cradek, what generated that?
got a url?
no, but it'll be in cvs one of these days I'm sure
I get it HALlelujah
don't know if he has any editing working yet but it wouldn't surprise me
wonder what K`zan is making here. http://wrlabs.shacknet.nu/~vw/MyMachineShop/ztmpwk/v200608261734-001.jpg
who has been working on that?
jepler wrote it yesterday and today
jeez you guys are awesome
I've been wanting to see this for a while... an evolutionary tree for all animals, to see how genetically they are related... based on best guess genetic information.
I was hoping by the time I needed to really play with hal there would be some kind of gui. :)\
it will definitely be nice
A-L-P-H-A: That was mostly an experiment to see what it took to make clamps. It was the only metal I had in here at the time. Just a bit short, but I just got some T6061-T6 to make some out of that should work OK. If not I'll make the next batch out of some kind of steel.
K`zan, clamps are affordable enough to just buy when on sale... I wouldn't suggest making them, if making them to save money was the option.
Getting that damn M10-1.5 die to start seems impossible :-(.
oh I see it's the tip of a clamp
cradek: Or a VERY short clamp :-).
wonder where jmkasunich2 is
What is the secret to getting a die to start rather than wear down the bar to min diameter?
got a shelf off ... lets see what I can do next... I need another large box to carry the crap outta here.
is he still in canada?
I meant virtually
jmkasunich is still in Canuckia.
I'm from Canuckia.
I mean he joined 20 minutes ago but hasn't said anything
cause you scare him cradek... quit intemmediating him.
I'm just a bit frustrated with this process, about to change gears and try it that way and throw the damn die set in the trash :).
what are you trying to thread? 10-1.5 seems easy
K`zan, you aren't trying to chuck the tap, and have the machine spin it are you?
cradek: Yeah, into 12L18
he said die
No, have a die holder in the tailstock.
die is fine on a lathe... at slow speeds.
This thing doesn't have the power to drive a die :-(.
Or anything of size anyway.
probably not... therefor... cut the thread roughing it... and then run the die over it.
can you single-point?
About all I do is use the power to back the rod out.
Yes, just got to change the gears.
Starting to seem simpler. I think these die things are for fixing buggered threads rather than making them,
Will do that when I go back in, taking a break ATM, don't want to toss the lathe out the 4th floor window :-).
K`zan, are they split?
it's a chinese 8x10 isn't it?
No, solid. I have this feeling there is a lot about threading that way I will need to learn.
Cummins, but still chink.
'giht... same diff... tiny. :)
K`zan... hey hey now... I'm Canadian Chinese...
No offsense, people describe me as a kraut, no biggie :-).
Shorter to spell :)
people say i'm a kosher dill
I have no clue what a kraut is.
kosher dill = Jew?
* fenn shrugs
German, but that ain't all.
Similar to a sour kraut :) LOL.
is there an online list of racial slurs? I'd be afraid I'd use them wrong
can you use them wrong?
not sure, that's what I would have to research
Only if they want to :).
[00:55:25] <A-L-P-H-A> http://en.wikipedia.org/wiki/Racial_slurs
cradek: do you still have the link to those small servos with encoders? some surplus place.
I think they were at surplus center (surpluscenter.com)
[00:56:32] <cradek> http://surpluscenter.com/item.asp?UID=2006092019540165&item=10-1834&catname=
was it this?http://surpluscenter.com/item.asp?item=10-1834
nice... only $13 bucks
I haven't even looked at them
original HP Deskjets have a nice motor & encoder (1 per printer)
tiny sucker,,, wouldn't do much though.
it would be just to play
depends on the gearing I suppose
* jmkasunich2 is back
my lathe motors are surprisingly small, and very fast
is there a DIY servo driver?
I wonder if it has a quadurture encoder
or just one channel
skunkworks, no... it has a 200 lpr
might just have to buy one and see. :)
A-L-P-H-A: I used an L298 with pwm right out of emc to drive two small servos
and a little confusing
did he draw it, then start hal to fill in the current values?
or did he use an ordinary hal file, the start the program and it drew the image>
jmkasunich2: it all came from hal, I don't think there's any editing yet
probably just the output of halcmd show
how's it decide where on the page to put things>
damn laptop keybiard
I bet image 5 is has things placed manually, it looks too nice
[01:01:11] <cradek> http://emergent.unpy.net/files/sandbox/hallelujah2.png
this is probably how it starts (this was from earlier today I think)
Wonder if anyone else here has the HF Micro-Mill?
my thoughts - I would hide the current value display
the screen is busy enough as is
click on a signal or pin to see the value
those are the params
all of that is SMOP
hide the params, click on a box to bring upi the params and their values
S for jepler maybe
IMHO this is a huge step
adding/removing blocks and lines is a bigger step
it's very exciting
right - the actual editor part
Yow, what is that, unreadable here.
click on it, your browser probably scaled it
(there's nothing wrong with the image)
Ah, it did, impressive whatever it is :-).
its just big - 1281 x 1132 pixels
definatly will help people like me that need the visual.
I was thinking of an offline editor, I guess he's going for the whole nine yards
I think having it live like CL is very valuable
like CL and halshow
true, but at the same time dangerous
no worse than CL
yes and no
think of it as just a fancy interface to halcmd
we don't call halcmd dangerous, even though it has exactly the same powers
I guess I was thinking too hard
suppose you are drawing a line - if the link command isn't actually issued until you complete the drawing, I guess thats ok
Wonder if that USB keypad could be hacked to control the machine rather than having use the regular keyboard, so much to learn...
although you still need to make sure you connect things in the right order
there is a lot to think about
for instance, deleting lines
if you click on a line and hit delete, what to do
delete the entire line (hal signal)?
but we all seem to agree it should go this way right?
delete just a line segment?
maybe you don't delete lines, you unlink pins
then it's not ambiguous
I'm sure any problems can be worked out, and I'm thrilled to see this progress
still could be funky
I am too (both)
look at axis.0.motor-pos-cmd
it is connected to a signal that goes two places
the split is right at the pin
that's an accident of how it's drawn
unlinking that pin has exactly one meaning
if you tell it to disconnect that pin, hwo do you draw the new version
how you redraw the line to connect the remaining three things is a minor issue
with the other two pins still connected to each other
you'd draw two lines to connect the three reamining pins
then the user can fix up the presentation however he likes
don't worry about the presentation, do functionality first
heh, this is all about presentation
I disagree, it's about making a gui editor
which is a (better) way to _present_ a hal config to a human
I do understand your point
anyway.. I guess the key is to distinguish between editing operations that are "presentation" and those that actually change the underlying netlist/config
yes that's two very different things
in a schematic editor, they don't have that issue, because they aren't online editors
you can draw anything you want in any order, split nets, recombine them, and all kinds of other stuff
_then_ you generate a netlist from it
although now that I think about it, powerlogic (the editor I use at work) keeps track of nets on the fly
if I try to connect two existing nets, it points out that I will be joining distinct nets
and I can choose to continue or cancel
there's not a way to join two existing nets with halcmd
I dunno how it picks the name for the combined net. I suspect that if one has a user assigned name and the other is $N00123 it takes the user assigned name
I think eagle lets you pick
I suppose hallelujah could do that too.
for hal, connecting two nets means remembering the pins attached to one, deleting the signal, and attaching the pins to the other
yeah I was just thinking that too
it can be done, but not pretty
and also not at all an issue for an offline editor
an offline editor has many other problems
like where do you get the list of pins and params
you mean for each module?
I really think we want online editing
I was assuming some sort of library, like a pcb or schematic shape library
you could even do your editing with hal stopped, right?
but this method certainly reduces the amount of duplicate information
you don't need to create the library symbols, they're automagic
this makes it match your hardware too
you could certainly edit with hal stopped, and wise people would do that for all but the simplest changes
the live hal gives you a ton of usefull info
think of all the pins that appear and disappear as you change the insert parameters
I can imagine the lines lighting up like with CL
but the drawing also contains other info that isn't in the hal - the "presentation" stuff, like block location on the page, and routing of the signals
you've convinced me that making the symbols on the fly is the way to go
otherwise you have to try to match some library with your C code and that would be total hell
touching on some SMOP stuff before hitting the tougher stuff...
I'd lose the params, right click the block to see them and their values (and to change the values)
put the pin names inside the boxes
I bet jepler will see this later or tomorrow
I dunno - the short names are fine outside, its the long ones that get messy
<troll>maybe it should be themable</troll>
I don't know what that means
I don't know what I think about the presentation of the blocks
Harry potter theme hal?
as you say, its not a fundamental issue
a not block might be nice if it looked like -|>o-
thats what I had in mind for the library shapes
muxes and some others would be nice as non-rectangles too
but who knows what axis.0 should look like
getting the info directly from hal and building the symbols to match is a net win
and maybe someday comp could be extended so you could include an optional drawing shape and hints on how to render it
although I have no idea how that info would get to the editor
yes, or it could just be in hallelujah, or maybe even somewhere else
some file, created by comp or the make process, that hallelujah reads
lets talk about a more fundamental issue
given a netlist, it is possible to determine dependencies
explain what a dependency is
pin A drives signal foo which drives pins B and C
you want the function that writes pin A to run before the function that reads pin B
right now, the human decides that stuff, by the ordering of addf commands
it can be automated, but there will need to be some tricks (like a way to break loops) and it will need hints
(like which thread a function should be in - it can determine order within a thread, but the user's intent determines which thread
well you could still make the user order them
one of those lists where you move things up and down
yeah, but that is error prone
and invisible on the drawing
yeah representing it graphically would be nicer
have to think about that one.
if we embed just a little extra info in each component it can be mostly automated
for each function, simply need to identify the pins it reads and the ones it writes
wish jepler was here
then you start at blocks that write no pins (output drivers for the most part - DACs, and digital outputs)
oops, no, you end there
start at the blocks that read no pins
dig ins, ADCs, hardware encoder counters, etc
add them to the beginning of the specified thread, and their outputs become valid
then find any block with all its inputs valid, and add that
I see what you mean
each time you add a block to the thread, its outputs become valid, and thus there is a chance that some other block will now have valid inputs
loops (like pos-cmd to pos-feedback) will prevent it from completing
but a NOT with its out tied to its in might be something you would want to write
you could have a dummy component that is a "unit delay"
one input, one output
anyway, you just insert the delay into any loops
(that reflects reality - what the component writes to its output on one pass of the thread is what it will read on the next pass)
the unit delay is like a digital in - its output doesn't depend on its input, its assumed to be valid at the very beginning of the thread
that will allow the auto-sequencer (for lack of a better name) to proceed thru and past loops
again, I was thinking of doing this offline, so you could do something like turning all signals red, then change them to green as the become valid. you could see the sequencer setting things up, and see at a glance if any signals never become valid because of loops or whatever
then add unit delays as needed
this is all about ordering in the thread, right?
it gets interesting for signals that pass between components in differnet threads
a parport output driver might be in the fast thread, but some of its pins are driven from fast thread comps (stepgen) and others from slow (maybe CL or iocontrol)
when sequencing the fast thread, the pins driven by slow thread signals would never become valid, so you'd never add the parport write function to the thread
you could toss unit delays in there to solve that
(unit delays don't have any existance outside the editor, they are _not_ real hal comps, and have no overhead
they just indicate a break in the dependency chain
heh, you can tell I've thought about this, the words are just spewing forth
if you have a chain and you break it wherever, you risk having some of your values be a period too old?
or is it more sinister than that?
yes I can tell you have
yeah, you can increase the delay thru the system
imagine a string of 10 invert blocks in series
if you run them in the right order, an edge propogates thru the entire chain in one execution of the thread
run them in the reverse order, and it takes 10 runs to propogate
the auto-sequencer algorithm would figure out the correct order automagically
if you inserted an (unneeded) unit delay in the middle breaking it into two chains, the sequencer _might_ execute the two chains in the incorrect order, adding a delay
but the fact that a delay block appears in the chain would at least be a tipoff to that possibility
other than these loops do you see any other architectural problems?
with the sequencer thing? not that I've managed to identify
I mean the whole works
live editing raises a whole list of things
some are probably SMOP, others may be more complex
any new problems that halcmd doesn't already have?
the graphical representation lets you ask things that halcmd can't do
splitting nets (SMOP: could forbid deleting segments)
we already figured that one out
we could add that capability
(nets will need _some_ name. most schematic tools just assign meaningless ones, then change them if you specify a name
or SMOP it by deleting and readding everything
we could just insist on an assigned name for each net - prompt as soon as you connect a line to a pin (starting a net)
yes that's an option
deleting and readding would be bad if the machine is running
but I bet often I don't care, I just want it to add that line
we could disallow edits without stopping hal
not that simple
suppose you have a servo machine, and you stop it while the DAC is outputting 0.01 volts
well couldn't we add halcmd rename-signal oldname newname?
not a lot, but with the loop open and the DAC stuck at 0.01 volts, the machine is gonna start to move
yes, that is the approach I'd take
what's that loud noise? why does my foot hurt?
stopping hal with machine-on seems like an all-around bad idea
so you don't want the editor popping up a dialog that says "you can't edit while HAL is running. <cancel> <stop hal>"
adding a hal API to rename an object would be pretty simple
need to take the mutex, verify that the new name isn't already used, unlink the object from the sorted list, change the name string, find the new position in the list and relink it, release the mutex
(the lists are sorted)
mine is ext3 and I think I manually partitioned this one
assume we solve the problem of "pins need to be owned by some component"
(thats what currently prohibits pins like "servo-thread.tmax")
you could have a pin called hal.lock
when set true, editing would be prevented
machine-on could be connected to it
thats not foolproof tho
no, you could change it
user space editor could check the pin, its ok, grab the mutex and start tweaking, then the pin gets set
lets think about this online editing thing
stopping hal with machine on is definitely bad
making changes to the system with machine on could be ok, or bad
depending on what you change
don't disconnect the PID feedback!
if you think about real hardware, very few people ever rewire circuits with the power on
sure, sometimes you might want to jumper a couple things, but its rare
you really have to know what you're doing to attempt that
the problem here is that turning off the power ("hal stop") can itself be dangerous
the known safe method is to shut down emc
but that deconstructs the circuit we're trying to change
I keep coming back to that hal.lock pin
you could connect it to the estop signal, or motion.enable, or something like that
but hal.lock is only usefull if the guy who sets up the system connects it right in the first place
the cases where it's safe to edit are going to be different depending on the machine
maybe there's not a general case fix.
thats why the machine designer would have to be the one to decide what (if anything) gets connected to hal.lock
oh, another partial fix!
I keep coming back to 'this is no different than halcmd'
if hal is loaded but not yet started, you can edit safely
you are right that this is fundamentally no different than halcmd
but at the same time it is different
halcmd _can_ be used interactively, but rarely is (at least for stuff of the scale we're talking about - I do setp all the time)
I came in late... are we talking about halui
the editor makes it easier, and therefore it _would_ be used interactively more often
dave_1: hal editor http://emergent.unpy.net/files/sandbox/hallelujah5.png
I would really like to see if the program that generated that flow chart.
jepler wrote it
I don't know if the layout on the page was generated automatically or he tweaked it
hmmmm . can't get it to open the link
the pins, params, and interconnections was automatic I think
I would be very surprised if the exact layout were automatic
the placement and routing was manual
very very cool stuff
Nice work jeff - very exciting.
dave_1: hum, do you get an error or does it just sit there?
just sits there... no response or browser
on a lighter note http://www.electronicsam.com/images/house/happycats.JPG
that isn't the only link
the happycats doesn't work either
maybe your browser is ill
what IRC client are you using>
don't know that one
gaim or FC5
skunkworks, our cats would never do that - always needed several inches of space
berry (the one on the right) is the social groomer.
they are best buds. never seen that either.
FC5 is fedora core 5
on xchat, I have to right click the link and then click "open in browser"
you are doing something like that, and get no browser?
gaim is the same irc client that is on 5.1 and maybe 6.06
I tried the open in browser
I never used gaim or fc5
jepler, read back
cradek, maybe deciding when its safe to edit isn't so hard
if hal is stopped, safe to edit (but do _not_ provide a gui button to stop it, that is risky)
jmkasunich2: wow there are a lot of words there
there are a lot of issues with online editing
lot of benefits too
so theres a lot to discuss
OK, gaim does things that are unexpected.... it does not bring up the browser but I switched to the window with a browser and the link was already open.
oh, its reusing a browser
I've had that mysteriously happen here too
sometimes I get a new browser window, sometimes an existing one gets hijacked
skunkworks ... when you get finished you can come do my tub.
other cat is kewl also ... doesn't look like it missed too many meals
No - he is beefy :)
OK. now for my question.... I just down installed 6.06 ... easy install ... and added gcc-3.3
Still have not finished the bathroom.. Starting back up again as the weather gets colder.
I haven't thought about any of these hard issues.
trying to compile ... my program needs "/include/sys/io.h" or equivalent and cannot find it. Do I need to install other stuff or has it moved?
let's see .. did you cover that it's not anywhere in the structure of the HAL component which pins are read or written by which function of a block?
hi ray ... late for you isn't it. ;)
dave_1: try apt-get install build-essential
This the box you want to compile on?
I mentioned that we'd need to add that info
why dont we take hallelujah discussion to -devel
I'll give that a try.
see ya later
I doubt you want gcc-3.3 for anything
Most of us got ext3 as our file system.
Don't know why you did not.
did you manually set up partitions.
Or did you use old ones?
so which gcc should I shoot for
I started with a clean install .... and took the defaults
actually didn't get offered a choice
cradek... still here?
The drive was clean when you started?
not really but I had it repartition, etc
the 3 or 5 dapper installs I have done where all ext3
just checked the last one today for cradek
I suppose I could wipe it and start over
I'm going to go try some things.
see ya later
see you dave
night - good luck with hallelujah
<filename says all>
A-L-P-H-4: Also, thanks for the idea to single point the stud and then hit it with the die. Worked well!
same as me then :/
DANEX: there are easier ways to change the Nick ;)
MAYBE HE"S YELLING!
I am sorry for the trouble, I was trying to set every thing back up, my computer burned yesterday :(
It was a first, I have never seen a power supply flame up
with actual smoke and flames?
more smoke than anything , a few sparks
DANEX: that's OK ;)
The most disturbing is that the power breaker did not work
do you mean the circuit breaker in the main breaker box?
I had to cut off the house main to kill power
oh, so you couldn't turn off the breaker - that's bad
you should have been able to unplug the computer though
I probably could have but....
Good thing is , it exposed a problem that really needed to be corrected
your breaker box? ;)
yep. that's definitely not something to messa round with
it's gotta be working perfectly
Well, time to go to work
I will be testing a HALUI set up later today
I'll let you know how it works
Good Bye All
enjoy. good luck with the electrical repairs :)
I used ubuntu 5.xx to resize an XP partition and it worked fine.
that's really neat
Last night I asked the new live to resize both XP and linux partitions and install the new and all three work fine.
rayh: whoa.. you're more adventurous than me ;)
*remebers trying to use the ntfs drivers back when they were very experimental*
ro well they still are ;) but more so
the userspace app. seems to be quite good nowadays
* anonimasu nods
I remember reading the caution during a kernel compile and decided I didn't want to go there.
IMO most of the Linux packages are getting pretty mature and predictable these days.
it's the unpredictability of the on-disk structures you have to worry about (for NTFS)
yeah, only downside is the package systems that dosent evolve as fast
* anonimasu is turning more anti debian lately
I'm not a debian pro, but I'm not sure how package management could be much better
though allowing multiple versions of programs in an easier way might be nice
I have loads of trouble running experimental stuff that there isnt packages for yet
Sure. I can see that.
that should be identical to running things without a package manager ...
(ie, ./configure / make / make install )
yeah, but as there are no way to tell apt/dpkg that you already have something installed
ah. true enough
I still like stuff like that all in a single directory.
unless you make a pseudo package..
I've had it remove my hand installed xorg because I installed a package that depended on xorg..
or when removing such a package..
but hey, as long as you obey it it works great *cynic*
it should work if you install xorg first, then do the manual install (from source), though it'll complain when there's an updated package
it would be nice to have a "mark this package as installed" button in synaptic
actually, "mark this package as externally provided"
I agree that "stupid" is true of many of the packages as well as the package manager.
Stupid in that there is only one acceptable way of doing things.
but when you consider the problems that opens up, I can see why it isn't there
it dosent have to be easy to do, but it should be possible
you could always just not use the package manager
install everything from source, like the "old days" :)
haha, I'd like the best of both worlds :)
in fact, gentoo might be the distribution for you ;)
I do think that over time you can see how the package developers' choices constrain the flexibility of that package.
rayh: what package
I'm working with the gnome package manager right now.
It's very different from anything I've seen before.
isn't Syanptic the gnome package manager?
oh sorry gnome partition manager
ah - that's definitely different ;)
I don't understand how to unlock partitions in there so that you can work with them.
rayh: real men use echo "" > /dev/hda
rayh: dont you have to unmount the disk first?
it might be a stupid remark, but you never know ;)
Most of the partitions are not mounted.
like the ntfs
I'm also seeing a version 0.1 under help but sf has 0.3 and doesn't even list 0.1 in the history.
Strange, every time I start with my custom config I get "Unexpected realtime delay" error
anyone else have that ?
I just mounted a dos partition and it turned from available to locked.
So there may be something to the notion that these locked partitions are mounted someplace.
try "mount" at the console
way strange how this install works when you have several partitions and drives.
Ah I figured out how to log in as real root.
sudo su -u root passwd
supply a password
actually, sudo su works as well, fora root shell prompt
run gdmsetup and select allow local administrator login
right - though you also need to change the root password (I think)
I think Ububtu doesn't seta root password by default (but they disable root login from local or remote connections)
damn, I need moresleep and coffee
I thought acemi described one of the processes hanging, but both of his 'strace' outputs show a program that exits just fine
I noticed that too
[174808.621969] RTAPI: WARNING: tried to delete task 02 while running
[174808.621992] RTAPI: WARNING: tried to delete task 01 while running
are these errors that only started to appear recently? I see them in the dmesg on install2
I don't know, sorry
no, must not be -- I got them with /usr/bin/emc too
jepler, how was the hallelujah picture made?
SWPadnos: sudo -s works too
does anybody know where to buy Roland vinyl cutting blades in Europe?
I am trying to recycle an old Ocè A3 plotter of mine to cut stickers
or using my router and emc as well
cradek: is it raining by jepler? :)
skunkworks: it is
(his site is down)
has been pouring all day
poor cradek too
[18:10:33] <cradek> http://radar.weather.gov/radar.php?rid=oax&product=N0R&overlay=11101111&loop=no
heh look at this
might be raining for a while today
did the phone company have a excuse?
I think he doesn't bother to ask them anymore
* alex_joni forgot how good music feels
I mean really listen to it
music is one of the good things in life
80's music does that to me.
as cheesy as it was
you should try to listen to "stratovarius - years go by"
it's a nice ballad
I mean - yeah ,me too
I have a soft spot for 80s stuff too, apparently we're all about the same age
[18:14:04] <alex_joni> http://www.youtube.com/watch?v=jjiqaXMV9uU
I have the same soft spot :)
not the video, but the music ;)
but not that crap hip-hop junk that came out at the same time as the good stuff (like Men at Work, U2, etc.)
* alex_joni came out when the 80s started
hmm that Yaruto song isn't my style I guess
heh - me either it seems
it's not Yaruto ;)
but I guess it's a matter of taste
I was thinking more of stuff that was released in the 80's, some by bands that were popular before then (Asia, Toto, The Cars, The Romantics, Men At Work, and even pseudo-metal bands like Twisted Sister and Ratt)
J. Geils Band, Steve Miller band, Yes (Big Generator and 90125), Supertramp (Cannonball) ,,,
[18:18:41] <skunkworks> http://www.1500videos.com/
same band: http://www.youtube.com/watch?v=RIdRqUWQtAY
you'll probably hate it :)
we just saw u2 in milwaukee - very good
oh yeah - Ah-Ha, one of my favorites
and who can forget Eirythmics
err - Eurythmics
Ah-ha is great
eurythmics has a couple of nice songs
SWPadnos: like Travis ?
I mean.. do you like travis :D
I really like the sound of songs like "It's Alright, Baby's Coming Back" and "This City Never Sleeps"
I'm not sure I've heard of Travis
SWPadnos: let me check youtube ;)
[18:22:16] <alex_joni> http://www.youtube.com/watch?v=PJ6fZmggzoA
SWPadnos: "here comes the rain again" is very good too
Violent Femmes - Blister in the Sun :)
A song once you get in your head you can't stop humming it
hmmm - I should buy a "Men Without Hats" album
(they are also a WI band
not Men Without Hats?
what about Kate Bush
oh right - the Violent Femmes
was Donovan earlier than the 80s ?
skunkworks: I like "Lies" better (same violent femmes album)
like the early '60s
;-) I still like him
My parents music :)
listening to 'colours' just now
I hate Donovan with a passion (due to the fact that his greatest hits album was played on endless repeat at one movie theater where I worked)
mellow yellow - ARGHHH
Can anyone help me setup the hal files for a type 2 quadrature output on the x axis
Can't seem to get it to work
I had it working once but have lost the configuration files.
isn't that a bit off topic? we're talking about 80s music here
just change the HAL line which says : loadrt stepgen step_type=0,0,0
to loadrt stepgen step_type=2,0,0
cradek: I was going to say something - then looked up at the channel name. Who would have guessed.
yes have that loadrt stepgen step_type=2,0,0
tfmacz: you also need to change the links from the stepgen outputs to parport
linkps stepgen.0.step Xstep
linkps stepgen.0.dir Xdir
change those to :
linkps stepgen.0.phase-A Xstep
linkps stepgen.0.phase-B Xdir
HAL:5: bad variable replacement
HAL config file /home/ted/emc2/configs/engraver//servo-test-core_stepper.hal failed.
Shutting down and cleaning up EMC...
bad variable replacement???
you might have some off  in the hal file
can you post the hal files to pastebin.ca ?
or you have some syntax error.. but I've never seen that message before :)
OK....dumb fingers....have it working now...Thanks..I knew it was simple
I won't loose these files this time...
tfmacz: even if you do.. 5 mins to ask in here ;)
now back to the real deal
80s music :)
80s music gives me the creeps
If ever anyone wants to torture me ,,, just play Eurithmics over and over and over
I never had mtv when I was a kid. It is so funny seeing them now. :)
good afternoon, all.....
anyone available to field an emc question?
does homing work?
pass. next question... :)
but I have a feeling that's not your only question
i was attempting to home the machine this morning....it didn.t work.....
homing to index pulse is iffy on some hardware, but that's the only problem it has
What emc you running and is it servo?
it jumps to the HOME_OFFSET value and thats it.....however it does think it is homed....
motenc lite board
Try reversing the polarity of the home sig in HAL
what emc version?
i will do that
Keep your hand on the estop when you try it.
and a HEAD version, hell the head ver won't even find my index pulse pins....
Can you see the switch change polarity?
and just where does one change the polarity?
are you trying to home to a home switch, a limit switch, or a home switch plus index pulse?
home switch plus index pulse.....
ummm, what file do i change the switch polarity.....
actually i can see the dio board find the switch......
in the halfile you can change pin-xx-in to pin-xx-in-not
i will try that tomorrow....
will the macine find home from any spot in its travel
you tell it which way to go to find the home switch
so your home switch should be on one end of the travel
and ideally it stays on until the limit switch on that end is hit
my home switch is in the center of both axis travel.....
if you have this setup, yes it's foolproof
so it only stays on briefly in the center of travel?
I mean the home signal itself
the original machine would find it from anywhere i did not have to tell it what direction to go, between the limits and the home switch and the index pulse, it found it....
how did it know which way to go to hunt for the switch?
it 's a prox switch, so are the limits....
IMO you would have to have the axes on the same side of the switch.
how did it know which way to go to hunt for the switch? it can't "know"
i thought there was some type of "reverse axis if it hit a limit command......
if the switch is ON for the left half of the travel and OFF for the right half, it'll work fine
ah I can see that would work
but not ideal eh?
I suppose another homing method could be implemented, for use only with systems that have separate high limit, low limit, and home switches
move in some direction until a switch is hit. if it's a limit, reverse direction
plus some logic to always approach home fromthe same direction
so go past it if you were "below" the home switch, and approach again from the positive side
yes that is exactly what the original did
then it searches for the next index pulse....
and that is home...
well, it's not that complex programmatically, I'd say that defining all the options for the ini/HAL files would be the real problem
in the meantime, i need to get the machine to do some sort of home function, can i just go ahead and use a limit switch to home it?
or should i add another home switch some where?
yep. you can set the home switch to be the same as one of the limits
then add in a HOME_OFFSET to make it move back to the middle of travel
kewl, i will do that tomorrow....
I think there's also a setting that tells emc that home is a limit switch, but I'm not sure of the specific name
(this is so you can move once you hit HOME, since a limit switch would also stop all motion)
yeah - that one ;)
I don't remember the specifics of HOME_OFFSET either - I'm not sure what sign the number should have
ie, does it tell the position when the switch is hit, or how much to move to get to home?
where do i assign a limit switch to be a home switch?
just connect the home input to the same signal as one of the limits
there are good pictures in the user manual
showing the signs for the various options (HOME_POSITION etc)
i also have a motenc lite question
when i start the machine and the computer is booting, the motenc dio board has random io enabled until the computer boots.....this is a problem
mdynac: page 74,75 in the 2.0.3 docs tell about those velocities etc
or must i delay powering up the dio board?
I don't know
doesn't the emc mamill at the fest use a motenc?
I think it uses one of each kind of IO board, heh
if the motenc has no means of controlling outputs before the driver initializes, there's not much emc can do
you may need to add some external hardware to keep the outputs all off during boot (something like the charge pumps used on some breakout boards)
yes it has a motenc
it's just not cool to have the flushing come on randomly at boot time, i realize i need a shower, but not at that time
there is some kind of charge pump setup on the mazak
it's easy to generate a charge pump signal with either hal or CL
but that is too late in time......
# 3) Parallel Port, driving PMDX-122 card. This provides a charge pump
# type watchdog, and also provides a small number of inputs that
# can be sampled at a higher rate. The jogwheel comes in thru this
# card and is counted in software.
mdynac: the charge pump is by default not active
only once emc has started you can activate some things
if it blocks or reboots, it'll stop again
everything is fine after emc boots, just before, whilst pc is booting....
mdynac: that's where the chargepump comes into the picture
argh.. sorry bout the typo
gotcha, ic now.......
disable the i/o power supply till emc is up and running ( a common side to all i/o ), then let emc enable the p/s output
must i use a pmdx-122 card to do so?
mdynac: any charge pump should work
but tomp's solution is even more general
i need a quick one to implement, money is no object.......
parport pin, relay, hal
what tomp said
tomp's is simple but depends on the parport being one way or the other before emc starts
that's why a charge pump is better
money no object? hire me :-) !
thx tom, missed you at IMTS, heard you saw Larry
hmm you can have tomp as an intelligent switch
when he sees that emc2 is up, he'll enable IO
yep hiya hiya got 300 chmer amp p/s for sale?
ehy, ya need one?
11 if price is good
and interface is acceptable
why would you want one of those pieces of garbage?
to burn big dies, retrofitting old machines
looking for a sinker?
no a 300 a ps
you making dashboards???
180 avg right now
and they wanna go faster
i'll ask larry and get off this thread
das schmeeer has some "booster" modules to add to thier p/s to increase current....
had a machine with 3 of 'em in a while back, looks like frankenstien lab, 3 more cabinets...
so i get a pmdx board and hook up the charge pump thingy to the motenc dio and do some config editing....is that it?
I'm not sure the motenc dio has support for the charge pump
do you have an opto board on the dio?
see you Lerneaen_Hydra
actually the pmdx 122 board looks like it would interface my specific i/o a bit better that the motenc board....
mdynac, We used the charge pump from pmdx on the Mazak.
can ya tell me a bit about how you use the two?
reading the pmdx manual right now.....
CP_OK is a +5 signal held low when it is working.
mdynac: something interesting to read: http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?MazakPowerOnOff
You can see the signals and pins used in configs/demo_mazak.hal
mdynac: there's a search field on www.linuxcnc.org it also searches inside the wiki
about line 170
k thx all BIG help....
mdynac: no problem
hmmm. was the Mazak updated with Ubuntu?
SWPadnos: think I heard so
Yes 5.10 before fest
I plan to 6.06 in a week or so.
I thought so. maybe I'll change that comment about KDE to a comment about Gnome ;)
I need to gather up info regarding the upgrade in the next few days.
hi jeff.. less rain now?
rayh: if you can, do a clean install
I can move stuff away to a stick and clean the disk.
alex_joni: I may not be here for long...
jepler: too bad
will cd /
sudo rm -rf * work?
but you have an option to format the disk during install
if you do an install, it'll blow away the partitions anyway
Right. I just thought I'd suggest the linux virus.
good plan :)
Just in case anyone here was x$#@ enough to try it on their machine.
rayh: copy off your ssh key before you do it, so you can get cvs again
since its config is in cvs I don't think you really need anything else
was the final config committed to CVS?
I think so
ok. I know there were a few testing things in there (like the big VCP panel with all the I/O)
I guess I don't see any big reason to update its OS
A-L-P-H-4 is now known as A-L-P-H-A
A-L-P-H-A: the hallelujah pictures are screenshots of a HAL GUI I have been working on a bit
'a bit' he says
oh I dunno -- 4 or 6 hours work, in all
it's actually complete crap
it's hacked together, no organization at all. and it has no editing support, which is what everyone wants.
no load/save support, so it doesn't remember where you put the bends
I have some ideas for it (and HAL components in general)
jepler, I would love to have a flow chart program, that's OSS, that keeps arrows, and stuff... and auto route stuff with minimal crosses.
another one of those travelling sales man problems.
graphviz... I'll check that out now
A-L-P-H-A: this ain't that
[20:26:39] <SWPadnos> http://www.graphviz.org/
thought it was. :)
[20:28:16] <A-L-P-H-A> http://www.bit-tech.net/news/2006/09/21/Charge_your_AAs_over_USB/
now if these ahd 2500mAh I'd be buying them in a sinch.
SWPadnos, know of a way to get an AVR chip to get caller ID info off the line?
[20:31:27] <A-L-P-H-A> http://www.nytimes.com/2006/09/21/technology/21caller.html?ex=1316491200&en=c0a676ba8f71a637&ei=5088&partner=rssnyt&emc=rss
cool little toy
not specifically, but CID is basically tones between the first two rings
A-L-P-H-A: tried avrfreaks ?
you'd need some hardware to prevent the chip from being blown up (a transformer)
A-L-P-H-A: hook its serial port up to a zyxel modem and plug the modem into the phone line
heh - I was going to suggest a modem ;)
all hail the zyxel U1496
best voice/callerid/fax/modem ever
I bet caller-id decoding chips are cheap
we've had some in service for like 10 years straight.
still going strong.
hey jepler: http://yapgvb.sourceforge.net/
"Yet Another Python GraphViz Binding"
there are a couple of other python/graphviz projects, linked from here: http://www.graphviz.org/Resources.php
[20:40:56] <A-L-P-H-A> http://www.infidigm.net/projects/dtmf2/
heh... this guy must be local to me, as e-sonic is local to me... :D
cradek, happen to know if the U1496 works with UK CID?
I'd seriously love to have a computer, that's always up... check the phone number... run it through asterisks, or something, and just hang up on telemarketers and private numbers.
robin_sz: let me see if I have a manual
I remember zyxels being the bomb.
A-L-P-H-A: you need to pick up before hanging up
UK cid is different to the rest of the planet
A-L-P-H-A: that way it'll cost them
alex_joni, have the machine do it for me... pick up, and hang up for me on telemarketers, and private numbers.
or use Asterisk :)
I would love to have asterisks voice mail system, to go through a shit load of systems.
If you are a telemarketer please press 8.
there is an autimated one called "TORTURE"
for astersisk for that very purpose :)
If your company name begins with "A", press two. "B" press two. "C" press one.... etc..
complex menus, never withthe option you want, and no way back
darn, if so I can't find it
I should setup my own asterisk server... I'm making a PVR already... why not right?
what hardware woudl be good with asterisks?
mmm ... for analogue?
If you name begins with A-M Press: 31256894583254685333322156****0##2###66544
Snom 190 phones are good and cheap
hmm, the "quick reference" is a meter long, fine print, both sides
I run several asterisks ... 40 odd phoens in the geneva office
a web to phone thhing
and the phones in my own business
hmmm .. cant find that modem on ebay ...
yeah we run hylafax too
but I have yet to find a decent way to get faxes out from Windows :(
faxes in via asterisk is good
but we have 30 lines of ISDN coming in to some expensive card
Eicon DIVA server card
"Setting S48.0=1 will cause the modem to report CND information in its ASCII coded hexadecimal raw data format. The DTE software is responsible for explaining the data."
so I think this means if it can't decode it, you can
UK does some weird line voltage reversal first to initiate the sequence
and uses a different tone standard
was that Zyxel UL 496?
or U 1496?
it says CND is available in "several parts of USA and some other countries"
if the UK standard is newer than about 1993 it almost surely won't work
its older than that .. but unique
I don't see anything else relevant in the manual
ok thanks for looking
I'll try and find out of my Digium X100P can handle it
* alex_joni is off to bed
robin_sz, this is 1990 again... :) right? Cause we don't ever see that happening... so such a rare occurance needs to be pointed out. :)
shrug... I just have to say it .. its a habit
I am usually tempted to go "splitters!" .. in memory of that scen from Monty Pythons life of Brian ..
holy crap CDrws are slow... especially at 10x.
but only means to make a boot disk atm... as the other computer doesn't support USB boot, unless I flash it with a bios update, which is what I'm doing.
lol... the irony.
frack... stupid errors.
frack? Battle Star Fan?
[21:45:29] <alex_joni> http://directory.fsf.org/emc.html
* alex_joni is not here
long asleep ..
sleep is good
what's the state of tool change moves in 2.0? I kinda thought they didn't work at all...
Anyone up for a question on the emc2-head version?
danex: I can try
O.K. I have installed the emc2- head and run in place option/ The sims appear to work,
It fails when I try to use my config from the current version at the point of HALUI , i think
I have the reports from the files
debug and print
have you made these changes to your config? http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?UpdatingConfigurationsForDevelopmentVersions
No, I will make and test