the datasheet for those parport cards show that they probably support the "X" mode for 9 outputs and 8 inputs on one port.
STROBE and other pins are shown as bidirectional, open collectors
is that a good thing?
for most people it doesn't matter
and for us ?
it can give you more flexibility in wiring switches
he has 3 ports.. shouldn't matter
well that's true
and he doesn't plan to use them.. even less :)
Well, I ahvne't installed the card, just bought it this morning
But I'll have it if I build another machien and go servos
ah, servos -- the dream of the CNC hobbyist
the holy grail
I thought that was finding a machining center for $200
including all the toolsing
hm I foundone before
but it was sold
when i called 5 minutes after it went on the net
Hey, an emc beowolf cluster in Real time!!!!
one cpu per motor is all we ask =)
anyone remember that RT system test thingy?
[01:49:04] <jepler> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?TroubleShooting
03jepler 07HEAD * 10emc2/src/Makefile: get rid of KDIR, we already had KERNELDIR in Makefile.inc
[02:26:19] <Jymmm> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?TroubleShooting
Speak to me, I want to see how the rt test with network activity =)
hi everyone im after some help i have been testing threading on my lathe but im getting different pitches with different spindle speeds im using g33 z-1 k.05 which i think should give me 20 tpi but its all over the place from 4tpi to 13 tpi the slower it runs the courser the thread getsany suggestions
harty: sounds like you might have a wrong spindle scale
how many counts/rev does the spindle encoder have?
short night this time :(
how do you mean?
was it only I that forgot about the perseids?
alex_joni: its a 1000ppr encoder and i tried changing the value in hal to see if i would change but no luck the thing that's got me stumped it that it changes pitch with different rpm
harty: I have no real experience with it unfortunately
is it repetitive?
and what emc2 version are you using?
2.0.3 and its repedative certain rpm's give me certain tpi
like 800 rpm gives me 13tpi and 100 rpm gives me 6tpi
harty: to me this sounds like a setup problem
could you pastebin.ca the relevant config parts ?
Lerneaen_Hydra: it seems some people were uploading all kinds of crap (php exploits and such to the wiki/uploads)
I changed some settings to allow only registered users to upload, can you test if you can still upload?
stepper pinout hal http://pastebin.ca/129007
[09:08:46] <harty> http://pastebin.ca/129008
this one has the hal and the ini file
that's not nice :(
harty: hmm.. why do you have Feed_override at .9 max?
alex_joni: same user/pass as before?
Lerneaen_Hydra: sure, if you didn't change it
any faster and it skips steps what should it be
seems to work
you can delete test.txt
harty: feed_override should always be 1.0, that way you'll really have the feedrates and speeds from the program
if it skips steps, then you need to adjust the values in [TRAJ] and [AXIS_*]
k will change that
alex_joni: upload didn't previously require user/pass?
Lerneaen_Hydra: apparently not
harty: it looks pretty ok to me, but like I said.. never used it myself
alex_joni: did the exploits do any damage?
harty: you could try to get cradek in here to help you, when he'll be around
Lerneaen_Hydra: don't really think so
alex_joni: ok. good
harty: cradek is the one who wrote this.. so he might know :D
he helped me set it up so i might try later thanks for your help
harty: can you output the exact command you are issuing?
03alex_joni 07HEAD * 10emc2/docs/.cvsignore: typo
alex_joni: the gcode im using is g33 z-1k.05 here is the gcode file text http://pastebin.ca/129015
it should produce a 20 tpi 1/2 inch unf thread i hope
hmm.. so you're running the spindle at 500 RPM?
even at 500 rpm it give me a 7to8 tpi
I mean that files sets S500 (spindle speed 500RPM)
I suspect there is something wrong with your calculations..
how did you get those values?
there is no spindel control on my machine so i get it close
I would also add an M3 after the S500
to actually command the spindle start
k is distance traveled per revolution to .05 is 20 tpi
i will add that and see what happens
whan running the g33 comand nothing happens until you start the spindle
are you sure you mean 20 tpi?
im not sure about anything at the moment :-)
then I'll try not to confuse you any more.. :)
.05*20=1inch if im working it out correctly
I need to run.. bbl
harty: keep bugging cradek, maybe he'll be around later
will do thanks
hmm, 0.05 inch/thread works out to 20 tpi (1/20=0.05)
'lo there harty
random question: why was emc moved from RTLinux to RTAI?
slow day today eh?
nearly bed time here
3 pm here
just gone midnight
Lerneaen_Hydra, simple answer, RTLinux was not ideologically pure or something ... to do with patents and the GPL and some other political stuff .. it worked fine, but there was some random licensing issues
eugh, I see
the rtlinux guys take a GPL kernel, add some of their patented stuff, re-distribute it and set conditions on what you can and cannot do with it
another purely practical issue: the rtlinux guys focused their efforts mostly on their proprietary version (that costs money) and the free (as in beer) version suffered as a result
neither version is free (as in speech)
doesn't that violate gpl?
non-free (as in beer) is gpl safe, but non-free (as in speech)?
there stuff isn't gpl
they made a proprietary patch?
the kernel is, and I'm not sure how they manage to merge their patches and the kernel
rather than basing it on gpl'd software?
was it a gpl violation?
all I know is that for a couple of years at least, the GPL world in general has preferred RTAI over RTLinux
that I can certainly understand
EMC2 actually uses something called RTAPI that is a wrapper around the RTOS
in theory, we can use either RTAI or RTLinux, just need to have a wrapper layer
does the wrapper eat lots of cpu or is it just pretty much pass-through?
in fact, EMC2 builds and runs on BDI-2.xx, which is RedHat 6.something with RTLinux, not RTAI
its pretty light
it certainly seems that trying to support rtlinux and rtai was more effort, as jmk says there is a wrapper layer, but, supprot for two things that do the same thing was taking up effort in places
it does some housekeeping - keeps track of things that can cause kernel oopses, and tries to clean up after programmers if they forget to free resources, etc
sounds like a good idea
before RTAPI there were hundreds of ifdefs all over the place to deal with different RTOSs
* robin_sz remembers that well
it was quite bad
like coding a webpage for a whole manner of nonstandard browsers
the current RTLinux version of RTAPI is for a rather old RTLinux, anyone wanting to use todays RTLinux would have to make a new RTAPI wrapper for it
it doesn't look like that's going to be a very likely movement though, I take it
bought that nice big linear slide for the tube laser project
£102 ... bargain
the encoder is almost worht that :)
linear slide + encoder for £102?
mmm .. 500 ct/rev
wait .. picture
[15:21:53] <robin_sz> http://cgi.ebay.co.uk/ws/eBayISAPI.dll?ViewItem&ih=002&item=120016762206
the ballscrew is a bit "fast"
I can imaging
but for the tube laser project, thats fine
with that thread
this wil be axis 3 ... across the tube
axis 1 is tube rotation, axis 2 is tube in/out
axis 4 is head up/down
any action today..?seems pretty quiet..?!
there was a short discussion of RTLinux vs. RTAI earlier, but not much since
hmm... do you know if someone successfully impemented a toolchanger with emc2?
yes I do
and it worked well
(the mazak config)
hmm... any web refs I can take a look at?
I think the configs are in CVS now, but I'm not sure
jmk or Ray may know
wow mazak project looks like a lots of work..
I believe that jmk committed the last of the configuration to the mazak directory at cvs.
there's even a video of it in action
If only i had a quicker connection.
I can mail you a CD :)
I saw it at the show.
then there's nothing to worry about - I haven't wadded anything to it :)
hmm.. there is only one cvs.. the files are 2 months old.. are these the ones..?
or added - I don't wad much anyway :
are you getting HEAD, or a release version?
[16:11:02] <ottos> http://cvs.linuxcnc.org/cgi-bin/cvsweb.cgi/emc2/configs/demo_mazak/demo_mazak.ini?rev=1.9;content-type=text%2Fplain
is it the head?
err - should be, I think
yep - CNC Workshop was at the end of May, right?
Yep head has it.
You probably won't be able to run it there.
It uses a Vital card, a parport card from PMDX, and an old ISA IO card.
But the .hal files and the cl files are interesting.
the big file: http://www.cncgear.com/Files/mazak.avi
You could move and start up the cl and see the ladder.
the small file: http://www.cncgear.com/Files/mazak-small.avi
yes that's what I'm after.. i hoep I can impement it om my vital card alone..
Sure you could comment out or remove the .hal references to the extra ISA card.
do you know what codec is the video running.?
whatever the Fiju E550 camera has :)
garh - Fuji
I was able to watch it on my Ubuntu machine
is it difficult to set up toolchangers?
(of course this depends on the mechanics)
hoping it's not that diff.. :D
There are many different kinds of tool changer setups. Each is a bit different from the other.
Biggest trick is to sort out each step in a tool change and define the sensors and solenoids that are relevant to each.
so the actual "programming" bit isn't all too hard?
once you know how the hardware is supposed to work, and how to tell if it didn't (or did), the programming is relatively easy
hey ottos, long time..
how do you do cubed or squared in oocalc
object oriented pen ?
ya.. been busy tunning the PID..
its not x^^2 or x**2
or something like that
oops - POWER(Base; Power)
ok - gotta run.
see you all alter
later, as well
nice.. toolchanger works.. another prob resolved.. :D
nice the large video works..
ottos: when can we see this? :)
i have one video .. but have to get the cd back..just couple of g0 with the old 5'axis..
still nice to watch I bet
can send you a copy once I have the cd back..
still have a lot to finish vfd with high speed spindle and the toolchanger.. I fuggure another 5 year and I'm done..
heh.. I'll be around ;)
ok gents I'm outta here.. ciao..
alex_joni, was it you that suggested some "light ubuntu" thing yesterday?
it's Ubuntu with XFC4 a lightweight window manager
darnit, can't find a shotshot of it
[17:07:20] <Lerneaen_Hydra> http://upload.wikimedia.org/wikipedia/en/thumb/f/f6/Xubuntu606.png/800px-Xubuntu606.png
ugh .. my trigonometry is rusty
just trying to work out the kinematics for the tip of the cutting torch on a piece of rectangular tube, with radiused conrers .. simple enough, but my trig was rusty
eventually, you have to lay it all out on a spreadsheet, and see if the graphs look liek a toolpath
im still trying to figure out how to describe cuts on a tube in gcode ... im sorta imagining the entire tube laid out as a flat sheet, slit up the back
Y0 is the middle of the top face
X runs up and down the length of the tube I guess
I guess what is needed is some way of describing the cuts, that recognises that the tube has rounded corners BUT doesnt get upset if you change the corner radius of the tube, ie a cut from the middle of one face to the middle of the adjacent face should still work, even if the program dosnt match the actual tube in some random way, radii of corners for example
which sort of implies using polar cooardinates .. but that just seems bad and wrong on reactangular tube ...
maybe cam would be better?
it eventually has to output a program for emc to run
imagine I move across the top of the tube
say its 60mm x 40mm rectangular steel
with a 5mm radius on the corner
60/2 - 5mm =25mm
so I start say in the middel of the tube, Y0 ...
move to Y-25
and now rotate the tube, following the radius, until I end up on what was the side wall
40/2 -5 =15mm
so I start off the rotation with Y=-25, A (the rotation angle)=0
I finish with A = 90 and Y=+15
I then go on to move Y to 0 again finishing the cut inthe middle of what was the side face, but is noiw on top right?
ideally, Id want some way of programing that, that was not dependent on the corner radius, so if by chance, the tube came in with 4mm rads, the same program would work
maybe it will look something like this:
still sounds to me like a simple python "cam" app would be better if you change input material a lot
no no no
the same "gcode" needs to be able to work with various tubes
so you can make sure the tip follows your actual tube without bothering the drawing office to rework the program for the shop floor
oh, this is a large scale thing?
with drawers and operators?
and operators don't code?
mmm .. not especially
that complicates things
they set the machine up and load the program
if the tube is not quite as ordered, ith should still work, reasonably the same
im not planning on using capacative height sensing on this, so all torch heights will be by "dead reckoning"
so, setting the machine up to exatly the dimensions of the actual tube is a must
sounds like you've got your work cut out
going to be fun
what if it's not quite round?
* alex_joni suspects you'll have problems on cutting bevels
Im expecting it to be rectangular
or the pipe?
oh, thought you want to cut round tubes too
round pipe is just a special case of rectangular
width = height
and corner radius = width/2
03rayh 07HEAD * 10emc2/docs/src/code/Style_Guide.lyx: copy code_notes to developers handbook.
03rayh 07HEAD * 10emc2/docs/src/Master_Developer.lyx: Added code notes appendex
im not going to thin about oval tube just yet :)
03alex_joni 07HEAD * 10emc2/src/po/Submakefile: added AXIS to the install target
03alex_joni 07HEAD * 10emc2/src/Makefile: added AXIS to the install target
03alex_joni 07HEAD * 10emc2/docs/src/Master_Integrator.lyx: added Master_Integrator.lyx .. please add info to it
03alex_joni 07HEAD * 10emc2/docs/src/Submakefile: added Master_Integrator.lyx / EMC2_Integrator_Manual.pdf to the doc build process
03alex_joni 07HEAD * 10emc2/docs/.cvsignore: new manual to ignore
scary isn't it
you've been busy
scary to write no doubt
Lerneaen_Hydra: it specifically says "please add info to it"
jmkasunich: not as much as I'd want
I'm going to be away for a week again (stupid training)
but I'll have a linux laptop, and if I have time today I'll load LyX on it
Lerneaen_Hydra: "Design for Six Sigma"
bleagh... corporate crap
them and their damn corp-speech
yeah, this is week three of four
jmkasunich: I don't envy you :/
hey.. that mazak. it look *very* familiar to the mill at my school
yay.. reached my 600th commit on emc :D
especially the toolchanger
alex_joni: thats a lot
jmkasunich: 1.91 years ago, for an average of 1.16 days between messages
1.16 days doesn't seem like a lot :)
well, its probalby more like a week between batches, and 10 messages in a batch
see.. a week is way far from a lot ;)
hi all, what's up ?
been thinking about doing something similar to halvcp with Python now that jepler made a python module that allows it
'lo there etla
L_H just downloading your movie...
is it an emc controlled machine ?
it's a mill at my old school with a toolchanger extremely similar to mazak
no, I doubt they would like me playing around with it ;)
it's an old (1980's) heidenhain mill
tnc 415 IIRC
Lerneaen_Hydra: good quality movie
I wouldn't have coolant run on the toolholder...
it's rather interesting becuase the original file is actually quite a bit better, and it was recorded with a tiny pocket sized digital camera
etla: that is probably the least of that machine's worries :/
it was generally mishandled all the time
and with flaky software too...
alex: is the current HAL_Documentation going into the new integrator handbook ?
etla: 2 different things
the HAL documentation will stay separate
but lots of it will go into the integrator handbook
HAL_Documentation is renamed to HAL_Users_Manual, btw
does anyone want some schematics for a DC servo controller?
similar to the G320
are they public domain ?
etla: I think so..
[18:58:20] <alex_joni> http://www.gertronik.de/cncecke/servo.zip
what is that? built in PID?
sounds like quite a biest : IRFP260N mit max.30A/150V
seems to have far too many components
Lerneaen_Hydra: it takes step/dir inputs I think
yeah. that would explainthe large board
I said like the G320 from gecko ;)
is there any real benefit to having step/dir from emc where a board does the PID bit?
compared to emc doing pwm?
If I could get some cheap good DC servos from somewhere then maybe it would be interesting to build a dc controller also...
(steptype 1 IIRC)
L_H: without PID in the drive you need encoder counting hardware in the pc
most people seem to prefer just parallell port for some reason
hmm, yeah. if you don't go too fast though the parport can do that
20khz, at 0.001mm accuracy you'd get 20mm/s -> 1200mm/min
which is enough for most people
of course you probably don't want to go over 1000mm/min
ok, but how many parallell ports do you need for 3 axes with A/B/index inputs each ?
1 output per axis, and 3 inputs=4 IO per axis
you should be able to squeeze all that into one parport
IIRC there are enough inputs for that
pwm for the motor?
oh, dir might be nice too
you probably need two - for an H bridge, you have PWM for each side
can either do left side and right side PWM, or PWM + DIR
heh. that would make it two parports probably
still, a parport is cheaper than special hardware
ok, well I'm well past the 'cheap' stage with this hobby ... ;)
haha, I see
I'm still stuck there
well if you count 1500 for a minimill, 400-500 for ballscrews, 100 each for servodrives, 100-200 for servodrive psu + misc parts then suddenly 200 for a mesa card or similar is not much
all of this spread out over some time ofcourse so it doesnt sting that bad...
uh, what was the logarithm function called when 10 or e isn't the base?
ln is log-e
yes, well log entails that the base is 10
and ln e
what about when the base is neither 10 nor e
usually the base is written as a subscript to log
so log_a(b) would be the base a logarithm of b
oh, so log_3 for example?
alex_joni you still awake?
well ... log (subscript n) has been used
'course my memory is fading at this point...
alex_joni: Do you know the diff between "desktop" and "alternate" xubuntu by chance?
Jymmm: desktop does the standard live-cd install, alternate does the install like in 5.10, text based
alternate also has better support for large network installs and so on
WEll, I dont' need cups, hp printing, or all the other stuff. (bells and whistles), so would alternate be a better choice, or no real difference
AFAICT only the install procedure differs
once it's been installed the system is the same
"The Ubuntu Alternate CD uses the text installer instead of the new GUI installer. It needs less system memory and permits advanced installs with preseeded options as well as LVM or RAID disk configurations."
I also found it a bit faster than the gui install
Yeah, I read that, but doens't tell be squawt =)
as it doesn't load all the live-cd stuff
the installed system is identical
better not be =)
how do you mean?
Jymmm: what LH said
well if you didn't need X as example
you can remove that later
or type server when installing
err. when you get the initial boot screen
and it will install the server base package
otherwise you can remove via apt/synaptic
apt is nice, but aint THAT nice =)
synaptic is nice too
apt-get remove xf86 is gonna leave residue
it's not xfree86 anymore ;)
x.org, but now I'm nitpicking
and really a waste of time to install when you know you're going to remove it before hand.
anyway, do a server install then
I just want better control of what's being installed. I dont have a soundcard, so dont install anything sound related sorta thing.
Jymmm: sounds like the old way (mid 90's)
alex_joni what you mean?
Jymmm: manual package selection
I used to be good at that back with SuSE 6.3
start with minimal system, add them as you need them
Why do I need a shitload of icons to volume control, cdplay, mp3 players, etc
Not the M$ way of installing everything
whether you need it or not.
Jymmm: yes, package selection would be nice
minimal installs are nice but everyone has their own idea of a min install ;-)
alex_joni: Yeah kernel only, no bootloader =)
alex_joni: Nah, I'm not *THAT* sadistic =)
oh.. too bad
03alex_joni 07HEAD * 10emc2/docs/src/gui/halui.lyx: updated with the latest pins
03jepler 07HEAD * 10emc2/docs/src/Submakefile: typo fix
* alex_joni heads to bed
* robin_sz returns
Anyone remember how to get X to rescan by chance?
you mean reconfigure by inspecting the hardware?
At least for video, yes.
I think that was it, lets see =)
* robin_sz finds something totally bizarre on porkolt.com