The adeos crew were bitching (rightly so) about Paolo's hacks to the adeos patch
In light of the discussion, I will drop the -adeos from all future kernel related packages.
because you're using the Paulo version?
using the RTAI branch tag in preference.
yes, and it also serves as a reminder as to which tree I used originally.
hence 126.96.36.199-magma for the current kernel.
WTF?.... synaptic is locked up
no response to any click, including attempts to close the window
Only had that with a CPU quiting on me.
Oh, and when a HD went in to thermal shutdown.
other windows are fine (like the one I'm talking to you in)
paul_c My only reason forthe addition of the BDIXXX is there have been so many update lately that it's difficult to tell what is what, plus on old systems it's easy enough to determine what version they are running.
IIRC, the kernel hasn't changed on every BDI-4 release
2.6.10-adeos was used on several, starting from 4.20 I think
jmkasunich I have a stack of BDI... cd's here, I have no clue what is actually installed =)
(and including 4.27 it appears)
2.6.9 up to and including 4.18
2.6.10 up to 4.27
jmkasunich on those hdd utilities... I personally have a habbit of doing a full destrutive test on brand new hdd's.
after this experience I will do the same
jmkasunich better to wait 4 hours and find a failurem than when critical down the road.
unfortunately.... I already did the install
jmkasunich eh, you get practice =)
4.29 will be out by the time your done testing it =)
I've had far too much practice lately
I hope to have 4.29 uploaded tonight.
paul_c why not wait a week between distros?
or a year ?
ok 14 days
I'd wait a week at least to flush out bugs
paul: seriously, about the uname -r tag
you seem to be using bdi.xx as the version in the repository
is it reasonable to append that to uname -r?
2.6.10-adeos-bdi.92, or 188.8.131.52-magma-bdi.101
The kernel modules are the only packages that get a bdi.xx tag
jmkasunich why bdi.92 ?
you running bdi-4?
That signifies which kernel build I'm using.
fire up synaptic, look up the kernel packages
Ok, what about BDI.91/4.29
BDI-4.29 is the complete distro
right, which is what *I* wanted to know.
uname isn't supposed to tell you about the distro, just the kernel
jmkasunich couldn't think of another plce to put it.
lemme run through some of the numbers....
First, there is the version number of the CD
paul_c: ditro isn't a biggy, only a wish list thing. dont put too much effort into it.
Then there are the kernel package revision numbers...
I'm just drain bamaged and can't keep up with them al =)
and then all the version/revision numbers on all the packages.
The bottom line is the revision number of the main emc package - This generally matches the CD version.
jmkasunich why remove the vga= part? http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?BDI4Tweaks
so BDI-4.29 will use emc-1.0-29
was there ever a BDI 3.xx ?
Jymmm: dunno, I copied that from the BDIFaq page (I think the original info came from Paul)
jmkasunich ah, ok. I like the framebuffer part, personally.
though my latptop likes 791
Jymmm: if you know what each part of those changes does, feel free to add more detail to the page
I just know that what I have there worked for me
jmkasunich I just removed the 'splash=silent' and that was it. But it shows itself upon shutdown
(I added the last part, editing /etc/default/bootsplash, without that you get a bunch of file not found messages during startup and shutdown)
OK... 4.5Meg save over the last 4.28 build.
paul: is the initial setup of KDE's toolbars and such something that you configure as part of making the BDI, or is that stock KDE?
Can add icons to it post install.
I know... I need to add a shell icon
I just think it strange that a shell (Konsole) isn't there by default
(mumble, mumble, curses beta bitches..)
jmkasunich it's back after you update
konsole will return.
the shell icon that is
* paul_c blames Jymmm for it.. ;)
paul_c hey you bitch... did you hear me?
I said once you update, the shell icon is back
along with a calculator one too
update as in apt-get update? I don't see how that will change icons?
kcalc added at the request of Ray.
paul_c oh you can add a clac, but bitch if I ask for the shell to return... you bitch!
now, now.... let's watch our language shall we ;-)
updating bdi-branding and the firstboot script will run again on reboot adding icons & stuff.
jmkasunich I WAS watching my lang
Jymmm's testing has been valuable.
whoa... booting up and a shitload of numbers are prepended to a ll the things
timestamps of kernel messages.
in seconds since boot.
that I mucked up something =)
thats a 2.6.12 thing isn't it?
kernel .config option to disable that... maybe a sysfs/procfs or boot line option too.
paul_c oh it's fine, I've just been editing a shitload of files since I updated and thought I mucked up something bad.
well... It keeps you on your toes.
paul_c you are doing a good enough job for that!
jmkasunich I dont have a /etc/default/ directory
ah shit... nm
on 4.27 it is /etc/default/bootsplash
damn tab completion
just removing bootsplash from init and setting the splash mode to verbose should be enough.
paul_c it's upon shutdown that you see the nag msg's
I do most of the CLI admin from an ssh session...
jmkasunich: I think you are going to... um... err... like the 2.6.12 kernel
the RTAI headers throw some compile time warnings with rtapi
I did the first three steps from the wiki (apt-get remove, update-rc.d remove, and removing splash=silent), and I got about 20-30 messages like "/etc/init.d/rc: line 34: /sbin/splash.sh: no such file or directory"
changing the line in /etc/default/bootsplash stopped that
I have to wonder if changing the line in the default file alone would do the trick?
re: rtapi and RTAI, I guess sooner or later there will have to be a distinction made between the different versions of rtai
with configure checking for the right one
right now we have rtl_rtapi.c and rtai_rtapi.c, might have to have magma_rtapi.c, and later fusion_rtapi.c
The magma branch is moving towards full 64bit support
if the APIs keep changing
The API remains the same
just a few ints become longs
not talking about the 64 bit thing, but the magma/fusion thing
if the only problems with magma are int/long, that's easily enough fixed
I think ;-)
back to bootsplash a sec... my tweak completely eliminates it
it sounds like you are saying change "splash=silent" to "splash=verbose" and you get the pretty picture and the scrolling kernel messages?
ok, I'll add that to the page, give people a choice
if they want the picture and the messages, they don't do the apt-get remove or the update-rc.d remove? or do they?
ok, plants watered!
Perhaps shoud add a section for those that want to use diferent bootsplash screens
feel free ;-)
paul_c Eh, how often do you really see it?
(bout time somebody other than me edited the wiki)
time for dinner here, back later
lot of work for 60 seconds
yah! my bashrc is back!!!
oh yeah, that reminds me of another thing to put in the tweaks... bashrc, or "exactly where do I put 'export CVS_RSH=ssh' so it will be done when I log in"
bottom of ~/.bashrc
anyone know what keystick is written in?
I havne't touched c++ in decades
C really, it doesnt use much (any?) C++ constructs ... if you know C it wont be hard
iniFind is depreciated
oh hell... my NUMPAD edits went bye-bye.
wehn I get some tiem I'll learn diff and patch as chris suggested.
I don't know C either =)
diff -uw old new > patch.diff
I like C over C++ anyway
paul_c what EXACTLY does that do?
or what is the result in patch.diff I mean
Jymmm: if you don;t know C++ or C, how do you know which one you like best? Just curious ...
robin_sz dont start
have you got your router going yet?
robin_sz I asked you nicely not to start.
I'll take that as a no then
robin_sz: Quit trolling.
now that's damned annoying... konsole & kcalc are not added to the toolbar..
paul_c did you check form an existing acnt?
oh fuck the lot of you ... I ask Jymmm a simple fucking question .. that I genuinely was interested in an answer from .. and you call me a troll ... still I guess it gives you a small hard-on from some weird sense of power ... anyway, fuck it .. bored now.
* Jymmm tolerance of antagonistic tendencies has diminished for the current moment in time
paul_c hows that for proper brit speak =)
paul_c: is only the actual difference between new and old show up in patch.diff?
old: a,b,b new: a,b,c,d,e,f patch.diff: ,c,d,e,f ????
less the typo of course
ok, so how would I apply the patch then?
patch -p1 < Jymmm.diff
or patch -p0 < Jymmm.diff
ok, that should give me a start. ty
the assorted diff & patch options take a day or two to get use to.
Well, since I'm the only one interested in the pseudo pendant, I won't submit the changes
to which file(s) ?
/plat/bin/tkemc, minimill, axis
it's basically adding binding to the NUMPAD only. Not even changing them.
so it doesn't affect normal usage ?
untested, but not that I'm aware of.
It seems that none of the GUI's even have the numpad mapped at all... not even to theri coutnerparts on the rst of the keyboard.
Or do you have cvs write access ?
I dont have anythign cvs at all.
I was gonna make an entry in the wiki when I get it doen
I want to map the extra keys to other things as well, but want to test it forst.
NUMPAD_ENTER == ESTOP, stuff like that.
I'd bug JMK and get it in the head of emc2
woohoo! axis doens't segfault anymore
let me write it up in the wiki, then if he wants it I can get it to him.
* paul_c hands Jymmm a free null pointer to use in kernel space.
most keys are obvious... HOME maps to go home (if it'll work), make '.' toggle between cont and incre
NUMPAD_ENTER == ESTOP
still leaves 5, /,*,-,+,DEL
ok, heading out to my Uncles bday dinner if a few. see ya later or tomoorow
#5 log current Pos. ?
poor man's probe function
yeah, I do all the mappings, I'll get folks input first and go from there.
wb LawrenceG we all missed you!
uponJoin('wb $nick we all missed you!');
tnx Jymmm... dam matrox g400 gl drivers.... trying to get a stable gl platform....
LawrenceG after I did the updates I got axis to work w/o seg faulting
and this is a ThinkPad i1500
there is a new version of ubuntu due out on the 10th of Oct.... maybe they have some better support in it
ok, gone to dinner folks! Toodles
Are there any video cards that have solid open gl acceleration?
nvidia have h/w acceleration, but closed source drivers that may or may not play nice with RT code.
I pulled my nvidia card and put the older matrox g400 in hoping for better support...
I just tried installing the matrox "improved" binary drivers, but after about 10 reboots and a very unstable system. I backed them out
running ATI cards here - Get 180fos with glxgears
wonder I have a display at all!
Do you play 3D games or watch DVDs ?
no games... sometimes dvds
emc is my GAME
HiRes 3Drendering ?
axis is what I am trying to get working
you probably don't need h/w acceleration then.
Are you running my kernel/emc-modules ?
on this box, ubuntu and using sim.... in shop I have bdi-4.20 with home built emc2 running great (mini display)
OK... 'cos there was a bug in the freqmod module that was killing axis
I mostly want a gcode viewer to check programs before I take them out to the shop and axis looks good, but has problems locking up window
I get there... its just kind of painful
supper time... got to go
there is an emcplot3d in the old EMC tree
compiles OK on this 2.6 box...
doesn't run though....
a newly-built emcplot3d works for me
from emc1 branch cradek_stable
I'll have a lok in a bit to see which branch this is on.
afaik, nobody has screwed it up on the main branch, but who knows
there's been some monkey business there
work seems to have dried up in head..
do you know if that is because it's done, or because whoever was doing it got bored?
possibly off doing other things. Certainly don't see KR doing anything about the bug reports on head.
you mean bugs concerning his changes?
bugs & undocumented "features" that are a direct result of his changes, yes.
ok, I don't really care about that then
just wondered if there was something I could/should be doing in HEAD.
I wish he had put his interp enhancements in emc2, so they could be hashed out and have time to stabilize along with everything else in emc2
branch HEAD to a KR-dev, and revert/merge cradek_stable.
funny thing is I branched cradek_stable for *me*
I never intended it to be the stable reference for everybody
it has at least one custom hack of mine (it's harmless for others)
the writing of var ?
no, being able to specify some emulated and some real IO bits (and sharing IO and step/dir on one parallel port)
it's required for my homemade driver box
when running bridgeportio, negative bit numbers give you emulation (like minimillio) while positive give real IO
Yes - I saw that when reviewing some of the change logs.
can you do step/dir + IO on one parallel port with emc2 yet?
my box needs two digital bits: chopper enable (~estop) and spindle forward
I should probably ask jmk
If all you need is two IO pins, you could do that with an unmodified bridgeportio
no, it'll step on the step/dir lines
just reading back
long time no type
you should be able to mix and match I/O and steps on the same parport with emc2
that kind of stuff is what HAL was made for in the first palce
or link minimillio, parport, simio & a custom extiointf...
last time I asked, someone (?) told me you couldn't do it yet
what specifically is "it"?
have the 8 normal step/dir bits, and the other 5? available for estop, spindle, lube, etc
right now there is very little support for the "io" functions at the high level
ok, maybe that's the issue
IOW, HAL can let you send anything you want out the parport, but right now we don't really have something that takes a "spindle on" NML and sets a HAL bit
alex was working on it, he may actually have much of it done
maybe I misremember the problem
I've been out of the loop, life got very busy
if there's no digital IO at all, I can't use it
ok I'm glad to hear that sharing the port is planned
there is hardware out here that requires it
yeah, once a signal is in the HAL enviroment, you can send anything anywhere
speaking of life being busy - the weekend isn't really over yet is it? surely there's a day or two left
I have a nice new 80G hard disk to replace the tiny one in my machine, but I hate to take it apart, it's so reliable
what brand? I got a nice new 80G hard disk a couple weeks ago, and it's making funny noises
(part of the life that got busy - rebuilding PCs)
ah, mine is a samsung
probably comes from the same place in japan
samsung is korea
uh, china here
I wish the americans could still make some damn thing or other
I'm about to trade in my ford for a volkswagen
lawyers, we make those just fine :-(
and illiterate kids
well I'm going to screw up my computer - be back later...
4.29 winging it's way to the seeding server..
"seeding"? Are you doing bittorrent now?
seeding so that the other mirros can download faster.
otherwise I would be all week uploading a single image to all the servers.
# MTBF - 87,600 hour(s)
I wonder if they sell some product where the MTBF is 1 hour(s)
one hour fuse for dynamite ?
drag race engines
Just one more CD to burn for ol' Ray...
paul_c: I still think we (you) should try bittorrent
that makes 3 - he probably hasn't even gotten the 4.27 one yet (mailed thursday)
14 disks for the complete ix86 Sarge, plus one security update
five for Sarge ia64 (partial set)
ah, thought you were counting BDI's
and a couple of misc data disks.
ok, I misunderstood...
when you said "one more to burn for ray", I thought you were talking to me - I burned and mailed him 4.23 and 4.27
I'll mail 4.29 from here along with the rest of the pile.
you're making him a "pack 'o data" for transatlantic shipment?
got to find a crate to pack the stuff in now..
I wonder what the bandwidth of a 40 ft shipping container full of DVDs on a slow freighter is?
bet its faster than Rays phone line
I think usenet was propogated like that a long time ago
100CDs is ~137.5 cubic inches.
[04:03:37] <cradek> http://www-mice.cs.ucl.ac.uk/multimedia/misc/tcp_ip/8702.mm.www/0200.html
How high is the bit density of DVDs compared to magentic media? I have to guess that it's lower.
in 1987, I bet magtapes were higher
looks like dual-layer DVDs are about .4 GB / cm3 if paul's figure is right
yeah, given that in 1987, DVDs didn't exist
what is a DVD-ROM? 5G or so?
while a 500GB 3.5" hard disk is 1.28 GB/cm3 (1 x 5.75 x 4 in3)
jmkasunich: dual-layer dvd is about 9GB, single-layer is 4.7GB (but those are probably GB, not GiB)
see you guys tomorrow
so what is the volume of a 40' container ?
696600 CDs or DVDs
8' x 8' x 40' outside, I assumed 7.5' x 7.5' x 39.5' inside
86 rows, 18 cols, 450 in each stack
works out to about 1.5 gigabytes per second, assuming dual layer DVDs, and a trip around the world (25,000 nautical miles at 20 knots)
of course, with the container holding ~6270 Terabytes, I'm not sure where you'd get the data to fill it
I think your calcs are out...
could be, calculator fingers might be slipping
one CD is ~5" dia
(7.5 * 12)/5 = 18
18*18 = 324
a spindle of 100 CDs ~5.5"
I measured a thin jewel box, 5.5" x 5.0" x 0.2"
(39.5 * 12)/5.5 = 86
bare or spindled would be denser
324*86*100 = 2786400
4x denser without the jewel boxes
single layer DVDs?
OK, double it then...
Say 30 days shipping
Looks like DLT tapes (40GB native) are much less dense than DVDs at .14 GB / cm3
Gives customs time to pick over the contents.
I recall reading a webpage somewhere talking about the aggregate amount of data the human race has collected (not counting duplication of course)
(40GB, 4.16 x 4.15 x 1.0 in3)
I think 12PB exceeds the sum total of human knowledge by several orders of magnitude
although there are probably pron libraries bigger than that ;-)
I'll email Ray a copy of Encarta then..
you mean, collected in digital form? 12 petabytes is only 121 bytes per day for 6 billion people over 45 years.
consider that the computer age isn't much older than that
and for most of that time less than 1% of the population had access
even now it's something like 10%
Digital computers date back to the 30's
Babbage is even older.
bet from 1AD till 1945 the total of digital data was less than 1G
then there is that bronze "thing" unearthed in Greece a while back
xrays show a complex set of gears inside..
yeah, but some days I "collect" 2GB of data in digital photographs.
How much data does GCHQ & NSA collect per day ?
you can capture a crapload of data from "nature"
perhaps the article I was reading referred to "writings", ie, text in some human readable language
if 1% of 6 billion people collect 2GB of digital photographs once, that's 60 petabytes
in the context of libraries, etc
Sorry Jymmm But that bugger stays banned.
one picture's worth about 2 megabytes, what with pixel inflation
paul_c just doing normal cleanup is all.
but really, goodnight this time
yeah, pics and movies and sound bloat things a lot
Those two were on a ban for a reason.
paul_c Ah. Well chanserv will usually clena em out once a month.
jmkasunich what were you getting from nasa?
or your 2gb I mean
you're mixing things up
paul referred to the amount of data nasa collects (things like hubble pics)
I have some high res nasa stuff.... 300dpi
and cradek (I think) was referring to the fact that a shutter-happy person with a digital camera can take 2G of pics in a day
HA! I collect that in 2 hours
I have to carry 3GB in CF cards with me and then the laptop to empty them
jeez man, haven't you heard of "quality, not quantity" ;-)
They are quality
how many shots?
ok maybe 3 hours, but about 300
the camera can shoot 3FPS
Not like I have a full frame camera or anything
maybe, but that ain't composition and framing and good photography, that's the equivalent of "spray and pray" with an M16
(exception would be fast action stuff like sports, then you need to spray and pray
lol, ok fair enough there. but when I need to take the time, I do.
But when I have an angry wasp... I do a LOT of spray and pray that I dont get stung!
Also, I get the hell out of there too as I'm usually in areas that I shouldn't be in.
Last year I first got stopped by MP, then chased across town by ythe cops.
but, that one wasn't my fault.
I like shooting abandoned places. Adn I was shooting an old aerospace factory where the weeds were 6 ft tall.
exotic taste in photo subjects eh?
Well, this property is adjacent to other proprety and no visable boundry lines/signs, and well lets just say that was no mans land.
I remember seeing a guys webpage that was full of that kind of stuff, abandoned mines, factories, etc
jmkasunich I got inspired from this website --> http://www.abandoned-places.com/
abiet, not much of that in this area (SFO)
I had to go to the OTHER side of Yosemite to find somethign old and with character.
annoying web navigation
his stuff is good, so I let the navigation slide.
I love the staircase with ivy (if you find the link, let me know)
neat old stuff there
I like the big old machines
new stuff is so sterile and boring
yeah. kinda why I got into the Del Monte packing plant here. been abandined for 30 years, around since the 1900's. The guard even let me get on the roof
this is pretty good... http://users.pandora.be/a-p/images/asylum14.jpg
I tend to take much less "artistic" pics - more of a documentary thing, "look at this big machine"
In my case... I try to make the viewer go "wait a sec..." and sort see something that makes them go back for a second look.
what's a GOOD sftp client for linux?
something nice and easy to use.
it's bedtime here, goodnight all
never used sftp myself.
nix or win?
hsftp or vsftpd for server side.
nono, I'm looking for a client.
I'm resorting to wine + filezilla.
one 40' container loaded with DVDs (assuming 30 day transit time) is equal to transmitting one DVD/Sec.
Hell... Faster than some DMA transfers.
that's what a zillion dvds?
Now, how long would it take to watch all those?
[05:08:36] <Jymmm> http://boss.streamos.com/wmedia/gecorp/wms/singingintherain_300k.wvx
* paul_c gone too.
AchiestDragon: you there ?
anyone around ?
this circuit is driving me crazy :(
do you remember the optoisolator circuit I build ?
do you have the diagram
[12:49:09] <Jacky^> http://www.roboitalia.com/modules.php?name=coppermine&file=displayimage&album=158&pos=0
i'm testing with a squarevave generator and scope
it work, i get the square wave on out
but the out are only in 5V state
i think high ..
so 0v off and 5v on
i tried to add a resistor between out and gnd in the uouput side
i always have 5V :\
using a resitor the voltage go down
but also the square wave !
ok ,, on the ouput (pin 6 of the opto ) you have a 1k resistor ,right
remove that form the one you are testing
i tried to remove it, but if removed won't work
i mean if i remove 1k resistor i get no signal in out
ok so you say always 5v so what level dose it change to
0 i suppose ..
in normally state the pin driver clock should be 0 not 5V , is right ?
0v and 5v are ok
not for the limit switches
pin 12-13-15 should be low
for how i'm running emc
for a logic 0 it should be between 0v and 1.25v and for a logic 1 between 2.5v and 5v
I know ..
the problem i cant get low state
you may need to invert the signal
by software should be easy
but if not then a 74hc04 should do the trick
but im not sure if it will work
when the software reads the port if its told to invert it then it should be ok
im not shure about emc but kcam lets you individualy set the port bits so you can invert any one
input or output
i've a small circuit to isolate the optoswitches
i've 3 optoswitches for axis
the output of 3 optoswitches are going to an IC 7432
all 3 home (for example)
and i get only one out
the signal are isolated
do you have diagram of that?
just a summary schematic for the 7432
ok , 1 min
anyway the output of any 7432 is inverted ..
that why i'm using an inverted config on emc..
7432 is a logic or , and not inverted
but.. entering in input with an high state i get the inverted state in out ..
on input 0, 0 = 0 1,0=1 , 0,1=1 and 1,1=1
on the input to the 7432 one pin goes to the opto isolater output where you taking the other input
this maybe because not all optos are on at the same time ..
if phisical limit = a logic 1 into the 7432 and you have 2 of them then it should be right ie any logic 1 input to the 7432 should give a logic 1 on the output
i think the optoswitches work inverted ..
just a second
looking for a link ..
if thay do then you will need a diferent chip
if the switches give a logic 1 normal and a 0 when at limit then use a 7408 inplace of the 7432 (need to check if the 7408 is open collector output or not )
[13:17:35] <Jacky^> http://www.temp.eleinmec.com/issue10.htm
its the small circuit in fig. 5
i've 3 of these for any axis
0v closed 5v open
so if its closed for limit then its 0v at limit
change the 7432 for a 7408 (same pin out diferent function )
sould work ok then
thanks for the patience :)
for it to work as is you will need all 3 to be closed to get a change
Hi Paul. How goes the software/package battle today?
Heading off out to the post office in a while.
1lb of CDs to mail out.
Wah! That'll cost more than a pound.
I'll owe you big time at next year's fest.
ksirc here doesn't want to show that fancy L
Any news on that ini yet ?
Have not heard. I'll call em after a bit.
When you do, tell'm I have a beta ready.
If the ini is OK as it is, I'll upgrade it to "final"
k -- this is with our changes from yesterday?
I see one error message "stepping type not found"
freqmod defaults to step/dir
right. Post Office. Be back later.
rayh: Box posted - Should be 7-10 days.
[Global Notice] Hi all. John Reeves Hall, a community hacker mentioned here previously, passed away on September 17. ( http://overcode.yak.net/
) He'll be missed. I know he would want to thank all of you who contributed to his drive for the American Cancer Society.
_AchiestDragon is now known as AchiestDragon
Not as expensive as I was expecting..
mailing out coasters eh?
wonder if ray is around...I am still getting quotes for ATC spindles but I would want to be confident emc could be made to do that
it's a lot of mony
it probably would handle it nicely
but better check to make sure
I tried to pm ray but he is away from the computer I guess
I got quotes from HSD and today I talked to Colombo
Also need to get quotes from Perske
The Columbo 10 hp or so water cooled spindle has a noise of only 72 dba at one meter
that would be nice
the porter cable is nearly 100 dba
oh yeah...everyone has to wear ear plugs
I need to go for a bit
be right back
Of course while cutting the Colombo would have more noise
yeha but it's still less :9
les just use a plasma cutter and be done with it!
* Jymmm just had an epiphany... encoder dont do shit for position
if there's backlash theres nothing an encoder will do for that
no, but if theres a following error..
it's measuring the turns in the motor, not the position
but a dro (scales) would/should do both of those
linear encoders do..
but you better forget about thoose..
linear yeah, not rotary
with some trickery you can do linear with a rotary encoder also...
you mean $l4I$n$e$a$r$ ?
unless it's thoose chineese ones..
but +/- 0.03
not chinese for $500
someone here gave a link last week, I called them for a price
ah well, I was thinking some of thoose scales robin posted...
but, if you have backslash thoose wont do you any good..
because your machine will slop along with each turn of the cutter
and ruin everything you cut..
breaking cutters/causing toolmarks..
that's the extremes..
it depends a bit which way you are cutting..
les: Hi. Was away working on an ini.
What you need.
anonimasu well, the control would have to compensate or stop things
rayh: cheeseburger, fries, and a coke.
Jymmm: dosent matter really
I have questions about emc and automatic tool change
you are the expert right?
* Jymmm chuckles *
I can try to help you.
Has anyone ever implimented ATC with emc?
I understand only part of atc functionality is written up in emc
Yea. Matt and I under BDI-4.20, soon the Mazak will with 4.29/30 and EMC2.
What INYO is missing?
A HowTo in respect to toolChanger?
Weel , not sure. I am getting quotes for VERY $$$ atc spindles and I want to make sure emc can be made to work with them
Yes it can.
for exmple...what happens right now when a M06 is called?
You will need either a drive that includes orient or add a spindle
rayh w/o being too much of a smartass.... have you personally tried it?
it seems to just pause a bit
rayh ok, fair enough =)
If you are using bptio it simply says that the tool you asked for is now in the spindle.
* paul_c vouches for seeing Ray's toolchanger code work.
* Jymmm wonders how much paul_c got paid to say that?
which io program has tool change stuff?
paul_c damn your cheap!
Matt and I modified tkio by adding the stuff to drag in the air spindle, watch the turret rotate and stop at the right tool.
actually paul_c paid to see it work.
oh yes tkio
I should try that
Well if yoy know c you can mod bptio and do similar things.
I can try
I se no place to store tool locations currently
If you wish, there are tkio/iosh chapters in the handbooks that will help
for pick and place in the work envelope
Oh pick and place.
That is a bit of a different story.
of course part of it could be in motion code
There are two ini variables. I forgot their names.
something about tool_change_position and tool_clear_position comes to mind.
Mxx hase to invoke a move to a specific place
one thing I don't have to do...
If they are in the ini, the machine will move to tool change at the start of m6
I like orienting spindle drives myself.
all that is in the handbook?
a single contact close will do orient.
Let me brainstorm a bit here on what I see.
I will go to the handbook too
This can be implemented with either HAL or tkio
First is a g code that tells next tool
any time after last m6 and before the current one.
I am not finding anything in the handbook so far
What this command requires is a tool prepped messaged into status.
can someone recommend a good dial indicator to use with my sherline as well as a source? it would be nice to be able to use a magnetic base or put it in the chuck.
mrallen: Western Tool has a cheapy for $15 and a mag base for $20
got a link?
mag base will be a problem with the alum sherline.
I used a 6" piece of ground precision steel
my machien is all aluminim too
enco or shars have plenty of cheap indicators...they work fine
the website sucks (for over a year) but you cn at least get the closest store to you http://www.westtool.com/
Hey Les. Let me put this in the wiki page on tool change.
ok even better
i was advised not to go cheap on the dial indicator
I have used them for many years...always worked fine
paul_c: Is it possible to change the value of a global like tool_change_position
mrallen: Well, for $15 if it isn't good enough, you can always get a good one
I have plenty of Starett, federal, and other indicaTORS BUT THE CHEAPIES SEEM TO WORK ABOUT AS WELL
rayh : FYI... I'm working on creating a 'pseudo pendant' my using an external NUMPAD
hit cap lock
I'll work at http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?ToolChangerImplementation
What do you talk to the atc plc with?
Oh you want a separate plc?
or perhaps emc is the plc
tkio is a plc so is bptio
no I don't need separate....
but do need a few bytes of io somewhere
A separate plc would work for carousels and arms and such.
not needed for pick and place.
Yes and it can be a parport
or a dio card
well I have to verify tool seating and stuff like that
Could use spare pins on the stg
24,000 rpm iso 30
I think I use all the dio
3 axis with all limit switch etc
Got an extra ISA slot
just cheap dio card?
rayh: Um.... ISA slots are about as rare as Full Service gas stations, or even $0.25/gal gasoline
I verify tool mount/dismount and spindle rotation
would use hall sensors
[19:22:52] <rayh> http://www.winford.com/products/crd155b.php
that looks fine
the bsoft board site seems to be gone. They were about the same price for 48 IO.
24 is plenty
winford has a breakout board as well. External pull up if you add resistor packs.
I'm really not trying to be a smartass here, just that ISA boards are rare and if you ever have to replace them in the future, you might get screwed.
one issue with pick and place is the tool location coords
idal place for that would be the tool table
yeah jymm, but i't 50 bucks
There are the two basic tool change locations.
already available in the emc ini file
from the ini file.
easy to expand?
first is a location that the machine goes to when it sees an m6
the second it goes to after it sees an m6 and returns from IO program.
it returns to the first after another IO return
the trick I want to ask paul_c about is can we change the tool change location between
if so we can do pick and place right now.
what i'll do is let you write this in the wiki for all to see rather that ask questions now
the tool clear position raises the spindle above the tool holder
Okay. I'll get that done sometime today.
But to answer the general question -- yes we can change tools.
Yeah at tool clear the hall effect could probe the area and check for proper engagement
What started this is the tp/tc bug fix
I've known folk to use a photo interrupter to see that the tool is clear of the spindle.
that meant I could go faster....
and run more power....
but it also means I have shorter tool change intervals
hence 100 a day....
The easiest tool changer would be a carousel or moves tools under the tool change, tool clear position.
or slide that ...
I could do a carousel
but a router has such a big work envelope...
I'd think so. If you do it is a trivial matter to implement
You would have to rapid to the same place for every tool change.
Well havee to have a meeting with the customer before I spent the big $$$
That rapid move would stuff the old tool into the carousel
then raise up to clean, spin the carousel to the new tool, and drop the spindle to pick up the new.
control carousel with emc or plc?
We could easily do this with tkio and add in the sensors you want
I presume tool clamp, tool release, spindle oriented
I don't have to worry about orienting, but I undrstand in general you do
Tkio can do it with a winford card or (jymmm) a pci
If you have a drive dog that you need to engage you would have to orient.
If the angle between tool and spindle is not critical then you don't need to orient.
right...thr industrial routers seem not to do this. Low cutting torque, and balance issues of drive dogs at perhaps 24krpm I guess
typically iso 30...without dogs
Okay then you just have to make certain that the holding shaft is in place.
Most spindles have locations for clamp unclamp sensors
speaking of which...just got the colombo electrospindle quotes....looking now
$5610.00 for the motor
$995 for the inverter
I can do better on the inverter
now...the manual er25 collet version is:
draw bars must be made od gold
Sounds like it or is it both the draw bar and the motor shaft.
How about the puller or pusher for tool release.
It"s a lot of money...but just imagine using hand wrenches for 500 changes a week
the other option is to palletize.....do tool one on ten pallets, then tool two, etc
but that is a bit of a pain as well
better than a tool change every few minutes
by hand, yes it is.
25 hours a week of just tool changing labor
$5k is not sounding so bad
note my attempted self justification.
les if it IS made of gold.... then you have soem blin blig =)
If you're doing the 25 hrs then it is much to much.
well, I'll go get a bite and leave you alone. Thanks Ray.
something strange just happened..
it seems like emc is ignoring max_accel and default_accel for a axis..
that would be bad.
my only stepper axis whirrs when I try to move it...
for rapids it may use the value in [traj]
I remember something is screwy about that
that's for the other axis'es..
and something is surprising - it uses the min of those values, or the max, or something
don't remember if I actually figured it out, or just noted that it's screwy
cradek: could you look at it if you have any time to spare?
sorry, I have to leave soon
oh, not <now>
if you figure out what it's doing, and can make argument on the list that it's wrong, I bet it would be easy to fix.
oh, it's just not obeying the max limit..
err accel limit
it may be using max(traj limit, axis limit) when you expect min(traj limit, axis limit)
try rewriting in terms of max and see if that works
(I haven't looked at the source)
does the axis limits superceed the traj.. or the other way around? in that case
depending on which is higher right?
I don't know, that's what I'm saying
I think it might (surprisingly) be max, not min
err max and default..
ah well, I'll have a look at it..
for g1 it works nicely
this per-axis accel stuff was put in for rotary axes (because they are in degrees) and in/sec2 and deg/sec2 are *very* different
so I should throw the stepper into the lake?
I don't think it was ever really debugged properly.
it's probably a simple bug if you can figure out what it is.
I have a spare servo off work.
might be a easy fix..
what file does all that reside in?
I dont have too much of a clue about locations in emc1
somewhere in /emcmot probably
the vars from the ini get passed around in /emctask
ah well since it obeys the max_velocity I can use g1 moves in my cam generated stuff
I am kind of short of time :/
going to be out with customers all week..
hmm just got a call for Bill Glen, the developer of the warthog cnc router product line.
want to work on some stuff.
The guy we were trying to get hold of back in May ?
yep. He is trying to resurect from the ashes
We have had a look inside some of the cnc routers we dicussed....
they are HORRIBLY misengineered
they look good on the outside.....
bet there will be a fair bit of us.gov money if you try to set up down on the Gulf..
for example a 1.5 meter y axis consisting of two 25mm unsupported round rails
well another storm is heading for it you know
just heard that on the news.
* Jymmm cries
anyway the routers are so misengineered that the only option is for us to design the machines...they just build them.
They seem to be amenable to that idea
Bill Glen sold a lot of machines...he just overextended the business too much
I am trying to convince hime that we cannot build machine tools in this country.
we can sure as heck design and sell them!
Like Pacer did
get a single supplier agreement for the US market ?
how do you enforce that?
Well if we design the machine it's not really an issue......
the chinese machines are junk....who would want to be a supplier?
lots of iron and aluminum...
put in the wrong places
chinese machine tools have improved a lot in the past couple years.
well, there's a lot of variation.. but the 'good' tools are becoming fairly decent.
sure I have lots of them...tons and tons...
but we are talking specifically about one cnc router
be careful with overseas production.
you have to go there.
I've only worked with it for semiconductors, but I've unexpectedly seen -my- silicon being sold through other channels
and have had to litigate
s. korea, hong kong
thats what i was trying to say.... icee said it perfectly.
they steal all of les' engineering as their own.
* icee nods
I have had infrigement litigation on one of my patents in japan
sell the same machines to another distributor
les china is a WHOLE nother story though
it's an issue
the ONLY thing you could enforce is non slaes in the US
Sounds like having them make subassemblies would be safer...
bolt it together elsewhere
and that's by US Customs, not the asian mfg
les fabricated in china, assembled in usa
customs doesn't really care, too often. you need tight distributorship agreements in the US so that when you see someone selling your machines, you can be sure whether they're authorize
actually, might be china for customs too
For example have em make various frames and assemblies....
and you need audit in your distribution agreements, too, so that if you find out someone has been purchasing directly from the manufacturer and bypassing you, you can prove it
icee les can get a pantent on the design, so if they sell inthe US, it's a ptent issue.
but don't teach them our quick methods for laying profile rail with great precision
jymmm: *nods* us customs, though, unless the particular problem is very widespread, isn't verifying patent compliance
design patents, invention patents, strong contracts, etc.. are all good things to have in your back pocket
icee: due process implied.
jymmm: so? you get an injunction against one firm, someone else brings it in
at least customs is a way of stopping the import to the US.
I am just filing on the air generator thing
icee the source would be flagged
icee at least it's one legal method, better than NOTHING at all.
Well anyway this thing was dead before due to the emc tp bug
that is not a problem anymore though!
Yes, emc had a bad error in it's trajectory planner.
* icee is just building his first cnc mill now
It rendered emc useless for any kind of high speed machining
we talked about it for years....Paul finally found and fixed it a couple months ago
What are you building icee?
I'm designing a high current stepper driver board
I intend to use it first just with a mini-mill I have sitting around, but will likely acquire a full size mill once I have pieces working
well old full sized cnc mills with fried controls are cheap...almost free
* anonimasu cries
a broken cnc mill(197x) over here is like 4000eur
get old clunkers for free even with working controls.
one in crap condition
les: I was going to ask you if you've seen the phenomen..
les: no clue
some kind of tax?
bp style mills are about 4000 eur..
sounds like a business opportunity to me ;)
Are clunkers free in some of europe paul?
they are here.
will someone send me one of these "almost free" CNC mills? I'm in the USA...
Some down here....
[21:11:37] <anonimasu> http://www.gnosjomaskin.se/productlist.asp?Objekt=Maskin&Status=Beg.&Kategori=3&Split=30&Sortera=ArtikelId&Action=Search&Submit=Starta+s%F6kning
BTW machine dealers are not the place to get them
that's a machine dealer..
anonimasu: holy crap, somebody umlautted all over that page
I don't know what a BÃ¤ddfrÃ¤s is but I want one
click the english flag on top
Rebuilders and auctions are the place
remove a zero and you have euro's..
I should throw 2000eur in some pocket for a bp style mill..
anonimasu throw it in mine, there's lots of room
but then put them out in a yard uncovered and let them rot
Jymmm: no way
Jymmm: in my own for sabving
Tools from schools are usually pristine
then maybe I can actually get the correct replacement parts for the 3rd time
les: can you limit one axis acceleration?
err rather does it work?
I tried limiting my max accel on z
but it dosent work..
anders: you can, but I don't
les: it seems like it dosent work..
I think not
I saw some stuff on the list about that
anonimasu: There are per axis accel & vel. limits
paul_c: they dont seem to work..
g0 bypasses the max accel..
when jogging the do.
yeah, but I am not jogging
I am trying to do g0x0y0z0
but cordinated moves, the TRAJ settings are used globally
paul_c: what can I do about it?
the z axis is a stepper, and it cant accel as fast as the servos axis:es
You still on EMC1 ?
no driver for usc
and I have no time to mess with it at all
BDI-4 has a working USC driver.
hm is that in the tree=
err main branch
bdi-4 branch... tagged bdi-4_27
would that compile on my box?
I need to commit the bdi-4.29 fixes.
[21:23:09] <anonimasu> http://www.blocket.se/view/5936461.htm?caller=nbl_s&l=0&c=1&city=0
but I need to fit it first ;/
If you are running a Debian install, you can download pre-built drivers.
anonimasu: You want to machine tank parts with that ??
looks big in the picture
it probably is..
thousands of kg?
for conversion ;)
with some large servos on that.
well 15000 Kr is getting better
drop something on your foot?
damn earth ive under the foots !
another parport rousteed
well, too large but damn.
that's a nice machine..
I dont care for size..
as long as it can fit
the trouble is that with my current mill tool steel is a dream ;)
you deserve big machines anyway
Ah so a nice parport roast in italy....mmm
* anonimasu pats jacky on the head
ahhhhhhh !!! umpf :\
well roasting parports till golden brown is hard work
damn I forgot to buy some coke..
almost quitting time for me....I have to go out and catch a nice fat bug for my pet mantid's dinner
a bumblebee would be good
a chicken that thinks it's a dog, and now a praying mantis...
It is wild to watch that bug
you been sipping that moonshine again, haven't you?
I'm not alone...I joined the mantid internet site!
a bug that grows to 6 inches long, swivels it's head and watches you....
and looks like an alien
and even eats BIRDS
those are cool insects but how can they kill a bird?
I had a site somewhere that has pictures of it
[22:06:47] <les> http://www.birdwatchersdigest.com/site/backyardbirds/hummingbirds/mantis-hummer.aspx?sc=birdwireJul2005
and that is just a small crolina mantis
mine is a muchlarger chinese mantis.
He just came in the house and set up station on my computer monitor!
so I put him in a nice terrarium
10 cm long!!
max size=15 cm
make a nice snack for Congo.
congo was scared of it
a cat that is a chicken.. ha.
here's a picture of a mantis I got a few years ago: http://emergent.unpy.net/index.cgi-files/sandbox/mantis.jpg
er, a picture I took
nothing special, but I was just reminded by this conversation
only one problem I have...after the first hard freeze no more food for it
you can always get crickets at a pet store
will it eat that?
yes they will
I will have to take some macro lens shots tooo
I'd like to see 'em if you do
I'll do that
I think this photo was with nikon 5700, it has a pretty powerful macro mode
it's wild how they turn their head to watch you
well, off to catch bugs. Later!
so i have a lot of questions about EMC, but i feel i should be using EMC2 before asking them; they're pretty "basic feature" questions that may be answered in the new UI, or at least I'll be able to contribute patches to an active codebase
'pends what the questions are..
well, there are several "simple" features missing in the UI; better reloading of modified files for one
e.g. i have to "open..." the file, and even then the editor contents (if open) doesn't get synced
but this is EMC1, so i want to use EMC2 with the assumption that UI may have been improved in some ways
the UI is mostly the same
sounds like you are using mini
i.e. these questions may be moot if i can upgrade; however, i'm still using non-RT, so the configure script stands in my way
i'm using 'sim.run'; whatever that gives me
let's have a look...
for what its worth, i think you need to provide trivial non-RT support, as that will get the program installed on many more computers (if only to use the backplot)
that said, i'm willing to help code a solution; i've been thinking about it since hitting this over the weekend
DISPLAY in the standard sim.ini is tkemc
zwelch: Yeah, I might keep more of an eye on emc2 if it had non-rt support
my thought would to be to add a "--with-simulation-only" or similar configure option, which selects a "nonrt-config" script that acts similarly to the other RT options; the new config script would provide the same information, but with "dummy" stub include files, the default compiler, etc.
that way, all the non-RT bits are then inside the package, but act just like a "real RT" installation
Much of the RTAI/RTLinux differences are hidden behind the rtapi wrapper
Using standard posix IPC calls shouldn't be a major problem in a sim layer.
but I suspect JMK's HEL stuff might cause major problems....
okay, i've now glimpsed at the rtapi dir
the build system is not what i would call transparent, though
It's a damned sight better that the rcslib/emc build system
btw, was it intentional for the emc2 configure script to no accept "--without-rtai"; it interprets that as --with
* zwelch can probably fix that
there are three --with-<realtime> options.
rtai, rtlinux-free & rtlinux-pro
right, but i want a build script to be able to explicitly specify --without-<the others>
the last not being supported at th mo.
I would suggest a --without-realtime then
well, that's not the standard way of using the options
because that would imply a --with-realtime twin, which duplicates the above
i think you meant --disable-realtime
(look at ./configure --help
A --witout-realtime would be a lot easier to type than three --without-<realtime>
--disable-realtime works too.. It's just symantics.
actually, it's more a matter of standard GNU autoconf coding conventions
and regarding the build script; i am thinking about gentoo ebuild common practices
they have bash functions to automate the --with/--without option generation based on their USE flags
not familiar with ebuild
but in the case of no RT use flag being specified by the user, the build script could just as easily decide to simply add '--disable-realtime', if that sounds good
.ebuild == rpm.spec, but more developer friendly; they are bash scripts that export 'unpack', 'compile', 'install', and other functions for Gentoo's Portage package manager to use
zwelch: Using that argument, why not embrace Debians debhelper scripts ?
Simple, easy to use...
* zwelch sighs
let's just not go down that road :)
leaving the philosophical realm of package management politics, let's go back to how we got there: consistent use of establised GNU autoconf conventions
well, it seems like you are advocating scraping the current config.in stuff..
regardless of which distribution packages this, i was simply advocating using the autoconf --disable and --with options as they have been established to be used
i'm not sure where i advocated scraping anything
i'm advocating fixing a few simple bugs in the configure.in script, and adding a new --disable-realtime option
Perhaps I misinterpreted "regarding the build script; i am thinking about gentoo ebuild common practices"
only so i can go do it myself an know that the patch stands a good chance of being accepted
i believe you did; i meant there that common practice is to always specify either --with-<option> or --without-<option>, if that option is controlled by a USE flag (i.e. a user configurable binary setting)
... and that practice stems from the fact that such are the sematics of --with-* flags, as defined by the GNU autoconf standards
* zwelch often hates irc, because despite his attempts to be as precise with his language as possible -- people still manage to read the worst in it ;)
Although a --with-rtai automatically implies a --without- for both RTLinux variants..
yes, but explicit it better than implicit
because an explicitly configured system more clearly identifies the options choosen; in the case of maintainers changing the "default" logic, the explicit flags should always give you the same features
The configure.in could probably do with a quick clean up - There has been some additional cruft added that isn't used/needed.
now, i do agree that the rt options are a special case that are somewhat outside that somewhat pedantic reasoning, but i think the principle is sound
well, that would be best to do first
for sure, --without-[tcl|gtk] makes sense.
let's think about this for a second
well, i just went back and read the relevant chapter of the GNU autoconf manual covering the use of --with flags
and my pause for thought was found to be invalid :)
i was wondering what the Book said about "disable vs with"
(or enable vs without)
the summary is that enable/disable is for optional (internal) features; whereas, --with-<package> specifically is meant for optional features that depend on an external package
so the --disable-realtime is correct; but the --with-<rtai|rtlinux|etc> are also each correct and valid as they already exist, except for their improper handling of the --without case
... which incidentally, should be solvable simply by checking for the value 'no' :)
(i had wondered if maybe the --with-rtai would be more properly be done as --enable, but they would not)
As I see it, ignoring the possibility of a --disable-realtime for a moment....
silently ignoring a --without-rtai is no worse than an explicit check
It saves a few lines.
from what i just read, --without-rtai gets internally translated to --with-rtai=no
so you have to check
[23:23:29] <zwelch> http://www.gnu.org/software/autoconf/manual/autoconf-2.57/html_mono/autoconf.html#SEC138
* zwelch shrugs
also, i admit that three --without-rt* could silently equate to an explicit --disable-realtime, that is only true until the configure script adds a fourth option (rtai/fusion, maybe?) and all build scripts would break
At the moment, some of the make files are starting to get on my tits - Perhaps a reason here to sort them out ?
well, what i'm looking at presently is simply a workaround for the most superficial of the problems; there are likely to be other problems as I get further along
i'm personally not ready to say "sure, how hard could it be to just rewrite the build system?" :)
The "build system" is nothing more than a collection of Makefiles
but i could look at coming up with a better solution in the process of getting things going
if i had my say, i'd use the full autotools suite (autoconf, automake, and libtool)
but i say that without having looked at all of nope
neat feature that... :)
Question: How well does automake handle compilation of kernel modules ?
paul_c: how the servos are working on emc, like stepper ? clock-direction pin ?
the automake system can be used for everything except the kernel module builds, simply calling the existing Makefile
(assuming that it can't handle it at all)
Jacky^: step/dir, quadrature (grey code), two phase, and four phase
those are the kinds of issues i'm sure to encounter along my way
zwelch: So a separate make file would still be needed for the kernel stuff...
automake is pretty durn flexible
the trick is figuring out which tool to use without loosing your fingers in the process
way too flexible, in some regards ;)
autotools can also get plain nasty to work with..
which is why my instinct was to say the above: use the right tools for the jobs
autotools are great, until you start to need to build things like kernel modules
Most of the code is pretty simple, and isn't really designed to run on non-linux platforms.
The RT patches kinda dictate a Linux install.
moreover, i'm looking at this from a "least development cost" too; it may be trivial to convert the majority of the code to full autotools, but why spend another 90% effort getting the kernel modules coverted when there's an existing Makefile?
kbuild does a pretty good job of hiding the evils of module make rules.
well, if a non-RT version can be build, you might be able to run the simulator on windows... why wouldn't you? (other than the "typical" porting issues encountered)
I can think of plenty of reasons for not wanting to support M$ platforms...
well, so can i :)
but others may not be as altruistic, and should we stop them from adding those bits (and risk a fork)? :)
One valuable resource that we have very little of....
Active & committed developers - Their time shouldn't be wasted on a whim.
well, i'll be blunt - the best quality developers are committed because it pays them to be, typically because they are paid by a company that is profiting from the work
Not happening with EMC..
so, e.g., sherline does not contribute to the development?
there aren't any paid developers actively working on EMC
* zwelch sighs
that's just so *wrong* :/
(excluding the NIST guys here..)
hmm, perhaps i have overestimated the use of this code in industry?
Oh.. You find parts of EMC being used in all sorts of oddball applications.
or perhaps - more probably - i am once again cursed with an inflated sense of optimism about the rate of "giving back" by profitable users of this free software?
Several universities use the code for various applications.
presumably, these academic and "oddball" users become contributers, in some fashion?
If you want to see a commercial application built on the original EMC code base....
[23:58:14] <paul_c> http://www.artofcnc.ca/
number of bug fixes and enhancements fed back.... F.A.