jymmm is going to be so mad at me.. ;)
[00:25:13] <skunkworks> http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&rd=1&item=320155952581&ssPageName=STRK:MESE:IT&ih=011
skunkworks: wow, you win
and hey, so do I, I just got back from an auction - machine shop in a nearby town closed
got a few nice things.
cradek: cool - what did you win?
how is the vacuum table coming??
the vacuum table is almost done - just the little holes to be drilled and then figure out a way to hook vacuum to it
at the auction we got a 2hp floor drill press, 12? inch rotary table (half of my future 5-axis project), tapping head, tooling, etc etc
cool - your thinking of a trunion?
actually the rotary table might be more like 16", it's big
actually this one will mount upright - at first I could just bolt an angle plate to it and mount another (small) rotary on it
I have a couple (very) small ones
exactly - most of the projects I can think of doing with 5 axis are pretty small.
yeah that's probably true
* skunkworks doesn't plan on making engine blocks out of billet aluminum ;)
there was no cnc anything in this whole shop
I could tell because the vises and machine tables were still in good shape :-)
a working bridgeport tracer mill sold for $300
yikes - was i tempting?
and I didn't bid on the 40"x108" lathe...
not in the least!
who wants a hydraulic mill
not I. have one already :)
the 40" lathe had a lot of interest - sold for $12k which really surprised me
"oh and sorry, we just sold the forklift that could lift this - you'll have to figure something out"
the guy who bought the forklift was helping people - very nice
good - there are still people like that.
he loaded our drill press into the pickup
and by sheer luck the drill press and tapping head are both MT3
floor drill presses are nice. I still have to put mine back togather - kinda was waiting for the garage. (in the basement right now.)
yeah we only had a little tabletop drill press - now we'll have to find someone to give it to
anyway, fun but tiring day
I like those
think you'll get more bids at the end of your engraver auction?
hope so - but I am happy where it is at right now. That many bids so far from the end of the action is good.
I am hoping for an ending push :)
yeah very good sign
you might get it!
I bet r***l comes back
it's apparent that for another $100 he could get it
oh I read it wrong - maybe not so
t***c's bid might be somewhat higher - but probably not too much since he hoped to get it for 15
if u***a wants it, he'll probably snipe in the last minute or so
or a few of them..
yeah and they're both cursing r***l right now
r***l may not have the experience for sniping - hard to tell
idiots bidding early. - But I like that.
hmmm. probably not. usually snipers put a bid in, then let others bid it up, then one laast one at the end - not the r***l pattern at all
heh - nice return for you at least :)
I have seen someone come in at the end and get it - one that hasn't bidded previosly
yep. I do that sometimes :)
its up to almost 50 watchers.
atomic clocks are so handy
dad does that.
I think that's the only winning ebay strategy
did I mention lately that I hate ebay?
you can win by just bidding a shitload of money also
no, I don't think you have
does a couple of refreshes to get an idea of the internet speed.
heh - I usually snipe around 5-10 seconds before the auction closes
SWPadnos: yes but that will cost you (much) more
I meant more by "win" than "get the item"
you don't have to get the page back within the auction limit, the request only has to get to them in time
cradek: you're really upgrading your shop these days
jmkasunich: yeah what the heck
some people have boats. :)
or motorcycles - oh wait
or old cars
2hp is a pretty beefy drill press
I just got these: http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&tem=110080947943
jmkasunich: it's definitely heavy, but I doubt the 2hp rating because it's 110v 15A
how many rpm?
I need to make adaptor plates for my motors, but it's a good start
is the motor?
skunkworks: not sure
sounds like shop-vac or router HP
SWPadnos: the link is dead
it goes down to 150? rpm with two pulley reductions so it's for big drills.
actually, if a drill press is rated in HP, that means it must be relatively new
usually they're rated by the size of the largest circle they can drill in the center of
[01:11:27] <SWPadnos> http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=110080947943
it has that rating too, it's 20"
ah - I had deleted the i from item in the first link
thats a BFD
yeah it's definitely big enough
mine is a 15" clausing, and its a nice heavy drill press - 300 lbs or so
yeah - I think I can now do the conversion, with all that spare time I'll have during the frigid winter
a 20" probably weighs closer to 600
cradek: varispeed drive?
150 - 4500 or something like that
step pulleys, not adjustabel ones
any idea who made it?
either this one or the other one said craftsman somewhere on it
FWIW, Enco has a 1.5HP 20" with varispeed, $2K and 750 lbs
craftsman eh? that explains the 2HP
you didn't pay 2K I hope
oh hell no
then why damn?
that seems like a lot, but varispeed adds a lot I'm sure
I couldn't tell from your tone of voice whether that was a regretfull damn or a "cool, I got a really good deal" damn
I got several good deals today
the latter is usually pronounced "day-am"
ok, fwiw, I don't know all my racial slurs either
when I vary from standard english I sometimes get lost
anyone tried stepconf yet?
hi speak of the devil
cradek: alex looked at it at least enough to spot a typo
but I think when he tested it I'd missed checking in some files ..
jepler: more than I've done, even though you've asked me several times to look at it
cradek: oh well .. now it's even easier for you to try it
cradek: I hope you bought a small crane to go with your rotary table
yeah that thing weighs a surprising amount.
the two of us carried it though
enco again, horizontal/vertical is 200 lbs
as short a distance as possible, twice
I'm sure this does weigh 200lb or more
lifting it onto the mill table is where it gets interesting
just lower the knee to the level of the table...
you want a cart that is roughly the right height, roll it to the table, run the knee up or down, and slide
you said it might be bigger than 12"
I think it said 230mm on it, but that' 9" so makes no sense
unless it's radius? but that would be odd
I'll go measure it
that is a good roi
and a load better than you'd make running PC boards for cheapskates
less hasle I hope.
jmkasunich: 320mm (12.6")
dyslexia kicking in again I see
nah I don't blame any particular disorder
that's about the size of mine, except mine can't do vertical
230 -> 320 = dyslexis
yeah I'm glad I can mount it vertical
I kinda wish I could, but oh well
much bigger (radius > quill travel) would be no use
my little 4" can be vertical, so it's OK
flat rotaries aren't all that usefull with CNC
right those are really for manual machines
no,but it has been useful when doing things manually the last 3 years :)
soon, it'll be useless :)
you just need a large angle plate
cheaper than a new rotary table
definitely need a crane then
(rotary + angle plate)
how about a 20" tilting rotary?
weight? well, if you have to ask...
slightly more than a Bridgeport :)
not that bad
I think the enco "phase II" is it, or virtually the same
but it will make your knee motor struggle
[01:42:39] <jmkasunich> http://www.hgrindustrialsurplus.com/images/products/08-001-313.jpg
$400 before shipping 250lb...
I saw that one
jmkasunich: wooo that looks nice
that looks nice - what was the pirce o
nah - have 'em hold it until the next time I drive through :)
$1k for a good five axis setup seems like a great deal
wonder how easily it could be cnced
here's the complete URL
[01:44:25] <jmkasunich> http://www.hgrindustrialsurplus.com/search-products/product-detail.aspx?id=08-001-313&searchtable=1&sortExpression=&SortASC=&pageSize=50¤tPageIndex=0
yeah - that does look like a good deal
this month's saturday sale was yesterday
(they're open one Sat a month)
20" is pretty big to put on the table of most machines
yeah 12 would be better
if the table is 20, I bet the overall size is close to 3 feet
but - what if you want to make engines from a single aluminum billet?
it looks like the table is the widest part, but the overall length does look to be 3'
I know - let's get Roland to put it on the Mazak!
so you could stick it on a 10x50 table maybe, but a 9x42 probably not
heh - that would be fun
it would fill up the table
I've seen smaller units similar to that one in the past
sure, but the table can move a fair ways on the mazak
usually pricy (by my standards anyway)
I think the mazak has 40" of X and 16" of Y
it seemed like more, but that could be because it's a big table you really want to get out of the way for
actually, it is 28 x 15.5 x 8
(just checked the demo_mazak ini file)
so it could be 30x16x8 or thereabouts, less some safety margin
I'd rather find a two axis table in a more reasonable size
well, with a rotary table, it kind of doesn't matter. you can rotate a larger part so the work area is within the tool envelope
though not with 5 axis
so I guess an XY rotary table can be useful with CNC, if only to swing large parts into the work envelope :)
that's what I've got
$130, delivered :) (though I think the guy overloaded his trunk with my table and 2 6" vises
yep - got the 12" rotary for $130 and 3x 6" machine vises for $85 each
err - 2 vises
it's uvc (par to x&y then rotates about Z ) 12" great deal
it sits flat on the table and rotates about Z (it's a C axis)
yep u v c
it is a good deal, those are usually $500 or more used, $2k or more new
about $1/pound ;)
the trick to a troyke is knowin that letting the cutter sit still will marr the part... the trick is "which handle do i turn now!?"
this only has one handle
except for the locks
not crossslide rotary?
nope, rotary only
scusa, yep only C
big fat heavy duty rotary
greek isnt it? greeks made helios calipers ( best non-dial caliper ) and troyke
how to adjust troyke rotary tables http://www.practicalmachinist.com/cgi-bin/ubbcgi/ultimatebb.cgi?ubb=get_topic;f=21;t=000113;p=0
btw built in cincinatti :-[
at least greek food is good :)
even if we don't get their rotary tables
you know 'oopa' means 'migawd the cheese is on fire'
[03:48:50] <LawrenceG> http://imagebin.ca/view/kpVsKrOF.html http://imagebin.ca/view/HepZ6K.html http://imagebin.ca/view/Mzzx6f.html
lathe spindle encoder
now for the electrical part...
neat! how many holes is it?
if your sensor shows 50% duty cycle (I think it might, you did a good job sizing the holes) you could try counting on both edges
yes... figured 90*4 should give a count every degree
intersted in the gcode?
oh you will have two sensors for quadrature?
3...a b and z
yep - I see the index mark - cool]
is this a shoptask?
looks like the Diablo version
yes... 95 or 96 version
(at least it's yellow like the diablo :) )
some photos and the gcode might be nice on the wiki
you're not the first or last to convert one of these machines
oh - eldorado, not diablo
very excited about getting threading running
gotta get to bed... goodnight
night. see you later
CNC threading is really really nice
hmm. when you get it working, could you make me some 1.75x1/2" shoulder bolts, with about 1/2" of 3/8-16 thread on the end? :)
[04:08:00] <LawrenceG> http://pastebin.ca/700315
... named variable, owords, subroutines... the works! thanks lerman
bolts would be fun
... tou will need cvs version of emc to run the file... 2.17 doesnt have named variables yet
I'd need a CNC too
that's what the bolts are for ;)
hey if everyone makes you a part, you will have a machine by next fest!
I got some servo mounts off eBay
they're new, and have bearings, belts, and sprockets, which is cool
they also have metric screws that don't fit the Bridgeport
and the X mount has 1.75" deep screw holes
and of course, the mounts are for NEMA34 motors, and mine are NEMA42, but that's reasonably easy to fix, even with a manual mill
np... get the stick welder out and tack them on!
I could just manhandle the metric screws into the imperial holes
will cap screws work or do you need a custom bolt?
they're available from MSC
at $3/60 each ($18 for a 5-pack)
if I want stainless or something hard (or non-import), they come in hundred packs for $loads
must be nice bolts :}
I hope so
trouble is they don't tell you how much thread is at the end
they only give you the shoulder length/diameter, plus thread pitch/diameter
only $998.05 for a 100-pack in stainless
yeah - and no singles ...
sounds like you will have to design them into your next project at the office
hmmm. actually, the 5-pack for $18 is Holo-Krome, so they're probably pretty good
grade 8 alloy steel
that should be strong enough to hold ~25 pounds of mount and motor (with 4 screws)
well... I think its time to go put my feet up and relax... goodnight Sir
oh shit. I think those bolts may be $18 each O_O
the latest rtai magma patch is clean against the current ubuntu gutsy beta kernel
it looks like git is enabling the ubuntu people to track the vanilla kernel pretty closely .. and signs are that the rtai magma patches are being generated from a git tree as well
Hey jepler tell me a bit about git?
it's another version control system -- an alternative to CVS and SVN
it was started by Linus, and intended to be better than the others for "distributed development"
Ah I remember something about Linus starting something else.
where I add some features and you fix some bugs and then a third person can easily incorporate all of those changes along with his own changes
and for the linux kernel it seems to be working
(well, I'll let you know in 10 hours when I can actually try this realtime kernel I'm trying to build...)
Better you than me! I'm glad I quit building kernels a while back.
I didn't fill my quota yet
so I might give it another go sometimes
I hope someone else does the next ones, I want out of the business
(or just save it for later)
in addition to the centralized model where developers check in patches to one server, git lets any individual developer have a revision control tree which can be private or accessible to others
cradek: I did the dapper though
and a good job you did too
with your help though :)
maybe you can do impetuous impala
The central repository knows about or can know about and use a private tree?
rayh: not until someone pushes the changes from the repository below
rayh: git (the software) doesn't have a "central repository" which is different than the others, though the communities which use git do
it's really very distributed
rayh: changes get from one tree to another by a developer doing a "push" or "pull" command
for instance, if I have write access to a server, I can make some changes in my private tree, then "push" them to a public tree
then alex can "pull" them from the public tree to his tree
That's similar to our current CVS, so what are git's advantages.
oh, my very own tree.. wouldn't that be nice
rayh: it's like having multiple cascaded CVS'es
with information beeing able to travel up or down branches
one advantage is that once you do the checkout (called a clone), each developer has the entire history on their local drive
so I can see who made what change when, while I'm sitting on an airplane
rayh: yeah you'd really like the thing that SWPadnos is describing, because "git diff" doesn't have to use the network at all
How much bigger would my initial clone be?
Than my checkout is?
it contains all history, so about 20-40% in emc2's case?
there's another neat thing, called bisect, which semi-automatically finds changes that cause problems
sticky bit has come loose
git isn't as space-efficient as cvs, so it may be double the size. jepler did some looking but I don;'t remember the numbers
"changes that cause problems" ??
You mean from one person's tree to another?
sure - if there's a bug in the latest version, but you knew that the version from last tuesday didn't have the bug, bisect can automatically find the change that caused the bug
changes that might have caused the bug
it applies half the intervening patches, you test to see if the bug is there, if so, it applies half of the unapplied changes ...
* rayh is a bit skeptical about finding rt bugs that way.
if the bug is there, it reverts half the changes ...
oo automated regression testing sorta?
you still have to test manually
or if not manually, then at least separately
no, automated tracking of which changes between certain dates have been apllied :)
Ah like a compile bug rather than a runtime machine issue.
but git-bisect automatically chooses what revisions to test, based on whether you say a particular revision is "good" or "bad
it could be either, but you do the testing
well.. can be both, as you test manually
it's like playing high/low
unless the Trojan horse ran out of hay
fresh checkouts comparing size for cvs and git:
okay. now for the loaded question. Are we already using git?
but with emc2.git I can look at the whole history of the project with the network cable unplugged
no, we're still using CVS
can you delete files from the git history?
can we parallel git and CVS?
rayh: it's possible to import the history from CVS to git, that's where emc2.git came from
"git-cvsimport" does a pretty good job of it
rayh: but it's impractical to assume people will commit to both systems
it's mroe likely they use one system, and the other gets synched systematically
git-cvsimport is strictly one way
It moves history to git?
while git is neat I don't know that it offers enough advantages to justify any disruption changes might cause
I mean, to change from CVS to git
and having two revision control systems publically is a recipe for confusion as well
fenn: I don't know how to remove something from history in git .. it's probably possible-but-discouraged
Thanks for the explanation.
i cant seem to find the manual page for git.. or the command :)
you may not have it installed .. on dapper the package is called 'git-core' though
[14:25:16] <jepler> http://www.kernel.org/pub/software/scm/git/docs/
^^ online documentation
weird, i did apt-get git and it downloaded... something
git is something else .. Description: GNU Interactive Tools, a file browser/viewer and process viewer/killer
so the revision control system package is called git-core
also make sure you install gitk to graphically view the history
alex_joni: thanks for fixing step
you can get a copy of the emc2 history in git format by: git clone git://git.unpy.net/emc2.git local-directory-name
but don't do it frivolously; this is on my home dsl and it takes quite awhile to do
is there a reason it takes so long? does it use compression?
it just transfers more data than a cvs checkout
and ity's uploading from DSL, which is generally pretty slow
[14:54:43] <rayh> http://wincent.com/a/about/wincent/weblog/archives/2007/07/git_notes.php
a link to why some folk have switched to git.
why some folks have not switched to git: http://log.ometer.com/2006-10.html#20.2
it looks like you can use git with a centralized model just fine
it should be noted that git is actually well-suited to maintaining local changes (such as config changes, or jlmjvm's local change to Axis), while still allowing you to grab updates
so one could use SCM on their own configs (lets you roll back something that fubars the machine), and also pull in changes made to the central repository
SWPadnos: what? you mean you arent already using a local repository to store changes to your config files? :)
well of course *I* am, but you know - normal people could too :)
that was an interesting use of CVS I thought. I read about some people using it to manage /etc
change the server config, if something bad happens at the next reboot, you can undo changesets until it works again
windows had something like that, they just backed up the whole system folder though..
* fenn spares you all a rant about versioning filesystems
not sure if I was competing with anybody, but 'git clone .../emc2.git' took about 10 minutes wall time over my dsl
you could set up a git repo on dreamhost i suppose
yeah if we switched I think that would be part of the plan .. git repos can be on dumb http if you want
it would pull from cvs periodically
"jeez, I can tell a (kernel, Haskell, shell) programmer wrote this."
alex_joni: from the link i posted
from fenn's link - a guy saying "huh? it's complex and I don't need this" about git
i dont necessarily agree with him on that point, but he's obviously a smart person based on what i've read of his blog
a simple 2 minute "intro to git" can probably get you 90% of what you need to know
haha - what a great spam subject line: "one-eyed monster-charger for you"
certainly beats "cwdz10j"
M2CW concering git. I'll support it if we decide to go that way.
there's a broken link there
(the 'read the documentation' part)
Gotta run. Thanks guys.
alex_joni: but the real docs are online too: http://linuxcnc.org/docs/devel/html/config/stepconf/index.html
jepler: I know.. just ran into it
thanks for fixing the url there too
* jepler <-- plainly incompetent
heh, why do you say that?
it happens to me all the time too..
I blame it on distractions
v2 could be called 'tillyconf' and add support for 12-strut oligopods
umm - sure.
skunkworks_: it's a SMOP
:) my engraver sold. Now I will see if I actually get payment :)
are you using escrow?
or is that part of ebay these days
I expect full payment before shipping.
fenn: I saw that phrase catched
i'm a filthy meme-hole
although to be exact it can only be a 9-strut oligopod in emc2
g-code restricts you to 6 axes anyway doesnt it?
hmm.. ok so i guess you could translate both hexapods together, but not rotate them together
you can rotate them together, not sepparately
you could have a mode switch command too.. ick
lock platform 1 rotation
i dont want to think about that
M62 (-> halcmd setp oligopodkins.switch-mode 1)
emc was compiling soo good untill
Compiling realtime emc/motion/command.c
emc/motion/command.c: In function ‘emcmotCommandHandler’:
emc/motion/command.c:1149: error: ‘NULL’ undeclared (first use in this function)
emc/motion/command.c:1149: error: (Each undeclared identifier is reported only once
emc/motion/command.c:1149: error: for each function it appears in.)
what data is your checkout?
the current CVS TRUNK has "#if 0" on line 1149
ok - I see it in rev 2.1.7 HEAD
using something other than the dapper realtime kernel package?
figure out what header your kernel defines NULL in, and include it explicitly in any file that gives that error
it sounds like some header file isn't being included - NULL should be defined
err - right :)
here's the real link for the hexapod pic: http://www.imac.unavarra.es/parallel/images/argazkiak/img_4336.jpg
alex_joni: how is that coming?
I have nfc
they posted on the devel list with problems running the kins
you know - I just responded to Patrick Ferrick, and it made me think of something
there's no reason why we can't have very long plaintext descriptions of ini variables, in the ini files
oh-oh, that can't be good :)
you don't need comment characters, the ini file format is meant to have lots of irrelevant stuff in it, since it's meant for multiple programs
(and everything for all the other programs is irrelevant to a specific program)
well, it would allow for local documentation in the ini file ...
I know a lot of that stuff was removed at some point, because it makes the file big and hard to navigate through
anyway, just something to keep in the back of your mind
weeelll.. we can always make it big again
but it's hard for users to look at it initially
the thing is, you load it up in an editor, and there can be these nice paragraphs telling you what the vars mean
let me guess... coloured paragraphs?
err - probably not
well, they'll have the same color as everything else :)
unless you use syntax hilighting and set it for "settings" files (or desktop or whatever your favorite editor calls them)
mine is mcedit
does it do syntax hilighting?
hmmm. that made me hingry
looks like near-vanilla linux kernels are delivering some pretty good latency results ..http://www.osadl.org/Realtime-Preempt-Kernel.kernel-rt.0.html
getting into the realm where emc/hal systems without BASE_THREADs could run
#Minimum latency: 0 microseconds.
#Average latency: 7 microseconds.
#Maximum latency: 39 microseconds.
#Total samples: 7069336
cradek: huh, another user reports "I do something, and then touch-off never works again through the rest of the session"
hmm - what do they mean "never works again"?
the touch off doesn't pop up - or what they type in doesn't do anything?
I have not seen that yet
Click Touch Off for say X axis and attempt to enter 0.125" when I type
the first 0 (zero) an error appears in the popup window saying "bad
number format" once this has occurred I can't add offsets to any axis.
From: "Andy Ibbotson"\
he is not entering a o?
I dunno -- that doesn't give the error message he reproduces there
good guess though
that's thinking like a user!
I have experience with that :)
Here's an interpreter error I haven't seen before: "Null missing after newline"
needless to say this first realtime kernel I built doesn't work
lot of these messages:
[ 2554.132000] Bad page state in process 'kswapd0'
[ 2554.132000] page:c11e8f20 flags:0x40000000 mapping:00000000 mapcount:-1 count:0
[ 2554.132000] Trying to fix it up, but a reboot is needed
hmmm. that seems like a bad message to get
RTAI Testsuite - KERNEL latency (all data in nanoseconds)
RTH| lat min| ovl min| lat avg| lat max| ovl max| overruns
RTD| 0| 0| 131866| 186847751| 186847751| 9921
RTD| 0| 0| 126707| 172760187| 186847751| 19899
talk about unpromising results .. and this is the 50% of the time that it doesn't hard-lock the machine
what kernel/RTAI versions?
18.104.22.168 and RTAI magma from the last 24 hours
22.214.171.124 is the basis for ubuntu 7.10
or at least the current betas
I had excellent results with cradek's SMP kernel, but I had to rebuild it to add SATA support, and the result is pretty crappy
it does boot and run though
hm the CPU speed shown in dmesg is wrong
grep MHz /proc/cpuinfo
cpu MHz : 2000.000
[ 725.704000] RTAI[sched]: Linux timer freq = 250 (Hz), CPU freq = 1193180 hz.
that looks familiar
is that just misprinted from RTAI?
I did have CPU scaling enabled, but disabling it (setting to "performance") didn't make any difference
ie, clock freq vs. CPU freq?
maybe I should remove it and rebuild
but the CPU can't go that slowly anyway - I think the bottom clock speed is ~1 GHz (at least for the A64 / Opteron)
oh it doesn't say kHz does it
[194617.013612] RTAI[sched_lxrt]: Linux timer freq = 1000 (Hz), CPU freq = 2134658000 hz.
timer tick frequency
^^ working machine (dapper w/linuxcnc official kernel)
lots of extra zeroes there
cpu MHz: 2134.658
the in-kernel variable is in kHz
at least it was in 2.6.16
or whatever this is
I wonder why it's sched vs. sched_lxrt
mine says sched also
the configure --enable-sched-<mumble> options seem to be different in current rtai magma
I think my configure-rtai script doesn't work right
or maybe .. something else entirely
ok. that could be. I think I used it here as well :)
man CRTs stink
clearly I need to blow a few hundred dollars and get rid of this CRT in favor of a nice LCD
wtf? ccache is getting nothing but cache misses on this third kernel compile
hit: 38 miss: 10174 and climbing