skunkworks: they are all lined up like ducks in a row Nice Job
thanks. I really don't like tiling :)
and I see some bookmatching
wife did that - I am colorblind
don't do any phone lines then
still needs to be cleaded a few more times.
I do network cable by the twist in the pairs.
is this your entrance to your house?
back door - kitchen area
cool, it will look nice when it's done I can tell
getting there. thanks
fucking never ending clicking
sometimes the computer just pisses me off because i cannot use it fast enough
get a bfh then
each time I wake im somehow fealing the groove that I cant handle
are you a bot?
who are you
thats what i say
well this is pointless so I'll just get a glass of wine and do something productive
we fucked on the bed fucked on the floor fucked so long i grew a fucking afro
grow up UNIT41
you cant blame me
igts the system
hardly my doing
im just part of the scene like all yall
don't think you like meall
its the SYSTEM OF SYSTEMS
so complex no one will ever figure it all out
thou shall not fuck the system
it makes no sence from a biblical standpoint but it seems thats the way were living
If you think thats a mess you should try and change things with the powers to be...
ide love to change the world but i dont know what to dooo.... so I leave it up to you
well can't waste any more time on line with useless banter...
wifey gave me 11 seasons of MASH on 36 dvd's.... WOW... ; )
m enter key
i thoght it was pretty COOL too... all the MASH one could ever want...
all I wanted was the Faulty Towers DVD's and a Jewell trigger
got the jewell trigger and it's sweet
we got that one for the father in law..LOL
Unit41 what makes you think I'll change it to your liking?
Faulty or Jewell
"I learn it from a boook"
well I bid you all a good new year including the EMC gods...
[00:22:26] <dmess> http://www.youtube.com/watch?v=ZkWWRTORMmg
from the mine i was in for 4 yrs...
stan's an old friend... what a freak.. lets see how many rules we can break underground ON video...
who is watching?
ALS I can allways change it later after there is more direction
someone has to take the initial plunge
either way someone allready planted the seeds of the future long ago
but he posted it on YOU tube... its a BIG deal.. around the mine..
the future is inevitable it can only be delayed
I only say that because i doubt it can be sped up much more
it can only be sped up as fast as you can create it...
and what are we speeding for
to keep up with demand
aha thats what were doing....
were fixing the world through commercial goods
if people in other country's want fridges and stoves they can come get our old ones
are you chinese
what would that even explain ?
we don't make goods anymore
we make burgers
I make arcball
is it possiible to read the value of a hal param of one component inside another one?
params can't be "connected" to other components
but the hal library is capable of getting at the data, of course
is there any official, non-hack method -- a hack being defined as halscope's SHMPTR method?
not to get at params
the "best" way to do that would be to change the param to a pin
params are by definition not intended for comp to comp communications
or decide that what you're doing is wrong, and use another method
yes, that's the "hack" i've been considering
well, you may need to revisit the reason you're trying to get at one component's params from another - that may be the basis of the hackishness of the solution space
I'm trying to get the current velocity of stepgen -- this is exported by stepgen.N.frequency
some params do have justification for changing to pins, and the setp command has been modified so if there is no param of that name, and there is an unconnected pin of that name, it will set the pin instead
you want it in steps per second, not user units?
both work fine, but I don't have access to either at the moment
ok, the actual velocity, as limited by accel/vel limits
the normal way to get velocity is to run the position through a ddt block
which you could do with stepgen.n.position-feedback (or whatever the actual name is)
stepgen more or less provides a canonical encoder output, doesn't it?
since it has pseudo-feedback
minus the index stuff
ok, aren't encoders supposed to have vel output now?
or was that only discussed?
(then again, you did point out my documentation errors earlier :P )
actually, encoders are supposed to have vel, but stepgen doesn't really provide an "encoder"
ie, lack thereof ;)
its reasonable to expect that stepgen would comply with the encoder definition, but I never thought of it that way
yeah, it's an output and a feedback component
no index, and you also can't implement the reset pin (I don't think so anyway)
hmmm - yeah, I'm not sure how that could be accomplished
though you can do a pseudo-index, byt using counnt mod <something>
man, I think I need new fingers or a new keyboard
there can't be a reasonable index enable, you can't home to it, etc
well, you could, but it would be totally fake
there are somewhat hackish methods of doing that, but I agree that it isn't all that useful
I don't want to do that, I tink its stupid
index is supposed to convey some information from the real world, not something fake
you can have a home switch input to the stepgen, which can either become index or be used to decide which mod-N pseudo-index to use
like I said, hackish :)
anyone have an automatic loob setup on there mill?
daaay-am - now here's a spindle: http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=170128715398
SWPadnos u ever dnc?
I helped a machine shop set up a DNC system about 15 years ago, but not since then
actually, I don't think they drip-fed it, they just downloaded programs from the PC (but DNC may have been involved)
I don't remember :)
im kidna confused on this rs232 cable
[01:10:53] <Gamma-X> http://www.aggsoft.com/cnc-dnc/anilam-crusader-ii-m.htm
SWPadnos if u can check that out and look down on the bottom of the apge
ok, what's the trouble?
what are the differances in the setup.
differant ways of setting it up
do you have a 9-pin or 25-pin port on your PC?
(the top two images are for 9, bottom for 25)
let me see
oh - the bluetooth adapter had a 9-pin connection
are you trying to use the BT dongle(s) still?
i doubt it
ok, good :)
im afraid to mess shit up
so, assuming that you have a PC made in the last 10 years or so, you probably have a 9-pin connector
i dont think my laptop has a serial port hahaha
then you'll need to get one - the USB ones will likely work
and they're very likely to be 9-pin as well
so, the diagrams on the left (the without handshake ones) are easier to make, but require software handshaking
that's a bit slower, but it looks like the software supports it
my home pc has a 9 ping, well im using a 1989 anilam controller so what would be faster handshake on the cable?
the diagrams on the right are for cables with hardware handshaking lines - harder to make, but may be better in the end
hardware handshake is better in most cases, but it does depend on the processor speed
of the controller or computer?
well controller is 386 based and my pc is 3.6 ghz
the PC shouldn't have a problem ;)
i figured so lol
the controller shouldn't either, but you never know
i wonder if i should get teh wireless ones...
in any case, you should be able to get a null-modem cable or adapter, and then an extension cable, and it should work
the pin numbers look normal to me (but I haven't looked at serial port pinouts for a long time, so I could be wrong)
whats a null modem cable i hear that and i think wow 1986
it connects the "output lines" for hardware handshaking to the "input lines" on the other end, and it swaps the receive data / transmit data pins also
SWPadnos would u trust a wifi connection for this?
if you connected e.g. two computers together with "straight-through" cables, then the transmit lines would be connected together, the receive line stogether, etc.
i mean i can only run at 4800 baud lol
I wouldn't use a radio between the G-code and the motion controller
i got a plan
in my view, the only place to use a radio is when a cable is very difficult or impossible
and I used to make radio controls for a living (I still get the checks from the company ;) )
you aren't going to put the PC in the house and the machine in the garage, are you?
i dont want the bluetooth ones though
that's - to put it bluntly - a dumb idea ;)
you can network from the garage to the house, no problem
but I wouldn't put any "thin air" between the PC with the G-code and the motors
use copper, it conducts much better
how much is a usb serial adapter
or pcmcia adapter
should be $30 or less for a USB one (probably $5 if you find a cheeso one)
ill just use my laptop
its got built in wifi so i can set it up on the network and transfew from pc to laptop to controller
but thats more amps bein pulled from garage lol
SWPadnos: I pointed them out
i only have a 30 amp box in my garage
I'm sure a laptop and/or PC won't pop a breaker
laptop and a cnc machine on a 30 amp box? lol
well, the CNC might be an issue
jmkasunich: re velocity fb: how much extra processor usage would adding a DDT block cause?
a few hundred nanoseconds
it's a single subtraction and a multiplication - should be about 4 cycles plus cache fill time
(plus some memory accesses, of course)
there is some moderate HAL overhead, but not much
and HAL execution
I think we saw that HAL overhead was in the tens or low hundreds of cycles
hm, so it might be better for me to hack the param into a pin
you are fscking nuts
it'll still be in steps-per-second
not in units per second
maddash: why are you even using emc?
anyone who cycle counts on a PC in these days is an idiot
so you'll need to multiply/divide that to get a usable number
it may not be a PC
fenn: b/c I don't want to have to program my own rs274 interp and a tp
I thought it was some powerPC embedded thing
or was that someone else?
maddash: why are you using rs274 to run a robot?
err - ARM
jmkasunich: ok ok, I'll do both and benchmark it
maddash, if you get the BASE_PERIOD down low enough (dependent on the PC and processes running), then you'll see a drastic speedup in thread execution time
jmkasunich: and anyone who cycle counts on a PC in the context of EMC could be trying to squeeze out the smallest *_PERIOD posisble
once you get to the point where the fast thread code/data isn't evicted from cache (at least, that's the theory)
yes, you told me about that last year
maddash: you'll see far greater effects from cache issues then from cycle counting
remember - if the other module goes away (which can happen), you'll be looking at nothing (or an access violation, in kernel space)
that's if you try to access params, that is
fenn: walking. running. attack.
fenn: evasive manuever patterns.
yes, rs274 is clearly suited to these tasks
hmmm. I wonder what changing that variable to a pin would do to execution time
since it would then be double-indirect access
right - compared to using a ddt block ...
ie, 6 of 1, sqrt(36) of the other
SWPadnos: any optimisations you might make on a machine might be void or less on the next one
I'm painfully aware of that ;)
wcet= worst case execution time
(which is what you should consider for RT operation)
oh sure - use a chip with the same core speed and 1/2 the cache size, and you're hosed
or not - it's almost random
or the other way around.. and you wait for the cache to fill :)
there's a nice nvidia specific instruction
I really like the way the core 2 duo gets better latencies when one core is bogged down doing nothing
(err .. not nvidia specific.. but I think the nvidia driver uses it)
(ie, while true ; do echo "nothing" > /dev/null ; done)
it's something that takes a couple of hundred thousand cycles to complete
like REP STOSB? :)
flush all caches and write back info.. or somethign like that
with a very very high count
it takes a bit to store into memory 4-8MB of caches
oh yeah - some invalidate cache and write back or something
what I don't understand is why the velocity wasn't exported along with the position in the first place, since there already is a variable for it
it's not velocity
it's step frequency
it's proportional to velocity, but you still need to scale it
which is proportional (and just as good as) vel
the velocity can change on a step-by-step basis, it doesn't change only at the servo rate
which would require floating-point math in the base thread
(it's not necessary for getting to the next frequency, but it would be necessary for converting that to a velocity)
well, I hope you guys are happy, because I've set up the ddt block
look at ddt.n.tmax
if my robot dies in combat b/c of some reaction latency, there'll be hell to pay
luckily, we're separated by a thick location barrier
he's sending the robot to get ya
I'll drive over it with my Jeep
ddt.0.tmax=809, but I'm running sim on for the moment, so is this value reliable?
I don't think so - the kernel can interrupt the userspace apps
but it won't be more than that in kernel space
also, try plotting ddt.0.time - you'll get a good idea of the distribution that way
I think you'll find that vast majority of the time its quite fast, and once in a while its slow
especially on sim
the slow ones are when something else displaces it from cache, or it gets interrupted, or any of many many other things that totally swamp out a few cycles
SWPadnos im realy tempted to do the wireless dnc...
it may even work
[01:55:38] <maddash> http://img252.imageshack.us/my.php?image=200801012053351400x1050wf9.png
wtf? ddt is supposed to be smooth
who want make money
i make robot take money
maddash: you are measuring feedback velocity, which is quantized - its a stepper after all
it moves some integer number of steps each servo period (sometimes zero)
it looks like the ddt is in the base thread
that's a 25 KHz sample rate
so the output will always be 0, 1 step, or -1 step
yep. I'm running all of this off a hal test setup
interesting - notice how the ddt execution time trends downward as the step frequency goes up
or it has more samples at a lower time anyway - maybe not a trend
quick comparison of usefulness of ddt vs. stepgen.0.frequency: http://img521.imageshack.us/img521/9562/200801012102381400x1050us3.png
my hal file (for those interested): http://pastebin.org/13789
one thing - you did set fp=1 for the base thread, right?
fp0 or whatever it's actually called
hey - what's that desktop theme called? those look like NextStep/GnuStep icons
oooooh, I didn't notice fp1=* in the manpage
thanks for that
does it matter in sim?
dunno - probably not
I'm using iceWM and rox-filer (file manager) with the theme, IceBuntu
though it might. I don't know how the context is changed between "threads" and "other stuff" in sim
the source icons and the gear thingy look very similar to some nextstep icons
I wish there were next drivers for modern hardware - I still have NeXTStep 3.1-ish
fp=1 by default, i think
that would be the safest
got some good news
found out my spindle on off is controlled by the controlller aswell as reverse
80-20 is now architecture http://www.tomahouse.com
yes, it's just alum extrusion underneath (flash heavy, user beware) ( and i was looking at recycled shipping container house when i stumbled into it )
i know someone who works with tomahouse
er, worked. now he's with jeriko house
I just hate sites that pop up maximized windows
it's very annoying when "maximized" means "spanning 3 1280x1024 monitors"
send them some hate mail
oh, even better, flash decided that some stuff should be right-justified on that 3840x1280 canvas
i just read a nice forum that had rogern talkin about a retrofit
tomahouse kinda reminds me of disneyland
how do i know if i have servos or steppers?
it doesn't look like it's appropriate for colder climates
jeriko house is 80-20 too
anone know the differance between servo and stepper?
sure, but it's hard to describe how *you* can tell the difference :)
if the motor isn't attached to a machine, you can turn the shaft. if it turns smoothly, it's a servo, coggy = stepper
there attached lol
if the wires are visible, then a servo will have only two wires (plus any for a tach and/or encoder), but a stepper will have 4, 5, 6, or 8
if there's an encoder, it's likely a servo
or a tach, for that matter
if you look at the drivers, they may be labelled for the type of motor they drive
(DC brush servo or stepper, or something else)
ill take off the back panel tomorow
you could also try looking at the motor nameplate, if one is visible
it doesnt say
im goin to bed i will talk to u all tomorow
what ratings does it have?
what do u mean
2400 rpm max.
if it lists things like "holding torque", "steps/rev", "max frequency", then it's a stepper
i know it says sem mt30 i think
if it has stuff like "Volts/1000RPM, and torque/amp" then it's a DC servo
I think that's a servo
but you can google it
i will lata
google's 3d warehouse is fun, sort of a vrml warehouse of objects
what is "lolcats" and why is every commenter on this page http://icanhascheezburger.com/2008/01/01/funny-pictures-no-fite-just-rubs/
praise to the chaos god
home for two weeks, no snow.... have to go to work in the morning, and we have close to a foot and still falling
hi quick question, where might I find the backlash compensation settings in emc?
in the ini file
[AXIS]BACKLASH = whatever
AXIS_n that is, where n = 0 to whatever axis
[03:55:47] <jmkasunich> http://jmkasunich.com/pics/star-test-1821.jpg
is that just in inches? so 0.004 would be 4 thou?
its in machine units, like the rest of the ini file
jmkasunich: whoa. how you make those? edm?
if you used inches, its in inches
maddash: milled it
for electrical diagram, what should I use to draw it up?
jmkasunich: but the bottom of your grooves are almost perfectly sharp
I ground a pointy tool
started with one of these: http://jmkasunich.com/pics/misc-tooling-2.jpg
jmkasunich: On the spiral, I see like \\\\\\\\\\\\\\\\ marks new the top-left, what are those caused by?
and ground a fairly shallow V on it
Jymmm probably just tool marks
I'm only running 1450 RPM or so
and that was a rather deep cut for that tool
jmkasunich: Ok, so what would you need to eliminant them in the first place?
sharper tools, lighter cuts, I dunno
higher speeds would probably help
one foot in Cleveland?
jmkasunich: oh, so not soething you have a LOT of exp in diag just by looking?
argh, gotta goto mentor
tomp: from illinois? thats a haul
jmkasunich: k. Les could look at that and tell you what your setttings were. I just thought everyone else had that too.
yep, just had 3600 mileson my 91 honda wagon
wrong time of year for that
well, happy new year to all, it's bye for me
What are some proggys for drawing up electrical diagrams?
SWPadnos: Does it take a few months to learn, cause you dont have that long.
I have it, and I've learned it reasonably well :)
SWPadnos: Is there something more TONKA based?
but I'd only be able to give you diagrams as PDF files, since I don't think you'll be spending $8k to $12k on your own copy
you can use corel draw if you want - you won't be making netlists
SWPadnos: I meant something with objects; fuses, switches, lamps, etc
yeah - I don't know. Eagle may have those symbols, but I'm not sure
I never have found a good "cabinet wiring" diagrammer
LawrenceG used something, but never asked what it was.
most people I've asked just use whatever CAD program they're used to - AutoCAD, Eagle, OrCad, etc.
Jymmmmm: check out kicad, it's much easier to pick up than eagle
the symbol library kinda sucks though
there were a few ladder diagram makers that had good syumbol libraries for this stuff, but they were all time-limited demos (until you pay $a few hundred for them)
fenn: does it have electrocal symbols?
boy does it
even a Win32 varianty too....
fenn: 3D too?
for some things
[04:18:50] <Jymmmmm> http://kicad.sourceforge.net/wiki/index.php/About_KiCad
its opengl too, not pov-ray
Just saw the 3D photo on that page
dont think you can export a model or anything
no need to, just slightly impressed with the featureset
i think its supposed to be like orcad
that's not a good goal
that might explain a lot of UI suckage
but maybe its just because they're french :)
fenn: is there an english manual?
those are pretty nice 3D previews
you can get to it from the help menu
[04:22:43] <fenn> http://kicad.sourceforge.net/wiki/index.php/Mini_tutorial
fenn: TY, I'll check it out.
Jymmm: been a long time.... last year I think
lerman: you here?
* Skullworks-PGAB is home again :/
fenn: Did you say that kicad had electrical components? breakers, relays, contactors, lamps etc?
lost steps suck: http://jmkasunich.com/pics/mahan-1.jpg
(also, unflat stock sucks)
jmkasunich: Sounds like you're beginign to appreciate my delima
this is quite nice... http://mp4cam2avi.sourceforge.net/index.htm#features
does net replace newsig and linksp
programming a g code generator for SHCS's counterbores
[12:08:00] <BigJohnT> http://i47.photobucket.com/albums/f163/johnplctech/Python/counterbore.jpg
yeah, net is equivalent with linksp and newsig
but newsig and linksp won't go away
they have to stay for backwards compatability?
and because there's nothing wrong with them :)
andd some people find them easier to understand
or is there situations where they still can be used
they are equivalent.. so both can be used to do the exact same things
linksp linkps are close to what is actually going on in the code
just what ever you get used too
what is linkps?
same as linksp
link pin to signal
but in reverse
instead of link signal to pin with linksp?
so I'm guessing that linkpp is link pin to pin
cradek_ is now known as cradek
yes but it caused some confusion because it creates a new signal with the same name as the first pin
good morning! happy new year! oh I was doing some year-end cleanup and I deleted all my data... [one of my users]
morning guys. happy all that stuff also.
tank yew, tank yew
What is that a terrorist's imitation of Gen Patton?
more like a lumberjack or bowmaker I guess
hmmm - what is a bowmaker called?
I give up.
I don't know
We thinking violin bow or bow-and-arrow.
longbow (and arrow)
but I suppose one could be wrapping presents or making instrument strummers as well
I wonder what culture invented the longbow.
probably the Chinese, in about 2500 BC
first recorded use was by the Welsh in 633AD
We see here the cultural definition of history. Which history do we read.
ah - at least two Neolithic longbows were discovered in England, radiocarbon dated to between 34040 and 36640 BC
err - 4040 and 3640 BC
that would have been a very wide error bar ;)
yes it would.
But when you get that far back what's a little error among friends.
yeah - not like anybody can prove them wrong (at the moment)
interesting page: http://www.edinformatics.com/inventions_inventors/longbow.htm
Bower is the only name I found associated with the craft of bow making.
Never thought of pronouncing it that way.
yeah - seems more related to "bowery"]
"A bowyer is a person who makes bows for archery. The term bowyer typically implies a skilled craftsman, though with simpler designs even a beginner can produce a functional bow."
ah - thank you
(someone who makes arrows is a fletcher, not to be confused with a fletcher)
er, not to be confused with a felcher
right - I actually knew the term fletcher :)
* jepler ruins his own joke
now im confused
I'm with tomp!
don't miss: http://www.anderswallin.net/2008/01/mowing-foam/
hi guys, I'm confused as to the meaning of emcmotconfig->servocycletime. the variable name sounds like the amount of time to be spent inside emcmotcontroller (before permitting an exit), but from the code in setservocycletime() in motion.c, it looks like the delay in seconds before each function in servo-thread is called (e.g, servo-thread period). which is it?
I haven't looked at it, but I suspect it's the thread period
not a delay between functions
the second one should read, "delay inn seconds before the execution of servo-thread begins again"
cycle time means "the time between cycles", not "the time spent per cycle"
time between invocations
so servocycletime is predicts the next time emcmotcontroller is called
that was clear
in the RT code, it should set the RTAI thread period
the calculations that setServoCycleTime does should be based off the 'period' argument when the servo function is actually called from some HAL thread. If you use 'threads' to create a thread with a different rate and put the servo functions on it, the results will likely be incorrect. (no sample configuration does this, of course)
yes, I get that
but all this assumes that the function call will take less than servocycletime
so what if the execution time of the function > servocycletime?
yes, that assumption is made for all threads
then you're fucked
your PC locks up
SWPadnos: was that sarcasm?
you never get back to non-realtime code because there is always a call to a real-time function pending. That makes it look like the PC has "locked up".
no, I'm serious
also, the PC could actually lock, depending on how stacks are managed
jepler: so the calls build up to infinity
have you watched that episode of I Love Lucy, where she is working at the candy manufacturing plant?
or whatever it was
they arrive at a fixed rate
haha I can see where this is going
if you can't finish with each item within that fixed time, hilarity ensues
it was chocolates
hmmm. there was one where she was adding the whipped cream on top of pies
I myself like the vitavitavegamin(sp) one.. but I digress.
so if a greedy bastard like me wants to squeeze out the last bit of performance from his machine and lowers servo-cycle too much, then he'd get what wwas coming for him?
yes - sync first
[17:35:33] <skunkworks> http://youtube.com/watch?v=MTsfVbN7UOM
hey - I spelled it right :)
wow, I never knew that black and white could be so funny
all this will be moot when the mesa TP is completed, right?
I don't know what you're referring to
are you working on an FPGA TP?
I know Colin Mackenzie mentioned the idea, but I'm not sure where he is with it
mesa? i520 something? housing the essentials of emcmot?
none of the core EMC2 developers are currently working on that, that I know of
hmm, i was thinking about it, but then I spent my savings on a dualcore asus instead
probably a more sound investment
the mesa folks have something called "SoftDMC" (which essentially places the entire motion controller within the FPGA) but that's an alternative to emc, not something to be used in conjunction with it. (I don't have any specific knowledge of SoftDMC, though)
I don't know how much you'd be able to get into the 5i20, and the 5i22 is $500, not $200
so what was the big fuss about the 5i20 6 months ago for?
more configurable I/O and other functions
not embedding motion.c
can't get past the TOC w/o crashing xpdf
works fine on evince
and Acrobat reader :)
segfaults are becoming the new, "this program has performed an illegal operation" for me
what do you believe will be improved by lowering your servo cycle until just before your machine locks up?
oh interesting, the "reason" that 525 lines was selected for NTSC -- only odd numbers that were the product of small numbers were candidates, and apparently 525 was the best candidate near 500. http://en.wikipedia.org/wiki/NTSC#Lines_and_refresh_rate
err - 5*5*7*3
I gather that 17 is "too big" a multiplier
or divisor as the case may be
plus it was a mistake ;)
that would have been 425 lines
well it would be a surprise if 5*5*7*3 was equal to 5*5*17..
indeed it would
It's so tasty too!
hm, check my work here. if you do a circle in N servo cycles, you do 2pi/N radians per servo cycle. The maximum distance from the arc of 2pi/N radians and the chord connecting the ends is 1-cos(pi/N). So if you do a 1/16" circle in 1/5 of a second (just about F240) you have 200 servo cycles per circle. (1-cos(pi/200))*(1/8.) is about 1/64000 inch. Acceleration in this circle is about 123 in/s^2.
er, whoops -- I was calculating for a 1/8" circle, not 1/16"
anyway I think the max distance from the circle to the chord is small, even at the "slow" 1ms servo cycle, but perhaps I went wrong on my math
it should be even closer, since the motors accelerate in a square-law fashion
on the real system I bet it ends up pretty round, but undersized
so the cusps are curved, which brings the actual path closer to the ideal circle
(which isn't ideal if you're actually trying to make a 200-gon :) )
normally polygonization of circles inscribes the polygon in the circle. But you can choose different n-gons which minimize the chord-arc distance or the distance between them..
I worked out the math once but lost it
now, seriously .. lunch
Anybody using IMC MDrive steppers?
[19:01:09] <lerneaen_hydra> http://cgi.fark.com/cgi/fark/youtube.pl?IDLink=3299360
[19:18:27] <Jymmm> https://bugs.launchpad.net/ubuntu/+source/mesa/+bug/109784
hmmm. it's getting cold here
then you need to be fired
I don't think I'll do that, thanks
yes, I got it
my hypothalamus is quirky -- I don't shiver (or feel cold) as long as my hands are kept warm
yeah - I tend to be OK as long as some part of my body is warm - I can almost stand in the freezing cold with a hand in hot water, and have it be OK
if I submit fixes to emc, is it better if everything is rolled into one patch, or if I submit individual, issue-specific patches that need to be applied in a certain order?
individual patches are better, but it's best if the system is sane with each commit
I mean individual by function, so if you need to change some interface to allow a new function, you should submit one patch (including all changed files) for the interface change, and another to add the new feature
s/function/purpose/ (maybe ...)
that assmes that the hypothetical interface change does not prevent EMC from working, without the new function ..
(clear as a blizzard?)
yes, but unfortunately, moot as well
I totally forgot to save a different doc after each set of fixes
what is the nature of the fixes in your new file?
a shitload of tings
incorrectly spelled var names,
some "TODO: Fixes,"
and a modification to limit switch rules during free mode
bummer. it's best to keep cosmetic things separate from functional changes
bummer. it's best to keep cosmetic things separate from functional changes
(in case you missed that)
I'm rebuilding my changes, anyway
maybe it won't take too long
are you one of the people running on an ARM board?
or some other embedded thingie
that would be nice, if I could get it to compile onto an embedded arch
but sadly, no
ok - must have been remembering someone else
doesn't sound very realtime like http://www.cnczone.com/forums/showthread.php?t=49134
quote LOL if I try to open internet explorer while mach is running a code it slows way way down to like one step at a time but still keeps going and don't miss steps just goes real slow and in chunks
and never try to run WinAmp at the same time
I really should play around with it.. I can be open minded..'
-_- and _why_ would you want to run mp3's on your cnc machine?
on a laptop...
I tried Mach 2 or 3 a while ago - not running a machine of course
all time functions got screwed up, systemwide. things like double-click speed, pop-up delays, etc.
I like to surf while cutting parts...
and talk to you goof balls.
I never bothered to look at the output with a scope. I probably should some day
skunkworks: you have balls..
- the machine was running fine until my email client autochecked..
skunkworks: if I felt that I could leave the machine that much without watching I could aswell go away from the shop :)
and sit inside..
I really don't think I have had emc make a bad part that wasn't my fault.
skunkworks: in 2 seconds you can machine a hell alot of clamps. :p
skunkworks: in 2 seconds you can machine a hell alot of clamps. :p
well - if I was machining a new part.. I wouldn't leave.
that's how I feel..
when a circuit board takes about an hour a side.. I have walked away.
as long as I have seen it run correctly once.
I were thinking about aluminium/iron/steel parts..
that too - as long as it has been proofed out.
that reminds me - I need to setup some video cameras ;)
jepler: I noticed that that joystick axles have "dead" area near 127. 0-111 is working linearly. Then value changes to 127. After enough movement it changes to 143 ja continues linearly to 255. Do you know were this is defined?
Martzis: if this is in hal_input, I think you can get rid of this effect by using 'setp' to turn down the 'flat' of an absolute axis to 0. You'll also notice 'fuzz' in the parameter listing -- that controls how many counts the axis must change before the HAL pin is changed to a new value.
Martzis: for hal_joystick, I think this involves modify the source to use JSIOCSCORR (set correction) ioctls to change some values used inside the kernel.
you might also want to read the description I wrote of 'fuzz' and related parameters in the manpage -- since it was all fresh in my mind at that time. http://linuxcnc.org/docs/2.2/html/man/man1/hal_input.1.html
or 'man hal_input' in the terminal
jepler: Thanks again. I'll test that.
(for hal_joystick you can also write a separate program that uses JSIOCSCORR and run it at startup or when starting emc; I think those corrections persist as long as the device remains connected to the PC, or until you reboot)
I can't seem to find where in the kernel the default 'fuzz' value is set for USB devices.
Ok. I do not know if I can use hal_joystick. It worked until I changed encoder inputs as relative axes. It seems that joystick with relative axes is not detectid as a joystick.
In windows it is working
hmm and now I see for hal_input that the 'fuzz' processing takes place in the kernel so just setting the HAL parameter for 'fuzz' won't have the expected effect.
can nc code execution fork?
maddash_ is now known as maddash
fork as in branch, or fork as in "spawn a new process"?
fork as in multithread
eg, 'g1z*' whilst 'g2x8'
ok, that's a loop, which is possible
not even if the two forks are in the same group?
err - wait a minute
G-code isn't executed like a program, it's interpreted by a single-threaded interpreter
you can only have one motion happening at a time
there are instructions (O-words) available to choose between differnet instructions (if, while, etc.)
but only one motion will happen at a time
damn. ah well, it was worth a try.
can I have 2 controllers on one machine?
why would you want that
for many reasons - but what do you mean controllers
only sensible if they did different jobs
right but it's a safe bet that's not what is meant here
so you can play two-player games, of course
Gamma-X, clarify please
* cradek 's eyes cross
i guess we will never know
* archivist checks his telepathy module, rattles it to see if its working, hmmm
archivist: can you see this now?.........now?............now?
damn battery is flat!!
haha, 'two-player games'
* skunkworks thinks maddash should go back to playing WOW ;)
i think what he meant was can he keep his existing control on his machine, and also have emc able to run it, I do not know if he is also thinking about simultanious operation.....
I play wow, cant help it if im a good warlock
I play irc
wow is like irc with real time monster devistation
i want 2 controllers so i can keep the anilam in workin order while i install an emc setup
* skunkworks saw gezar's wow screenshot..
gezar wins a cookie
that was my herbage on my mule
gezar: that was amazing
the guild im in is only like 596th in the us as far as progression rank
Gamma-X, yes, you can do that, if you use the same interface between the control and the machine (ie, analog servo control + encoders) - but you'll need to disconnect one of them
cradek : my ability to reach inside of gamma's brains and find the answer?
you could use something like the MIL circular connectors
but it's likely to be a lot more trouble than it's worth - you have to do the retrofit for EMC2, plus make sure you can still use the Anilam
and conjure up some means of switching between them
omg, MIL connectors are expensive
I disagree with SWPadnos. you have estop, limits, and other safety considerations
you don't want to rig this up in any temporary way
sure, those also have to be double-connected, and switched out along with the motors
I just didn't list all the connections
if you can't commit to a full retrofit and the associated downtime (and I definitely understand) I suggest not doing anything to the machine but use it
skunkworks: what makes you think I play WoW?
(I agree it would be a lot more work and expense)
gamma-x : im with cradek on this one, and on top of that, you need to learn how to use your machine, learn how to use it on your existing system
maddash_: just a guess...
i am currently learnin how to use it
yeah, if it works, you can learn to cut some things first
its simple to be honest
dont hate the player, hate the game
but i want more features than what my 1984 controller does
yes, MIL-C-xxx connectors are very expensive - I spent more on connectors and backshells than I did on my motors
SWPadnos: did you use the crimp style or solider?
the crimper I think is 70 bucks by its self
but the crimp tool would have been several hundred (for the AMP one anyway)
solder is much better, the crimped water sealed stuff, if you miss pin one, its pure madness to remove one, yeah I know the little white, blue, yellow removal tools, but its hell, unless you use the right size wire shield, then its not so bad
and yeah, im thinking amphenol mil connectors zomg
my connectors are type 24-7 - 16 pins each. Getting at the center pins with a soldering iron is a pain, and repairs are nearly impossible (once the outer ring of conductors is soldered in place)
yep, MIL-C-5015 or something
MY PATCHES ARE GONE!!!!!!!!!!!!
I think that was "oops, I killed my X server"
DIE DIE DIE DIE!
hmmm. this seems to be a reoccurring problem :)
yeah haven't you done this a few times before?
those were "sudo rm -rf / somedir" type problems before
omfg my robot code
it's all gone
* SWPadnos wonders how one can accidentally hit CTRL-ALT-BACKSPACE
I've done it
rewrap-region in emacs is ctrl-alt-\
it's fucking 50F and I wore MITTENS while coding
ah - ok
I know that feeling
but, emacs autosaves, like any sane editor does
yeah - I'd imagine that compiling would also require the data to be saved
I can't use emacs!
vi is wonderful
or should I say "pterodactyls roc!"
I've been coding for 8 hours straight and intentionally did not test any of my code, so that at the end, when I compile at the end, everything'd click together and I'd get a massive high
I guess then there's a risk of a massive low then ;)
looks like u got a massive wah wah wahhhhh
so there must be saves in between
my editor is geany
never heard of it
the hot girl here (http://www.emmaalvarez.com/2007/12/top-best-50-ubuntu-opensource.html)
showed it to me
oh - it's up to version 0.12 - how stable
for future reference: add Option “DontZap” “yes” to Section “ServerFlags”
or save your work to disk
oh look, terminal emulation built into an editor
wonder if it has a movie player too
or maybe kill X while leaving the processes in :0 alive? what a dumb design
using aalib, of course ;)
no, just suspend the processes (like ^Z)
may i recomend vim as a fun and friendly code editor
now i feel bad, afk cutting self
you should get help
when I'm done with this bot, I'm going to replace the non-lethal airgun with a 50MW laser and go after the guys in charge of X
brb, I have to go cry
whats all ur opinions, is emc better than a industry standard controller like my crusader II? or is there flaws that may make it not as good in a few ways?
my bet is that you can find ideosyncracies in either EMC2 or a commercial control
especially one from 1989
Gamma-X: if you feel like your control works, use it.. if you feel it's too crappy replace it..
it's likely that EMC2 on a newer computer will have better performance, but that may not matter since the machine itself is probably the limiting factor in either case (though it's probably close in the case of the 1989 '386)
but that's just what I think..
yeah, if it ain't totally broke, don't take it apart :)
yep, my point also
rather if it's good enough to do what you want dont bother.
emc is really nice :)
thanks for rubbin it in! lol
one of the best reasons to replace a control with emc is that with emc you have control over how it works.... dont like something... change the program
LawrenceG: that's a nice hobby
LawrenceG: making parts is another one :)
yes... there are some problems with giving out the keys to the car