i think i just aced my calc test
fragalot1 is now known as fragalot
archivist_ub: Well, I just ordered a load of µC's with ethernet capabillties, EEPROMs, FET drivers, and PWM modules,... Time to attempt to have fun with ethercat? :D
hmm the cutter compensation seems odd
trying to get my hands on some atmel devboards too
adding compensation to two intersecting curves I expect the intersection to move out and up but it went up and the curve center point looks like its changed as two curves no longer parallel
fragalot, I don't think you'll be able to do ethercat in software. It relies on inserting data into a packet and recalculating the checksum on the fly
SWPadnos: Yeah, that's basically how ethernet works.
ethercat is a little different
the packet is sent from the master, but it has "miissing information", which the slaves insert on the fly
by "on the fly", I mean without receiving and retransmitting - the packet is being output while it's still being received
fenn http://www.archivist.info/cnc/with.png http://www.archivist.info/cnc/without.png
then I'll do it differently, I guess
have EMC calculate the pulses, put that in somesort of array, push it over to the controller, have the controller start on that array, and sortof keep a buffer..
like send 1kbit worth of movements over to the controller, have it start on that, but keep feeding more as it's buffer starts to clear out
ah, so you're trying to make a SUB / Ethernet connected step generator
err - USB
SWPadnos: Yeah, anything that works, basically
ethernet, as it's full duplex, sounds more attractive
it's only full duplex in point-to-point toplogies
direct connection to the host is what I planned on
could do it via a router, but why make it more complex than it has to be :p
OK, so a single board, like the smoothstepper, G-Rex, ncPod, or others?
pretty much, yes
I might make it have fan control, spindle control, various M-codes,.. basically till I run out of pins on the 100-pin controller :p
are you trying to make something that works with EMC, or something that does the motion control itself?
dno yet,.. if i can, i'll let EMC do most of the work
like write somesort of plugin for it,.. I dunno
they're called "HAL drivers"
there you go
what kind of OS do you think we run here?
yes. we don't do "plug-ins" ;)
shared libraries are OK though
regarding your project, the problems we've come up with when discussing this kind of thing (which happens every month or two) is that feedback has too much latency
so more advanced functions (like spindle-synchronized motion) must be implemented on the device
Ofcourse, even if i can't get that to work, it 'l still be nice for people that don't need that
since HAL allows you to do other interesting things (like running N motors from one joint output, possibly with some offsets, and maybe all synchronized to the spindle), we usually drop the idea :)
(since feedback latency is a problem)
Ofcourse, even if i can't get that to work, it 'l still be nice for people that don't need that
remember that feed override is also a "real-time" function, as is feed hold
and feed override can change the path shape, based on blending and tolerances
so there's some balance between buffer length and responsiveness, even for functions that most people will want
for what definition of realtime as its all digital and has software delays
uh - there are two types of feed override that are done in realtime - adaptive feed override and feedhold. Those are always responded to in <1 servo cycle, so 1ms by default
every user input i could move over to the controller, i'd move over, and somehow feed that back to EMC.
even if it doesn't work, it 'l still be a nice attempt
sure, the more people who beat on it, the better chance we have of getting a functional solution :)
thats the spirit
what assortment of microcontrollers did you end up with?
I knows its going to be a bar steward, but cant see "impossible"
ewww. bummer :)
SWPadnos: (picked at random for testing)
SWPadnos: also got some atmel "ethernet starter kit" board sampled,.. waiting for their confirmation :p
heh. PICs happen to be my least favorite microcontrollers (with the possible exception of 8051)
I like the AVRs. I usually use assembly on micros, so having a uC that knows how to use a carry is a bonus
SWPadnos: I just go with PIC because they're a) easy & fast to sample b) I have more experience with them c) got the programming hardware..
QuickWriter II here
ICD2 clone here
SWPadnos: I think you have a huge appreciation error there
or something like that - haven't used it in a while
heh PIC carry in the program counter os so broken in some of them
8051 are waaaaay better and nicer than PICs
well, there's no ADC instruction ...
it's always add / btfsc C / inc
not if you use Keil :P
that's C, ain't it? ;)
yeah, it's a bit of a hassle doing that in the registers all the time
SWPadnos: yeah, I know.. sounds like cheating
it does have the advantage that you can directly address any bit/byte in the current memory bank, but other than that, I really don't like them
my question is,.. whats wrong with using C if it works and increases productivity during development?
I thought about usign ethernut for an emc2 device
I have one AVR project that should be changed over to C. It's about 800k of assembly source at the moment, so it's getting hard to manage
lol what would be funny is if I ended up with a device that's basically just a dedicated EMC2 computer.
fragalot: nothing, except that you need to buy a c-compiler most of the time
nothing, I just don;t usually work that way
for the older uC;s
alex_joni: that's why they have "student edition" compilers
and support for gcc in the newer ones
yeah, gcc works afaik
actually the best way to do things would be to get an arm9
and stick a linux+rtai+emc2 in there
I wonder how small it could be
|-------| depending on your font
I hates ARM a little
SWPadnos: I bet small enough to fit inside the kernel
I'm not kidding
I guess I could break out one of these AVR32 or AT91SAM boards I have here
alex_joni: I thought you meant the hardware. :/
SWPadnos: DO IT
I did something like this for S3mumble
some arm7 or 9, I forget
* fragalot breaks his old PS1 open
oh come on, S3 made video cards
no, it was a chip designation
hang on.. googling
I thought about doing that, but I only have 3 characters to go on
[11:13:10] <alex_joni> http://www.samsung.com/global/business/semiconductor/productInfo.do?fmly_id=229&partnum=S3C2410
3 characters was enough for google :)
have you seen the Etrax?
SWPadnos: I built a 2.6 kernel for that and using buildroot I built a / unionfs thingie, which I linked into the kernel
so I only had one 2MB blob which I had to upload to the board to make it work
ah, so you'd still need external flash
I'm a bit fuzzy on the details
I think I uploaded it directly to RAM
there are some ARM chips that have 2M flash on them, but most top out at 512k or 1M
it also had 16MB NAND Flash
well, if you want it to power on this way ...
the board I used
right - external
but I was only trying to make it work initially
and some support for boot flash
wth is that called, lots of terms elude me atm
it used to be on older industrial mobo's, before CF replaced that almost completely
some Flash memory
it's some flash in a DIP28?
solid state disk modules
not like we have now ;)
32-pin chips, very thick, look like EEPROMs with a few extra programming pins
yeah, something like that
also called DOM - "Disk On Module" or something like that
anyways.. not really important
yeah, that's it
[11:18:13] <SWPadnos> http://shop.trenz-electronic.de/catalog/product_info.php?currency=USD&osCsid=c1a4ae1c1d787a6437d1d0075d256598&products_id=213&language=en
axis is ok
and the spartan there.. just perfect to fit some stepgens in it :)
this one has the ETRAX FS + a spartan 3 500k
a bit pricey though :/
if it were < 200$ it would be great
although I doubt there is RTAI support for the Etrax
yeah. I'm not too familiar with the CRIS architecture
but smaller systems have less need for lots of RT extensions, since they usually run optimized (and stripped) kernels and fewer apps
and you'll have interrupt handlers that are actually interrupt handlers instead of "interrupt signalled tasks" :)
otoh, I don't see an eth there
the etrax has 2
oh, cool.. no Phy then
it has one with phy, the other without. you still need magnetics and the jack of course
hmm.. 91 I/O from the FPGA
and 72 GPIO from the Etrax
not bad for a small system ;)
should be plenty :D
which comes down to the main question
which one of you will build an embedded system for EMC?
well, I charge $75/hour, and I expect it would take 1000 hours or so. so if someone has $75k to spend, I'll do it ;)
otherwise, it'll take a while longer :)
my first uC had 85 IOs
instead of figuring out how to make it communicate nicely,.. just.. do it on the spot itself, fix all "timing" issues :;p
they didn't have chips that big when I started :)
alex_joni: my first had 35 :o the 2nd had.. 3?
google doesn't even know that one
yeah, those came a decade or so late for me :)
fragalot: try without the 68
it's 68HC12 as a family
it was typoed
and HC812A4 is the model
m.. I wonder how well my Xilinx devboard actually performs (Digilent Basys board)
same as anything you'd design - it has no electronics between the FPGA and the I/O connectors
heh, they still have the page from where I bought it
[11:28:14] <alex_joni> http://elmicro.com/de/kit12_r.html
if it's the spartan dev kit
it's already got VGA & PS/2, and outputs for 4 axis..
oh, that one :)
the only major difference between the starter kits and a custom design is that the clock may not be waht you'd use
[11:29:06] <fragalot> http://www.engsoftkorea.com/home/02_product/image/BASYS_p_m_2.gif
hmmm. I think that's different from mine
I've always wondered what that lil' DIP socket on there is good for
ok, it's mostly the same, but the pin headers are different
and the digilent version has a serial and VGA port (or at least 9-pin and 15-pin)
oh, my board only has the VGA port
you appear to have USB instead of serial
you must be thinking of another board?
no, it's the Spartan-3 starter kit with a Digilent board
but it's a year or two old
[11:33:55] <fragalot> http://www.digilentinc.com/Products/Detail.cfm?Prod=S3BOARD&Nav1=Products&Nav2=Programmable
<-- that one?
yep, I think so
different board :p
SWPadnos: the s3c2410 has an ARM920T core
how good is ARM compatibility? is a program compiled for ARM9 binary compatible with ARM11?
apparently, this board supports 400Mbyte/sec transfers over USB ... I'd say that's pretty good, beeing 8 times faster than the host you're going to connect it to...
fenn: beats me..
fragalot: any USB2.0 is max 400MB/sec
alex_joni: USB2.0 is max 480Mb/s not MB
not even USB 3.0 is that fast afaik
A Super-Speed (3.0) rate of 4.8 Gbit/s (600 MB/s).
I stand corrected.
same here for 2.0 :P
lets go lick our wounds in the local pub
meet you half way there?
where do you live
I'm 45 44 01.03 N / 21 12 40.46 E
* fragalot grabs his GPS
(that works on maps.google.com)
* fenn wonders why we dont use decimal coordinates
fenn: because base60 is easier to do math with
alex_joni: meet you at 48 11 42.98 N 13 38 55.93 O
alex_joni: thers a lil' place called "the loose moose"
well.. it's close to the A8.. so that should be ok :D
'sonly about 850km away
fragalot: where are you located?
oh, cool :)
sounds like Prague is closer to the midpoint :)
mm, somebody promised me a cingular 8125 .. I wonder what customs 'l do to rape me this time
SWPadnos: you're not invited!
between you two, silly
alex_joni: y'know,.. We've got a fairly big airport... lol
[12:11:04] <fragalot> http://www.trevoroldak.com/uploads/evolution.jpg
heh, yeah. that's some good scientific evidence right there
anyone up for a physics question?
if im accelerating something that weighs 60kg to 0.08m/s in 0.1second
F=60kg*0.8m/s^2 = 48N
according to my calculation, if I neglect friction..
I made the question up to see if I got my physics stuff down good enough :p
heh. well, the math and units check out
* anonimasu has this issue with mixing units all over the place
sure. good thing it was an easy one ;)
haha, next thing will be to add friction to it..
my HP 2-C or 28S have saved my butt on many occasions
since err - 28C
and gravity.. but that's for tonight..
since they have unit catalogs built in (which gives you the base units for just about anything)
I'm thinking that the friction will be F=48+(mg*f) if I havent forgotten :p
if you neglect the change from static to sliding friction, it should be something like that
that ends up as 165.8N with friction, assuming the friction dont change with velocity..
I don't remember how friction coefficients are specified though - I think it's just a fraction of weight
yep that's right
(weight being mass * gravity)
friction would be (m*g)*f
I figured this stuff would be more interesting if you plug it into a cnc machine calculation ;)
thse are 2 questions from last years exam I had
I don't know the gooeyness equations though (for way oil and stuff)
* anonimasu nods
fragalot: what the hell are you studying :p
scary questions; )
anonimasu: Electronics-ICT.. you?
uh.. preparatory stuff for getting into a university
I've decided to go for mechatronics engineering :p
robots and stuff ;)
but im getting old :p
(I'm 24 already)
meh, i'm 18 and my lab partner is 56yo..
I've got to leave for a bit, going to head to work and rest/work um.. look busy
fragalot: does my calc add up?
dno wasn't following o.0
fragalot: in short 60kg at 0.08m/s in 0.1sec
fragalot: I get that to 48N
you mean 0.08m/s² ?
sounds about right
I mean velocity
a = 0.8m/s^2
and with friction and a gravity coefficient of 0.2 I get that to 165.8N
all you do really is calc the weight in N for the friction, ofcourse thers a difference between the static and dynamic friction coefficient, gravity only comes into play for converting the 60kg to newtons (* 9.81)
oops, 9.82 :p
(im in sweden -_-)
it's been nearly a year since i've had this stuff
one newton is 9.81 though, and the gravity coefficient is 9.82 here :p
it's kind of disturbing.
irl they work with 10 anyways.
I'd bet that consistent rounding would fix that problem
* anonimasu nods
thanks for the help!
most engineers after they graduate work with 10 instead of 9.81, since (following the rules for scientific rounding) they won't gain any accuracy anyway
fragalot: im going to have a shot at your airplane calculation later tonight..
anonimasu: have fun, I flunked it.
it's probably way over my head :p
anonimasu: take the fuel consumption into account aswell!
haha damn.. that gets hard ^_^
* anonimasu is only learning trivial stuff yet
it's like introductory physics..
prep stuff -_-
wel, that was the question for 1st semester in uni. lol
I ended up changing schools, .. it was madness there
im trying to prepare for math madness..
I like math when I can calc real stuff.
don't do what I did and take a calculator with dead batteries to the exams
my calculator's forbidden I think :/
it's a casio cfx9850gb-plus
after the first 10 minutes, I got a nice " Dead battery, going byebye."
anonimasu: I use the TI-84 plus
in uni most calculators are allowed
hehe, in the us this one is dissallowed
aslong as their main goal is really beeing a calculator,.. :p
anonimasu: that's.. madness
I think it's because the solver is really good
doesn't seem to have stuff onboard that my TI-84plus can't do..
lol, I even put the periodic table on it :/
they banned calculators on us, but the better professors had problems that you could do in your head
you could leave answers with pi in them and that kind of thing
SWPadnos: that's plain silly imho
SWPadnos: oh, we /HAVE/ to if we can
but you were expected to be able to multiply/dive by 2, 5, and 10
so you can use a calculator,.. but it's only purpose is to verify what you've solved.
well, in some ways it's easier if the problems can be done in your head. If you get something that you can't do, you probably set up the problem wrong
SWPadnos: Aye. Same thing can happen with a calculator too tho :p
like I said, you could leave certain constants in the answer (pi, h-bar, R ...)
well, with a calculator, you get a number and don't know whether you did it right
I've oversimplified a problem once, and I ended up seeing the answer in the question itself :/ somehow, magically, I gave the right answer,.. I still don't know how I did it
[12:55:50] <anonimasu> http://www.educalc.net/246080.page
with your brain, if you're trying to get the sqrt(24579.3), you might have serewed up
(with well-formulated questions anyway ;) )
anonimasu: oops, mine isn't allowed either it seems
mine either, and it's 20 years old
owell, it's more fun,.. checking if you did the integrals right,... lol
hell, the calculator solves them for you, plots them out,... you still need to do it yourself, but it's nice to know that your end result was correct
I lost my normal calculator in chem class
* fragalot decides to study for his data communications class a bit
vauge question of the day http://www.cnczone.com/forums/showthread.php?t=65598
err - spaces in the directory name?
oh - your good...
tree file error, that's a new one
dunno if that's it though
ok, shower time. bbl
take it easy
oh wow, nice question
when is the re-balling of the screw?
backlash .0012 now - can't decide if I'm disappointed or not
I might let it wear in for a while and try again
just a bit bigger?
no, I'd have to go smaller again, and put them in both circuits I think
I must not understand something about how the balls fit in there, because my math keeps giving me answers that are too large
it could be that there is no size of ball that will give perfect results, because the inside of the nut is bad
I got the quote for the bearings..
500eur each :p
any1 here good at networking? (extended subnet masks basic stuf..)
nvm, got it
[14:36:57] <anonimasu> http://www.roymech.co.uk/Useful_Tables/Tribology/co_of_frict.htm#coef
hm, seems like they didnt manage to measure dry friction of cast iron
heh cast can be odd
well, machines are made out of cast iron :p
hmm my best bit of cast iron work is not on my webserver
[14:47:41] <archivist_ub> http://www.archivist.info/collection/showresult.php?searchv4page=1&errlev=0&searchstr=pendulum&srcdata=title&srcprog=searchv10.php&Type=PD&Accn_no=5695&dir=2006_09_07_pendulum&file=P9072699.JPG&subject=14974
[14:48:32] <archivist_ub> http://www.archivist.info/collection/showresult.php?searchv4page=1&errlev=0&searchstr=pendulum&srcdata=title&srcprog=searchv10.php&Type=PD&Accn_no=5697&dir=2006_09_07_pendulum&file=P9072701.JPG&subject=14976
mm, that's about the same size chuck i had flying over my shoulder back in highschool
continuous cast rod
archivist_ub: is it me, or is that a rather.. big pendulum :p
took a wonderful polish
looks like it
was for a mercury temperature compensated pendulum so its bored out
also, that script you use as the gallery shouldn't be using HTML only to resize the image >.> it takes 50 seconds to load the thumbnail
script is designed for local use I am adding thumbs to it
very nice! what is the material?
it would be easy to have php to resize the image tho
and how did you turn that taper? I don't see a compound
I make thumbs with php
rotated the topslide
ooh it's pretty and shiny
it was far too large for the lathe took some bodging to do
like carbide tipped masonry drill ground to get the bulk out
<-- those, or actual stone drills?
and reamers modded to drill and milling cutters abused on the end of a bar
no stone drills
one reamer got its own back, it was only case hardened
I think I just had (kinda) an idea again for the networking thing.. What if we do it the way IBM used to do it with their ring topology?
eg. pass a token arround, who'mever has the token is allowed to talk,.. and just have them take turns between each package
how about RTNet instead?
I thought you were the one claiming that wouldn't be possible?
wait, diff. thing
* fragalot looks into RTnet
USB is not so great. RTNet could be fine
no, i meant ethercat before
rtnet looks simpler
there is a low cost system that uses a similar idea to mitsubishi on that commands have a timestamp so the drive can interpolate they also pass /delay the clock so they are synced
mm, sounds good
just wondering,.. is it possible to "bitbang" the ethernet port? :p
no thats hardware
I hate the term bitbang
it's just a term
would be nice if you could tho :( would instantly have solved my problems, :p
heh - even 10baset would be in the 10mhz atleast.
skunkworks: which is why I wondered if there was just a way to basically use it instead of the parport w/o the need for additional hardware
* fragalot starts to randomly ponder if the µC couldn't just parse the G-code on it's own
* anonimasu_ is still into this physics stuff
trying to get this friction stuff down..
anonimasu_: do you have to be able to calculate if it topples over, or slides away?
I only need to know that friction ends up 0.25% of the normal force on a surface..
or whatever the friction coefficient is..
48N to accelerate the table up to speed.. and 41.2 newton more to get it moving..
89.2N to get the table moving from a standstill acceleration 0.8m/s^2
if im calculating it right
and a fat facemill chomping the work
however it dosent quite match up
if I get this right, you need to overcome the static friction, then the resultant of the forces becomes > 0 and whatever you are moving starts to move
and then you start another calculation for the dynamic friction..(dont you)
get a spring balance push/pull against the axis in question, measure
uh.. im doing physics homework.
er wot maffs
anonimasu_: you are correct, dood
toastatwork: at what point?
making it split
µ_s for f = 0, then µ_k for f = ma
because you dont need more force then to overcome the static friction at first..
then you need the force to accelerate up to speed..
fragalot: i think the general idea is to port EMC to ARM chips
so in reality it's just the higher value you need to take into account..
if the static friction is higher then the kinetic friction
or sliding(whatever they call it)
if kinetic was higher than static i'd lol
I'd say that static friction is always higher than sliding, but there's probably some exotic material that would prove me wrong
I added them togther and that ended up too insane..
toastatwork, thixotropic grease and camera lens threads then it is
camera lens threads?
toastatwork: I still miss one piece of this..
you need to overcome static friction, or you get zero acceleration. Once that's overcome, you can reduce force since sliding friction is lower (or you get higher accel, since more of the force is going into accel instead of overcoming friction)
I put the wrong grade on once near locked it up
toastatwork, yes focus action, its the grease thats doing the work also on potentiometers on hifi
if the static friction ends up as 0.07 for cast iron with oil film that gives me 41N in friction against the slide..
bloody hell wireless
* anonimasu_ really needs a friction vs force graph
toastatw1rk: if the static friction ends up as 0.07 for cast iron with oil film that gives me 41N in friction against the slide..
toastatw1rk: however, as the friction dosent go away when I start accelerating that would mean I need F + ma + 41N to overcome the friction against the slide
F=MA + 41N to start
when you start accelerating the 41n isn't 41n anymore
hehe, gravity dosent go away -_-
once you are at speed oil film floats the mass and friction drops and viscosity effects start to matter
yes but the static coefficent does
and the sliding friction for cast iron is the same as static friction
well, most places never mention any static coefficient for cast iron
[16:23:01] <anonimasu_> http://www.roymech.co.uk/Useful_Tables/Tribology/co_of_frict.htm#coef
* anonimasu_ googles about oil
top of that page mentions some of the error sources and uncertainty
yep I saw that
hm, I wonder where I can find a reference about kinetic friction
speed vs friction
friction is kind of a nonsense thing anyway
it's a gross oversimplification
but it earns you a living dosent it :p
surface finish and oil effect greatly, cast iron would be the off one out as it contains carbon for self lubrication
surface finish is very dependant and oil
I guess this is as good as I can calc it..
i missed the part where this matters for something
what are you doing this for?
I forget things I think are boring.. and physics without any application are boring
so, im just theorizing about a mill, to make it interesting
like what force I need to make my mill go zip zip zip ;)
at 5m/min and 0.1s to reach that speed
without oiled slides?
but with stationary table and doing a rapid.. as example..
once a slide starts moving it's pretty much entirely viscosity, not friction, that provides the opposing force
hydrodynamics and whatnot
* anonimasu_ nods
and that's way beyond the scope of my physics :p
i think i got 100% on my calc test
*makes a mental note*
also isn't viscosity a simple thing?
it's linear, iirc?
not for some oils
hm, not nescessarily
[16:34:56] <anonimasu_> http://en.wikipedia.org/wiki/Viscosity
I got bored and mocked up something that's probably impossible, but still fun as a concept
what do I hold on to
as long as you can
[16:46:39] <fragalot> http://omploader.org/vc2t4/P1000962.JPG
dno if it's readable
or if it makes any sense for that matter
rather hard to read, can't you just type it?
whats the fun in that
* archivist_ub detects tippex
one problem is you are confusing axis with joints - there might be more than 9 joints
from the homepage: It can simultaneously move up to 9 axes and supports a variety of interfaces.
anyway.. why send step/dir and dir/rpm in the same packet?
for powered axis's
like on lathes
fragalot: what kind of binder do you have that has 6 rings, 2 pairs right next to each other
what's a powered axis?
fenn: basically, a lathe can have a powered tool
ok, why is that not a spindle?
agreed, it's confusing. and fairly pointless, should be a diff. packet
fenn: because i'm tired? :)
i dont know how that's generally handled in g-code, but i dont think there's any way to do it in emc
beyond custom m-codes or the analog pin stuff
then that part just drops out and becomes part of the M-codes, :p
if you're going to let the device do arbitrary stuff then you force the user to learn how to program some wacky PIC
basically just have it translate all those things to outputs..
take a look at how the hostmot arranges pins and functions, i think its fairly similar (but you hve no feedback of course)
oh wait, duh
you can't do it that way because it's USB right?
yeah, USB takes too long to reply
* anonimasu_ needs a formula collection badly
so you have to send time-stamped commands instead of a basic step/dir packet
well, that's tomorrows project buy one
because the host needs to ask for it first,... all in all, it takes too long before you get the response.
fenn: that's why i put the "WHEN" thingy in, kinda,.. as a quick "save" :p
fragalot, setup the synchonous mode
archivist_ub: for USB? o.0
fragalot, read usb2 spec its long but its in there
and packet size is configurable
* fragalot reads
I would expect regular timed packets going both ways
damnit ic an't find the full USB2.0 spec
so packets are both ways in the servo period
only various quick Q&A
I have/had it on my old box
9MB zipped pdf
took me a week of skim reading to a get a percentage in the brain
nighttime reading, here I come.
I hate reading specs, even skimming it hurts the eyes
why did we just jump back to USB btw? :p
ethernet already has the full-duplex system, all we need to figure out is how to effectively put it to good use
sorry i thought you were doing USB
1) define the problem
2) solve the problem
* fenn vows never to eat kimchi again
oh come on - kimchee is great (several kinds)
tell me, is it always rotten?
it's "fermented", not rotten
it should be a bit like spicy sauerkraut
heh funny definition of rot
there are radish based ones as well, which are also quite nice
yes i had the radish kind
if it's rotten, I can imagine it tasting like crap
i'm pretty sure it's "fermented"
well, it can go bad too
there's a brand you can get in the store that's pretty good, but I don't remember the name
the kimchee eater thought it "tasted like kimchee"
hmm I want a debugging gcode interpreter
archivist_emc: that would be called "employee"
it's expensive tho,.. and prone to mistakes
* archivist_emc hating arc some more today
but it's very very good at covering those mistakes up,... till some point
[17:20:20] <fragalot> http://www.ebsnetinc.com/software-documents.php?id=88
I not breaking endmills till I get sense from axis
I just had a nice idea
archivist_emc: do you have a mill?
er yes see youtube where its in action
k, mind cutting this out & mailing it to me? :p:p http://www.instructables.com/files/orig/FAU/Z4M9/F3KRXJ2T/FAUZ4M9F3KRXJ2T.dxf
believe it or not, a DXF does not a part specification make
fenn: no, but he won't do it anyways, shipping is too much.
luckily /me is dxf blind on this box
[17:27:01] <fragalot> http://www.instructables.com/id/SXCIEH6F3HY3SOB/
<- I just want this cut out,.. I wish he put dimensions on that
No, he wont cut it for you.
looks like some kind of cnc mill parts
that's laser work :p
or waterjet ;)
really close together too, couldnt do it with endmill anyway
scroll saw apparently
from your desciptions in here, ask useless boss here to use piercing saw
it should work..
theres approx .3" between each part if i'm not mistaken
7mm is doable.
ESP. in mdf xD
7.62 to be exact..
* archivist_emc uses 2mm endmills
how are you going to hold it down?
lol, no, just clamp it to the table
uh, no tabs or nothing?
just let them fall out?
is there an online ral pic I can see
yes in a sec.
[17:30:19] <fragalot> http://www.instructables.com/files/deriv/FCK/BMSH/F3HY3SK9/FCKBMSHF3HY3SK9.MEDIUM.jpg
[17:31:12] <anonimasu> http://imagebin.org/27838
I could pretty much cut it out by hand,.. if i had proper dimensions. or if i could properly print that out in a tile
unfortionately, i'm unable to do either
heh just a cheap chinese jigsaw and a bit of effot
fragalot: just build a cnc machine to cut it out :P
fenn: bastard :p
uh.. you still need to make it square..
whoever that made it had laser.
welcome to boostrapping 101
WE CAN REBUILD IT, STRONGER, FASTER, WITH BLINKING LEDS
anonimasuhe did it on somebody else his CNC machine
he said so in his instructions "I cheated"
That doesn't change the fact that I don't tho
no belgian EMCers here? :(
fragalot: how about you make some thing like "hektor"
two strings attached to stepper motors
good enough for a plotter
I somehow don't see that working for engrave work
just to draw the outlines on your sheet of plywood
then you cut out with woodworking tools
now that is a mighty fine idea
even better would be a cable hexapod, but i digress
fragalot: can't you open the dxf in qcad to put the dimensions in?
yes qcad reads it just fine
BigJohnT: that takes forever, and it's still in inches, would have to convert every single one by hand,... as the silly app won't do it for me :p
nothing wrong with inches
can't you just scale the drawing?
you can switch inches/mm
archivist_emc: k, i'll send some complex thing to you in metric, but won't put the measurements on it, and won't give you any measuring devices that work in metric, see how much you like it ;)
fenn: orly, where?
BigJohnT: that would be an idea..
edit->current drawing preferences-> units
uh it's metric..
af far as I can see..
fragalot: Im swapping units all the time
archivist_emc: so am I, but I still prefer to work in the native units,.. esp. when all your tools are metric
inch is native :) to me
i for one prefer planck length
I have tools in both and the inch/mm button on digital calipers is really useful
* anonimasu nods
archivist_emc: only calipers I have at home are 0.02mm dial calipers
hmm play time
I WANT PLAY TIME!
off out to archive now
archive.. play time?
tutti"stranieri" o c'è anche qualche italiano?
I just managed to print it out in a tile
Guest879: I do understand a little bit :)
parlare e piu dificcile per me :)
cradek: sorry the ball screw didn't come out better :(
stuste1: I could keep trying... but I think the nut is worn on such a way that I can't get it perfect
turn a new one ;)
the ball sizes required to take up the lash just don't fit in the nut
probably - if you are better by .0003 on one side you could expect another .0003 and be at .0009 - still not as good as you want
oh it's much better - from .0036 to .0012
ok - then I would probably try the other side
it would mean ordering more balls again - I'm tired of it for now
I thought you were at .0015 when you started
besides I soldered it all shut so nicely
no it was .0036
does it have any tight spots anywhere?
nice - I would definitely go for the other side
it has one spot that's just a tiny bit more loose - way away from the chuck, I think it might be the home position
where it sat and oscollated? ;)
I just put the pieces together, and it's pretty damn big :p I got my work "cut out" for me cutting these pieces out by hand o.O
hi jeff :-)
* skunkworks thinks it might have been quicker to download the latest ISO
scale it down fragalot :)
BigJohnT: No. :p
the work table is 30"x30" and,.. Well, even if i scale it down, in terms of cost it practically won't change a thing,.. Only concern I have is that my steppers are only .4Nm
the parts won't be so big :)
same work cutting them out tho ;)
the part count doesn't change.
and if i use a 5mm drill, or a 3mm drill... what's the difference :p
* toastatwork is winrar
I'm off, cya
it is nice to be paid overtime to watch a cnc bridgeport and play xmoto
xmoto, imho, has some nasty physics
YOU CAN SHUT YOUR DIRTY MOUTH
yeah i agree sort of
i enjoy the bouncy physics
lean back,.. do a wheely, put the front wheel down, and the rear lifts off O.O
i do not enjoy the rough control
yeah, it's too crude
other than that, it's great fun.
lots of maps too
i get frustrated fast though
i prefer playing rrootage if i want to get frustrated
first time i played that game, i had /NO IDEA/ wtf was going on, with all those thingies flying everywhere
hm, as for the bearings I asked for a quote of they are 500eur..
it looks like I'll be using a thrust bearing that I'll preload with a known torque and a radial bearing just to center the ballscrew
@ 500 eur
the other option is angular ball bearings but they are hard to get too
they are sold by "SKF Multitech"
the place that makes/renovates spindles :p
toastatwork: it really looks like looking into air-bearings are worthwhile
yeah that sucks =(
well, for spindles atleast ;)
it feels damn complicated
diy air-bearings -_-
apparently all spindle bearings start at 500 eur and more
(for 20mm :P)
a 100mm id bearing = $$$$$$$$$$$$$$$$$
yeah, replacement bearings on some of our machines are like 5k
why dont you use the air bearings you make?
[20:43:13] <anonimasu_> http://www.aspe.net/publications/Annual_2000/PDF/POSTERS/EQUIP/DESIGN/SHANG.PDF
because we don't want to have to put 20k+ in man hours, matrials, and machine time into making a custom bearing we're only going to use one of
do you invest that much in making a single bearing?
I thought you mass produced them
we do mass produce them
btw, have a look
we don't make "Standard" bearings like you'd expect with ball bearings, though
if we wanted to make a ball bearing style housing for an air bearing, that would be a ton of work.
have a look where
no pdf reader
at the moment
ah your at work too, hold on..
[20:46:13] <anonimasu_> http://imagebin.org/27856
the tolerance seems pretty slack too.. about 0.0127mm.. for 400lbs of radial stiffness..
haha, they don't show you any of the important stuff
it's a whole paper..
but it takes a lot of work to make an air bearing
then do it man
it talks about line feed correction coefficients and stuff..
i have no idea what that is
to make the pressure more uniform over the bearing surface to get the maximum stiffness..
oh, because oriface bearing
yeah we don't have to worry about that
i am going to wget that paper and read it later
* anonimasu_ nods
well, you'r NDA:ed, though you might be able to tell me if im wasting my time ;)
i am nda'ed on the porous part
which you don't have the capbility to do anyway
but i don't see any reason why you can't do an oriface bearing
it seems like +/- 0.01mm is about the limit of precision you need
err no.. less then 0.01mm..
yeah dude, less.
be careful because a lot of research spindles are really sloppy and not rigid
but they're accurate as long as you don't load them up any, i.e. try to mill something
this seems rigid..
*shrug* just our experience.
there's a radial load vs displacement graph
well you just said it wasn't .01 mm
so that would indicate it isn't sloppy
10 um is about what we make ours to
how big is a microinch?
* anonimasu_ has no idea really
that's a millionth of an inch
when you breathe on something
the surface height will change by a couple microinches
when you turn a light on, everything will grow a microinch or two in a few seconds
well, they measured the stiffness to 100µinch with 600lbs loading..
measured with capacitance probe
that's not very stiff for a machine tool spindle.
it's stiff, but machine spindles are usually stiffer than that.
10 displace it 100µinch..
100 ui is .0001"
hm.. that's pretty bad
but like i said, 10 um is a pretty good gap
we do it smaller, but i would expect it to work
im not sure about how precise my lathe is..
the dro does 1µm..
not that precise, trust me.
we have a very, very good lathe that will barely do 10 um over about 4"
our air spindle lathes are more accurate but they're not useful for general machining like that
well, if you want to do 1µm stuff.. you need temp control for sure
you can do geometric features to 10 um
without temp control
as long as you are careful and do it quickly
if you look into lapping, which you will have to do
if you want to make a spindle
you should be able to build some laps and get your sizes at home.
turning stuff down to 0.01mm isnt usually a problem
but I cant measure less ;)
the problem is not measuring it, not so much
if you lap it well, you should be able to fit the rotor and stator without any measuring
btw, load capacity > 2180N
both axial and radia
porous seems cute in compairsion to orfice bearings ;)
uh, machine spindles have ratings in the thousands and ten thousands of pounds
im not interested in lathe spindles btw..(not as much)
it's obviously NOT a slouch of a spindle, but it's not a spindle built for machining.
or doesn't appear to be at my obvious first glance
maybe he's using it for some diamond milling or turning
does it say in the paper?
oh, it's a chemical mechanical polishing spindle -_-
the spindle might be 2" wide..
err 1" shaft ;)
does he go into the sizing of the orifaces and the windows
i've always wondered about that
i.e. how to size them and why
only that they have to be really tight
i am thinking that it's totally possible to build a diamond turning lathe at home
out of aluminum
* anonimasu_ nods
did he build the spindle in two halves
that's what we've been doing lately for a bunch of people, rather than building spindles we build pads that have a radius
and that way it can be off a little geometrically
it was just the bearing I think
because the rotor matters a lot more than the housing
like, a lot
you know what I hate?
not having a engineering degree.
aren't you cs?
there are all theese nice papers on everything on IEEEE..
like if you google for how to mount lasers to sharks there's a paper about it..
and you cant be member of IEEEE, unless you are a engineer or at a institution..
or throw enough $$$$$$$$$$$$ at them..
whenever I search for something cool I find "A paper describing the optimized design for air bearing milling sindles"
or something alike on there..
and no dice?
i don't think my school qualifies
being a 2 year
what's the link
slow internet is slow
i'm going to go out to the lobby real quick and see if precision machine design has any of the equations for air bearings
i know it has a chapter on them
and a lot of math
so maybe that can help
it probably does
though I cant do the math :/
(that's why I dont have a engineering degree)
[21:15:55] <anonimasu_> http://mdrl.mne.psu.edu/docs/PSMChap.pdf
another cute paper
dangit somebody moved all the stuff around in the lobby
and the person who did it went home for the day
[21:23:28] <anonimasu_> http://www.grc.nasa.gov/WWW/RT/RT2001/2000/2200bream.html
we have a cute bearing that works, but we cover our orfices.
it'll be loaded in a half hour
which one of them
the one i got right as i joined
my internet is freaking out atm
think it's faster if I send it to you
or use filebin.ca?
neither, i'm dropping random packets
i don't know why irc is staying online like it is
freenode is liberal with dropped packets ;)
google won't load
I've had the same issue
lol and i appriciate that
my link quality is 3%
i don't know why there are windows in air bearings
it doesn't make any sense to me
but then i've never built an oriface air bearing =/
probably because with a small hole you have a limited area for the air to spread out from..
some mentioned earlier about a ball bearing that didnt move much and its axis 'oscillated'. when the balls dont turn over 1 full turn thats called 'fretting' and requires special grease.
they are fancy devices
toastatwork: but that's just what I guess
anonimasu_: but why does that matter
it may also have to do with centering
tomp: sorry I cant find a good picture
toastatw2rk: I know why now
when you load a air bearing you close off the orfice creating a pressure rise and as the F=p*A ends up being large enough it makes the bearing fight to center itself the more you load it
[21:38:14] <anonimasu_> http://www.loadpoint-bearings.co.uk/products/how_air_bearings_work/
anonimasu: re fretting : http://www.tribology.co.uk/services/investigate/pb05-0.htm
i think maybe it was cradek's lathe near the home position.
that shouldnt happen
toastatw2rk: does that seem sane?
I think it makes sense
as if you close off just a small hole, you get no surface area and very little force
it makes some amount of sense, yes
for looking it up/explaining it
it's something i've wondered for awhile
haha, np im just obsessive with stuff I dont understand
also i don't know if you got these messages but someone moved the book
so i'll look for it tommorow
im in no hurry
I need to get better at math so I can make stuff make sense..
Hello All, I wonder if you could help me with regards to the error below. I get this when connecting via tightvnc to a vncserver running on the EMC server.
Application initialization failed: this isn't a Tk applicationunknown color name "Black"
When starting emc.
[21:49:22] <alex_joni> http://ubuntuforums.org/showthread.php?p=858323
most vnc servers don't have opengl support
so even after you solve that error, axis is not likely to work
you may have to use tkemc
or use ssh -x
(tkemc is also affected by the 'unknown color name "Black"' error)
Thanks Guys ...
good night all
OK just installed xming and tunneled X via ssh / putty (X11 forwarding) I can run xclock ok but am getting the following error when starting EMC....
_tkinter.TclError: Could not allocate font
Shutting down and cleaning up EMC2...
theres a version of caelinux with brlcad available as a vmware image http://caelinux.com/CMS/index.php?option=com_content&task=view&id=45&Itemid=40
that means that some font required by axis is not available.
the other lines of the error message that you didn't show might give a clue as to what font that is
if you haven't, I'd try installing "Xming-fonts"
(or bin that copy of windows and just use linux everywhere :-P)
Not a bad idea... lol
Weird... I unable to view web pages at moment. But can ping and use IRC .
ladder logic compiles to pic or avr , src available http://cq.cx/ladder.pl
oh cool -- I think I found that project once, then lost it again
only problem is, writing ladder logic in impeccable ASCII art is probably just as hard as writing correct "C" code
.. or does it only *look* like you have to write the ASCII art by hand?
jepler: i just figger his input code code be changed to read a classicladder file
that is just too bizarre, im on a single box and clone me just left the room
yah, tomp3 is me at home, thats ok
dareposte is now known as tom2
tom2 is now known as dareposte
must get a lot of work done with a clone...
do you leave him at home to load the cnc all day?
get up and put the clone to work, then go to work, and come home to a finished pile of parts and all your beer missing...
heh, remember laury anderson's clone... http://www.youtube.com/watch?v=27aLils4m_8
herself as a midget with a mustache
she used to host this great show called off-center from minneapolis