* alex_joni just fired up his devel machine...
is it ok to use: cvs rtag -r auto_configure_0_1 -b autoconf_install_alex_1 emc2
keep the tags reasonably short....
the purpose of the fork can be put in the comments of the first commit.
do I need the -r ? or is the -b enough?
* alex_joni is a little lost between tags and branches... :(
just the -b flag
and I can do this from within the last branch dir?
* alex_joni is back..
so to get things right:
cvs rtag -b auto_configure_0_2 emc2
cvs update -r auto_configure_0_2
looks about right - Although....
If you have any uncommitted changes
they could get trashed...
I don't have any
just did a cvs up...
can I put the comment to the commit?
to explain what & why?
in a second...
Evening Paul. How are things there today?
* paul_c curses dependecies that don't show up untill install time...
alex_joni: You're weather turning cold yet?
paul_c : try ncurses dependencies :D
* rayh goes for more coffee so he can get the correct words.
rayh: still pretty ok... around 40-50
paul: the cvs update checks the branch against the main module?
Hmmmm.... You might have wanted the -r flag after all...
can I tag again ?
before I commit?
You'll need another flag....
Nice weather here today as well about the same temp. Saw some ice on the lake last weekend.
cvs rtag -r auto_configure_0_1 -b autoconf_install_0_1 emc2
cvs rtag -r auto_configure_0_1 -F -b autoconf_install_0_1 emc2
I see.. to retag
to Force the re-use of an existing tag
yeah... it's all there ... in the manual.. just got to know where to look (or what to look for)
* paul_c uses the index ;)
* alex_joni wonders what to say in the comment to the commit...
don't think: "don't want to step on anybody's toes so I'm using my own branch" is ok... is it?
I think it's past H
I know the feeling!
Just keep saying to yourself, we will have an improvement worth the effort.
well... that's what keeps me going ...
These abilities are critical to it's success.
paul: I did the commit...
didn't see anything to happen
Hey Paul. How's the new BDI install set? My Knoppix is beginning to do goofy stuff...
I can see a cluster f%$# on the horizon.
rayh: what cluster?
Well It's my fault for trying to apt-get a minor update from debian unstable.
* alex_joni wonders... everybody seems to be talking about debian lately...
I ran RedHat 5,6,7 then switched to Mandrake for 8,9. Then both of these tried to charge some way for their upgrades and such.
I use SuSE...
They also made changes to the pristine kernel and other packages that made a mess of what I was doing with EMC.
I don't really like kernel updates ...
I've got a SuSE box here next to me. 9.1 server.
I like to do my own compile / install
True. That way you can get only the stuff you need.
It's a novel/SuSE thing.
I do like what I see of YaST
yast is pretty cool...
I especially like that they have GPL'ed it
it used to be proprietary...
Although the Synaptic package manager from deb is good.
I didn't know that.
it was... till some point around suse 8.0 I think
or even sooner
What debian needs is a good/helpful configuration system.
I also like YOU
Yast Online Update
I've not tried it.
I see it there.
Seems to me that it brings with it all of the RPM problems.
I didn't need to set it up...
Have you had a good experience with YOU.
it came preinstall...
it just does it's job by itself
I get e-mails from time to time, what has been updated
[04:46:35] <alex_joni> http://news.com.com/2100-7344_3-5175682.html?tag=nefd_top
I've been messing around with SuSE, trying to break things.
I'm at the point of trying to decide between it and Paul's new BDI.
Hi Alex, Hi all
* Imperator_ wants to find out how homing works
Somewhere deep in YaST I changed the cdrom eject settings and now it stays open whenever I'm not using it.
I did dislike something about yast
I am using 8.2
Imperator_: I can try to answer your questions.
and during install it installs from /dev/hdc (which is my cdrom)
but after install it detects the cdrom as a writer and does the ide-scsi workaround to be able to burn
Ok, i want to know if it is possible tolet a hal module know that emc2 wants to home a joint
because i want to write a module for ganry axis
after that yast doesn't find /dev/hdc anymore ;)
and i have to home this axis twice
Oh. That's way beyond my kin.
one for every axis of the gantry
Imperator_: I think some new NML messages need to be written (defined) for that to work
hm, thats the question
but I think a more general approach is to have emc know that it has 4 axes
and emc needs to home them all
then Y, Z
or you don't want emc to know that it has 2 X-es?
that is also the question if emc has to know that, because during operation it makes no difference
and even in free mode that two axis don't have to move individual
well.. you have to define the axes somewhere
I would go with the ini file
i thought or better john thought to write a hal module that does this
you gotta define the axis somewhere
you don't want to have things hardcoded
so my thinking is: you define the axis in emc.ini (or whatever filename)
therefore you will have it in TkEMC
or whatever GUI
If you do define xx in the ini, both will show in the standard guis.
but I'm not sure what names you could/should use...
don't think it'll work to use X twice ;)
We could fix the gui fairly easily.
the gui yes, but I am more concearned about emc
anyways.. let's say you name it X'
you would have 2 different X-es in the GUI
X1 and X2 are the usual names
Imperator_: Does your gantry design permit each X to move separately?
but the operater don't want to know this and he dont want to see two axis
it is not build at teh moment, but that have not to happen rayh
i think a hal module is a dood idea
So both motors must act exactly the same.
I've worked on a couple of machines that had four motors for Z so I understand the problem.
Imperator_: then the HAL module is the right way to go
That would also be my opinion.
ok, then i have to find out how homing works
is somewhere a diagramm how all the files are working together ?
there are some diagrams in the pdf files
IMO you would command a home in X and your hal module would drive both until it saw the home switch and an encoder if there was one.
about homnig too... but not to this level
problem is.. right now the driving is not done from a hal module.. right?
YES !!!!! BDI now has KDE.
up to 16 CD's with KDE ?
:-) thats nice Paul
Just the one CD
got a link?
or not yet...
Got Python-openGL plus a few of the VTK libs
Need to work on building EMC now...
Imperator_: as I see it hal is connected through the ext_ functions to emc
in which file are they
paul_c: This single cd is intended for users. Does it also include devel stuff?
src/emc/hal_intf/exthalmot.c and extintf.h
fsckit... exception error during install...
rayh: There are a number of dev tools & libs included
LyX will take it in to a second CD for sure...
alex_joni: that is the old interface i think
what do you mean?
These additional packages can be gotten by apt-get though.
* alex_joni has got to go home... (still at work at 9 PM is not good)
alex_joni: Slacker ;}
I'm reasonably certain that developers will have their PC connected so that should not be a serious issue.
See you alex.
i think the files in src/emc/hal_intf/.... are not used anymore Alex
slacker = Part time worker.
part time sleeper
Go home and get some rest ''
anyways... give me a link when it's half-way working... I wanna try it out ;)
well.. catch you later... bye
Imperator_: I don't think HAL will help with homing an XX axis
The homing routine is coded to work on a single joint at a time.
otherwise i have to change alot
was machining something with emc just now that shows that velocity is miscalculated somewhere
I was making a rosette...kind of like machining a doughnut with a spiral tool path
given point to point xyz the greater the z component the slower it went
sometimes almost stopping
I think velocity in 3 axis blends is miscalculated
but 2 axis blending works
...this is the original traj planner
I am gonna have a close look at the code...this has given me some clues
the spiral has a very small slope
and I think it is applying the slowest axis velocity to all axes
It still gets from point to point
but with drastically wrong velocity
are the points made by a CAM software ?
it consists only of g1 lines ?
very close together g1 yes
i have made sometimes two test tool path which only move in x and y with very smal g1 lines. if you want i can send you them for testing
xy blends do fine
xyz blends slow improperly
you could time your XY program before and after replacing all the Y with Z
even if the vector is tiny fractions of a degree from the xy plane
good idea cradek
is it anything containing Z that's wrong, or three-axis vectors in particular?
that file i send you are boththe same spline only the lenght of the g1 lines are different
blended moves in particular
single moves in xyz have the correct velocity
of course it is supposed to slow down if allowed accel is insufficient
that is not the case here
if you need test files, say please what you need. I have a very good CAD/CAM software at work
mine are pretty good too
i can also import nc code and maybe turn it with a given angel, or so. but only G0 and G1
what do you have ?
Millwrite, artcam, mastercam
do you have a cnc router there?
i know only mastercam. that is one of the biggest competitor of the one i have
at work we have a Digma 500GC it is a HSC machine www.digma.de and two Deckel machines
I may call Fred on this one...this particular job really showed the bug
making the top (flatter) part of the donut speed was 120 ipm
as it reached the more vertical parts speed dropped to about 30 ipm
and was irregular
can you send me your file ?
I'd like to see it too
my pleasure...it's big...1.3 Mb
email for you and cradek?
send it with dcc please
do not have dcc
what is dcc ?
file transfer over irc
les: your irc client doesn't do it?
hmm I am using mIRC...let me see if it has that
rightclick on my nick and maybe it has "send file" or something
yes has DCC
working on it...
ok it's transferring
be back in a bit
ok now I will try a DCC to imperator
i have chatzilla don't know if that works ???
but interesting to know
I sent a send request but you did not see it I guess
give me an email address
there was the message "DCC Send rosette2.tap"
but don't know what to do
oh...ok....you need to click or say yes or something
hm but how
check some stuff to see
this is pretty neat
did not know about it
have you recived my mail ?
I will do a send request again while you look
let me check..
you can send your file back to this adress
ok will do
ok have it
sorry forgot to zip
this is in inch
z0 is top surface
cuts are negative z
jep i have scaled it by 24
diameter is about 7.5 inches
it starts in the middle is there the speed too low ??
at the outside it is ok ?
no...because that has a raised dome fearure with a flat top
extreme outside is ok as it is only xy moves
because at the middle the angel between the lines is about 20�
maybe emc detects a edge
you have some errors
what do you see
this is an artcam file...
you have some problems with your surface i think
one moment i make a screenshot
it made the parts..just with the wrong spped and feed
the file looks fine to me
I took a quick look and it seemed ok
sometimes repeated points occur
but I didn't see any
still might be some
look from the side to the aereas where you have about 45�
you have mail les
latest (a screen shot?) has not come yet
any second now though
with which software do you have made that tool path ?
it only uses g0 and g1
same as mine
les: on the rounded parts, Z is going up and down
you mean short term?
look around line 3982
let me open it
z-.3158 z-.3160 z-.3169 z-.3139 z-.3156
look at the screenshot
I agree the file is bogus
look at lines 4220-4225
Z is reversing on every line and moving .004
so you're running into your acceleration limit
hmmm I see
let me calculate it...
do you have a IGES or VDA surface ?
it was created in the cam program by sweeping a polyline
since x and y should be most of the vector speed....
look at lines 38182-38189: the biggest component is Z
I calc the z dither observed is at about 4E-5 in/sec
rats word does line numbers on a page basis
must use something else
repeat after me: Word is not a text editor
ahem I see that haha
I was forced to just go to xp because I had to write code running on it
don't think there is a real text editor on here
you can see that problem even in the flash promotion at the artcam homepage :-)
i use context. it is a very good and free editor
let me paste these lines
I think I do not have accel limit problems (calculating) but I do not think emc blends on reversals
even in one axix only...
I have seen that in the code...but not sure whether it is in joint or task space
the vector angle is small though
.004 in .02
>>> atan2(.004, .02) * 120 / pi
er, that's wrong, it's * 180 / pi = 11.31 degrees
you got my screenshot les ???
>>> atan2((2.8179-2.8225), (0.3763-0.3677)) * 180 / pi
the screenshot has not shown up!?
that's lines 38182-38183
what do you mean les ?
haha got it via irc
I sent that one...
the highlighted segment in the picture corresponds to the highlighted program line
what program generated this?
I had calculated only.016 in/sec^2 needed for some moves but let me check this
looks nasty but still very small blend acceleration I think
...running at 2 ips
i think artcam has problems with its surfaces on which it creates the tool pathes. the errors are atarting in the middle like beams
the errors are starting in the middle like beams
yes...I had never used a shape generated by sweeping a profile
normaly that has not to make problems
It's bad...but my machine/emc should be able to plan those motions and run them at full speed
and i saw the same effect at the flash demo at there homepage
the work looked ok due to the tool convolving the path
if i create tool pathes on STL surfaces i have the same problems with our Digma. It has a Siemens 840D controller
I still wonder why emc is slowing and jerking with this
it wants to follow precise the tool path. but with this edges it can do that only slowly
accel required=.03 in/s^2....accel available=15 in/s^2
by my calculations
I can turn it up a lot more...0.5g
make correct tool pathes that makes more sense :-)
better tool paths would be good
if you send me that plyline i can make you the tool path. But i can't make that spiral moves
in IGES or VDA format
it does have an accuracy/ chordal deviation parameter...
it was set to .001
may have to play with that
well I made the parts...it's just that this stuff didn't do the expensive carbide tool any good
chip load too low
it was a one time job
les: How large is the final part ?
about 7.5 inches
A reasonable size for a test file...
looks like quite a few very short segments
I was running it with the old planner
typical artcam file
do you really need 0.0001 moves in Z
you can clearly see a .0001 scratch in polished wood
if you are very accurate you have to sand less
sanding=75% of product cost in woodworking
anyway cradek sent me a screen shot of a blowup of the path and it is full of nasty little wiggles
tha accel reqired to do them is very vvery low
talking of wood....
* paul_c has to go and beat up some MDF
segmentqueue only works with some jobs...
be back in a bit...
this was mdf
$75 is what I charged for 2
oh...1/8 slot drill is the tool
ball end would have been better...but didn't happen to have one
mdf = medium density fibre boards.
usually comes in sheets of 8x4.
uuse it a lot for spoilboards etc
les, if you got an ikea near you, that kinda stuff.
I do not use it for furniture....only solids
is ikea known in the US ?
big furniture store/importer
www.ikea.com :) or was it www.ikeausa.com
afordable furniture (I am being nice here)
half of my furniture are bought from them :-)
ikea in Canada has resorted to luring you in, to eat their breakfast, and shop at the same time.
their breakfast is cheap...
I cannot make afordable furniture
I cannot even afford the stuff I make
I don't like ikea... cheap stuff. looks nice, but I just don't like it.
I feel that it'll fall apart any moment.
it is for $$$ rich people.
same here alpha
in the past it was that way, but now i think it is not bad
helped this girl move into her new appartment. And she had a couch form ikea... I swear, if it wasn't so bulky, I could have lift the thing myself... as you could lift up one side with 3 fingers.
But I do like MDF for fixtures and the like
the stuff I get is usually consistent to about .04 mm in thickness
I've seen some wall shelfs from ikea... they were decent. $40 for a shelf, with the front face as a mirror. It wasn't bad looking... just felt cheap to me still. Particle board.
les? laminant? or are we talking sheet metal?
well heheh...buy my stuff alpha
no just MDF 0.75 in
not a big fan of ikea's stuff. Looks good, feels cheap.
it is accurately sanded
Les, what do you make? got pictures?
ok chiao folks late here
[08:51:01] <les> http://lmwatts.com/gallery.html
thanks for help
les, very elegant work... wanna send me perhaps a 'sample' jewelry box? :D
I think some girls would love me for it...
haha thank you
wanna send me some sample money? :)
Take monopoly money? Canadian money? [they look the same]
the cnc mostly does signs
[08:54:29] <les> http://lmwatts.com/signwp.html
if I ever get around to making the backlash dissappear on the Y Axis (it's right now at 0.020") I'll start doing stuff with it.
my mills is messed up right now.
the lathe works mint though. :)
I am searching for an extra mill
bridgeport interact...something like that
auctions are you friend.
seen a bridge port, go for $700CDN. Series 2.
yeah they are pretty much free if you can find them
was it decent shape too... just that shipping it from the stixs was the issue.
cause everyone's buying a CNC machine instead...
a new CNC machine isn't really that bad anyways.
old cnc with burned controls is very cheap
a ABS plastic free form lithography type machine is only $24,999USD.
shit, I'd take one with burned controlls.
throw some geckos in there, wham. done.
it's pretty easy
A-L-P-H-A: the dude organizing the CCCCamp sells those in .de
sxpert, which the ABS plastic machine? or the geckos?
actually, the lying guys never sent me the information I requested on that machine.
A-L-P-H-A: the ABS 3d sculpting machine
sxpert, lots ofplaces sell them now. :)
from $75K down to $25K in a matter of 2-3 years?
the issue is recapturing that investment... you'd need to charge like $50/hr for how many hours to recapture your initial investment back.
just not worth it...
I put $20k in materials and over a year full time into my gantry
...and I must get $100/hr to run it
500hrs just to cover the initial investment. RE: abs machine. PLUS materials.
really not worth it when I think about it. Though, I still would like to have the machine mind you. :D
About to start a big production run that will make ok money
les... how big is this machine?
does is handle 8x4 sheets with ease?
5x5x2 feet 1000 lb moving mass metal capable
May I ask what's the spindle?
it is removable
right now a 3.25 hp router
looking to upgrade to 5-7 hp
cool... what are you using as ways?
so, moving head? or moving part?
moving head I take it.
[09:06:42] <les> http://lmwatts.com/cnc.html
do you find you get enough clearance with a router? I mean the base of the router won't allow you get deep pockets if your bit is short.
it can be a problem yes
2" steel sq. tubing for frame?
that's nice an solid.
I designed it to do metal but it is doing wood and plastic currently.
frame is a 2x2x1/4 space frame modeled with finite element analysis
but the ways rest on 2x6 solid 1018 steel
how hard was it to weld the frame square?
about 2x6 including the ways
pretty easy...but I have made many.
I use a surveyor's transit to line it up for welding
I've got magnetic 90deg clamps, but I trust those like I trust American president.
no I used welders heavy duty c clamps and braces
hmm... I guess I could always just take a piece of string, and measure corner to corner. :D
if they're the same, it's square.
atleast in that 2d plane anyways
I got the frame good to about 1/32 with very careful tack welding
then it is scraped in where needed to 0.0005
I'm a beginner welder... I can fix metal chairs, and the welds are solid... just not pretty.
I couldn't smack them off with a hammer.
trick with big frames like mine is to avoid later warping due to stress
how'd you measure that over such a large distance?
I have an article on that...let me get it
[09:16:11] <les> http://lmwatts.com/shop.html
link on the bottom of the page
ABOUT LONG FLAT SURFACES
oops hit cap lock
my thoughts to this, could be avoided, if I just bought some preground flat stock. The cost is more expensive, but not prohibitive.
depends on the accuracy you need
that stuff is usually 0.0005" isn't it?
I know preground drillrods are around 0.0005"
if you weld up ground stock two ways might be only 1/32 of coplanar
also it warps loads when you drill way mounting holes etc
and if I just got stock that large? I could probably get the stock up to 8' without much issue. 12' may be harder.
drilling mounting holes could be an issue.
hmm... 3/4" tack it to the frame? shouldn't the size be enough to hold it's shape? [I'm just asking, cause I dn't know]
with the grind/scrape method you can just use ordinary hot rolled
what you can do is grout ground pads to the frame with epoxy
the y axis on mine is that way
but I have not written an article about how to do it yet
it is also 0.0005 flat
You have to calculate deflections...I needed 100lbf/.001in at the spindle and that is what it took to get it
neat way of doing it without a grinding
I learned part of that from old machinery rebuilders in Chicago
takes about 2 days work
but easier that trucking it to a big way grinder
saves a few k$ too...
I must go to town for a min and get some fast food....refrigerator is kinda empty
back in a bit
so how similar is bdi rc46 to whatever the next one will be (rc50?)?
I'd like to get my software to work with bdi, but I have no experience so far with bdi
the man that could answer this, is probably asleep right now. as he's in Europe somewhere.
jepler, but as far as I know... not much of a difference, functionality is there... no major changes in Gcode tha tI know it.
* paul_c is fighting with python exceptions
paul_c: I'd be happy to try to help, though I'll be called away to dinner very soon.
paul_c: /msg me your traceback if you like
It's a case of finding what cased the exception...
Don't you get line numbers in the traceback?
well, dinner time .. I'll drop by later, to complain that I can't get BDI to work.
* paul_c thinks he has the answer...
The last error reads:
TypeError: not enough arguments for format string
part of the string has a % in it...
I'm back. Did you find the problem with your format string yet?
Is having a % in the string enough to raise an error ?
paul_c: "%d %d" % (a, b)
this is like printf
the number of % have to match the number of items in the tuple (a, b)
yes, I'm actually here. Guarding the keyboard from the cat...
Did you manage to get any photos for this new CD ?
I got some good photos of Hal May's shop.
I can send you a few via email
any closeups of "Made with EMC" ?
a shot of one of his clocks
Send them over and I'll add it to the list.
do you want *.png *.jpg ?
I can work with most formats.
it's afternoon ;)
I'm at work since 7 am... that was a loooong time ago
is there any way to delete not used branches?
I hope those are not important in any way... (like some max_branches)
If we ever find out, I'll log a support request.
how's the deb stuff going?
Sucessful install last night..
nice... can I download a image?
It installed, but a number of issues need to be fixed before anyone else can attempt a test install.
just remember to put me on the test-list
A minor problem with the kernel that needs fixing...
* alex_joni is tired/bored...