jlmjvm: any luck?
So, when you do the upgrade, what else do you have to do to switch? Do you have to run an install script or anything?
I followed the instructions on the wiki, but now when I start up, all I get is the splash screen.
MASEngr: just this: http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?UpdatingTo2.2
MASEngr: next you need to update the configs
according to the link on the bottom of that page
Those files haven't moved, have they?
They seem to be identical and in the same location.
The config files.
the files in ~/ are not touched
their contents need to change somewhat
only the sample configs in /etc/emc2/sample-configs have been changed by the install script
you need to change the configuration, not the location or name of the configuration files
probably all you need to do is copy a new var file to your folder
and name it accordingly
linkpp could be an issue too
halcmd: linkpp parport.0.pin-01-out parport.0.pin-10-in
<stdin>:1: Warning: linkpp command is deprecated, use 'net'
(that's what I get with 2.2.0)
OK. I guess it's CVS that errors out?
it does work though
ok, I thought there were some errors reported
no idea :D
didn't see it error yet
Alex_Joni: Wait, are you talking to me, or looking for help yourself?
MASEngr: not really looking for help :)
MASEngr: you need to copy a .var file from /etc/emc2/sample-configs/stepper/stepper.var to your config folder
MASEngr: sounds complicated?
No, I just didn't have that file before.
maybe it was called differently
but there must be a .var file in your config folder
the new one must overwrite the old one (same name), or else it will not work
note that any coordinate system offsets will be lost when you overwrite that file
alex_joni:no luck yet
jlmjvm: same error?
All I'm getting is the splash screen, then nothing. When I restart, it says, "EMC2 is still running. Restart?"
[00:20:02] <jlmjvm> http://pastebin.ca/764184
MASEngr: open a terminal (applications->accessories->terminal)
MASEngr, try running from a terminal. that should give you a meaningful error message
and run emc by typing 'emc'
jlmjvm, you also need a new var file
jlmjvm: can you pastebin the output of "ls -al /home/jlmjvm/emc2/configs/stepper" ?
SWPadnos: he already replaced it, if I understood him correctly..
doesn't look it, from that error message
didnt replace the var file yet,only the emc.nml
replace the var file, then try running again
Okay, I now have some info. The linkpp command is deprecated, and I should be using net instead.
[00:24:01] <jlmjvm> http://pastebin.ca/764187
Also, the configuration is now out of range. Did the ranges change?
I'm not sure what's wrong with the nml file, but EMC certainly won't start without the var file being right
configuration out of range?
parameter 5221 out of range?
Sorry, Coordinate system index parameter 5220 out of range.
right - 5220 ;)
I'm using sneakernet to cut and paste.
Sorry about the lag.
is the EMC machine on the net?
MASEngr: that means the var file is the old one..
Yeah, that would be easier, wouldn't it.
MASEngr: you need to replace your old var file with a new one from /etc/emc2/sample-configs..
Adam1 is now known as MASEngr2
There we go.
see.. wasn't that hard :D
I was getting instructions via a different machine while this one was being worked on.
It wasn't *total* clulessness. ;)
Anyway, yeah, the errors I'm getting on startup are: Starting EMC2...
m5i20_io.hal:83: Warning: linkpp command is deprecated, use 'net'
emc/task/emctask.cc 312: interp_error: Coordinate system index parameter 5220 out of range
Coordinate system index parameter 5220 out of range
emc/task/emctaskmain.cc 2629: can't initialize interpreter
MASEngr: whenever you see the coordinate system .. out of range message, it means you need to replace the damn var file
Right. And the stepper.var you pointed to earlier is a known good configuration.
jlmjvm: your stepper.var is dated from today.. I don't think you replaced it yet
(actually any of the var files in the sample configs are just as good)
you have to remove stepper var and replace it with emc var
jlmjvm: that depends on how it's called inside his config
Okay, that error is gone. Now I've got: Starting EMC2...
m5i20_io.hal:83: Warning: linkpp command is deprecated, use 'net'
emc/task/emctask.cc 312: interp_error: Unable to open file
Unable to open file
emc/task/emctaskmain.cc 2629: can't initialize interpreter
thats what i just did
thats what im getting now
Wait, why I am asking for help with that?
MASEngr: what's your config folder called?
is it m5i20 ?
who owns the var and/or nml files in your config dirs?
[00:34:05] <jlmjvm> http://pastebin.ca/764200
Who, me? It's a 5120, yes. The folder is called m5120
jlmjvm: a reboot might help
MASEngr2, can you pastebin the output of ls -Al in that dir?
MASEngr: ok, for you the var file is actually called m5i20.var
so you need to replace that one
Same content, I think. (I renamed the stepper.var to m5120.var after a cursory glance.)
But I've replaced just to make sure.
m5iiiii not 1
head, meet desk
ok.. just pointing out :P
ah, we meet again
Uh, I can't tell if this is a common error or if I'm going "above and beyond" here.
MASEngr: not really, so far :)
alex_joni:no change after reboot
Oh, good. It's hard to tell sometimes.
jlmjvm: ok, lets step back for a bit
jlmjvm: try to run a sample config
Now the latest error is: Starting EMC2...
m5i20_io.hal:83: Signal name 'iocontrol.0.tool-prepare' may not be the same as a pin.
Shutting down and cleaning up EMC2...
EMC terminated with an error. You can find more information in the log files
Which tells me: 8217
PID TTY STAT TIME COMMAND
Stopping realtime threads
Unloading hal components
there's the error with linkpp ;)
MASEngr: you replaced linkpp with net ?
Well, I thought that might work.
net has a slighly different syntax :P
if it's an error now to have net names the same as pin names
you were close..
net needs a net name
net <name> pin pin pin
ah, an error with the net command - I get it (duh me)
Okay, so what's the fix for that?
add a name
net somename <the smae list of pins in the linkpp command>
samd, not smae
net masetoolprep iocontrol.0.tool-prepare ...
in m5i20_io.hal line 83
same, not samd or smae
what about smee?
what about smee? smee's me - what about me?
Ahah. Yeah, I just added "toolprepped" and "toolchanged" in front of the two lines.
after net, I hope ;)
Now it boots.
MASEngr: good, now try again :)
one we can kick out of the channel now
Yes, I already changed it to "net" from "linkpp"
MASEngr: don't sweat it
What about the old configuration parameters? Can those just be copied over, or will I have to recalibrate the machine?
they are already there
config is different from var files
the var file has coordinate offsets, so those will be 0 now
Ah, excellent. I'll set it to run 500 parts and leave for the evening.
* alex_joni whines silently
but things like scale and pin assignments are in the hal and ini files, which you shouldn't have changed (except for hte net thing)
sim_inch is running
people are overconfident in software
ok - let us know how many of them come out right ;)
jlmjvm: ok... try a stepper sample
Alex, I've been programming for too long to trust ANY software.
disconnect your USC/UPC first
Disconnect the milling machine first, you mean. ;)
that too ;)
unless it's the other machine
SWPadnos: then why is his config called stepper_inch ?
there are two or three machines there, I don't know which this is
the stepper inch sample comes on,but gives this mesg:RTAPI: ERROR: Unexpected realtime delay on task 1
that's OK - but this computer may not be ideal for realtime
that only points to possible problems with your machine
didn't you have that problem before?
machine was working
(I thought we discussed putting in a PCI or AGP video card)
anyways.. take it as a warning for now
i did get it when i tried a different vid card once now that you mention that
but there has been no recent change
ok, moving on..
back to the config issue
jlmjvm: if the sample config runs, that means emc and co are OK
why are the x and y jogs backwards?
because it's not your machine
it's a *sample* configuration
never mind,im losing my mind,lol
now.. lets try to fix your configuration
there are 2 possible ways to do it
should i just build a new 1
1. fix your old 2.1.7 config
2. build a new one based on the new sample-configs
2. has the disadvantage you need to set a lot of parameters
wouldnt i be better off with new?
you need to set steps/inch for all axes
limit switches if you have them
(might easily turn into a couple hours job)
lets try 1 again
I will type some things and you will tell me if you agree or not.. ok?
your config is in /home/jlmjvm/configs/stepper
the config file you run is stepper_inch.ini
jlmjvm: ok so far?
can you copy a fresh var file?
from the samples?
cp /etc/emc2/sample-configs/stepper/stepper.var /home/jlmjvm/emc2/configs/stepper/stepper.var
yes, from the sample
i deleted the stepper.var earlier and replaced with emc.var from the samples
didnt find a stepper.var in the common samples
rename emc.var to stepper.var
only the emc.var
mv emc.var stepper.var
or copy as alex suggested
jlmjvm: ok, now lets try to run it again
but im still getting the realtime error
glad it worked
the realtime error is a problem with your PC, and won't be fixed with software configuration
unless you use an older emc2
which simply doesn't tell you there's a problem
are there new hardware requirements
right, where it doesn't bother to tell you
this is all new stuff
no, this is a better diagnostic than before, to prevent possible problems
no, but the newer version of emc2 is more exigent? in finding and reporting errors
but will it still cut a part
should, but it may not look so great
yes.. just as good/bad as before
SWPadnos: the task 1 error is a rt deadline missed reported by rtai
right - the same as before, but it's now telling you that there could be timing issues with this PC hardware configuration
right. I'm not sure if the code waits a little while after startup to start checking, or if it starts up immediately
the emc2 determined error (% of lateness) is a different one
and that one is more severe
the task 1 error usually appears more easily during startup
let me reboot the computer
no need to
shouldn't be necesary
unless you really want to :P
this is not a software problem
the video card is a common culprit, but it could be just about anything
what is the CPU speed on this PC?
jlmjvm: what you could do..
jlmjvm: open a terminal, and type 'latency-test'
(make sure emc2 is NOT RUNNING when you do this)
amd2.2gh socket 754,gig ram
running latency test right now
hmmm. onboard video?
(or did we ask that already?)
what video card?
are you using the nvidia display driver?
don't think so
* alex_joni goes to bed
is nv ok?
nv is the OSS one
have a good 2-hour rest ;)
nvidia is the bad one
ok. I thought they both had issues
and that it's best to use vesa
SWPadnos: I might squeeze about 5 in
, but I'm not sure
I run nv all the time
what should the latency numbers be
jlmjvm: ideally 0
but there's no such system :)
I get about 11-12 usec here
but numbers below 10000 are probably very good, between 10000 and 20000 are probably OK
that's quite bad
140143 is terrible
rebooting won't help
especially for a parport based machine
maybe he changed to vesa?
although that was too fast for that :D
yeah, I was going to say that that's unusable for steppers
might still work for an USC
although I have nfc
could be, but it's still pretty bad
that's 14% of a 1KHz update - lots of phase noise
* alex_joni sticks fingers in his ears
hi there. long time no see
what was that latency test command
didnt have the -
did you change video drivers or something?
nope,something must be dying,lol
hard drives can also be a problem
did a full power reboot and my latency is much lower
open some windows and drag them around
it did that once before
run something like glxgears
but dont open emc,lol,like i just did
it really didnt like that
8877,7502 is what im getting
I hope that's two different numbers mashed together
with the test running (but without running EMC), start some programs and drag them around
open a terminal and run glxgears
expand and contract the window, drag it around, etc.
als, just for the heck of it, go to a terminal, change to the root dir, and type `find . -type f`
its 2 numbers
but i still get the emc error when it starts
what kinda latency numbers do you have
none at the moment - I'm not on a RT EMC computer right now
figured you had some super low numbers
one computer I use has RTAI test numbers below 500
I'm not sure if the HAL test you're running gives the same numbers as the RTAI test though
damn,my low was 4200
the max on that particular PC is about 2000 or so
but it isn't running a GUI
no X at all
did you have any realtime errors when you changed to 2.2
I don't have a machine that I routinely use EMC on
my Bridgeport retrofit is a "very long term" project :)
could you get that error if you dont have any hardware hooked up?
mine is gutted right now
yes, it's telling you that the rtaltime system noticed that a deadline was missed
it's possible for that to happen with only a disk drive attached to the motherboard
thats what it is then
that's what the message tells you, yes
id bet money
so the trick is to try to find what piece of hardware/driver is causing the problem
i bet it will go away when i hook up my hardware in my mill and connect my parports
I wouldn't bet that
but you can if you like :)
well im hoping
I don't think that pays, actually
you should try to find the problem, especially with a software step generation system
this is less of a problem for a system that uses something like the USC, since it doesn't need to do thigns quite so often
how do you find a realtime problem?
that's a very good question
the way it's usually done is more or less by trial and error. Try changing hardware and/or drivers until it goes away
try running applications that might cause the problem to occur more often, to try and narrow down the cause of the problem
the machine gods smiled upon me today
and i got the 11" blower covers finished
i was expecting something to go horribly wrong but it didn't
a first for this project
if i use the stepper inch sample i dont get the realtime error anymore
may need to just do a new config anyway
SWPadnos:it was my base task period set too low
sure - if you try to run too fast, you'll get errors
but, that very high latency number (143000+) is unusable for stepper systems
was 10000,changed to 20000
at least, for stepper systems that you want to run faster than 3IPM or so
ok. if those exceedingly high numbers were just anomalies, you may be OK
im sure they were
but I'd want to know where the anomalies came from before cutting anything :)
i prolly had it jammed up earlier
where do you get to the new stuff that i was seeing earlier,like the port setup
I'm not sure what you were looking at earlier
was it the stepconf script?
dont know,alex sent me a link earlier that showed a box where you set the pins for the parport
try running stepconf from a terminal
without EMC running (just in case :) )
gave an error
I don't know anything about it other than the name and some of the pretty pictures. sorry
will worry bout that later
see you. good luck
what's going on this evening
mostly tech support for people who have upgraded to 2.2
I haven't even tried yet.
it's easy, but you do need to be sure to update a couple of things in your configs
[03:00:24] <SWPadnos> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?UpdatingTo2.2
and then follow the link near the bottom, UPDATING
I have been busy trying to learn Python and C and C++.
I will try it. thanks
oh, that's a good evening's work ;)
or three hundred
I was going to say 'or maybe two'
just being realistic ;)
one thing I've noticed about computer languages - the syntax is generally pretty easy to pick up (a computer needs to understand it, after all)
I am not wanting to be an expert - just understand some of what is going on
it's knowing what's already been done for you (library contents) that's the real kicker
my main problem is trying to figure out the new symbols and short cuts
so for you, the hard part will be when you look at a function name, you need to look up where things come from
whether it's printf or inifile.find
(the former from standard libraries, the latter from the emc ini routines)
I have the same problem with EMC, though I'm usually handicapped by using a Windows system on which I don't have the latest code, so I can't search ;)
go to www.mpm1.com:8080 and check out the link there. tell me if it is working and what you see
cradek - what is the computer is hosting EMC? Surplus sales has some quad core 3 gig w/2 gig ram and 160 gb sata hard drives. These are IBM Intellistation (M50). I think they are model 6830. They were manufactured in 2006. Is the host comparable?
it's hard to tell if it's "working", I have a project named "unnamed"
with a bunch of main main main main on the project page
did you click on "unnamed"?
that's where the main main main ... (29 times) is
yes, it does seem to have the data
incidentally, when you ask about "hosting EMC", which part do you mean?
the CVS server is one computer, and the wiki and linuxcnc.org are on a hosting service
the server - the computer doing the serving
serving of what?
(there are multiples :) )
ok I was just wondering if the project could use a good fast computer
heh - as many as you'd like to donate ;)
but it wouldn't help much for servers. the CVS server was a laptop in Jeff's basement for a while, I think
I'm not sure what it is now, but I suspect the limitation there is upload bandwidth on his DSL line
probably - I was just trying to find out where I could contribute a little bit.
much appreciated. I'm sure we'll think of something :)
does the application on my website look helpful?
in flat form, I'm not sure. If there are more tree-like options, then absolutely
though the flat form is also useful because you have only one place to search for identifiers
pick on some of the identifiers and then scroll the left window up and down
ok - it looks like there's quite a bit hiding behind those links :)
I don't understand the difference between flat form and tree-like.
oh, flat is where there's a list of more or less everything, regardless of where it comes from
tree has things grouped in a hierarchy, like a file browser (a directory tree)
this then is both flat and tree-like
yes, it seems that way
first impression is flat because there are so many identifiers
and all the "main" links are from various files that have a main() function
so it looks a little weird, because there's no way to differentiate the different mains
in a tree, you might have "emcsh" which has a node under it called main()
ah, ok. the link does show the file name, so if you can see links (like firefox shows in the status bar), then you can differentiate between the mains
I haven't picked on any of the mains. I picked on some of the file names farther down on the main page. Then you can pick on identifier in those files. It will give you a lot of information about the identifiers.
This is my first attempt at using this software. It may not be as complete as possible.
it's great, especially for a first attempt
I am trying to understand the data flow through EMC. This helps me a LOT.
there's something similar on the CVS server, called lxr. I'm not sure how much duplication there is between the two
heh, then it's useful :)
I need help, I'm a slob
clean up your room!
just found a robot
I was wondering how this compares to open source. This is from a man in Slovakia. It is called Xrefactory. I have only downloaded the trial version. This is the output from compiling emc2-trunk this morning.
this is an example of lxr (and what it was written for): http://lxr.linux.no/
you can click on identifier search, freetext search, or file search in the top right, to search for things
otherwise, it's similar - get to a file, and inside the file, more or less everything is clickable to see where it's used and/or where it's defined
I am looking at it. The Xrefactory is mainly written to refactor the code. I don't have ANY idea if that is useful or not. I was looking for something to cross reference the project so I could follow the flow.
I'm not sure we want to tackle a real refactor before EMC4 or so ;)
What is refactoring? Will it rearrange the files. Write new files in another logical arrangement?
I'm not sure it can be done automatically or semi-automatically
I will try lxr tomorrow.
refactoring is basically restructuring the code so it works better and is more maintainable
I'm attempting to get emc up and running on my old laptop... got ubuntu installed and emc2 running.
But i don't see any ticks on the parport.
It's a stepper config.
it might be needed due to long-term feature creep (like EMC perhaps), bad initial design, changed requirements, etc.
Any suggestions on where to start for figuring out why there are no pulses on the parport pins?
I read that about it but what does it do? I may have to try it and see what it does.
I'm not sure I understand the question
I can't hurt it any more than having to erase and reload
Then I will be able to see what it does.
wowbagger, does this help? http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?TroubleShooting
what lxr does?
Does it rearrange the includes, defines and other things or does it rewrite new files in another logical format?
no - refactoring
refactoring is the software word for redesign (from the mechanical world)
if you have a machine that has a part that often fails, you'd redesign that part, and therefore likely the surrounding parts, and then maybe more of the parts, until you eliminate that problem
at some point, you go back to the drawing board and decide to do a refresh of the design, because you have many small changes, and you may see that there are some more major and/or funcamental changes that can improve things even more
so even though the current design works, you "fix" it anyway, for whatever reasons
that is refactoring, for software
did I get that rightenough?
a correct (enough) description of what refactoring is)
I don't believe in refactoring, so I'm a bad one to ask
SWPadnos: Indeed. I had to add loadrt probe_parport. Thank you!
err - great!
but also greaet!
the idea is you can change the code all around to make it more pleasant to work on, without introducing any bugs or changing functionality
or at least that's my understanding
sounds like a good idea
I have never seen it done :-)
note that a laptop is probably not going to be the best computer for actual machine control, unless you're very lucky
I was thinking more like turning frankenstein into - um - some good looking guy
I'm still trying to load www.mpm1.com:8080...
I'm going to give it a try and see how painful it is...
just keep your fingers out of the way ;)
it's a small, weak thing.
it won't load for you?
what does it do?
pretty much no response, only a spinning icon thing and "Loading..."
ok it timed out now
interesting. I get a timeout with FireFox, but Mozilla loaded it fine
i would like a bumper sticker that said "my other mill is an integrex"
My other mill is a Krups
my system administrator only lets me have one web browser
i would hang it on the shitty Tong-il we have
wow - they still make the same coffee mill I have
Chris - www.mpm1.com:8080 is from my desktop computer. it seems to work real fast some times and not at all at other times. This is the first night of serving from that computer.
and have had for close to 20 years
stustev: how is your conversion coming?
In my old emc.ini file I had a input/output_scale setting of -8000 to make one of the axis go the right way. Is that how I should still do it? Or should I invert that stepper's dir pin somehow?
negative scale is still the easy way
No progress yet this week. Randy should be in tomorrow to install some hardware. I have been trying to learn Python, C and C++ in the last two weeks. It may take me another day or two.
or three ;)
stustev: I'll let you know how long it takes me (remains to be seen)
I was being optimistic
have you been working on it a month or so?
yeah something like that
if I knew python and a lot more about C++ (and C), then I'd be there already
I still know squat about C++. I can (barely) edit it and fix bugs in it.
i feel the same way about g-code.
I've spent 5? years trying to care, but I don' tyet
from the experts and presentations I've seen, I'm very impressed with it, even for embedded work
however, I'm not sufficiently proficient to actually use it yet ;)
I am sufficiently deficient though
I'm a little excited about the new probing stuff
it will be nice to have a smart probe algorithm
I have a couple bugs left though it seems (at least one I didn't recently add)
maybe a refactor will fix the bugs
I mean, hmmm. could be
[04:02:07] <cradek> http://www.google.com/search?q=define%3Atroll
refactoring is like building yourself a new VMC from scratch, instead of replacing a dull end mill
they had to do that to solve chatter problems on modern vmcs
that's them, not us
what's that - put a dull end mill in the spindle?
well, if the old one is bad, you should stick a new one in ;)
a dull end mill will solve chatter problems
no end mill should too
no end mill - will cause no noise
indeed. no power is even better for that though
you can feed at like 40000 ipm
with no endmill
like air guitar
exactly but different
NO EXACTLY THE SAME
well, OK then.
air guitar has noise - the person singing
the mill makes noise!
and believe me, that is noise
not without power
the ultimate quiet solution
i think someone's crashed the VF-0 at school one too many times
mill noise is music to MY ears
for all power equipment
the spindle load sits at about 40%
this xref thing is working for me now. it reminds me a lot of lxr.
anytime it is on
and the damn thing stalls
I'm impressed that the worst I've done to my bridgeport so far is to pull the work up out of the vise
i was at 500 rpm @ 7.5 ipm
3/4" five flute roughing mill
nothing too crazy
it was like, plate
and the spindle just stops after a second
what's driving the spindle?
what was the depth and width of the cut?
and what kind of overloads are there?
swpadnos: i suppose so, whatever Haas puts on their VMCs
i think it's an integrated spindle/motor with a vector vfd
it's supposed to be like, 5-10 hp?
unless there was a massive current surge at some point, I'd be very surprised if the motor is damaged
the depth of cut was .375, the thickness of the plate
width was full slotting, at .750" wide
it seems much more likely to be the drive or protection systems
5 hp should cut that - even full width
a bridgeport would cut that
but yet, the spindle just... stopped
and was bogging until i dropped the feed down to 5.0 ipm
i think somebody crashed real bad
and didn't tell anyone
40% load is not normal with an idling spindle
how does the spindle feel when you turn it by hand?
not as free as I would expect, but it is smooth and without catches
there's no neutral on the mill, so i don't know if i'm feeling the motor or the bearings
you'd think a crash would cause runout
the inside of the taper is a little buggered
or i'd look
this mill gets a fair amount of abuse
it shouldn't be free and easily turned. It will have to spin the spindle and the motor. When you spin the spindle and release it does it coast to a stop or stop immediately?
government jobs only?
stustev: it stops fast and bounces back a little
eric_u: it's a VMC at a college
from just a spin by hand?
ah, student jobs only
eric_u: no, students don't touch it, it's the inept "technical staff"
is your spindle brake stuck?
stustev: no, it turns too freely for that
and there's no lock option
hmmm. could be dragging though
which makes me doubt it has one
the only think it's got is a detent thingy for spindle orient
eric_u: i saw one dude crash the mill six times in the span of ten minutes trying to make, i shit you not, a t-nut
40% load into a stuck brake will be evident to the nose
and to the ears
it's funny retrospect
but at the time i wanted to kill the guy
especially since he gives EVERYONE shit about very tiny mistakes
but he goes through two endmills anytime he sets his hands on the thing
and might not get a good part out of it
i don't know what's up but when i see the department head i'm going to mention it
the spindle issue
also, after you do any cutting on this thing, the tools get like, FIRED out of the spindle into the tool carousel
during a tool change
very loud "BANG" and the whole carousel shakes
I think I'd stay far away from this machine
and Mr. T-Nut
i don't think he's allowed to use the machine by himself anymore
nobody besides me can touch it without the department head being around or pre-approving the program/setup
i still want to know why the spindle stopped
it didn't even throw an alarm or an error
i just hit "go" again and it turned back on
and continued cutting, ablit bogged down a bit
programmed 500, was turning 390 or so
the taper is shot - it is bell mouthed really bad. put some prussian blue on the taper, do a tool change by hand and see where the taper touches the tool. it will touch on only about 1/2 to 3/4 of an inch on the top of the adapter. but that is not why the spindle load is so much
stustev: i will definately do that
i have a feeling the root cause of both issues lies in the same ballpark, though
could just have many problems all caused by a different kind of ineptness
check the runout on a tool - do a tool change and check it again - tap the side of the tool holder and check it again
like, gently, or a little tap with a lead mallet?
(gently by hand)
yes tap it with a small hammer
sure, what am I looking for?
not only runout, but change in said runout?
you are looking to see if the adapter will move sideways in the spindle.
i was kind of surpised at how stiff the spindle was though
I have some Haas machines. I have had the spindle load problem before. I will ask tomorrow what the solution was.
i was expecting something like a bridgeport or a little heavier
this is like, the big 30 hp mori seiki at work when it's in gear
does not turn freely
smooth, but not free
does it now make a lot more noise than it used to?
the spindle has some imbalance in it for sure
but i don't know about more noise
when you get it up around 3k it starts to give a little dirty "vrrrr"
Would installing vmware server cause base thread to be longer ?
like a real high speed, low magnetude imblance
what does it sound like at 100 rpm. it should be silent
or near to it
you could hear a little motor whine but otherwise the spindle should be silent. Have anyone called the Haas dealer?
nah, nobody uses it anymore but like me
i'm going to bring it up
"call haas, something to do with the spindle/motor/load limits is not right"
if no one uses it and management has no interest they will not be willing to spend any money to fix it.
well that's the thing, it's a "show" mill
they're like, "we have this fabulous program look at our technology"
how old is it?
so if something is really fucked up, they're going to spend to save face
five, six years old?
past warranty - you would have better luck trying to get Haas to replace it for free.
to save face and show technology
haas has done things like that in the past
not sure what the particulars were or what they did though
penn state is Haas' showroom
and they will again - it will take 6 months or a year - someone there knows - ask around
eric_u: penn state?
they actually sell machines off of one of the floors at the IE school
are you serious, i was looking at going there
the school gets new machine tools continuously this way
eric_u: i haven't looked that hard yet =)
which one has the machine tools, i'll pick that one
University park, of course
by "looked" i mean "is on my list of schools to review"
they have a really nice new building for industrial engineering
mechanical engineering has gotten pretty good
damn, that's hot
you can look up the "learning factory"
good night all - must get my beauty sleep
stustev: we are a really tiny local community college
so dunno how accomodating haas wil be
and goodnight sir
goodnight to you too, if you are leaving
i guess i will take a shower.
[04:34:15] <eric_u> http://www.lf.psu.edu/
grad student told me today they have a waterjet
anyone have a mesa they want to get rid of?
i wonder if i will have to take saftey/training classes once i have my journeyman's card
as a machinist
to use college facilities
probably, us Ph.D.
's arent too smart
judging from what i've seen a Ph.D. try to pull on a lathe, i'd have to agree in some areas
at least after all that training, you should learn to stay away from situations that will make you look stupid
I try to learn something from everyone I meet, even the janitor
toastyde1th is now known as toastydeath
dude janitors are awesome people to know if you have a question about how to clean something up
"how do i clean this up, steve"
"here, i will show you."
never helped me yet, but they like it when you treat them with respect
yeah, i make a point to say hello and ask how they're doing
to all the maitnence staff
yahoo's spam filter stinks
they are sending yahoo groups emails to spam
either that or they're SUPER EFFECTIVE
that's just goofy, wonder if they sent themselves a sternly worded email
which was then caught by their own spam filter, and generated another sternly worded email in reply
Should 2.2 's max interval be close to the ovl max in 2.17?
I've got an ovl max of 15882 in 2.1.7 and Max Interval of 37976 in 2.2 .
SWPadnos_ is now known as SWPadnos
SWPadnos_ is now known as SWPadnos
SWPadnos_ is now known as SWPadnos
SWPadnos_ is now known as SWPadnos
Another day - off to face the mid-week crisis at work.
user_: (yes, I know you're gone :-P) Max Interval is not the same as 'ovl max' in the rtai latency tester. the column most similar to 'ovl max' is 'Maximum Jitter'.
cradek_ is now known as cradek
Good morning. I just popped in to say thanks for the help last night. The machine appears to be working with the update.
Yes, it is. Thanks again. Bye.
where to make the changes to stepspace for double step /INI?/corestepper?
also how can I satify my Geckos 20us hold time? /DIR_HOLD of 2 periods?
al: in 2.2.1 there will be a wizard 'stepconf' that will write this kind of configuration for you. it might be smart to hold on for a few days until we get it released. I think jepler wants to do it soon.
ok I tried to fire up the wiz but it didn't start
yeah it's got a bug in 2.2.0 unfortunately
it will be fixed soon
will I have to reconfig the whole setup ?
the stepconf wizard asks you everything (scales, velocities) and writes a complete configuration
what about graphics cards the latancy site rec. g400 450 matrox but I have no APG slot? will any pci work?
yes a lot of pci cards work fine. in fact those matrox cards are (were) available in pci
al: if you feel engergetic - there is a lot of info here http://axis.unpythonic.net/0118813410
here I mean http://axis.unpythonic.net/01188134101
to use doublestep, the modifications to the ini/hal file are roughly: make BASE_PERIOD larger than max jitter + datasheet steplen + datasheet stepspace + a few more ns. setp parport.0.reset-time to the datasheet steplen in nanoseconds. for each stepgen, add or change these parameter settings: 'setp stepgen.N.stepspace 0' 'step stepgen.N.steplen 1'. For each pin MM which is a STEP signal, add: 'step parport.0.pin-MM-out-reset 1'
yeah, the data sheet for your stepper driver boards
bbl, it's lunchtime here
can I still use a DIR_HOLD time of two period?
I'm using a BASE_PERIOD of 35000 just know at 31000 I keep geting reltime delay error I have ran tha latancy test for 3-4 min. surfed copy and pasted and the worst numbers I got were 23000ns
jepler: you there
user_: Max Interval is not the same as 'ovl max' in the rtai latency tester. the column most similar to 'ovl max' is 'Maximum Jitter'.
al: yes, but it's better to just ask your question; maybe other people can answer it even if I'm not around.
setp stepgen.0.stepspace 0
setp stepgen.1.stepspace 0
setp stepgen.2.stepspace 0
setp stepgen.0.steplen 1
setp stepgen.1.steplen 1
setp stepgen.2.steplen 1
I put this in the corstepper and I get parm or pin not found
step or pin sorry
welcome to the world of troubleshooting hal files. Here are some techniques I use when faced with errors like that:
* read the relevant documentation for the component I'm having trouble with. In this case, "man stepgen" will show the documentation, and then I can verify that the pin name is spelled right
* in the hal file just before the line with the error, add "show param" (or ("show pin", etc), then look at the output to see if I can spot what the right name was
error mess hal 83 is this the line
core_stepper.hal:48: parameter or pin 'mumble' not found
this line means that the error was encountered in the file "core_stepper.hal", the line number was 48, and the explanation of the error was "parameter or pin 'mumble' not found"
I just installed EMC 2.2... thank you for all the improvements :-)
hi Hugomatic -- I hope the upgrade works for you
jepler: thanks ! I was reading the archive
I do have a question about the tool table: what are "machine units"? And what happens when I run a g20 file and then a g21 file: do I need a metric and an imperial tool table file?
Hugomatic: "machine units" are the ones you use in your inifile
jepler: I like the on screen limits and the image-to-gcode improvements
Hugomatic: the changes in 2.2 to the tool table are specifically so that you don't have to have different tool tables or tool table entries for g20 files and g21 files
Hugomatic: in 2.2, if your inifile is in inches, then your tool table is in inches, and tool offsets will work right in g20 files and g21 files
Jepler: excellent: my Sherline mill is metric, so if my tool table file is in mm (even though most of my end mills are inch), then I won't have to worry. Thank you for the clear up, it all makes sense now :-)
Hugomatic: oh good, I wasn't sure if my explanation was any good
al: any luck yet?
This weekend I am bringing my mini mill and emc laptop to a train/car modeler meeting in Montreal. Do you know of good ngc and png (for image-to-gcode) sample files?
i just got some diametrically polarized magnets for use with the magnetic encoder chips, and boy are they fun to play with: http://fennetic.net/pub/irc/magnets1.jpg http://fennetic.net/pub/irc/magnets2.jpg
usually it only contacts the surface at three points so the whole thing jiggles when you poke at it
Hugomatic: I'd also like to hear about good images for image-to-gcode
random question; wouldn't it be very easy to read servo-encoders at high speeds with a quadrature -> step/dir logical unit and a full adder (with a suitable number of bits), connected in parallel to the parport? (where the quad->step/dir and adder is physical logic)?
jepler: I found a fabric "bump map" at this location: http://hea-www.harvard.edu/~fine/POV/fabric.html.
Have a look at the 100x100 black and white weave pattern.
lerneaen_hydra: so for instance as the motor turns clockwise the value on the parport pins d0..d7 would go 0, 1, ..., 255, 0, ... and when going counterclockwise it would go 0, 255, ..., 1, 0, ... ? I think that would work -- elson's usc and the pluto are just very glorified versions of this.
jepler, yes, exactly like that
and you'd make it so that the number of bits is large enough so that it can't go past 1/2 the total number of states per base_period (otherwise you won't know which direction it turned)
so in reality it might go 0, 30, 55, 100, 150, 220, 31
going from 0 to 130 directly though is not good
one minor annoyance is that up/dn counters only come in 4-bit varieties it seems
well, you would have to interpret that as turning counterclockwise 126 units
fenn, yeah but you can daisy chain them
General question: is EMC2 compatible with the CNC steppers interface from fpga4fun.com?
jepler, yes, that's what the program would interpret it as, in reality you wouldn't know what happened
next you'll need some circuitry to have several counters with only one driving the data lines at a time .. soon you'll have invented EPP
Hugomatic: short answer: no
how fast are hardware counters?
in the tens of MHz
fast enough then
Hugomatic: longer answer: emc 2.2 has fpga firmwares for the currently-unstocked pluto-p board for control of servo systems (well tested and stable) and stepper systems (alpha-quality, some users have reported serious problems)
alternatively, as fast as you are willing to pay for
lerneaen_hydra: one thing i'm investigating soon is "magnetic encoders" which sense the rotational angle of the shaft, so you dont need fancy hardware to count ticks
or optical encoders..
fenn, they measure velocity?
no, they measure the magnetic field with 8 hall sensors and figure out which way the magnet is facing
so only 256 steps/rev?
they come in 8 10 and 12 bit from what i've seen
the 8 sensors is just a coincidence
I've never seen a "standard" encoder, are they glass-optical?
or something else?
jepler: thanks. I'm fiddling with fpgas in my spare time, so I'll have a look at the pluto firmware. I'd love to have a USB controller, because par ports are hard to find these days.
sometimes, sometimes they are a thin sheet of photo-etched metal
oh, with holes and optical sensors?
yes the etched part goes all the way through, but 'holes' might be the wrong word
ok, small slits far longer than their width
are there any velocity based "encoders" (they wouldn't be called encoders probably though), not that I can easily imagine hardware that would be velocity based
at least not an accurate one
sure they're called tachometers
and there are many varieties
lerneaen_hydra: this is one flavor http://www.electronicsam.com/images/KandT/servostart/Endoder1.JPG
cradek, aren't they relatively innacurate
they're used for measuring velocity, at which they do a good job :)
I'd at least think that they're not completely linear
what with the induced current and drag
skunkworks, ah, nice
skunkworks, nice re-use of a computer chassis led connector/cable
lerneaen_hydra: there are times when the quantization of an encoder is a problem, and you need an analog tachometer
also they were often used in combination with glass scales on the table, to account for backlash (and prevent the motor from hunting)
yes :) http://www.electronicsam.com/images/KandT/servostart/Encoder.JPG
yes, true, but I doubt that they can be good at standstill, what with the rotation being the integral of the speed (only a small error leads to a large movement over time)
lerneaen_hydra: nobody uses them to measure position
perhaps you are thinking of something like an inertial measuring unit
hmm I can't say I've heard of that before
integrate the signal from an accelerometer to get velocity and position
ok, so you'll need to null/set position and speed from time to time
not very useful in machine tools
skunkworks: how goes it
#double step changes Al's
setp stepgen.0.stepspace 0
setp stepgen.1.stepspace 0
setp stepgen.2.stepspace 0
setp stepgen.0.steplen 1
setp stepgen.1.steplen 1
setp stepgen.2.steplen 1
setp parport.0.reset-time 4500
setp parport.0.pin-03-out-reset 1
setp parport.0.pin-05-out-reset 1
setp parport.0.pin-07-out-reset 1
changes to my hal file
now I only get 60% step ratio
is it my gecko 20us hold time getting missed?
steplen = 1 is surely wrong
[21:16:08] <cradek> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl/emcinfo.pl?UPDATING
i'm there now It doesn't mention steplen ony step space
The units for steplen, stepspace, dirsetup, and dirhold have been changed from BASE_PERIOD units to nanosecond units. Internally, the values given are rounded up to a multiple of BASE_PERIOD, and at runtime the parameters are changed to display the rounded up value.
you have set them to 1 ns
use my gecko 4us pulse time here?
yes sounds like
i'll give it a shot
ok good luck
i put 4100 in there and got the same results
al: what do you mean by "60% step ratio"?
it only moves say 40% of the intended move
.4 out of 1."
did you addf parport.0.reset?
I bet that's the problem
addf it to the base thread
I'm not sure what the ideal order is...
(none of the sample configs appear to use doublestep, which might be considered a bug)
the configs we distribute as samples
cradek: for doublestep, the correct stepspace is 0 (step pulses may occur on subsequent BASE_PERIODs) and the correct steplen is 1 (smallest nonzero step length). This assumes that steplen + stepspace + latency is less than one BASE_PERIOD
oh did I give crap advice? sorry
Iwent back to one
cradek: 4000 or 1 would have the same effect, as long as BASE_PERIOD is above 4000
are geckos the most "needy" common drives?
if so, maybe our stepper sample config should do gecko timings with doublestep
forget it - stepconf will fix all of this.
.reset should be late in the thread. I think that stepconf places it last.
in theory you want to fine-tune it by looking at the typical run times of the other things in the function, and placing it where .reset-time has probably passed, but not by much
(because when reset happens late, it has the effect of decreasing stepspace)
do you think im missing my 20us hold time?
that's controlled by .dirhold and .dirsetup
i'm also using 70000 DIR_HOLD
then without evidence to the contrary, I'd assume it's fine
missing the "addf parport.0.reset" is enough to explain the machine moving wildly wrong distances
al, do you have step multipliers in your drives (G340 or G210)?
ok. how recent are they?
I think the 20 us dir setup time may be overkill
(though it won't matter much, since that's only applied at direction reversals)
the data sheet call for it
hmm. ok. I don't know if those use the newer pulse multiplier circuit
best to be safe then
tomp really here?
increasing dirsetup and dirhold will not have a huge effect on step rate unless you make them really crazy values
right - won't matter much
the addf parport.0.reset want an argument reset time?
spun steppers using emc for the first time today, very exciting
reset time should be set to the step pulse width of the geckos, 2 or 2.5 us
al: the full line would be: addf parport.0.reset base-thread
and it must be after the addf parport.0.write function, preferably at the end of the thread
so the order of functions is somewhat different - you start with the parport.0.write, then do "all the other stuff", then the parport reset
eric_1: great news :)
jeff you threw me a curve with the step and not setp
al: that's a typo I make all the time -- sorry for the confusion :(
thank you so much
al: is it working yet?
and she's sailing
were your velocities limited by step rate before?
dinner si calling my name thanks to all you EMC Guys
I wish dinner would call my name...
vending machine has already called and been answered
cradek, I thought you were a vegetarian :)
SWPadnos: I was thinking of saying a very similar snark..
need a third stepper drive
I wish I'd noticed that Mesa makes a screw terminal version of the 7i37 and 7i33, can't really justify buying them now that I have the idc versions
well, it's not as rosy as you might thing anyway
the screw terminals don't come with them. When I had to buy them, the terminals were $22 each (*2 per board)
right, it's got headers, not terminals
that adds up
Pete W said that they're now $10-12 or so at DigiKey, since he mentioned tha Avnet had them at that price
I know. I have 5 of them
and I even have extra terminal strips
you guys got your boards cheaper than I did though
the other problem is that the terminal headers are a little bit close together
you have to use a 180 degree one on the inside connector, so the wires point up
I wondered how that worked out, not a lot of space
the screws aren't accessible when it's plugged in, because they're blocked by the outer terminal strip
but 4x50 pin ribbon cables are driving me nuts
I tried putting the 7i33 and 7i37 inside the computer case, it was impossible
at least with the header versions, you could mount them to any available case area, and run the ribbons out from them
the way I had it, I had to flip the 50 pin cables on themselves twice, it was about 3 inches thick
that wouldn't work with the terminal versions
yeah, connector orientation is pretty important when making those cables
you wouldn't want the terminals inside the case
especially if you want strain releifs
heh - nope
too bad you can't just plug the 7i33 and 7i37 into the ports on the anything board
yeah, though it would still be pretty thick
definitely a 2-slot solution at that point
I was up to 4 slots
and it didn't fit
hmmm. it seems you may need to optimize that a little
I couldn't get the cables out the back of the computer
I have a 5i22 in an embedded system that has almost no space, and it'I have 3 cables coming out just fine
If only I knew someone with a milling machine
the fourth would be a little cramped, but I could do a cable with no strain releif at nearest the bracket, which would solve that problem
I'll have to look at that again
it seemed like it would be fairly easy until i tried it
you have to plug in all the cables before installing the card
then thread them through the slot, then plug in the card
that's no problem, i wanted the 7i37 and 33 inside the case, then cables running to breakout boards
the 3 cables I have are fairly tall, and have strain relief on them
oh, do you not have connectors on the breakout board end?
eric_1: wait - your running steppers and you have a mesa card? 2 different projects?
I'm helping a guy at work build a pcb router
mill is in my basement with servos
does flash work on 64bit linux?
I don't think so
eric_1: you can install 32-bit firefox and proprietary flash plugin if you try hard enough
hmm - I don't remember having problems playing youtube videos on dapper
it's a work computer, not having flash will keep the grad students from wasting time on youtube
hmmm. unless this is a correct solution: http://www.darronschall.com/weblog/archives/000258.cfm
the emc2 dapper install isn't 64-bit
I'm running fedora
I thought it was a linux thing :)
I may have followed something like these directions: http://ubuntuforums.org/showthread.php?p=1174435
I'll check it out at home
just noticed I'm logged in at home too
[22:38:53] <eric_1> http://www.futurlec.com/StepperMotorController.shtml
considering that ST lists the maximum supply voltage at 46V, and the max current at 2A, I'm not sure I like that futurlec board
people who inflate manufacturer specs don't get much respect from me
could be a problem
didn't see any docs either
i hate to spend much on a controller, I always end up thinking i should buy a gecko
50V is the "ABSOLUTE MAXIMUM RATING", but 4A certainly isn't (3A non-repetitive is)
well, you could just buy a gecko and save the consternation
the web site listed 46V as the max V
though that was probably "max operating" vs "max ever you dummy"
that would be pretty nice of me to give away a gecko
I'd have to look deep inside myself and see if I'm that nice of a person
I don't see anything that looks like an L297 or other translator+chopper chip on there
I didn't either, must not have step and direction inputs
pretty crude implementation
if it's not step dir, wouldn't it be quadrature (like step phases are)?
datasheet here: http://www.es.co.th/Schemetic/PDF/ET-SMCC.PDF
at least it looks like the same board: http://www.es.co.th/Detail_eng.asp?Prod=ET-SMCC-V2
hmph, calling it a datasheet is not exactly correct
itit's hard to tell what it says
but I think you've cracked the code, futurlec has most of those boards
looks like they may be using a 555 for a chopper?
oh, well, at least I found a substitute for the zeropower ram that dallas semi discontinued orphaning some of my brushless motor drives
in this file: http://www.etteam.com/download/et-smcc%20v2.zip
you will find .ASM files that indicate the interface is direct to the L298 control lines (6 of them)
e.g., ;/*=>Motor1 : EN1,EN2,INA,INB,INC,IND */;
;/* Signal : PA0,PA1,PA2,PA3,PA4,PA5 */;
not much use then
but that doesn't explain the function of the other IC
no, I wouldn't recommend this board
you might be able to rig it up with 4-phase outputs from emc, but you wouldn't have an effective chopper and you'll use up all your output pins that way
don't want to do that
too bad I only managed to scrounge 2 stepper drives
what drive though?
yikes - wonder what this will go for http://cgi.ebay.com/Parker-Compumotor-OEM650-Stepper-Motor-Drive-Nema-34_W0QQitemZ180177336680QQihZ008QQcategoryZ78196QQssPageNameZWDVWQQrdZ1QQcmdZViewItem
that's gonna hurt in the end
650 would be nice
I have had good luck with oem650s
I have the servo version somewhere in my basement
you can usually get them for under $50 if your patient.
how many amps are your steppers?
[23:10:41] <eric_1> http://cgi.ebay.com/Applied-Motion-Drive-Motor-HT23-397D_W0QQitemZ220167136233QQihZ012QQcategoryZ78196QQssPageNameZWDVWQQrdZ1QQcmdZViewItem
that's what I'm using on the first two axis
actually, the version that only has one l298
cool - have you thought about xylotex?
just don't look at them sideways..
do they have a single axis version?
* skunkworks was just wondering that..
yes I think so
the 3-axis version has a header to pass through two step signals, power, and gnd to some other board
I have a xylotex, haven't pushed it too hard
I don't think we will push this system too hard
hm, maybe I'm mistaken -- I don't see a 1-axis board on the front page
pcb mill mostly
I thought they had one, don't see it either
connector is just pp passthrough
eric_1: I think it's after the signal conditioning
they have a pdf for 1 axis http://www.xylotex.com/1axlayout.pdf
this refers to a 1-axis board with no stock: https://xylotex.sslpowered.com/secure.htm
someone is selling oem650 for 48$ buy it now
(approximately the same page as http://www.xylotex.com/OrderEntry.htm)
that would kick a xylotex around the parking lot without breaking a sweat
Yes - I really like the oem650s
it's in singapore
well, as long as it's not DOA
even the older s drives work good
[23:20:11] <skunkworks> http://cgi.ebay.com/Parker-Compumotor-S-series-S6-Microstepping-Drive_W0QQitemZ110085900057QQihZ001QQcategoryZ71394QQrdZ1QQssPageNameZWD1VQQcmdZViewItem
granted - that is way too expensive... but internal power supply
for that much I could buy another mesa and go brushless
* skunkworks ment to keep an eye out for them
mmm servos .. it's like the SLR fof hobbyist machining
not a camera guy, but I think I know what you mean
however, ebay allows one to have a huge collection of brushless motors and drives in your basement, ask my wife about that
[23:22:53] <eric_1> http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&rd=1&item=150180080660&ssPageName=STRK:MEWA:IT&ih=005
calls it a servo drive ;)
750 is servo?
they look exactly the same
750 is also stepper - but It has mid band resinence(sp) dampening.
JymmmEMC has some 750s.
I don't know
looks like the shipping from singapore would be at least 27$
apparently the "OEM770 series" is servo. http://www.compumotor.com/catalog_2002/pgs142_152_oem_servo.pdf
just reminded me that all those manuals are on the disk of a computer that doesn't boot
you can still get them off of compumotors site.
I pulled some of them off of there by booting with the emc liveCD and using scp, but it was a pain. for some reason I couldn't use scp
I have a nice collection of manuals, it would be a lot of work to reproduce all those downloads
bye, thanks for the help
skunkworks: 650 and 750's
Still trying to find a good breakout board, but haven't been looking hard lately.
JymmmEMC:What kind of machine you gonna be using?
why is it that, when i seemed to be having a wonderful "take you kid to work day".... my other 16 yr old kid was out getting charged with assault...... I hate days like this...
sorry just NEEDED to vent that
dmess: it's all good =)
dmess: I had a pretty shitty day last night at work too.
I completely understnad the venting aspect.
little shitt's been eatin' off the STUPID plant lately
what's work entail for you JymmmEMC:
nice looking machine
yes.. most guys just lay plywood or MDA for the table
dmess: One customer owe's us LOTS of money, so anytime someone needs to enter the bldg, we have to get authorization from the comptroller. I had to call him five times last night, Then one thing we've never done was get credit card, so had to make up an authorization form and have him sign and fax it back, so the first time he......
but theres no EMC2 option ... opps.. duhh
you gonna run emc or mach/