cradek: is the driver for the mesa board done?
JymmmEMC: the new modular one isn't, but the old one works like it always has
the big new feature will be step generation, which Adam won't care about anyway
cradek: which is how? (the old one I mean)
JymmmEMC: just great for servo systems
cradek: I am thinking for small stuff like #10's... problem is the existing spindle can't instantenously reverse
ds2: it doesn't have to, no spindle can
then what's the requirements to be able to rigid tap?
it should reverse "pretty responsively" when switched from say +300rpm to -300rpm
and the slower you tap - the faster it will reverse ;)
i.e. it should neatly slow and reverse within a few turns
some machines would have to go pretty slow I think.
so you are saying I can do it with the normal DC brushed motor + SCR based controller?
we used 300 on the mazak.
if that's the case, then all I need is an encoder
I understand any VFD can do it, but may need a braking resistor
yeah if you can command S300M3, then M4 and get a nice reversal, you're entirely set
this is just plain old DC (threadmill motor)
I found a bit of info on my servos... max 2400rpm, volts 140, pulse-amps 30
tacho 9.5volts / 1000rpm
cradek: is the encoding the same as for single point threading, 1 pulse per rev?
ds2: your question is not right. An encoder (or many pulses per rev) are needed for threading. An encoder is need for tapping because it reverses.
an index is required for both threading and tapping.
you should have at least low hundreds of pulses per rev so an encoder is the obvious way to get it.
did you drop a word or a typo in the first line?
'An encoder is needED for tapping' is the only typo I see
ask another question if it's still unclear
my understanding so far is -
an index is required for both threading on lathe and rigid tapping
what I don't get is what is the difference between an encoder and an index
'an encoder with an index channel'
many encoders don't have index, only AB/quadrature
okay so for threading, only an index is need but for rigid tapping both index + encoder is needed?
for both, an encoder is needed
for both, an index channel is needed in the encoder
(it's strictly true that for threading, on a spindle that never reverses, you can get by with only one channel plus index, as long as you have many pulses per revolution)
for rigid tapping you certainly need an encoder - how else could you tell when it reverses?
I am thoroughly confused...
let's go back to a reference point - the lathe
well they are different issues
when cutting threads on a lathe the spindle does not reverse, so you don't need a signal that tells you about reversal.
so is that why it does not need more then 1 pulse per rev?
one pulse per rev is not enough. You need hundreds.
you can not adequately follow the spindle orientation with one pulse per rev.
I am almost I found a reference that says otherwise... looking for the page right now
so drive that notion entirely out of your head!
emc has never done that. other controls do.
then how much angular resolution is needed?
the other controls will cut "drunken" threads if the spindle speed varies AT ALL during a threading pass. That's no good for emc.
I would not want less than one pulse per degree or so
okay, I think this actually clears it up for both tapping and threading... cuz I was under the impression it was done with a single pulse
for tapping that's clearly impossible
for threading it's possible but it would require a lot of "faith" or tolerance for bad threads
I suppose for threading if your speed control is rigid enough....
yeah if you have a very heavy spindle and very light cuts, it might possibly work tolerably
I am guessing for tapping, the angular resolution required would depend on the pitch as any error translates into stress on the tap
yeah that seems true
on the mazak we tapped entirely rigid (in a jacobs chuck) and there seemed to be a little stress when backing out
a holder that gives a few thousandths would be nice
but what pitch and size thread?
the spindle encoder on the mazak is 360 ppr I think
I mean the tap
it was 3/8-16 I think??
but we successfully tapped (a lot) in Al
that's pretty course... I am thinking of stuff like 6-40, 10-32, etc
we made something like a tooling plate
you probably want plenty of resolution and low speed - if you play with it, let me know how it goes
got to think of a good way of encoding it
yeah it can be a challenge
I wonder if the 400line US digital encoders would tolerate 10KRPM
on my BP there's the top hole where the drawbar would have gone - I can just stick it there
good question. 10k is very fast for an encoder.
but the bridge port spindle isn't capable of going very fast... the spindle I am thinking of can do 10K
if you get the kit type (no bearing or contact) you'd be fine as long as the wheel doesn't fly apart
bearings assemble is good to like 12K on it
jmkasunich__ is now known as jmkasunich
jmkasunich: Did you say you have an IEEE account?
jmkasunich: know anyone that does?
jmkasunich: ok, thanks
SWPadnos: yeah? can you grab a doc or does tht cost more?
I wonder if I have a login for online docs. I should have one somewhere
just wow - Hot wire foam cutting - http://www.youtube.com/watch?v=nVCS2JX-Dww
the hektor robot in ecmascript and svg. pix http://imagebin.org/10017
runs in opera9
i havent figgered how to capture the little vectors yet, maybe java can do file io from a browser
if the html doesnt load, load the svg, then reload the html
the segments should be able to go into halstreamer, getting quadratic beziers into emc
okay, I think I am ready to try moving heavier machinery
why does one need an index channel to do rigid tapping?
fenn: figure it and tell me
Damn that took forever !!!!
"that the strength of programmers' opinions is inversely proportional to the amount of data they have"
alex_joni: your ego is way too biug is what fenn is saying
no it ain't
lol, yeah it is
and you know it ;)
well.. then I'll hide for a while
actually going to france for 2 weeks+
why hide, it's publically know you have a ego the size of the goodyear blimp
alex_joni: dont do anything I wouldn't do, and if you do, name it after me =)
I'll be back on the 10th of sept
depends on TZ
9/10 there, 9/11 here
it's always 9/11 somewhere
Jymmm, waddyameean slumming in #mysql, I run the bot in there
archivist: Okey, then your slumming in here ;)
cradek_ is now known as cradek
* skunkworks__ doesn't have any new videos - sorry
good morning skunkworks__
you have a tail
yah - got busy last night and left a few computers on. I almost got trunk installed but a hole in the wifes tire derailed that project
skunkworks__: trying to be pythonic?
I am sure everyone is getting sick of me posting the same video :) tough
skunkworks__: show off.
skunkworks__: i hope you g64'ed, because all those line segments are bad for the steppers
heh - do they blow up?
should I be wearing safety glasses?
and the last question - does that look like it is running exact stop mode?
heh, im not sure..
depends on your accels ;)
exact stop mode goes zip zip zip with a fast machine :p
the machine is set to 25in/s/s
or 635mm/s/s - I think
for you metric people.
how do convert to metric seconds?
what metric seconds
* skunkworks__ is feeling a bit goofy this morning. sorry
"(why are they called axes in HAL and in the ini file and joints in the manual ?)" <- he's got a point, but its a bit late to fix that now
easy stuff ;)
fenn: for "peck tapping"
cradek: i still dont really see the need.. it's not like you have to worry about counters overflowing or anything
actually a spindle can overflow pretty easily. but you're right it would be possible to do without index.
during the tapping operation
if it can overflow, it can overflow during tapping, you just have to be unlucky
counters overflow at 2^31 counts (signed 32-bit int), and resolution starts to be lost at 2^24 counts (float)
if the basic types of HAL were 64 bits I wouldn't worry about it
for a 32 bit int, a spindle with an 8000 line encoder at 6000 rpm will take 5000 years to overflow?
but unfortunately they're not
units '2^31/(6000/min*8000)' years
Definition: 2684.3546 s
ah pesky order of operations
Definition: 20.97152 s
2^31 ~= 2 billion. a 4000 count encoder at 500 RPM gives 2 million counts/minute, so only ~1000 minutes are needed for overflow
units does a/b*c differently from most programs you're familiar with ..
confusingly, 2/3/4 and 2/3*4 are the same in units ....
if you were turning a long screw, you could easily have a program that lasts 16 hours
ok, so, it's not "required" but "suggested"
a much worse case is run at 10krpm for a while, then slow down to thread
fenn: the current implementation requires it.
that's the "unwinding"" problem?
how about a "spindle position reset" signal from emc?
err - I think I'm missing a piece of the puzzle for this conversation :)
nevermind - gotta get to a meeting anyway
i guess you could do that with halcmd
why make it more complicated than it needs to be?
if the internal types were signed 64-bit int and double, you'd get 356 years of 6000RPM before loss of precision, and 365ky before overflow
encoders without index are a lot more common around these parts
jepler what do you mean "loss of precision"?
is spindle position not an int?
it should be a double, not an int
spindle position is a hal "float"
for scale=8000, 16777216 and 16777217 counts convert to the same position. In C syntax, (float)(16777216/8000.) == (float)(16777217/8000.)
float was used instead of double because storing and loading double-precision numbers (64 bits) is done by nonatomic instruction sequences in at least some cases, but access to HAL types must always be atomic
Can semaphores be used within HAL?
rtapi includes semaphores (rtapi_sem_*)
they are not actually used in HAL
HAL only uses mutex'es from rtapi
So access to non-atomic variables could be controlled by mutex'es.
it would be a big change in HAL to make access to pin data require a mutex. If 'p' is a HAL pin then '*p' reads the pin and '*p=x' writes the pin (in "C"; comp dispenses with the "*")
you would also have to change the thread priority model: right now, the priority is simple: the fastest thread has priority. But if you use mutexes to control access to values in HAL, you have to let a slow thread preempt a fast one when it holds a mutex for some value to be read in the fast thread.
lerman: what's the goal?
To permit the use of doubles or long longs in HAL.
easier to wait for rtai or another realtime kernel to become stable on 64-bit machines; then 64 bit primitive types can become default without any changes to source code
Also, you can make pretty serviceable plastic nuts with a tap and some cutting boards.
^^ what a great idea
* fenn hopes thats not sarcasm
no I'm serious, that's a good source for ... stock
yeah i used to use them before i discovered a plastic supplier's dumpster
now i've got more HDPE than i know what to do with
could be worse.
(wish I could fine Al that way)
i think emc.so is pretty much what i was talking about for "python bindings"
yeah, people keep insisting that we should have things we already have
loving/hating gcode is the other common thread
(I also love/hate gcode fwiw)
the /lib dir is always confusing
How do you identify the "live" connector on AC inlets that look like "A" on this page? http://www.mouser.com/catalog/631/943.pdf
i'm usually grepping in src/ and i'm like "wtf i know its here somewhere"
(on the back / terminal side)
I always check how a power cable is wired
(the small prong on the outlet is the hot one)
use a meter on one of the cords, test for continuity between the plug and the other plug
i guess that's what cradek said
the other interpretation is "always check how a power cable is wired cause it might kill you if you're wrong"
power doesn't kill us here in low-volt land.
when it bites us, it just makes us mad
we should all switch to 10KV 10KHz AC immediately!
The router will ship today. I am sorry for the delay. I usually ship the next day, or at the latest,2-3 days after order, but I ran into some unforseen delays. I am going to send you some motor cables with the router. Thank you for your patience.
</email> well at least he wrote me back promptly after I asked him about it
have anyone seen the limit of the new tp?
[14:43:36] <anonimasu> http://www.youtube.com/watch?v=Wr5GywChdd4&mode=related&search=
high speed milling
that doesn't look high speed to me
internet news flash: some videos on youtube may be boring or pointless
pretty lethargic actually
that was the point..
I watched a sodick machine..
zip zip zip
is it actually making any chips? http://www.youtube.com/watch?v=matwPj6DKY4&mode=related&search=
it makes dust..
[14:47:22] <anonimasu> http://www.youtube.com/watch?v=JsWVTQy2rYg&mode=related&search=
air for coolant and chip clearing I think
video title: "DATRON High speed milling machines" Related video title: "I pimped my dog" -- wtf?
[14:49:27] <anonimasu> http://www.youtube.com/watch?v=h5w3yGcEevw&mode=related&search=
there they actually make chips
the speed that the penguin ran was as fast as it would go.. I think. for me
with the constraint I had
[14:53:21] <anonimasu> http://www.youtube.com/watch?v=ITeRJCkMQ90
I have not tried the g64 pX.XXX yet to see how much I can speed it up.
[15:12:13] <cradek> http://www.youtube.com/watch?v=nibzhPfOF1Q&NR=1
^^ another emc2 spotting
wow he has his acceleration set low. I wonder if the machine really needs it that low
Nice. Looks like a stout little mill.
I wonder what that part is. It looks like a marble holder
no clue. He doesn't mention emc though - shame on him :)
that looks like a very big mill for steppers.
although it seems to work OK.
wow, not a good hole milling algorithm: http://www.youtube.com/watch?v=uiyzDZhcqP8&mode=related&search=
I think people think end mills are drill bits. They'll plunge without thinking twice, but are afraid to cut much of anything with the side
looks like he could have done this in about three cuts if he'd use helixes instead of plunging
depends on the mill - some are center cutting - but you still want to be carful with speed. S
sure, but even center cutting end mills don't plunge well
they only barely work
well is reletive I guess. We have done it a lot. maybe it depends on the size of the machine. ;) try it on your bridgport now.
cradek: so you prefer a helical plunge to the maximum diameter, then a single spiral in to the middle?
true maybe I'm overly gentle with plunging.
jepler: for that hole I'd rough it without plunging (not quite full diameter) and then do a finishing pass around the outside
needs a solid machine, a plunge here on the old stuff vibrates a "little"
that is awallin's
hrm scrollback.. yeah.
[15:52:39] <fenn> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Videos
hmm - I should probably add some links
Is there a way to start an interactive emcsh from *another* xterm when emc2 is already running?
MichelG: yes -- give me a second to do it myself so I can give you the right directions
MichelG: in the shell, change to the same directory where your .ini file is
MichelG: then run it this way: emcsh -ini yourinifile.ini
jepler:Thanks, I am trying to do that for a couple of hours without success.
MichelG: do you get an error message?
jepler: I am doing that with inifile sherline-mm.ini, but I get libnml/cms/cms_cfg.cc 624: cms_config: can't open 'emc.nml'. Error = 2 -- No such file or directory
libnml/nml/nml.cc 369: NML: cms_config returned -1.
* BufferName = emcStatus
* ProcessName = xemc
% emcsh -ini axis.ini
% emc_joint_homed 1
* Config File = emc.nml
* error_type = 0 (NML_NO_ERROR)
MichelG: are you sure you changed to the right directory and gave the right argument after "-ini"?
MichelG: I can get exactly that error *if* I am in the wrong directory or I give the wrong name for the inifile
jepler: I am in emc2/bin, my command is: ./emcsh -- -ini ../configs/stepper-xyza/sherline_mm.ini
MichelG: change to the same directory where your .ini file is
It is correctly loaded by emcsh, but trynml() gives an error
MichelG: then run it this way: emcsh -ini yourinifile.ini
the first step (change to the directory where your ini file is) is important
you may need to give a path to emcsh in this case .. e.g. ../../bin/emcsh -ini sherline_mm.ini
jeppler: OK, that was the trick; THANKS!
MichelG: good, and you're welcome
MichelG: if you run into questions about python's 'emc' or 'hal' modules, feel free to ask them here too
jeppler: thanks, I am experimenting for creating "druids" ...
jepler: the only think I would say would be nice in your image to gcode is actual size of the output. So you don't have to do the math. ;)
skunkworks__: I agree
(the only thing ?!)
* skunkworks__ has only used it for a little bit so far..
i should have added 'so far' ;)
skunkworks__: did you cut something from it yet, or just look at the gcode?
just looked at the gcode - hopefully there will be another video tonight :)
I know your excited
it looks like arcs now also ;)
did you find a good image to work from?
not really - do you have some that you would think would work good?
there's the torus which is in nc_files
yes - I have actually dry run that one also.
it's not too interesting though
in the installed version
I am converting one of my car right now for the heck of it - but the computer here at work is a 500mhz celeron and it is running really slow
image-to-gcode is pretty inefficient / slow
here's another image I ran into on the internet: http://emergent.unpy.net/index.cgi-files/sandbox/SampleDepthMap.gif
oh - that is cool.
I thought a nice lithograph would be a good selling point
if it cuts it decent
I looked at the bits I have - and the ends are differnt metal than the shafts.. They may be carbide. That would help with the circuit board demo.
I was going to check them with a maganet tonight
skunkworks__: good news -- I have a tracking number for my cnc router
" In FedEx possession " it says when I track it
does it give an estimated delivery date?
Well - I hope you get it before the weekend.
that would be nice but I doubt it
they normally deliver on saturday also (fedex) atleast they do around here.
just go get some cutting boards and we'll throw one together
it takes a good 15 to 30 minutes to convert a image on this computer. :) painful (using mostly default settings - no roughing passes)
the mouse moves every second or so.. I thing something may be horribly wrong ;)
how big's the image?
if the mouse is moving erratically that usually indicates inadequate RAM
it get rt error instanly when emc starts. - this is the same model computer I use at home except it is a 500 celeron vs 1ghz pentium III.
could be BASE_PERIOD too low
the torus takes a long time also
it is the default 50us
03cradek 07TRUNK * 10emc2/src/emc/motion/control.c: fix jog wheel velocity mode's interaction with feed scaling
Does anyone know if anilam 5volt glass scales are compatible with the Mesa 7i33?
what's their output?
i have no idea I cant find any info on them
you may have to get out the 'scope then
well I believe that they output "Square-wave voltage signals, channels A and B, in 90° quadrature relationship"
if the signals are single-ended TTL compatible they are very likely to work with mesa.
or they are "differential sinusoidal current or voltage output"
hmm, not so good then
with a scope you should easily be able to tell if they are sinusoidal or square
I am not sure if they output digital or analog though
we're not going to know better than you are :-)
so were are good with square and not good with sinusoidal?
actually you're way better with any of them vs. some digital serial protocol
like on heidenhain
but if they're the only positional feedback, they won't be that will they?
with sinusoidal, you could add an analog comparator per signal to convert into a clean digital signal -- but this sacrifices resolution
the glass scales I've seen are absolute encoders basicly
and they transmit some packages with the absolute position
something designed to read a sinusoidal signal would get many different "position levels" during one rotation of the signal, while converting it into a square wave with a comparator will give you only 4 levels
and all other kind of strange stuff
jepler: maybe a resolver to quadrature converter might help there
basically what I have right now is an old anilam crusader M with glass scales,and SEM servos with tacho's. I am just trying to figure out the hardware I need to purchase to convert.
the one I've seen are 1024 counts / sinusoide
I would think glass scales would be digital.. but what do I know. ;)
I think they actually work with some magnetics on the glass
first record a sine wave on an 8-track tape, pry it open, stretch the tape out, hot-glue the tape head to the axis
feed the output into the line-in
they are square wave ttl quadrature output afterall
that's really an interesting idea -- DS floppy disk as high resolution magnetic encoder
with index! (the hole)
phase A on the top, phase B on the bottom, and the index hole for index
hmm I like those ideas
oh I guess it was obvious.
did the 3.5" disk have an index hole?
I think it oriented one way
oh that's right, there's a notch thing in the center metal disk
can floppies handle the workload
15 sector/track * 512 bytes per sector * 8 bits per byte / 2 bits per cycle = 30720 cycles per revolution (5.25 "HD" disk)
no, you'll probably have to change your encoder once per week
anbody have any ideas on how to switch a bridgeport style induction spindle motor to a DC motor, and have it controlled with EMC? I will be getting the Mesa 5i20 and 7i33 and breakouts.
just run it with a VFD. if the VFD has +-10 input for speed control, wire it up to a spare 7i33 output
Adam0: is it 3phase?
yeah it is a 3 phase
vfd would be the way to go.
ok what extra hardware do I need?
[19:58:24] <cradek> http://web6.automationdirect.com/adc/Shopping/Catalog/AC_Drives_-z-_Motors/GS1_(120_-z-_230_VAC_V-z-Hz_Control)
right now the spindle is on a 3phase converter because I have no 3-phase in the shop
for 2hp you can get a single-phase 230v input vfd
[20:00:12] <cradek> http://web1.automationdirect.com/adc/Shopping/Catalog/AC_Drives_-z-_Motors/GS2_(115_-z-_230_-z-_460_-z-_575_VAC_V-z-Hz_Control)/GS2-22P0
the DC bus transformer will probably be 3 phase too, so this isn't the only work you have to do for single phase conversion
alex_joni: you around?
what do you mean by that cradek? I am not that good with the electronics side of things.
these look pretty nice. configurable IO like "at speed" output
Adam0: you can run the spindle motor, using a VFD like the second link I posted, using single phase power. But I bet other parts of the mill currently require 3 phase too.
but I'm just guessing of course.
alex_joni: I remembered your astro-photo from a few weeks back. I'm thinking about building or getting a tracking mount.
cradek the spindle 3 phase is on a different 3 phase system.
oh ok, then it's easy
you can run it with a vfd and emc can control it directly
alex_joni: how long was the exposure for the star-trails pic?
cradek the vfd is wired to the 7i33 then?
yes the vfd has, among other inputs/outputs, a 0-10v analog input that can be used to set the speed
ok well that makes the spindle mod easy
awallin: about 45 minutes iirc
wow 10 lathes on the poll
at this rate lathes might surpass "nothing"
cradek: by "gcode works so
differently for different machine configurations" i think you mean "gcode is utter shite and has to be hacked to work properly in every instance"
well, I don't really want to talk about that
so, got any better ideas for how to do UVW in a universal sense?
it's not a problem of UVW, it's a problem of gcode dialects in general
there is no universal sense for anything beyond XYZ, and there may not be for XYZ either
at least, not for the machine controller
i'm sorta frumpity about W being ambiguous (could be yaw too)
everyone agrees on G0,G1 X,Y,Z. *anything* else is more or less variable.
it's possible to decide that XYZABC means TOV, but there's no universal mapping of that to machine joints
T O V?
(TOV=tool orientation vector)
or is it tooltip orientation vector?
rotary axes are not commutative anyway
even XYZ is somewhat subject to debate, when you throw in lathes and front/back cutting
i think a right hand coordinate system is pretty standard, and Z parallel to the spindle
so lathes should all be the same?
unless your cutter is on the back side of the stock
the main spindle
and on a foam cutter?
if there's no main spindle it's not really a lathe is it
there are no (or barely any) universals
is it that way on a lathe? (X=diameter/radius, Z=distance from chuck)
ok - I keep forgetting that
well foam cutter would be XY because Z is generally considered depth
maybe I should get a lathe and not retrofit it for a few years, like my mill :)
the guy I bought my mill from had a nice turning/milling center for sale too
SWPadnos: sounds like a plan
it was a little bigger.
get a cnc lathe though
cnc lathes are so much nicer than manual lathes.
well, there's a nice one for sale here, I think. I just haven't been able to get the folks to sell it to me (though I haven't tried all that hard)
the extra 30-50 years of development since wwII helps a bit
it must not be very "for sale" then
well, it's for sale, but I haven't pushed much to get them to sell it to me. they're still using it AFAIK
I also had a typo/thinko in the last email I sent - I asked about a CHNC, and the unit I had looked at was an HNC
but little did I know they also had some CHNCs for sale ...
(= later, more expensive unit)
re: the floppy disk/tape idea, you could have a pseudorandom code to give absolute position
yeah you could have lots of channels on a floppy
i was thinking more for the tape, where you'd only have like 2 or 4 tracks
need the heads to read them though
2 is plenty then
I'm not sure a floppy read head could handle that - I think it needs rapid magnetic transitions, so slowing it down (liek an encoder) may not let the head detect any bit transitions
all you need is some of these http://www.electronicsam.com/images/KandT/conversion/accupins.JPG
but then there is modding a hard disk.....
swp slowing it down?
archivist: your hard disk is nothing compared to my holographic versatile disk!
well, with encoders, you want to see how far something has moved, so you run the thing at some rate other than the 300 RPM of a floppy drive
(or whatver a floppy runs at)
yes could not read the magnetic at slow/zero speed
soooo burn special cds
have you seen dvd-ram?
also, the clock rate is probably somewhat fixed - trying to get data at any other rate might be difficult
bah now i cant find a decent picture
anyway, it has built-in sector markers
fenn: so does lightscribe DVD's
that's also a good idea
lightscribe gray code
[20:43:30] <alex_joni> http://dsplabs.cs.utt.ro/~juve/dropbox/lightscribe/IMG_9642.JPG
CD might object to sitting on a 30k rpm spindle
52x is pretty fast
yeah, close to 10k or such
27.5krpm says google
i'd bet on the 10k
1x = 200 rpm
I'm pretty sure CDs are 300 RPM
[20:44:54] <alex_joni> http://hypertextbook.com/facts/2000/LawrenceFung.shtml
03jepler 07TRUNK * 10emc2/src/hal/utils/comp.g:
'modparam' declaration allows the documentation for the module parameter to be
automatically integrated in the generated manual page. 'modparam int' actually
generates the declaration, while 'modparam dummy' just inserts documentation.
swp maybe on those fancy 66khz lp's
any last feature requests for emc2?
03jepler 07TRUNK * 10emc2/src/hal/drivers/pluto_servo.comp: use new 'modparam' to improve documentation
it depends on whether it's CAV or CLV, I guess
* alex_joni goes away for 2 weeks with a laptop :)
alex_joni: have a good trip
don't use the computer too much
jepler: only when I finish my stock of books
well, then you could add a lightweight SVGALIB/FB version of AXIS then :)
oh, and HAL hardware interrupt threads
(if you have time)
please fit the FB version in 640x480 =)
fix keystick so it doesnt need an x server
make an aalib port of axis so it runs on the terminal
oooooh - that's a good one :)
use libnethack for a nethack interface to emc
make EMC runnable on a 386 w/16M of RAM and 32M hard drive
you shoot the part files you want to run, and when you have to change tools, the big boss monster comes out and yells at you
no no -- the palm 100
ds2: i think it already does
fenn: I been told otherwise
ubuntu sure won't
ds2: emc runs on crappier hardware
I think the timing bits in RTAI/RTAPI need a TSC
just not the latest distro
which implies 486, no?
don't care. I just want to embedded into a system that fits in the original machine chasis!
rtai probably bitrotted for pre-pentium CPUs
TSC implies 586
you'd have to hack out the bits that use TSC (they're only for statistical uses anyway)
well in that case ...
reporting of thread / function invocation times, for instance
* alex_joni packs more books
primitive latency checking in task
and so on
alex_joni: rtai hacking for dummies? :)
I don't suppose these super-duper processors actually allow you to use the TSC or some other register to generate interrupts, do they?
and you'll at least need 386+387, there's no FP emulation in kernel space
removing the TSC dependacy would be good anyways...then you can use WinChip's and other early pentium clones
okay, I'll settle for 486+
there are a lot of 486 laptops floating around
Elan SC520 (though the interrupt controller really sucks)
ds2: patches gratefully accepted
when winter comes, I'll do more code hacking
nothing major, just like last year...little bits here and there
don't sell yourself short just because you don't do anything "major"
well.. see you guys
yeah - look at me ;)
almost any project will benefit from someone who sees an area that needs cleanup or tiny improvements, and then follows through with patches
because too often the developers are concentrating on the next big feature
what might help speed things up is if someone can set up UML images for compiling the RTAI stuff
if you are improving code quality and adding polish to the system .. that is really valuable
my biggest stumbling block to going deeper is no time to setup an RTAI build system
compiling rtai cant be that hard - even i can do it
it isn't that it is hard, it takes time and normally requires yet-another-machine which means another HD + Box + etc
UML gets around the last part
have you seen this? http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?RtaiSteps
you assume people are using apt
i'm afraid i dont see how uml images would help. that's like flowchart diagrams right?
User Mode Linux
they should have chosen a different acronym then
basically a Linux kernel + image virtualized as a user process
ds2: actually when i wrote those instructions i was using redhat
qemu/vmware is a bit heavy weight
i doubt rtai would work with UML anyway
or any virtualization technology
but compiling it should work, I'd think
uh. but then why would you need UML?
i thought uml was to test the code you just compiled
didn't say work
just to let me compile as I understand it, wants special headers
no, it just wants linux kernel source
what I want is a cross build enviroment
Oh? so I can boot the Ubuntu EMC2 CD and steal the headers from tehre?
you should be able to compile with some incantation like ./configure --with-realtime=/path/to/RTAI --with-kernel=/path/to/soem/kernel or similar
maybe.. but the safe route is to get sources from kernel.org
I am trying to avoid the patching time sink =)
but I will look in that when things wind down later this year
patching is pretty trivial - it takes like 30 seconds once you have RTAI
plus download time ;)
it really goes in cleanly?
mv linux-2.6.10 linux-2.6.10-adeos
patch -p1 < ../hal-linux-2.6.10-i386-r9.patch
most patches I deal with don't and it is hours if not days of resolving conflicts, etc
if you pick a matching vanilla kernel and rtai patch it will go in without conflicts
okay, even better
but if you choose anything but an exact match (and that may not always be obvious from the rtai patch naming) then you may get reams of conflicts
or if you choose a distro patched kernel as a starting point
yeah, dont do that
I like starting with K.O sources
it is just patches that go in cleanly have been rare lately
don't matter if this works cleanly, I'll just do a chroot and be set
the only other stumbling block is making sure you use the same gcc version to compile rtai as you use to compile emc/hal
are there explicit checks or will it fail in subtle ways?
i think there are checks now
looks like i'm going to be messing with openvz soon
"The unique feature of User Mode Linux is that you can run it under standard debuggers for studying Linux kernel in depth."
and it is lighter weight then full virtualization
03jepler 07TRUNK * 10emc2/src/hal/utils/comp.g: make TRUE, FALSE appear in manpages
im having a heck of a time d/loading an iso... 3rd try now...
im on a win-blows box
[22:04:19] <fenn> http://gnuwin32.sourceforge.net/packages/wget.htm
i'll pull it in and try to install IT
how come I get this running head with a 2.1.7 config? I don't see anyting on this page http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?UpdatingConfigurationsForDevelopmentVersions
emc/task/emctask.cc 312: interp_error: Required parameter missing
Required parameter missing
emc/task/emctaskmain.cc 2653: can't initialize interpreter
trunk I mean
wai t - that is the var file issue isn't it - the new axiss
i´m having problems on my EMC
it was running OK
today i turned on the PC-CNC machine and EMC turned off
anyone can help?
very odd... how has EMC turned OFF
well, i shoosed the machine configuration (anyone) an then EMC goes out
are you hooked up to a machine???
i run EMC, choose a Machine conf, and EMC close itself
but i have my own machine conf
ok.. emc or emc2 ??
running on debian 4
any switches down... e-stops or such??
not hardware problem
have a back-up?? from when it worked??
EMC2 don´t work, XYZ screen don´t load
im an old tape guy...
last words on lots of screens are:
libnml/os_intf/_shm.c 257: No shared memory buffer exists for this key and the I PC_CREAT was not given.
wanna see dmesg? I can put on a pastebin
any storms in your area recently??
hm, PC crashed this week one time...
sounds like corruption....
what should i do ?
do you grab an image once in a while??
look, my install is not ubuntu
is you maschine config file still there and ok??
a have not fashio interface lol
yeah its still there
i have a very small install
even better you can dou your own thing then
my graph interf is only: XVESA, AEWM++ , XTERM and EMC2
im using putty now yes
cool.. you can fatten it.. LOL
i can put on pastebin all you need
have you got a puppy cd??
nope, i got DEbian 4 netinst
then installed openssh-server just to sendo files
installed xvesa and basic fonts
it worked great
why?? .. hardware???
but today its not opning
i´m from Brazil
where are you??
shite i'll packya a box a stuff.... ut i'll need a mailing addy...
huh?? Will ya send me what? lol !!!!
what are you running??
on machine? or on this PC ?
well im on XP now
spare parts... i have a storage unit of OLD machines..
but what dont work??
using putty to send commands to mey R.I.Ppped emc2 lol
EMC2 dont work dude
linux is up
i command: xinit; emc2 runs automatically
opans upto 2nd penguin
but then emc2 stops and crashes, X simply goes out, them text screen
i hate when that happens..... i feel for ya...the BRAINS in here might help ya
yeah i hate it too
the will.. wait for it... ;)
cradek or alex should be able to help...
If X crashes, sounds like it might be a video config issue
maybe, but i can see EMC screens , are you sure?
or, maybe not... i dunno
emc spalsh screen is just a single image
yeah a GIF
i will check the xorg.conf again
might try starting emc from shell and see what the log shows
thx Jymmm... ; )
emc from shell? heare on putty ?
pls show me how, nice aproach
i think its the problem
i mean, i´ts not
cos im seeing the VGZ config
* robin_sz bounces
i remember that it was working good
Please check my DMESG on:
[23:42:05] <Degas> http://pastebin.ca/667414
I will try again oto run X from putty and pastebin it too, one sec
* JymmmEMC duct tape's robin_sz (in a non-sexual way)
Stepper_mm runs OK
i will try to re-send the config files...
ok ... i'll believe that Jymmm..... need 911 robin_sz???