#emc | Logs for 2008-04-17

[00:04:47] <ken> ok...been looking around and what is "stepconfig" ?
[00:10:34] <skunkworks> * skunkworks introduces ken to the docs.. ken meet docs - docs meet ken.. http://linuxcnc.org/docs/
[00:13:18] <ken> hi doc and thanks skunkworks
[00:14:14] <skunkworks> :)
[00:25:31] <ken> ok...i found stepconfig info...but it don't tell me how to get to it(start it)
[00:25:34] <ken> :)
[00:27:01] <ken> never mind...found it :)...guess i need to open my eyes :0
[03:16:38] <SWPadnos_> SWPadnos_ is now known as SWPadnos
[05:47:25] <alex_joni> good morning
[05:49:39] <sweeper> y helo thar
[06:19:15] <micges> good mornig
[06:27:59] <sweeper> yo
[08:26:51] <lerman_> lerman_ is now known as lerman
[12:18:57] <skunkworks_> cradek: http://www.cnczone.com/forums/showpost.php?p=440291&postcount=146
[12:18:58] <skunkworks_> :)
[12:39:57] <skunkworks_> cradek: you only had 5 levels of pwm when you used pwmgen - right? what kind of following error where you getting? Do you remember?
[12:41:07] <skunkworks_> (for your lathe - before the pluto)
[12:55:25] <alex_joni> hi samco
[12:55:47] <jepler> it was probably more like 20 levels -- 1ms/50us
[13:00:36] <skunkworks_> morning
[13:04:12] <cradek> skunkworks_: I don't remember if it was +-5, but it was pretty darn low like that
[13:04:37] <cradek> the FE was a few encoder counts
[13:04:48] <cradek> but I doubt it's a good test bed, considering how light the machine is
[13:06:14] <skunkworks_> still damn cool :)
[13:06:51] <cradek> the pluto setup is so much better!
[13:10:34] <skunkworks_> heh
[13:36:18] <skunkworks_> hmm - I may have upset some engineers.
[13:37:40] <skunkworks_> http://timeguy.com/cradek/cnc/lathe
[13:37:43] <skunkworks_> heh
[13:38:05] <skunkworks_> I mean.. http://www.cnczone.com/forums/showpost.php?p=440411&postcount=147
[13:39:37] <cradek> wow, I got strawmanned in a thread I'm not even participating in? that's amazing.
[13:39:53] <skunkworks_> That happenes a lot ;)
[13:40:20] <skunkworks_> there are probably contracts out on your life.. :)
[13:40:38] <alex_joni> well.. when you start having hitmans
[13:40:44] <cradek> that Tom guy asked me whether emc would reverse command to the motor in order to decelerate faster than coasting to a stop
[13:40:44] <alex_joni> then you should be worried
[13:40:56] <cradek> the answer is definitely yes, and I tried to explain why
[13:41:24] <archivist> wee a hornets nest
[13:41:34] <skunkworks_> yes - I had said the same thing when they where worried the h-bridge wouldn't brake fast enough.
[13:42:05] <cradek> kreutz was being sarcastic, but there is a kernel of truth in his sarcasm: pid can probably cover up a lot of crappiness in a drive
[13:42:43] <skunkworks_> Yes, - in the last post I posted - I tried to smooth it over.
[13:42:44] <archivist> some people allways get the wrong end of the stick, cradek's explanation was plain, to the point and correct
[13:44:01] <archivist> the correcting loop in this case just happens to be in software, and not an "engineers" electronic loop
[13:45:13] <cradek> people who have been in motor control for a long time might not believe how well a fast single pid-ff position loop and a high resolution encoder can do
[13:45:50] <cradek> surely velocity mode amps are still king, but you can get by with less if you throw computing power at it
[13:46:01] <cradek> * cradek shrugs
[13:48:41] <cradek> The guys I admire in these threads are the ones like Tom and Sam who want to actually try stuff and they work for it
[13:49:02] <skunkworks_> aww - shucks
[13:49:06] <skunkworks_> * skunkworks_ blushes
[13:49:08] <cradek> it's easy and cheap to sit back instead, and say it won't work because it's not how everyone else does it
[13:49:26] <cradek> that doesn't contribute much to anything
[13:49:27] <archivist> I only read the stuff you lot point at
[13:50:04] <cradek> I'm excited about the low-cost open hardware designs you're trying to make - combined with free software how can you go wrong?
[13:52:44] <archivist> hmm the Canon colour printer had h bridge servo years ago with a 6809 driving it
[13:56:30] <cradek> well we know a very basic motor drive can definitely work, at least for small stuff
[13:57:12] <cradek> on old machines with a 100Hz servo cycle, not so much
[14:03:53] <archivist> just been digging in my archive, cant find the old source for the printer, should be at home somewhere
[14:36:08] <alex_joni> maybe it's not archived properly
[14:37:29] <archivist> nah archived deeper
[14:41:29] <SWPLinux> many printer motors use a DSP directly controlling several H-bridges
[14:42:35] <sweeper> yup
[14:42:41] <sweeper> all the new ones I've taken apart do
[14:42:48] <SWPLinux> in fact, we did a board layout (15 years ago) that used a square-wave trace around the motor shaft as the encoder - the magnets would excite the trace, and we got a tach signal from that
[14:43:18] <sweeper> only one I've taken apart that DIDN'T use bipolars was an epson dotmatric
[14:43:56] <SWPLinux> the think Kreutz missed is that you still need a good H-bridge design, it doesn't matter if the corrections come from software or hardware
[14:44:05] <SWPLinux> thin
[14:44:06] <SWPLinux> g
[14:44:40] <sweeper> aren't h-bridges like, really really easy?
[14:45:01] <SWPLinux> pretty simple, yes
[14:45:03] <cradek> I think no high-current circuitry is really very easy
[14:45:08] <archivist> we bought in the A1210 and doubled the resolution and speeded it up a bit (burnt a few motors)
[14:45:11] <SWPLinux> but you still have things like dead-time
[14:45:15] <SWPLinux> and switching loss
[14:45:31] <sweeper> yea
[14:45:35] <SWPLinux> and nonlinearity near the switch points
[14:45:39] <SWPLinux> etc.
[14:45:52] <sweeper> I think I want to throw together a few h-bridges to drive the stack of tincans I've got
[14:46:40] <sweeper> the guy building tommeleise (sp?) says he's gotten decent results with them
[20:25:56] <alex_joni> * alex_joni drops a pin
[20:26:28] <skunkworks_> what kind of pin? roll? ;)
[20:28:53] <alex_joni> a voodoo pin
[20:52:16] <fenn> detonator pin
[20:53:50] <alex_joni> safety pin?
[20:54:02] <SWPLinux> bobby pin
[20:54:09] <SWPLinux> firing pin
[20:54:17] <SWPLinux> clevis pin
[20:57:01] <alex_joni> HAL pin ?
[20:57:10] <SWPLinux> oooooh. tricky
[20:57:19] <SWPLinux> cotter pin
[20:57:52] <cradek> lapel flag pin (very big deal, I hear)
[20:58:06] <SWPLinux> yes, in some states