I ended up using another mail server to get round some of the email probs I was having.
some interesting reading
[00:04:53] <fenn> http://www.evworld.com/archives/reports/lynxmotion.html
the only bad thing about sourceforge being a free service is that you can't call them up and bitch at them
[00:04:59] <fenn> http://lynxmotiontechnology.com/e813.htm
does jmkasunich want the phone number ?
174hp continuous rated motor in a 20" dia 6" thick package
for SF? they go out of their way to force you to do everything thru "support requests"
but if you have a number, I'll call it
corporate HQ for the parent group.
(google reverse search on the number came up with VA)
that number is described as a support number for "Sourceforge Enterprise Edition"
8/5, so nobody will be there (might call anyway for grins)
or get Jymmm to go over there....
phone msg: "Please note, this is not support for sourceforge.net"
fenn: Those are the ones that are for the diesel-electric city busses.
Phydbleep, is it 3 phase or what? I couldn't figure that out from the info provided
this is the electric bus program in brazil you're talking about?
somewhere in the EU I think.
wanna do an electric motorcycle ala "AKIRA"
ceramic tesla turbine genset powering two in-wheel motors
fenn: use pmdc motors with the magnets in the rim of the wheel.
at least that was my take on "Ceramic double-rotor two-wheel disk drive" as in the movie
Yeah, sounds about right. :)
wanna get the script in japanese to see what it really said
not enough days in the year
* Phydbleep has the DVD. :)
the new one or the old one?
Pioneer TXH remaster.
* fenn spent way too long trying to download it off the net.
SWP_Away is now known as SWPadnos
paul: you around?
I don't see Paul in the user list
oh he left while I wasn't looking
he does that sometimes :)
trying to hunt down a mail problem... wanted to see if he could check the headers on some list admin messages for me
my ISP and sourceforge are having a pissy fit at each other
would you like a GMail account for this type of problem?
I have a yahoo account, just subsribed it to the lists
and have been using it to test things
your isp thinks sourceforge is spam?
either that, or the other way around, or something even more complex
well - if they run exchange server, it would make sense ;)
probly isp thinks its spam, bounces a few messages, sourceforge deactivates account due to bouncing
my users.sourceforce.net alias (which normally forwards to my ISP addy) now reports that my isp account is boucing
i had that problem with yahoo groups once
meanwhile, messages sent to the lists (which don't have anything to do with my SF account) are being bounced
sent from yahoo?
no sent from my regular isp
from yahoo to the lists works OK
from yahoo to my regular addy works
from yahoo to my SF alias doesnt
and from my regular addr to the lists doesnt
from the lists to my regular addy work
this is funny - I just signed into GMail for the first time in a few months, and I now have 50 invitations available to me
it used to be hard to get just 6 invites
what's the big deal with gmail
why not just get a real email account?
free, fast, 2.2G storage, google search tools are used for searching it, and the interface is pretty good
can you do pop3?
forwarding, that is :)
I have regular email as well, but the GMail account can be used for IMAP-like access from the web
there are also other tools for accessing GMAil like a regular MUA
dirty script hacks, you mean
well - that could be :)
I understand they do offer pop3 access now
i think i need to get a C++ book from the library.. too bad tomorrow's sunday
cradek, that's pretty amazing
if people weren't so stupid and webmail-crazed, i'd say they made a mistake
It's basically a somewhat better (IMO) version of Yahoo mail
but only 0.5% will ever use it
people definitely do like webmail. and web-bbses. I have no idea why.
and no crap "portal" to get to the mailbox
webmail has a single valid use for me
in addition to the not-too-important "consistent address"
I use webmail for 90% of my mailing (It's provided by my ISP, not yahoo or gmail)
it's accessible anywhere, with the same folder layout
so is my mutt over ssh
it's basically got the same functionality as IMAP
I can use it from any of multiple linux boxes, from the doze box at work, from a hotel, from my wife's doze box, etc
and I'll have the same email address for as long as I want to pay to keep the domain name
most windoze boxes don't have ssh (can you believe it?)
then google for putty and click on the link
sure - mutt over ssh is great - if you know the IP address of your home machine
or if you can get dyndns or an equivalent to work
I have a static IP address of course.
how else would I host my web pages?
installing putty on the work computer would get me in trouble (IT folks)
and there are firewalls and proxies and all that
jmkasunich: you can run it directly from the web
webmail just works
that helps a lot - I do now, but haven't looked into mail stuff yet (I haven't set up my server yet)
i've got a free unix account.. freeshell.org no dyndns or anything fancy
well, technically they charge $1
I fire up a "real" email client every week or so to empty out the webmail inbox, but just about everything else is done with webmail, even when I'm at my main linux machine
jmkasunich: that does sound like a simple solution if you can tolerate the interface (and editor)
err, s/can tolerate/like/
didn't mean that to sound so negative
the big trouble I've had with webmail is the lack of an outbox, though that just changed at my ISP
if emc's gui crashes, what happens?
does it go into estop or just keep on cranking
generally, emc exits
I bet it varies
because when the gui exits, the run script continues on to clean everything up, as if you exited on purpose
I'm pretty sure that Jon Elson or somebody mentioned that the GUI had croaked one time, and the machine kept on going
depends on the nature of the gui croak
SWPadnos: if it hangs, that might happen, but if it exits, emc will exit.
if the process ends, the script will shut everything down
actually, I think the shell had crashed or some such
but if the gui hangs...
right - if the script is still running :)
yeah what he said
as I said, I think it varies
okay, so another hypothetical situation, what happens if the gui hangs on a remote machine?
the mill keeps milling
(in theory you can have several guis, and can start and stop them independently of the main
technically you should be able to fire up another GUI (on the same machine or another), and keep on watching the progress
of the main emc program ("task")
I've actually had several GUIs on my machine, attached to EMC running on a machine that Paul had, while he also had a GUI on his machine
I think the NIST folks may have done that - one PC in front and another behind a large machine, both running GUIs and both able to control the machine at the same time
saves walking aroudn the machine
the only thing that doesn't work wel lin that instance is that the G-code following doesn't work, unless the file name and directory structure are identical on the two machines
we did it at Fest
(so it was at NIST :) )
the gcode doesn't go straight into task?
task is the gcode->nml interpreter
the display dhows the text from the file, not the "actual" code executing
so if the file isn't in the exact same place, there's no display of the code
I don't see any good way around that either, with the idea of Gigabyte G-code files being thrown around
well, you could do it incrementally
jmkasunich: can one build and run (sim) emc2 without realtime yet?
the GUIS have no knowledge of each other
i mean, task sends a chunk of what it's doing right now to the gui
no - task has a status buffer that shows its state
(and a command buffer)
cradek: no, and it will probably be a long time
jmkasunich: that's a shame
several GUIS attach to those buffers, and can issue commands and look at status
but they can't talk to each other
jmkasunich: that's how I do all my debugging
well, could you implement an input buffer in the same way, and have the gui look at that
attempting to compile the motion and other realtime code in user space gets very ugly
jmkasunich: how do you debug the rt stuff?
jmkasunich: you can't use the debugger, can you?
dynamic linking, I'd bet
breakpoints in RT code are usually not a good thing
I'm used to doing realtime/embedded so I don't miss the debugger (never really learned how to use gdb, even in user space)
fenn: it's possible (though probably not recommended) to put the text of the current line of G-code into the command path, and just pop it back into the status path, but it's a lot of work for not too much gain (I think)
I use halscope, printk, etc to figure out what is going on
jmkasunich: I'd feel pretty naked without gdb...
I haven't used that kind of debugger much
not since dos debug
oh, whole different world.
I found that velocity bug without changing a single line of code or compiling. I just poked at it with the debugger.
different fools, different tools ;-)
at one time I wanted to try to make a version of RTAPI that would allow you to buiid systems that could run without a RTOS
Hey! I resemble that remark!
but it got ugly, and I didn't see enough value to support the ugliness
folks still using printk are more likely to be the fools
well in emc2, that may be the only available tool.
for the realtime stuff anyway...
no reason you can't use gdb in userspace
is there a possibility of using kgdb on the RT stuff?
what is kgdb?
SWPadnos: is that the scheme with gdb over the serial port?
That's one way, I think
jmkasunich: I think the theory is that you hook to it with another machine, so setting a breakpoint doesn't disable your interface to the debugger
remember, hitting a breakpoint in RT code is unwise, especially on a servo machine where it can cause a runaway
I know that you can use it on a serial port on embedded systems, but I don't know if there are better tools for debugging on the target the system
jmkasunich: sure, you don't do it with the machine turned on!
I've found halscope to be a usefull tool for working on the RT stuff (but then I'm a double-E, so I naturally reach for a scope first)
cradek : Awwwwww, you're no fun =(
jmkasunich : Damn, you're able to find a bra that size?!
jymmm has a filthy mind
I imagine halscope is good for finding out when output values are wcrewy, but finding out *why* they're screwed up is a different story
double E = EE = Electrical Engineer...
jmkasunich : Hey, you said Double-E, ***NOTHING*** wrong with that at all.
I just add some "test points" inside the module in question, and probe them
there is no 'geek' without 'EE'
G (.) (.) K ?
I assumed I was among fellow geeks, who would understand my meaning...
sadly, Jymmm has proven himself to be at least somewhat normal
jmkasunich: the other geek here with a scope at least as good as yours understood you fine.
quit talking abot Les behind his back =)
cradek (unless you have a scope too)
they was talkin about me, jymm! just got my tektronix 502A yesterday
I wish I could upgrade mine - the new HP MSOs are really fantastic
fenn ah, cool.
* jmkasunich has two
* cradek is not sure how many he has
Tek 2465 analog, and Tek 2440 digital
i don't know if it's even worth it to see if it works.. when am i gonna use a scope?
* cradek has only one that works right and is any good...
cradek B&K =)
I have a thing for vintage test equipment so I have some old scopes.
is a sound-card "virtual scope" good for most everything, or is it rather limited for everyday applications?
I guess my 100 MHz 200 MSa/s 2 analog + 16 digital channel with 2Mpoints depth is a little more modern than average, huh? :)
Agilent 54622D - great scope
definitely more modern than the average home scope
SWPadnos : s/100MHz/1GHz/ and we'll talk =)
tho I use significantly better ones at work
yeah - I may trade up to the new 6000 series (I think)
they have 4+16 channels, color screens, and higher memory depths available
2GS/s, 500MHz, 100K samples, 4 chan
the screen updates are so amazingly fast, it's unbelievable
I have more use for a spectrum analyzer than a scope.
the Tek TDS scopes are really sucky in that area
turn the attenuator knob three clicks, and it takes a couple seconds to get it right
A co-worker bought our old DSA-602 for $1
I told him to put a piece of glass on top and call it a coffee table (thing is huge)
Jymmm, run the signal into the sound port and use ffitw or equivalent software
the sample depth is the kicker for me - I love being able to zoom in by 9-10 clicks on the timebase knob (3 orders of magnitude)
fenn (for antenna testing)
yeah, that can be handy... not too fond of the way Tek does zooming on the TDS scopes tho
fenn nice thought though
if you get a chance to see (or get) a demo of the new HP, do it. they're amazing
sorry - Agilent
it's odd, we seem to be a 100% Tek shop for some reason... I don't think anybody set out to do that it just happened
SWPadnos nah, it's HP!
yeah - says so on my 28S and 28C calculators :)
and on my 20s too
Ok, why does the picture of this scare me? http://www.madelltech.com/m1-7.html
it's funny that a calculator I bought nearly 20 years ago can do things that the most modern spreadsheet can't
SWPadnos What's worse, is that HP doesn't make any GOOD calcultors anymore.
nor does Agilent :)
they only have 3 calculators now, all look like shit too.
here's a scope for you Jymmm: http://www.home.agilent.com/USeng/nav/-536897734.536894886/pd.html
trigger jitter less than 500 fs rms - not bad
for $100K it BETTER come wiht a $10K calibration ref too!
my main need in a scope (or probing system really) is common mode rejection
the kicker for me is often lack of isolation
it's funny that for work on high voltage systems (or non-common-ground), a tekscope is still the best tool
I need to measure a +/-15V gate signal (at 5V/div) that is sitting on top of a 1000V, 20KV/us common mode signal
tekscope to the rescue - isolated ground, even probe to probe (though it may not be 1000V)
I think it is - maybe 2500V isolation
what is a tekscope? (I assume you don't mean just any scope from Tek)
nope - they have a series of portable DSOs called TekScopes
SWPadnos This isgetting warmer --> http://www.home.agilent.com/USeng/nav/-12093.0/pc.html
oh, kinda like the fluke scopemeters?
yes but better - more like a real scope, but similar
There is a specialty RF company out there, I wish I could remember theri name. All their gear has a med blue paint on it.
(and way better than the HP logicDart)
used a scopemeter once, retch
the scopemeter is a glorified multimeter - the texscope is a slightly pared-down scope
we had a demo of a new tek scope with isolated inputs, the TDS(TDA?) 2024
at about 5KV/us, the isolator got upset, generated off screen artifacts that lasted hundreds of nS
completely unacceptable (and we told the sales guy that in no uncertain terms)
we usually use P5200 differential probes
when we need really clean measurements of high side gate voltage, we use a 3034 battery powered scope and float it (carefully)
OK - it looks like the tekscopes only have around 600V isolation - oh well
[02:03:56] <SWPadnos> http://www.tek.com/site/ps/0,,3M-10566-INTRO_EN,00.html
any of you have a porter cable router?
probably safer than floating the 3034, but basically a single channel scope in our application... it would need to be floated to work
Ryobi here, sorry
jymmm why don't you just gear up a standard ac induction motor?
SWPadnos 1/2" and 1/4" collets on it?
1/4 only, unfortunately
I may get P-C one later
fenn: gears or belts to go to 20K rpm are noisy, inefficient, and expensive
Jymmm, sorry i was assuming you had a lathe, that may not be true
(or at least two of the three)
well, there's always airmotors
I just bought a craftsman 2hp with 1/4", but I'm thinking it might not get the depth I'll need, and if I get a PC one that has a 1/2" collet I get make/buy and entension (2-3") to use 1/4" bits in, or if that's a bad iadea at 25K rpm
probably a bad idea, but you'd have to look at the maximum bit length they recommend
the bearings should wear faster the longer the bit (assuming similar cutting force)
I wouldn't make an extension for 1/4" collet, but thinking the 1/2" might be "ok".
OK - it looks like kdb is a better tool for debugging the kernel/RT side of things
[02:10:42] <SWPadnos> http://oss.sgi.com/projects/kdb/
fenn np =) I'm making a gantry router, and trying to learn from others regrets/mistakes =)
don't know how recent it is
OK - looks up to date - there are patches for 2.6.11
Jymmm, I was thinking somethg like this http://www.buildyouridea.com/cnc/pcb_drill/pcb_drill.html
well - I'm off for the night - I should be around for the dev meeting tomorrow though (yay)
see ya all later
SWPadnos is now known as SWP_Away
well, just hit three postmaster addys and two abuse addys with a long eamil about my problems... they'll probaby kill me now ;-)
jmk or anyone who knows: Are there some more current instructions needed to compile emc2?
I thought I read that somewhere in the channel backlog..
what system do you want to compile on?
you have realtime installed?
RTAI I mean?
I meant the BDI 4.20
oh, good ;-)
[02:49:32] <jmkasunich> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?BDI-4_Install
Yeah, I've read and done that
I thought there were some extra steps that you told weyland
depends on what you are trying to do
I was just going to check out the latest CVS of emc2 and try it out
the instructions on that page get you ready to compile any version, and the tail end of the page actually does compile the bdi-4 branch
I had weyland compile a tagged version of the head of emc2 instead of bdi-4
checking out the head should work
heh apt-get install kdevelop: After unpacking 275MB of additional disk space will be used.
nice lightweight package
the package is fine, it's the 35 other packages it wants to download that are the problem
i wish there was a graphical tool that would show you RPM dependency trees
can't be done... impossible to display them on a 2D screen, need 4D at least
huh? just draw it as a tree
do it all the time with phylogeny software
just being a smartass... too many packages depend on to many other packages...
* jmkasunich doesn't like dependencies
actually i find it's usually ONE package that needs five hundred others
and that turns out to be something dumb, like a music jukebox
I've tried RPM's, and don't wanna go back
vs what? debs? .tgz?
I can cope with .tgz, but I find debs to be much more agreeable
I haven't tried everything - not enough hours in a week, or days in the year - but I'm really liking Debian
* Phydbleep wanders back and notices that asdfqwega is here.
well it's time for this early bird to go to bed
it's been a long hard day of typing for me
* Phydbleep has to go rebuild the vibratory cleaner.
* mshaver is back
* mshaver is back
* mshaver is away: mshaver
* asdfqwega read www.buildyouridea.com - Damn, now I want to make a 18" x 24" machine
A-L-P-H-A: :D Did you get your spindle fixed/done?
I fucked up a line of code in my g-code while making, and made the hole too large... but like 0.010" :(
I'll work on it in the morning.
A-L-P-H-A you have a PC router by chance?
no. I looked at those... they seem really nice. So do mikitas.
with the soft start.
the 1.5hp ones?
yup, with variable speeds.
* Jymmm looks at makita.com
I wonder the price http://makita.com/Tools_Item_View.asp?id=776
PC and Mikita's are like around $300 CDN.
they ain't the cheapest things, but they are good.
PC ~$130 USD
hey! not bad then.
just look at the duty cycle of them.
make sure they are 100% duty...
otherwise, you'll have burnt out coils.
or worn bearings.
I bought a craftsman 2hp (peak) router, but it's only 1/4" collet
I can still return it though
I think I need to ask Les, but what I'm think is that I won't have enough clearance to cut deep, so I want to get a 1/2" router buy/make a 2-3" entension and use 1/4" bits on it.
ER-11's are good for that.
I've got a DA300 I'll let go cheap.
types of collets.
and they'll hold 1/4" router bits?
DA300, with1/8", 1/4" and three other collets.
and didn't you just get that?
no, I just got the ER-11.
I remember you having someone look up the bevel on DA300 for you
that was prior to mine.
mine current on.
if not, I'll toss on it on eBay.
ah, they why are ytou getting rid of it, ?
don't need it.
I guess I could keep it for an extensio.
and it'll fit into a 1/2" router collet?
STUPID kb is missing key strokes.
oh, WAY too long.
you "could" just chop it.
to whatever length you want.
oh that wireless kybd has come back to haught you again, huh?
it's low on juice or something, so I just changed batteries.
now it's being fine again.
stupid freak'n KB.
Well, that's what I have to ask Les, I'm not sure if the ext will casue issues or not. Someone mentioned premature bearing wear as well.
premature cause you're putting an extended sideload.
yeah, and I'f really not trying to that, just extend it's reach a bit, especially on curves... maybe 1 or 2" would be enough.
[sighs] bloody wireless connection
Well, that's creepy - Jymmm is greeting me
asdfqwega would you prefer I don't?
== $250 USD
emc2 - compiled and working
And now...to break it!
asdfqwega : I'm still interested in knowing what cut that deer out.
Jymmm Precision shotgun. :)
asdfqwega: your nick reminds me of something, could it be hitting a qwerty keyboard?
ValarQ: With a 10kg cat. :)
Or maybe someone beating their forehead on the kb. :)
[06:48:24] <Jymmm> http://www.mlcswoodworking.com/shopsite_sc/store/html/smarthtml/pages/routcollet.html
I just got a set of these, very nice --> http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&category=67243&item=7519183342&rd=1
Jymmm: a thousand million little homiculi weilding machetes
ValarQ: my nick comes from my left-handed tendencies combined with my need to hold a soda with my right hand :)
Oh bloody hell...connection's going again?!
ROTF!!! The set I just bought off of ebay for $68 (less the silkscreen) --> http://www.mlcswoodworking.com/shopsite_sc/store/html/smarthtml/pages/set66.html
asdfqwega: i see
reminds me about the code i wrote which contained a lot of the words "database" and "request"
one can get pretty tired in the left hand ;)
Speech to G-Code... Turn left 135.2 degrees for 12.7324 mm
asdfqwega: You still alive/awake?
Damn, I scared him off.. :)
While I'm waiting for apt-get to download lyx...
Paul, I use M62/M63, and I know what they are
I was looking through emc2, and saw that there is also M64 and M65
M64/65 are also in EMC and the bdi-4 branch..
They do the same thing as M62/63 except.....
a) the pins are on the nonrealtime port
b) On/Off is not coordinated with motion.
Hm...I'm trying to remember where in the emc documentation there was information on the gcodes
The User's section.
Or are you after a more detailed "What should this G word do" type doc ?
Heh...my connection is flaky...cvs checkout never finished
[09:32:25] <paul_c> http://www.isd.cme.nist.gov/personnel/kramer/pubs/RS274NGC_3.pdf
No wonder I wasn't finding what I remembered
is steppermsegmod.o the new version of steppermod.o ?
i'm using the latest vers. of BDI
can't find steppermod.o on realtime/lib/ dir..
steppersegmod is (was) an experimental module in EMC
hi paul_c , which should i use ?
It uses some code that is broken and is unlikely to be fixed.
i'm using 3 steppermotor
maybe is better freqmod.o ?
On the BDI-4, for stepper motors, you only have one choice.
that is freqmod.o , right ?
tnx a lot
just drop the extension, so it becomes freqmod
i've a problem to connect togheter 3 opto switchs in parallel mode
* Phydbleep jumps for joy.
I fixed that piece of crap vibratory cleaner. :)
Hi jacky^ :)
i found that maybe the mode i connect the switc is wrong
i'm using a 2 resistor for any switches and all 3 switch are connected in parallel
maybe... i should use only 1 resistor for all 3 opto switches
beacuse so, the resistor for any switches don't go in parallel mode
jacky^: There should be 1 resistor for each of the 3 pins 12,13,15.. 3 switches in parallel should use only 1 resistor.
instead, i was tring with one resistor for any opto switch !
i'm stupid :-(
And I'm not psychic so I can't see it from here. :)
your suggest was ok
but i was missing the schematic
jacky^: did the resistors solve the problem with the mechanical switches?
but the switches won't work yet in parallel mode
OK.. So then ignore the resistors for the opto-switches.. Make sure to leave the current limiting resistors for the LED's though.
5V for the power?
yes, abou 5 V (4,8)
use a 1k or 1.2k resistor in series with each LED.
so, that limit current on pin right ?
Do that for each of the leds. Yes, That will limit the drive current to the LED.
No problem. :D
* Phydbleep impatiently waits for 10AM so he can re-start the parts cleaner.
Kerap!.. 5 hours to go.. :\
Phydbleep: what is 10 AM ?
why are you waiting ?
Phydbleep: I make it 12:01
Late enough on a Sunday that I wont be worried about waking/disturbing/pissing-off hte neighbors. :)
It's 5AM here now. :)
And since I changed the motor on the cleaner it wants to run at ~ 90db. :\
Okay, that's enough trying to read code...time to pass out
Phydbleep: stepper oer servo ?
* Phydbleep passes asdf-qwee the 'out'.
...and I thought we were out of 'out'.
Phydbleep: why don't build a case for your machine ?
jacky^: 1/3HP, 110V, 1725 RPM, AC motor.
too big ?
jacky^: No place to mount it to the bits of steel and the plastic buckets. :)
Phydbleep: You don't need a sound cabinet, you need a blast shield ;)
Hehehe... The motor is from an old treadmill, The steel plates were part of a Kubota tractor shipping crate, The springs are from a V8 Chevy motor.
And my abrasive buckets are square ~5 gallon buckets that we get cat-litter in. :)
* asdf-qwee has a vibratory tumbler, AND a sandblasting cabinet
Phydbleep: there is some picture online of your machine ?
Hehehe.. It'll shake 20 pounds of glass beads and 20 pounds of parts like mad.
* jacky^ curious
jacky^: Not yet.. It's really simple.
ah, if you are here after
2 steel plates with 4 exhaust valve springs to hold them apart.
i can connect my webcam to my linux box
so can show you ma small cnc
if you like
Mount 1 bucket, the motor and the shaker weight on the top plate.
jacky^: Sounds good to me, I need to get a small camera for the shop too.
Mount bottom plate to San Adreas tectonic plate
Fill the other bucket with parts/abrasive and drop it into the mounted bucket.
Plug it in and cover your ears. :)
It was a case of 'build a cleaner' or spend weeks cleaning parts with emery paper. :\
Phydbleep: Sounds like something I might try for a casting sand sifter
asdf-qwee: For a sifter use 2 contra-rotating weights... It allows you to maximize the shake and minimize the jump.
asdf-qwee: Are you going to build a ball-mill and a muller as well?
Phydbleep: I don't really need a muller - I'm using ordinary clay greensand, not oilsand
Phydbleep: ...and what's a ball-mill?
tumbler, rotating drum full of marbles to mix/crush clay.
Hm...probably don't need tumber
For good steel balls for a ball mill go to the shops where they re-build car CV joints and dig in the scrap ball bucket. :)
Okay...I'm leaving for real this time :P
They're also really bitchin for use in a slingshot. :)
Sure, Sure, A likely story. <JK> :d
G'night asdf-qwee :D
asdf-qwee: Hey! When you go for 'real' can you get me a pizza?
And some breadsticks, Don't forget the breadsticks. :)
and a strong black coffee...
* Phydbleep wonders if he'll get shot for "And stop tracking mud across my nice clean kitchen floor!."
12" Bridgeport rotary table Currently $0.99 http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&category=41943&item=7519721556&rd=1
image maybe is poor... sorry
can't be better
my fantascientific machine :-) hehe
can you see it ?
wget -crk http://*
it's an inexpensive cnc machine
my budget was ... :-(
just an experiment
i would like to upgrade to a real cnc machine in the next future
This NURBS stuff is starting to look more and more interesting....
i found an italian website that have something of interesting for me..
but, can't find the price :\
[11:51:25] <jacky^> http://www.ronchinimassimo.com/assi_lineari.htm
what do you think about ?
are good ?
* paul_c has been researching the use of splines in order to reduce the amount of information that gets passed to the trajectory planner.
jacky^: Congratulations it's a machine. :D
* Phydbleep is off to fall over for a while..
G'nite all. :)
Ahh..... fenn..... A word please.
Re: "Rumley" chastisement (sp)
a) the changes made were not discussed beforehand
b) They break with existing documentation.
c) They directly affect how the package responds to the usr input.
So.... If the programmed reponse differs to the documented behaviour, there will be plenty of others that may be upset.
this is why i think emc should be more configurable
maybe I was overly blunt in my comments....
but I'd already seen some of the changes long before they were committed
in either case, they should have been committed on a branch, not HEAD.
i don't remember everything he committed.. only the G92 "fix"
agreed about using branches for non-standard stuff
multiple M words from the same group on a single line....
G92 with a P modifyer.... Certainly wrong when you can use G10 to the same effect.
Yo Ray, me old fruit.
paul_c, where should something like "M02 don't_erase_G92's" configure option go?
Hi Paul. How's the island today?
fenn: But as you pointed out, discuss first, change when agreed on.
M02 clears G92 offsets - It is clearly defined in the docs.
but sometimes you'd rather it not do that
If ya don't like it, get the docs changed first. !
but then the docs dont match the code :)
which is worse, undocumented code changes or documentation for things that dont exist?
I don't think that g92 is that far of from Tom's doc.
Worst case - Code that does not do what it is supposed to do.
rayh: There were a couple of small errors in the code...
G92 was only being cleared on X after an M02
IMO x is treated different in a couple other places as well.
all axes are created equal, especially if you're working with a hexapod
fenn: .... 'nother thing..... Just because sme mickymouse outfit (Centriod?) does something, it does not mean we have to follow suit.
no, but you should pay attention to user requests
Hey centroid runs on Linux now.
pity their bloody web site doesn't
don't slavishly adhere to what the user requests, implement something intelligent
First - Define "users".
uh, people who use your software?
ISD -- Intelligent Systems Division -- where've we seen that sign.
fenn: I have three (maybe four) classes of "users".
yes of course
You don't think you're getting out that easy, do you fenn.
Those that have payed for the software.
(none, in this case)
Those that have downloaded the source and actually use it.
Potential users that have downloaded and want to use it, and have made an effort to read the docs.
"Those that have payed for the software" Your tax dollars at work.
Dept of Defense.
Then there is the final group.....
Dept of Commerce
rayh, they don't use emc anyway, so they dont count
The ones who just hang around in the background and whine, and will never use it.
Naval shipyards uses a stewart platform that can travel 300 feet.
Airforce uses one that can travel an entire b52.
One USAF base use(d) EMC core for materials handling.
And then there's that little mars rover.
Yep. They got a hexapod as big as a blimp hanger.
using emc or are you talking about rcslib?
Some of each.
well, that's cool. havent seen anything about that in any docs or links
And you won't.
Less than 5% of emc installs show up on anyone's listings.
Parts of EMC & rcslib gets used for stuff we never get to hear about in detail.
how do you know about it, then?
I mis spelled the Czech republic a few years ago in a post.
Got several very nice notes. One is in the archives.
Whole shop floors of abandoned Russian machines retrofitted.
fenn: Off the cuff comments overheard in private - Some of which allude to hush-hush projects.
Part of any open source project is a lack of systematic marketing.
Hell Alpo dog food was advertised for a year before the first can was manufactured.
it's actually quite disgusting when an open source project tries to advertise
We tend to get used to vaporware
In commercial environments.
I agree. My job round here is loudmouth...
but mostly to counter stupid claims.
We gain 42 new users each month. And never hear a word from 41 of 'em.
fenn: Didya know the G200x & Mach1/2/3/x all have their roots in EMC ?
funny, mach2 is emc's main competitor in the hobby market
As does the servo sherline retrofit from IMS.
Or whatever Fred Smith calls himself.
(or what ever we call him...)
And mach2/3 is a reasonably good product within the range of machines it can run.
I don't believe I'd trust it to a platform 100' in the air with me on or below it.
but don't try it on a servo'd machine or a high speed mill.
Now that is not to diminish your criticisms of EMC in any way fenn.
Just a bit of big picture.
yes, people will use crummy software if there's nothing else around
Gotta go for Dinner - fenn, Don't let Ray scare you off.
boy dinner already huh
Catch you in a bit Paul.
You voting today?
fenn: Tell me a bit about yourself, background, experiece and such.
i thought i already did that
* rayh digs through his alzheimers head.
learned c++ and spanish in high school, studied microbiology, jet turbines, japanese in college
after college i decided to make a lathe because i wanted to make a paintball gun
and couldn't afford a lathe
didn't know at the time you could get a lathe for dirt cheap used
too discouraged from looking at industrial catalogs
based on every other mass produced product, one would think that a lathe from the 30's is a piece of crap
but it turns out that machine tools aren't mass produced in reality
anyway, i've spent the last 2.5 years educating myself on what most people just grow up with
namely, how to run a lathe, welding, metallurgy, etc
There is some great old hardware out there.
yeah i really want a late 30's south bend benchtop
or a myford super 7
but those don't exist in the US
i'll never be satisfied with the lathe I built, since i can always make an "improvement" to it, no matter what
right now it's comparable to maybe an atlas from the fifties
Well that's a lot better than nothing.
gingery style... 7x14, box ways, plain bearings, belt drive
no screwcutting yet
it's a hell of a lot better than nothing
but what you really get out of building tools is an appreciation for how to make something from scratch
We have a lathe waiting for single point threading.
Yes and to be able to make a machine that will hold 0.0002 for 20 years.
Is way beyond my comprehension.
hmmm 20 years of sitting on its ass maybe
actually i retract that statement
I've got three lathes in the garage. All ran 3 shifts for 10+ years
this design should last well into the next eon: http://demg.penton.com/content/am/out_of_the_box.pdf
and isnt that hard to make with the proper equipment
Um. I'm on the end of a very long wire. It'll be a while.
actually thats the wrong one
look at this one instead - http://pergatory.mit.edu/rcortesi/portf/grinder/machdsgn48-51.pdf
uses granite way surfaces
basically you make a box way out of granite tile, like a surface plate
then align a porous hydrodynamic bearing to the way, and fix it there with epoxy
the slide never touches the way, slide/way interface is constantly being flushed with lubricant/coolant
and even if a chip manages to get stuck in there, it's made of granite anyway
he uses the same type of bearings for spindles too
The first I heard of these kinds of bearings was on the 200" reflecting telescope
We toured some of the NIST facilities during devFest.
They were studying cutting on an inexpensive lathe
Watching harmonics, spindle load, and figuring out how to
they were "studying cutting, and using an inexpensive lathe" or, "studying inexpensive lathes"?
specify a cutting technique along with part drawings.
Seems the military can spec a projectile
build it exactly the way they want during testing but
when the parts come back from manufacturers they do not perform the way they did during tests.
same as with any manufacturing
The intent was to be able to take an inexpensive lathe and by sensing several
variables be able to cut material exactly the way the product was specified.
but then it is no longer an inexpensive lathe
I guess that is a matter of definition.
The lathe that they had modified there was no longer inexpensive.
well, you can buy a corvette, or you can buy a camaro and put a porsche engine in it
They had several hundred grand in the sensors
But the idea was to be able to measure things like cutting torque and
calculate deflection based on chip load?
and vibration in resonance frequencies
active vibration damping?
Their results were showing that faster is better
yeah cutting forces decrease with surface speed, until you hit the speed of sound in metal, and then cutting forces go _negative_
and that there were bands of speeds and feeds where very good results can be had.
The project supervisor and I went through some of the plots and the
results were very interesting.
what's the name of the project?
Not a clue. More military stuff.
"smart machining systems" ring a bell?
wots #HALCMD1 = save neta ?
a command in the ini file that would (if uncommented) display the configuration of signals on stdout
Ah. I knew a lady named neta once. Thought maybe you did also.
Mornin Matt. Good to see you here.
save net displays "nets" (signals and connected pins)... the a on the end tells it to include arrows that make data flow easier to see
hey youse guys: I got jmk's cool AB computer going last night & tried to build emc2 on it...
the reason it is "save" instead of "show" is that the information is displayed in the form of commands suitable to re-create the config
capture it to a file and you can later restore it
* paul_c switches to lurk mode
tried... that doesn't sound good
* rayh neds coffee. brb
I had a couple problems: I installed a fresh bdi 4.20, then I got the head of emc2 & tried a "./configure;make"
the first problem was that I had to untar the kernal sources :)
did you do the steps on http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?BDI-4_Install
that's a one time thing you need to do before you can build
then it would build a while... steps? let me look at that... hold on a minute...
SWP_Away is now known as SWPadnos
I winder if there should be a link to that wiki page (or a snapshot of it) somewhere in the KDE menu on the BDI
depends on whether you see the BDI iso as a way to get a particular version of EMC, or as a way to get a distro on which to compile the latest EMC
no, I didn't do those things ;), I did install the kernel sources, but that's about it...
run thru those steps then try again, and it will probably work
I guess I'm thinking of an item under the EMC menu like "How to upgrade EMC"
I'll go downstairs & put the CD in & do it now
i maked a mistake with original generic.ini of BDI, so i download a new generic.ini from cvs, but now stepper on axis X wont work well :\
apt-get install --reinstall emc
controllers are ok
are you running BDI-4.xx?
jacky^: hint, use RCS locally to check in and check out your .ini files before editing.
do what Paul said
I favor simply renaming the ini files... leave the supplied ones alone and name your customized ones something else
mymill.ini or such
yeah, but even then I always use RCS or CVS, as its nice to be able to go back to how it was last week etc
true - but how many BDI users are gonna set up a local repository?
thats why I said RCS, it needs no repository
co -l foobar.ini
it just makes a foobar.ini,v in the current dir
from the command line?
assuming you have RCS that is
robin_sz: :-s i have no idea
man ci and man co come up empty on my BDI-4.20 box
what is ?
it's unlikely you have RCS on your computer, so you can't follow robin's suggestion
don't worry about it, just do what paul said
i would like to know what is RCS :\
revision control system
dont worry about it ...
keeps track of changes you make to a file
something like test program ?
its just a way of saving stuff so when you screw up, you can go back to how you had it before
ah.. right ! revision.. sorry
no - it is usually used by programmers to keep track of changes they make to a program
saves your ass from time to time :)
can install via apt ?
of course ..
apt-get install rcs
hmm, to install rcs, it wants to install libc6
bah.. i've another prob
The following extra packages will be installed:
libc6 libc6-dev libc6-i686 locales
Segmentation faulty Tree... 64%
might just be that libc6 has been updated, ,, theres been lots of updates on Sarge recently, since the freeze
, oh, yeah, Sarge is now frozen, if you hadnt heard
I wonder what the risk level of those updates is?
heh - I just started an apt-get upgrade
opnly 352M to get :)
So - got lots of time to talk now :)
A word of caution - Don't upgrade grub....
jmk i tried to upgrade libc once before i knew it was a no-no
what's the grub issue?
result: system totally trashed beyond all hope of repair
fenn: thats unusual on a debian release
it was redhat
dedrat I can imagine ..
debian is usually better at calculating the dependency tree
they had it in bugzilla all along too, just hadn't bothered to do anything about it
The last grub version I tried didn't like background images....
So if you upgrade, comment out the splash line(s)
why don't upgrade grub ? it's dangerous ?
didn't like as in "refuses to boot with a background image" or didn't like as in "the image looked funny"
will not worl emc after a grub upgrade ?O_o
Fails to boot with the splash image (any splash)
OK - good to note
ah.. np.. for this
ooh, looks like Sarge just got security updates too ...
I had a problem with the image anyway - too dark, so the hilight bar didn't show up
btw how do i turn off that annoying penguin with a flyswatter image?
if you find out, let me know
grub splash is not the same as the bootsplash image
that's not annoying - it's cute :)
(my only real peeve about BDI-4)
it's annoying when you have to look at it for six hours
(its the background image for all consoles)
apt-get remove bootsplash
well well well.... bonus side effect of upgrading libc - my computer is finally off of British Summer Time!
ooh, 80mb of upgrades here ...
jmkasunich: While you're at it.... Can you fix the BDI-Live slot on the farm to use configure & PLAT=nonrealtime|realtime
similar to what we did the other day for rcslib? yes, it's on my list, just didn't get there yet
Sombody propose an EMC topic?
* jacky^ mumble mumble
That's a tough one.
trajectory planners & NURBS curves.
hmm.. removed bootsplash but it's still there
Help jacky to start with his cnc :P
mmmm NURBS ...
question: if we use libnurbs++, will that cause problems in the realtime interpolator?
due to c++/c differences
you can not use C++ in kernel/realtime space
there was some discussion last week as to whther trajectory planning even neede to be doen in realtime
oh i thought it was non-realtime already
not quite realtime ... ;)
basically it happens as the job runs
kernel space is the same as realtime space
fenn: which is part of the reason the segementqueue runs into trouble, it just cant keep up sometimes with very small segments ...
fenn: in theory at least, it could be done in advance, before the job runs ...
that sounds like a good idea to me
les was considering it as a way to save his machine ...
why don't code everything in C?
but it doesn't solve how to implement the realtime nurbs interpolation
ValarQ, haven't found a nurbs library in C yet
* paul_c has much of that solved already
converting NURBS to points is trivial.
but can you access the nurbs data structures in order to do that?
opps should I wake up?
les: no, quiet now grandpa, nothing to get excited about. drink your cocoa ;)
les: Nah... we ain't on the math yet.
fenn: Yes, you would need a list of knot points to process.
is (les) really considering nurbs as a way to fix motion in emc2?
or is this all speculation?
no, consdiering pre-calculating the trajectory
or discussing it anyway
I had been looking in to the possibility of using splines to get over some of the limitations of the existing trajectory planner
and reducing the amount of data being passed to the RT code.
but also transferring trajectory info from user to RT as coefficients, not a large number of lkine segments
so the RT section would be calculating from NURBS or whatever
in theory, its fairly simple to just do it all in advance, and then transfer the points into a stack for the RT stuff to eat through
plus, you get to go along the path in both directions if you do it in advance,
your decels become accels
and predicting where to start slowing down
just goes away
Yes - Think of it as following on from Rogier's work.
i never managed to read more than five pages of that paper.. bleh
yeah, its a bit heavy
or as Les would say "enjoyable light reading"
NURBS in general do not guarantee smooth kinematics
But there is a lot of stuff out there for them
even a sourceforge projest
but CAM systems?
that canoutput nurbs for toolpaths?
the NURBS curve fit would be done by the TP
I don't know... have a look...but it's a (gulp) c++ class...
which then passes the curve data to the motion code (right?)
Agreed, NURBS in it's self doesn't guarantee smooth motion, but it does go a long way over what we have at present.
I suspect anyone who could afford a NURBS capable cam system wouldn't be using emc ...
there are provisions in RTAI for using some C++ in kernel modules
though I know nothing more about it than that
les: If you are looking at libnurbs++, steer clear - It also uses STL.
99% of users will still have the "I sent it a load of small segments, it went slow and jerky" problem on gcode, even with nurbs
I am, and I will.
My BS meter went way over the top with "I suspect anyone who could afford a NURBS capable cam system wouldn't be using emc"
SWPadnos: C++ and RTAI in the context of emx[x] is a no-go.
is there some application like a cad on linux to produce gcode ?
rayh: Where did you hear that comment ?
if NURBS can represent any path (including arcs and lines) then g-code can be translated to a NURBS path in user space, you don't need NURBS capable cad
websys is cad/cam on linux and unix
paul_c: is there somewhere that has specific reasons why? (I know you've looked at it, and I don't want to waste your time explaining it to me)
tnhaks rayh , ill try it :P
(and yes, I do know the general rule of "no C++ in the kernel")
jmkasunich: ahh, so you where thingking of fitting a nurbs curve to the gcode, rather than the CAM producing its commands as nurbs knot points?
I agree with Ray. Having nurbs in readsble g-code for emc is impractical....only the highest end cam systems can do it.
RTAI & C++ won't work in the context of emc[x] unless you are willing to forego compatability with RTLinux
Nurbs as an internal TP calculation perhaps
les: Bob over at Webersys let slip that there is a post-processor that can export NURBS
Ah. I though that development was going forward with RTAI - didn't realize the RTLinux support is continuing
But only certain subsets (if any ) would match position, velocity, and accel when passing control points
well.... You could scrap the rtapi layer and align yourself with RTAI....
One of the big reasons the interp does what it does is that there was a fixed size memory allocation for it so that embedding was possible.
rayh: considering that an embedded machine often has 8-32 MB RAM these days, maybe that can be changed :)
even more for a PC
(used as a machine controller)
No reason I can think of that we need to hang to that spec.
at the interp level, NURBS support would add little to the memory requirement.
steves_logging is now known as steve_stallings
We could allow traj to stack in advance.
rayh: careful now ...
But we will have to produce better ways to manage stuff like estop and feedrate override.
It would have to be the whole program, no?
feedrate override would mess up the accels
if we pre-calced the traj
calc at max feedrate...others will be valid
but that may be excessive ...
for machiens with a large dynamic range
But it is really the only way to allow for non-deterministic things like feed over ride during execution without the TP being RT
feedrate override is handled in usr space, so that isn't really an issue.
I don't think that it has to be the whole program at one time.
it is an issue if you've pre-calced the entire path assuming a specific rate
you can always run over it again ...
but that would be horible if it had to happen for each change in feedrate override.
the premise is that if you change the feedrate, that change should take place "immediately", not after you recalc a large chunk of the path
I think pre calculation of the whole program is no good idea !! My programms have usually some MB !!!
or just stick an FIR filter prior to the rt stack
hm, if you can pre-fit the nurbs curves in userspace before running the program, you would be all set..
although that would require mods to the interpreter
Les' point is that if you calculate at max rate (according to machine parameters, not the maximum found in the file), you only have to slow down t o get a programmed rate or an overridden rate
it's a simgle multiply/divide (per axis) to slow down
right which is just time scaling
always realisable if the only direction is slower
but if you calc at the max rate, you are going to round some corners and such... if you slow down, the machine can make the corners sharper, and in fact that might be why you slow down
simply following the fast path slower won't help
and you ignore the programmed rates when calculating the trajectory, then apply a slowdown from max speed to get the programmed rate
* anonimasu nod
SWPadnos: good point
jmkasunich: yeah, but that's why you need to have somthing like lookahead
jmkasunich: calculate the angle change of the next s[t] segments
if the angle is greater then x slow down the feed..
what is the package containing emc on BDI ? can't find it
I understand jmk.....but I think it is a fundamental thing you must do if the TP is not done in RT
it's not a matter of ahving the full speed of the machine, it's the max feed possible for a given feature
my point is that IMO maybe the TP should remain in RT
SWPadnos: good point
RT vs non RT is a good thing to talk about.
SWP: for a sharp corner, the max feed is zero, unless you are willing to accept rounding
it was the speed thing I was talking about
that's where exact stop, exact path, and cont contouring come into play
true - so the path following modes need to be honored
in exact path and exact stop mode the path can be precalced, and just speed scaled later
or slow down until the machine can cope
unless you have a CAM program that does loops for outside corners :)
in contouring mode, the amount of rounding is speed dependent
you can't just speed-scale the path
sounds like plenty of user choice...so what's the problem?
jmkasunich: you could specify the maximum alowable G-force during a sharp corner..
at the Siemens controller you specify a tollerance: TOLSL (0.003) and TOLSR (0.02) (SR:roughing; SL:finishing)
actually - you already have a max accel in the ini file
if the feedrate is high, max_accel still means a large rounding radius... slow the feed and the same accel gives you a tighter curve
[15:12:36] <anonimasu> http://www.whisqu.se/per/docs/math2.htm
anon: what I think folks have been talking about is doing the lookahead in user space, precomputing the path assuming a particular speed...
but you need to recalc the lookahead if you change the speed
jmkasunich: hm, I wouldnt think lookahead could be done in userspace..
since it has to be realtime..
agreed... that's what I've been getting at
you can calc everything of the path in userspace if you want, but you still need to check your corners/speed, in realtime if you want lookahead
besides, I can think of applications where accelometers are used for that..
gotta be able to do some of the TP in RT if you want adaptive machining.
paul_c: have a eye on the url I pasted, might help out a bit
OK so tightening corners with non deterninistic stuff like feed over ride would have to be real time
as would slowing down if the spindle load increases, etc
and I don't think you want to be parsing an entire file and pre-planning the TP in usr space.
I can't think of a scenario where I would want to do that though
others might I guess
if you are milling ferrous stuff, it would probably be greta
if you are milling woodish stuff, slowing down when you hit a knot is nice
For me...I want a precalculated path to be followed regardless of feed OR
hm, wouldnt that give you nice burnmarks
can't just slow down without slowing spindle too
at least in my work.
I was assuming that the reason for the slowdown is that the spindle is bogging down
well, when you mill ferrous, and it starts to be too heavy, you can slow down, in non ferrous stuff like alu it'll gum
so yes, keep feedrate proportional to spindle speed
* anonimasu nods
belts... (IOW, approx speed match, not absolute position match)
Would there be some serious calc time issues in a trajectory level curve in RT?
depends on how many segments need looked at
the problem is that more segments need looking at the faster you go
and the smaller the segment
fenn: I just read your post..
so you get a square law (or so) increase in workload at higher speeds
right, so the RT calc load is unknown
another line of thought I have is this:
1) emc has s-curve motion planning
2) if the velocity adaptation worked there would be no stutter
:\ can't find a deb package on BDI, im dowload rmp package to convert with alien ..
bad idea ?
3) to make the s curve work well traj/servo pperiod has to be big...then it coarse
it's too course
wish I could type.
jacky^: Just what are you trying to install from an rpm ?
paul_c: i maked a mistake on ini files, so i'm tring to reinstall emc on BDI
On BDI-4 ?
The emc deb package is already on the CD
can't fine nothing using apt-cache search emc
did you do "apt-cdrom add" ?
ah.. should i add the cd rom source so..
So as I see it two things would make the existing TP much much smoother......
fix vel adapter and increase servo rate
increasing the servo rate won't improve the TP....
les did you mean increase planner rate?
it won't improve what is there....
but it will allow a higher traj/ servo period ratio
and that kicks in more cubic s-curve sub interpolation
heh sounds like a dirty hack to me
according to fred it needs to be about 10:1
it is a dirty hack.
just needs to be said though.
It is not as good a solution as splining at the trajectory level.
if you increase the traj/servo period ratio, will that reduce data starvation?
for the lots-of-little-lines scenario
the ratio increases s curve smoothing
the starvation is an algo or math error
s curve smoothing is in the servo interpolator, right?
when the requested points are too fast the TP should always slow down until they are NOT too fast
s-curve is at the servo rate
two distinctively separate problems.
the verify currently checks a program to see if the machine limits are going to be exceeded, right?
(ie, axis limits)
I think so
it should also check to see if either (a) there are any features that the machine can't follow (due to kinetics)
but when unrealisable speeds or accel are requested there are only two choices....
or (b) there are any features that are too small in time (that would cause choking)
IMO we need to s%$# or get off the traj. The nice thing about it is that more than one traj planner can coexist in the code and get used or not at the whim of the integrator. So who's gonna start this traj planner?
yes - refuse to run, or slow down
\slow down to the point where they are possible or refuse motion.
sorry - I had it backwards :)
The choice is almost always to slow...for obvious reasons
not obvious if finish is important
slowing down gives correct geometry
If finish is important one might need to write a better g-code program so less velocity adapting occur
refusing to run tells the operator to hire womeone with a faster machineif they want a nice finish
I agree that making that speed ceiling higher is aa good idea though
It is about the only way if you have over rides and the TP is not done RT
wow you guys just covered almost everything I said in the feature request (re: segmentqueue update)
oh yeah? heh
it's been under discussion for a month or two (or is that years? :) )
rayh, are you referring to a adhere_to_path and a velocity_adapt TP option?
non-debian cdrom-detected the cdrom drive contains a non debian cd
you can adhere to path and velocity adapt both
adhere to a planned path
as opposed to adhere to a g-code path
I think we are calling the latter "exact path"?
er sorry i meant constant_velocity not adapt
"exact stop" ?
well there is stop at each point.....
and an exact path where at least one axis must stop if there is a reversal
exact stop will stop between each line/ark endpoint, even if the lines/arcs are tangent at that point
jacky^, mount /dev/cdrom /cdrom; dpkg -i /cdrom/ (dist?)/cl/ (something) / emc*.deb
exact path stops when lines/arcs are not tangent, but doesn't stop if they are tangent
continouus contouring doesn't stop at all, it rounds corners
jacky^, uninstall emc first
apt-get remove emc
ok that's clear
fenn: No - bad advice.
paul_c, does that remove dependencies also?
apt-get install --reinstall emc
his apt is broken, see above
I have no idea which of those three modes, if any, are implemented completely correctly in EMC ;-)
I suspect only exact stop works under all cases
So, I'm starting on those procedures to get compiling to work. I had installed development packages when I installed 4.20, so the first step yielded:
it probably takes a pathological program to break exact path
and many programs break continuous
$ sudo apt-get install cvs g++ gcc gcc-2.95 ncurses-dev pciutils-dev tk8.4-dev
Reading Package Lists... Done
Building Dependency Tree... Done
cvs is already the newest version.
g++ is already the newest version.
gcc is already the newest version.
gcc-2.95 is already the newest version.
Note, selecting libncurses5-dev instead of ncurses-dev
libncurses5-dev is already the newest version.
pciutils-dev is already the newest version.
tk8.4-dev is already the newest version.
0 upgraded, 0 newly installed, 0 to remove and 0 not upgraded.
W: Couldn't stat source package list http://ftp.debian.org
testing/main Packages (/var/lib/apt/lists/ftp.debian.org_debian_dists_testing_main_binary-i386_Packages) - stat (2 No such file or directory)
W: Couldn't stat source package list http://webersys.com
./ Packages (/var/lib/apt/lists/webersys.com_debian_._Packages) - stat (2 No such file or directory)
W: You may want to run apt-get update to correct these problems
is this ok?
well arcs are not blended....they are exact stop even if tangent in all modes
run apt-get update like it says
ok, so exact path is busted
mshaver: Tip - Comment out the teo source lines in /etc/apt/sources.list
and the velocity adaptation as well
ok - doing that now & continuing the procedure
which ? comment or update ?
commenting out the two source lines in sources.list and continuing the procedure in the wiki
darn - there's no emacs on this 4.20 machine...
apt-get install emacs ;-)\
apt-get install vi
apt-get install pico
nano is installed though...
i love pico
so why would mshaver want to mash his keyboard just to edit a couple of small files ?
I can't apt-get install emacs because of the problem with sources.list... trying nano...
apt-get install openoffice
SWPadnos: scroll up two lines.
openoffice for editing text files??
(just adding to the general hysteria)
some newbies might not understand that >:P
newbies wouldn't understand emacs or vi either
i _still_ can't figure out vi
ok, problem with sources.list is fixed - actually I was going to try to run Microsoft Office under Wine for editing text files ;)
:wq is all I need to know :)
I don't understand what the arguement is all about re vi or emacs.
It all boils down to "I don't care what *you* like - I'm right and you should listen to me"
they both require keyboard mashing and generally suck. (my own opinion)
i'd at least expect interrupt to work so i can quit if i accidentally type vi
instead i have to close out a perfectly good console
all editors require keyboard mashing - it's inherent in the function
word is even better - it requires keyboard and mouse mashing to get anything done
Wrong, the mouse must be fondled. 8-)
hey - get a room ;)
or stamped on when you get a bsod
fenn: to leave vi ... <esc>:q!<return>
btw - I like nano!
sombody know what is the root passwd on BDI Live ?
jacky^: yes, its written on the CD
on CD ?
Doesn't BDI4.20 install KDE with Kate and KWrite?
i've an iso
jacky^: Live doesn't have a root passwd
yes (KDE, etc.) but I'm ssh'ed in on a console so need text mode editor
to get root, type "sudo su"
jacky^: sorry, just having fun ;)
uh.. it ask me for a passwd :\
and set your own.
ssh -X should do it on your system, matt. It did for me.
maybe because i already have installed bdi on hd ?
but it won't accept the passwd of root i use on hd :\
same applies sudo passwd to set one
You will need to be root or sudo on the box and get root on the ltsp.
i'm going to try..
If you booted from CD, Live won't look at the HD for passwd
but it's strange, it ask for root passwd from live cd..
The passwd looked for by ssh -X is on the remote box.
just press enter
auht failure with enter
so when you did sudo passwd what did it do?
robin_sz: it work with sudo passwd
thanks a lot
now can mount /dev/hda :D
ok, emc2 HEAD builds for me on bdi4.20!
jmkasunich could you tar/gzip the files from your old website and dcc it to me?
or post it somewhere
there's no directory listing
can't you just wget the website?
nowhere to begin?
i dont understand
(if you have wget)
[16:34:25] <fenn> http://home.att.net/~jmkasunich/
I get :no contect" too
the files are still there though
I mean: I get "no content" too
[16:36:29] <jmkasunich> http://home.att.net/~jmkasunich/EMC_Docs/EMC_Home.htm
I don't have a top level index.html, that's why the no-content msg
well, some of them are
page url not found
that link works!
interesting that they block access in that case, versus providing a directory (like apache defaults to)
better for privacy, I suppose
fenn: One downside to wiki... It doesn't translate to pdf docs for printed manuals to well.
Defaulting to a directory listing sounds like a security risk.
Oh hey. We should specify what options we want with indent.
rayh: already done, built into "make indent"
that was quick
fenn: wget -rk http://home.att.net/~jmkasunich/EMC_Docs/EMC_Home.htm
it worked :)
it said "not found" when i tried to load it in a browser
it worked in Mozilla
hmmm... I'm looking at it with konqueror right now
hrmmm... worked for me.
and in Firefox on an EMC machine
works for me now
could be some flakiness at my ISP
wouldn't surprise me a bit
did you do those diagrams with LyX?
guess i just have to read with my own eyes, huh :)
easycad (windows cad pkg) then print to postscript, and used ghostscript to make gifs out of them
I might be able to find the original easycad drawings and convert to dxf or something, but no guarantees
(that was a couple computers ago)
probably easier to re-draw it
and update it at the same time
coming up with the mental picture is the hard part
well, not if you wrote the code :)
now that I've got emc2 to compile, is there a way to run a Servo To Go system with it? Or do I need to write STG HAL modules?
notice figure 3... that is a HALish way of describing a circuit, drawn before HAL work even started
you or me or someone needs to write a STG driver for HAL
ok, need to look at that...
is there a good tutorial on how PID loops work anywhere?
don't know of any specific one... just have to google it
[16:55:53] <paul_c> http://lorien.ncl.ac.uk/ming/digicont/digimath/dpid1.htm
ewwww... mathy stuff
[16:57:05] <paul_c> http://newton.ex.ac.uk/teaching/CDHW/Feedback/index.html
much nicer... bookmarked for future reference
* mshaver is back
* mshaver is away: mshaver
[17:01:47] <paul_c> http://zone.ni.com/devzone/conceptd.nsf/webmain/F0A8687D6706719686256D6D005F9C56
nuts... RTAI now has ia64 support in one of it's branches.
ooohhhh - cool
It won't do you any good though....
that's the branch without shm?
how much resolution do you need for adequate servo control voltage? is 4 bits enough? 8 bits?
4 isn't likely enough
12 min. 16 is better
could i get by with 8?
probably, but the "dithering" at steady-state would be higher
A Delta 20 control had an 8 bit DAC (1) and an analog mux with sample&holds to store the output voltages
If you were running a 30 year old system, 8 bits might be enough.
I wonder if a 6-channel sound card could be coaxed into being a multi-axis servo output...
what if i used a "logarithmic" DAC instead of a linear one?
I looked into that, problem is sound card outputs are AC coupled
since you dont need that much fine control at high speeds
fenn: I would disagree... High speed is where you want tight control
(and also low speeds)
the PID loop will make the average speed corect, but the lower the resolution of the output, the more stuttering you'll get
a higher servo rate can in part compensate for that
12 bit dacs are cheap enough anyways.
lower esolution output should also make the system more prone ot oscillation, since small control changes make large feedback changes
SWPadnos: More coffee
(now wouldn't *that* be a great sed feature :) )
working on it now
paul_c, i'm planning a one-parallel port servo control bus
(the coffee that is)
4 bits used for addressing, that leaves 8 for data
fenn: you should review the pico systems stuff
fenn: Jon Elson has been there already.
are you planning on muxing a pseudo-DAC to different axes?
no i'm addressing data to multiple devices, be they stepper step patterns or servo voltages
you'll have a speed problem, I bet
especially with step signals
that's the tradeoff
you can pare down the number of devices on the bus, and only multiplex through the ones actually present
i.e 6 devices will go at half the rate of 3 devices
pare it down to 4 axes, and just use 2 bits each (like freqmod :) )
the issue will be the outb instructions.
I/O runs at a nearly constant rate, regardless of CPU speed - it's around 1 microsecond for an outb
which is a reason to wonder if 8 bits might be enough ;-)
freqmod just hits step/direction once each cycle, right?
yes - all bits are calculated, then a single output is done
you can't do that with address / data outputs
* jmkasunich goes to do yard work
jmkasunich is now known as jmk_away
that reminds me - I should be fixing the deck stairs (before the rain starts)
hmm pico-systems stuff is pretty expensive
the whole reason i wanted to do it my way was to keep the cost down
you'll limit yourself to roughly 10k pulses/sec if you do a multiplexed step output, maybe less
(that's just an estimate - so you may fare better)
anonimasu: Can you resend that email please.
do you think is it possible to use an ATX power supply for 3 stepper motor/controllers of 2,1 A anyone ?
about 30 V
30V*2.1A=63 watts continuous
from 12 to 30..
not much for a 300 watt p/s
so, a 250 W power supply should be adeguate right ?
i really dont know
250 watts is fine, performance with only 12 volts from supply will suffer due to insufficient voltage to force current into the inductance of the motor windings
then, will not use atx
actually, can't you use the -12V as ground and get 24V?
or will that blow up the ps?
-12 supply is typically very low current on an ATX supply
actually i'm using a small transformer about 2 A with a capitor of 2200 UF
i don't know if is adeguate for 2A motors
well, at least you dont have to worry about frying your motors
but fryng also performance :\
good rule of thumb is providing 2/3 of the motor coil rating as supply current, e.g. 3 motors with 2 ampere coils need 4 amperes total
don't forget that transformers are rated RMS and a rectifier/cap is NOT and RMS load, derate to about half on transformer
SWPadnos is now known as SWP_Away
tnx steve, very clear
i was looking the powers supply used here.. http://www.roboitalia.com/modules.php?name=coppermine&file=displayimage&album=101&pos=30
it seem strange to me.. that can work well
hard to comment, picture does not tell me any specs
power supply looks to NOT be a PC style, more likely something like plain 24 volt single output
alex_joni_away is now known as alex_joni
hey steve.. anything interesting today?
lots of discussion springing from Fenn's comments on dev list
just reading that now
will sweep through the logs aswell
jacky^: you that guy from italy with the limit switch problems?
alex_joni: yeas, maybe i solved
i'm stupid :\
i was connecting optoswitches and resistor togheter
2 resistor for any opto siwtch
someone tell me to use only 2 resistor
and the 3 optoswitch in paraller togheter
ok, heres my question, sorry
i found a transformer that have in output 20-0-20 5 A
it should be fine for my controllers (required 12-30 V)
but i don't know how to connect the diode
20+20 give 40 volts that are too high
i think is unusable right ?
40 volts probably wont hurt the motors
getting close to the limits for mosfets though
40 is too high voltage
also for the capacitors
you can use the transformer as center-tapped secondary with two diodes to get about 1.414 x 20 = 28 volts
should i use, 2 diode for 20 - 0
also this configuration needs less derating on transformer since only one half at a time conducts, about 0.7 x rating
it is called half-bridge, ground the center tap, use one diode from each remaining lead to the cap
epanorama may have some schematic..
i'm going to look (half-bridge) tnx steve_stallings
[18:11:40] <steve_stallings> http://www.inform.umd.edu/EdRes/Topic/Chemistry/ChemConference/Software/ElectroSim/FullWave.html
paul_c: what mail?
I'll be back later to chec.. need to work
cradek: you here?
great wiki... a few comments, if you'd like
1.the 2.6 stuff works now with emc2 HEAD, so you don't need the branch
2. you said you want to controll flood?
more of a hardware thing I'm thinking
well.. standard emc2 uses simio in the emc.ini
I hacked one called io
which exports some HAL-pins for flood, mist, spindle, etc
replace simio with io
should make it fairly easy to use
right... I was thinking that it was probably jsut goona ba hardwarething
I'll likely get to it, but right now, I'm just concentrating making chips, and getting the interface where I want it
got axis to run?
I'd like to run the sherline interface, but John wasn't sure how hard it would be to get it in there
so I'm looking at axis
but am having problems there too
I did run mini on emc2
difficulties not problems
got axis to compile for emc1?
I think part of the problem is that i didn't do the make install
and its linfving in a different directory
no, never did that
it's easier to do that
then.. when you get axis running for emc1 (with make install)
what? install it to emc1?
just place a symlink inside the bin dir to emc2
in which bin dir?
weyland: that's how I did it the last time, but will check again
give me the dirs where you have emc1, emc2, and axis
emc1 is stock BDI4.20, emc2 lives in ~weyland/emc2/emc2jmk/, and axis is unpacked in ~weyland/downloads/axis-1.0
you probably need to get emc1-bdi from CVS
didn't look at the latest axis
I used a previous version
the main problem with emc2 was getting it compiled
??? why emc1?
I can't go bnack to emc1
that's how jepler did it
that's why we upgarded me to emc2
I didn't say you should
problem with backlash comp
then I'm unclear
I agree emc2 is what you need to use
but emc1 is ok in order to build axis
once you have it compiled it can be used with emc2 too
without any problems
if you have some patience, I'll have a look at running axis on bdi-4
maybe send you the bin
so you're saying to compile axis like i was using emc1, but continue using emc2...?
I can try that
weyland: Email addy ?
and I can send you the latest build for axis
weyland: just for the future.. don't spell your email addy in clear
use smthg like alex DOT joni AT robcon DOT ro
paul_c: at one point you said smthg about including axis in the BDI
got dropped because of space?
axis is included.
package name - python-axis
just thinking out loud...
Hello all, I got upgraded to emc2 as per Weylands wiki page. But still am having trouble with bachlash compensation. Seems to only add it in 1 direction.
since we now know that emc2 is good for me, should I just go ahead and do the make install?
bpmw: are you for sure running emc from inside the emc2 directory?
brb, going to check mail
apt-get install python-axis
E: Couldn't find package python-axis
Weyland, I was able to get mechanical lash down from 0.020 to 0.004 and now am trying to get 0.000 through lash compensation but does not seem to work.
weyland: you shouldn't run make install on emc2
Any body else have any thoughts on this matter?
at least not for now.. maybe in the future
bpmw_: watch for jmk_away turning into jmkasunich
then bug him
okay, got the mail, but it wasn't 2gig
installed the 2.3axis
I think he's gone till after supper!
weyland: sorry, it's not there yet.. (the 'make install' )
bmpw_: when he gets back then
errr... what isn't ?
make install is not functional on emc2
oh, you mean in emc2?
at least not to the extent it should be
right now it only installs man pages ;)
I changed my ini to say DISPLAY= axis
and when I tried to start emc2, I get the splash, but no emc...
lemme try from command line
Can't execute DISPLAY program /home/weyland/emc2/emc2jmk/bin/axis
weyland: you need a symlink to axis
i guess i need a symlink
placed in emc2/emc2jmk/bin/
I also wanted to thank Weyland for putting the upgrade instruction on his wiki page for my behalf. upgrade was long 4.5 hours on dial up. But went extreamly well otherwise. THANKS again Weyland!!!
jmk_away is now known as jmkasunich
bpmw_: you're ver welcome
was the least I could do for the help I received
bpmw.... you said something about backlash probs?
nad if it helps someone, all the better
alex_joni: where's the symlink need to point to? I can't find it
depends on where the axis binary is
so what seems to be the problem bpmw?
I was trying to calibrate Mill this aft,
alex_joni: think I found it, hold on
Have 0.004 of mechanical lash, so added 0.004 into backlash and it only seems to work in one direction.
what do you mean only works in one dir?
alex_joni: OOOOOOOOoooooooooo... Nice......
each time you reverse direction, it should move by 0.004" in the new direction (plus whatever move you are making)
paul_c has kicked alex_joni from #emc
??? what the heck was that?
* paul_c showing that it can be done ;)
for example, dial indicator set to 0, make a +0.100 jog dial only moved to 0.096. then made a -0.100 jog and dial moved back to 0.096.
bpmw_: make more than move in the first direction
THEN, reverse and check
you mean, move +0.150 then set dial ind to 0 and make a -0.100 move?
move anotehr .100 again in the same direction
set dial to 0
THEN reverse direction
playing in axis
Sorry had to gas up lawn mover and start it for wifey
hmmmmm... not getting plot display in axis...
Any way, did as you said and on -0.100 jog actual move was -0.101
try it aa few times and see how much repeatability you get
weyland: what's the bin name for axis?
the links name?
or the real ones name?
bpmw_: could just be your screws aren't that great
steve_stallings is now known as steves_logging
alex_joni: which one?
the real ones
you might want to copy it to /usr/bin , makes it easier to use
fenn is now known as fenn_afk
Weylan, did same test 4 more times (this is with 0.004 set in backlash) and same result. on making the -jog actual move was -0.101
bpmw_: is it possible you have actually .003 backlash and are over compensating?
bpmw_: does it return to 0 the other way?
alex_joni: still no love, or display
also, backlash will vary over the length of the table...screws tend to wear more in the center than near the ends
not according to dial indicator with no backlash
yeah, good point, john
I'm trying cal in the middle. with out backlash set it is consistantly 0.004
you got me, but I don't know anything about the softwware side of it
one electrolitic capacitor has made a coffe :\
John's the man, there
weyland: works ok here (axis on emc2)
With that set in backlash it seems to add it self up on each direction change but only in the - direction!
to see if baclkash is doing its thing, try making incremental jogs of 0.0001
alex_joni: damn... figures
lemme go try using mini and see what happens
weyland: what's wrong with axis?
bpmw_: what jmk said, and maybe reversing the direction once in a while
alex_joni: just want to see if mini lets me see the plot, that's all
Ok John, brb.
weyland: change tkemc to mini
alex_joni: just did, getting nothing again, checking from command line again, probably another sym link
weyland: my bad
give me 5 minutes to fix emc.run for mini
alex: should we just copy mini over into the emc2 tree?
it is there, I already did that
changing emc.run now
John, trying to do a 0.001 jog I get "joint 0 following error"
but there are a few more hacks I need to apply mini to make it work
yeah, now its looking for emcsh, brb makinf more sdymlinks
sorry, bpmw, I have to go...
jmkasunich is now known as jmk_away
weyland: works now, commiting the stuff so you can check out
bpmw_: Post your ini file somewhere.
Sure, such as?
03alex_joni * 10emc2/scripts/emc.run: added mini support to the runscript
weyland: how do you check out the CVS?
paul_c, Can I e-mail it to you?
Ok, thanks I need your address!
Ok thanks, will take a few minutes though.
weyland: it'll probably be a while till the fix I just commited will be visibly in pserver CVS
it seems that sourceforge can't keep the CVS servers synced better than this
up to 24 hours.
lately I've seen under 12.. but that's not a rule
alex_joni: Okay. Got it running. Bunch of symlinks
displays plot fine
* paul_c is away: feeding the animals.
mini and axis?
haven't rechecked axis yet
just got mini running and it displays the backplot fine
will go recheck axis now
axis should display by default
alex_joni: No... jsut a black box where the plot should be....
maybe your openGL isn't working ?
not sure how to check that tho
paul_c , A'm working on it having trouble with floppy drive.
btw.. as I said a few months ago (axis is great)
ran it just with emc2 on the latest BDI
thanks! glad you like it
hey if you got it working with emc2, maybe you could put up some simple instructions on the wiki
just recently I added some nice g54 support to cvs
are you using the cvs head?
I'm using the one on the BDI
* alex_joni is lazy today
oh so paul did all the work? you fooled me
paul included it for emc1
which version is on the bdi?
I just symlinked it
Ok paul_c have sent .ini file to you.
bpmw_: paul_c is away for a bit
there are important bug fixes in 1.0 and I've added some new features since.
ok.. getting it now
and documenting it in a wiki
cradek: ok like that?
I don't remember
but I did *lots* of testing before 1.0 and fixed several (6-8?) bugs
and if you use g54 offsets, you will like the cvs head now, it shows both the g54 and the machine origins and allows you to set them up easily
hmmm.. looking where to find cvs
start at axis.unpy.net
yeah.. seen that
hmm, not sure how to get it with anon cvs
but can't find "user" instructions to download from cvs
lemme just try
you can definitely get a 1.0 tarball from the web site
yeah.. cvs kinda refuses the connection :)
connect ot craie.unpy... failed
(jepler runs the cvs)
will bug him when I'll see him the next time
for now using the tarball
I can send you a tarball
mail it to me
can you receive 560k?
it'll take a while, but yes
do you have bz2?
alex.joni AT robcon DOT ro
ok, it's 250k, on the way
it still says 1.0 but it's from cvs after 1.0
when jepler gets back into town I will bug him about setting up anoncvs
right.. how about emc2 stuff for axis?
he may or may not want to
mind if I send you a patch or similar?
it would be very nice if it worked with both versions out of the box
getting the mail now
(I will not accept breaking emc1 in favor of emc2 but would be thrilled for it to work with both)
it should be very close to working. just an install issue if anything.
the compiled binary works with both
ok.. got it, looking now
hey boys, what should I look for if I can't seem to get g41 to wrok?
probably, but I don't know
did you set your tool?
and you read the manual?
but it doesn't seem to make a diff
many times now
and followed all the stuff about approache paths etc?
it worked fine baq in emc1
then its probably broken :)
actually, it shold be OK, the interp is pretty much unchanged
I somehow lost the gcode file that i had working, and had to rewrite from memory, so its VERY possible that I munged it up
I dont think so, but it is possible
I just cant get it ti recognize tehe offset
* robin_sz thinks its possible
wanna look at the file?
driving me nutz
short trip, but still
* robin_sz is on the phone
I think those "4G native, 8G with hardware compression" advertisements for tape drives are always lies. I got 5.9G on this one.
cradek: that only works for text files
ever heard of those 100 MBit wireless lan thingies?
they actually are only 54 Mbit
and they get up to 100 Mbit by compression
it only works if you send LARGE text files
so, like I said, it's a lie.
well.. it's not for people like u & me
trying to copy a movie
well I got my backup down to 7.3G, thinking it would fit on one tape... Not even close.
don't trust 'em for backups.
a double layer DVD can house 7.3
unproven technology and all that
I recently restored a bunch of tapes that were made in 1985 and kept in the safe deposit box since
* robin_sz passes cradek a bunch of punch cards
a wagonload of them
so in the year 2025, I think it's likely my DAT tapes will work better than your DVDs, but who knows.
How you doing with that setup bpmw_
Not good, cant calibrate!
* rayh knows a bit about punch cards. 2700 in one project.
Have sent paul_c a copy of my .ini file.
I actually have a tape punch and reader here too - I've recovered very old tapes
I've got a punch and three readers on the back of a shelf out there.
Not done anything with them. Thought about recycling the steppers.
I have two readers - one is a crazy thing that reads at *1200 baud*
stand back when you run it!
a ribbon of paper clear across the room.
we made great steps forward in our bridgeport conversion today
we hacked all those cabinets off :)
there was a tape reader in there too
replaced it with 3 geckos :)
it made *significant* amounts of room in the workshop too
the machien looks MUCH smaller
neked comes to mind.
the Baldor rep wants us to try one of his controllers on it too
I'd think the Gecko would be a bit limiting.
why's that ray?
I was able to get about 65 IPM with geckos on a series 1
thats fine for us
the original would move at 120 or so.
The current into the motors in the factory version was about 9+ per phase.
The gecko can only handle seven max.
And there was not amp turn down on the factory setup.
we just need it to make a few parts
speed is not an issue
I really liked the way that the gecko drove the bpt. Nice and clean.
Much smoother than the original which hammered a lot.
microstepping must have accounted for that difference.
cradek: fighting with axis ;)
better you than me
robin_sz: Can you tell me where you got the encoders to work with the Gecko G2002?
asdf-qwee: there are no encoders on the G2002
you mean G200X?
when is that out?
well .. I have one :)
* A-L-P-H-A gives robin_sz a cookie, and sends him off to the short yellow school bus.
you're north american... you know what that means.
I am most certainly NOT north american
robin_sz you're really touched. :)
actually Brits have short busses too!
aj knows what I mean.
what *do* you mean?
robin: I thought the G200x used encoders for feedback from the steppers
it can .. or will be able to
right now it has encoder inputs
but there is code missing :)
actually, the pid loop might be closed inthe fpga
I'm north american, and I know what that means
you can d/l the code from sourceforge
Still, I do know you were quoting some Digikey part number for an inexpensive encoder
so where's _my_ cookie?
asdf-qwee: oh that
US Digital .. 19 dollars
* Phydbleep is in New Mexico where they have the short yellow school busses with the welded on window screens..
My memory may be on the fritz (haha), but I'm not that far gone
[20:54:55] <robin_sz> http://www.usdigital.com/
Phydbleep: Mad Max and the Thunderdome?
And a gun rack. :)
[20:55:20] <robin_sz> http://www.usdigital.com/products/e4p/
Oh wait.. That's over the border in Texas.. Everything in Texas has a gun rack or three.
comes down to $10 in 10K quantity :)
actually, its too low res for the G200X
we need 4000 cpr
robin_sz: what about 4096ppr?
but there are other cheapies on there
Phydbleep: we are sticking to multiples of the pulses per rev
I have some 500count encoders, just need to adapt them
robin_sz: Fine, Roll a couple of bombers and leave them on the side table..
thought 5 o clock is tea time
not 2200 ;)
Hi thomas, you a fellow Canuk
* anonimasu yawns
didn't notice you around ;)
oh, been working
just having a break..
how's the tree thingy?
Waiting for the laptop to charge..
oh, it's good
but, I still need to tweak some code..
do androids dream of electric sheep?
yes, they do
Scottish androids do anyway. :)
SWP_Away is now known as SWPadnos
I thought that was beef
I liked that book and the related movie.
PK Dick, right? (and of course Blade Runner)
There are a couple of books set after the movie - they're pretty good
by K. W. Jeter (or D. W. Jeter)
Heard of em never read em.
They were on the mark-down rack, so I bought them
hmmm.. books :)
what are those?
I have time for man pages and technical manuals .. reading for pleasure is denied to me at the moment :(
I need a bit of help with networking and 4.20
are they like.. floppies?
no - floppies are dead dinosaurs
oh.. sorry for them
I'm trying to reach through this box with another.
right.. you need nat
I thought Irish androids dreamed or British sheep.. Or is that just Tory MP's?
this is 4.18 the other 4.20
ppp0 is the way out.
ppp0 is his default connection to the world
why iptables and nat for two boxes to see each other?
alex_joni: There's an echo command to get it to work.. Hang on a sec.
This box is named henrynet.
Oh. I can see the other with ssh.
<rayh> I'm trying to reach through this box with another.
notice the "through"
* alex_joni is more used to american than robin
rayh: you need an iptables script
that does that
I don't know of any standard debian tool for iptables
echo 1 > ./sys/net/ipv4/ip_forward
okay... going nutz here... need g41/42 help please
ok .. post yer script
Phydbleep: yes, but in order to do smthg usefull he needs NAT
post here or mail you?
just paste it ... its small right?
yeah 20 lines
that way we all get a laugh
lol I'm game
oh .. *game*
either would work for me.
careful... you've not seen me with me spear
N01 T1 G40
N03 G42 D1 F15 G01 Z-.5
N20 X-.7773 Y1.5375
N30 X2.5401 Y1.5375
N40 G02 X3.2224 Y1.0445 I0 J-.7188
I think we may have two conversations going on.
N50 G02 X3.2224 Y-1.0445 I-3.2224 J-1.0445
N60 G02 X2.5401 Y-1.5375 I-.6823 J.2258
N70 G01 X-.7773 Y-1.5375
N80 G02 X-.9961 Y-1.3187 I.0 J.2188
N90 G01 X-.9961 Y-1.0204
N100 G03 X-1.0216 Y-.9227 I-.2 J0
N110 G02 X-1.0216 Y.9227 I1.6466 J.9227
N120 G03 X-.9961 Y1.0204 I-.1754 J.0977
N130 G01 X-.9961 Y1.3187
N140 G02 X-.7773 Y1.5375 I.2188 J0
N150 G01 X-.750 Y1.5400
N160 G0 X-3 Y2 Z1
N170 G40 M0
weyland: That better not be g-code porn... <JK> :D
uh you can't g0 while comp is on.
so, it ignores the comp?
or is it that the g01 cancels the comp out?
didn;t make any diff
I took out the g01
you need a positioning move dont you?
I don't understand
before the plunge?
weyland: You need to home the machine maybe?
are you suggesting, or asking?\
The tool is off to the side for changes, so is clear of the piece
move the G40 up a line.
Close enough for NASA. :)
rayh: I did, no diff
rayh: Did you get nat to run?
see spot run!
I'm making some progress.
see spot do fandango-on-core!
rayh: Firestarter is your friend.
Okay. I'll look there asap. Thanks.
I notice that if I click for tool info, emc thinks that I'm using tool 0, not tool 1 even tho I soecify T1 in the file
And there's Debian packages. :)
and you have set up T1?
in the tool file? yes
well, MDI it
what? just mdi "T1"?
rayh: no change
weyland: What's it NOT doing?
* Phydbleep cant read g-code worth a damn yet.
It is NOT changing the path to compensate for the cutter
that'll be my favorite g-code
could you explain me a thing about limit switch please
tthe pin 12-13-15 in normal state are up to logic level 1 (5V) ?
and when the switch close go to 0 ?
some of the parport pins are inverted
you can define the polsrity
can't remember which
i can see using tck tool ioshow that pin are red (i suppose 5V) right ?
when green shold be 0, i suppose
not sure ... why not try with a voltmeter on the pin?
robin_szi tried, pin are 4,8 v
well, ground it and watch the GUI change then
ioshow are reading right i suppose
well, ground it and watch the GUI change then
if its working, you'll see it change
if its borked, it wont
anyway, i can change polarity from generic.ini file ?
and if it's the parport .. ;) get a new mobo :=)
rememebr to switch the ioshow to the right port AS WELL as the right labels too
I guess what I'm asking is what could stop a g41/42 from working?
jacky: you can change default polarity from generic.ini
i'm using an old isa card 0x278
but that doesn't ensure that the pin actually works
or you limit switches
i also have pci card, but i prefer use the old isa for security now
because no matter if I use it or not in that file, it's not cutting any differently
weyland: and you made the mods we suggested?
thanks a lot
sorry, been baq and forth a bit - which mods?
jacky: no problem, that's why we are here for ;)
errr.. to where? there's a lot of text there
move G40 to the line before?
Yep, did that in MDI and put it in the file too
yep did that too
and emc still thinks its tool 0
go abck to emc1
is it possibly not updating something about tools?
weyland: you broke it
now fix it
got me a .45 and a shovel, here...
should be enough to tame the biest
.45 ... girls gun.
actually... hold on... lemme fire up the old emc1 and test the same file...
weyland: what you want is a pistol chambered in .300 win mag
weyland: You're using emc2?
trying to ...
rayh: yes he is, had a bug in the backlash comp in emc1
jmk fixed it in emc2
I didn'tsee a tbl file with 2
rayh: yes, emc2
it works in emc1, same file
I can confirm.. no tbl file here neither
I saw one specified in my ini
so, I went through the steps to add a tool in emc2
it is specified, but it doesn't exist
I copied one oever
try creating it (or copying from emc1)
plus that doesn't address the fact that emc2 doesn't recognize tellign it a tell change
the m6 t1 thing
I wonder of two is capable
Who wanted to see pictures of the junque-pile vibratory cleaner?
[22:20:34] <Phydbleep> http://18.104.22.168/lathe/DCP_0754.JPG
Need to use round buckets though.. The corners tend to collect parts.
I *lurvvvv* that hi-tech eccentric weight onthe motor
did you plamn that in 3 d cad?
robin_sz: That sir is my 'adjustable' "hi-tech eccentric weight". :)
you add extra washers?
Yeah, I have some that are 1/4" thick, 2" od, 5/16" id, hardened washers from the tractor crates.
make your own out of unobtainium.
best substance not known to man.
unlike cheap chines socket sets
which are made from chrome-plated shittite
hardware, I usually order spaenaur.
robin_sz: You're buying the wrong Chinese crap then.. Find someplace that sells the 'good' Chinese crap. :)
dunno if you people know who spaenaur are.
I buy "Beta Utensli" sockets
and brittool spanners
hmm... time to bore out that hole to close to size, and then ream it. 1-1/8" ID here I attempt (again)
time for bed ...
* paul_c is back
SWPadnos is now known as SWP_Away
SWP_Away is now known as SWPadnos
well... that wasn't ny fun... linux/emc barfed and rebooted while messing with the tool tbl
Gee, I didn't think EMC even supported lathes and here you have it running a bar feeder. 8-)
steves_logging is now known as steve_stallings
tried running my file thru emc1 and it's not responding to g41/42 at all
sure would like to know why
can't make the part i was making just two days ago
weyland: What has changed in the last 2 days?
Bridgeport Boss machines can be quite picky about approach moves. Expect EMC is same.
not sure, to be honest... nothing that I can think of
EXCEPT, that I installed axis...
but that shouldn't have done anything
steve_stallings: i solved with the power supply :) tnx
weyland: I thought you said you erased your program and rewrote it from memory...
steve_stallings: can you see anything like that in the file I postd?
yes, you are correct, but it's a very simple program
and I just can't find anything wrong with the way I';m using it
but to say you can't think of anything that changed...
I've tried everyone's suggestions, and they haven't changed anything about its behaviour, when in fact they should
well,... okay... I'll give you that nit.
it still doesn't explain why I cant change tool numbers
weyland: Thought you were on BDI-4.20
BDI4.20 CVS\d to EMC2
in that case, you can not be running emc1
the bdi-4 branch in emc2, yes.
or even emc2 head.
we installed it to a seperate directory
and didn't "make install"
paul_c: he's using the emc2
but also emc on the BDI
emc2 running locally
did you do a "make modules_install" ?
weyland- Boss requires approach move to be in the compensated plane. Z move will not enable comp in XY.
Also, if you cannot select tool, how do you set tool diameter?
Zero diameter will result in no viable change to path.
paul_c: no, I did not
steve_stallings: I';mn not clear on what you mean in the first sentence
steve_stallings: and I agree about the tool diameter
I mention it as a possible symtpom of the real problem?\
gimme a few mins to compile from SF
I used the branch that John sent me, and is on my wiki oage
Boss needs comp move to be G01, G02, or G03 with no change in Z and it is typically the next block after the one containing the G41 or G42
weyland: you can safely use the emc2 HEAD now
it has been merged
and any new bugfixes will happen there
has anyone checked emc2 for the g54/g0 problem I mentioned on the list (and fixed)?
steve_stallings: AHHHHHH.... okay, lemme go check my code...
cradek: Probably still there.
argh, curse TCP
just copied half a G to find that it's corrupt
try UDP.. then it's cursing time
weyland: I think you might stop looking ;)
I just opened tkemc
and did a tool table load
just tell me you found it
issuing EMC_TOOL_LOAD_TOOL_TABLE (...., emc.tbl)
unknown command EMC_TOOL_LOAD_TOOL_TABLE
can't be good :(
heh - oops :)
so it's not me?
or my code?
not this time
don't think so
nice catch Alex
now GO FIX IT~!!
although .. the code seems to be there
* alex_joni is looking now
weyland: Tool offsets are working in the bdi-4 branch
paul_c: okay... *and tyake this wrong - it's text) but what does that mean to me?
is it my stuff, locally, or this version of emc
does that mean I need to do another upgrad?
bdi4 is the emc that came with the BDI
and if so, from where?
the one you installed by default
from the CD
okay, but even that doesn't seem to work anymore
or is that not valid now that >I've upgraded
then it may be your code
*throws hands up in air)
tool_load_tool_table should work (it's NMLtype 1107)
I can't use thej BDI as it comes because of that backlash issue
at least, it should be recognized
and the NML gets sent out
the receptor doesn't recognize it
so what to do?
weyland: with a little time, maybe alex_joni can fix emc2/head.
weyland: but I think it's his bedtime!
cradek: nah, way past that
which reminds me - where's the instructions to upgrade via the "head"?
[02:48] <alex_joni> cradek: nah, way past that
18:48:11 < alex_joni> cradek: nah, way past that
weyland: the same stuff, just leave out -r tag
too bad you're there instead of here
Im assuming that its jsut a diff directory than john gave me
cradek: see, even by your standards
03paul_c * 10emc2/src/emc/task/emccanon.cc: Fix a bug with the external position update routine - Another one found by Chris Radek.
03paul_c 07bdi-4 * 10emc2/src/emc/task/emccanon.cc: Fix a bug with the external position update routine - Another one found by Chris Radek.
alex_joni: would that include the "P" given me in those instructions?
weyland: If you do a "cvs up -dPA" in your working directory.... (after a make clean)
in other words would I use the "P" or not?
OK - time to go for a walk with the wife. see you guys later
SWPadnos is now known as SWP_Away
most ppl walk the dog.....
quiet - I miight talk
Ooops... just re-read...
alex_joni: would you send me those instructions, so I can make sure not to fsck it up?
im not clear
sorry that was for paul
seems the message from the GUI ends up in the wrong place
weyland: Which directory did you download the emc sources to ?
OK, cd ~/emc2/emc2jmk/src
(we will do this line by line...)
OK, cd ~/emc2/emc2jmk
cvs up -dPA
should be doing loads...
fast connection ?
OK... now for the fun stuff...