#emc | Logs for 2008-08-14

[09:47:27] <pjm> morn, just took my 1st cut with my new cnc based on emc2 !!
[09:51:06] <archivist_ub> happy boy
[09:51:13] <pjm> yep i'm well pleased!
[09:51:17] <pjm> it actually all works
[09:51:49] <pjm> http://pjm.dyndns.org/cnc1.jpg is my machine
[09:52:35] <pjm> i just have some counter weight to add for the Z, then it'll be perfect. I used a smaller weight but need 25Kg really
[09:55:06] <archivist_ub> nice looking
[10:04:36] <fragalot> * fragalot jealously stares at pjm
[10:09:49] <archivist_ub> fragalot, the cure is to start building :)
[10:10:47] <fragalot> archivist_ub: i should do a fund raiser first
[10:10:48] <fragalot> :)
[10:11:24] <archivist_ub> I know the money problem, I want to get on with axis 5..........
[10:12:05] <archivist_ub> and the boss is threatening to sell the site here (no job then)
[10:12:08] <fragalot> i'd be happy with 3
[10:12:08] <fragalot> :p
[10:12:13] <fragalot> ocuh
[10:13:14] <archivist_ub> I also need a bigger machine to take on more common work
[10:14:28] <fragalot> what do yo uhave now?
[10:26:48] <archivist_ub> home brew
[10:27:04] <archivist_ub> hmm g83 doc error
[10:29:30] <archivist_ub> R is unexplained in the G83 doc
[10:31:56] <fragalot> archivist_ub: got any pics of it?
[10:32:28] <archivist_ub> www.archivist.info/cnc
[10:32:34] <fragalot> *click*
[10:33:14] <fragalot> ah yeah that one, looks fun
[10:33:29] <fragalot> 4 axis machines rule
[14:41:22] <toastydeath_> ffffff
[14:49:45] <Dallur> white ?
[14:54:23] <skunkworks_> in 24 bit color
[15:17:41] <jymm> monochrome
[15:35:40] <archivist_ub> * archivist_ub adjusts to yuv
[15:39:29] <fenn> still white
[15:40:18] <fenn> or maybe pink
[15:44:37] <skunkworks_> * skunkworks_ is color blind
[15:47:22] <skunkworks_> fenn: for your high powed hexapod.. http://www.electronicsam.com/images/KandT/servostart/boardfinal4.PNG
[15:47:24] <skunkworks_> ;)
[15:47:59] <cradek> make one out of copper already, sheesh
[15:48:21] <skunkworks_> well - I was waiting for jepler to get his machine straitened out.. ;)
[15:48:32] <cradek> I think he successfully made a board yesterday
[15:48:55] <skunkworks_> Sweet :)
[15:48:57] <jepler> yeah, after setting the trace widths to 87mil
[15:49:32] <skunkworks_> why so thick?
[15:49:40] <skunkworks_> depth issues?
[15:50:18] <jepler> no, I think it's due to the runout on the dremel spindle
[15:51:04] <jepler> I think it would have been OK with slightly narrower traces, but I went to the max I could without getting drc errors
[15:53:36] <skunkworks_> was this for the breakout board?
[15:53:47] <jepler> yeah
[15:54:30] <skunkworks_> Neat
[15:57:20] <jepler> it ended up looking about like this in eagle: http://emergent.unpy.net/files/sandbox/switch-breakout.png
[15:58:49] <cradek> does that plug go right to some part of the pluto?
[15:58:52] <skunkworks_> those are some fatty traces :)
[16:00:26] <jepler> cradek: that 5x2 connector is the same pinout as the shrouded header on the pluto, you could connect them with a ribbon cable. I actually used the sparkfun female jumper cables, though.
[16:01:42] <cradek> old serial port thingies are 5x2
[16:54:07] <tom1> archivist: your worm and ring gears are really nice.
[17:01:12] <archivist_ub> tom1, its fun making, with 5 axis it will be easier
[17:02:49] <archivist_ub> they just went out the door 15 minutes ago
[17:25:52] <ra3vat> ra3vat is now known as dimas_
[18:44:45] <cradek> I wonder how small a thread you can cut (in spec) with a plain old TCMT insert with 1/64" radius
[18:45:16] <cradek> hm, they may not have the necessary relief
[18:50:49] <skunkworks_> does anyone know how to change the 'x' size here
[18:50:50] <skunkworks_> http://imagebin.ca/view/s-OWMC4.html
[18:50:50] <cradek> hmm, not very small, since 32 tpi would be the first one you could even think about cutting - and you would be completely on the radius
[18:50:59] <skunkworks_> I cannot for the life of me find the setting.
[18:51:11] <cradek> 'x'?
[18:51:26] <cradek> the thermals?
[18:51:38] <skunkworks_> I want the traces connected to the pad bigger. More substantial
[18:51:56] <cradek> look in drc for thermal?
[18:52:12] <cradek> maybe even just disable them
[18:52:27] <skunkworks_> I will look again.
[18:53:26] <jepler> change the width of the polygon
[18:58:41] <skunkworks_> well cool - I can turn it off.. (I didn't know it was part of the polygon) duh
[18:58:50] <skunkworks_> now lets see if I can make them bigger
[18:59:29] <cradek> that's obscure
[18:59:58] <skunkworks_> you can be sarcastic if you want.. ;)
[19:00:16] <jepler> result of changing width of a polygon: http://emergent.unpy.net/files/sandbox/thermals-12mil-50mil.png
[19:01:23] <skunkworks_> I am seeing the effect. the issue is that it doesn't fill in where I need the 'fill'. I may just turn them off. I will have to play with it a bit.
[19:01:52] <jepler> you could rip up the polygon, then route from the pad into where you know the polygon will fill, then ratsnest. it will keep the wider trace that you added manually
[19:02:12] <jepler> you can do the same without ripping anything up if you use this command: wire 'signalname
[19:02:19] <jepler> (change 'signalname to your signal name, e.g., 'gnd)
[19:07:24] <skunkworks_> Hmm - neat idea. I will have to try that. The thermals are nice because they keep the solder around the pad better. but those are just a little too small.
[19:07:30] <skunkworks_> (in my mind)
[19:24:27] <pjm> evening
[19:38:56] <skunkworks_> jepler: when I rip up the polygon - I cannot run route from the pad. Can I force a route?
[19:45:31] <jepler> skunkworks_: rip up polygon. set polygon_ratsnest off. rats. now you'll get the airwires
[19:45:47] <jepler> then when you're done, set polygon_ratsnest on. rats. that'll put the polygon back
[19:45:55] <skunkworks_> Cool - thanks
[19:48:03] <skunkworks_> that works
[19:50:47] <alex_joni> rats?
[19:50:54] <skunkworks_> rats are good
[19:51:38] <jepler> alex_joni: "ratsnest" is a command in eagle that makes it place "airwires" between pads and traces with the same signal name that aren't connceted by wires.
[19:52:46] <jepler> but by default "ratsnest" also recomputes polygons, which caused the board to end up without any airwires
[19:53:04] <jepler> the "route" command requires you to pick an airwire as the first step
[19:53:12] <jepler> I am sure this is confusing if you're not an eagle user ..
[19:53:14] <jepler> lots of crazy non-words
[19:54:50] <alex_joni> they are words..
[19:55:00] <alex_joni> but yeah, confusing :)
[20:09:06] <alex_joni> good night all
[20:12:07] <OoBIGeye> night alex
[22:22:24] <skunkworks> does this make sense?
[22:22:26] <skunkworks> http://www.cnczone.com/forums/showthread.php?p=489174#post489174
[22:33:34] <cradek> I agree it would be cool, but it can't really work, and it's not a control issue. consider the angle at which the tool meets the workpiece when you're doing this.
[22:35:18] <skunkworks> right - but simple ovals (that are almost round) would work
[22:36:54] <cradek> maybe, barely
[22:38:17] <skunkworks> heh
[22:54:51] <toastydeath> that's called slow tool servo
[22:55:00] <toastydeath> or slow slide servo
[22:55:28] <toastydeath> freeform machine tools do it, and some custom aftermarket controls do as well
[22:56:49] <toastydeath> there's a primitive version of it called form relieving
[22:57:01] <toastydeath> for manual machines
[22:58:34] <toastydeath> it's how camshaft grinders work, too
[23:04:18] <anonimasu> isnt the big issue having enough torque to move your machine fast enough :p
[23:04:38] <anonimasu> and a servo controlled spindle/rotary table
[23:04:56] <toastydeath> what some of those machines do is just keep a constant spindle speed and make the X axis do all the hard work
[23:05:20] <anonimasu> yep.. requires alot of torque
[23:05:21] <toastydeath> so they don't have to deal with having to have full control over the C axis
[23:05:32] <anonimasu> or slower spindle speed ;)
[23:05:38] <toastydeath> lol
[23:07:18] <anonimasu> at 1krpm you need to move the slide in and out 8.3 times per second :)
[23:07:31] <anonimasu> if you are making a lobe ;)
[23:07:37] <toastydeath> depends on the tool system
[23:07:45] <toastydeath> some are short stroke, high frequency
[23:07:47] <toastydeath> but most aren't.
[23:07:48] <anonimasu> *sighs*
[23:07:49] <anonimasu> that
[23:07:51] <toastydeath> most turn at 60-100 rpm.
[23:07:59] <anonimasu> that's besides the point really
[23:08:02] <anonimasu> :)
[23:08:04] <toastydeath> which is?
[23:08:11] <toastydeath> 1" stroke is more than enough for some applications
[23:08:22] <toastydeath> not all forms need the whole X axis stroke
[23:09:03] <toastydeath> i'm not talking about the really high frequency thing, i'm saying there are tool systems that can do 700-1000 hz over moderate ranges
[23:09:18] <anonimasu> where do you find that kind of actuators at a sane price?
[23:09:36] <toastydeath> you don't
[23:09:45] <toastydeath> lol
[23:10:00] <toastydeath> they're using speaker voice coil type things as linear motors
[23:10:20] <anonimasu> you are talking about diamond turning machines right?
[23:10:22] <anonimasu> err lathes..
[23:10:24] <anonimasu> and stuff..
[23:10:32] <toastydeath> no?
[23:10:39] <toastydeath> i mean some diamond turning lathes use that type of system
[23:10:53] <anonimasu> Yes, but for a normal lathe?
[23:10:55] <anonimasu> :)
[23:11:03] <toastydeath> okay well, define "normal lathe"
[23:11:18] <toastydeath> because plenty of places consider 8+ axis lathes "normal" now
[23:11:48] <anonimasu> I'd say diamond turning lathe's arent normal.
[23:12:02] <anonimasu> it's like comparing a cmm to a caliper..
[23:12:03] <toastydeath> but i'm not talking about diamond turning lathes - you brought it up
[23:12:23] <toastydeath> whatever, the bottom line is it's possible, and if you don't care about speed, you could do it on a hobby machine
[23:12:30] <dmess> they are SUPER PRECISION 2 axis only... the ones ive seen anyway
[23:13:04] <anonimasu> modulating stuff at 1000hz is damn fast :)
[23:13:13] <dmess> yes... i WILL do it on my hardinge no problem
[23:13:41] <toastydeath> you don't need a 1000 hz slow tool system to get started in asymmetric turning
[23:13:44] <dmess> slow down... the diamond will work slower..
[23:13:48] <anonimasu> no
[23:14:02] <anonimasu> toastydeath: though voicecoil's will do that kind of speed, but they are fairly shitty in regards to torque.
[23:15:07] <toastydeath> i'm sure that was a design issue for those kind of systems
[23:15:13] <dmess> drive torque with diamond is relativly LOW if you use a small TNR
[23:15:24] <toastydeath> and this isn't diamond turning!
[23:15:44] <anonimasu> yep
[23:15:46] <toastydeath> you can take any old dumb lathe with a good enough spindle encoder and do asymmetric turning if your control supports it
[23:15:59] <toastydeath> those are the only two requirements
[23:16:44] <anonimasu> I wonder if you can turn flats if your machine is fast enough
[23:17:01] <toastydeath> yep
[23:17:19] <toastydeath> or if you slow your machine down enough, all you need is to stay inside what you consider a reasonable following error
[23:17:20] <dmess> broaching is the same.. jus co-ordinate the c-axis ... i made a 16 lobe X .06" rise and fall on a DMG linear about 5 yrs ago..
[23:18:42] <toastydeath> the trick is getting the control to synch the C axis without requiring a bajillion lines of code
[23:18:50] <dmess> the linear would do a 1/2 " flat true to .003 flatness
[23:19:33] <toastydeath> but since the emc folks are already trying to do some pretty neat 5-axis stuff to enable hand programming those forms
[23:19:39] <dmess> i had a pretty neat macro to map the c axis
[23:21:19] <dmess> but the big guys want CAM integration.... so new and exotic g-codes are Bad to their thoughts
[23:21:41] <toastydeath> yar
[23:23:19] <dmess> i have programmed a DMG evolution AT the machine to make a Visitors part @ a show.... tell me what you want... i'l make it come true
[23:23:58] <dmess> siemens 840D LOADED
[23:24:27] <toastydeath> hahahah
[23:24:28] <toastydeath> A+
[23:29:36] <dmess> but i can do the same thing on ANY machine that can keep up with my 12 lines of code