I think so. But it might be harder than that. Is that for the gcode we are interpreting? Does it include the .ini files?
[00:00:16] <cradek> http://cvs.linuxcnc.org/cgi-bin/cvsweb.cgi/emc2/src/emc/rs274ngc/interp_read.cc.diff?r1=1.10;r2=1.11
it's for a different purpose, not for changing the way g-code files are parsed
this is the change that broke it (using strtod)
(That might be a pain.)
[00:02:05] <cradek> http://www.isd.mel.nist.gov/personnel/kramer/pubs/RS274NGC_3.web/RS274NGC_311a.html#999272
this shows that . is to be used for the decimal point in rs274ngc
...so currently it doesn't conform to the docs with certain locale settings
right, it's just a bug
Whew. I don't like the 47 < c or the c < 58. S/B '/' < c ; c < ':' or better yet -- '0' <= c; c<='9'.
lerman: yeah, it was hideous - probably why jepler changed it
But that's the code with the '-'; so it is gone (I hope).
but he surely didn't think about commas
Yup. The language definition says a decimal point is a '.'. But heck, we've messed with the language a lot (well, I have), one more change shouldn't be a big deal.
If the users want the language to allow comma in the gcode where decimal point is, I'll add something to the .ini to allow that it the gcode.
Making the .ini file allow that might be harder (I haven't looked at the code to read it yet -- nor do I plan to).
Changing the gui(s) is a separate issue that is out of scope (for me).
I think allowing comma, whether or not it's optional, is unnecessary
Is that a higher or lower priority than internationalization of the error messages? :-)
they already are - but only AXIS uses them
(I just tried today)
(If god had meant for us to speak all of those funny languages he would have given them nuclear weapons). :-)
[00:10:51] <cradek> http://timeguy.com/cradek-files/emc/german-error.png
hmm, looks like we are missing a few strings in de
I thought it was pretty complete...
As long as the framework is there, the strings will be translated.
That will take care of the canned messages. I've got to do something in the interp, though, to handle the new variable messages I've added.
(Shit. Did I say 'I'. I should have said 'someone').
But it looks like it isn't a big deal.
I assume that today the interp gives english errors and the gui translates them - but it sounds like you've made that scheme not work anymore
haha I try to remember to say "someone" too
No. Yesterday, the interp gave numeric errors with a unique number for each FIXED string. Today it has a number meaning VARIABLE string and
ah, numbers, the international language
gives a string which might have embedded values. So, today we can say 'multiple definition of o200'. Before we had to say 'multiple definition of oword' and let the user figure out which one.
(That's not a good example because the line number is available to see which line it was on).
"on line 200: redefinition of O word"
yep was just thinking that
I mentioned line numbers in my recent email
But 'undefined named parameter' would be a good example. 'undefined named parameter #<foobar>'
There could be multiple usages of named parameter on a line.
Even before named parameters, there could be numeric parameters out of range ##1 is legal.
There is no way for you to find out that #1 contained the value 12345 and that #12345 is out of range. (at least there wasn't in the past).
Today I could add a message saying that #12345 is out of range on line...
NIST was working with a very limited context... slow machines limited memory... etc. That excuses some of the limitations in the implementaiton.
so you are going to have to get the interp to do the translation of those printf format strings now right?
Sure. I would just call the proper function to translate the format string from english to the current language. There is a utility to scan the code and find all of the format strings.
Then each of the strings has to be translated.
I've never written the code, but it doesn't look too bad.
The translation routines apparently take and english (source language) string as a key to look up the target and return it.
yeah I bet finding the dictionaries at runtime, installing them properly, etc, is the hard part
I don't know much (anything) about i18n
I think not. It uses the locale. There is an init call that tells where they are. And now that I know how to read a .ini file, we can just put the path to those files into the .ini
no, paths do not belong there. configure/make install need to set the paths correctly.
I've "ported" g_ascii_strtod from glib 2.12.4 to plain C: http://pastebin.ca/349719
the ini is for configurable options, NOT options that have one correct setting that the program should know
maybe too much code, feel free to throw it away :)
I was wondering if you were going to call me on that. :-)
I realized after I said that -- I am wrong.
Probably an environment variable set by the script.
erDiZz: wow! that is a lot of code
Depending on where its installed.
lerman: yeah (without looking) I bet so. or maybe it's compiled in.
man gettext -- does the translation.
yeah, that's a lot... even too much to read it and check for validity
Compile in the relative path. .. That would be OK. But you still need something to give the abs. path.
locale_data = localeconv ();
decimal_point = locale_data->decimal_point;
decimal_point_len = strlen (decimal_point);
wow, decimal points can be multiple chars
that's ... amazing
were they one char, the'd be represented as a "char"
a multi-byte character maybe
gotta run - bbl
cradek: just saw your email. Let's take a look at where the interp is giving the wrong line number. If you can make some sample programs, I'll take a look at them.
The easiest thing would be to create a subdir under nc_files called tests. Then generate test programs with errors containing comments with the proper line number.
(Easiest for me, I meant) :-)
That way I would have tests that I could repeat. I'll run them one at a time and fix things.
(I'm not sure that there is no need for traceback -- but I agree that there is a need to have error messages identify the correct line).
Another interest I had is trying to get MDI to be able to call subroutine. I think it would neat to be able to mill a surface by MDI ing.
o_surface call  (where 1,2,3,4 are lower left and upper right corners of surface)
In principle, that should be easy to do, but my attempts have all failed due to my lack of real understanding of the interface between the interp and the rest of the world.
I'm trying to interface it "to the rest of _my_ world" right now :)
I'd say the interface is of a retro kind
say, I can't have several instances of the interpreter in one process
(at the same time)
Yes (in principle). I'd be surprised if it worked.
and is it possible to have a different program source than a file?
When there is an underlying assumption that an object is a singleton, the coders sometimes get sloppy and use statics when they shouldn't.
in MDI mode, as I understand, it treats errors differently (and maybe something else differs)
Then things break if you have more than one.
say, I'd like to pump a huuge program over a network connection directly into the interpreter
That will NOT work. (Perhaps it used to, but not any more). The interpreter must have an input source that is seekable.
When it loops or calls a subroutine, it saves the current position and then seeks back to it as necessary.
ok, good to know
But we could make it work. Tell the interpreter to copy its input to a file as it reads it. Then when it seeks, just loop back to the file.
If it is reading past the end of the file, use the actual source.
Right now, the seeks and tells are in a variety of places. But most of it (maybe all) is in the file interp_o_word.cc
I've imagined some special kind of a file, when it doesn't return EOF until explicitly told so
Sure. A fifo or a pipe will do that.
but they' re not seekable
Even a serial terminal will do that. (Correct. Not seekable.)
But any file could look like that. I'm not sure what happens if you try to read off the end of a file in non-blocking mode.
We can make it work if we need to.
well, this setup would do: a proxy character device which redirects all read requests to a user-space daemon, which receives the program over the network, writes it to a temporary file, and waits for the data if the request is beyond what's already received
simple enough, and no changes to EMC at all
Works for me.
how do i search for user groups on freenode ( looking for which linux to try on old laptop toshiba 730cdt pentium 150 )
guess I saw a web site with irc channel search
[00:58:16] <erDiZz> http://irc.netsplit.de/channels/?net=freenode
thank you :)
woo :) http://mync.sourceforge.net/gncedit.png
Argh! I just tried to use the scroll buttons on a screen capture. Again.
erDiZz: Just realized - gncedit shows you the tool path as you write the code. Cool!
it is supposed to. Currently that "refresh" button has to be pressed each time
that's a matter of a couple of lines though
well, I'm off
ooh an original pentium?
what distro does BDI use now?
well, BDI uses debian I think..
though the livecd is ubuntu
BDI is its own distro but it uses mostly debian packages
the emc2 livecd has nothing to do with BDI
however at one point, there was a BDI-Live
seems like It was Knoppix at one time? <shrug>
yeah, BDI-Live was knoppix
been a while
before that it was RH7 (BDI-TNG) and before that it was RH6 (BDI-2.xx)
I have bad memories of rh6...
6.2 was the last time I used it
bye guys, I'll be back on friday
alex_joni; back on friday?
I wonder where he's off to
say I want to perform linear motion in XY plane, and move by an arc in UV plane at the same time, like "G1 X.. Y.. G102 X.. Y.." in a single frame
any way of doing this with EMC? :|
...or move by two different arcs in two planes, to be more general
XY and UV are parallel
If I understand correctly, there was a "VGER" version of the interpreter
but EMC comes with NGC version
[13:09:43] <erDiZz> http://www.isd.mel.nist.gov/documents/kramer/RS274VGER_11.pdf
...by T. Kramer
erit's a limited subset of the normal interpreter..
[13:24:53] <anonimasu> http://imagebin.org/7252
anonimasu; don't tell me that you did it manually
lerneaen_hydra: haha, no I didnt
anonimasu; how'd you do it? a plugin?
erDiZz: that interpreter is a limited rs274 interpreter
erDiZz: I dont know why you would want to run it or what ir provides over the standard interpreter
I do have an interpreter that parses G1xx functions, and there is appropriate code for trajectory planning
I'll just keep it for certain kind of machines
why dosent the normal interpreter work?
cus it doesn't do to different arcs simultaneously
isnt that a trajectory planner limit+
I don't know
all I need is an interpreter
I use my own trajectory planner
heh, gotta love a standard temp-check i2c chip that powers down the system when probed :/
sounds like a smart design moce
03jepler 07TRUNK * 10emc2/src/emc/motion/command.c: "linear move in MDI out of range" is a much more helpful error message than the old one.
ooh, very exciting news posting on tuxcnc.org ... no news since november and then some rivettign read about err, well, somethig to do with applying patches for something to do with an AMD64 and err, well, whatever, I'm sure its really important
03cradek 07TRUNK * 10emc2/src/emc/motion/command.c: make a few of these errors clearer still
03cradek 07v2_1_branch * 10emc2/src/emc/motion/command.c: backport better error messages
good morning all. question? how can i increase the number of samples that halscope captures?
change a variable and recompile?
SWPadnos: there's a bit of interest in the apt programming system, and i added a link to it in the emc wiki. I gather that this is your's and Steve Stalling's host, and dont want to fill it up with stuff thats off topic ( tho it makes gcode for emc ). I'd like you to let me/us know your thoughts. I been prepping a couple tutorials and wondered about loading them onto the wiki. Thanks
after i run halcmd and get the parallel port to blink leds and read buttons (ala owhite's experiments) i try to clean up so i can run emc, but have no luck. how do i clean up after creating pins and threads, and functions? (scripts/realtime stop --> ERROR: Module rtapi is in use by threads,hal_parport,hal_lib etc )
halcmd unload all
Roguish: I believe there is an arg you can pass when loading the RT part of halscope
* jmkasunich looks
i've been getting used to halscope. getting better pictures now. but think i have discovered a BUG in emc2.1.0
loadrt scope_rt num_samples=<somenumber>
the default is 16000
whats the bug?
taking notes again.
well. kinda interesting. let me attempt to explain what's happening.
got my system tuned much better. f-error is -5 -6 or so now. good. except when i go above 35 ipm. then the f-error just increases.
background: steppers, servos? what hardware device (if any), or is it software stepping?
really increases.. very fast to really big numbers. i have tried different tunings, different accel and vel maxes.
servos. m5i20 with yaskawa ac drives and matching motors.
pretty stock emc2.1.0
analog to the drives?
yes. i am using a mesa 7i33 out of the 5i20
are the drives running in velocity mode? (ie: 0V tells the drive "stand still", +10V tells the drive "go right at XXX inches/min"?
or torque mode = 0V tells the drives "make zero torque", and +10V says "make full torque"?
yes. the drives and motors are tuned seperately with 'simawin' a yaskawa gui setup program.
tomp listens cuz hes got exactly same hdwr mesa m5i20 7i33 yaskawa motors and amps
so, velocity or torque mode?
and what is the velocity scaling?
10V tells the amp to go XXX inches/min... what is XXX?
+-10 v = 4500 rmp
which is how many ipm?
scaling : 416153.6
( thats the sigmawin viewof things, not the emc view )
i have a 10:1 gear head
thats counts per inch?
that numbers seems crazy high, lets go thru the math
yes. counts per inch. it is scaled pretty good. only off about .016 in 1.00 inches by dial indicator
5mm per rev.
eh only off?
0.016 in an inch is not good
Roguish:thta's a shitload..
especially with servos..
0.00016 might be considered good
that's lots also
ok, we have screw pitch (5mm) and gear ratio (10:1)
we need counts per rev
what is the encoder resolution?
(are you the one with resolvers that are converted to simulated encoders by the drives?)
and your machine units (ini file units) are inches?
inches all the way
ok, 5mm screws means 25.4 / 5.0 = 5.1 turns per inch
the gearbox means 5.1 * 10 = 51 motor turns per inch
( the yaskawa can scale what it reports to outside world, it has a 'pass thru encoder, but can scale the output, so the specific resolution may be misleading, butting out now)
the encoder means 51 * 2048 = 104448 counts per inch
please don't leave yet. i'm not proud....
if the 2048 is actually cycles per rev, not counts per rev, throw another x4 in there
104448 x 4 = 417792
that is the proper scale value I believe
which is 1.00393 times the value you have
that would account for 0.00393 inches per inch error
dunno where the rest is coming from
how did you get the number you used?>
i cant recall the math at present,
ok, don't worry about it for now
your value is in the right ballpark anyway
but why would the f-error go crazy at a particular feedrate?
nearly a half-million counts per inch is a lot but it seems to be correct
not sure yet, I'm working my way throught the troubleshooting process
that's the gearhead effect.
3how large are your servos?
ok, lets figure max speed....
we have 51 motor revs per inch, and 4500 rpm max motor speed
4500/51 = 88.2 ipm max speed
200 w motors with 400w drives on a small, new test x-y stage with 10' x 10' travel.
10" x 10"
88ipm on a 10 foot machine would suck
last machine i made was 30' x 12' x 8 '
880 woulf be sane..:)
ok, do you know for a fact that the motor/drive combo will actually do 4500 RPM when commanded with 10V?
that is the max speed i figured also.
35 ipm * 51 revs/in = 1785 rpm
if something is making the motors top out at that speed, then you'll get following errors if you try to go faster
yes. but i am depending on the sigmawin program to tell me so. i will jog the motor to 4500 and they really scream.
not to jmk specifically
jog in sigmawin can go faster than limits prescribed in other params, which are in effect at runtime
the break point for the f-error to start increasing is about 35 ipm
how difficult is it to decouple the motors from the screws?
either before or after the gearbox, doesn't matter
not too hard after the gearhead.
(I want to be able to spin the motors without running the table into a stop)
we can use halcmd interactively to send whatever voltage we want to the motors
and verify the resulting speed
are you able to do things now? or is the machine somewhere else?
ok, got a motor off.
ok, I'm starting the m5i20 config of emc here so I can walk you thru things
go ahead and start yours
and open another shell
which axis did you disconnect?
ok, start halcmd with -kf
if you are running an installed version, its just "halcmd -kf"
if you have run-in-place its a little different
type "show sig Y"
it should list about 7-8 items
Yenable Yfault Yhome, etc
this one is the connection from the PID loop to the DAC:
float 0 Youtput
we're gonna disconnect the dac so we can drive it manually
don't type the " "
I'm using that do distinguish commands from me talking
if you do "show sig Y" again, the Youtput signal should look differnet now
float 0 Youtput
the pid output is still driving the signal, but its not driving the dac
now we have to fake EMC out so it won't throw a following error when we move the axis
that disconnects emc's position feedback
"linkps axis.1.motor-pos-fb Ypos-cmd"
that connects the feedback to the command
"show sig Y" should now include this:
float 0 Ypos-cmd
now you can bring EMC out of estop
and do "machine on"
if you jog Y in emc, the display should change, but the motor should do nothing
(we've disconnected the motor from EMC, and we are lying to EMC and telling it the motor is doing exactly what EMC tells it to do)
ok, now we can make Y move manually
"setp m5i20.0.dac-01-value 0.1"
we're seperating emc from the drve to see where the problem might lie.
did that setp command make the motor start to move?
that command should be putting 0.1 volts on the DAC output (I think)
depends on dac scaling
so the motor shouldn't be going that fast
0.1v out of a 10v full scale is only 1%, or 45 rpm at the motor
4.5 rpm at the gearbox output
does it look like thats what its doing?
yes, pretty slow
can you verify with any accuracy the actual speed?
count how many times the coupling setscrew goes by in a minute, etc
random question: are most spindle controls linear with voltage/rpm, or exponential? IMO it sounds more sane to have an exp. control
4.5 rpm = one rev every 13.33 seconds
though more difficult
lerneaen_hydra: I suspect most are linear
more like 30 secs
ok, that means the motor is running only about 44% of the commanded speed
and that means it will top out around 2000 rpm, not 4500
"setp m5i20.0.dac-01-value 1"
that should make it go 10 times faster
it should be 450 rpm at the motor, and 45 at the screw
ok, i gave a dac value of 0 and i show .001 v on a meter.
but I bet you'll see 200 at the motor and 20 at the screw
a millivolt of offset isn't terribly bad
it can be corrected for if needed (the dac has an offset parameter) but its not the cause of the problem we're seeing
dac =1.0 meter shows .386
"show param m5i20.0.dac-01"
Owner Type Dir Value Name
07 float RW 1 m5i20.0.dac-01-gain
07 bit RW TRUE m5i20.0.dac-01-interlaced
07 float RW 0 m5i20.0.dac-01-offset
you get something like this?
yes, but gain=.3937
ok, thats why you're not getting the expected voltage
something in your config must be setting that
ya, that's the output scale.
mm/inch .3937 is 1cm
Roguish: wrong is in the eye of the beholder - there are different ways to look at scaling
but you need to pick one and stick with it
I think you might have a mix of two different approaches
(which isn't surprising, because we don't document that well at all)
you managed to get the PID tuned pretty well with the output scale set as it is, right?
i'm trying to read all the current docs as i am a firm believer in RTFM.
yes, i think it's tuned pretty well.
well, tuning and scaling for servos isn't in the manual, so its not your fault ;-)
"show param pid.1"
that should give you a fairly long list
with all your gain values
"show param pid." will give you an even longer list (all three axis, not just Y)
can you pastebin the result of "show param pid" ?
sure, give me the link to pastbin
oro? alex_joni is not here??
[19:10:38] <jmkasunich> http://pastebin.ca/
a-l-p-h-a: alex is traveling I think
[19:12:17] <Roguish> http://pastebin.ca/350816
yeah he said he was going to be back friday
got any ideas for a valentines gift?
not much, coding some java stuff
hmm, serious or one-off (the receiver)
IGain of 1 million looks high to me
SWPadnos: it has a very wide range, up to 2 mil, down to 100k.
is that the range you've experimented with?
P goes unstable at about 11000.
I suspect that most people have no gains over ~1000 or so
yes, i've tried several tuning approaches.
11000 is huge
SWPadnos? got a server with java running anywhere?
a-l-p-h-a, only dreamhost
i have found it works best to keep the drive/motor tuning fairly soft. and increas the emc tuning.
it's unusually difficult to tune a PID inside another PID
yes it is
which is what you're doing
took awhile to figure out.
SWPadnos, so no java there...
a-l-p-h-a, dunno. there may be. other than some local workstations, those are the only "servers" I have access to
okie doke... was looking for a place to dump a whiteboard.
my hosting doesn't have java
you may want to choose torque/current mode to remove the loop inside a loop ( both yaskawa and emc are trying to control velocity ) ( thus the softening trick works when both are velocity )
maybe not until solaris is up and running there.
a-l-p-h-a, what's the whiteboard app? (is it generally availale, or did you write it?)
fsck - got disconnected
Roguish: you still there?
did you get the pastebin address?
[19:29:50] <SWPadnos> http://pastebin.ca/350816
I mentioned that the I gain looks very high
ok, you have two choices
you also missed the fact that there's a PID loop in the drive thats tuned "soft" ...
it does seem high, but I'm no tuning expert - if it works it works
anyway, lets deal with the scaling issue first
the easy way is to change [AXIS_N]MAX_OUTPUT in the ini file to 25.4
that will allow the PID output to go to 25.4, and the OUTPUT_SCALE of 0.393 will knock that down to 10V
but that means that the PID output is in arbitrary units
the (not much) harder way is to change OUTPUT_SCALE to 1.0
that way, the PID output is in volts, and when it goes to 10 (the current limit) the DAC will put out 10V
the reason its harder is that changing output scale also changes the gains
you will need to multiply your existing gains by 0.393
dgain, igain, and pgain
like I said, its not much harder
as it is right now, the PID output is limited to 10 units by MAX_OUTPUT, and the OUTPUT_SCALE of 0.393 means that 10 units = 3.93 volts
which yields 1768.5 rpm, or 35 IPM ...
so the DAC will never put out more than 3.93 volts, the motors will never go faster than about 1800 rpm, and therefore you get following errors if you try to go faster
won't EMC complain if requested speed > maximum speed?
not even a graphical thingy in axis?
the only maximum speed it knows about is the speeds in the ini file
it's up to the integrator to tell emc what the maximum speed is
it doesn't know about motor limitations, dac scaling, etc
its the integrator's job to figure that stuff out and tell EMC in the ini file
i would like MAX_OUTPUT=10 to give 10v.
output scale is not well documented, and in this case, using something other than 1.0 is what caused the problem
jmkasunich; the max speed I meant was the maxvel in the ini
Roguish: I think thats the right way to go
so set OUTPUT_SCALE to 1.0
if the speed in the gcode is > maxvel it should issue an error, right?
and multiply your PID gains by 0.393
lerneaen_hydra: I dunno if it issues an error, or just runs at the lower speed
can do, will try immediately.
well, it doesn't result in a following error at least
it just limits the requests to the ini maxvel
no, it doesn't
but that's a different problem
(result in a following error)
the problem in this case is that the ini maxvel doesn't accurately describe the actual maximum attainable from the motor (+ driver, DAC, ...)
dang - there is an omission in the G202 datasheet
the actual maxvel could be less if cutting or something
it shows the overall dims, but not the mounting hole pattern dims
is that one of the ones with the built-in capacitor?
or are those the G21x?
I thought he left the hole pattern the same, but I could definitely be wrong about that
I can measure the dims, but I was hoping to get the actual numbers from the datasheet
it probably is the same
but do any of the other datasheets spec the pattern?
[19:42:22] <lerneaen_hydra> http://loadingreadyrun.com/videos/view/49/How+to+Talk+like+a+Pirate
I think they do - lemme check
the 201 does
yep - 2-3/8 x 1-3/4
jmkasunich: you are a HERO!!!!!!!!!!!!!!!!
glad I could help
halcmd in interactive mode is usefull for digging into problems
i know you guys a bustin butt coding and developing and all, but please, please don't forget to do some documenting.
ONCE, AGAIN, THANKS TO ALL.
Roguish: the default value of OUTPUT_SCALE in that config was 1.0
SWPadnos, jmkasunich, tomp, et. al.
if there was some doc that prompted you to change it, point at it and I'll try to fix it
oh sure - list me first when aksing or documentation :)
no that was for thanks.
yeah - that too :)
don't axe me fer it
you guys making fun of my typing?
no, SWP's typing
yeah -mine. it's all mine :)
jmkasunich: where should i put the loadrt to increase halscope samples? i don't see it loaded in any of the config .hal files.
you can add a HALCMD line to the [HAL] section of the ini file
Roguish: turns out its easier than that
no loadrt needed
you can just add the number of samples to the halscope command line
when you start halscope:
i start halscope from inside axis, but i can do it via command line.
would give you 30000 samples instead of the default 16000
if you are doing it from within axis, then the loadrt approach is the way to go
background (if you care):
the realtime part of halscope that does the sampling is called scope_rt
what would the exact command in the ini file be?
you can load it with "loadrt scope_rt num_samples=blah"
once its loaded, when you start the user space portion, it will use the number of samples reserved by the realtime part
did that, didn't seem to work.
if the realtime part isn't running when you start the user space portion, then the user part startrs it
in the [HAL| section of the ini file...
# list of halcmd commands to execute
# commands are executed in order, HALCMD1 to HALCMD<HALCMDMAX>
#HALCMD = save neta
yes along with 3 HALFILE lines
add HALCMD = loadrt scope_rt num_samples=30000 (or whatever you want)
emc crashes with that command.
fails to start.
works fine here
"HALCMD = loadrt scope_rt num_samples=30000"
you added that exactly, to your ini file, right after the line that reads:
#HALCMD = save neta
set it to 32000
what error message did you get?
works here with 32000 too
HAL:0: ERROR systemv failed, returned 1
HAL:0: ERROR: insmod failed, returned -1
ok, need more info
do "sudo dmesg -c"
then try again
and pastebin all output from emc, as well as the output from "dmesg"
(run dmesg after you try, and pastebin that output, not the dmesg -c output)
you want the full debug info?
[20:25:45] <Roguish> http://pastebin.ca/350874
ok, I lied
the ability to change the number of samples is a post-2.1 feature
(I was testing here with a CVS version)
jmkasunich, i gotta run. thanks for the help. i can use the terminal halscope.
for some reason I thought that was in 2.1, but its not
good future feature. i like it.
03jmkasunich 07TRUNK * 10emc2/configs/demo_mazak/demo_mazak.ini: remove out-of-date comments
03jmkasunich 07TRUNK * 10emc2/configs/m5i20/m5i20.ini: remove out-of-date comments
03jmkasunich 07TRUNK * 10emc2/configs/motenc/motenc.ini: remove out-of-date comments
03jmkasunich 07v2_1_branch * 10emc2/configs/demo_mazak/demo_mazak.ini: backport: remove out-of-date comments
03jmkasunich 07v2_1_branch * 10emc2/configs/m5i20/m5i20.ini: backport: remove out-of-date comments
03jmkasunich 07v2_1_branch * 10emc2/configs/motenc/motenc.ini: backport: remove out-of-date comments
is that a rt command?
yeah, it causes a RT hang for up to 162000 seconds (45 minutes or so)
I can't find a freak'n box to box my gift in... anyone got any good ideas on making a box?
cardboard, alu angle stock, sheet...
what's it supposed to look like?
lerneaen_hydra, dunno... need inspiration.
an origami box could look cool
very simple to fold
lerneaen_hydra, I'm asian... origami is for white people now... so not asian to do something like.
I'm Canadian born Chinese.
with a UK citizenship...
hmm, origami isn't a good idea then
also, it's gonna hold candy too... and I want to fit an ipod in there.
the ipod box is, something like6"x6"x1.5"
is this the valentines day box?
something like that.
I was about to say that if it was for a (guy) friend then you could weld together a 3mm steel box around it and let them open it best they can....
I did that once
took a while to open
hmmmm... I'm thinking right now. of alu, with black plexi glass as walls.
I did that too
lerneaen_hydra, I don't think the GF would appreciate that.
hmmmmmmmmmmm.... just thought of an idea.
except I put an access cover on one side held on by 32 #4-40 screws
damn... the tapping would be such a pain in the ass.
CNC dude... I'm lazy
red plexi glass, with a alu frame.
Skullworks; oh, that's cheating
Skullworks, I don't have an tapping attachment...
used a air-butterfly to put them in...
the real way involves welding everything shut so you have to use an angle grinder or similar to open it
yeah then open with a plasma cutter and hope the static don't hurt the iPod
or just use lots and lots of ablative sheilding
the ipod box was gonna be surrounded by cinnamin herats.
thinking of making a box...
alu... would need something like 12-16 little cubes of alu...
gonna draw it in cad now.
need to make one of those boxes that if you slide the proper part out the whole thing comes apart like a puzzle
used to see them sold in novelty shops made of wood
yeah... those are cool
had a few when I was young
saw on that was made of stainless...
one key piece, the rest puzzle pieces.
was pretty cool... let me see if I cant find it.
maybe there is some page out there... fire up ggogle
hi all... does anyone know how to use rtai_ulldiv?
oo i like the velocity display in axis
can't find it now.
it was on cnc zone
[21:10:06] <a-l-p-h-a> http://www.cnczone.com/forums/attachment.php?attachmentid=20640&d=1154742127
re: rtai_ulldiv , not me, very few are that deep, that they are familiar with the rtai calls, i suggest asking cradek/jepler/joni/jmk
this was the thread... http://www.cnczone.com/forums/showthread.php?p=167235
it's about me!!! :D
or at least my work heh
that's some fun... machining that sucker.
trying to get it aligned properly and stuff... not fun unless you have a rotary table I would assume
maybe the rtai or comedi ppl ?
I found a mail talking about that and the guy who asked managed to get it done, but did't explain how
I'd like just to be able to perform a division
is that unsigned long divide?
beetween RTIME values
not familiar, just kinda guessed
I have a question about dallurs config. I see he is not here, but maybe someone else can help me.
in dallur-advanced.hal, in the section "connect all limit signals to relevant axes". What is the logic of "pos" and "neg" there?
sigh ... how I hate cnczone
you have to scroll down a whole page before you are clear of the adverts ... theres another entire pages worth at the bottom ... sigh
someone has to pay for the bandwidth
atleast thats cheaper these days
thats only part of the problem ...
the other two parts are:
1) avergae IQ of the users is in single figures ... a few brighter ones on there now, but it does tend to attract pondlife
2) censorship. try responding to a question about someoen askign if something is worth buying .. .it happens to be crap and you say so,
no problem ...
unless the stuff is advertised on there ...
there were some really REALLY bad stepper drives a guy was advertising, and I took the trouble to explain to someone why they were not worth having .. it "disappeared"
environments where you can't say anything negative about anything breed idiocy
name of said drive?
* robin_sz forgets
basically, it was 4/8 fets ... and a 3V supply ...
since there are no edit nazis here
the performance one was a 5V supply and some big resistors in series with the fets ...
no chopping, just ... clunk, clunk, clunk
ah - high tech ( if you were living on the 70's still).
my mill used to run like that (half stepping) - a pretty naive drive system
my 60s Bridgeport had saturable inductors in series with the coils ...
no, no no ...
cradek: no ... not half stepping
you misunderstand .. 3V PSU ... just switched across the coils ...
well you could half step with that setup
that wasnt the issue
depends how you drive the transistors
it was the 3V and total lack of current control
sure I understand
coil saturation times - max step...
people complained it didnt go very fast ...
and hot too I would guess
even my 196x bridgeport ran 70V to the motors and had a saturable inductor thing to limit the current
still kept your shop warm I bet
(I've seen the fins on those old stepper motors)
1st CNC I was trained on was a Bridgeport with a Bandit series 2 CNC.
still have the paper tapes...
no radius comp
such a joy!
[22:46:46] <robin_sz> http://www.cnczone.com/classifieds/showproduct.php/product/385/cat/9/date/1140378910
there you go ... theres the shite drives :)
and for the ultimate fun .. http://www.stepperworld.com/FET3dynamo.htm#4
... oh yes, this is the top of the line system with a 5V PSU ...
Skullworks, ours had a BOSS 5 control ... up until we ripped it off and put a Baldor and some Geckos on it :)
we've run some tricky 20 hour plus contouring programs on it and its not missed a beat
robin_sz, what's a baldor?
wonder can I get these sourced locally. W0QQitemZ190026120528QQihZ009QQcategoryZ41403QQcmdZViewItem or http://cgi.ebay.com/4-C1375Z-Large-ATA-Flight-Case-Trunk-Ball-Corner_W0QQitemZ190079961578QQihZ009QQcategoryZ41403QQssPageNameZWDVWQQrdZ1QQcmdZViewItem
USB on one end, 8 axis of step output or servos and encoders or whatever on the other
oops. first link was http://cgi.ebay.com/Lots-of-8-C1180Z-Mini-Flat-Flight-Case-Trunk-Corners_W0QQitemZ190026120528QQihZ009QQcategoryZ41403QQcmdZViewItem
robin_sz, sounds like a GREX almost
a-l-p-h-a, yeah like that, but with real servo capability ratehr than just step pulses ...
the GREX will be able to do servos eventually, but not just yet
thos ball corners are very very common
robin_sz, as opposed to the mesa, which does have real servo already
how does the mesa connect to servo geckos?
I didnt know MESA did a stand alone controller?
thought the servo geckos are step/dir
so... the mesa + gecko 3xx series doesn't play nice with each other then?
jmkasunich, or SWPadnos would be able to answer my Qs then.
the mesa is encoders in, +-10V out right?
shit... just put away the groceries, and now I'm hungry... wasn't when I was putting them away
and it doesnt have a motion controller on board
you need emc or some such to make it work
the Baldor and GREX have the motion control on board
well, to be accurate, the GREX has emcs motion controller on board :)
the G-REX has a highly modified EMC controller on it, and that in only one of the configurations
not the generally used one (with Mach), or the GeckoMotion one
the ones I have have it on anyway
There is a Linux client for it though (out of date, I think)
mmm ... "out of date" ?
are you using it with custom software?
or with Mach?
no, the linux/win front end and the emc-like software
the frontend seems to work just fine on linux or windows
I mean out of date in that those clients haven't been updated in a while, AFAIK
not that they suck
I compiled one only a month or so ago ...
steve hardy hasn't released anyting recently
right - I haven't noticed anything anyway
not that I've been looking though
ive not tried the Mach4 thing .. dont enjoy that much pain
the baldor stuff is good, but pricey
$700 for the USB box thing
I have Baldor motors, but I can't afford any of their stuff new
I buy the odd servo drive when I see one cheap
well, the problem with the G-Rex is that it isn't designed for industrial safety standards
we have a full baldor rig on the DIY laser
it's just a nice step generator, with extra stuff
well, it winks current instead of sourcing, there is no isolation for the digital outputs or the analog I/O
until the second revision, the startup state was random
isolation should always be done with optos, off baord
rev2 was better yes
the outputs sink to a common ground
but I'd imagine the Baldor has some of that stuff built in ...
but the Baldor is closed source
so the $700 price tag isn't as bad when you add in all the little "bits and bobs" you need with the G-Rex
for the avergae user, a GREX is fine though as the geckos have the optos in em
you just need an opto or relay or something for the additional outputs
and lets face it, most hobbyists dont even have their spindles on/off under cnc control
sure. too bad the average user doesn't know the meaning of "closed loop control"
the grex can be closed loop
yeah, but 95% of the users have no clue why they would want it
it can be, but what percentage of people do you think use it with anything other than the Mach firmware?
(which is not closed loop)
mmm .. zero
yeah - zero enough for our putposes :)
which is a shame
in reallity though
the average small stepper powered router gets away witout it
infact, plenty of commercial stuff is open loop
smal sign routers up to 48"
yep. I see posts on the CCED and Gecko lists talking about Mach being closed loop because someone wrote a plug-in that will stop the machine if the error gets too high
(using encoder input on the parport, counted in software)
that's a moving limit switch, not closed loop control
at least ow they shold be able to count on the grex
they got annoyed with me when I pointed that out :)
wait ... did I show yo the servos I spotted on eBay?
I wonder how good mach really is compared to emc2 ;) how a compairsion would look
emc2 running with a full setup
and mach with one..
mmm ... dont go there :)
I was thinking of just such a comparison
as ever, performance wise EMC would kock spots off it
I also think it would be cool to write a converter so EMC could load their XML files (at least the port pinout and tuning stuff)
and we are catching up fast on config/interface
but I don't think the format is documented, and there was so much crap in there, iI can't tell what's important
* robin_sz nods
SWPadnos: post a video..
and I have to reboot my machine every time I run it, since it screws up the timers so badly
probably the mach users would shit themselves...
I do still like the ability to create your own displays/layout in a drag'n'drop way
yeah yeah ;)
the py widgets will soon be there though
thats coming on very fast now ...
of course, its all wasted effort
I expect paul_c to be releasing his tuxcnc ting anytime now and emc2 will just be so poor in comparison ... ;)
just remember to tell me the sloccounts when he does :)
read the mm .. 3rd article down where he umm "evaluates" emc2
thats the one
I'd like to see tuxcnc and the trajectory control and stuff.
well, yeah .. its just emc oldstyle
well, tuxcnc is an emc offshoot, so I wouldn't expect it to be all that bad
SWPadnos: already forgot about how the old planner behaved?
and Paul is a smart guy - he may have made some nice improvements somewhere
no, I haven't
too stuck with the old stuff though
it was damn bad.
but the solution we have probably isn't the only one out there
and there's no reason why Paul can't use the emc2 planner - it is GPL, after all
andmaybe we can pick up on some of his bits
should we ever see the source ;)
as he's hellbent on GPL issues..
if he ever releases anything :D
liek we ever saw the source packages for his .debs?
[23:28:03] <anonimasu> http://www.tuxcnc.org/pivot/entry.php?id=11#body
Three CNC seminars at Cabin Fever this year - one mentioned only Mach, the speaker for another was a Mach developer - the third I missed.
d. Anyone contemplating using the code for any other purpose other than personal use needs to conduct a full audit and read the user/developer mail archives (would be wise to consider the commit logs also).
Sherline was there runnig their CNC machines - BDI, I think?
back in a bit - just got the 2-minute warning for dinner :)
anonimasu, yeah, that bit REALLY wound me up ... if he thinks theres a copyright issue, he should say where, file, line and version
put up, or shut up
well, he did in the mailing list..
that there are ones..
and what was the bits of code he identified?
not smallmath again?
nope, I dont think he talked about that..
he just complained about it..
well, I'll pay more attention whe nhe releases the source packages to the ,debs for BDI
that copyright stuff is nothing more than FUD
I just wish he would stop being so concerned over something he's so interested in pointing out that it sucks..
jmkasunich: I wish there was a way to make him stfu.
well lately he hasn't said anything on list
probably because he knows we'll call him on it
isnt he working part time for some american outfit these days?
so .. these servos I spotted on eBay ...
[23:36:32] <robin_sz> http://cgi.ebay.co.uk/ws/eBayISAPI.dll?ViewItem&item=320078627767
and he has another pallet of 5
they're JMK sized servos they are :)
I want them, but have no clue what for :)
robin_sz: build a mill
anonimasu, you saw how big they are?
13kw ... 20hp
big ole gantry mill, toothbelt drive on both sides
so .. 40hp of X axis ...
wait, not a mill ....
thought it might make a nice lift motor for my sheet feeder
needs a 1000kg lift
I presume they do servo drives that big?
how do you think they position machinetool spindles?
err mill spindles?
of course they make them, but you better be sitting down when you read the price
I'll wait for one cheap on ebay :)
I guess you can forget about it
I bought a 15A 300V servo drive on there recently for ... £50
not played with it yet
yay laptop backed up.
what are these "backups" of which you speak? are they a custom amongst your people?
I dropped my lappy 5 cm into my bag with the power on, and windows ate something
thats one of my favourite snowclones
dont run windows on laptops?
I need it for work.,..
it's my work laptop
to be fair, I'm just about to order a win laptop
so I were getting ready to kill myself.. :D
dell precision M90 :)
"no no no no! GIVE ME MY FILES!"
are the the smallest ones they have+
price tag like a small house
but, it has to be done
more like 3k eur
well, it's not insane :)
but I am
stapling machine mrs polansky
390eur is what the cheapest dell costs now
they do a basic desktop for 120 GBP
SWPadnos, you around?
no monitor, but what, less than 200 eur for a new desktop machine?