuratively and literally) of life
finally after work :)
cradek: I fixed import
in a few minutes targz will be available
micges: know anything about building debs?
ok, then we'll see how to make it work
I'm thinking about making a deb package for it, so it can be installed easily by emc2 ubuntu users
then you only have to add it as an AXIS filter, and open dxf's with it :)
Alibre Design Professional is on sale...
(it's probably not that advanced/stable yet, but I'm sure it'll get there)
BigJohnT: I'll wait for the Expert version, thank you
what is in the expert version?
alex_joni: I think that I can obtain building debs :)
I have the CAM
for 30 days
only had a few minutes to try it out the other day
can't comment on it yet
[16:27:06] <BigJohnT> http://www.myvirtualnetwork.com/mklotz/
[16:27:20] <BigJohnT> http://findarticles.com/p/articles/mi_m3101/is_n5_v63/ai_9000692/pg_1
we have a big old oak desk that we got from from a friend. It came out of a school. It has a door bell type switch drilled into the top. I thought - that would be a cool button to start the computer. totally mount it within the desk somewhere..
(motherboard and supply
we wonder what the button was for - we think a trap door for bad kids.
[19:02:53] <tomp2> http://www.ozbricks.com/bobfay/om4.htm
5 axis lego mill
hah! and a cnc toast burner ( elvis on toast )
Using EMC 2.2.3, I cannot use the numpad on my keyboard to set the numbers in the "Touch Off" dialog. Is this a valid bug candidate?
yeah though I don't think it's a new bug
I'm looking at the axis.py. I'm having a hard time trying to understand the link between the 3D lines and the gcode. Does every line have a list of attributes, that includes the gcode line number?
Or is there a high level doc about axis.py?
Hugomatic: no, there's no documentation about the structure of axis.py
it is what it is
jepler: And it is quite nice, I must say. Is there a way to run axis within a debugger? I use Pydev and I don't think I can attach to a running AXIS.
Sometimes my screens are full of lines, and I can't see the details (image_to_gcode with multiple passes). I'd like AXIS to be able to "hide" certain lines, based on the gcode line number, or time, or % (show only lines between 2367-3000 for example, or between 2min3s and 5min10s). Its like an extended selection, I guess.
I believe I could hack it, if I tried
Hugomatic: beats me, I don't use pydev
jepler: are you like Linus, and you don't use a debugger either?
I don on C/C++ -- them's hard languages
I do on C/C++, that is
jepler: all right, then, I'll stop whining. gedit and 'print' will be my only friends. I'll let you know how it goes.
cradek: around ?
i have problem with english
i'm to configure home
by index pulse
parallel not work?
two parallel port
sure.. that works too
all other signal are ok
i need one pin for home switch + one for index pulse?
and another 2 for the encoder
i use servomotor dc closed loop
step dir as stepper
billy_kid2: I don't think you can home to a home switch and index pulse without an encoder feedback
yes, because you have to know the exact machine position from encoder feedback at the moment the index pulse is seen -- you don't get that if something outside the PC is actually handling the encoder, or if you're using a stepper.
not possible only index pulse
used as a switch
billy_kid2: you can use the index pulse as a switch
and2 between the index pulse (with a latch) and the home switch
but you'll need to go very slowly for this to make sense
not sure about the index only homing which is recent
maybe that can work without encoder.. but I doubt it
jog by hand near the switch (or use the switch like a G38.2 probe)
then home with index only.. slowly travel towards index pulse
index pulse is very short...
that's why I said very slowly
and you can put a latch on the index.. but it won't be as accurate
(the idea with index pulse is that you usually have a hardware counter for the encoder,
then when the index pulse comes the hardware counter resets the count
even if emc2 didn't notice the index pulse yet
when it will notice, the machine might have moved on
but the encoder has been reset at 0, and now it know the precise position)
i leave home switch
billy_kid2: how fast are you going?
input scale 200
billy_kid2: max speed?
5000 mm minute
83 mm / sec
-> 16600 steps /sec
is that the same as the encoder resolution?
x and y have 1000 ppr
y and z sorry
x 800 ppr
1 rotation = 1 mm ?
so you need 1000 * 83 / 3,2
if you run the base_thread at 10-15000, you can count that in software
with encoder 500 pulse double velocity
i can post the link of my servo?
x axis now has a clue.
does the index work?
index does work
seeks home at max velo
and then goes klunk, klunk
HOME_LATCH_VEL is set for -0.01 or -0.001 (tried both)
it starts looking for the switch first.
later the pulse
how configure index pin?
billy_kid2: dave_1 is running a motenc card.. and counting encoders too
[22:37:17] <billy_kid2> http://elm-chan.org/works/smc/report_e.html
doesn't work here
elm site comes up fine here
very very good servo
i have 4 driver
alex ... I'm wondering if I have the ini file set right for home on index
dave_1: it sure seemed like that
then something else is weird
dave_1: what emc2 version ?
is it 2.2.3?
ok, shouldn't matter
can you describe what happens?
tanks i go
billy_kid2: see you, sorry I can't be of more help
billy_kid2: where from?
we used to have some italian guys in here
xemet from sicily, and giacus from calabria iirc
xemet great head!!
maybe you can ask them when they're online.. easier with the language
dave_1: there are a couple of states in the homing mechanism..
position axis in middle so there is plenty of travel
ok.. you hit home on X
mouse home button
it starts to travel towards home switch
x travels negative at 90 ips
hits index and seems to oscillate a bit then turns green
hold your horses
no home switch .... yet
ah.. index only?
then I think you need HOME_SEARCH_VEL 0
ah! I'll go try that ... brb
good nigth :-))
"17:45:10 <billy_kid2> xemet great head!!" wtf?
try not to be so narrow minded
bet your italian is way worse than his english
mille de graze
why would anyone want to go for a VFD nowadays?
cost-saving or pricey?
they're crappy character displays, with a backlight
VFd in here is normally Variable Frequency Drive
you mean CRT?
I mean, vacuum fluor disp
cathode ray tube
imo they're not really cheap, and they use a lot of power
and take up too much space
no crappy green 7 seg
fuck this, I'm going make him use an LCD
VFD are more readable, and waaaaay brighter
LCD are the same resolution in general
are you kidding? 320x240 vs. ~4x20?
there are text and graphic lcd's
besides, can't the LCD contrast be turned up to match the VFD?
oh, then the comparison is between graphic modules and character modules, not between LCD and VFD
and there's OLED displays.. also quite nice
alex_joni: but no graphic VFDs
what's the "O"?
maddash: dunno.. I like graphic LCD's with 2-3 colours
like these: http://www.futaba.com/products/display_modules/module_products/graphic/index.asp
organic? or something like htat
O means Organic
[23:06:04] <alex_joni> http://en.wikipedia.org/wiki/Organic_light-emitting_diode
beight but lower power - they almost glow without any power applied
* maddash has a picture of asians being crammed into an led
POLED, TOLED, SOLED and all other families
* alex_joni goes to bed
SWPadnos: can you help dave_1 ?
I don't know :)
motenc and homing types aren't my specialty
but I can stall until jmkasunich gets here ;)
good night all
just got back from shop
damnit, mouserrr won't let me sort by $
things are really erratic
it acts like it wants to do a standard home on switch + index
takes off neg stops and comes back a ways and homes.
No home switches are hooked up so I think it is seeing noise
I really intend to use home switches but was trying the home on index for fun and to see if it works
dave_1, I don't remember the ini parameters you need to do index-only homing
there has to be a "don't do that" joke in there somewhere, but I'm worn out right now
basically I think ... home search vel = 0, home latch vel 0.01 (i.e. slow) home ingnore limits yes
I guess tomorrow is dedicated to wiring the home switch stuff ... or at least connect the opto-22 board and pull stuff high
you just have one index?
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2008-01-29.txt
I'm only working with the X axis .. and yes there is only one index pulse/rev on the encoder
so you're just going less than a rev
it should go less than one turn ... yes
ideally with a home switch you set the cam so it moves 1/2 turn after it hits the switch
why is that?
you really don't want the switch and the index on top of one another ....
are you using mechanical switches?
I have some nice AB oil tight limit switchs with roller
can you buy the rubber covers for those things?
actually X is automation direct and Y is AB
they are already sealed around the piston
ok, mine had rubber bellows, forget what brand
I'm using proximity switches for home, don't really want to hear the table running into a switch over and over
encoders are Koyo 2500 ppr ... automation direct
I have a couple of prox switches but have not tried them
I've tried mine, they seem really repeatable
I don't have them mounted on the machine though
I'm supposed to build a gage for tool length with them
that would be a good use
Ray H says ther are very repeatable
the cool thing is that they only register out the front
I'm going to thread them into my table
SWPadnos: why are the graphic lcds with resolutions >= VGA so much more expensive than if you bought a pre-built LCD panel?
where are you getting your lcds from?
digikey and the like are always outrageous for things like that
anyone know the link for the "daisy" video on youtube?
ok, they are also going to be unreasonably expensive
as will all the electronic distributors
it's just the way their business model works
er, then who would be reasonably priced?
they know if someone like me needs an lcd panel for work, we will pay their outrageous price
and if someone needs a lot, they'll go somewhere else
surplus is the only place
there used to be a guy I bought lcd panels from surplus, don't remember the name
ebay isn't an option?
doesn't ebay sell used merchandise?
should be some nos, i just don't know what your application is
i have to shoot my neighbor's cat in pitch dark
"new" or "reasonable price" choose one, unfortunately
well, used should be fine then
steal his lcd and use that
who was the person that was using emc to fly santa around his front yard? That would be excellent anti-cat weapon platform
seriously, though I just bought this thing (http://www.embedded-store.com/index.cfm?fuseaction=product.display&Product_ID=313&CFID=568209&CFTOKEN=51094331)
and it cost me a fortune
that is a little expensive
don't know that $80 is a fortune though
yeah, and it sucks up >10 i/O pins on my mcu
maybe I could disassemble it and hook it up to some lcd driver ic? there is such thing, right?
there are serial displays available for a little less than that
so you want a graphic lcd that would basically display 2 lines of text?
well, that's the other thing -- 122x32 isn't enough; i need something like >=128x64
did you look at sparkfun?
gotta go, bbl