[Global Notice] Hi all. We're having issues with an attack. We're restoring a previous services backup. Please bear with us.
[Global Notice] Hi all. As you are now aware, services are back up. This would be a good time to change your channel password if you can. Please check with first level support (/stats p) if you need help, and bear with us, as we might be a bit bogged down. Thank.
[Global Notice] This would be a good time to mention that freenode needs help with out 'freenode-registry' project, which would certainly help us avoid some compromises and recover more effectively. If you're interested in helping out, stop by #freenode-registry. I may have spotty availability for the next few hours or day, but will want to talk to you personally. Thanks.
K`zan: normal threads or a real screw?
[Global Notice] (and by the way, that channel name is #freenode-registry without the dots, thanks!)
anonimasu: Err, not sure what you mean, it is a 12" length of 1/4"-20 "all thread" rod I picked up from Home Depot. Normal threads and a real screw (esp at those prices :-).
1/4" 20TPI threads
hm, no idea..
normal threads.. such as fasteners..
err as used..
thoose threads are crap.. :/
they take away 40% of the torque..
Take away torque? How?
yeah you need precision ground ballscrews made from solid silicon carbide
This was a bust :-( : http://wrlabs.shacknet.nu/~vw/CNC/videocamview1with+2lens.png
not really, but proper threads..
fenn: I might need them, but this is what I can afford to work with for the moment...
Yeah, it sucks.
rolled threaded rod is fine
I got an idea of making another and then using the drill press for the Z axis...
how will you automate the quill movement?
Still need a driver though, unless I want to try to program the AVR to handle it, which strikes me as a BIG job.
yeah, but not the threads used for fastening stuff..
Same device would work for the quill, just replace the handle with the stepper ?!?
anonimasu she is using steppers scrapped from floppy drives
First off get rid of, or at least reduce the tension on the quill (heluva spring on that).
fenn: I can see the budget perspective, but I've tried running with fastener thread in a prototype at work..
you will need some sort of spring to take up the backlash in the quill rack and pinion
fenn: the 1000$ servo stalled ;)
yeah cause you had it geared for high speed operation
Working on a holder to contain two nuts with a spring between them.
Unf, gorilla glue wouldn't hold the nuts to the wood frame :-(.
Afro engineering at it's finest here :-).
anonimasu: well, my threaded rod machine works fine
fenn: how much torque are you wasting at moving your machine?
probably about 50%
fenn: wouldnt that torque be better to have while cutting ;)
Looking at the positioning, it is rather amazing that as close as I can see using pencil marks, it is right on.
yes but the point is it works, and i didnt spend hours and hours worrying about drive efficiency
Ran it back and forth a lot last night in one inch increments and it comes back to the pencil marks every time. At least until I reduct the inter-step delay too much.
I have GOT to get a drive together...
K`zan: what rpm does it lose steps at?
Byte the Bullet and just get the HobbyCNC 3 axis one.
fenn: Dunno what the RPM would be, but lemme see, when it gets down to 6 second/inch value, it can get nasty in terms of just buzzing.
Not always BAD, but it would seem any would be unacceptable.
I need to run back up to the thrift store and grab the 35V/2.5A supply...
fenn: I am not worrying about effiency, but if you have small steppers ;)
Wall wart :).
so... 0.166 in/sec and .05 in/turn = 0.
* anonimasu really needs to set up some avr dev stuff
Looks about right, I changed the value to 10 sec/inch so as to be "safe".
anonimasu: Run linux ?
If not even winavr is decent if m$hit holds together.
I run windows on my workstation since I use it for work
The new AVRStudio deals with the avrgcc stuff quite nicely.
and linux on my ircbox..
so winavr it is
I were talking about the hardware..
Got it running on both and while I loathe win$hit you probably won't hit the box hard enough to break it.
There is a new (V5) avrdude that works nicely under linux and wincrap.
* anonimasu has a stk500
Works fine with that also :) -c stk500v2 if you have done the recent upgrades to the V2 firmware.
and well, I cant progeam anything from this box either way..
* K`zan need to make up 2x10 to inline to get stuff off the stk500 to the protoboards...
need to use the workbox for programming
since it has serial port..
err has a
It will do, just back up early and often with m$crap.
and a parallel port for programming plc's..
Not that that will help since you still have the registry crap to deal with, but you will feel better :)
I dont think my workbox has crashed in 12 months..
Hummm, all my machines have at least 1 parallel and 2 serial ports ?!?
Some multiples of that.
K`zan: cant really do fieldwork with thoose boxes..
[Global Notice] Hi all. Just to make sure everyone knows, we are reasonably-certain your passwords have not been compromised, but password changes are a very prudent precaution at this point. Thanks!
I use m$hit as little as possible anymore and the less you use it, the longer it is between wipe/reinstalls...
Been almost 2 years now and it is getting to be about time.
I use 3 apps on my workbox..
Used to be 6 months between when I was actually trying to use it for real work.
I find it to work well for my use..
Hate that registry SHIT with a passion that denies words...
Although there are a number of things (m$fs) that will run unregistered...
I don't mind so much re-installing but it gets to be a real PITA to re-install all the apps.
[Global Notice] Specifically, for your information, services passwords are hashed, and no direct access to our database files was involved in the attack. So any possible compromise would be in the form of changed passwords. Thank you!
03paul_c * 10emc2-auto/wiki/ (diff_log keep/E/EmcIni.kp page/E/EmcIni.db): "Auto update wiki from a cron job. Sat Oct 15 01:15:02 BST 2005 "
* mshaver is back
* mshaver is back
* mshaver is away: mshaver
"Many introverted hackers who are next to inarticulate in person communicate with considerable fluency over the net, perhaps precisely because they can forget on an unconscious level that they are dealing with people and thus don't feel stressed and anxious as they would face to face."
Hang them all and charge for tickets...
* K`zan wonders if I am the only one who will have the HobbyCNC kit...
i've got a hobbycnc board
i didnt solder it though.. but i imagine you are more than competent enough :)
hi everyone :)
I've been researching the linuxcnc.org project and like the look of it... is there someone here that knows a bit about the source code and how i can set up a debugging environment for it and maybe develop in the future?
i'd like to get my head around the code a bit more first of course
Not a emc programmer, but hi anyway
From all the responses it looks like I am going to be the only one with a HobbyCNC driver board...
Moot point anyway:
The minimum is just that. Running steppers rated at 400ma at 500ma will
quickly overheat them and demagnetize them in short order.
Back to the drawing board...
I assume from that response that the L297 doesn't go below 500 mA.
K`zan: the L297 will give you whatever current you want - you just have to choose your sense resistors and reference voltage accordingly
That was the impression I had, but according to hobbycnc, you can't adjust it on their board ?!?
I'm sure you can "adjust" it by changing the sense resistors if nothing else.
of course I have no idea what drivers they use, but the basic idea of a chopper is to use a feedback loop to regulate current
it's all determined by those sense resistors
Aha, check this:
[05:11:11] <K`zan> http://www.hobbycnc.com/driverboards/4aupc/X%20Y%20axis.jpg
I had assumed it used 297s, but that doesn't match up # wise.
no, since it microsteps, I think
just a thought, has anyone here ever tried using EMC on a distributed control system? i.e. where the control loop for the servo not being done by the emc program?
SO, looks like that isn't going to work for me, perhaps it will for Jim.
He is going to have to spend a chunk on servos.
to drive that Taig
err Big steppers.
I have the build running with a couple of problems.
I have a .ngc where I set feed to 16 and move, then set it to 2 in a g2 line.
After that I can't seem to set it back to 16 to move to the next cut
can you paste the relevant lines?
g1 z.1 f16
#1 = 0
#2 = 5
Up to the f3, back plot is scooting right along
g2 x[#1-.127] r.127 f2
g2 x[#1+.127] r.127
g2 x[#1-.127] r.127 f2
g2 x[#1+.127] r.127
g1 z.1 f16
#1 = .7175
g1 x .7175 y-1.7323 f16
When it gets to the lines with f16 (or 10 earlier, it doesn't speed up
I think g1 #1 is a syntax error
There is much more to the file, but that is enough to demonstrate
yeah, sorry, I should have asked you to send it to me
In that case, what would be the correc syntax for use of a variable in that context?
can send if you wish. Seems to run fine including the change of center for the circles, just no feed speed change
sure, send it over
actually, I don't understand what you want that line to do
On the way
looks like for these steppers I am going to have to figure out what is coming out of the parallel port and use that to command the steppers through the AVR...
What I have is a series of 16 indented circles.
Instead of editing the code and offsets for making the circles, I found where I can use a variable to hold the circle center
g1 x #1 should move x to the position held in variable #1
I get line 311: unexpected end of file until I add m2 before the %
It would have been nice if the gcode designers had read postscript or at least basic when they did it.
JimHurlburt: you could have also used relative coordinates, or a coordinate system offset
Didn't seem to bother here, but I can add it.
Verify didn't whine anyway
I agree that f16 on line 31 doesn't seem to do anything
(why do you not just use g0 there?)
My sherline mill doesn't seem to like g0, sometimes the steppers just sit there and quiver.
I think there may be a line in the config file that will slow g0 down a bit but haven't dug for it
Thanks, will adjust that one tomorrow. However, f16 should work after an f2
Anyone know of a reference to the parallel port signals for each axis ?
To change the subject a bit, I found and downloaded axis last nite.
JimHurlburt: have it working yet?
definitions I mean. What does the step expect and what does directions expect.
K`zan: step is active low
After a bit of fildling got it to start to run, but it segfaults and dumps out
cradek: that commands one FULL step (1/8th turn) ?
Would be quite willing to help tinker but it would help to have some idea where to start
not even a python error?
K`zan: depends on your setup.
The way I have this set up now when I hit the step switch, it makes one full turn.
How one would keep track of that might be interesting.
K`zan: that's not what you want, you want it to make one step (the smallest rotation possible)
I'll stop the run on backplot -- still no joy on fast feed and re-set if or axis and send the error message
cradek: ok a step and not a turn. So I will need to keep track of the last step and direction taken to deal with the next command. Not difficult,
K`zan: hook step to an interrupt. When you get the interrupt, read dir, move that dir in your state table, output it
running EMC DISPLAY PROGRAM -- axis...
Xlib: extension "XFree86-DRI" missing on display ":0.0".
./generic.run: line 979: 16868 Segmentation fault plat/$display_plat/bin/$display -ini $INIFILE
can you run any GL program on your machine?
Running gentoo & x,org -- it looks like maybe an Xserver error
do you have glxgears?
JimHurlburt: you got RTAI set up on gentoo ?
Gl programs, not sure.
K'zan No, running in sim mode to test and debug gcode before taking it to the shop
JimHurlburt: Ah, guess I'll have to put DBI on a spare partition :-).
cradek : Yes, and they run zipping right along
K'zan : Perhaps could, but didn't feel the need
Rtai may not be a 2.6 option anyway and some of the guys have been working on low latency stuff till it's pretty good I hear
JimHurlburt: at a shell run python -c 'import OpenGL.Tk'
JimHurlburt: Gentoo freak here :-).
JimHurlburt: I still have not spotted the bug in your gcode. maybe it's an emc bug.
cradek : Python, thence import OpenGL.Tk pops up a little tk window
hmm, I figured it would segfault
there are lots of bad pyopengl packages out there
did you buidl pyopengl yourself?
Gentoo, there is no choice. It always builds on my machine
do you have more than one python version?
have a look at the first line of .../emc/plat/.../bin/axis
to make sure you're testing the right python
I think 2.3 and 2.4. The running one is 2.4.1
% head -1 emc/emc/plat/linux_2_4_20/bin/axis
#! /usr/bin/python as you said
Which when run from the command line pops up 2.4.1
ok and that's the one you did the import in, right
Correct. In addition, when gentoo moved from 2.3 to 2.4 it rebuilt all the python dependencies as well
If I start 2.3 and try import OpenGL.Tk, get the no module error. I installed it special for this
Did have to go to plat/linux_2_6_11_gentoo_r6/bin/axis if that matters
do you have /usr/lib/python2.4/site-packages/OpenGL/Demo/tom/checker.py?
or wherever it would be on your system
try /usr/bin/python ..../checker.py
Shows a checker board, then the Xlib error same as axis
Seems to be an xorg.conf problem
or it could definitely be your python
[06:06:46] <cradek> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=282942
you might be able to fix it with this method... but I don't understand the details
but wait, I thought we worked around this
are you using the latest snapshot of axis?
oh wait, you get the Xlib error, but no segfault?
I don't think that DRI error is a problem
may not be. the gears ran well
does the checkerboard work? can you navigate with the mouse?
you can fix the dri error by changing Load "dri" in xorg.conf
I think I'm back to having no clue if the checkerboard works
Can click on the circle and wiggle the checker board around
I assume axis didn't dump a core?
Can do the xorg.conf change. Will take a bit, will need to make the change and restart x
I don't think it will help, but I guess it can't hurt
No core that I can see in emc
cradek: Does EMC (or controllers in general) time the step output or do I have to buffer it ?
K`zan: I don't understand the question
cradek: In that case, I'll log of and tweak x. See if you're still here in a while
Thanks for the time and attention
cradek: In my little program, I have to delay a given amount between steps or the stepper sits there and buzzes, I would think that the parameter file in the controller should know (be told) what that delay should be.
oh I understand
all that is timed by emc
you configure it by telling emc the maximum velocity your machine can handle
that, taken with the scale (steps per unit distance) give a maximum step rate
Ah, OK, is it something you adjust in a parameter file - ah, thanks, so all I have to do is just do it :-))
right, the driver just has to step when emc says step
Much to learn :-) Thanks so much!
ALmost got it, just need to save state and direction and deal with that on a new step/dir command and I can hook up the parallel port to it :-). Toooo cool.
more interesting is that your stepper can turn faster than you think, if you accelerate it smoothly up to a speed above which it was buzzing for you
emc handles this acceleration for you too
direction polarity also something you specify in the parameter file?
OK, too cool!
don't worry about it, and if your motors turn backward, you just reverse it
OK, well I have to keep track of direction to get the right value out of the table...
So I will assume a given state as CW or CCW as the case may be.
there might be a standard, I'm not sure
but it's not clear that CW is + or - since you can have RH or LH leadscrews
so I think it's not worth worrying about
if (bit_is_set(PIND,2) dir = CW;
Ok, just so I deal with it.
I may end up with a controller that works with my funky steppers and save the coins for that Taig :).
err DRIVER :-)
jepler wrote a three-axis step/dir to halfstep phase drive for the avr
I used it for a while
err, four axis
Only need 3 here, but that's OK :).
C or ASM ?
he managed to get four interrupts, using INT0, INT1, T0, T1
Was just sitting here debating that, Use a small AVR per axis or a big one for all...
Cool, any chance he might part with code ?
a big one for all is possible, but it's too clever by half
you have to set the timers so they're about to overflow
then set the external timer input pin to increment the timer
There are a number of interrupt on change pins, will have to see.
then catch the SIG_OVERFLOW interrupt
just two, the other two are timers
No need to use the timers if you have 4 interrupt on change pins handy...
do you have that? I'm sure we didn't (atmega16)
Might make more sense at this point to do all three separately rather than toast one biggie on a screwup / misunderstanding...
I suppose it will be easier to get right.
Lemme check, never really paid attention to those as I never needed them.
Cheaper if I screw up :-) too :-)
Maybe this might be a good thing for that old M8515 "I will never use" :).
If you get a chance, rent "The longest Yard" (2005) w/ Chris Rock and Adam Sandler. It's really good.
And I'm not a Adam Sandler fan at all.
Looks like that PCIE thing is only for the really tiny Tiny AVRs. Not helpful :-(.
working on some haskell code
yum yum.. haskell is tasty
yeah, it sure is :)
where did you go?
hello ValarQ :-)
alex_joni: i spent a couple weeks in the ancient forests in oregon, then spent a month on my bicycle riding around oregon and the pacific coast
what about this encoder ? http://www.usdigital.com/products/e4p/
is ok to use with a servodrive, 250w servomotor ?
what kind of drive?
is the resolution ok ?
ive no idea..
well, yeah, since you have g340..
how many counts ?
is it like 200 step ,or better i suppopse
you can scale it at the drive if it's a bit high to get high speed..
200 is pretty low
you gonna connect it to the motor?
480 to 1200 pulses per revolution (PPR)
Jacky^: what resolution do you need on the machine?
alex_joni: great question :\
what do you want to be able to do with the machine?
alex_joni: how long will you be here today?
anonimasu: will go away for a while, but I'll be online later today
alex_joni: i planned 1x1 mt wood machine
I'll be around for an hour or so
les helped me ..
Jacky^: not size, accuracy
do you want to cut parts with 1mm steps?
or do you need better finishing?
say you cut a circle
1mm is huge...
better finishing would not bad..
do you want to see 1mm steps?
so you'd need smthg like 0.05 - 0.1 mm at least
or even better accuracy
1 mm seem a lot
but once you decided that, you can fgure out what encoder to use
depending on the resolution of your leadscrews
you do use leadscrews, do you?
othe leadcrew will be 20 mm diamter 5 mm lead
ok.. 5mm lead
probably abba leadscrew
that means 5 mm / turn
ok, so if you have an encoder with 200 counts / turn
that means 200 counts / 5 mm
= 40 counts / mm
pretty nice ..
that is 0,025 mm accuracy
not accuracy, stepping size
should be fairly ok for a homebrew machine
what fenn said
so you will be able to position the machine at multiples of that resolution
will you drive the leadscrew directly?
I mean motor connected to leadscrew ?
but ill use twin screw on x axis
to move the gantry
some strap too
i think should be 1:1
so the above calculation is ok
if you do 2:1 then you get twice as good resolution
and so on
so basicly you start the other way around
you got the leadscrews, got the resolution you need
then you choose the encoder
if you got the G340 then maybe choosing a better encoder is not a bad idea
alex_joni: it seem geckodrive seel no encoders..
the G340 has a multiplier from 1 to 10
that way you can compromise a bit
i need to do a new order, so ..
when you find out how fast you can crank out pulses
and what resolution you really need..
so you can have a 2000 counts encoder and use it like it is a 200, 400, 1000, 2000 encoder
i know gecko has step multiplier
depending on what you need
but ive no idea how it work ..
you need to see the step multiplier as a resolution divider
hmm.. you dont need a step multiplier for a servo based system
if the step multiplier is 1, then the motor needs to turn 1 encoder step for 1 commanded step
step multiplier is much more complicated and funky than a resolution dividier
fenn: why not ?
if the step multiplier is 10, then the motor needs to turn 10 encoder steps for 1 commanded step
fenn: it turns out to be the same thing
even if it's more complicated to do
and the G340 already has it in there
for about 60$ more money
fenn: i'm buyng G340 right for the multiplier :(
right, but you dont need to have a step multiplier to divide down resolution
isint good ?
just use a binary counter chip
yeah, sure.. but you can use it for that
fenn: not with signals from the encoder
right, anyway, it's probably better to just buy the G340
if you don't mind the resolution division, you might get better performance of your machine
as in 10 times faster ;)
and it's only 39$ more pricy
Jacky^: definately worth the money
I still have some G340 on stock
but I just heard of a project on that I'll use them ;)
alex_joni: should the encoders be connected to parport too ?
no eh ?
no needed right ?
I'd stick with a simple design in your case
i asked for some quote in italy too ..
just a moment..
ie. thi axis http://www.ronchinimassimo.com/rm_tl_y300.htm
300 mm travel
don't like it
573 EUR for such a crap
it seem to me too ..
Gioco sulla madrevite [mm] 0,1 +/- 30%
also.. 300mm is a bit little
yeah that's a expensive slide..
i always thought parma was in spain
600 mm travel, E. 1339,00
and not even ballscrews from what I've seen
same axis ..
the R type seems a bit better
mmhh.. ive no quotes for that
if it's the same price, then it's a bit expensive
but only a bit
anyway i will not use rounded bars..
low profile rails
for woodworking though i think you can get away with just aluminum extrusions like 80/20
* alex_joni goes away
* A-L-P-H-A pokes alex_joni
in the eye!
almost chilly enough to turn on thr heat
hm I should look outside
it might be snow..
about 2 degrees with 50 km/hr winds on the mountain tops
but no wind in the valleys
* anonimasu ponders going outside to have a go in disassembling the machine
morning les :)
was asking for encoders on new machine
what ballscrew pitch?
ballscrew, 20 mm diameter 5 mm lead
i was looking at miniature encoders by usdigital
they seem to be cheaper
about 500 pulse/rev if the gecko does 4x interpolation
I use us digital
hm relatives on visit
what you think about ep4 miniature encoder ?
ah well at my parents appartment..
I hope they dont come to visit me
anonimasu: i will came with all my family ]:)
I could fit 23 people.. :)
but I dont want you & your family here :)
I use e3....ep$ would be ok too
i will came the same
les: good ..
or well anyone for that matter
just need to know if gecko 340 does 4x interpolation
I don't kow....but it is important to find out
les: mm.. i will ask
hm my little cousin said I had a mess here..
* Jacky^ give a cup of strong italian coffe to anonimasu
*puts on some tea*
its a good brand
maded in naples
yesterday i machined a beautiful star box
with a nice inlay on the top side ..
after 2 hours i finished to machine it .. it curved
how did that happen
im going to try again with another wood type..
anonimasu: the piece was 2 cm thikness
i machined at 3 mm
that seems thick
not enough ..
*craves for making parts*
bad quality wood too ..
I am going to finish the machine today..
or well pick it apart so I can get bearings next week..
I've got projects stacking up
to build a machine ?
anonimasu: are you able to machine ballscrew with your lathe ?
no eh ?
on my old small lathe that was..
I machined a flat then crimped a -- on the end of the screw..
for machining bearing surfaces off..
I didnt feel like cutting too much screw..
if I had more experience I would have ordered the screw pre machined..
but I cut it at my micro lathe by the time ;)
i wonder if anyone has tried les's idea for putting nubs on the end of a ballscrew
well, i guess les has, but he's just a nobody :)
I think I am going to skip building a robot style robot
and just build it as a manipulator..
you mean the screw on caps?
put ball bearings in the grooves of the screw, then add a "cap" with ID equal to the diameter of the balls
imagine that's more accurate than trying to cut a funky ID thread
wonder if endplay is an issue though
Many don't realize that machining the ends can cost as much as the screw
les: right ..
as for my next machine I am not buying skf screws but isel ones..
wonder who makes the isel ones
they are 1/5 the price ;)
but not nearly as accurate
Soon my machine has to go down to fit the new spindle
but it depends on the app..
I would have bught the screws pre machined..
and bought real gears with < <
I machine ballscrew ends...but it is a lot of work
I think skf beats you in precision ;)
I'll bet they don't ;)
and ordered the bearing mounts off the local machine shop..
I machine and grind ends to .002 mm TIR
it is expensive to do though
i know i already asked this, but how do you center the ends? just a 4-jaw?
good question. It is often a problem, because on some screws the outside lands are not good datums
a good 4 jaw will work....indicate the bottom of the ball grove while in threading mode
on some screws the lands are good for datums
then a collet is fine for holding
*reads about kinematics*
I wonder how to make toolchanges easily.
1. unscrew drawbar 2. remove tool 3. insert tool 4. tighten clamp bar
fenn: oh geez I didnt think about that
glad i could help :)
I wonder if I could build that manipulator to pull the tool out..
alex_joni is now known as alex_joni_away
alex_joni_away is now known as alex_joni
alex_joni is now known as alex_joni_away
alex_joni_away is now known as alex_joni
Hi Alex. How you doing on this fine weekend?
as you said.. fine
Good. Beautiful weather here in western Illinois.
I slept late today
Good. So did I.
Got a good download of 4.30 and will install today.
[16:13:49] <Jymmm> http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=8335653541
hang on .. my browser is busy :D
White Trash Survival Kit... http://www.in-depthangling.com/forum/showflat.php/Cat/0/Number/242993
did I miss the meeting?
oh, wrong day
not much going on today ;)
heh.. too early ?
not enough coffee?
no coffee yet!
right.. there you go
[17:02:28] <Jymmm> http://www.joltcola.com/
dont tell me you've never had jolt..
something with lots of caffeine
I drank jolt for years -- long ago!
[17:05:52] <Jymmm> http://www.geocities.com/SiliconValley/9692/joltcola.html
not available around here..
Never realized that Transylvania is in romania
alex_joni So yes, all the bad dracula jokes will begin shortly
and will include Frankestein, Werewolf, the mummy, and all the other hollywood horror creatures.
at least till November 1
* alex_joni prepares
alex_joni here ya go... you can order yourself some http://www.frozencpu.com/caf-05.html?id=mQaodxEt
but shipping might be a bit much
wow. you guys don't know how lucky you are with high speed connections. This is awesome.
heh.. seems rayh got addicted
* alex_joni wonders how long it will take till rayh will have a highspeed at home
rayh: At my gf's... we tried microwave, cable, dsl, nada. Until they put in fiber optic cable in 3 years later, then we could get DSL
Thanks for that report, Jymmm.
rayh wasn't meant to rub it in at all
One phone company put in fiber optic along the county road this summer.
only 1.5 miles to go.
that's not far
Got 4.30 installed on athalon 64, compiled emc2, and running test program.
what are your plans with it?
Got a customer that wants to run with this PC configuration.
I'll do a bunch of testing for stability during this week and if it passes, we'll use it on the Mazak.
so it's emc2 all the way
You bet for this Mazak.
I'll start testing classicladder Monday.
nice, let me (us) know how it works
We are using a vital motenc lite for the motion
and another card for IO
TomP wrote up a big set of web pages with much of the mechanical.
I'm working on wiring diagrams now.
pen & paper
vital motenc lite
Rough sketches right now dxf when I figure out how.
<Jymmm> vital motenc lite???
what is it?
it's a board
with emc2 driver for it
google it up
[19:03:03] <alex_joni> http://www.vitalsystem.com/
darn this wiki is driving me crazy
anyone knows how to put a picture in a page?
What's going on with it?
I used the html <img src= tags
I have not tried the upload menu item but it is supposed to return a url to the uploaded image
width="120" height="120" border="0" alt="Playboy Corp, All Rights Reserved">
That url can be put right into the text.
There was a sample image url in the sandbox.
the problem was that my image was using UPPERCASE
[19:07:40] <alex_joni> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Alex_Joni's_Toy
I've got a new problem that I need to see if anyone else is experiencing
Oh. Hadn't thought of that. I think there is a markup that could get you around that issue.
rayh: I simply renamed the image
ok, so what is the pulse rate of the vital motenc lite?
it's analog Jymmm
It is a servo only board.
new installation from latest BDI, with EMC2 installation within that, using the exact method described on the pages
alex_joni you ever get that USC card up and going properly?
EMC2 is VERRRRRRRRRY slow to respond to key strokes and...
I can't seem to get it out of estop
Jymmm: I have no USC card, anonimasu has
weyland: that's strange
latest emc2 from CVS?
EMC1 responds fine
anonimasu you ever get that USC card up and going properly?
Jymmm: he's not here
just did another fresh install today
but I think that he did, only with emc1
weyland: and it builds just fine?
yep, no complaints
just VERRRRRY slow, and can't get movement
hmm.. I only have 4.20 over here
didn't put the latest
didn't se any, but to be honest wasn't keeping an eye out for them, as it's always just worked
try make -s
be right baq
WARNING: Symbol version dump /usr/src/kernel-source-184.108.40.206-magma/Module.symvers
is missing; modules will have no dependencies and modversions
thats all so far
weyland: I don't see any problem with responsiveness here.
man... it's REALLY SLOW to repsond to keystrokes
I times it -
takes 6 seconds from the time I click a button or keystroke it
instant in EMC1
try increasing PERIOD
the one in the .ini
Does emc2 take is fastest thread time from PERIOD?
it's (typoing 101 again)
that helped a little bit
still can't get it out of esrtop
what have I forgotten?
(in the last two months)
load an halmeter
and look at iocontrol.0.estop-in
Okay I see three BASE_PERIOD, SERVO_PERIOD, and TRAJ_PERIOD.
RTAPI: ERROR: could not open shared memory
HAL: ERROR: rtapi init failed
BASE is the most inner loop
weyland: that sounds bad
when did that happen?
when I tried to load the halnmeter
try halcmd show
I navigfated to the directory and ./halmeter
try ./halcmd show
I suppose emc2 is running all the way
I mean, you didn't try to run halmeter with emc2 stopped
got it now
ok, at least it's not that borked :)
what am I looking for?
look for iocontroller.0.estop-in
if you can find it :)
* alex_joni is typing from memory
might be iocontrol.
iocontrol.0.estop-in it is
errr.. ~# ifconfig
eth0 Link encap:Ethernet HWaddr 00:02:44:1F:23:AF
no, I don't se anything like that
look at the emc.ini
and check what
# Name of IO controller program, e.g., io
#EMCIO = simio
EMCIO = io
EMCIO should be set to io, not simio
and all I can find is the estop sense line
it's set to 1
shold that be 0?
thought I recalled something like that
emc2 shouldn't have any estop connected
but.. it troubles me that you don't see iocontrol.0. running
how does the overall system behave while emc2 is running?
also slow? or only emc2 is slow?
everything else seems to be humming right along
ok.. lets rewind a bit
can you post me the messages during load?
maybe I screwed the pooch here
has anything changed recently with regard to the ini?
I took for granted that everything was teh same
not sure, but it's definately different from emc1
try to cvs up a recent one
so I grabbed the old ini I had from my last emc2
so still emc2, it "should" be ok..
can you send it over?
lost yer addy, though
can you hep me to install kde ?
sent the addy in prv
yes, close before all apps ..
* Jacky^ is quit*
Jacky^: wanna quit?
I mean if you want to quit :)
i wonder why kde wont run if disk is full ..
we could arrange that
need for temp files?
100% in use
only 1 gb ..
weyland: the same problem when running tkemc?
dunno, haven't gotten to think that far ahead
lemme go try
hmm.. iocontrol must be running
or else you would have gotten errors during loading
as iocontrol.0.spindle-on is connected to parport.0.pin-09-out
weyland: there is a problem in the ini
change the line: TOOL_TABLE = emc.tbl
to TOOL_TABLE = configs/emc.tbl
okay, did that, it's a little better
now I cvan control the estop, but am geting joint errors
put the original period back in place
it was a wild goose
damn, still getting following errors
what could have changed?
ok.. looking at your settings
slamming me head now
everything should still be the same
its not like I changed the ini before I dsopped it in the direcrtory
how about renaming it, and getting a fresh out of cvs?
just to make sure it's not an emc2 issue?
I'm kewl with that
how do I get it?
rename the emc.ini
and do a cvs up
is the new one there?
It's seems to be commonplace, but to step generators *HAVE* to have some sort of driver? I mena, the paraport itself doesn't actually need a driver, why does a stepGen?
depends on the step gen
there are 2 kinds
and hardware ones
and otoh, the parport does need to have a driver
just got it, but it's till sroloing things bvy
let it scroll, let it scroll, let it scroll
alex_joni I wouldn't mind the USC board (as example) if it wasn't specifically emc based. I'd like sort of hardware only (no drivers) step gen.
Sing it, man, Sing it... You rock. Houston, we have movement...
alex_joni I've been burned far too often to get/buy/use propritary hardware.
Gotta tweak a few things, but looks to work.
Lemme try that
weyland: probably something borked on that ini
Hadda be, but can't imagine *WHAT*...
* alex_joni would suspect too low accel along with backlash
lemme play and let ya know
afaik nowadays backlash takes teh accel limits into consideration
anyways.. glad it works
why is it that computers only blow up when you need them?
weyland murphy is a dead man when I catch him
because at other times they are switched off
I'll be baq inna while with a report
l8r and thanks - baq l8r
* Jymmm pokes alex_joni
alex_joni I've been burned far too often to get/buy/use propritary hardware.
but I cant find a stepgen like I desc
thats why I was asking *IF* it even exists
(technically that is)
what do you need?
a pure hw based stepgen that can connect to paraport and give me at least 100KHz
sounds like the USC
but also the G-rex
but that's not fully done yet
so there are no others out there?
probably a bunch
these are the ones I know of
and USC is probably the best choice if you want to use emc
actually a wide range of devices from PICO
[20:15:18] <alex_joni> http://www.pico-systems.com/motion.html
again, I'd prefer a pure HW device, not HW + driver/SW
well.. you need smthg to drive it
if you want pure HW get a controller
like a burny
yeah... but still you need some software to talk through the parport
well, take the xylotex board I have... anything can talk to it, no drivers for it.
and? does it do step generating?
I just need a clean pulse train, why does that need to have a driver of some sort is what I dont understand.
ok, so where can I find such an animal
it needs smthg to tell it what the pulse train should look like
and for that there is no standard
some use parport, some usb
setting up switchs or shell to it is no biggy
but even for parport: some use epp, some direct connections, etc
the thing is: the HW device is usually inteligent
and uses a protocol to talk to the PC
or to get the commands
and the "driver" is actually the software that implements that protocol
why is a protocol needed at all? can't a stepGen be a passive device (once configured)?
it gets data
then it needs a protocol for that data
even rs232 is a protocol
but no driver is needed for RS-232
right, because it's done in HW
that's all I'm asking for
the assembly of bits
ok.. then you need a new mobo
with your pure HW driver on it
ok, where do I buy
before the parport
you build, you don't buy
there is no such thing
ok, so what is the limitation of paraport the protocol?
ok, so what is the limitation of paraport, the protocol?
but that's not a real limitation
the only thing is to figure out what you want
and look at the existing solutions
so a PCI/ISA device could do 100KHz pulse train?
if it has a HW step generator on it, surely
it could do 10 MHz pulses
but why can't the paraport?
but that same HW gen. can be in the PC (on a PCI/ISA) or externally
but you said a pure hw dev doesn't exist
the parport by itself can't do that speed
because of timing limitations
ok, timing... of the chipset?
of the ISA bus
because the parport is hooked up to the isa bus
so now the ISA bus can't do 100KHz?
but PIC can?
PCI doesn't work that way
PCI can be set up to use "smart" transfers
that might increase performance
DMA for instance
you get throughput..
but if you still send commands one by one (set pulse high, set pulse low, etc)
then you get bad performance
ok, hold on a sec...
BASICALLY... what is a stepGen that's connected to a paraport actually doing, that the paraport alone isn't?
it is an entity by itseld
if is somehow inteligent
ok, so a stepgen is sorta like a debounce circuit
it gets a prescription, and then generates pulses
the parport needs to be written by software
lets take an example
to send a pulse out of the parport
1. set the parport pin 1 to 0
2. wait a bit
3. set the parport pin 1 to 1
as long as some signal is present, a clean output will be produced
4. wait a bit
1 and 3 are done by some software
0 == ___________________
by writing the data to a memory location (inside the PC)
1 == --------/\/\\\////\\/\_-------++++
the parport usually doesn't need a driver, because way back it was mapped into the PC memory
so, as long as ANY ugle signal is received, a continous clena pulse is generated, correct?
so outb(0x00, 0x378) actually turns all pins to 0
I'll get there
did you get how the parport stuff works?
a loop like this:
paraport == hw mapped
that's a pulse train
the actual output on the parport switches from 0 to 1
[11:57:47] <alex_joni> that's a pulse train
the whole sequence is a pulse train
right, out addr 0x378
in this case about 1 usec between the pulses (that's how long the outb takes)
ok.. now imagine you have a black box
the external HW stepgen
it is connected to the parport
and there is a software talking to it
and the software says: I need 1253 pulses at 12.34 kHz
then the black box starts generating those pulses
after that it says.. I finished
and the software says: now I need 234 pulses at 1.5 kHz
and so on..
got it so far?
and the way they talk to each other is the protocol
but you're expecting full duplex communication
not necessarely.. but I used that example
it can be single sided
the xylotex is Rx only.
the xylotex is not a HW stepgen
the xylotex is a motor-driver
does the stepGen HAVE to be RT? Could it just play tivo ( instead?
I think your mixing things here
the stepgen is a custom hardware
usually there is a counter in there
and a comparator
but that's not important
the HW stepgen is doing one thing: outputting pulses
those pulses you connect to your xylotex
the same pulses can come from a PC running some sort of RT software (be it emc, or any other)
it is certainly better if the pulses are generated in HW (not in SW) but it's more expensive
as you have dedicated hardware doing only one thing (generating pulses)
so.. the USC is basicly a hardware that outputs steps
but as it is it is pretty dumb
it can only do that
someone needs to tell it what steprate it needs to generate
So, the limitation of a paraport is dues to the way it's (legacy) mapped to chipset. It was never designed to put out a 100KHz pulse train.
though with ECP, DMA, and a small amount of external circuitry you could do very well
icee: yes, but you'd still have a proprietary solution
alex: hmm? why do you say that?
all you'd need is an external clock and a latch
So as long as the USC sees a pulse train be it 0001.11111.111 it'll output a CLEANEDUP pulse with the same counts as it received: 000111111111 ???
. == garbage
the USC doesn't see a pulse train
it generates one
it gets a command: generate 3x0 and 9x1
alex: If you're building a piece of hardware to generate steps
with a frequency of 2300 pulses / second
you really ought to keep track of position (step count), and send that back periodically
and take commands with acceleration rates and target velocities
icee: I know
e.g. pretend you're a servo
ok above you said outb(blah,blah) x number of times. But with the USC it's a single command not multiple ones to achive the pulse train?
that's why that solution is faster
becasue it takes 2uS per command
yes, and because that command contains what would have been 200 pulses
and 200 pulses would have been 400 commands to the parport
ok, I get it now
Ok, now... what about my ugly pulse == clean pulse thingy?
that is generaly done by a debouncer as you said
but usually that part gets skipped
right, but is that so bad for mot ctrl?
output from a parallel port is usually relatively clean
it is very bad to have lost pulses
espc if the cable is of a reasonable length
and shielded properly
in emc, what is the paraport "driver"?
no, it's some name in the ini file
sorry, I misunderstood.
jepler thats ok, so did I =)
Jymmm: we get to a nice topic
what kind of emc?
emc1 has a few of "drivers" that write to parport
well what is talking to my paraport then?
but basicly freqmod
ok, that's it freqmod
and the difference between them?
steppermod runs on RTLinux only (not on RTAI)
freqmod wraps a PID around the steppers, and you can tune it like a servo
steppersegmod is the steppermod extended with segmentqueue
in case you're wondering: segmentqueue is an advanced blending software, that never worked as it should
Jymmm: any time
man, using the ext numpad as apendant really works out well.
it does? cool
could you set up a page describing that?
a short wiki.. with a few pics would be great
* Jymmm hates wikis
bounch o' electronic post-it notes.
how about an email to me? and I'll do the wiki then?
yeah.. but others benefit from it
actually, if you're using axis, he put it in there
what are you talking about?
[21:10:39] <Jymmm> http://axis.unpythonic.net/index.cgi/
jepler added the NUMPAD mapping to axis
for him to do that ;)
in minimill it's ok, by tkemc, well JUST SAY NO!
K`zan: I've posted that AVR program that cradek mentioned last night
jepler: Got it, thanks ***VERY*** much :-)!
Was looking at using a Tiny13 to do a single axis. Or possibly a Tiny2313 so I'd have some extra pins for whatever.
Now I have something to look at, so I may well change my approach :)
jepler: that's an old version - a lot of things are different in the final one
wondered when you're gonna show up
seing jeff around..
seeing jeff around...
and you two usually running in pairs :D
well, I saw my name go by.
congrats about beeing mentioned on opengl.org
jepler: seems there's a small problem with one of your links
[21:34:09] <alex_joni> http://axis.unpy.net/index.cgi/http:/sourceforge.net/mailarchive/forum.php?thread_id=7989777&forum_id=33140
I dont belive this crap... lastnight the stepper stalls, today it's fine. and I didn't do/change abything?!
maybe the air is less moist
actually, MORE moist today (rained this morning)
then maybe it's that
so I can only run my router when it's rainging?
lol.. or you need to set up a sprinkler
cnc rainfall huh? Whats the g code for that?
lol... that's 'make air moist' huh?
rain on it
Jymmm: what emc did you test?
alex_joni: what page is that on? I'll fix it.
[22:39:20] <alex_joni> http://axis.unpy.net/index.cgi/
down at Paul C releases AXIS packages for BDI 4.20 By Jeff Epler
thanks for noticing. fixed.
cradek: do you have the newer version?
cradek: nevermind, I just read my e-mail
new version of maxnc.c now on that page
got it, thanks much!
K`zan: good luck with that code
It really takes 16 MHz to keep up with EMC ?
depends on the step rate you need
Heh, with these, it doesn't appear to be much, best I can get is about 1" in 10 sec.
Well, I can get it down to about 7 sec before the stepper starts chattering.
[23:02:26] <alex_joni> http://www.thedailywtf.com/
based on my year-old recollection about the cycle times, I think that it's about 50x more speed than necessary for the 12 ipm * 8000 steps/inch mill it was used on...
and that's assuming that you have 4 axes all stepping
gads, people still use IE ?!?!?
K`zan: so I've heard
so even using the on-board RC oscillator at 1MHz would probably be OK
scary, reminds me of lemmings :-)
There's a UI WTF right there- you have to enable the disable option, rather than disabling an enable option. Sure, they're syntactically the same, but the second doesn't make yourself ask a question like "now do I not want it not to not show up?"