Machine configuration file is 'minitetra.ini'
Error in startup script: expected integer but got "100.0"
"after $imgtime exit"
(file "/usr/share/emc/tcl/bin/popimage" line 72)
starting the hexapod-sim configuration..
it seems to be harmless
in my tree I have the pnp_unregister_driver removed
03cradek 07TRUNK * 10emc2/src/configure: new version string
03jepler 07v2_2_branch * 10emc2/tcl/bin/popimage: work properly if INTRO_TIME is not an integer
cradek: can you pastebin the probe_parport patch (or check it in) so I don't have to do it again?
03cradek 07TRUNK * 10emc2/src/hal/drivers/probe_parport.c: this breaks the parport on the second run of emc on my oldish (P3) machine.
* cradek stares dumbly at the plug on his probe
03jepler 07v2_2_branch * 10emc2/src/hal/drivers/probe_parport.c: merge rev 1.5: pnp_unregister_driver was harmful on some systems
cradek: thank you
03cradek 07TRUNK * 10emc2/configs/scara/scara.ini: at least it runs this way...
03cradek 07v2_2_branch * 10emc2/configs/scara/scara.ini: backport: get scara to run again
03jepler 07TRUNK * 10emc2/tcl/bin/popimage: show emc version number on splash screen
freenode seems a little shakey tonight
hmmm. you think so?
:} hi SWPadnos
I havent tried 2.2.x yet... but I am running cvs from about a week ago.... nice work.... I am putting together a new controller for my lathe so I can use all the fun new features
cool. I think there was a rash of fixes just before release, you may want to update (either CVS or installed)
yes... I keep a pretty close eye on this channel http://imagebin.ca/view/PFWDdrQ.html
is the parts collection
yes... I keep a pretty close eye on this channel http://imagebin.ca/view/PFWDdrQ.html
is the parts collection
looks like it's all done except for some wiring
throw it in the box and hit the switch.... hi Chris
and maybe a little drilling ;)
true, just spray in some epoxy and toss the components in
wire them wherever they fall
a collection of bits from the junk box... I have been hoarding some of theat stuff for 25 years... time it got some power on it!
its not to practical from a new parts point of view, but it will be nice to have in one box... a real control power relay, spindle relay, mist coolant and a real watchdog loop with emc
heh. just make sure to replace all the electrolytic caps first
(at least that's what Yaskawa recommends for their servo drives, every 5 years)
5 seems excessive if they are in use
I can see it on servo drives... they work caps pretty hard and hot
I think they're supposed to be inspected every year, and replaced every 5 anyway
a good excuse to get the service tech in every few years... repeat business
or in my case, no business (beyond the first test units)
most of which was bought on eBay anyway :)
right half of box is low voltage side, left is 120vac... plugs for computer (non switched), one controlled via control power, and one for the spindle
[04:35:59] <LawrenceG> http://imagebin.ca/view/r7Bc9rzT.html
power wiring dwg.... close to final version
interesting - there's a link at the top of that imagebin page to a "voicebin"
it doesnt seem to go far using my browser.... wonder what they are planning
audio samples I'd bet
or at least audio
kind of a slow version of skype!!!!
maybe the next thing is a binbin - just take any file ;)
or the french version bin-la-bin
or a disney one - a chitty-chitty-bin-bin
to stay on topic... ngcbin
email machines around
would be great if I could email my chips to someone.... the garage needs a good sweep
that would be the dustbin
SWPadnos_ is now known as SWPadnos
SWPadnos_ is now known as SWPadnos
sourceforge and linuxcnc.org announcements made
cradek: sorry about last night.. got home and went to bed, slept about 13 hours till this morning :)
alex_joni: in rt, i don't see that probing problem. It might be a problem with -simulator
hm, anyone gotten this since installing 2.2.x?
mandb: warning: /usr/share/man/man3/hal_unlink.3hal is a dangling symlink
what shows that?
a nightly e-mail
I guess I don't get those
must be another unpackaged file .. hm
install -m 0644 -o root ... ../docs/man/man3/hal_signal_new.3hal ... /usr/share/man/man3
aha -- emc2-dev.files:usr/share/man/man3/hal_signal_new.3hal
03jepler 07v2_2_branch * 10emc2/debian/ (emc2-dev.files emc2.files.in): hal_unlink manpage should be in -dev package
03jepler 07v2_2_branch * 10emc2/debian/changelog: note packaging bug fixed
heh - my picture is in the paper. Last night the astronomy club had a 'show and tell' on what telescopes to buy and what not to buy. I look like a gangster with my hat backwards looking thru my scope..
skunkworks: lol, thought you were looking backwards through your scope
I am really starting to be annoyed by mariss. http://www.cnczone.com/forums/showthread.php?t=46205
because his opinion is [surprise] that his own drives do it the best way?
"That is the way things are done now." <- heh
wonder why I haven't seen any industrial stuff doing it like that
hmm, isn't it silly to be annoyed by that?
all I can see when I look around are AC servos with resolvers
and CAN to talk to them
I agree those drives usually have PID inside
but that doesn't mean they don't report position back
in order to synch multiple joints
with gecko servo amps can you disable them, move the machine, and get position feedback to the pc? or when there's a crash/following error, can you resync position somehow?
(I don't see how these things could work with step/dir)
also load/overload sensing
of course you can't
you just drive them and hope they won't error
Don't get me started on my experience with Mariss.
it seems like the traditional system is better in lots of ways.
oh hi ray :-)
I think I know how you feel about this...
cradek: well.. that depends who uses them
if it's home users with no/little expectation.. imho they should go ahead and do what they please
I have some apps where I used G3xx drives, and am quite pleased with them
but never to replace the need for closed-loop
wonder what would one of us have to pay cnczone to make our opinions as important as his
probably a banner for a year or so :)
another way to make your opinion count is to just write a lot and do it LOUDLY ;)
I'm off to find coffee
I'll be heading home
cradek: hope I'll be around tonigh.. but not sure yet
you can run feedback to both the gecko and EMC
SWPadnos: yes, you can
but not FF
but most people don't see the need, because they spell EMC "Mach"
and you can either have 2 loops
or no loop in emc, and use it for feedback only
the geckos are great, but you can't tune I, and there is no FF
hi chris, alex
a graph widget for pyvcp to display following-error history would be nice...
and I think the P is the same as Proportional, but the D pot on the drive is Damping, not Differential
(which may be the same thing, I don't know)
Gotta run guys. Catch you all later.
maybe I am a bit crabby today. :)
I sure am (now)
SWPadnos: how is the project coming?
it's truly amazing when phone salesmen refuse to get off the phone
so now it's time to log another complaint with the FCC
my part is basically done
I'm going to add a couple of features
which I guess I need to do today and tomorrow, since I think I'm leaving on Sunday for a photo shoot
maybe I'll actually find out who we're shooting sometime ;)
hah "Other servo systems rely on black boxes and technical jargon to keep public inspection and comment from affecting opinions."
03cradek 07TRUNK * 10emc2/nc_files/smartprobe.ngc: better probing algorithm
cool, that was fast
03cradek 07TRUNK * 10emc2/src/emc/task/emccanon.cc: 0 vel/0 acc moves take a long time
[16:37:32] <cradek> http://timeguy.com/cradek-files/emc/smartprobe.png
jepler: if we do what we were talking about, you could easily (print,G1 x#5061 y#5062 z#5063) etc
that would be a pretty neat trick
those numbers are now always in machine coordinates, aren't they .. ?
so this might make no sense: G0 X#5061
no, they are in the chosen coordinate system
so, unlike offsets and so forth
they're not useful otherwise
in my hemisphere they go from about 0 to 1
can one of the new G38.x codes be abused to store the current location in #506x without doing any motion?
g91 g38.3 x0 f10
so now the probing codes are: 38.2, 38.3, 38.4, 38.5
.3 is "until probe not in contact"?
.4, .5 are like .2 and .3 but without an error if no change in contact?
2,3 are probe 'down'; 4,5 are probe 'up'; 3,5 suppress errors
'down' means it expects to go from no-contact to in-contact
all report tripped or not in #5070
looks like you can get thousands of points without getting bored watching it now
3-4 a second
(if they're close together)
so after a "successful" g38.5 where the probe went into no-contact, what value is in #5070?
1 = the probe changed state as requested
0 = the move completed but the probe never changed state as requested
so the "up and over" algorithm is while [#5070 NE 0] up, over to target, endwhile
it will eventually be high enough to get 'over' to the requested target, at which time the probe move will complete without changing state
jepler: the handshaking between the various bits is definitely broken in -simulator. everything works perfectly in rt
cradek: that's too bad
yeah but it's not that important
we just have to remember not to look for bugs that aren't there
I'm writing docs for the new g38.x; want to come read them before I check them in?
[16:59:22] <cradek> http://timeguy.com/cradek-files/emc/smartprobe2.png
I think this is 2500 points, it took about 20 minutes
hmmm. I wonder if it's possible/desirable to have AXIS show probed points with a little glyph
03jepler 07TRUNK * 10emc2/docs/src/gcode/main.lyx: g38.x documentation
again, you're my hero
03jepler 07TRUNK * 10emc2/docs/src/gcode/main.lyx: explain debug, and print, messages. add more labels
cradek: is that in sim-mode or with a real machine?
[Global Notice] I apologize for the trip through splitsville the past couple of days. Apparently we've annoyed a kiddie and he doesn't have anything better to do than packet a few of our servers. We're working with our sponsors to block the traffic.
bounce bounce bounce
there's a server message explaining...
and, I didn't know "packet" was a verb
jepler: it's interesting that you could probe more points in places where the slope has a discontinuity
(I'm looking at the edge of my circle)
anywhere the radius is less than the probe radius you mean?
not really, I mean anywhere there's a sharp feature
if you look to the 'left' and 'right' and see wildly different slopes, you know something interesting happens 'here'
so, do some recursive search until you're sure you've found it to whatever tolerance you want
yeah, sort of
[19:31:47] <cradek> http://timeguy.com/cradek-files/emc/probe-results.png
see how well this grid defines the surface of the sphere, but the sphere/plane intersection circle is bad
isnt it "equally bad" over the surface of the sphere?
and, you could just do waterline instead of a grid
i can see your point though, for something with a complicated parting line
no, I think it's worse when the surface is nearly parallel to the probe
but yeah, waterline would be cool for some stuff (except you can't find inside features)
for a convex shape it would be great
nice stuff on probing :D
looks really fun :D
makes me wanna build a probe :D
or get one somewhere
gene had some ideas - maybe it'll start another discussion
CIA-18: are you broken again?
he has a homemade probe I think
* alex_joni kicks cradek
* alex_joni kicks CIA-18
seems like it
[20:26:57] <jepler> http://www.indoor.flyer.co.uk/probe.htm
btw, error reports in #cia didn't do much good
likewise reporting at code.google.com remained unanswered
* alex_joni feels the same way
One thing I realized is that the scanning process can be very slow. The height at which you begin the scan which I the feedplane must be higher than the highest feature within the area to be scanned. If object has a combination of high and low features then the mill can spend a lot of time raising and lowering the probe.
it wasn't much different for emc2
played with another gutsy today.. works very nicely
could the new "smart probing" stuff have been done with M66?
probing is special because it records the position when the probe trips, even though the machine is still thundering along at full speed
fenn: M66 waits for an input
but motion is not happening during that
could you do a custom m-code to write the values?
fenn: I doubt that
actually you could, but it would suck, there are only two arguments to custom m codes
m101 P#5220 Q#5221 etc
I think enhancing (debug,...) or (print,...) to write to a file is what we want
(log,) (logopen) (logclose)
fenn: I don't see: 1. how you can command moves from a custom M-code
2. condition that move to stop when an input happens
alex_joni: you arent commanding moves from the m-code, you're just printing values to a file
3. write to a file (that might be easiest of all 3)
fenn: ok, printing works.. but only 2 params
(log,G1 X#5061 Y#5062 Z#5063)
alex_joni: so there's no way to stop a move with m66, only start a move?
alex_joni: do you think more than one open file is needed?
jepler: who knows..
I wouldn't limit it to one
g1 / m66 / g1 would move to the first position, then do the input action, the move to the second position
if we write this from scratch
sorry, that was a troll
(log would do append.. right?
then you could do just (log, "</path/to/>file.txt", ...)
I don't want to have to mention the file each time
maybe (log, ..) would write to the last one opened?
(log,I don't like that idea, but I can't explain why)
you're no fun :P
would you prefer a file handle?
yes that's more or less my idea
but it assumes one file is enough
one at a time
ok, I can live with that
who wants something else, can do O-functions
to close/open various files
can you pass strings to o-subs?
fenn: don't think so
but you can call O100 call log1
O101 call log2
alex_joni: I'm not sure what that means, O100 call log2
that's a procedure call : to something that does logclose, logopen file2
but I said O101 call log2 (different procedure name)
not sure if I'm making much sense..
jepler: answering your previous question, yes, but you have to move: g91g38.3x0.0001f1
any amount of motion I think
z.0001 seems like the better move
a 'move' that doesn't move is not issued
OK, I updated my bog entry about the new probing stuff to reflect that
maybe (debug) should be (message)
mmmmaybe but unfortunately we released with (debug)
hmph, in lyx the line on the right-hand side of the table is present, but in the html version it's gone missing. http://www.linuxcnc.org/docview/devel/html/gcode_main.html#sub:G38.2:-Straight-Probe
message is too close to MSG which pops up a window
oh, debug pops up a window too
cradek: doesn't the probe have something circular at the tip usually?
or does it get idealized as a point?
they usually have a sphere at the tip
well, I was wondering about the "cutter radius compensaion must be off" part
but as I think about it, it's quite hard to reason about the actual contact point
I think this is better done in some analysis program afterwards..
you could use a point if you assured that it was always pointing in the direction of motion
and wear wasn't a concern
(find the shape which goes through some spheres around the logged points)
fenn: but it seldom is (pointing in the direction of motion)
think about probing a sphere from above, with strictly Z moves
use a probe with a sphere end that matches the tool you'll cut with -- then you're fine
in general, I suppose it's a 3d version of polygon offsetting -- you take the points you got and offset them by the diameter to find a possible surface that matches the probed points
yep a simple matter of CSG
jepler: can I ask you a simple thing?
I can probably do it myself, but it'll take me lots longer..
but I reserve the right to give an evasive or useless answer
the "only?" user of teach-in.py asked me if we could add the line number to the written file
also spindle, mist, coolant, lube status
I'm sure all of that is possible
someone uses lube status?
'flood', 'mist', 'lube', 'lube-level', 'spindle_direction', 'spindle_enabled', 'spindle_speed' seem to be the python names for the items in the emc status buffer
you'd just add them to the 'print' statement
to keep track of line number, you'd put at the outermost indentation level: "linenumber = 1". In a function that modifies linenumber, put "global linenumber" at the top. e.g., "global linenumber; print linenumber; linenumber += 1"
and string joining is done with . ?
'print a, b, c' prints a, then a space, then b, then a space, then c
then a newline
otherwise, if a and b are both strings, then a+b is also a string
"." is not used for string concatenation in python
is it 'print a, b, c' or without the '' ?
without, of course
ok, cool.. I think I can manage now :D
alex_joni: soon you'll forget all about that c++ silliness
fenn: what does tracking-test do?
No manual entry for tracking-test
can't say I didn't try :P
* alex_joni doesn't understand cvs anymore..
teach-in.py is in src/emc/usr_intf/axis/scripts , right?
find / -name teach-in.py
(it seems like that by looking at the cvsview), but when I do a cvs diff I get this:
juve@taurus:~/emc2.TRUNK/src/emc/usr_intf/axis$ cvs diff
juve@taurus:~/emc2.TRUNK/src/emc/usr_intf/axis$ less scripts/CVS/Entries | grep teach
if it'sa new file, does cvs diff give useful output?
cvs diff doesn't say anything
I guess the answer is no then ;)
do you have that file in your checkout?
and I changed it..
oh. then I'm no help
I think I know what happened
I put the file in my local repo there before jepler checked it in
after that cvs didn't want to check it out because it colided..
ah, so your cvs Entries doesn't show it a s amanaged file yet
nope.. but I renamed it, checked out the one from the repo
then replaced it, and it seems happy now
heh.. we can sync ntp servers after jmk
cvs diff never does what you want it to do
oh, it does the trick for me..
fenn: so what does tracking-test do?
it tests the Track class in vismach.py, and is a fun toy that shows off hal_input with the mouse
you're supposed to do halcmd -fk configs/sim/tracking-test.hal
I always do -kf , I thought that sounds politically more correct than -fk
anyway, tracking-test shows off a bug in Track that i was hoping someone else would figure out
jepler: YM G90 HTH
it "mostly" works, except for at low values for z
fenn: I saw that.. but I thought it's kins
alex_joni: does the cone follow the mouse when you run tracking-test?
* alex_joni plans a special move
wonder if a probe will trip on my way to bed
good night all