SWPadnos_ is now known as SWPadnos
emc 2.1 released?
it's ready but we're debating how exactly to release it
we could make it appear in the update manager just as any other new version, or we could require the user to do something to choose to switch from 2.0 to 2.1
(this is an issue mostly because a few little things need to be changed in any customized configurations)
03jmkasunich 07TRUNK * 10emc2/configs/nist-lathe/nist-lathe.hal: fix lathe spindle sync signals
03jmkasunich 07TRUNK * 10emc2/configs/sim/lathe.hal: fix lathe spindle sync signals
03jmkasunich 07TRUNK * 10emc2/configs/lathe-pluto/lathe-pluto.hal: fix lathe spindle sync signals
03jmkasunich 07TRUNK * 10emc2/configs/max/max.hal: fix lathe spindle sync signals
03jmkasunich 07v2_1_branch * 10emc2/configs/lathe-pluto/lathe-pluto.hal: backport: fix lathe spindle sync signals
03jmkasunich 07v2_1_branch * 10emc2/configs/max/max.hal: backport: fix lathe spindle sync signals
03jmkasunich 07v2_1_branch * 10emc2/configs/nist-lathe/nist-lathe.hal: backport: fix lathe spindle sync signals
03jmkasunich 07v2_1_branch * 10emc2/configs/sim/lathe.hal: backport: fix lathe spindle sync signals
03jmkasunich 07v2_1_branch * 10emc2/configs/nist-lathe/nist-lathe.hal: remove inaccurate comments
03jmkasunich 07TRUNK * 10emc2/configs/nist-lathe/nist-lathe.hal: remove inaccurate comments
03jepler 07TRUNK * 10emc2/src/hal/utils/halcmd.c: fix 'net arrow' bug
03jepler 07v2_1_branch * 10emc2/src/hal/utils/halcmd.c: merge rev 1.113: fix 'net arrow' bug
ve7it is now known as LawrenceG
Jymmm is now known as Jymmmmm
That one had to have been shot in seattle :-/.
not that anyone wants to know this ...
but I just found the old video of the time the basement of our rental house backed up with sewage
and poo was floating around in it
sounds like youtube time to me :)
Had that happen in two apartments I;ve lived in - will no longer live in an apartmet except on the top floor - shit does indeed flow downhill...
so do machine tools, unfortunately
announcing the self levitating Vertical knee mill...
would anyone by chance know SPICE?
try in ##electronics
but that question gets asked a lot and seldom garners a response
better to ask a specific question
I just want to see if someone knows what does it mean when spice says thereis negative current through a diode
I'll try them
charge recombination maybe
Oh... the complicated stuff :/
that's why you're using spice in the first place
I just want to see if I can get a laser diode driver to work with a 1MHz pulse rate
is that all?
unless you have a pointer for a pulsed laser diode driver ? =)
i know that laser diodes will die if the drive exceeds the rated current for like 1 picosecond
so that makes them fiddly and expensive to design
according to SPICE, Idiode never exceeds Imax, but it does go negative even though Vdiode is never negative :/
oh, no that would change the voltage too
heh... it is a puzzle.
hmmmm I have no chance of seeing it if I build the circuit :(
joomla killed all my changes, mercilessly
evil evil thing!
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2007-01-28.txt
good morning alex_joni
cradek: how did you install the upgrade package? did you use dpkg -i ./emc2-foo.deb? if so, I noticed that doesn't work ok for updating. But creating a local repository (as I described on the wiki page) makes it update without problems.
[09:57:04] <alex_joni> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Emc2.1.0
you can even update using the update manager after you set the repo up. I think dpkg sucks at pulling in dependencies, and apt does the right job
alex_joni, any fun stuff at work lateley?
hmm.. depends on your version of fun :)
got an old bot working again
1988 old :)
what is the use?
welding .. it's all I do at work :)
Is it going to be used at one of your customers?
this one is at a university
it can be :)
anything fun happening?
* alex_joni goes to make some sushi
alex_joni, send me some when your done, plz
DCC RECV MAKI
[10:33:59] <fenn> http://www.catb.org/~esr/writings/no-sushi/
apt-get -install alex_jonis_sushi /dev/myMouth
Climbing Mount Fuji,
Cherry blossoms in the wind,
I see a Coke stand.
todays poet, alex_joni
* alex_joni plays to read that
Martin_Lundstrom: it was from fenn's link
fenn: got my comments on the licensing?
hope they made some sense :)
basically, you audited everything but forgot to write it all down :)
2-3 times :)
i coulda swore the emc front page said "we are making the transition to GPL"
it said a few years ago :)
no this was yesterday
hmm.. you had a very old cache hit?
i logged in to fix it and that's when i asked if it was ok to do so in irc
I didn't touch the fron page..
Last Updated ( Sunday, 01 October 2006 )
* fenn thinks joomla is possessed with evil voodoo spirits
but I bet that was picture related
i poked around at the about page today
fenn: care for some nice german tunes?
seems good (the about page)
hmm.. seems I need to upload them.. will take a while
alex_joni, is the industry in Romania growing?
at what rate, do you have any clue?
Martin_Lundstrom: no idea :)
but it is growing steadily
ok, but people is getting it better or?
yup.. for 16 years now
but we went about 50 in the wrong direction
so it'll be a while :)
imo mentality is the biggest issue.. and we still nees 1-2 generations to pass for that to change
I think it is too slow
it can't go faster
people won't accept it
would more external monetary help, speed up the process you think?
hi, anyoen awake yet?
Martin_Lundstrom: money always helps
I messed around with the python+opengl stuff yesterday, but nothing really nice to show yet
I did make a mind-map to get a bit organized: http://imagebin.org/7046
anonimasu, Dallur, did you hear that, go check out the image and see what you think
heh nice blobbies
blender has alot of open gl python stuff
I need to get the hierarchy of classes set up properly... then I expect progress will be much more rapid when the basic framework is there
now I worked with raw OpenGL, since VTK documentation is so sparse
sounds hardcore :)
cam software will be mostly about data structures and algorithms anyway
it would be nice to not have to duplicate effort though for the cad system
yes, but so far I got some help from Julia Todd/freesteel, who is eager to help with the algorithm part
yeah i talked to him earlier but it didnt go anywhere
som most of the cam part is handled, as long as there is a working UI around
feel free to assign me a chunk if you think i can handle it
well, I want to make it go somewhere now
what would you say that your strengths are?
lots of free time :)
(I'm not a professional programmer by any means)
oh, wow, that's rare
i just revised my sourceforge profile today
fenn: are you not selfemployd?
so right now my choices are: wxPython for GUI toolkit and OpenGL for viewport, can you live with that?
I'm on WinXP, but it should work under Linux also
yeah i would have chosen wxpython too, but i guess python/opengl support is wishy-washy?
maybe we can just steal jepler's minigl library
I've had no problems yet
I use the glCanvas widget from wxpython
anyway, one of the architectural design problems is how to separate geometry from how geometry is drawn(which might change)
I guess I'm just having a hard time grasping how the components of this system communicate with eachother
well if you have a relational geometry system you are probably going to use implicit geometry (primitives, csg) rather than a mesh
I'm sure I could write the geometry class fairly easily. But how does the UI communicate with it, how does the viewport draw the geometry
there's something called the command pattern, maybe that's what I need?
or some kind of central "controller" through which everything goes
[11:12:07] <awallin> http://en.wikipedia.org/wiki/Command_pattern
must confess i never really understood what that stuff meant
just how objects talk to eachother and how you build a good class hierarchy
how about we make something simple, see how it goes, and then write it correctly the second time :D
are you more interested in CAD, or CAM?
i want a full system.. dunno which will be more useful to start with :\
just drawing two kinds of objects would be needed at first: STL surfaces, and Lines
if we want to start with the toolpath generation
cam seems like a simpler target
just input configuration and output
input would be an STL file
IMO CAM seems like a more applicable target to begin with, as there are lots of (OSS) apps that can generate models
then the toolpaths are generated, they will all be Line objects (segment after segment after segment)
then the toolpath may be output in some suitable format (maybe filtered first)
do you have time/energy to look at this now?
yes, i may go to bed in a couple hours
ok, I will put what I have on my website
beware that I broke it yesterday so it doesn't do anything useful
I'll add comments before I post it so you might make sense of it
should we set up some kind of cvs or darcs or something?
yes probably, but the project needs a good kickstart first
it gets out of hand quickly
a core group needs to come up with the basic architecture for the system, and the architecture needs to be good from the start
oh, and we need a name to set up cvs on sourceforge... naming is really hard
i'm pretty anti-sourceforge lately
are there other free places where this could be hosted?
I'm sorry now that I look at the code it's a complete mess... I'd like to clean it up before showing it so it will not be before evening (in 10h where I am)
oh, someone is ddos'ing berlios right now :(
ok i'll wait til tomorrow then
if you think savannah is OK, I might setup a project there
do you have experience with setting up a prject on savannah
hrm. "The review we do can be long and tedious for both the submitter and the reviewer."
I wonder if it could be hosted on linuxcnc.org
back when i was going to use brlcad libraries i was going to call it "cam-o-blammo"
that's a bit long
I think I'm going to ask the guys about hosting this kind of project on linuxcnc.org - that would be the easiest
i seem to have forgotten all of the potential names
yes it would
it could be thought of as a .stl filter for axis
or maybe emCAM
yes, but I would not consider integration with emc at this point, I don't mind having the toolpath generation in a separate app
if tight emc integration is what you're going for
its really easy to make a filter for axis
just a python script that prints gcode
and, if we are going to use wxpython on pyopengl+glut then the dependencies are different from emc
and, I'd like for it to be cross-platform, i.e. run on windows and linux (+mac?)
yeah and that's not a bad way to go for emc's gui either
how's cross-platform capability wrt opengl, wxpython and the like?
wxpython shoudl be really good
lerneaen_hydra: that's the point of wxpython
it calls gtk or win32
opengl might be harder (require some tricks)
so essentially it's only the other deps that can break
and, the CAM stuff from Julian is currently only in a dll file, which works under windows
It's written in C or C++ and then compiled, and somehow mangled with SWIG so that the functions are callable from python
what is he releasing anyway?
lerneaen_hydra: no, not yet
surely not the cash cow adaptive clearing algorithm
awallin; think you can get it?
fenn: what I have now is a z-slicing algorithm, take an STL file and imagine a plane cut's it at some Z-value. it will output the contour where the plane cut it (offset by the tool radius/diameter)
but as he said, most cam algorithms are basically commodities so it wouldnt be a disadvantage to open source them
lerneaen_hydra: I think Julian has a positive attitude to this, but he said releasing the source for everything right now is not possible
hmm, I see, NDA's?
waterline seems like pretty standard geometry
== z slicing
but those two guys are pro's on the CAM stuff so we definitely need their help, wether it's ready C-code, or just advice on how to do things
lerneaen_hydra: it's their own code, but they might have licensed it to cimco + others
waterlining seems to be very easy to implement
reading their blog, it seems that many people want their source code...
anyway, our program should be able to work with an algorithm from anyone
i dont think that basic stuff like waterline and constant stepover will be hard to write from scratch
but I think if we get something going, the freesteel guys will contribute.
fenn: so, if you want to start with something, learn how to interface between C and python, using swig (that's what julian used=
its the full 3d and HSM stuff that's hard
I know very little about how to write a C program that's callable from python
look at gdepth
jepler should be a big help with the C<->Python stuff
what about jeplers gdepth?
actually we can probably use a lot of gdepth for doing constant stepover
well, basically you subtract the depth of cut until you get to the surface you want right?
yes, there's a thing about this on freesteel, finding the contact point btw the cutter and a triangle
take the maximum z value of the surface under the tool, and that's your tool's Z height
that might be an algorithm to start with also: cutter shapes(cylinder, ball, bullnose) and find the contact with a triangle
the contact point stuff is more useful for pencil milling
as you said, I think the basic cam algorithms will appear pretty quickly once we have a framework to work with
so, what have you written so far?
a wxpython frame which has a glcanvas inside it
using simple commands I can draw stuff inside the glcanvas
using STLtools from Julian I can import an ascii or binary STL file and display it
commands like GL_TRIANGLES?
I can rotate and zoom with the mouse, panning is a bit crap
this is in "direct" mode (if I remember the terminology) not using a display list (yet)
then I can call Julian's zslice and get a list of points that lie on the zslice contour. but I haven't drawn these yet
because I wanted to write the architecture right from the beginning
another approach would be to do a simple program that does model to toolpath, as you suggested before
but I'm confident that the cam algorithms work, so I wanted to work on the UI stuff first
it looks like dallur wants to write a cad library first and work on the actual program later :D
yes, he started on something like that, the geometry tree in my mindmap
I need to go now, but I'll be back later. As I said earlier, if you are interested, a small example of how a C-program can be called from python using swig would be useful
i dont understand the bead/ring/magnet
a snake is like a 1d magnet?
er.. something like that
[11:57:30] <awallin> http://aerohydro.com/products/rg/rg.htm
a bead is a point that lies on a curve
how can you patent a conceptual framework!?
like beads on a string
fenn: in the USA you can patent almost anything...
fortunately not everyone lives in the USA
a ring is a bead on a snake
a magnet is a point on a surface
a snake is a curve on a surface
anyway, this relational geometry would be for the CAD part of the program. CAM will work with points, lines and STL surfaces
i dont see any need for a distinction
well, a bead would be defined as Bead(myLine, t)
shouldnt it be the other way around?
where myLine is a reference to the line it lies on, and t is a parameter from 0 to 1 which says where along the line the bead lies
the Line is supported by two points Line(startPoint, endPoint)
and the bead is supported by the line Bead(Line, t)
a simple surface is made from two lines Surface(Line,Line)
a magnet: Magnet(Surface, u, v) u,v= [0,1] are coordinates within the parametric surface
I really need to go now, catch you later
How is your table?
SWPadnos_ is now known as SWPadnos
Rugludallur; why do you switch names between Rugludallur and Dallur so often?
or are you not the same person O.o
two personalities :)
they sometimes claim to inhabit two different bodies, but I wouldn't bet on it
does anyone know of an app to go from a heightmap to a dxf?
``My wish for peace and harmony among all who work for freedom -- whether in software or elsewhere.''
lerneaen_hydra: to what in a dxf?
stl would work too I guess
I was hoping that by just throwing enough memory/disk space at it the aliasing won't be too nasty
this is some great reading (thanks fenn) : http://www.catb.org/~esr/writings/no-sushi/
you have a grid? wouldn't be very hard to translate but it's two triangles per grid point
yeah, I have a bitmap/whatever format image
and I want to make an STL/DXF based on it
"Kyoto's pizza (a Japanese imitation of an American imitation of Italian food) turns out to feature toppings like squid, eggplant, and pear. Very weird."
heh, in sweden there's tons of so called kebab pizza
cradek; any ideas?
[14:11:49] <cradek> http://timeguy.com/cradek-files/stl-to-dxf/stl-to-dxf.py
here is a very basic program that shows how to generate a DXF of 3DFACE entities
instead of starting with STL you could read the image
* lerneaen_hydra can't code
not yet at least
nothing like a project to get you started
that's true ;)
[14:13:51] <cradek> http://timeguy.com/cradek-files/image-to-gcode/image-to-gcode.py
this reads an image and accesses the pixels directly
you just have to massage the two together :-)
do you understand conceptually what the triangles are that you would output?
at least I have one possible model of it in my head
I'm imagining a grid like a crossword puzzle, with diagonals drawn one way through each square
oh, ok, so four triangles per pixel
your height map is one height for each intersection
no, just two
hmm, that would also work
well to be more clear, two per "square"
why wouldn't square = pixel?
ignoring the boundaries, it is one pixel per square
ack, boundaries too
I wonder what I'll do about them
stop generating triangles when you get to them :-)
I was thinking more like if my cam app will throw a hissy fit if I try to import a shell
oh you think it should be closed?
I'm not sure ATN
it would be only a little harder to close it - you would make triangles "down" from the edges, and then a big pair of triangles to make a square on the bottom
the lower surface should be a bit lower than the lowest data point
actually, dxf lets you do quads, which makes this whole thing easier
I'm not sure if they're required to be planar - if so you better use triangles instead for the map (but you can still use quads for the side/bottom)
a four-sided triangle
isn't that an oxymoron?
I think a 3DFACE can have 3 or 4 vertex
hmm, it turns out my 3d app can do this, although it's more than slightly convulted
sometime I'll probably try to connect the two together
installing you ubuntu/emc with a resolution of 640x480 is really crappy
you cant see the buttons of the installer
rcsu: hi, you might mention that on the #ubuntu channel, and maybe ask how to file a bug report so they can fix it in later versions
yea, should really do it
I've seen others say that too, so they should know about it
is 640x480 a limitation of your monitor? EMC might not work very well unless you have 800x600 or more
no, my monitor can drive 1280x1024
* rcsu is installing now
there's an option on the initial menu to pick the vesa video driver - don't remember what it's called - but you might try booting with that
you may have a problem with your particular video card - again, maybe asking in #ubuntu would help
alex_joni: thanks for the M-code tip
lots of interest in cad/cam lately
was discussing this with fenn, and we came to the conclusion that it would be nice if a cvs repo for cad/cam could be set up
I wonder if there is room/bandwidth for that on linuxcnc.org?
Should be. But a whole cad/cam system is a big project.
sure, need to start small. ability to import stl files and export basic toolpaths would be v.0.1
So not CAD, just CAM?
yes, I think cam would be the more interesting part to start with.
Hi Dan. awallin just brought up one of your favorite subjects.
there are very few open source cam programs
there are lots of little bits and pieces though
awallin: Dan is the original owner of linuxcnc.org
ah, I didn't know that.
rayh, currently on my ubuntu dapper machines, I am using Qemu
to run win98 cadcam
and it works fine
but, it's not ideal
he paid for the name and steve_stallings served the site.
so, I have rhino3d, vectorcam, and autocad running
Someone was saying qemu was darn slow.
not too bad for me, I got a darn fast machine :)
have you tried vector
sorry I wasn't reading right.
I have a 1gig AMD processor in the garage and a 3 gig Intel here in the house
who would I talk to to set up a new cvs thing?
connected with ethernet
awallin: How do you see us developing the material and tool cutting database for CAM.
awallin, have you checked this out?
[15:52:40] <DanielFalck> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl/emcinfo.pl?Cam
rayh: you mean a database of material hardnes, and a tool database with suggested feeds/speeds?
DanielFalck: yes, I've looked at most of the links, but nothing is really close to an opensource version of mastercam of surfcam or similar
awallin, that's true
rayh: I guess these databases would be built up slowly as people with real machines and experience contribute material and tool parameters
awallin, you are using python now right?
DanielFalck: yes, with wxpython for gui, and opengl for graphics. and the idea is to write cpu-intensive stuff in either c or c++ and call those from python
aside: in tkinter with Ubuntu/gnome, if the cursor is replaced by self.config(cursor="watch"), it doesnt use the stock tcltk watch, instead the gnome animated busy cursor
awallin, my idea would be to start with a good editor (master gui thing) that calls other scripts to create geometry/subroutines
have the text editor be the heart of the program
then have an opengl window called from it
ok... I'm not sure it should be a text editor, but scripting a la rhino/autocad is good
awallin, I kind of like the way ray and the guys did it with the tcl/tk program called 'cp'
but take it a step further
add a way of harnessing mark pictor's surfacing routine
and add the Aptos stuff....
?msg nickserv identify SnPa51n2
and the true type font stuff that Chris or Jeff did
be right back...
I guess it depends on your typical machining job. for 2D or 2.5D conversational might be ok, but for 3D moulds I don't see it as very helpful
aptos and mark's stuff is 3d
what I'm saying is, don't limit yourself to thinking you have to start a 3d project from scratch
there are already plenty of those that you might be able to harness
but most of them don't start with the machine tool in mind
and I think that's a mistake
the g-code is where the rubber hits the road
The machining process is why I asked about speeds and feeds. To me that is the core of a good CAM.
a good feed/speed calculator would be simple to do and very useful
The geometry stuff is not trivial but it is certainly a lot more available that speeds and feeds.
I have a simple bolt hole circle calculator that I did in python
and I think I've seen a cradek/jeppler version too
calling up something like that and a feed/speed calc would be useful
varkon is another good geometry engine that could be harnessed for our purposes too
it can do anything from simple 2d drawing to surfacing
and it has a scripting langauge built in
Gotta run guys. IMO it would be possible to start an additional module in the cvs.linuxcnc.org. Ask the board for something official.
there's always the question if learning a new (possibly complicated) system is as slow as making your own...
see you later ray!
but, if there was a master program like emc2 that could be built upon to accept other people's scripts...
that could be written in C/C++, python, tcl/tk....
yes, it needs to be modular so new developers can learn how one part of the system works quickly, and contribute to that part
trying to dig through some poorly conceived mess that you didnt write is much harder than writing your own poorly conceived mess :D
that's probably true
swig is easy enough.. basically you include the header file, run swig, compile, and link
you tried it?
it generates 20 pages of garbage that somehow connects the C code to the python shell
bits of which i can recognize from _gdepth.c
i guess python modules start with _
so it wasn't that hard to get running? I might try swig later
no it's super easy
did you try passing some data btw python and C?
gcc -fpic -c example.c example_wrap.c -I/usr/include/python2.4/ ; ld -shared example.o example_wrap.o -o _example.so
swig generates example_wrap.c from example.i
which basically just has the headers in it
awallin, can I try and dcc a small program to you? very small.
yes, but I'm not sure dcc works on my machine... try it
"waiting for transfer to begin"
it might be a firewall thing in my end?
might be on my end.
can you pastebin it?
how about a pastebin?
where should I put it?
[16:30:27] <DanielFalck> http://www.pastebin.ca/331152
it just sends output to screen
yep, it works
I had it set up as a macro in 'nedit' and I could paste the output to my text
did it just for fun
anyone experienced with opengl?
Twingy, jepler, and cradek are
I don't know twingy
* fenn vaguely remembers poking around with opengl ten years ago
Twingy wrote a CAM program : http://gcam.js.cx/index.php/Main_Page
and he was a developer for BRLcad
hi Twingy, we were talking about opensource cam programs
can you tell us what gcam can do?
make stuff using sketches and bolt holes
you extrude stuff
are you the only developer working on it? or is there a cvs repo somewhere?
I'm the only developer
this is not going very fast
it's all relative :)
Twingy; so essentially 2.5d only?
right, it's geared for 3-axis mill's and 4th axis stuff later on
it's primarily geared toward people that have a small home made or inexpensive COTS mill that they want to use to design stuff using free software
hmm you dont even need the .i file in some cases
going now, but I'll be back later
I saw the mindmap
Roguish_ is now known as Roguish
I frequently get null pointer errors and the like in mythbackend. I suspect it might be connected to the dvb card. It seems from the backtraces that its dvb_core that bugs out. is it fixable?
Sorry, wrong channel...
could be worse :)
python renders a 3d wireframe from .obj format http://www.3dartist.com/WP/python/tkinter3d.htm
I've fried a stepper driver today
what kind of driver?
* jmkasunich takes notes of what not to do
not remembering to set the current limit.
the stepper probably melted toghter and the output transistor blew..
along with the .50Ohm resistor.. and the trace on the board..
so really, you fried a stepper motor, and the gecko was just an innocent casualty
hm, it was my own fault for not setting he current limit..
how is the current limiter set?
a auxilary resistor..
lerneaen_hydra: you wire a resistor
well, I have a g340 and another servo
with a broken encoder..
err, I was thinking more like jumpers, flashing a program etc
anonimasu; err, you forgot to connect the resistor?
I were going to before I ran this test..
but I were going to try contouring a part..
with some more Z axis movement..
I found my psu issue
my rectifier is way too puny..
lerneaen_hydra: well, I were pissed the first 30 minutes..
as I've avoided Z movement before I've never had that much current going through the stepper.
and my psu has been dipping..
so, now that I had excess power.. :)
you must have been very lucky before
lerneaen_hydra: well, I thinkt the psu has saved me
to have exactly enough power not to burn the motor yet still enough to drive
though it's a good reason to move to a servo for that axis..
I just need to make/machine a gearbox..
or a bigger gear.. as the servo has higher rpm and lower torque..
images will be online later this week :)
of the victim.
Roguish_ is now known as Roguish
though it sucks to be without machine for now
03cradek 07v2_1_branch * 10emc2/configs/stepper-xyza/stepper_xyza.hal: don't display halscope by default
03cradek 07TRUNK * 10emc2/configs/stepper-xyza/stepper_xyza.hal: don't display halscope by default
your diagram looks nice
now only the implementation remains :)
I'm reading a bit on design patterns to get some sensible architecture for gui-opengl-model
I figured it's a good idea to get the basics down right the first time. That means it will probably take some time before I have any toolpaths to show....
but if the basic architecture is sound, it should be easy to extend into a full-featured cad/cam program
although Julian T wants to see cam algorithms integrated with emc. I don't quite see how this is going to happen... Julian: "No point in having yet another stand-alone CAM system that does all the usual"
how about for shops with Fanuc controls that are converting their computers over to Linux?
integrating CAM functionality in the controller (emc) is a much larger topic. I would be happy if we could create an open source standalone cam program.
stand alone seems a much better plan to me
programming in the back office, run the program on the shop floor
that's the right way to do it- to keep the machine running while programming
julian mentioned that toolpaths could be adjusted in real-time based on feedback from the machine, for example the step-over could be reduced if the cutting forces are too high
programming at the machine wastes too much time
we are always about 3 days ahead of our laser in programing
but that's pretty advanced stuff
that's good to hear from you, I imagine you are 'industrial' cnc users?
what does Julian T do? is he a CAM programmer with a commercial product?
at work and home too...
let me give yo an example from real life
a friend has a Haas Mini Mill
learnt to program it online on the machine console
DanielFalck: Julian is one of the guys behind the freesteel blog: http://www.freesteel.co.uk/wpblog/
pretty interesting reading mostly - if you have time to go through it all
has no offline programming
this forked fine in year one
he wrote stuff and ran it ..
now , hes got busier
a lot busier
he has stuff he wants to program and the machine is running
he has downtime when it shoudl be workingm because hes writing the next part ...
robin_sz: that's kind of odd, even ancient controls allow you to do it while running
he'd like to take time out to learn an offline package ... but hes too busy waiting for gaps in the machine time to program up the next part
yep, I understand this argument.
he tends to write a bit, air-cut it, write a bit more .. .
so at least in this channel people agree that an opensource cam package would be nice?
programming at the control is pretty inconvenient too. Lots of chances for 'fatfingering' and crashing.
well, I'm sure 90% of all users agree
laser is my main interest
and plasma to a lesser extent
milling and turning is mine
also, the issue of programming at the control is that it's hard to do complex shapes easily..
robin_sz: mine too, but I do micromanipulation at the nanoscale with lasers...
awallin, Ithink mylasers might be a bit too big for you :)
I would be interested first in 3-axis milling. later turning
robin_sz: 4W CW at 1064nm is the most power I've worked with. We have a 10 W CO2 in the lab, and some new 30W diode thing at 808nm also but I haven't worked with those
awallin, 800W Yag and a 1700W CO2
03jepler 07TRUNK * 10emc2/src/emc/motion/control.c: make only one pop-up appear when the "real-time delay" is encountered
another thing to think about in the debate is file control. If things come from a server, there is more control over quality/rev level etc...
if you have a bunch of guys programming at the control in a large shop- it's very bad
robin_sz: oh... nice. But I bet the beam-quality and pointing stability etc is not really research grade
the YAG is pretty good, but fibre delivered so spot size is not super small
but it does pierce 10mm steel plate in about 0.6 seconds :)
robin_sz: you should post some laser-cutting videos on youtube sometime! are your machines controlled by emc?
a little known fact is that molten steel is extremely transmissive to 1nm light ... it produces a waveguide effect, delivering energy deep into the work .. so pierce times are quicker than co2
Could anyone point in the right direction,i need to know how to compensate for different diameter tools in emc and dont know how?
Guest659: have you entered the tool diameters into the tool table?
you define your tools in the tool table, then use G41/G42
[20:10:56] <cradek> http://www.linuxcnc.org/handbook/RS274NGC_3/RS274NGC_38a.html#999268
but i might seems stupid but what next
^^ radius compensation mode instructions
or see chapter 20 of http://linuxcnc.org/docs/2.1/EMC2_User_Manual.pdf
no time to chat I guess!
03jepler 07v2_1_branch * 10emc2/src/emc/motion/control.c: merge rev 1.87: fix unexpected delay pop-up
03jmkasunich 07v2_1_branch * 10emc2/docs/src/config/emc2hal.lyx: fix docs to match code - jog wheels are now done in realtime, not halui
03jmkasunich 07v2_1_branch * 10emc2/docs/src/gui/halui.lyx: fix docs to match code - jog wheels are now done in realtime, not halui
03jepler 07v2_1_branch * 10emc2/debian/ (emc2-dev.files emc2.files.in rules.in): to decrease the mani package size, get rid of the HAL manual (everything in it was repeated) and move the developer manual to the -dev package.
03jmkasunich 07v2_1_branch * 10emc2/docs/src/config/emc2hal.lyx: index-enable is IO, not OUT
03jmkasunich 07TRUNK * 10emc2/docs/src/config/emc2hal.lyx: fix docs to match code, jog wheels are now done in realtime, not halui
03jmkasunich 07TRUNK * 10emc2/docs/src/gui/halui.lyx: fix docs to match code, jog wheels are now done in realtime, not halui
03jepler 07v2_1_branch * 10emc2/src/Makefile.modinc.in: -fPIC is necessary on some systems (x86-64) to compile external components
03jepler 07TRUNK * 10emc2/src/Makefile.modinc.in: merge rev 126.96.36.199: -fPIC; also, provide LIBDIR which is needed to build non-rt components standalone
Hmmmm are the commits suggesting I can just hook up a quadrature encoder on the parallel port directly and EMC can decode it?
yes, you can do that in 2.1.x
you don't need to run head to do that
you could do that in 2.0
jogwheels in 2.0?
there are speed limits of course, software can't count as fast as hardware
no, not jogwheels
so there isn't much of a reason to build a PICboard to convert a quad encoder to a USB HID device for a jog wheel?
you asked if you could hook up and encoder and emc could decode it, which 2.0 can do
ds3: that would make it much harder to use with EMC2
using the decoded value for jogging is a newer addition
the docs said jogwheel though ;)
cradek: how so? I'd have the PIC emulate a keyboard, like left and right arrow keys
I use a jogwheel hooked directly to the parallel port and it works perfectly
jogging with the wheel direct into emc is going to be _much_ smoother and more responsive than faking keystrokes
the idea of building that was before I decided to go EMC and fake keystokes works in more controls
and the arrow keys can't move the machine .001 without changing to incremental mode
pretending you're a user pressing keys is just not going to work well.
is there a recommended number of steps per rev for use with a jog wheel?
whatever you want
I only have 16 and it works quite well
and it's not steps per rev, it's distance per click/rev
had planned to gut a mouse and use the scroll wheel
I have 0.05mm/step and can reach 400mm/min without problems
oh I see what you mean
a real jogwheel usually has 100 for obvious reasons
unless I am looking in the wrong place, digikey encoders are like $20 a peice and I can get a mouse for a lot less
IMO if you have 100steps/rev then you really need detents
yes you do
right, I've used the wheels on a commericial control; just didn't know how well it'd work if I deviated
btw, wasn't someone making a servo-as-detented-jogwheel thing?
how many steps/rev does a mouse have? 50-100-ish?
think the datasheet said 16
you mean on the scrool wheel of a mouse?
or the ball?
fenn was talking about it
from the mice I've looked inside they've seemed to have lots more
lerneaen_hydra: you need torque feedback.. for that..
my mouse scrollwheel is detented, 20 detents per rev
this is in the $1 mouse I disassembled
bought it cuz a MiniDin-5 (?) connector + cable costed more then $1
jmkasunich; oh, scrollwheel, right
I was thinking ball-mouse ball encoders
anonimasu; why not just a nice big encoder on the servo?
those are gonna be higher, maybe 64 or 100 counts/rev
but no detent
yeah, that's what I was thinking
there is a guy that used a stepper for an encoder
stepper for encoder?, hmm, you'd need an integrator to get a constant signal, right?
I fiddled with a "virtual detent servo" but it really felt kinda crappy
my melted stepper dosent have natural dents..
or can you send a current through and check the position?
jepler: hm, I have a servo that will do that..
[21:16:37] <ds3> http://www.webx.dk/oz2cpu/20m/encoder.htm
(PID forces the commanded position towards a "detent", but weakly)
jepler; what's crappy?
lerneaen_hydra: it just didn't feel very pleasant to use
hmm, that's too bad
was it 1:1 or geared
anonimasu; what's the swedish word for tacky?
the type of tacky thats corny and 2nd-hand-flea-market-esque
lerneaen_hydra: that's the place you are thinking off..
I'm thinking of an adjective :/
I have no idea..
that's not a word I use often..
do you know what word I mean?
* anonimasu is crying over the burnt stepper
was the stepper worth more than the gecko?
about the same
the gecko died too.
there's a hole in the pcb..
one of the output transistors exploded.
and took the .50 ohm resistor with it..
through the board..
no big deal..
im not machineless for long..
going to order a new encoder for my spare servo..
and machine a beltbox..
still it feels shitty
and I did some cool 3d contouring at the time also
anonimasu: take a picture
I'll put it on my blog under magic smoke escape
jmkasunich: I cant get pictures out of anything tonight
jmkasunich: yeah, SET A CURRENT LIMIT RESISTOR ON YOUR GECKO!
hmm, does that mean that you're completely SOL if you get a short between the stepper output terminals?
lerneaen_hydra: yes.. completely
more like POP!
TO-220 transistors aren't big enough to BOOM
I'm just exaggerating.. :)
it felt like boom
we had a BOOM in the lab friday
IGBTs in a 300HP drive went
jmkasunich; the cap you had that blew, that must have been a shit-your-pants boom if you were next to it, right?
out teacher blew a cap at physics once..
a small one..
I wasn't around for that one, I think it happened in the field
the one on friday sure made me jump
I was 10-15 feet way minding my own business
jmkasunich; oh, too bad ;)
atleast I found the psu trouble
and I increased my accel for X/Y to 600
for big boys
for real men
yes that too
hmm, real men use purely mechanical control, or better yet, only musclepower
that would be quite a feat..
300 horses all turning a rotating whatchamacallit
imagine how much they must eat
how manu units/s is one G?
im slow as hell.
oh, that type of G
depends on where you are though ;)
well, machine acceleration
I wonder how much I can crank up the accel for my servo axes..
IIRC it's 9.82 in sweden/europe/up north and 9.81 in the US
I wonder if my machine would throw the table away if I set my accel to 9820
well the servo drive must have a max_current
and as current controls torque controls accel
hm, yeah I know
I'm pondering if it'd work..
I wonder what emc will do if it assigns an accel it can't acheive
hm, my geckos will fault..
that's what will happen..
? wasn't this a servo you were driving?
yeah, with g340's..
that's another matter
I think my issue is gearing/table weight
though I dont need 1G
I'll own a machine that will do HSM loops someday.
ever seen thoose little loops?
I cant find a video..
[21:58:17] <anonimasu> http://video.google.com/videoplay?docid=-923619711177709820&q=5+axis
at the end when machining the T
but that's a bad example..
emuge had a better video
so small machining in general?
very tiny movement..
with very high accel..
yeah it seemed to be somewhat trochoidal
that's the name..
but there's hsm loops on direction changes also
shouldn't it be called ham and not hsm?
[22:02:11] <anonimasu> http://www.emuge.com/news_events/imts06_videos.html
high accel machining
they use it for really hard stuff..
LH: for your question earlier, Rugldallur == @home Dallur == @work
Rugludallur; aha, ok
LH: been working on the table on day, hoping to upload some pictures tonight :)
err all day
but I fried my laser diode last night so I am unable to fine-adjust the sides
memo to self: don't mix up the negative/positive leads for diodes :P
03cradek 07v2_1_branch * 10emc2/debian/changelog: 2.1.0 release
03cradek 07v2_1_branch * 10emc2/VERSION: 2.1.0 release
[22:17:39] <Rugludallur> http://rugludallur.com/fileadmin/user_upload/100_5427.jpg
that one is from today, the whole things looks loads better after galvanizing
wow, that's quite a thing
all hail 2.1.0
cradek: remember to put pre-2.1.1 into Version
I've done this a few times :-)
this is a happy moment
cradek: I know :)
* Rugludallur raises a glass to everyone
Rugludallur: have any pics of parts that you have cut?
awallin: not many, I did very little cutting before tearing everything to pieces to have it coated
ok so you just got it assembled after coating?
awallin: still assembling, need another week or two
awallin: assembling the whole thing takes 80 hours or so :P
awallin: im encoding some videos right now, I will post link when I have uploaded
great, I hope to do some videos too once our mill upgrade is done... that migth take a month or two
awallin: if a photo says a thousand words then a video says a million :D
03jmkasunich 07TRUNK * 10emc2/configs/common/ (core_sim.hal core_stepper.hal): don't need to load scope_rt or halscope by default, they can now be loaded from the GUI menus
03jmkasunich 07TRUNK * 10emc2/configs/dallur-thc/ (dallur-classicladder.hal dallur-core_stepper.hal): don't need to load scope_rt or halscope by default, they can now be loaded from the GUI menus
03jmkasunich 07TRUNK * 10emc2/configs/demo_mazak/demo_mazak.hal: don't need to load scope_rt or halscope by default, they can now be loaded from the GUI menus
03jmkasunich 07TRUNK * 10emc2/configs/hexapod-sim/core_sim_6.hal: don't need to load scope_rt or halscope by default, they can now be loaded from the GUI menus
03jmkasunich 07TRUNK * 10emc2/configs/lathe-pluto/lathe-pluto.hal: don't need to load scope_rt or halscope by default, they can now be loaded from the GUI menus
03jmkasunich 07TRUNK * 10emc2/configs/demo_step_cl/classicladder.hal: don't need to load scope_rt or halscope by default, they can now be loaded from the GUI menus
03jmkasunich 07TRUNK * 10emc2/configs/max/max.hal: don't need to load scope_rt or halscope by default, they can now be loaded from the GUI menus
03jmkasunich 07TRUNK * 10emc2/configs/nist-lathe/nist-lathe.hal: don't need to load scope_rt or halscope by default, they can now be loaded from the GUI menus
03jmkasunich 07TRUNK * 10emc2/configs/ppmc/ppmc_load.hal: don't need to load scope_rt or halscope by default, they can now be loaded from the GUI menus
get out your wading boots, its flooding
03jmkasunich 07TRUNK * 10emc2/configs/demo_sim_cl/classicladder.hal: don't need to load scope_rt or halscope by default, they can now be loaded from the GUI menus
03jmkasunich 07TRUNK * 10emc2/configs/stepper-xyza/stepper_xyza.hal: don't need to load scope_rt or halscope by default, they can now be loaded from the GUI menus
03jmkasunich 07TRUNK * 10emc2/configs/sim/ (servo_sim.hal tripodsim.hal): don't need to load scope_rt or halscope by default, they can now be loaded from the GUI menus
03jmkasunich 07TRUNK * 10emc2/configs/univpwm/univpwm_load.hal: don't need to load scope_rt or halscope by default, they can now be loaded from the GUI menus
03jmkasunich 07TRUNK * 10emc2/configs/univstep/univstep_load.hal: don't need to load scope_rt or halscope by default, they can now be loaded from the GUI menus
here it comes again!
03jmkasunich 07v2_1_branch * 10emc2/configs/demo_sim_cl/classicladder.hal: don't need to load scope_rt or halscope by default, they can now be loaded from the GUI menus
03jmkasunich 07v2_1_branch * 10emc2/configs/hexapod-sim/core_sim_6.hal: don't need to load scope_rt or halscope by default, they can now be loaded from the GUI menus
03jmkasunich 07v2_1_branch * 10emc2/configs/lathe-pluto/lathe-pluto.hal: don't need to load scope_rt or halscope by default, they can now be loaded from the GUI menus
03jmkasunich 07v2_1_branch * 10emc2/configs/nist-lathe/nist-lathe.hal: don't need to load scope_rt or halscope by default, they can now be loaded from the GUI menus
run away, run away!
03jmkasunich 07v2_1_branch * 10emc2/configs/dallur-thc/ (dallur-classicladder.hal dallur-core_stepper.hal): don't need to load scope_rt or halscope by default, they can now be loaded from the GUI menus
03jmkasunich 07v2_1_branch * 10emc2/configs/ppmc/ppmc_load.hal: don't need to load scope_rt or halscope by default, they can now be loaded from the GUI menus
03jmkasunich 07v2_1_branch * 10emc2/configs/sim/ (servo_sim.hal tripodsim.hal): don't need to load scope_rt or halscope by default, they can now be loaded from the GUI menus
03jmkasunich 07v2_1_branch * 10emc2/configs/stepper-xyza/stepper_xyza.hal: don't need to load scope_rt or halscope by default, they can now be loaded from the GUI menus
03jmkasunich 07v2_1_branch * 10emc2/configs/common/ (core_sim.hal core_stepper.hal): don't need to load scope_rt or halscope by default, they can now be loaded from the GUI menus
03jmkasunich 07v2_1_branch * 10emc2/configs/max/max.hal: don't need to load scope_rt or halscope by default, they can now be loaded from the GUI menus
03jmkasunich 07v2_1_branch * 10emc2/configs/demo_step_cl/classicladder.hal: don't need to load scope_rt or halscope by default, they can now be loaded from the GUI menus
03jmkasunich 07v2_1_branch * 10emc2/configs/univstep/univstep_load.hal: don't need to load scope_rt or halscope by default, they can now be loaded from the GUI menus
03jmkasunich 07v2_1_branch * 10emc2/configs/univpwm/univpwm_load.hal: don't need to load scope_rt or halscope by default, they can now be loaded from the GUI menus
03jmkasunich 07v2_1_branch * 10emc2/configs/demo_mazak/demo_mazak.hal: don't need to load scope_rt or halscope by default, they can now be loaded from the GUI menus
* lerneaen_hydra drowns
03cradek 07v2_1_branch * 10emc2/VERSION: bump after release
guess that makes it official :D
* alex_joni lends cradek some extra bandwidth
I could use it!
dapper packages at the moment?
I notice the beta packages are gone
will those who installed the alpha or beta pkgs get auto-upgrades to the real thing?
ok, then I'll leave the laptop running a bit longer
jmkasunich: 2.1.0~foo < 2.1.0
2.1.0-foo > 2.1.0
you're so clever
not me :D
EMC 2.1.0 is released!
cradek has changed the topic to: Welcome! EMC (Enhanced Machine Controller) is a linux-based CNC control. | Latest release: EMC 2.1.0 | http://www.linuxcnc.org
darn.. too slow :D
* alex_joni thanks all that worked for this release
now I have to run :-)