MDI: G10 L2 P1 X0Y0Z0
so g10 is implemented? I saw a place holder in the doc
what is L2?
jepler: your computer may be slow, but at least it boots
in EMC 2.3, there will be an AXIS menu item to zero a coordinate system, but I'm not sure of the specifics
jmkasunich: uh oh, you having troubles?
G10 is definitely implemented, what docs have a placeholder?
not with one of my regular PCs
the embedded board?
I dug a 433MHz celeron industrial PC out of a drawer, trying to test it
the cpu fan spins
thats about it so far
interesting. are you providing 12V, 5V or both?
the use of a KVM switch for this kind of testing probably isn't wise
but I'm lazy
actually, the board shouldn't care if there's a keyboard or monitor attached, but having it on a separate monitor is probably a better idea
the lithium battery is stone cold dead
the board might not care about kb/monitor
bummer. I haven't had good luck replacing those (when they're soldered on)
but the only way I'll know it lives is if the video works and I see bios stuff
emc2 user manual, section 17.3.2 "(add here)" under the title, setting coordinate system values within gcode
what version of EMC2 did that come with?
so does the g10 L2 param mean anything?
Guest735: have a look back at section 11.6 (page 75 in my pdf). That is the specification for G10.
looking on the web, I see G10 in section 11.6, and there's a description there
yes, I found the doc on the web, when I was looking earlier, I was passed 11.6 so when I searched I found the 17.3.2. I had to run out, just got back and tried it
oh, ok. "add here" is after the statement that the description is in the general programming section
Guest735: in a future version, G10 will be able to set other kinds of things -- different P-values will select the things
for instance, I think that in 2.3, "G10 P1" will set the tool table
but the L2 param?
in 2.2.8, there is only G10 P2
different L-values will select the things
for instance, in a future version "G10 L1" will set the tool table
like tool 20
now I get it
In G10 L2, the P-number selects which of the several fixture offsets will be set -- you'll usually use P1 (which corresponds to G54, which is the coordinate system usually in effect if not set by a G55..G59.3 code)
sorry I confused things by getting my Ps and Ls backwards
so backing up -- are you on a system with home switches or without?
what would be a good used model of an apple computer to run Ubuntu and EMC2 on? something cheap?
I guess it would need a parallel port.
heh. here's a great line in an ebay description: "This lot consist of Three (2) pieces considered new or unsued"
Guest735: I don't know of any Apple computer suitable for running emc.
it would need to be one with PCI slots and an Intel CPU, unless you want to do a lot of work
k, someone mentioned earlier they found a $50 apple that runs ubuntu and emc2
so the M2 command must reset the g92 offsets, correct?
Guest735: yes, and that behavior is one of the main reasons I recommend to use G54 offsets instead of G92.
(the other is that the axis "touch off" screen, which is very useful, only sets the G54 offset, never the G92 offset)
I think I have it figured out now. thanks, I had always used g92, will use g10 and/or touchoff
can you pass params to g54?
I'm not sure what you mean
I see this written -> g54 x0 y0 z0
without a g0,g1, etc
no, the x0 isn't "going to" g54, it's going to g0, the modal motion code in effect (set one line above)
I'd never write that myself
so g54 x0 y0 z0 is effectively g54 g0 x0 y0 z0
(exception: I'd write G53 G0 X0 y0 Z0 to move in the machine coordinate system; G53's not modal, so it applies only to the line it's on)
or if the line above was a 'g1', then it's equivalent to g54 g1 x0 y0 z0
one last bug, I am still getting "following" errors on my axis even though I am using servos. should I just increase the parameter in the ini? Can't remember which one.
servos with g320's so I guess they look like steppers to emc2
the loop is closed in the g320, no feedback to the PC
did you configure with stepconf or othrewise
both, stepconf first and then tweaked it, it seems like it showed up when I added .005 backlash to one axis
ohh ohh pick me
BigJohnT: oh are you going to take over for me?
that's great because I want to get away and have some dinner
no but http://www.linuxcnc.org/docview/html//common_Stepper_Diagnostics.html#r1_2_1
I'm skipping dinner so my new nick will be littleJohn
Guest735: ok -- you're almost certainly right that it started when you added backlash
Guest735: er, when you added backlash compensation
Guest735: look in your inifile and double the STEPGEN_MAXACCEL value
Guest735: do it for any axis where you also added a backlash value
I'll try that. that is the ini for axis or hal?
there's only one ini
it only gives me the error when jogging, usually when switching directions
stepconf sets up a maximum acceleration for the motion planner ([AXIS_0]MAX_ACCELERATION) and for the step generator ([AXIS_0]STEPGEN_MAXACCEL)
without backlash, it's enough for the step generator acceleration to be just a few percent above the motion planner acceleration
but with backlash correction, emc uses "extra acceleration" at a reversal to take up the backlash
that's above and beyond what's in [AXIS_0]MAX_ACCELERATION -- in fact, up to double
so you double the STEPGEN_MAXACCEL and all should be well
I'll go look at it
I'm going to go have that dinner now
I hope I didn't get too technical on you
so max was at 10, stepgen max was at 10.5, so I took stepgen max to 21, correct?
that's the way I understand it now Guest735
JustinXJS is now known as JustinXJS2
KimK_ is now known as KimK
Guest901 is now known as skunkworks_
canned cycles are very cool
I didn't know they are so flexible and usable
dear jesus what would i do without canned cycles
what are you do (produce) ?
heh... write a lot of redundant code
jensor: how is it going?
Well. I just discovered a phenomen, I added a loop on axis.ngc to repeat itself 1000 times and it made it run very jumpy
and doesn't follow the track very well
the machine doesn't follow, or the cone on the screen doesn't follow?
the cone on the screen
that's normal. for a huge program the display can get slower. the machine will still work the same.
you can put the magic comment (AXIS,hide) in the program after the first loop, and it will stop loading into the preview at that point.
The way its written I can't get into the first loop
I don't understand what you mean
you said add AXIS,hide
after the 1st loop
you could put it after line 416
I'll try that
how is the latency tests?
scripts/emc: line 537: 1721 Segmentation fault $HALCMD loadusr -Wn iocontrol $EMC2_BIN_DIR/$EMCIO -ini $INIFILE
Can I use source of emc126.96.36.199 with rtai modules in *.deb ??
cradek, it runs ok after addin g that command, but i do have to ok the g -code error that pops up
ya313: I believe so
eric_unterhause1: ok. then i need to install other system to be clearly sure i still have this error
you downloaded a source package?
skunkworks, still working on it. I decided to let axis repeat itself while waitng for the rtaip error to occur.
It seems that just sitting over nite it didn't occurt
but i get a mistake running "realtime" emc. I think i need to install everything agai, then compile python2.5, then only emc2... I will do it next 45 minutes
The idea is to hopefully have a log of what is going on when it occurs, to see what causes it. I have moved the baseperiod back to the original setting of 50 uS
and let emc run axis.ngc over and over
I understand that the rtaip error message only appears once per session even though it may occur later. This tends to mislead in that one thinks that it is not reoccuring.
If my kernal had been recompiled, to reduce the latency when emc was first set up (2-3 yrs ago) have anything todo with my problem.
if you had a problem, and you solved it by recompiling the RTAI kernel, and then you replaced that kernel with one that has teh original problem, then yes, it could have something to do with it
would updating em c to the newer version replace the kernal
I tend to think that it does
yes, if there's a newer kernel
it doesn't remove your custom kernel though
so you should still have a GRUB boot option for your custom kernel
how do I determine that
Hi guys, I purchased some servo drives, they can do step/dir or +/- 10V, but the can also do "Master Command Control (RS-232/485 and MotionLink Communications)" can EMC output MCC or MLC?
no, use the +/-10
My other servo drives loose some steps in step/dir mode if I drive the quickly, so serial communication sounds sweet
step/dir servos are an abomination
hehe, I do agree, and I have a pluto laying around here which I haven't played with yet
you need a analog servo/encoder interface: mesa and ppmc are two good ones
is that how you interface with +-10V?
jensor, look at /boot/grub/menu.lst
pluto can't do +/-10
I just bought two of these for a direct drive trunnion table I will build: http://cgi.ebay.com.au/ws/eBayISAPI.dll?ViewItem&ssPageName=STRK:MEWNX:IT&item=170292174954
mesa and pico have boards that have DACs on them to generate +/-10
cradek, no, but it can do pwm which can be converted to -+1-V
jensor, or reboot and press ESC for the grub menu
arg, to late to type fast
The_Ball: yeah, depends how much circuitry you are willing to build
I wouldn't consider running anything but a small machine with a pluto - I just don't trust it
I already cut some op-amp circuit boards to do the job, but lost the "puff" at the end over a year ago
I know that feeling
how much for a mesa or ppmc board that can do five axis?
$199 + some I/O cards
are they parallel port based?
PCI for the Mesa
they have parport too
pico are parallel port
oh right, mesa has a parport version now
(pci is better)
the IO cards are signal conditioners that convert from PWM+dir to +/-10V
parport costs less than half
and do differential conversion for encoders
you can probably do 8 servos with a 7u43, but you'll have very little I/O left over
I have pci
and you need 2 PPMC boards for the Pico solution
I'm responding to "parallel prot costs less than half"
I'm talking about mesa only
that's true, but it may not be ebough for a >4-axis system
SWPadnos, I'll look at the grub setup
problem with 5 axes is that the boards all seem to have 4, so you need two (8)
mesa doesn't have that problem IIRC
output boards, yes
jensor, ok. I didn't realize that you had made a custom kernel. if you can figure out why, please let us know :)
ok, there is an 8 servo version of the 7i43 firmware
wow, mesa have a lot of cards
but once you use 5 of them, you're left with only 18 I/Os for everything else
and a PITA wiring up the second connector to use them
a digital data stream equivalent to a dac seems like it would be a good thing, but I've never seen one
SWPadnos, only 18I/O's?? only! I'm running three axis now only using the parallel port and have plenty pins left
well, I think in Bridgeport and HNC terms these days :)
and most of my experience is on the Mazak
you must not have many home/limit switches or spindle/coolant control
or a pendant
hehe, then 18 would be 'only', hehe
can't live without a pendant
but it's true, 18 is more than the parallel port gives you, total
cradek, correct, only spindel and coolant control
SWPadnos, my pendant is a wireless rumble pad, haven't figured out how to use the 'force feedback' yet, hehe
I wired all my limit switches in series
going without home switches is just too painful
home switches will have an input each
you could share, but you can't home everything together then
I'd like to stay off my limits
the way bp set them up, they are right up against the stops, 3/8" or so away from metal to metal contact
Which mesa board should I get, cradek your saying PCI is the best way to go?
some people say...
in my opinion only
here are a few differences between PCI and parallel port:
they are more expensive but also seems like they are less trouble
well over the years you have had some good opinions in here
probably is better, less chance for an unknown 3rd party to screw things up
PCI: $200, parport, $90
you may need to use a pci parport too
SWPadnos, I just spend $1000 on those drives, can't stop now
PCI: fast I/O, could run the servo loop at 5-10 KHz. parport, 100 uS or longer update, so you shouldn't go much beyond 2-3 kHz
ah yes, pci it is, and are all these boards supported by EMC2?
PCI: more I/O, 3 or 4 connectors * 24 I/Os per (=72 or 96 IOs). parport, 2 connectors fro a total of 48 I/Os
which is plenty fast enough for most systems
in my opinion, the only major advantage to the parport card is price
SWPadnos, waht I see is 3 boot options: Ubuntu,kernal 2.6.15 magma, Ubuntu, kernal 2.6.15 magma (recovery mode), and Unbuntu, memtest86+
or for slotless machines
there is a minor advantage in that you can connect it externally to some PCs where PCI wouldn't work
jensor, ok, I'm not exactly sure what that means, but it looks like you have only one kernel installed
SWPadnos, Is the memtest86+ unique to mine
did you upgrade EMC2 by reinstalling from CD?
the add on boards start hiding the price advantage
ok, then your kernel probably hasn't been changed by the upgrade (or at all, by the look of it)
also, the fact that you need a very expensive parallel cable :)
don't price all the 50pin cables then
those are the same between PCI/Parport, unless you need more I/O than the 7i43 can provide
oh, wait :)
wow, these cards are amazing!! I can wire all the servo fault and enable pins individually, this opens up a new world
in which case PCI wins due to capability
The_Ball: didn't you have a video of your mill running around with a gamepad?
yes i did, hehe
oh interesting. our thermometer has an end stop at -30C
no wonder why it didn't move much last night
cradek, when you have hooked up your servos using +-10V what sort of feedback to you use for possitioning, quadrature from the drives?
resolvers on the ballscrews -> resolver to quadrature conversion -> mesa card -> emc2
The_Ball, are your servos DC or AC?
is the +-10 input a velocity command?
can you get quadrature output from the amps for emc?
cradek, all my servos are direct drive, so I guess using the quadrature output from the drives is ok. I see there is an acessory for the 5I20 card that is a RS-422/485 interface, would that alow the card to get the possition straight from the drive?
cradek, +-10V is a velocity command yes, the drives can output quadrature or step/dir to mimic the resolvers
I don't think you can use RS-4* for emc currently
velocity in, position quadrature out is perfect
ok, i was hoping it was just interal in the 5I20, but I have no idea :)
mesa 7i33 will hook up directly to that
in theory, don't you risk loosing steps with the quadrature feedback? That's the problem I'm fed up with right now (loosing steps)
Ok, so if I order a 5i23 and a 7i33 I'll be alright
for 4 servos
if you have an op-amp board all designed for the pluto, it should also work for the 5th servo you want (on the second connector)
differential quadrature is reliable but if you have bad wiring, grounding, etc, of course you can lose position. there is no magic bullet.
SWPadnos, actually I lie, my quill is driven by a gecko drive (step/dir only), i was planning to build a H-Bride for the pluto on that
oh. in that case it's easier :)
sweet, calling mesa now :D
there's a ervo+stepper FPGA firmware, so you can use a stepper on the one axis and servos on the others
mesa can do step/dir too
The_Ball: if you get the "T" (screw connector) versions of the interface cards, don't forget to ask for the connectors too
oh, and (I shouldn't say this until I've bought more, but) ebay is a good source for opto-22 I/O boards and modules
pretty close to the cost of a 7i37T+connectors, but more flexible
there's a 1M ballscrew going on ebay that would be great for the new lathe, but it has a 30mm pitch :(
that seems a bit coarse
whats the diameter
about a foot
sounds a biggun
Total lenght=1130mm Travel=940mm Screw Pitch=30mm(High speed) Diameter=20mm Accuracy grade=C7
What is C7, any good?
it's slightly better than C6
or slightly worse :)
ah, now it's clear as mud
excellent. my work is done
i'm guessing it's a little worse than C6... "There is a little rust on the screw"
C7 is down amongst the rolled screws
ah, that explains the cheap price
seems C3, C5 are better than C7. I haven't spotted even numbers yet.. ttp://www.misumiusa.com/CategoryImages/Metric_2006_pdf/METRIC307-308.pdf
(see lower left side of page)
[17:34:52] <archivist> http://www.nookindustries.com/profilerail/ProfileRailGlossary.cfm#Accuracy
archivist: ah that's better than what I found
C7 is a sod to google
that mainly relates to rails but I assume the numbers cover more as ballscrews are in the same table
[17:42:35] <skunkworks_> http://www.youtube.com/watch?v=qY-FCN5ZXkg&feature=channel_page
skunkworks, sweet, yours?
a h-bridge I have been pulling my hair out with http://www.electronicsam.com/images/KandT/servostart/schem/newcurrentlimit/close.jpg
The_Ball: do you have a link to that video?
skunkworks, nice, what's your double sided alignment trick?
2 work coordinates.
skunkworks, i'm trying to find the video, while you wait, the third video here might amuse you: http://wigen.net/hobby/f-16/
skunkworks, hahaha, here's the video, makes me embarresed to watch it now: http://wigen.net/workshop/cnc/jogspeed.xvid
The_Ball: would you have a problem if I put it up on youtube?
skunkworks, no not really, people will watch anything these days
hehe, i need to get some ballscrews on there
with this mesa board that i have now ordered and with the trunnion table and spiral machining, that bed will see some serious work
The_Ball: is that you flying the jet?
thanks, good fun
Is that scratch built?
heh - I could just read the website. ;)
If my max step requirement (using quadrature) is 4167 steps/sec what would be a recomended base period?
no, don't read, then I can pretend it's scratch built
jansor - quadrature gives you one transision per base period. (very nice) so lets say 5000steps per second 1/5000 is .0002 or a base period of 200000. did I do that right?
yes, assuming you're talking about outputting quadrature
for input you should have a faster period because of jitter
also note that the longer the base period, the worse the timing resolution from step to step (it's always an integer number of periods between steps)
So as the timing resolution gets worse, the greater the velocity error between what is specified and what one gets?
time for bed, it's only five in the morning
stustev: can I pick your brain for a minute over on #emc-devel?
stustev: specifically, I'd like you to take a look at http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?WrappedRotaryAxes
and point out what I got wrong
jepler: sure - will do (or at least try) :)
i think you should allow it to wrap up as many times as needed.. WITHout having to unwrap when the machine is referanced again
will be back in a minute
jepler I must watch as Im interested (hobbing)
jepler: would you give me the link again
[20:05:31] <jepler> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?WrappedRotaryAxes
what do you mean by "enable relative rotary motion" I can see a relation between an X or Y not just another rotary
i will see if i still have the wrapping macro for the Toshiba
again, sim run in place - works, realtime run in place - doesn't work
ya313: what does uname -a tell you?
only one, № 15019 mistake
"uname -a" ?? in which folder? b-t-w, i can not attach internet in rtai mode
ya313: typing "uname -a" will tell you the kernel name
to make sure it has rtai extensions
alex_joni: ok, please, wait an answer withing few minutes after my logging out
but if you're sure rtai testsuite works ,then we can go past this
did you try the rtai testsuite?
testsuite? b-t-w, "dmesg" shows "unregister rtai" after the faulty launch of simulation mode in real-time emc2 compilation
ya313: sorry.. gotta run for 30-40 minutes, but there are others here, just post more details on the error you get
alex_joni: no, and where's its documentation?
alex_joni: ok, but how can i run its test suit?
ya313: you go to the folder where you compiled rtai (usually in /usr/realtime..)
"man rtai" doesn't help
there should be a folder testsuite/latency/kern
there you have a .run if my memory serves me right
alex_joni: i'd used .deb -files from linuxcnc.org
ah, then it should be ok
can you pastebin the exact error from emc2?
ok, will see. i'm about to quit and restart pc
just a second
did you install using the install script? or from the LiveCD?
ya313: if you used the install script (or you installed packages by hand), then you should install a couple more packages
alex_joni: well, script, then some packets from synaptic, then manual compilation of python 2.5.2 and emc188.8.131.52
alex_joni: yes, but --sim + --in-place already works
RTAI doesn't let run axis in simulator mode
ya313: install "linux-ubuntu-modules-hardy-rtai" and "linux-restricted-modules-2.6.24-16-rtai"
ok, wait a second, i can check if it are already installed
those might make your network work with the RT kernel
ya313: ok, gotta run now.. hope you can pastebin the error message emc2 generates during startup
then someone in here surely will help debug this
yes, i can install it.
ok, i'm restarting pc
good evening. i want traslate the error message with axis interface.what is the module for text the error in emc2.2.8?
motioncontrol: which language?
a little bit of information here: http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Internationalization
ok, where to paste a mistake? e.g. paste.org?
we like pastebin.ca
it doesn't "work" in my browser
use any you like - doesn't matter
[21:04:55] <ya314> http://paste.org/4864
read this to kids at the evening if they've got a bad grade in school ;)
well, it's on a laptop
RTAI or RTLinux. Hope, both let commands to be sent thru the network interface?
RTAI and RTLinux are mostly irrelevant to the question of sending commands over a network
and what about the fact simulation mode doesn't run in copmpiled 'in-place' emc2. option --enable-simulator let it work. But realtimeprocesses fail. Should i use emc2 from .deb -files? I'm compiling it after the python 2.5.2, without the last one emc2 doesn't run, too
that's why i think i just need to use debian emc2, not a tarball since i'm using RTAI modules as .deb
run-in-place and enable-simulator are independent options
if you want a simulator build that is also run-in-place, you need to specify both
ok, then it doesn't work as a realtime process
simulator is not realtime by definition
well, it doesn't work as a simulator in RTAI -kernel
but it doesn't even run in that kernel-RTAI... Is it right/wrong behavoiur?
I don't understand your description of the problem
[21:20:52] <ya314> http://paste.org/4864
, but i'm trying to run the "sim-axis" module
that looks like a non-simulator build
err, no wait
yes, it's non-sim
i want traslate in italian language.
...but... i still must be able to run the sim mode, axis GUI, right?
you don't need the make setuid step for simulator, and the make process shouldn't tell you to do it
we already have an italian translation - if it is incomplete you could help improve it
axis needs to communicate with other modules that are RT for a non-sim build
so you need to reconfigure: ./configure --enable-run-in-place --enable-simulator
and make again
i want only modification and customization the my emc versio.
SWPadnos: ok, but i wanted to run sim module of the EMC2.
SWPadnos: he wants to run RT, but the sim config
ya314, you will also need to source the emc-environment script: from the src directory: . ../scripts/emc-environment
ya314: _please_ try to be carefull what you mean/write
SWPadnos: then all i need is to adjust the settings of particular machine type and start to work! Right?
I sure hope so
depends what you want to do
i see the txt in emc2.2.8/src/PO IS THIS THE FILE FOR TEXT?
motioncontrol: it's a bit offtopic, but i can give you free CAM Expert license if you would have verified its USER Reference's translation
good night all
goodnight alex (lucky)
heh. see you Alex
does anyone knows the steps of work in Code-Aster.org?
never heard of them
well, a free FEM suit, which seems close to ANSYS's capabilities.
the emc when start open the window for select the driver(m5i20,pluto,sim ecc.), is possible select the configuration in one file and not open the windows select driver and open when emc start only the axis interface?
motioncontrol: like --myconf='PATH_TO_FILE'? Could be...
[21:33:03] <SWPadnos> http://www.caelinux.com/CMS/
SWPadnos: oh, yeah, i'm just out of my internet money this month to search too much sites.
yes , i want load the configuration driver in one file(i thing .ini file) and not open the window for select driver.
that's a distribution that's target at FEA and that kind of thing
SWPadnos: the main point of this all is usually a fatigue of metal, esp aluminum => right borders' curves of particular details
sure, modeling before building and breaking things :)
also, impact.sourceforge.net can handle a crush test of a car, coefficients of strains of sheet-metals
SWPadnos: i'm modelling 2D for simple details, then using brlcad.org suite to dxf-g and g-'some_format' it. then loads and so on. Worked in ANSYS (Ed license).
yam314 is possible you can the example for the load the my configuration in one file and not open the window select driver when emc start?
motioncontrol, there are two ways to make EMC just start up:
actually sort of 3
all this must take 10-15 minutes, i mean most of 2D details. When we're talking about a plant or a firm, 40 working hours a week
you can make a desktop icon for the selected config, and just run EMC from that
motioncontrol: emc ~/emc2/configs/sim/axis.ini
from a terminal, you can run "emc -l", which will run the last selected configuration
or that one, specify the ini file directly
motioncontrol: no, NO, I've just imagined what are you trying to say
ya314 i don't understand. if i write in one termila ./emc -1 the window selsct driver not start?
motioncontrol: "emc -l", not "emc -1". if you can't read the difference between "l" and "1" (ell and one) then use a different font.
in your irc client
ok thanks jepler
ok, quitting. Will try it tomorrow
quick question...what is the proper way to setup a slave on the X axis in emc, i just set P8 and P9 to match P2 and P3 but it wont jog both motors at the same time