#emc-devel | Logs for 2006-07-22

[15:30:19] <alex_jon1> alex_jon1 is now known as alex_joni
[15:57:51] <anonimasu> jmkasunich: it's a logitech :9
[20:15:11] <jmkasunich> hi all
[20:15:42] <jmkasunich> only here for a bit, busy weekend
[20:19:09] <cradek> hi
[20:19:15] <cradek> it's quiet today
[20:19:26] <cradek> alex was here for a bit, he's having a nice vacation
[20:19:36] <jmkasunich> I need a vacation
[20:24:02] <jmkasunich> ot connect to us.archive.ubuntu.com:80 (, connection timed out
[20:24:02] <jmkasunich> http://us.archive.ubuntu.com/ubuntu/dists/breezy-updates/Release.gpg: Could not connect to us.archive.ubuntu.com:80 (, connection timed out
[20:24:08] <jmkasunich> any idea what thats all about?
[20:24:38] <cradek> domain name pointer alpaca-ii.it.anl.gov.
[20:25:25] <jmkasunich> interesting... is that one of the ubuntu servers?
[20:25:29] <cradek> I guess it's some mirror in .gov, which is down
[20:26:49] <jmkasunich> hmm, that "us.archive.ubuntu.com" is in my sources.list
[20:26:56] <jmkasunich> is it in yours?
[20:27:17] <jmkasunich> perhaps I should be pointing to another mirror?
[20:27:21] <cradek> yes it's in mine
[20:27:56] <jmkasunich> and when you do an apt-get update, you get timeouts?
[20:28:17] <cradek> yes
[20:31:21] <jmkasunich> so, when are you gonna post a movie of your lathe?
[20:31:42] <cradek> well I'm about to break it (rewire it to use the new parport hackery)
[20:32:32] <cradek> then I'll have some outputs available for spindle control
[20:32:38] <cradek> (assuming that parport will do it)
[20:32:51] <jmkasunich> maybe test the parport first?
[20:32:51] <rayh> Hi guys.
[20:32:56] <cradek> definitely
[20:33:00] <cradek> hi ray
[20:33:03] <jmkasunich> did jeff commit the parporrt hackery?
[20:33:06] <jmkasunich> hi ray
[20:33:09] <cradek> no it's in patch form
[20:33:21] <cradek> we would have to figure out how to turn it on and off
[20:34:09] <jmkasunich> I'd be tempted to tackle that now, but I'm going to have to leave soon (anywhere from 5 mins to a half hour)
[20:34:51] <cradek> http://www.ubuntuforums.org/showthread.php?t=220938
[20:36:12] <cradek> jmkasunich: in the short term, just picking another word (like "in" and "out") is probably fine
[20:37:09] <cradek> jmkasunich: with proper hardware it's possible to have bidirectional handshaking on those pins (like estop could be asserted internally or externally)
[20:37:20] <jmkasunich> hmm
[20:37:32] <jmkasunich> I was thinking in terms of having either an in or an out on each of those pins
[20:37:39] <jmkasunich> but I guess you might want both
[20:37:41] <cradek> i.e. there's a pullup, either open collector driver can pull it low
[20:38:26] <cradek> rayh: did you read any of the discussion about tool changing?
[20:38:30] <jmkasunich> the only downside of that for Estop is that "open" means "safe to run", it would be better if estop was normally closed
[20:39:07] <jmkasunich> time for me to go
[20:39:08] <cradek> ok, I don't really want to talk about estop :-)
[20:39:13] <cradek> ok later
[20:39:36] <rayh> No I haven't. Got a link or a copy?
[20:40:03] <cradek> unfortunately it was in irc and I think not logged
[20:40:16] <rayh> Ah.
[20:40:19] <cradek> let me get my wits about me and summarize
[20:40:32] <rayh> Okay.
[20:41:17] <cradek> there's a [EMCIO]TOOL_CHANGE_POSITION that says where to go in absolute coords when you encounter an M6
[20:41:56] <cradek> the ngc spec for M6 is here http://www.isd.mel.nist.gov/personnel/kramer/pubs/RS274NGC_3.web/RS274NGC_34a.html#1000531
[20:42:10] <cradek> it says: During a tool change, the machining center may move the axes, but when the tool change is complete, the axes should all be back to where they were before the tool change began. If the new tool is a different length than the old one, the tool tip will be in a different location.
[20:42:59] <cradek> I think this is unsafe because if you are safe above the work, and switch to a longer tool (even with length compensation on), you can crash into the work
[20:43:14] <rayh> Sure.
[20:43:28] <cradek> so I'm thinking we need a better behavior
[20:43:35] <cradek> I see two options and I want your opinion:
[20:44:01] <cradek> 1. move to TOOL_CHANGE_POSITION, change the tool, and stay there when done - this lets the programmer put in the new length comp (G43 Hx) and then G0 to the desired location
[20:44:33] <cradek> 2. move the controlled point (tool tip) to the same location as before the tool change when possible (might not be possible if you were near a limit before the change)
[20:45:19] <rayh> I thought that 1 above was how I remember it working the last time I ran tool change.
[20:45:43] <cradek> TOOL_CHANGE_LOCATION has been broken for a long time, but 1 is what it does today (I fixed it a couple days ago)
[20:46:15] <rayh> Oh. Okay.
[20:46:20] <cradek> another problem with 2 is that you would have to program G43 for the NEW tool before switching to it
[20:46:42] <cradek> which seems a bit counterintuitive
[20:48:01] <cradek> I'm just nervous about how G43 and M6 interact, and the ngc spec seems like one of the worst of all possible choices, but I need to hear from people who use G43 and toolchangers before I pretend like I know what I'm talking about
[20:48:04] <rayh> Sure. I really favor move to tool change, change tool, set length, move.
[20:48:30] <cradek> ok so you think #1 is best
[20:48:37] <rayh> I believe that a post to the user list would be a good thing.
[20:49:08] <rayh> In the long run, I favor m6 canceling any m43.
[20:49:23] <cradek> after how it's been going recently I'm afraid LOTS of people without any understanding of the issues will respond without taking the time to understand the question
[20:49:58] <cradek> right now I *think* you can change tools with G41/42/43 in effect
[20:50:02] <cradek> I'm not sure what that causes to happen
[20:50:13] <cradek> but maybe we should consider disallowing it
[20:50:14] <rayh> Could use a, "If you are using m6, would you describe your system and comment"
[20:50:51] <cradek> maybe, but still I'm wary :-)
[20:51:00] <rayh> There should be common ways this is done on other controls.
[20:51:17] <cradek> I'd rather hear from you and dave-e and a couple other clueful folk who you might know to ask
[20:51:19] <rayh> Sure. Lot's of crap goes by these days.
[20:51:59] <cradek> I know dave-e uses G43 on his
[20:52:09] <rayh> Let me look in the mitts books I've got and ask roltek how his fanuc handles the problems.
[20:52:21] <cradek> that would be great
[20:52:34] <cradek> tool changing is a can of worms since complex IO and motion can both be required
[20:52:44] <cradek> we don't have a good way of dealing with this now
[20:53:10] <cradek> and the spec itself seems bad (prone to crashing when changing tools)
[20:53:42] <cradek> from what I can see, the spec does not say you have to turn off comp (radius and length) before changing tools
[20:53:53] <cradek> but maybe we could/should require that.
[20:53:57] <rayh> I agree with the increased IO and such that we do need to address issues with it.
[20:54:20] <cradek> but I really hate to break any existing programs, even if we think they are bogus
[20:55:30] <cradek> ok thanks for your thoughts, I'm going to leave it as #1 right now, I think it's at least safe
[20:56:16] <cradek> jepler points out that if you change tools you're probably going to be cutting in a new place anyway - going back to the same location is of limited utility
[20:57:17] <rayh> certainly seems like it to me.
[20:57:47] <cradek> ok glad to hear that
[20:58:34] <rayh> That is pretty off the cuff right now but I'll report back soon.
[20:58:45] <cradek> thanks
[21:00:05] <rayh> Note to todd sent.
[21:04:43] <rayh> Here is one approach I hadn't expected. "Wjem tje tpp; ;emgtj pffset axos os retirmed tp tje referemce [pomt. tje pffset pf tjat axos os camce;ed.
[21:04:48] <rayh> sorry
[21:05:16] <rayh> When the tool length offset axis is returned to the reference point, the offset of that axis is canceled.
[21:08:02] <rayh> This is from meldas50 and 500
[21:08:27] <rayh> both controls allow for multi axis offsets for tool length.
[21:09:00] <rayh> What that means to EMC is that the move to tool change position would cancel any length offset.
[23:18:23] <cradek> rayh: and also cancel radius comp?
[23:41:14] <rayh> Let me look at that.
[23:44:46] <rayh> I don't see them shutting off radius comp or nose shape comp with a tool change.
[23:54:31] <rayh> I don't see anything in either manual here.