got an error that says file ended with program end...
but at the end I hae a m30
did you use a % at the beginning?
begining of the file?
no... should I have?
if you did, then you need to end the file with % also, I think
make sure there's a CR on the line with M30
hmmm... seems its not a problem with the end of the file... but the if statement I just added
o302 if [#19 lt #4-#5]
#20 = [#5*#9]*360
#16 = [#1+COS[#20]*#15]
#17 = [#2+SIN[#20]*#15]
#19 = [#4-#5]
oh. then I'd fix the if statement ;)
I see it
o, not 0 ;)
03jepler 07TRUNK * 10emc2/docs/src/lyxtree.py: make include and input inset types work
Are variables pass values outside of if statments?
Ziegler: yes, parameters above #30 are "global" to all parts of a .ngc file, while parameters from #1 to #30 are "local" to a sub
im in the same sub...
but specifically from an if statment.. one sec paste bin
I don't believe the concept of "passing variables" has anything to do with if statements. there should be no locals to an if
[02:22:45] <Ziegler> http://pastebin.ca/732539
alright thats what I figgured...
for whatever reason I am getting arc errors after 303 sets the variables
gcode has a local scope?!
yes, subroutines do
I'd add brackets here: o303 if [#19 lt #4-#5]
[#19 lt [#4-#5]]
not sure if this is a cause of any problem
SWPadnos: well subroutines are like functions I'd hope ;)
yes, but they have both locally- and globally-scoped variables
no change jepler
and you don't get to choose which are which ;)
I manually calculated the number
Radius to end of arc differs from radius to start: start=(X0.2722,Y0.2722) center=(X0.3085,Y0.6555) end=(X-0.0362,Y-0.3833) r1=0.3850 r2=1.0945
aren't I and J always relative?
ahh... I J is center from start of curve isnt it
yes in emc
meaning that you don't want to add #1 ad #2 to the calculated endpoint
SWPadnos: only if you live in the ozarks
relative, not relatives! ;)
ok.. I was doing distance from end point... which did not matter with the rest of the routine because it began end ended at the same point!
SWPadnos: your uncle is your father, mkay!
the sign may be wrong actually
you do need to add in #1 and $2, because they're subtracted out in the G2/G3 lines
but you may want to change the sign of either the entire equation (multiply by -1) or just subtract the COS/SIN terms instead of adding them
on lines 30 and 31
and 8 and 9 (did you mean I "dont need"
hmmm. no, I'm just looking in that if statement
but now that you mention it, it would make sense for the equations to be the same ;)
if statement is fine
ok, let me rephrase. I was only looking at the final endpoint calculation
yeah I think I am going to do it different
go from "start degrees" to "end degrees first" and then finish the rest of the arcs to the bottom
that should be an infinite loop
as long as #19 is >= [#4-#5] ... if #19 < [#4-#5] #19 = [#4-#5] ... endwhile
well... maybe the way I explained it... but I let me re-do the code to see if that makes sense
(degrees != depth)
what's this program do?
or it will... at some point
I don't see how that loop can exit, unless there's some infinite loop detection code in there
I wish we had a way to do more than one turn of a helix at a time
SWPadnos: let me fix the code
canon has a number of turns parameter in the helix call
but I don't think the interp uses it (or can with G-code as input)
yeah I think there's support in the low levels (like a lot of things)
it shouldn't be to hard to add that though - throw in a K parameter like threading has, and - err - step 2 ... and it's done :)
I'm thinking of this: http://www.sciencecartoonsplus.com/gallery.htm
[02:40:11] <cradek> http://en.wikipedia.org/wiki/Underpants_Gnomes
under 'The Underpants Gnomes have a three-phase business plan, consisting of:'
yes. ever popular on Slashdot
oh ok, I didn't know whether or not you use the internets
i has herd tell of tha intarwebb
ouch, $50-60 for a shell mill
yeah probably so
a few "cheap!" things add up to actual money
why do you want to mill shells?
I sure know that
because they're so hard
hi jmk :)
I have several decent shell mills, but the one that fits the arbor I have doesn't end-cut (and that's what I want to use it for)
Ziegler, the end angle calculation is probably wrong, but there should be no number that can be given to the endpoint calculation to cause a radius error
the sin/cos functions will always give you a sine and cosine of a point on the circle - they do modulo on the angle
cradek: I'm sure you can find one in here: http://www.hgrindustrialsurplus.com/search-products/product-detail.aspx?id=10-287-004&searchtable=2&sortExpression=&SortASC=&pageSize=50¤tPageIndex=0
counting gas I'd probably be over $60.
my I and J were the distance from end point to center... instead of start point to center
better box: http://www.hgrindustrialsurplus.com/search-products/product-detail.aspx?id=12-247-052&searchtable=2&sortExpression=&SortASC=&pageSize=50¤tPageIndex=1
how soon do you need it
I'll get back over that way sometime :-)
oct 20 is their saturday this month, and I was planning a visit
I can poke around
that would be cool, but it's a pain to ship stuff like that
(just to save me a few bucks)
one cutter isn't a pain
(I wasn't suggesting buying the whole tub)
sure I understand
my arbor requires a 1" hole in the shell, and I just want something good to square up blocks with
(so it has to cut on the end, but of course not plunge)
like a smaller version of my big one
[02:51:07] <cradek> http://www.use-enco.com/CGI/INSRIT?PMAKA=320-4370&PMPXNO=950861&PARTPG=INLMK32
hard to tell if those teeth have bottom relief
[02:52:18] <cradek> http://www.use-enco.com/CGI/INSRIT?PMAKA=320-4845&PMPXNO=951479&PARTPG=INLMK32
this one sure does but $108 and discontinued
you really want a face mill, not a shell mill
[02:53:10] <jmkasunich_> http://www.use-enco.com/CGI/INSRIT?PMAKA=365-4020&PMPXNO=951538&PARTPG=INLMK3
seems like those are all indexable
You can often find resharps like that on ebay reasonably priced
(shell mills are for horizontal milling machines)
face mill + 1" hole = all discontinued
SWPadnos: this works now: http://pastebin.ca/732579
does it exit from the loop?
duh - nevermind
need to check what happens if #19 happens to be equal to #4-#5
[02:55:37] <jmkasunich_> http://cgi.ebay.com/2-1-2-INDEXABLE-SHELL-FACE-MILL-4-FLUTE-SD-322P-753-4_W0QQitemZ130161391481QQihZ003QQcategoryZ25260QQssPageNameZWDVWQQrdZ1QQcmdZViewItem
the loop should never execute in that case
but roundoff error might cause an extra loop on occasion - probably with no Z motion, but definitely not helical
right... but trying to decided if it needed to run one more time after line 26
(I think it does)
msc has a bunch of them
also a question for you, how big is your mill
2hp and "rigid ram"
SWPadnos: can I exit the while statment early??
you do mostly alum right?
yes right now
Ziegler, I don't know. you can certainly stick an if statement in to conditionally do the "rest of the loop", and insure that the next check will exit
put another o301 endwhile inside of o303
no, I don't think so
cradek: what are you looking for in your face mill
a specific cutting dia, etc?
just wishing there's a faster way to square up work than a fly cutter
anything a flycutter won't do?
I want a flycutter with 10 teeth or so :-)
sounds like a 6" face mill
that might be a bit heavy
"Inside a while loop, O- break immediately exits the loop, and O- continue immediately skips to the next evaluation of the while condition."
2.5 or 3" would be very nice
3" is a good size for bport
4-5 teeth in alum
Ziegler, you could just not set 19 to 4-5
remove line 32, change lines 35 and 37 to use Z[#4-#5]
and the loop will exit
also, here's a sneaky thing. set parameter 21 to 2 or 3, and use G#21 for the circles - gets rid of the ifs :)
cradek: i actually put a negative radial, positive axial rake on a flycutter
and that kind of grind will whomp material off, especially aluminum
err - 22, 21 is in use
you know ;)
I think so
without giving the machine the brigeport shakes
so I can do G#11
so a single if at the beginning to choose between left/right, and then all G#23
no, you want to be sure it's going to be an arc move ;)
so check for it at the beginning
Cradek, Check cutting tools on http://www.cdcotools.com/
and ctl+f 46917 3inch cutter
more face mills
you could probably use #11 though, but it seems icky to do if [#11 EQ 1] #11=2 else #11=3
[03:03:32] <toast> http://www1.mscdirect.com/CGI/N2DRVSH?SISECT=0002006527&SIS0NO=00000674838
looks like cdco is all R8 stuff
those kinds of large cutters are where the skinflint in me starts screaming "surplus or ebay, surplus or ebay!"
yeah no kidding
Didn't know your shank.
(I paid under $20 for my 5-1/2" one)
believe it or not, those kinds of cutters are the kind of thing I haven't bought yet, because they're too expensive
but I have one arbor, and it's for a 1" hole, so I'm pretty picky
SWPadnos: you aren't missing much
1" hole leaves you with lots of choices
most people don't push them hard enough and constantly wear the teeth out
flycutter works fine with patience
especially without a CNC to do the pushing
or power feed, for that matter
most of the bridgeports at school get the shakes real bad
I had that, until I put in the ballscrew
We had a few at school, but anything harder than al you couldn't really use it to potential
when you get those things moving
the ones at work do, as well
nice thing about the insertable ones is that you can get the RPM up
ok. time to get back to those DACs (or modbus or whatever I'm supposed to be working on at the moment)
(4" dia hits a couple hundred SFPM pretty quick)
Ziegler, let me know if that G#23 thing doesn't work
yeah even 500sfm at 2.5" isn't much
but you have to feed it at .005-.010" per tooth
I run my big one at 70 RPM in steel, 140 in aluminum
but with 10 teeth * .010 that's a decent feed
not sure I can even do 70rpm...
that's pretty darn slow
that is slow
with coated carbide
toast: mine has HSS teeth, not carbide
and I have no source for replacement teeth
good reasons then!
WHat do the teeth look like? HSS m2 ?
i don't know though, dude
[03:08:58] <jmkasunich_> http://www.metalworking.com/dropbox/_2001_retired_files/Facemil3.jpg
you can run hss pretty high
but the no replacement teeth thing
[03:09:27] <Ziegler> http://pastebin.ca/732594
SWPadnos see line 7
toast: I run 140 RPM (and sometimes a bit more) in aluminum
that works out to 201 SFPM
they make carbide teeth
I have no idea how to even get the teeth out
i've never looked to see where to get them though, but if you are interested you might want to call up kennametal or valenite
no screws or anything
[03:10:32] <jmkasunich_> http://www.metalworking.com/dropbox/_2001_retired_files/Facemil2.jpg
I run HSS at 350sfm successfully so far
Ziegler, what if I make #11 -1?
is pretty safe
hehe... wont work
toast: my handbook's in the garage or I'd argue :-)
might do some funny stuff, no? :)
toast: I thought that was pretty high for hss (but I bet you know better than I do)
toast: I'd be afraid to even ask what carbide teeth for my mill would cost
my lathe's at work or i'd argue
the teeth are big
by actually showing you the cut and that the cutter takes it just fine =)
ok time to id 10 t proof it
they're pricey, but regrindable and they're really tough grades of carbide
I'm perfectly happy with the HSS teeth actually
as it is I spend almost as much time setting up the cut as I do actually cutting
a good balance
a cut like cleaning up those plates in my pic takes a couple minutes
Speaking of tough carbides, I got a replacement blade for my cable trencher recently and the new design uses three 1/2 inch carbide brazed square tip lathe style cutters brazed onto the bar. This thing will cut through rocks. What carbide is this?
10x10x3/4 plates, stack of 6
03jepler 07TRUNK * 10emc2/docs/src/ (l2h.xsl lyxtree.py):
* Support book layout: new tags include Author, Title, Part
* Support tableofcontents, printindex to place the toc and index at the desired location
* Remove debugging messages
skinnypuppy34: titanium carbide
standard carbides, probably like an ISO K30 or K40 grade
Cool, it it tough as hell i give it that much
carbide is amazing stuff
i am still trying to get my head around hard cutting
hard feeds/depth of cut
all the sounds and vibrations are new
like starting all over again
Yeah I didn't do too much hard cutting and wasted a few cutters getting dialed in on realistic feeds.
"this is so cool! *thudthudthud-bang* Damnit."
YEP, Now milling UHMW at 80 ipm was fun
Waiting on my motors/&P.S.... sigh
I've bought my geckos, limit switches, wire, cabinet, oh crap button to wall mount, and have ordered my motors. Need to make some pulleys and brackets, got some al plate recently for the job
Sooo just about finished scrounging parts
make sure your oh crap button stops the spindle!
Yep, I was going to put it on a motor starter to interupt and shunt
How close to the travel ends should the e-stop switches be on a bp clone retro?
dunno, never set one up
far enough to stop in time
did you know a BP clone doing a 100 ipm rapid
can totally knock the head out of tram
if you hit what you are working on
it mills for a little bit though!
I'm anxious to get emc working on this so I can tighten the following error limits a lot
I don't trust these amps to fault
fortunately I haven't made any big mistake yet, but it's just a matter of time
DId jlmjvm get that pico card working with steppers + encoders?
On the Hurco KM3 ( Kondia made BP clone ) the limit switch allows for about 5.5mm travel before hitting HARD
another one i pulled was doing a toolchange on a Haas SL-20
without moving the turret away from the part
another time i forgot to do a program reset and attempted to use a roughing boring bar as a drill
in a big cnc lathe
that was frightening
but the limit switch also tripped the amp to short the motors to breaking resistors
I've heard of people milling with dial indicators
they don't cut very well I suppose
oh man that would suck so bad
i'd go home, if i did that
Ha seen that done on an sl-20, seen another guy break off a live tooling head crashing it into the chuck on the sl-20. Last time he ran that machine
stays in the machine and nobody gets hurt = good crash
to be fair the sl-20 crash wasn't my fault
Never start the spindle @ 12000 rpm with an edge finder in it.
it was my program and i was on the machine, but the change that caused the crash was someone else
did the tip come out like a projectile
of course it did
don't think it was there after it passed 4K
the worst crash i have been involved in, or standing near, was a 20-hp cnc in low gear
I accidentally ran my wiggler (the pointy one) at 2000 and it stayed in - probably wouldn't have much faster though
was running copper, when the crap rewelded
and managed to halt 180 lbs of chuck
it made a ridiculously loud "boom"
then a jet turbine sound
until we hit e-stop
people came from all over the shop to see what had happened
had one like that a few months ago
but i didn't do it, i was just standing there.
Wow. now the crash that broke the live tool head off sounded like a helicopter crossed with a jackhammer.
I was running a huge Mazak and the spindle tach feedback card fried and I had a runaway
nothing quite sounds like a spindle that is running away
It was heading for 7000 rpm with a 4000 rpm rated 12" chuck
you lucked out
before it blew the mains breaker
that was lucky
i don't think the enclosure can stop the jaws off a 12" chuck
much less the chuck itself
had half the building come to see what all the noise was
what did it sound like
like a spooling turbine?
well a few years back a guy crashed the same machine - Z workshift was to be 115.0mm and he put in -115.0 - rammed the turret into the face of the chuck and took almost 1/2" off the chuck face after tearing the jaws off
they had to remove all the sheetmetal starting from the back because the door was so exploded - and take it all apart to get the door off/open
repairs cost over $40K
toast is now known as toastydeath
alex__ is now known as TheBall
* skunkworks wonders if fenn has seen jros's videos
* fenn looks
looks like a ufo hovering
they sure dont move it very far
i'm working on a hexapod visualization with vismach.py
right now i've got a cylinder tracking the mouse through hal_input
did you look at pumagui.py ?
there's a lot of fiddly business with the order you build stuff and where to call Capture()
and it took a while to figure out i forgot to put tool_coords into the main loop call
so now i need to implement a PointAt transform
to connect the struts to the platform
then i can just grab the current tool tip position and move the platform to there
halui should export cartesian commanded and actual positions
why should halui do that?
I wanted initially to do that.. but it should be available from motion
well, what i really would want is cartesian commanded and actual from each kinematics module
and motion just exports cartesian commanded instead of joint.*
then you can feed other stuff into kinematics from hal
beyond just teleop velocity commands
the reason i said halui is because i'm doing a gui-ish thing and i figured that whatever hal module does it would have to use NML to get those values
[13:58:23] <cradek> http://translate.google.com/translate?u=http%3A%2F%2Fwww.imac.unavarra.es%2Fparallel%2Femc2.php&langpair=es%7Cen&hl=en&ie=UTF-8&oe=UTF-8&prev=%2Flanguage_tools
I think they like EMC
:) why wouldn't they? I think the only people that don't like emc - have not tried it ;)
i just noticed that teleop acts rather strange around the edges of the work envelope
for example in hexapod-sim, when z is near zero, both page up and page down cause z to go negative
instead it should say 'reached soft limit' i think
maybe this is just how hexapod kins behaves
for some reason when i command g0 z0 it thinks its at z=20
is that the issue where there is more than one solution per position?
from the rev/forward kins
wonder if the guys at unavarra made their updated kins available
* skunkworks really doesn't understand it very well.
skunkworks: something like that. i think its in a co-planar configuration at z=0
and the kins code doesnt know how to 'flop'
i dont get where the 20 comes from though
something to do with home position
there was an list email from ray talking about kins on a big machine. sounded like the solved the problem by picking the kins solution that was closest to the last position.
yeah but its nowhere near home position
there are some other problems with hexapod kins that i'll have to fix somehow - offsets for cardan joints and account for linear motion when you rotate a nut around a screw
since a cardan joint doesnt rotate around a point like a spherical joint does
hm. actually maybe it does sometimes
cradek: not that I know..
fenn: I think the kins usually have a flag to remember which of the multiple solution is appropriate
at least it does that for puma
the code is pretty crusty and definitely not newbie-friendly :)
fenn: it's not necessarely a place to beautify code at the expense of performance
part of it is i'm hacking on python right now and its just toooo easy
03fenn 07TRUNK * 10emc2/lib/python/vismach.py: added Scale/HalScale for a visualization i was doing
i think kins returns the starting position if it cant find a solution
what sick EE designed that laser with 24V 16 bit BCD..
probably the same sick EE who originally used thimbwheel-type switches to set the power digitally
wait - so I can setup emc so that I can used thumbwheels to do mdi input? http://www.electronicsam.com/images/KandT/conversion/frntcontroller.JPG
sure, if you have enough I/Os ;)
just use weighted_sums on each wheel
(you may need to increase the max number allowed)
Thats what I need is tons of contacts that can get dirty - No, I ment Z10.3 not Z0.3
and I think you'd need two Mesa cards (the 5i22, with 4 connectors) for all those
or just one bad bit - in the top digit: Z80.0000
good mornign. which kins is appropriate for a 5 axis machine? 3 regular translation + 2 rotational on the z for a spindle?
but you need to use a CAM post for the geometry of the machine
err - that knows the geometry of the machine
oh yeah. mastercam.
03alex_joni 07TRUNK * 10emc2/tcl/ (TkEmc tkemc.tcl): fix handling of 9-axes
what is the 5i22?
a newer version of the 5i20
it has 4 connectors, and a 1M or 1.5M gate Spartan 3 chip
instead of 200k Spartan 2
same driver? or new driver?
no driver at the moment
in the works?
shouldn't the old one run?
03alex_joni 07TRUNK * 10emc2/tcl/TkEmc: revert default fontsize, didn't mean to change it
I have several of the cards for a different project, but once that's done, I'll be working on the new driver
alex_joni, I don't know. haven't tried. I think not, because the driver looks for the PCI IDs of the 5i20, which are different from the 5i22
also, it has the bitfile built in, which is wrong for the other chip
"I knew that"
jmk and I have gotten the new driver utilities to work with the 5i22 - I use bfload daily :)
I have a hostmot config that loads on the 5i22 as well (from PeteW), and it has tons of stuff in it
it's just impossible to map the functionality into HAL very well. The config is very configurable, but HAL drivers are not - as far as exported pins go
oh, how about 12 stepgens, 12 PWMgens, 6 quad counters, 6 SPI ports, and 6 serial ports?
also, all bits can be used as I/O or as the special functions (so you can have just 96 I/Os if you want)
so like a 12 axis setup?
pretty much, except you'd need more quad counters :)
oops - only 4 SPI and 4 serial ports
but that's pretty easy to reconfigure
oh forget it then, MUST have 256 serial ports!
to run a 5 axis (really 6 'cause it's a gantry) go with a 5i22 or dual 5i20's?
the two 5i20's will cost about the same as a 5i22-1M
but you'll get 6 I/O connectors
then again, a single 5i22 has 4 connectors, and 2.5x as many FPGA gates as the two 5I20s combined
3 each on two boards = 6 total
and emc2's configs?
assuming that there are drivers for both boards, that doesn't matter much
alex_joni: Yes you just got hitched recently, but lesson 1 of the parenting handbook... http://news.yahoo.com/s/ap/odd_joy_ride;_ylt=AjcHP9t2LevGwYurz5FdZ5XtiBIF
the only thing that would matter is that you would have two update functions running in the dual 5i20 case, which takes a little more time
no, this will not be in 2.2.0, as far as I know
it may come later, because we're allowing new drivers and modules to be added to the release branch in the 2.2 series (I think)
refering to the 5i22?
you can use multiple 5i20 now, I think
there's been discusison on that in some of the mailing.
I think you may have to pre-load the FPGA configs using a utility, and tell the driver not to re-load it
thanks for the info.
sure. (I hope it's correct ;) )
gotta cobble together a simple gantry system and work out gantrykins.
and simple homing.
I think that will be easier in 2.2.0
how is it that the 5i20 doesnt have enough i/o for a 5 axis system? am i confused?
it just doesn't have enough hardware PWMs
what about a 'firm' pwm?
the 7i33 (?) uses one full connector for 4 axes, so you get 4 or 8 (or 12), but if you need 6, you lose the 24 I/Os on the second connector
basically a counter and a comparator
if you want to do custom FPGA configs, then of course you can do it
well, slow i/o is much cheaper than buying extra 5i20's
I thought the plan with what jmk was working on was a swiss army knife controller. able to setup whatever you wanted for each pin. Or not?
you still need the fast quadrature inputs
skunkworks: that's what it looked like to me
yes, that's in the works, and will support all 3 flavors
it's not quite "any function on any pin", but it is more flexible than "I/O or an alternate function" like the 5i20 has now
it's more like "pick one of these four or eight functions for each pin"
SWPadnos: HW PWM's? is different from programming some in the FPGA?
well, sort of. once you stick something in the FPGA, it's hardware. whether you call it hardware or firmware is up to you
So why not more PWM's on it? limitation of ________________
limitation of "it's not in the FPGA code that's used in the current driver"
the card has 96 I/Os, and I think there may be enough gates to do 48 PWMs, using all the pins (PWM+dir or up/down)
Ah, so limitation of current design, not a REAL limit has been reached
err - the 5i20 is only 72 IOs, so that would be a limitation of 36 PWMs
03alex_joni 07TRUNK * 10emc2/src/emc/usr_intf/emcsh.cc: fix for UVW axes
right - the question wasn't "what can the card do", it was "what can I do with the current driver"
(and the current driver assumes, and loads, a particular FPGA config)
SWPadnos: Can't one lego block an FPGA?
"Drag and Drop FPGA!!!"
at least, not in any general sense. you can't do the equivalent of dynamic linking on bitfiles
Jymmmm, let me introduce you to LabView FPGA
SWPadnos: your favorite piece of software in the whole world, right?
SWPadnos: you forgot to mention something
SWPadnos: No thanks, I only want their robot arm (to play with)
yeah yeah, which I hate
you said that surprisingly seldom in the last few months
which manages to use 1.5 million of the 3 million available gates (in a virtex chip no less) to interface to fewer I/Os than a Mesa has, and still can't get loop rates above a few hundres Hz
* Jymmmm lol @ alex_joni
can anyone else get here: http://www.silicondust.com/zxc/content/view/5/26/
whereas I have a 5i22-1.5M, interfacing to two 6-channel ADCs (plus I/O on any pin), scanning at 200KHz+, and using less than 10% of the gates in the chip
Anyone seen anything else like this? http://www.dvrplayground.com/article/13944/
I'm looking for HW based IPTV boxes
Jymmmm: hey, did replacing that bad part in your setup fix your stepper stalls?
ds2: what bad part?
Jymmm: that mount with the croaked bore I think
SWPadnos: (I wish)
uhh - I meant hte X axis ball nut mount
ds2: Can't replace it, needs to be machined
Jymmm: have you done the remachining yet?
ds2: Only 100 more hours of sanding to go
was hoping to get an indication if system tuning on a stepper is worthwhile
and I figured you were further down that road; prehaps enough to get a good indication
ds2: you got some parker drives?
ds2: what are you tuning, geckos?
I'm adding a 4th axis and it seems to stall with any motion
Jymmm: The stock taig drive
ds2: did you swap out motors or drives by chance?
A to x as example
Jymmmm: no, the 4th axis is all new... X/Y/Z are tolerable
Jymmm: I can't swap between A and X
at least not w/o a lot of work; seems taig loctited a lot of their motor coupler screws together
I suppose it may just be a preception thing, the 4th axis does have a worm reduction
ds2: how fast are you thing to run it?
ds2: how fast are you trying to run it?
fenn: how is pete v.'s trick any different from 'placement new' which is built into C++? http://www.parashift.com/c++-faq-lite/dtors.html#faq-11.10
jepler: you're asking the wrong guy :)
Jymmm: short answer is "I don't know" - I just figure it should have a edge (a 4inch RT) speed comparable to the speed on any of the linear axis
the long answer is I am still using the taig supplied dos software which I haven't figured out how to change the steps/degree thing
ds2: You mean TurboCNC?
Jymmm: no. MSP2007
I'd switch over to EMC but I am waiting to finish my new stand for it so I can have a monitor that next to it that can last more then 1 milling session ;)
right now it is on a crummy laptop that seems to be taking the chips in stride
ds2: I can't find software on their website
Jymmm: you can't download it... it comes on a floppy
ds2: Then how do you update it?
Jymmmm: you don't!
hence the strong need to move to a better controller; but the desire to be able to make parts in the mean time has kept me on that thing
jepler: i think the difference is that pete's code allocates new memory as normal, but then it doesn't get de-allocated when the class destructor is called, so you end up using the same piece of memory over again (and you dont have to specify where to put it the first time around)
ds2: have you tried Turbocnc?
jepler: but i'm really foggy on most of this stuff
Jymmmm: no... should I?
I thought of TurboCNC but then figured, time is better invested in EMC2
ds2: It's DOS too, as long as you have hdd space, can't hurt anything
Jymmmm: that crummy laptop is there cuz the IDE controller is blown so it'll be a floppy job if I try it
i used it and was appalled
ds2: it's 342KB
fenn: in what way? does G01/G00/G03/G02 work?
it doesn't blend arcs, so it stops at every arc
this is a problem when you're doing a contoured surface
Jymmmm: I could try it but from your experiences, isn't stepper stalling more of a physical problem then software?
ds2: Yes, but at least I can tell you where to change settings in it =)
oh that's fine... TurboCNC would just be a intermediate til I can get EMC in there with the proper chip shielding
Jymmmm: oh hahah
what do most people use a chipshields around a mill? plexi/acrylic or polycarbonate?
use polycarb, not acrylic
cracks as easily as glass
polycarbonate is much more abrasion-resistant
as a matter of fact, you cut acrylic just like glass..... score and snap
polycarb is more impact resistant too
will the polycarb be frosted with scratches after a few runs on the mill spraying chips at it?
till you wipe off the chips =)
ds2: I say polycarb from a safety standpoint compared to acrylic (not to mention a bit cheaper)
Polycarb great, you can wipe mud off your cycle visor and not be very scratched at all
ds2: not flying chips, but flying broken tooling
there's a reason they make classes from polycarbonate (and it sure isn't the optical quality)
where? polycarb is about 1.5-2.0x more expensive when I got it last time at OSH
right, even glasses
ds2: you're buying from OSH, of course it's more expensive =)
I've had some badly scratched safety glasses so...
ds2: but you can still see, can't ya =)
they make glasses out of it because it's lighter
Jymmm: that's relative to acrylic (plexi)
you would have had badly scratched eyes if the glasses had been acrylic
I should get some safety glasses
* cradek heads for a tangent that nobody else cares about
polycarbonate needs scratch-resistant coating
fenn: I think that's not true
if you're making glasses out of it
it's practically bullet resistant in all forms though
Jymmm: got a better source? tap doesn't have polycarb often in their cut offs
polycarb is very puncture-resistant (it's lexan, which is touted as unbreakable), but it does need coatings for scratch resistance
we buy ge lexan with a scratch resistant coating only if the customer wants it.
ds2: wuit being a cheap bastard!
ds2: how much you need?
bare polycarb is more scratch resistant than CR-39 with a scratch-resistant coating (eyeglasses)
I have some margard
probally a total of 2'x4' but it can be in small sheets
for my shields at work
ds2: you want polycabr or lexan?
Jymmm: I need/want a enclosure for the mill as I get chips 4-5 feets away
Jymmm: polycarb should be fine
go dumpstering at a plastics distributor
Ha, I've got a nutty friend that calls it the "green store"
even if you dont find polycarbonate there's bound to be something good
4 years ago, they were fine with that... as of 1 year ago, they were unhappy about it
Jymmm: I think either 1/8 or 3/32 stuff should be fine... normally tooling is 1/4" dia so the chips aren't too bad and that should tolerate a one time blast from a broken tool
Jymmmm: worse come to worse, I'll get more 8x11" panes from osh
ds2: Lexan .125" 24x48" $24.45
Jymmmm: tap or ?
Polycarb, glazing grade .125" x 24" x 48" $32
pier's dxf2g new code announced today on freshmeat... tarball at http://pier.unirc.eu/software/dxf2G.tgz
thanks pier :)
tomp, got a link for dxf2g?
[21:09:44] <tomp> http://pier.unirc.eu/software/dxf2G.tgz
tu psrle Italiano? ;)
tomp, Yes I can say "Italiano" other than that, I'm illiterate. :)
* skunkworks can barely handle one language
I can say "Cappuccino" and "espresso" also
actually, if you only count ordering coffee, I think I know about 6 languages ;)
SWPadnos: it's food, of course you can.
hmmm. that makes me hungry
SWPadnos: and I wouldn't expect anythign less from you either =)
of course not!
how long would a 7meg xfer take at 115200? ( i got a synce-serial xfer using ttyS0, and i cant get any idea of how long it will be before i can go to the bathroom :(
115200 / 10 = 11520 bytes/second
thx, saved me from getting out the calql8r ... bye >now<
SWPadnos: /10 ?
oh, when all done it gets chatty "File copy of 6819656 bytes took 11 minutes and 7 seconds, that's 10224 bytes/s."
((filesize in bytes/8)/bitrate in seconds)/(60*60) = minutes
filesize/10, this is serial with start and stop bits
err - *10
ah, forgot about them.
RAW baby, RAW!
ethernet also has overhead ;)
only if you use very small packets
ethernet frames themselves are required to be >= 64 bytes in length, and 19 of those are framing information
raw transfer for sheer speed test
but it's the same 19 bytes for a 64k packet as well
(if your MTU can go that high)
Internally too lazy to check, externally 1470
fscking dsl and pppoe
oh, i got some communications with the labjack and hal vis python
.. very messy... setup a stackframe, call a few checksums for parts of the frame, call a write, call a read, examine a returned array.... different data for each type of query or config
good night all
see you Alex
Anyone using Cad Expert? Is it worth the dough?
Cam Expert that is.
Isn't there something under nix that has wordlists? Like words that beign/contain the leeter Z (as example) that I can view (not just search for)?
ok, how about something that has verbs, nouns, etc?
sorted by v/n/etc?
don't think there is anything like that
nice twist on rack&pinion ... http://www.nexengroup.com/rps/
a roller rack
like a gear train
[22:54:25] <Ziegler> http://xkcd.com/327/
[23:01:39] <SWPadnos> http://xkcd.com/325/
I have an infrared camera at work, if you lie on the floor for a few minutes, there is a heat image there for about 1/2 hour
good trick for lab tours
hmmm. too bad it leaves evidence when you nap on the job
[23:10:28] <SWPadnos> http://xkcd.com/310/
about 6 months ago... I managed to actually go through them all
I came out acting quite nerdy the rest of theday
well more so than usual
hmm - an extra 512mb of ram makes the wifes laptop usable again.
nice isnt it
no more complaining I hope. :)
jmkasunich_ is now known as jmkasunich