well, it only took me an hour and a half to get back to where I was before
(although I now have a beefy diode in the circuit to prevent a repeat)
What is the cause of "Cannot change units with cutter radius comp" errors? Does "units" refer to inches vs. mm? Or what?
I pick one and never change it. How am I getting this error?
you sure you're not changing it?
I'll check, back in a bit
G20 only appears once at the beginning. Lots of G2's, though.
can you pastebin the code?
Wait, let me restart axis/emc2, sometimes that fixes things.
do you get a line number for the error?
reload the file..
Yes, "Near line 43 of... concave corner with cutter radius comp". Then "Cannot change units with cutter radius comp". http://www.pastebin.ca/1337607
* JymmmEMC hands jmkasunich a 1,000,000 volt 2Amp Diode =)
KimK_: if you get an earlier error with cutter comp on, cutter comp is not being reset. issuing m2 in mdi may let you avoid restarting emc entirely.
JymmmEMC: I had some 1200V 75A ones ;-)
jepler: OK, thanks
jmkasunich: Well, that's probablt better than using half a bridge =)
hmm, I guess there is such a thing as a "megavolt diode". http://adsabs.harvard.edu/abs/2002AIPC..650...57P
well, at least if it turns out that "Compton electrons" and "bremsstrahlung radiation" are also actual things
I've heard of both of those things
jepler: Sure, 1MV no probelem, just no current though =)
and have a tenuous (and probably incorrect) grasp of what bremsstrahlung radiation is
This might help too: http://imagebin.ca/view/TeOIlQ.html
* alex_joni is getting bremsstrahlung from this laptop
good night all
alex_joni: That's not called a laptop, it's called a luggable
good night alex
crap 4am :/
alex_joni: No way, means I have to be at work that much sooner
alex_joni: other than that, sure =)
today is sunday though..
alex_joni: See, that's what you time travelers get!
a day off?
a day less
I thought it's many nights less
KimK_: by turning off the "start cutter comp" instruction, I see the helix down in a different location than in your qcad image, and then on line 48 I get an error about arc endpoints. I think maybe you should look at how you are generating arcs, and whether you're getting the right centers -- work on a proper path without enabling cutter compensation first, the cutter comp second
dont be fooled by the time travel agents, the future is no more fun than the present
jepler: OK, I will do that, thanks
need 5 digits for your center ij?
Radius to end of arc differs from radius to start: start=(X-6.7765,Y6.1977) center=(X-6.7495,Y6.2147) end=(X-6.7235,Y6.1977) r1=0.0319 r2=0.0311
the error is much more than in the 4th or 5th digit
Is there a handy arc-checking tool in emc2 that I might not know about?
that message above is simply the one shown by axis when I load the program
(or in other GUIs it's the one that I would expect to see when you verify or run the program)
Really? I don't get that here. Did you change the G64 value?
all I did was comment out the G42 line.
OK, thanks again, I will try those things. I'll be here for awhile if you think of anything else.
maybe bigjohnt's arc buddy woul be useful to spot check? else its pythagorus and accounting (to get startpoint if g90)
arc buddy: http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Simple_EMC_G-Code_Generators#Arc_Buddy
(I haven't used it myself)
we have motor rotation!
rather sad that it took all day to get that far
thanks archivist, i'll check it out
jepler: In using axis's single step function, I have noticed that it highlights the current line only while it is executing it, and once it's done, the current line number is gone. Is there anything that can be done about that? Add subtle highlighting that remains after execution? Add a "last line executed: n" or "next line no: n" display in the margin?
beachsurfin2 is now known as ibuffy
good morning at all
Anyone up and around?
Would you happen to know of any stemmer motor driver IC's found in a HP-Printer?
i'd think there is a couple , probably onyl low current
although they might be integrated to a point where u cant get at the pins you need
yeah, I just wasnt sure what their location in the board or their ID's
I am making a prototype system from scrap printers steppers and the hopefully the driver in the printer.
should be easy to ID, usually they'll have 'fins' in the middle pins or some other factor that makes them look like they can handle some power
do u have a list of what IC's are in there?
no, that was something I was looking for
I cant find a schematic or anything for the printersd
u got the driver PCB out? I'd just google the relevant IC's
yeah its right here on hand.
I really just wasnt sure what was on hand, if the the driver system was even in an IC or some other integrated (usesless to access) chips
what is the diff detween an oem650 and oem750? which would you choose for single z axis on small plundge edm?
I have one of each and have yet to see a diffenece between them
JanVanGilsen: one of my pupils found your 5 axis machine somewhere at the Paardenmarkt
It in C3 in between the Brigeport and the Kuka robots :-)
so it's a bit small compared to the others =)
C building... That's the one on the left when you enter through the left main gate?
when you enter, you have walk straightforward until you bump into it
ah, in my days that was the building for elektricity
milling equipment was located in the building next to the toilets
That building is part of the highshool (SISO) =)
Wouldn't <label font="Helvecia" size="20">This is a Label:</label> be a nice way to create a label in pyvcp than the current method?
looks good to me
* BigJohnT wanders off to see if the coffee is done
since I'm goin to rewrite the whole parser =)
can you make the leds centered?
I've done some testing with the xml-schema, it's verry promissing :)
cool, is this for 2.4?
when is it going to be released?
a year or so?
I guess but I have no clue
that should work (we need some time to test it on trunk) and I need to make it first =)
What dou you mean with getting the leds centered?
they are always left justified
no idea, i can look into it :)
[12:15:46] <BigJohnT> http://www.linuxcnc.org/docview/devel/html//hal_pyvcp.html#r1_6_3
The parse wil validate the xml, e.g: It wil print an error if <size> is less then 0, or if size isn't an integer ...
if you leave it out then you get the default?
[12:17:39] <BigJohnT> http://www.linuxcnc.org/docview/devel/html//hal_pyvcp_examples.html#sec:GS2-RPM-Meter
take a look at what I had to do to center the leds in the above example
If you leave out a attribute it can return a default value, you can also have a default value for elements, but thats only returned if the element is empty (not when it's left out)
You can make an element obligated ect.
[12:25:25] <JanVanGilsen> http://www.w3schools.com/Schema/schema_intro.asp
* BigJohnT wonders if there is a way to clear dmesg before starting emc...
It might be better to make the <size> attributes instead of elements (i.e. <led size="20">) and define the default value in the schema...
if that allows you to leave them out then that would be better imho
found it in "man dmesg" =)
I googled it but didn't get far
man was next :)
the led could also be made a single element entry <led size="20" halpin="myled" />
can't you do that now?
that would be one of the goals of changing the parser
now the parser is quite loose, settings could be a attribute or an element, thats against xml philosophy
what is the difference between an element and an attrigute?
I think elements should be used when multiple childs are possible... e.g. <radiobutton><option>One</option><option>Two</option></radiobutton>
so you can have more than one element of the same kind but only one attribute... that even makes sense to me at this early hour
The max and min number of elements can be set in the xml-schema, so it's possible to have a element that can occure only once
That behaves quite like an attribute, but cannot have a default value when its left out
Then its value would be "None", so it could be tested in python if the value of an element = None and assing it a default value...
I don't know whats the best method to use, with the xml-schema you don't have to be a programmer to alter the default values so that might be better
KimK_: I never use "single step" to relate movements in the preview with the source code. Insted, I just click on the line in the code pane, or on the movement in the preview pane, and see how they correspond.
pyvcp feature that I just added it to my todo list <pyvcp updateperiod="100">
it seems to me that 90% of programs with 'dxf support' do no such thing
and also, whats the deal with the gui/interaction design on most CAM programs?
many seem to look and behave like something from 1995
G'morning.... Question on interference
I moved my mill into my basement last week for various reasons. When done, I was leveling my table by routing the 3/4: skin done til all was cut. In the process, my Z plunged without command and tore up an endmill
Just now, I turned everything on and while the computer was booting, the y axis started to jitter real bad
after a minute, it went away, but when I f1/f2 the machine from Axis - the jitter started up again real bad
I never had this problem before. Might this be rf or other interference in the house?
have u tried a mains filter in line with the PSU for the steppers?
The psu is a medical type with the filters built in
a good filter, or ones built in to IEC sockets?
you have me there - I do not know
jepler: Good morning. OK, thanks, I didn't know the code lines and the displayed moves were "buttons" too, nice. Good for previewing and editing. But I still think there should be something to tell the operator what g-code line will be executed next when single stepping through a program. (Sometimes operators are distracted or called away.)
I will take a closer look at the docs on the psu
KimK_: the highlighting in the gcode follows stepping
cradek: Yes, but it seems only to be highlighted while actually executing, then it disappears. Or is that just in plain sim mode (what I'm using for this.)
fixed in 2.3
This is a 2.3 fix?
ah, too slow
Another reason to celebrate 2.3's eventual release. How goes the bug removal?
I think it's in good shape but it's hard to know for sure until a lot of people start using it.
I think alex is testing packaging already
I bet advanced users who know how to build from cvs are already testing
shrdlu-_ is now known as shrdlu-
hm, so I've got this thing raster printing, kind of.
Problem is, since I've had to max out all the acceleration settings to get it to smoothly pass over the code transitions, when it comes to the end of the line and has to turn back, it's not smooth at all
how am I going to get some bits to have a relaxed accel, and some to just 'pass over'?
[16:10:48] <shrdlu-> http://www.youtube.com/watch?v=VDupEuulxgA
I want the head to move like that
shrdlu-: this has more point using a halscope trace
so we can see what's happening
what should I trace?
well.. it would be nice to start with a program
say you have the raster mainly in X (left right), and advancing one line at a time in Y
that's what I have
cept I had to do something weird
because it stops so abruptly, and this cut has some very small sections, I had to put a line at each end of the whole thing
so it wouldn't go insanely fast over small sections
what do you mean by 'start with a program'?
start by writing a program, which you probably have
then pastebin it so we can see what you're talking about
oh, ok. It's pretty big. I'll paste a few thousand
hundred should be enough
[16:34:52] <shrdlu-> http://pastebin.com/d75a67ec3
cam.py allows you to fill vectors with lines
shrdlu-: I notice you have all G0 moves
and you have :"M63P00 M62P00" one after the other
that has no effect
hm. cam.py uses z axis commands to presumably turn the laser on and off, I just did a find/replace
replace with m63 etc
[16:45:53] <jepler> http://emergent.unpy.net/index.cgi-files/sandbox/velocity-dip.png
captured somewhere between line 200 and line 250 running on sim/axis.ini. I had changed G00 to G01, added G21 at the top and M2 at the bottom.
emc's trajectory planner follows the rule that it never goes so fast over a segment that it can't exact stop at the end of it. On short segments, this means it limits maximum velocity. I suspect that's the effect I'm seeing in this trace.
yah, I changed the G00s, but I'd been moderating the speed with 'max velocity' anyway
It does do some strange mini halts in between sometimes
but the main issue is it stopping too abruptly
that's causing the head to miss steps
I think anyway
all stops obey your acceleration setting
yah, but what am I going to do about that?
how can I have it both ways?
what do you mean both ways? (I didn't read back)
well, say I have a bunch of small areas where I want the laser to fire, in order to make it not jump when transitioning
I have to max out the accel settings
get motors that can do the higher accels you need
but when it comes to the end of that row of small areas, I want it to transition smoothly back
these motors are capable of it, it works with my plotter card
I really dont want to issue any stops at all, except when the head moves y up one unit and goes back the other way
what do you mean by "transition smoothly back"?
[17:08:49] <shrdlu-> http://www.youtube.com/watch?v=VDupEuulxgA
all I see is something moving back and forth quickly. it doesn't tell me much
passing staight over the middle seconds, then smoothly turning back when it gets to the end
are you really doing a raster scan and you want each scanline to be constant velocity?
I'm doing raster yes
so you really want one long continuous move, with the laser going on and off an arbitrary number of times during the move
SWPadnos, when my laser does that, when it gets to the end of each pass, it jolts
unfortunately that's not what EMC does. you could modify it to do that though. anything else is only an approximation of what you want.
what would the modifications consist of?
somehow in gcode specify the ons and offs for a move, get that information into the trajectory planner and then do so
alex just wrote this, but with two limitations: one on/off per move, and the on/off is only at the beginning of the move. you would have to generalize it.
can I read off the number of pulses taken or something and just trigger it at the right number?
gah, sorry. I have to go to a stupid obligtory personal training session :/
I don't see how you would do it outside the trajectory planner
cradek: and even in the TP.. it's pretty hard to define how to do it
alex_joni: I wasn't saying it's easy, I'm saying that's a good approach and it would actually give you the desired behavior
you could get an on/off every servo cycle during the raster scan if you wanted it
yes, I agree.. the results jepler plotted are probably the best shrdlu- can get without changing emc2 extensively
maybe it could be done by defining M62 P0 Ixx Jxx
to define turn on/off locations
(or move percentage, or something like that)
yeah not sure how you would represent it in gcode
I'm not even sure it belongs in EMC-the-mill-and-lathe-control
ideally the g-code would only define the areas where the laser needs firing
and the control would scan over an extents of the program , and use the g-code only for triggering the laser
good evening at all
With his laser control - it appears he is operating a 2 axis machine. Could one use the z axis to trigger on and off the laser? the z move could be infinitesimal to avoid acel issues and then the only time accell and velocity changes would be at the end of a scan line? Just a thought
and if he is using all three axes, then add one for the laser on/off switch
bglackin: you still need to slow down in XY if you move in Z
back to an earlier Issue I have - I am getting jitters in my machine from outside interference of some kind. I am using an Astec iMPseries ps. Looking at the specs - its a switching PS (I have seen others say to stay clear of these). ITs also suposed to have low emi
Alex - I relocated the router to my basement - trying to get it back up and running - I am encountering interference issues - my motors were squealing when the cumputer was booting up
PacSci 205 oz unipolars
bglackin: computer booting sounds like parport wasn't initialized
that can be "normal"
the jittering was before I hit the power button on the PC
the motors were actually drifting back and forth
when you relocated, did you disconnect anything? maybe bad connections after relocation ( its usually the last thing you did --old saying )
I wouldn't have my system on before the pc if it causes problems
The location in my basement is below the kitchen (roughly under the fridge) and near an upright freezer in the same room - will these pose possible issues?
tomp - Yes I disassembled the whole unit - I did go through and recheck everything
Eric - I was under the impression I should have my driver board powered up before the PC
control power yes, just not enable, sounds ok
I did run the unit last weekend for 1.5 hours using a large endmill to scab the sacraficail board so the table was level
got a scope? check mains, wiggle connections. (dont cut anything yet )
it crashed by an inexplicable Z move (not programmed) which dove the endmill down and tore it up
I have installed labview8.2 on debian before i have convet .rpm package in .deb package.Exist in emc a library .so for read and write parameter?
No scope unfortunately
I can certainly recheck all connections
bglackin: what drives?
if mains look ok, move down to other ac and dc supplys/transformers, if thats ok check any heartbeats/clocks.
hobbycnc - modified to allow EMC to drive the idle current circuit
check gnds to a good cold water pipe going deep into earth
without isolating the problem, and observing well, its all a bit voodoo
are the computer and control on the same ground?
meaning, "theres things you can try"
I have grounding rod off my breaker box
same as upstairs?
Eric - Yes - I have them on a fused gangbox
we have some grounds in my lab that are at different potentials
"grounds" is what you get when your building is "rennovated" by some mafia types from Pittsburgh
I have a spare slot on my panel - I can rewire a single circuit to that plug
I do run my router and vaccum off different circuits - never fired them up today though
i'd run a cable upstairs and then discard the idea if it didnt make a noticeable diff.
I will go recheck all leads and run a plug to a different circuit before diving off
you really needd to look close at the power sources and supplies, look for 'real gnd', loo for AC on top of DC... best o luck
and at connections
urfh 2am, bye bye
ciao alex, i search the library .so for read and write the parameter, but the emc not use the dinamic library and i thing not is possible comunication labview whith emc.
Most copier sized lasers use Corel to convert to HPGL and send to the laser via a printer driver. Settings in the printer driver (based upon the job) can entail: Speed, Power Level (%), DPI, PPI, Air Assist (on/off), Frequency (KHz) (?), and Error Diffusion (?). I think the laser being constantly on is done only with cutting operations, and PPI with engraving operations. Samples of various jobs and their settings can be found here: http://www.gccworld.co
Alex_joni wins - parallel plug was loose! THanks guys
How could the parallel plug come loose? Cause we know that you wold always securely connect the cables so that wouldn't happen.
acemi: installed your lenny packages
alex excuse for close immediately of yestarday, but in Romania the restourant close fast and i don.t do Dinner for work on maschine.
and I'm missing rtai_math
bglackin: do I get a price?
I have the cables "rigged" for the moment - that one cable is my limiting one - I may have pulled it off then stuck it back in without securing it adequately - sure seems to be the case
Oh btw... I was talking to Marriss once and he said the one guranteed way to kill a drive (even his vampire drivers) was for the paraport cable to come disconnected.
Ok - give me $50
alex_joni: is rtai_math module enabled in the default RTAI config?
acemi: not sure
I know we enable it in the ubuntu package
alex_joni: Maybe it was a motor cable it was a while ago. I'll let you test it to determine which one.
I compiled RTAI with the default
but oddly emc2 doesn't run for me... not sure how/why it runs for you
trying to load motmod fails because of missing symbols
lenny was released yesterday
ceil I think
btw.. if anyone installs lenny.. 4G is pretty close
for a standard desktop install
4g is the next generation of wifi over celluar service at wifi speeds
I meant 4GB
4gb install for lenny?!
alex_joni: is that desktop install?
20:11 < alex_joni> for a standard desktop install
after manually cleaning /var/cache/apt/* it's about 3.2G
+ emc2 stuff -> tiny bit under 4G
(after removing gimp, openoffice, inkscape)
alex_joni: rtai_math is not enabled in the default config
but I can run emc2 without problem
maybe it was 2.2.x ?
I'm testing TRUNK
that's crazy, not even Vista is that bad
oh, vista is way worse
without apps too
Not THAT bad, I have a few VM's of vista
Vista's Home Premium and Ultimate editions both require 15GB of disk space to install, 10 times that of Windows XP.
Right "to install" as in temp file space needed during install, but I mean after that.
[18:19:07] <alex_joni> http://www.nvnews.net/vbulletin/showthread.php?t=85744
"Eh, My windows folder is 23-24GB on my laptop(Ultimate 64 bit). You are fine. Vista is just big."
You're gonna make me vpn into work... hang on....
JymmmEMC: I'm sure a tailored system is less
these are people who leave automatic updates and restore points and all that crap on
I'll be damn... 12.3GB VM image
win7 is supposed to be _much_ smaller
We have a full gigabit network at work, so I guess I never noticed too much.
I have Win7, only played with it for like 60seconds
seems like that raster style ability would open up a lot of machines to emc. China are outputting these cheap lasers with plotter cards by the supertanker load
shrdlu-: Why would you want to use raster on a laser? That has to the slowest operation ever... Rastering a 5x7" image on to acrylic takes like 50 minutes.
shrdlu-: it's probably up for someone with such a machine
JymmmEMC, raster output is a standard for every non industrial laser I've seen
it enables you to do things like print photographs on glass, marble
profile cut acrylic
shrdlu-: You might want to look at the link I provided early. Majotity is vector, not raster.
majority of what?
alex, should I start looking at trajectory control?
I'm happy to learn it now. Since that stuff you added is good enough for the moment
shrdlu-: You are so adamant about using a laser in raster that I give up. Good Luck.
JymmmEMC, I have a desktop laser for art and crafts style stuff. Every similar laser by universal, epilog etc feature raster as a major part of their driver offerings
why is this a strange thing?
probably needs it's own trajectory planner?
JymmmEMC, what was it in the link you posted that I was supposed to look at?
that sounds, daunting, eric_unterhausen :)
shrdlu-: how big is this machine?
desktop lasers I've seen are 60x60cm maybe
I have a vast laser bed of about 22x23cm ;)
cradek: small enough for overnight shipping :P
it was quite cheap. Chinese job, cost about $1200
alex_joni: true. if someone would give me a machine like that I'd write the EMC extensions I described to shrdlu- earlier.
[18:54:50] <shrdlu-> http://www.hxlaser.com/products/chapter/sp.htm
well, after rebuilding rtai, emc2 works like a charm on lenny
alex_joni: did you change anything in the RTAI config
is there a way to pwm those 62/63 codes?
huh, 2.3 version of stepconf looks good
acemi: I used the debian/ folder we used for ubuntu
shrdlu-: here's an idea.. you can use p0 .. p3 for different laser power outputs
the output only enables/sets the power for a pwmgen which connects to the laser output
hokay, great thanks
BigJohnT_ is now known as BigJohnT
hm, heh, the emc site just turned spanish
I can't really tell the difference, written
there are links bottom left for different languages
For turning the power on and off and setting different levels - why not use a spare axis different heigt/rotation = different power and home position =off - unless your trying to avoid gcode altogether
I have done that bglackin, and used the spindle output
Ahh - sounds like a simple workaround
the problem is, in order to get it to raster scan, without slowing down after each move, I have to put the accel at max
there are thousands of tiny moves involved
so it'll stutter without that
but that means when it reaches the end of the row, and has to come back the other way, there's one big jolt
Alex - I just did a grid on my router and am trying to work out the natural skew - I will put my measurements up on pastebin in a few min
which means i can't run at high speed
why not allow it to "overrun" with no power on the laser - it cuts down you effective work envelope but avoids the bump
just have a perimeter in your work envelope which is used solely for accell and decel
I can dynamically change the accel?
max your acell on the laser axis and when its on the perimeter - it should not matter since that axis is "not moving"
bglackin: won't work
Ok - I am a dummy here just throwing a few thoughts out
no problem with that :)
shrdlu-: I don't think there is anything config wise that can make it better
yah, I've kind of given up with tweaking it. I'm going to have to learn how to change things in the code itself, I guess
considering I can barely make sense of the hal files, this may take some time
I lost my EMC icons when I did something stupid trying install KDE last night, so I just reinstalled the package via the sh script
now I'm getting "insmod: error inserting '/usr/realtime-2.6.24-16-rtai/modules/emc2/motmod.ko': -1 Operation not permitted"
hm, maybe it's because I used the new stepconf
shrdlu-: check dmesg
ah, trying to run a new stepconf config with 2.2.8 won't work
gah. weird random lockup again
(not emc related)
mshaver: hi, any progress?
Alex - Was goiong to send a priint to pastebin, but qcad is acting up on me Here are my 4 corner measurements
On EMC LL=3,3, LR = 43,3, UR = 43,21, UL = 3,21
LL to UR 46 5/8, LR to UL = 46-17/32
looks like 3/64's of skew?
or do I need to go back to geomtery?
did you measure the sides as well?
or were those the EMC position readouts?
yes - sides are 18X43
yes and yes
X travel ok and y travel ok - just skewd a bit - better than it was before I moved the router
I'm not sure how the length of the sides can match the DRO when there's skew ...
oh, maybe they can
no millkins yet - used a straight rule from the drilled points
(maybe I need to go back to geometry class ;) )
looks like I'll be alone in spelling though
yep. I don't need that
remedial typing sometimes though :)
brains already onto the 3rd thought past the present typing
This just bites... I have a nice 1U opteron box that has 3x 15K RPM fans in it and screams like hell. It's a Rackable (tm) so custom enclosure/connections/PS so just can't grab off-the-shelf mid tower case and toss it in. argh
Even grabbed a piece of corrugated signboard and mounted a 80mm fan (what I had handy) on it blowing directly down on the cpu. It's 'okey' but just not enough
Looks like I can safely ignore the skew - less than 0.001" per inch on the y axis. Reassembling the machine has made a big difference
good night all