#emc | Logs for 2007-01-18

Back
[00:09:11] <lerneaen_hydra> testing 123
[00:10:26] <CIA-8> 03cradek 07v2_1_branch * 10emc2/bin/.cvsignore: new file
[00:10:51] <lerneaen_hydra> could someone test writing something with lerneaen_hydra in the line?
[00:11:28] <cradek> lerneaen_hydra is bothering me
[00:11:48] <A-L-P-H-A> lerneaen_hydra: sucks at UT2k4.
[00:11:58] <lerneaen_hydra> hmm, ok
[00:12:00] <lerneaen_hydra> err
[00:12:02] <anonimasu> lerneaen_hydra:
[00:12:05] <lerneaen_hydra> * lerneaen_hydra sobs
[00:12:27] <cradek> did you know you can kick someone off irc if you type /quit and then their name
[00:12:37] <A-L-P-H-A> what happens when you put lerneaen_hydra the emo kid into a circlular room?
[00:12:51] <A-L-P-H-A> he dies, cause he has no corner to cry in.
[00:12:54] <lerneaen_hydra> my status is currently set to away, right?
[00:13:13] <cradek> 18:13:02 [freenode] -!- away : Away
[00:13:19] <A-L-P-H-A> yes you are away
[00:13:30] <lerneaen_hydra> and now?
[00:13:46] <lerneaen_hydra> * lerneaen_hydra is testing IRC in gaim
[00:13:54] <A-L-P-H-A> normal
[00:14:01] <ejholmgren> that's possible?
[00:14:01] <lerneaen_hydra> ah, excelent
[00:14:05] <lerneaen_hydra> yeah
[00:14:07] <A-L-P-H-A> lerneaen_hydra: test highlight... which probably doesn't work.
[00:14:09] <lerneaen_hydra> I just found it today
[00:14:14] <lerneaen_hydra> A-L-P-H-A: yep, it works
[00:14:18] <A-L-P-H-A> lerneaen_hydra: ,there's always gaim.
[00:14:17] <lerneaen_hydra> nice :D
[00:14:33] <A-L-P-H-A> oh you are using gaim.
[00:14:39] <lerneaen_hydra> now when FF dies becuase of flash IRC won't die with it
[00:14:43] <A-L-P-H-A> I was using gnome x-chat...
[00:14:44] <A-L-P-H-A> that is nice...
[00:14:52] <cradek> irssi
[00:14:57] <A-L-P-H-A> but w/gaim it is one less app to open
[00:15:04] <lerneaen_hydra> I need a shiny gui too :/
[00:15:28] <A-L-P-H-A> beryl project, google it
[00:15:38] <A-L-P-H-A> I need a new vid card to run toys like that.
[00:15:41] <lerneaen_hydra> the integratedness of irc in gaim is very nice actually
[00:15:43] <A-L-P-H-A> but I'd get bored of it quickly I think.
[00:15:46] <lerneaen_hydra> beryl is t3h shit
[00:15:54] <lerneaen_hydra> yeah, nice to play with to begin with
[00:15:57] <ejholmgren> is there a better way to find specs for electrical components other than just typing the part no into google?
[00:16:01] <A-L-P-H-A> lerneaen_hydra: not much scripting I can do, as I don't know other languages that are support with gaim.
[00:16:03] <lerneaen_hydra> it played nasty with some apps though
[00:16:18] <lerneaen_hydra> A-L-P-H-A: oh, that could be an issue
[00:19:06] <ejholmgren> I'm trying to find the constant current capacity of the bridge rects I got at the surplus store for $1
[00:20:33] <lerneaen_hydra> gah, efnet doesn't like gaim :(
[00:20:35] <lerneaen_hydra> Xchat it is
[00:20:48] <A-L-P-H-A> lerneaen_hydra: yes it does.
[00:20:53] <A-L-P-H-A> I'm in #autocad on efnet now.
[00:20:58] <lerneaen_hydra> in gaim?
[00:21:07] <A-L-P-H-A> no... your sister.
[00:21:38] <lerneaen_hydra> 1) I don't have a sister 2) I don't have a sister
[00:22:21] <ejholmgren> he's going to pout
[00:23:13] <A-L-P-H-A> if he did have a sister, I'd be scared for her...
[00:23:27] <lerneaen_hydra> A-L-P-H-A, any particular reason?
[00:23:32] <A-L-P-H-A> lerneaen_hydra: don't touch your imaginary sister like that!
[00:23:41] <lerneaen_hydra> O.O
[00:23:57] <A-L-P-H-A> lerneaen_hydra: you're not in the midwest of america anymore!
[00:24:04] <A-L-P-H-A> :)
[00:24:15] <lerneaen_hydra> where touching sisters is fine if not encouraged? O.o
[00:25:31] <A-L-P-H-A> hey, I'm not the americant... I don't know.
[00:26:43] <A-L-P-H-A> hmm... this is fun... says it's going to take 1h to transfer my files from one spot to another, via samba... this is not the best I idea I think to use samba to do file transfers.
[00:27:14] <lerneaen_hydra> * lerneaen_hydra read 1h as lh several times
[00:27:29] <lerneaen_hydra> what's the data throughput?
[00:27:34] <lerneaen_hydra> and how many files?
[00:27:42] <A-L-P-H-A> umm... how can I tell?
[00:27:48] <lerneaen_hydra> (inside a lan)?
[00:27:50] <A-L-P-H-A> files... ~300.
[00:27:55] <A-L-P-H-A> yeah, inside lan.
[00:27:58] <lerneaen_hydra> hmm, not that many then
[00:27:59] <A-L-P-H-A> eth0
[00:28:09] <lerneaen_hydra> what's the size of all of them?
[00:28:13] <A-L-P-H-A> 300 files, ~100gigs.
[00:28:16] <lerneaen_hydra> and is there extreme fragmentation?
[00:28:17] <lerneaen_hydra> oh
[00:28:18] <A-L-P-H-A> I don't know, guessing.
[00:28:20] <lerneaen_hydra> 100 gigs
[00:28:33] <A-L-P-H-A> 3 gigs = 5 miutes or something like that.
[00:28:47] <A-L-P-H-A> guessing.
[00:28:55] <A-L-P-H-A> don't know anymore.
[00:28:59] <lerneaen_hydra> assuming 10mb/s that gives you 10k seconds -> 2.7 hours?
[00:29:05] <A-L-P-H-A> 100mb/s
[00:29:12] <lerneaen_hydra> oh, gigabit lan?
[00:29:28] <A-L-P-H-A> people use 10mb/s lans?????
[00:29:43] <A-L-P-H-A> I could understand if it's 11mb/s wifi...
[00:29:47] <A-L-P-H-A> but there's G and N
[00:29:49] <lerneaen_hydra> oh, 100mbyte/s or 100mbit/s
[00:30:00] <lerneaen_hydra> I meant mbyte/s
[00:30:25] <A-L-P-H-A> 100mbit/sec that's how things are...
[00:30:35] <lerneaen_hydra> yeah, so roughly 10mbyte/s
[00:30:36] <A-L-P-H-A> normal 100 (whatever) ethernet.
[00:30:40] <A-L-P-H-A> with a switch inbetween.
[00:30:44] <lerneaen_hydra> yeah
[00:30:56] <A-L-P-H-A> I doubt I'm getting 10mB/sec
[00:31:00] <lerneaen_hydra> that should max out at around 8-11mbyte/s
[00:31:47] <lerneaen_hydra> and that's around 2.7 hours for 100gigs
[00:33:31] <lerneaen_hydra> A-L-P-H-A, you alive?
[00:33:59] <A-L-P-H-A> no. I'm accended to an energy being. So technically no... but only in essence.
[00:34:26] <lerneaen_hydra> :p
[00:38:36] <A-L-P-H-A> hmmm...
[00:38:41] <A-L-P-H-A> gaim has no /ctcp command
[00:42:31] <paragon36> Goodnight All ... Thanks for help today!
[00:42:37] <lerneaen_hydra> 'night
[00:51:31] <A-L-P-H-A> stupid transfer is still going.
[00:51:45] <A-L-P-H-A> eta is now 28.5mins
[01:49:25] <fenn> oo.. they cnc-cut carbon fiber using a blade with ultrasound
[01:49:38] <A-L-P-H-A> that's neat.
[01:49:43] <A-L-P-H-A> why not with water? no heat then.
[01:49:51] <A-L-P-H-A> sudo apt-get update
[01:49:57] <fenn> i'm watching "how its made" clips on youtube
[01:50:09] <A-L-P-H-A> how it's made is CANUCKIAN! :)
[01:50:12] <fenn> waterjet will cause the fiber to fray apart
[01:50:24] <fenn> great show..
[01:51:35] <fenn> http://www.youtube.com/watch?v=FwPi80mjlKM&mode=related&search= helicopters
[02:08:19] <A-L-P-H-A> argh... stupid gaim.
[02:08:24] <A-L-P-H-A> can't get beta 5 to install. :(
[02:09:20] <CIA-8> 03cradek 07TRUNK * 10emc2/src/Makefile: install debuglevel, although it doesn't work yet
[02:09:20] <CIA-8> 03cradek 07TRUNK * 10emc2/debian/emc2.files.in: install debuglevel, although it doesn't work yet
[02:09:29] <CIA-8> 03cradek 07v2_1_branch * 10emc2/src/Makefile: install debuglevel, although it doesn't work yet
[02:09:29] <CIA-8> 03cradek 07v2_1_branch * 10emc2/debian/emc2.files.in: install debuglevel, although it doesn't work yet
[02:26:50] <A-L-P-H-A> I think I know what to make now... a back scratcher... none of that bambo stuff... but alu arm, with maybe a brass scratcher...
[02:27:37] <skunkworks> titanium and carbon fiber
[02:29:22] <A-L-P-H-A> I don't have any carbon fiber
[02:29:31] <SWPLinux> carbon fiber could get itchy
[02:29:44] <SWPLinux> kind of the opposite of scratchy
[02:41:29] <ds3> wheeeeeeeeee my $12 200OZ NEMA23 stepper works
[02:53:11] <A-L-P-H-A> compiling your own shit sucks.
[02:53:30] <SWPLinux> I tend to flush from time to time
[02:54:05] <A-L-P-H-A> if you need courtesy flushes, while in the middle of a #2... you need to go more often.
[02:54:51] <SWPLinux> oh - *com*piling your shit ...
[02:55:15] <A-L-P-H-A> brb
[02:57:03] <A-L-P-H-A> ahhhhhhhhhh.... after an hour of compliing, and tinkering.
[02:57:07] <A-L-P-H-A> gaim 2 beta 5!
[02:59:42] <A-L-P-H-A> frak... ssl is broken.
[03:16:41] <A-L-P-H-A> http://rescuehumor.com/video/TASER_to_Balls.htm
[03:16:48] <A-L-P-H-A> it's flash
[03:29:28] <ejholmgren> is that the one where he totally misses
[03:29:47] <ejholmgren> and shoots the prongs right into the guy's nuts?
[03:34:34] <K`zan> A-L-P-H-A: Per the link you sent me yesterday, that is common and clintoon replaced them ALL, not just some of them...
[03:41:09] <ejholmgren> http://www.youtube.com/watch?v=AafSvzXD27c&mode=related&search=
[03:41:17] <ejholmgren> cnc hard liquor :)
[03:44:58] <CIA-8> 03jepler 07TRUNK * 10emc2/src/rtapi/sim_common.h: restore printing semantics from before msg_handler changes: rtapi_print goes to stdout, rtapi_print_msg goes to stderr
[03:49:46] <CIA-8> 03jepler 07v2_1_branch * 10emc2/src/rtapi/sim_common.h: merge rev 1.8: restore printing semantics from before msg_handler changes
[03:54:54] <tomp> fen pointed out the helicopter video, then i finally found how to snag youtube vids... google for this python script youtube-dl.py ( it's a commmandline tool)
[03:55:00] <tomp> fenn
[03:56:59] <CIA-8> 03jepler 07v2_1_branch * 10emc2/share/axis/tcl/axis.tcl: make 'debuglevel' work for run-installed
[03:57:25] <CIA-8> 03jepler 07TRUNK * 10emc2/share/axis/tcl/axis.tcl: merge rev 1.11.2.4: make 'debugglevel' work for run-installed
[04:01:28] <K`zan> 3 steppers plus 20' extension cables $144...
[04:02:16] <K`zan> 3 Geko 201's for about $350.
[04:02:36] <K`zan> Then we need cabinet and power supply 24v or greater.
[04:02:52] <fenn> ah thanks
[04:02:56] <K`zan> Seems a better deal than the xylotex stuff in the long run...
[04:03:17] <K`zan> Will also probably need a break out box too.
[04:05:03] <owad> Building your own power supplies?
[04:07:12] <K`zan> Could, got a 24v 10A wheelchair battery charger here I can hack, with a voltage doubler circuit it would give me about 50v..
[04:08:06] <K`zan> Thinking about ordering the steppers and extension cables one month and the a geko a month for 3 months.
[04:08:27] <K`zan> I may be able to mount them in the CNC machines case - lots of room and not much in there.
[04:09:30] <owad> I've never seen an affordable unregulated linear power supply. I don't know if I'm must looking in the wrong places, or what.
[04:10:00] <K`zan> Considering this is free, that is pretty affordable in my book :).
[04:10:37] <K`zan> Finding serious 100V or 150V caps might be expensive though.
[04:11:15] <K`zan> NOW I wish I had saved all those IBM mainframe power supplies I had :-(.
[04:11:36] <ds3> wonder if there is a source for the telco rectifiers
[04:11:48] <ds3> cheap source that is
[04:13:57] <K`zan> Dunno. I'm getting braver about a number of things these days :-).
[04:14:14] <K`zan> Building some of this stuff rather than throwing $$$ I really don't have at it.
[04:14:55] <K`zan> Gonna see if k-mart has any of the 8x16mm skate bearings tomorrow while I am out.
[04:15:30] <ds3> one time I got a pile of those at the dollar store, 4 for $1
[04:16:10] <ejholmgren> K'zan: http://stores.ebay.com/VXB-Bearings-Skateboard-and-Slotcar_W0QQssPageNameZstrkQ3amefsQ3amesstQQtZkm
[04:17:57] <SWPLinux> K`zan: are those the Automation Direct steppers?
[04:18:55] <ejholmgren> how much do cheapo skate bearings go for at kmart/wallmart/etc?
[04:19:04] <SWPLinux> if so (or the Xylotex ones), you probably want a 60V or supply
[04:19:11] <SWPLinux> or so, that is
[04:19:24] <K`zan> SWPadnos: Yes, the ones you pointed me at.
[04:19:46] <SWPLinux> ok. for geckos, the recommended voltage is 20-25x the nameplate rating
[04:19:50] <K`zan> ejholmgren: Dunno for sure if they even carry them...
[04:20:09] <SWPLinux> those are both around 3V steppers (which is a little high, I gather)
[04:20:09] <K`zan> SWPadnos: lemme see what that was...
[04:20:51] <SWPLinux> they're both something like 3.13 ohms - the AD motors are slightly higher current and slightly lower resistance than the Xylotex, IIRC
[04:20:56] <K`zan> SWPadnos: Well, I am not sure what I can get out of that 24v charger, 10A should give it something at 3A
[04:20:59] <SWPLinux> err - 3.13 volts ...
[04:21:09] <SWPLinux> voltage = speed, current = torque
[04:21:13] <SWPLinux> for a motor
[04:21:15] <K`zan> SWPadnos: Is that good?
[04:21:40] <K`zan> e.g."better" than the xylotex steppers?
[04:21:42] <SWPLinux> 24 V will work, but the corner frequency will be lower than with a 60V supply
[04:21:58] <SWPLinux> I think so - the slightly higher current is what gives you the slightly higher torque
[04:22:08] <SWPLinux> (for the most part)
[04:22:24] <K`zan> SWPLinux: Well, I can grow over time with this, can't do it all at once.
[04:22:45] <SWPLinux> true - you may need to wait until you can't stand waiting for the slow machine ;)
[04:22:54] <K`zan> No idea where I would look for a 75V 10A supply and I suspect it is not going to be cheap.
[04:23:09] <SWPLinux> 10A may be more than you need for those steppers
[04:23:10] <K`zan> SWPLinux: Gotta start somwhere.
[04:23:13] <SWPLinux> yep
[04:23:16] <K`zan> 3A each?
[04:23:18] <jmkasunich> grrrr
[04:23:21] <K`zan> ~
[04:23:27] <SWPLinux> no - 3V each. that was a typo of mine
[04:23:47] <K`zan> Current?!?
[04:23:58] <SWPLinux> they're around 2.8A, and the rule of thumb is 2x the rated current for 3 motors
[04:24:24] <SWPLinux> though that's for servos, I think. for steppers, it may be much lower since the Geckos are basically switching power supplies
[04:24:37] <K`zan> 2.8A/phase * 2 phases per motor * 3 motors...
[04:24:42] <SWPLinux> then again, you multiply by 2 because there are two windings :)
[04:24:59] <K`zan> 18A :-(
[04:25:03] <SWPLinux> that comes to around 10A, for servos. I suspect that it can be lower for steppers
[04:25:05] <K`zan> ~
[04:25:09] <SWPLinux> you forgot * 2/3
[04:25:19] <K`zan> 2/3?
[04:25:36] <SWPLinux> sorry - my explanation was ambiguous
[04:25:44] <K`zan> * K`zan confuzed...
[04:25:56] <SWPLinux> what I had meant to say was that you take the single motor rating and multiply by 2 to get the supply current for 3 motors
[04:26:06] <SWPLinux> so you can run 3 motors for the current rating of 2, basically
[04:26:20] <K`zan> "OK" :-).
[04:26:34] <K`zan> Still 10A :-(.
[04:26:46] <SWPLinux> now, the Geckos are switching power supplies, so unless you're running at full speed and full torque, you don't need the full 10A
[04:27:05] <K`zan> But shouldn't the PSU be scoped for full load?
[04:27:22] <K`zan> What happens when you have all three axis running (not common, but...).
[04:27:28] <SWPLinux> only if you expect to run all 3 motors at top speed and top torque for any length of time
[04:27:35] <SWPLinux> running at speed is the key
[04:28:01] <K`zan> Sigh, MUCH to learn.
[04:28:05] <SWPLinux> the gecko is a nearly 100% efficient switching power supply
[04:28:50] <K`zan> For now, I'll take your word for it and try to understand it in time.
[04:28:54] <SWPLinux> so if it needs to provide an effective 6V at 5.6A (2.8 * 2 phases) to the motor, it takes (6V/24V) * 5.8A from the supply, or 1.4A
[04:29:02] <SWPLinux> err - 5.6A
[04:29:22] <K`zan> Cruise speed?!?
[04:29:23] <SWPLinux> you need to look at is a power thing: power out of the gecko = power into the gecko
[04:29:25] <K`zan> ""
[04:29:47] <SWPLinux> if you need 20V * 5.6A, that's 112W
[04:29:51] <SWPLinux> (output)
[04:29:57] <K`zan> I got the data sheet for it, skimmed it briefly, will dig into it more deeply as I have time.
[04:30:19] <SWPLinux> so you also need 112W input, but 112/24 is only 4.666A
[04:30:24] <K`zan> Ah, I think I got a vague clue here, but may have to dig for it :)>
[04:30:52] <SWPLinux> the higher the supply voltage, the more that works to your advantage for "normal" speeds
[04:31:23] <K`zan> OK, but I can start with ~24V at around 10A right?
[04:31:36] <SWPLinux> (ie, that same 112W output with a 60V supply is less than 2A)
[04:31:35] <K`zan> How finely does that need to be filtered?
[04:31:56] <SWPLinux> not too finely. I think Mariss recommends 80000 * I / V microfarads
[04:32:17] <SWPLinux> hmmm - that's a lot for you, 33000 uF
[04:32:39] <K`zan> Used to have piles of those :)
[04:32:59] <SWPLinux> luckily, you can go the cheap route and get a 50V (or 35V) cap for this supply
[04:33:04] <K`zan> See my plan above?
[04:33:27] <K`zan> Steppers & cables next month then a geko a month for 3 months.
[04:33:31] <SWPLinux> yes indeed
[04:33:59] <SWPLinux> you should get the (a) power supply working soon though, so you can play with it :)
[04:34:04] <K`zan> Unless I get lucky and win the lottery (which I don't play so the ticket would have to blow into my hand ;-)
[04:34:23] <SWPLinux> after 2 months, you'll have a 2 axis CNC mill, and can actually make 2.5D things
[04:34:44] <K`zan> Will do, got to look into that charger, might be able to use an old set of Jean's wheelchair batteries for filtering to start with.
[04:35:02] <K`zan> Yep :-), looking forward to that :-).
[04:35:18] <SWPLinux> http://cgi.ebay.com/4700uF-35V-Radial-Electrolytic-Capacitors-10pcs_W0QQitemZ290071538190QQihZ019QQcategoryZ36336QQrdZ1QQcmdZViewItem
[04:35:20] <K`zan> This month, if I can swing it, get the extended tables for it.
[04:35:37] <K`zan> Long way from 33,000uF :-(.
[04:35:48] <SWPLinux> 10*4700 = 47000
[04:35:54] <K`zan> True...
[04:35:55] <SWPLinux> that's a lot of 10
[04:36:02] <K`zan> Yep, it is.
[04:36:04] <jmkasunich> what voltage?
[04:36:07] <SWPLinux> 35
[04:36:13] <SWPLinux> for a 24V supply
[04:36:27] <K`zan> With balancing resistors and I don't know if I'd need diodes there or not.
[04:36:41] <SWPLinux> bridge before bank of caps
[04:36:48] <K`zan> Could get expensive in a hurry.
[04:36:49] <SWPLinux> no balancing resistors - they're in parallel
[04:36:59] <K`zan> Ah, right...
[04:37:07] <K`zan> * K`zan shifts brain into gear :-)
[04:37:28] <K`zan> Thinking of HV amplifier supplies for ham radio :)
[04:37:32] <jmkasunich> how about 6 pieces of 6800uF 50V snap mount?
[04:37:53] <SWPLinux> that would do it
[04:38:04] <K`zan> 35V each...
[04:38:36] <jmkasunich> I have three cap banks here, removed (with a band saw) from a larger PC board
[04:38:40] <K`zan> Wait, caps in series increases value, so I would need the resistors...
[04:38:46] <SWPLinux> no
[04:38:51] <jmkasunich> each one is 6 pieces of 6800uF 50V, all in parallel
[04:38:57] <jmkasunich> K'zan - you want parallel
[04:39:10] <SWPLinux> caps in parallel increase capacitance, caps in series increase voltage but decrease capacitance
[04:39:10] <K`zan> Time to dig out the book again, sigh...
[04:39:25] <K`zan> * K`zan is stuck in HV amps....
[04:39:27] <SWPLinux> and may need balancing resistors
[04:39:52] <SWPLinux> http://cgi.ebay.com/4-Marcon-10000uF-35V-Radial-Electrolytic-Capacitors_W0QQitemZ7569832297QQihZ017QQcategoryZ111603QQrdZ1QQcmdZViewItem
[04:39:55] <K`zan> Heh, just how much is an 80V 10A supply ;-)?
[04:40:20] <SWPLinux> actually, there are pretty inexpensive transformers in that power range
[04:40:34] <SWPLinux> I have a giant one I'd sell you if the shipping wouldn't remove all reason to do so :)
[04:40:36] <jmkasunich> 80V 10A?
[04:40:43] <jmkasunich> I thought you were talking 24V?
[04:40:53] <K`zan> I need to talk to my old ibm sysprog buddy in chicago and see if he has any of those supplies left.
[04:40:59] <ejholmgren> there are some fairly stout toroidals lurking about on ebaty
[04:41:00] <SWPLinux> 24V == cheap, 60-75V = better performance :)
[04:41:21] <K`zan> * K`zan on a *BUDGET* :-(
[04:41:33] <jmkasunich> SWPLinux: understood, but I thought you had selected the lower voltage because it suited the motors
[04:41:45] <SWPLinux> nope - it suits the parts lying around
[04:41:54] <jmkasunich> spending _any_ money on something you know is inadequate is not the way to live on a budget
[04:42:07] <K`zan> jmkasunich: No, I just happen to have a 24V 10A wheel chair battery charger at my disposal...
[04:42:21] <jmkasunich> so if you're gonna do 24, then don't look on ebay
[04:42:23] <jmkasunich> do it without spending money
[04:42:26] <jmkasunich> (if you can)
[04:42:31] <K`zan> jmkasunich: Thus the 4 month acquisition schedule...
[04:42:55] <K`zan> Getting good stuff at 2x the price (aprox) over time rather than cheap shit NOW.
[04:43:00] <jmkasunich> is this 4 months to build the 24V system, or the good system?
[04:43:07] <K`zan> Yes, I suffer from IG as badly as the next person.
[04:43:32] <SWPLinux> 4 months for motors + geckos, power supply not included ...
[04:43:39] <jmkasunich> sorry if I sound confused, I missed the first part of the conversation
[04:43:45] <K`zan> jmkasunich: THis is going to take me time, I'll have steppers and drivers in 4 months and in the process I can be looking for a better power supply.
[04:43:49] <jmkasunich> are these motors that will want the nicer power supply?
[04:43:59] <K`zan> 24V was pretty good for the xylotex stuff...
[04:44:15] <SWPLinux> automation direct steppers, 276 oz-in (?), ~3V, 2.8A/phase
[04:44:22] <K`zan> jmkasunich: whether they do or not *I* do :).
[04:44:44] <jmkasunich> the reason I was asking is because I was doing some scrounging to see if I have stuff you could use
[04:44:54] <K`zan> Wow, thanks!
[04:44:55] <SWPLinux> so going with Mariss' suggestion of 20-25x, 60 is better, but they should work somewhat at the 8x level of 24V
[04:44:59] <jmkasunich> but it doesn't make sense for me to send you 24V parts if what you really want is higher
[04:45:26] <K`zan> jmkasunich: I do have the 24V 10A charger, when I can I'll get better.
[04:45:39] <jmkasunich> ok, you don't need 24V caps?
[04:45:49] <K`zan> Might be able to do a voltage doubler or trippler on it ??!??
[04:45:50] <jmkasunich> because I'd be happy to send you one of these little cap banks
[04:45:56] <SWPLinux> pull that thing apart - it's possible that the transformer can be rewired for a higher voltage
[04:46:00] <K`zan> jmkasunich: I need to look inside it again.
[04:46:06] <jmkasunich> probably less than a pound, so even shipping isn't much
[04:46:12] <K`zan> Will do that now, hold...
[04:46:23] <SWPLinux> actually, it probably has nig caps in it anyway - it's a DC supply already
[04:46:27] <SWPLinux> s/nig/big/
[04:46:35] <jmkasunich> its a battery charger
[04:46:41] <SWPLinux> true
[04:47:00] <jmkasunich> automotive type chargers don't have much if any filtering
[04:47:16] <jmkasunich> how many amps did you work out?
[04:47:28] <SWPLinux> well, if the 2/3 rule applies, 11A or so
[04:47:55] <jmkasunich> 2.8*2/3 = less than 2A per axis, times 3 axis = 6A
[04:48:03] <jmkasunich> am I missing something?
[04:48:03] <K`zan> Gonna drag laptop into the shop, brb
[04:48:05] <SWPLinux> * 2 phases / motor?
[04:48:14] <jmkasunich> no
[04:48:16] <SWPLinux> ok
[04:48:29] <jmkasunich> one reaches max current when the other is at zero
[04:48:33] <jmkasunich> besides its a power thing
[04:48:54] <SWPLinux> ah - even better :)
[04:49:15] <jmkasunich> * jmkasunich looks for mariss's white papoer
[04:49:16] <jmkasunich> paper
[04:49:32] <SWPLinux> yeah - I can never remember what he says about current
[04:49:44] <SWPLinux> and I read his answer every month or two when someone asks on the geckodrive list
[04:50:23] <jmkasunich> The drive will always d
[04:50:23] <jmkasunich> raw less than 2/3 of the motor’s rated
[04:50:23] <jmkasunich> current when it is parallel (or half-winding) connected and 1/3 of the motor’s rated current when it
[04:50:23] <jmkasunich> is series (or full-winding) connected.
[04:50:37] <jmkasunich> so 6A
[04:51:01] <SWPLinux> ok. 6A 75V (max) is 450W
[04:51:29] <jmkasunich> I bet 3V motors will do great with anything from 48V up
[04:51:44] <SWPLinux> * 1.8 for rectifier duty, and you get a 800-ish VA transformer
[04:51:57] <SWPLinux> yeah - that's the max you would want
[04:52:01] <SWPLinux> (that = 75V)
[04:52:10] <SWPLinux> of course, on a minimill, that may be too much :)
[04:54:35] <jmkasunich> I have several 24V (regulated) 7.2A power-one linears, and could spare at least one, but even if you tapped the unregulated DC its probably no more than 36V or so
[05:00:42] <Anastasia> This is K`zan in the shop
[05:00:56] <Anastasia> ok got the cover off the supply, checking...
[05:00:57] <SWPLinux> this is a good transformer: http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=270078806725
[05:01:04] <SWPLinux> $59 though
[05:01:30] <jmkasunich> plus shipping
[05:01:35] <SWPLinux> $10
[05:01:41] <jmkasunich> and assuming it goes for the starting bid
[05:01:56] <SWPLinux> true. they always have them though - it's a store
[05:02:38] <jmkasunich> actually, 28 x 2 * 1.414 = 80V
[05:02:42] <jmkasunich> no margin
[05:02:54] <Anastasia> 26.5 vac to the rectifier.
[05:02:55] <SWPLinux> parallel the secondaries
[05:03:08] <jmkasunich> oh, 40V
[05:03:09] <jmkasunich> that works
[05:03:30] <jmkasunich> although I thought you were aiming a little higher than that
[05:03:41] <jmkasunich> Anastasia: full wave bridge rectifier?
[05:03:57] <SWPLinux> 26.5 is great. I wonder how far it drops
[05:04:01] <Anastasia> 24.7 out of the full wave bridge
[05:04:02] <SWPLinux> under load
[05:04:08] <jmkasunich> something screwy
[05:04:15] <SWPLinux> err - yeah
[05:04:17] <Anastasia> No load, just sitting here plugged in and on.
[05:04:24] <jmkasunich> 26.5AC thru a full wave bridge with a cap, should be making 37V
[05:04:41] <Anastasia> Gonna disconnect the stuff on the other side of the bridge.
[05:04:41] <SWPLinux> that's either the worst rectifier ever made, or the measurements are wrong ...
[05:04:48] <SWPLinux> was that on DC both times?
[05:05:27] <Anastasia> No, ac first (output from xformer) then DC across the bridge.
[05:05:44] <jmkasunich> is there a big cap across the bridge output?
[05:06:00] <jmkasunich> if not, the meter might be reading average
[05:08:58] <Anastasia> Not sure, values are conistant
[05:09:13] <SWPLinux> stick the meter on AC on the bridge output
[05:09:18] <SWPLinux> see how much it varies
[05:09:24] <jmkasunich> can you see much of the interior, or is it pretty packed?
[05:09:30] <Anastasia> 24v out of the rectifier makes sense with 26.5 - .7 x4 ?
[05:09:41] <jmkasunich> nope
[05:09:42] <SWPLinux> (that will tell you how much it varies, I mean)
[05:09:58] <jmkasunich> there are only two diode drops with a bridge
[05:10:19] <jmkasunich> if there is a cap, it will charge to the peak of the sine wave, which is 1.414 times the rms value
[05:10:24] <Anastasia> Nope, largish transformer, rectifier and the battery charging stuff (currently disconnected)
[05:10:39] <jmkasunich> that explains it
[05:10:44] <jmkasunich> you need some Cee
[05:11:01] <Anastasia> Cee?
[05:11:06] <jmkasunich> capacitors
[05:11:15] <Anastasia> Ah
[05:11:18] <SWPLinux> big "C"
[05:11:19] <Anastasia> right.
[05:11:50] <Anastasia> 26VAC useful at all ?
[05:12:30] <jmkasunich> with caps, that should get you 36VDC or so
[05:12:47] <jmkasunich> not great, but not bad
[05:12:48] <SWPLinux> so 50V would be good ;)
[05:12:51] <CIA-8> 03jmelson 07TRUNK * 10emc2/src/hal/drivers/hal_ppmc.c: correct handling of encoder index-reset for homing sequence
[05:12:56] <SWPLinux> 50V caps, that is
[05:13:01] <Anastasia> Looks to be wound with about 18ga wire...
[05:13:08] <jmkasunich> the transformer has only 2 secondary leads, going to the bridge?
[05:13:13] <Anastasia> 100v caps :)
[05:13:18] <Anastasia> Yes
[05:13:23] <jmkasunich> bummer
[05:13:28] <Anastasia> 2 secondary, checking primary
[05:13:40] <jmkasunich> I was hoping it was center tapped with two diodes instead of a full bridge
[05:13:49] <jmkasunich> then you could rewire for twice the volts
[05:14:02] <Anastasia> 4 primary, 2 connected together.
[05:14:16] <jmkasunich> probably dual input voltage, 120 and 240
[05:14:16] <SWPLinux> and it runs in 120V?
[05:14:21] <Anastasia> center tap arrangement - not that would help.
[05:14:27] <Anastasia> Yes 12vac
[05:14:35] <jmkasunich> hmm
[05:14:44] <jmkasunich> you mean 120Vac, right ;-)
[05:14:44] <SWPLinux> the series config should be for the higher voltage
[05:14:48] <jmkasunich> yeah
[05:14:50] <Anastasia> Input (from plate): AC input 12v 3.7A 60 hz
[05:15:02] <SWPLinux> err
[05:15:02] <Anastasia> 120v
[05:15:09] <Anastasia> hate this keyboard...
[05:15:15] <jmkasunich> laptop?
[05:16:11] <jmkasunich> for most dual voltage primaries, there are four wires, and the 120V connection ties two of them to one side of the 120V line, and the other two to the other side
[05:16:11] <Anastasia> yes, wireless is all it has going for it - gateway solo PII/266 384M RAM :)
[05:16:40] <jmkasunich> the 240v connection ties two leads together, and not to anything else, and the other two go to the two sides of the line (one wire to each side of the line)
[05:16:42] <Anastasia> two wires tied together AC line to the other two.
[05:16:52] <jmkasunich> thats odd
[05:17:06] <Anastasia> Suspect I could turn it into a 12V supply if I wanted (NOT)
[05:17:19] <SWPLinux> only if you don't mind melting the transformer
[05:17:38] <SWPLinux> (the higher voltage would saturate the core and cause problems)
[05:17:55] <jmkasunich> converting to 12V would mean lower voltage
[05:17:59] <jmkasunich> but irrelevant anyway
[05:18:12] <Anastasia> xformer ~ 3.5x4x4.75
[05:18:21] <jmkasunich> the input wiring seems screwy, but without knowing a lot more, you don't want to mess with it
[05:18:36] <Anastasia> dual primaries...
[05:18:49] <jmkasunich> yeah, but they're wired in series
[05:18:57] <jmkasunich> normally that would be the connection for 240V input
[05:19:04] <Anastasia> IIRC, wire parallel for 240V and series for 120...
[05:19:12] <SWPLinux> backwards
[05:19:12] <Anastasia> Yes
[05:19:13] <jmkasunich> nope, other way around
[05:19:17] <SWPLinux> (like the caps ... :) )
[05:19:33] <Anastasia> LOL
[05:19:52] <jmkasunich> think about it this way - each winding is 120V, regardless of the input
[05:19:56] <Anastasia> No ID on xformer
[05:20:06] <Anastasia> hummm
[05:20:12] <jmkasunich> for 120 input, you connect each one directly to the line, each one has 120V and they're happy
[05:20:17] <Anastasia> * Anastasia grabs pen and paper
[05:20:36] <SWPLinux> (and inductors are like resistors - they sum in series and divide in parallel)
[05:20:43] <jmkasunich> for 240, you connect them in series, each one has 120V, the series combo has 240V and they're happy
[05:20:50] <Anastasia> * Anastasia thinks she'll leave input alone...
[05:21:03] <jmkasunich> good plan
[05:21:08] <Anastasia> :-)
[05:21:24] <Anastasia> Being attched to the wall is no fun :-)
[05:21:44] <jmkasunich> now - if you add some 50V caps to the output, you will probably get 36V or so DC
[05:21:45] <Anastasia> THing is bloody ancient.
[05:22:31] <Anastasia> looks like there may be a selenium rectifier on the charging circuit plate...
[05:22:35] <ejholmgren> just don't irc from the bathroom :/
[05:22:51] <Anastasia> Why? some people have a head phone :-).
[05:23:17] <jmkasunich> ?
[05:23:27] <Anastasia> telephone in the bathroom...
[05:23:37] <jmkasunich> Anastasia: selenium - that is ancient
[05:23:43] <Anastasia> perfect place for a laptop :-)
[05:23:49] <Anastasia> If that is what it is.
[05:23:52] <jmkasunich> and whats this plate stuff? does it have tubes in it?
[05:24:24] <ds3> what about a pair of theses:
[05:24:30] <Anastasia> No, just an aluminum plate the charging circuitry is build on. I'll take pix and put them up...
[05:24:30] <ds3> http://www.mpja.com/productview.asp?product=7846+TR
[05:24:40] <ds3> 10A each peice
[05:25:07] <jmkasunich> ds3 - still only about 36V output
[05:25:22] <Anastasia> essentially what U have ...
[05:25:25] <Anastasia> I
[05:25:30] <jmkasunich> right
[05:25:43] <jmkasunich> except the one you have you don't have to pay for (or pay shipping, transformers aren't light)
[05:28:40] <Anastasia> True :) prolly cost more to ship than to buy.
[05:28:54] <Anastasia> Got pix, lemme shut down in here and put them up.
[05:28:58] <jmkasunich> ok
[05:33:37] <ds3> is it a bad idea to make stepper motor mounts out of steel?
[05:37:25] <K`zan> Should have them up in just a sec...
[05:40:04] <dead_ted> hey folks: getting this error on make. Only been able to find one obscure reference to it in an old emc irc log:
[05:40:11] <dead_ted> In function `rtapi_task_new':
[05:40:15] <dead_ted> warning: passing arg 2 of `rt_task_init_cpuid' from incompatible pointer type
[05:40:39] <dead_ted> http://www.linuxcnc.org/irc/irc.freenode.net:6667/emcdevel/2006-02-11.txt
[05:40:42] <jmkasunich> what OS are you running?
[05:40:55] <dead_ted> Gentoo
[05:41:10] <dead_ted> compiled the kernel according to the instructions provided on the EMC site
[05:41:11] <jmkasunich> its some kernel header thing
[05:41:26] <K`zan> http://wrlabs.shacknet.nu/~vw/MyMachineShop/tmp/battcharger24v/images.html
[05:41:39] <K`zan> Click on thumbs for larger (BIG)...
[05:42:37] <jmkasunich> http://jmkasunich.dyndns.org/pics/capbank50v.jpg
[05:42:50] <K`zan> black rectangular thing is a 6" ruler...
[05:43:31] <K`zan> loading, slowly...
[05:43:33] <SWPLinux> dead_ted: what CPU? I'm assuming you're using a 32-bit Linux
[05:43:34] <K`zan> :-)
[05:43:42] <jmkasunich> dead_ted: I remember that error, I just don't remember the details
[05:44:04] <dead_ted> heh... that's what you said in the irc log I found ^_^
[05:44:12] <dead_ted> Intel(R) Pentium(R) III CPU family 1133MHz
[05:44:17] <jmkasunich> it was something like one kernel/rtos version used int and one used unsigned int or some such
[05:44:43] <jmkasunich> no...
[05:44:49] <jmkasunich> lemme look at something
[05:44:51] <K`zan> jmkasunich: NICE bank of caps !
[05:45:14] <jmkasunich> I'd be happy to send you one
[05:45:19] <jmkasunich> you have soldering equipment?
[05:45:33] <jmkasunich> you'll need to scrape off the solder mask and attach wires
[05:45:37] <K`zan> jmkasunich: Sure, just let me know what you want for them!
[05:45:44] <jmkasunich> you can see that the caps are already wired in parallel
[05:45:47] <K`zan> I can handle that :).
[05:46:03] <jmkasunich> I didn't pay anything for them, I don't expect anything (they're going to a good cause)
[05:46:13] <K`zan> Shipping at least?
[05:46:35] <SWPLinux> maybe we should set up a developer food fund for the CNC workshop ;)
[05:46:59] <K`zan> Mine could use it ;-)!
[05:47:30] <jmkasunich> does the usps still do $2.95 for small priority mail packages?
[05:47:37] <K`zan> Gurgle, dunno.
[05:47:40] <jmkasunich> K'zan, what is your zip code?
[05:47:50] <K`zan> 98103-3362
[05:47:51] <SWPLinux> I think so - though it may be $3.50 or something now
[05:48:04] <SWPLinux> dead_ted: could you stick more of the make error into pastebin?
[05:48:23] <jmkasunich> looks like priority is $4.05
[05:48:35] <K`zan> NO problem :-)!
[05:48:45] <dead_ted> rtapi/rtai_rtapi.c: In function `rtapi_task_new':
[05:48:45] <dead_ted> rtapi/rtai_rtapi.c:693: warning: passing arg 2 of `rt_task_init_cpuid' from incompatible pointer type
[05:48:45] <dead_ted> make: *** [objects/rtrtapi/rtai_rtapi.o] Error 1
[05:48:52] <dead_ted> get quite a few warnings before that
[05:49:05] <jmkasunich> K'zan: email me your shipping address, I'll send it out
[05:49:15] <jmkasunich> no need to pay me, but if you feel the need, $4 is fine
[05:49:17] <K`zan> your email addy?
[05:49:19] <SWPLinux> that was http://pastebin.ca/ , not "paste in" :)
[05:49:28] <jmkasunich> jmkasunich at att dot net
[05:49:37] <K`zan> rr copy.
[05:49:42] <dead_ted> hah... was wondering what that was... been outta irc for awhile, thought it was some new hip term
[05:49:50] <SWPLinux> heh
[05:50:13] <jmkasunich> you'll need a hot soldering iron
[05:50:19] <jmkasunich> that is 4 oz copper
[05:50:35] <dead_ted> hey, that's pretty slick: http://pastebin.ca/320238
[05:50:50] <K`zan> jmkasunich: Sent!
[05:50:54] <jmkasunich> the mounting holes are connected to the negative side of the caps, which is convenient
[05:51:11] <dead_ted> I have two processors if that makes a diff
[05:51:24] <SWPadnos> are you running SMP+RTAI?
[05:51:47] <K`zan> jmk, yes, handy! Got LOTS of room in that case.
[05:52:54] <jmkasunich> dead_ted: /usr/src/linux/include/asm/adeos.h:25:25: irq_vectors.h: No such file or directory
[05:53:03] <jmkasunich> thats not a warning I don't think
[05:53:11] <K`zan> And it has LOTS of cabling for both in and out.
[05:53:26] <K`zan> LOL, probably a deal breaker :)
[05:54:36] <jmkasunich> dead_ted: I'm afraid I haven't build a RT patched (or even a regular) kernel
[05:54:55] <jmkasunich> there is some problem in your system's kernel related headers, but I don't even know where to start
[05:55:13] <K`zan> * K`zan just installed from the live ubuntu... Simple (thanks folks!).
[05:55:20] <dead_ted> ok, no worries... I'll do some digging
[05:55:25] <jmkasunich> dunno where you are, but its almost 1am here
[05:55:35] <jmkasunich> most of the developers are asleep
[05:55:38] <dead_ted> Ubuntu... heh
[05:55:49] <jmkasunich> perhaps you should send an email to the developers list
[05:56:00] <K`zan> Kernel headers from the looks of it:
[05:56:00] <dead_ted> well thanks anyway, I'll do some pokin' round and maybe drop 'em a line
[05:56:02] <jmkasunich> either include the make results, or point at the pastebin
[05:56:03] <K`zan> /usr/src/linux-headers-2.6.17-10/include/asm-um/irq_vectors.h
[05:56:03] <K`zan> /usr/src/linux-headers-2.6.17-10/include/asm-i386/mach-voyager/irq_vectors.h
[05:56:03] <K`zan> /usr/src/linux-headers-2.6.17-10/include/asm-i386/mach-default/irq_vectors.h
[05:56:03] <K`zan> /usr/src/linux-headers-2.6.17-10/include/asm-i386/mach-visws/irq_vectors.h
[05:56:33] <dead_ted> yea, that's the problem, I have those header files, but not exactly sure which one it is looking for
[05:56:43] <jmkasunich> I don't know if anybody has tried to build on gentoo - if we can improve our configure to find the headers, we'd like to do that
[05:56:58] <SWPLinux> SMP may also be causing some issues
[05:57:10] <SWPLinux> that's not a supported thing (unfortunately)
[05:57:17] <K`zan> I've got gentoo here (32 and 64 bit) in case that helps.
[05:57:27] <K`zan> Don't use it anymore, but I have it.
[05:57:27] <jmkasunich> not supported by rtai that is
[05:57:33] <dead_ted> ok, I'll try a recompile with out SMP and see if it changes anything
[05:57:37] <SWPLinux> by "unsupported", I mean that we're not sure it works
[05:57:51] <SWPLinux> actually, the RTAI / SMP stuff is quite old (and possibly mature)
[05:58:04] <jmkasunich> its rtai/64bit that is flakey, right?
[05:58:07] <SWPLinux> yep
[05:58:28] <SWPLinux> I know SMP has been there for a long time, but I don't know how well it works
[05:58:54] <SWPLinux> my only (functional) SMP machine is Opterons, and I can't bring myself to install a 32-bit OS on it :)
[05:59:01] <jmkasunich> dead_ted: keep track of your results, capture both make and configure output,and post to the list
[05:59:08] <K`zan> 64 bit everything is flakey IMO...
[05:59:10] <dead_ted> sure thing
[05:59:16] <SWPLinux> K`zan: :P
[05:59:30] <jmkasunich> I have two smp P3 boxes, but neither is in use right now
[05:59:38] <jmkasunich> and I'm not about to build a kernel just to try it
[05:59:45] <K`zan> Hey, I'm running 32bit on my amd64 box now because of all the apps I either can't get or won't run...
[05:59:49] <jmkasunich> too many other things on my plate
[06:00:05] <SWPLinux> I have one probably dead forever SMP Pentium Pro Overdrive machine (woohoo!)
[06:00:23] <K`zan> Maybe in another year or two... :-/...
[06:00:26] <SWPLinux> Flash is the only one that annoys me, I think
[06:00:34] <SWPLinux> and of course EMC
[06:00:50] <SWPLinux> (though that works great in sim mode on this machien)
[06:00:54] <K`zan> Runs fine in 32 bit and does one really need 64 for that?
[06:01:25] <SWPLinux> no, but I have 4G ram in this machine, and 32-bit compiles either can't use it all, or are slower because of it
[06:01:35] <SWPLinux> err - 32-bit OSes
[06:01:58] <K`zan> You need 4G to run EMC - I'm in BIG trouble with 512M in that box...
[06:02:24] <SWPLinux> I hae 4G to do imaging and FPGA stuff (while running Windows XP / 200 in a VM) ...
[06:02:28] <SWPLinux> s/hae/have/
[06:02:33] <K`zan> Ah, gotcha.
[06:02:35] <SWPLinux> emc is just a sideline on this machine
[06:02:58] <K`zan> Never got into FPGA stuff, looks real interesting and handy though.
[06:02:59] <SWPLinux> the bright side is that I can do a full compile of emc (after a make clean) in around 20 seconds ;)
[06:03:13] <K`zan> :-) ALWAYS *nice*!
[06:04:17] <K`zan> Gonna go clean up the shop, place is so small and cluttered that anytime you do something it is pull-it-all-out do it and put-it-all-away :-/.
[06:04:31] <SWPLinux> heh - see you.
[06:04:37] <K`zan> bbiaf
[06:04:40] <SWPLinux> I should get to bed, it's 1AM here
[06:04:45] <jmkasunich> ditto
[06:04:48] <jmkasunich> goodnight all
[06:04:55] <SWPLinux> see you all later
[06:05:26] <dead_ted> well thanks for the help.. doin' a reboot. I'll post to the list either way. Or maybe dual boot this Ubuntu all the kids are talking about these days
[06:05:37] <SWPLinux> you may like it
[06:05:47] <SWPLinux> (unless you really enjoy recompiling stuff a lot ;) )
[06:06:04] <dead_ted> yea, yea ... later
[06:13:40] <Anastasia> back on the other box.
[06:22:14] <K`zan> Okdokee, all cleaned up and eagerly awaiting the next project :)!
[07:36:21] <A-L-P-H-A> damn it, and it was the perfect day...
[07:57:50] <A-L-P-H-A> hey A-L-P-H-A
[07:57:54] <A-L-P-H-A> heh
[07:57:56] <A-L-P-H-A> hey alex_joni
[08:09:20] <ds3> anyone know if you can braze/silver solder 12L14?
[08:10:32] <A-L-P-H-A> ds3? for?
[08:10:44] <A-L-P-H-A> context is a good thing...
[08:12:48] <ds3> I have a rotary table with a collar that is smaller the the closest peice of steel tubing I want to use to make a stepper motor mount. So I was thinking of silver soldering in a spacer ring on to the tube
[08:13:11] <ds3> the tube is probally 1018 or similar mild steel
[08:16:37] <A-L-P-H-A> ds3 what's the size difference?
[08:16:38] <A-L-P-H-A> vibrations?
[08:16:45] <A-L-P-H-A> misalignment?
[08:16:53] <ds3> let me get my figures
[08:17:29] <A-L-P-H-A> I wouldn't use silver solder, as it'll be subjected to some stresses... silver solder is for like pipes, and electronics... not really to hold things together.
[08:17:39] <ds3> not much... around 50 thou
[08:17:49] <A-L-P-H-A> that's huge... 0.05" is a lot of space...
[08:18:05] <ds3> that's why I want secure a spacer ring
[08:18:15] <A-L-P-H-A> oh...
[08:18:24] <A-L-P-H-A> oversize the spacer ring, to get a compression fit.
[08:18:27] <A-L-P-H-A> won't go anywhere.
[08:18:30] <ds3> it'll be too thin to make it just stand alone
[08:18:31] <A-L-P-H-A> ever.
[08:19:01] <ds3> wouldn't the oversize ring collapse when I press it in?
[08:19:08] <A-L-P-H-A> are you just making a coupler between the rotary table, and stepper motor?
[08:19:30] <ds3> this is the mount frame to support the coupler
[08:19:36] <A-L-P-H-A> os it this the stepper motor mount...
[08:19:39] <A-L-P-H-A> os=or
[08:19:42] <A-L-P-H-A> got photos?
[08:19:55] <ds3> not really, I am studying the table at the moment to design things
[08:19:55] <A-L-P-H-A> cause I'm having a hard to time visualizing from your desriptions... plus I like photos. :D
[08:20:08] <A-L-P-H-A> I'm doing this with mine...
[08:20:13] <ds3> hmmm let me find a picture of the one I am copying
[08:20:49] <Jymmm> ds3 tube of what exactly? you said motor mount, and I haven't seen a tube on one of them yet
[08:20:50] <A-L-P-H-A> C-channel, and making the mount from that... table stub= |'''''''| = motor.
[08:20:57] <A-L-P-H-A> that's my ascii drawing of a C channel.
[08:22:36] <ds3> this is what I am trying to copy, except it will be to a rotary table instead of the mill table:
[08:22:37] <ds3> http://www.cartertools.com/mounts.html
[08:23:11] <A-L-P-H-A> ds3, hahah... why make it complicated like that?
[08:23:30] <A-L-P-H-A> avoid the tube, and just go with a thick C channel... unless you really want to go that route for some reason or another.
[08:23:42] <ds3> A-L-P-H-A: got a better idea? only reason is cuz I have a taig and I understand how that goes together
[08:24:00] <ds3> but how do I secure the C channel to the rotary table?
[08:24:23] <A-L-P-H-A> C channel... make it wide enough and the faces large enough... so that you can making mounting holds on either ends... then put a journal between the two shafts.
[08:24:39] <A-L-P-H-A> no welding, minimal fuss.
[08:25:01] <A-L-P-H-A> or box tubing to do the same thing as above...
[08:25:02] <K`zan> Night folks
[08:25:06] <ds3> so basically a hose clamp on the table side?
[08:25:19] <A-L-P-H-A> no.
[08:25:31] <ds3> I don't see how it will secure to the table
[08:25:37] <A-L-P-H-A> ds3, see on the left size... http://www.cartertools.com/cncmt05.jpg there's two screws?
[08:26:06] <ds3> this is not the exact thing I want to mount it to
[08:26:19] <A-L-P-H-A> ds3, I understand... principals the same.
[08:26:23] <ds3> let me find a picture of the other half; I used that link just for the mount (blue part)
[08:26:29] <ds3> this table has no screw holes
[08:26:42] <A-L-P-H-A> oh
[08:26:52] <A-L-P-H-A> would you be willing to drill and tap some holes?
[08:27:10] <ds3> i prefer not to drill/tap
[08:27:33] <A-L-P-H-A> ds3, what type of machine is htis?
[08:27:34] <A-L-P-H-A> this
[08:28:09] <ds3> the mill is a Taig but the part I am doing is the 4th axis
[08:28:25] <ds3> this is the rotary table I am trying to CNC:
[08:28:27] <ds3> http://www.littlemachineshop.com/products/images/480/480.1927.jpg
[08:28:47] <Jymmm> ds3 you are trying to mount that to your table?
[08:28:49] <ds3> the knob and dial is coming off to attach the motor to
[08:29:04] <ds3> Jymmm: no, I am trying to mount a stepper to that
[08:29:06] <Jymmm> ds3: ok, and the problem is?
[08:29:52] <Jymmm> ds3: mounting the motor, or coupleing to the rotary table shaft?
[08:30:13] <ds3> mounting the motor
[08:30:23] <ds3> I can copy the taig couplers
[08:30:53] <Jymmm> ds3 : Couldn't you just use a long 'L' bracket?
[08:31:19] <ds3> Jymmm: I don't want to drill/tap the rotary table itself
[08:31:27] <Jymmm> ds3: Yes, I know.
[08:32:23] <ds3> that opens up a can of worms; there are very few surfaces that are prependicular to the shaft and the shaft needs to move for the backlash adjustment
[08:32:40] <A-L-P-H-A> ds3, I know you don't want to drill/tap... but it would make things trivial to get mounts on... would you be willing to reconsider?
[08:32:48] <alex_joni> hi A-L-P-H-A
[08:32:52] <A-L-P-H-A> hey alex_joni.
[08:32:55] <alex_joni> just recovered form a power loss :)
[08:32:58] <A-L-P-H-A> alex_joni, shibby
[08:32:59] <alex_joni> fscking power company :)
[08:33:06] <alex_joni> well.. I recovered ..
[08:33:14] <alex_joni> my systems were all up and running in no-time :)
[08:33:22] <ds3> A-L-P-H-A: not unless I can be sure I don't break their backlash adjustment
[08:34:15] <A-L-P-H-A> alex_joni... UPS?
[08:34:18] <A-L-P-H-A> no UPS?
[08:34:28] <A-L-P-H-A> which reminds me... I should test my UPS sometime soon.
[08:34:30] <Jymmm> ds3 doesn't the rotary table have t-slots on both sides of it?
[08:34:53] <A-L-P-H-A> Jymmm, many only have two sides... a bottom, and side mount.
[08:34:58] <ds3> Jymmm: nope.
[08:35:24] <Jymmm> ds3 then can't you rais the RT off the table by 1/2"?
[08:35:30] <Jymmm> raise
[08:36:01] <ds3> Jymmm: not really, this is a Taig... Z axis is limited as is :(
[08:36:14] <Jymmm> ds3 you only have two choices: mount to the RT, or mount to the table.
[08:36:43] <ds3> Jymmm: if I can solve my spacer problem I can do this
[08:36:55] <ds3> okay, loctite 690 it is, forget about soldering
[08:37:22] <Jymmm> ds3 Silly Putty!!!
[08:37:32] <ds3> =)
[08:38:09] <Jymmm> ds3 oh yeah, almost forgot.... WUSS!
[08:38:25] <Jymmm> ds3 tap it and easy the issues
[08:39:00] <Jymmm> ds3 besides, how does tiag do their cnc converstion?
[08:39:43] <Jymmm> Ewwwwwwwwwwwwww, a set screw?!
[08:39:48] <ds3> Jymmm: Taig sells the Sherline RT as the 4th axis; that is designed from the ground up to be CNC
[08:40:33] <Jymmm> ds3 I still say drill and tap (or add a plate)
[08:41:06] <ds3> i'll think about it
[08:41:39] <A-L-P-H-A> Jymmm... I agree with Jymmm. :) but hey... it's your toy. :)
[08:42:13] <A-L-P-H-A> Love the song "wake me till september ends" - greenday.
[08:47:41] <Jymmm> ds3: Are you going to mount the RT upwards?
[08:47:49] <Jymmm> ||---
[08:48:52] <Jymmm> ds3: You COULD make two sets of brackets, using the other holes on the RT to mount againest. One set for vertical, another for horizontal.
[08:48:54] <ds3> Jymmm: I want to maintain that option
[08:49:16] <ds3> I still like to find a way to grab that collar
[08:51:45] <Jymmm> freeze the RT, heat the coupler
[08:52:00] <A-L-P-H-A> you know, that coupler isn't a terrible design... could be improved on... but hey... not bad at all.
[08:52:03] <A-L-P-H-A> http://www.cartertools.com/mounts.html
[08:52:40] <ds3> whoa... the problem is now how to grab the coupler.... my problem was if I can solder 12L14 to mild steel so I can use my scrap that I ahve already
[08:52:59] <ds3> Shrink fitting onto it is definitely an option
[08:53:32] <ds3> the loctite idea I was contemplating was for a spacer between my oversized (slightly) tube
[08:58:16] <A-L-P-H-A> ds3, call loctite... they have some stuff... but really depends on a lot of factors... call their engineering/tech guys, and they'll tell you what to use...
[08:59:06] <ds3> n/m this is a non-problem. I miscounted my digits reading my mic.
[08:59:14] <ds3> it is only 0.005 off. ARRRGGGG
[08:59:20] <A-L-P-H-A> oh. that's fuck'n nothing.
[08:59:44] <ds3> exactly. I hate reading mikes this late at night. almost an excuse to get a digital mike ;)
[09:00:17] <A-L-P-H-A> get a mitoyo :D
[09:00:21] <A-L-P-H-A> I want one.
[09:00:32] <ds3> that one that reads to 0.00005?
[09:02:56] <A-L-P-H-A> ummmmmmmmmm. I think 3 zeros, not 4.
[09:03:52] <ds3> some one showed me one that read to the millionth
[09:04:05] <Jymmm> ds3: freeze the RT, heat the coupler
[09:04:14] <A-L-P-H-A> mitutoyo sorry
[09:05:01] <ds3> Jymmm: rev 2 will probally be that; going to try to throw together a set screw versions first
[09:05:17] <ds3> the amount of backlash may make this not very viable anyways :/
[09:05:34] <Jymmm> ah
[09:05:39] <A-L-P-H-A> no adjustments in the backlash?
[09:05:49] <A-L-P-H-A> there is even in teh crappy india one I got
[09:06:02] <ds3> A-L-P-H-A: there is an adjustment; that's one of the reasons I did not want to drill in
[09:06:22] <ds3> i just don't know how tight i can get it before the motor stalls so i need a proof of concept first
[09:06:23] <A-L-P-H-A> the adjustment is in that ring?
[09:06:58] <ds3> elsewhere but from the docs, it causes the worm to shift orientation and I think that ring follows the worm
[09:07:23] <A-L-P-H-A> ooooooooooooooooh, it is 4 zeros.
[09:07:25] <A-L-P-H-A> damn
[09:07:34] <ds3> nice ain't it? ;)
[09:08:10] <A-L-P-H-A> I want this. http://cgi.ebay.com/Mitutoyo-293-344-DIGIMATIC-SPC-Coolant-Proof-Micrometer_W0QQitemZ130068382128QQihZ003QQcategoryZ92085QQrdZ1QQcmdZViewItem
[09:08:34] <A-L-P-H-A> http://cgi.ebay.com/Digital-Mitutoyo-Coolant-Proof-1-Micrometer-Brand-New_W0QQitemZ180074540679QQihZ008QQcategoryZ92085QQrdZ1QQcmdZViewItem
[09:09:57] <ds3> nice but too rich for me
[09:10:16] <ds3> that thing is probally more accurate then my gage block set ;)
[09:10:21] <A-L-P-H-A> hahah... probably
[09:10:31] <A-L-P-H-A> I don't have a gage block set.
[09:10:35] <A-L-P-H-A> so you're one up on me there.
[09:10:48] <A-L-P-H-A> don't need one either though... so... :)
[09:11:10] <alex_joni> A-L-P-H-A: big UPS
[09:11:17] <ds3> i kind of regret getting that... should've spent the money on a gage pin set
[09:11:22] <alex_joni> but it was 3 times today .. about 45 minutes each time
[09:11:28] <alex_joni> my UPS doesn't handle that :)
[09:11:31] <alex_joni> 2000VA though
[09:11:35] <A-L-P-H-A> alex_joni, oh... damn
[09:11:47] <A-L-P-H-A> alex_joni... at home laready?
[09:11:53] <alex_joni> at work :)
[09:11:56] <A-L-P-H-A> oh
[09:12:02] <alex_joni> 11:11 < alex_joni> at work :)
[09:13:15] <alex_joni> 67.5% charge in the UPS right now
[09:13:28] <A-L-P-H-A> heh
[09:14:15] <ds3> anything think theses couplers are a worthwhile design:
[09:14:18] <ds3> http://www.cnconabudget.com/coupler.JPG
[09:14:46] <A-L-P-H-A> ds3, those are pretty trivial to make if you have a lathe, mill, and slotter.
[09:15:20] <ds3> A-L-P-H-A: slotter? I'd think a slitting saw on the mill should do
[09:15:35] <ds3> but is it any good ;)
[09:15:42] <A-L-P-H-A> dunno... I use gears instead.
[09:15:50] <A-L-P-H-A> XL gears... works great...
[09:16:07] <ds3> came across a lot of places suggesting a 'soft' coupling (belts, springs, etc)
[09:16:54] <ds3> think even the current (last?) issue of MEW suggested soft coupling for a CNC conversion
[09:17:13] <A-L-P-H-A> some people have great success with radiator hoses.
[09:17:17] <Jymmm> OMG.... Corel bought WinZip and WinDVD, I'll be damn
[09:17:31] <A-L-P-H-A> really? were the PR?
[09:17:32] <ds3> whoa...
[09:17:33] <A-L-P-H-A> were's
[09:17:46] <Jymmm> http://www.corel.com/servlet/Satellite?c=Content_C1&cid=1152796555474&lc=en&pagename=CorelCom%2FLayout
[09:17:51] <ds3> who's having a pool on how long before corel spins it off?
[09:17:55] <Jymmm> "Current Product Releases"
[09:18:28] <A-L-P-H-A> Jymmm, are you sure they aren't reselling?
[09:18:48] <A-L-P-H-A> cause winzip.com still says nothing about corel
[09:19:05] <Jymmm> click on ABOUT on winzip page
[09:19:05] <alex_joni> A-L-P-H-A: but nut works really well
[09:19:19] <A-L-P-H-A> oh
[09:19:29] <alex_joni> A-L-P-H-A: http://www.networkupstools.org/
[09:19:31] <A-L-P-H-A> Copyright © 2006 WinZip® International LLC | Privacy Policy
[09:19:31] <A-L-P-H-A> WinZip is a Registered Trademark of WinZip International LLC
[09:19:31] <Jymmm> Thank you for your interest in WinZip Computing! A lot has changed over the last few years, and we're now part of Corel Corporation.
[09:19:38] <Jymmm> Thank you for your interest in WinZip Computing! A lot has changed over the last few years, and we're now part of Corel Corporation.
[09:19:52] <A-L-P-H-A> yeah yeah.
[09:19:56] <A-L-P-H-A> but they should update this shit then.
[09:20:11] <Jymmm> A-L-P-H-A they did
[09:20:23] <A-L-P-H-A> Jymmm, bottom copyright.
[09:20:46] <Jymmm> A-L-P-H-A so they're a seperate legal entity, bfd
[09:21:11] <A-L-P-H-A> * A-L-P-H-A pats Jymmm on the head... good boy. go rewards yourself with a cookie now. :)
[09:22:04] <Jymmm> A-L-P-H-A I am tired of you're constant mockery. It is NOT funny, and anymore just annoying.
[09:22:30] <A-L-P-H-A> Jymmm, ignore me then... don't abuse your op powers.
[09:23:03] <Jymmm> A-L-P-H-A Just start behaving.
[09:23:12] <A-L-P-H-A> Jymmm, grow a thicker skin.
[09:23:28] <Jymmm> A-L-P-H-A I'm serious.
[09:23:38] <A-L-P-H-A> then ignore me.
[09:25:42] <Jymmm> I've had him on ignore for two days already. I'm sorry but enough is enough. I've directly told him it's not appreaciated and he continues his same actions. One of you are welcoem to talk to him, but it's getting really old now.
[09:26:18] <Jymmm> I like the guy, but damn.
[09:27:24] <Jymmm> Maybe now I'm being a bit vendictive, but was planning on removing the ban in 5 minutes anyway.
[09:31:16] <alex_joni> Jymmm: please unban him
[09:31:26] <Jymmm> in 3 minutes
[09:31:54] <alex_joni> fine by me :)
[09:32:11] <alex_joni> didn't read back.. just stuck my head in here :)
[09:32:41] <Jymmm> alex_joni np, just totally tied of the dis-respectfullness (if you know what I mean)
[09:32:56] <Jymmm> err tired
[09:33:25] <alex_joni> * alex_joni notes not to pat Jymmm on the head .. ever
[09:33:48] <Jymmm> alex_joni that was the straw, this has been an on going thing for months.
[09:34:10] <alex_joni> * alex_joni notes not to be ongoing for months before he does that :)
[09:34:25] <Jymmm> anyhow, I'm going to unban him now, and just not speak to him anymore.
[09:35:43] <alex_joni> now behave and shake hands
[09:36:12] <A-L-P-H-A> :)
[09:40:27] <alex_joni> http://www.bigsister.ch/demo/main.png <- funny PC names
[09:41:28] <A-L-P-H-A> dun get it. :)
[09:42:00] <anonimasu> heh
[09:43:37] <Vq^> "rejectix", lol
[09:45:59] <alex_joni> disastrix :P
[09:51:45] <alex_joni> nice car on auction tomorrow
[09:52:02] <alex_joni> Ford Model A from 1903 (the 3rd car built by Ford)
[09:52:12] <anonimasu> :)
[09:52:15] <anonimasu> cheap
[09:52:17] <anonimasu> I bet
[09:53:59] <alex_joni> http://www.rmauctions.com/FeatureCars.cfm?SaleCode=AZ07&CarID=r206
[09:54:06] <alex_joni> not that expensive
[09:57:09] <Jymmm> #30, not #3, but still =)
[09:57:34] <anonimasu> *yawn*
[09:57:42] <Jymmm> I'd suspect Jay Lenon will bid on it.
[09:58:05] <alex_joni> the expected price is 400k$
[09:59:01] <alex_joni> wonder how accurate that is
[09:59:31] <Jymmm> Depends on how many are still in existance, looks like it's in immacualte condition.
[10:23:21] <A-L-P-H-A> hey lerneaen_hydra
[10:23:34] <lerneaen_hydra> 'lo A-L-P-H-A
[10:23:47] <lerneaen_hydra> anything happening?
[10:24:00] <lerneaen_hydra> :(
[10:24:10] <lerneaen_hydra> I'm scaring everone away
[10:24:21] <A-L-P-H-A> yuppers
[10:56:04] <rafa> hello friends
[10:56:21] <rafa> yestday i quetion about..
[10:56:54] <rafa> how to make M code for turn on one motor or led for exemple
[10:56:57] <rafa> whell
[10:57:12] <rafa> in my computer i have 2 parport
[10:57:39] <rafa> and in my satndart_pinoult.hal i delare only parport 378
[10:58:05] <rafa> is possible write other parport?
[10:58:19] <rafa> whitt emc runing?
[11:01:00] <alex_joni> yes
[11:01:10] <alex_joni> what is the address of the second port?
[11:01:59] <rafa> 0xb00
[11:02:16] <rafa> and i have one more
[11:02:19] <rafa> 0xd00
[11:02:27] <rafa> i made it?
[11:03:52] <rafa> being thus a378 sera used for the o emc and to another one I can use to bind the engines
[11:04:07] <rafa> ok?
[11:04:26] <rafa> being thus a378 is used for the o emc and to another one I can use to bind the engines
[11:04:32] <rafa> ok?
[11:07:33] <rafa> friend
[11:07:38] <rafa> its ok?
[11:08:40] <rafa> friend alex_joni, its ok?
[11:08:44] <rafa> please
[11:15:26] <rafa> please
[11:46:29] <rafa> thanks for your resposta
[11:55:53] <anonimasu> 3hello
[11:56:12] <alex_joni> huh.. he'll be back :)
[11:56:15] <alex_joni> I was busy :D
[11:56:20] <lerneaen_hydra> anon!
[11:56:22] <lerneaen_hydra> alex_joni!
[12:03:05] <anonimasu> :)
[12:03:40] <lerneaen_hydra> death note was licenced :(
[12:03:58] <lerneaen_hydra> err, licensed
[12:04:04] <anonimasu> lerneaen_hydra: www.baka-updates.com has even licensed stuff..
[12:04:06] <anonimasu> ^_^
[12:04:11] <lerneaen_hydra> ooh, yummy
[12:04:30] <lerneaen_hydra> hopefully some subbing group is still subbing it
[12:04:36] <lerneaen_hydra> live-evil dropped it though :(
[12:07:02] <anonimasu> :/
[12:07:06] <anonimasu> officially yes
[12:09:04] <anonimasu> go to irc and check :)
[12:11:48] <lerneaen_hydra> oooh
[12:12:23] <anonimasu> or well check baka-updates regularily
[12:12:31] <anonimasu> it's like the official no-licence tracker..
[12:12:34] <anonimasu> everything ends up there..
[12:16:04] <lerneaen_hydra> anonimasu, hmm, I see
[12:16:38] <lerneaen_hydra> I'm just afraid of saying something like "where's death note" and getting a !kb
[12:17:22] <anonimasu> well idle for a while
[12:18:42] <lerneaen_hydra> yeah
[12:22:30] <A-L-P-H-A> you guys are anime whores?
[12:22:59] <A-L-P-H-A> last thing I got was ergo proxy
[12:23:10] <A-L-P-H-A> alex_joni abusing ops too? :P
[12:23:13] <lerneaen_hydra> A-L-P-H-A, more or less
[12:23:45] <A-L-P-H-A> what was that other one that was popular... chevron cavalier?
[12:23:45] <lerneaen_hydra> too bad ergo proxy crapped out and lost all semblence of a good plot midway through
[12:23:47] <A-L-P-H-A> or something like that.
[12:23:55] <A-L-P-H-A> I haven't watched it yet...
[12:24:00] <lerneaen_hydra> oh
[12:24:01] <lerneaen_hydra> >.<
[12:24:14] <A-L-P-H-A> it's in a giant queue of eps.
[12:24:24] <A-L-P-H-A> I had honey to clover as well...
[12:24:34] <A-L-P-H-A> dunno what I did with it... it's either deleted or backedup
[12:24:39] <lerneaen_hydra> wasn't that some ecchi series?
[12:24:47] <anonimasu> no
[12:24:49] <A-L-P-H-A> huh?
[12:25:07] <lerneaen_hydra> oh, I'm mixing stuff up again
[12:25:13] <anonimasu> though it may be as one, on the review site..
[12:25:18] <anonimasu> but, not remotely..
[12:25:19] <anonimasu> :)
[12:25:23] <A-L-P-H-A> Kenshin... and neo genesis evangelon are still my favs.
[12:25:31] <A-L-P-H-A> full metal panic is fun too...
[12:25:36] <A-L-P-H-A> but that second second was milked... :(
[12:25:38] <lerneaen_hydra> neon genesis?
[12:25:42] <A-L-P-H-A> neon
[12:25:42] <lerneaen_hydra> bleh
[12:25:50] <anonimasu> hm, it's good..
[12:25:55] <anonimasu> too bad they ran out of production budget at the end
[12:26:00] <A-L-P-H-A> lerneaen_hydra... learn to like the story foo! :)
[12:26:07] <A-L-P-H-A> anonimasu, I heard the guy went crazy...
[12:26:16] <A-L-P-H-A> the writer... so it got fucked up near the end...
[12:26:20] <lerneaen_hydra> It wouldn't surprise me at all
[12:26:25] <anonimasu> yeah
[12:26:26] <A-L-P-H-A> after he got out... he finished it off proper.
[12:27:43] <anonimasu> yep
[12:28:56] <A-L-P-H-A> anonimasu... have you ever looked at an anagram of your nick?
[12:29:34] <anonimasu> no
[12:29:42] <A-L-P-H-A> http://wordsmith.org/anagram/anagram.cgi?anagram=anonimasu
[12:30:14] <anonimasu> lol
[12:30:22] <A-L-P-H-A> hehe... MI ON A ANUS
[12:30:46] <A-L-P-H-A> alex_joni, jail en ox
[12:31:05] <A-L-P-H-A> A-L-P-H-A, ha alp
[12:31:11] <lerneaen_hydra> HAHAHA, lerneaen hydra -> A DARN LEERY HEN
[12:31:19] <A-L-P-H-A> A-L-P-H-A, ah lap
[12:31:36] <A-L-P-H-A> A HANDLE YEN ERR
[12:32:03] <A-L-P-H-A> A RANDY REEL HEN
[12:32:13] <A-L-P-H-A> A HYENA LEND ERR
[12:32:38] <A-L-P-H-A> A HYENA LEND ERR
[12:32:48] <A-L-P-H-A> lots of anal stuff as well...
[12:32:59] <lerneaen_hydra> O.o
[12:33:15] <A-L-P-H-A> SWPadnos, pass down
[12:33:25] <A-L-P-H-A> was ponds
[12:33:37] <A-L-P-H-A> http://wordsmith.org/anagram/anagram.cgi?anagram=lerneaenhydra
[12:33:47] <A-L-P-H-A> LEARNED YARN EH
[12:34:12] <A-L-P-H-A> HARDEN ANY REEL
[12:34:16] <A-L-P-H-A> sooooooooooo many
[12:35:13] <anonimasu> * anonimasu slaps A-L-P-H-A
[12:35:21] <anonimasu> your nick is too stupid to have any sensible anagrams :D
[12:35:40] <anonimasu> Ah pal!
[12:35:46] <anonimasu> :S
[12:36:36] <alex_joni> alex_joni is now known as jail_oxen
[12:36:39] <jail_oxen> :-)
[12:36:42] <lerneaen_hydra> haha
[12:36:57] <jail_oxen> jail_oxen is now known as alex_joni
[12:37:12] <A-L-P-H-A> hahaha
[12:37:19] <A-L-P-H-A> :)
[12:41:02] <lerneaen_hydra> A-L-P-H-A, seen elfen lied?
[12:41:14] <A-L-P-H-A> someone mentioned it to me... but nope
[12:41:16] <A-L-P-H-A> don't know...
[12:41:28] <lerneaen_hydra> 's good
[12:41:29] <A-L-P-H-A> been more facinated chasing girls, then watching stuff.
[12:41:42] <lerneaen_hydra> haha, true true
[12:42:14] <A-L-P-H-A> haven't watched anime in a while... downloaded it, but not watched
[12:42:52] <anonimasu> A-L-P-H-A: yeah we've noticed
[12:43:12] <A-L-P-H-A> anonimasu. :) quit stalking me. :)
[12:43:20] <lerneaen_hydra> O_O
[12:49:04] <A-L-P-H-A> lerneaen_hydra, what's elfen lied about?
[12:49:13] <A-L-P-H-A> an elf that lied?
[12:49:42] <anonimasu> heh..
[12:49:45] <anonimasu> A-L-P-H-A: dont ask..
[12:49:49] <anonimasu> guro.. and stuff.
[12:49:59] <A-L-P-H-A> "guro"?
[12:50:05] <anonimasu> google pictures
[12:50:09] <anonimasu> for it.-.
[12:52:08] <anonimasu> ^_^
[12:52:16] <anonimasu> not my cup of tea.
[12:54:51] <A-L-P-H-A> hmm... sounds interesting... if it was like only 12 eps, to watch.
[12:54:57] <A-L-P-H-A> but not like 50 or something.
[12:55:08] <A-L-P-H-A> short term stuff... none of this narutu crap.
[12:55:18] <A-L-P-H-A> gad, I hate that anime.
[12:57:11] <fenn> A-L-P-H-A: want to waste like 10 hours of your life?
[12:57:19] <lerneaen_hydra> guro?
[12:57:42] <A-L-P-H-A> ooooooooooh. http://www.google.com/search?q=define%3A+guro&ie=utf-8&oe=utf-8&rls=org.mozilla:en-US:official&client=firefox-a
[12:58:08] <A-L-P-H-A> fenn, depends with what.
[12:58:10] <lerneaen_hydra> ... that is *not* what elfen lied is about
[12:58:27] <lerneaen_hydra> elfen lied is 12 eps or so
[12:58:30] <A-L-P-H-A> fenn, battlestar galactica has wasted more than 10 hours of my life... and pretty happy about it... it's such a good show.
[12:59:35] <fenn> A-L-P-H-A: well.. here is the first episode... http://www.youtube.com/watch?v=-goXKtd6cPo
[13:00:10] <A-L-P-H-A> lonely girl.................. ummmmmmmmm.
[13:00:57] <A-L-P-H-A> her voice is pretty annoying.
[13:02:37] <A-L-P-H-A> fenn, well... that's 3-4 minutes of my life I can't get back.
[13:08:24] <A-L-P-H-A> \bob?
[13:09:13] <marley> hi
[13:09:26] <marley> god evening
[13:10:21] <marley> good evening. haw to start "emc" to an file .ini
[13:18:12] <alex_joni> hi bob
[13:18:18] <alex_joni> where is the file?
[13:19:48] <jepler> morning guys!
[13:19:53] <alex_joni> hi jeff
[13:20:52] <marley> I want to start "emc" through icone without the menu of configurations.
[13:21:13] <alex_joni> marley: you need to link to "emc /path/to/your/inifile.ini"
[13:21:35] <alex_joni> for example (works on my machine) "emc /home/juve/emc2/configs/stepper/stepper_mm.ini"
[13:22:15] <alex_joni> I right clicked the Desktop and used create Launcher if I remembered correctly, then entered the line above as the executable, and I clicked the run-in-terminal checkbox
[13:22:57] <marley> grateful
[13:23:05] <alex_joni> dead
[13:23:25] <alex_joni> marley: http://en.wikipedia.org/wiki/Grateful_Dead
[13:23:28] <lerneaen_hydra> alex_joni, any reason for an end-user to run in terminal?
[13:23:30] <alex_joni> (just kidding)
[13:23:41] <alex_joni> lerneaen_hydra: it helps in the beginning till he gets it right
[13:23:48] <lerneaen_hydra> oh, that's true
[13:23:48] <alex_joni> and you can spot some errors, other than that .. no
[13:23:52] <marley> tanks you are a credit for one beer with mi...
[13:24:16] <alex_joni> I drink only virtual beer :D
[13:24:42] <marley> :b
[13:28:46] <alex_joni> skunkworks: good morning :D
[13:28:54] <alex_joni> logger_emc: bookmark
[13:28:54] <alex_joni> Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2007-01-18.txt
[13:29:10] <skunkworks> Good morning alex. Have not had time to play with the debs yet.
[13:29:23] <alex_joni> oh, ok.. just wanted to ask :D
[13:29:26] <skunkworks> * skunkworks would rather be.
[13:30:28] <skunkworks> I think I almost ran out of gas on the way to work this morning.
[13:31:10] <skunkworks> I looked down and the gauge was on empty - but no light. Then the thing started running funny. Going to have to see if we have some gas here at work.
[13:42:41] <Dallu1> Dallu1 is now known as Dallur
[13:47:31] <danex> Good Afternoon, alex_joni
[13:52:12] <paragon36> Hello All!
[13:52:48] <alex_joni> hi danex
[13:54:05] <danex> alex_joni: I am attempting the deb package install
[13:55:54] <danex> It is a little odd on my laptop
[13:56:07] <alex_joni> odd? how do you mean that?
[13:57:08] <danex> It is an issue with the docking station/harddrive
[13:57:14] <alex_joni> ahh.. ok
[13:57:50] <danex> The docking station has a harddrive in it/ this is where I put the Linux install
[13:58:40] <alex_joni> heh
[13:59:37] <danex> I just rebooted into the 2.1.o alpha
[14:00:55] <danex> No problems as yet
[14:01:39] <alex_joni> rebooted?
[14:02:00] <alex_joni> or you mean .. started emc2.1.0 alpha0 ?
[14:02:56] <danex> After I swaped the hard drive to the computer from the dock, A nice setup my laptop has
[14:03:22] <alex_joni> ahh.. ok.. because there shouldn't be a reason for the PC to reboot because of 2.1.0
[14:03:55] <danex> The harddrive from the dock will fit in the slot for the CD/DVD drive
[14:04:20] <danex> the effect is two harddrives in the laptop
[14:06:32] <danex> I can work with the HD in the dock but it must be in the computer to boot up in linux
[14:13:00] <danex> I open the sims, this looks good. When is the 2.10 to be released?
[14:16:19] <danex> I have to go, doctor's appointment, I will let you know how this works in the test
[14:24:36] <anonimasu> alex_joni: are you awake?
[14:24:40] <alex_joni> partly
[14:25:05] <anonimasu> what kind of amperage should you use for 1mm stainless
[14:25:10] <anonimasu> on a tig..
[14:25:32] <anonimasu> I know plasma is better ;) but.. I dont have one :)
[14:26:46] <alex_joni> 30-40Amps
[14:26:51] <alex_joni> what electrode?
[14:27:14] <alex_joni> 1.6 should be good.. do you have filler wire?
[14:27:16] <anonimasu> 1,6mm
[14:27:23] <anonimasu> yes
[14:27:28] <Dallur> if you are having burn through problems you can also try reverse polarity (if your electrode is big enough)
[14:27:48] <anonimasu> hm, dunno how you do that..
[14:27:51] <anonimasu> btw, while at it..
[14:28:01] <anonimasu> what's the MIG/MAG button..
[14:28:04] <anonimasu> or whatever it's called..
[14:28:16] <anonimasu> knob..
[14:28:40] <anonimasu> let me find you the name..
[14:36:23] <anonimasu> the symbol looks like
[14:36:24] <anonimasu> |
[14:36:25] <anonimasu> \
[14:42:27] <anonimasu> I think it's "MIG/MMA"
[14:43:24] <Dallur> MMA = stick welding , probably for flux core wire
[14:44:06] <Dallur> I would guess if you turn it to MMA the gas solenoid is turned off
[14:44:49] <anonimasu> hm, let me check what it is..
[14:45:20] <Dallur> MAG on the other hand means that there is oxygen present in the gas shield, which changes arc dynamics, only thing I can imagine such a button would do is to slightly change the current
[14:46:27] <Dallur> usually the oxygen is in the form of CO2 btw
[14:48:09] <anonimasu> mma/tig
[14:48:11] <anonimasu> arc force regulatino
[14:48:14] <anonimasu> regulation
[14:48:36] <Dallur> stick welding vs. tig welding
[14:49:02] <Dallur> just turns off the high freq. start (if it has it) and the gas solenoid
[14:52:17] <anonimasu> it's a knob.
[14:52:21] <anonimasu> theres another switch for it..
[14:52:42] <anonimasu> ah
[14:52:47] <anonimasu> arc force works only in MMA
[14:58:26] <alex_joni> anonimasu: sorry.. was away
[14:58:51] <alex_joni> anonimasu: reverse polarity is a good idea.. but you need to decrease current for that
[14:59:01] <alex_joni> or you'll ruin the tip :D
[14:59:16] <alex_joni> 1.6mm probably works poorly in reverse polarity
[14:59:32] <alex_joni> (reverse polarity- simply swap + & - at the machine)
[15:01:43] <anonimasu> that's not easy to do..
[15:01:59] <Dallur> there might be a toggle switch on the machine though
[15:02:01] <anonimasu> hm, yeah..
[15:02:02] <anonimasu> it is
[15:02:04] <alex_joni> anonimasu: you should have 2 plugs..
[15:02:06] <anonimasu> gas is separate..
[15:02:08] <anonimasu> :)
[15:02:15] <anonimasu> alex_joni: why is 1.6 poor at that?
[15:02:17] <alex_joni> yeah.. gas and signals are usually separate
[15:02:28] <alex_joni> you need bigger electrode for reverse polarity
[15:02:28] <anonimasu> ideally I'd have 1mm..
[15:02:31] <alex_joni> or it will melt
[15:02:32] <anonimasu> ah..
[15:02:36] <anonimasu> at 20a.. ;)
[15:02:39] <alex_joni> yes
[15:02:44] <anonimasu> hm.. ok
[15:02:50] <anonimasu> I wont try it then
[15:02:55] <alex_joni> at reverse polarity the electrons are travelling the other way around
[15:03:07] <anonimasu> wont the tungsten end up in the weld?
[15:03:11] <alex_joni> not melting the workpiece, travelling to the tip, but the other way around
[15:03:18] <alex_joni> unlikely :D
[15:03:41] <anonimasu> I'll just do it the usual way..
[15:03:44] <anonimasu> low amperage..
[15:03:49] <anonimasu> and patience
[15:03:49] <anonimasu> :)
[15:03:57] <alex_joni> ok.. start at lower currents, and increase untill you actually melt stuff
[15:03:58] <anonimasu> * anonimasu needs 1mm tungsten
[15:03:59] <anonimasu> yep
[15:04:01] <alex_joni> use thin filler
[15:04:10] <alex_joni> you need a good grinder for the tip
[15:04:12] <anonimasu> hm, I have some 0.04
[15:04:15] <alex_joni> to make it sharp
[15:04:18] <anonimasu> yeah..
[15:04:29] <anonimasu> I've got a valve grinder that i use..
[15:04:38] <alex_joni> then don't forget to grind the pointy thing down
[15:04:44] <anonimasu> ok?
[15:04:44] <anonimasu> why
[15:04:55] <alex_joni> pointy tip means too much current / square mm
[15:05:00] <alex_joni> overheats the tip
[15:05:04] <anonimasu> ah
[15:05:04] <anonimasu> ok
[15:05:07] <alex_joni> lasts less
[15:05:16] <alex_joni> ideally you grind at about 30 deg
[15:05:19] <alex_joni> from d to d/3
[15:05:30] <alex_joni> if you get my meaning :)
[15:05:34] <anonimasu> yeah
[15:09:32] <tomp> reverse polarity could be 2 contactors, one straight thru, other crossed, turn on 1 or other thru hal
[15:09:48] <alex_joni> this is not emc2 related :P
[15:11:53] <lerneaen_hydra> alex_joni, all the more reason to take it to #emc-devel
[15:12:41] <alex_joni> :P
[15:21:43] <alex_joni> whoa.. seen the new scl600?
[15:22:41] <alex_joni> http://www.300cforums.com/forums/off-topic-discussion/20683-mercedes-benz-scl600.html
[15:23:34] <lerneaen_hydra> most everyone here runs *nix, right?
[15:24:03] <Dallur> brb
[15:24:13] <jepler> lerneaen_hydra: seems like a fair guess, since the channel is about a unix-only piece of software
[15:24:38] <alex_joni> * alex_joni ran it on doze a while ago
[15:24:40] <lerneaen_hydra> jepler, hmm, that wouldn't stop them from running something else on their main workstation
[15:24:46] <lerneaen_hydra> as I did before
[15:25:24] <alex_joni> lerneaen_hydra: you were going somewhere with that question?
[15:25:31] <lerneaen_hydra> no, not really
[15:25:43] <lerneaen_hydra> how so?
[15:26:00] <alex_joni> just asking
[15:26:31] <alex_joni> * alex_joni loves http://www.networkupstools.org/
[15:28:03] <alex_joni> bbl
[16:11:10] <awallin> lerman: I found a few more papers on cnc simulation, let me know if you are interested
[16:22:41] <alex_joni> awallin: how about putting them on a wiki page?
[16:26:20] <awallin> alex_joni: sure, that will work, but the links are mostly to the abstracts only, downloading requires subscription
[16:28:57] <alex_joni> sure they do :)
[16:35:03] <lerman> awallin: yup. I'd be interested in looking at them. You still have my email address?
[16:39:17] <awallin> lerman: the same I used three days ago, I have that in my outbox
[16:39:33] <awallin> there are a couple of them, so hope you have 5-10Mb free in your inbox
[16:40:03] <lerman> Yup. I host my own mail server. So, I have a few gigs free.
[16:43:45] <awallin> I've sent you three emails now
[16:44:03] <awallin> it seems that the 'matchstick' model is the most commonly used simple model
[16:44:29] <awallin> good for 3-axis only, but that is mostly what people use anyway (how many EMC'ers run 5-axis machines?)
[16:45:24] <awallin> Julian T of freesteel suggest an adaptive approach: if the z-value between two points changes too much, insert more points in that area, run the last tool motion again, and again check if neighboring z-values are within a set tolerance
[16:47:26] <paragon36> Has anyone come across any sites showing how to use pc atx / at power supplies piggied back together to produce higher volts (30V) at around 15A?
[16:49:40] <awallin> paragon36: I think that was described on cnczone's electronics forum
[16:50:39] <awallin> lerman: did you receive the emails?
[16:51:19] <lerman> Got them. Haven't started reading yet. Thanks.
[16:53:19] <awallin> if anyone has better acces to journals than me I'd be interested in http://dx.doi.org/10.1016/0010-4485(83)90082-9
[16:53:45] <awallin> our library's subscription doesn't go that far back, it's from 1983...
[16:55:37] <awallin> maybe I'll ask that on emc-users
[17:03:59] <tomp> about publications... yes if anyone is ieee member, they might be able to get at some publications the rest of list cannot get to
[17:06:56] <anonimasu> agreed
[17:07:58] <SWPadnos> paragon36, don't do that!
[17:08:01] <paragon36> Thanks for the power supply info :-)
[17:08:33] <SWPadnos> paragon36, you have to be very careful if you want to try using stacked PC power supplies, since they invariably have their cases tied to ground
[17:13:12] <paragon36> Hello SWPadnos, I have seen an article that shows how to issolate there case etc but I can't seem to find it at the moment.
[17:13:35] <SWPadnos> yeah - lots of cardboard / plexiglass plus electrical tape, probably :)
[17:13:54] <paragon36> Probably lol
[17:15:34] <lerneaen_hydra> SWPadnos, duct tape
[17:15:53] <SWPadnos> duct tape is sometimes conductive (has metal wires in it)
[17:15:55] <lerneaen_hydra> duct tape + sledgehammer = solution to most problems
[17:16:31] <lerneaen_hydra> and if it doesn't work, apply more duct tape and/or get a larger sledgehammer
[17:16:44] <paragon36> Forgot the wearing of Wellington Boots there Lerneaen!
[17:16:54] <SWPadnos> yep - rule #1 of the "brute force" approach: if it isn't working, you're not using enough of it
[17:17:02] <lerneaen_hydra> exactly ;)
[17:34:56] <maddash> I'm getting a funky error when I try executing /etc/init.d/rtai-dev : ": invalid option" -- I copied and paste the script from the wiki (http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Debian_Etch_Compile_RTAI#etc_init_d_rtai_dev) Any clues?
[17:36:18] <cradek> I suspect a copy-paste error
[17:36:32] <cradek> run it as sh -x /etc/init.d/rtai-dev and see what line is wrong
[17:37:10] <SWPadnos> does that need to be run sudo or suid?
[17:38:00] <maddash> that's odd - returns an error on line 7
[17:38:20] <maddash> funny, b/c the script worked on my previous compile of rtai
[17:38:53] <maddash> is it just me, or is line 7 = "for n in `seq 0 9`; do" ?
[17:39:38] <SWPadnos> it looks that way to me
[17:39:38] <jepler> $ for n in `seq 0 9`; do echo -n "$n "; done; echo
[17:39:38] <jepler> 0 1 2 3 4 5 6 7 8 9
[17:39:46] <jepler> this works fine in my shell
[17:41:42] <maddash> uh oh - can anyone else confirm this? it could mean that something's wrong with my rtai kernel
[17:42:20] <SWPadnos> confirm that the sequence works?
[17:43:07] <maddash> that, or the script from the wiki verbatim
[17:43:23] <SWPadnos> the script has nothing to do with the kernel - it's a shell thing
[17:43:40] <cradek> pastebin your script.
[17:43:50] <SWPadnos> is sh linked to something (like dash, bash, csh, ksh ...)?
[17:44:00] <jepler> $ sh rtai-dev
[17:44:00] <jepler> mknod: `/dev/rtai_shm': Operation not permitted
[17:44:00] <jepler> mknod: `/dev/rtf0': Operation not permitted
[17:44:02] <jepler> .. etch ..
[17:44:03] <jepler> (sh is bash)
[17:44:22] <SWPadnos> don't you need to be root to mknod?
[17:44:32] <jepler> SWPadnos: yes that's why I got that error
[17:44:32] <SWPadnos> (in /dev)
[17:44:34] <SWPadnos> ok :)
[17:44:39] <jepler> but all the lines of the shell script were accepted by sh
[17:45:34] <jepler> $ ./rtai-dev
[17:45:35] <jepler> /bin/sh: - : invalid option
[17:45:52] <jepler> make sure the file is a proper Unix text file and that there is no whitespace after -e on the very first line
[17:46:33] <jepler> when I used my editor to save the script in text mode, I got the error I pasted above
[17:46:34] <SWPadnos> it would be really funny if it turns out to be a CR/LF error
[17:46:40] <jepler> er, in DOS mode
[17:47:29] <maddash> http://pastebin.com/862189
[17:48:13] <SWPadnos> argh. can you repost that to http://pastebin.ca ? (it tends to work better)
[17:48:59] <maddash> dillo freaks out at *.ca
[17:49:07] <maddash> hold on, let me install firefox
[17:49:16] <SWPadnos> so you can't go to any canadian websites?
[17:49:52] <SWPadnos> it finally loaded
[17:49:59] <maddash> i know, such racist (jingoistic?) software, huh?
[17:51:44] <maddash> http://pastebin.ca/320625
[17:54:03] <CIA-8> 03jepler 07TRUNK * 10emc2/src/rtapi/ (rtai_rtapi.c rtl_rtapi.c): enlarge message buffers so that the 'unobtainable step rate' message fits (backport candidate)
[17:54:22] <CIA-8> 03jepler 07TRUNK * 10emc2/src/hal/components/stepgen.c: improve unobtainable step rate message (backport candidate)
[17:58:01] <awalli1> .
[17:58:45] <Dallur> ..
[18:00:01] <maddash> ...I just realized that this could be a problem with $PATH
[18:00:17] <lerman> awallin: I just read the Choi paper. (Without following all of the details.) It seems like a reasonably doable technique for modeling the cutting of a mold. Unfortunately, it does not handle undercuts -- you can't have them in molds.
[18:01:40] <lerman> I suspect that will be a problem with a lot of the simulation techniques.
[18:02:32] <lerman> Automobile makers don't CNC parts; they CNC molds to make parts. The auto makers do a lot of the research -- they have lots of money.
[18:03:11] <lerman> We need papers from the airplane manufacturers. They can afford to CNC individual parts that might have undercuts.
[18:03:38] <lerman> And also have the money to spend on fancy algorithms.
[18:04:39] <SWPadnos> this is a place where a NURBS modeler with boolean functions would come in handy (as long as it's fast :) )
[18:05:53] <SWPadnos> except that actual 3D motion isn't so easy with odd things like slot or keyway cutters
[18:06:11] <SWPadnos> motion parallel to the XY plane is pretty easy for even those though
[18:06:51] <awalli1> lerma: a 3-axis simulation that doesn't support undercuts would be a start...
[18:07:28] <awalli1> I can try throwing something together in matlab during the weekend. I'll use matlab since it has the graphics part "for free"
[18:08:50] <awalli1> bbl
[18:11:10] <maddash> "undercut"?
[18:11:38] <SWPadnos> if you have something like a keyway cutter, it can cut a horizontal slot in the material
[18:13:20] <maddash> by "undercut," are you referring to one-sided contour approximation?
[18:13:38] <SWPadnos> I don't think so, but I'm not sure whaty you mean
[18:14:26] <SWPadnos> consider cutting with a bit like this: http://www.rockler.com/findit.cfm?page=1572&cookietest=1
[18:14:33] <lerneaen_hydra> I assume an undercut is something that is not representable with a simple heightmap
[18:14:49] <SWPadnos> currect, hence the issues with the "matchstick" approach
[18:14:54] <SWPadnos> err - correct
[18:15:00] <lerman> An undercut exists where a line parallel to the z-axis intersects the surface in multiple places (excluding the bottom).
[18:15:04] <lerneaen_hydra> matchstick?
[18:15:12] <SWPadnos> matchstick = height map :)
[18:15:22] <SWPadnos> like a bunch of sticks sticking up from the table ...
[18:15:31] <lerman> I've found references to 'dexel' representation. Anyone familiar with that?
[18:15:31] <lerneaen_hydra> oh
[18:15:41] <maddash> so, basically a concave surf
[18:15:52] <lerneaen_hydra> what's wrong with an STL-like surface? hard to work with?
[18:16:03] <SWPadnos> hard to generate on the fly, I'd bet
[18:16:31] <SWPadnos> lerman, I saw a dexel reference as well, but no definition of the term
[18:16:46] <lerman> NC milling simulation and dimensional verification via dexel representation (US patent #5,710,709) The source code seems to be available.
[18:17:47] <SWPadnos> ah - "depth element"
[18:17:51] <lerneaen_hydra> has the patent expired?
[18:17:55] <SWPadnos> gah - friggin IEEE
[18:18:02] <lerneaen_hydra> or maybe not applicable to OSS?
[18:20:10] <lerman> Looks like a linked list at each pixel.
[18:21:42] <SWPadnos> have you seen this one: http://findarticles.com/p/articles/mi_qa3685/is_199501/ai_n8724848/print
[18:21:57] <SWPadnos> lots of references in the bibliography
[18:25:39] <SWPadnos> have you seen the textbook "Computer Graphics Principles and Practice" (by Foley, Van Dam, Feiner, Hughes) ?
[18:26:07] <SWPadnos> http://www.amazon.com/Computer-Graphics-Principles-Practice-2nd/dp/0201848406
[18:26:11] <lerman> The patent has a Jan20, 1998 date; but you really have to look at the claims and the references. There might be an earlier dexel patent which has expired.
[18:26:14] <lerneaen_hydra> there isn't by any chance an OSS parametric modeling app?
[18:26:20] <SWPadnos> brl-cad
[18:26:32] <lerman> I've seen the orignal Foley and Van Dam.
[18:26:42] <SWPadnos> ok. chapter 12 is relevant to this topic
[18:27:01] <lerman> (Actually, I've met Foley.)
[18:27:15] <lerneaen_hydra> it seems to me like all moves in a mill can be easily represented in a parametric modeler
[18:27:17] <SWPadnos> col
[18:27:19] <SWPadnos> cool
[18:27:43] <SWPadnos> lerneaen_hydra, they can be, but the idea is to show material being removed from the stock as the machine is cutting
[18:27:48] <SWPadnos> at least, that's one goal :)
[18:27:56] <lerneaen_hydra> oh
[18:27:57] <lerman> Yes. But one of the papers suggests that the CSG approach runs in N^4 where N is the number of moves.
[18:28:13] <lerneaen_hydra> hmm, you could still do that, if you have a moving cut extrude
[18:28:54] <lerneaen_hydra> afaict everything you use in a mill can be described as a sketch with a certain cut-extrude depth
[18:29:36] <lerneaen_hydra> and if you sweep that along with a sketch that will keep the old material away then you'll be able to do any type of mill, 3 axis as well as 5 axis
[18:29:40] <lerneaen_hydra> lathe too
[18:29:44] <lerman> I'm (easily) convinced that the dexel model would work. Not much more cost than the simple depth model for the case where there are few undercuts. At each X, Y, store either the depth or a link to the list of dexels.
[18:29:45] <lerneaen_hydra> edm too
[18:29:53] <tomp> offCurrentTopic: a yahoo group 'multimachine' builds machines from engine blocks ( to get accurate stabile bores for spindle,overarm, bed)
[18:29:57] <SWPadnos> you need a circular sweep at each end, plus two sweeps of the cutter profile
[18:30:12] <lerneaen_hydra> yeah
[18:30:58] <SWPadnos> lerman, hmmm. kind of like a vertical run-length-encoded set of heights
[18:31:07] <SWPadnos> that could work
[18:31:12] <lerman> Yup.
[18:31:25] <lerman> Seems pretty cheap to implement.
[18:32:05] <lerman> Need memory that is proportional to the surface area (more or less).
[18:32:15] <SWPadnos> other than the memory issues it represents - many allocations and non-uniform arrays
[18:33:08] <lerman> Linked lists are straightforward.
[18:33:56] <SWPadnos> yes, but many thousands of mallocs will likely bog down
[18:34:01] <lerman> Depending on the allocator, allocation can be pretty cheap. I've written one that is very fast.
[18:34:21] <SWPadnos> since the need for a new type at a given point will always leave a smaller space unused ...
[18:34:56] <lerman> Malloc a large block that is then broken into smaller fixed size nodes for the lists.
[18:34:59] <SWPadnos> sure. I guess there can also be a pool of 1-, 2-, and 3-height structs, and anything higher gets malleced later
[18:35:35] <lerman> The big gotcha is the threaded holes. Each point on the flank of the thread is an area of undercut.
[18:35:45] <SWPadnos> yep :)
[18:36:57] <lerman> The single height is represented by a constant. Anything else gets a pointer to a linked list of dexels. Since most points do not have undercuts, that's a big savings.
[18:38:16] <SWPadnos> one extension: have an explicit "bottom height" instead of assuming it's zero
[18:38:34] <lerneaen_hydra> but do you really need to simulate threaded holes?
[18:38:52] <lerneaen_hydra> that takes lots of power for something IMO just aesthetic
[18:38:53] <SWPadnos> that actually gets rid of some types of undercut, and allows you to flip a part to show multiple processes in the same model
[18:39:05] <lerman> You would want to if you are milling (rather than tapping) threads.
[18:39:26] <SWPadnos> yeah - a tap is a pretty hard cutter to profile anyway
[18:39:35] <lerneaen_hydra> hmm, sounds difficult
[18:39:39] <lerneaen_hydra> you'll need to model the cutter too
[18:40:19] <SWPadnos> the assumption for modeling cutters (without doing full physics simulation) is that the spindle speed is much higher than the vertical travel, so you get a complete sweep of the entire profile
[18:40:25] <lerman> Yes. You will need to model the cutter. I would use the profile of a 'round' cutter.
[18:40:40] <lerman> That's what I (meant to) said.
[18:40:53] <SWPadnos> "cylindrical" :)
[18:41:06] <lerneaen_hydra> SWPadnos, how accurate do you plan to make this sim?
[18:41:15] <SWPadnos> I plan to have lerman make it very accurate ;)
[18:41:20] <lerneaen_hydra> haha
[18:41:38] <lerman> So... No ridgid tapping.
[18:41:42] <SWPadnos> heh
[18:42:21] <SWPadnos> it looks so easy in the wiki: http://en.wikipedia.org/wiki/Constructive_solid_geometry
[18:43:46] <lerman> It will be 100% accurate. To the specified *precision*.
[18:44:53] <lerneaen_hydra> well, no, not if you gloss over things like cutter deflection, machine stability, cutter position, gravitational pull of machine operator ;)
[18:45:17] <maddash> do any of you live in NY?
[18:45:23] <SWPadnos> I'm sure those will be defined as "outside the specified precision" ;)
[18:45:24] <maddash> NYC, USA , that is...
[18:45:34] <SWPadnos> nope
[18:45:47] <SWPadnos> I don't think anyone who's here frequenctly is from NYC
[18:45:55] <SWPadnos> frequently
[18:46:23] <lerman> I'm from Connecticut. Close enough to NYC. (And far enough) :-)
[18:46:26] <maddash> haha.
[18:46:28] <anonimasu> hello
[18:46:37] <SWPadnos> Vermont here -also close and far enough ;)
[18:46:44] <maddash> freakin' lame-ass traffic officers
[18:46:57] <lerman> The simulation will be of an 'ideal' machine.
[18:47:08] <SWPadnos> it's the parking thieves^W officers I can't stand
[18:52:33] <maddash> brb going out to vent...
[19:01:20] <lerman> maddash: why do you ask about NYC?
[19:02:31] <SWPadnos> I gather it had something to do with "freakin' lame-ass traffic officers"
[19:02:59] <lerman> Ah... That explains the need to vent.
[19:05:15] <lerman> If I ever get around to it, the simulation will probably use the dexel model. Initial version would be orthogonal XYZ axis.
[19:06:17] <lerman> I don't know that I'm qualified to do it, but with the help of jepler and cradek, it would be nice to be able to have it as an alternative display pane in AXIS.
[19:06:31] <SWPadnos> single or dual height-maps are probably sufficient for the part representation
[19:06:56] <lerman> Not for my milled threads, though.
[19:06:59] <SWPadnos> yeah - there's a second step we haven't really discussed - converting the internal representation into something displayable
[19:07:32] <lerman> Probably some sort of zbuf algorithm.
[19:08:24] <lerman> I'd want to be able to rotate the part to various orientations.
[19:08:33] <maddash> for the curious (lerman, SWPadnos): some traffic officer who had nothing better to do than patrol the empty industrial sector just wrote my father a ticket for (direct quote here) "blocking the sidewalk" by momentarily leaving his car on *his* clearly marked driveway.
[19:09:37] <lerman> Was that in NYC or Yonkers? I understand there is some sort of job action going on in Yonkers.
[19:09:53] <maddash> nyc. it's just as bad over here.
[19:11:25] <jepler> lerman: it would not be technically difficult to, say, draw a zbuffered triangle mesh in the AXIS plot window. however, the whole plot window is redrawn whenever the machine moves, so it also has to be fairly fast or the user will complain,.
[19:12:43] <jepler> even a "modest" 20000 triangles (from a 100x100 heightmap) will probably be slow with only software opengl, especially if you're doing software step generation
[19:13:16] <lerman> That wouldn't work. The 3D objects would have to be stored in graphic memory and only those that changed would be updated (I think).
[19:13:34] <maddash> why not utilize hardware acceleration?
[19:14:24] <lerman> Yes. That was the intent of my graphic memory comment. The hardware would use the updated display list.
[19:15:00] <lerman> Something would have to convert the dexel representation to efficient display list objects.
[19:15:04] <jepler> maddash: many people report "Unexpected realtime delay" errors while using the nvidia drivers. We recommend not using those cards.
[19:15:50] <lerman> Does current display hardware accept 3D objects to display?
[19:16:36] <lerman> Or does the model have to be converted to triangles or whatever by software outside of the display system?
[19:16:37] <jepler> if by "3D objects" you mean "triangles", yes.
[19:17:02] <lerman> What culls the invisible triangles?
[19:17:11] <cradek> the zbuffer
[19:17:53] <lerman> Triangles are two dimensional; not three D. So, the display system takes triangles representing the surface(s) and displays them.
[19:18:01] <jepler> opengl has lots of ways to control which pixels (if any) of a triangle are actually drawn to the framebuffer.
[19:18:03] <jepler> uh
[19:18:10] <jepler> triangles can have any dimension
[19:18:38] <cradek> the triangles are in 3d space. typically the zbuffer shows you the "front" triangles
[19:18:53] <lerman> No... three points determine a plane -- a 2D object. But yes, the triangles ARE in 3 space.
[19:19:09] <jepler> opengl usually uses homogeneous coordinates, so a triangle is defined by p0 = <x0,y0,z0,1>, p1, and p2
[19:19:27] <lerman> And a set of triangles can define a 3D surface.
[19:19:42] <cradek> what's your question again?
[19:20:16] <jepler> the triangle is transformed according to some matrices. then there's an optional "facing" test, then "zbuffer", "stencil" and other tests on the individual "fragments" (pixels) of the triangle. if all the desired tests succeed, then that fragment of the triangle can affect the color buffer
[19:20:26] <jepler> I'm leaving out a lot of details here
[19:21:03] <lerman> At any rate, the problem is to convert the dexel representation to a set of triangles defining the surface. Then update the dexel representation as each move takes place AND update the set of triangles corresponding to the dexel representation on the fly.
[19:21:38] <tomp> jepler: whats the 4th data in the point def "<x0,y0,z0,1>"
[19:23:31] <awalli1> lerma: by dexel, do you mean one heightmap vector, or something more fancy?
[19:25:32] <lerman> awallin: damn you've got to stay with the program. I haven't been able to get access to the papers, but essential, at each point in X,Y, you record a sequence of depths representing the Z distance to EACH surface (in the case where there are undercuts).
[19:25:33] <jepler> tomp: that extra "1" is tacked on to all coordinates in opengl (and usually need not be given explicitly) because transformation matricies "work better" (I think it allows them to express translations and the perspective projection) when coordinates are 4x1 and matrices are 4x4.
[19:25:54] <tomp> k
[19:26:23] <lerman> tomp: so called "homogenous" (I think) coordinate system. Needed for perspective rendering.
[19:27:06] <jepler> tomp: http://www.glprogramming.com/red/appendixf.html if you're a mathematician
[19:27:18] <maddash> so in summation, the problem at hand is the lack of a 3d rendering system that is fast enough for RT...
[19:28:35] <maddash> wireframe?
[19:29:02] <awalli1> I don't think RT would be a requirement for a simulation
[19:29:09] <tomp> or http://www.ipb.uni-bonn.de/DAGM/Tutorials2001/Session-1/img60.htm not necc to ddefine, point, but used for this purpose...
[19:29:17] <lerman> No. I don't think that's a real problem. It might be that some of the video boards interfere with the RT stuff. I, personally, view the simulation stuff as something that would be used with EMC in simulation mode. If you are running real hardware, just look at the material you are cutting.
[19:30:05] <lerman> tomp: Why trade a headache for an upset stomach? (When you can have both.) :-)
[19:30:24] <tomp> lol
[19:30:32] <Dallur> IMHO, I think the simulation is most useful when you are creating the g-code, not when running the g-code
[19:32:08] <awalli1> lerman: if you send me links to articles I can see what I have access to from here
[19:32:26] <maddash> so what's the problem, then?
[19:32:35] <cradek> the naive conversion from height map to triangle strips is pretty trivial - getting the height map (and picking its size/resolution) is the hard part I think
[19:32:56] <awalli1> maddash: we're just brainstorming what stock model to use and how to implement it...
[19:33:18] <lerman> google: dexel representation
[19:33:20] <jepler> maddash: the problem is that the nvidia driver DOESN'T WORK WITH REALTIME and no one outside of nvidia knows why
[19:33:28] <awalli1> cradek: you are probably right...
[19:33:38] <jepler> maddash: if you use the nvidia closed source driver with accelerated opengl YOU WILL GET THE "UNEXPECTED REALTIME DELAY" message
[19:34:07] <jepler> maddash: at that point it's easier to say "oh fuck it" and put in a working card (e.g., matrox) and forget all about it
[19:34:26] <cradek> yeah, that's what I did (matrox)
[19:35:28] <jepler> this post seems to explain that nvidia is doing some deeply gross things, but it's all over my head: http://www.nvnews.net/vbulletin/showthread.php?t=67430
[19:35:37] <jepler> e.g.,
[19:35:38] <jepler> e.g., During my experiments, I figured out that the nvidia driver uses
[19:35:38] <jepler> semaphores in interrupt and tasklet context, but sleeping functions
[19:35:38] <jepler> are basically strictly forbidden in those code areas!
[19:35:40] <jepler> eek
[19:37:26] <lerman> awalli 1: http://ieeexplore.ieee.org/search search for dexl give lots of papers (that I can't access).
[19:37:37] <lerman> That's dexel.
[19:38:43] <lerman> See:
[19:38:44] <lerman> 4. A fast and efficient projection-based approach for surface reconstruction
[19:38:46] <lerman> Gopi, M.; Krishnan, S.;
[19:38:48] <lerman> Computer Graphics and Image Processing, 2002. Proceedings. XV Brazilian Symposium on
[19:38:49] <lerman> 7-10 Oct. 2002 Page(s):179 - 186
[19:38:51] <lerman> Digital Object Identifier 10.1109/SIBGRA.2002.1167141
[19:38:52] <lerman> Summary: We present a fast and memory efficient algorithm that generates a manifold triangular mesh S with or without boundary passing through a set of unorganized points P/spl sub//spl Rscr//sup 3/ with no other additional information. Nothing is assumed abo.....
[19:38:53] <lerman> AbstractPlus | Full Text: PDF(840 KB) IEEE CNF
[19:38:56] <lerman> Rights and Permissions
[19:39:12] <awalli1> lerman: wait a minute, I need to switch on the VPN...
[19:40:08] <awallin> ok, now I'm running VPN...
[19:40:23] <tomp> awallin: maybe you can access ieee stuff, for example depth buffering display techniques... http://ieeexplore.ieee.org/Xplore/login.jsp?url=/iel5/38/4056977/04056985.pdf?tp=&arnumber=4056985&isnumber=4056977
[19:40:59] <tomp> oh you guys already said that while i typed :|
[19:41:08] <SWPadnos> um - is VPN really useful in this context?
[19:42:12] <awallin> SWPadnos: with VPN my laptop, now connected to my ISP at home, aquires an IP address from work (univ. helsinki) and I get access to whatever the library at univ. helsinki has subscribed (which is pretty much)
[19:42:28] <SWPadnos> ah - now I get it :)
[19:42:37] <awallin> tomp: that paper is by ROssignac and Requicha?
[19:42:56] <awallin> oh... 10Mb
[19:43:33] <Jymmmm> SWPadnos lo
[19:43:33] <tomp> yep, and maybe stuff from ccrp would help, sounds like you can get in there too
[19:43:43] <SWPadnos> hi Jymmmm
[19:44:16] <tomp> and inria ( the french are good at math )
[19:48:49] <awallin> hmm 'dexel' gives only three hits on ieeexplore
[19:50:32] <lerman> I got 14 hits.
[19:50:36] <tomp> http://www.math.tau.ac.il/~danha/papers/nc_cad04.pdf dexels in multi axis machining
[19:51:15] <lerman> Whoops. I've got to get to work. BBL
[19:55:22] <ds3> if nvidia is not recommended, what is a good 3D card to use (wrt to the RT properties/behavior)?
[19:56:04] <cradek> there isn't one that we know of - seems like it's best to use software rendering
[19:56:13] <SWPadnos> if by "good 3D" you mean "a card that has modern 3D rendering performance", I'm not sure there is one that's really good for RT
[19:56:33] <ds3> okay, just thought I'd ask given the comments
[19:57:05] <cradek> I've had a couple ATI cards that work with the free "direct rendering" code and don't mess up realtime - my laptop is one of these
[19:57:19] <cradek> but modern performance? nope
[19:57:26] <cradek> it's old hardware
[19:57:56] <jepler> software opengl is not bad if the program using it is not written with the assumption that you have a 1 gigapixel/second fill rate -- axis is opengl, and it performs "well enough" on sub GHz machines and a software renderer.
[19:58:22] <jepler> (though "well enough" may mean 5-10fps)
[19:58:45] <cradek> yep, it works fine, but if we add polygons, it won't
[19:58:46] <lerneaen_hydra> good enough for something like this though IMO
[19:59:04] <lerneaen_hydra> btw, would there be anything that would make AXIS SIM and hardware 3d not work?
[19:59:13] <cradek> no
[19:59:26] <SWPadnos> that works great for me :
[19:59:28] <SWPadnos> :)
[19:59:44] <cradek> but you're talking two separate machines then - most people aren't going to have that.
[19:59:46] <SWPadnos> in fact, AXIS RT in a VM on good hardware works pretty darned well too
[19:59:56] <lerneaen_hydra> cradek, yeah, two machines
[20:00:03] <SWPadnos> (but I do get "Unexpected Realtime Delays" :) )
[20:00:06] <ds3> tangential question - any plans to turn the sim mode into a G code verifier (clean up the UI and other minor tweaks)?
[20:00:13] <lerneaen_hydra> how utterly unexpected
[20:00:15] <lerneaen_hydra> :p
[20:00:42] <jepler> ds3: what do you mean by "verifier"? With emc2.1, you can run axis on any old linux machine (the SIM mode that lerneaen_hydra was talking about above) and preview g-code.
[20:00:59] <SWPadnos> or get an error if the G-code is invalid ...
[20:01:56] <lerneaen_hydra> jepler, are there any 2.1 debs in the rep yet by any chance? (for a pure SIM based install)
[20:01:59] <jepler> if your sim ini file has the same machine constraints as your real machine you can even preview the exact machine path that you will get due to blending
[20:02:08] <ds3> jepler: something that would maybe let me define a stock size for one; prehaps ability to save the images to an STL file for comparism; and a tightly integrated editor
[20:02:58] <lerneaen_hydra> * lerneaen_hydra que integrated editor debate ;)
[20:03:01] <jepler> ds3: no, axis is none of those things
[20:03:10] <awallin> ds3: that would be nice, let's start with the cutter simulator...
[20:03:26] <anonimasu> hm..
[20:03:50] <ds3> awallin: 2 minor problems - I donno much about doing UI/userland stuff
[20:03:55] <awallin> jepler: could such a program use the EMC interpreter without too much trouble?
[20:03:58] <jepler> ds3: this sounds more like what lerman is digging into research to do
[20:03:57] <anonimasu> awallin: how much code do you have so far?
[20:04:21] <awallin> anonimasu: nothing yet, just reading for a few evenings
[20:04:23] <anonimasu> I know some opengl, drawing voxels/similar/whatever lerman should be simple..
[20:04:32] <anonimasu> lerman said..
[20:04:36] <ds3> jepler: It would be nice to be able to do the equiv of what the CAM systems can do except this will verify the output of the post processor
[20:04:42] <jepler> awallin: sure, the emc interpreter is available as a shared library, and during each release cycle (e.g., in all the 2.1.x versions) the API is the same
[20:04:51] <jepler> awallin: but the API isn't documented by anything but the source code
[20:05:28] <awallin> jepler: so, a seprate program in python or C or C++ can use the interpreter without having to "boot" the whole of emc
[20:05:35] <jepler> awallin: yep
[20:05:46] <awallin> that's good!
[20:06:03] <anonimasu> hm
[20:06:12] <ds3> also, is there something half way between --enable-run-in-place and normal compiles? (prehaps something that takes a EMC_ROOT enviroment and references everything off that)?
[20:06:41] <awallin> so now I have about 18Mb of pdfs, is anyone else than lerman interested?
[20:06:45] <awallin> that's 12 papers in total
[20:06:46] <anonimasu> awallin: I am
[20:07:09] <awallin> ok, I'll put these somewhere on the web
[20:07:28] <anonimasu> awallin: I think a good way to test this is to write the "matchstick" class first..
[20:07:37] <anonimasu> in 2d.
[20:08:13] <anonimasu> and write the subdivision(to add more detail)
[20:08:17] <anonimasu> and add the 3rd dimension
[20:08:34] <awallin> anonimasu: one thing to figure out, if we want adaptive accuracy, is how to easily insert more matchsticks at selected locations
[20:08:39] <anonimasu> then work on the yeah..
[20:08:47] <anonimasu> that was the idea of doing in 3d first..
[20:08:54] <anonimasu> 2d..
[20:09:41] <anonimasu> awallin: matchsticks are similar to voxels right?
[20:09:49] <lerneaen_hydra> um, am I just silly or why not "just" do 2d based sim as a heightmap? (greyscale image, variable resolution)
[20:10:03] <lerneaen_hydra> by 2d I mean no undercutting
[20:10:12] <lerneaen_hydra> so most 3-axis mill operations
[20:10:14] <anonimasu> yeah, for getting the adaptive accuracy working..
[20:10:22] <SWPadnos> one suggestion (made earlier, but bears repeating): have a "top" and "bottom" value in each matchstick struct. this allows a simple operation to "flip" the part, and visualize multiple milling setups (as longas the Z axis is parallel for all setups)
[20:10:31] <anonimasu> SWPadnos: yeah
[20:10:36] <SWPadnos> rather than assuming a bottom of 0
[20:10:38] <anonimasu> SWPadnos: you can do rotation on matrices.. also..
[20:10:43] <anonimasu> to offset your workpiece..
[20:10:56] <anonimasu> \
[20:11:11] <anonimasu> SWPadnos: that's a voxel :)
[20:11:16] <lerneaen_hydra> the hieghtmap based implementation would be quite similar to jepler's image to gcode, except in the opposite direction (gcode -> image)
[20:11:31] <anonimasu> http://en.wikipedia.org/wiki/Voxel
[20:11:36] <SWPadnos> well - rotating the array is one thing - I'm thinking of machinging a motor mount (for instance), which has one program for the top and another program for the bottom
[20:12:06] <anonimasu> yeah, but that wouldnt be related..
[20:12:10] <anonimasu> or would it?
[20:12:16] <SWPadnos> yes it would
[20:12:22] <anonimasu> well, then flipping would be possible..
[20:12:32] <SWPadnos> the matchstick array can be written to disk. it's a trivial operation to "flip" it over
[20:12:35] <SWPadnos> exactly
[20:12:37] <anonimasu> yeah..
[20:12:44] <SWPadnos> then you can see the results of multiple (parallel Z) operations
[20:12:47] <awallin> on an apache server that doesn't show directory listings by default, can I enable the directory listing for one dir somehow?
[20:12:55] <SWPadnos> .htaccess
[20:12:56] <anonimasu> awallin: yes
[20:13:12] <SWPadnos> something like allowDirectoryListing Yes
[20:13:14] <anonimasu> I'd be happy to start hacking on a matchstick class..
[20:13:19] <anonimasu> if we can define what we want..
[20:13:28] <jepler> awallin: the interpreter has C++ and Python interfaces. This program shows all the python-level "canon" calls made while running the ngc file named on the commandline: http://emergent.unpy.net/files/sandbox/gdump.py
[20:14:37] <awallin> jepler: looks very simple, thanks, I'm not there yet, but I'll ask you later again:)
[20:15:05] <jepler> not quite everything makes it out to the Python layer -- just the things that axis uses in the preview, basically. I notice that "spindle synch on" and "off" messages aren't passed, for instance.
[20:15:50] <awallin> right, so now someone needs to tell me how to do the .htacces thing for http://el-servo.physics.helsinki.fi/personal/awallin/cnc_papers/
[20:16:02] <lerneaen_hydra> http://pixelcomic.net/014.shtml
[20:16:03] <lerneaen_hydra> haha
[20:16:07] <lerneaen_hydra> fitting :D
[20:16:20] <anonimasu> haha
[20:16:24] <lerneaen_hydra> (all the ususal "charachters" are pixels)
[20:16:30] <anonimasu> lerneaen_hydra: what do you think about it?
[20:16:36] <lerneaen_hydra> think about what?
[20:16:40] <anonimasu> voxels..
[20:16:49] <lerneaen_hydra> oh, something octree based?
[20:17:07] <ds3> awallin: ls | awk '{print '<a href=',$1,'>',$1,'</a>\n';}' > index.html
[20:17:18] <ds3> that should be an alternative hack
[20:17:25] <awallin> ds3: ok, Ill try that...
[20:17:29] <lerneaen_hydra> seems to me a parametric solution would be easier to implement, though that may just be because I'm more used to thinking like that
[20:17:58] <anonimasu> hm..
[20:17:59] <anonimasu> yeah
[20:18:22] <awallin> -bash: a: No such file or directory
[20:18:49] <anonimasu> lerneaen_hydra: perhaps you should have the matchsticks with Zt and Zb values..
[20:19:00] <ds3> oops
[20:19:10] <ds3> replace the first and last ' with a "
[20:19:15] <anonimasu> and then on a undercut you split, the matchstick, into smaller voxels if you would like..
[20:19:27] <anonimasu> *rant*
[20:19:31] <awallin> ls | awk "{print '<a href=',$1,'>',$1,'</a>\n';}" > index.html
[20:19:31] <awallin> awk: cmd. line:1: {print '<a href=',,'>',,'</a>\n';}
[20:19:31] <awallin> awk: cmd. line:1: ^ invalid char ''' in expression
[20:20:23] <anonimasu> ls|awk {'print
[20:20:25] <lerneaen_hydra> it's just that when you do matchstick based stuff you're immediatly locked to resolution and aliasing issues, whereas something parametric has "perfect" resolutio
[20:20:43] <anonimasu> lerneaen_hydra: this is just for rendering..
[20:21:03] <lerneaen_hydra> yeah, of course
[20:21:03] <anonimasu> lerneaen_hydra: keep in mind that when you intersect a matchstick you chop it up into smaller sticks..
[20:21:08] <awallin> lerneaen_hydra: the matchstick model can be made adaptive, just insert more sticks where you think the Z-height changes too much
[20:21:12] <anonimasu> that's the idea..
[20:21:15] <ds3> okay let me run it
[20:21:35] <lerneaen_hydra> awallin, are the matchsticks very "long" and "thin"?
[20:21:46] <anonimasu> yes
[20:21:46] <SWPadnos> awallin: add to the .htaccess file (or create one in that dir) the line Options +Indexes
[20:21:56] <lerneaen_hydra> essentially rectangular voxels?
[20:21:57] <anonimasu> yeah
[20:22:08] <lerneaen_hydra> where the bits that intersect with the tool are removed?
[20:22:12] <anonimasu> yep
[20:22:25] <SWPadnos> awallin, you can also add IndexOptions +FancyIndexing if you want icons and stuff
[20:22:32] <ds3> ls | awk '{print "<a href=" $1 ">" $1 "</a>\n";}' > index.html
[20:22:38] <ds3> there this one i have tested :)
[20:23:07] <lerneaen_hydra> do you plan to make matchstick dimensions dynamically changable?
[20:23:10] <SWPadnos> that generates a static list, and doesn't include HTML headers
[20:23:21] <ds3> *nod* hence the term 'hack'
[20:23:28] <SWPadnos> heh
[20:23:38] <ds3> works against facist admins who turn off .htaccess
[20:23:40] <ds3> =)
[20:23:45] <anonimasu> lerneaen_hydra: check for cutter intersection..
[20:23:48] <awallin> IndexOptions +FancyIndexing by itself does not seem to be enough
[20:24:15] <lerneaen_hydra> anonimasu, say you start with a block a,b,c in size, is that one matchstick or lots of small ones?
[20:24:20] <SWPadnos> you need the earlier one as well: Options +Indexes
[20:24:23] <anonimasu> it's one
[20:24:31] <lerneaen_hydra> ok
[20:24:33] <anonimasu> or well a fixed set..
[20:24:39] <lerneaen_hydra> hmm
[20:25:22] <anonimasu> lerneaen_hydra: then keep track of the closest sticks and subdivide them..
[20:25:36] <anonimasu> as you have the cutter diameter and stuff..
[20:25:37] <awallin> yeah, now it works... I'll let you guys know when the upload has finished
[20:25:38] <anonimasu> and position..
[20:25:49] <lerneaen_hydra> so essentially it's still octree based
[20:26:12] <SWPadnos> quadrtees, with the height as the data in the block :)
[20:26:17] <SWPadnos> quadtrees
[20:26:17] <anonimasu> yeah
[20:26:22] <lerneaen_hydra> oh
[20:26:34] <anonimasu> but quadtrees makes it hard to do voxels, on undercuts..
[20:26:34] <lerneaen_hydra> what about multiple matchsticks in the same XY coord?
[20:26:41] <anonimasu> oh
[20:26:45] <anonimasu> that would work..
[20:26:48] <anonimasu> for undercuts..
[20:27:17] <lerneaen_hydra> the Z-buffer would have to do a shitload of work when rendering the result
[20:27:42] <lerneaen_hydra> s/would/will
[20:27:51] <SWPadnos> like I said, the representation of the material is only part of the problem. making it displayable is a whole other issue
[20:28:02] <anonimasu> * anonimasu nods
[20:28:14] <anonimasu> though a quadtree can be sorted for speed.
[20:28:28] <anonimasu> if I remember it right..
[20:29:23] <anonimasu> SWPadnos: previewing stuff is never lightning fast..
[20:29:38] <anonimasu> not with mastercam or anything else..
[20:29:47] <anonimasu> the faster you go the less detail you get..
[20:30:33] <anonimasu> lerneaen_hydra: how would a parametric approach work?
[20:30:45] <lerman> Use quadtree dexels. quadtrees where the data is either a single depth or a linked list of depths.
[20:31:13] <anonimasu> dexel?
[20:31:39] <anonimasu> never heard that
[20:31:48] <lerman> dexel: depth pixel. Apparently from Van Horne.
[20:32:31] <anonimasu> ok?
[20:32:49] <lerneaen_hydra> anonimasu, I was thinking that you start with the base material, and for each move you make a feature that you subtract from the main mass (like extruded cut for example). if you even want to simulate the cutting procedure then you can have a part feature whose shape changes (gets longer)
[20:33:20] <awallin> now there are a bunch of papers at http://el-servo.physics.helsinki.fi/personal/awallin/cnc_papers/ these are courtesy of university of helsinki and myself - please DO NOT redistribute(post on websites or other public places).
[20:33:31] <anonimasu> you just did :/
[20:33:32] <awallin> I will take them away tomorrow morning
[20:33:33] <anonimasu> move it to another subdir
[20:33:37] <anonimasu> ah
[20:33:38] <lerneaen_hydra> that would be quite nice because you could very easily have a quick & dirty sim (steps through the code, removing each line of code from the material at a time), and a 1:1 sim that simulates the motion and lets you see it
[20:33:38] <lerman> Scrolling up a ways:
[20:33:40] <lerman> <lerman>awallin: damn you've got to stay with the program. I haven't been able to get access to the papers, but essential, at each point in X,Y, you record a sequence of depths representing the Z distance to EACH surface (in the case where there are undercuts).
[20:33:54] <anonimasu> lerneaen_hydra: well, that's essentially the matchstick we are talking about
[20:34:35] <lerman> With dexels, you have matchsticks with (possibly) multiple hunks cut out of their middles.
[20:34:36] <anonimasu> lerman: how do you do subdivision?
[20:34:42] <anonimasu> yeah..
[20:34:45] <lerneaen_hydra> the difference is that instead of modifying the matchsticks you have a lump you start with and then you subtract volume by volume, so you never loose information as such
[20:34:45] <anonimasu> sounds like a sane deal
[20:35:12] <lerneaen_hydra> it's kind of like the difference between a mesh-based 3d app and a parametric 3d-app
[20:35:19] <lerman> With dexels, when you cut a match, you split it into multiple hunks that can be linked together.
[20:35:29] <anonimasu> lerman: that
[20:35:35] <anonimasu> that's a horrid algorithm :)
[20:35:57] <lerman> The CSG approach is functionally much better. But... it is very expensive in run time.
[20:36:23] <anonimasu> hm
[20:36:38] <anonimasu> sounds like the difference between the normal and truesolid simulation in mastercam
[20:36:45] <lerneaen_hydra> could be
[20:37:41] <lerman> IN principle, the CSG approach is a simple as recording the fact that you are subtracting the tool at each position on the path. Rendering could can be done off line; but then it isn't real time.
[20:38:00] <SWPadnos> lerman, you mentioned CSG as being O(n^4) - is N the number of objects, faces, ...?
[20:38:04] <maddash> lol pixel comics
[20:38:43] <lerman> N is the number of tool montions. Which mightnot be too bad if the motions are long lines.
[20:39:20] <anonimasu> hm
[20:39:30] <anonimasu> what we are talking about looks like that voxel simulation paper.
[20:39:35] <lerman> In a past life, I implemented the 2D analog of an round tool cutting along a pixel plane.
[20:39:47] <SWPLinux> use the input G-code for the number of motions, not the feedback positions during machining
[20:40:05] <anonimasu> * anonimasu nods
[20:40:17] <SWPLinux> and/or, aggregate out the machine position feedbacks into larger single motions
[20:40:26] <SWPLinux> -out
[20:40:53] <anonimasu> I dont think we should concern ourselves with the machine position.. really
[20:41:01] <anonimasu> rather commanded position..
[20:41:29] <SWPLinux> I'd like to keep the idea in mind of having the preview (in AXIS or something) show the material
[20:41:32] <anonimasu> compensating for a machine that cant keep up in the cam program isnt a sane thing to do..
[20:41:39] <SWPLinux> that's true
[20:41:59] <anonimasu> though for the emc preview ofcourse, but that's not the same thing :)
[20:42:06] <anonimasu> all the cam program should do is spit out the optimal path..
[20:42:19] <alex_joni> hi guys
[20:42:25] <anonimasu> that the that the machine should be able to follow, within reasonable limits
[20:42:36] <lerman> Well, for CSG using the commands is natural. For a dexel or voxel approach running off of the steps from a stepper would probably be easier.
[20:43:13] <SWPLinux> hi Alex
[20:43:30] <lerman> At any rate, you would like the arcs broken into lines so you can just Bresenham them (yes, there is an arc type of bresenham).
[20:43:43] <anonimasu> lerman: yeah..
[20:43:53] <alex_joni> darn you guys talk a lot
[20:43:54] <SWPLinux> you can even do ellipses with a Bresenham-like algo :)
[20:44:04] <alex_joni> and a lot of 'crap' .. hard to filter quickly :D
[20:44:20] <anonimasu> alex_joni: forget it :D we still arent going anywhere close to a conclusion ;)
[20:44:29] <SWPLinux> though I did have problems when trying to do non-aligned ellipses (with 16-bit fixed-point math)
[20:44:30] <lerman> Think 3D. Don't be a piece of filter paper; be a sponge.
[20:44:34] <alex_joni> can you tell me the topic at least?
[20:44:38] <anonimasu> cam
[20:44:45] <anonimasu> haha
[20:44:55] <SWPLinux> 3D visualization of the result of machining operations
[20:44:56] <anonimasu> lerman: that's a good quote
[20:45:10] <maddash> psht. a filter paper IS 3-D.
[20:45:19] <lerman> (Copyright 2007, Kenneth Lerman).
[20:45:25] <lerneaen_hydra> maddash, oh no it's not
[20:45:29] <lerneaen_hydra> I say it's 2d, therefore it is
[20:45:30] <lerman> Not in our model.
[20:46:05] <lerman> I think therefore I blather.
[20:46:11] <lerman> :-)
[20:46:23] <maddash> silly engineer. a filter paper is several million atoms thick. :P
[20:47:21] <anonimasu> maddash: as lerman said not in our model ;)
[20:47:32] <lerman> Is there a decent platform that we could use to cobble these ideas into a proof of concept?
[20:47:42] <anonimasu> hm, opengl? and c++?
[20:47:44] <lerman> decent == easy to use
[20:48:02] <awalli1> lerman: do you have matlab?
[20:48:04] <maddash> visual basic. lol.
[20:48:07] <lerman> No.
[20:48:53] <lerneaen_hydra> python & opengl?
[20:49:09] <anonimasu> * anonimasu nods
[20:49:15] <awalli1> or vtk?
[20:49:21] <anonimasu> though you would really want your stuff to be in c++
[20:49:28] <anonimasu> lists and stuff..
[20:49:33] <awalli1> but vtk might be harder than simple opengl...
[20:49:42] <lerman> Well, part of the proof of concept involves timing. So I'd want the hard part to be in C.
[20:49:56] <anonimasu> opengl is very simple, if you dont complicate it..
[20:49:57] <anonimasu> :D
[20:50:04] <lerneaen_hydra> haha
[20:50:09] <anonimasu> um..
[20:50:13] <anonimasu> I just contradicted myself.
[20:50:14] <anonimasu> :D
[20:50:23] <lerneaen_hydra> <insert topic> is very simple, unless it's difficult
[20:50:23] <anonimasu> slap me
[20:50:28] <lerneaen_hydra> that's almost irish
[20:50:28] <awalli1> bbl. lerman: did you catch the link to the papers, I'll take them away in a few hours
[20:50:37] <SWPLinux> no you didn't. opengl is very simple if you don't use any of the complicated stuff ;)
[20:50:41] <lerneaen_hydra> t'will be a wonderfull day tomorrow, unless it's not
[20:50:56] <anonimasu> heh
[20:51:10] <anonimasu> if all you want to do is draw polygons it's pretty simple..
[20:51:11] <anonimasu> :D
[20:51:12] <maddash> those aren't contradictions
[20:51:18] <anonimasu> shaders and stuff gets nasty
[20:51:24] <maddash> those are tautologies
[20:52:03] <lerman> I caught the link. Could you make a tar.gz out of the whole set so I can grab them in one swell foop?
[20:52:18] <maddash> ughhhhhhhhhhh I still can't get my rtai-dev script working
[20:52:25] <SWPLinux> wget -nH -r <the linked dir>
[20:52:56] <SWPLinux> plus or minus a couple of "exclude=" options, for the various index files
[20:52:59] <maddash> lerman: spoonerism?
[20:53:10] <SWPLinux> are you spooking in speenerisms?
[20:53:44] <maddash> haha that last one wasn't a spoonerism..."ea" vs. "ee"...
[20:54:02] <lerman> One fell swoop.
[20:54:17] <lerman> Whatever that means :-)
[20:54:23] <SWPLinux> sure it was - speaking / spoonerism ...
[20:54:27] <lerman> Neither one makes any sense.
[20:54:40] <maddash> "Fighting liars in the quadrangle"
[20:55:01] <SWPLinux> a fun book: "Lady Slings the Booze" - by Spider Robinson
[20:55:03] <lerneaen_hydra> * lerneaen_hydra removes the bottle of hallucagens from #emc
[20:55:20] <lerneaen_hydra> * lerneaen_hydra gives them to some other, more needy, channel
[20:55:21] <SWPLinux> http://en.wikipedia.org/wiki/Spoonerism
[20:55:43] <lerman> SWPLinux: I'm wgetting awallins stuff now. Never did learn to use wget.
[20:55:57] <SWPLinux> me either - I had to "man" it :)
[20:56:02] <maddash> * maddash berates lerneaen_hydra for removing the wrong bottle of *hallucinogens*
[20:56:11] <tomp> join #ubuntu
[20:56:14] <SWPLinux> and you still end up with a directory tree underneath the current dir
[20:56:20] <maddash> tomp: LOL
[20:56:20] <tomp> duh
[20:56:24] <SWPLinux> tomp: no thanks :)
[20:56:50] <tomp> my tts reader stopped reading what you guys say...
[20:56:52] <maddash> just wget [url]
[20:57:01] <maddash> your WHAT reader?
[20:57:09] <maddash> oh, "tts"
[20:57:28] <anonimasu> hm
[20:57:38] <anonimasu> text to speech?=
[20:58:15] <tomp> yes
[20:58:33] <lerman> Mine has trouble pronouncing SWPLinux. :-)
[20:58:35] <maddash> I was thinking more phonetically...
[20:59:11] <tomp> 'do' is always 'doh"
[20:59:48] <SWPLinux> oh, now this is too funny: http://www.fabelbish.com/
[21:00:07] <anonimasu> heh
[21:01:07] <maddash> "Fame lucks"
[21:01:28] <lerneaen_hydra> haha: what a shiny fish -> shat a whiny fish
[21:01:57] <SWPLinux> you and your fish ... :)
[21:02:15] <lerneaen_hydra> * lerneaen_hydra slaps SWPLinux with a trout
[21:02:34] <SWPLinux> Swplaps winux trith a slout
[21:02:44] <SWPLinux> whatever that's supposed to mean
[21:03:07] <lerman> awallin: got them all.
[21:06:18] <anonimasu> hm
[21:06:24] <anonimasu> did all sensible discussion die :/(
[21:06:26] <lerman> The Blasquez paper discusses UNDO in this application. It's a good reminder that we need the ability to single step forwards and backwards. After the simulation shows that we are cutting in the wrong place, we will want to go back and determine the line of gcode that was wrong.
[21:06:45] <lerneaen_hydra> * lerneaen_hydra whines about parametric modeling
[21:06:50] <lerman> No. We are just busy typing long winded comments. :-)
[21:07:07] <anonimasu> lerneaen_hydra: that's not the same thing as machining or is it?
[21:07:21] <lerneaen_hydra> ? how isn't it?
[21:07:25] <lerneaen_hydra> cd ..
[21:07:26] <lerneaen_hydra> oops
[21:07:27] <SWPLinux> lerneaen_hydra: this is inherently not a parametric application. the idea is to see what's left after a series of steps have been performed
[21:07:37] <lerneaen_hydra> yeah
[21:07:55] <lerneaen_hydra> you perform various tasks, straight cuts or arcs for example
[21:08:10] <SWPLinux> we're not trying to generate models, rather to see if the movements generated from a created model will actually result in something similar to the model
[21:08:12] <lerneaen_hydra> hmm, let me sketch up an example
[21:08:12] <anonimasu> lerneaen_hydra: yeah, but you dont want a _arc_ you want to move in a arc motion with the cutter..
[21:08:24] <lerneaen_hydra> just a few minutes
[21:08:38] <anonimasu> lerneaen_hydra: while a parameteric representation would be perfect, I dont think that's what we want
[21:08:59] <lerneaen_hydra> becuase?
[21:09:05] <anonimasu> it
[21:09:08] <lerman> 'parametric' just refers to the fact that you can scale the elements...
[21:09:08] <SWPLinux> we're not representing the material. in a sense, we're representing what's taken out of the model
[21:09:12] <anonimasu> it's a model of the material/part being cut
[21:09:20] <SWPLinux> s/model/material
[21:09:22] <anonimasu> not the part itself
[21:09:22] <SWPLinux> /
[21:09:34] <lerneaen_hydra> hmm, then I'm not expressing myself correctly
[21:09:40] <lerman> Most parameteric modeling systems use either CSG or BREP. I think that might be what LH is referring to.
[21:09:45] <lerneaen_hydra> let me make this screenshot
[21:09:48] <anonimasu> ok
[21:10:05] <SWPLinux> a vector representation, rather than a raster one, would be nice (and that may be what you're trying to say)
[21:10:29] <lerneaen_hydra> yes it's inherantly "vector" based
[21:10:37] <lerneaen_hydra> no rounding/lossy stuff at least
[21:10:52] <lerman> Hey. The Blasquez paper has a good description of the dexel approach.
[21:10:56] <SWPLinux> right. that gets computationally expensive with moderate complexity
[21:10:56] <anonimasu> yeah
[21:11:44] <lerman> I referred to Van Horne, earlier. That should be Van Hook.
[21:11:54] <alex_joni> Van Wilder
[21:12:14] <SWPLinux> Van Morrison
[21:12:22] <alex_joni> Jim Morrison
[21:12:30] <anonimasu> awalli1: hey
[21:12:34] <jepler> Jim Jones?
[21:12:45] <SWPLinux> Tom Jones
[21:12:54] <anonimasu> http://www.springerlink.com/index/G2581W79U7752707.pdf
[21:13:11] <alex_joni> Tom Petty
[21:13:24] <anonimasu> * anonimasu sighs
[21:13:34] <anonimasu> fsck.
[21:13:49] <SWPLinux> heh - I thought of that one, but then couldn't think of another "petty"
[21:14:02] <anonimasu> I HATE THEESE PAY SITES THAT LETS GOOGLE INDEX THEIR*
[21:14:04] <SWPLinux> except the racer
[21:14:58] <alex_joni> Richard Petty :P
[21:15:04] <anonimasu> hm.
[21:15:19] <alex_joni> there's an actress Lori Petty
[21:15:25] <alex_joni> http://www.imdb.com/name/nm0001614/
[21:15:44] <lerman> If I were google, I'd be charging them to index that stuff.
[21:16:18] <lerneaen_hydra> http://lerneaenhydra.shacknet.nu/images/files_temp/part1.png
[21:16:21] <lerneaen_hydra> for example
[21:16:32] <lerman> So... At least someone besides me thinks that a dexel approach is a good idea.
[21:16:42] <anonimasu> look at the choi pocketing simulation
[21:16:43] <lerneaen_hydra> and if you want to simulate the cutter moving and cutting away material
[21:16:49] <lerneaen_hydra> (2 seconds)
[21:16:59] <anonimasu> paper..
[21:18:17] <SWPLinux> lerneaen_hydra: what software did you use to generate that, and how long does it take to "rebuild" if you change the path?
[21:18:26] <lerneaen_hydra> I did it in solidworkd
[21:18:29] <lerneaen_hydra> solidworks
[21:18:35] <lerneaen_hydra> it's fast to generate
[21:18:40] <SWPLinux> ok. I can benchmark it myself then
[21:19:05] <anonimasu> lerneaen_hydra: yeah but that will be different from what you are cutting.
[21:19:14] <anonimasu> really different
[21:19:29] <lerneaen_hydra> http://lerneaenhydra.shacknet.nu/images/files_temp/part2.png <-- while cutting the slot
[21:19:35] <SWPLinux> yes, but the problem is that you may end up with many thousands (or at least hundreds) of motions of that profile, and many of them will be "cuttin air", and can be optimized out of the representation
[21:19:55] <anonimasu> lerneaen_hydra: yeah, when modelling/making toolpaths.
[21:20:02] <lerneaen_hydra> anonimasu, yep
[21:20:05] <anonimasu> lerneaen_hydra: but when simulating, it's not remotely like the final thing
[21:20:32] <lerneaen_hydra> SWPLinux, very true, but that doesn't mean you can't optimise this way of viewing it
[21:20:34] <anonimasu> there's also the Z vector approach
[21:20:35] <SWPLinux> lerneaen_hydra: one thing to try with that: make the path move in all 3 dimensions
[21:20:45] <anonimasu> SWPLinux: that's easy ;)
[21:20:56] <lerneaen_hydra> any particular reason?
[21:21:11] <SWPLinux> ie, change the depth of cut, but keep the cutter aligned in Z (but still along the vector in XY)
[21:21:11] <lerneaen_hydra> it'l just be the same thing but with varying depth
[21:21:26] <anonimasu> lerneaen_hydra: no guide curves ;)
[21:21:31] <SWPLinux> I know, but it changes the resulting geometry a bit, and would probably change the rebuild time
[21:21:40] <anonimasu> not by much
[21:21:43] <lerneaen_hydra> the guide curve would just be a 3d line instead of a 2d one
[21:21:44] <anonimasu> atleast not in solidworks..
[21:21:49] <lerneaen_hydra> * lerneaen_hydra shrugs
[21:21:51] <anonimasu> yeah..
[21:22:08] <anonimasu> SWPLinux: you cant see what it makes the block out of anyway..
[21:22:13] <lerneaen_hydra> and the guide curve is *exactly* the gcode we have to begin with
[21:22:20] <SWPLinux> I find that a cylinder is easy, but a tailpipe (for instance) takes a bit of time
[21:22:29] <lerneaen_hydra> tailpipe?
[21:22:46] <SWPLinux> like a bent tube that would be used on a car ...
[21:22:45] <anonimasu> http://www.almaskin.se/imptest.jpg
[21:23:23] <lerneaen_hydra> oh, for a car, what does that have anything to do with this?
[21:23:48] <SWPLinux> the complexity of the resulting visualization data depends on the model
[21:23:56] <lerneaen_hydra> oh, sure
[21:24:05] <SWPLinux> a mostly 2D model is relatively easy to convert to visual data
[21:24:14] <lerneaen_hydra> thats true for anything though
[21:24:28] <lerneaen_hydra> btw, what is this type of modeling called? is it parametric?
[21:24:39] <SWPLinux> right - that's my point. I'm trying to get a gauge of the time it takes to make models of more complex things
[21:24:41] <anonimasu> hm, for proof of concept I'd go with the Z depth trimming..
[21:24:51] <anonimasu> SWPLinux: it's realtime..
[21:24:54] <anonimasu> fairly..
[21:25:00] <SWPLinux> solidworks is parametric, because you can edit the sketch and rebuild the model with the new data
[21:25:04] <lerneaen_hydra> yeah
[21:25:06] <anonimasu> SWPLinux: adding a operation in solidworks goes _flip_
[21:25:06] <lerneaen_hydra> that's what I though
[21:25:22] <lerneaen_hydra> SWPadnos, I can do a test torture test and see how long it takes
[21:25:39] <SWPLinux> heh - import tort.ngc as the path ;)
[21:25:53] <SWPLinux> even chips3D would be of interest
[21:26:00] <SWPLinux> err - 3Dchips
[21:26:04] <anonimasu> SWPLinux: you would have to convert it to a model..
[21:26:12] <anonimasu> I'd save your model as a step..
[21:26:16] <anonimasu> nd import it again :)
[21:26:24] <anonimasu> or something that make it a mesh ;)
[21:26:48] <SWPLinux> well, I mean use the same method (of sweeping a profile along a path), but use an NC file as the path somehow
[21:26:51] <SWPLinux> even a simple one
[21:27:29] <SWPLinux> I'd specifically want to see how that bogs down in SolidWorks, rather than exporting a model of the intended surface ...
[21:27:42] <anonimasu> SWPLinux: I dont think you will see that happening
[21:27:48] <lerneaen_hydra> SWPadnos, the sweeping bit is exactly what I meant
[21:27:53] <SWPLinux> I know :)
[21:27:59] <lerneaen_hydra> SWPadnos, now I'm making a huuuge spiral
[21:27:59] <anonimasu> SWPLinux: it never bogs down.
[21:28:01] <SWPLinux> anonimasu: see what happening?
[21:28:05] <lerneaen_hydra> just insanely large
[21:28:23] <SWPLinux> you need to change something in the path, and rebuild the model. redrawing is nearly instantaneous
[21:28:49] <anonimasu> let well, if we change the taper?
[21:28:51] <anonimasu> would that count?
[21:29:27] <SWPLinux> it's rebuilds that take time, and that's what has to be done every time the path changes (like a path simulator would do several times per second, ideally)
[21:29:35] <SWPLinux> anonimasu: probably
[21:29:42] <SWPLinux> change the sketch or the path
[21:29:47] <lerneaen_hydra> you only have to rebuild the bit you're working on
[21:30:06] <lerneaen_hydra> you'll subdivide each line of gcode into one contour
[21:30:09] <anonimasu> it takes a long time with 1000rev's..
[21:30:26] <anonimasu> didnt we have this discussion about contour's with the tp issue before?
[21:30:35] <SWPLinux> pretty much :)
[21:30:36] <lerneaen_hydra> * lerneaen_hydra cires, solidworks is bitching at me and doesn't let me draw the spiral
[21:30:55] <anonimasu> it locks up because my computer is slow with a _huge_ spiral..
[21:31:07] <lerneaen_hydra> anonimasu, how do you get spirals to work?
[21:31:13] <anonimasu> you sketch a circle..
[21:31:15] <lerneaen_hydra> I have a sketch with a circle
[21:31:18] <anonimasu> then push spiral..
[21:31:23] <anonimasu> select it and push spiral.
[21:31:27] <anonimasu> that's all there is to it :)
[21:31:27] <lerneaen_hydra> oh
[21:31:32] <lerneaen_hydra> i am dum
[21:31:38] <anonimasu> nah
[21:31:46] <SWPLinux> are you guys on SW2006?
[21:31:52] <anonimasu> 2005
[21:31:55] <lerneaen_hydra> it's still bitching
[21:31:58] <lerneaen_hydra> 2006 here
[21:32:03] <anonimasu> lerneaen_hydra: what does it do?
[21:32:04] <lerneaen_hydra> greyed out
[21:32:08] <SWPLinux> ok. I have 2004 in a VM - we'll se how it goes :)
[21:32:10] <anonimasu> select your circle..
[21:32:12] <lerneaen_hydra> insert -> curve -> spiral
[21:32:17] <anonimasu> close the sketch..
[21:32:26] <anonimasu> first of all :)
[21:32:33] <anonimasu> as it's a feature
[21:32:57] <lerneaen_hydra> oh
[21:33:03] <lerneaen_hydra> I was doing it as a curve
[21:33:07] <anonimasu> :)
[21:33:36] <lerneaen_hydra> heh
[21:33:53] <lerneaen_hydra> SW + VMware + 1000 rev spiral = not fun
[21:33:55] <anonimasu> heh
[21:33:56] <anonimasu> I made a 400
[21:34:05] <SWPLinux> you forgot to add in the Opteron 275 :)
[21:34:08] <lerneaen_hydra> though that's not becuase of intrinsic problems
[21:34:10] <anonimasu> calculating takes a bit..
[21:34:23] <lerneaen_hydra> I'm running on an X2 3800+
[21:34:30] <anonimasu> im out of ram.
[21:34:43] <anonimasu> 744 mb swap..
[21:34:49] <lerneaen_hydra> I've got 512mb allocated to vmware though, should be doable
[21:34:54] <SWPLinux> lerneaen_hydra: that isn't likely to be a good indication of performance on a CNC-class machine then ...
[21:35:11] <anonimasu> heh..
[21:35:13] <anonimasu> wait a sec.
[21:35:18] <lerneaen_hydra> it's just drawing the guide curve now
[21:35:25] <lerneaen_hydra> no swept profile
[21:35:40] <lerneaen_hydra> so IMO seems to be an implementation problem
[21:35:47] <alex_joni> anonimasu: around?
[21:35:49] <anonimasu> yes
[21:35:58] <alex_joni> ever did threads in alibre?
[21:35:59] <maddash> are you guys having problems with accessing journals?
[21:36:06] <anonimasu> no
[21:36:29] <anonimasu> alex_joni: I think I read a tutorial about it..
[21:37:10] <anonimasu> arc length 50422.27mm
[21:38:38] <anonimasu> doing the solid now
[21:38:54] <anonimasu> it calculates quite a bit..
[21:39:31] <anonimasu> locked up..
[21:39:32] <anonimasu> out of ram :)
[21:40:02] <lerneaen_hydra> I really doubt SW is as optimised as it can be
[21:40:15] <anonimasu> it's pretty well optimized..
[21:40:18] <anonimasu> but this is a extreme.
[21:40:23] <lerman> I'm back. Firefox locked up on me. That happens occasionally when I'm looking at pdf files.
[21:40:36] <anonimasu> you arent supposed to make 50 meter paths, with surfaces..(3d)
[21:40:45] <maddash> what the heck is a bi-arc? or a biarc? or bi arc?? google/wikipedia refuse to let me know...
[21:40:47] <lerman> Did I miss anything important.
[21:40:52] <maddash> bi/arc vs. arc?
[21:41:01] <anonimasu> lerneaen_hydra: comparing solidworks performance
[21:41:09] <anonimasu> err lerneaen_hydra
[21:41:11] <anonimasu> lerman
[21:41:14] <anonimasu> damn tab
[21:41:30] <anonimasu> lerman my firefox does that dtoo
[21:41:39] <CIA-8> 03cradek 07v2_1_branch * 10emc2/src/emc/task/emccanon.cc: fix blending last move if it's a g1
[21:41:41] <SWPLinux> I wouldn't expect us to come up with anything better than SolidWorks for quite a while (at least not in the performance realm)
[21:41:50] <lerman> OK. Since I know nothing about solidworks, I guess I shouldn't comment.
[21:42:02] <CIA-8> 03cradek 07TRUNK * 10emc2/src/emc/task/emccanon.cc: fix blending last move if it's a g1
[21:42:03] <anonimasu> SWPLinux: it's very well optimized for working with large models..
[21:42:17] <anonimasu> but it's pretty dependant on good/fast hardware.
[21:42:19] <lerman> Doesn't solidworks use CSG? We can do better by not using CSG.
[21:42:20] <SWPLinux> yes - it's also used as a benchmark for video card performance ...
[21:42:24] <anonimasu> my issue is the graphics card..
[21:42:30] <lerneaen_hydra> SWPLinux, it's not optimised for stuff like this
[21:42:48] <anonimasu> SWPLinux: you would need a reference model..
[21:43:16] <SWPLinux> I think the point I was thinking of was that it's not a simple process, and the task we have is essentially like a SW rebuild, not a SW render / view rotate
[21:43:28] <anonimasu> yeah..
[21:43:32] <anonimasu> true
[21:43:39] <lerneaen_hydra> hmm, IMO it's not a rebuild
[21:43:49] <anonimasu> we never got that far.
[21:43:53] <lerneaen_hydra> at least not the implementation I'm thinking of
[21:44:14] <anonimasu> lerneaen_hydra: you would need to run the entire calc for the surface either way wouldnt you?
[21:44:20] <lerneaen_hydra> each Gcode would be its own guid curve
[21:44:27] <lerneaen_hydra> and it's own swept profile
[21:44:36] <anonimasu> lerneaen_hydra: this is not related to the other problem then
[21:44:49] <lerneaen_hydra> which means you only need to redo the bits that the line comes into contact with
[21:44:58] <anonimasu> the simulation must represent, the cut away material
[21:45:10] <lerneaen_hydra> sure
[21:45:10] <anonimasu> what happens when you move a spling?
[21:45:14] <anonimasu> spline?
[21:45:33] <anonimasu> or guide curve?
[21:45:33] <lerneaen_hydra> guide curve that's not the complete part of the gcode line
[21:45:45] <lerneaen_hydra> like my second image
[21:46:05] <anonimasu> really, this isnt gcode related at all..
[21:46:06] <anonimasu> either
[21:46:11] <lerman> You might want to look at brl-cad to do a quick implementation. It has all of the CSG stuff you would need.
[21:46:15] <anonimasu> this is whatever internal format for points(targets) we use..
[21:46:24] <jepler> maddash: this pdf seems to explain what a biarc is. http://mat.fsv.cvut.cz/gcg/sbornik/sir.pdf
[21:46:26] <anonimasu> you will be posting it in the end..
[21:46:42] <anonimasu> or splines/motion
[21:46:42] <jepler> maddash: a biarc is a pair of arcs with a joint at which their tangents "match"
[21:47:21] <anonimasu> lerman: I think the Z depth intersection stuff they are talking in the second paper is the fastest way to do 3d..
[21:47:25] <anonimasu> proof of concept..
[21:47:27] <lerman> You might be able to create a way to drip feed brl-cad with a hook from canon.
[21:47:41] <anonimasu> as the toolpathing stuff is the primary thing
[21:47:44] <lerman> Which one is the second paper?
[21:48:01] <anonimasu> contouring simulation something
[21:48:06] <anonimasu> by the same guy as the voxel paper..
[21:48:20] <anonimasu> there's a description of some algotithms briefly
[21:48:36] <anonimasu> I think we should start at a sane level to get toolpathing/visualization working
[21:48:40] <anonimasu> simulation..
[21:48:51] <SWPLinux> gah -I wish I knew how to make a sweep :)
[21:48:52] <lerman> What's the file name?
[21:49:13] <anonimasu> wait
[21:49:17] <maddash> * maddash profusely thanks jepler for unwrapping the biarc enigma
[21:49:21] <anonimasu> choi - pocketing simulation.pdf
[21:49:22] <anonimasu> wrong..
[21:49:26] <anonimasu> not the same guy
[21:51:16] <anonimasu> sounds like a _easy_ algorthm
[21:53:04] <lerman> I would start with a uniform x, y grid based depth approach with a clean interface. Then I add the dexel notion. Finally, I would modify it to use quadtrees on x,y. The interface to the module should be the same in all cases.
[21:53:22] <anonimasu> yeah
[21:53:31] <anonimasu> I agree
[21:54:27] <SWPLinux> the interface would need to include unsupported features, or it would need to change, I think
[21:55:00] <SWPLinux> going from a single solid (matchstick) to a multiple solid region (dexel / voxel) represents a change in how the cut is described (I think)
[21:55:11] <lerman> Well, I would probably *try* to design the interface for the hard case to start.
[21:55:30] <lerman> In all cases, I'm assuming orthogonal xyz.
[21:55:31] <anonimasu> hm, SWPLinux cutter motion is how you would describe your cut..
[21:55:31] <SWPLinux> unless the simplistic approach just uses the lowest point as the cutoff, and ignores the actual profile (of a slot cutter, for instance)
[21:55:41] <SWPLinux> cutter motion and shape ...
[21:55:44] <anonimasu> cutter as a solid body..
[21:55:44] <anonimasu> yeah..
[21:56:04] <SWPLinux> so a T-slot cutter has a different notation than a plain endmill
[21:56:05] <anonimasu> or subdivide your matchsticks.
[21:56:21] <lerman> No. In all cases, I would have to define the cutter shape. Even without undercuts, you need to handle ball end vs square end cutters.
[21:56:27] <anonimasu> yeah
[21:56:45] <SWPLinux> cutter end shape is a different animal than cutter profile
[21:56:47] <anonimasu> subdividing ends up _close_ enough I think..
[21:56:59] <SWPLinux> heh - it can also be done with matchsticks though - coming in from the side :)
[21:57:01] <anonimasu> you end up with smaller matchsticks with different z then the other..
[21:57:31] <SWPLinux> this is the kind of shape I'm talking about: http://www.discount-tools.com/tslotindexmain1.htm
[21:57:45] <anonimasu> it will get computing intensive
[21:57:52] <anonimasu> yeah..
[21:58:10] <SWPLinux> that's what has to be represented in the modeling classes, even if it's ignored until the resulting models can handle it
[21:58:12] <anonimasu> undercuts ends up hard, until you go for the dexel approach..
[21:58:18] <anonimasu> yeah
[21:58:18] <lerman> I might make it object oriented. So if I tell the cutter to move, it would in turn tell me which new voxels have been cut. Then I would translate voxel cuts to do the right thing in my material model.
[21:58:43] <SWPLinux> hmmm
[21:58:47] <anonimasu> hm, yeah
[21:58:56] <anonimasu> lerman: split them up to represent your model correctly
[21:59:24] <lerman> It would also tell me what non cutting space it touches, so that I could detect when the slot cutter crashes against the model.
[21:59:30] <anonimasu> yeah
[21:59:36] <lerman> Split what up?
[21:59:38] <anonimasu> that's essentially what you want..
[21:59:43] <anonimasu> lerman: subdivide for more detail..
[21:59:47] <anonimasu> insert more matchsticks..
[22:00:08] <anonimasu> lerman: if you cut 1/100 of a matchstick/round the corner
[22:00:23] <lerman> Yes. But the initial version would not do splitting. I'd start splitting when I got to the quadtree version.
[22:00:26] <anonimasu> lerman: though it may be more optimal to start out with tiny matchsticks..
[22:00:44] <anonimasu> as rendering takes less time the computing changes..
[22:00:52] <anonimasu> yeah
[22:00:56] <lerman> I would do that. The question is how much time/memory you can deal with.
[22:01:07] <anonimasu> lerman: in cam programs that's a user setting..
[22:01:14] <anonimasu> atleast mastercam/visualmill..
[22:01:25] <anonimasu> more detail(slower) and less detail(faster)
[22:01:31] <lerman> I would let the user specify (in an init file) what resolution to compute at.
[22:01:42] <anonimasu> yeah..
[22:01:44] <anonimasu> slider or whatever..
[22:02:01] <lerman> That works.
[22:02:03] <anonimasu> lerman: im just on the cam approach, not the realtime display..
[22:02:32] <anonimasu> where realtime performance isnt as critical
[22:03:23] <anonimasu> lerman: ideally, when cutting you scan the sticks close to the cutter and manipulate the height accordingly to the cutter model..
[22:03:31] <lerman> I've mentioned this before, but my view is that realtime isn't that important. The simulation doesn't have to run in realtime.
[22:03:34] <anonimasu> as you have the xyz from each stick as reference..
[22:03:44] <lerman> Correct.
[22:03:48] <anonimasu> and the z value of the cutter offset from it..
[22:04:22] <anonimasu> that handles ballnose tools also, and plunging with odd tooling
[22:04:31] <anonimasu> and edm..
[22:04:36] <lerman> Scanning sticks near the tool is why an XYZ orthogonal arrangement is desirable.
[22:05:22] <anonimasu> what kind of toolkit do you use?
[22:05:23] <lerman> Also why quadtrees are harder than an XY matrix. (Well, not harder; but slower.)
[22:05:31] <anonimasu> hm, really?
[22:05:43] <lerman> Also quadtrees have to be split and that slows things down.
[22:05:50] <anonimasu> the xyz matrix is also piece of cake to code..
[22:06:11] <anonimasu> the big question
[22:06:15] <lerman> I'm basically a C programmer (or objective-C -- but not recently). On unix.
[22:06:26] <anonimasu> hm, im thinking python, and c++ for the backend..
[22:06:33] <anonimasu> everything computing intensive..
[22:06:56] <anonimasu> I'd rather not use c++ for setting up viewports and stuff..
[22:06:57] <lerman> I'm learning python (a little) now as I work on GWiz.
[22:07:18] <anonimasu> or well window's rather..
[22:07:27] <anonimasu> viewports is easy, but the actual window to render in is not
[22:07:28] <anonimasu> :)
[22:07:38] <lerman> The language is easy. The libraries are hard (non-intuitive to me).
[22:08:06] <anonimasu> yeah..
[22:08:14] <anonimasu> that's the deal with c++/opengl
[22:09:03] <anonimasu> also, lots of people here know python
[22:11:08] <lerman> The two parts that are at the heart of the matter are the tool/toolpath/model manipulations and the conversion to triangles (or other surfaces) for opengl.
[22:11:29] <lerman> I think I understand an approach to the tool/toolpath/model part.
[22:12:08] <lerman> Converting to triangles could be as easy as converting each rectangular face to two triangles.
[22:12:25] <lerman> That would NOT scale at all well.
[22:12:41] <maddash> That would NOT scale well at all.
[22:13:10] <lerman> An alternative would be to combine rectangles within some tolerance and produce rectangles.
[22:13:42] <lerman> I could probably come up with some sort of greedy algorithm that would work.
[22:14:41] <lerman> I suspect. though, that it might be fairly difficult to ensure that the surface remained closed.
[22:15:12] <lerman> I'm not familiar with the art in that area. It could easily be a well solved problem with a neat solution.
[22:15:26] <SWPLinux> SW question: how do I get a sweep to subtract instead of add material?
[22:15:30] <lerman> neat => one that is easy to understand and code.
[22:15:58] <lerman> Mutliply by -1 if in three dimensions of -i if in four. :-)
[22:16:46] <lerman> Or should I have said, "multiply by e to the pi i power"?
[22:16:53] <SWPLinux> heh
[22:16:57] <anonimasu> lerman: well, if you abstract each matchstick..
[22:16:59] <SWPLinux> I thought I'd imagined that
[22:17:08] <lerman> really
[22:17:24] <anonimasu> as a Z X/Y depth.. matrix..
[22:17:42] <anonimasu> where x and y are the positions in the global grid,
[22:17:47] <anonimasu> of matchsticks..
[22:18:27] <anonimasu> SWPLinux: you make a path extrude instead of a cut
[22:18:40] <lerman> That inverts part of the problem. But you really need to access the matchsticks by XY. (and Z in the dexel case).
[22:19:15] <lerman> So the XY for each matchstick can (should) be implicit in where it is in the matrix.
[22:19:16] <SWPLinux> got it
[22:19:21] <anonimasu> yeah
[22:19:42] <anonimasu> lerman: though I'd want a _reference_ in the matrix..
[22:20:07] <anonimasu> so we can call up another array of matchsticks..
[22:20:25] <anonimasu> lerman: for subdividing..
[22:20:31] <anonimasu> tell me if im way off..
[22:20:50] <anonimasu> where you have a coordinate system localized for each matchstick..
[22:21:06] <anonimasu> so you can split them in 1/10th or something
[22:21:28] <anonimasu> *rant*
[22:21:28] <lerman> Well, I think in pointers, not references.
[22:21:40] <anonimasu> yeah, you are right..
[22:21:48] <anonimasu> do you think it's a bad idea?
[22:22:01] <lerman> That would be OK if you are in the quadtree case.
[22:22:28] <anonimasu> hm, I'd rather skip quadtrees.. and just use a matrix..
[22:22:44] <anonimasu> but that's personal perference.
[22:23:06] <anonimasu> lerman: I think we can do away with it.. really
[22:23:15] <anonimasu> if we add a vector for each stick..
[22:23:25] <lerman> I'm a cheap bastard. So, in the initial case, I would have a matrix of words. Each word would have either a float representing the depth at that point or a pointer to a linked list of dexels.
[22:23:29] <anonimasu> yeah
[22:23:48] <anonimasu> if we add a vector to each stick, we can have them non planar..
[22:24:02] <lerman> On the one hand you talk about matrix and on the other you talk about dividing. How would you do that? Quad trees do that.
[22:24:18] <anonimasu> so you can actually see edges where your ballnode endmill goes..
[22:24:24] <anonimasu> even at cruder detail..
[22:25:12] <lerman> why do you need vectors at each stick to see the edges?
[22:25:37] <anonimasu> not edges top surface..
[22:26:04] <awalli1> here's something about drawing height maps with opengl http://www.codeproject.com/opengl/OPENGLTG.asp
[22:26:12] <lerman> So when you "cut" a voxel, you would record the normal to the cutter?
[22:26:24] <anonimasu> lerman: yeah
[22:26:28] <awalli1> I'm taking away the pdf's soon, so get them now if you want them
[22:26:35] <anonimasu> I have them
[22:26:39] <lerman> I got them. Thanks.
[22:26:41] <SWPLinux> some time ago, jepler created a test program to do endmill height mapping using OpenGL libraries
[22:26:44] <anonimasu> lerman: that way you end up with smooth surfaces when contouring..
[22:26:53] <anonimasu> without lots of sticks..
[22:27:00] <SWPLinux> that results in essentially what you're talking about now
[22:27:05] <anonimasu> yeah
[22:27:28] <SWPLinux> and in the case where there's hardware to do the raster ops, it goes very quickly
[22:27:30] <anonimasu> as I said before tell me if im entirely off track..
[22:27:37] <anonimasu> im ranting..
[22:27:40] <anonimasu> :)
[22:28:05] <anonimasu> lerman: I have no idea how you would do that, quad trees may be the easiest way
[22:28:14] <anonimasu> but as I've never dealt with them I dont know
[22:28:14] <lerman> The example doesn't allow undercuts.
[22:28:28] <anonimasu> lerman: yeah, though if you extend it to the pointer it would..
[22:28:31] <anonimasu> err
[22:28:38] <lerman> I don't know if quadtrees are easy, but they are elegant.
[22:28:40] <anonimasu> to the other example with subdivided sticks..
[22:28:54] <anonimasu> with pointer to external sticks.
[22:29:07] <anonimasu> that would handle undercuts..
[22:29:53] <lerman> Perhaps not with the algorithm used, though.
[22:30:17] <anonimasu> I think it would, as you keep checking for tool collission
[22:30:20] <anonimasu> aginst your tool model..
[22:31:07] <anonimasu> and if you find a stick with a pointer.. you check if the Z of the tool intersects with anything in the other matrix..
[22:31:31] <anonimasu> I need a pen and paper..
[22:31:32] <anonimasu> :D
[22:31:36] <lerman> I mean the model in the opengl example. It seems to trace along the x axis generating triangles.
[22:31:44] <anonimasu> yeah
[22:32:25] <anonimasu> lerman: I think I can draw this up in a little bit
[22:32:28] <anonimasu> I just need to put the dog to bed
[22:33:13] <lerman> Are you dyslexic 'dog' == 'god' ?
[22:33:59] <lerman> The fact that you need to put it to bed leads me to believe that we know which is the superior being.
[22:34:08] <lerman> :-)
[22:34:18] <anonimasu> hehe
[22:34:20] <awalli1> bbl
[22:35:05] <anonimasu> lerman: the easiest approach is a fixed size matchstick matrix..
[22:35:16] <anonimasu> with a z depth for each stick
[22:35:33] <lerman> Agreed. That's version one.
[22:35:38] <anonimasu> and a cutter object.
[22:35:57] <anonimasu> with a depth and size in 3d
[22:36:15] <anonimasu> that's actually much harder then the sticks..
[22:36:26] <lerman> I think of the cutter as having a profile.
[22:36:44] <alex_joni> anonimasu: figured it out..
[22:36:59] <anonimasu> alex_joni: nice
[22:37:06] <anonimasu> lerman: yeah but that wont handle edm..
[22:37:12] <alex_joni> you need to draw the V which cuts the thread :(
[22:37:29] <anonimasu> alex_joni: It's the same in sw.. I think
[22:38:08] <lerman> It is especially hard because you want to be efficient. So, when you move it in a particular direction, it only cuts on the side towards the motion. In three dimensions since the motion is in 3D.
[22:38:18] <anonimasu> yeah..
[22:38:40] <anonimasu> well, may be easier to assume we wont be doing edm yet.
[22:38:59] <alex_joni> anonimasu: bleah :(
[22:39:11] <alex_joni> how about standard screws?
[22:39:14] <anonimasu> alex_joni: you usually mark your hole as _threaded_
[22:39:21] <alex_joni> actually all I need is threaded rod
[22:39:27] <lerman> One trick you can do is break all motions into steps in either X, Y, Z, or combinations (both positive and negative). Precompute which voxels will be cut.
[22:39:32] <anonimasu> alex_joni: you dont need thread to make thoose stick in a hole ;)
[22:39:39] <lerneaen_hydra> aaargh. my brain is poisoned
[22:39:50] <anonimasu> lerman: yeah that speeds up calculation quite a bit..
[22:40:07] <lerneaen_hydra> I read "you usually mark your hole as _threaded_" as "you usually mark your asshole as threaded" :(
[22:40:07] <lerman> I did that in my 2D version.
[22:40:29] <lerman> Reamed but not threaded.
[22:40:36] <lerneaen_hydra> O_o
[22:40:44] <lerneaen_hydra> wtf have you been up to
[22:40:50] <lerneaen_hydra> prison?
[22:41:07] <anonimasu> lerman: you must be old as hell..
[22:41:22] <lerman> Older. :-)
[22:41:31] <lerman> Why do you say that?
[22:41:35] <anonimasu> if you were around in the days when humans were 2d
[22:41:47] <anonimasu> err "if you were around when humans still were in 2d"
[22:42:00] <lerneaen_hydra> haha
[22:42:07] <lerman> Ahhh. clever.
[22:43:05] <anonimasu> :)
[22:43:15] <anonimasu> I like this conversation
[22:43:29] <anonimasu> the cam stuff
[22:44:02] <lerman> Yes. Better than work.
[22:44:09] <anonimasu> hehe
[22:47:51] <anonimasu> 78haha
[22:47:55] <anonimasu> yeah
[22:49:00] <alex_joni> good night all
[22:49:10] <anonimasu> night alex..
[22:49:15] <alex_joni> happy CAM'ing
[22:49:37] <anonimasu> hehe
[22:49:59] <lerman> see you alex.
[22:50:12] <anonimasu> im busy drawing some stuff, I'll be back in 5
[22:52:53] <alex_joni> SWPadnos: got the boards today
[22:53:03] <SWPadnos> great!
[22:53:30] <alex_joni> only 3 more months and I'll have them :)
[22:53:34] <SWPadnos> heh
[22:53:49] <SWPadnos> I can bring them to you faster if you pay the airfare ;)
[22:54:14] <alex_joni> is that a 2-way business-class ticket for 2 ?
[22:54:36] <lerman> Plus the dog.
[22:55:00] <SWPadnos> no business for me - it's first all the way! :)
[22:55:19] <alex_joni> eeek
[22:55:39] <cradek> I'll bring it but my five cats need tickets too
[22:56:09] <SWPadnos> there's an FAA limit of two animals per flight (in the cabin)
[22:56:14] <alex_joni> * alex_joni sticks cradek and the cats in a big crate
[22:56:16] <SWPadnos> one for smaller planes
[22:56:33] <cradek> so you can't mix three cats and make an explosive?
[22:57:02] <alex_joni> btw.. did you guys see Mythbusters recently? they just had a nice salami-rocket :D
[22:57:33] <SWPadnos> hmmm - that makes me hungry
[22:57:47] <anonimasu> hm
[22:57:54] <anonimasu> machinists handbook alex.. look what you made me do :D
[22:58:25] <alex_joni> anonimasu: told you not to bother
[22:58:41] <anonimasu> haha
[22:58:44] <tomp> last nite i read an old paperback scifi mag (Dec 1958), back cover had advert for poster, address to send to was 'Geek Systems, Inc, suite 1114, 501 Fifth Ave, New York, 17, N.Y.,
[22:58:49] <anonimasu> it's a good excercise..
[22:59:04] <SWPadnos> cradek, 3 cats are by definition explosive ;)
[22:59:41] <lerman> No. It's three wives that are explosive. Hence the suggestion that you have four.
[23:00:02] <SWPadnos> hmmm - a spare. good thinking
[23:00:09] <alex_joni> lerman: I always thought it's a wife + mistress that gets the spark going
[23:01:18] <lerman> You know what you get when you divorce the wife and marry the mistress? ... A job opening for a new mistress.
[23:07:27] <anonimasu> iab
[23:07:31] <anonimasu> :D
[23:13:31] <anonimasu> 4hm
[23:13:40] <anonimasu> alex_joni: I made a thread :)
[23:19:12] <A-L-P-H-A> what's a good way to get your food to be "un" asleep?
[23:19:29] <SWPadnos> cook it
[23:21:06] <A-L-P-H-A> SWPadnos... okay "pass down" :)
[23:21:40] <anonimasu> damn I *yawn*
[23:21:44] <SWPadnos> heh -no problem "Pal - HA!"
[23:21:54] <A-L-P-H-A> lol
[23:23:36] <lerman> I'm reading the patent that uses dexels. It's worth the read. US patent no. 05710709
[23:25:12] <lerman> Apparently, for the 5 axis case, more time is spent scan converting the tool so that the dexels are parallel to those of the work than for anything else.
[23:27:12] <anonimasu> hm ok?
[23:27:41] <lerman> In the 3 axis case, that would not be a problem.
[23:28:02] <anonimasu> yeah
[23:30:25] <maddash> jepler: what's the date of publication of the paper you sent me?
[23:32:34] <maddash> did jepler go to bed?
[23:36:58] <tomp> 'dexel based geometric reasoning and visualization for die configuration', Cai Li, Ohio State University, 2002, (referenced at a uni in Taiwan)
[23:38:57] <anonimasu> * anonimasu nods
[23:40:01] <anonimasu> hm
[23:40:08] <anonimasu> I wonde what comp alex has..
[23:40:11] <anonimasu> this will kill it.
[23:49:03] <anonimasu> *yawn*
[23:49:07] <anonimasu> lerman: let's continue this tomorrow
[23:50:22] <SWPadnos> I wonder if this would be helpful: http://gts.sourceforge.net/
[23:52:15] <anonimasu> night
[23:52:26] <SWPadnos> see you
[23:52:39] <anonimasu> going to look at that tomorrow :)
[23:52:41] <SWPadnos> here's a c++ octree library (sort of): http://www.hxa7241.org/articles/content/octree-general-cpp_hxa7241_2005.html
[23:52:42] <tomp> Real Time Simulation and Visualization of NC Milling Processes for ... http://www.cg.tuwien.ac.at/research/publications/1998/Koenig-1998-NC/TR-186-2-98-04Paper.ps.gz
[23:53:17] <tomp> ...for low end computer systems
[23:53:43] <SWPadnos> hmmm - I wonder if that's the same Koenig that did some spline book (I think. I'm pretty sure that name came up in the path planning discussions a year ago)
[23:55:42] <anonimasu> damn
[23:55:53] <anonimasu> if we had a mathematica licence we could try this algorithm
[23:55:57] <anonimasu> http://www.wolfram.com/products/mathematica/tour/page8.html
[23:55:59] <SWPadnos> heh
[23:56:00] <anonimasu> ;)
[23:56:07] <SWPadnos> try octave instead
[23:56:10] <SWPadnos> or is that matlab?
[23:56:14] <anonimasu> no idea..
[23:56:24] <anonimasu> I have neither
[23:56:29] <anonimasu> and I dont know math :D
[23:57:20] <SWPadnos> octave is a FOSS version of either matlab or mathematica. thinking about it, I think it's matlab
[23:57:29] <anonimasu> hm, yeah..
[23:57:33] <anonimasu> mathematica seems heavy :)
[23:57:35] <SWPadnos> then again, other than the name, I don't know the difference :)
[23:57:36] <anonimasu> night
[23:58:20] <SWPadnos> ok, well I know one useless difference. I think Matlab was based on the MAPLE library way back when, and Mathematica had its own. Or it was the other way around
[23:58:24] <SWPadnos> see you :)
[23:59:07] <SWPadnos> http://graphics.cs.uiuc.edu/surface/doc/annotated.html