* chinamill is away: fixing mill
looking with interest at lerman's measurements
I think 100 blocks per second is not a brick wall....
suspect trajectory cycle time was set at default .01 sec
We have gotten several times quicker than that
but still far too slow for HSM
normal ratio of cycle time to servo time is 10:1
but it can be set to 2:1
the lower the ratio, the less cubic sub interpolation occurs
Since block processing (tp) occurs in the real time thread servo rate ultimately determines block rate
les_w: could the lookahead be altered ? to process blocks faster?
well, two philosophies....
one...do the tp in real time
that way when dynamic scaling occurs the plan remains time optimal
do the tp before program runs
yeah but where would the brick wall end up like that?
so scaling must be done by changing servo rate
with the first....path generally gets more accurate as you slow down
with the second it does not
BTW emc does the first....
so max vel and accel are not scaled
however this requires that the tp is part of the real time thread
with all that computational overhead
currently the brick wall is simply servo rate.
That controls block rate
What happens when you raise the servorate?
the interpolation gets messed up isnt that right?
too short segments?
at some point user space stuff doesn't get any slices I guess...I observe that the display and io go nuts
then it locks up
that's a odd behavious
shouldnt there always be segments unless they take to long time to calculate?
well at some point thr rt thread is hogging all resources
hm, could you give the RT thread a own processor?
like with hyperthreading or something
and in emc it was chosen to do the tp real time
that should provide lots of cpu time overhead for the TP
yeah, I guess it could be multi
others would know better than me
I think Paul and I managed block rates in the area of 500/sec or so
still far too slow for HSM
how fast would HSM be?
several thousand per second would be good
is that possible without dacs
consider a 20 mm radius circle made up of g1 moves
at a slow 50 mm/sec speed
at 1000 bbs what would the segment length be?
that'd eb 50mm/1000
I get .05mm
so that is starting to get pretty good at that speed
but at 500 mm/s
not so good
at .5mm segment length
and at 100 blocks per second...
Got my machine all back together last night....
Have to do the test today....waiting for a more socially acceptable time to do cutting...
Did scribe lines on plexiglas before putting it back together....had 20 inch diagonals within .01 inches...best I could see by eye...
Should be about 3 times better than it was...at least...hoping for more....
this is servo with motenc?
Yep...but the problem was my assembly of the X/Y saddle...
Tightened up the double nuts while I had it appart too...X table seems to run allot quieter now...y is the same...
My hardware is still down for rewiring
so you might be able to shed light on ken's 100 block/sec limit
Possibly, I guess....
your sero priod?
.0005 as I recall?
I have my Traj Cycle Time = .001 , and the axis cycle times = .00025
If so, you should be able to get to 1000 blocks per second in theory
so yes you should be at 1000 bbs
700 MHz PIII here...
pretty decent numbers
motenc may be faster than stg
I can't go that fast
Well, should be, PCI ...
ever try .0001 servo?
The difference between my stepper machine running standard parallel and PCI parallel was about 3...
3 times that is...
(hand near big red button)
.0001 would be very good performance
Nope...not yet....might try .0005 and .000125 ...
it would disprove some theories about what the brick wall is in emc
After I do my squareness test later today, I ll give it a try...will have to double check all the following errors...
do try it if you can...it would help
with .0001 you should get block rates of 5000/sec
Yea, but the machine only moves at 240 in/min...not in 1000's like a HSM machine...
And the spindle is only a few thousand RPM...not in the 10K's RPM....
I know...but the issue is whether the program can run at that speed
Not that all machines need it
Those of us with HSM or moldmaking desires are dying out here
if the motenc has a dramatic performance improvement it would be good to know
Ok...well it has to be somewhat better, at least twice as much, due to the DMA in the bus....
But, that might be somewhat simplistic view...
well depending on latency vs just raw calculations in the thread
big thread since the tp is done in real time
The fact that I saw 3 times improvement in the Stepper system switching to PCI bodes well...
We are in mid January mode here...Gotta think this is going to be a -20 F year...
if you can run at .0001 in a 700 P3 that pretty much trashes the theory that rt/user comms are the bottleneck
oh...coldest I remember was 82 I think
-98 chill factor
and that was in evenston!!
I could go try the .0005 now...and set it back if it changes for the squareness...not too noisey, unless it oscillates...
you are at .00025 now though right?
Yea...I was out to the Wheaton Hamfest that sunday morning,,,got a ride from a guy from West Palm Beach...now that was scarry !
We were in his little station wagon, he had the radio turned way up, and each time the wind would gust he could not hear it, and the car would move a lane to the left...
it was deadly
car quits...you die
Turned down the radio...and pointed to the source of the vortex noise, and told him when you hear the noise, counter steer....
We made it ok...yea...it was -26 when we driven down new Rte 53 at 60 mph..and the wind was gusting to about 30 ....
It was a deadly situation....
I lived in north evanston at the time near the lake
we had just moved up from florida in 81
Off to try that experiment...back in a bit...
good time for me to eat breakfast
Welp...simple first impressions at TRAJ cycle time = .0005 and AXIS 0-1 cycle time =.000125 is that it works the same as before...Will run some following error tests next...
Seems to work...x has +/- .00015 following error...y has +/- .00045 (needs to be retuned since addition of tooling table wieght) ...
But otherwise no significant change...I ll go run the penquin and see if anything craps out...
now since you don't need the speed...you can turn that into decreased following error
BTW wb( what servo amp mode are you using?
running some tests here cradek
I think he is at 125 microsecond servo time in a p3700 with the motenc
is this emc2?
following up on ken's posts
I think so
if wb9 can run at 100 microseconds that pretty much debunks the theory that user/rt comms are the brick wall
should give 5000 block read rate
rather than ken's 100
This would indicate that isa latency is a big deal for emc
wb9 is running a motenc at 125 microseconds servo
in a p3
My 100 number was not intended to show anything other than the fact that the log has the time in it and that we can do at least 100. Also, that's on a slow machine.
I think your trajectory cycle time was just set to default .01
Its EMC 1...
No. -- That is NOT data. Yes. It is EMC 1.
I m running the verify now....if it clears, I ll let it run and it ll take about a half hour...
Have Traj Cycle Time at .0005, and Axis Cycle Times at .000125 ...
because logging perturbs ken?
wow. 8 khz servo rate.
should do 4 khz max block read
We never broke 400 microseconds with an stg
I ve started it...wont be back for a 1/2 hour (probably)...it started nicely...The machine is in another room...
ok...and the times are?
does SF work ok for you guys? the webpage I mean...
yes but it's a little slow
it's very slow over here :(
seems slow most of the time
Hi Alex, Morning Les.
* lerman realizes that he didn't acknowledge greetings from others.
pretty exciting stuff this morning
np = no problem
wb9 running at 8 khz servo
nm = never mind ;)
just noticed that I'd disconnected during the night
jepler: thanks for your advice about my machine - I did get it working
had to rebuild the kernel and rtlinux
cradek: You got it to boot on the Pentium machine?
too bad about your motherboard and/or CPU
yeah, it was fast
modern power supplies suck
the failure mode was strange
for 10 minutes or so my monitor's picture was wavy
I figured I would have to repair that monitor *again*
then the machine turned off
and apparently fried itself up
Ok...it dry ran the 3D_Chips problem with no apparent problems....so seems to work...
program that is...
that's 8 khz servo and 2 khz block read.
again: what are your amplifiers?
Galil....will have to look up the p/n's...DC brushed servos...
no tachs right?
ok. torque mode. that's fine
I think everything is running torque mode at the moment...Z is for sure...
you will want some ff1 and ff2 to take advantage of that servo rate
I have FF1 and FF2 tuned in...
Dave Engval and I have been back and forth on the tuning...
does sound like some serious HSM to me ;)
Now all I need is 60 K rpm spindle, hi...
les_w: I think you'll want to upgrade the STG ;)
wb9mjn: very nice work :)
or vanishingly small following error...can take your pick
stg -> ebay
60k spindle? 15 bucks at HF. Air powered
lives a few hours haha
Nope...no significant air flow, or noise budget for such...has to be electric...
So...from this experiment thanks to wb9 can we conclude that rt/user comms is not the over riding brick wall to servo and block rate?
we can conclude that in your case probably ISA was a bottleneck
It ll happen...just takes time....problem with 60 K RPM spindle is the tool has to be small <1/4 inch...so the reach down into the work is limited without a 4 and maybe a 5 th
not sure about rt/user comms, that might still appear on very short segments
although .. I still have hope
very very short in wb9's case
he should be able to do bblock rate at half the servo rate max
This computer was on the network during the test. and the only activity was the IRC traffic...
although that will pretty much turn off cubic sub interpolation
Not sure if the $10 buck router isolated that activity or not...
at some point you will reach a limit...but I'll bet it's below 100 microseconds
This is by far the fastest I have ever seen emc run
by a wide margin
We need to celebrate chinamill had a great success this morning(afternoon there).
Well, at least I m getting some things right, on this project, hi...
rayh: he did?
He was able to edit up a 4 axis HAL for his stepper driven mill
details are on a very crude page of the wiki.
I ought to try that too
cradek: edit a wiki page? :D
* rayh is looking for the url
it was always a shaky setup with emc1 because the units were so different
seems like the accels never were right or something
wb9: helped a bunch thanks. At some point you might want to find out where the hairy edge really is, then lengthen times 25% or so for safety.
[02:42:11] <rayh> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl/emcinfo.pl?SampleParport
We don't know what kinds of issues he will have with rotational speed yet.
but he got things turning.
wb9mjn: That is great news. Two things to celebrate.
this is a great leap of progress day huh ray?
I wonder how jmk is coming with the stepgen bug
and I just put a very crude HAL show script in the dropbox.
If you've got an emc2 running I'd appreciate comments.
Your welcom Les...
gives hope to us HSM/moldmaking types
alex_joni is now known as alex_joni_away
Hi Martin. What's the latest progress on the fpga
Hi Rayh, i still have much to learn
i will spend the next hours for that, hoping to get the DAC running
We've had a couple of success reports already today. I'll look for yours as well.
steves_logging is now known as steve_stallings
ah...Chris "chairman of the board" Radek
Hi Ray, see you tricked Chris. 8-)
We had a very good first meeting. Long though.
Mostly discussion. We'll be putting up a wiki page that describes actions.
We chose to affirm the major directions taken during the spring meet at NIST
You saw the "chairman" decision
yep, it is announced on the history page
We decided to get into emc(1) and produce a "final" stable release
and we chose to merge the lerman interpreter stuff with emc2 head.
lerman adds a "o or O" command for looping, subroutines, and conditional execution.
We are planning a first release of emc2 very soon.
candidate of course.
cradek: uh oh, looks like we'll have to figure out how to make those interp changes play nice with axis' line number mapping...
jepler: lots of work coming your way, things are starting to move :D
hey you guys
cradek: but actually as long as xemc still shows the right line of code, we're probably fine.
just thought i'd drop in before going to church - read ray's mail this morning, need to reply with a few assenting comments
cradek: (congrats on being simultaneously appointed chair and person)
thanks, I think
Matt, as former president will you continue to get secret service protection?
rayh: do you really think it's better to keep linuxcnc.org at Steve's place, rather than renting web space somewhere - I'm just worried about server security if we switch to a linux server...
I was chairman right until ray sent that mail, and then something happened to my gender that I don't quite understand
les_w: I just want my pension & health insurance ;)
guess so, sorry to say
I knew something seemed ... different
yes, your highness.
alex_joni_away is now known as alex_joni
mshaver: My intent was to say that it should stay at Steve's as long a Steve wishes.
My wishes are to keep my life simple. Current server arrangement is fine, but cannot support downloads. The bandwidth is 384K SDSL and frequently saturates due to downloads from Metalworking.com site.
Adding EMC images would not be workable.
rayh: Ah, OK. I was going to say that if you wanted, I'd pitch in for a server - How about Sourceforge? Do they host your web site as well?
SF offers hosting for every project
steve_stallings: That's what I'm trying to save you from ;)
we've never used it, but its there
but afaik SF won't host huge images, like CD images
Don't they limit storage? Not room for compiled images?
no, not CDs
jmkasunich: I thought so - Steve's connection is somewhat limited & we could stress it too much with .isos & stuff
for large images we could use bittorrent
very few folks can/will host CDs
I believe that there are several options out there for hosting of iso images
jepler: I say that every time this comes up but nobody is interested
Sure they will, just add money. 8-(
sorry, I came in in the middle, and thought you were talking about the normal web content rather than the isos
we could put isos at ibiblio - some distribs do that
$10/GB/mo seems like a "good" price for web hosting, but that gets expensive fast
tbl's site was hosting a lot in the old days.
(of storage, not bandwidth)
cradek, jelper: correct me if I'm wrong,but doesn't bittorrent only buy you something when people are doing simultaneous downloads?
jmkasunich: those of us with a little bandwidth to contribute could leave our clients running, and the downloaders would pull from all of us
jmkasunich: but you're right, it would be especially useful right after a new release comes out and many people want it at the same time
We should really consider ISO image downloads and regular web presence as two different issues.
that leaves non-peak hosting, which is a little easier to handle
Gotta go for a bit, back around 1:30EST
don't hold your breath about new BDI releases
mshaver is now known as mshaver_away
agreed with steve - two different issues
At this point, is there anyone who plans to put together new iso files (installers or live CDs)?
as far as emc releases, sf can handle those easily
not that we know of
my mirror can handle lots of traffic
jepler: not that we know of
dunno if anyone is doing more with the puppy linux thing
no limit that I know of, about 100MBps
matt was ;)
There was interest in "puppy" but it has not taken root.
alex: your mirror doesn't have any monthly limit? you don't want to be surprized with a $200 internet bill....
I've run both the emc puppy and regular and it is a nice compact system.
there are a few things that I don't like but hey...
jmkasunich: it's alocal university, and no there is no limit
so far I have a few thousands downloads on BDI's
The community has limited resources. The current BDI seems to work well. It is widely available. We should concentrate on EMC stabilization and web presence enhancement.
i'm with steve_stallings on the directions to focus.
I've found emc2 to run well under 4-30.
my only caution is don't try to update the real time from what is on the cdrom.
* cradek would never update realtime on a working system
Question: BDI 4.30 is latest production Linux kernel? Real time system is current, but changes loom?
it's a 18.104.22.168, pretty recent
you can't have the LATEST for more than one month :D
OK, sort of latest, 2.6.x.x
And we are past all the build system issues with the new kernel?
Wait. There are at least two different 2.6.12's out there and they are not compatable.
and the apt stuff allows you to get them wrong.
One is compiled with the older gcc and one the current.
If we are reasonably current, put a stake in the sand and say this is where stand, these are the issues, and we will be staying here while we work on other parts of the system.
steve_stallings: I'll say emc2 builds OK on any 2.6.x
given that RT stuff is patched ok
with the disclaimer, alex_joni, that the kernel and all were compiled on the same system throughout.
Fenn, are you awake? Want to chime in on web presence issues?
what is the latest kernel rtai supports? 2.6.10?
but we don't
with increasing risk of bugs as you use the newest stuff
rayh: fusion is something else entirely
it is a flavor of RTAI, has nothing to do with the kernel
I know but fusion will build with the latest kernel.
It has everything to do with the kernel we use.
and in fact there was recently (maybe a couple months now) a split in the RTAI communith, and many of the former Fusion folks are now working on Xenomai)
and paul has a fusion patched kernel out there.
ok, but to say that fusion is the latest and we don't work with it is misleading
fusion is a branch of RTAI, going in a different direction from main RTAI
I've got one box here that I am not able to get back to developer status because of kernel issues.
we do work with the latest mainstream RTAI
magma it's called
it's actually a developing branch
not a release version
the latest kernel paul did was 22.214.171.124-magma
am a bit confused by naming convention on rtai webpage, is rtai 3.2 vulcano or rtai 3.2 test 2 (magma) more recent?
and a test kernel 2.6.13-fusion (I think it was 13, nto very sure)
delacroix: you don't want to go there ;)
there are so many I lost count, vulcano, vesuvio, magma, etx
RTAI naming is very confusing
and they are always walking on the bleeding edge
which is why I use a BDI kernel
i only ask because magma only seems to have patches up to 126.96.36.199, and vulcano has patches for 2.6.10
the latest BDI kernel is kinda bleeding edge
the patches are not in the place where you would expect them
the latest patches are on Adeos (different project now I think)
cvs.gna.org or something like that
I lost interest in that :( .. too hard to follow
OK, OK, where was that stake in the sand?
[03:56:52] <delacroix> http://download.gna.org/adeos/patches/v2.6/adeos/i386/
that's the one
great, now im slightly less confused
delacroix: glad YOU are ;)
i said slightly :)
even slightly is ok ;)
am trying to set rtai up on suse oss 10, and wasnt looking forward to trying to roll back the kernel, looks like i dont need to now
delacroix: how so?
well if i can get and adeos patch for the kernel version suse10 uses, then cant i just apply the patch to the kernel sources that come with it?
ok, make that slightly more confused then :)
it's always a bad thing to use a distro kernel source
use a stock kernel from kernel.org
I used to do rt patches to mandrake and it was a pain.
note - the patches *may* work, but there's no guarantee. you'd have to look carefully at the output from patch, and use a stock kernel if any hunks didn't work
in the best case the patching will fail, worst case it won't
but something might break lateron :(
steve_stallings is now known as steves_logging
* chinamill is back
Jacky^afk is now known as Jacky^
[05:42:37] <Jacky^> http://www.repubblica.it/2003/e/gallerie/scienzaetecnologia/scooterbuca/ansa72004780412162142_big.jpg
[05:42:49] <Jacky^> http://www.repubblica.it/2005/l/motori/dicembre05/scooterpiaggio4/scooterpiaggio4.html
good morning all :-)
hi alex, did you get the Heid code?
mshaver_away is now known as mshaver
Jacky^: I translated the text of the Piaggio article to English. The first sentence starts, "After thousand indiscretions..." - I am laughing so hard I can barely type! :)
Piaggio is going to become a zombie ..
but the latest seem a nice idea
lets see ..
* mshaver started riding on a 250cc Benelli, I pushed it farther than I rode it, but I sure like the looks of the machines made by Aprilia!
many things are happening
I want this: http://www.aprilia.com/portale/eng/magnet.phtml
ducati seems to came back from usa to italy (partially)
piaggio-aprilia are fused togheter
scooters are designed here and assempled in India (low cost..)
its hard to find a very good product today
hmmm - they're missing an important part of the steering mechanism there ;)
piaggio designs in Italy & bulds in India? like the bullet?
oh wait - it's a trike
for the low cost humane resource
* mshaver ...is talking to dave engvall on the phone...
oh nice, heard he and wb9mjn tuned a nice emc machine
with 8kHz servo rate
ugh. worst CNC article ever in this month's Nuts & Volts
the guy runs his steppers in unipolar mode, L/R drive, with a Basic Stamp program that is so primitive it can't even perform fewer than 4 steps at a time (since that makes for complicated bookkeeping)
guess: stepper, open loop, nonrealtime, basic interpreter
oh, it's probably got very dependable timing, since it's all running on a single-tasking microcontroller
only a few things worse than a pic, but the basic stamp is one of them
8051 is great
OUTA = %0010 / PAUSE SPD / OUTA = %1000 / PAUSE SPD / OUTA = %0001 / PAUSE SPD / OUTA = %0100 / PAUSE SPD / RETURN
but at least the 8051 knows what a carry bit is
you got C for that, but basic? cmon
our RS274ngc interp is a choke on the emc... its the only way we have to describe a path now, but the NML wouldnt mind another
inside we have line circle/arc and nurb... has anyone looked at the nurb code inside?
the gcode wont touch nurbs, so no one's played yet
hmm. i told Dave E about it... lemme go look for the hook
canon.hh NURB FUNCTIONS i guess the java-tool makes cc outta the hh
NURB KNOW VECTOR NURB FEED...
* alex_joni looks
I'll have to disappoint you...
just grepped the whole source, no place else but there..
seems some definitions were made, but that's it, no functions to go with that
oh yeah i see 'extern'
how to ask Fred?
they are there since 2000 at least..
ok, something to do in my spare, thanks
so its ok to declare the function external if you dont call it :-)
not very nice..
linking might give a warning
but it's not called, so no serious stuff
lint was a tool to find unused stuff, and unused stuff is confusing in largish projects, but yes, benign
nurbs on EMC???
time and direction / unit vector
look at canon.hh, seems to be dead end code tho ( declared extern but nobody home )
lots of heat and light this morning
tomp: how's your translation coming?
fenn, my neice didnt wanna work on it, so I began last nite
dunno.. both spelling look weird
i before e except after c
I beagn with babel fish & your stuff :-)
hard, its hard
actual translation software is 30% intelligent
tha same word can mean 5-6 different things
how it go to associate that to the phrase
I know, but its harder to find a Mandarin machinist to critique it :-)
software must be more intelligent, to examine entire phrases
but it required a lots of algoritms and cpu use, i think ..
tomp: you don't know any mandarin machinists??
I do, but thats another problem: $$$ gottta go visit!
the mean of a single word is important !
think here they changed a word in a law and changed all
1 word ! damn
jacky what are you talking about?
and now we have a law, that if a guys, teen, 14 years old, download an mp3 and listen it, got 20k euro to pay and 4 years in jail
just changig a single word
reflect his mean to 360 �
so, the world is moving around the mean of the words
there was a (c) law before.. right too
but they changed 1 word ]:)
i dont believe in slavery to language.. that is one reason why i am an anarchist
s lucre with profit
is not the same thing
profit mean, if you donwload an mp3 and you heard that
you do not buy a disc
then youve a profit..
because you saved on the price of a cd?
cause du damaged the autor of song
before was different
if you was copyng and selling multiples copies
thats why thay changed a word in a law
soon they will change it so that you will be penalized if you didn't buy or download anything at all
because you must have been up to something
I like an italian interview released from a music band some week ago
there is precedent.. a utility company charged a man for not using any electricity after he ran a generator
they want to directly sells theyr music in internet without pass through any major
here a disc price is around 30 Euro
when the starting price is about 6-7
why 30 ? just because it pass trough a lot of hands
and anyone want to lucre up
but is going to finish this story ! damn
just tryng to use terror on peoples
anytime go to a cinema, before the film, start a spot that seem want to eat you
* fenn gets back to work, maybe
fenn: did you try the Lerman interpreter or Jeff Epler's ellipse gener8r?
I would like to try the lerman interpreter - have not yet. You would not happen to have the exact dirction on how do it would you?
I would like subs and conditional statements
you gotta use cvs to get the lerman fork of emc2, i was walked thru it by Ray... you got emc2 running?
sounds like all I have to do is issue cvs update -r lerman-interp command and then recompile
yes - goofing around with it - right now I have axis running
then try the examples,and look at Jeff's ellipse code from cad-cam-dro-emc group msgs
sounds neet ;)- (I have it hooked to a machine with stepper) t
i posted an ellipse prog ( hack of Jeffs ).. this comes from Dave E's thread on different paths ( not G1 G2 G3 )
I remeber reading about a while ago - there was some standards for being able to input funcions into a gcode progra
I think Ken Lerman put a lot of effort into giving us this stuff & it makes for much smaller code & nice macros
I think Jeff's macro generators are a good idea toio, an interpreted Pythin script
I am going to have to hook the emc2 machine back up to the network and update then try the new interp.
i would like to be able to call a python script from g-code with some arguments.. would m10x be able to do that?
I think all this makes Jon & Ray's cp1` program stronger
is there any documentation on the new interpreter? Like new commands and such ? THe wiki page touches on it but that is about it.
no there isn't any documentation
I called bash, but python will work too
well the 'docs' are the 3 examples
feel free to write some :)
did! Jeff & Ken wanna get some feedback
I posted to cad-cam-edm-dro I can copy to you Fenn, it was inline code in post to Jeff about Dave E's thread
well, just a link would be fine
oh that was emc-developers
that would be great
different mailing list
i thought you were talking about documentation for the new interp
we where - I guess all there is a thread on the cad-cam-edm-dro yahoo site
the docs are the examples in lermans src code, in his 'nc-prog' (sp?) folder, shows vars, loops conditionals
skunkworks: i think tomp is talking about this thread http://sourceforge.net/mailarchive/forum.php?thread_id=9137238&forum_id=6435
shouldn't I be able to get those file by navigating though sourceforge? I tried it a few times and got lost (cvs)
yeah, you have to select the lerman-interp branch from the pull-down in order to get files from that branch
no idea which files were modified though..
not sf, the cad-cam-edm-dro group! if it's echoed elsewhere , well.. ok, going to find it now...
sorry.. (i avoid yahoo groups like the plague)
search more dammit!
ok, me wrong,.... firstname.lastname@example.org, now what was the topic??
[Emc-users] Need a thread cutting boring bar for mini-lathe ( gawd, people never chg the title on these idea changes )
the topic was [Emc-users] Need a thread cutting boring bar for mini-lathe
here are the nc files - http://cvs.sourceforge.net/viewcvs.py/emc/emc2/nc_files/?only_with_tag=lerman-interp
i got the ellipse python external gcode gen., hacks to cp1 for heidenhain & an M104 that turns on external device & wiat for it... where should I post & how?
well.. i think CP1 needs to be in some sort of "utils" folder in the cvs.. but others disagree for some reason
what sort of hacks to cp1?
I had Heidenhains with dialogs limited to exactly 13 questions, needed 14, no way to link, used CP1 to put a paper bag over the Heidenhain.
output is not gcode, is Heidenhain code
geez no idea on that one.. is there a heidenhain users group?
it's an example of how to collect data and xlat it into machine tool cmds, not necc for Heid
i would like to collect all these little cam scripts floating around and put them in cvs somewhere
I gotta go, but the idea of not-cam, but tiny scripts is good. Jeff used Python, Jon&Ray used TclTk, hell I used Hypercard years ago!
tomp: do i have your permission to mark ellipse.py as GPL?
(need to get jeff's permission too)
and java/ecmascript more recently
fine fer me, but it's ElPokt.py pronounced El Pockety Pie :-)
SWPadnos_ is now known as SWP_Away
how's it going?
(try that again) Hrmmm...
looking for JMK.
he's been around...
Is anyone going to do a deb for emc2 ?
not that I'm aware of
do you mean eventually, or soon?
distribution methods are on the topic list for the board
for now it is source only (right now it is CVS source only, but tarballs are hoped to be in the near future)
any plans for a sane install & config file ?
depends on the definition of sane
alex is working on make install
03cradek * 10emc2/src/emc/kinematics/ (tc.c tp.c): I don't think anyone was using this debug and it fills up system logs.
jmk - Hello. I have not gotten to being at the machine and having irc contact. You had said I would need to edit hal files also - Is it something you could explain here - or would it be just too hard to do on my own?
If you are to continue down the pin/signal path for linking all those ****** modules together, you need to look at XML.
skunkworks: I dunno, right now I'm buried deep in something else
that is ok - I am not in a rush.
did you see these? the gantry is what emc2 is running - the k&t is the one that we would like to convert. http://www.electronicsam.com/images/KandT/
Anyone have any tricks to regain position between tool changes when the controller crashes?
Jymmm: You hack the gcode to remove everything up to just before the crash, reset/home the machine and hit go.
the usual, chasing bugs thru the forest that is emc
gave up on that for today, working on some new stuff
(new bugs maybe)
im too busy cutting metal to worry about emc anymore ... but I stop by from time to time to say hello
K`zan: This was between code/bit changes =(
hi, im trying to follow the rtai install instructions in the wikki, but cant seem to get mkinitrd to work, should this be "mkinitrd /boot/initrd-2.6.10.img 2.6.10" as i keep getting an error
Jymmm: The couple CNC machines I used, dealt with tool changes automatically, so I don't know anything about that aspect.
Jymmm: Assuming non-automatic tool change, why not just put a pause in the code at the point you want to change a tool?
* K`zan running around going crazy too.
delacroix is there a particular reason why you aren't using BDI stuff?
(i wrote those instructions btw)
and what's the error you're getting?
SWP_Away is now known as SWPadnos_
RED SALMON CAVIAR!5 oz!
how much are 5 oz in kg ?
ebay is nicee :P
Jacky^ is now known as Jacky^afk
Some new goodies in http://wrlabs.shacknet.nu/~vw/Tools/
0909,01do the STEP_INDEX and DIR_INDEX lines in the ini file controll wich pins send the setp and dir, or do they control something else?
are u using EMC2?
yes, in EMC1 the INI file is it
it's much more hard coded than EMC2
you can configure some of the IO, but the stuff that goes through motion
like limit, home, amp enables/faults is hard coded
0909,01to be honest all i ever use is the step/dir on the motors anyway
0909,01i never get around to wiring up and home/limit switches
I don't think you can configure step and direction in the ini file with emc1
I haven't seen any settings named STEP_INDEX or DIR_INDEX
it's true there are no such config options
0909,01on the bdi install i used it gave me two ini files by default, one named generic.ini and one named emc.ini. while the two files are pretty much identical, under the axes section in the emc.ini file there were STEP_INDEX and DIR_INDEX lines with number that happen to be the same as the pins that control the step/dir while there were no such lines in the generic.ini file (the one that was actually loaded)
This may have been something that Paul added on the EMC1+ branch for the BDIs
try it and see if it works
ok - it looks like there's some test code that looks for the direction and step index, but it doesn't look like the parameters are actually used
0909,01my mistake, the lines weren't the pin numbers... example the x axis had:
DIR_INDEX - 0
STEP_INDEX - 1
while Y was dir 2 step 3, z was 4 and 5 and so on
0909,01yeah, swapping the 0 and 1 didn't seem to do anything
yep - I looked at the emc.ini file, and saw that
looking at the code, I see where the index isloaded and checked, but all the function does is print the setting - it doesn't actually change the code
(or the pins the code uses)
0909,01ah, so i guess me hacking up thst parallel cable to swap those pins wasn't a waste of time after all ... that kinda makes me feel better
SWPadnos: this fortal is really nice stuff.. glad i bought two boxes of it
cool - glad you like it
it is a bummer that it isn't availabe any more
SWPadnos, votes come in yet?
03paul_c * 10emc2-auto/wiki/ (17 files in 11 dirs): "Auto update wiki from a cron job. Mon Dec 5 05:30:01 GMT 2005 "
Hello, any emc2 users awake?
:) I have some problem with the axis, I jog one way, and when I stop, the axis go a little the opposit way. Do You know anything about it? (i'm running a week old emc2 from vcs)
sounds like some backlash setting
but i'm not sure... :/
I think backlash compensation is done before the ordered movement
Would You know anything about inverting a output parprot signal in hal?
yeah, there is a parameter or pin named something like parport.0.pin-5-inverted
Think I saw something about going backwards after a jog
Ah... Thanks... I'll try that later
skunkworks was seeing that.
halcmd setp parport.0.pin-01-out-invert TRUE # for example
jtr: saw where, bugreport?
here - might have been a bug report as well
I guess I should try updateing to latest CVS
I think they figured it out - let me check the logs
trying to figure out where the logs were...
ok, maybe I should try to get the mill online first and update CVS
* chinamill is away: Will be running between the computer and the mill
My axis is still in the mode of I jog one way, and when I stop, the axis go a little the opposit way. (with latest cvs update). Anyone got a clue?
* chinamill is away: eating
* chinamill is away: will be checking on and off
Does anyone know good values for a metric style ini for FERROR and MIN_FERROR (for emc2)?