alex_joni_ is now known as alex_joni
Morning...+1 F here today...
about the same here
48 and heavy rain here
well yahoo says currently 0 ;)
les - wb9mjn - did you see these? http://www.electronicsam.com/images/KandT/
the gantry is what has emc2 on it - the k&t is what needs to be retrofitted
(still running a germanim transister controller) ewww
sounds like i ought to run a motenc on the gantry and perhaps use the stg on a vertical mill
unless I but a new one
looking at Haas
[01:07:17] <les_w> http://cgi.ebay.com/Haas-VF0-CNC-Mill-with-4th-axis-board-machine-like-new_W0QQitemZ7567923571QQcategoryZ12584QQssPageNameZWDVWQQrdZ1QQcmdZViewItem
now if I didn't buy a new car right now...
Newer computers do not have ISA ports...
Nice machines...but if they have been used hard for 10 years...might be some service costs...
you can get em. thing is I wouldn't want too
It is unreal how cheap machining centers are going for on ebay - compared to how expensive they where new
your p3 700 may run emc faster than a new box
I think you should put together a MotenC-100 based PC/Controller, and when its working swap out...Then sell the PC and STG as a set....
currently wb9 you are the fastest emcer alive!
125 second servo update
no one has done that before to my knowledge
I imagine the fastest PIII boards are still out there...just...
Oh, I meant WHY, newer computers slower at EMC ?
Ran my square test....came out as close as I can see to the squares I have...so that part is fixed...
because hard real time cannot use caches, piplines, and other statistical methods that the newer boxes depend on
Is there an explaination to that? I have a p III 500 and out of all the computers I have tried emc on - I can get the fastest period on the 500mhz machine. That just doesn't make sence
ah - thanks les
I don't think you can disable that stuff on a p4...so it really bogs down hard real time
There is a place in town here...that has a bone-yard...Tiger ? they keep changing their name....I got the PIII board out of the boneyard 2 years ago....built up the system from
what was the fastest p3?
Maybe 3 years ?
No...It was in the boneyard because the new stuff was faster....
I have some 800 p3 - so atleast that
I think the fastest PIII s are 1.2 GHz ?
This motherboard also had a ISA port, and all the new stuff did not...At that time, the MotenC-100 was just discussions by Abdul on the newsgroup...
So, I hedged my bets...and got the older mother board with the ISA port...
i see 1 ghz
is yours the 8 axis?
Still, good advantage over 700 MHz...
Yes, its the 8 axis board...first one he released....
He has the 4 axis board now too....for a little cheaper...
i'm gonna give abdul a call
from the reseach I have done - the motenc seems to have less complaints than the stg card. Plus it is pci
It will go wacko if the PID's are not close...Once you get the PID's close, and do a clean restart, its solid...
you know your performance is now right up there with a 2500 dollar galil accelera motion card
Then you can tune it into low following error...
actually 4000 dollars for 8 axis
Could be a EMC or driver software issue...
Do not have enough background to know, or figure out here...
well, yeah, servo tuning is always critical....I think that's normal
also, emc does not have a band reject filter in the pid so it is a little touchy
well, a lot touchy
Dave Engvall never got his dialed in close enough, and went back to the STG...I think he has it now though...
I think he has the MotenC-100 working now, though..that is...
Might email him for his experience...
some folks use velocity or voltage amps and mistakenly use the d
which makes for an unstable fourth order system
dave called last week...
he wanted me to test something, but the machine is down for new spindle
would like to see an impulse response of your machine wb9
need to do it with an accelerometer though...logging perturbs
[Global Notice] herbert Hi all. The server you're on, herbert.freenode.net, is experiencing routing problems. We'll need to shut it down. Please reconnect to irc.freenode.net .... thanks!
Lost it there for a bit...
Looks like a nice machine...Too bad its only a day away...no time to fly out to AZ and look at it...
Kinda allot of money to buy something not inspected...Look at his sales...Only small stuff ...Yes..the Haas...
yeah that one is too far for me
plenty more around though
Doing laundry...back in 5...
Got booted that time...
They must be maintaining something...
[Global Notice] (Oops, ignore previous message. :)
those Haas machines are about 40k new
They have a wide range of sizes/options....
the mini mill is nice too
price is a little too high on the ebay ones
more than new price at the time
$15k or so would be more in line
they are 4 axis though
I just need to enhance the prototype capability
I am finding I need to make multiples of things
typically runs of 5 or so
burdensome on a manual mill
Probably the smallest machine with a tool changer ?
and a 4 th axis...
yeah I think so
10 hp and 7500 rpm would do some pretty good work
don't need much more...not doing manufacturing
well, who is...in this country
A few...but fewer daily...
however I am doing well with development and prototypes
...waiting on that ups truck still
i'll go make a late breakfast
quick question - Is the escape key suposed to be an estop?
gosh I don't know... my estop is hardwired
I only use the keyboard for homing
No...that an abort...
Its the F1 that is Estop...
Just wondering as I don't yet. when I hit escape - the machine instead of stoping it deaccelerates
good - thanks
I like the escape for an non panic stop - as I would not loose steps.
now that I know f1 is an estop - will have to try it
Thought we were having an earthquake....but they are digging up the street closer to the building now..were a block down...now they are about 50 yards down ....
blah. i'll fill out nov invoices while i'm waiting
let's see....grossed......36,600...not bad huh?
I really out to use some of that for a VMC.
better than giving it to the feds.
who knows cvs?
not I...other than lookng at code on a windows box
I am trying to install the lerman-interp branch
jepler gave me "cvs update -r lerman-interp" which seems to download stuff ;)
yep, that's the trick
yeah, I would like to try that. Does ken have documentation?
but then when I do a ./configure then a make clean - and a make it doesn't work
have you previously built another version?
actually it stalls at the make clean - something about a directory not existing. So I deleted the emc2 directory and re cvs-ed it
you may need to cvs up -dP -r lerman-interp
otherwise you may have any of lerman's new directories missing
thanks - I will give it a try - sourcforge seems to be a little flakey this morning
hey - gradek - what will axis do with programs containing subroutines?
credek - sorry
there we go
skunkworks: it's not tested but we think it will work
;) I think the -dP wroked - thanks
we're working on building it here too
skunkworks: Didn't you have a problem with jogging? axis reversing at the end?
chinamill was here earlier asking about the axis reversing at the end of a jog
told him I thought you had experienced that and thought it had been fixed
Yes - still do - this is with steppers - there seems to be an issue with the stepper code.'
oh, ok thanks - must be thinking of something else.
probably the following error at the end of a jog - is that fixed?
jmkasunich would be the one to ask - I think he understands it better than me. ;)
better than me, too - just trying to make sure I give out good information.
jtr is now known as jtr_away
[03:15:29] <skunkworks> http://solaris.cs.utt.ro/irc/irc.freenode.net:6667/emc/2005-12-02.txt
here is part of it
I was asking about acceleration
cradek - I seem to have the lerman-interp installed - I can run do-while loops and such. I will now install axis.
ah ups came...sack of thomson bearings.
has anyone used the editer inside emc1? it definatly seems to have some querks
in the bdi 4.3 install
the find and replace doesn't fine and replace everything. Plus the find and replace window closes when you click into a different field.
axis does work with the lerman-interp
I will have to try some of my old programs from turbocnc. - does it actually predraw?
and the interactive preview works right (clicking on a move in a subr highlights all the moves done by that line)
that is pretty cool
I could say cool one more time ;)
what the hell. go ahead...
Cool. I've been stalking. I've seen that I've been requested to merge my stuff into the HEAD. Unfortunately, I have not the foggiest idea how one does that. (As I've mentioned before, I am not a CVS person.) Is there a web page or tutorial somewhere that would tell me how to get two versions, compare them, and merge them? I'd like to get that done ASAP.
lerman: I have always been able to get by with man cvs
lerman: the merge of a branch is semi-automatic if you get the incantation right
hey jepler - this command worked for me to get the lerman interp. cvs up -dP -r lerman-interp had to add the -dP
skunkworks: glad you got it to work.
A quick look at the man page didn't show me the word merge -- except for rcsmerge.
lerman: It's done with update
lerman: cvs up -A; cvs up -jlerman-interp
[03:50:41] <jepler> http://cvsbook.red-bean.com/cvsbook.html#Merging%20Changes%20From%20Branch%20To%20Trunk
lerman: you will probably have to manually fix any merges that fail
So, what directory do I want/need to do the cvs up -A ... in.
lerman: do all those commands from the top level
lerman: those commands work in the current directory and subdirectories
lerman: are you sure you want to do the merge right now...?
or are you just going to practice and not check it in on head?
are there any driver examples in emc2 for cards which do firmware pid, and take a target position? i have a pair of isa motion controllers, and am debating buying the driver source for them and writing a driver, or putting the cash towards a new controller board thats supported
Probably not this minute, but I should practice first. Also, when would be a good time?
question - what am I getting when I get the lerman-interp? Am I getting a whole differnt build of emc2 or just the interp and support?
lerman: that's a good question
All I've done with lerman-interp is the changes to the rs274ngc directory.
skunkworks: but if there is a fix on HEAD after lerman's branch, you will not get it when you check out lerman-interp
skunkworks: at least that's what I think you're asking
The real merge issue is what has changed in the HEAD while I diddled the interp.
cradek - yes
That's why sooner is better than later for the merge.
lerman: right, but I bet the rs directory has not changed, so it'll be easy
So you should get that merged ;)
I wonder if we should tag head before your merge
lerman: let me send an email to the rest of the Board about this merge
lerman: we need to pick a good time
When I asked what directory, before, I meant, does this need to be done from a tree with the head in it? And how do I get that? (If I don't have it in writing, I forget it, and I didn't write that down.)
like I said before - I have only played with emc for a few week. It took me 2 hours to figure out the reason why conditional statements where not working is they didn't exist. ;)
yes, you check out head with all sticky tags removed (cvs up -A)
then in that same directory you use cvs up -jthe_branch
Well, I suspect that almost anyone could do the merge better than I can -- because it is all CVS stuff and not much else.
btw - nice work lerman. Is there a standard gcode model that you used to to the loop/conditional stuff? Is it online?
I doubt we have anyone who could do it without also wading in the documentation
The wiki has a page name something like AdditionalGCodeFunctionality that is the 'spec' for the changes.
Yes - I have read that. it doesn't really give syntax - I suppose I will just use your sample gcode files.
Remember that they are interpreter test files -- not code that is to be run.
right - I just need the syntax of how the lines are constructed
It looks pretty strait forward - thanks
lerman: my dry-run of the merge worked perfectly
modified docs/AUTHORS, added nc_files/interp*, merged a bunch of changes in rs274ngc, added rs274ngc/interp_o_word.cc
lerman: I sent an email to the rest of the board asking when we think a good time would be.
cradek: you said something earlier about emc documentation in lyx format. where is it? I don't see anything in the emc2 cvs tree
it might be under emc/docs or something similar
Well, I just did a merge. And it seemed to work. -- although the following is disconcerting:
Merging differences between 1.8 and 126.96.36.199 into interp_internal.hh
cvs update: move away `src/emc/rs274ngc/interp_o_word.cc'; it is in the way
I've got to go to a meeting now. Later...
yeah, I didn't get that at all
the C means you had a local change
I assumed that the Board had already discussed when, because the minutes show:
Motion by JMK:
that we contact Ken Lerman and ask him to move the 'O' changes into HEAD, offering him as much help as needed to do so
Second by Ray
JMK, Ray, Alex, Jon, Cradek: AYE
But, it appears that I'm ready to go. I just need to do a 'commit'.
did you fix your C (changed) file?
03rayhenry * 10documents/fest04.txt: removed an obsolete file
lerman: let's tag before and after your merge
Dumb question... What does stepover do/allow?
does the question make sense?
are we talking stepover where you have lets say 2 parts - they are identical except one gets and extra hole. You can you the same program you just step over the extra hole. I don't know how it is implimented in emc though
No, in respect to toolpath... if you are using a .25" diam bit and you set the stepover to be (lets say) 0.125"
I know it's overlapping the tool, but what do you benefit from doing this besides a better finish?
In this case it's 50% overlap, but when do you want more, and when do you want less is what I'm asking.
toolpath is some piece of software you're generating gcode with?
I guess it's an obvious name for a cnc-related program. chris started and we both worked on a dxf-to-gcode converter called toolpath, but never released it. I didn't know there was another one.
oh, no. This is generic CAM question I'm asking... Like stepdown as an example
[05:01:04] <jepler> http://unpy.net/~jepler/toolpath.png
stepdown is how many passes to make using a specific tool to get to the desired depth.
jepler 3d cad?
Jymmm: no cad, just dxf to gcode conversion. you have to make your dxf files in some other software.
jepler ah, cool.
are there any advantages/disadvantages to using emc2 to do pid control loop over using an external pid controller?
rolf - where you the one that used to work for kerney & trecker?
oops - sorry - it was roltek - nevermingd
jepler: Does it work ("Toolpath") ?
Jymmm: it works, but it needs polish
jepler: Do you select tools in it?
Jymmm: among the things you tell it are the tool diameter and the shape of the end (ball, vee, or flat)
jepler: Ah, cool. Can you selct/choose/use multiple tools on the same paths?
Jymmm: No, we hadn't gotten around to adding that feature.
Jymmm: you could create several outputs from the same model and different tool configurations, though
jepler: It was more a 1/4" to rough out, and then a 1/8" to get into the details.
without duplicate work.
No, the program isn't that smart.
it'll duplicate work
delacroix: I don't think there are any drivers for cards that do onboard PID. If you can get the register set for your card, it should be possible to make a driver.
delacroix: also, the only advantage to running the PID in hardware (assuming that it's an FPGA or DSP) would be speed, but the average PC processor is pretty fast these days
swpanos: the issue i have is that i got two pid controllers dirt cheap off ebay, but its going to cost me to buy the driver source under NDA from the manufacturer. am just trying to make sure there are no major obstacles before i pay for the driver code, as money could go towards a new card
Jymmm: did you actually get your "stepover" question answered? (it looked like the conversation changed midstream)
do you know what the hardware is? (DSP, microcontroller, FPGA, analog hardware)
SWPadnos: Not really. I understand the benefit of stepdown, but not stepover so much.
Jymmm: ok - one sec and I'll tell you everything I know about it ;) (this much > <)
swpadnos: the card is a tech80 5650a and its dsp based
do you have a link to some hardware specs?
ok - I found one at ServoSystems.com
oh, u beat me to it :)
the source license has got to be expensive ;)
I don't see it listed on the Servo Systems website (I only see the development kit)
its $100 admin fee + signed NDA
hm - that's not bad ;)
the devkit is $595
so not too bad, considering the boards cost me $40 for 2
there will be one problem though - NDA + GPL = trouble
i know, i am unsure of the licence implications of using their drivers
problem is all i have to go on is a chapter of the manual on low level interfacing, and its missing loads of info
that's not the only end you have a problem from ;) You can't use any portion of the GPL source code for a closed-source driver. you may be able to use it yourself, but I wouldn't expect any help from others in fixing it (in fact, the NDA would prevent you from showing your code to anyone else anyway, wo you'd really be on your own)
if you have register information, then an open source driver can be made
i have register information, but am not sure if its complete, have docs for the cards io map, as well as the dsp which performs pid
have found one person who wrote linux drivers for this card like 5 years back, and am hoping he will let me get a copy of his code
as he didnt use the nda source
Jymmm: that would be ideal
no - not Jymmm , delacroix
well lots of people signed on...but pretty quiet!
shhhhh -we're sleeping
happen to know how to get emc to log
not off the top of my head
ie to write to emc.log as Ken did for his timings for the interp
(I suppose I could just have said no, since emc2 doesn't have logging anyway ;) )
it's on the gui
it writes a random named file
no ... that is logging....and does different things...
Although I often have problems with writing garbage
that one i can make work...
I just checked my emc.log file...in emc dir and it is 0 length
also it seems logging perturbs the rest of emc a bit
well i've been lucky I guess...my logging seems to work most of the time
do you want logging, or debug messages?
oh dave...yesterday wb9 ran with the motenc at 125 microsecond servo!
seems it blows away stg
I wonder what he can do for axis speeds?
only about 250 ipm
yep...but you can get 400 uS ?
400 is the best we could do with stg
stg is ISA, right?
that is about my experience.
that is for all axes (I think)
yes ... stg is isa
ok - that's the bottleneck
seems to be
I think I had better go to motenc...if everything works
does it dave?
Don thinks that the wider bandwidth make tuning more sensitive
like index, etc?
yes index works...
Don's system and a couple of others are working...
I've had problems but will try again soon.
sure it would be more sensitive....emc has no notch filter
You might check with Don to see if everything you need is operational.
I could retire the stg to a vertical mill
unless I buy a new Haas!
I'm working toward putting the vital board on a Cincinnati tracer
Well the haas would certainly do wood nicely. ;-)
I just need another mill in the prototype area. Trouble is I have little time to convert anything
I need this for metalworking
big chunk of my income comes from engineering development/prototyping
I had limits and homing working on the vital but motion was a problem....and I don't know why...
how much metal do you have to remove for prototyping?
so I need to throw some money at that end of the shop
well the haas is a good machine just not as stiff as the Mazak
Also...I think spindle is direct drive not geared.
Just typical stuff...last job made a trash bin of aluminum chips
and some 01 tool steel chips too
I am finding I need to make several of something
well O6 machines better
more often than not
do you generally design in CAD, then go machine from a print, or go machine something, then try to duplicate it?
O6 machines better than O1
I draw it up....customer usually needs docs
ok - just wondering if conversational CNC was something you'd need
what do you use for cad?
just autocad mostly.
I'm on autosketch now but planning on going to synergy
only 2.5 D
The cnc router is metal capable with a spindle change
some cut aluminum with the colombo but I don't think I will
a set of bearings for the thing is 800 bucks
so anyway, yeah I need to get a small vmc in the toolroom area
soon it will have it's own space
you have 3 phase?
or use a converter
I use a large rotary
can go bigger if needed
mine is supposed to start a 30...and run 80 total. I've never pushed it
It does blink the lights when i fire up
the smaller Haas is 10 hp spindle
that would be plenty
5 would even do.
even 5 hp at the spindle will do a lot
I rarely go over about 2.5 demand
I need decent rpm too
that is steering me away from old BP boss etc
what do you consider decent
that pushes the limits on a lot of machines
yeah I know
I think the Haas mini will do that
can't always get what you want!
it is only $$$$$$$$$
yes the Haas can
I need too long a reach to use many small mills
[06:53:02] <les_w> http://cgi.ebay.com/Haas-VF0-CNC-Mill-with-4th-axis-board-machine-like-new_W0QQitemZ7567923571QQcategoryZ12584QQssPageNameZWDVWQQrdZ1QQcmdZViewItem
datron dynamics, baby!
[06:53:32] <SWPadnos_> http://www.datrondynamics.com
- the high speed milling people
(I think you've seen the video before)
it's in the same price class as that Haas on eBay (within a factor of two anyway ;) )
the one I showed is a little high
has 4th axis board but no rotary
clean looking though
gotta run...catch you later
7500 pounds - probably $3-5k in shipping
Thoughs pic dont look 1995
I noticed that
Well, maybe TAKEN in 1995 =)
even the controller kybd is perfect
so...7500 rpm 10 hp ought to be fairly fast compared to a BP
yeah - those look like photos taken at delivery, not after 10 years of use
am a bit confused, have been following rtai-steps from wiki, and have gotten an error whilst configuring rtai that the kernel src i have provided does not have hal patch applied/enabled. I applied the adeos patch, do i need both?
10HP helps, compared to 4 (max on a BP, I think)
looks brand new
delacroix: what distro are you using?
back to the shop for me. Later
ive compiled a new vanilla kernel for suse 10
188.8.131.52 with the 2.6.13 adeos patch
and the patch applied cleanly?
no errors while patching
have you rebooted to that kernel?
hm - OK. can you run the RTAI test programs?
no - of course you can't
are there any settings in xconfig i need to do for rtai other then specify my kernel source path
probably - I'm checking now
so - you got through step 9 with no errors, and step 10 is a problem (making RTAI)?
just to be sure, you *did* enable ADEOS when configuring the kernel, right?
erm, better check
just a sec
is it meant to show up in make xconfig for the kernel?
cant seem to find it, should be in general right?
How does the home work - if all axises have only one parrelel port input? does it run them all home at the same time until one of the switches is tripped then backs each axis off until if figures out with axis. or does it home each axis one at a time.
with only one input, you'd have to do them separaately
delacroix: it's just under "Processor Type and Features", in the top-level menu
if you don't see it, the patch didn't work correctly
(not in the Processor type menu, but under it)
dont see it, something must have gone wrong with patching
ok - try that again, and if it doesn't work, then download a stock kernel from kernel.org, and do it again
i am using a stock kernel from kernel.org
I thought it was a suse RPM or something
kernel is 184.108.40.206 from kernel.org patch is adeos-ipipe-2.6.13-i386-1.0-09.patch
I ws looking at the 2.6.12 version - I'll download the ones you have and try those
lerman interp installed ;) http://www.electronicsam.com/images/KandT/Lerman.JPG
delacroix: I just noticed on the wiki page that you're supposed to use the RTAI HAL patch, not the ADEOS patch
(at least, there's no ADEOS patching in the instructions - I guess the HAL patch includes ADEOS)
I thought someone said recently that HAL is not needed
i saw that, but hal patch is the old way of doing things, adeos is the newer more supported interface
dunno - I've only used BDI pre-built kernels, or ADEOS
this is about where I got tripped up the last time I tried to build the RTAI kernel
I've only used rtlinux
I'm sad that our documentation is SO bad
yeah - it's hard to keep up, but it's especially hard when you start from almost nothing
bdi has probably allowed us to be lazy about this particular part of the documentation
true, but the wiki was actually started after BDI (4.20 at least) was released - anything there should be at least from the last 7 months or so
(I'm not sure if it was started before or after Fest)
the rtai-steps was last edited a month back, but there is a typo on the patch line, and its missing a call to depmod before mkinitrd
delacroix: what processor are you compiling on?
sorry, was afk
the machine is an athlon
have family set as athlon/duron/k7
I see referenced to lathe functionality in emc - is that something that is being worked on?
paul has something working I think skunk
on a boxford ME10 or something
les_w: Any particular clearcoat that does NOT absorb stain better than others?
hmmm you do not want it to absorb dye?
is this pigment or dye?
les_w: No, actual wood stain.
hmm minwax....let's see
prob a pigment stain
MinWax brand - oil based wood stain
so you want a basecoat that does not take the stain?
les_w: I know it'll absord some of it, but what it does now isn't very contrasting.
and you want it to be?
I put on 3 coats of PU, and it still absorb much of the stain combared to bare wood.
stain is over the 3 coats?
Well, I rubbed the stain on by hand
after 3 coats?
PU, Carve, Stain
les_w make sense?
I'm using the PU as a barrier
ohhhh you are staining the letters dark....
finally. I'm slow today.
lol, tis ok, I'm slow all year
ok I would make a stain from thinned latex paint
thin with water
till it's like milk
how is turkey call boy today?
les_w: So thin out colored latex paint and use like a stain?
right Jymmm...that's what I do on most signs
turkey boy is ok
turkey boy needs a vmc I think
hows the project? cluttereed with patent lawyers yet?
lerman - Great interp. http://www.electronicsam.com/images/KandT/Lerman.JPG
les_w: Ok, will give it a shot. One more question.... PU or any clearcoat.... what can I thin it with enough to be able to dip coat some really tiny pieces?
I need to up the productivity of the metalworking area
hey Jymmm, made any parts on your router yet?
les_w: me too!
jymmm....thin with high flash naptha
Hey, we aim to please :-) . Thanks for being the beta test site. It's time to get this stuff merged.
les_w: had a zero productivity day today ... blew a lens
les_w: size a dollar piece, somewhat thicker, pyrex with a fancy coating ... guess how much?
les_w: "high flash" (?!) napatha
les_w: I'm not making pyrostech stuff here ya know (wait till March)
robin_z: Hi, yep
yep cut parts?
[09:19:43] <Jymmm> http://static.flickr.com/33/61765799_f610fa36cb_m.jpg
* Jymmm is off to the store.... bbl
not bad ...
high flash means that is it less combustible! high flash pint
So...I need a vmc robin...not just a cnc BP
Lots and lots of time is in front of a mill making 4 or 5 of something
buy one ^_^
costs about the same as a new Audi....gulp
? vertical machining center?
les_w: you could buy both couldnt you?
yeah. buy both and pay tax....not sure!
How come people like vertical verses horizontal. - with the k&t being horizontal - chip removal is less of a problem.
No reason really....they are just out there.
Looking at the inexpensive Haas machines
make sure you get a _good_ example
some haas are supposedly horrid..
I've seen lots of horror stories about the machines not keeping within spec
ways? spindle? control?
quality in general
sounds in the gearbox..
I'll have to check that. The price is very low...
not making ballbar tests within spec..
bit I've bseen them about better machines too
I'll research that
although most posts I've seen are at cnczone but, check it out
I would much rather do a conversion on an older machine
but may not have the time
I like to do that
hard to justify in business though.
once you have a cnc a retrofit is easy to build
I would have a lot more free time if I put some serious power in the metalworking area as well as the router
hey les_w, /msg me .. I have a photo
I need bigger everything basically
robin...not getting messaging?
robin_z is now known as robin_sz
how's life robin_sz?
blew a lens on the laser today
$1400 up in bits
what kind of laser?
co2? how many watts?
a giant "laser"
we have a 1000 watt laser here - I "blew" a lense in it by accident. tried to cut some metal without the assist air on. the sparks pitted the lense oops
that was maybe 10 years ago
oopsy .. gone
so.. how's family robin?
fine ... but a bit stressed
[10:33:50] <robin_sz> http://www.electricstuff.co.uk/uktesla2004.html
lots of work ...
children being err, well, children really
assuming that gettingup at 6am is consdiered normal, I guess they are normal
* alex_joni has been sick lately
so catching up on work
anyways.. going to bed now
skunkworks: so you play with lasers too?
We have a cnc laser at my real job. PRC
oh, yeah. me too
use it for cutting steel rule die boards and plastic
a die-board machine ...
I have one somewhere at the back of the shop
funny - lost back there?
as you say, good for wood and plastic
right - we have cut shim stock and maybe .06 steel but that is about it.
im always amazed at the amount of tar-like residue in the extraction from cutting plywood
have any fires?
that is exciting
not recently ... its been unplugged for a year or more
what laser do you have now?
so ~ same as 1500W CO2
and a 1700W Optomic
in "need of TLC" :)
ah - I have been out of it for a while - trained the die shop how to run it and let them go. This is a 1000 watt PRC laser.
1000W for dieboards?
thats a lot for wood
lets us run about 30 ipm and make a .028 cut
It is I think rated 800 but we get 1kw. 3/4 inch die board
all the way though
all the way through (threw) whatever
the 800W yag cuts 10mm steel just fine
right frequency ;)
well .. it did till it blew a lens this morning]
sparks? something got on the lense and it went poof?
it was not the material side of the lens, but the beam delivery side ... so something dropped in
the lower lens is OK
its the upper thats fuxxorred
"doublet" lens, a normal convex lens, sandwiched to a plano concave
[11:52:23] <skunkworks> http://www.empirescreen.com/die/laser.jpg
quite .. err
[11:53:36] <robin_sz> http://www.rapidcut.co.uk/images/LY2500.jpg
Have not blinded anyone yet ;)
"caution laser .. do not starte into beam with remaining eye"
This one the table is what moves - the laser is stationary
same as my ferranti
[11:55:50] <robin_sz> http://www.hw.ac.uk/mecWWW/amu/amu9s.jpg
same as that one#
aww fsk .. .now this is funny
AXYZ .. the router/cnc maker .. claim to have a CNC laser ...
now, what I want you to bear in mind is that the ferranti MF series has a very very long folded tube, and HATES vibration
its ratehr delicate and not happy when shaken
so .. with that in mind ...
[11:58:44] <robin_sz> http://www.axyz.co.uk/CNC-laser-system.html
spot the idiot :)
the tube in our laser is about 4 ft long - with a blower
this is 30m long, folded
slow flow, no blower
the YAG is solid state
I remember that term - slow flow. boy its been a while
means no big fan to go wrong
blower change at 20K hours
right - about the time I got out prc was trying to make a solid state powersupply for there laser - they ended up having problems with them (ours is an onld hv transformer and tubes for control)
had it replaced a few years ago
yeah, theres is a DC discharge thing IIRC
the Trumpf is just some arc lamps and a crystal
no nasty gassess or blowers
right - I always wanted to build a ruby rod laser when I was a kid. ;)
big krypton lamps and a ruby
you wont want to see this picture then ... ;)
too damn cool, found a zip100 USB at the thrift store for $3 and can get my autocad stuff off the old zip disks! YEA!!!!!!!
* K`zan does happydance!
am having problems setting up adeos still, have downloaded 2.6.14 from kernel.org, and applied patch adeos-ipipe-2.6.14-i386-1.0-11.patch, but still dont get an adeos package listed when i do build xconfig
do i have to do anything else before it shows up?
sorry - I got sidetracked looking at that
np, you have already been a great help with this
all the options in the patch are CONFIG_IPIPE, rather than CONFIG_ADEOS
it looks like it should add "-ipipr" to the name of the kernel if it works - what does uname -a give you?
ipipe, that is
it does change the extension to -ipipe after patch, but no extra options are added to make xconfig
unfortunately, I have no experience with this - I'm afraid I can't help you much (until I have time to learn about this)
uname -a gives the kernel i have atm, which is 220.127.116.11-adeos, the previous failed attempt
thanks for the help so far, its a real shame noone at rtai has written install instructions, would make life a hell of a lot easier
can you boot to the new kernel, or did it not get made correctly?
the newest one hasnt been made yet, was trying to find out why adeos isnt listed
might give it a try
ok. I was trying to se ethat, but I can't even select i386 as the CPU - it defaults to X86_64 for me, and there's no place to change it
am compiling now, may be a while
heh - that's why I was trying to test on my main machine - 15 minute full kernel compiles
is it ok to use make install on kernel rather then messing around making the initrd images myself?
delacroix: that's what I've always done in redhat
I'm not sure that make install will make an initrd for you (don't know - I don't use initrd these days)
when i do a make install i see mkinitrd output fly by
why do you need an initrd? (just curious)
hey delacroix wanna update the wiki page once you get it working? :)
SWPadnos_: in redhat, you need it for ext3 (or scsi)
odd -can't you build in the dirvers, rather than using modules?
fenn: its write protected, but im keeping notes of my experiences, will write em up when i get it working
of course you could
make that if i get it working :)
(there's a super-secret password, but most emc users can probably figure it out ;) )
building filesystem drivers as modules seems really dumb if your root filesystem requires those drivers
unless you're happy with initrd ;)
as long as you dont fiddle with anything
btw thanks for keeping notes delacroix
well - that's why you use an initrd - the drivers are there, so you canmiscompile your kernel and install the modules, but you'll still be able to read the root partition so you can fix it (unless you've messed with the initrd, in which case you're screwed)
fenn: they do it so they can distribute a relatively small kernel, and you can put your root fs on any filesystem
yeah but ext3 is like 25k ?
* cradek shrugs
it works great
nobody has to build a kernel anymore
'cept us poor buggers
finally, after ten years or more of having to always build custom kernels
RT all the way, baby!
the real benefit of initrd comes when you have root fs on scsi.
the initrds are kernel-specific though (right?), so don't you need to make a separate initrd for each kernel (even if you're installing an RPM)?
yeah, stuff like raid (where you want to do significant setup before mounting the root fs) is where initrd makes sense
SWPadnos_: yeah, but it happens automatically even with make install
but why not have a small filesystem (I think it's called /boot) and make the magic "switch the root now" work even without initrd
sure - it's just another way of doing it (rather than building in the SCSI / RAID / fs drivers)
cradek: you can put scsi in the kernel though, just like everything else
jepler: all fifty supported cards?
no, all one cards that are in your machine
jepler: then you're building custom kernels again
it's a distribution time-saver, not a user time-saver
Then why not modularize IDE too?
jepler: sure, why not
RedHat can make a small kernel, and ahve the make install script make the niitrd with the correct modules
jepler: the answer is, I'm sure, that it is used on 90% of machines
for generic installs, it's great. for a customized install, it's no better (and one more file to maintain)
I guess my point was that even for someone building their own kernel something like initrd is desirable (but then I went on to contradict myself)
no wonder you lost me then...
jepler: jmk ran into that same axis bug when trying to run Chips
cradek: the ESC Shift-HOME one?
jepler/cradek: how do you feel about collecting your cam scripts into cvs so we can work on a conversational-programming wizard?
speeking of ze deveel
heh - I guess I scared him
fenn: What scripts of mine do you have in mind?
the ellipse pocket one
tom ok'ed the idea already
fenn: Oh. Well, that's about all there is. I only wrote it to illustrate my point about the power of Python..
actually - that raises a good point regarding CAM and emc "helpers"
I think conversational programming borders on silly in the gui age
not if you don't have a multi-billion dollar cam program already
fenn: it's been in the back of my head to write a gui g-code editor
plus a multimillon dollar CAD program to feed it
there are exactly zero free/open source cam programs
so start one
ding ding ding - there's the problem with EMC :)
I might even help
what's the problem with emc?
well, i'm lazy and really good at procrastinating
i've been meaning to write a .svg to g-code converter for a month now
well - this is from some discussions with a friend of mine who's a mechanical engineer
vector graphics format
cradek: it's xml
scalable vector graphics
cradek: but for drawing gradient-filled polygons
cradek: it's the standards-based thing without sound that will never displace flash
or even work right in microsoft's browsers
didn't know it was for animation
fenn: yeah, you're supposed to make it interactive with DOM and ecmascript
* cradek watches the acronyms fly by
er, you're supposed to be able to
i use it because autotrace and potrace use it
anyway - the problem with emc, from a machine user's standpoint is that it's not a "complete system"
autotrace? cradek once wrote a gcode back-end for one of the bitmap->vector converters
kernel panic not syncing, vfs unable to mount root fs. seems to effect all kernels, looks like i broke something
on the low cost end of the spectrum, you have Mach, DeskCNC, and TurboCNC
jepler: yeah, it was autotrace
is it in the autotrace sources?
Mach and DeskCNC both come with all the helper programs that let you convert from STL, DXF, or bmp into G-Code
no, it's on one of my machines somewhere
hacked into some old cvs version of autotrace...
iirc autotrace could already generate dxf or hpgl or something like that - simple matter of chaging some strings around
all these formats are so simple
now that I think about it, I think I started with the hpgl output
hpgl is pretty close to gcode.. for some reason i couldnt get it to output hpgl
PU and PD need to be changed into something more appropriate
kinda a good news bad news day, have got sourcecode to work my motion hardware under linux, but have trashed my linux box and need to reinstall. figures :)
also HPGL doesn't care about line ends, does it?
also straight autotrace output is not good.. i need to use something like inkscape to correct the algorithm output
email the source to yourself, just in case ;)
delacroix: seems like you should be able to boot your install or recovery CD and fix it
SWPadnos_: what do you mean about line ends?
HPGL can have no \r or \n characters, right (all on one line)
it also has no concept of feedrate
it usually doesn't have linefeeds but I don't know if they break it
if you put them after the ;
I don't think they do - they're ignored (AFAIR)
I'm no hpgl expert
it's been around 20 years for me as well
SWPadnos_: neither does autotrace (have a concept of feedrate)
ok - just thinking about the conversion from HP-GL to G-code
also being able to do pockets from a spline shaped blob would be good
and that is more than just file conversion
SWPadnos_: I've never used a cam package so I have no idea what they're supposed to be able to do
well - that's the problem
emc is made by eggheads and computer people, not machine people
(at least, that's what it looks like from the outside)
the other problem is most machine software really sucks
so people expect it to suck and they get away with it
SWPadnos_: I'm constantly told that as a programmer I have no clue what a machinist needs
no - I'd say that not all machine software sucks
right - me too ;)
I see people say crazy things about what is and isn't programming or what a machinist can or can't (will or won't) do
you know what you need, and you can write the utilities that get you there
[looks crazy to me anyway]
it makes me what to throw up my hands
i think part programming is programming, end of story
but the commercial companies don't throw up their hands
you can't say "a machinist wont get it" and then try to make it like running a manual mill
you just have to teach the machinist to program
and you can load up deskCNC, go to file / Open, and load a BMP file, then engrave it
fenn: lots of people who hang out here will disagree with you
or file / open, and load an STL file, and create toolpaths, then mill it
fenn: but then the next day they'll figure out how to run o12345 #1 #2 subroutines in g-code
fenn: I don't understand it
well, why syntax like o12345 and #1 #2 then?
really random question, but have a spindle for my mill which needs 420 psi to actuate tool changer, any idea what a x7 pneumatic intensifier is worth?
most people don't give a flying fsck about g-code, because most people don't use it directly
and don't want to
neither do i
(dont want to)
SWPadnos_: that's not true in my experience.
SWPadnos_: why would they want subrs and loops in the interpreter then?
you need to go from CAD -> CAM -> controller in a smooth way
well - emc tries to be all things to all people
cradek: because they've got no cam software
fenn: ah, maybe
Who wants to know about HPGL specs?
nobody - really
somebody was asking, just could tell who
it's true - nobody
* fenn cackles
SWPadnos looks guilty to me
* SWPadnos_ muahahahahahaha
SWPadnos: May the fleas of a thousand camels infest your bed!
cradek: the interpreter is mostly a non-issue for the "average machinist", as long as it can run the parts their CAM outputs
so extending it isn't a problem, but removing features is
Jymmm: they may
SWPadnos: ...and 10,000 gallons of slat water flood all your tools and stock!
who cares, with all those camel fleas around, I'd be gone
I'm going to go do something productive... later all
while on holiday for 14 days
good luck cradek
thanks, I'll need it
700mb down, 3.3GB to go!
backing up your porn collection?
fenn: Nah, that's on a cron job
and it's 4TB, not GB
downloading the Project Gutenberg ISO?
Heh, I wish, along with wikipedia.
along with the entire LOC
one line of code - that's not much
LOC = line of code (or LIbrary of Congress)
I have one liners that are 283 chars long, but yes... Lib of Congress.
the new unit of measure of storage devices - "how many LOCs can it store?"
that works for me. =)
I have SO much data around here, it's sad. If I have some massive storage it be great
I need like 100GB DVD-R's
only 32GB iirc
jepler it be cheaper to make a RAID array
I think Blu-Ray can goto 100G, but HD-DVD is in the 30G range
till it hits the streets it does me no good.
My biggest bitch is there are no DVD librarys
like 5 disc DVD drives
jymmm forget blu-ray go holographic cube
well - they're not too reliable, as these things go
which isn't reliable?
optical storage in general isn't considered archival quality
cdr's die after 20 years or so
even less, in my experience
I actually have pressed CDs that have dies in less than 5 years (though those were some of the earliest releases)
use punch cards
they'll be readable in 40 years
ive lost loads of data due to bad dvdrs, think im going to set up a raid fileserver and use that instead, as chances are it will outlive my dvdrs by a long shot
100 if you use good paper stock and keep them away from mice ;-)
they do make archival quality cdr's though..
depends on your definition of "archival quality"
it's a real problem, actually
there's a lot of information which, at the rate of technological advance and obsolescence, won't be readable in 10-20 years
I chisle my important data on granit.
I have some pretty early pressed cds
over 20 years old
consider that we can still read documents that were made 1000 or more years ago, and you see the dilemma
the problem is bloat
swp phthalocyanine cdr's last 200 yrs
I've got 3 that I can think of that actually have visible holes in the aluminum layer
reliable storage methods aren't dense enough for todays exploding amounts of data
hell mother nature can't even make a reliable data storage medium
fenn, that may be true, but I'd bet that's only if you keep it in an environmentally secure location, and also keep a cvomputer that can actually read a CD around
books are pretty good
I don't use cdrw much for anything
rocks work -look at the fossil record
I suspect they are not long lived
I guess sputtered gold on some non organic substrate would be best.
Hmm reading josh's post...
about parametric toolpaths?
i like that idea.. no idea how feasible it would be
interesting points - I was going to respond with something about latency and throughput, but haven't had the time
the real long term issue for any high-tech storage method is long term availability of readers
thoughtful post...but some misconceptions i think
is he here?
maybe he got sad that nobody replied to his post and went away
no - it's from today. He's posted before
I will I guess
maybe he doesn't spend every waking hour on IRC like we do ;-)
about the stutter
that's just an error...pure and simple
you take the mathematical side, I'll take the hardware side ;)
oh i was remembering the one from a month ago.. daily digest not in yet
"the reason for joint space tp"
as far as conic section g code commands...great but not supported by cam systems
well...no inverse kinematics for one
think we're talking about different things
I find nothing wrong in that post
it is true that trapezoidal is faster than cubic
it also beats up ballscrews and other hardware
It is also true that higher order polynomial paths with minimum jerk will be less accurate to an arbitrary desired path
like an s path made by a tangent g2 and g3
i can see that on my "flop-o-matic"
cubic or higher is gonna make a flat spot at the junction
er, wait i had that backwards
hmm josh seems like someone is on his case or something
no one is is there?
probably paul said he had a dumb idea
just a guess
why do you say higher order paths are less accurate?
Well I would just mention that if no cam system supports such things....it would limit it somewhat
that concerns the latest post
My old robotics book has a bit on joint vs machine space planning
well if emc is to support multiple kinematic architectures it would be a nightmare to maintain joint space planning
Fenn: specifically higher order paths that are jerk limited only
higher order paths in general can be more accurate
like a taylor's or fourier series
in the limit they can be exact
but in the real world acceleration has derivatives
we need to limit them some in machine tools if we want them to live a long time that's all
skidding ballscrew balls are expensive!!
* fenn tries to finish reading the latest post before he gets confused
I need to re read it too
I agree that things like on demand feed over ride turn the tp into a real time problem
you want to keep the path time optimal.
i think feed override is more for slowing down than speeding up
so you dont break stuff
I would argue that you could just scale servo rate, simultaneously slowing vel AND accel...it would be ok
doesn't servo rate have to be an integer multiple of something or other?
it would not be time optimal, but so what? you are slowing it down
of a timer interrupt I guess
but that is fast enough that it should give enough resolution right?
dif between rt tp with feed over ride and servo rate scaling is this:
slow down with servo and the path does not get tighter
slow down with a recalculated tp in real time and it does
is the current tp realtime?
i thought i unsubscribed from yahoo groups antigravity but it still comes anyway
"brandons people will worship me , you will see.
and no one can do antigravity becuase they are not crazy like only crazy
people like me discover and master alins technology , i am the one , and you
are nothings .
i am now upgraded to the god of antigravity and time travel .
kosol the one .
time travel forward is easy: go fast.
well you simply find a wormhole that has been traveling at relativistic speeds and go through it
personally i dont believe in all that relativity crap
rather i dont believe that C is a constant
I had to take a bit of modern physics in school
I found no argument
made sense to me
but I am no cosmologist
back to non-relativistic physics
funny was just reading up on olber's paradox last night
is a fifth order polynomial sufficient to describe most cam output paths for greater than (n) line segments?
to improve the block read rate
where blocks are written in polynomials that is
really big test of relativity is coming up, actually
whether gravity probe b detects frame dragging
not gonna happen
fifth order is required simply to have finite jerk
so things like servo current will be a ramp
not sure about that either
well, no one's very sure about that ;)
never understood the need for a graviton.. you don't need a particle to describe the attraction of opposite charges do you?
although special relativity pops out of maxwell's stuff
ratio of permeability to permittivity
yah.. special relativity isn't so 'weird'
c has to be constant for that to work
you take maxwell's stuff, assume constant speed of light, and poof, you have s.r.
the derivation is left as an exercise for the reader
well, icee, I was shown that c==constant pops out of maxwell's eq
(think I had a test on that)
that sounds wrong to me
but what do i know
well find out
I had those courses 25+ years ago so I might be forgetting things
well, from whati recall, maxwell's stuff works with non-constant c but you get all kinds of other 'interesting' contradictionish-sounding-things
e.g. similar to what you get with non-euclidean geometry
it just sounds wrong, but it could actually represent the universe fairly well. or not.
well non constant c would kinda screw up classical electromagnetism I guess
regardless whether you use non-euclidean space or non-constant C, you still have light bending around stars
changing the speed of light (and hence index of refraction of the vacuum) makes more sense to me
Well general relativity predicts that
photons have mass!
in georgia they do.
the concept of "mass" doesn't mean anything at the speed of light
at the phase velocity of light
but at the group velocity...
group velocity is just semantic sugar
like saying a square wave travels faster than a sine wave
because you get the threshold sooner
funny you should say that...I am working a bit with non linear acoustics right now on a project
lots of quantum effects
like solitons and coanda effect stuff?
things popping from one energy level to another
but it's a common effect
non linear vocal chord spring rate
pops from one freq to another
isn't that just resonance?
just a demonstration of quanta...that's how those guys figured that out
drop a rock in a still pond.
watch the waves spread out
watch the waves bounce off the edge of the universe..
little wavelets vanish at the leadinf edge of the wave train
and reappear at the back
kinda hard to figure out
* fenn looks around for a still pond to violate
ah i have one right here
can only see one wave at a time though oh well
stop messing with the toilet
i have a sheet of cloth hanging from the ceiling
if you hit it you can watch the waves ripple
if you tug on it you can vary the speed of transmission
the 10^5 pascal overpressures I am getting in my "thing" I made does weird stuff
law dogs say I can't talk about it though.
like explode? or like slowly melt apart
les_w " things popping from one energy level to another" like being kicked in the nuts?!
I would love to say....hate patents
les_w instant acapella
tenor to accapela in 0.00002" seconds
hmm that's only 1 atm
When youfile for international patents, you cannot publicly disclose anything.
yes 1 atm
I have to do what my patrons say
they provide the funding
and I like to eat and stuff
glad i told those suckers to piss off
i'd be slaving away in a lab all day instead of eating
This research project isn't helping the hole in my stomach...I can say that
maxwell found that the speed of electromagnetic waves is a constant, i.e. independent of the nature of the electric and magnetic fields
in linear media, yeah
but if "spacetime" or "aether" is not uniform then it doesn't mean anything
and it's obviously not uniform, or we wouldn't be sitting here
* fenn hates physics discussions
now don't make me vanish
I don't remember enough for a good physics discussion
* fenn generates a les_w-shaped EM pulse 180 deg out of phase and *poof*
I remember on a test I had to derive special relativity from no more information than "there is no aether drift"
25 years ago
can't do it now
without cracking a book
that's because you were just repeating what the teacher said
yes I was
writing a proof doesn't prove it
science is founded on the fact that there are no proofs
probably the only thing we can be sure of
yup...guess that's why I ended up an engineer
how many qc toolholders is enough?
more than I have...that's for sure
standard kit of 5
Buy the cheap 5 pc kit just to get 5 more
i made two sticks of "toolholder stock" but one of them is under by 5 thou.. i'm thinking of making another that's to spec
cheaper than individually buying them
they are a real time saver
the messed up one might end up working anyway though
just so they repeat
that's all thgat matters
good for cnc stuff
I just use them on my manual boxford
i'm actually making two at once so i can swap between the cnc lathe and the manual
"cnc lathe" is a bit of an overstatement at this point
piston type post and 5 holders 100bucks...I'll just buy em
i like being able to make my own tools.. that's why i made 'em
I do too
nobody will pay me $20 an hour either
can't do it as much these days...commercial pressures
poor les has concentrated too many assets in his person
hire an indonesian kid to do your engineering
cost of an ivy-leage education is only $120k thes days
I have a retired engineer as a part time employee now
where do you live?
well something like that
I just went to a state university...but I did ok.
was that estimate low or high?
low I think
including room and board
les_w (Sorry... did you ever say a clearcoat earlier? I had to hit the store)
the money ain't worth nothin
pity since I am making a bit.
take a wheelbarrow full to buy groceries.
btw try adding salt to help with stomach ulcers
meat attracts sodium ions and that degrades the stomach lining mucus
fenn: the origin of almost all stomach ulcers is microbial
the degradation of mucus theory is passe
Having MBAs make a gant chart on "when I will be innovative" makes mine more I think
yeah yeah heliobacter pylori
* fenn cackles
les_w: "sulforaphane--a phytochemical in the broccoli sprouts--killed off any H. pylori that was exposed to it."
oh...I like brocolli
and licorice root is supposed to help regenerate the stomach mucus also
I went to the doc. lots of tests. He finally said..."when you were younger you could take more stress. Your body can't do what your brain wants anymore. Here, have some dope."
"here this will make you lazy and apathetic"
plus...dope prescription== no pass flight medical
doc's can help you if you're already lazy and apathetic
03cradek * 10emc2/nc_files/3D_Chips.ngc:
jmkasunich says he doesn't think these changes were intentional,
and they break work offsets.
well, they were intentional, I just don't recall the intent ;-)
I think it depends on how you like to use work offsets
well IMO they broke the file
yeah, maybe a matter of opinion/practice
jmkasunich: because of the unrelated commit log, I bet ray didn't intend to check in that change to Chips.
it's easy to cvs commit the whole tree, forgetting about a change you made somewhere else
dave is now known as daveland
03paul_c * 10emc2-auto/wiki/ (17 files in 12 dirs): "Auto update wiki from a cron job. Tue Dec 6 05:30:02 GMT 2005 "