* MattyMatt crossposts to #reprap
andypugh something tells it it may be capable of both
it looks hung up on the side though in the picture
[00:03:03] <Valen> http://wpcore.wpe.s3.amazonaws.com/wp-content/uploads/2010/05/MtOrab-windgearboxpart01.jpg
mach3 users spare no expense! http://bdyoutube.com/video/bXxKAXbg3SA/Cheap-5-CNC-axis-mach-3-controlled.html
Would have been a lot cheaper with EMC2
my fave was they couldn't evne use a clean piece of paper
so, is there any FOSS CAD software even worth bothering to look at?
I wasted so much time checking into EDA stuff I figured I'd ask first this time
did you find eda stuff? :P
What's HeeksCAD like now?
celeron55 none worth wasting the time on
I tried kicad and geda
i've used gEDA to order boards from a fab
well to its credit you could finish the demo before it crashed
unlike my experience with kicad
which I believe crashed repeatedly at step 3
i've not experienced any stability problems with geda, it's very stable software
I probably ought to pull down a more recent HeeksCAD and see if it is less flaky.
HeeksCAD quickly failed in simple drawing for me. It needs more blender's ctl keys
kicad seems to work ok for me
it didn't here
for CAD, HeeksCAD is the only "big package" which can do some stuff
I'm blendered. I'm incapable of using anything with the mouse buttons the normal way around now
so check out heeks?
I kind-of lost interest when I bought Alibre, and that might not see much use as it seems I am going to be given a free home-use Autdesk Inventor license.
I've tried qcad
yes Heeks is good
and getting better rapidly
and long ago blender
well when i tried qcad that wasn't exactly yesterday anymore either
heeks has some unstability and lackness of features (not very surprising?) but people use it to generate g-code for some real stuff
and development is active
nah i don't need it for any g-code
just need it to draw with
g-code would be cam wouldn't it?
pfred1: 2D or 3D?
it's not very good in drawing, targeted to machining at the moment (though intended to be general purpose at the end)
andypugh 3D would be nice but
I think it's a lot better at 3D than 2D
honestly now i do my computer drawing with xfig
I'm just looking to try to make a step up from that I suppose if I can
that's more like visio tho
The 3D packages aren't great for 2D drawings, unless you make a 3D model and let them draw it. Of course, they get it _right_ so it's worth the extra step :-)
Flash is still the best 2d drawing prog, for freehand stuff especially
...there just is no open source software that does it
I can draft on paper so I can get plates correct
if you want animation
Right, time I was off.
I just never had much luck drawing with computers
I should be making my gcode for tomorrow
how about inkscape? not targeted to mechanical drawing, though
yeah I've used inkscape
to make like elipses and whatnot
Flash is much nicer to draw with than Inkscape
'to me it almost seems random
like plunk something down try to move it where you want it
its not really how I approach a drawing project
well I'll give heeks a try
you don't want heeks for drawing
that bad huh?
qcad is better at 2d
I mean wth? can't anyone make a decent drawing software suite?
yeah qcad seemed like it had some potential if I spent the time to learn it
qcad seems a rip of auto
mine is a rip of Flash. did I mention I like Flash?
but the freebie qcad is pretty crippled isn't it?
just what is this flash?
the editor for making SWF
pfred1: i'm not sure whether the commercial one has a lot more useful features or not
I've really been leaning more towards physically modeling my machine lately anyways
I have version 2.0 licence. macromedia were giving away full licences for 2.0 when 4.0 was released
at least the grouping system of qcad lacks stuff (i couldn't find a way to copy stuff between groups (or layers, whatever))
the free version at least
some softs you have to cut paste
I know eagle is like that
took me a bit to figure out
cut then hit the traffic lite
and qcad's UI is very unintuitive and slow
well group select cut traffic lite
like group select cut traffic lite is intuitive
heck til I learned about it I was convinced it wasn't possible myself
btw, could you list the main requirements of the software you're searching for?
really I love to draft but the state of computer drafting still seems crude to me today
I want to design my CNC machine
and I figure drawing it out can't hurt
I made a reprap mendel in blender. it doesn't move yet but it could
and I drew a design for mine in blender too
the last time I used blender i manged to create a sphere in the dark after much fiddling about
yep that's blender. steep initial curve, and then it's all uphill from there
like olympus mons
I found blender a bit less user friendly than povray
actually, i sometimes have had a need for 2d drawing software a bit like qcad but faster to use and better - but i haven't found such
honestly its the best stuff I've used drawing on a computer
the name sounds good at least
you don't know what xfig is?
haven't used it and haven't seen it
but after 10 seconds i have!
celeron55 you don't run Linux?
xfig is the default Linux paint sort of
my default linux paint is kolourpaint or mtpaint or something more random
well back in the day it was xfig or bust
gimp here ;->
though lately distros haven't been shipping it by default so much anymore
xfig predates gimp by quite a few years
if i need to scribble a few digital lines its xfig for me even today
my first distro was mandrake 9.2 - i guess it didn't ship xfig
well if you run X you can run xfig
ldd `which xfig` is about as basic as it can get
the gui seems a bit odd - and i can't find wheter i can enable a big crosshair cursor
its the snizzle for basic 2D drawing though
I just can't believe its the pinnacle of CAD
I swear to god I can slap in a linux distro from 15 years ago and get almost the functionality as i can with the latest and the greatest
probably - though youtube wouldn't work 8)
i use icewm as a window manager
that'd only aid productivity
I like windowmaker if I want a really bare bones WM
well this is fast, stable and has everything i need - and has a very familiar UI
ever tried Windowmaker?
not really, but it does look a bit odd
add on wmakerconf and I forget there's another configurator and you're all set
it looks like next
well its supposed to
it was what came after afterstep
so how is it better than icewm, for example?
I've run ice but not enough to know its features it wouldn't suprise me if it wasn't lighter using resources
i don't use any menus, i just launch terminals with win+t and programs with win+space + write launch command
wmakers menu is pretty handy
i have tested icewm and fluxbox on an old machine - fluxbox was way slower :P
you can write the commands then just launch off it
yes wmaker is better than any of the boxes
but really, i find it very unnecessary to change WM, the current one works just fine
does ice have configuration tools?
no, i don't need such things
yeah/ you just take what you're handed do you?
there are config files
not much to change in default config, though
by the time you open one up I'll be using wmaker
i have my configs already available and machines set up - no need to start using it again
because i am already using it
you're not going ot sell me on the theory that hand editing config files is superior to custom gui configuration tools
i'm not trying to
then what are you trying to say?
i have just answered your questions
having run linux for over 15 years now and having run more WMs than I can even remember i don't have any questions at this point in time on this subject
you can run X with no WM at all you know?
"does ice have configuration tools?" "yeah/ you just take what you're handed do you?"
i wonder what are those if not questions
celeron55 they're leads
they lead you down a garden path
just like this isn't really a question either 20:52 < pfred1> you can run X with no WM at all you know?
when you do that its called running weave
i can tell you one thing: i have once tried wmaker, some years ago, when i was finding a good wm to stick to
I've no WM loyalties
i find this being a bit like the same thing like you and gEDA - i find gEDA very useful, you find wmaker. you find gEDA not very useful - i find wmaker not very useful.
I ran KDE for about 6 years then when i uppoed to ubuntu I hated the KDE they shipped so I'm using gnome now
not that i have any great love for gnome but at least it doesn't have ugly menus
i've also changed away from kde - after they dropped support for 3.5 in debian unstable
kde 4 is awful and i don't like gnome either
I think i was running 3.5 for a long time
was whatever I got with suse 8.2
but that machine literally blew up so it was time to upgrade
I can't design circuits in geda like i can with eagle
that is a valid reason for not changing software, if you can stick with eagle
in 10 years I'll give geda another shot see if its any better
i have to admit they made some progress in the last 10
celeron55 you got any stuff you've made with geda online?
anyone know when iso's with the emc will be made?
tlab 2.4 ?
well ubuntu 10.04 + emc
is 10.04 a LTR?
I believe so
well whatever they call it
yes it is
I still imagine changing versions is a lot of work for the EMC maintainers
they are having issues getting RTAI working with the kernel in it is my understanding
but it has been running
I'm usually off by a factor of 2 in my guesstimates and I think 6 months isn't enough time
pfred1: what kind of stuff?
tlab what is so great about 10 ?
I dunno, I was just wondering
celeron55 boards you've made with it
here's a photo of the one i ordered http://celer.oni.biz/~celeron55/random/2010-04/dcdrive2/sdc10239.jpg
i've also designed lots of DIY boards with it
here is the prototype for that thing http://celer.oni.biz/~celeron55/random/2010-02/dcdrive2/sdc10161_m.jpg
beats what i managed to do with it
it's a general purpose motor controller for random stuff that comes to mind, also i have half working software for driving small bldc motors with it
but anyway, doesn't really matter what i do with it, does it? :P
yes I've seen people using the stuff i cannot imagine how they do
it worked like ass for me
there are other people too using it for real stuff - but indeed there might be a steeper learning curve than in eagle for example
yeah i don't have the time now
but i love it how everything is scriptable, for example
I needed my stuff done yesterday
and now it is
if I stuck with geda I'd still be trying to get 2 resistors together with it
I think the biggest problem I had was the lack of integration between the parts
i find it being a good thing, and it is a goal in the project
well that is counter to how I design
it's nice there are different programs with different approaches to EDA
something for everyone :-)
as i design the physical board the schematic can change to make things flow better
kicad doesn't work too nicely for that situation either, i guess?
I mean I do have to eventually build ths stuff
kicad crashed on me on step 3 of the tutorial
so whatever it does is moot
well, i found it being rubbish in usability, so didn't get to try too much after i found geda
did you try the hal libraries for geda?
hal libraries for geda?
yes, the hal libraries for geda ;)
look on the wiki
ah, i see (google helps)
haven't had any use for those
there is an impressive example done in kicad too but how anyone managed to do it is beyond me
they do look cool, though 8)
the thing that surprises me most is that some people say they like kicad better than geda
I've met them
which was why i gave it a shot
it did look promising right up until is bombed out on me
kicad has supposedly more users because it works on windows
geda ran as far as I remember it is just so all over the place i couldn't deal with it
I have gschem running now i wanted to check what version I have
the gsch2pcb thing surely is a dirty hack but it does work for me well enough
looks to be one from last year
pcb is a very impressive piece of software - they say the first versions were written for some amiga
well I really did run geda over 10 years ago and I have to admit I was a bit dissapointed it hasn't improved more in the time between uses
I guess I was just expecting a bit more
probably lack of developers - there really are not many that want to use EDA on linux
linux is marginal stuff, EDA is marginal stuff -> linux+EDA shouldn't practically even exist 8)
people who run Linux are the kind of people who do EDA though
well, a bit more often, yes
lots of "normal" people run linux on these days, though
to its credit I'll say geda has attacked EDA the UNIX way i just prefer a more integrated solution
or rather it is a lot easier for me to work in a more integrated suite
my first schematic is never the one I run
i guess one *could* make scripts to provide better integration between pcb and gschem - mainly automated updating of parts and netlist. i guess it has been discussed by the developers already, though
in eagle i just hit a button
and if both windows are open it auto updates too
and that is EXACTLY what I want
huh, I found kicad fairly easy. I've made several boards with it now.
there is a graphical glue program for gschem and pcb: http://geda.seul.org/wiki/_media/geda:screenshot_gerbv.png
but not close to eagle
it was the first EDA program I'd ever used as well...
mozmck kicad couldn't have been any easier for me about 6 mouse clicks and it froze up
heh, haven't had that! used it for a couple of years now on and off in windows and linux. haven't tried it on ubuntu 10.04 yet though...
9.10 here could have been a factor
when i find the time I plan on upgrading to Debian
looks like i installed geda around 2010-02-27
heh i liked kicad so much I must have delled its source tree
I see I have kicad-2010-02-28-RC5-UBUNTU_9.10.tgz lying around
geda is a bit like any unix-style software - basic installation isn't very special, but gives you most you need for setting up a nice toolset for doing stuff
not good for someone just wanting to get something quickly done the first time
well i have put in some time with eagle now
maybe it'd be nice to have someone make a package with a configuration that will get everyone started fast
a bit like a "geda distribution"
I found a few tutorials
well... dunno :P
stuff ran but I'm just addicted to the IDE
hmm... i'm too addicted to IRC, it's 5:00 AM now in here
pfred1: hmm, I found you need to use wxwidgets from ubuntu, not from the wxwidgets repository (if I remember correctly)
There is a ppa for kicad that I use and it has worked well for me.
mozmck is that what was killing my kicad?
the wxwidgets thing? could be.
it just ticked me off that like on the third step of the tutorial the stuff was dying on me it didn't instill confidence
here's the ppa I used: https://launchpad.net/~paxer/+archive/ppa
Makes it real easy to try.
I can never keep the ubuntu names straight
I wish people would just use the numbers
karmic is 9.10, lucid is 10.04
but when I see them I'm like yeah i knew that
which only serves to further tick me off
soon as you put up karmic is 9.10 I was like yeah thats what i have
I know they got the idea from debian but it just doesn't work as well for ubuntu
there are too many versions of ubuntu
yeah thats what I was thinking debian doesn't release what but once every 2 years?
and ubuntu is on a twice a year cycle or something
something like that
so yeah its just too many names and numbers flying around
in debian the numbers are meaningless the names matter
I've heard some terrible rumors of what to expect in ubuntu in the future
You can stick to the LTS releases, one every two years I think.
yeah, I'm not sure at all about this gnome shell.
Tasks: 155 total, 1 running, 153 sleeping, 0 stopped, 1 zombie
thats what I'm not too keen on
I may be scrounging around for another desktop environment again.
for what I'm doing i should be getting by with 2/3rds of that
108 processes max
I don't mind xfce. I have tried kde several times and never have really liked it.
hmm, i seem to have 193 taks running
ubuntu just has too much overhead
try xubuntu. I use it on my cnc router in the shop.
though i've probably installed most myself, as this is debian
I'm gonna bite the bullet here and just config a vanilla debian distro for my desktop
that's what i did
yeah its a pain to config but I don't change PCs that often
in a couple of months i should have an rxvt term going in X :)
hmm, i can recognize most of the tasks that are running
seems that many programs run many tasks, not a single one
I'm thinking KDE has to die because kernel tasks start with K too
what was funny was when I ran KDE i liked gnome apps better by and large now that i run gnome there's a couple KDE tasks I prefer in it
for example, i have many sftp connections in konqueror at the moment
every tab has it's own kio_slave task and also an ssh task
yeah for a file manager i prefer konq over the gnome deal
though gnome fm deals with file thumbs better
kde4 has a new file manager, dolphin
i think it sucks, though
I use konq
I like the location for copy paste into terminals
also, amarok2 sucks... i guess i'm sticking with 1.4 to the end of the universe
I swear half the time I think we're going backwards with software
pcs are 10X faster but they're 2X slower using them
and they don't appear to be doing much more to me
I agree. They make the software flashier, harder to use, and more buggy.
they're making text harder to read with transparencys and wasting screen space with different kinds of effects
mozmck when you got that 486 DX4-100 it was the most smoking CPU on the planet
and everything takes cpu and graphics power out from real use
heh, yep, and cost a fortune too!
now everyone has quad CPUs and they don't do much more!
uh, 6-core now.
* mozmck kicks mozmck-6core
each one of those cores is like 10X the 486 what is going on?
more like 30x
or even more
what made me upgrade was quake the machine gun was so slow on the 486
It does compile and build packages for a ubuntu kernel in 18 minutes instead of 1h30min or so on my dual core.
ok dev softs is still good its userspace that has bitrot
mozmck: but did they wait 10 hours for compiling stuff on 486's?
no they didn't, software was quicker to build, too
actually I built X world once and that took like 8 hours
on a pentium 100 I believe
when it finally finished i fell to my knees
I was so happy
celeron55: uh, some software. I knew developers working on things like pbx software that said it took 2 hours or so to recompile.
And that was just part of the software, years ago.
there was a damned bug in X and i had to build it
8 hours is a long time. I made some mistakes on some of my kernel compiles and it would quit with errors after it had made most of it, or while building packages.
that was aggravating.
I'm afraid to try to compile a Linux kernel anymore its so complicated
it's not that difficult
the gentoo folks do that all the time - just be sure you've enabled support for everything needed to access your hdd partition 8)
if you build everything as a module no it isn't but if yo utry to just ubild what you need it can get to be a bit much
you can use that module loader
after that, you can add stuff gradually - it doesn't take long to build a stripped-down kernel on recent machines
even less a few modules
it was a lot easier in the 2.0 days
the first kernel i've built was 2.6
best linux kernel of all time was 2.0.36
thing was a rock!
why is that?
I net there's still machines running it that haven't rebooted since the day it was released
I think I ran that when I first started using debian.
installed it from 6 floppies I downloaded
I donno it was just a magical version
I think the bulletproofed version is .38
but when .36 came out it was unstoppable
crap never crashed it just went and went and went
the cool deb install is the 2 floppy net install
then you can IRC in another term while you're installing the rest ;)
you can use it still
celeron55 if there is one thing running slackware for 7 years taught me it was the value of package management
best linux distro of all time was RH 7
apt works very nicely, too
broken package management sucks a lot
no package management sucks worse
how is that?
try it someday
then you have to keep track of all versions
of all libs and userspace binaries
trust me it can get to be a bit much to manage quickly
ah, i read wrong... you're right
which back in the day running slack was pretty much where yo uwere at if you wanted anything they didn't ship which was just about everything
like i heard about KDE and I was like oh oh oh i gotta try it
but damned if I was gonna run that pussy red hat stuff
where i could have just rpmed it
oh no i was going to compile it
yeah well 3 weeks later i got it to work
spent 15 minutes in it and that was that
that was KDE version .4 on slack 3.4
see back then slack was libc5 based and KDE was stricktly glibc2
and getting slack to play nice nice with glibc2 proved to be less than trivial a task for me!
amazing how many shells are dynamically linked though
those were heady days nothing is so exciting anymore
what ever happened to World Domination?
I suppose so in the world of supercomputers
linux reigns supreme there
something like 90% of the worlds supercomputers run Linux
but Linux is a hard fail in userspace
Android is doing pretty well...
eee's did well
'till MS started twisting arms...
probably find that the linux on eee was mainly a negotiating tactic to get MS to drop the $130 price of windows in a $400 item
but I got one
pissed off that crap linux it came with and put ubuntu on it
heh Microsoft Shifts Robotics Strategy, Makes Robotics Studio Available Free
got a touch screen I need to wire up to it then i'll put the netbook remix on
[03:31:29] <L84Supper> http://spectrum.ieee.org/automaton/robotics/robotics-software/052010-microsoft-shifts-robotics-strategy-makes-robotics-studio-available-free
healthcare robotics powered by microsoft, sounds like a self sustaining industry :)
oh snap ;->
Yep, could make the BSOD be the real thing...
PCW_ is now known as PCW
is it possible, to use emc2 for a 4axis foam cutter?
I think so
but the cam i use (inventorcam) does not export files with xyuv
there are still only 2 axis used?
maybe you'll need to get a better cam then
gcode file looks like this
[11:07:02] <numen_> http://nopaste.info/fd960b3a47.html
celeron55 but which one?
i need for cutting model airplanes
or the wings of them^^
numen_: if you have only parallel movement of the 2 sides, then you have a 2-axis machine
even if there are 4 motors
what is D215 ?
maybe you need to tweak (change) the postprocessor from inventorcam
alex_joni nope not parallel movings
the M6 looks kind of lost, the G43 is at the end by itself too
look at all post processors it comes with and modify the one that is already closest to making a usable program, not sure alex_joni agrees, the program looks quite messy/unstructured to me, i'd want a post processor that's closer to start modifying
or just write your own g code
JT-Dev it is a 4axis wire cutter
I saw that :)
a hot wire, like a edm? then the m6 and g43 wouldn't really be used and could just go bye bye i guess
it's just two parallel axis machine so not to complex to hand code
you should handcode something simple, then adjust the post to replicate that
a plane wing sounds like something i'd like a cam software to post me something
where is the D215 coming from alex_joni asked you about?
hmmm. apparently I did a "what does Ubuntu 10.04 do when the battery runs out" test last night
cool! apparently it goes to hibernate mode or something, so it starts up where it left off
hey SWPadnos, so I was pondering putting my newfound AVR stuff to use for the purposes of PWM out and encoder counting, I was wondering what the best way of interfacing it to EMC would be
I'd prefer a dual port memory sitting on the PCI bus, but it looks as though thats a pain in the butt to do
yes, yes it is :)
parallel port 8)
a parallel port or other simple digital I/O card would be the way to go I think
unless there's a hardware SPI chip on a PCI card, all ready and waiting for you
it seems to be a pain mainly because nobody makes stuff like that for PCI anymore
hmm there are lots of I2C busses floating around inside motherboards these days
parallel is probably still better though, even SPI will still have latency, which HAL doesn't deal with very well
USB would be really nice
but yeah latency is a pita
yeah, especially for USB
USB3 might be different, but that's not likely to be easier to implement :)
SPI latency would be set, and not nescaserialy higher than parallel port
PP bandwidth is 2.5mbit isnt it?
I think with EPP cycles it's around 1MB/sec
not bits, bytes
a bit faster with DMA, maybe up to 2MB (but I'm just guessing on that)
hmm, I spose mesa can do it
[13:25:15] <Valen> http://en.wikipedia.org/wiki/Light_Peak
would be a nice interface ;->
a mesa FPGA card is the simplest solution there, I think
drop the AVR
oh thats the easiest off the shelf, but for a lot of people mesa is too expensive
are you using one of the XMega or whichever it is that can actually do quadrature in hardware?
I was thinking theres alot of people who like mucking about with AVR's so if there was a way to enable that as a backend for EMC it might be handy
I/O and "backend" aren't really the same thing :)
i know but you get my drift
well, it really depends on where in the system you want the connection to be
encoders/PWM/step-dir is one thing
any motion control is another
I own the complete mesa tool chain and use it for my mill, its great and works out of the box, its just a little spendy and not "student friendly"
nah just ecnoder/PWM
EMC does all the other stuff fine
right, so you need some other hardware so the AVR can talk to the PC in a non-latency-screwed way
thats the bit of glue I'm pondering at the moment
I wonder if RS232 might do the job
a $15 PCI parallel port card is probably it
last time I looked it was in the ballpark at the higher bandwidths
no, serial is sucko
why hate on serial?
I use serial all the time, but in this case, I think it ends up being too slow
and I don't know that we can use devices like that in the RT code
how many bits are the encoders typically?
as in their representation to EMC
115200 baud is 11520 bytes/second, which is 11.52 bytes every servo cycle
if you want to support 3 or 4 axes, you have 3 or 4 bytes per channel, including overhead and other I/O
encoders are 32-bit internally, but hardware can be whatever part of that it needs to be
some are 16-bit, others 24, maybe mesa is 32 - dunno
SWPadnos: 115200 bauds = 14.4 bytes per sec
maintaining the 32-bit value internally is the responsibility of the driver
micges_work, until you add start and stop bits
PWM out is going to be 8 bit, encoder in can be 4 bytes then
those can happen simultaniously
but yeah, reaches the limits rapidly
and remember, it takes the full millisecond to transmit, so there's still that USB-like latency problem
how many I/O bits are available on a parallel port?
I think it's 5 in, 4 out, and 8 data, which can be I or O or I/O in advanced modes
and theres enough bandwidth for a simple address + data/command system to be implemented fairly easily
there have been discussions about using serial, and though it could work OK-ish, it seems like it's not a great idea, nor is it very expandable
also, serial ports are at least as rare as parallel ports these days
I think the problems with the serial stuff is they are not dramatically faster
so your not seeing the latency drop as a result of brute force speed
considering that all the Pico Systems boards, and the 7i43 from Mesa use the parallel port, I'm sure it's fast enough
yeah, and its easy for the punters to play with too I spose ;->
well, if you could use a serial port at 921600 baud, it might be different
SWPadnos: serial ports are more 'industrial' than lpt, at least here
and there are serial ports that go that fast, just not over very long wires
I was thinking of a stackable architecture so you can just add a "module" be it stepper or servo
not common/cheap though is the problem
micges_work, yeah, thinking about it, I was wrong to say they're as rare as parallel, they're more common, and you can use RS422 or 485 if you need better environmental performance
Valen, see the pico-systems PPMC :)
bring the PP wires out to a stackable set of pin headers, stack up modules containing an AVR and suitable drivers all in one
parallel port connected card cage, add DAC, motor drives, I/O, etc. to it
[13:38:53] <SWPadnos> http://pico-systems.com/PPMC.html
yeah looking at it now
they still need external amps though
they're not student-cheap, but that's more or less what you're talking about
yeah, similat idea
SWPadnos: we are testing 1kHz half duplex com rt driver now, when it will be working we paste all info at wiki
oh, I thought he had an amp that would sit in the cage, but maybe not
I wouldn't have that photo on my "buy my stuff page" ;->
it looks like it was cut with a jigsaw and bent with a phone book ;->
the cage itself
the web store photos look better
micges_work, hah, put 8 of them on the parallel port 8 RT serial ports
except that you need to use all your CPU time bit-banging
lol it was a joke yaknow ;->
Valen: 10 bytes on each frame (simmilar to modbus)
I'm thinking if it can be made to work with the hobby brushless motors it could be pretty nifty as well but I'd save that for mk2
ok so the outside the PC stuff is sorted, parallel port interface, some kind of addressing and command data protocol
so you can stack as many cards as you feel like onto it (say 256 ;->)
take a look at how Jon did it on the PPMC
there's a 256-byte address range for EPP
oh I'll stand on the necks of every giant I can find lol
he split that into 16-byte blocks, and uses one or two blocks per device
you might be able to do something interesting with wiring the high address bits, similar to the way PCI shifts interrupt lines around between the slots
I was basically thinking of using the 8 data lines to xfer data, then use the 4 output lines to command all the cards into a simple "listen for your address" "read command" "report result" kind of mode
via bit banging
probably better to use epp if you can, I think it lowers the CPU wait time
use one of the I/O's for data ready from the cards perhaps
probably I haven't looked into it in too much detail
since the address cycle and increment is automatic
I was also thinking of trying to keep it as conceptually simple as possible
I think you set the address register once, then when you read from some location, it automatically increments, and the hardware deals with the address/data cycle
it matters little at this stage anyway
both systems would probably work and as mesa is already pumping the data around without too much overhead i think we can assume it works lol
the AVR card might need a little help from some external logic, or possibly one of the chips that has an external data interface
why do you say that?
I don't know the exact timing of an EPP cycle (you can look it up), but if it can transfer 1-2 MB/sec, that means that the address, read strobe, data valid, etc happens in 0.5 - 1 microsecond
ok EPP doesn't seem to outrageous by the look of things
the AVR is going to be running at 20mhz with nothing much to do other than twiddle its thumbs
which means the AVR has to respond pretty fast when a read happens for example, and since the interrupt latency is 7-10 cycles before you can output anything, you'd need a very very fast AVR to keep up
at least until it gets to running a brushless controller
encoder counting isn't twiddling its thumbs
is if its hardware
so an XMega?
or an AVR32?
the avr32's still seem scary
since the normal AVRs don't do quadrature
yeah. I need to design a couple of boards with those in the next week or so, and haven't gotten to the point that I like them yet
eh worst case i'll do it with PIC's they have quadrature inputs and some kind of parallel port interface
not to mention the software
that would be a worst case :)
i'm still on my first AVR project at this stage
heh. not me :)
I'm normally aPIC person
but the wind seems to be blowing the other way so, I'm sampling the dark side
at my old company, we were one of the first places to design in the AT90S1200 when they first came out
and 12 months later when you got stock?
we actually discovered an error in Atmel's temperature testing methodology
no, we got them right on time
oh wait they were probably really good clones ;-><
I didn't say we were the fastest kids on the block ;)
we moved over from the PIC, and I have never gone back
anyway, what's involved in the PC side of things?
software, similar to existing software (if you use a parport)
I'm dabbling with an avr32 right now, with ethernet. What don't you like about it SWPadnos?
hell, you could make something that uses the same register structure as the PPMC or USC, and have the driver finished already
I've got a driver (mostly) working for a ks0108 lcd so far.
mozmck, my experience with more complex microcontrollers/DSPs hasn't been all that great
yeah, but people get shitty when you clone their stuff directly
since there is so much initialization to do, and the software support for peripherals is more complex, it's harder to hold all that stuff in your head
or my head anyway
atmel have an app note showing how to get 60kbytes/sec over EPP in a 4mhz AVR
with code for both sides
I like to know exactly what's happening in a small embedded system, and the AVR32 is a little too complex for that so far
I want to have an idea about everything but not worry too much about the details
that's 300 kbytes/sec at 20 MHz (assuming it scales linearly), which is still 1/3 or less of the theoretical bandwidth
but more than we need i think?
wheres that calculator
the USC from Pico has a 32-register map, separate for read and write
it's a bit different, but I'm getting the hang of it. I don't find it too much harder than the PIC chips I've been using (I've also done one project with an atmega)
I think a full update is about 50 microseconds on that unit (I don't recall if that's read and write or just one of the two)
we have 4 bytes of encoder, call it 2 bytes of PWM, and we want to push it out at 1khz?
oh, I've been using ATMegas for - well since they came out :)
Valen, that's per axis
i get a data only rate of 22kbytes a second for 9 axies
6 * 9 * 1000
I'm betting address setup is more significant when you only transfer 6 bytes at a time
and 6*9 is 54
dunno where that other thing came from
mozmck, I have one AVR project that's over 800k of assembly source
Mega AVR, of course
even if you double that its still in the ballpark for the parallel port and that app note to handle assuming it scales
but I can still look at the code and know exactly what can affect what else, and when
Valen, yes, it's fast enough, it's just not "super-fast"
doesn't need to be ;->
the beauty of running on a fixed clock, provided you are done before the next tick it don't matter if its by a mile or an inch
weeeelll, yes and no
if you read feedback data, then calculate a new command and send it, if that command doesn't arrive until just before the next bit of feedback is captured, you have a phase problem
true, but provided that is consistent one would assume that it would tune out
the pico and mesa products work well at 1 kHz because it takes <10% of the servo period to read, modify, write everything
if they took 90%, it would be a problem
I would presume though if they can do the same job then I should be able to be at least half as good as they are
sure, you can tune some out (I don't know the specifics), but I think you lose performance when you do - you can't have it as aggressively tuned
sure, it's a very good starting point
oh, take a look at the stuff - uh - some guy recently posted
he made a parallel port I/O expander that had EPP handshake circuitry on it
oh that thing about the stuff?
yeah, that one
or the other one
nah not that one
oh, you're right
that was crap
I was going to get you a beer, I'm starting to think you dont need one
I don't drink them anyway
not enough coffee in them
you can get coffee flavoured beers
because some people feel the need to destroy anything that is good in the world
you can buy me a coffee though, that's always OK
If i ever return the favour you can get me a girly cocktail, the kind with lots of milk sugar and chocolate in ;->
but now I can't visit Australia - they might see all the porn I have on my laptop and phone
and sieze me or something
on the way back to america?
no, there was something yesterday about AU customs being allowed to search laptops and phones for porn
oh like what america has had for ages, where they take your laptop, dump the disk and arrest you if its running linux
no, that's for terrrrrrists
yeah the ones that use linux
[14:11:22] <SWPadnos> http://yro.slashdot.org/story/10/05/20/1313213/Australia-Air-Travelers-Laptops-To-Be-Searched-For-Porn
We use serial to drive our 8I20 (and will for newer amps we are working on to save I/O pins)
but our default baud rate is 2.5 Mbaud
yeah just spotted that
SWPadnos: that's a lot of assembly. I have only used C.
good luck finding all my porn hidden in the random noise in the encrypted volume with all my work stuff on it
pcw_home, doesn't the hardware scan all the time and present the latest results to the PC in a register?
mozmck, yeah, it's about time to make that one C
pcw_home what are you using to get the data out of the PC? some kind of high magic hardware I'll wager
SPI in FPGA
yeahp, that'll do it
Well not quite, for interfacing to the 8I20 with EMC we use the sserial HM2 module which has a simple 32 bit register
interface on the PC side and does all setup an protocol internally
8I20 is RS422 asynchronous serial (LBP protocol) to minimized isolation cost
if one wanted to move child porn from one country to another, it seems like carrying it on a laptop through airports would be a method that's pretty far down on the list
Valen: smart multiport serial in FPGA
cradek: it'd be pretty easy to do it anyway, the amount of encryption stuff thats floating around you could have kiddy porn sent via email through every spy network in existance and they wouldnt know it from a video of 2 dogs playing.
same for the plans for a nuclear bomb or the latest goss from parez hilton
the only people laws like this affect are law abiding citizens who aren't actually the problem in the first place
I guess I should have read the reviews better before purchasing this mobo http://www.newegg.com/Product/Product.aspx?Item=N82E16813131647
the rear ports don't line up with the insert :/
SWPadnos: where in the source would I start looking for driver code?
IE how to interface with the outside world
's a good place for em lol
oh well I'd better get to bed
its tomorrow here already
thanks for the chat, with luck something might come of it
ok, see yuo
O K C Y 2 R A P N
SWPadnos: for amp interfaces the smart serial does nothing unless the torque/angle registers are written (other than setup at reset)
to avoid jitter in the outgoing torque commands
good night :)
rofl night random dude
Other smart interfaces (like the resolver interface) free run asynchronously with the host
and in fact had to have special hardware to avoid a beat between the internal sample rate and the host
Ok, nobody got it, eeeesh. Y'all are asleep or what?
got what ?
O K C U 2 R A P N
say it out loud
are /you/ asleep? 8)
(I was trying to come up with something like that for those that like to use aolspeak all the frickin time)
I hv 160 chrs
wow I just got a BIG problem on our server
It's on fire?
* Jymmm grabs the marshmallows
during the upgrade to 10.4LTS
SWPadnos: Got the graham crackers?
window manager crashed
one confirmation window is on the back, another on the top, and I can't switch betwenn windows
no GC n my cfee
Eh, 10.4 is as stable as a cracker whore with DT's
so the question is : do you know some magical idea to have a working system at the end ?
I've had great success with a clean install of 10.04, but not with an upgrade
'cause if I reboot now, I've somethng half-upgraded
sound not really working...
I think I will try reboot on a 10.04 live-CD and try to do the upgrade again
SWPadnos: Clean install, 4 machines. Video resoltion detection is iffy, return from suspend is a crap shoot, C+A+B is disabled, can't switch to C+A+F1 shell
do you guys think this could work ?
bricofoy, no idea. you did back up your data first, right?
Jymmm, hmmm. interesting
yeah, but not the config file for the backup software...
so I REALLY don't want to reinstall the whole thing...
on my laptop, resolution detection is perfect, suspend/hibernate is perfect, CAB is disabled, but it can be re-enabled, switching to a shell doesn't work
but I can switch back to X with C-A-F7
and this laptop is weird because it has 2 video cards
I have had less success with the nvidia drivers, but Intel seems to be working well
ok so ni ideas ?
a few oddnesses with ACPI-related things, like the backlight brightness adjustment buttons (just got that fixed last night), but things like the play/pause/volume etc. all work perfectly
I'll try upgrade from the CD...
bricofoy, not really. you could always plug the hard disk into another PC to get the config files you need
and then you can mess around with it a bit more "aggressively" :)
yeah, I'll do that at the last thing, if upgrade from cd fails
oh, so this is an upgrade through the update manager?
that you're doing now
ok, well that's probably a problem. I think the update only works from the last version or the last LTS version, which are 9.10 and 8.04
well, it's only specifically meant to work from those, but I guess you could get lucky
in fact, try this on another PC before, worked good
but ther, window managed crashed, the upgrade is waiting for me clicking on "OK" somewere, but I can't reach th damn button
well, that's annoying
a little :/
did you notice how far it got before going bonkers?
I tried re-lauching the window manager, bu got segfault
there aren't many places where it asks for user input
something relative to the servise it as to restart after PAM upgrade
usually only at the beginning or the end of the actual installations, unless some package is waiting in the hidden terminal for user input
yeah that's it : a package needs some config
but the window goes on the back...
well at the moment, X just crashed
I think this is the end :/
yep, sounds like a reboot may be in order
can't do anything, no C-A-Fx works
even numlock is frosen...
but file are still online with samba
I guess I'd try rebooting from hard disk and then re-running the update manager
it's possible that it will continue from where it left off
yes, but I fear it won't reboot
if not, you've got the liveCD to fall back to
i'd guess it will boot
I go to try
thanks for the help
incidentally, #ubuntu might be a good place to ask this kind of question :)
Or #CentOS or #Debian
as in how to remove 10.4 and install _______
"throw your hdd out of the window and get a new one"
the hdd with all the worshop files, I 'm not sure it the good idea...
celeron55: That's a waste of perfectly good magnets!
ah, use a sledgehammer then
I don't thick visable (vertical) seams is a good thing on 6 TON jack stands... http://sfbay.craigslist.org/sby/tls/1751795522.html
that's how they are made Jymmm
well I also have a question MEC-relative :)
JT-Work: Maybe so, still don't think it's a good thing.
I use M62 to control a glue valve on my machine
anything stick out that would make you not want to buy any of these mobo's http://www.newegg.com/Product/Productcompare.aspx?Submit=ENE&N=2010200022%201070946178%201072710689%201075707618%201071048377%201125627400&bop=And&ShowDeactivatedMark=False&CompareItemList=22|13-157-196^13-157-196-TS%2C13-128-441^13-128-441-TS%2C13-130-274^13-130-274-TS%2C13-128-438^13-128-438-TS
but if I stop a program when G62 is active
the outputs stay to the "ON" state
and this is a BIG problem, as I use it to control a glue flow
bricofoy, you can use an AND block in HAL to make the actual output depend on both the digital output and the machine being on
JT-Work: check which RAID chip they're using
JT-Work, man, that's a tough one
ok will try that
I wouldn't get the first one, since it's the prior generation of chipset
all right one down :)
the second one is the only one that has esata 6GB
but I like the fact that the last two have dual ethernet ports
but will it induce some delay in the output switching ? beacause turning it ON or OFF during the program needs to be really synced with motion
That last two have 3yr warranties
that's also a plus
two down now
well no, #2 has esata 6G, but 3 and 4 have dual ethernet, so it's still a 3-sided die toss :)
#2 has 8 SATA connectors
[15:08:13] <JT-Work> http://www.newegg.com/Product/Productcompare.aspx?Submit=ENE&N=2010200022%201070946178%201072710689%201075707618%201071048377%201125627400&bop=And&ShowDeactivatedMark=False&CompareItemList=22|13-128-441^13-128-441-TS%2C13-130-274^13-130-274-TS
they all have firewire, but it looks like #2 has no header for a front port, which is annoying
down to ^^ those now
and I would check the manufacturer specs and the photos too, just to be sure
yes, I all ready have an Athlon II cpu so I need to check
well, look at things like USB/FW/audio headers
Look at the RAID chipset
the Jetway mini ITX board I got has 3 USB ports, not 3 USB headers for 6 ports, as the description says
which is NOT listed
I think RAID in in the 890FX chipset
they may not have external RAID chips
Dont know, doesn't say
the gigabyte one AMD SB850 provides 6 native SATA3 storage interfaces with superfast 6Gbps link speed and RAID 0, 1, 5, 10 support
No RAID6 ?
nope, what is that?
no, almost no motherboards do RAID6
maybe a couple of Tyan or SuperMicro server boards might
SWPadnos: No, they do =)
SWPadnos: Tyan and SuperMicro
yes, I thought so :)
the other one has this SATA 6Gb/s: RAID 0/ 1/ 5/ 10
JT-Work, I guess the thing I'd do is maximize the number of accessible "latest-generation" ports available
that means that #2 from the original list may be the best bet, since it has esata 6Gb
since dual LANs aren't usually all that useful
SWPadnos: Maybe not to you
note the word "usually"
Great for firewall/router, media lan
it sounds like this is going to be a workstation, not a CNC controller or media server, or at least that's my assumption :)
yes a general workstation
When it retires as that, can be reused as somethign else.
...though it does have lots of space for additional LAN cards
Dual eSATA is good, dont know what they mean by eSATa/USB combo though
there's a type of connector that you can plug either a USB or esata cable into
so it's only one port, but could be either type
my laptop has one.
no firewire ports on the MSI
at least not on the back pane
JT-Work, what's your current motherboard?
The whole PCIe notation bugs me. Not saying that the MSI doens't have the same issue, but jsut isn't documented
[15:20:01] <JT-Work> http://www.newegg.com/Product/Product.aspx?Item=N82E16813128441
Better to read the manual on that
brand new but the rear plugs don't line up with the insert :/
ha wrong link
I was wondering :)
[15:20:50] <JT-Work> http://www.newegg.com/Product/Product.aspx?Item=N82E16813131647
is this an upgrade or a new PC?
firefox now inserts a new tab instead of adding one to the end
brand new pc
yeah, I noticed that
ok, in that case you got the wrong CPU :)
[15:22:08] <JT-Work> http://www.newegg.com/Product/Product.aspx?Item=N82E16819103702
you would almost assuredly get better performance from the cheaper motherboard and a 6-core (or even 4-core) Phenom CPU than you will from an expensive motherboard and a dual-core
it's a quad core cpu
ok, athlon II x4
the memory I have http://www.newegg.com/Product/Product.aspx?Item=N82E16820145260
still, I'm betting that spending the money on CPU would be a better performance boost than on the motherboard
the MB just needs to have the I/O you want
ok, that's fine memory
Yeah, they are just up the street from me
so what would happen if you just didn't use the ATX I/O cover? :)
I'm really reluctant to buy a new mobo/cpu, just doens't seem worth it for what you get
that other motherboard looks like it's got most of the I/O of the others, and the performance difference between the 870 and 890 will be pretty low (like maybe 5-10%, if that)
Jymmm, it depends on what you're upgrading from
the main reason to get a new mobo at the moment is for fast USB3 and SATA 6GB (plus new CPU support, if you need that)
I'm sending the Asus back because of the alignment problem
SWPadnos: From 3GHz.... I ned raw speed, not band-aid'ed "let's just add more cores"
does the shield line up when it's not in the case? (ie, just popping the cover over the back panel)
yes it does but it looks like the motherboard is 1/8" too low
cores aren't a band-aid, but they're only useful if you have software that will use them
when mounted in the case
you did use the standoffs, didn't you? :)
I have two cases the same and the other one is fine
SWPadnos: They are good for multi-tasking, but when you have a single app that needs raw speed, they haven't done anythign abot that for at least 5 years
it came with 1/4" long stand offs and it looks like it would take 3/8" to line up
you need different apps
weird. that sounds like a case hardware issue though
SWPadnos: It's some cryto stuff.
I have two cases the same and the other one is fine
it's not like the motherboard could affect the alignment by that much, they're only 0.090 in any case
We really should be near 6GHz by now
Jymmm, why not go work for AMD and make that happen ;)
they're just down the street
the case http://www.newegg.com/Product/Product.aspx?Item=N82E16811119196
i've understood they have problems increasing frequency
* Jymmm checks stock prices...
AMD == $8.42, no thanks.
JT-Work, I don't understand what you mean by "I have two cases the same and the other one is fine"
my other motherboard in an identical case the plugs line up good in the back
so I don't think it is a problem with the case atm
have you tried swapping which case the motherboards are in?
they're supposed to be identical, but ...
unless the second case is messed up...
ding ding ding :)
however there was 2 cons for the motherboard that mention alignment problems with the rear plugs :/
so I assumed I had the same problem...
I think motherboards are only 0.090 thick, so even if the motherboard weren't there, that wouldn't lower the connectors by 1/8 inch
check the reviews here http://www.newegg.com/Product/Product.aspx?Item=N82E16813131647
The shield plate is kinda weird and the metal indents get in the way of some of the USB ports and ethernet port.
ok, that's a nice mid-range case from a well-known brand, I wouldn't expect it to suck
The ports don't quite match up on my silverstone grandia htpc case, not really a con for this mobo, could be the cases fault.
but Asus is also a well-respected name in motherboards
yea that's what I thought
might be an issue with the insert not the motherboard
yeah, could be
or misalignment of the insert because of grounding tabs
What's the diff between Atom D510 and Atom 330?
it fits in nice and tight into the cutout
it does look like there's one hole that is a bit low
Jymmm, the D510 is newer, and I think it uses less power and is a little faster
and the 330 is discontinued, whereas the 510 is not
all of the plugs are low on mine compared to the holes on the insert
510 is also fanless (nice feature)
[15:36:35] <L84Supper> http://ark.intel.com/ProductCollection.aspx?familyID=29035
click the boxes for comparisons
grommit: I belive in fanage tyvm
Just one more thing to fail
fan/no fan is motherboard dependent
you ppl and "It's too loud" need earplugs!
but since the D510 uses less power, you need a fan less of the time
surprising that the D510 max TDP is higher than the 330
Less power is good, sounds liek it's throttling back the cpu when there's no high usage liek the mobile cpu's
but it also includes the GPU
all modern cpu's save power when idle
all modern CPUs *can* save power when idle :)
the 3GHz P4 on my server goes down to 375MHz
SWPadnos: I'll double check the case and wait to see what Asus says... I put in a support ticket this morning
I know nothing of AMD cpu's is this any good? Athlon II x4 ?
less cache than the Phenom-II's x4
2TB Seagate $120
how much does the cache affect performance?
celeron55: BIG TIME
many of the Phenom-II x3 may be unlocked to x4 and sell for <$99
celeron55: Think cache like taking a shower in a 18" x18" shower stall, the more the better
the x6 is the one you (I) want :)
[15:49:16] <Jymmm> http://www.newegg.com/product/product.aspx?Item=N82E16883103244
On sale at Frys for $530
well, the thing that gets me about those cheap systems is that they always use crap components on them
HD3200 is crap chipset-intergrated video
1TB is OK, but is probably a slow drive, not a good one
skunkworks_: CPU usage looks like a cayola drawing =)
6GB is a stupid amount of memory for a Phenom, since they aren't triple channel like the i7
HD3200 means 78x chipset
SWPadnos: lots of memory good when workign on graphics and video
oh right, and as one of the reviews mentions, the PSU is 300W, which is on the edge if you want to add some hard disks or a video card
6GB means 2x2GB and 2x1GB - seems like a silly way to go ...
DDR3 + 3200 is wierd, let me see that spec
could be socket AM3 with 780
yes, it's DDR3 1066, with ATI HD3200 integrated video
Well for $530, not expecting too much, but at least it can upgrade PS and ram to 16GB
the thing is, you can put together the system you want in the first place for about the same price
Phenom-II x6 ~$300
usually, when you price out the separate components (other than Windows), the total cost is lower to put it together yourself
SWPadnos: Please spec one out
L84Supper, that's the 1090T, the 1055T is $200
Jymmm, that's left as an exercise for the reader :)
and I'm not the reader
the cost difference is even worse if you have to toss out anything from the integrated build
SWPadnos: No, No, go for it, including mouse and kybd for less than $500
well even the latest 880G or 890G only have integrated HD42xx
Im having the damndest time with coordinates and cutting a part. I have a ngc file (that works fine typically), but every time I try to cut this part the cutter is several inches higher then it should be (so not cutting). I have cut this in the past so I don't know what I am doing wrong this time...
the last one I bought, which was about a year ago, was about $750 including a monitor
so for any intensive graphics application you'll want a GPU PCIe card anyway
SWPadnos: mobo, cpu, ram, case, PS, optical drive, card reader, kybd, mouse for $500ish
quad core, faster than the 810, 4GB (not 6), 1TB, speakers, mouse, keyboard
I wanted to use this gcode file to make another part and editted it do this. Bu t when I go to home things, the same thing happens with my new part. So I must be doing something bascially wrong.
When I start EMC, I home the three axis (and I send them back to zero before I turn the machine off). Then, at the top of the file it says: ( g49 / t0 m6 before touching off!), (touch off with 0.2 edge finder at 1.4 *right*, 1.1 *back*, -0.375 bottom), ( material size: 1.30"+/-0.010" x 1.00"+/-0.010" x 0.375"+0.025/-0.010 )
So, I use the edge finder, go to the right, choose X axis and Touch Off, enter 1.4, P1 G55, OK. Then to the back with the edge finder, choose Y axis and Touch Off, enter 1.1, P1 G55, OK. Then bring the edge finder down to where the bottom of the part would be, choose Z and Touch Off, enter -0.375, P1 G55, Ok. Then st art but the first tool is way too high.
I have a tool.tbl file that lists all the tools and their offsets.
And as I said this has worked before. What am I messing up? it's driving me crazy.
shouldnt you touch off your z with a tool instead of edge finder
that level of MB is $80 or less, CPU is $120 or so, case+PSU is $30, DVDRW is $25, kb/mouse is $30, card reader is $30, memory is $100, HD is $100 - total of $1\515, without working hard at all
hmm, I wonder... I thought I had done it with the edge finder when it worked but now that you say that I am not 100% sure...
i edge find the edges,touch z with a tool that im using
that's off the top of my head, some of the prices may have changed (lower) in the last few weeks
SWPadnos: It's hard just getting a PS alone for $30, much less a PS+case combo for $30 imo
for those who spoke about my upgrade crash some time ago : happy ending :) apt-get upgrade after rebbot, then reboot with "failsafe" the chose "dpkg" in the menu to repair what apt- get upgrade had lef wrong, and the server side works ok (samba and backuppc)... but still somethink wrong : keyboard and mouse does not work on X. and as C-A-Fx is disabled, and ssh not in use, I've no ways to interact with the PC...
nope, you can get a cheap mini-tower with PSU for $30, I've done it
not great stuff, but neither is the acer :)
SWPadnos: 300W ?
Rosewill maybe, 350 or 400W. actually it's pretty nice
SWPadnos: Then that's kinda useless if that's what I'm already getting pre built
x4 or x6 cpu is 125W
well, that's my point
you can get the same shit that's in the pre-made one for the same price
as long as the 300w spec is accurate
SWPadnos: So I'm still not seeing where the benefit is other than I have to build it?
but instead of throwing money away when you replace the crap components with good ones, you can just spend the money (plus a little more) on good stuff in the first place
jlmjvm: I am sure I didn't touch off Z with the tool last time. It is a short tool and I would have remembered trying to do that. I suspect if I do that now, I will crash it...
Heh, this is kinda cool see pic #8 http://www.newegg.com/Product/ImageGallery.aspx?CurImage=83-103-244-TS&ISList=83-103-244-Z01%2c83-103-244-Z02%2c83-103-244-Z03%2c83-103-244-Z04%2c83-103-244-Z05%2c83-103-244-Z06%2c83-103-244-Z07%2c83-103-244-Z08%2c83-103-244-Z09%2c83-103-244-Z10&S7ImageFlag=1&Item=N82E16883103244&Depa=0&WaterMark=1&Description=Acer%20Aspire%20AM3300-U1332%20Phenom%20II%20X4%20810%282.6GHz%29%206GB%20DDR3%201TB%20ATI%20Radeon%
heh, CD storage :)
Yeah, dont you hate when you can't find that disc
grommit:dont wanna do that
"ATX case features: place to hide your weed"
Pac Man is 30yo Today!!!
and my 8yr old still loves it
that must be from the release date for pacman
we had pacman and space invaders on a test rig in 79'
a buddy of mine here reversed pacman and the original cpu, they sold 10k's of pacman consoles using an FPGA
Ot you can just install MAME =)
I brought up emulation on a cheap PC when they were working on it....
This is funny... http://money.cnn.com/2010/05/20/news/economy/arizona_los_angeles_boycott/index.htm?hpt=T2
they said that it did play the game "exactly" like the original :)
did not play
I support the boycott personally, reminds of <German accent> "Your Papers Please!"
if anyone flags down a passing spaceship please let me know
L84Supper: Nah, that's just Sedonna, all kinds of freaks there
could be nice to create a pac-man g-code
to play pacman with a router and a remote pendant :)
where pacman actually gets eaten
I wonder if "Pak Men" would be a TM infringement?
depends on your lawyers :)
L84Supper: Very true! lol
how much memory is needed with a stripped down kernel (~1MB) and EMC running just what's necessary?
wondering if it could all fit in a few meg of cache
yeah, no gui or X
you would need at least sshd
or I would
L84Supper: But I'd use debian, not ubuntu if you want to do that
I'm just thinking of a dedicated embedded app for EMC
L84Supper: Ubuntu has this WEIRD bundled dependancy crap
I'll have to see what services EMC uses
yeah, that is why we got EMC on ARCH Linux
L84Supper: Once I uninstalled Totem movie player, and it totally wiped out all of X
ubuntu does all kinds of poorly documented changes to just about everything
It gave a warning, but i did it anyway to see what would happen =)
Nice to know that a single app would kill the whole GUI!!!
we realized that the ubuntu devs just don't understand DRM
As in TO NEVER INSTALL IT?
DRM in the kernel not digital rights management
NetFlix blocks me using my PicoITX board because it has a DVI iface without locked drivers =)
the 10.04 kernel had lots of last minute patches from upstream to fix kernel mode settings and few other DRM problems
[16:35:35] <L84Supper> http://www.linux.com/news/software/linux-kernel/287804-ubuntu-1004-may-backport-more-kernel-drm
DRM = Direct Rendering Manager
[#emc] "This is the #emc channel - talk related to the Enhanced Machine Controller and general machining. Website: http://www.linuxcnc.org/,
wiki at http://wiki.linuxcnc.org/"
vs digital rights management
Especially if you come from a heavy M$ world and hear it all the time.
[#emc] "This is the #emc channel - talk related to the Enhanced Machine Controller and general machining. Website: http://www.linuxcnc.org/,
wiki at http://wiki.linuxcnc.org/"
Apparently the DRM includes monitor identification routines etc... ?? The confusion with rights management is a pain..
Been doing any more cool work with your router?
And finding any buyers for your work?
nah and yes
first customer is my supplier of oak
a new sigh for his place
he fell in love with the deer scene and wanted that in too :P
Is it made out of oak ?? ;-)
got some pics of your work ?
he will glue some oak for it
so I dont even have to supply material :P
well... I did the thing in aspire
That one deer scene you have is really nice ... I think that would sell well in the states
trial atm :P
so It wont do much good :P
what is wrong with aspire?
it's only us$2500
Dave: the price
compared to artcam though....
gotcha ... frallzor just needs to sell more pieces ..
what does artcam go for?
I would be interested in an 'enhanced' demo version of aspire or vcarve pro if someone had such a thing.
have to sell a lot of work to pay for that software ..
Ive tried to find "enhanced" versions
me too, no luck :
but "enhanced" cant be found :P
frallzor: of artcam?
I thought vcarve pro was suppose to be cheap?
$600 I think
so, cheaper than other stuff, but hardly cheap for me.
of aspire and such
enhanced of those =)
I have "enhanced" trial of artcam
but boy does it suck compare to vectrics stuff
$600 isn't bad if you can recoup your costs by making 10 pcs or so..
I'm not selling anything though, just hobby stuff
I'll probably pay for cut-2d, but I just bought another boat and seem to be more broke than usual.
If you can make stuff like frallzor has, that stuff could sell easily at a craft show ...
Im considering getting Aspire pretty soon though, pretty sure It will make me money
and If I sell the machine in the future, it will not set me back =)
So Aspire is that much better than v carve pro ?
Or different animal?
[16:55:16] <frallzor> http://www.vectric.com/WebSite/Vectric/software_comparision.htm
frallzor : do you have some pics of your work ? I'd like to see
[16:56:05] <frallzor> http://www.pici.se/p/aBwpEvRwD/
just a model I bought from vectric =)
but It shows the machines potential
planning on visiting the states for a roadtrip this summer, bring along milled stuff and sell roadside? :P
wow ! great
how long to machine this ?
and what kind of bit[s]?
6mm end mill and 3mm ball end mill
frallzor: How big is that piece?
what is the starting to generate the code , some picture ? or 3D analisys of a hand-made carving ?
if these are simple parallel finish toolpaths then they are doable now or in the very near future with open-source CAM
for that model, cash :P
awallin: what packages?
I've tried everything I could find, but nothing that was workable.
atmega: ocl for calculating toolpaths, heeks, blender, or your choice of GUI
atmega: do you have a picture of the kind of toolpaths you want?
no, I just wanted some signs originally
nice size frallzor ... If I were you I'd visit some antique type stores in the US when you visit ... I think you could sell those rather quickly ....
wonder if one can bring along wood without permits
You should put a machine in the back of a Sprinter van and travel around cutting them :-)
Mobile machine shop
I've made door panels, inserts, fillers for my boat the last week or so. I ended up doing gcode by hand because it was simpler than getting OS stuff to be workable
if I'm melting plastic while cutting, should I slow down the spindle or speed up the feed or what?
frallzor: Go online and check US Customs. I don't think there are many limitations on bringing finished wood pieces into the US as long as they are not from endangered species etc... Bring in raw wood is probably a different story ....
soo.... how about that guy Obama, right?
I think he's more of a mach3 kind of guy.
Jymmm, check your gmail account
yeah Obama doesnt do emc2
"Yes mach can"
SWPadnos: WHAT DO YOU MEAN NO CUP HOLDER?!?!?!?! WTH good is a computer w/o a cup holder?!
it has a cup holder. it's even got power eject :)
the little button next to DVD
Ah, ok. that's better!
Where do I put the 5.25" floppys in?
same place you would on the Acer
in the cd slot
that's a cup holder
SWPadnos: and the 8" floppys?
those go next to the 120 mm fan
in the shredder, 5.25 are the new thing
grommit: PHUK U, I like my 32 lbs boat anchor 8" FDD!
oh hey, you can make wish lists public (but unfortunately can't have combo discounts in them)
not nearly as cool as the 1000lb card punch.
SWPadnos: 2x2Gb + 2x1GB ???
yep, same as the stupid acer
(I told you it was dumb :) )
Yeah, I think so too.
almost better to save the money spent on the 2x1GB
and just have 4GB total
so you get a faster CPU, faster DVD-RW, better motherboard, more expansion slots, and a 480W PSU, for the same prive
2x1GB is nearly free
with the downside that you have to put it together
$50 or so
actually $58, it's DDR3
I'd rather spend the $60 on a bigger hdd, than on 2x1GB
sure, that was kind of my point too :)
you stick the money where you think it'll do you the most good
and you don't waste any by throwing stuff away
but I dont think it can do that.
SWPadnos: I got a response back from Asus telling me that nothing is wrong and perhaps my standoffs are too short :/
yeah, that's $250 or so
it's going back to NewEgg!
[17:33:08] <jlmjvm> http://pastebin.com/1uGapNyq
SWPadnos:can you see if my pid is linked up,im trying vel mode on axis 0
get joint error when i try to jog
you need to connect the command output from the motion controller to the pid input, the pid output to the stepgen command input, and the encoder feedback to the pid feedback input
all the coefficients are correctly set from the ini file though
oh, and since you're using stepgen 0 in vel mode, you need to connect velocity-scale and velocity-cmd, not the position equivalents
oh hey, there I go: http://secure.newegg.com/WishList/PublicWishDetail.aspx?WishListNumber=9401429
and i thought i was close
sweet little machine
eh, it's OK
I'd spend a bit more and upgrade most everything :)
beats my laptop, so it's sweet
but it's a good approximation of the pre-built Acer Jymmm linked to earlier
anyone want to cuddle?
Heh that link says $512, but the eail said $479
frallzor: My pirahna does
yes, there was a $15 discount on a combo deal, which can't be put into a wish list
I emailed you a cart, which can have combos in it
there was a great case deal available, but since it had a rebate, I didn't include it
end price $9.99, including PSU
Yeah, thanks. I hate rebates.
yeah, I know :)
it is pretty cool though: http://www.newegg.com/Product/Product.aspx?Item=N82E16811166049
(though the reviews suggest it sucks mightily)
[17:51:15] <frallzor> http://pici.se/p/large/fxFjaANXY/
is it just me or this this look odd to you too?
it looks intentional, perfect clean cut there, but never seen one like that
Question: I open a ngc file in AXIS. I see that part. I touch off on the various sides of the actual blank. Now the axis is below and to the right of the part showing in AXIS. Is that how it should work?
by axis, I mean the cone that represents the head
[17:54:28] <SWPadnos> http://wiki.linuxcnc.org/emcinfo.pl?CoordinateSystems
ok, I will read it again.
also, take a look at the coordinate systems sample G-code file (I don't remember the exact name)
it's possible that you have a G92 offset in there, or something else entirely
how can I tell if I have a G92 offset set?
I don't recall exactly
you can look at some vars in the var file (don't remember which ones)
you can issue G92.1 and see if the tool cone or axis origins move in the preview
(I think it's 92.1, could be 92.2 or 92.3 come to think of it)
you need to change the HAL signal connections, represented by lines, from the top arrangement to the bottom arrangement
* archivist scores 1800 on the google pacman
Are we the only two here right now?
All the usual suspects still at work?
its a 24hour channel
Yeah Just seems kinda quiet in here this afternoon. Last time I was here, it was like Grand Central Station... ;-)
Got a wee problem when I try to run a patch of G Code on my two axis (X,Z) machine - get a warning window when it hits the G55 G0 X0 Z0 line.
I get a warning window that pops up saying the line has exceeded the negative minimum of the Z axis.
Then I get two choices, continue, or cancel. If I select continue, the program runs fine, moves to the G55 X0 Z0 line and comences the next moves according to the file.
No soft limits set up on either the Z axis or the X axis
ini file, emc.var and the little ngc file are up at pastebin at http://pastebin.com/NktuFG3W
I don't know what happens if you have no soft limits set up
that seems like a bad way to go IMO
I've tried it both ways, with the soft limits set up I get even more errors.
It gets even more errors of exceeding the limits of both axes I mean.
The X travel is physically almost 58" and the Z travel is a little over 2.5"
The G code is set to run a cut length of about 12" on the X and about 2.2 inches on the Z.
Forgot to say hi Steve!
hi Mark! :)
My manners suck today. Been a long busy week at work... ;-)
eh, no worries (yet!) :)
and he STILL hasn't said it yet =)
wendtmk: it is probably a later line in the program that exceeds limits. the gcode interpreter reads way ahead in a lot of cases.
wendtmk: (you implied it, but never said it =)
If I eliminate the G55 line, the program runs just dandy
then I'll bet you have a problem with some offsets :)
That's what Im thinking, but I'm still kinda flailing away trying to figger out how to write G Code and how to use the offsets in the code.
wendtmk: try shut down emc, remove *.var file and run emc and your buggy program
My machine is pretty simple, and the G code should be pretty simple too. It makes a one pass cut of a strip of bamboo.
I see an offfset of -2.251949 in the var file, and it looks like it might be in the G55 system
* frallzor ponders
Mike, I've reset the G55 params back to 0.0 and it runs fine. However, when I touch off on the work piece again, for both the X nd Z axes I run into the same problem.
so it is some offsets problem
Steve, yes, that was what the Z indicator in Axis told me I was at when I touched off the workpiece. Am I doing that right?
Mike - Yes. It's my beginner-ness showing through.
And this if for jymmm - Hi Steve! ;-) I'm a little slow today...
it looks like the limits should be +/- 1e99 if you don't specify them, so it does seem strange that you could get an error like that
LOL, it's alright wendtmk! We all have "bad hair days" as I like to call them.
jymmm - ROFL!
ewww. that's a terrible mental picture
SWPadnos: Hey, $20 cash money if you spike YOUR hair!
When the test_strip.ngc loads, the cutter image in axis looks like it's supposed to, above and too the left of the workpiece (using the G55 coords).
SWPadnos: Hell, I'll even pay $50 for a 8x10 of that!
SWPadnos:my latest attempt
jymmm - my hair ain't long enough to spike. Probably ain't got enough of it to spike either...
[18:42:21] <jlmjvm> http://pastebin.com/Bki3GVj9
Jymmm, you've never seen my basic training photos :)
wendtmk: No, but SWPadnos's hair is =)
SWPadnos: Nu uh, I wanna see that spiked as it is now!!!
jlmjvm, looks like you connected the command wrond
My basic photos show a head of hair only a tad shorter than it is now!
you have net xpos-cmd pid.0.command => stepgen.0.velocity-cmd
that connects the two inputs together (but not to the motion controller)
you need motion -> pid -> stepgen
SWPadnos: I think i might re-use an existing case, it has a window in it, but I'll remove that, and I want to add in a furnace filter as we have a lot of dust around here. and maybe some big fans too.
feedback goes from encoder back to both pid and motion
Jymmm, seems reasonable
SWPadnos:could you give me an example line
net axis.0.motor-pos-cmd pid.0.command
net pid.0.output stepgen.0.velocity-command
I don't know the name of the stepgen velocity command, take a look with halcmd or halshow
could be vel-cmd or somethign
and then there's feedback
like net axis0feedback encoder.0.position pid.0.feedback axis.0.motor-pos-fb
G42 doesn't work very well up a narrow angle the tool won't fit down
is that intentional or could it be fixed?
do you have a picture of the situation you're talking about?
When you touch off an axis, what value do you put in the parameter window - the one that shows the actual postion of the cutting tool?
the number you want the current coordinate to become
Okay, so lets say I move the z axis so my touch off tool is sitting right on top of the work piece. If I want that to be my Z0 under G55, I'd enter 0?
Okay, let me give that a try.
this is the hole I'm cutting http://imagebin.ca/view/U22GY9eZ.html
I'll go and get a screeny off Axis now
if you're asking what I think you're asking, then the behavior is intentional
I have no idea what you're asking yet :-)
inscribe a circle that's the diameter of the cutter you want to use inside the acute angle, tangent to both legs, and that's how much "rounding" you'll get
in the acute angle, I'd like the tool to stop and turn around where the tool just fits, whatever size it is
yes, which will be further and further from the end the wider the tool
that's fine by me
is that not what you see?
SWPadnos: why not wait for the second picture
oh, right :)
Any of you happen to have a b&s 618 surface grinder
Okay, so I touched off the "workpiece", and set 0.0 in both the X and Z axis G55 window. Now I get a double error window when I run that says: "Program exceeds machine maximum on axis X" and "Program exceeds machine minimum on axis Z".
Still no soft limits set
that is interesting
Axis screeny http://imagebin.ca/view/ewr1a1Ri.html
Let me give it a whirl with the soft limits set and see what I get.
hm, I don't see an image there
it doesn't download for me
wendtmk: you always have soft limits. maybe they are not what you think they are.
I think I posted an all-white image :)
he commented them out of the ini, and it looks like the defaults are +/- 1e99 (at least from iniaxis.cc)
that's gotta be the wrong approach
Okay, now with the soft limits uncommented, I get one warning - "Program exceeds machne minimum on axis Z"
set up homing, set up correct soft limits, home and jog around to confirm they are right, then set your coordinate systems with touch off
Should the Z axis minimum be 0?
How does that work if my home location is set to 0?
on Z, usually 0 is the top of travel, and the bottom of travel is negative
[19:14:35] <MattyCNC> http://imagebin.ca/view/cAl2FQv.html
Okay, but if my home switch is set to machine 0, then the minimum limit should be some other number than 0?
yes, it should be a negative number
if you home to zero at top of travel, MAX_LIMIT should be +0.001 and MIN_LIMIT should be -10.0 or whatever
MattyCNC, you might try making the code into a subroutine, then call it with G42 on and off
So MAX_LIMIT is what the limit switch above the home switch should be?
that lets you see the actual outline it's trying to cut, as well as the compensated path
I have separate home and limit switches
that path does look wrong to me, but I can't see exactly how wrong :)
MattyCNC: or put a / in front of G41 and G42 lines, and use block delete to toggle it on and off. you can backplot without compensation, then turn compensation on and see how the compensated path compares
it's the extra bit at the left. the rest is fine
And MIN_LIMIT would be where theoretically the tool just misses the table, roughly?
wendtmk: no, MIN_LIMIT is the bottom of travel of the axis
is there a tiny arc or line at the corner?
can you post the G-code on pastebin?
give us the gcode?
Okay, let me swap the values and the signs and run it again.
oh you had MAX_LIMIT < MIN_LIMIT?
didn't notice that
no, they aren't swapped in the posted ini
Chris, yes. The signs were correct for MAX and MIN, but the limit distances were swapped.
And now the program runs perfect! Woo Hoo!
[19:22:54] <MattyCNC> http://pastebin.org/262263
Thank you guys very much!!!!!!! Leave it to me to make a bone-headed beginner's error like that.
at least it wasn't a boneheaded masters error, those can be a bear to figure out ;)
Boy, am I glad it was something simple.
MattyCNC: that can't be the whole program because it doesn't end in M2
and it's only 16 lines instead of 208
sorry yeah that's just the first layer. it spirals down 20 layers the same
Steve: Give me time. Once I figger my way around this stuff, I'll be prime for making a bone-headed non-beginner's error... ;-)
..and then draws a few other toolpaths in G40
MattyCNC, can you make your post output only a few layers of just the contour in question?
also, is it meant to be a V cutter?
hmmm. no nevermind that
there would be concentric paths if that were the case
you can see I selected 3mm endmill in the blender script :) that's for show tho, D1 is set in emc's tooltable
I am confused that it does not give you a gouging error. I will study it.
You can get a correct path by removing the line G1 X34.869311 Y10.919794 Z-0.268930
ah, it looks like I gotta write my own compensator for really complex pockets. it might not be just 2 edges in the way
this should cause an error becuase the cutter can't touch that line without gouging the path
ah if I have to hack the gcode, I'll just lop the angle off in blender. I wanted toolsize agnostic gcode
you need to use no compensation in blender for that
[19:31:07] <cradek> http://timeguy.com/cradek-files/emc/bad-line-removed.png
when I remove that line I get this correct path
blender isn't adding compensation
has blender added gcode support ?
hmmm. then why is there a diameter setting?
SWPadnos: , becuase it will do compensation soon :)
ok, well in that case :)
and yoyoek's baking functions use it
also, your entry and exit moves are both wrong and will gouge the corner
If I select G54 coordinate system when touching off a part, the part moves (in AXIS) to the location that I am touching off. However, if I select a different system, say G55 instead, the part stays where it was when AXIS opened the file. This doesn't seem to be correct behavior from what I am reading at: http://wiki.linuxcnc.org/emcinfo.pl?CoordinateSystems
you'll only see things move if the program uses the coordinate system you touch off, I believe
a program that doesn't use G56 doesn't depend on where you touch off G56
The program I am using doesn't appear to have G5x at all in it...
it will need a g5x to "move" it
G54 is the default coordinate system, so unless your program specifically uses a different one, it will only depend on G54
Meaning that if I use G55 it will ignore G55?
though I don't know what happens if you issue a G55 in MDI and then run a program
touching off doesn't select the new coordinate system for use, AFAIK
it just changes its offsets
it should automatically get the entry move right now. I hadn't checked that
note the "AFAIK" there :0
so if I touch off using G55, then issue G55 in MDI and run program it may work...
you may not even need to run the program
or did you mean reload?
meaning if g55 is selected in the program it will appear where g55 origin was touched off at
here's the latest version of the blender script http://filebin.ca/uqwqpt/mGcode175.py
just in case my hard drive dies :) that's probably less usable than the original ATM :)
for blender 2.43-2.49
Ok, you were right SWPadnos. If I touch off G55 system, go to MDI and enter G55, the part moves to the right location in AXIS and the right thing happens in the program!
turning cuts into python functions, like pycam(or heekscad) did seems like a good idea to me right now
no it doesn't :) I'll just build TC into the exporter
when the exporter is really sophisticated, it can use gcode subs
but not tonight, josephine
JT-Work_ is now known as JT-Work
frighteningly, i think I am beginning to understand the coordinate system finally :-O
all the TC you see on the other cuts in http://imagebin.ca/view/U22GY9eZ.html
is drawn by hand
mattymatt: do you use blender and the python script a lot or is this a first?
I use blender a lot, but I'm hacking the script as I go
I was playing with blender and saw the script (but haven't done anything with it yet) and was wondering if it would be a good (free) alternative for cad/cam
Are you sending changes back to the guy who wrote it originally?
if you are comfortable with blender, it can be made usable
NewEgg Standard Return Policy 1. * Return for refund within: 30 days * Return for replacement within: 30 days * Restocking Fee: Yes
I went through a couple tutorials and it seemed straight forward enough...
* JT-Work wonders what the restocking fee is?
I should send to the email in the script, but it looks to me like his domain has been taken over. it could be his GF or momma's domain tho
Yeah, there was some strangeness with his website/domain.
there was someone posting about it in cnczone (I think)...
about the script I mean, not his problems
I've seen him in here
JT-Work, about 15%
[19:59:18] <jlmjvm> http://pastebin.com/jSWwHwqi
hmmm, return or make some 3/8" standoffs on the Hardinge
That link you sent for the script, is that the original, or + your hacks?
or trim the adapter plate so it will fit
+ my hacks
or don't use the shield plate
I'm calling it version 0.175.matt until I get in touch with yoyoek
Or just use washers
jlmjvm, only two things pop out at me: 1, you have the names of the X vel and pos commands attached to the wrong things - this isn't an error, but it would be confusing to look at xvelocity-cmd and wonder why it's actually a position
and 2) you didn't connect feedback back to the motion controller
it's only going to PID, but it needs to go to both
do i need 2 net lines for that?
no, you can list as many pins as you like on one line
MattyMatt: is his latest 174 or 175?
his latest is 174
I usually version my scripts by the version of blender they work with :p
jlmjvm, oh, you're still using position-cmd and position-scale on the stepgen. those should be velocity pins
to 2 decimal places usually :) 2.43 to 2.49 is an unusually stable period
You might want to update the wiki page with your version of the script. I found it very convoluted to find the thing to begin with....
I will, but this is not a release
I've broken tings and added a bunch of empty promises in the gui
and I still don't fully understand yoyoek's stuff
SWPadnos:like this net axis.0.feedback encoder.0.position pid.0.feedback => axis.0.motor-pos-fb
empty promises are par for the course ;-) Perhaps if you post it somewhere he would pick it up and incorporate your changes
jlmjvm, yep, but if you're going to put the arrow in there, it should be to the left of pid.0.feedback :)
it's not an error, but it could be confusing
I am running on a Mac (blender 2.49b) any reason it wouldn't work?
none I can think of
ill give it a try then
3 of his bake functions work as before, 1 I've made generate curves instead, which is how I represent toolpaths. he uses meshes
there are advantages to using curves, apart from the ease of distinguishing them as toolpaths
well, they can be curved, is one :)
cubic beziers can be straight lines and arcs
blender has a poly curve which has no curves at all, but my script can't handle them yet, still hacking on that. I have to convert them all to bezier
and when I export, I don't detect beziers that are arcs yet. if they aren't straight lines, I chop them up into G1
I need an is_bezier_arc() func
I can make a grungy one
MattyMatt: Can you make it pink with fluffy bunnies?
take normals of the contol points, assume where they meet is the center, walk bezier and see if all points are same distance, within tolerance
that is how you do it know, or your function?
that is how I will do it, if I don't stumble on a more elegant way
MattyMatt, take a look at a bi-arc conversion (I think those are for beziers)
it'l be just in XY, although I suppose I should check Z at all points too to make sure a constant Z move during a G2/G3 is acceptable
look at the center point and radius of the two resulting arcs, and if they're the same (or close enough), it's an arc
though I'm sure there are easier ways, by looking at the control points and "force vectors"
(or whatever all that stuff is called :) )
easier on the cpu maybe, but silicon is cheap, my brain is rare and precious :)
um. yeah :)
well if I break it, this won't get done
the control points form tangents to the curve
yeah, so that's why I assume the normals intersect at the center
if I just leave the insert out for the rear connections on that Asus motherboard will there be any problem?
it's not a whole solution, because the control point vectors can vary in length
I'll do it my grungy way. some wiki reader with a CS degree can fix it later :)
There is a parametric equation for the cubic curves in wikipedia.
meh 9.36pm. it's already too late to mill this tonight
it's friday night, but next door has a baby
no friday nights for them :-)
jlmjvm, that looks better. have you been trying to load these at all?
and you don't get errors?
none,except for joint error
huh. that's odd
oh right, there is still a position scale with the stepgen in vel mode
i didnt change the pos to vel like you said
everytime i do it wont start
right, you may need both, if there's actually a vel scale (which there may not be)
having prob with vel scale
you should be using halcmd and/or halshow to see what the pin names are
there may not be one on the software stepgen, I was thinking of hesa
looking for pins,right?
params I think
I don't think scale is a pin
halcmd is a little better for looking around like this (IMO)
halcmd show all stepgen.0.vel
halcmd show all stepgen.0.v
will show anything (pins, params, functions, and signals) that start with that
or halcmd show all stepgen.0
which will show everything having to do with stepgen.0
says program show is not installed
the program is halcmd
the word "show" tells halcmd what you want to do
after halcmd i typed show all stepgen.0
halcmd show all stepgen.0
halcmd by itself does nothing
if you want a shell that you can type commands into, you need halcmd -kf
the k makes it keep running if you make an error
lets do hal show
the f means read from file (which means the terminal if you don't tell it a file to load)
ok, that's fine then. position scaling is used for velocity (duh me)
I thought passwd would make a new account
or adduser maybe
[20:49:27] <frallzor> http://www.pici.se/p/large/fxFjaANXY/
is this right? =)
or its supposed to have that missing part?
the back (or right) cutter looks broken off to me
unless that's not a second blade
I thought that too but its brand new and the surface is spotless
where it COULD be broken
seems allmost intentional but Ive never seen this then =)
well, it's weird that the tip of the cutter doesn't look like it's aligned with the center of rotation
that it is though
SWPadnos:so the hal file looks right?
I think so, at least as far as I can tell without knowing anything about the machine
SWPadnos seen cutters like this before?
I've seen router bits that look similar
I can't tell from that photo whether it really looks broken to me
well it is a router bit for v-grooves
where would be a good start point for the pid values
id say it was broken if it wasnt "broken" så exakt without signs of it :P
jlmjvm, I don't know
Since SWPadnos doesn't know, surely that's faulty memory and requires a cold boot a la kick in the ass!
does anyone have an example of the "gtom" setup using steppers with velocity?
I'll bet gtom does :)
joined the forum to ask for it,somebody else wants to see it also
yeah, sorry, but anything I tell you would be completely unsubstantiates
seb_kuzminsky:do you have a copy of the gtom setup?
just wondering,he was talking to ya in the forum
alex_joni:can you take a look at this,getting a joint error,thanks http://pastebin.com/8rtYW0AZ
If you haven't visited google today do so, playable packman with "google" as the maze
that is too funny
SWPadnos, quick question pls? Do I need to debounce the signal from an optical switch in my home made encoder ra is that handled in the HAL?
(or not ra)
cradek, are you around?
Anyone know this please? Do I need to debounce the signal from an optical switch in my home made encoder or is that handled in the HAL?
I remember reading in the Integrators Manual that EMC can debounce somehow
Thanks pfred1 I figure it might be simply the result of taking only one sample per polling interval..?
well that doesn't give a step by step
OK but it is a start, thanks. I might just add a 555 or something to ensure a clean signal anyway.
there's 35 instances of debounce in the Integrator manual
Wow. OK I should have done my homework first before asking! Thanks.
well that doesn't make finding your answer easier
That's OK at least it's probably there somewhere. :)
maybe page 112 of the Integrator Manual?
halcmd loadrt debounce cfg=1,4,2
I'm reading it and it makes no sense ot me looks like keep plugging away at it until something works
I think when i used to make electronic switch debouncers i used 2 nand gates but it was a long time ago
think i got the circuit out of the ttl cookbook
yes, that works for switches but you need a dpdt switch, set / reset
as i can recall I used momentary contact spring loaded pushbuttons
With a 555 and a couple r's & c's you can disable the signal after the first step, then it resets itself after a set short time interval
yeah but I'm always short of 555s
but nand gates I got by the bucketload
with a push button i've only seen rc filters used...?
I've debounced with a schmitt trigger, with an rc
yeah i htink its really just an rs flipflop you sort of make out of almost any logic gate that does it
the rc is dependent on the switch
but there is a cap and a resistor involved
The only nand gate method I've seen needs to be reset by grounding it after each button push...
a stiff switch may need a bigger cap to stop all the bouncing
that's why it's called an RS (set/reset) I think...
don't matter you can delay in software
I have a 1uF on my reqular flip switches
and a .1uF for some push buttons
I'll have a look at the Integrator's Manual.. Is that where you suggested I look pfred1 ?
yeah page 112 if you can figure out what they're talking about
there's some commands to play blind mans rubix cube with at least
[22:44:46] <tlab> http://www.ganssle.com/debouncing-pt2.htm
tlab that looks terribly familiar
it's a ... umm damn it
it is spdt though
jk flip flop I believe
what kind of switch you need?
he said he has optos
opto switch is still a switch yes?
heck I'd play with hal and get it eventually I'm tired of soldering over here ;0
it's on or off
I boxed up all my drivers and port board and junk yesterday I'm onto designing my mechanics now
if you look at part 1 of that guide it shows the bouncing in analog and digital
[22:48:55] <tlab> http://www.ganssle.com/debouncing.htm
theres zip in getting started
the digital still shows bouncing, so you still have to add more capacitance to filter it out
* pfred1 sees soft limits in his near future ...
pfred1, bounce is no problem with limit switches
davidf you doing home switches?
The home sequence only looks for the first pulse and ignores anything else
but if it was home switches
I made some really nice ones with optos from old printers. :)
yeah i have a bunch of thouse out of cash register printers
F style emitter detector pairs
that's the right thing.
I also have a pile of the microswitch arm switches
but i think I'm just going to be lying to EMC for a while over here when I get it all running
You can make a very simple comparator with an LM324 op amp. need only a 5 v supply. compare the detector voltage to a diode-drop .6v or so. feed that as the home pulse.
only of the manuals that mentions debounce is the integrator one
By default, the input devices may not be accessible to regular users--hal_input requires read-write access, even if the device has no outputs. To change the default permission of a device, add a new file to /etc/udev/rules.d to set the device’s GROUP to "plugdev".
With that system. I got .001 or better repeatability. :)
what do you name the file?
I figure the optocouplers I am using on the input of my buffer board are so slow they should cancel out any bounce
you'd be surprised...
But like I said, for homong switch bounce doesn't seem to make any difference at all. The home sequence looks for the first home response, then it backs up. and approaches slowly until it sees the first voltage step again.
well thre's no way I'm dealing with any more electronics over here right now I just got done building 5 stepper motor drivers and a port buffer board
home. not homong whatever that is!
I have to make chips and sparks for a bit
just dont crash
no by hand i have to make the mechanics of the machine now
I'll be cranking the handwheel by hand so to speak
probably be using saws mostly
isn't a handwheel made for a hand?
yeah I made pretty cheesy motor drivers and have pretty cheesy motors so I figured double up on the X and Y axises with it all
tlab yes it is but this is a CNC centric channel
really for the sort of metal working I do CNC wouldn't be of nay use to me
though I do have a power X axis drive on my mill which is handy now and again
well I guess I'll have a look at the manual re debounce.
I can't believe the hal manual barely makes mention of it
JT-Dev, waht you said above... was that re debounce in the HAL?
I thought he was on about something else
davidf here hit me up on youtube :) http://www.youtube.com/watch?v=cgbeyNNBZ68
you can be lucky number 100
I have a feeling it will work just fine as is. I'm thinking the developers headed off that problem from the git go.
davidf I was trying to get another hit on my youtube page
cool I saw it. 100. :)
yeay i broke triple digits!
you're practically famous now. :)
I made 3 motor drivers like that one
I bought mine on ebay. I have three parker zeta4 drives.
well usually I enjoy electronics except right now I'm a bit tired of it is all
they are really nice. No power supply needed. They run off the 120 v line power.
supply 90 V dc to the steppers.
you have AC steppers?
oh so they have built in power supplies of some sort
no. just regular nema 23 stepper motors.
I have a couple stepper motors that run off 120 VAC
they're both nema 34
yes. built in
you supply a phase capacitor
then they just run one speed
they rectify the ac, and send chopped dc to the stepper motors.
they're pretty powerful though
steppers run best on pure sine waves
maybe someday I'll make a christmas decoration with one of those plug in stepper motors i have?
they do supply sine waves. it is filtered and output as sine
[23:16:45] <davidf> http://www.compumotor.com/manuals/zeta/zeta.html#ZETA4
davidf I think I've seen those drives a few times
$1100 new. I got them for $200.00 each used
well some compumotor drivers they're pretty common
they got like a 50,000 step a rev mode?
my drives aren't the top of the pops but they only cost me $10 a piece to make
well for around twice that yo ucan get them assembled from china
but I'm not so sure of those boards are even as nice as mine are
did you build from plans or your own design?
I basically went with the typical applicaiton schematic in the datasheet
just threw on some optos and filter caps here and there
datasheet for the chips?
I got the schematic drew online let me find the URL
[23:22:24] <pfred1> http://img695.imageshack.us/img695/5841/tb6560ahqa.png
I pretty much threw on the LED parts of the circuit to tie off unused gates in the schmitt inverter
but hey whats an electronics project without at least a couple blinkin lites?
what is that TB6560 - Complete one chip driver?
made by toshiba
yes bipolar chopper
they're $4.73 each from digikey
havent heard of it b4
they replaces the ta8540 or something
davidf, you can debounce in HAL if you need to, but it's always better to make the electronics as clean as possible
current capability / winding?
MrSunshine__ is now known as MrSunshine
their biggest drawback is you can only run them to about 30 volts or so
SWPadnos, thanks! Hi btw. :)
after that they blow up like chinese new years
the reason for that is that noisy electronics introduces extra jitter into the timing of when the edges are detected
good to 3 amps a phase
SWPadnos anyone ever tell you that you can be a real party pooper sometimes?
SWPadnos, OK I'm going to use a 555 timer to debounce and figure out a timing delay that will still allow my max desired rpm.
pfred1, no, never
why do say that? Because he's OT? LOL
no because now someday I'll have to make another board for my cnc project
I guess by the time i get to it I'll be more up for it than I am now
how will a 555 timer do debounce for you?
using it as a monostable?
you cna make is one shot
yeah, monostable multivibrator :)
really I'm forever scrounging for a 555
but that's not great for an encoder, since it needs to be debounced on both edges
you'd be better off using a $0.75 AVR or PIC for debounce
hmm, the plot thinkens...
gotta run, bbl
you have got to be kidding
we can discuss the microcontroller option later if you're still around
ok, party poopr. :)
a microcontroller to switch debounce?
thare has to be an easier LSI solution
seems like overkill to me, but SWPadnos usually knows what's best.
davidf here is a rich LSI solution for ya http://img692.imageshack.us/img692/646/hsoptos.png
4 LSI chips making a half stepping sequence
good to 50 MHz too!
I made 2 motor drivers like that
I didn't think about having to debounce on both rising and falling edges. that kind of complicates things. Maybe I'll go back to the throwing money at it mode. I can get a nice encoder for $30
davidf here: http://www.electro-tech-online.com/attachments/general-electronics-chat/33434d1252818202-hex-schmitt-trigger-invertor-use-debounce-switch-how-my-set-up.jpg
you can get 3 out of one chip
[23:34:24] <davidf> http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=102-1307-ND
I thought there was a feedback wire in it but I'm not seeing on on that schemo
most seem to favor this approach: http://www.internetarchaeology.org/www.geocities.com/Templarser/digital/CLK4.gif
anymore I'm a big believer in the schmitt effect myself
I used them to square up my optocouplers they worked a treat
that looks familiar. it relies on the propagation delay of the triggers. I have a textbook from my EE days that shaows several like that. Forgot all about it. You can gang up several logic gates, strung together as many as needed to get past the bounce interval...
yeah sometimes delay is a wonderful thing
schmitt triggers are sort of the Mr. Magoo's of the digital world
they ignore a lot of stuff for you
OK you are full of good ideas. thanks.
I'd try that before i blew $30 on what will probably be that in a cheap plastic box
no no. the $30 was for the rntire, out of the box ready made encoder, debounced, and all set up. just attach it to the shaft and away you go.
or programmable from 1 to 4 k or something like that check that digikey link I posted above.
SWPadnos:what do you think about these motor-drive units http://www.machmotion.com/Drives-and-Motors-TECO-Servo-Drives-&-Mot/c47_53/index.html
like a quadrature encoder?
jlmjvm I think SWPadnos said that they were leaving
pfred1, yes, exactly that.
davidf never heard of it ;)
maybe he will see it later
[23:42:58] <davidf> http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=102-1307-ND
* pfred1 is all about the open loop
guys, there is a solution to this problem?
[23:44:37] <penguin> http://www.cnczone.com/forums/showthread.php?t=93949
davidf I'm htrowing 440 oz/in on an axis and if I lose stepps I'll have ot reevaluate
pfred1, me too. this is just so I can use the G76 canned threading command. It needs to sync the Z axis on the lathe to the spindle, so you need a spindle encoder signal, and an index pulse (one index pulse/rev and position signal.)
davidf ah you are encoding your spindle?
where'd JT go he fixed his Fred Flintstone encoder on his hardinge
it used hall effect devices iirc
i have a delay problem with the m64/65 commands
G01 X1.2 M65 P0
this turn on the digital output 0
but the movement stop a few ms
i'm using correctly the command m65?
M-codes and G-codes can be in the same line?
penguin, sorry you're over my head there. That would be a q for SWPadnos or cradek or jepler or someone more knowledgeable than I.
andy knows his gcode pretty good but he must be out this evening
pfred1, yes I'm encoding the lathe spindle. No prob though. I think that schmidt trigger or similar low key solution will be fine.
SWPadnos, will be back later penguin - check back in awhile and ask again.
thanks david hasselhof
I think my electronics books were written by weirdos tho, like Clive Sinclair
penguin, One thing you might have a look at... Your problem might be due to too slow accelleration on that axis. Maybe it is taking much longer to stop and start that it needs to. Try increasing the max accel variable on that axis?
davidf, I already set the acceleration to the maximum
i start again
G01 X1.5000 M65 P0
there is a pause there
a tiny pause
i'm working with a laser, and this pause burn more the material
what is the m65 for