Jepler: What is the correct command to do build-essential? sudo apt-get build essential?
sudo apt-get install build-essential (I think)
jmk: thats doing it - thanks.
Hopefully the last question - Tried sudo comp --install millkins.c and got a FIle not found error. I did download the file from the wiki - where should I place it on my computer?
then go to that directory and sudo comp --install millkins.c
how bout ontop of the monitor ).... Actually - it is on my desktop atm - still not being seen though'
not being seen where?
are you in the desktop directory?
(its a hidden dir)
how do I get the terminal to point there?
I would not recommend using the desktop
open to suggestions
mkdir a new directory, and copy or move the file there
open the terminal and then type "cd ~/.Desktop" (without the quptes)
* robin_z ponders what quptes might be ;)
actually, it's not hidden, it's just called ~/Desktop
robin_z, they're similar to quotes
still gettin error
grr capitalization counts
needed cd ~/"D"esktop
yep, its case sensitive
Broke 3 mother fscking SS screws by hand! eeeeeesh
looking for help with opengl issues on my emc box - getting vertical stripes down the screen
have a deskpro EN Compaq
cannot figure the exact vid hardware - can't seem to find sysinfo command
lspci will tell you a lot
one of the lines will have the word "VGA" on it
along with other words that may be descriptive
ok - will give it go
intel 82815 815 chipset
I tried putting a outboard video on this box and all I could get was 600X800 resolution
what kind of video card? (out of curiosity)
found it in my junk box - cirrus Logic chipset - noname on it - holding it right now
if it's that old, it won't matter :)
sounds like I should put it back in the junkbox
(they had great cards in the ISA and some of the VLB era :) )
maybe even PCI
I think I still have the programming manual for one of their chips somewhere
this is a pci board
CL-GD5424 or some such
ah, 5440, that may have been it
ok, there was a 5464 chip, that could have been the specific one
but it's been 15 years or more since I've looked at that particular manual
no need to start )
oh, it was the 546x
(looking at the wikipedia page, that one looks most familiar)
anyway, there have been several issues mentioned lately with the Intel 810 or 815 chipsets
I don't remember the specific fix(es), but I think someone put what they found in the wiki
Aye - was searching gmane and google and that is all I saw - apears I have the same type issue though
tried searching the wiki and did not get any hits of interest even on the 810
can you put the contents of /var/log/Xorg.0.log on http://pastebin.ca
perhaps they did not add the notes yet
yeah, could be
or I'm totally wrong and nobody has found a fix :)
I'd like to see a log from the i815 chip, even if it's in 800x600 mode
also, you can try changing the resolution while at the desktop by pressing ctrl-alt-keypad+ and leypad-
you may need to use the left hand set of ctrl/alt keys
(I don't recall)
getting the log atm - brb
-o and -(-0)
[03:36:45] <bglackin> http://pastebin.ca/1324153
The keypress works - still have the lines (
ok, two issues here (of course)
I don't know how to fix - err, well either one of them
but I may be able to figure out the resolution problem
this is cute: (II) I810(0): Not using default mode "1024x768" (unknown reason)
what's the resolution of the monitor?
is this 8.04 or 6.06? (or something else entirely)
also, does the login screen come up at one resolution and then the desktop at another?
at max now probabl 1024X 1200 or somesuch
ok, please pastebin /etc/X11/xorg.conf
Hard to tell - I just plugged this monitor in (one for router is out in the garage
this is a bigger unit - resolution for login was at 1024X 1200 but when I got it - it dumped to 600X800 - was a bear getting axis aligned (and the reason its in the house atm)
getting xorg.conf brb
it looks like a 1280x1024 17" LCD from the X log
or maybe 18"
the login screen comes up in the first usable resolution, the desktop goes to whatever resolution it's set for (or it starys the same as the login screen if you don't set it)
[03:45:35] <bglackin> http://pastebin.ca/1324156
ok. what monitor do you intend to use back in the garage? (is it this one?)
viewsonic E70fb will be the one
work affectionately called them "Boat anchors"
well, let's try the easy way
I was happy to offload them )
with the correct monitor attached, hit ctrl-alt-F1 to get a text console
dpkg-reconfigure -phigh xserver-xorg
sudo dpkg-reconfigure -phigh xserver-xorg
you may have to do this first:
sudo /etc/init.d/gdm stop
but I think it may be done for you
note that tab-completion will help you a lot
sudo dpkg-r<tab><tab> rec<tab> -phigh xser<tab><tab> (or similar) will do most of the typing for you
never knew that regarding tab completions - new linux user (<6 months
doing the submarine strokes atm
it works in a lot of places (but not everywhere)
I have a second monitor of the type I plan to rehook so I am going to reboot and see if I can get through it
there's probably a basic tutorial somewhere on the system. it could help to familiarize you with some of the concepts that may be different from waht you're used to
swimming is used comp boxes and monitors atm
CRT or LCD?
crts aka boat anchors (by others)
yep, I agree :)
I call it a shop and kid's computer - and beats buying new stuff )
though my center monitor is still a nice CRT - itll suck when it dies
I have a E790B on this atm - 21" display - not too shabby
you have to spend a lot of cash to get an LCD that's as good as a reasonably good CRT
oh, 21. hmmm
your fonts will look weird
so will I if I can't get something working )
the display reports that it's 360z270 mm, which I think is 18"
<looks for a ruler>
I think 360x270 comes out to 18"
oh, I'm used to LCDs now ;)
that may be the visible area :)
15X11 - I need glasses and a perspective thats accurate
be back shortly
executed that command it it tossed me to text mode - reran it with a start command to get back
it should put in text mode
sudo /etc/init.d/gdm stop
yes, that would result in text mode
gdm is the Gnome display manager
lol - ok - no idea what to do then
I think you can't reconfigure the X server with the X server running ...
you log in at the text console (if presented with a login prompt)
ahh - so hit text mode and run the reconfigure then?
and then type those commands into the termina;
kk - brb
that's why I mentioned ctrl-alt-F1 to get a text console ...
(that makes it less surprising, since you did it yourself :) )
he might want to use screen and a text mode irc client while he's doing this
if I knew how to get him through that, I might suggest it
but since I don't use screen yet, I'm no help there
I really should, it would be very helpful for me
it ran and saved a"possibly previously customized version
screen is the bee's knees
did not clean up the lines but perhaps it will clear up the resolution issue
yep. I'm afraid you'll have to google for the i815 problem fix
and it's likely that the first thing people will suggest is to switch to a later distribution
if you can put your config infor on a USB stick or something, you may want to try the 8.04 CD (unless you did that without success)
ack - I can't just use the update manager for 8.04?
no RT kernal automagially right?
guess I should have done the 8.04 update before emc-dev install
well, you may be able to, but I don't know if it will jump straight from 6.06 to 8.04
along with the millkins (
it may make you go through a few in betweens (6.10/7.04/7.10) ...
if you're on dial-up, it'll be painful ;)
cable - it could be reasonabley painful
the CD is 700M, there are 300M or so of updates, and then there are the extra packages
(emc2-dev and build-essential)
aye - perhaps I should just bite the bullet and redo this box
good practice loading all that stuff over
incidentally, is the problem you're having actually skew?
and perhaps I can get my wireless working at the sametime (g)
well, reloading isn't the preferred way to fix things most of the time, but it may be best in this situation
what kind of woreless?
linksys G - the one that does not work automatically
a PCMCIA card?
is that a carrier with PCMCIA or an actual card?
looks like a carrier
there is some deal that several of the librarues do not automatically load with the RT kernal
I only ask because I had about the same thing for my old laptop, and the card hardware had problems
yes, you'll have to install linux-ubuntu-modules or some such (for the RT kernel)
there was an issue that prevented it from being able to connect to secure hotspots
this was under Windows or Linux
I think sullivan on the list found the solution - I asked him to post his notes to the wiki - havent checked for em yet
I think I eventually found out that the early version of the card I had was a defective design
regarding the secure hotspots issue - I think there was a worm that exploited the defect - my company's IT locked down all lappies hard recently - half the time - I can see the network but get a catch 22 on the handshake - and I travel alot so it sucks
with that card, it made no difference
I even used the Linksys wireless manager application when using Win2k
and this box ) Not gonna lug this thing around - not too mention my sneakernet is starting to wear out
burning the cable line now - Comcast's super highway appears to have been parked (1mb/sec verus advertised 6)
funny. they're supposed to have some boost thingy that makes the first XX amount go really fast, but it looks like they're throttling instead :)
I am getting ready to dump em and just get DSL - I do not do heavy downloading so no point and thier basic cable stinks - time for direct
DSL isn't all that speedy
I've been looking at cable instead ;)
as to your earlier question regarding SKEW
but Comcast is 110 a month now and they are cutting channels - all the kiddie channels and mine are still little - need those toons
direct +dsl =60 bucks
My gantry is cocked - not sure whats causing it - I want to rework the mahine but I have several other things I need to rebuild before pulling it apart
gantry is cocked?
it wasn't clear whether the offsets you mentioned were on the axis you moved or the other axis
when I pull it against the "-" X Axis stops - one side of the gantry hits .25" before the other
why is that a problem
my gantry does not wrack at all so it sould just be along the Y and uniform
ok, so as you move the head across the gantry in Y, the X position changes
that's a problem.
its a hobby machine for woodworking so less of a problem
is one of the legs of your gantry free to unbolt so you can tap it in
I could probably shim one side - but I figure if I do the corrections - it should work just as well and takes less measurement time
do you have home switches?
No - stepper machine - was planning on limits - just not enough time
limits and home that is
what do you mean by corrections, specifically
you must have home switches to make non-trivial kinematics work
gantrykis needs to know its absolute position along the Y axis
using the millkins module - it will do a correction to take out the skew
right - that one ;)
ack - looks like at least one switch is in my future
so it's going to move Y to correct for x?
it may be easier to tap tap :)
my experience is that you're going to get a ratty finish if you rely on that kind of error correction
i would humbly suggest taking the hour and shimming or tapping it in
hmmm - wasted enough time already - guess it can't hurt
only because very few machines have the step resolution to make it work properly, most machines move in quantized steps, not a smooth line
if it won't tap out - no wories on the finish - at least for now since I am just building stuffy for the kids out of wood and mdf
of course there's a valid rebuttal in that i could be nitpicking for most folk's purposes
the machine has that resolution anyway, so it doesn't matter too much
well -to your favor - had I spent an hour today - I would have already finished it and not be here waiting for another distro to download...
but it does get me to tune up this box more )
while you're waiting, you could fix it
garage is 18 degrees - heck no )
speaking of tuning. I am using the hobbycnc baord - it uses Allegros' sla707X version chips - I found a spec sheet - what do I need off it to figure the base period -
not sure how to translate the jargon they use with what is described in the integrator's manual
does the board have optoisolation?
(that can change timing)
I believe it is optoisolated
ok, then you need to ask hobbycnc
or take some measurements with a good scope
Queries to Hobbycnc were met with little response - the canned response they give (mainly Mach users) is to turn down velocity and acceleration
well, you can try to use the spec sheet numbers, but maybe add a little to wach measurement
I could probably get a buddy at work to scope it for me - he's heavy into instrumentaion and RT control (cement palnt stuff)
what happens with optoisolators is that if they aren't symmetric in turn-on and turn-off, you need longer timing to account for the difference
know Bob Dawson?
Nay - I am new to the industy - jumped into it 2 years ago
and I work on the dirt side - digging rocks
Bob was a professor of mine in the late '80s. he wrote some concrete plant control software way back when
originally got into this trying to figure a way to cut terrain models
Perhaps he is downstream of what I am in - Concrete is the batching plants - cement is the burning process that makes the powder for the concrete plants
his stuff was used to control the mix recipes
I waste too much time teaching upper managers how to read topo maps - I was looking for a way to translate them quickly into a model
Cool - similar to batching software for asphalt plants then - I used to deal with some of that stuff
I think he got the second software patent (ever) on that process
last place I worked the IT manager cried everytime a batch computer crashed - he would take a new box and dump MSDos 3.10 into it
everything was dos based and still is to my knowledge (at that operation
nothing wrong with that
very stable - just not fancy - stable is good
comcast is cooking along at 9.1kbs - glad i am spending 100 bucks a month on "high speed" internet
SW - what was command to provide hardware details?
that's one of them anyway
one of many =)
Night - thanks for the help SW
good mornig.i have sometimes the error message axis: encoder.00 <= dt how you can this....This allarm is present sometime for axis 00, axis01 and 02. because?
afternoon all, would someone here have a simple example of closed loop spindle speed control? I have a VFD now hooked up and of course an encoder on the spindle, but I'd like to see some examples of how others have done this
pjm__: some of the sample configs have that
[13:55:14] <JymmmEMC> http://failblog.files.wordpress.com/2008/12/fail-owned-price-tag-placement-fail.jpg
pjm__: I belive the stepconf wizard will do that automagiclly
BigJohnT: assuming you connect the encoder to parport
if he has some other hardware to count the encoder, than he'll have to figure it out
pjm__: did you look in the Integrators manual?
BigJohnT yeah i have that, i have the encoder connected and linked to a bargraph display in axis and it reads the proper spindle speed
but i will do some further digging
pjm__: here is how I did the GS2 example http://www.linuxcnc.org/docs/devel/html/examples_gs2_example.html
and this is the pyvcp part http://www.linuxcnc.org/docs/devel/html/hal_pyvcp_examples.html#sec:GS2-RPM-Meter
BigJohnT: I think he already has that
ok, didn't know
or remember :(
ok thanks, well i shall read and digest!
pjm__: hang on
you just said you have the spindle speed in a bargraph in vcp
ok.. there is not much else to do
so this was just to check if I could do it first, it only uses a 4 ppr encoder tho
something i made quickly
you can use the at-speed input to the motion controller
that way you can have some movements wait for the spindle to reach the commanded speed
for example transitions from G0 to G1
ah yes interesting
another interesting thing to do is to do rigid tapping (if mill) or threading (if lathe)
so for example what i'm aiming for is say i command 3000rpm and take a cut, i want to keep at 3000rpm under load
yeah rigid tapping is on my list, not sure on the encoder res needed for that, i think 4ppr is too low
i have a nice heidenhain 125ppr that may do, this is using parport tho, nothing fast..
well for the first part you probably want to implement some PID
motion.spindle.speed-out -> pid
yes pid's will be a new learning area for me
encoder.speed -> pid
pjm__: probably the closest is looking at some servo configs
ok will take a look and digest
what do u recommend for spindle encoder ppr for reliable rigid tapping?
sorry.. no experience
pjm__: The "best" spindle encoder would be the one that has the highest resolution possible, but just low enough that it will not be above its speed (frequency) specs at the highest anticipated RPM. Does that help?
pjm__: what's your max rpm? and the current BASE_PERIOD ?
fwiw the mazak has 1440 counts/revolution according to its hal file
yeah, but it doesn't use parport to count it
Now I have a Q: Is there anything that I have to do in AXIS to set up the perspective view? I have a g-code program that shows up fine in ZzXY but is blank in P. I can go load chips_3d and it shows fine. Reload my g-code and I can't see anything in the perspective view, except for the XYZ numbers changing. I've tried zooming out a few steps. Any ideas?
KimK: pastebin your code
OK, here you go. http://www.pastebin.ca/1324548
I wonder why you have all those (AXIS,show) comments
Just because I'm writing it by hand from Qcad and fooling around with it. The plot gets pretty busy-looking when everyting is enabled, so sometimes I hide parts of it while debugging,
indeed, removing those I see something
Oh, did I miss a "hide"? Oops.
I had to make multiple changes -- getting rid of cutter comp too (presumably my tool table doesn't match kim's)
but you said that it was different in perspective view from other views. are you prepared to recant that now?
KimK: I don't think you missed a hide..
no, it's all (AXIS,show) comments in that file
No, but I can restart AXIS to be sure, I think I did already though. I have run the code, no errors, I can push ZzXY and see those plots just fine. Nothing in P.
but the show/hide feature was written by a programmer
it turns out that if you write (AXIS,show) without a prior (AXIS,hide) then your plot disappears
I guess that rates a patented #emc "heh". "It's impossible to make something foolproof because fools are so ingenious."
axis keeps track of a quantity called "suppress", and it's always 0 at the start of a preview. When it encounters (AXIS,hide), it increases "suppress" by 1. When it encounters (AXIS,show) it decreases it by 1. If the value is not zero, then the preview plot is hidden
if used as intended, this works just fine
but "as intended" includes "never write show without an earlier hide"
maybe you should check > 0, not !=0
that will still be odd if you have 3 show's you need 4 hide's to hide things again
why not make suppress = 1 when it finds a hide, and 0 when it finds a show
at some point you have to say to the human: suck it up, this is how the program works
alex_joni: because they should nest
you think someone actually does tha?
(sorry.. my spelling is crap..)
imagine you have some subroutine you don't want to see, so you put (AXIS,show) and (AXIS,hide) around it
I bet you can hide it even without (AXIS,hide) :P
now you want to also hide part of your main program
yeah, I get the point
maybe it calls the subroutine, maybe it doesn't; with my scheme you don't have to worry about it
then I guess it's a job for BigJohnT to fix it :)
fix what? who me?
From a programmer's standpoint, why does the "supress" value affect P display differently from ZzXY displays?
KimK: it doesn't.
Why doesn't it hide all or none?
KimK: there's something else going on that we haven't discovered yet
OK. I'm happy to help if I can
BigJohnT: yeah, the docs
specify that you _always_ need to use a pair of (AXIS,hide) <-> (AXIS,show) and in that order
I was relying on the very bottom of the page http://linuxcnc.org/docs/html/gui_axis.html
well... it does say "Anything between the (AXIS,hide) and (AXIS,show) will not be drawn during the preview."
even though it won't be surprise-free for users, I made alex's proposed change
the > 0 check
jepler: that, and a small docs fix to emphasize that you always need to use the pair of hide/show ..
should take care of most surprises
Would there be any advantage to having (AXIS,show) be like G40, you can say it as often as you like?
no. stacking in one direction only is the worst of all worlds, and I want it to stack in the "hide" direction.
G40 doesn't "stack" in either direction
Stack? As in subroutines?
As in multiple invocations?
if you 'hide' when you're already hidden, it's not an error; but the preview doesn't reappear until an equal number of 'show' are encountered
I think of this as nesting or stacking
the motivating example I gave earlier also involved gcode subroutines, but that's not a requirement
And what will it do if you 'show' when you're already showing?
you mean after the change I made on TRUNK?
or do you mean now, in 2.2.8?
I was just talking about the complement to your example above
in 2.2.8, if the net number of shows minus hides is nonzero, it hides
You said "if you hide when you're already hidden..." I was just asking what if you show when you're already showing
so if the only thing you have is an '(AXIS,show)' then everything subsequent is hidden
now to me that's nonsensical, like writing the following and pretending it's algebra: )x+1
so I didn't write the software to do anything useful in that situation
actually, your part program said something like this: )x+))1
it's nonsense, and so you got garbage
I see. I think of the shows and hides as start/stop buttons. You should be able to push stop if it's already stopped, and the reverse. But you're the programmer, so OK. We'll "suck it up".
yes, the problem with software is how it's written by people who think like computers
Any thoughts on why I can still see ZzXY ?
Shall I take out all the hides and rerun it? (Maybe I'll do that...)
Oops, I meant shows
ah, what you didn't tell me is that what you see is not the preview plot but is the backplot
axis sizes the perspective plot according to the extents of the preview plot. when there is no preview plot, it uses a default that is bad.
that's fixed in TRUNK, by the way
Taking out the four shows seems to have restored the perspective view. Both preview and backplot.
Since I have the opportunity to ask, will there be a 2.2.9, or is 2.2.8 the end of the 2.2 line due to the coming 2.3?
KimK: if there are bugs in 2.2.8 we'll fix them in 2.2.9
there are 2-3 minor things in v2.2x (after 2.2.8) atm, but nothing to justify a new release
OK, I was just curious what you developers are planning for us mere mortals
(in the past we didn't have new versions after a enw majour was out.. but that's not a rule)
depends on you
list of 2.3 features, subject to change: http://cvs.linuxcnc.org/cvs/emc2/debian/changelog?rev=HEAD;content-type=text%2Fplain
Quite a long list there. There is also this (a summary?) http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Emc2.3Status
that's a todo list for 2.3
the things in the changelog are already done
alex_joni: look at last mail at emc-list from brian, is there any warning to worry about ?
no, I don't think there's anything to worry about
it's just that it's a more modern compiler, that is easier to freak out :)
Greetings all - reinstalled 8.04 and emc2-dev - added a millkins component (not activated yet and trying to fix a few last items before packing this back to the garage. I checking screen resolution, my monitor is not recognized? Any suggestions?
I did stop GDM but my computer hung and I could not access the text terminal mode to reconfigure
if anyone is looking,the log and config files are uploaded - xorg.0.log at http://pastebin.ca/1324636
and xorg.conf at http://pastebin.ca/1324640
bglackin: switching to a text console via ctrl-alt-F1 should always work
(before you try to kill/stop gdm)
looks like the monitor is read just fine to me
I need to find a list of all the shortcut keystrokes )
it picks 1024x768
what's the symptom?
perhaps not any - went into the reolution settings under preferences and saw monitor unknown - perhaps I am worrying too much - last eve this same setup had the hashes running across it must be a little twitchy now that that is gone with new install
so it's working fine?
my monitor shows as unknown too. Some sort of programmer incompetence involved because the driver knows what it is
Cr - yes - it appears fine - just unecesary worry on my part perhaps. Thanks for the clarification Eric
(II) intel(0): EDID vendor "VSC", prod id 8209
last Item - my linksys wireless G card does not appear on my network connection list - I have added all the restricted rtai modules
sure, if you think it's incompetence that someone didn't know what vendor VSC id 8209 is
card is PCI wireless
what do you mean by all the restricted rtai modules?
bglackin: does it work on a regular kernel?
there is linux-ubuntu-modules-2.6.24-16-rtai
I don't have the actual name - I cut and pasted it from an older Alex Joni email to the list - Added it in the hopes it would show up
cradek: good point, but unknown monitor is less than helpful
thats the one
seems like wireless is not going to be a good idea with a rt kernel
bglackin: there is also linux-backports-modules-2.6.24-16-rtai
eric_unterhausen: why not?
non-rt safe modules?
I didn't understand that..
bglackin: did you install from LiveCD ?
AJ - Yes
hmm, ok.. then it's a bit harder to test
one thing you could do is try a regular Ubuntu 8.04 LiveCD and check if wireless works
if it does, look for the module (driver) it uses
New to linux so not sure how to switch kernals
bglackin: that's why I said it's a bit harder to test
but you can download a 8.04.2 CD from Ubuntu.com, and boot with that one
looks like I need to invest in a new pair of converses
if wireless works, then we can look what's missing from the -rtai kernel
I'll burn a copy and try it out
bglackin: if wireless works, get online and go to pastebin.ca
open a terminal on your PC, and type dmesg
put the output from dmesg on pastebin.ca, and we can see what goes
running it now - brb
[18:24:10] <bglackin> http://pastebin.ca/1324658
is the dmesg dump
that was fast
bglackin: not the right CD :)
I am learning
that's our LiveCD
I said you should try the Ubuntu LiveCD
ahh - I thought you meant the Linuxcnc Livecd
sorry if I wasn't clear
I did understand - just got wires crossed - gonna burn the non emc distro now
btw - I am on a hardwire with the box right now
allowing me to work over the net
that's always better if available ;)
Aye - Pulled the machine in last eve to get it sorted out
deciding whether to move everything into my basement - would certainly be warmer
I will report back later - download is crawling along on this "high speed connection"
anyone out there good with AC 3 phase drives?
got a 10hp dirve that runs a plunger type machine.
good on the up stroke, but on the back stroke the motor becomes a generator and oveloads the the DC bus.
think i need either something to suck up or burn off a lot of energy.
external braking resistor
resistors or capacitors? not sure.
but big and hot
Haas uses electric stove elements
there is a place on the drive to attach
on almost all drives
gotta figure the actual amount of energy.
there are terminals straight to the dc buss on the drive.
how about diodes between the buss and the resistors?
there should be a braking resistor internal to the drive which you then bypass
there is a braking transistor
yeah, could be. not sure. gonna speak with ACTech tomorrow.
look at an0008
application note? 0008
uses a programmable output, looks like a relay and a braking resistor may be needed, hard to say without more thought
hope it is an external option. i don't want to have to replace the drive. too much $$$
the AN0008 mentions an external module
I'm confused.... How can a 10HP 3PH AC motor muck up a DC bus?
most AC motors are driven by a servo drive
those are usually powered from a DC bus
what do you mean?
and it sounds like BEMF is what he's dealing with.
which is usually "balanced" out with the energy going to other motors
the motor is powered through a 3 legged H-bridge directly from the DC bus
what if he only has a spindle?
then there are no other motors to "balance"
otoh, I think it's unlikely to think that other motors are to balance things out
usually the motors start/stop pretty together
I'm only speaking from multiple stepper drives as Mariss had explained it
unless you have a multi-machine DC bus, which is uncommon
get a motor from the junk yard and use it to turn the garden gnome ?
at 10HP it's gonna be a big gnome
no doubt =)
Mariss had some schematics how to dump BEMF into resistors
that's probably ok to do on his drives
I would be scared to do that myself on bigger servos though
Ok, garnden gnome, pink flamingo, water fountain pump, wooden orinmental windmill, and porch frog gribbiting! How's that?
sounds better :)
the servos we use take 380~ and rectify it to about 560V DC
that's the servo DC bus, which powers all drives
@ 10HP that would be like a crap load of 500W resisters
or like someone mentioned stove element
they're already rated for 220VAC
and gawd knows they suck up the juice!
Maybe use them to heat up lunch or something =)
they're sure nice in winter
yeah, could make some kind of radiant heat thingy and a fan
good night all
I am pretty new to EMC2, just did a quick simulation with the liveCD yesterday
Did anybody try to run EMC2 on an embedded CPU ?
such as ARMè/9, AVR32, ... ?
not a lot of people around here ...
don't think anyone here has done it
looks like the Atom will probably work
there are people using ARM processors with RTAI, but that might not be enough processor for step/dir generation
so the limitations are more on the processing power side ?
EMC does most of its work in floating point - on a PC, FP is just as fast as integer math, not sure if that's true on ARM and such
the biggest hurdle is probably building the realtime kernel for your platform of choice
that's true, often the floating point is much slower on these processors
the kernel does seem to be a bit of a project
so EM2 is not dependant on CPU architecture, given that sufficent floating point computation power is available ?
I wouldn't go that far
you would be treading on unexplored ground, I'm sure there are some traps
it's probably fairly close to being hardware independent as long as RTAI runs
I ask because it is not easy to convince some people that a PC can belong to an actual workshop
and they would be reassured if an embedded system is used
use an embedded PC ;-)
PC-104 or something like that
my opinion of embedded pc's is in the tank
too much wasted time
embedded bios are very badly tested
the pc ones
speaking of which, I should stop talking and start milling - we're having a warm spell, and I'm way behind on making gears
and PC104+ is a nightmare
and I understand laptops are not ideal either ...
not for emc
the problem is mostly the energy saving nonsense
ah yes, I read something about latency spikes on the doc
different question : the parallel port driver seem quite easy to use
is there an advantage to use dedicated control boards, ie. for steppers ?
one hardware I have in mind would need up to 4000 steps/sec
is that doable with the parallel port ?
I did the latency test, and it said 15 microseconds jitter, looks doable
I have 8us and am doing about 10k steps/sec right now, your results my vary
JustinXJS2: what mobo?
yeah 8us seem very low :)
this computer has done that, but I'm not going to run a mill with it
I was in the 14us range, until I disabled the onboard LAN and pulled the wireless card
do you have an add-in lan?
so far the motors are just on the bench, but I am getting 2100RPM out of them with 100% accuracy
no, no lan for now, I have more than one computer in my garage, so it is no big deal
I should try that
I was shocked how much it lowered the jitter
I bet one of those 3com server cards would be good to put in the machine, I bet they would have less effect on the jitter
it is good to know anyway
yes, basically my machine has all the hardware that is not being used disabled
and I have a matrox videocard in it now, I had bad luck with ATI
putting a matrox g450 didn't help my jitter at all
P650, it was the only matrox card I had laying around
thinking about putting the nvidia back in
some people suggest that digging out an old pci millennium is the best choice, never tried it myself
and same for unnecesary software, you disabled extra running programs ?
haha, i have not seen one of those for years
well, I am just using the default install off the livecd so far, I have not changed any of the software yet
does the number of CPU cores help reduce latency, or it doesn't matter ?
It sounds like it can or cant, depending on the chipset
so it is not clear cut
from my understanding, no, but I am a newbie at all this
it can increase latency if cpu affinity is not setup properly
I also have an old robotic arm, that was designed for teaching uses, 25 years ago
6 step motors (5 joints + clamp)
driven by parallel port, but with a strange protocol
I wondered if EMC2 is modular enough to accomodate this ?
pic of the beast : http://michel.roemhild.free.fr/IMG/jpg/image-youpi.jpg
it doesn't have a "strange protocol" module, but you could write one... you would be better off converting it
it may be, but you would have to write a component that turns emc's position commands into whatever it needs
what block would that be ?
emc produces commanded positions every servo-period (typically 1ms) .. most people use that to drive a step generator or a PID loop
you'd also have to write kinematics, which describe how to turn cartesian positions into joint positions
I saw a Puma example, with 3D model
I should be able to modify it to suits my needs
[22:47:04] <jepler> http://www.linuxcnc.org/docview/html/motion_kinematics.html
this document describes kinematics, which translate between cartesian and joint coordinate systems
yep, I read that
jepler: Thank you for the work you did on biarc's!
ugh, is this the system for communicating steps with the robot? http://youpi.forler.ch/prog.htm
a system which is designed to move only one motor at a time is going to work very badly with emc
jepler: I don't know if you saw my post on the emc lists, but I've been hacking on your inkscape exporter a bit lately, and incorporated your biarc.py script into it
jst_home: yes I did see it
jst_home: nice work
jepler: Thanks! Took a while to get my head around a couple of things, but it sure seems to work nice now
I don't think I'll be doing an updated version soon, but if anyone asks I'll point them at the mailing list posts
@jepler: what has to be done separately for each motor is only the direction choice
the steps can be done for all 6 at any time
ZbuffeR: ah, rusty french + only skimming, I am not surprised if I misunderstood
jepler: Feel free. I'll probably be posting more as I make more progress
sorry, have to go
I am still not sure where in EMC2 I would plug my custom step generator module ?
anybody can point the relevant part in the documentation ?
do I have to build a custom hardware driver such as the ones described in Chapter 12 in the Integrator Manual ?
or would that be a HAL component ?
brief update on loading a wireless card on 8.04
it appears that the card I have (linksys wmp54g v4.1) needs a restricted driver and firmware
therefore getting it running is at my risk....
got the firmware (search ralink in synaptic) loaded - still struggling with the driver portion
New sneakers are looking good atm
did you try the rt2x00 ?
[23:23:01] <ZbuffeR> http://rt2x00.serialmonkey.com/wiki/index.php/Hardware
ok bye guys, time to sleep
jtr_ is now known as jtr