vacuum table dit google and look at instructables, cnczon, and some people using old air hockey tables. i make 'em outta graphite, but i machine graphite.
jmkasunich: OK, back. Decided coffee was probably smarter than Jack at this point :)
Not sure where I am confused at...
x and y run from 0 to 8 and 4 respectively. Z runs from 0 (top) to -1.8 (table) but as to why 0.5 nor -0.5 are out of limits evades me.
ok, lets back up a bit and just talk about coordinate systems
those numbers you just gave me - those are in the ini file, right?
well, stepconf which I am not sure is the same thing exactly?!?
Lemme load the actual ini that stepconf makes...
stepconf makes ini files
Right, but it does things that you don't see or something...
all stepconf does is ask nice friendly questions and do the math for you
it doesn't do anything you can't do yourself
all its answers go into the ini file
OK, that is what I entered in stepconf but looking at axis_2 does not correspond (x(1) and Y(2) do)...
MIN_LIMIT = -0.01
MAX_LIMIT = 0.01
well thats fscked, no wonder it can't go anywhere
That might explain it. Lemme check stepconf, I am sure I entered 0 and -1.8...
you keep running off to do things....
first, you wrote "X(1) and Y(2)" a few sentences ago
Uh, well, yes.
was that a typo?
X is axis_0, Y is axis_1, and Z is axis_2
x seems to correspond to [AXIS_1]
well then thats fscked too
is there an axis_0 section in your ini file?
Yes, typo'd it and kept going with it :-(.
thats what I suspected, and why I asked
MIN_LIMIT = -0.01
MAX_LIMIT = 8.5
Correct for X.
Ditto for Y (1)
But 2 is hosed..
Number are right in stepconf (if not in the correct order),
ok, I seem to recall you changing them a few minutes ago - from 0 and 1.8 to -1.8 and 0
0.0 to -1.8
And back again...
I wonder if stepconf wants the lower number first?
Lemme chang it and then look at the .ini again.
just remember - the ini is what counts
Yes. It has been suggested I quit using stepconf and get used to the ini...
MIN_LIMIT = -1.8
MAX_LIMIT = 0.01
now, let's not rush off and try your program again
lets talk about coords
LOL, caught me :)
if you had home switches, your machine coordinates would _always_ correspond exactly to machine positoon (after you home)
. == yes, understand
and that is how it should be - so the soft limits can keep you from running into the end of the table, etc
since you don't have home switches, you can home anywhere you want, and you _could_ use machine coords for your parts - but its a bad idea
that is the idea behind the "touch off" thing?
when you write your g-code, you want to be able to put the zero at some convenient part
maybe the edge of the blank material, or the center of a hole, or something
if you jog to that spot and then hit home, the machine coordinates now start at that spot, and you could run the program
That is what I have been doing.
but the machine limits no longer match up to the real machine, cause "machine" coords are no longer at the right place relative to the machine - you've turned them into "part coords"
instead, if you jog to that spot and hit "touch off", you create a new set of coords that are relative to the part
and machine coords stay where they were, so limits and such still work right
I have been shifting the home to where I want to start - so what happens is that whatever the home I set is, is outside of "real" home. If this is so, then I am more confused...
there can be only one home - when you hit home, the machine forgets everything it knew prior to that
I am under the impression that wherever you set home corresponds to the min_limit so the rest follows, but apparently not?!
machine zero is the absolute frame of reference for the entire machine
In other words, without real home switches, essentially, home is where it sits when you press "home axis"/
well, there are some tricks with home offset and such, but if those things are zero in your ini (probably are, don't go looking right now), then wherever you hit home becomes "zero"
OK, the problem wasn't what I set, it was the sequence of the numbers I used in stepconf that caused the problem.
the best way to work it without switches is to make a mark somewhere - when you turn on the machine, jog to the mark and hit "home axis"
Yes, been trying to do that.
always leave it at "home" when I quit.
well, that part doesn't matter, other than saving some time when you start up again
Last thing I run when shutting down: g0 g53 x0 y0 z0
thats probably a good habit
Understand. When I get to real switches, then things can be done differently.
so, you start it up, and if needed jog to the marks (if you did the G53G0x0y0z0 you won't have to jog), and you hit home
thats the last time you should hit home until you power down
I can see why people have told me to avoid stepconf. Without access to this and only looking at stepconf I would have never found the problem.
So just hit home axis to be sure once I load up emc. Procedure coming in as well as going out.
make sure you are in the right place, then hit home
the procedure coming in is the important one
Thanks, this has really been helpful and informative!
the G53 thing at shutdown is a convenience
lining up to the marks and homing at startup is what makes the limits work right
after you are started up and homed, you load your program
then you need to set part coordinates - that is what touch-off is for
OK, got my notes on the desktop put all that under startup and shutdown procs :).
I always put comments at the start of my program telling me where to touch off
Will give that a try today. Rebuilt the Z axis again and it is a bit stiffer now.
for example, lets say that X=0 for your part is at the left edge of the raw material
you can jog the center of the tool till it lines up with the edge, then hit "touch off" and enter zero
Going to extend the Y by moving the stepper back about 2" on an extension plate.
Getting there - slowly but getting there. May not be much more than a PipeDream, but I am sure learning a lot!
Again. MUCH appreciated.
Going to play with soft wood rather than continue to chop up PCB stock :-).
Now to rewrite my gcode :).
now to cut wood (unfortunately not with CNC)
This is all adjustment and evaluation :)
Wish I had CNC to cut the wood I am using on this - a jigsaw is not renowned for square cuts :-(.
Cleaning it up with a belt sander also leaves much to be desired.
Really tempted to order the combo belt / disc sander from HF.
a nice table or radial arm saw would be nicer, but no place to put either :(.
I have a small table saw, but I'm cutting large parts
ripping 56" long x 48" wide sheet of masonite is not fun when the saw table is 20" long x 16" wide and you have no helpers to support the sheet
Hi guys: I'm trying to get a postprocessor for apt360 and 5 axis machines going. I was wondering if anyone has some resources I could use. I know some of you have some 5 axis machines, and jepler did an animation of a gantry machine. I could use some way of testing my code. Any ideas?
I would be interested in knowing how jepler and cradek are getting 5 axis g-code.
cam of some sort
or writing very simple stuff by hand
jmkasunich: circular saw and clamp a straight edge on it?
If I could have a cl file and the corresponding g-code from the cam system, I could make a post for that machine.
using the table saw, clamping a straight edge to the work and running it along the side of the table
jmkasunich: ew, sounds like a pita
jmkasunich: If you have a trim router, just clamp the sheet to the bench top, and have at it.
only got a couple more cuts
crotchetyGuy: were you able to get the 5 axis simulator going?
DanielFalck: no - you mean in emc?
How do you go about that?
stuart probably has the visualization set up for this
let me look on the wiki
Stuart is the main reason I want to do this.
this might be a start
[01:49:05] <DanielFalck> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?FiveAxisGantry
I can run the vismach from this box in simulated mode (not connected to a machine)
pretty impressed with the videos from jepler and cradek
crotchetyGuy: the post should be the same for any EMC machine
K`zan_emc: if you have any suggestions of how to make stepconf better i'm sure they'd be appreciated
crotchetyGuy: in the startup menu for emc2 (the gui configuration menu) there's a 5axis example that will bring up Axis and the 5 axis gantry machine
DanielFalck: thanks- I'll check that out- I'll probably have to update my code.
fenn: Stuart implied that also- however, I don't quite get it. how can the same code for a dual rotary table work the same as a AB head, for example?
you program the tool orientation vector
and tool tip
that's how it's supposed to work at least, not sure if it actually works yet
so the axes don't put out angle information?
if you want, you can run 'blind' i.e. with no kinematics
but that's stupid
can you give me an example of how some actual g-code would look?
[02:02:50] <DanielFalck> http://imagebin.org/18343
[02:03:36] <DanielFalck> http://imagebin.org/18344
[02:04:00] <fenn> http://www.youtube.com/watch?v=fWKYQUj5AOs
this is a simulation of http://cvs.linuxcnc.org/cvs/emc2/nc_files/cone.ngc
uhh.. perhaps that's not the most clear and readable g-code ever
once you have the sim running on your desktop, you'll figure it out
but notice the g1 w-100 f2000 (drill)
those are the angled moves
so how do I get the configuration dialog?
when you choose the emc icon from the Applications menu or when you run "emc" with no arguments, you get the configuration picker dialog.
the 5-axis stuff that we've been talking about is not in the stable (2.2.x) version, though
ok, thanks, kinda dumb on emc stuff.
ok, so you are doing the calcs via macro g-code-
a, b, and c represent angular position, though, right?
b and c; a is not used in this setup
b and c specify an orientation for the tool (any two angles would do, it could have been done with ab or ac instead)
ok, so it seems like you will need a new post for each machine config.
no, you just need a different emc "kinematics" module to specify how to turn a pose into motor positions
you could skip having emc do the kinematics and do it all in the cam postprocess instead, if you like
(you can do 5-axis with machine specific posts just fine in the stable version of emc2 for that matter)
crotchetyGuy: see if Stuart can supply you with his sim vismach before you get too deep into things
since you're setting up a post for him anyway
ok, I will send him an e-mail.
how are you starting emc on your computer now?
jepler: I don't understand- so you can use tool vectors or angular positions?
DanielFalck: I suppose I will have to compile from source?
I don't think so
you have Axis/emc2 running now right?
click on the 'Help' 'About' items in the top menu
see which version you have
crotchetyGuy: you specify the angle of the tool
I'm running running pre-2.3 CVS Head
it's been a while for both of us
the 5axis configuration and supporting files are only in the CVS version, not in any stable released version.
jepler: so what would the g-code look like for that? say I had the tool vector 0,0,1 (3 axis orientation)
crotchetyGuy: you have to express the tool direction in terms of two rotations. If 0,0,1 means the tool points down along Z that would be an easy case, B0 C0.
ok, so you can't specify the tool direction via a tool vector- it has to be an angle
crotchetyGuy: yes, but by using different kinematics modules in emc you can have B0 C0 mean the same thing for any machine that can achieve that orientation.
jepler: what if the machine has an A, B axis?
A,B is functionally equivilant to B, C
on a mill, anyway
crotchetyGuy: are you asking me if emc will force you to write programs with B and C specified, even if your machine has A and B instead? No, it doesn't force you to do that
but because of the mathematics (two angles are enough to specify the tool orientation uniquely) you could do it that way if you wanted
even though the angles are functionally equivalent, the x,y,z positions will depend on the kinematics, though? or does emc do that too?
Perhaps I should do some homework on emc. :)
i don't know how emc handles it.
I am struggling to explain how it works...
B,C is a vertical mill, A,B is a horizontal.
you specify two angles to get the tool orientation -- BC in the case of the 5axis demo configuration in CVS. This specifies a "tool coordinate system". You can move the origin of the "tool coordinate system" by specifying XYZ. You can move the tool inside the "tool coordinate system" by specifying UVW.
The kinematics module inside emc2 turns all these numbers (8 of them if you've been keeping count) into positions for 5 motors.
that's a pretty cool way to make hand-coded 5 axis stuff doable
If you've got a rotation in play, then specifying a move like G1W-1 probably moves 3 of those motors (the linear ones)
so, any word on g99 quaternion mode?
ok, got you- that is how you did the drilling on the cone?
jepler: is that something in a standard or from a commercial control, or is it something unique to emc
because that's darn cool if it's emc original
toastydeath: I think it's unique to emc
well, if not it's a separate invention
I don't know what's in all the commercial controls
i don't think they have as elegant/seamless of an implementation
but one of cradek's goals, if I may put words into his mouth, is to make hand-coding in 5axis a real possibility
i know at least haas has some kind of "tool coordinate system" command
but it's awkward at best, from looking at it
'night guys, it's getting late here
thanks jepler- nice work-
crotchetyGuy: I didn't do much of the work -- I just made one video for the youtubes.
can I do that without any special tools?
make a video I mean
of the simulation
i might want to move a backslide parallel to Z that is called W, and would not want any other axis to move. I hope this is not disallowed in EMC2. I hope that is unique to only a certain set of kinematics.
yeah that would suck for dual z axis machines and also wire cut machines
didn't think of that
i'm pretty sure it's unique to a set of kins, but i'd like that clarified.
this is an example Stuart started- http://imagebin.org/18345
a beveled slot
green is tool vectors
! what have you guys been doin! cool!
tomp2: kins are what maps axes (ABCXYZ etc) to joints (quill, knee, saddle, table, rotary table, head tilt, etc)
if you want W to directly control one motor with no interaction, you can make a kins module that does exactly that
(actually trivkins already does)
yes, and I hope) one kins could exist where there was no relation between joints at all, each joint is an axis.
that is what trivkins does
like a 8/9 axis trivkins
"trivial kinematics" - what goes in comes out
thanks, phew! i worried
I'm pretty sure trivkins is already 9 axis capable, but I can't swear to it
yes, it is 9 axis
no space bar on that... xyz abc uvw in trivkins.c
yep - just looked at it :)
those emc2 guys, they're pretty good
Are blank lines OK in gcode files ?
well, I'm going to try to get emc 2.2.5 going- cool stuff.
blank lines should be ok
OK, lost my live plot and wondered if that caused it, thanks!
they could interfere with toolpath stuff
depending on how the control handles them
i don't know how emc handles them.
K`zan_emc: you should be able to get the plot back by clicking on the view buttons at the top of the window
If I used the touch-off button, g49 clears that right?
jmkasunich: Lemme try it
I don't think so
maybe it was g54
g49 is tool offset, must be g54
clearing a touch off is non-trivial
don't go issuing g-codes when you don't know what they do
wouldn't g53 set it back into the machine coordinates
and g54 sets it into the first work coordinate system?
g53 is temporary: "for this move, use machine coords"
oh that sucks
Easire to just make the zeros what you want each time.
toastydeath: how often do you really want to move in machine coords?
the fancier the program is, the more often
Heh, thought I had a clue.
[03:03:16] <jmkasunich> http://www.linuxcnc.org/docview/html//gcode_coordinates.html
I realize that page is long, but it is important stuff
The morez I learnz the less I knowz :-/.
if you aren't gonna use G92 ( I never do), you can skip section 1.4, which is darned near half of it
g92 is mostly for old school lathe controls
with no work offsets
though there is one guy at work who uses it pretty frequently
the "touch off" button sets G54 by default, but I believe you can use it to set any coordinate system
Gonna take a brain break and go get some chow, bbiab.
if you clear all the coordinate systems, and only touch off the G54 one, then you can switch to G55 if you want a modal "use machine coords" command
(G55 or any of the other 8)
darn - somewhere in the docs is a paragraph that tells you how to clear the various offsets
I thought it would be on that page, but either I'm blind or its not
there's no offsets page?
on the control?
what do you mean "on the control"?
[03:11:02] <fenn> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?CoordinateSystems#So_if_you_re_lost_what_should_you_do
you mean the GUI?
the traditional thing is to press the "offsets" button on the control
we have several different GUIs - most of the time we're talking about Axis here
and it brings up a management page that shows you all the various offsets
all the work, tool, and other things that are set
a single screen that lists the coords behin G51,2,3,child systems etc
I don't think axis has such a thing
i agree with toastydeath here - there should be a page showing all the offsets in effect
build it and they will look
also not in effect, so you can set them.
fenn: thanks - that should be in the regular docs too
i think the data is available in a big list of variables but it's not terribly clear
toastydeath: yes, but setting them would involve new nml commands because you arent just typing in g-code
somebody should start Axis and poke around in the "View" and "Machine" menus, it might be in there already
nah i'm too lazy :)
fenn: i dunno how it works in emc
all i know is i am used to walking up to a machine and being able to arbitrarily change any offest to any value and have it take effect immediately
even during a run
gosh the fonts look even worse over ssh
i loaded axis in 2.2.5 and dont see it
machine->show emc status has some xyz triplets but not sure what they represent
i recall having this conversation before with someone
and the end result was there is no such thing in axis or any other interface
but that may be incorrect, obv
in mini there is an offsets tab
it only shows the values of the g54..g59.3 coordinate systems though
and it doesnt seem to work (at least running over ssh)
"can't save sim.var" and the teach button just sets the field to 0.0
I don't understand the subtract/add
(your var file is probably not writable)
oh its running from /etc
ok I sort of understand the subtract/add now
you might use them with an edgefinder somehow
and the 'teach' is sort of like axis's touch-off
yeah mini has the coord systems and yoiu can edit them view them and 'teach' them ( say the value is 'here' )
[03:25:12] <cradek> http://sourceforge.net/tracker/index.php?func=detail&aid=1763005&group_id=6744&atid=356744
I see that alex has plans to work on this
[03:25:29] <cradek> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Alex_Joni
making a scheme that lets us view/edit coordinate systems and manipulation/viewing of other variables in the gui will take a lot of effort.
and since our interpreter reads ahead of the machine motion it will probably not ever be very useful to watch while a program runs (except if single stepping)
could be very useful to see variables while single stepping.
I am talking to myself but that's ok.
i think the idea was to see the origin/reference value of each coord system. i think these are stored in the variable list.
yes it'd be nice to watch variables in single step debugging
stack trace with expandable objects, too much to ask?
re add subtract " When you do this you may need to add or subtract the
radius depending upon which surface you choose to touch from. This is selected with the add or
subtract radiobuttons below the offset windows."
tomp2: got it, that's sort of what I figured out
np, i had to go read it, handy for simple shape bumping, like a wiggler and a block
I'd like to put some edgefinder specific stuff in axis's touch off, but I can't come up with a simple, clean, self-evident user interface for it
the benifit of having a dedicated screen to set offsets is the user can decide how they want to touch off
and elimiates those kinds of issues
the truth of it is that with my .5" edgefinder I can probably type .25 or -.25 easier than I can figure out a bunch of buttons
cradek: i think you'd need some kind of wizard with a diagram
"am i adding or subtracting, what's the dimater of the edge finder"
fenn: yeah I'm afraid of that too
I crank the handle .25" over after i touch off
and just read off the position screen
that way I don't screw up by going -.25 when i should have gone .25
I have to admit that I usually go up and then move to zero, to see if it's right
touch off, move up in Z until the MPG handle is at the 0 position (physically)
then crank the machine over using the graduations on the mpg until you've gone .250 (or .1)
that's an interesting approach
if it looks/feels right, type the number in and no room for error
the mpg on the BOSS doesn't work that well unfortunately
I'm sure that's why there are no marks on it
do you have an operator/posi coordinate
on the machine
this may not exist on yours but on most controls, there's a DRO-like coordinate system
so when you touch off you hit Z-esc (or x-esc or whateveR)
sure there's a position readout
and it zeros
so you can just move over .25 using that
graduated mpg is not the rule and if theres a tall standing boss, i cant move over it
yeah I could do it. it's a bit tedious though...
these are good ideas but solve some setups, not all
if you don't have the equipment to make it fast, then it will be tedious
fortunately 1/2" is easy to see so I don't screw that up.
gaah - I so very carefully cut those slots and dados in the right place
on the wrong piece of wood
keep the primitve tools simple and available so the user can get the job done. dont make him work some predetermined way. let him use the tools that suit his need.
tomp2: so you think edgefinder specific stuff in touch-off would just confuse the issue?
interesting, I didn't expect that
no, fenns guided gui is a good idea, but hte low levl tool of splitting, keep that avialable, the subracting D or R keep that available. buid the fancy out of the primitive, and have all available
"I just touched off on the edge that is zero... I'm left of the edge, so I must be negative... enter -0.1 (I have a 0.2 dia finder)"
i think it's a good thing to have around for new users who may not know all the particulars yet
make a cursor move like the number entered? a visual clue to what + or - means?
but i would never in a million years want to use anything but typing my number in by hand
(for me it's more like: now I know I have to enter .1, .2, .25, .5, -.1, -.2, -.25, or -.5)
as a check, run Z up to clear, jog till finder is visually centered on the edge (+/-0.020 is fine) and make sure the DRO reads near zero
heh, Heidenhain avoid any user number input, too many poeple type lek me :)
must just be me :-)
they use an idea like RayH had, the 'here' button
RayH called it 'teach'
no dysl;exia, no sign error
i think all those are a good option to have if the user wants to use them
no misplaced decimal
but i have noticed that all the machinist i have worked with and spoken to so far who work on large and expensive parts refuse to use those features on the controls
and have taken up the practice myself, because you can train yourself to re-check and catch your errors
i went to a place in Maine, they srapped a 100k$ turbine part because .0045 became .450 in an offset page
yes you have to be careful
the teach/here scheme combined with an edgefinder would be nasty. the current position is 1.0732 and the 0.5 edgefinder is on the left. what number do I put?
with axis touch-off, you always put -.25
it's relative to "here" instead of relative to the machine origin
how does the user skip that?
how does a user of emc put in numbers right into the tool and work offsets?
with axis, he would have to MDI g10
or change the tool table for tool offsets
toastydeath: what's the advantage of having to re-check and catch errors?
yeah I guess I don't believe in putting these numbers in directly - sounds like I'm with fenn
"all the machinists do it that way" could just be stubborn old habits
and that's cool with me if you have that viewpoint
I never measure anything from machine origin - how the heck would I even do that?
the problem is you are forcing your users to work in the way you feel is best
you always do that
rather than the way they want to work.
rather than the way YOU want them to work...?
this is a silly discussion now
i don't think it's silly, i am telling you what I observe in a machine shop
and you're telling me it's stupid
YOU can write YOUR own python, cuz AXIS is cool
if you can explain why it's a good idea, we might learn something
no, I said "I don't believe in it", not that it's stupid
no, I think they're telling you that either way, users will have to use either "your way" or "their way". either way, somebody will do something they didn't expect
fenn: it puts a whole bunch of extra steps in that force you to look at the number again
that is my personal experience and the experience of my co-workers
and it catches a lot of mistakes
regardless of how I, or you, or anyone else touches a tool off
there's no way to just put a number in for a tool offset or work coordinate in the GUI
but if the mistakes are caused by the fact of the number not representing an obvious quantity, rechecking it is a workaround for a bad user interface
to me when I'm touching off, the obvious quantity I can readily measure using every type of device meant for touching off, is the distance from the spindle center to the workpiece
where the work is in a space 40" long subdivided in ten-thosuandths of an inch
it's obvious to me that THAT number is the one the user can get right the most easily
is not an obvious quanity
that's the bad gui
toastydeath: have you ever used Axis's touch off (or even seen it used)?
toastydeath: sorry i'm not following? what "puts a whole bunch of extra steps in that force you to look"?
cause it sure don't sound that way
i haven't, jmkasunich
i don't use emc
are you sure you understand what it does?
i think a visual hint of the effect before submitting is handy, and reduces errors
it would be neat to touch-probe a block of stock and see it in the part preview
here is what i am trying to do
i am trying to give you the methods used by people who have more experience than any of us in the room regarding machining
if you don't agree, then fine!
i don't use emc, it doesn't affect me.
toastydeath: if the left edge of the part is "zero", I put in my edgefinder, jog around till the finder says its just touching that edge - I click touch off, enter -0.1, and the DRO says "-0.1000"
toastydeath: would you do me a favor? ask why they prefer entering numbers vs letting the machine record the number
then if I want to be double sure, I raise the spindle, and either jog over a bit till the finder is roughly centered on the edge, or MDI G0X0
fenn: i've asked
"it makes me feel comfortable.
i asked how
I guess I need to get a CVS version of emc to get the gantry 5 axis sim?
well, that's just stubborn old habits
and that's cool if you think that
if they had only ever used the thumbwheels instead of a CRT readout, they would be most comfortable with thumbwheels too. That would NOT mean that no other way might be better.
but the guy who said it is the most experienced CNC machinist and works on the most expensive jobs to come through the shop
i'm sure it works fine for him
yeah, and it works for a lot of other people
he's also had thousands of opportunities to learn from his mistakes
so your users are stupid?
also, if it's a task that he has to check carefully each time because it's easy to screw up, the procedure probably could use improvement
he's a skilled worker
i think making it possible for relatively unskilled users to do what they want, is a worthy goal
i think so too, but in the same vein you are limiting your audience to unskilled users
no, that's not true
not entirely true, no
he can use 1940's era g-code if he wants
g10 l2 p1 x1.2345
oh hey you can even add N001 to the line
then check the DRO and see if it says -0.1, and if not, try another G10 number until it does
the thing that is bugging me about this
is that it's a simple thing to ask
provide a way of communicating with the control in a very basic way
is g-code not basic enough?
the most basic way is actually the gcode, and it's already possible
any gui touch off will be built with the primitives that the smart user could use directly. Tthere's different tools for different users. take the power and the responsiibility, or use the simple gui.
i wouldnt put the options like that - a gui can be as or more complex than a command line
imean that the lo level thing is the powerful dangerous thing, the gui is the safe but limiting thing
i think you are off in never-never land chasing cool stuff in control and ignoring features that every commecial control has and that most production machinists use.
i have a macro that finds a surface. does nothing with tool dia or length or taper. i have higer level finding routine that use that lo level and do account for tool dia/len/taper. differnt tools a, all available.
toastydeath: yes you're right that emc's gui should show the offsets, but the more features and methods you add to a program the harder it is to learn and maintain
and that is very true.
that doesn't change my basic complaint
emc's feature list goes into very cool stuff
but, compared to 'basic' and more or less standard interface features among all commercial controls
is completely deficient.
if that's perfectly fine with your userbase and the dev team, i will shut up and never mention it again
the reason i mention it is because it seems like people are interested in having a open source control that really does compete with commerical offerings
but i can very easily shut up and never mention what is standard in commercial controls again
that's not a dramatic statement, i just don't want to argue about this crap and i don't think anyone else wants to either
i value your opinion
its just not "the final word" hah
we're always interested in what you say, esp because of your experience, but you gotta allow that others wont always agree or even understand
toastydeath: I don't know what to say except I understand the complaint. I'm not trying to be dramatic either. But I know I personally get interested in adding something to emc when an emc user says "hey I can't do this task I want to do because something is missing"
or 'but please allow that...'
fenn: i know it's not the final word, and i don't pretend it's even the first word
I don't care about chasing all the features in commercial controls. Part of the reason is that I don't know all of them.
but I do care about people using EMC to do real work, and one of the real things they do is touch off a workpiece, and I feel like this is currently handled in AXIS in an excellent way.
I only regularly use one other control, and it's pretty old. It's bad at a lot of things that EMC does better. It's better than EMC in a few ways too though. I may add those things to EMC if I can later.
(fwiw, it has no way to set the work offset in machine coordinates. The only way is "relative to here", just like AXIS)
that is valid
I have my Centroid /Bridgeport running in the garage and it has a few cool features that I would like to incorporate into EMC- later on
my experience is that I would have a hard time touching off in EMC at work
from what you describe, obviously
and I think I can hack those features in with some help
so I guess I just don't agree that emc is "totally deficient" because it doesn't show the offsets in this way. It's just different, and you can do the same tasks with it.
not having used emc to do real work.
's not "totally deficient"
touch off doesnt seem to do anything in sim? i enter 5000 in the box and nothing happens
toastydeath: I bet you would be totally comfortable with AXIS's touch off after using it about twice. I'm totally serious.
i think it lacks features that home users won't notice, but that commercial machine shops would miss.
fenn: probably your varfile isn't writable again, just like last time
cradek: what about the fixtures we have where we touch one arbitrary point
on the fixture
silly var file
and then all the offsets are relative from that
like g54 might be x-4.22 y1.563 from that point.
for a home user, that's never going to happen
how can you touch just one point?
fenn: the fixture was made accurately
easy enough, touch off and enter those values
dont you need to touch at least 2 points?
fenn: picking up a hole
uh, with a dti-spinner?
then find Z when you bring the fixture down
bring the tool down
last month there was a case where I had 18 tools touched off, and the fixture had to be re-mounted
it was the same in X and Y because of how it mounts
but I had to enter a Z offset to compensate for that error or I would have to touch all the tools off agian
so all the tool lengths changed by a certain amount? you'd just touch off Z again
(with any of the tools)
one place we could improve emc is online tool length measuring. the BOSS does that nicely.
the big shop I visited recently has an optical tool length measurer. the length is marked with a sticker on the tool.
but I usually want to measure them in the machine
yeah those things are awesome
i don't know how this works
but apparently, with tool presetters you use G42
and enter the offset as positive
and make up the difference with the Z offset in your work coordinate?
i might have that wrong somewhere, but it was different than what I do at work
do you mean G43
i mean G42
subtracts the tool length rather than adds it
however it works out, rather than getting the length of the tool
in emc you can have lengths negative or positive
yeah, you can in commercial controls too
there's something funky about it that i don't understand
on the BOSS they are always "like" negative offsets in emc
TLO=0 is your longest tool
bigger TLO causes the quill to come down
this is the reverse
it goes farther up
and you compensate by setting a huge Z in your work offset
in emc with positive tool lengths it goes up
so emc can do whichever you like
yeah, i'm not saying you can't
i was just saying there's this tool touch off thing you can do with optical presetters that i don't understand
maybe you'd heard of it/knew what it was and could enlighten me
the setup I saw was simply a device that measured from the gage line to the tool tip. they marked this value (which was several inches) on the tool
seems like for some tools this length might even be more than the quill travel!
so you would need other offsets to compensate ... or something
a bigger machine to compensate
there's a 6" facemill at work that is 18" long
if your tools were marked 6" to 9", maybe you would put 0-3 in your tool table, I'm not sure.
i saw it and was like "what"
who knows man, work won't buy a presetter
so i guess i will never know
why not cut a hole in your surface plate, find something that matches your taper to go over the hole, and use a height gage
or if you have a grinder, just grind it
the problem is the gage length
the tools don't register on the flange
that's why you need to set it on the taper
er i got ahead of my self
just make it a fixture or something
i now have no idea what i am talking abouit
cradek did you ever make that taper thingie to measure tool lengths?
just got a couple of these beasts. they are the same as my spindle.
on the mazak, pull one pod out of the carousel, bolt it next to a surface plate. now you got a taper to seat on, and a height gage to measure differences to some standard.
i thought your spindle was nmtb-30?
NMTB30 with a special nut that grabs the flange
looks just like those
oh, so that's a tool holder holder?
2" straight shank to QC30 adapters ... or something like that
no idea what they were for
but the taper will be right!
fenn: I did already make a ring that the tool holders sit in. its inside bore matches the taper. I can set this on top of 123 blocks and measure heights that way.
I don't use it though - it's easier to measure on the machine if I'm just doing one or two.
is the wxWidgets installed on the system from install 6.06 emc live?
is there a easy way to tell the version?
synaptic package manager
dpkg -l "*libwx*"
i'm looking at the pm and is it listed as libwx?
Synaptic tells me the widget set is not installed but 2.4 and 2.6 are available.
there are two packages on 7.10, called libwxgtk and libwxbase
I'm not sure what 6.06 has
the hardy live cd wouldn't install so I installed dapper live cd all went well
what did the hardy livecd do?
it said bad cd or hard drive mid install move cooler place ,it's cold enough here!
it would boot fine and it installed on my dell
I've noticed that cvs server is not responding
any ideas ?
it works ok from here. I think the internet is not very healthy this morning. there is probably something wrong between you and it.
I cannot get there either.
you guys should try a traceroute/mtr
what is the ip address.. it isn't resolving the name..
[14:44:17] <BigJohnT> http://www.cnczone.com/forums/showthread.php?p=453151#post453151
I get there with traceroute.
in about 18 hops.
ah I'm lucky -- just 7 hops from where I am
Sitting next to it;)
nah I'm miles away
From here it's "unknown" to detroit, chicago, denver, then into ALLTEL, then nebraska.
Ping request could not find host http://cvs.linuxcnc.org/.
Please check the name
skunkworks_: you should probably enter just the host name: cvs.linuxcnc.org
the odd thing here is that the last 3 hops are inetnebr, then unpythonic.net, then back to inetbr
different IP though
mtr gets a complete trace, but "traceroute" on bsd ends with !Z packets -- I wonder if I firewalled something I shouldn't have.
(!Z is apparently "administratively prohibited")
this is how things look from here: http://emergent.unpy.net/files/sandbox/trace-cvsserver.png
and here's the apology from my ISP: http://pastebin.ca/1022795
jepler: emergent.unpy.net unavailable :(
micges: maybe my ISP is a bit optimistic that the problem is fixed :-P but the words I am typing right now are going through the machine that hosts emergent.unpy.net and which sits next to cvs.linuxcnc.org, so it seems to me that it is OK :-P
back to books :)
[15:10:25] <skunkworks_> http://pastebin.ca/1022814
Hey skunkworks I see a .218 on the end of that and am able to get through.
whois says they both belong to Jeff.
Maybe I am on jeff's bad side today :)
actually - now I am timing out past the 126.96.36.199 address
seems to stop there now ^
I think my firewall blocks the packets of some ping and traceroute programs from going to the very last hop (.217)
I think mine does also.
(if it is still turned on)
I see the 217 and then 2 of 218.
Maybe my traceroute needs a debounce ;)
Hello. I'm having problem running EMC this morning, after booting. I get the message: insmod: can't read '/usr/realtime-2.6.24-17-generic/modules/rtai_hal.ko. I only have a /usr/realtime-2.6.24-16-rtai directory on my system.
Hugomatic: seems like you've booted the wrong kernel
has anyone bought this (probably crappy) little air compressor from HF? http://www.harborfreight.com/cpi/ctaf/displayitem.taf?Itemnumber=95275
if you want to do anything more than blow chips away, I wouldn't bother
I'm tired of hooking the mill up to the big one to change spindle speeds, and I'd like the spindle brake to work again. I wonder if this is enough
0/6CFM@90PSI is crap
yeah I know it's crap, but I bet I don't need much at all
I guess if it completely sucks, I can take it back
if you look in the sale flyers for local auto parts stores, they'll have somewhat better ones on sale from time to time
I got one that's a bit better than that for $89 I think (115PSI max, 4.2CFM@90PSI)
SWPadnos: thanks I'll consider that
this is odd - to say the least.. http://www.youtube.com/watch?v=N9u2pDqOKZg
(you have to read the whole discription)
the other option is to get a splitter for the big one, so it can be connected to the mill and other things at the same time (unless there's another reason to not do that)
Hugomatic: if you installed another kernel after you installed the linuxcnc.org -rtai kernel, ubuntu may have automatically chosen the newer kernel on the next reboot. Either make sure you pick the -rtai kernel when you boot by using the menu, or edit /boot/grub/menu.lst to indicate the -rtai kernel.
cradek: thanks... I can now run EMC again. You were right: I was running the generic kernel, and my grub menu was hidden. It now works as long as I run a "failsafe gnome session".
I don't understand what gnome session would have to do with it
anybody have gcode handy to cut out a DB25 connector for a panel?
I don't think my ISP has resolved the connectivity issues..
* jepler curses their name
jepler: I'm sure I saw some g-code for db25
unfortunately can't remember where..
yeah I'm sure it exists out there somewhere
ah, this is what I remembered
[16:15:35] <alex_joni> http://timeguy.com/cradek/01166390211
ah I've heard of that guy
but I think it's for the layout, not the slot
jepler: you did?
ir2 is now known as rave1
rave1 is now known as ravennb
jepler: looking at the g-code... I think it's what you need
(couple of lines and arcs..)
I've one question... I build a cnc milling machine an use emc2 as software
(heh, the db9.ngc is just as long)
now my question is there a way to use a tool length sensor?
ravennb: there is example gcode called tool-length-probe.ngc
ahhhh right... I've seen this code before
[16:19:23] <cradek> http://cvs.linuxcnc.org/cvs/emc2/nc_files/tool-length-probe.ngc?rev=1.4
I think it's also on your system already
but how do I do the mechanical connection of the switch?
you can connect it to any digital input on your system.
at which pin do I connect it? I know i could program this in the *.hal file
if you are using a parport, there are several inputs available
yes I use the parport
ok I'll search for an open digital input
in HAL you hook the input to the pin motion.probe-input
if you this way: parport -- switch (NO) -- gnd
then you probably need to use the pin with the -not suffix, because it will be low voltage when the switch is closed and high voltage when the switch is open
jepler: do you need a pullup? I think mine has one but I'm not sure.
yes thanks... thats great... now I need only a tool length probe switch
if the open condition of the switch is not properly sensed, then you may need a "pull-up" resistor to ensure that a high voltage is present
to be honest I am not sure whether parports typically provide the "pull-up" resistor
it also depends on the pin you use
tomorrow I try to build my own "milling depth regulator" it's not quite so hard to build as the ones you culd buy but I think it should do his job correctly
P.S. sorry for my bad english
I simply screwed a microswitch to a block of wood and it worked quite well
I have been meaning to make a better one. I think I will use a dowel pin with the ends ground flat for the plunger.
I'll try to "drehen" (rotate is not the right word^^) a sleeve of Teflon
ravennb: your english is pretty good, don't worry about it
turn, as on a lathe
yes, "turn" is the word I was finding too
yes right that is what i mean
parport pin 15 could I use it for the tool length sensor?
yes pin 15 is an input
I looked at page 122 of the integrators manual to find that information
integrators manual? I didn't find that manual
on your menubar at the top of the screen: applications / cnc / integrators manual
cool ubuntu 8.04 with emc. I used only 6.06 up to now.... ok but I haven't bootet on emc ubuntu right now
becaus it's installed on my work und I am at home
you can find the same information here: http://www.linuxcnc.org/content/view/5/5/lang,en/
FWIW, the IEEE1284 standard's "recommended termination" for the receiver includes a 1.2k pull-up to +5V. If your computer's BIOS offers EPP and ECP options, then the port should be confomring to IEEE1284. But nothing in this world is guaranteed :-P http://www.fapo.com/1284elec.htm
Make the plunger trip the switch as it slides past the switch rather than into the switch. That way it will not trash the switch when it does not stop.
it would be cool to have a second switch (before a hard stop) that is wired to limit/estop
The mazak tool measuring stands use two but the first to contact is a slow down, the second to contact is length.
But it would be good to have a limit before it crashes.
Nice idea cradek
the "slow down" sounds smart
I was surprised when I calculated the feed rate I can use while still getting full resolution - the switch is read every servo cycle so I can move one step per cycle, which is F15 on my machine
I did a double probe at first (first one fast, then the second one slow) but this did not improve the results
with the new probing codes you could go fast to the switch, and then slowly go up and measure the length when pulling off
it would be cool to add an "aux latch" input to one of the FPGA configs, so a probe input could latch the encoder values
you'd want a spring in the tool sensor though :)
you'd probably want to probe in the same direction every time. some switches may have well-defined trip points but not as well-defined release points
(but I don't know that for sure)
sure could be
switch hysteresis (trip to release distance) is a function of $$, microswitches as much as carpenter values (1/32) , best (eg mikon) is 1um
Mazak has a spring that goes into the hollow plunger.
ive suggested voltage drop before, its very reliable, just need on element insulated, tells hal when you touched, no moving parts in the sensor at all, 9V at 20mA is fine
What would you use as the points for that, tomp2?
no pint , a surface, like a 123 with a 1/32 greenglas plate beneath it
have a pint
and what happens during decel
no point, a surface, like a 123 with a 1/32 greenglss plate beneath it
I don't see how you can not have provision for moving past the contact.
hard glass under the 1-2-3 block ;)
dunno ray i do it all the time, i get close, i go in slow, and the forget that value and go in slower, i use delicate graphite tools that easily break and dont break 'em
I've never seen such a thing on a commercial tool measuring stand.
dont go past, stop when you touch, do it slowly
Even the renishaw versions have a slider breaking a light beam.
emc can capture the posn fast enuf to allow position capture on the flay, if lfy is < 1"/min
What happens when the switch fails or dirt prevents the voltage drop.
same thing as if motor runs away or renishaw swx break and you probe fast. i beguin safe and slow, i short just before to hear a beep signifying the circuit is working. nothing is failsafe
It has. I vaporized at .25 drill when the stand failed.
if it weren't for broken machines, where would we be :) check before use
Switched to 250 ipm rapid when it saw the first switch.
I suppose I did make a few bucks on that one.
dimas__ is now known as dimas_
is there a male-male extention lpt-cord available to purchase?
male-male wouldn't be an extension
I was wrong to think I have one which fit to connect my interface board
correct not an extention but one which I could connect interface board with
dimas_: sure.. male-male, female-female and male-female are the common ones
(and obviously female-male..)
* alex_joni thinks a laser-distance meter would be nice for probing
just go to XY, and measure Z from the tooltip
(might even work with M66 & co)
alex_joni: thanks, seems I could buy it tomorrow if not find in the garbage
[18:51:16] <BigJohnT> http://www.cnczone.com/forums/showthread.php?p=453764#post453764
What do you think he's trying to say? Mach is cheap and EMC takes time?
I'm formulating my response now LOL
Sounds about right, you pay one way or the other, but open / free stuff is WELL worth the effort :)!
Ah finally found your note that he seems to have responded to.
Wish I could figure out how to use that site in some consistent way.
seems like Tom likes to bash EMC dunno why...
[19:01:37] <BigJohnT> http://www.cnczone.com/forums/showthread.php?p=453829#post453829
remind me not to buy anything from him
not that I do, being too cheap to pay for anything
looks like the mail server at sf is in trouble again.
we get what we pay for, I suppose
You mean like, "What's a bird say when it flies over jepler's house?"
DO NOT CLICK THIS LINK. It is a trap for badly behaved bots. If you follow this link, you will be
IMO we get what we work for.
unable to use the site afterwards.
when I asked some questions about building a THC he basically told me that was to advanced for a hobbyist
hobbyists are most underrated
I wonder who's life it is when he says "free upgrades for life?"
without emc commercial soft would cost 10times the actual price.
it does cost from 10 to 100,000 the actual cost
rayh: with emc you get free upgrades for your lifetime, but no guarantee you don't have to write the upgrades yourself...
BigJohnT: in your video it looks like the cutted strip bends heavy while cutting. Do you justify whith squeezer?
[19:38:00] <alex_joni> http://www.youtube.com/watch?v=uuGaqLT-gO4
that's quite nice
that is neat - nice find
if people want to compare the number of axes, let's add some more. we're out of letters though, so we'll have to call them $, %, and &
gefink: I just got back
I'm not sure I understand your question?
it looked like the strip bended from heat
oh Tom as in CaudleT, not me ;)
gefink: no, it bends down because I did not have the supports under it yet. I was cutting the support slats in the video
haha @ people plugging mach
with mach, screen blank, screen saver, opening files, running apps, a whole lot of shit would dwell the machine
and your out $160
emc is like, run a cycle and browse the web, on something as bloat as gnome, xscreensaver kick in, machine is like whatever
gefink: the machine was making itself!
at cnc workshop we repeatedly hear that only programmers can run EMC. I heard it from two different people last year. I also noticed that neither of them came to see the EMC demonstrations.
cradek: you hear that all the time from mach people
axis is easy as fuck to use?
but it doesn't bother me one bit. they can do whatever they like.
BigJohnT: nice machine
proof in point Gamma-X did it
sometimes people think editing a text config makes you a programmer
lol i see my name but what context was it used in?
you ran EMC LOL just checking to see if your awake
and with the competing prduct, there's no way you'll get free high quality jmk chew-outs
yes but pid is wacked!
ok your just not done
but you did it :)
u guys gotta come up with sumtin like an auto mated thing.
like an auto tune for PID?
* skunkworks_ has gotten chewed out by jmk.. It is a right of passage.
that stands for "auto tuning PID"
is that a python script?
that stands for "your machine might vaporize"
no, it's the experimental auto-tuning PID component
doesn't always work, and hasn't been futzed with much
SWPadnos, that thing does not work lol
I have played with it a bit.. (at_pid) but not on iron.
well, you can't tune out backlash
i think emc would be better with a good auto pid
Gamma-X: are you familiar with our policies regarding contributions?
well, let me clue you in
write it yourself, dammit! :)
your in trouble now
am I about to get my ass chewed?
send us your shiet, and we'll include it\
if it's nice
id have no clue what the hell id be doin, think of what i just said like putting a comment in the wendys suggeston box....
but especially if it turns AXIS buttons pink
* alex_joni hides
I have a silly question about shaft couplers. (i can hear u snickering already). I saw a wired HOWTO an they used vinyl tubing to connect the motor to the lead-screw. Does that sound lkie a good idea?
I,m looking for a simple solution for the couplers, any reccos?
Guest370: lots of people have use garden hoses for couplers
it really depends on the application if that's enough or not
no use hydraulic hose it's much less flexible and has a wire braid
like how much torque you need, etc
how much twisting is acceptable
I've connected up to 5 hp with it
the trick is not to leave much space between the shafts
outboard motor alex
I am worried that the motor shaft and the screw aren't quite parrallel and a metal tube might really make a mess.
Guest370: can't hurt to try
I took a lawnmower engine and mated it to the lower unit of an outboard
(and it's not that expensive either)
yeah, try an see after all the only thing that should break is a little cheap hose.
you won't break the hydraulic hose
hydraulic-something from a hardware store or Harbour Freight--Home dPot, etc.?
o'reilies auto parts
oh yeah! auto parts.
airhose and hose clamps
or any auto parts store that makes hydraulic hose
have them cut it to length cause you won't like to
depending on the torque.. Most air hose is fiber-reinforced
(easier to work with)
i'm not running it very fast, and cutting through very light material (linoleum) the friction from the screw beearings seems to be where most of the torque requirements will be.
in fact, I'm kinda worried about that, seems like the screws should move very freely, but these take a little work to move by hand.
that does sound bad
I'm using a xylotex board kit and bought a pre-built kit.
did you check alignment and gibs etc?
checking alignment--good idea
ways = the bushings the screw is running through?
* BigJohnT going to my fab shop see you guys later. Oh Gamma-X when you get the auto tune done Alex will put it in for you
the kit is mdf--bye BigJohn and thanks!
Guest370: ways are the parts that slide against one another when the axis moves
come on you will get it done Gamma
i think i'll contact the designer and ask him bout that--he should have some specific advice for me. But generally, your machines have minimal friction or binding of the lead screws?
how big a machine?
if it's small, and there is no motor hooked up, you should be able to freely turn the screws with two fingers
but a stepper motor can cause significant drag especially if it's wired up to a driver
* skunkworks_ bets he clicked on a link.
cradek--sorry--i lost the connection. it's a tabletop machine--http://www.modularcnc.com/index.php?p=detail&pid=3&cat_id=0
neat. yes I think the screws should turn easily. The movement should feel free.
hmm--i must check that out!
-Lead screws are 1/4-20 with new double backlash compnesating drive nuts.
that's the one.
hm maybe not. these may cause some drag depending on the design.
it seems like i need to have the nuts fairly tight to the bushings or bearings to work properly, i think.
oops--i gotta go now--thanks all for tremendous help!
Heh, only about $100 or so more than what I am building. Might have been smarter...
pminmo: what's up?
hey, questions on cncworkshop and emc
any specific days
for emc set aside?
[21:43:43] <skunkworks_> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?EMC_Fest_2008
[21:43:52] <skunkworks_> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?CNCWorkshopClasses2008
gives a bunch of info. Normally the emc gurus are there all week.
thanks, I'd sure like to get EMC2 running, but I keep running into wireless connectivity issues
Is this a desktop?
I can't have a wired connection where my workshop is
lots of people like the ethernet <> wireless bridge.. (it took me a few tries to get the wireless card working on the laptop and dapper. I ended up using fwcutter that uses the windows driver in linux)
not sure if your referring to a piec of hardware or software. I'm linux clueless
hardware - you hook it to your wired network card... It is pretty transparent to linux then. (normally the wired network cards work just peachy)
I am trying to find a link.
I can hookup wired temporarily, and get things running
but I keep hitting a stone wall on wireless, have tried three different adapters
I downloaded the non release 8.04 live cd and have it running wired
that went pretty smooth
something like this http://www.linksys.com/servlet/Satellite?c=L_Product_C2&childpagename=US%2FLayout&cid=1134692497433&pagename=Linksys%2FCommon%2FVisitorWrapper
except find something cheaper. :)
pminmo: How do you like emc2?
I've not gotten to the point of thrying to run it because I can't transfer files
have you looked at the ubuntu forums and searched for your hardware?
I am sure others have had your issues. http://ubuntuforums.org/
oh yes, the three adapters I have all are supposed to be able to be installed nbd. but each one I have tried some installation issue comes up and I can't find a solution
Yeah, first pass with the revised Z worked quite well dispite the crap fibreboard I used to run it on :).
Now to figure out how to generate gcode. Appears to be nothing under linux to do that (draw, generate gcode).
K`zan_emc: try qcad and http://188.8.131.52/translate_c?hl=en&langpair=de%7Cen&u=http://www.christian-kohloeffel.homepage.t-online.de/dxf2gocde.html
or whatever fits your fancy on http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Cam
unfortunately svg->rs274 is still crappy
* LawrenceG finally got around to documenting my breakout board for Jymm http://members.shaw.ca/swstuff/bob.html
Anyone used this: http://www.gcad3d.org/
or is everyone using eagle and something to gen gcode from it?!?
K`zan, adding another picture of control box..... just a minute
Gotta do chow, bbiab.