jepler: I _think_ that backlash only uses an additional 50% of accel
jmkasunich so do you need to double it?
EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/hal/pyvcp.lyx: add more info on widgets
probably 1.5x would do the trick
opps wrong window
note: I think, and probably ;-)
then I'll CMA and give a suggested range :)
EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/common/Stepper_Diagnostics.lyx: add info on using backlash
EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/config/ini_config.lyx: add info on using backlash
say goodnight Gracie
EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/gcode/main.lyx: shorten some index entries
jmkasunich: did you see my exciting video? http://www.youtube.com/watch?v=qY-FCN5ZXkg
that looks ... big
that's your driver?
it is not a 'servo' but it was light enough I could carry home to test ;)
it is off of a large conveyor belt.
hammering the piss out of it.. mosfets get warm with not heatsinks. (driver chips still working)
seems like you've accomplished your goal
next is the big ass servo. (and I need to make the larger power supply)
[03:58:35] <cradek> http://xkcd.com/37/
dang, your hermes is so cool
what a great engraving machine
heh - yah - it is now the garage floor. ;)
soon the building above it.
I had a heck of a time catching it - but the input of the comparator was going negative. the spec says never go below .3v and if there is a possiblity of it happening - it could put it in an unknown state. I put a snubbing diode across the input and so far so good. (time will tell) I need to get some schottky diodes
the comparator is the current limit sensor?
that's much less than .3
I meant -.3
heh, funny comment: http://www.youtube.com/watch?v=iDX_fa1KFI0&NR=1
wow - welcome back
especially if it gets to the east coast by Monday/Tuesday
wait - maybe I was the one gone
Lerman_______ is now known as Lerman
and back again
KimK_ is now known as KimK
I saw on http://cvs.linuxcnc.org/cvs/emc2/debian/changelog?rev=HEAD
that is written "converted many pins to parameters for added flexibility"
there should be parameters to pins
EMC: 03alex_joni 07TRUNK * 10emc2/debian/changelog: fix thinko (parameters changed to pins, not the other way around), thanks to micges for pointing it out
EMC: 03tissf 07TRUNK * 10emc2/docs/src/gui/halui_fr.lyx: french translation update
EMC: 03tissf 07TRUNK * 10emc2/docs/src/config/ini_config_fr.lyx: french translation update to follow John
EMC: 03tissf 07TRUNK * 10emc2/docs/src/hal/ (pyvcp_fr.lyx pyvcp_meter_fr.png): french translation update to follow John
EMC: 03tissf 07TRUNK * 10emc2/docs/src/common/Stepper_Diagnostics_fr.lyx: french translation update to follow John
EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/ (Submakefile docs.xml index.tmpl): add system req to html and fr one too
[13:02:24] <micges_emc> http://imagebin.org/35960
I think this is issue that joint velocity jump above limits
this is the config file http://www.pastebin.ca/1310016
joint limit is 80*60=4800
and axis.0.joint-vel-cmd is 83.3*60=4998
oh and gcode was F1000 FeedOverride is 5.0
EMC: 03tissf 07TRUNK * 10emc2/docs/src/ (4 files): french translation update changed copyright to 2009
EMC: 03tissf 07TRUNK * 10emc2/docs/src/common/Document_Header_fr.lyx: french translation update changed copyright to 2009
EMC: 03tissf 07TRUNK * 10emc2/src/po/fr_axis.po: french translation update changed copyright to 2009
EMC: 03tissf 07TRUNK * 10emc2/docs/src/common/Stepper_Diagnostics_fr.lyx: french translation - fix typo
in the docs on linuxcnc.org in section about G73 there is letter L in description, but when I'll MDI it error shows up "L word with no G10, cutter compensation, or canned cycle"
[15:00:39] <micges_emc> http://www.linuxcnc.org/docview/html//gcode_main.html#sub:G73:-Drilling-Cycle
cradek: that URL corresponds to the v2_2_branch documentation
on that same page in section canned cycles there is no mention about G73
hmm, L is not listed in 1.3 Word table
micges: it is a bug, thanks for noticing it
umm I don't think G73 is actually in v2_2_branch
that's also a doc bug then
but trunk's g73 has a bug with not accepting L
RELEASE_2_2_8: 1.13, but G73 was first added in revision 1.14
is "L" not a valid word for G73?
BigJohnT: it should be, but due to a bug in trunk, it is not accepted
"L" (repeat) is accepted for all canned cycles
hm, something is wrong with all canned cycles in trunk - rapids aren't
EMC: 03cradek 07TRUNK * 10emc2/src/emc/rs274ngc/interp_check.cc: allow "L" repeats with G73
micges: there are no joint constraints in emc 2.2 or in TRUNK. joint constraints remain unimplemented.
only cartesian constraints are implemeneted
I thought I'd made this clear in the past
boy did I screw cycles up. looks like my regexp coding replaced all traverses with feeds.
cradek: argh, oops
yes I want to make sure that is it, rarely I see it on real machine (only from notes from operators)
I got an PM from a guy on cnczone saying he has an excellent servo drive schemantic and he would send it to me. It ended up being the hacked gecko drive schematic. That thread on cnczone is not gone. ;)
It does interestingly enough have a diode across the input of the comparator. And it is schottky ;) I get there at some point usually.
no heat here at work.. I meant 'That thread on cnczone is gone'
omg u copeid marisss?
marisss schematic has a big block where the power electronics are anyways.
yep I recall that
cradek: G73 is working now
EMC: 03cradek 07TRUNK * 10emc2/src/emc/rs274ngc/interp_cycles.cc: put back the traverses I accidentally changed to feeds
skunkworks_: I'm not sure what about that thread merited deletion
for instance, I'm pretty sure that producing and disseminating a schematic that corresponds to a circuit board you own is not copyright infringement
no - but cnczone gets paid advertising.. So mariss probably has a big say in what gets deleted.
yay web bbses
there was a big stink recently when a tormach thread got deleted.
that's easy; just delete the stink thread too
jepler: stinks cause ad revenue
heh - both attachments are also missing from cnczone. ;) I guess mariss didn't want his schematic out there.
sometimes you later realize a post was a mistake...
(same goes for a cvs commit)
[15:47:19] <skunkworks_> http://bbs.homeshopmachinist.net/showthread.php?t=32670
oh dear god
what disease or chemical imbalance makes people produce GUIs like that
Hey, that's great David! I like TurboCNC's GUI a heck of a lot more than the gauche Mach3 GUI.
dave seems like an interesting character..
kinda hard on the eyes
the gcode readout should be punched tape that scrolls by.
ooh I like that idea
What made me wonder though is if this is a true windows application what don't they use a standard font instead of the dot matrix font used.
that's great, but what about when the interpreter goes faster than 1200 baud?
that UI would translate well to a VFD though (Vacuum Fluorescent Display)
or a car radio or something
yes, a 5x7 vfd could do a decent job at displaying paper tape (7 bit only)
I like the first comment though :)
and the continuing rants about how programmers don't know a damn thing about anything
oh, I haven't noticed those yet. should I just top reading?
I bet I'm a much better machinist than those ranting machinists are programmers.
quite possibly, yes
that seems likely
(also, I bet not one of them has ever given me anything)
hmmm. looks like someone is misremembering their conversations here
you didn't stop reading, did you
that's the kind of remark I'd expect from him, given past experience
now I just have to integrate it with AXIS:
now I just have to integrate it with AXIS: Originally Posted by MTNGUN
To my taste, most of the gcode interpreters are written for hobbyists who like have lots of whistles and bells and eye candy to play with..
yes, because interpreters are so pretty
I meant to paste this: http://emergent.unpy.net/files/sandbox/tape.py
hmmm. it's 52 degrees in my office. I wonder if it's time to turn on the downstairs heat
jepler: you fool programmers don't know that nc tape isn't punched in ascii
makes me dizzy
for some reason, that made me think of the engineering senior who burned assembly code into an EEPROM, and expected the computer to boot
"see, it's right there" is what he said when we finally read out the PROM
(actually, it was probably just EPROM, since we had to use UV to erase back in those days)
this is kinda funny... http://www.cnczone.com/forums/showthread.php?t=71528
cradek: what encoding is nc tape in?
it must be EBCDIC, just to keep things interesting
I don't know... I don't know if there's a standard
seems like the machines always have a key on the front panel...
[16:31:29] <jepler> http://faculty.etsu.edu/hemphill/entc3710/nc-prog/images/Fig04-01.gif
this calls it EIA, which I see skunkworks_ said half an hour ago
skunkworks_: wow, what a thread
there, now using the EIA encoding I found on the web: http://emergent.unpy.net/files/sandbox/tape.py
jepler: when will you put it in AXIS?
showing it should be optional, of course
just add another display checkbox
eia was just ascii with even parity wasn't it?
skunkworks_: no, have a look at that gif I linked above
they're quite different
for instance, even ignoring parity, 'J' is not 'I'+1
ah - I guess it has been 16 years since I had to play with that.. :)
that is what the k&t interface I made did - converted text to eia - hacked it into the tape drive.
heh. you could make an LCD-based fake tape (if the reader is optical)
I think I had thought about just using led's into the detectors - but I think I just hooked right into the photocells
good night all
[19:30:08] <cradek> http://timeguy.com/cradek-files/emc/wtf-splash.png
hm, I wonder if I had the "invert X" option enabled when I made the splash screen
how do you do that again? it's not a view option
some undocumented inifile thing
it's easy to add toggles in the view menu
oh it's actually something in my ~/.axisrc
cradek: where do you get the at speed light?
it's in trunk's sim/lathe.
sim/lathe has simulated spindle inertia, and I think I used 'near' to say when it's at speed
EMC: 03flo-h 07TRUNK * 10emc2/src/emc/rs274ngc/ (interp_queue.cc interp_queue.hh): includes for strdup and free; removed typedefs, not needed with C++
lol, the emc users list is getting weirder every day
"The machine that says NAK, didn't Monty Python write about this?
NAK is an AKnowledgement response, isn't it?
I think it was NI
NAK is neagative AK
jepler: do you want me to respond on the wiki page or here?
stustev: I'd be happy to discuss it here, then I would update the wiki page
but if you'd prefer to do it another way I'm flexible
I will go to the machine and try things as necessary
the machine will not show a position of 360
it will show 359.999 the next .001 movement will show a position of 0
OK, so the DRO shows from 0 up to but not including 360
g91 is incremental (relative)
I will go check to see what the DRO shows
I tentatively assign a new code for this wrapped axis mode because I don't think we want to change what emc does for angular axes in G91 mode now
would it be more sensible in some cases to allow separate axis control of wrap mode rather than all 3
archivist: cradek wondered the same thing
from the standpoint of coding it I don't like that idea -- it makes the code 8x as complicated
a trunnion user has a +- 90ish and a full rotator
but I have never played with the expensive toys like others, just trying to think of coding gears and the like
myself, I don't have any machines with rotary axes
what I really want is to do a move to 1xxxxxx degrees and move directly to 0 or some start value
the machine moves as follows
g90 g00 b45 - moves the machine to 45 degrees (the tip of the tool would move in a counter clockwise direction when viewed from Y positive) (looking in the Y negative direction)
g91 g00 b-10 - moves the machine 10 degrees opposite - the DRO shows 40 degrees
the DRO shows 35 degrees instead
I can't understand 40 degrees
g91 g00 b10 - moves the machine back to the 45 degree position - the DRO shows 45
when the DRO stays <360 I call that a rotary scale
I follow so far, makes perfect sense
when the DRO shows ex. +720 or -720 - I call that a linear scale
so far this sounds like what emc does now
I thought there was a way of commanding angular movement that was different than emc
that you told me about once
stustev: what happens with the sequence G90 G0 B0; G0 B-0
is there any way to get one full turn?
jepler: I think he is describing that, but hasn't hit on any of the cases that are different yet
360 would probably be one full turn - I will try it - 720 would be two full turns
the difference would be g90 g00 b45 and g90 g00 b-45 are the same position
stustev: so G90 G0 B30; then I want to turn 360+10 degrees to end up at B40 - is it impossible in G90?
(these multiturn or exactly one turn cases are the ones that are baffling to me)
maybe you would command B400 (360+40)?
jepler: sorry for taking over the questioning
you should maybe just forbid anything outside <= -360 and >= 360 in G90 mode
I bet nobody would program that anyway
for multiturn you'd use G91, which is a lot clearer
a bit of good news is that 2/4 of the concave corner situations work in lathe mode
I spent minutes and minutes of my life making test cases
I cant wait to play cnc lathe as well
g90 g00 c330 - moved to c330 - tool tip moving counter clockwise when viewed from z positive
c340 moved to c340 in the same direction
c-330 moved clockwise 10 degrees to c330
c700 - (360 + 10) - moved 10 degrees to c340
you are correct - g91 would be the way to do 360 degree or more motion
700-(360+10) is 330, should c330 have moved to c=340?
that wasn't the question :)
c330 moved to c340
that sounds a bit incorrect to me, as a design goal
10 degrees of motion on an absolute scale
330 should be either 330 or 30 degrees of motion
or it should move to a position of 30 or 330 absolute
depending on the sign of the motion command - from zero c330 is either a 30 degree move or a 330 degree move
the resulting position is the same place - the direction of motion depends on the sign of the command
from zero a c330 command and a c-330 command is the same place
or was that move actually c720-(360+10) ?
the resulting motion is 30 degrees or 330 degrees
the c700 command was from a starting point of c330
from C=330, issue command C(700-(360+10)), which evaluates to C330, if I'm doing the subtraction right
which means you are at C=330, and you issue the command C330
that makes me surprised when the machine moves 10 degrees to C=340
since 340 is either 340 or -20, and 330 is either 330 or -30 when looked at modulo 360
I was at c330 I told the machine to move to c700 (one 360 degree rotation + 10 degrees)
the resulting motion was a 10 degree move
ok, so that wasn't an equation? :)
it moved to (700%360)
using the +/- direction rule I guess
if this is the formula I can understand it
sure, if the command was 700 (=340 in base 360), then I'm with you
if I had told it to go to c-700 it would have stopped in the same place after a rotation of 350 degrees in the other direction
stustev: you actually typed G90 G0 C700, right?
so if I understand right, in G90 mode, you can do no more than one full turn
ok, it was a reading error on my part :)
just less than one full turn
the target position is word%360 and you get there in the specified direction
I saw "c700 - (360 + 10)" and read it as an equation that evaluated to C350 :)
ok, I see that, agree that was not very clear
must run - bbl
looking at the wiki page, it sounds like I got the first 3 bullet points right
in G91 mode, the machine moves 'word' degrees relatively
what about the next thing I said: # The motion is always greater than 0 degrees but never greater than 360 degrees
the DRO is %360 and shows the position
yes, seems like you can not get a zero degree move, you get a full turn if 'word' is the current position (numerical instability problem here)
we have not yet explored the behavior of offsets
* If 0 is the result of a calculation or is taken from a #-parameter (not a literal "0", "+0" or "-0"), then the sign is undefined. The equivalent +360 and -360 can be used instead.
-360 and +360 are equivalent to -0 and +0, right, except that you can reliably get them as the result of operations?
I think that's right
OK, that leaves the two questions on my list:
* What does G10 L2 A- mean?
* Should G91.2 modify the behavior of non-wrapped axes
what is G91.2?
G91.2 is the code I propose to switch into this new mode
I don't like that
we'll retain compatability with old gcode except when this mode is turned on
you don't like the specific number, or you don't like having it be a mode?
wrapped or not wrapped should be inifile
per axis I think
I don't care about the number, and I think we have to keep compatablity with old gcode somehow
compatibility is don't set up your machine this new way
I don't like that. It means that when you install this emc you have to throw out all your old programs if you want to enable this mode
default is definitely not wrapped rotaries
I guess I see your point, but I don't like either option now
how will you do one wrapped and one not, if using g9.2?
that needs an axis as well
that way lies madness
ini config for which axes should be wrapped, and aG-code to select old or new behavior
certain axes can't wrap, regardless of a G-code switch
certain axes on certain machines can't wrap
I don't know if it's possible to have more than one axis that can travel i
maybe if you can run a lathe spindle as an indexer or spindle, and you have a rotary table
if your machine has B wrapped but you code something that fits in my machine's non-wrapped "B", I'd like to be able to run it
sure there is
but you're saying I can't, if you used the wrapped mode
think of a gantry's C (spindle turns around and around) and A on the table
I can think of a hobbing machine set up with two infinites and a non infinite
yes, that's what I just said :)
jepler: I don't think that kind of code is generally going to be portable, no matter what you do
jepler, you have a point - if wrapped code still fits within the envelope of your machine, you should be able to run it
that's a limit problem though
oh, wait :)
I wonder if anything of value will be available in the Circuit City going out of business sales
we don't have a circuit city in lacrosse.
I don't like them, but I'm willing to set that aside if I can get something useful out of their demise :)
EMC: 03bigjohnt 07v2_2_branch * 10emc2/docs/src/gcode/main.lyx: remove g73... must have been a senior moment when I added it
I'd rather have a senior moment than no moment at all
ain't that the truth
I would just like a moment
type of chocolate over here :)
well I have no idea why I'm still at the office, an hour after I loudly proclaimed it was friday...
* jepler leaves