hi again folks.
is this right? in tp, during a single time slice that is has goalPos.tran.x, y, z... the x y z amplifier voltages calculated for a servo system are just vectors ( signed magnitudes ) describing this immediate goal position. _During_ this very short time slice we have no checks, we 'point and shoot' ( checks occur after the time slice ends ).
tomp: obviously you cant check stuff faster then the fastest thread on the machine
i was just wondering if during this small time slice the system was unchecked, not worried about thread speed
im going insane.
yes, yes you are
and I hate HF, I hate HF, I really really don't like HF, .....
anonimasu: short drive
tomp: yes, emc produces just one commanded position command per execution of the motion-controller function (typically, 1ms) and checks the feedback position just as often
jepler: thanks, now i have to understand the execution queue, meaning... why is there one ( this q is from a slow machine point of view where calculating future tasks may not be desirable, where alternate destinations may be desirable ... for sinking edm )
tomp: to mill something, you have to control the position of the machine over time while respecting acceleration and velocity limits. one way to do that is to divide time into discrete units that are small compared to the time constants inherent in the system being controlled.
I imagine the same is true of edm, though acceleration and velocity limits are not the highest concerns
cq cq cq dx
killar is now known as klickrr
jepler: i should dig out this edm trajectory planner. in edm you dont calc the new posn based on time & velocity. the new velocity is calculated , based on the difference between desired gap voltage and actual gap voltage ( not actual posn & desired posn). the new posn depends on gap. and disregards any history or trend of velocity. looking for pdf now.
tomp: Well, what's taking you so long?
lotsa crap to wade thru, just in edm ;)
tomp: CTRL+A, SHIFT+DEL =)
tomp: CTRL+A, SHIFT+DEL, ENTER =)
tomp: Oh, in that case.... 350 deg F for 5 minutes
or 30s in microwave
Heh (movie plot), all computers should be stored in microwave ovens. Then when the feds come raid your place, just hi tthe start button (in not automated that is)
GPS w/ touch screen.... it looks MUCH better in person than on their website... $500 http://www.miogps.com/US/products_h610_features.htm#1
very responsive touch screen, and was picking up a signal indoors too.
very nice UI too, much better than my Garmin.
* Fritz wishes for an x-acto knife
Milling PVC on a Taig mill - burrs out the wazoo
overrated, use a machetti instead
Fritz: Oh, green scotch bright bad
or scrape with edge of box cutter
Small internal features...I don't think a pad will reach
cut to size
jepler: jmkasunich: I found the paper (cnc traj planner for edm ) , I added it to the wiki as a link so anyone can download it http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?ElectricalDischargeMachining
i built a comp that decides the new velocity, and it works ok, i think too complex now, but simpler than the PID ( cuz its only P )
the link is at the bottom of the wiki page
tomp: you made me wait for that thing, EEEEEEsssshhhhh ;)
read it yet?
havne't even clicked the link =)
whats taking you so long :)
doing much much important work (scratching my butt)
too much detail (jym)
did not need to know that
You know you want the details, photos coming
Skullworks-PGAB: for you.... 8x10 glossys
03cradek 07nineaxis * 10emc2/src/Makefile: 9 axis sim hal
i have three working stepper motor drivers (step,direction) and i wanna use them with emc. Without a physical parallel port, like using a laptop can be done using a converter? (usb->parallel ie)
like using ftdi chips
not easy from a laptop
there are non-paraport options - but not much workable from a lapto[
might try USBcnc - it uses an earlier version of EMC in a PIC processer driven via usb
runs in windoz
it does all the realtime functions i the PIC so Windows and USB buffering are not an issue
in the PIC
[05:10:20] <Skullworks-PGAB> http://usbcnc.com/
damn i am familiar with atmel avr ... thanks i'll check it out, din't know it
do you know if it's working ok ? it looks very beautif
I've heard good things - I was considering it at one point
but I decided to skip steppers and go full servo
personaly I like his version 1
the guy have done quite a job, he wrote real-time needs of the software in the uc
i would need 3 years to do it on my own
ver 1 had a DB25 output port to hook your breakout board to - like a remote paraport
that's not very difficult to manage, if you can make pcb's at home
BTW you can download schematics and EAGLE board files for them
just need to buy his preprogramed PIC
it's not that expensive, if i need it i'll buyit i geuss
i don't wanna raise entropy ;-)
(bad english i know)
also it's a 4 layer board, it can't be done at home (i think i saw a 4 somewere)
his PC software is free too - only the PIC (or complete unit is sold)
yes i dnl it just now, and looks perfect
thanks for the link again
it's a complete solution, it has lot of input (run, stop) and even more output
quite good !
yeah - I was thinking about a mini machine to take to shows and do engraving onsite
that would work from a laptop just fine
i am an amateur with cnc, i am just building my machine (comparing to u)
I run big iron at work everyday - but I'm building my own small machine as a hobby
:-) nice, and good luck
same to you
when you say small, how small exactly ?
I'm converting a SEIG X3
can you give x * y dimensions? on the working table? i don't know seig x3
thats a mil
isn't it like 12x4
epic wrong maneuver
yes - here is a link to the U.S. Imported version http://www.grizzly.com/products/G0463
i am going for a table router (i don't know how can i say it)
6 1/4" x 21 3/4"
like 16" x 6" x 18"
well i am building a table cnc router with 60cm by 100cm working area
naybe too much for a starter
how much Z travel (height)?
i think router end to working area is about 40 cm
40cm vertical cutter movement?
i am not building it my self, a friend who has the tools to do it (he has tools for cutting inox steel for food machines)
it will be fun
yes, but you wont cut that long, there is a router going up and down
i hope i understand what you ask
no - but its OK most routers are near the same size
(my part is to build the motor drivers and communicate with pc)
i guess i'll find out on the way
getting there is half the fun
laugh out loud
yes but at me ? :-S
no, just in general
not at you.
what are you going to be cutting
and my friend vinyl for signs
vinyl - that USBCNC has support for tangental knife cutting too.
tangental knife cutting?
wait to find what tangental is
[05:45:45] <takeda> http://www.summa.be/pages/nfo_tangentialordrag.html
you mean this
yes - uses like an X-acto blade on a pivit - used to cut vinyl, foam, and balsa for plane parts or signs
also in foam cutting i think
niiice ( <--borat's nice )
basically - not sure how he does it - check the manual
my guess would be requiring a spindle encoder
not for the drag metod
well tell me something, until now (we have all the parts) the machine cost is about 3,500 euros (i think about 5,000 in $)... is that a sane value for a table with that dims?
sounds a bit high - but thats more based on local economy
yeah man, that's approaching a used commercial cnc knee mill
but - you may also have chosen only top grade components
we ordered ballscrews and stuff ftom marchant dice (ebay shop :-S )
NSK liner rails are not cheap
good ballscrews cost too
i dunno if i'd use linear rails in a router
[05:53:15] <takeda> http://stores.ebay.co.uk/Marchant-Dice-Ltd
1900 euros cost was from ballscrews/shafts and all those things...
unless you're using linear rail in a different sense
I see HIWIN rails and other goodies there...
I am looking at ABBA ballscrews as a first choice - and HIWIN as second - only because HIWIN seems to cost more/take longer on custom orderes
everything done now...
your vender has a good selection
i don't remember which he bought
skull wait plz to look at something
check your takeda window (priv)
sorry was away for a bit
i upload a pdf but i delete it again
most of the thing we bought are HIWIN
I think its time for me to sleep - been a long week. - be back later.
pew pew pew milling
it would be nice to have bed mills at work instead of bridgeports
birmingham seems to have some very nice hybrid bed mills
hey Martin, how are things going ?
I have some feed back, but I have to reboot now bb i a flash
Interesting, reading up on GPL v3 and Linus stand on it, I predict loads of kernel code will be released under v3, Linus will refuse to include it in his branch of the kernel and then in 6 months his branc will seize to be the "authoritive" branch
Hi again Rugludallur
hey Martin, you said you had some sucess
How are you?
martin: pretty good, I'm working on the table, with the new MP1000B set up and fully configured (even serial port)
Yep, my tiny filter is tester with part success
martin: but I still get massive HF/RF issues :(
OK, what kind? Limit sw prob?
Martin_Lundstrom: THC locks up, shows wierd stuff on display and then it needs a reboot to recover
Martin_Lundstrom: I'm starting to think that perhaps Thermal Dynamics just has the most overkill HF unit ever made and I got unlucky enough to get it
Im gonna look at some pictures of your machine
martin: there are a couple of new ones in my gallery
I got a few first fotos too, but they are on the HD here
Martin_Lundstrom: hmm you need some web space :)
I think my plasma cutter does not have a HF module that is powerful enough
Sure, web space is nice
martin_lundstrom: I'll try to remember to create an account and put up a virtual host for you on Monday :)
martin_lundstrom: if I forget just prod me while at work
? can you arrange top domain?
that cost money right?
martin_lundstrom: but it's not a top level, it's a .com domain :)
martin_lundstrom: top level would be .martin
and I can't arrange those :)
martin_lundstrom: I get em for almost free
cool, I can send you the money
martin_lundstrom: no worries, keep the money and send me a patch when you are done :)
anyway, I did some more cutting trials
martin_lundstrom: how did it go ?
and there is still some strange stuff going on
good news is that I learn more HAL thins and I understand your schema better and I can do basic debugging
So I seen my filter work by checking signals
but still when I fire, most of the time Z drops in height anyway!
How is your arc start?
Martin_Lundstrom: It would be so nice to have an editor that would look like the scematics and show live signals :)
I cant find a picture of your plasama hose sheilding, do you have one?
damn, freenode got the long nicks
At least it would be nice with a saveable setup in the HAL config window
martin_lundstrom: no shielding on the plasma leads, but I have tracked the issue down to HF
Yes and your problem aperes when HF arc starting right?
martin_lundstrom: the HF is in a separate glass insulated wire though
martin_lundstrom: runs parallel to the leads
You have a HF torch
martin_lundstrom: industrial control (machine) torch
cool, my bullshit pc does not have a terminal for that
? bullshit pc ?
my bs plasma cutter
martin_lundstrom: there is no magic, it's just a dc powersupply with a bit of control logic
Build my own arc starter mmm
I like that
martin_lundstrom: that's easy, even building your own plasma cutter is easy
martin_lundstrom: just need to have safeties and sensors to make sure gas is flowing before current flows
I have some big IGBT modules laying aroung, they itch my fingers sometimes
Anyway your HF wire, glass insulation, what is that? Is it a good shield?
martin: no shielding on it, it would probably be almost impossible to put shielding on the leads inside the cable chain though .. hmm
The wires to your steppers also have seilding?
I guess you can strip it down
and change the HF into something better
martin_lundstrom: yeah, but getting something that can handle several thousand volts and has shielding is non-trivial ..
hmmm, Are you sure? I have a diferent setup
martin_lundstrom: regarding the voltage of the hf or ?
yep, I think a shielded battery cable should work?
what do you think?
martin_lundstrom: I'm sure about the voltage numbers, not sure what battery cables are rated for though ..
hmmm, I dont think theres too many with 10kV rating, but its not for continous use right, the cable cool down etc?
martin_lundstrom: it's only during start but it's a question of the insulation preventing the arc from conducting directly to the table through the insulation
ok, but I think thins cable is what is your problem
martin_lundstrom: It's not my first thought but I'll look into it
afk for a bit, working on the table :)
Another thing you can do to reduce HF is to get a shield cup for the torch, that will reduce the problem somewhat
rayh: good morning
tp.c & tc.c manage the motion elements on the queue, what file reads the struct & creates the new voltages for a servo? (ref: http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?back=Simple+Tp+Notes
from that wiki page: tpRunCycle? is called every TRAJ_PERIOD. It updates currentPos
I rember that working out how everything is hooked up and works in principle is not super hard.
it's the blending code that's non-trivial to understand
anyway cradek's opinion was that a TP with better/deeper lookahead could most likely only be written entirely from scratch
so understanding the current blending scheme is perhaps not required
if that is what you are thinking about tomp?
awallin: i was looking for who executed the needs of the next item on the queue. I 'm not directly interested in hi-speed lookahead, but may need to insert,remove & abort queue elements on the fly for edm.
btw: i did read your explanation and it's very useful, thanks
uh, i thought you wrote http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Trapezoidal_Velocity_Profile_Trajectory_Planner
btw: just fixed the borkd gif link to Leslie Watts' gif
thanks for the wiki fix. Les Watts page was about the only documentation available when I looked into it. simple_tp_notes is what I did a while back
what's a reasonable base_period on a good RT machine??
I traced the code to see joint->vel_cmd is an lvalue only in src/emc/motion/control.c (besides motion.c nulling it( but have not found where that value is picked up, converted to a suitable voltage andapplied to a dac ( mising stg srcs here :(
base period? I dont know, would need to convert that to servo update rate and compare to some industrial controllers
base_period is for step generation with the parallel port
tomp: the TP outputs position. A PID in HAL deals with velocity
thanks, I had a note suspecting PID
the TP just makes sure that the sequence of position commands respects the accel and vel limits. so internally the TP worries very much about accel/vel, but to the outside (HAL->servocard->drive->motor) it does only one thing: outputs a sequence of positions every traj_period
is the PID comp called for this purpose? (not sure all PID requests go thru hal )
all PIDs reside in HAL
so the pid get's the position command as input (TP says go to X=1)
and then the PID knows where the machine is (say X=0.9)
and then it tells the drive to do something to correct this situation
the PIDs can run at a higher rate than traj_period. I believe PIDs run at servo_period in the default setup
thanks, i intend to send the 'PID' ( a new comp that is similar ) slightly different info.
The current PID uses position desired and position actual and outputs position corrected.
It would have no I and no D as they dampen the response to the process.
the new comp has process voltage desired and process voltage actual, and outputs process voltage corrected ( as velocity command to an analog amplifier )
this what how an 'rc-servo' acts
this is how edm works during an individual position command en-queue
ok, I don't know very much about edm
but I guess you need to keep constant discharge conditions for optimal material removal rate
anyone finding gmail incredibly slow lately?
yes (constancy of discharge conditions), thanks
sorry, I dont use gmail
receiving a message takes 10minutes. at work it is instantaneous
* chr0n1c has only used his gmail account once to play around with the google im thingy
kanzur1 is now known as kanzure
03cradek 07TRUNK * 10emc2/configs/common/core_sim9.hal: merge nineaxis branch (UVW changes)
03cradek 07TRUNK * 10emc2/share/axis/tcl/axis.tcl: merge nineaxis branch (UVW changes)
03cradek 07TRUNK * 10emc2/src/Makefile: merge nineaxis branch (UVW changes)
03cradek 07TRUNK * 10emc2/src/emc/canterp/canterp.cc: merge nineaxis branch (UVW changes)
03cradek 07TRUNK * 10emc2/configs/sim/ (14 files): merge nineaxis branch (UVW changes)
03cradek 07TRUNK * 10emc2/src/emc/ini/initraj.cc: merge nineaxis branch (UVW changes)
03cradek 07TRUNK * 10emc2/src/emc/kinematics/ (gantrykins.c rotatekins.c tc.c tc.h tp.c trivkins.c): merge nineaxis branch (UVW changes)
03cradek 07TRUNK * 10emc2/src/emc/rs274ngc/ (11 files): merge nineaxis branch (UVW changes)
03cradek 07TRUNK * 10emc2/src/emc/sai/saicanon.cc: merge nineaxis branch (UVW changes)
03cradek 07TRUNK * 10emc2/src/emc/task/ (emccanon.cc taskintf.cc): merge nineaxis branch (UVW changes)
03cradek 07TRUNK * 10emc2/src/emc/motion/ (command.c control.c emcmotcfg.h usrmotintf.cc): merge nineaxis branch (UVW changes)
03cradek 07TRUNK * 10emc2/src/emc/nml_intf/ (canon.hh emc.cc emc.hh emc_nml.hh emcglb.h emcpos.h): merge nineaxis branch (UVW changes)
03cradek 07TRUNK * 10emc2/lib/python/rs274/ (glcanon.py interpret.py): merge nineaxis branch (UVW changes)
03cradek 07TRUNK * 10emc2/src/emc/usr_intf/axis/extensions/emcmodule.cc: merge nineaxis branch (UVW changes)
03cradek 07TRUNK * 10emc2/src/emc/usr_intf/axis/scripts/axis.py: merge nineaxis branch (UVW changes)
03cradek 07TRUNK * 10emc2/docs/man/man9/.cvsignore: new pluto step docs
03cradek 07TRUNK * 10emc2/src/emc/task/emccanon.cc: don't lose UVW when we do the external sync
sudo apt-get update failed... how do i clean up ?? Reading package lists... Error!
W: GPG error: http://ftp.debian.org
testing Release: The following signatures couldn't be verified because the public key is not available: NO_PUBKEY A70DAF536070D3A1
E: Dynamic MMap ran out of room
E: Error occurred while processing r-omegahat-ggobi (NewVersion1)