it looks like the screw comp data is not stored in shared memory
so I don't see any real obstacle to enlarging the limit
(its a #define constant)
of course my machine is using an installed EMC 2.2.2.... if I go with a larger comp file I'll have to run CVS until 2.3 comes out
that would be unfortunate I think
although I'm guilty of it too
hm, exiting emc doesn't turn off the spindle. I wonder when that re-broke
regarding running CVS, it may or may not be bad, depending on your POV
chris... thats what the watchdog is for
if it means the upcoming code gets more testing, its good
yea baby... it works a treat
sure, but it used to turn it off before exiting.
why did you break it?
isn't spindle controlled through motion?
does the GUI send NML messages on shutdown, to tell motion its going away? I thought the runscript did the cleanup and shutdown after the GUI exited
I wish I could remember. alex fixed this last time it broke.
(nothing in task seems to stay fixed)
yay, I have home switches
fucking shit man just prepare for the worst and never stop
its like folding laundry
somehow, it's repeating to within about a step
tenth indicator shows sometimes 0 and sometimes .0001
ouch thats horrible backlash
hey guys... take the rest of the year off Thankyou for all your good work on emc
cradek: I don't quite remember either
LawrenceG: can I do that too?
this bites, my w_tool_length branch is diverging, and I'm not sure what to do about it
diverging because of changes on trunk? or just your stuff?
the combination of the two
would merging your branch have bad consequences for trunk?
yes it would break TLO for 'normal' machines
alex_joni: opps... not sure what you are asking...... I think you have already expired your year off :}
cradek: how normal?
it hackishly applies the Z TLO to W instead
we need a more general solution
not sure what else is important in there - looking
and can we say W is Z for regular machines?
'regular' machines don't have a W axis
or is that restricted to XYZA machines only ?
hmm.. I'm not beeing clear..
I mean.. let's have UVW always exist
TLO gets applied in UVW (which matches XYZ usually)
hot wings, cheese fondue and home made french bread for new years feast... supplemented with rum and coke... that seem kind of international in scope
when a machines uses ABC to rotate UVW so does the plane where TLO gets applied
no you have to move W to apply it. That breaks all gcode that applies like G0 G43 Hn Z0
it's not quite clear to me.. but I'll take your word :)
TLO doesn't actually get applied to an axis until you move it
* alex_joni dislikes g-code :)
gcode is great. it lets me move my machine around
I mean some of the specs :)
I know, I'm just being silly
but I actually think gcode is fine
it's complex, but that's because the task at hand is complex
I'm still trying to find a way for g-code to do what I would want with a non-trivkin
and I don't think it's possible..
so I'll probably end up with a different interp anyways
man today I got a flash card the size of a potato chip crumb
kins_both definitely works better than inverse_only
for one thing it doesn't give a warning at startup
assuming you have formulas for both directions ;-)
and assuming you dont get lost
woo whooo - finally - a utillity tub in the basement.
good night all
heh, you can hear the periodic error correction
change in tone as it is running thru its travel?
yep - I have a once per rev error of about 0.0006, thats the one you can hear
its about a 1.2% change in velocity
looks like screw comp works quite nicely
less than +/- 3 tenths error and less than +/- 4 tenths lash over the full travel range
less than two tenths lash over most of the range
wow that sounds great
next I gotta do X - that will be much simpler, since I'll only need to check for scale error and overall (average) lash
probalbly take a measurement every inch
SWPadnos_ is now known as SWPadnos
EMC: 03cradek 07w_tool_length * 10emc2/src/emc/usr_intf/axis/Submakefile: stuart stevenson's visualization of the viper machine
jepler: in halgui, in file canvas,
in HalCanvas.sig_event theres a call "realwid.move(..."
i put debug prints in file data Signal.move and Component.move.
The widget moves, but neither debug print occurs.
( trying to make siglines visible before they are dragged )
tomp: if your print isn't printing, then probably the line is never being reached. you must be mistaken about which code you are triggering when you try to move something.
yes, i'm sure i'm not following the real code flow
I don't know anything about that "halgui" that is in emc's cvs .. somebody else wrote it and I don't think I've ever taken the smallest glance at it.
ok, btw happy new year
is that a 5 axis mill?
yes another 5 axis configuration
I did a gantry, I think this is called a bridge (the table moves in one direction)
I need to thank all of you guys who do development of EMC2. I think it's come a long long way very quickly.
cradek: both rotaries are still stacked on the spindle? or is one on the table>?
stacked on the spindle
DanielFalck: thanks for the kind words
its a shame that config is so hard to actually build (what with motors not being infinetsimally small and such)
yeah it's a very nice configuration
spindles are heavy too
at least heavy-duty ones are :)
I haven't weighed the spindles I have here
yeah I don't know what's so hard about it, I made a 5 axis machine :-)
I'm guessing 10-15 lbs by the time I put a housing on it
SWPadnos: dunno if you heard - I found some potentially nice spindles at HGR
oh - cool
I didn't see that (that I recall)
[16:53:05] <jmkasunich> http://jmkasunich.com/pics/surplus_spindle.pdf
I got 14 of them in varying conditions for $120
oh - is one of these the one that you took apart (with the custom tool you made)?
* jmkasunich <-- the spindle baron
I'm hoping I can convert them to use tormach tooling, and sell a few
I've got small motors with encoders, does that count for a knighthood or something?
I'm intrigued by the 5 axis machine I saw that had one of the rotaries at a 45 degree angle
fixed 45 degrees?
yeah, I'm trying to find the youtube vid
I'd have to see it to be able to hypothesize why could be "better"
woohoo - read or deleted all the email from the last week!
[16:58:44] <jmkasunich> http://www.youtube.com/watch?v=nFqUMKlCCFY
the orientation of the rotary axis lets the tooltip remain near stationary
about 0:50 in the vid you see it rotating
so you can easily change the orientation, but not so easily in a contouring move
works for contouring too (there is also a C, a rotary on the table)
that's pretty cool visualization software they've got there
I'm sure its not just viz - its the CAM
at 3:38 you see 5 axis contouring
it's pretty cool for either
yep, I see it
I have a 1Kw 5000 RPM motor that is 5 x 5 x 9, and a spindle that is ~8" from nose to back bearing, ~9" from nose to where I'd put a pulley
I've been trying to figure out how to make an assembly that allows at least one angular DOF with that
EMC: 03jepler 07TRUNK * 10emc2/src/hal/components/ (mux2.comp mux4.comp): improve documentation
EMC: 03jepler 07v2_2_branch * 10emc2/src/hal/components/ (mux2.comp mux4.comp): from TRUNK: improve documentation
EMC: 03jepler 07TRUNK * 10emc2/src/hal/components/ (and2.comp or2.comp xor2.comp): improve documentation
EMC: 03jepler 07v2_2_branch * 10emc2/src/hal/components/ (and2.comp or2.comp xor2.comp): from TRUNK: improve documentation
cradek's 5 axis visulation, for real: http://www.youtube.com/watch?v=L5937BZiMmo&feature=related
EMC: 03jepler 07TRUNK * 10emc2/src/hal/components/ (lowpass.comp steptest.comp timedelta.comp): add missing license statements
EMC: 03jepler 07v2_2_branch * 10emc2/src/hal/components/ (lowpass.comp steptest.comp timedelta.comp): from TRUNK: add missing license statements
EMC: 03jepler 07TRUNK * 10emc2/src/hal/components/Submakefile: Use --require-license to make the build process complain if .comp files do not specify a license
EMC: 03jepler 07TRUNK * 10emc2/src/hal/utils/comp.g: Use --require-license to make the build process complain if .comp files do not specify a license
EMC: 03compile-farm 07BDI-4.51 (18.104.22.168-rtai) * 10emc2head/: build FAILED ; see http://linuxcnc.org/compile_farm/emc2head_slot6_log.txt
EMC: 03compile-farm 07Ubuntu 5.10 (breezy) non-realtime (2.6.12-10-386) * 10emc2head/: build FAILED ; see http://linuxcnc.org/compile_farm/emc2head_slot1_log.txt
uh oh did I get it wrong?
EMC: 03compile-farm 07Ubuntu 5.10 (breezy) realtime (2.6.12-magma) * 10emc2head/: build FAILED ; see http://linuxcnc.org/compile_farm/emc2head_slot2_log.txt
serport.comp, no license
EMC: 03jepler 07TRUNK * 10emc2/src/hal/drivers/serport.comp: add license declaration
EMC: 03compile-farm 07Ubuntu 5.10 (breezy) non-realtime (2.6.12-10-386) * 10emc2head/: build PASSED
EMC: 03compile-farm 07Ubuntu 5.10 (breezy) realtime (2.6.12-magma) * 10emc2head/: build PASSED
that's what I get for compiling on a sim system
EMC: 03compile-farm 07BDI-4.51 (22.214.171.124-rtai) * 10emc2head/: build PASSED
isn't it nice to have a fast compile farm?
jmkasunich: it's so cool to have the compile farm
and yes it sure is fast these days
jepler: fine by me
darn, running andy's config and gcode works fine for me
EMC: 03jmkasunich 07TRUNK * 10emc2/src/hal/user_comps/devices/hal_joystick.c: raise limits - some joystick-like devices have a _lot_ of buttons
jmkasunich: I believe that joystick should be deprecated in favor of input
jmkasunich: for one thing, joystick doesn't call hal_ready, and I don't see how it can
I'm not opposed to that (I think)
I think that input covers any device joystick does, and more
I wrote joystick for an engineering week project 2 years ago, and haven't used it since
Martziz (on #emc) says he tried input, and had funny results, so he's using joystick
* jepler scrolls back
skunkworks: works for me (yes some names changed)
That is what I figured.
cradek: truetype tracer does single straight lines of text only, right?
if I want to do curve text or something I'd have to dxf it, then rotate and shift the letters in cad, and then somehow do cad to gcode
with only a little work you could assemble a series of lines using offsets in gcode
yes curving would be another story altogether
probably easier to modify truetype-tracer if you want curved
multi-line output, centering, and multiple font sizes would all be nice - would you consider those to be good features for TTT, or things best done some other way?
obviously scale handles sizing, offsets can handle multi-line and centering one line under another
I've almost written multi-line - it's very easy
more complex when you want each line a different size, or worse a differnent font
for each argv[n] reset the x offset and go down in y by the amount specified in the font
yeah that stuff is not going to fit in the 'very simple at the command line' paradigm
you need to parse xml and ... oh forget it
I need to learn how to use offsets now
I just thought of a handy feature, maybe I should find the source and add it
"print the height and width of the text as a comment"
helps figure out the offsets and such when composing a multi-line item
so - you don't need reverse kins because of no feedback?
cradek: or is it forward kins?
I thought it would not be necessary - but it's nice to have, so you can still move in joint space whenever you want
ah - I remember someone mentnioning that. did you write the kins in both directions now?
yes, it's very easy for this machine
was there multible solutions?
it's just very simple trig
you can tell since I was able to write it :-)
nice. (I don't think you have any math issues)
cool, I have home switches now, so I dialed in the center of the horizontal rotary
I pointed my wiggler at a speck of something on the table and it definitely stays pointed there
* skunkworks checks cradeks youtube space.
yer wiggler eh?
yeah it's funny - it's about as long as the entire Z travel of the machine :-)
I'm glad I got the one 3/8 holder - could never cut anything with it, but it's good for stuff like this
cradek: are you using the lathe-pluto setup on 2.2.2 from the cvs?
I'm not sure
I wanted to point the guy to the cvs lathe-pluto.. should be safe - right?
the hal file does look like it matches what I'm using
the ini is slightly different
ok - thanks
I think the hal file is what he really needs - that and a point to the docs.
thanks for helping him
least I can do
damn - this floor is looking nice.
one more wash and I think it will be done.