JymmmEMC: according to wikipedia, the bandwidth of single-link dvi is about 4 Gb/s, with dual-link being twice that. I suspect you'd have trouble writing data at that high a rate on commodity storage devices. For instance, the 15krpm SAS disk MBA3300 claims 179MB/s, so you'll need 3 of them running full-out to store 4Gb/s of data. The 3 drives'll fill up after half an hour.
jepler: I wonder if there's a a subset of DVI that could be used, though not sure it would do any good as there's no ausio on DVI iirc
Hey JymmmEMC ... Mr Gates doesnt want you near that stream!
wow, *here* is the best schematic I've seen in a long time. I love the antialiased fonts. http://www.epanorama.net/digipack/Digi_Pack/Board.jpg
LawrenceG: I know =)
jepler: you haven't used good schematic software
jmkasunich, what software to you use? Cadence is kicking my butt
at work we used to use PADS PowerLogic
now they're moving us to Mentor Graphics Expedition
haven't done a Mentor schematic yet, dunno how they're gonna look
do some editors clean up your schematics for you?
not that I know of
the look of a schematic is a combination of the library symbols, fonts, and the guy who drew it
ok, so all the schematics I draw would still look like crap
the symbols in that one jepler posted are hideous (IMO)
yeah mine always end up cluttered messes
the transistors with really fat vertical lines, ditto the caps
those are the normal eagle symbols so I've become immune to them. it's the jpeg artifacts that I was really bitching about
I don't know if the libs we get from cadence are normal, I'm tired of drawing footprints and schematic symbols for things like dsub connectors
it didn't register for me (at first) that it was a jpeg
that doesn't help the readability
IMO anything like drawings or schematics should never be turned into a bitmap until it is about to hit a display device (printer or screen)
fwiw I agree with you
I tried to plot my schematic to a pdf today, they told me they couldn't find my licence
you need a license to plot to pdf?
that's what I wanted to know
these days it sometimes seems you need a license to take a leak
solidworks does that with the educational license as well
you can do whatever you want except print the drawing without a educational use statement on the dwg
I don't know if cadence has the same limitation
the thing that I find pretty amazing is that universities don't teach how to make circuit boards
on a completely unrelated note, I absolutely hate milling tough steel with the side of an endmill
judging by the online tutorials, they just make chips
* jmkasunich picks thousands of tiny sharp steel splinders out of his fingers
I hate milling steel period
splinters that is - needles even
hoping to get flood coolant to help with that
* jepler notices that he's mixed eu-style and us-style resistor symbols on his schematic
(I started from somebody else's schematic then added us symbols)
what's wrong with steel
toastydeath mind your own business
ooh, it's late for me .. night folks
i like steel
you probably have flood coolant
and rigid machines
or a bridgeport?
toastydeath: if you use the side of an end mill to clean up a saw cut edge on a piece of A2, don't you get those fscking little needle chips?
I tend to get chips like that on side milling
the tougher steels are worse
jmkasunich: he likes the pain
12L14 you get regular chips
jmkasunich: i get beefy needle chips
if it's chattering i step it down a couple times
I was cleaning up the edge of something 3/16" thick
all i know is feed rate cures needles
if the machine will do it obv
thats my problem
i often drop the spindle speed a good deal to make up for it
on like a bridgeport
otherwise, i feel bad for your hands and can identify =(
full depth (3/16), 5/16" end mill, 1800 RPM, taking about 0.020 off, feeding at 2ipm
rpm too high?
I made a mental math error
yeah man you need like a 7-10 ipm thingy with a 4 flute em
I thought I was around 80-90 FPM, but I think I was close to double that
rpm is about right esp. if it is carbide
HSS, 2 flute
then you're about 5 ipm minimum
i would keep dropping the spindle and bumping the feed
I meant to go slower - I thought I was only doing 80-90
I also thought I was doing about 0.002 per flute, but it was only 0.001
anyway, the milling (and drilling) is done
nothing left but pseudo-lapping and polishing - think I'll start that tomorrow
anybody know any secrets for getting a really smooth and flat surface?
..other than lapping or scraping?
these pieces are already ground flat stock, 1" x 2"
I'm planning on sticking some 600 silicon carbide grit side up on the machine table and rubbing them around
then go to 1200
if they came from Enco, send them back
they come from slices off a piece of 3/16" x 2" x 18" GFS
kidding, I'm disgruntled with how the ground stock I got from them looked
I think I got this from McMaster - its been here for years, left over from some other job (about 1/3 of the original piece)
I think somebody's kid deburred them and then started making abstract art all over the surface
the stuff I got from mcmaster was much better
my goal here is to be able to stick these pieces on either end of a stack of gage blocks and use them to set inside mics, etc
I know I won't get gage block finish or flatness, but I want to get as close as I can
(they won't wring of course - my gage blocks are 1" square with a 1/4" hole thru the middle, I'll be gently clamping these things to the stack with a 6/32 threaded rod thru the hole
I know how to make them shiny, flat not so much
(how to make I mean)
i dunno man, without a legit lapping rig i am not sure how that will turn out
the trick with shiny is to go in 45 degree increments
so you knock down the ridges
toastydeath: I'm hoping that since they are only 1x2 it won't be too curved overall
I know the edges will get rounded a little, but I'll avoid them in use
turns out if you polish to 1/2 micron, the light bounces off
and you can't see the surface in your microscope
using oblique lighting
jmkasunich: i don't think it will be curved, i think you're going to get a bunch of high and low spots
but the only way to see is try, eh?
I have a zygo surface measuring machine that will measure how flat it is for you
toastydeath: yep, just gotta try
to less than a nanometer
thats a bit more than I need
the gage blocks are in 0.0001" increments, and I believe they are accurate to 50 microinches
it makes everyone look like a piker "you call that flat?"
jmkasunich: if you know anyone with an optical flat and a monochromatic light you can check your results
actually, I have an interferometric flatness tester that works pretty well
sitting under my desk
I have a surface plate and dykem high-spot blue - that will have to be good enough
fair enough man
heh, I'm glad I'm not going for parallelism
these pieces are off by 0.00015 over 2"
never quite figured out how to do the parallelism thing
parallelism scraping or lapping
lapping you just weight the high end
how do you measure?
comparator of some sort
I have a very nice mix
this one, only the -A model http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=140214555994
wow - 30 seconds of "lapping" with 500 grit took off 0.0001
I think it just took off the high spots from grinding - the grind marks are still _very_ visible
that is good
cradek: i think you liked pminmo's l297/298 for small steppers, any opinion on his A3977 pcb ( down to 1/8 step ) ?
tomp: sorry, no idea
jmkasunich: I know how to get a black polish on steel - but not how to make sure it's flat
how do you do the polish?
I have also fit parts using 500-600 grit cloth paper, and you can do .0001-2, but it never turns out at all flat
oilstone dust and then diamantine on brass
(and I've only done tiny parts)
a lap IOW
wonder why it doesn't wind up flat?
I think you get a sphere, but I can't say why I think that
because you are only using two surfaces
one is going to wear concave, the other convex
yes but why?
if the radius is sufficiently long though.....
because the rates of wear are different
even in the same bar
plus you are not starting with two falts
but two irregular surfaces
if one piece is bigger than the other (part is usually smaller than the lapping plate) the bigger one will go concave (less wear at the edges)
I think you can get .0002 - just mic it all over, and rub the high spots more
that will just tell me that the sides are parallel to 0.0002
but obviously, I'm describing scraping
I didn't read back far enough to catch what your actual goal was :-)
I'm already parallel to better than that
one side flat and smooth
other side doesn't matter
if you have three parts
you should be able to actually lap them flat
no sandpaper, no guesswork
I think that's only strictly true for three round (or maybe square) parts
mine are 1x2 rectangles
you should still be able to get them flatter
via three surfaces than without
can you just "borrow" a surface grinder somewhere nearby?
besides, the pieces were ground to start with
I'm trying to go from good to very good, not from crap (milled) to good
out of my league/skillset I think
i don't think you are going to actually beat ground
without some prior practice and verification
but obv. that depends on the quality of the initial grind.
I'm getting much smoother than the original grind
don't think I'm getting any flatter
SWPadnos_ is now known as SWPadnos
SWPadnos: among the living?
[04:15:43] <tomp> http://www.mini-lathe.com/Lapping/Lapping.htm
has some info, but they ended at 600 grit, which seems to be where you started =-O
lapping with paper tends to round the edges - the paper is too squishy
I started with paper, but then I took some of the oil/abrasive slurry and I'm rubbing the two pieces of steel on each other
lapping plate? ( cast iron thingy with squares, actually with grid of slots & lapping paste )
a toolmaker friend may have one
heh - I've used my table saw table (cast iron) for that
in fact, right now I'm alternating between rubbing the plates on the saw table, and rubbing them on each other
* alex_joni yawns
Power supply is done, gives me almost 34V!!!
Next we get EMC figured out for the HobbyCNC card...
Wheee, do that tomorrow, I'm beat.
Anyone in here use anything Sherline, or similarly mini, or is it all Big Toys with you guys? :)
i have some vaporware
here's some pics of my lathe, i guess it counts as mini: http://fennetic.net/machines/lathe_modifications
nice, you've done a lot of modding
with servo's: http://fennetic.net/pub/camera/DCP_0785.JPG
ah, the dream
or is that necessary with a lathe?
well, there's no electronics yet
but it will be closed loop
I just finished my first real modification to the Sherline mill
[06:39:55] <gfixler> http://www.flickr.com/photos/garyfixler/collections/72157604713088268/
a built-in vacuum system with 3/4" loc-line tubes
i think i'd just use a regular vacuum hose and a positioner like for dial indicators
yeah, I was just feeling like being fancy
I spend far more time building things to help me build things than I do building things
that's a good thing
I figure it's all experience
i like your rubik cube. did you ever sell them?
thanks - no, I've been too busy
the Make crowd picked up on it, and sort of went to town with it, and I've since seen someone from China selling versions of them on Ebay for about $16/ea.
how does it feel to be a post-industrial non-millionaire? :)
my ship will come in, I just know it :)
new CAM for EMC2 has been releas on gcnccam.sourceforge.net
Guest104 is now known as skunkworks_
jepler: step-config is awesome
I don't know how well this thing will work for circuit boards.. It is cable run and seens to need around 650 steps per inch with the driver set to 1000 steps per rev.
it flys though.. I was running the axis at 540ipm acc set to 50in/s/s
one thing - the fastest I could run the arcspiral was around 130ipm
that would be 133 full steps per inch
(I think it probably should be 666.666 steps per inch(I just did a quick measurment to get it close)
skunkworks_: what hardware are you playing with now?
[12:52:44] <skunkworks_> http://www.youtube.com/watch?v=zKZYj51Exn0
I just hooked the steppers into some parker drives I have.
that was the old control in the video?
'fluidmove' :) dos based - serial to the controller in the machine.
I had to take it pretty far apart to get to the steppers to see current and such.
love your compaq 386 machine
ah, newer than I thought
It is the only place that the software for the thing resides.
ooh it is?
hmm, maybe a backup copy (on a floppy?) is in order
it was a trip down memory lane.. It has windows for workgroups on it - and wolf3d
666.66 steps per inch would give .0015.
but I don't know how acurite those would be.
(1000 steps per rev driver)
might have to borrow an indicator. I could up the stepper driver to 2000 steps per rev)
skunkworksThere was no overdrive, there was DX2 and TURBO though =)
I am looking right at it - Intel overdrive POPD5V
Time to upgrade to Hardy Heron, hopefully I'll be back soon :)
skunkworks I dont care if I have an Overdrive CPU in my hand (literallly), DX2 & Turbo
That is just goofy talk
skunkworks Qty: 2 actually
has anyone played with this? http://gcnccam.sourceforge.net/
we are waiting for you cradek :))
saw some nice toys at the big UK machine tool show yesterday
cradek: Do you know the developer?
cradek: I dont recognise the name... Arnd Begemann
cradek: I guess not to shabby since it's rated as beta and only been around since February.
Though, I suspect he was working on it longer than that.
to set up a computer to use with the CVS version am I better off just installing the live EMC cd or unbuntu?
the live cd is easiest
yea! third one is the charm it actually boots up from the live cd
the dload of gcnccam is behind the dload from HardyHeron ;)
sourceforge is running a bit slow today
jmkasunich: you there?
* dimas just finished soldering L297/298 driver board...
anonimasu: he's never around on weekdays except evening
heh, I never realized that =)
I dont know the time there
* anonimasu is digging into bearing tolerances and stuff
trying to grasp what kind of precicsion you need to make spindle housings to
I'm not having any luck building gcnccam
Sure don't know, but when I upgraded the uMill to belt drive it came with a new set of bearings. I would thing really tight.
I'm using oak and skakeboard bearings for the head on the Pipe Dream :).
im talking about p7 machine tool bearings :p
that's abec 7
Power supply is finished - ~34V. Next I need to get EMC going for the HobbyCNC drive and then hook up the steppers and see what happens.
and they probably cost around 800$
im looking at it because I'd like to get a quote how much a spindle housing and a shaft will cost to grind
anonimasu: Sorry, can't be any help with it, WAY over my head at this point :-(.
* skunkworks_ normally just wings it.
skunkworks: so you'd wing it if you put 800$ into a set of bearings?
cradek: i tried dbl clking the deb and the 'package installer said there were 6 xtra pkgs needed. i ok'd that, and its still hung on file 0 of 6.
not like the xfer wont occur, but is like it didnt start
tomp: it may work for you then. ubuntu.com is extremely slow today for the obvious reason
but, I wanted to build it myself
ah! didnt see the files were from there, best o luck
* skunkworks_ wouldn't probably ever pay $800 for bearings..
I wonder how good "standard" bearings are
just not extremely good
i hate it when firefox just suddenly dissappears, blowing away a dozen related tabs
archivist: the outside tolerance of bearings are the following: from skf's website
Normal(cln/cl7c) = 0 to -13µm
and P5 = 0 to -9µm
or buy balls and grind ways on the spindle
[16:03:34] <anonimasu> http://www.skf.com/skf/productcatalogue/jsp/viewers/tableViewer.jsp?tableName=1_0_tt3&maincatalogue=1&lang=en
[16:04:38] <anonimasu> http://www.skf.com/skf/productcatalogue/jsp/viewers/tableViewer.jsp?tableName=1_0_tt1&maincatalogue=1&lang=en
the first is p5 tolerance the second is normal
Kia is the runout
if you look at a 30mm id bearing 5µm
for p5 tolerance vs 20µm
that's 3 times more runout
and 15µm = 0.015mm
cradek: fenn: gcnccam nicely dies with just one rectangle from qcam in either r14 or 2000 dxf format, when BUILD is chosen.. i'll send the pm to the author.
dies nicely means he has a I died, wanna restart button
As reported yesterday, I filled in a bug-report on the parport in-mode stuff
did you ever get your inputs to work?
nope, I bought an other PCI parport, now i have 15 inputs that do work :)
i find it a stupid sollution dough
it's nice that computer hardware is so cheap - but it's too bad it's such a crap shoot
while reading on the parport subject i discovered that on some ports you'll need to set bit 7 instead of bit 5, I added that to the report
did that fix your troublemaker parport card?
don't know, i'm not running a source built version on that PC
since a parport in "in" mode still has 4 outputs, I bet few or no people have run into this bug. but, I agree that it would be nice if it was possible to run as input-only.
Yeah, it even wouldn't slow down the parport component since it only has to be set once
it would make it even faster, not having to set the outputs to zero
sent email report to abegeman re: gcnccam. be advised the app has a report error button which uses bugbuddy, but bugbuddy didnt list his app, so cant be used.
Rats, no emc config for the HobbyCNC drive, back to the manual.
fenn: did you get gcnccam to work with dxf from some other app than qcad?
DXF is a moving target
I wrote a cleaner for one programs dxf out so we could send files to a laser cutter
fwiw linux parport_pc only sets bit 5 to enable input on the data lines. http://www.gelato.unsw.edu.au/lxr/source/include/linux/parport_pc.h#L146
wheres the md5sum on the hardy heron ( before i burn it ) wasnt offered at ubuntu.org?
tomp: official release?
[17:00:57] <tomp> http://www.ubuntu.com/getubuntu/download
doesnt mention it
[17:01:45] <alex_joni> http://ftp.astral.ro/mirrors/ubuntu.com/releases/hardy/MD5SUMS
tomp or burn and there is a self test
thx, i got the md5sum of what i dloaded from k3b, need the reference to see dload was ok
tomp: save the link I gave you.. you have the MD5SUM there
yes, got it thx
cool, they agree, now if they agree after baking ;)
was looking at some tiny machine designs, this is tiny http://www.youtube.com/watch?v=gVQpDFVyPGg
the xy is belt drive, the z is screw
hey tomp it's been a while
how are things going?
crazy. looked like i had a lot of work, now clients are backing out of some deals. hmmf!
but interesting :)
gotta keep your head up ;)
* anonimasu sighs
the more I dig around the more I want a datrondynamics machine :p
i saw that, it's amazing. love the sound
it's make nice graphite electrodes
I crave a machine that sounds like that..
on the other hand
there are lots of nice machines
[21:20:55] <anonimasu> http://www.youtube.com/watch?v=LLC25KbCTxs&feature=related
I saw the Citizen machines in the flesh yesterday....dribble.....
got a video of what you saw on the web?
we need a 12000rpm spindle
that you can DIY
faster than DIY?
as in .. buying?
faster spindle ;)
he wants 40krpm
[21:25:16] <anonimasu> http://www.youtube.com/watch?v=ZPqJTPvvFMI&feature=related
on the other hand thats not bad eitehr
[21:26:05] <archivist_win> http://www.youtube.com/watch?v=gOil5JKjP_4
[21:27:47] <archivist_win> http://www.youtube.com/watch?v=WxW2rlsgxU0&feature=related
another night hooked on cnc movies.
[21:29:30] <anonimasu> http://www.youtube.com/watch?v=uAcBU2iqkKE&feature=related
<- that's probably as good as it gets
the other fraisa videos too
hm, that video of the mach 3 machine
3000rpm and 6000mm/min feed
with a 2 flute cutter
i got a response from Arnd Begemann already, he's author of gcnccam.
he says you need a couple named layers in the dxf file, will try, the report back :)
i have named layers, its just that nothing happens when i click on them
(3000 * 2) / 6000 = 1mm per tooth
actually it's 0.9906.
3000*2/6000 = 0.9906?
no, but the online calc I found calculated it as such..
but that's in "
what's the advantage of doing high feed low depth of cut?
faster metal removal
seems like you'd wear out the tip of the cutter
I have no idea
that chipload seems huge..
good enough explanation :)
dormer recomends 0.06mm/rev of feed..
also it would tear as it comes off
for most endmills of theirs..
fenn: I think it has to do with what kind of spindle speed you've got
and machine rigidity
there IS a tutorial for gcnccam on sourceforge, a pdf.. so i rtfm 1st :-[
tomp: did you see http://www.youtube.com/watch?v=ZPqJTPvvFMI&feature=related
(I dont get how it works with that chipload)
Bet that would raise hell with steel !
I dont get how it works in alu
it'd say Schlurp for me and it'd gum all over the place
shutting down all youtube till i test the fix for gcnccam !
* anonimasu is jealous of that chipload
what a bastard -_-
I want the small live tooling off the Citizen
* anonimasu nods
[21:56:38] <anonimasu> http://www.youtube.com/watch?v=OT6vobT9mAw
is it just me or does that servo sound non healthy
yeah sounds like a stepper
seems like it's the drives that makes them sound like that
yow! i got gcnccam to produce gcode. i need maybe 40 or 50 more whacks at it to document how. http://pastebin.ca/997857
there's work to be done, that was a 10" by 7" rectangle and a 1/4 inch dia tool . told app to cut inside, there oughtta have a bit more motion
I'm getting some gouging error running the pastebin ,
tomp: what was the raduis used for t1 ?
i cretaed square corners ( on purpose, to see what;d happen )
i think i posted the dxf earlier...
Drive is finished, got the heat sink drilled, tapped and installed along with some standoffs!
Once I figure out the configuration for EMC, I get to take my steppers for a spin :)!!!
I took the basic stepper config and hacked it to correspond to the HobbyCNC inputs. Not sure about a lot of things, but assuming reasonably defaults, it might work!!!!
K`zan, get 4 running and play daisy
Tooooooo Coooooool :)!!
archivist: Huh? You speak Greek to me :-/.
Only got 3 steppers, the extra axis is a spare.
daisy.ngc in with the demo files
Daisy is in the examples, plays daisy on your steppers
sounds ok with 3
Hummm, now I wonder if having no stepper on the A axis will blow the driver.
Doesn't day anything in the docs, could that be because they assume that if you ordered a 4 axis you are going to have steppers on all axis?!?
SO much for any tests today, gonna have to ask about that. Not willing to toast a driver chip, they are hard enough to get in, I can imagine that out would be fun too.
Check on cnczone , you may find an answer
good night all
My search critera doesn't seem to help, will try some more.
nite alex, thx again, & HardyHeron is running :)
SkinnYPupp: did you check the code using AXIS?
I loathe google, snarl, growl, snap...
No idea. Posted a message to the HCNC group and we'll chill until we hear something. Need to roust up some ethernet cable for the steppers for once I am done testing.
K`zan: drivers do no like running without load
re: gcnccam, after fixing ( circumventing ) the cutter comp problem, it's doing quite nicely
tomp: Yes using axis, gouging at line 59 unless I comment out the tool
I see he has a tool table and AXIS/EMC has a tool table, making them agree didnt help, but removing G41 fixed mine
I figure it's my understanding for right now thats the problem ;)
anonimasu: Thanks, looks like I'll have to come up with another stepper. Bummer, gonna be a week before I can play :-/.
[23:03:14] <tomp> http://imagebin.org/17059
K`zan: no way to disconnect one driver?
looks good! I'll have to install it tonight .
K`zan: what does the jumpers do?
Whats the stepper winding resistance, wonder if you could get away with using a speaker as a dummy load
now that would make a racket! daisy at 120dB
lol , gotta run
Wow, got several responses to the query already - OK to run with no stepper connected, just do NOT connect / disconnect with power on.
Cool, off to do a temporary wire up of the steppers, no real point to hacking in the ethernet cable to the stepper before I know how long they need to be!
re: gcnccam to delete a layer ( operation) delete it's name in the dialog :)