wednesday we go to Capri
tomorrow tour of naples
a tour of niples?
the other would be better... but that's just me.
beer and a steak... in front of the computer... serious... as a guy,... the only other thing I could want right now... is well... maybe not right right now, later.
one penality for A-L-P-H-A
ahah A-L-P-H-A SPAGHETTI NO SPEGHETTI
K4ts: you have to explain me who is <K4ts> Trosi
ah ! you missed the name too
another penality for K4ts
you are happy giacus tonight?
not really ..
is that an offer to make him happy? ;) ;)
I try to be
if this is webcam based... can I watch K4ts? hahaha
nah.. K4ts isnt a cibergirl
think we do not use audio conversations ^_^
"cybergirl" you're negative one pts now too!
i go to sleep
with who? ;)
tomorrow I work
go for the tour or work ?
K4ts are you going to naples for wine tasting?
have a fun tomorrow
* A-L-P-H-A wonders if I tease people too much... probably... but it's fun
A-L-P-H-A: she already live in naples
I can believe she already has niples.
have a good wine right under her home :)
even you and I have niple.
red wine from vesuvio
don't know what vesuvio...
for heaven's sake... you can buy wine in plastic bottles in IT... that scares me.
A-L-P-H-A: YOU DRINK WITH ME?
K4ts... only if you're around my age group.
I'm 27 as of yesterday.
[00:27:20] <giacus> http://www.giacus.org/photo/vacanze/pompei/img082.jpeg.html
see the mountain in that pic ?
that's the Vesuvio vulcan
is that pompei?
no A-L-P-H-A i 43
hmm... maybe as friends then... not to get you drunk and take advantage of you. :) hehehe.
so... where are all the fossilized people from the village of pompei?
in a closed museum
got any photos of them?
no sorry ..
those are the excavated ruines?
many peoples died during they we're loving ..
maybe you know that
giacus? say again?
night night = nini
cc= ciao ciao
nini = sort for good night.
I thought cc was = yes yes.
or ci ci... however it's spelled
ciao is short enough. :)
why not stay, and keep us happy/entertained? :)
lets play a game... heheh... how long we can keep K4ts here to play.
so are you going to stay?
K4ts... do you play any instruments?
guitar, piano, harp, harth, violin, flute?
I wouldn't mind a nice bottle of wine, and some good company for laughs and good times.
so do you know her in real life?
got not much time her
yes, she's my gf
wait. wait. wait.
is it correct fg ?
have I see her?
in one of the photos? where I said that was a nice shot of her?
yes, she is
[00:37:23] <A-L-P-H-A> http://www.giacus.org/photo/vacanze/floridiana/img008.jpeg
can't see other many woman around here
yes, and the children is her son
the little girl is your daughter?
one of the daughter
how BIG is this cake? http://www.giacus.org/photo/famiglia/annamaria_9/img044.jpeg.html
they are 5
hehe not so much
I went an saw a client today.
they were a bunch of wide boys
he makes wedding cakes.
DAMN wedding cakes are EXPENSIVE!
wanna try with a cnc ?
hmm... maybe for icing the cake...
that would be NEAT.
having a pump to maintain constant pressue on an icing bag... to squirt out icing.
in a given pattern.
the guy just looks... gay.
pretty funny photo. hehe.
whats the problem with the gays ? :)
I've some friends gay , good people
nothing... unless he's noy gay, and looks that way.
then people could get the "interesting"/"wrong" idea about him.
unless, he's likes to be called a 'her'
A-L-P-H-A: you should quit while you're ahead (?)
but it's fun.
people might get the "wrong idea" about you
sorry was watching edmonton being beaten to death. :(
okay... problem... no one said being gay is bad. gay's fine... to each their own. but when you're not gay, and look it... that's wrong.
* cradek bites his tongue
does it hurt?
A-L-P-H-A: What, you've never heard of Metrosexuals before?
anybody have a good greyscale image that is suitable for engraving?
one with variable depth, not just two depths
ooh, I'll try this .. http://www.philc.net/images/SampleDepthMap.gif
it makes a ~5000 line file with this improved image-to-gcode.py
(resized first to fit within 300x300 pixels)
line simplification algorithms are my favorite new hammer
I can use them on anything
03cradek 07lathe_offsets * 10emc2/src/emc/rs274ngc/ (6 files): checkpoint of lathe tool offsets. currently radius comp in XZ is working for lines only. needs lots of cleanup.
yeah it's a good start I guess
I'm afraid I made a mess of it, but maybe I can clean it up later
[02:29:40] <jepler> http://emergent.unpy.net/index.cgi-files/sandbox/image2gcode.png
the problem is that image-to-gcode doesn't have the nice stuff like cutter shape compensation
yeah, it probably needs that to do a good job
that looks like a decent result though
depends how much relief you're cutting and the tool size
you could add image support to toolpath - just load the image right into the z buffer
no idea how you'd display it on the screen
with 2*m*n triangles
so this isn't a new idea I had?
or maybe just display it
can't gl put an image on a quad?
03jepler 07HEAD * 10axis/scripts/axis.py: rework filters. Use new [FILTER] section, and pass the input filename to the filter. have a different filter for each extension
03jepler 07HEAD * 10emc2/configs/sim/axis.ini: example filter section for axis; you probably don't have these programs installed yet
yes, you could texturemap a quad
* jepler computes that the average line in this generated file is about .11 inches
03cradek 07HEAD * 10emc2/nc_files/comp.ngc: very basic cutter radius comp test file
that seems pretty long, that's nice
yeah but there are a bunch that must be about .7 or more
huh, I never noticed before that my laptop has an emulated scrollywheel on the right of the touchpad
(or do you hate that?)
no, it's fine, if I've never noticed it means it doesn't trigger accidentally
jepler: A self-portrait?
fresh off the 3d scanner
the knobs on the toolchanger are called retainer knobs right?
wanna see a nice emc machinist in action ?
[12:05:23] <giacus> http://www.giacus.org/files/gnutux_emc.avi
im at work :)
ordering linear rails :)
yeah, it seems blender can do nice things with animations
also could be useful to make small movies showing how emc do axis homing
I just designed a algorithm to make toolpaths off drawings
building a cutting machine :)
how it work ?
tried it ?
going to do it tonight
cant use emc for the control as it's too complex :)
ah well first thing is to make coordinates out of shapes
going to try it tonight
* jepler starts adding a GUI to image2gcode.py
wow - that's the opposite of what I'm doing
jepler: how do you intend to use gui programs as filters in axis?
damn it... a 2hr drive to waterloo and back... that is my day. :(
asdf23r is now known as A-L-P-H-A
there was a power outage in my basement^Wdatacenter. the power is back but the CVS server didn't start automatically. I'm looking into it now.
jepler: I think nobody but me had noticed
* alex_joni didn't
jepler: how do you intend to use gui programs as filters in axis?
cradek: it has to be a program that will write its gcode to stdout and exit when it's done
argh, my girlfriend^Won-call engineer isn't at the house to resolve the CVS server problem
I'll have to go myself
I'd say that it's reasonable to use a GUI to load a file, and be presented with a GUI that lets you choose how that file gets machined
I doubt there's any hurry jeff, everyone's at work
graphically choosing things like the height scale for a bitmap (and seeing a preview) would be most excellent :)
SWPadnos: I might suggest some software I tried, but I'm afraid I'll motive jepler less
then please don't ;)
SWPadnos: it's not GPL software, so I don't think it should even matter
I have stlworks, which is relatively OK
I tried some 60-days limited version of MeshCAM which seems nice
but only runs on Windows (though I haven't tried wine)
it does STL & bitmaps
oops! s/at the house/in the datacenter-complex/
CVS server is up again
any idea why it didn't start?
alex_joni: yeah, because it doesn't start after the power's been off
so crappy BIOS..
it's not exactly server-grade hardware
I think that's the same for dsplabs ;)
to have ATX and not implement that function in the BIOS is a waste imo
this isn't any ATX though
it's a laptop, and maybe even from the pre-atx era
oh, then it's understandeable
btw.. when I had issues with wireless networks, I was thinking of putting a Nut-device with some relays there (to cut/restore power to stuff)
I have a USB relay on my dsl modem which sometimes requires an external reset
CPU: Pentium II/Pentium II Xeon/Celeron (300.68-MHz 686-class CPU)
haha stand back!!
cradek: good enough ;)
I guess that means it doesn't require a fan running all the time, which makes it more reliable
jepler: I can't figure out how, in axis, to cause the tool table to be reread
cradek: hm, I bet you can't.
have you UTSL yet?
why would I do that when I can complain instead?
it looks like it uses 'tool_table' from the stat buffer
well, I guess now is a good time to reboot. sonce the power went out, and it's still hailing.
see you all later :)
Anyone heard of a wirewrap tool/gun for 22ga wire?
Nm, just found em... telecom usage
[17:30:54] <jepler> http://emergent.unpy.net/index.cgi-files/sandbox/image2gcode.png http://emergent.unpy.net/index.cgi-files/sandbox/image-to-gcode.py
Jymmm: umm if you want to think so
jepler oh I didn't look.... thought it was the one from last night.
that's very cool... how did you generate the image?
nope, it's different
I found the image on google image search for "depth map" or "depth image"
what I'm showing off this time is the GUI for the image->gcode conversion
is that the one that chris wrote?
not that it's perfect
another tooling adaptor finished :)
it started with the one that chris wrote, but it's now much more featureful
5this time for a facemill
it does tool shape compensation and path simplification
jepler is is primarily greyscale to z-depth conversion?
and integrates nicely with CVS versions of axis .. you can just open a .png file in axis, and the conversion window pops up
yes, it treats the greyscale value as z depths
jepler: BIG SUGGESTION (if easily done)... estimated running time
the cruder the estimate, the easier
roundign off to the nearest leapyear
I think 15 minute increments would be fine
or is that too much?
well it's easy to determine the XY distance moved
jepler what is "y step" ?
it's harder to determine the Z distance moved before actually generating the output
Jymmm: how many pixels in the image does it move down each time
like skipping a few pixels?
if you are doing a 500x500 image in a 2x2 square, and enter 4 for y_step, it will step down .016 inches each time
because each pixel is 2/500 inches
so it controls how close together the rows are
hm, something about the shape compensation isn't behaving like I expected..
Just strange to see it in only ONE axis adn the output would be "stretched", but I've seen it in other img2gcode apps too.
The X "step" is variable, depending on the complexity of the image in that area. That's set indirectly by the tolerance value
but just because your image is 500x500 pixels, you don't necessarily want to scan across it 500 times. you might want to cross it 250 times (y_step = 2) or 125 times (y_step = 4)
y_step doesn't distort the shape -- a circle will come out a circle, not an oval.
Yeah, i was just thinking that you would lose resolution in the one axis is all.
of course python
aha, just a sign error
it would have been correct if there was any such thing as a negative-diameter ball-end mill
jepler hey now, not breaking the laws of physiscs now, wait till next year
I think there would be a lot of good things about a negative-diameter ball-end mill
it adds material =)
or make (
but you would have to mill around
jepler nice job btw =)
has anyone built a SLS with emc as control yet?
* jepler googles "SLS"
"A layer manufacturing technology in which the layers are formed by using a laser to bond the surface of a bed of powder material in the desired shape." ?
'santa clause machine'
laser sintering ;)
cool stuff :)
sounds like a modeling machine
Jymmm: never heard about them?
rapid prototyping machines
jepler mind if I share that screen shot in another channel?
(my sample depth images I found on google image search: http://images.google.com/images?q=depth+map)
[18:06:16] <anonimasu> http://www.materialise.com/prototypingsolutions/laser_ENG.html
Jymmm: sure -- what channel is that?
jepler some of the folks in ##php
I'm ragging them on python =)
I don't think I'll join you
nah, not a good thing to do =)
"numarray" makes Python great for this type of thing. You can juggle 1000x1000 floating point matrices just as easily as you can add 2+2
for instance, these two lines of code perform tool shape compensation for one position on the image:
m1 = image[i:i+tw, j:j+tw]
d = (m1 - tool_shape).max()
take a rectangular portion of the image from i..i+tw in one direction, and j..j+tw in the other; then subtract from each element the corresponding one from the tool shape; then find the biggest element in the result
jepler: very nice work
anonimasu pretty cool (the link)
alot better than what I was talking about
clearly I need a new digital camera. http://www.theinquirer.net/?article=32520
jepler: The sad thing is that no place develops prints > 5MP
even big ones ?
etla: if you wanted 16x20, it won't be a print, a poster maybe, but not a print.
maybe some of the commercial houses might, but none of the common ones.
The largest I can get is 8x10 on Fuji Crystal Archive paper
but you do get nice effects with a big sensor + lower noise (I have a canon 20d)
I have a Nikon D70, and I can get larger, but not on archival paper
these people go up to 20x30" with a photographic process: http://www.ezprints.com/ezdirect/products.asp?Type=Print
I've never gotten anything that big from them
my camera's a 6-megapixel 300d
etla: The ONLY problem I have is they don't say what paper they use.
I don't have any of their prints not in frames or I'd see
I think it may be fuji paper
"laser clock": http://www.youtube.com/watch?v=pIwrkqQeYcs
calling them now...
[18:31:28] <jepler> http://www.hackaday.com/entry/1234000860073758/
You're right... they do use Fuji Crystal Archive paper for small and the large formats... VERY COOL!!! Would love to find some place locall y that does .
until you do .. give these guys a try
FWIW... Walgreens uses Fuji Crystal Archive paper, $0.19 for 4x6 online, and they'll store all your images indefinatly as long as you buy something once a year.
AND you can dl the original full resolution too if you want to.
I love the fact that I can have a print made at 3am or even have it processed in a store 2000 miles away
ezprints expires the uploads after awhile .. a few months, I think
Well, I'll use them for large format. But check out walgreens.com it's pretty damn good.
pictures of my conversion will be up soon
grabbing them now
jepler: you've used EZ Prints a lot?
Jymmm: I wouldn't say "a lot". I've been ordering prints from them for 4 years or more, but I only order a few prints a year.
jepler: no problems? quick srvice?
Jymmm: I don't remember any problems
I've gotten prints that were darker than I intended, but then I never followed their calibration process either
hiya folks. how are things?
they seem good from here
[19:23:11] <jepler> http://www.youtube.com/watch?v=y9c3V0q8cgk
if you hate automatic transmissions and ABS, you'll hate this.
[19:24:09] <jepler> http://emergent.unpy.net/index.cgi-files/sandbox/image2gcode.png http://emergent.unpy.net/index.cgi-files/sandbox/image-to-gcode.py
SWPadnos: I've been showing off this: ^^^
a big upgrade of cradek's old script, with GUI and integration into AXIS 1.4
hey, a gui
oooohhh - cool :)
does it do tool shape compensation now?
and path simplification like we (I) added to toolpath recently
the same way toolpath does it?
I'd fiddle with it, but my house still has no power
hmmm - I wonder if that python code will work on the Windows python interpreter ...
SWPadnos: yes, it should. it requires PIL and numarray, but I think those have windows versions available
well, maybe I'll try to find out
I can't get any work done anyway ;)
[19:30:13] <cradek> http://youtube.com/watch?v=veIU0Jwu54w
ok I'm sure everyone but me has seen this already, but it's still pretty funny.
jepler: that volvo should call up the insurance company so they can raise my rates, if it senses I would have plowed into the other car
I hadn't had the courage to watch it yet
that's pretty terrible
I think one of them is about not eating shellfish
can you think of ... ANY BETTER building?
jepler: can toolpath/image/etc be installed next to axis without any confusion now?
jepler: I assume both toolpath and image use zbuffer.so, but that's probably ok?
image, yes. it is just the one .py file
image doesn't use zbuffer.so
did you find a better way to do compensation?
oh just with numarray I'm sure
most of zbuffer is polygon scanconveresion
one part of it does speed up the repeated tool compensation operation, but I don't know how big a difference it made
it was nice to write this without any custom "C" extensions
pixel size is the output units per pixel?
that looks really great
I was hoping someone else would write us some filters, but I guess you had to do it
maybe if there are a few, it will get people excited about the capability
for me, a toolkit or documentation about integrating into emc (/ AXIS) would be very helpful
(I'm excited about the capability, FWIW)
the filter outputs gcode to stdout, you specify the filter's name in the ini
and sim/axis.ini shows the ini items necessary for filters
the "toolpaths" one doesn't actually work, because that software doesn't write to stdout
is the filter passed the filename on argv?
sure - that part's easy. but getting a UI that looks similar to the others (as they're created), and things like 3D preview, will be common necessities
SWPadnos: yeah, it's the same problem faced by gimp plug-ins
oh, that kind of thing
eventually they've developed some common stuff like a preview window ..
that's a very different story
obviously, the more lazy a filter writeer can be, the more filters we'll get ;)
seems like hacking up an existing one will be a good strategy
that's sure where I'd start
well for output there's this "gcode.py" (which I just inlined in image-to-gcode.py)
unless I knew some other toolkit well
a "preview" mode could be added to it
presumably using tk+minigl.so
axis is a pretty good previewer
sure, though the filter needs to be finished with the conversion before you get a preview in AXIS
jepler: we really need to tackle REALIZE don't we
cradek: that would be another nice one
SWPadnos: yes if it's slow, that's not going to be so adequate
jepler: I think it won't require a gui at all
it takes several seconds to a minute to generate the g-code in i-t-g .. so while you can hit ctrl-r (and get the same settings as last time as a starting point) it's not fast
jepler: after all, it doesn't have one now
but I don't know how you'd do the preview faster
how do you get the same settings?
cradek: REALIZE is different than i-t-g because it doesn't have an input file per se
jepler how slow are we talking?
cradek: I pickle the settings
jepler: sure it does, the dxf
hmmm - can we design a near-minimum set of functions / utilities that a filter writer would need, with the hope that such a library will be implemented if there's a spec? :)
cradek: oh, forget it -- I thought you meant ttt
hey, another nice candidate
jepler: cradek is volunteering you for everything =)
* cradek gleefully sees jepler finishing all his projects
jepler: see what I mean
err - work projects or emc projects ;)
* Jymmm is surprised that cradek hasn't asked jepler to paitn the house or do the laundry (yet)
Jymmm: e.g., 39 CPU seconds to convert a 1024x1024 image to g-code using y step 4 (so 256 scans) and various other settings
jepler: Hey that's fine... anything under 4 minutes I think is pretty good.
are you hitting disk, cache, or memory limits there?
it's surely cpu bound
SWPadnos: CPU bound
the tool shape compensation
ah - ok
what CPU is this?
shape compensation or path simplification
SWPadnos: 1.5GHz mobile intel
do I hear any votes for a 200kHz 6502?
286 in TURBO speed at 12MHz
ok - that's a pretty quick CPU
jepler Hey, level my Apple IIe out of this!
even the Bell&Howell apple was ~1 MHz
you guys may drool over i-t-g while I'm gone
* Jymmm hands SWPadnos a mop bucket
I'm at a tea shop - they'll clean up after me
oh, well carry on!
the wifi seems slow today
so does my dsl
SWPadnos: in other news, I have radius compensation in lathe mode (XZ) almost done
SWPadnos: after that I'll try to figure out gouge protection
any chance of that translating over to mills as well?
did you see all the info LH wrote up for me?
lots of good stuff there
I think there's no parallel for lathes, since the tool is round
ok. I"m probably thinking of something else entirely anyway - more anti-crash I guess
with a lathe, a tool can cut only in particular directions
right, dependent on which orientation is set
I haven't thought through all the details I'm sure
just radius and XZ offsets will be a big step
I'm trying to keep the interp compatible, that's going to be the hard part
so far my plan is to expand the tool table with optional fields that specify the extra things a lathe needs
like a beer holder
are there sane defaults if the optional fields are blank?
with vibraton dampening, so your beer doesn't go flat.
yeah, and a chiller to keep it cold
now you're thinking.
actually.... I wonder if you connect an AC pump to the spindle motor
SWPadnos: the extra fields will be things like allowable cutting angles - if blank, don't restrict the direction, like a mill
ah - ok
ok, co where do we put the condenser to prevent swarf from silling it up
also tool shape - defaults to round (specified point is the center of the tool)
hmmm - center of the tool, like a ball end mill?
btw, lathing graphite is messy.
just so you know.
bill203 and icky in beer too
then the beer holder needs a cover.
warm ! darn weather
I think we can justify an additional three axises for the purposes of a beer cover.
are coocking at evening
err... automatic beer cover
a BIG cooler
it probably be far cheaper to buy ones of those new robots I've seen on TV lately to just fetch us a beer
homemade beer ?
giacus: No, a lathe with built int beer holder/chilller and automatice cover
not really if you get a good strong beer from germany
I like du demon, really its french
mission completed http://www.giacus.org/en/emcmania.html
feel free to use as you like, they're copylefted :P
I stop here .. too much warm
giacus: You SHOULD copyright, then license as you see fit. I've read where someone abused something and the create was liable.
whats the right term ?
a creator isnt god here :(
you have on the page copyright, buy you just said copyleft
make ? made ?
you == creator
and who otherwise ?
did you read the license on the link ?
nope, just waht you said 30 seconds ago
Jymmm: but copyleft do not mean public domain
can't understand what you mean ..
maybe 'creator' is wrong ?
"creator of works", it's a generic term for author, artist, composer of music, chemist, etc
ah, ok, not here
I'll change it
ther person who created the works in question.
the it equivalent is creatore
giacus es El Diablo
well i'm not sayng I created emc
but the images you see, yes
we talk a bit yesterday about this.. very annoyng
I could apply the CC license
but to be honest.. I hate it
Free Art license its ok
anyone here use open office?
hello A-L-P-H-A :)
I use emacs, but sometime oo too
A-L-P-H-A: yes. I use OOo
SWPadnos, know how to use a datasource, to populate a label template?
I spent a while trying to get it to go.
never done it before, but I'd say the label wizard is the easiest way to start (assuming OOO 2.x)
I'm using 2.0.2
unfortunately, I have no data sources set up, so I'm not sure I can easily try it out
SWPadnos, I could send you one... CVS, or odt. or something
well. it looks pretty trivial, if the sample data source is ainy indication (the biblio database)
no it's not trivial.
i spent 2hrs looking like an idiot at my clients, while I was trying to figure it out.
the documentation sucks arse.
well - I'll play fo r abit, and let you know what I find
cradek: how goes the lathe programming?
radius comp in XZ works for lines now (not arcs)
arcs are just more of the same, slightly more complex geometry
any idea of how it will compensate? as the result path isn't an arc will it divide to small line segments?
or does the interpreter support curves?
no it will do the same thing it does for mills now
there's a picture, just a sec
[21:04:22] <cradek> http://www.isd.mel.nist.gov/personnel/kramer/pubs/RS274NGC_3.web/RS274NGC_38a.html#999413
a line is translated to an arc (around the corner) and then a parallel line
an arc is translated into an arc (around the corner) and then the concentric arc
ooh, roll over is good
lathe will be the same except the cutting angles are restricted
high speed, less stress on machine and faster times
so currently just line -> arc,line is working
I was thinking, an a g41 in a mill goes from arc -> arc
On that pic... top-right corner of the "interptripter does it thisway" you think that's a artist booboo?
but isn't that behavior different in a lathe?
yeah those are tangent
I mean, emc makes them tangent of course, the pic isn't great
the path (if it in an arc) isn't concentric at least
ok, just had to ask =)
Lerneaen_Hydra: it is if the control point is the center of the tool's radius
oh, center is different
I was thinking tool position 3 or similar
[21:08:25] <Lerneaen_Hydra> http://wiki.linuxcnc.org/uploads/cutting_orientation.png
yes but imagine a circle inscribed in the tool radius
with the center of the circle being the controlled point, now the compensation is just like a mill
you just have to restrict the cutting direction for gouge protection
yeah, so emc will do the little math that's done to convert from pos 3 to what would be 0
cradek: by direction you mean the plunge angle?
yes the user will specify the origin in your photo, but emc will use the center of the radius
yes I guess plunge angle
cradek: oh, ok
I haven't thought through all those details yet
cradek: if emc does that then there shouldn't be that much to change
you don't know in gcode which is a "plunge"
so I think you have to restrict the allowed directions of movement for g1/2/3
cradek: yes, so it doesn't go over the trailing or leading angle
however, in some cases you want to disable that functionality
but for g0 you allow any motion right? you have to be able to back out of a cut
when doing a face turn for example
cradek: yes, I guess g0 should be free
for facing tool 3 can cut straight "down", why do you have to disable it?
if you have a tool type 3 (the type of lathe I'm thinking of, the type I mainly use) has the tools behind the spindle, so when I say three I should actually be saying 2) and you define a trailing angle, it should stop you from diving deeper than said angle. doing a face turn however would be an invalid move, even if the tool is made to do that
hmm after looking at the picture some more I don't know what angles are safe to allow
the difference seems to be which side of the tool you intend to cut with
why not specify two tools with different allowed angles?
in the lathe I've used trailing/leading angle compensation is only used with canned roughing/finishing cycles. support for manual code would be good too though
you would use one for facing and another for turning
need to have a primary plunge direction vecot
cradek: hmm, that could be done, I guess the g-code bloat would be worth it
i used alot of iscar groove turn tools...
2 offsets.. and primary plvectors...
dmessier: Looks likes your stalker has found you... RUN! RUN FOR YOUR LIFE!
need a hug.. ; )
Lerneaen_Hydra: obviously I need to think about this some more :-)
Lerneaen_Hydra: or, I might say the spec is incomplete and leave it to you to figure out!
cradek: WE have confidense in you =)
you have edge cam right??
Lerneaen_Hydra: at first I will not have gouging protection, only radius comp and XZ offsets
cradek: okay, I'll think some more ;)
check out the lathe module tool setup
cradek: the best solution (although I think this is far out of the current scope of emc) is for one to describe in the start of the file the dimensions of the billet, and where X0 Z0 is located. then in the preview window one can see the billet (semi-opaquely) and the toolpath
maybe even a tool shaped in the same way as defined per tool-type and leading/trailing angle
ive seen it ALL done in gcodes..
um yeah I don't think we're going to have that soon.
parametric tool display from gcode ddefinition
cradek: It may be a possible thing to add later though ;)
arent we just ALWAYS making a wish list???
no, some of us are programming
and some days.. someone makes it so.. like YOU
anyway, if there is an easy way to enable/disable gouging/diving protection (an unused g-code) I think that would work
yes that should be possible
doesn't really follow any other specs, but they all really suck in that respect
I'm worried about being incompatible with cam programs (that I don't have)
like only support when doing canned rough/finish and so on
cradek: typically (afaik all) cam programs output moves as G40, that is to say pre-compensated
canned cycles from cam is always a best wish from any good post maker.. dont sweat it
so all this tool config is also in the cam
the cam takes care of gouging protection
so emc is perfectly usable with lathe/cam already
the only thing that the cam may want to use is g42, if only for readabilities sake
just touch tools and r& R
I've made several parts in edgecam with it
that video I uploaded for instance
no comp in cam...
dmessier: what do you mean?
point to point..lines and arcs
pre compensated for toolnose rad..
you mean that cam doesn't compensate for the radius?
oh, all the cam I know of does/can
yes it does..
but YUO cant change tool nose rad at the machine without reposting it
Microsoft OLE DB Provider for ODBC Drivers error '80004005' </edgecam.com>
where i could the other way... its all about how YOU wanna work..
dmessier: afaik you can set up most cam programs to output G42 and the desired contour
I wanted to check the price on edgecam, but their web site gives an obscure microsoft error
alex_jon1 is now known as alex_joni
cradek: oh, ok. It's rather expensive
that's all I needed to know
15000 or so canuck bucks
for any capability
alpha.. call back..
cradek: are there many unused gcodes?
Lerneaen_Hydra: a whole infinity of them
G1xx -> Gxxxxxxxxxxxxxxx
oh, over 100? are those acceptable to use?
acceptable for who?
for... the one that wrote the spec ;)
in all seriousness, it's just that I've never seen a gcode over G98
we've had g501's on some machines
Lerneaen_Hydra: in all seriousness, there is nothing that can stop you from using one
and the specs are pretty vague
even 30 years later ;)
I mean they don't cover all of it
probe macros were g'codes...head orientations were as well with plc inputs for feedback
so there's no spec that says "...and it was decreed, and hued down in stone, that there shall not exist any gcodes with a number greater than 99..."?
hm is that really important?
nope, and M-codes already exist up to M200
if there's no space for g-codes you need to keep adding them if needed
you cant stop adding functionality because some _old_ spec wont allow ti
* alex_joni saw some really old specs today
series 1 g codes are pretty defined... 2 & 3 as well... but the word is your oyster in factory installed options
ther interpreter as it stands can't go above g99.9, I think
oh, one thing I was thinking of, is there any plan to "split" or seperate the lathe and mill specific code from each other? so a Gxxx would enable mill mode and Gyyy would enable lathe mode in the interpreter (tool positions, compensation, and so on)
hm, that'd be a nice thing
keep it separate..
<- these ones;)
it gets multiplied by 10, then truncated (if I remember correctly)
it seems that things could get confusing otherwise
SWPadnos: not truncated
alex_joni: old specs ;)
Lerneaen_Hydra: I hope that's not necessary, and it can all be decided from the tool table
Lerneaen_Hydra: a config is either for a lathe or mill, not both, so there's no need to have gcodes to switch modes
that's right :)
I thought the multiplied number was converted to an int
cradek: a config? you mean the main .ini?
in emc2 a config is the set of files in the directory
so actually we have G010 (instead of G01)..
G999 for G99.9, and so on
or, G1 is actually item 10 in the command table
oh, stuff not seen in the normal userland
yeah, 999 for G99.9 ;)
technically, we could hide some nice stuff under g0.x ;)
how much lathe functionality is there?
but you can take each G-code and append .x (where x = 1..9)
anonimasu: threading works,
sure, but the command (code or lookup - I don't remember) uses an int as an index
anonimasu: threading, radius compensation is half done
well, regardless of how it's done, it seems that mixing lathe and mill can cause issues, so as long as there is an option to choose one or the other, then that is good
Lerneaen_Hydra: it might be part of the ini: lathetask instead of milltask (if that's ever needed)
Lerneaen_Hydra: I will not require separate compiles for mill/lathe, but some other means will probably be necessary
is the ui different when you run a lathe?
or will it be?
AXIS already has a lathe mode
anonimasu: which one? :-)
AXIS does ;)
doubt that any other will be expanded..
hard to say
do we have a feedhold button yet?
not sure how much different they need to be
maybe axis is the way to go for a ui :)
you can feedhold with hal now
it's actually adaptive feed: 0.0 to 1.0
* anonimasu cant wait to get the mill togther
emc is turning more and more into a real controller :)
ok guys.. off to bed
night alex :
cradek: different compiling is (IMO) way out of the question, I was thinking more of a bit in the main ini set to mill or lathe, to choose the functionality
yes, I'll do that if necessary
it's not necessary yet
radius compensation is based on the active plane
for a lathe you'll always be in XZ (g18)
yeah, if need by
a canned roughing cycle probably wouldn't be all to good in a mill if it's made for lathes
just like drilling canned cycles aren't appropriate for a lathe
well maybe some are, I don't know, you could use them to center drill etc.
but live tooled lathes only need g-code for tlaxis vector selection
we'll stumble over that bridge when we trip over it
cradek: you can also have lathe/mills
drillind canned cycles are AWESOME on a lathe...
they're lathes but three axes of movement and prescice controll of the spindle
I think you could use cutter comp on a lathe/mill right now, switch with g17/g18
face groove... or od groove
dmessier: and not too hard to implent too ;)
ive used it since the fanuc t6
ok I have to go now, bye all
g91 g71( if i recall) x,z,l,r
dmessier: I'll be glad if I dream anything that doesn't have to do with work now... so mostly anything will be good ;)
packing bearings is not exactly fun
Lerneaen_Hydra axel grease?
if you can chk out the siemens powered lathes by DMG ( deckel maho guildemiester)
not really, it's actually just the rollers for roller bearings (SKF)
have an AWESOME front end that is a dos application...
dont ask... i found out.. DR dos
what front end?
im trying to remember their trade name
turn plus... i believe...
no that was a heidenhein front end driven off g-codes too
can't find it on google, no biggy.
chk out the heidenhein.. it was easier to operate...
Heh, this looks like it is from 1950's NASA command center... http://www.quinx.com/images/hh415.jpg
[22:06:12] <dmessier> http://www.dmgcanada.com/us,turning,ctx
.... might be close
its a heidenhien tnc control...420 by my best guess
currentl set for network link.. i believe
nice handy operator control... has balls if unlocked... i used it on a 5 axis nutating table... High school in montreal...must be nice
someone had to crash test it... ; )
i autographed a fixture... ; )
out of the gooness of your heart
* robin_sz wonders what a 5 axis nutating table is
rock, sway, or nod; usually involuntarily
[22:28:37] <jepler> http://numeryx.com/cnc/nutating.htm
Univerally... sometimes the x axis was pointed in the rhubarb...
pointed in the rhubarb? http://www.urbanext.uiuc.edu/veggies/images/rhubarb.jpg
NOOO.. machine looked like a vertical.. had a nutating table on a 45 degree knuckle ALLA baove mentioned head.. but it held the tabl.. over a vertical spindle and spun the coorinate system
coord moved with tool axis vector...kinda screwy on a 5 axis swarf cut
tool doesn move... table dances under it... frame of referance is ODD
seem to move..
mmmmm rhubarb crunch
Dallur2 is now known as Dallur
Anyone have any clue on the spindle_forward / estop problem ?
:( my message never made it through it seems
I lagged out
I connected a signal to the spindle_forward pin and if I do that it seems that estop no longer affects spindle_forward
nope, nothing from you before that
anyways I need to detect when the spindle is turned on to initiate a homing sequence for the plasma torch
I know abort turns off the spindle, I think estop does too, are you hitting estop (f1) in the gui, or is it external?
but when I hook up the signal to spindle_forward it does not seem to work
the estop is through classicladder
but it reset it perfectly before
before hooking up a signal to iocontrol.0.spindle-forward
and I checked the config before I did hook up the signal and I noticed a funny thing
iocontrol.0.spindle-forward ==> iocontrol.0.spindle-forward
is in there ,, so I was thinking there might be some sort of wierd hack there to make estop reset the spindle_forward
newsig SpindleFwd bit
linksp SpindleFwd <= ppmc.0.dout.00.out
linksp SpindleFwd => iocontrol.0.spindle-forward
here's an example from univpwm
well I think the arrows are backward
but they don't do anything
are you saying hooking a signal to the output from iocontrol changed your estop behavior?
estop does not reset the spindle_forward anymore
did you hook it through classicladder?
estop was hooked up through classic before
and I just hooked up the spindle_forward through classicladder
I don't know how to say it nicely, but I bet you messed something up then, it's easy to get confused with the classicadder inputs and outputs
I just found it really funny that before hooking anything up to iocontrol.0.spindle-forward and having nothing in my config for iocontrol.0.spindle-forward there was still a line when I did halcmd show which showed iocontrol.0.spindle-forward ==> iocontrol.0.spindle-forward
if you use linkpp, it makes a signal with the same name as the pin
that must be what you're seeing there
sounds like a linkpp to me
ok makes sense, I did not realise that ,, thanks
do a show sig iocontrol
welcome, good luck finding it
it should show you what (if anything) that is connected to
* cradek looks at arc compensation and sweats
you have my sympathy
(but not my help, I haven't a clue what you are doing)
thanks I guess
some of the interp is just crazy
int turn; /* 1 for counterclockwise, -1 for clockwise */
why is that so crazy?
why not int clockwise; /* 0 or 1 as is so bleedin' obvious */
maybe they multiply by it
positive angles are counterclockwise
I saw some comment about indent when I was reading back on -devel
what was that about?
you mean me bitching?
indent makes code less comprehensible by tossing any formatting done by the author
some of it's very hard to read now
that assumes that the formatting done by the author was done well
like Interp::find_ends() which was virtually destroyed, I assume, by indent
what file is that in?
no, that code was destroyed by the use of ?:
although somewhere along the line someone took out the SET_TO and ISNT crap, and I thank him for that
and lines that are two fscking long
uh-oh, are you one of those anti-?: guys?
personal bias, I don't like ?:
if you're gonna code a branch, admit it and code a branch
don't try to pretend its just an assignment
let's not talk about it :-)
so it sure was hot out today
coding it as an if would allow a full line for each expression
not here, we had a line of storms go thru last night and cool things off
supposed to get hot again in a few days
101F after work today here
yay air conditioning
to be honest, that code would be a bitch to read no matter how you right it
yeah, it sucks
but the totally bogus linebreaks make it harder to understand
so de-bogofy them and commit
no, I won't do that because it messes up cvs
much of that is because there is so blasted much whitespace before the wrapped portions of each line
I don't buy that CVS argument
suppose you did a commit that simply fixed punctuatiuon, with a comment saying "fixed indenting, no functional change"
I realise thats probably not possible, you have other changes going on already
that's ok, just a preference, I just would rather deal with bad formatting than useless cvs diff/cvs annotate that covers that version from now until the end of time
especially since there are branches
you can never ask "what's different on this branch" because the answer is "every f-ing line"
heh, I've been known to commit fixes to spelling or typos in comments....
that's fine, it's just that changing every line is BAD
IMO people who do that (with indent, for instance) are just inexperienced with revision control
I certainly fit that description ;-)
the worst violators are the ones who make functional changes, run indent, and then check it in
I dealt with that in emc1 emcmot and I still remember the pain
I've done that (very early on)
* jmkasunich hides
in emc1 there were two TPs, with copies of huge files with a few lines changed
theres a lot of that
someone ran indent on one of them, which screwed up its formatting like indent always does, but also now when you diff them, every line is different
then someone fixed bugs in one or the other a few times
lets just switch to subversion and get this over with :D
it took me hours (really) to figure out the simple bugfixes and resynchronize them
it should have been 5 minutes
well, if it was EMC1, that wasn't me
I'm really not blaming anyone
even though cvs would tell me who did it :-)
thanks for distracting me from arcs, btw
get back to work!
:P the whole spindle_forward not reset is actually a bug in the CVS checkout from last night, same problems with the stepper_mm hal :P
getting latest to compile and try again :D
Dallur: can you describe the problem a little more?
when I hit estop the spindle_forward should be reset to spindle_off
but with the current version it is not
I guess that logic needs to be in iocontrol
I really hate the current estop logic (and the io handling for that matter)
It worked the last time I checked, and I know it works in the last release
jmkasunich: I just forwarded everything to classicladder and I do custom estop logic there
I tested it 5 min ago in release btw
yup, verified with latest
when you say "release" you mean 2.0.1?
looking at diffs now, there haven't been that many changes since then
[23:54:52] <jmkasunich> http://cvs.linuxcnc.org/cgi-bin/cvsweb.cgi/emc2/src/emc/iotask/ioControl.cc.diff?r1=1.31;r2=1.32
gotta be in that one
can you walk me thru reproducing the bug?
just start any config (stepper_mm for example)
you said it happens with the stock stepper_mm config?
take estop off
and hit estop
spindle turned off
hmm the spindle turned off for you ?
that does not make any sense,, hmmmmm