because ive never used these pyramid tip tools before
haha i broke one because i left feed overide at 200%, 20ipm with a .005" tip
lasted awhile too, haha
worry about them then cut the acrylic at 4-5 am... and it'll work out
yeah thats the plan
the alumnum goes fast, the taig eats it
and it dont gotta be pretty
just gotta have surface area
okay enough rest 4:20 is 40min over bbl
good luck... and if you cant make it right .... make it shinny
thanks again for tip
[00:02:30] <renesis> http://www.flickr.com/groups/electronics_efnet/pool/
pics of a laser cut parts for acrylic open baffle speaker
someone did that in san jose for me, is all for maker faire
haha, is weird seeing my design in clean cut acrylic 500mi away =\
anyway gotta go bye
I feel I failed to find the correct way
I need some kind of vector
so, I'll search the solution tommorow
haha i think i accomplished nothing with emc for like a week
its pretty neaty neat tho once you kinda get where all the files are and what the settings do
it's not a problem really
problem is hmmmm
specific russian equipment
you just have weird hardware?
heheh, ok yeah, my thing is easy so i dunno
my thing i can just drop my one config file over a default setup and im done
i wonder what GCAM will do if i do outside compensation and set the pocket option
yeah, its microstepped tho
so its pretty hard on the software step gen
tho once i figured out what minimum time i needed for the realtime loop, its worked perfect
like, its totally diff than windows machine control
windows, you breath on the system wrong, the machine dwells
the screensaver pops up, machine dweels, monitor blanks, machine dwells
emc runs on default ubuntu, which is bloaty as hell, and i can like browse web and look at docs while machine cycles are going
machine doesnt stop, ever
performance is like night and day
and axis visual feedback makes turbo seem silly and mach seem ugly
renesis: like this? http://imagebin.org/17222
yeah windows people dont understand
like, i can edit images and listen to music on my cnc machine
i cant even do that all the time on my windows xp general use box =(
linux cad/eda needs to be like 5 years from now already
geda makes me cry =(
if lunix could do 3d cad/cam and a better integrated eda suite, i could go back to blackbox and slackware or whatever and be happy for the rest of my life
i guess gcam/eagle/qcad kinda works, but it feels so comprimised compared to eagle on windows and autocad and sw (i use gcam on the windows too)
ren you discovered VAPT yet??
mind you you too young
wtf is vapt?
visual APT ( automatically programmed tool)
its for gcode output?
gcam is something im helping someone develop
yes of course .. the only thing that matters
he codes, i either go yay, or eh, or wtf
well, i was gonna say no boards matter too
but gerbers are rs274 too =\
any welders here?
please don't giggle too loud: http://jmkasunich.com/cgi-bin/blosxom/index.html
my first welding
haha i got bored at mig and stop going to class
i told that old man id get bored with the metal crayon
the 4 hours of tig i snuck in at the beginning was some of the most awesome shit ive seen in my life
pretty bird shit... whats the other side looke like.... did you get pentatrration
dmes1: which one? the middle one?
ha thats not bad for random metal
it was not great, probaby 50% of the back was closed, the other 50% you could still see the gap
oh, yeah dude thats good
1st weld is the bests
definitely for your first try, that shits hard, haha
I also did that one from both sides - the back didn't look as nice first time around
they made us oxy-gas weld and oxy-jet cut first
it made mig really easy and boring
all others are too cold.... your trying to take your time... welding isnt like that.... you work at the speed of the process
I got the torch because they claim it can welds lots of different stuff
the pyramid thing looks good
heh, thats really your first time torch welding?
most people make more of a mess
heh - after stick welding as a kid - mig is awesome. I have only watch someone tig.. Looks even cooler. I have only brased with a torch.
VERY nicely done my friend
I spent about two hours, and the flame was on most of the time
jmkasunich: do you use some sort of filler? or do you just reflow the existing metal?
small circles.. and wags... :)
and tap tap tapp
this torch has a nice small soft flame
skunkworks: dude its like, add rod to godlight plasma ball
I have no experience with "regular" torches so I can't compare
its alot like torch welding
the way you kinda push the pool around
renesis: you talking about tig?
* skunkworks watches the video
from what I had been reading, I figured I wanted to learn TIG, cause it seems the most versatile (different metals, etc)
if you can gas torch, you can prob TIG ok too with a little practice
but lots of money
same thing only hotter flame and smaller filler
mig is pretty easy compared to gas or tig
like, the heat goes exactly where you want
mig is a metal glue gun...
how big is the flame on a regular torch?
gas torch is like, heat everwhere, reflecting back at you, getting in your gloves
kinda sucks, fire is neat tho
jmkasunich: comes in all sizes
even a 000 tip is .008-.010" i believe.... and its imposible to work with
they even make tiny micro torches
yeah, my ex-wife had a jeweler's torch
I brazed with it a few years ago
i did alot of 000 thru #2 tip stuff,
never welded anything with it
its weird how bigger torches throw heat everywhere
the flame on this thing is about 1/4 to 5/16" long
(the main blue part)
like, youll be welding towards a corner, and the flame tip kinda bends arouns and blows heat at your neck or gloves
tig... is all a matter of grinding the 'TROde to fit the hole or poop your fixin' ive had moulds micro tig'd
thats an oxidizing flame...
not the flame tip really just the whole blast of heat coming off it
have you ever burned with HHO???
nope, just acetylene
man wtf, 'dont go over 15, it blows up at 15psi' WTF HOW ABOUT, DONT GO OVER 13, OR SOMETHING
ive seen demos.. HHO seems the BOMB for fusion welding
hehe welding class is fun
yeah it looks neat
how much do the little units cost?
i seen vid demos of them welding off a little square foot box
that torch I got is expensive compared to the basic oxy-acet torch - $369
i'll build 1 for under 500 bucks im sure
jmkasunich: just for the body, or like valves and tips and stuff?
body, valves, tips, etc, but not regulators, hoses, or tanks
not J's torch set-up
I got the kit on the left on this page: http://www.cobratorches.com/11.html
kids are just damage control waiting to happen... probable broken thumb and wrist in the last 1 minute
not mine i dont feel a thing...
sorry it may seem heartless but my 2 boys do more damage to themselves than is inflicted on them...
i was just tuning a system and it has velocity mode amplifiers and a velocity loop in the cnc ( not EMC).
The command signal from CNC always went to 0 soon after speed was achieved ( becuz the amp was the speed controller, the cnc was just the monitor/corrector ).
I put the amp into current mode, and the cnc command became constant, and did not return to 0 even tho the velocity was achieved.
I know these actions are correct, but I had never directly observed them.
thought it may be of use ;)
the point is: velocity mode amps and velocity mode control loops fight each other
(two people driving the car is not good )
wait - what about fire trucks? how does that fit in? ;)
what begins with FU and ends with CK? ( this got soupy sales fired ) the answer: fire truck
why does every house that CSI inspects have no lights? They always have to use flashlights?
because the cops turn them off to conserve electricity
to protect and to serve, remember
ah, snopes sez soupy never said that
dude, snopes is a conspiracy
[02:28:51] <Sweeper> http://xkcd.com/250/
toastyde1th is now known as toastydeath
Any one else around now that has used Gcnccam?
I can add tools and layers, but it won't draw any profile or feature lines , nor will it make more g-code than a bit of opening and closeing code
tomp: you there?
SkinnYPup: sorry.. all I know is from the wiki
alex_joni: thanks I don't have the problem running from the launch icon only from terminal
took me a minute to figure that one.
Today I'll make another effort to make something with EMC
I overlooked manuals yeasterday
Users and Integrators one
If I understand everything correctly, I should write driver for my Mill's controller
is it correct?
depends on controller
archivist, it's not one from list
what sort of controller
it some black box, which connects to computer by LPT cable from one side
and to axes from other
You can use some kind of protocol
to catch some information about robot's state
and this information is quite simple
i.e. "state of position encoder"
1 or 0
state of endstops
1 or 0 for every one
hmm EMC has the servo control in software
there are a lot of information about stepdir configuration
if all step dir then possibly treat as a stepper system and use stepconf
I can emulate stepdir programmaticaly
that's why I thinking about driver...
the system can do step dir without writing drivers
archivist, system can do it if my hardware is stepdir-based
and if stepdir lines controlled with LPT bits
but my system don't
I should use 2 outport instructions
to send a command to my robot
and 2 outport+1/2 inport instructions to catch an information from it
1/2 mean 1 or 2
do someone can provide me with some kind of efforts vector?
remember emc expects to control the servo, you have external loops, so it wont be easy
can you reformulate your words?
my english is poor
do you mean emc lack some kind of "more direct" control of my servos?
there are others in here who could explain in your language possibly
emc has direct control
is doesnt do indirect
do you mean EMC expects some lines in my LPT
archivist: he has some box which uses a serial connection to control to
alex, LPT, not serial
ok, the port is parallel
but there is some protocol
or do you simply have bits set?
protocol, not bits
but has internal servo as far as I can see
assargadon: sending packages over the parallel port is still a serial link :)
well.. ok, parallel link.. but anyways
really you have BITS, but after some kind of request
assargadon: ok, so protocol
the problem is that emc2 wants to do motor control by itself
what you want means you need to hack up emc2 pretty good
yes, I understand it now
even LPT driver cant change data direction
instead of the motion controller in emc2, you want your own component which sends things to the control box
yes, is it "driver" in HAL therms?
no, you won't be using HAL
I mean, I don't think you can use HAL
what you can use from emc2 is task/GUI and g-code interpreter
is HAL used fo things like
thre rest you'll have to redo
"modelate the way how endstop parraleled" ?
assargadon: basicly there are 2 approaches
1. keep GUI/task/g-code interpreter and write a new component which talks to your hardware
2. open your control box, and try to connect the motor controllers directly to emc2
option 1 means lots of software
option 2 means lots of hardware
alex_joni, how it possible?
I mean there are just no anought wires in LPT
assargadon: obviously not through LPT
and how, then?
throught 2-3 LPT's?
either with a motion control board, or using a couple paralel ports
yes, 2-3 LPTs might work (alas slow)
documantation told me, motion controller should get commands in some specific language
I don't remeber how it called
of course I select 1 ^)
assargadon: depends on you if you want option 1 or 2
ok, option 1 means you need to get really comfortable with the emc2 code
as far as I already wrote this code, but not adopted for EMC of course
that means, first download it
look through the files, etc
after you did that I can further point you in the right directions (hopefully)
assargadon: but basicly you want to intercept things at the CANON level (canonical commands)
the reason I select 1 is this black box can serve more than 1 axis
5 or 6 maximum, as far as I remeber
CANON is a language neutral layer, where g-code is already interpreted and converted to simple commands like: STRAIGHT_TRAVERSE (which is like G0)
alex_joni, don't I got this sources with ISO?
and STRAIGHT_FEED (like G1 with F-word)
assargadon: no, you only get compiled packages
so awfull :)
but you can get the sources with: 1. apt-get source emc2
assargadon: there *might* be room for the sources on the ISO, but there surely isn't room for the compile tools
btw, sudo apt-get build-dep emc2 will get you all packages needed to compile emc2 by yourself
(should be 1-200MB of packages)
ok, I should reboot myself into EMC-powered linux instalation
and run the sources installation procedure
alex_joni, I only doubt if it's all harder than writing my own software :)
from the other point of view, cool experience, isn't it?
assargadon: it won't be a simple task
assargadon: if you already have something going, then maybe taking the rs274ngc sources and putting the interpreter in your software will be easier
integrating a GUI shouldn't be too hard.. but you need libnml for that, and that also requires getting used to it
assargadon: I'm sure a good developer could hack it to work in 1-2 months (if he has no emc2-code knowledge when he starts)
* Vq^ has also been looking a bit at libnml
in particular the communication with the GUI
alex_joni, 1-2 months souds awfull, don't it?
by the way, what is the major difference between my controller's and other supported hardware controllers?
Is this boards provide you with some more direct control?
alex_joni (or anybody else), newby question
where sourced should be situated after "apt-get source emc2" ?
assargadon: in the folder where you ran the apt-get source
ah really, thanks :)
the axis python-modules seems nice
hollywood-like view :)
not just awfull numbers
hello. freenode keeps giving me this 'erroneous nickname' error and booting me
fenn ask a staffer in #freenode
is your nick setup fully
alex provide me with some help in my effort to utilise EMC
i think its because i changed real_name to freenode.net, but the error message wasnt entirely helpful
I try to understand the position of HAL
in all this system
I mean, I got the idea about main 4 parts
hal is between emc and the hardware
fenn, don't I/O controller is behind EMC and hardware?
hal is what lets you compose the modules in the realtime-system
I see the CUI, Motion Controller, IO Controller and Task Executor
assargadon: have you tried to run halcmd and list the components and how they are connected with signals between pins? if not try it
assargadon: i dont think there is any IO controller anymore
so all this staff I read is outdated and describes EMC things, not EMC2 ?
so pity :)
the docs are at linuxcnc.org/docs
I loaded manuals from there
and one of the first topics is describing of this 4 parts
hmm yeah i think EMCIO is dead
* assargadon do not understand anything more :)
I thought I should write my own version of EMCIO
oh, wait, maybe it's not dead
and use :)
and use all ather stuff
as far as other
assargadon, no you need to do more as you have external servo loops
hrm. io is still not RT? wah.
archivist, my external controller is not really control servos
for example, you cannot ask the coordinate from it
what is the use of the 'mdi' script?
Vq^: mdi.py lets you issue a single line as if you were typing in the AXIS mdi box
it only used for transferring data via limited number of LPT pins
fenn: do i need to specify it as my main user-interface?
according to the CVS log: bin/mdi configs/sim/emc.nml g0 x0
assargadon, I think you need to really think how your current controller works
one would expect the documentation to be in the code, however :)
ah, that works quite well
archivist, what do you mean?
I know for shure how it works :)
assargadon, whether the servo loop is in your controller or nor
archivist, no, there are no any loop in this controller
if word "loop" means controller use encoder information somehow
but it is byte serial controlled to servo motors!
archivist, all you can do with this controller is "direct control" of servos
ie forward/backward/reverse commands
are there any docs on the web we can read
and "get the encoder status" commands
archivist, yes, there are several docs on the web. And no, you can't read them, as far as they in russian.
I think I should write some kind of resume
to obtain some help from this community
is it correct?
that would help you understand the internal working as well as help us to help you
archivist, I understand internal working - finally I already wrote the code, working with it :)
Using infinit loops instead of realtime core :)
One only ask before I begin writing the resume
assargadon: You have a userlevel-driverm but want to write a realtimedriver for this device?
Do "shared memory" architecture used for pins values transferring?
gefink, sort of this. I have a userlevel-driver and I want to use/adopt it to do some work
I wonder what I should do for it
maybe it's realtime-driver
My program now maling some sort of test stuff:
1) moving to start endstop
assrmaybe a block diagram/circuit
2) moves to end endstop, measuring lenght of axis in steps
3) after reaching finish endstop, it moves to the middle of the axe
assargadon: have a look to serport.comp. its a simple driver with IO and realtime. easy to modify
gefink, thanks, maybe it will be usefull
I switching to writing a resume about my machine
I'll kill admin on my job
SkinnYPup: re: gCncCam, did you read his manual? did you read the wiki page? as assurance, I can get a full path to the details and away from them. maybe you need to uses the 'entry path' exit path' menu items?
anonimasu: just in now, hello
Heya folks. do many of you all do woodworking on your machines?
Jessica: i think most can do woodworking but most seems to run metal...
I turn metal into vapor with mine...
I just made my first wood cut on my new machine. my old machine (made with floppy steppers!) didn't have the torque to do wood.
BigJohnT: don't get wood to close ;)
so, now that I CAN cut wood, I was wondering if anyone had tips for routing a nice clean cut.
Jessica: insane rmp is the way to go :)
what diameter of tool and spindle rpm do you have?
generally, a metal cutting machine spins way too slowly
It turns wood into vapor and charcoal... hmmm how do I know?
BigJohnT: unless you're making diamonds, it's not interesting
chris, I'm just using a cheap trim router I picked up at the hardware store. 1/4" router bits for the moment.
BigJohnT: we (vq and me) tried to burn an old bed with our plasma :)
oh ok, so you do have a spindle made for cutting wood, that's good
direction of cut, rpm, and the wood type itself
I wonder if a woodworking group might have more help
I sure don't have any experience with wood
my first little test cut I got a lot of char on the edge of a piece of junk plywood. I also got a lot of wood fuzz on the edge of the cut. it was a nice straight cut though.
char means the tool has to move faster
isn't plywood hard to machine, the woodgrain isn't in one line right?
Chris, all the wood working forums I've found so far are empty or still sleeping! That's why I came back home to you all...
I think the fuzz is normal
should come off with just a bit of sandpaper
fuzz, could be wrong direction or poor wood, or rpm too low
you can decrease the amount of "fuss" if you run the cutter against the cut...
tool has to move faster? you mean RPMs or cut speed?
or dull tooling
rmp just floor it :)
20k rpm plus
Jessica: look up "surface feet per minute wood cutting" on google you should get a wealth of info
dunno what this cheap menards trim router puts out. it's a junk tool i just bought for testing the machine.
it needs to slice the wood before it has time to bend away
I'll make my own spindle eventually, so I can cut PCBs
Arch, what do you mean?
tearing is related to the inertia of the cut
hence wood working machines have high spindle speeds
ok, so faster is better?
and a dull cutter at speed soon produces smoke and black wood
I put a rheostat in line with the router so I can cut plastic eventually, but for the wood tests I have it opened wide up
i wouldn't be surprised if I had a dull cutter
try cutting your fingernail on it... thats what i do...
what is the best type of bit for wood work?
oobi: what? scratching the nail?
scratch the top of the nail...
if it cuts then the bit is fine...
With this manicure???? you must be kidding!
then get a volunteer to lend you a nail
Jessica: sorry i got "gender" specific or something :D
or cut a piece of paper
manicure and machine shop!! I dont believe you
archivist: i'm on your side
Arch, you caught me... my nails look like garbage most of the time. oil and dirt... chips.. ugh!
it is a bit of a stretch
you can borrow my nail if you wan't to...
[14:56:29] <OoBIGeye> http://hgsystem.no-ip.org/~kent/14.jpg
now, that is disgusting!
it hurt like hell to...
so, any thoughts on the best cutter for wood?
I think wood cutting tools are almost always carbide
wood dulls everything it touches
how called the device in english
which can be enabled or disabled
and when it enblet, it hold something
the robot "hand" i mean
Jessica carbide is the best but there is good and cheap carbide tools
gripper is what assargadon means
* OoBIGeye 's english sucks
no, gripper is correct word
I'll finish my rewsume in several minutes
I wonder you guys provide me with some kind of roadmap :)
Do someone know how to make table of contents in OpenOffice Writer with links to chapters?
for each chapter, make a chapter heading/title using the "heading 1" style
if you have sub-chapters, use heading 2
on the page where you want the table of contents, go to insert/tables and indexes, and pick "table of contents" or similar
there are options for which things will be included (like outlines and heading styles), and how many levels deep you want to go
gotta run. good luck
thanks, but problem was hyperlinks, not table of contents
but I find it already
well, good luck, gotta run :)
ok, resume is ready, I should publish it somewhere
What FTP client should I use in ubuntu? :)
I find one in MC :)
I usually use "ftp"
lftp is great commandline ftp. I haven't a clue what a good graphical ftp is.
if you want graphical, I think you can just connect with the regular file browser
File > Connect to Server, select Service type "Public FTP" or "FTP (with login)" as appropriate.
lftp and nautilus also have an option to connect over ssh which is much better than ftp if it's available -- in lftp, use fish://hostname, and in nautilus (file browser) select Service type "ssh".
[16:14:22] <assargadon> http://headache.h1.ru/EMC/RobotResume.html
I needed some way to upload my files to hosting
nothing more :)
ok, as far as my resume ready
I can torture a gooroos here
to force them read my resume
and give me some kind of guideness
alex_joni, you are first :)
fenn, you are second
in line to torturing, I mean :)
archivist, you are third :)
[16:17:31] <assargadon> http://headache.h1.ru/EMC/RobotResume.html
* archivist hides
* assargadon lought with diabolic laughting
It has 3 pages only :)
and 1 of them is almost completely consists of photo and table of contents :)
heh...I should to go now
so pity ;)
I'll back soon, I hope
assargadon: looking at that I would rip the box out, buy some motor drives
and a motion board
and let emc2 do PID loops, and you can then control it to you rlikings (with accel/vel control, etc)
assargadon: that's just me personally, but I wouldn't want to waste time on this
you mean replace both motors and control software?
no, keep motors and encoders
ask awallin about his setup, I would probably chose something similar
looks like you can use up/down pwm on the parport. you could hook the encoders to a second parport and read them with software encoder counting
cradek: yeah, and probably do a bit better than that control
do you know the encoder resolution?
cradek: no, I know what you've seen there
oh, he left, duh
his 'encoders' are not encoders, they are pulsgebers, from his description, he cannot get direction ( like 1 sender detector and 1 track )
that simply can't work
once he removes the stuff that is a problem to EMC, and add what he needs, he has little of what he wanted to salvage ( just a power supply and dc motors and 'machine' ) sorry for the bad estimation :(
tomp2: hacking emc2 to make it work with that is not worth the effort imo
his pdf didnt load on my old debian here
archivist: i could read his 'resume' and my friend Vladimir read the Russian :)
heh we had trouble this morning getting sense
make that archivist_loss
but im curious http://headache.h1.ru/EMC/RobotResume.html
* archivist sends to me
heh double loose gpdf on debian and the winbox cant read his pdf
I read it here on xp and adobe
mine are both old systems adobe 5 on the winbox
i'm sorry he cant get access to parts, he is interested and willing to work on it
free pizza for everyone that want's to come by and have lunch with us
hehe damned long way to dive for a pizza
archivist: you're closer.. don't complain
you can all come and watch the steam engine I shall be driving sunday and monday
meh.. no testdrive?
actualy once we start it, we mainly stand back and keep the visitors happy and safe
[17:32:32] <archivist> http://www.middleton-leawood.org.uk/leawood/images/outside/index.html
ah, thought it was a car
er slightly bigger
I can see that..
Those steam people talk that way.. 'driving' an engine..
expecially when you have to run the valves manually ;)
ahhh, time for a nap now
archivist: cool engine
800 english gallons per stroke
how far does it raise the water?
hmm 20-30 ft I must measure
what is the rph?
lol ok sph
non rotating about 4 strokes a minute ish
pretty quick then
must be awesome to run
there is a youtube vid
[18:34:06] <archivist> http://www.youtube.com/watch?v=PdIgEQ-r0nU
amazing, but I can't understand what they are saying...
what language are they speaking?
lol I knew that I was just teasing...
most obvious voice is Ian not speaking to camera, a derbyshire accent
I worked a number years in west africa with some expats from Britian on an oil rig
some with the heavy accents were hard to understand
hmm did I here me at the end
I'll run it again
may have been me at the handles at the end, or John
on the valves?
not me i think I just listened again
could be our apprentice
alex_joni, can you repeat you opinion about EMC using with my robot?
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2008-05-02.txt
I examine several quite pesimistic words
And I see my PDF isn't as good as I expect :)
But I still do not understand, what is the problem.
As far as I understand, there are some "interface cards", EMC can work with
a few yes
what is the main difference between this interface boards and my controller?
yes, several of as
Or maybe key problem is my "encoders", which is not really "encoders" ?
Yesterday (I think), someone was looking for an ftp client. Do you know who that was?
yes, I know
lerman: that was Assargadon
I used Midnight Commander's one
Ah. I just installed fireftp. Seems pretty neat. Looks like filezilla's interface.
to upload short description of my robot
lerman, thanks :)
[19:25:53] <Assargadon> http://headache.h1.ru/EMC/RobotResume.html
It seems highly rated.
I'll test it next time I'll need an FTP :)
I use filezilla on windows, nice and easy
I use Filezilla on Windoze and like it a lot. (I wasn't sure what platform you needed the client for.)
alex_joni, cradek and tomp2 agrred my robot is bad and it will never works with EMC
Assargadon: For each type of interface, a different HAL driver is required. These can be difficult to write, and require a lot of knowledge. And sometimes it's impossible -- there is simply no way to make a toaster into a CNC machine by writing software.
Assargadon, I have not read the pdf yet
I can't see it either
jepler, this drivers are difficult to write due to EMC arhitecture?
I can publish it as ODT or DOC documents :)
Assargadon: you should do html :)
Writing drivers for any environment requires a lot of knowledge about the environment.
Assargadon: I believe emc's architecture is good, but that doesn't mean that all types of hardware are easy to write drivers for.
Hello alex_joni : Did you notice that I fixed "your" bug? (Well, actually, my bug.).
lerman: yeah, cool
talk about fast fixing of bugs :)
Even faster creation of bugs. :-)
also, some designs are inherently bad, like an encoder that does not tell you which way it's turning. you cannot position correctly with these, no matter how much time you spend writing the software.
when you have hardware like this, and you ask for advice, people will tell you to not bother writing the software - instead you should replace the problematic hardware - and these people are correct
If I can't do thorough testing of code, I owe it to the users to at least fix them quickly. Unfortunately, I'm not always very good at that because I don't always keep a testing environment handy.
and fast fixing of the manual...
cradek, only problem I see with this encoders is the lowering of resolution...
maybe I did not understand correctly but I thought they have only one channel, so you cannot tell which way they are turning
Ah. That's harder (for me). The manual development environment is not part of my tool set.
Assargadon: what does this mean ? "So you can count steps, but you can't measure direction. "
cradek, it's correct. And yes, it's not as good as two-channel one.
But if only you know what way you moves, it's enought
Assargadon: frankly speaking they are crap.
suppose you are cutting from X0 Y0 to X1 Y0. During this move, Y should stay stationary. But, in fact, the cutting forces cause a small motion in Y. How will you know which direction it is?
Assargadon, useless for machine control
Assargadon: no, that assumption is incorrect
for the reason jepler describes
many forces other than from the motor act on a cnc machine
cradek, situations as this is making this kind of encoders crap...for "serious" project
it is the job of a servo control loop to keep the table at the commanded position, but you need better encoders for this to be possible
but if I want, for example, some kind of painting device
it still appendable
Assargadon: no, crap for EMC. EMC wants to know the correct thing has happened, not just the correct thing was asked for
my goal just now is get some experience, not making a usable tool
Well, stop and direction servo controller have no feedback to EMC. If you have faith, it works.
Emc cannot tell the correct thing has happened without directional information
Assargadon, incorrect feedback could be positive feedback and the machine will violently hit the endstops or you
tomp2, driver can loopback directional information to EMC, isn't it?
if only I assume there is no significiant forces other than motors
driver can only use what the hardware supplies as feedback, it cannot invent the sign of a vector
and really I cannot imagine the force which will turn the wormed axis...
tomp2, why not? :)
archivist, machine cant hit endstops due to achieving endstops stops the axis
only force which can rotate the axis is my fingers :)
(except the motor of course)
just have an e-stop handy and whirl away
Assargadon: if it invents it, then it is not a reflection of reality, just an invention. perhaps you can 'create' a sign that is true, i dont know how.
I don't know what estopis, despite the fact I saw button with this name in sim-EMCs I run
estop is short for emergency stop
it occurs when axis moves too far, amplifier overheats, user panics...
yes, my controller performs estop in case of overload
or called an oh shit button
and it has button for it
it should stop all possible harm ( motion, spindle, pressure.... )
red one :)
tomp2, driver can remeber last direction commant it receive from EMC
Assargadon: i dont know how you can use your hardware with EMC without alex's suggestion ( 1) lotsa hardware changes (2) lotsa software changes
which is normally preceded around here with "hey, hold my beer and watch this"
Assargadon: yes, i had that thought, but that is what SHOULD be ( ISTWERT ) not what DID happen (SOLWERT)
tomp2, I understand limitations of this approach
Assargadon: BUT it may get you moving, and is #2 ( lotsa software )
yes, if I stop my machine by catting the wire
and spin axis by fingers
it provide me with incorrect result :)
I test thee :)
the idea is use the sign of the command ( like DIR of a step and DIR system ) to be the sign of the 'encoder' pulse. And, make the analog motor do discrete 'steps' to the 'next' pulse and stop....
I understand unpleasure about "real world" information inaccessible
Assargadon: so leaving aside the problem with the position feedback in your system, it is possible to talk about what you have to write to get a system like this to work
You will have to write a hardware driver which uses inb() and outb() to read and write specific bit patterns to the parport
I really think I should republish my "resume" in readable format
to have less theoretical discussion
jepler, I already accomplish this
Assargadon: as a HAL component?
as far as I accomplish this
I was told EMC exists
your HAL component will have at least two "functions": one will run frequently (e.g., every BASE_PERIOD), do this I/O and update internal data
and instead writing all other machine-independent stuff
I should use EMC
the other will run less frequently (e.g., every SERVO_PERIOD) and update the HAL pin outputs.
For instance, one of those outputs would be for a limit switch. You would simply set that HAL value to 1 if the switch is closed according to the latest value read from the hardware, or 0 if it is not
jepler it's exactly I wrote in resume :)
and asking if this true :)
so, I should write some kind of "driver" (in HAL therms)
except for all the problems, yes. you should start reading the documentation and some of the existing hardware drivers.
or I/O controller (in general EMC therms)
you should understand what emc will provide (e.g., position command) and require (e.g., position feedback), then figure out how to provide that
in some cases you may wish to take advantage of other HAL components (e.g., pid + pwmgen to convert a position command into a PWM or PDM signal)
jepler, but as far as I understand, using "advantage of other HAL components" is not the problem of writing driver, but the problem of building HAL system
from the blocks
Assargadon: it guides your thought as you decide what the inputs to your component will be. Should your component take a position input and create a motor drive signal based on that and an algorithm you develop? Or will you save your own effort by using pid+pwmgen instead?
either one is possible, because emc is modular
jepler, of course I want to make as less work as possible :)
Is it correct, that minimal work I can do is something like
HAL module with encoder pins and direction pins?
This is a different, stepwise approach to using your hardware...
hal (not EMC) could move the dc motor until N counts were seen from the 'encoder'
and then some clamp/brake applied to the motor.
The count would be added/subtracted to start position
according to the sign of the Vcmd sent to the motor amplifier,
(or use the fwd/rev signal sent to the hbridge).
There is no existing component which accepts "pulse + direction" as input and outputs a position.
It would not be using EMC much, and HAL a lot.
IF this was accomplished
THEN i would begin writing the 'driver' for EMC to use with this hardware.
(guaranteed to not give your guaranteed positoning :)
so, it will be wise to make HAL component, which will has 3 pins
2 inputs: pulse bit and dir bit
and one output: integer coordinate
and maybe one more input bit: reset (setting current offset to 0)
floating-point coordinate is more typical. that is the type of pin that stepgen and encoder use, and the type of pin that emc requires.
as far as I'l accomplish this, I'll get some experience with writing HAL modules
jepler, floating point un user-defined units?
but the main idea is correct, so it?
I reboot myselfinto Linux
[20:19:21] <assargadon> http://headache.h1.ru/EMC/RobotResume.html
is now in HTML format.
er wot I just see a link to a pdf in an htmlpage
archivist, I think it's your cache
I test the link before posting
to abandon cache effect, try http://headache.h1.ru/EMC/
I can see ok now
I just copied RobotResume.html to index.html
So, the development vector is making HAL module, which counts current position from step strobs and dir bit
and testing it with halmeter :)
best to change the opto encoders so you get real feedback
archivist, maybe - but does it helps SO much?
I just abandon neednes of this additional module
but I still need driver, isn't it?
and I absolutely not sure that adding of second optoswitch is easier than writing this module :)
in the future it can be good addon :)
control PCB has the place for second optoswitch, so MAYBE all I should do is find similar one and sold it
hmm trying to save hardware effort by software wont be easy, its the harder way
archivist, it's depends :)
software cannot allways replace hardware
Writing the code for my machine take me approx 2xtime compared to time af attaching marker to it :)
archivist, of course
I'm happy it's now not just "impossible"
but "you need to write some kind of driver" :)
I was little desperated after reading your first log, quite pessimistical one
the basic hardware is a good starting point, its the optos and motor control that you should look at
you mean it's chapters/topics I should looks manuals for, do you?
by the way, some promotion pictures of EMC2 showed something\
which looks as GUI for HAL combinating
is it really so?
spending a week or two learning how opto encoders with two outputs to give direction, and how pwm can control your motors
archivist, is it so hard? I think learning of how compile and test HAL modules is much, much harder :)
trying to mess with hal with the incorrect hardware will be hard
archivist, I thought I miss the point
Do you mean I should look at "good" hardware descriptions
to understand how it SHOULD works
yes then you will see emc is a good match
before trying to cheat with my crap hardware?
yes your getting it
archivist, I have a good understanding about how good software should works
pertially it's becouse I spent a lot of time to readint RepRap reports :)
how many wires from the opto encoders
but I have no other machine to run...and I have no skills to build it
machine itself is ok, its a good starting point
but encoders are bad, isn't it?
how many wires from the opto encoders
for every encoder
gnd pwr a,B
one second, I accumulate my english
There are black tiny device I call Optoswitch
some have A,B outs
U-like thing with led in one "leg" and sensor in other "leg"
those switches has 4 soldering legs
yes I know the simple types
is it thing you asked me for?
the better ones have a light source and the sensor side has two
or you asked about encoder to be sure they REALLY has one switch?
like mices :)
archivist, I swear, there is ONLY pulse counter
a mice only needs 3 wires
I can say it for sure
1) I can clearly see empty place for second optoswitch
2) Manufacturer specification told me htere are two variations: one with 2 optoswithes for measuring direction, and other with one
3) controller box has a plenty LEDs
one for every status pin
so there are 4 leds for every axis
2 for endstops
and 2 for encoder
and only one of encoder's leds blinking when axis turning
There are no mistake - it's really one-channel pulse-counter-like encoder
if possible get the better opto
archivist, thing you tald about mean total replacing the electronics
adding additional LPT port
and using the "ordinary" EMC software
is it correct?
could well be the best option
it's too hard sentence for my english
can you reformulate?
"Maybe it's best option" ?
Except for my skills
yes, I have soldering iron
and even multimeter
but I completely unskilled in "real-world" electronics making
I mean I experimenting with software IC emulators
but not with real ICs
And I should rebuild all the electronics
even if it's just some kind of relays for motors
and adapters for encoders/endstops
if you NEVER making ANY electronics circuit - it hard task, isn't it?
assargadon: you usually dont ever run motors on relays
anonimasu, and how then?
assargadon: you use some kind of driver circuit like a H bridge
is it some kind of stuff
with blinking leds-sensors pairs?
preventing computer from burning?
* assargadon google for h-bridge now
ah, thanks to wikipedia
anonimasu, do you mean there are existing boards with (several) H-bridges on its
and EMC has drivers for those boards?
usuall you have a motor driver that accepts some kind of signal..
step/dir or pwm with a amplifier, or a card with analog outputs and a servo amplifier
anonimasu, but I'll beed HAL driver for this boards, isn't it?
step and direction servodrives require special servo drivers
err step and direction require special drives..
that work like stepper drives
I'm doubt ANY existent board will use this russian motors as effectors
I see only 6 "internal HAL components" listed in manual
is it really ALL internal components?
there is no others?
I dont know
Is there GUI utilite for visual adding/removing/interconnecting HAL components?
I dont think there are
[21:28:19] <assargadon> http://linuxcnc.org/images/stories/screen_thumb.png
that's the classicladder plc
so the picture at left bottom corner is not the utility I talking about?
so pity, isn't it?
hey, guys, you scare me so much with this hardware difficalties
but source tree even has template for new drivers
and it's look clear
You really fullish me :)
I just find "Counter" component
Instead of using a real encoder that outputs
quadrature, some lathes have a sensor that generates a simple pulse
stream as the spindle turns and an index pulse once per revolution.
I should only change several strings of code...
to make it bidirectional
has any one ever done a do-ability study on a relatively new machine tool???
say late 80's early 90's
well that's annoying
power failure took my computer down after 100+ days of uptime
I have a ups, but the outage must have lasted longer than it did
now I get to restart the compile farm, etc
we had a bomb blow thru wipe me out after 243 days.... took out the drive on the new fidia installation too ..... then again this morning... 20 G a piece... sill on installl...
blew 3 125 amp fuses today... and still the drive smoked up..... any ideas...Les??
what kind of drive?
what is that?
lightning hit or something?
125 amp fuses take some blowing
I assume this is in an industrial environment - three phase, etc
not today im still up.... i droped about when the 1st drive hit the dust.... the 2nd is a mystery for the factory that watched it to finger out
is the spindle drive AC or DC?
The sources of both drivers and components looks very, very friendly
i still think/suspect improper grounding
blown fuses directly upstream of the drive, or on a feeder that runs multiple loads?
ac as it a high speed spindle
well if fuses were nice and cold and the drive developed a short then the surge to blow the fuses could smoke the drive
if the drive developed a short that alone could blow the drive
there ain't fuses that are fast enough to protect IGBTs
all the fuses can do is limit the damage
I should try some kind of compilation tommorow
drive blowin the drive.... SLOW down
Pirelli man here has trouble with some italian stuff
fuses protect cables
ok what helps the drives..
so far I know next to nothing about the situation
I don't know how big the drive is, I don't know what if anything else is on the same feeder, etc
good quality parts in the drives
vmware is installed, but it has not been (correctly) configured
for this system. To (re-)configure it, invoke the following command:
bloody hell, it was working just fine for the last 100 days
jmkasunich: I also get a message that vmware-tools is not installed
on the -RT kernels
even if it is
this is just running vanilla dapper
yeah, but the vmware-tools are installed in the guest
not on the host
oh, I haven't gotten that far
I usually ignore it anyway, on the farm I don't need the services that vmware-tools provides
they allow for easy moving the mouse in and out of the guest
I usually start them, hide them, and forget them until the next reboot
ok, this is quite annoying
it wants to build modules
I bet I know what is going on
some time ago ubuntu probably updated to a newer kernel, but I didn't reboot
and I don't have the matching kernel headers package
the older ones should still be available
so maybe just rebooting and chosing that one is easiest/fastest
I'm running 2.6.15-51-686, I assume I want the headers for that
I just installed the header package, but I probably still have to do the reboot to fix all the symlinks
well, there's probably a way to do it without the reboot, but the boot is simpler
* alex_joni wonders what a boot does to symlinks
I would hope that things like this would get pointed at the running kernel:
jmkasunich@mahan:/usr/src$ ls -l
lrwxrwxrwx 1 root src 27 2007-09-12 18:34 linux -> linux-headers-2.6.15-29-686
oh, and if all this true about "comp"...
all this will be ever easier than I can dream :)
is it true that new components may be added without recompiling the EMC2 ?
sudo comp --install
Then whydo you scared me so much?
assargadon: I still don't think you can get away with writing 1,2..10 comps and HAL only
because we see problems in the future
but that probably depends on what you want your machine to do
if you only want the motors to move uncoordinated.. then you might be ok with comp's only
alex_joni, I heared several times about motor coordination problems
bah - /usr/src still points to 2.6.15-29
and I partially understan them
jmkasunich: that was probably your doing
but I doubt any kind of coordination is possible, as far as you have motors with uncontrolled speed
I don't think installing headers creates any /usr/src/linux -> symlinks
wonder why they are there?
but this motors has no any speed control...
assargadon: I understand.. and that's why I think getting it directly under emc's control would be better
I firstly try what is possible with several comps
but that means needed hardware
assargadon: if you don't have speed control.. then *maybe* it can work (very poorly)
alex, do you speaking about something like
software-emulated servo-based endstops
assargadon: no, I mean if you want speed control (and good control/performance of the machine) then you need special hardware
if you go with what you have, you might be ok only using HAL components
something with varying output voltage, for example?
but there are some kind of software PWM in EMC
do you mean LPT rate is just too low to satisfy PWM requirements?
by the way, I can make step-direction interface, isn't it?
can your hardware understand pwm
as far as I understand the idea, PWM is different-width impulses
i.e. when you circuit is turned on 70% of all time
it is expected you motor will run 30% slower
than if current is uninterrupted
is it true?
something like that
if EMC will turn on and turn off the motors via controller
its why someone said earlier not to use a relay
we should obtain our result, don't we?
PWM only gives you 70% if the turning on and off is fast enough to average out at the proper level
if you turn it on for 7 seconds and off for 3 you won't like the results
thats why I asked
software PWM is very limited in speed
do you mean LPT is too slow
software is too slow
to make needed rate of turn on/turn off commands
heh...my machine is very limited in speed as well :)
it needs to be in the kilo hertz range
with software PWM you can probably do 140uS on and 60uS off to get 70%
but you won't be able to do 72% - the next step might be 80%, at 160uS on and 40uS off
if so, why PWM component included into EMC?
because sometimes slow PWM is good enough
jmkasunich: his "controller" uses some protocol to talk
maybe it's my case
so I bet pwm won't work
if your controller uses some protocol, why are we talking about PWM?
protocols making thing little slower
or MUCH slower
but it don't make things impossible
isn't it? :)
I'm sorry, I have no idea what you are talking about - PWM usually goes direct to the power stage of a motor drive - protocols to go some drive intelligence - you cannot switch from one to the other without digging deep into the drive
jmkasunich, as far as I understand, you don't saw my machine short description?
no - my computer was down, then I was rebooting to fix vmware, etc
[23:34:27] <assargadon> http://headache.h1.ru/EMC/
my controller is not too intelligent
it can only turn on and turn off drives
or turn on reversed
do you know how the motor is turned on and off? relays, mosfet h-bridge, etc?
good night all
some kind of IC h-bridge, as far as I can inderstand
alex_joni, good night
good night alex
you need to get information about that IC
I have a russian-language documentation, and therms are different
what information especially?
how fast can it be switched, etc
ic number is same, start with ic number of h bridge
it might be a russian chip
do you realise it's RUSSIAN ICs?
it's russian chip for sure :)
really foundered in russia, ok, and noo roman numbers, but has russian numbers, ok
maybe there are roman numbers, but how it helps?
assargadon: do you think it would be possible to replace the single optoswitch "encoder" with a real one?
either made from a mouse, or purchased
jmkasunich, I HOPE it can be done
assargadon: I'm trying to figure out if you could control the machine using software PWM and software encoder counting
not with changing optoswitch
but with adding one more ;)
there are place for it on PCB
and even LEDs on controller to monitor it's status
if you can get two channels (and they are quadrature) then you can at least count to see where you are
then you will need to find out if the driver chips can be used for PWM
if they are designed to turn the motor on and off a few times per second, they might overheat and burn out if you try PWM at a few thousand times per second
jmkasunich, don't it will be easier to test experimentally?
LPT connection is just not fast enought to make several thousand switching per second, isn't it?
the LPT can do a few thousand
really? Greate :)
ok, i'll try to find this information
the limit is not in any way the LPT
is it the software PWM and the motor driver ICs
s/is it/it is/
about blowing chips :)
If I do not find it
I can always increasing the rate slightly
measuring the temperature of IC ;)
its not that simple
well, maybe in your case it is
from the other point of view
the machine normally runs 100% on or off, never in between?
I can just made driver to represent my robot in stepdirection form
why not :)
many motor drives (especially larger ones) don't like to be turned all the way on abruptly
they draw very large currents as they try to accelerate very fast
you mean moving around at low speed?
they have some inertia...
both electrical and mechanical inertia
but it's looks like mechanical inertia
but I can test it with my multimeter tommorow :)
electrical inertia = motor winding inductance
I don't think some kind of slow voltage raising procedure are used
inductance determines how long it takes the motor current to ramp up from zero to whatever current the motor draws when you apply power to it
yes, I understand ;)
mechanical inertia determines how long it takes the motor speed to ramp up from zero to whatever speed the applied voltage makes it want to go
yes, i understand :)
as far as I understand, you trying to say that using servos in step-like kind
if you suddenly apply rated voltage , the current will rise much faster than the speed, and will probably go higher than the rated current of the motor - then once the motor speeds up, the current will drop back to a safe level
can be failed idea
due to this two inertias
is it correct?
I am having trouble communicating
mechanical inertia will cause the back emf to be low and the current high
and yes, I understand it, too :)
archivist - right
(about induction caused by current "jumps" in solenoids)
however, I don't know if assargadon knows what back emf is, etc
assargadon knows ;)
I haven't made dinner yet
Assargadon know school phisics very well :)
do you understand motor control?
(maybe the problem is just the language barrier - but it seems like you don't understand, so I'm oversimplifying)
I do not understand question
if you mean some special receipts
used to start up 3-phased motors
or back emf compensation systems
than no, I don't
i have "feeling", but not "understanding" of listed stuff