is there a requirement to have pullups/pulldowns/drivers on parport pins?
I'm trying to see anything from the parport on my Opteron, and having no luck
the update function is in the base thread, I've tried the parport_probe trick (may not have done it correctly though)
if anyone has any suggestions, I'd love to hear them
parports are usually totem-pole output 20ma of sink, 3ma of source
should be enough to drive a scope
actually, they can light an LED - some are high, and I get lots of nice white light
yip, thats pretty good, mostly their source capaability is crap
actually, that's sourcing - there's no +5 on a parport
these are bright as hell LEDs though - 130000 mcd at 20 mA
(yes, 130k )
but a 0.8 degree beam?
nope. 10mm, 130 Cd, ~20 or 30 degree spread
* robin_sz wonders how many foot-lamberts that is
these are pretty brght - I can read in my office if I have two of them pointing at the ceiling
it's a 180 ft^2 office
[00:40:15] <SWPadnos> http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=7602852708
that's not the auction, but it's the same thing, I think
though I'm pretty sure mine are 25 degree
or 20 or something l(
led technology has come on a lot
nope - I was remembering the operating temperature in degrees :)
yes it has
you know IR solid state laser diodes?
what you thing the max cw power is from a single device?
on ebay, 150 mW
gotta have a lot of cooling for that, I'm sure
solder them direct to a cooled lump of copper
silver - even better
alex_jon1 is now known as alex_joni
robin_sz: seen one that does 454W
* Jymmm is exhausted... G'Night Folks!
good morning alex_joni
indeed.. same to you
I managed to compile new emc2 version eventually!!
fumbled a bit with headers and got them to find what they needed
the program starts with spalsh and a machine configuration menu
but then complaints
may I paste the message?
it runs in place
[08:01:33] <pier> http://pastebin.ca/145984
* alex_joni looks
ok, you're missing an utility program called fuser
bet slack has that somewhere, although I don't know where ;)
I've already used it several times
ok.. it seems you discovered something we didn't expect
I see the script is checking for /sbin/fuser and /bin/fuser
perhaps it is located in the wrong place for emc to search..
pier: 2 ways to fix it: 1. make a symlink /bin/fuser to /usr/bin/fuser
I'll try with a symbolic link
2. edit emc2/scripts/realtime (you can find the check easily)
ok... now the rt module pops up :(
[08:09:17] <pier> http://pastebin.ca/145990
sorry.. my puter is a bit slow now (have a virtual dapper loaded)
takes a while to open that page :(
pier: this one is easy ;)
but you should read the docs :))
the module hal_parport can't be loaded, because you already have a parport module loaded
my my ashes on my bald head .....
try 'rmmod pc_parport'
ok lsmod gave me a hint
or check with lsmod what's loaded
ok already done
ok rmmoded lp, ppa and parport_pc
God am happy!!!!!
thanks so much!!!
pier: no problem ;)
glad it's working
was too sad working with bdi on a different distro
pier: it seems everyone has their preferences
yes... that is true... it is much a question of laziness on my part though
I used to be a SuSE fan myself
alex_joni: do you think that once the parport have been removed the pins of lp0 should be clean... I mean no signals out of them?
but they might stay at the last state
mmmmh that really puzzles me a lot
pier: what does?
because I wrote a cnc code to cut balsa pieces for my aeromodels
and lately, to get rid of dos for good I ported it unde linux
but when it comes to writing to paralel port
pier: it's a bit of a PITA ;)
it look I always have some noise on pins
and steppers run randomly
enable pins work thoug
I wrote that code to handle just dxf files
without the hassle of writing G-code
03alex_joni 07HEAD * 10emc2/scripts/realtime.in: check another plausible location for FUSER (hint: slackware)
pier: thanks for the fuser finding ;)
you are wellcome
It always happens to me
I lately discovered by chance
a (I wouldnt say bug)
strange behaviour in BRLCAD handling triangles produced in dxf files by autocad
VMware server is nice sometimes
as in the case of noise from parallel port....
pier: what drives are you using?
your own design?
no... just cut and paste from datasheet
I made the board hatching
pier: I suggest you looking at the PminMo design
has some nice filtering on the inputs
[08:30:36] <alex_joni> http://www.pminmo.com/l297-8/l297-8.htm
how come then they work under dos...? Mistery!
they work OK under dos?
not a hitch
then I suspect they should work OK under emc2 too ;)
yes... I tried BDI and it worked perfectly
pier: no reason why it shouldn't work under slack with emc2 then
hehehe now I'll move to the garage pc to get new kernel patched and emc2 working... I am at home now
well, you can always drag the PC along
and move the kernel && modules if you have the same version
alex_joni: Thanks again!
pier_garage: no problem
pier_garage: where are you from ?
mind if I ask you something?
if you haven't already, can you put a pin on www.frappr.com/emc2 ?
you are welcome!
pier_garage: how's the sea now?
ehehe my flat is at first floor!... wait a sec
I drove not far from there this year (through Frosinone)
ah... prety near here
yeah, went to Siracusa (Sicilia) .. :)
pier_casa: asking a friend overlooking the shore how sea conditions are
had one night in Ceprano (a small village near Frosinone)
pier_garage: not that important really :)
not here as far as I know
calm and smooth as oil
[09:41:12] <alex_joni> http://www.ngline.ru/fun/1000000.swf
alex_joni: now I remember... I gave you my e-mail to get in touch to when in Italy?
don't think so ..
not that I can remember anyways ;)
ok I mistook you
for another person who asked
might have been someone else
does anyone think that patching a kernel on a 1.3Ghz pc would take more than 35 minutes?
ok... solved... my fault
alex_jon1 is now known as alex_joni
03eric-johnson 07HEAD * 10emc2/src/emc/usr_intf/emcrsh.cc:
Add 'Help' command to telnet interface, update source documentation, and fixed
a couple of minor bugs.
jepler: I got a problem...
I bought a dell to run ubuntu6,emc2, and AXIS on - it works, np.
Brenchtop real estate is at a premium so I want to xopen to the laptop.
I installed BDI 4.5 on the laptop (couldn't get sarge or ubuntu6 installed on it)
I setup an ini under BDI to use AXIS as a test - it works!
Now, when I try xforwarding or xopen to the laptop, I get a opengl error. TkEMC, no problem.
Axis V1.2.1 is on the laptop, and on emc2 AXIS v1.4a0
bdi may not have gl in the xserver, but ubuntu does
does a forwarded glxinfo give an error too?
Jymmm: I frequently use a similar setup, but the displaying system is not bdi (redhat 9, ubuntu 5, fedora 4) .. I have never seen that behavior.
I do that with all combinations of X too, but I have one very old X setup that doesn't work
on that X server I get a glxinfo error like I pasted the other day
so I suggest trying anything but bdi
you could even try with the dapper livecd without installing it
Finally found a fast mirror for 6.06.1
6.06.1? is there an updated CD?
[16:54:17] <Jymmm> http://de.archive.ubuntu.com/ubuntu-releases/6.06/
cool. that's better than ~300M of updates right after install ;)
[16:56:31] <SWPadnos> http://ftp.wayne.edu/linux_distributions/ubuntu/6.06/
- haven't checked speed yet
I'm at 285KBs
hmmm - dooooog sloooooooowwwwww
SWPadnos: Any reason you didn't use the link I posted?
he's not in germany?
I didn't notice it until after I responded ;)
cradek and you rpoint is?
You dl where there sleeping =)
that's my theory, and I'm sticking to it!
they've been awake for many hours there ;)
it's ~6:00 PM there
SWPadnos I never said I actually look at the clock =)
well, be that way then
See, maybe they're eating supper =)
could be. and the kids haven't gotten online for their games yet
I dount know, I keep trying till I get 40minute dl, instead of 3hour ones
so, does anyone have any suggestions on how to get the parallel port to work on my fast machine?
it's a new install of dapper, plus updates and emc2
you tried the probe-parport module and various bios settings?
I've tried the probe_parport module
I haven't changed the BIOS settings - oh right. I should be sure it's EPP instead of ECP
maybe you (we) are unlucky enough to have another kind of probe that has to be done
are you supposed to unload probe_parport after using it?
no I don't think so
you load it before hal_parport
ok. the port is set to EPP version 1.9
I can try 1.7 instead
or set it to "normal", though that won't allow inputs
surely "normal" does inputs (AT style)
hmmm - I wonder if I'll get a graphical login after all the fiddling I did yesterday with the X server
I thought you needed EPP for the bidirectional stuff
cradek: is the glxinfo suppose to display remotely?
Jymmm, it should display anywhere you hope to run GL apps
then displays in the term window
I typed it from
I got glxgears to run remotely
by the way - I think I noticed a bug in halscope, though it may have been me
if scope_rt is loaded, and the scope.capture function is already in a thread, halscope seems to think that it isn't running
[17:11:59] <alex_joni> http://www.aish.com/movies/PhotoFraud.asp
Hmmm, no confirmation within TkEMC if you hit ALT+X
jepler anything I should check out?
Drove 140 miles yesterday looking at MH parks; I'm tired.
hmmm. I've tried all combinations of probe_parport/no probe_parport, with the parallel port set to epp 1.9/epp 1.7/normal mode in the BIOS
the only thing I haven't tried is changing the base address, which is a longshot since I've verified it's correct (278)
SWPadnos what kind of machien is this?
dell? Asus? model# ?
SuperMicro motherboard, H8DCE
dual NForce Pro chipsets (2200 and 2050)
I suppose I should try plugging in a printer, to make sure the port is good.
not quite the latest
That be my FIRST thing to do
is the latest
but the only differences (noticed in setup anyway) are related to NUMA memory addressing
I went through the entire BIOS setup with a friend who has the latest BIOS
BIOS controls EVERYTHING, period.
I know that
Well, JUST DO IT!
does his work ?
I tend to treat workststions as though they were servers: if it ain't broke - *DON't UPDATE IT!!!*
I'don't expect him to use a parallel port at all
but the machines work OK
If there's a BIOS update... dont ask questions, just upgrade.
Drivers/SW/OS, that's another story.
on the contrary - changing the BIOS can do weird things like change the PCI scan order, which can cause disks to move around unexpectedly
SuperMicro basically says that if you have a system that works, and a BIOS update doesn't fix a specific problem you're having, you shouldn't upgrade
So, you have to reinstall... worse case scenario.
I agree with that
And you have a problem.
but the update says nothing about "fixing parallel port to work with EMC"
or the parallel port at all, for that matter
I just dl'ed it, it doesn't say anything. and I've had LOTS of BIOS that dont actually say anything, but resolve the problem too.
so there's no reason to think that the update will help this problem
sure, I know that happens, but this is a server motherboard, and the way SM works is very different from Asus or ABit or any of the other consumer manufacturers
I may do the update (though there are problems that my friend had that I don't want, and may be due to the BIOS update)
SWPadnos: when you load the real parport_pc module, what is printed in dmesg?
durned good question. I'll check
SWPadnos: is this to use one of Elson's boards, or hal_parport?
I saw nothing in dmesg
I think I'll reboot, and check to see whether parport_pc is getting loaded automatically
weird - it actually rebooted with the magma kernel
maybe the next thing to try is to disable the "exclusive" code in hal_parport and see if it works with all the linux probes done
I still need to transfer my PGP key to that machine - I don't have a checkout there
or I could add another key
no sense crowding the key ring :)
dmesg reports "ppdev: user-space parallel port driver"
/proc/interrupts shows no IRQ7 user
dmesg dod show a spurious INT7
lsmod shows parport loaded, in use by ppdev and lp
/proc/ioports shows nothing at 278, and nothing for parport or lp
you think ppdev is some kind of replacement for parport_pc?
did you just do 'modprobe parport_pc'? You have to do 'modprobe -i parport_pc' to really load the module
this si the line from lsmod:
steve@opteron-32-ubuntu:~$ lsmod | grep par
parport 29640 2 ppdev,lp
03jepler 07HEAD * 10emc2/docs/src/ (.latex2html-init Makefile Submakefile lyxdep.py):
build HTML version of documentation -- requires latex2html be installed
(configure test for latex2html should be written but is not yet)
I wonder what would happen if I plugged in the other SATA drive right now (the one that has my key on it
03compile-farm 07Ubuntu 5.10 (2.6.12-magma) * 10emc2head/: build FAILED ; see http://linuxcnc.org/compile_farm/emc2head_slot7_log.txt
well that was fast!@
uh how did I screw that up?
missing endif - make clean didn't even run
03jepler 07HEAD * 10emc2/docs/src/Submakefile: duh
forgot to save before committing?
no - that can't be
03compile-farm 07BDI-Live rc46 (2.4.25-adeos) * 10emc2head/: build FAILED ; see http://linuxcnc.org/compile_farm/emc2head_slot4_log.txt
03jepler 07HEAD * 10emc2/src/ (configure configure.in): check for latex2html and disable documentation building if it is not present
03jepler 07HEAD * 10emc2/debian/ (control control.in): latex2html is required to build the html format of the documentation
03compile-farm 07Ubuntu 5.10 (2.6.12-magma) * 10emc2head/: build PASSED
Hey, I have made good progress learning how the BASE_PERIOD etc affect MAX_VEL, etc,
But now I'm seeing something real strange.
Maybe you or someone might have some insight?
davidf: so far.. no :)
maybe after you explain the problem ..
I found I could lower BASE_PERIOD to 12000 nano with no prblem, and get 130 in/sec velocity nice and smooth.
But now, my mill overshoots by about 10 thou on a 9 inch x move.
I played with step_gen kinematics (re the need for headroom)
Made no difference.
Funny thing is,
you get overshoot with steppers?
Funny thing is,
I did this several times:
davidf: overshoot usually happens on low accels
03jepler 07HEAD * 10emc2/docs/src/Submakefile: must create the docs/html directory
davidf: it's not something you want to live with..
Anyway, let me tell you this weird thing:
cradek: usually if the stepgen has too low accel
and it can't keep up with the accel TP is putting up, it stays behind
I did a move to x=9.
meter said x=9.0000
I turned off emc.
(ie, closed it.
I restarted, and the x position came up at start as 9.0104.
Now that is just about the error I had.
Did that 3 or 4 times, same result each time.
oh you mean it's stopping too far? not going past and backing up?
this sounds more like commanded vs. actual
and the problem with step size
do you have backlash configured on the axis? what is the scale and deadband?
It doesnt seem to be backlash either. Several 1 inch moves in same dir give same result, after repositioning my ruler after first one to eliminate backlash measurement error.
Backlash=0 on all axes.
OH! wait a sec... I was playing with that in another file. Maybe I left it. hang on...
No, backlash = 0.
what's input scale on that axis? deadband?
davidf: can you pastebin the ini ?
he's got 8000 or 16000 steps / inch, so that's a pretty big error as a number of steps
we may also need to see the hal files
[19:07:44] <davidf> http://pastebin.ca/146628
which axis is doing this?
hmmm - no deadband at all
except for the home signals, Halfile is the standard_pinout. But I'll put it up anyway...
I notice that too
wonder if it defaults to something crazy
where should that param be?
and FERROR is set to 0.010 with MAX_FERROR at 0.050
I edited that from inch.ini as supplied /c emc.
so a G0 move will allow 0.050 of error and still consider the position to be within tolerance
but that wouldn't explain a wrong endpoint
I would still increase STEPGEN_MAXACCEL.. just to be safe
ah... thats waht I was sometimes seeing. at least .0325.
I guess we don't have deadband in stepper configs now, "max" doesn't have one
there should be no PID, so it's true that the endpoint should be exact
davidf: you only should get FERROR, MAX_FERROR is at full speed
so I think I might have messed with that when I was getting all those following errors the other day...
if it's a G0, that's full speed ...
g01 x9 f60
davidf: which axis causes troubles?
no we're talking about a wrong stopped position - I think ferror doesn't have anything to do with it
what feed rate, or does it matter?
see 3 lines up.
does it matter though - if you o F30 is the error lower?
I can get 130 in/sec now SWPadnos !
how do commanded and exact position compare when it's stopped?
03compile-farm 07BDI-Live rc46 (2.4.25-adeos) * 10emc2head/: build PASSED
Thanks for the help there. BASE_PERIOD=12000 @733 MHz P3
davidf: can you humor me and increase STEPGEN_MAXACCEL to 1.2 for X ?
sure, that doesn't seem too neurotic alex_joni
I'm still not clear on what you're seeing that's wrong
I need to start again at the beginning
I'm not sure (maybe I missed) on what axis the error is..
I see Y has 15-20% more STEPGEN limits then the normal ones
cradek, : Actual pos after 9 inch move = about 9.01 to 9.035
on X, the MAX_ACCEL is 1.0 and STEPGEN_MAXACCEL = 1.05 (which seems a bit low as headroom)
have you put a mark on your motor shaft to see if it turns an integer number of turns?
yeas, forget y z and a. I have been playing with x and the traj stuff w/o editing the others, so there will be that kind of thing.
are you sure your screws are that accurate?
that's why I asked about the motor shafts
right - I was just phrasing it anotherway :)
Before now, I had extremely tight tolerances. The mill is pretty good, & almost new.
but still, this test is simple and eliminates a lot of possible problems
But, maybe a few excursions at 120 in/min loosened it up.
But i can't hardly imagine it woiuld have done much that fast.
is this a sherline type machine?
SWPadnos: at 120 IPM? I doubt it..
heh - good point :)
No re the motor shafts, Taig re the mill. 12 on x, 5.5 on y.
.02 inch is half a turn of the motor shaft with 20tpi screws, you should be able to see it in the dark
I won't be doing that often... :D
humor me please
268 oz-in motoer at 2.0 amp, rated 2.8 amp
oh yes. just a second...
mark the motor shaft with a sharpie, G0 some integer number of inches, look at the mark each time
ok, goo idea. Another thing, maybe the motor connectors are pulling out/in too right?
good, not goo. :)
on a sherline the leadnuts are held with some set screws
anything can get loose, .010 is hard to see sometimes
let me check that out give me a few minutes to look at all that stuff.
davidf: remember to test one bit at a time
don't mix & test more than one thing, or you'll never know what the cause was
haha, as we're all telling him to try different things
yeah, I know.. but a list should do :D
cradek: I know I've got bitten more than once by fixing machines like that
turn 5 screws, and it works
then suddenly it doesn't again.. and you're back at square 1
or looking for software bugs when you have loose setscrews instead
yeah, that too
have to start at the beginning and test one thing at a time
I understand not introducing more that one var at a time if possible.
ok, test 1:
My mill has indexer indicators marked off at 50 / turn, 20 TPI so that's .001 / graticule.
moved 5 inches, back & forth, right on the money.
BTW I didn't mean that to sound the way it might have re the single var thing. Just read that again.
sounded like I was sarcastic or something. didn't mean it that way at all.
oh not at all
so 5" moves work right
hard to tell on line sometimes.
to the tenth.
On the motor shaft that is...
any move is off a bit on measurement with my steel rule.
so you have a mechanical problem after the motor shaft?
I have a 1" dial guage. Maybe I should set that up.
you can see .010 on a rule?
Well we don't know at this point if its mechanical or not do we?
sure, no problem.
if the motor shaft is in the correct position, then it's mechanical
I thought you said the motor shaft turns exactly the right amount
1/32 = .0325, it's about 1/3 of that...
oh, of course. dumb me.
you can preview the HTML version of the emc2 documentation here: http://bald.unpy.net/emc2-docs/
(it's the same stuff that's in the PDF-format, but organized a bit differently, and of course in a different format...)
OK cradek & SWPadnos , I see the light. If the motor moves exactly right it has to be mechanical.
ok good :-)
ding ding ding
Hmm... I did have a crash at something like 100 in/minute... ouch.
bonk bonk crunch ding.
ding ding ding DONG!
somebody should paint over those oscilliscope images ;)
but it's al tables & small, not a real strong motor, and shaft attachment is via 4 small nylon shear pins. Hmmm...
Beter check those to start with, eh?
swpadnos, does the above mean I rang swp-etc when I typed swp tab?
what is the screw pitch?
swplinux any rel to SWPadnos ?
my two machines are very close to each other - I just type on whichever one I'm at at the time :)
they're me and I - myself is out at the moment
Here's one to tease the noodle:
Brothers and sisters have I none...
But my uncles nephew...
Is my Father's son.
What relation is he to me?
he is me ;)
he is me :)
SWPLinux: I was me faster
darn you guys are smart.
you are the only you, alex
100 units of threaded inserts AKA tee-nuts, 10-32, 5/16, 3prong, .75" base, 1/4 barrel dia - $10.25/CDN + 14% tax [if there]
SWPLinux: couldn't handle more than one..
not too bad.
threaded inserts aren't T-nuts
from Readers Digest IQ test- Average time alloted something like 2-3 minutes.
they never were too smart ;)
Pretty good for old guys.
Well, consider the readership.
SWPLinux, it is on mcmaster.com
ok, to shear pin checking. Lucily it's just nylon 1/6th id plastic tube.
the 3-prong thing is a clue- those are press inserts that are supposed to go into wood or plastic, I bet
Need those T nuits for a mill table or what?
ok - "tee nuts" there isn't the same as T-nuts for machine clamping applications
SWPLinux, look up tee-nut on mcmaster.
anyone with good google skills around?
I'm looking at the pictures right now
alex_joni, not me.
* alex_joni is looking for a daasheet for MG200J2YS45
(IC, motor, ...)
www.enco.com maybe too..
TRANSISTOR IGBT POWER MODULE
www.littlemachineshop.com for small mills, etc.
for the T nuts that is.
how about the MG200J2YS50?
I bet the end is the current/ voltage rating
SWPLinux: I have that one
these are old.. the YS50 is from 1997
alex_joni, MG200J2YS45 is obsolete.
MG200J2YS45 # 197470 IGBT MODULE FOR MILLER
A-L-P-H-A: I know that, but I want to know what I could use instead
you'd need to check with toshiba.
well, the ...45 is a 600V/200A part
if that helps
can 300 amps be okay?
probably, but the switching time and power may be different
[19:53:54] <A-L-P-H-A> http://www.monstermarketplace.com/Electronics/ListingM1528a11-1.html
A-L-P-H-A: if you have snubber resistors around them, then it won't really work ok with some other transister
alex_joni, call toshiba, and ask them what's a replacement model.
appearantly the part is obsolete, but there's still tons of people that have them for sale.
Guys, Shear pins, linkake are fine... Lead screw nut, shaft feel tight & all good...
I'd use something better than a ruler to measure then :)
you're talking about an error that's about the size of an amoeba
You know, I just realized I can't see any way that I could OVER shoot if the lead screw and nut can't slip.
the ruler can move, I bet
So I'm guessing that ruler is off!
actually, if there's backlash, you can overshoot
Had it C-clamped to the table.
since the screw stops quickly, but the table has momentum, it might coast until it hits the other side of the screw
Right, but I made a 1 inch forward move, then repositioned the ruler.
it also depends which way you stopped before placing the ruler
yes see above.
get out the indicator - that's my vote :)
First I have to reassemle this pile of stuff.
dial indicator! get even more accuracy.
or the electron microscope.
SWPadnos, electron microscope would give you better resolution, and accuracy... that is if you had a scale small enough to measure it
SWPLinux: Come on, hardly accurate...
true, but not as easy to put a machine into one :)
yeah... plus the distance it can actually view is very very very small.
like the POV I mean.
SWPadnos: measuring millionths of a micron?! Might as well use elephants as a unit of measurement!
Jymmm, I like hedgehogs instead.
millionths of a micron -I'm not sure a SEM can do that
back to microns ?
but isn't there a real measurement unit called hedgehogs?
thought you guys live in the inch world :D
no, hedgehogs here... 1/16th hedgehog, 1/8 gedgehog, ...
% units inch hedgehog
Unknown unit 'hedgehog'
gaah - I can't seem to get the SATA HD to show up when I hot-plug it. bummer
mh (milihedgehog,) ch, (cubic hedgehog...)
[20:15:11] <Jymmm> http://www.unc.edu/~rowlett/units/dictH.html
maybe you're thinking of hogshead (a volume measurement)?
hogshead unit of capacity equal to 63 gallons
Heh... mutchkin old unit of liquid measurement equal to 3/4 of an imperial pint
lol... butt unit of volume equal to two hogsheads or 126 gallons
How did they come up with this one.... dol unit for measuring intensity of pain
"douleur" is french for "pain"
so some french guy made it up?
cradek jepler Ok, I tried booting off of ubuntu 6.06.1 and can't do it. any other ideas on teh axis thing?
MArque De Saude ?
he was spanish, I think
Marquis de Sade ?
[20:28:33] <Jymmm> http://www.kirjasto.sci.fi/desade.htm
Donatien Alphonse François de Sade was born in Paris into an aristocratic family.
I see. my bad
* SWPLinux never liked history anyway
The original publisher of BDSM
hmm.. XOrg-core updated today on Ubuntu for me
careful - my wife's reading the scrollback
because her computer died, and she's chacking email on mine :)
err - checking
you need a backup laptop
uh-oh - she's typing
nothing fancy, just enough of the basics.
I should stick her with the small panel PC :)
(800x600, celeron 500)
Hey, can you guys tell my sweet, adorable hubby to fix my computer or I'll have to bug him more?
SWPLinux: stop wasting time, and start fixing puters
I mean - could one of you guys tell my sweet, adorable wife, that these things take time and a large budget :)
Ingrid knows not to complain about a computer being unreliable or slow, because she knows I'll just buy a new, faster one...
jepler: oddly I lost that urge lately
alex_joni: good for you .. I still fight with it, however
That's ok as long as it's FOR ME.
at least 3 times a week I go look at the price for AMD "X2" CPUs
I think I will - fix hers fast (with one of mine), then "replace" mine with a newer faster one
WifeOFSWPadnos: I'm sure he'd be happy to let you have his computer, and he can pick up another one.
x2? I don't even know what that is..
they're pretty cheap - the dual-cores
alex_joni: It's their "dual core" CPU
$100 or so for an X2-3700
I might consider getting a very good dual processor PC soon
might be 2 x dual-core
for some simulation purposes
I haven't heard of X2=3700 .. I thought the slow one was 3800, and about $150
maybe I'm having a thinko
Intel's cheapest dual core is about $100 at newegg
alex_joni: make sure your software will take advantrage of multiple processors
SWPLinux: even if it doesn't.. room for other apps
no reason against a multi-processor system really
ah, there's a new -3600 part .. http://www.xbitlabs.com/articles/cpu/display/athlon64-x2-3600.html
true, but not as good as you'd think
if it's fast enough I think
'This processor is not yet included into the company price-list, however we tend to believe that it should sell at about $130-$140, which is $15-$20 less than the price of Athlon 64 X2 3800+.'
SWPLinux: for this one I'd want some serious GL accell
if you get a dual-socket MB< the best CPU you can get for non-multi-CPU apps is the Opteron 256
and 2 x 19" LCDs
for multi-cpu-capable, the best is a 285
and the most affordable would probably be the 265
any good page nowadays to get back in synch ?
I used to read tomshardware a while ago ;)
heh -like my machine. dual 244s, 7800GT video, dual 24" 1920x1200 LCDs
hey guys, looks like my travel overshoots by about .004 per inch.
Not entirely consistent, but fairly.
cradek, was it you who mentioned something about FERROR & round-off?
I think that was me
cradek mentioned DEADBAND
that's just backlash I bet.
but with stepgen, you should have zero "software error"
your handwheels will tell you for sure
backlaskh shows up when you reverse directions.
yes and no
you're moving so fast you may be "throwing" your axis to the other "side" of the backlash
I made a positive move first, then set the dial guage at 0, then a 1 inch move.
while your dial indicator is ... indicating, push the axis back and forth with your hand
g01 x=1.000 F10
could be your leadscrew?
.003 backlash by that method.
ok there you have it
are these acme threads or triangular?
even if it's not a reversal, huh?
davidf: the motor stops faster than the table
Triangulr I think, unfortunately.
table will still move, so it will stop at the other edge of the screw
.003 is great then, be happy
oh. momentum. Forgot that.
but you'll probably find it's different in different places along the screw :-/
somebody mentioned that earlier ... ;)
yes they did.
I have no excuse.
you can try software backlash comp, but down that path lies madness I think
try the old "two nuts" method
I can see how that might be a problem. Especially since it changes over time anyway.
yeah two nuts with a spring between can work nicely for low cutting forces, but they increase drag (and wear) on the screw of course
I have an adjustable nut. Two set srcews in a bronze nut, 1" long.
well, get adjusting ;)
Split nut, one side.
before adjusting you might want to measure the backlash at several spots along the screw
you can get a feel for how tight you should make it then
today, i have had a good day slicing stainless
* alex_joni heads to bed
robin_sz: seen my message this morning?
the laser is cutting really good now, much better than usual
robin_sz: read about a nice 454W laser diode lately ;)
no, in here
lots of cool needed to operate it :D
I knew 150w was do able ... but thats 3 times ...
a few of those and YAG rod ...
with 450W you can already cut some 1-2mm sheets
maybe 3mm even
now I'm gone..
you cant actually use laser diodes directly, as the beam quality is poor, but fired into a YAG rod, they are very efficient
Thanks for the help earlier...
Tightened nut, switched to 2000 uSteps. Big help
Looks like main source of error now is non-linear lead screw only.
Guess I need a real one for better accuracy than this.
But it's pretty good now.
Interesting, you can see the non-linearity easily.
is it a periodic error?
Set for 0 error at x=0, it goes up to 1.5 thou error and back down to 0 at 0.1 intervals, till x=.5 inch. Same thing on the way back.
Yes it is.
I replaced my (triangular) leadscrews with 1/4-16 acme for very little money
looks roughly sinusoidal
Really interesting. I can go back & forth from 0 to .5 at 60 in/min and you'd never know it was there unless you stop in between.
.5 is an integer number of turns
go to .45 (half a turn away) and I bet the error is at its max
How much for a 13" acme do you think? (1/2 inch)
your screws are 1/2-20?
ok i'll try that...
36" of 1/2-10 is $28
is that odd or something?
seems like a very fine thread for 1/2"
would 20 be about the same?
I don't think there is such a thing
Remember it's triangular.
Probably cant fit a 20 tpi acme on a 1/2 inch rod. :)
But 10 would work.
you can get 3/8-16
36 inches for $28 sound like a wiener to me.
weiner I mean.
yeah you'll have to come up with some nuts of course
you could get delrin anti-backlash nuts for some really nice results
Hmm, this could get bad.
pun-wise I mean.
Delrin will hold up ok huh?
depends what you want to cut I suppose, I'm no expert
How much for those?
I'm trying to figure out where I bought mine
mostly Al & brass, soft steel, plastic.
So re the integer # turns, you're thinking round-off error, or machining error in the screw?
oh definitely screw error
again you can see emc is doing the right thing by looking at your handwheel
yes thats right of course.
duh .45 is also an integer number of turns (.05/turn)
you need to check .475 to show the error I think
[22:37:17] <cradek> http://cgi.ebay.com/Brass-Anti-Backlash-Acme-1-2-10-RH-Nut_W0QQitemZ230018414330QQcmdZViewItem
this is pretty
but for low cutting forces I bet a spring-loaded setup will work better
if you have a lathe and an angle grinder or somesuch, you could use some of your extra screw to make a tap
that's what fenn did to make his own delrin? nylon? nuts
actually the error is 0 at 4.75.
0.475 I maen.
maybe just wear.
that wouldn't explain a sine error profile though
well I just took a swag. Only checked 6 points, & just saw it got up & back down
is .003 important for the parts you want to make?
No big deal. At least it's close enough for most stuff till I get a new lead screw.
why invert don't work on parport? I'm using xylotex pinout ( I'm on a XP now EMC is on slack on a diff PC)
Mhel: you can use inverted parport outputs
cradek, yeah the enable-out-invert doesn't work
Mhel: see http://linuxcnc.org/EMC2_User_Manual.pdf
around page 49,50
I'm sure it does...
cradek, that's how I learnt about the invert, but emc gives an error if i put invert.
it's on page 50 actually
what command exactly do you have in the halfile?
it's the default xylotex pinout, I just added -invert after -out, is that correct?
no, look at the page 49 diagram
the pin's name doesn't change
the invert bit is a parameter for each pin
so you use something like "setp parport.0.pin-02-out-invert true"
no, I didn't
Ill have to read more
one more thing, where is halconfig.tcl lcoated, axis can't seem to find it
halconfig.tcl isn't in the release, it wasn't finished so we didn't include it
there is halshow, which lets you look at the hal but not change it
ok, I just get an error halconfig is not there.
if you're not running emc 2.0.3 you might want to update (I think the halconfig/halshow problem is fixed in 2.0.3)
I'm using 2.03
maybe you didn't update the emc2-axis package to the latest too?
in my 2.0.3 with axis, the machine menu says "show hal configuration" and correctly runs halshow
actually I'm a bit confuse with the version of axis, 1.32a is what I got
but, there's a mention of 1.4 for debian
oh you're not using the ubuntu packages? that explains it I guess
I'm not sure if it's right everywhere (anywhere) else
I'm using slackware, I happen to have it already from a magazine, and I'm still on dialup :(
lerman_ is now known as lerman
it was a struggle to keep emc2 and axis in sync, now it'll be better since they are together
sure, I see
yeah, I couldn't install axis separately, but it's easy building it inside emc
I think the Ubuntu deal where they will send you CDs for free is really nice for dialup folks
you have to be a bit patient though
Ill try that
the thing is I'm new to linux too, and didn't know any better
once that's going, updating emc over dialup will be quick and simple
if you got rtai and emc working on slackware by yourself, you're no longer "new" in my mind
building a realtime system is not a trivial task.
but I'm confident with Linux now, once I learn how to go online with it, it's a big leap
just patient with reading docs
just curious - what do you have to invert for the xylotex?
I'm using gecko201, I have a homebuilt router
oh you said xylotex pinout, not xylotex
in Axis directions are inverted, then I looked for info to change it and found -invert
easier is to make INPUT_SCALE negative in the ini
either way will work fine
atleast there's a way ..:)
what's your timzone btw, most of the time noone's here
I'm central time US
people are here all the time, just sometimes quiet
the regulars are mostly in US and Europe
where are you?
I'm in montreal, originally from Philippines
~80 miles from me
(I'm just outside of Burlington, VT)
I thikn so - 7:06 or so
we're mostly -7 to +3 I guess, you're in the middle
gtg, and see if I could make the invert works thanks
ok good luck
I remember dialup when I had to disconnect when not using it... yuck.
[23:07:24] <cradek> http://www.ubuntu.com/FixForUpgradeIssue
wonder what they got wrong
03jepler 07HEAD * 10emc2/configs/dallur-thc/dallur-advanced.hal: wait-for-pin is no longer needed
i may pull this landing gear plan off WITHOUT any machines or manpower of mine own... for the same margin.. ; )
spindles seem to be lining UP.... like the stars do some time.. and PRETTY ones too..