libf2c2 - Shared libraries for use with FORTRAN applications
libf2c2-dev - Development libraries for use with f2c
apparently they have a surprising '2' on the package name
still couldnt find the package
do you have universe enabled?
(not sure if that's needed, but it may help)
how do i do that???
yes it is in "universe"
in Synaptic, go to repositories, click on universe on the two main Ubuntu repos
I don't know exact steps beyond that
looks like this article says how to do it from the terminal: http://paulstamatiou.com/2006/05/13/ubuntu-quickie-enable-dapper-repositories/
there's also a way to do it in the synaptic (package manager) GUI
dmess: are you trying to run apt360?
univers wasnt selected..
dmess: try this if you haven't got it yet: https://help.ubuntu.com/community/Repositories/Ubuntu#head-5bbef89639d9a7d93fe38f6356dc17847d373096
steves_logging is now known as steve_stallings
apt does this funny close-match search when you add a * at the end
apt the package manager i mean
er, but that wouldnt help in synaptic, nevermind
sweet... ; )
i got thru the configure i think..
now you need VOLTRON
"... And I'll form the head"
your powers combined..
dmess we are secretly plotting to invade canada, be afraid
nothing scares us... ; )
[00:34:21] <jepler> http://pbfcomics.com/?cid=PBF186-Guntron_Alliance_Force.png#172
[00:53:09] <fenn> http://pbfcomics.com/?cid=PBF192-COP$.jpg#178
I'm looking to do a cnc conversion on a Taig mini mill and using EMC. Is there a specific controler board I should look for?
There is so much stuff available.
what's your main goal? reliability, performance, economy, flexibility...
where reliability in this case means "user friendly"
anyway if you dont mind the expense, and they're on sale now, g203V and steppers works well for many people
steve stallings: I need to talk to you
[00:58:34] <jepler> http://pbfcomics.com/?cid=PBF077AD-Disassemble.jpg#144
who makes the g203v?
That's not a bad price.
then I just need stepper moters that I found for $50 here:
A PSU and a case.
[01:01:30] <sonicmook> http://www.xylotex.com/StepperMotor.htm
btw you need 3 of the gecko's for a 3 axis machine
just to make sure you know -- gecko drives are for 1 motor
yeah what fenn said
also, more mototrs here: http://web2.automationdirect.com/adc/Shopping/Catalog/Motion_Control/Stepper_Systems/Motors_-z-_Cables
276 oz-in, $39
From what I have been reading the Gecko stuff is the best..
swpadnos:is the pwm controller from pico closed loop
works good with EMC?
so that 1 is closed loop
but the usc isnt
the USC isn't, but the UPC and PPMC are
well, the USC is, but the gecko also has a position loop, so it's really dual loops
what does the usc do,just give ferror for position
the USC makes EMC think it has a velocity mode servo
the USC does probide encoder feedback, and it has a step rate generator, so EMC gives it a velocity command and it converts that to a step frequency
it won't work well with steppers, because when steppers start to lag, you can't get more torque from them by increasing the command output
you can with servos
so the usc +g320+us digital would be closed loop
and PID controllable
it's not an ideal setup though, because you need to tune both the geckos and EMC
and gecko tuning affects EMC tuning
would pwm be better
just get a real servo controller
FWIW, that's the setup I originally was going to use, and I have a friend who uses something not even as good and it works well
actually i'm surprised gecko doesn't have analog in
the PWM drives are equivalent to analog drives, and should provide superior finish
thats what i want
well, it does if you lift pin 2 on one of the chips (I forgot which one)
SWPadnos: without having to get out your microscope and hot air pencil
oh, well then no
I'm also surprised mariss doesn't have an analog-in drive, but I think the world of things that can produce step+dir is a lot bigger than analog
mach, any microcontroller, plcs, and so on
I don't think that's true except in the hobby market
wish mariss would put an error output on the stepper drives
uh.. analog would be way better
you'd have to get the position out of the drive though
i mean if the drive faults or dies the software would stop,not talking about position
the gecko servo drive does have a "exceeded max following error" output
or am I talking out my ass again?
yes,but not the steppers
I can't read apparently
actually, it's got an "I've shut down because something happened" output
jlmjvm: Yes, jepleris talking out of his ass, or yes, it does have that feature?
which doubles as the "I'd like you to shut down now' input
can emc use the fault signal from a g320
look at the pico systems gecko servo interface
for the usc
does it hook the fault signal to axis.0.amp-fault-in?
or the add on board
SWPadnos: how do you know who shut down first then?
that's what the GSI is for
it takes up to 4 gecko fault inputs, latches whichever one faulted, shuts down the rest, and has an output to tell EMC that a fault has occurred
there's an input to request the fault(s) be reset
[01:16:58] <jepler> http://pico-systems.com/gecko.html
I was unaware of that card
I think I bought the first one :)
it also powers the encoders, so you don't lose position if you manually crank the machine while it's shut down
ok - if all axis can go 400ipm - how come a combined move is only going 469ipm
Also IIRC the Pico Gecko board lets EMC control following error so your not depending on the internal 128 count following error check.
normally a fault would remove power from the geckos (if you have a properly designed estop system), so you'd lose encoder power
the internal 128-count comparator always runs
but that's between the gecko and the motor, not between the 1 ms servo thread and the encoder
skunkworks: did you set [traj]? (just checking)
and how did you measure that btw
also doesn't the pico board provide a +5VDC source for the encoders since the Geckos max out around 50ma
say the pwm setup would be a superior finish
or whatever it's called
axis has a velocity readout
Skullworks-PGAB, yes it does
so - in estop with the Pico - position isn't lost
skunkworks, try scoping or halmetering instead - axis may have sample rate issues (unless it's a long straight move)
Skullworks-PGAB, I think you need the extra card for that, or at least an external power supply
lemme look at it
there aren't dedicated encoder power outputs on the USC
there's just a 12-pin header with A/B/Z for 4 axes
[01:26:04] <Skullworks-PGAB> http://jelinux.pico-systems.com/gecko.html
yes, that's the GSI I mentioned
he used to call it the Gecko Servo Interface, hence the S in the middle ;)
"The encoder is powered from the USC board, so encoder power is constant, whether the Gecko drives are on or off." - its a little vague about E-stop - other that the Geckos get reset
usually an estop would cause a relay to kick out the gecko power supply
so if the geckos are supplying the encoders, they would also get turned off
wow - youtube is taking a long time to proccess a video
skunkworks: show off
"When in the E-stop condition, a braking resistor is applied to all the Gecko drives."
that does a load dump
I thiknthe resistor is undersized (and the traces leading to it), but for small to medium systems, it's probably OK
but if encoder power is from a seperate source...
ba who cares...
then the encoders remain powered, and EMC won't loe position ...
[01:31:57] <skunkworks> http://www.electronicsam.com/video/PICT1179.AVI
X looks a lot faster than Y
assuming Y is on the bridge
x is left to right - y is front to back
ok so the other way around
y is only 11 inches - x is 18
could that be?
no, it's reaching max velocity
i bet your accels could be a lot higher
it is 30 right now
are the scales correct?
30in/sec/sec would take like 12 sec to get to 400ipm
yes - both x and y is 4000 and z is 2000
oh damn, math argh
why cant people just say in/sec
* Skullworks-PGAB was thinking Jon E. could make alot of amps if he made a version that was turnkey for the pluto.
Skullworks-PGAB: i dont think he'd go for it
* maddash reaches for jepler's vicodin
his pwm is basically the same hardware as the pluto
Skullworks-PGAB: and cannibalize sales of his own servo controller board? fat chance
and has an arguably better interface (? dont really know enough about pluto-servo)
the USC is actually designed to go into industrial settings - it has transient voltage suppression, optoisolator sockets, etc. the pluto is meant for hobbyist experimentation
you choose which one to put in your machine - they have different markets
I'm aware of the pros and cons - and I would use his full sytem on my bigger mill - but the little SEIG - its a hobby class machine - not really something for serious work
jepler: how does interp_convert.cc know when to use gcodemodule.cc and when not to?
maddash: because of the way the various binaries and shared libraries are linked
gcodemodule.cc is not linked into milltask
no, it isn't. emccanon is.
right, so in milltask it is impossible that a call from interp_convert go into gcodemodule.cc, since gcodemodule.cc is not linked in
and yet, gcodemodule.cc is used -- how else can the live plot do its thing?
gcode.so contains gcodemodule.cc and is linked to librs274. gcode.so is imported by axis.
the GUI (e.g., axis) is a separate process from milltask
yeah. i just discovered the "import gcode" line after grepping the results from `grep -rn gcode emc/*`
grepping grep -- how pathetic is that?
anyway, a mystery has finally been solved
grep is pretty awesome if you ask me
what's -rn do?
-r = recurse subdir, -n = show me the line numbahs!
i think its past my bedtime
OR IS IT
* fenn looks behind him
you know what's even cooler?
sharks with lasers?
lasers with sharks?
`alias grep=grep --color $@`
yeah you can set the color too
i once set the color.
that way you dont mess up piped commands
sharks are overrated. http://www.joe-ks.com/archives_aug2007/PiranhaOnSteroids.htm
now imagine THAT little fellow with a laser -- not that he'll need one.
that's a big piranha but humbolt squid would eat him for breakfast
yeah, and then the piranha would eat the squid from the inside out
but.. razor sharp clawed tentacles!
or blast an escape hatch through the squid's digestive trac with that trusty laser
sharks.. with gps locators!
maddash: Not really, you're more than welcome to hang if you like.
actually, I was taking advantage of your typo
maddash: I know =)
grarrr i hate g-code
youtube video up http://www.youtube.com/watch?v=KxzTPXBbxXY
that is getting 28000 steps per second out of a base period of 30000 :)
skunkworks: just sell the damn thing already
Jymmm: getting there. :)
skunkworks: Yeah, uh huh... I'm getting my leggings on as it's getting a lil too deep in here =)
still would like to find the official software also...
skunkworks: You know you're gonna keep it, just pay for the garage some other time already
but so far I have been coming up with nothing.
jymmm: how is your machine running? ;)
skunkworks: Really, don't sweat the original software... but if you can get from CorelDraw to engraving (somehow), then you'll be good to go.
skunkworks: on hold, needed to take a break from it
well - if I can't get the software - I will really push the emc2/axis route on ebay.
I think if I was smart I would buy 3 geckos while they are on sale and sell it with them.
skunkworks: are you really that desperate to pay off the garage? You got a GREAT deal on that machine.
I really don't think I will use it.
skunkworks: Well, fair enough then.
btw - concrete should be down by next week :)
oh well - have fun.. I need to go to bed.
jmkasunich: Didn't you mention one time about some 5 conductor oil tight inline connectors?
I just need 12v@500mA max
yay i made a spiral: http://pastebin.ca/673063
that should be r+=dr
i guess i need start and end radius and calculate dr from that and number of turns
i couldn't figure out what you two were talking about for a minute or two
[03:28:51] <fenn> http://fenn.dyndns.org/pub/irc/spiral.png
that's some impressive nonsense
and the new debugged parameterized source: http://pastebin.ca/673078
i'm quite pleased how quick that went together
fenn: you should use arcs
(it's a very easy change if you use R format arcs)
cradek i'll make two versions
then i can generate my own spiral.ngc and arcspiral.ngc files :P
a version that does constant(ish) arc length would be nice
i always get confused by g-code's arc format
since arc length is r*theta, I think you could vary dtheta inversely by r
R format is just X,Y endpoint and R radius
theta += dt/r
yeah (be careful at the beginning of course)
theta += min(dt/r, pi/2) or so maybe?
heh, sorry to come in when you're done and tell you how to rewrite it :-)
it's a bad habit
no it's nice
ok now i have constant segment length
cool, I never quite wrote that, even though the math is easy
is the radius of each arc segment just hypot(start) + hypot(end) / 2?
the arc isn't centered on 0
I can't quite see where the center is, that's why I think you should use R format
i think its some sqrt(2) in there somewhere
are you sure it's not what I said? (I haven't tried it)
no, i'm still wondering why messing with dt changes the size of the spiral
I should go to bed, even though this is interesting. Goodnight...
the solution of course is obvious once you see it
you only need to calculate the number ofsteps, then divide (endR-startR) by that
well, that's not hwo i did it
i have a parameter L the length along the spiral
and i add dl to that each time through the loop
that's the other way to do it :)
then i calc theta from l, and r from theta*slope
right, L/totalLength*deltaR + startR
and slope is endR-startR/(endtheta-starttheta)
i think i have to start from zero
oh, no i dont, just mess with l
use L+l and R+r in the final coordinate output, then l and r are the offsets, and L and R are always 0-based
well it doesnt look right - the length isnt constant
i mean it looks like a spiral, but there's more segments toward the center
are you trying for constant path length or constant angle?
(CAV or CLV?)
contstant path length
er, constant segment length really
wow this is more of a pain in the butt than i first thought
there's no real formula for calculating the length along a spiral, you have to add up approximations
maybe there's a calculus way involving logarithms or something
well the calculus formula is too damn complicated to be useful
it would be a linear integration, you're drawing a line
its something like this, with an extra ) somewhere in there: l= ( (0.5*ln ( sqrt ( dt**2 +1) + t ) + 0.5*t*sqrt (t**2 + 1) )
the distance is R*dT(in radians)
i cant use R
you shouldn't need a log
of course you need a log
i dont get people who blog about math and can't write an equation correctly
the length of an arc is r dTheta. the length ofa spiral should be integral(r*dR*dTheta)
spiral segment, that is
so you get 1/2 r^2 *dTheta
1/2 * r^2 *dTheta
(maybe easier to read)
you need to use a min of either some angle or some length, so the 0 radius case doesn't screw things up
both i think
well, if the angle you calculate is too small, then use some minimum angle
if the angle is large enough, then use it (maybe you need max instead of min ...)
r and theta have "negative feedback" and so they oscillate back and forth each time through the loop
I'm assuming you want a "linear" spiral - R increases linearly as theta increases
I'm too tired to think about this much more (I should draw some pictures, but I can't find my pen)
[05:33:50] <fenn> http://fenn.dyndns.org/pub/irc/spiral2.png
<- this is obviously not right
radians vs degrees, anyone?
its all radians
that's the oscillation because r=k*t and t=l/r
so i need to get rid of one
hmmm. if you're integrating more than one dimension (or more than 1 variable), then there's something wrong with the concept
I see you'd have a log if you integrated the l/r
the incremental length is (r+dr/dt)*dt
= r*dt + dr
is slope == dr/dt?
that was more of a difference equation
err - limit
hmmm. I need to think about it when I can actually do calculus. which is about from 2:00 to 2:03 in the afternoon, most days :)
i'd like to think about this as a simple sloped line in polar coordinates but it won't let me :(
actually it worked ok enough to draw a spiral, it just didnt have a constant arc length
r and t change in proportion, so yo uonly have to calculate one of them
the total distance should be proportional to the product of the two (even if it isn't exactly that)
and that's as far as I can take it right now - time for bed
heh thanks for your sympathy :)
I could ask my mother, but that would take the fun out of it
in more ways than one
yeah she would probably say something like l= ( (0.5*ln ( sqrt ( dt**2 +1) + t ) + 0.5*t*sqrt (t**2 + 1) ))
and that wouldnt help me :)
no, she'd prove that there are no integers between 0 and 1
(she's done it before, you know)
(or more specifically, that there are no integers between the additive identity and the multiplicative identity)
boy, that makes me sleepy!
how are hal pins mapped to physical m5i20 connectors & pins ? ( eg: what connector and pin is "m5i20.0.enc-00-index" ? ) I donn see it in m5i20.pdf/m5i20_motion.hal/m5i20_io.hal
hey i solved it, thanks to my total lack of calculousness
[06:02:42] <fenn_> http://fenn.dyndns.org/pub/irc/spiral3.png
fenn: doesnt load (3rd timeout)
timeizzle my nizzle
my router must have changed its ip
fenn_ is now known as fenn
ok now its good
i still dont understand what i did though :(
[06:18:09] <fenn> http://fenn.dyndns.org/pub/irc/lengthspiral.py
if anyone is interested
m5i20 has io0 thru io72, divied up as 3 sets of 24io on 3 connectors ( P2,3,4)
the m5i20 hal files speak of "m5i20.0.out-nn" and "m5i20,0.in-nn" ... was P2 deemed in and P3 out? whats P4 then? (this connector intentionally left blank )
"The assignment of the pins depends on the FPGA configuration."
i wonder why there are two 5i20 folders
fenn: yes, how do i get the physical info from the FPGA files ( i'll have to ask jmk ) and if your interested http://www.dcs.gla.ac.uk/~jhw/spirals/index.html
we ridin' spinnas
well if you can read vhdl one of these files probably has the info you need http://cvs.linuxcnc.org/cvs/emc2/src/hal/drivers/mesa_5i2x/firmware/
will look, (domani) thanks
heh that last one is totally useless for my purposes
the number spiral is quite interesting
i thought last one was a loop ( from -4pi to 4pi step 0.01 ) and ran out polar points at a radius ( a power function) , some polar to cartesian transform should yield xy
well my thinking is trash now... goodnite
# I/O pins (from connectors)
# IO0 is P2 pin 1, IO1 is P2 pin 3, etc
so you need to follow the IOBits<0> ... IOBits<71> through the vhdl
yeh, got that, what m5i20.0.in-00-not? is it on P2 or P3 or P4 where does in end and out begin... i thought a lot of people bought these cards for emc... i think awallin or anonimosu have one running a servo system with emc2, gotta ask them
the files show io0-71 as NET "IOBits<0>" LOC = "p96" | DRIVE = 24 | IOSTANDARD = LVTTL | SLEW = SLOW ; not informative so far
5i2x_gpio.c might be more interesting
btw i think jmk's stuff is new, and there's another config written by peter wallace that has been around a lot longer
maybe jmk's driver only does steppers?
nah that's silly
tom i bet 00 is pin 1 on P2
not having the card, i dont know exactly what pins it exports
its in 5i2x_gpio.c not a list , you gotta ture the code to see it
hm i seem to remember this bug where emc doesnt execute the last line of gcode
i dont remember why it did that
hello mr kwajpol
i have one question abot velocity of setting of emc. How is it calculated. I have 4 axes and want the velocity of the tool to be calculated yust for two axes.
it's calculated by sqrt(x^2+y^2+z^2)
or do you mean feedrate?
yes feed rate
a want it to be calculated ony using x and y axes
well, too bad
use inverse time mode
a rotary move can't happen instantly
is the feedrate also calculated using x,y and z axes
what's the problem exactly?
i have 4 axes hot wire cnc machine
ah, so A isn't a rotary axis then
you should be using UV for your other linear axes i think
ther ate two pairs of parallel axes
there are two pairs of parallel axes
what's not possible?
to use x,y,u,v
dont know...a bug
i think U,V axes have been upgraded recently
yeah there've been people through who were doing hotwire
ok, will try once again, but i am skure it was not possible few month ago
does emc have stuff to synchronize movements?
i meant in the "dual turret" sense, which is what I believe his problem is
maybe i should set 5 axis machine (x,y,z,u,v) and not use z axis
where you essentially have program for each haed
with synch points, and both heads run so they complete at the same time
kwajpol: proper UV support has been added in CVS and won't be released until 2.2, but you can get CVS easily enough
i believe that is necessary for wire edm/hotwire/dual turret machines
toast: it's not necessary for wire edm/hotwire
just dual turret?
you would need two emc's running at the same time handing off synchronization to each other
i swear i've heard my co-workers talk about synch move
since emc only controls one tool at a time pretty much
when they do wire edm
its a limitation of the g-code language, not emc
they use M-codes
yeah for handing control from one control to the other
to define start/stop synch points and time
its fucking retarded
stupid mailing list
kwajpol: read this guy's posts http://search.gmane.org/?query=hulsker&author=&group=gmane.linux.distributions.emc.user
someone needs to create some kind of object-language for cnc
yeah like.. python
or something, yeah
like something between APT and python
i wouldnt mind python bindings for apt functions
all those CUTER RNDFC BLPRT
there's no nice C portable APT library though
well i guess more what i'm suggesting
is a machine that's sort of like a live interpreter
Worth $20 (kit) http://www.awce.com/rs1.htm
where you can define tool objects, machine profiles
Jymmm: get a breadboard
fenn: that was out the door price
its a waste of money
you still have to hook your wires to the board dont you?
might as well just stick the DIP in a breadboard and plug your wires into it
fenn: I'd still have to breadboard it all. PCB already made, just solder it up
OMG 4 capacitors its so fucking hard
fenn: perm usage, not testing
actually i think it only needs 2
I'd still need to get/find max232
well its not very common, you'd have to buy it
Ironically as old as the max232 is, it should be available at 7-11
apparently they are in modems
Let me have a big gulp, a slurpy, and a tube of max232's
heh maybe in china
in china they have them by the barrel full, grab a handfull
there is a TI/stm whatever knockoff of the max232, it's about half the price iirc
there needs to be a IC co-op
its digikey 296-1402-5-ND whatever that is
aw jeez "max232n"
ok it's 4 am, i guess i'm done
does anybody know when emc 2.2 is going to be released?
how do u use this??
this being what?
i need help with emc
is this the place i can get help with emc?
hello? anyone there?
yes, but help available depends on time of day and help needed, have patience
and state problem
what wud be the best time?
anytime you would get to asking a question about EMC I would guess
i have compiled kernel 2.6.19 on ubuntu6.06 with rtai3.5, and then i compiled emc2 from source, but when i try to run it, everything freezes, does anyone know why?
Guest512: what computer
i have sucessfully passed the latency test
what was the OVL MAX ?
it was around 7000 - 12000 i think
i can check again
i am checkin again right now
ovl max is 10390
base period is 100000ns
or even 25000
how can i do this?
edit the ini file
could i ask why?
ask why what?
why change to 50000
it's too high
where can i find this ini file? i am not familiar with rtai
[09:42:07] <Jymmm> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Configuring_EMC2
but the problem is that everytime i try to run emc in terminal everything freezes, including the mouse and keyboard...
you are telling the system to wait 50,000ns todo anything
it's too much, change the value
err change it to 25000
oh found it, shall i leave the servo period and traj period as default?
servo period = 1000000 and traj period=10000000
does anyone else have any other suggestions? bez changing the base period didn't solve the problem
steve_stallings is now known as steves_logging
03cradek 07TRUNK * 10emc2/src/emc/kinematics/tp.c: undo half-baked idea to eliminate the need for stepgen headroom
[14:04:32] <cradek> http://www.youtube.com/watch?v=QYCrpEX7yLI
^ even more complicated tool change than the mazak
heh - cool
just a tiny bit more complicated
yeah, just one more motion
the entire tool-mover is additional. slide, lift, slide, lower, slide
otherwise it's the same
holy crap http://www.youtube.com/watch?v=7zsRvaSV734&NR=1
Le carbone s'enflamme!!
is the coolant carbon (graphite?) or does he mean the material is "carbon steel"?
[14:22:36] <cradek> http://www.youtube.com/watch?v=uAcBU2iqkKE
wow, they have some neat tooling
Unfortunately we can't give more details about the vulnerability because
the German Law (§202)
reboot time. bbiab
a-l-p-h-a2 is now known as a-l-p-h-a
heh - funny marketroid statement in the Java update installer:
"By installing Java, you will be able to experience the power of Java, brought to you by Sun Microsystems, Inc."
I wonder if this will require another reboot
[16:04:40] <cradek> http://www.youtube.com/watch?v=arlS6zPESZA
I think everyone should have cnc
I was threading a part at 1000rpm this weekend
cookies, that is :)
1000 RPM - what thread pitch>
right up to a shoulder, like he's doing there
that's what makes it so hard to do manually
plus the fact that it's - you know - manual
its threading thats making me cnc a lathe
[16:08:30] <cradek> http://www.youtube.com/watch?v=0q_q53wsyHU
manual threading isnt too hard either
this guy either doesn't have a half nut, or doesn't have a threading gauge, or just likes running it backward
the little round indicator thingy that tells you when to reengage
at least I think that's what it's called
I dont have one
a right royal pain without one
threading with emc is very cool...
i bet he just doest know how to use it
I think jmk said he has to thread that way too, because there's no halfnut, or was it something else obvious missing
I had to do that on the sherline - no halfnut
well, I have one, but I have no gauge so I dont know where to engage it..
and no compound slide either, so it was iffy at best
no gauge? bizarre
small lathe :)
no half nut on this crappy lathe of mine, and 1 mm pitch leadscrew
though I have a DRO now
you dont really need a gauge you can just put a little flag on the end of the leadscrew to show you which way it's pointing
not for some ratios...
the odd silly ones like I have to do
now i forget how i did it without one
my al should have cooled down
I've been welding some :)
our monarch has a neat little thumb nut on the cross slide so that when it is turned in - the cross slide comes to a stop - so you use the compound slide for increasing the thread depth then you can run the cross slide out quick and back in to the same exact spot. if that made senes. (I hate manual treading.)
but the monarch makes it pretty easy.
so there's a stop on the cross slide so you can repeat the depth?
[16:51:01] <cradek> http://machinerepair.com/scrapers.html
yeh - it is actuall a mechenism on the actual cross slide leadscrew
gives you like 3 turns back out
only $1250 for the straight edge - not bad
that includes the nice wooden box, you know
heh: "Note that straghtedge prices DO NOT include shipping. You will be billied for actually shiping cost at time of shipment."
you will be billied!
1250 and no straighness spec!!
8-10 week delivery if in stock, too ;)
I think someone needed waaaaaay more (or less) caffeine when that ordering page was written
I don't know if I would have the patience to scrape ways.
let alone the abillity
I'm pretty sure I wouldn't have the patience to scrape ways very well
[18:40:34] <skunkworks> http://www.youtube.com/watch?v=w0-tXDEvAqg
Cool vid skunkworks
I think if I make a trunnion - that is how I will make it. (except a bit bigger)
I mean - If I make a 4th and 5th axis.
I was showing a friend some 5axis vids on youtube last night, no matter how many times watching its just mindblowing
this is one of my favorites http://www.youtube.com/watch?v=EdWWvnrFA9k
I am a turbocharger kind of guy.
I like the one making a scale model of a W12 VW engine
I'd helped him replace the bearings in his supra turbo a few weeks ago.. smart guy
My bad it was a W16
[18:47:34] <skinnypuppy1334> http://www.youtube.com/watch?v=GU32Q6QXtWQ
[18:47:37] <skunkworks> http://www.electronicsam.com/images/turbo/DSC02765.JPG
[18:48:52] <skunkworks> http://www.electronicsam.com/images/turbo/DSC02768.JPG
Nice work, what kind of cutter is that?
heh -carbide wood working round-over bit
only thing I had that had the right radius
Interesting I had wondered how router bits might work on aluminum, didn't expect it to cut cast iron/steel
it did both turbine housings - worked ok. (twin turbo)
I've been in a few stealths haven't owned one though.
it hasn't been driven in a while..
premium gas is expensive ;)
Sure is, I've got an 04 TDI golf thank god. It has enough torque to not ever get mad at it. Not a rocket though
nice - we had a tdi jetta for a while - shouldn't have sold it. 50mpg on average.
or was it 40mpg.. I don't remember
it had electrical problems all the time.
Yeah the older vw's were a bit buggy behind the dash.
I just finished building a 2,2 liter isuzu diesel to put in my 4door sidekick.
[19:01:44] <skinnypuppy1334> http://www.youtube.com/watch?v=Q34xIzTj8fQ&NR=1
Out to the shop...
skunkworks: I want that 5 axis table!
I may have to build one of those.
I suppose you might have room for it on R2
with only 5" of quill, it couldn't be huge, but it could be big enough
I shudder to think about generating gcode for it though
actually I don't shudder too much. It would definitely be interesting.
seems like offsets won't be useful
well tool length and XYZ would be
no idea about the rest
consider what happens if you you move the material an inch to the left while the "C" rotary axis rotates around the same spot in the machine coordinate system
you'd set the origin with the tool tip at the center of (both) rotations
oh hmm, the centers are inside the device
I'd try to build it so A rotates around a line in the table's plane
so 0,0,0,X,X is the center of the table
I think you get something like regular spherical coordinates then
* cradek scratches his brain holder
Brain hurts thinking of the many arc centers there would be
I think this is why programming in RPW is desirable
"I want to poke down to this spot from this direction"
let the machine do what it's good at, figuring out what motor positions give that result
heh - make a ball-ended bar so you can map the center offsets of the rotary axes
you mount it at what you think is the center, and then probe around it while rotating the two axes
in some fashion
and then do some math or something
yeah - I'd go for the "or something" option
what's slightly less than half-baked? third-baked?
Smart thinkin SWP I can see that very well
the trouble is that alignment has too many variables
you can have the axes of rotation be non-perpendicular (though good parts should mitigate that), but even if they are perpendicular, they may still be offset in either of two dimensions
(offset meaning that there is no point on both center lines)
Seems like would be best to set runout of sphere rod on rotary axis first, could be made into a jig after that
anyway - that's my geometric thought for the day
cradek: you already have 1/2 of that from the maxnc. (your rotory table)
might want to be carefull with it on the bridgeport though... one oops and it would be crunched :)
na - pop-crunch
that w16 is sweet :)
a trunion table really wouldnt be so hard to make.. you've got a cnc mill after all
gotta use it for something
the 5 axis thing everyone's yammering about
the hard part is getting it accurate enough
I was thinking BA servos with a BA timing belt - encoder mounted on the table.
that would work
1kw geared 10:1
why are they CMM'ing the counterweights?
on the veyron?
fenn: they arent cmm:ing they are measuring the torque..
for the crankshaft bolts..
or did I miss something?
uh, the stick with a ball on the end that touches the metal.. that's a CMM
I didnt see it..
its amazing how quiet everything is
fenn: I guess the engines are so expensive that it's justified..
fenn: to make sure that everything is within spec
btw, it takes 3 days to machine a engine
they dont cast them
and they take 5 days to assemble :)
they cast some parts
not the block..
[20:09:06] <anonimasu> http://www.youtube.com/watch?v=GU32Q6QXtWQ&mode=related&search=
i like that video, real simple, no stupid music
not the block machining one, the assembly
I like the machining..
though the music sucks.
on the W16 video I think you can see that the center of the table is also on the A (?) axis like I was proposing
I guess that makes the math easier too
cradek: why would it matter? you have to do kinematics either way dont you?
it just seems nice to get a hemisphere
if the table surface were further out from the center line you'd get more than a hemisphere of reachable angles
that 5 axis mill is slow compared to this: http://www.youtube.com/watch?v=Sebko0n6uJ0&mode=related&search=
I keep counting but I see 12 cylinders on this W16
5 axis brings up another interesting problem - how do you clamp the workpiece? :)
seems like there's room for two more, but that makes at most 14
bolt it from underneatH?
and registration for a second pass (to get at the 6th face) seems like it would be a nightmare
SWPadnos: is it any harder? just a vise or clamps still, I think
fenn: well, that's a test.. not production
it's just that 5 axis theoretically gives you the ability to reach more of the workpiece, but the clamps would then be in the way
huh, more cylinders just appeared at 07:30
cradek: look at 6:35
yeah I don't understand where they came from
cradek: keep in mind it's a 3 day process to cut them
oh they turned it around. I see.
SWPadnos: I think they have the same pallet mount points at both sides..
maybe they forgot to put it in on v 1.0 when they were filming the first half :P
SWPadnos: so they can flip the part and align it to another pallet :)
or another pallet with other holes/locating pins
though I may be wrong :)
right - but you still need flat spots or complex jigs to hold it after you've made those nice 3D curved contours :)
its so silly that they use these huge 5 axis machining centers to do finish work with 1/8" ball end mills
[20:23:19] <anonimasu> http://www.youtube.com/watch?v=Wbbt3L1zty0&mode=related&search=
wtf.. fast machine slow cuts.
that's slower then the old mill at work..
fenn: yeah that's silly
we need more robots!
ten of them at once, each with a micro finishing tool spinning at 100krpm
harbor freight has a 56krpm air spindle with 1/8 collet I've been wanting to try, it's $19
You don't just need ear muffs - you need whole head muffs.
that's what she said?
yeah I bet it's loud.
you need garage/barn muffs
and an air compressor with enough cfm to run it.
56k el-cheapo air spindle: $19. ultra-efficient noise canceling headphones: $499
[20:30:53] <cradek> http://www.harborfreight.com/cpi/ctaf/displayitem.taf?Itemnumber=47869
the box says it takes only 2CFM to run full speed
I see that. cool
ah.. i have some of those, they're real crap
I wonder how long they last
if you ever wanted to feell like you're at the dentist, but cheaper!
if you honed the bearings or something it might work better
its not too loud, louder than a dremel, not as loud as an angle grinder :)
heh - so what ipm could you do pcb with that?
skunkworks: 60 maybe
the tools I use are meant for .0005/tooth
i'd think a geared-up dc motor would be better
with a round belt
fenn: yeah probably so
I don't know where to find the spindle though
you've got a lathe dont you :)
lots of places..
you sure have a lot of projects for me
see if you do it first, then its less of a risk for me
fenn: where's the fun in that?
anonimasu: having projects work the first time is fun, actually
fenn: did you see my feedrate issue last night wasn't all me?
did you ever figure out what was causing the random motion?
skunkworks: tell me about it
no - I try to remember to shut the drives off before I exit axis.
fenn: well, doing the same thing as everyone else gets old..
still forget thoguh
* fenn shrugs
I was going to see at what point it starts at bootup.
skunkworks: is that the feedrate issue?
I don't know about you - but this doesn't look like 800ipm
[20:41:00] <skunkworks> http://www.youtube.com/watch?v=2CuqnT0a6t8
DOC .050 IPT .030 SFM 800 IPM 135
I believe 135 ;)
looks like they need a gooseneck
I don't concider that hsm then
[20:44:00] <anonimasu> http://www.youtube.com/watch?v=h5w3yGcEevw
though they just make dust
no real chipload there
[21:07:51] <anonimasu> http://www.youtube.com/watch?v=OL0nf19g5vk&mode=related&search=
Fast but sharp poke ya
i wonder if that pokey thing was done with ultrasonic machining
maybe once they get spindle speeds over 1 million rpm we wont have to hear annoying spindle noises
03jepler 07TRUNK * 10emc2/src/po/ (fr.po fr_axis.po fr_rs274_err.po): additional updates from Francis TISSERANT
fenn: feel free to be the first one
magnetic bearings, permanent magnet motor.. what's the big deal?
fenn: im sure there's one
[21:27:57] <anonimasu> http://images.ucomics.com/comics/db/2006/db061105.gif
fenn: if there werent a deal we would see 100krpm spindles all over the place..
i think you'd have to build the tool directly into the motor shaft
apparently there are 500krpm "spindles" although the tool itself is the only thing spinning
yeah i've read that numerous times :)
check this out: http://www.gizmag.com/go/6245/
it's not like I care too much about it..
I need a 10krpm spindle max..
* fenn chuckles at the thought of a gas turbine spindle
* JymmmEMC hands anonimasu a dremel
JymmmEMC: real funny..
Hey, I only saw 10K, no mention of HP
or mouse power either =)
* JymmmEMC hands anonimasu a lawnmower
im off to bed now again..
it coulda been a generator, but the lawnmower is easier to bolt up =)
LOL.... we had an 8000-60000 rpm machine... same thing no HP rating...
dmess: 1mm cut at 10krpm in alu.. >p
at 8000 i grabbed the collet and stopped it..
that's what I want..
3d contouring with small tools
anonimasu: talk to skunkworks
6.5mm dia cutter max for high speed stuff..
anonimasu: he's selling an engraving machine
Well, he keeps SYAING he's seeling it, but he plays with it more every day =)
I dont want a engraver..
it does 1mm at 10k in al iirc
I want a mill..
the spindle is easy.. the housing.. another issue
dmess: not really getting close to zero runout is hard..
or expensive :)
look for a good iron 3 axis machine going to scrapper...
that's not the issue..
oversize the bearings and order a spindler from bagg or the sorts..
did you check the price of a hf spindle lately?
if ya wanna PLAY ya gotta PAY.... for HSM
nothing cheap is good... and .. nothing good is cheap
I guess we should give up this emc stuff..
spindle wise my friends...
rigidity and kinematics is 1 thing most of you dont consider..
well, im off..
now taking a break for dinner .. more updates later
jepler: FWIW, On my xylotex board, I'm using a total of six fans.
jepler: One blowing down directly on the heatsinks, and the others to disapate the heat from the enclosure
I need to add a MAX232 to a gps, and extend the cable about 40' or so. What I was wondering which side should I extend the cable? between the gps and the max232, or between teh max232 and the computer?
JymmmEMC: are you running a lot closer to the current limit than I am? right now my current is set to 1.5A
right now it feels like it will be fine without any active cooling
seems like power dissipation should be less than 2W/package -- "Low rDS(ON) Outputs, 0.45 Ohms Source, 0.36 Ohms Sink Typical"
which is in line with what I feel with my fingertips
the m5i20 connector/pin assignments are in HOSTMOT54-E.PIN. now waiting for DAQstuff to show up.