4 am :)
good coverage zone :)
[00:01:28] <jmkasunich> http://en.wikipedia.org/wiki/Brushed_DC_Electric_Motor
read the section called "Speed Control"
mostly the first two paragraphs
the next two are less important
also good: http://lancet.mit.edu/motors/motors3.html
* jmkasunich seeks food
jmkasunich, you know, it's food supply crisis all over the world
so your task isn't simple ;)
I read both...and?
past his bed time
Hey, anyone know how fast you can spin smallerish servos that (basically) have no load - more like PTZ ?
that depends on the motor
NEMA23 or smaller
still depends on the motor
doens't need any torque, just good speed acceleration control
like what kind of motor?
all motors have a maximum speed depending on how they are built
some would break if over sped, others the magnetics will give a max speed
Hmmm, I'm sorta kinda thinking like gutted printers
google the motor if you can 5k to 30k it will be very variable
what about those pancake steppers?
a lot slower probably
any speed to them? I dont want to toss a lot of cash into this project, would like to use hobby/surplus stuff if I can.
I dont have a tach so nfc there to even do a comparison
what are you trying to do?
Guess I need to make a handheld step generator that's battery powered and I can take into stores =)
pan tilt zoom?
yeah, less the zoom part =)
how big is the camera? how fast do you want to move it?
small, and very fast
What about those RC servos, are they fast?
a bit crappy
ones I played with years ago wernt that quich and had some play
but you dant want that fast for a camera
how accurate - 0.5 minute of angle, or is 1.8 degree OK? What is fast - 360 deg/sec?
or are you trying to do camera shake control
Not sure on the accuracy, more is better. Mostly speed is what I need. I need to be able to move 90 degrees and back within 2 to 10 ms
easier to just move a mirror
I'm considering that as well.
also look at the mirror drives for xy laser
[01:21:49] <archivist> http://hackedgadgets.com/2006/12/02/ten-dollar-xy-laser-scanner/
two thirds down http://www.laserfx.com/Works/Works3S.html
[01:25:13] <archivist> http://elm-chan.org/works/vlp/report_e.html
I totally forgot about galvo's
the last link is the one found some time ago and show a lot more on making one
Even shows it being battery powered.... cool
mass greatly effects servo response though
Has anyone else bumped against this problem? Qcad's trim function doesn't do a good enough job at detailing the arc sart end and center points well enough to use in emc?
well running setup in wine it takes over the display manager . and won't install directx and .net
Whatta day :-). Ran all over hell trying to find 6 conductor cable for the steppers. Zip... Gonna use ethernet cable, these are 1.6A steppers so it should be well withing the capacity of the wire.
Need 6 so perhaps it might be wise to double up on the common wires since I have two spare conductors.
And I have 3 different colors :).
Amazing that the best I could find was telephone wire around here.
4 sets of PMinMo's stepper drives have near 60 smd resistors , that took me nearly 3 hours to solder! too damn small for my eyes ;)
SkinnYPup: what happens to the arc ends and center? ( the start should not be a problem, as you dont describe start in gcode )
is it precision ( number of decimalplaces ) or does the math of start radius != end radius fail?
best o luck
lewin2 is now known as lewing__
I begining my working about my robot and EMC just now :)
good luck assargadon
rayh, thanks :)
Keep us posted with pictures.
where are "realtime" command should be situated?
I need it for "realtime start"
I see some parameters becomes inputs during the times of writing manuals :)
don't pathes can be used for "save" command?
I mean, is it only can save in directory it currently running in?
halcmd save > /path/to/file
aha, I try exactly this
halcmd: save all ~/Desktop/test1.hal
<stdin>:28: Can't open 'save' destination '~/Desktop/test1.hal'
what can be a problem?
try absolute path
ah, ok, thanks
it just do not recognize ~
assargadon: if you run halcmd interactively, it won't
if you run each halcmd command, then it will
alex_joni, I understand, but why different parsers are used?
is it just a bug
or have some sence?
in the case of halcmd: the internal parser is used
in the other case the bash (or whatever shell you run) is used
halcmd will not replace a shell with a full parser
you mean, if I use IO redirection...
I just thought there are some command line parameter for saving
halcmd save > file (isn't that enough?)
<assargadon>you mean, if I use IO redirection...
how can i add additional instance of already loaded component?
you need to specify the needed instances at load time
so if I makes mistake, its my fault :)
assargadon: you can unload the component, and load it again
I understand I can save it, modify .hal file vie text editor and reload it again
alex_joni, of course i can unload/load - unless tons of specifications are amde
heh, EMC really need some kind of UML hal declarations with hal-files autogeneration :)
feel free to improve it wherever you see fit
save netlist won't do the job
by the way
is it possible to connect pins to parameters?
As you mentioned earlier that it looked like some params had become pins. IMO that is the case for a few of the modules.
And you may be reading a version of the docs that use the newer approach.
But then I could be way wrong about all this.
I read older manuals, but I see newer results :)
is impossibility of connection parameters to anywhere is only difference between params and pins?
Pretty much I think.
You can change the value of params using setp but they don't connect like pins.
so this differentiation is just made for understandability?
That certainly is an acceptable way of thinking about the difference between them.
Some parameters will cause havoc if changed during a run.
but this havoc can still be caused with changing parameters during the run via setp
Yes it can. stepgen.0.position-scale is a good example if an axis is not at zero count.
If the raw count is not zero you will most likely see a joint following error.
Where compilated files with HAL components are stored?
I mean, when I using command like "loadrt flipflop"
file "flipflop.something" should be loaded, isn't it?
probably flipflop.ko (.ko = kernel module)
and where this file should be situated?
I don't recall where they are stored, you can search for it though
I want a list of those files
find / -name flipflop.ko
to understand, what components are there
when that command finds flipflop.ko, go to that directory and look at the rest
Hello, this is my first time on an IRC. Too bad my RT typing is lacking. Are there really 33 users watching right now?
yes of course
Probably not. I often forget to show that I'm away when I leave my computer.
But if you write my name, a tone will sound and I might notice that there is something for me.
This is going to be interesting. Although, the last thing I need is another time sink.
33 people connected. I'm sure many are sleeping, away from the keyboard, etc
kirk, are absolutely sure you are not a bot, inspectin channels activity?
No. He's not the bot. I am.
no, me! I'm the bot
Why, do I give some kind of clue?
Is it common to to have conversations with a person's surrogate on IRC?
kirk_wallace: we're just fooling around
the only bot in here is logger_emc
Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2008-05-03.txt
* jmkasunich kicks logger_emc
that's odd, I thought it used to say "ouch" when you did that
It's developed an immunity to your kicks, jmkasunich
I don't get no respect
I just make my first installation with HAL :)
UCK Ubuntu Customization Kit, 'add any package to a live Ubuntu cd' may be of use http://uck.sourceforge.net/
One siggen with frequency controlled by another siggen :)
next step I want to make is implement some kind of pyGUI (or how it called)
and then I'll be rady for myself coding :)
anyone know what the 'PCI Latency Timer' setting in the BIOS does? I'm having trouble with realtime-delays popping up every now and then...
Is there information I should look at to get up to speed with IRC on Mibbit?
what is Mibbit?
google says its an IRC client
It was mentioned recently on the e-mail list.
well look at that
What? Can you see me in my shorts?
my realtime-delay problem might be related to the graphics card. There's onboard video that is disabled, and I have a separate pci-e card. Ubuntu does work with it, but should I install some card-specific drivers? For example glx-gears runs painfully slow which suggests the OpenGL acceleration of the card is not working
I just see a frog
how you do that jmk? is it in FFox? ( google not an application to my knowledge )
Was that mibbit or ribbit?
tomp: mibbit is apparently a web based IRC client
awallin: you are probably using a software based open-GL driver
instead of a proprietary one specific to your card
I don't recall which is better for RT
Good news, I think! May have found a way around EMC / Ubuntu not having wireless capabilty!! If it works it will beat hell out of 100' of ethernet or running wincrap stuff :-) :-) :-)!
Good start for the day :)!
If it works ...
the latency-test shows fairly normal values, but when I do get a realtime-delay it says the motion-controller hasn't been called for something like 400 ms. This doesn't happen regularly - only once in 5-10 min quite randomly. My processor load runs at around 60% when AXIS is running a program
K'zan, I just use WRT54GL's with PPTP.
K`zan: nothing beats ethernet
awallin: 400 milliseconds? ouch
jmkasunich: Heh, normally I'd agree, but to get 20' (air distance) it is going to take moving a LOT of furniture and much jerking around to get ethernet in there - with all the routing of cable, 20' turns into about 100'. Wireless it is :-/...
The bad news is that it's that long. The good new is that a long latency might be easier to find.
jmkasunich: yeah, I ran it in sim mode and it showed something like 2.6 - 2.7 ms between calls to the motion controller with the error always more than 400ms
this machine has been running a stepper system through the parallel port for 2 years without problems
sim mode? that doesn't do realtime at all
or do you mean a sim config?
so this might be related to the m5i20 pci card
when Ubuntu boots up there is some error message related to PCI and allocating memory but I didn't think it was important since everything works
I believe that the parameter motion.servo.last-period is in CPU cycles, not nanoseconds
the only thing changed between the working stepper system and now is the physical installation of the 5i20 card?
jmkasunich: a new install of ubuntu 6.06 and an upgrade from 2.0.something to 2.2.4
oh, lots of things
did you install from our livecd, or did you get vanilla 6.06 and then run the install script?
the old hdd was an IDE-one (parallel-ata), now there's a SATA drive
I installed from the vanilal 6.06 and then ran the emc2install script
yes. 2.4 GHz celeron, single core
maybe an SMI problem?
was there a message on the list about that? SMI could be disabled somehow?
when I experienced that it was either 3mS or 300mS (don't recall), and it happened every 64 seconds exactly
I had a failed install with SATA, so I just avoid them now. Should I?
this is much rarer. 3-5 minutes at least
kirk_wallace: I don't know about SATA - they're going to get harder and harder to avoid
I have an 8.04 realtime system with sata, that one works fine. I don't believe I've done 6.06 realtime with sata.
awallin: SMI can be disabled, but I don't recall exactly how - alex walked me thru the process
at the time it required compiling a module, but I'm not sure thats still true
if there's something newer than this, the wiki should be updated. http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?FixingDapperSMIIssues
this box is SATA, core2duo, but I'm not running RT on it
I did try it with an experimental SMP RT kernel that cradek built a while back, and that worked fine
Should I avoid x64 too?
K`zan: may be of use http://www.ezlan.net/Distance.html http://www.ezlan.net/antennae.html
and tin foil reflectors
kirk_wallace: when you say "avoid" do you mean you don't have a PC yet and are trying to decide what to get?
tomp: Got the hardware I need, problems is ubuntu 6.06 / EMC release doesn't do wireless.
you mean 6.06 doesn't have drivers for the wireless hardware you have?
Anything else does for this dongle.
can you get different wireless hardware? a nice old cheap wireless NIC?
I'd be better off running EMC on 7.10 without the RTK
not if you want to run a machine
About $50 or so less expensive.
The other alternative is to run wincrap stuff and I'd really rather not.
I guess I don't have anything constructive to offer - I hate wireless with a passion, and would rather hang CAT5 from the ceiling
In theory 6.06 sucked majorly for wireless. Having asked elsewhere everyone suggests anything later.
Not a real fan of wireless myself, but it has its place.
If the solution I have in mind works, it will be a non-issue. Basterdized to hell but cheap and effective :).
And it will be, essentially (as far as 6.06 is concerned), ethernet. Keeping toes crossed.
I avoid x64 and SATA for new EMC PC setups. I look for P4 systems from Geeks.com .
I'm running a P2 from the dumpster on my Shoptask
kirk_wallace, there's no technical need to avoid x64
you can just run a 32-bit kernel on it anyway
I've got wireless running on several 6.06 around here.
did you have to jump thru hoops?
I did have to twiddle a bit.
asa for SATA, I tried an experimental SMP kernel (on a core 2 machine, incidentally) which didn't support it, but other kernels did
rayh: Interesting. Maybe jmkasunich was right, just missing the module for that hardware. Sigh :), never a dull moment.
I know the feeling.
But I need to install from the Live CD.
And I know better than to think the kernel source is on the CD...
One was a linksys wireless-g notebook adapter
"Nobody builds kernels anymore, it is discouraged"...
kirk_wallace, will you have an internet connection after you install?
the kernel source is on the internet.... which you can't get to
I had to run some sort of netwrapper for it on the laptop.
Yea SWPadnos that too.
I am a fan of having Internet on all my computers.
you could always get an external wifi bridge
IMNSHO every installable distro should come with the kernel source tree the supplied kernel was built with (yeah, just a bit frustrated with all the roadblocks lately).
they're $50 or so
SWPadnos: THat is what I think I'm doing, will know once the hardware arrives.
heh - ok :)
Right. My first wifi experiment was the little box that is on the Mazak at fest.
Gonna take a break and beat my head on a corner :), refreshing ATM ;-).
The source is easy to install with Synaptic, isn't it?
strangely, no it isn't
I don't know that there are source packages that are installable via synaptic
app source I'm sure is, kernel source one gets some form of some kernel which may and probably does not bear any relation to the configuration used to build the kernel you are running.
but you can get the source to any package (including the kernel) with `apt-get source <packagename>`
apt-get source gets the exact source and patches used to make the package you name, even if it's the kernel
Sone at least get the config under /boot or /proc, which relally helps.
Yep, but the configuation you don't always get.
I think normal ubuntu has it someplace IIRC>
that is possible, though you can always start with the config of the current running kernel
from /proc or /boot (I forget which)
Much more common that it used to be, thank the ghods :)>
FWIW, I have had good luck with the Madwifi drivers.
ok, there are some kernel source packags available (as normal packages, vs. apt-get source), at least on 7.10
FWIW, I also use these biquads, http://martybugs.net/wireless/biquad/
. They work well and are easy to make.
Oops, I thought kernel-dev was in Synaptic, but I don't see it. Can you do it from Yum? That is the kernel source isn't it. It has been a while since I have done anything with the kernel.
if you want the kernel source that corresponds to the linuxcnc realtime kernel on ubuntu 6.06, install linux-source-2.6.15-magma
if it's not available in the package manager then you have got an incorrect /etc/apt/sources.list or you need to run apt-get update
is it possible to use #include in .comp files?
is it possible to "preprocess" .comp file and get .c code, not executable?
I don't need the kernel source, but out of curiosity sudo apt-get update didn't put kernel-dev in my Synaptic.
ubuntu calls it linux-source
yes, you might mean one of kernel-source or kernel-headers
depending whether your goal is to compile new modules (kernel-headers) or a new kernel (kernel-source)
assargadon: yes and yes
"comp --help" will help you with one of those things
jepler, greate and greate. How and how? :)
ah, thanks :)
and the documentation will help you with the other: http://linuxcnc.org/docs/html/hal_comp.html#r1_13_2
jepler, thanks again :)
jepler: Jon Elson just asked on email how hard it would be to fix the spindle start problem (on start in middle of program). Where would I look to answer the question?
Do you know the general path through the code?
lerman: in emctaskmain.cc, look for programStartLine.
lerman: when doing "run from line", this variable is set to the target line number. based on comparison between programStartLine and the line of the program is being read, something different happens
// we're stepping over lines, so check them
// limits, etc. and clear then out
Thanks. I'm at that line.
Well, I'm not committing to anything. I'm just looking.
pyVCP is cool
is it only way to open it is running it on other terminal session?
I mean, I have .xml file, describing some kind of VCP
and I want using it in my HAL
I should start HAL session, and then I should run one more terminal session and start myVCP command on it
it looks....not cool :)
halcmd can load pyvcp panels
something like halcmd loadusr pyvcp panelname.xml
thanks, I'll try
you probably need some other options there, like wait for component ready and some naming options
hmmm, works funny, thanks. loadusr really just starts outer program :)
I thought it load some kind of "module"
only problem is waiting option
loadrt loads kernel modules, loadusr runs usespace code
loadusr -W described as "wait till component becomes ready"
something like `halcmd loadusr -Wn mypanel pyvcp -c mypanel myxmlfile.xml`
what means "ready", I wanna know :)
the program calls a function named "hal_ready()" when it has exported all its pins
yes, I understand it now, my project works fine :)
good :) see you later
Is there activity on being able to change the appearance of pyVCP features, like more colors, shapes, textures?
not much. it's a fairly light wrapping of TkInter, so anything you find in the tkinter documentation you can do to widgets should be fairly easily to do/modify in pyVCP
A proper documentation of the XML syntax would be useful. Now the error messages are very cryptic if there is something wrong in your XML file
maybe that's called a DTD?
As far as I understand, the "machine" itself is number of configuration files
.xml for pyVCP, .hal, *.ini and so on
is it some list of needed file somewhere?
tool table, var file
the example configs show you what is needed
there are often several machines described in one directory
which makes it hard to understand
well, you need one ini file, one or more hal files, a tool table file. the var file may be autogenerated
an xml file for pyvcp is optional
K`zan: still having trouble with wireless?
K`zan: you can try installing linux-restricted-modules-2.6.15-magma (or something like that, I'll look it up, maybe it helps with your wireless)
K`zan: that name is correct it seems (for dapper)
"it's a fairly light wrapping of TkInter, so anything you find in the tkinter documentation you can do to widgets" Just to understand better, if it is a light wrapping, then everything in tkinter is not available?
I think it more or less passes properties off to tkinter, so anything that isn't specifically disallowed should work
it makes it hard to follow the code, or figure out what options are available, I can tell you that :)
kirk_wallace: everything I, jepler or tomp bothered to wrap is available
That's what I thought.
SWPadnos: but it's the lazy man's way to code it :)
heh, and it works :)
but idjits like me can't figure things out easily
we should have the pyVCP documentation inherit from the Tkinter documentation just like the code does
How hard is it to wrap more? Any links to what I can study (like I have the time)?
the code should be pretty readable. what widget/functionality are you thinking about
things like making the font size get bigger the closer to max speed you get :)
seriously though, color would be a good one
change text color depending on input value, or have a pin for color
I'm not sure if enable/disable are implemented, but that's another good one for a hal_bit input
kirk_wallace: look at LED or Button or some other simple widget here http://cvs.linuxcnc.org/cvs/emc2/lib/python/pyvcp_widgets.py?rev=1.35;content-type=text%2Fplain
latching buttons would maybe be useful?
(not the real NIST way.. but who knows..)
yeah, we have checkbutton which stays in one state, but something that looks like a button and latches?
or multistate flipswitch
I would like stuff like shading and shape of LED's, background textures, etc., just to class things up a bit and maybe reproduce the appearance of existing panels. VCP is such a good feature not to take to the next level.
that's what I thought about the original VCP :)
I don't recall well, but I think I used a couple of HAL components to get the VCP buttons to do toggling. (FYI: http://www.wallacecompany.com/machine_shop/EMC2/dro_vfd/files/
if you study the TkInter docs you will see what is possible and what is not.
(Sticky note on monitor)
kirk_wallace: but the pyVCP Button widget doesn't toggle by itself. It wouldn't be hard to add a new ToggeButton widget that stays down once clicked
# Load xor's and flipflop's for spindle buttons loadrt xor2 count=2 loadrt flipflop count=2 # Load toggle for main power ##loadrt toggle count=1 # load PyVCP loadusr -Wn fred pyvcp -c fred pyvcp-dro2-bp2.xml
Oops, that didn't work well.
I made a component that connects a push-button to 'nist-logic' : http://www.anderswallin.net/2008/04/toggle2nist/
K`zan: or maybe try madwifi-ng (apt-cache search madwifi-ng..)
awallin: is that the same as this? http://www.linuxcnc.org/docview/html//man/man9/toggle.9.html
I just remembered, with my toggle setup, short pyVCP button clicks didn't work. Maybe a shoot of contact cleaner would work ;)
BigJohnT: nope, toggle will take a push-button input and toggle the output between 0 and 1 whenever you push the button
BigJohnT: with toggle2nist I wire first the push-button to a toggle comp. then the output of the toggle to toggle2nist and then the outputs from toggle2nist to the three 'nist-logic' pins in emc: on, off, and is-on
this is more like a 'radio button' but decentralized
so you can tell with toggle2nist if the g code turned off the coolant?
yes, it works even if other stuff like MDI or g-code changes the state of whatever I am controlling
why is not in EMC? It is a comp right? I don't know much about comps yet
or is it there and the manual is lagging...
because nobody had written it yet
as a comp?
or the manual?
it's very easy to install from source. I just made it and posted it on my website. I have commit rights I guess but I'm too damn lazy and work on emc too seldom to remember the cvs commands...
comp: a tool for creating HAL
BigJohnT: hal modules written with comp have automatically generated manpages
I write the cvs commands down when alex tells me
awallin: if you pass on the commit rights I'll give it a whirl...
BigJohnT: doesn't work like that :)
BigJohnT: you are free to take my code off my website. cradek can give you cvs access if you're friendly with him...
awallin: it's not a matter of beeing friendly with cradek
or I can get alex to do it
[18:22:42] <alex_joni> http://www.linuxcnc.org/content/view/50/13/
BigJohnT: you can try :)
ok, but I prefer that you check it first
it works. you can test it with a pyvcp button if you don't have a hardware button
I prefer to test with both as I've seen some things work with pyvcp and not with hardware...
I have buttons galore LOL
are you using debounce on the HW buttons?
no it was a issue with code but that's been fixed a while back
I just hooked up a MPG3 pendant from CNC4PC today on my test computer
maybe "momentary toggle" would be a better name?
er, crap, no
this whole daytime thing doesn't really work for me
fenn: go back to sleep
or "smart toggle"
BigJohnT: added (toggle2nist)
alex_joni: thanks. what would we do without you...
do it yourselfs?
the process would have been a lot harder without "hints from Alex" from time to time as needed
"yourselfs", we are Borg.
alex_joni: I think i Quit with the Duron700, latency 600000 while Harddisk-io , with swapping disabled the same
no ubuntu8.4 on this Box
I suppose some sort of lock-out should be considered with toggle2nist?
to lock out what?
I haven't thought about how toggle2nist enough yet. Lets say something on the screen is locked out in program run mode. It would still be locked out to toggle2nist? I am just trying to prevent inappropriate actions. Or, at least with EMC2.
EMC2 will complain if you try bad things via halui
in the toggle2nist code, 'is_on' is always an r_value (right side of '=' in an assignment ), so it only get values from the hal 'wiring'?
ok, so you can toggle it and somebody/thing can toggle it , and you can tell state by the gui indicator (is this the old joke where the guy closes the drawer and it opens back out :) ?
or 'marvin the manic depressive button, who wont become un-depressed' :)
yep. obviously the push-button doesn't have state so you can't tell if it's on or not by looking at the push-button.
oh, i thought the gui could do that
but an led could
a year or so ago i think we wrote a button the was 'state' aware
gefink: sounds like a showstopper
Yes looks like there is a difference in the kernels
well.. obviously 2.6.15 is different than 2.6.24 :)
also the lost soundinterrupts ar not bounded to swapping or gnome
gefink: sounds like you're stuck with 6.06 :)
or maybe try building a special kernel for youself
i think stay on 6.06. All works good
by time i can try to find the kernel where the problem happens, but i think its not worth the time
kernelbuilding on a duron700 is not fun
True, BTDT :-)
gefink: I know :)
but maybe you can experiment with boot options
like noapic, nolapic, irqpoll ...
yes i thought to thy the polling-thing because i think its an interruptproblem
apic apm and apci was the first try
gefink: so far it proved usefull on some machines which didn't want to boot at all
acpi and apm are disabled in the kernel.. so no real use trying them
boot is no problem
yeah, I know.. but maybe it's something related
its less complicated to use the bootoption as make sure its disabled :-)
by time .....
now i use the 2000+ No Problem, The D700 is conected to my PCB-driller
gefink: if 6.06 works.. why change it?
alex_joni: just curiosity
gefink: heh, I know how it is..
maybe a candidate to replace my over the years updated Suse 6.2 working system
hmm.. I have a 6.3 somewhere.. loved it :)
I think it was my first really serious use of it
yes, but nearly all things are replaced, yast is disabled since years.
I liked yast
I not for all. it killed my sendmail.cf twotimes
but in the "old days" it was nice and fast
automated tools like that are nice if you don't use something else (like manually setting things up..)
btw. yesterday someone wrote about using O call to use a external ngc-file
Is this a 2.3 feature?
gefink: it works in 2.2.x too
but a bit different
you can o123 call
and have o123.ngc
which contains o123 sub
ahh. it must be an number
I tried with a name, got illegal Char.
did you do o<myname> call
yes, because i found nothing in the docu
it's been added to 2.3 docs
BigJohnT: that's only for 2.3
not for 2.2
ok, 2.2 is numbers only?
now its clear
ok, back to my pendant...
I tried 2.3 with prerelease of 8.4 and missed something...so changed to 8.4+2.2.5
I have three inputs and I need to set a different value in axis.0.jog-scale depending on which input is on. Is there anything better than mux4 for this?
mux4 uses two inputs to control the outputs...
BigJohnT: what should happen if no input is on?
it won't matter
so i think you have 4 values
then use mux4 and 2 of the inputs
in this case there is a three position selector switch and one of the three will be on all the time
value 0 and 3 are the selection for the unused bit
and a mux2 for the third
Value 1 for the first, value 3 for the second....
why using an additional input for a signal you know from the other 2
it would be value 1 and 2...
slap on forhead
i wrote "Value 1 for the first, value 3 for the second..." the 3 is wrong. 2 is correct, but i think you have it
yes it's done
actually I'll use value 0,1,2 with no input for 0 etc...
the guilty till proven innocent ploy: value =2 (the assumption) if in1, then value =1 ,elseif in3orwhateveryouliketocallit then value=3
assume the value, then prove otherwise, always saves 1 line of code
yep must be time to start drinking when you forget the basics
a little coke? or beer?
homemade wine I think
it's time to make more beer though
that might make ya forget the basics
but beer usually makes ya smarter
* archivist sips 7 up, and retains brain (I hope)
time for bed. by
as long as you don't forget your name you can walk home
been there done that... ;|)
archivist: you ran the steam pump today
thats tommorow and monday
cool, I was interested in the mechanics that kept the connection straight to the rods but could not see the other end of that rod where it connects
as the beam goes through it's arc
is that shown in the pics you posted the other day?
its the Watt parallel motion
someone took the measurement from that engine and put it on cad, its within 3/16 in its stroke iirc
cool , I see it on wikipedia now
but its probably better than that
I have also caded the watt motion from an engine and its within 10 thou
clever boy was James Watt
pretty smart fellow that Watt
in wikipedia it says Watt's design produced a deviation of about one part in 4000 from a straight line
specially when you see the early form, the links are not equal length
Ive also seen a 1980's article in Model Engineer on the subject with similar good results
this one says it does a perfect straight line Peaucellier-Lipkin linkage
hmm Peaucellier was later iirc
yes 1864 they say
bit of a cheek describing watt as Peaucellier!!!!
* archivist looks in book "Ingenious Mechanisms for designers and inventors"
sweet! a pendant with a mpg for 3 axis in 30 lines or less...
BigJohnT: how long did it take you?
btw, now you can sell your hal file to AramK
about 3 hours including installing the hardware
works pretty sweet
should there be a section in the integrator manual for "Other Hardware" or something like that?
it was part of my homework assignment
keeping the cod clean is a SWEET idea
that took most of the time working on keeping it short and sweet
alex_joni: or would it fit better in an existing section?
* alex_joni looks
yes it usually does.... but spagetti programming is alot of time for a long time...
BigJohnT: that was my guess now too
maybe Machine logic too
but Advanced topics I guess
machine logic to me is ladder and programming
where have you implemented it??
why not add a section in hardware drivers
have you not added ladde and some programming??
I was just commenting on what the topic machine logic would mean to me
* BigJohnT looking for hardware drivers
as was i agreeing
rayh: it's not really a hardware driver
as the MPG is connected to a parport
so it's a component that connects a parport to xxx.
rayh: no, only HAL files explained
BigJohnT: can you pastebin the hal file so we can comment?
so "how do do xxx with HAL?
[21:12:37] <BigJohnT> http://pastebin.ca/1006365
For most of this sort of thing we send folk to configs.
But you are correct that we need a place in the integrator manual.
BigJohnT: small issue I see
you want to connect the mux4.0.out to all axis.*.jog-scale's
same for encoder.0.counts
only the one that is enabled will do anything
but you still want the connections there, otherwise only X works
* alex_joni is off to bed
good night all
good night alex
thanks for the help
Stuff like BigJohnT, wrote is a lot like some of the examples of gcode that used to be in the programmer manual.
rayh: programmer manual?
Programmer section of the user manual.
I had a programmer manual hanging around for a while.
I didn't mean to remove things.. I mostly tried to move all things as good as possible to the new manual
if you find things like these, feel free to add them (they should probably still be in documents/)
We seem to have tossed most examples. Except the hal tutorial.
i still have a pdf of the old programmer manua somewhere
examples are easier to learn from sometimes
Why not create a new integrator chapter and name it HAL examples.
I think a full blown g-code manual could be valuable
tell me about it then show me an example and bang it sinks in faster
with examples and common programming techniques, etc
sounds good to me
That's my impression also
but that's probably too much for the generic User Manual
BigJohnT: you've been nominated :P
sure the integrator manual is a better place.
I don't have a running lyx 1.3 right now.
yes the integrator manual is the place
LyX 1.5.3 (Mon, Dec 17, 2007)
* alex_joni isn't so sure about the Integrator Manual
yea and I don't know the effect of editing there and then asking make to do the work.
I was thinking a standalone thing..
rayh: there is lyx2lyx
lyx 1.3.7 here
Yes there is but I understand it is flawed going backwards.
rayh: then I'd try this.. open the file, save
then make your edits, save as a new file
diff the two files, send the patch to someone with 1.3.x
the few times I tried a diff it was a real mess with the lyx code. Have you had success?
but you surely still have a 6.06 somewhere around :)
rayh: if I do how I said, then sometimes yes
um. yea there is an itx board laying over there with it.
but now I'm really off to bed..
Haven't installed the developer stuff there though.
Catch you later alex. Sleep well.
rayh: lyx should be enough
BigJohnT, do you have LyX on your system?
I added a bit to the manual the other day
on the o words
We could start a new file in docs/src/hal for examples.
that's what I was thinking
Then we can argue over whether it belongs in the integrator manual or a stand alone example book.
should be both LOL
I've actually started a file in docs/src/hal/hardware_config.lyx
Sounds good. Think we should use a chapter name for the start of it.
That way it will mix into most anything.
here you go http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Hooking_Up_A_MPG_Pendant
BigJohnT: thx, I got a real Fanuc hand pendant here waiting to be de-gunked & put into use ( with the little hand cranky pin in the mpg )
Just starting to learn G code. Don't know how to restart the pgm after a M0
Manual says use cycle restart button. Don't have one afaIk. Example says "S', no go.
that would be the same way you start from the begining
When I hit the "run" buttor, it starts over.
I'm using the mini and TkEMC gui's
ok, I only use Axis...
Ill try it.
Goslowjimbo, Mini shows restart if you press abort
So I need to press abort first?
The idea is that the widgets show if you've aborted a program and need to restart at the line executing during the abort.
in Axis you press the || button to continue
the pause button
you can also select restart from the view menu in mini
Thanks, I'll try all of these out. Was there a place I could have searched to find these tidbits?
They might be in the relevant sections of the users manual.
maybe we should add a link in the g code manual or something...
or add a bit to explain...
Let me look at mini's chapter.
this is a real/fast project.... what does it affect??
what need update....
how will it effect installations..
Just tried the mini restart. can't do that EMC_TASK_PLAN_RUN in auto with interpreter paused
i was in LEAN-6 SIGMA training last week.... ; )
did my first real welding today
the workbenches I got from HGR surplus have a stringer at the bottom front - ground it off and welded it back on about a foot back to provide leg room
mig, tig, stick?
tig and gas are the same more or less
except gas is cheaper (to buy the equipment)
and easier to see what you are doing - #5 shade instead of #10
if you have a good fit up you don't even need a filler rod
more distortion though
with gas or tig
gas I'd expect
gas and tig are the same process
just different forms of heat
yeah, but I suspect tig heat is more concentrated
depends on the tungsten size and the distance to the work
I can get real small with tig but you can get just as small with a jewelers gas torch
did you see my practice welds on Thurs?
no I missed them
[22:58:40] <jmkasunich> http://jmkasunich.com/cgi-bin/blosxom/index.html
the key is clean metal with a close fit. oxides melt at a much higher temperature and you will blow through your metal after you melt the oxides or rust
the stuff I was messing with the other day was pretty cruddy
jmkasunich, yes practice, I spent two weeks training at work before they let us out on production
today's stuff I ground clean first
did you have better luck with it today?
I need lots of practice
today I had two welds that came out nice, two not so nice
the nice ones were 1/8" thick angle meeting 1/16" sheet at a 90 degree angle, with both parts ending at the joint
watch the puddle not the arc, clean the parts with a flap wheel, position yourself so you can hold the torch steady
the not so nice ones were 1/16 sheet meeting 1/8" angle at a T
I move the torch away then dab the filler into the pool of molten metal
on different thickness focus the heat on the thicker part
I tend to have trouble with the rod getting stuck in the puddle
more than the thinner part
just dab it in
that part takes practice
my puddle is usually about 1/8" diameter or so
the rod is maybe 0.040?
the back and forth of the torch and the filler rod take some practice...
well, back to adding on to my deck....
talk to you guys later
ohh your 1 of me
i loke to burn pretty hot and quick if i can
dab-a-thon is the name
hit the back of the pool
I either need a bigger pool or a smaller filler wire
or a faster hand...
time to kick back and watch a movie...
its alot of coordination or randon vibrations,,( usually alcohol related ).... but seems to work..