#emc | Logs for 2005-06-01

Back
[00:06:09] <anonimasu> night everyone
[00:07:24] <jacky^> night anonimasu
[00:11:22] <fenn> geez pukiwiki was ridiculously easy to set up
[00:11:55] <fenn> tar -zxvf pukiwiki*; links localhost/pukiwiki
[00:12:04] <fenn> *boing*
[00:14:48] <CIA-8> 03rayhenry * 10emc2/configs/emc.tbl: add tool table
[00:15:28] <jacky^> night all
[00:21:01] <fenn> fenn is now known as fenn_afk
[00:42:11] <jerry> how do I deal with radius to end of arc errors ?
[00:43:59] <asdfqwega> Double check your program and geometry...and increase the level of precision you're working with
[00:44:15] <Jymmm> asdfqwega : laser?
[00:44:32] <asdfqwega> ie: Instead of 0.063, go to 0.06250
[00:45:05] <jerry> if i use sheetcam to process the dxf, it works fine, if i use camexpert, i get the arc errors
[00:46:02] <jerry> asdfqwega: so i need to edit the nc file and adjust where ever the error occurs ?
[00:47:21] <Jymmm> Jymmm is now known as Red70sShow
[00:47:21] <Red70sShow> Red70sShow is now known as Jymmm
[00:53:40] <Jymmm> logger_aj bookmark
[00:53:45] <Jymmm> logger_aj, bookmark
[00:53:45] <Jymmm> See http://193.226.12.129/irc/irc.freenode.net:6667/emc/2005-06-01#T00-53-45
[00:55:59] <asdfqwega> * asdfqwega is wrapping his brain around CVS usage
[00:56:10] <asdfqwega> [cranial smoke]
[00:56:29] <Phydbleep> * Phydbleep catches a flying sprocket.
[00:56:57] <Phydbleep> And 17 marbles.. You'll want these back. :)
[00:57:12] <Jymmm> 17? should be 3
[00:57:40] <Phydbleep> Jymmm: 3 of them are yours. :)
[00:58:33] <A-L-P-H-A> wth are you two smoking, and why are you both bogarting it!
[00:59:03] <A-L-P-H-A> cranial smoking, is probably really nasty tasting.
[00:59:16] <Phydbleep> * Phydbleep notices A-L-P-H-A's name on several of the marbles.
[01:01:15] <A-L-P-H-A> anyone wanna right some g-code? make a 1/4" slot, that's 5mm ~ deep, with a 3/16" endmill?
[01:01:27] <A-L-P-H-A> 3 degree ramp down.
[01:13:21] <Jymmm> A-L-P-H-A : Sure. CBD.
[01:13:38] <A-L-P-H-A> CDB?
[01:13:46] <Jymmm> Cash Before Delivery
[01:14:02] <A-L-P-H-A> I'm writing it by hand right now.
[01:14:04] <A-L-P-H-A> hush
[01:14:15] <Jymmm> $20, or $2000 if you want it accurate =)
[01:14:45] <Jymmm> $10,000 if you want it before 2010
[01:38:59] <Jymmm> Jymmm is now known as Red70sShow
[01:38:59] <Red70sShow> Red70sShow is now known as Jymmm
[01:40:55] <Jymmm> A-L-P-H-A did it work?
[01:50:31] <A-L-P-H-A> which?
[01:50:35] <A-L-P-H-A> my computer locked up...
[01:50:40] <Jymmm> your gcode
[01:50:43] <Jymmm> oh
[01:50:55] <A-L-P-H-A> a layer of dust on my heatsink caused my system to lock up. I vaccumed that up... and I should be good.
[01:51:03] <A-L-P-H-A> I just wrote it... haven't tested it yet.
[01:51:11] <Jymmm> layer of dust?!
[01:51:18] <Jymmm> in the case?
[01:51:27] <A-L-P-H-A> just over the head sink... nothing else.
[01:51:33] <Jymmm> DOH!
[01:52:07] <A-L-P-H-A> anyways, my computer locked up, so I took 5 minutes looking for the backup copy.
[01:52:15] <A-L-P-H-A> and that didn't save for 10 minutes, so I had to redo it.
[01:52:19] <A-L-P-H-A> which wasn't much of an issue.
[01:52:19] <Jymmm> I'm tempted to buy this thing @ $1400
[01:52:28] <A-L-P-H-A> thing being?
[01:52:42] <Jymmm> http://k2cnc.com/KT-2514_detail.asp
[01:53:08] <A-L-P-H-A> you know, it REALLY isn't that expensive.
[01:53:37] <Jymmm> The ONLy thing that bugs me is it only has a 3" clearance under the Z
[01:53:46] <A-L-P-H-A> how much do you need?
[01:53:54] <A-L-P-H-A> can't do any deep pockets. :D
[01:53:56] <A-L-P-H-A> <shrug>
[01:54:18] <Jymmm> I want 6", but will deal with 4" - jewelery boxes
[01:54:45] <A-L-P-H-A> Assembly: Some assembly required. Due to the every changing modifications to impove all machines, All assembly of all machines are by phone or email only. Documentation is currently not availible. <-- wtf?
[01:55:00] <A-L-P-H-A> Jymmm, that's Les' business. you wad.
[01:55:12] <Jymmm> what is?
[01:55:18] <A-L-P-H-A> jewelery boxes.
[01:55:23] <Jymmm> engraving them?
[01:55:27] <A-L-P-H-A> yeah, and making them.
[01:55:35] <A-L-P-H-A> though his are like $600!!!
[01:55:44] <Jymmm> I thought he did furniture and turkey calls
[01:56:01] <A-L-P-H-A> and jewellery boxes, and wood signage.
[01:56:40] <Jymmm> Somehow, I SERIOUSLY doubt I'll be any threat to Les' business in any way, shape of form.
[01:56:59] <A-L-P-H-A> I still personally think it's uncool.
[01:57:19] <Jymmm> what exactly?
[01:57:38] <A-L-P-H-A> infringing on someone elses product, that's someone a friend or something.
[01:58:37] <Jymmm> You're kidding right? He's East Coast, I'm West Coast. He's far far better than I'll ever be in respect to wood working.
[01:59:11] <Jymmm> The way you're saying it is that there should only be one jewlery box maker in the coutnry.
[02:00:01] <Jymmm> or one sign maker in the country.
[02:01:31] <Jymmm> A-L-P-H-A Seriously, when Les is around next, ask him what he thinks about it.
[02:02:10] <A-L-P-H-A> <shrug> not my business... though I still wouldn't do it, if I found out. It's between Les and you.
[02:02:16] <Jymmm> A-L-P-H-A If I lived next door to Les and all, that's another story.
[02:02:20] <A-L-P-H-A> Jymmm, I'm not saying it should be one.
[02:02:30] <A-L-P-H-A> I'm saying it should be someone you know's product idea.
[02:03:22] <Jymmm> Signs and jewlery boxes are nothing new. If I design something with style A, and les design one with style B, they never clash.
[02:04:31] <Jymmm> A-L-P-H-A Let me ask you this. What to you make typically?
[02:04:43] <Jymmm> s/to/do/
[02:04:54] <Jymmm> on CNC
[02:05:02] <Jymmm> brackets?
[02:05:07] <Jymmm> tooling?
[02:05:46] <A-L-P-H-A> Me? I use to make Car Antenna's... but the profit went from $25 to like $5. So I stopped. Now I'm doing someone elses design, so I just make their shit for them. Their design, their product, their sale. I just make them.
[02:06:17] <Jymmm> ok, so if I made antennas would ti bother you?
[02:06:46] <A-L-P-H-A> If I were still making them. If they were the same / similar design, yup.
[02:06:57] <Jymmm> what if I made ham radio antennas?
[02:07:13] <A-L-P-H-A> nope. But I see a jewellery box as a jewellery box
[02:07:52] <Jymmm> So there should only be one great composer? one great painter? one great sculptur?
[02:08:52] <A-L-P-H-A> Jymmm, you don't seem to get the point. You know Les. You shouldn't do it, as courteousy. If you didn't know him, that's fine.
[02:09:56] <Jymmm> A-L-P-H-A : Tell ya what, lets pick this conversation back up when Les is around. k?
[02:10:36] <A-L-P-H-A> nope.
[02:11:10] <Jymmm> Well, I'll mention it to Les myself next time I see him.
[02:11:31] <Jymmm> I'll bbl
[02:11:42] <A-L-P-H-A> it's also like, asking can I have the last peice of something at the dinner table.
[02:11:47] <A-L-P-H-A> who's going to object?
[02:16:26] <asdfqwega> Dang it...I need to talk face<->face with someone who has experience with CVS
[02:16:45] <Phydbleep> asdfqwega: What's the problem?
[02:16:49] <asdfqwega> I can puzzle this out, but...grrr
[02:16:57] <Phydbleep> * Phydbleep has cvs experience
[02:17:03] <asdfqwega> I'm trying to set up a local CVS
[02:17:10] <asdfqwega> ...which is easy enough
[02:17:27] <Phydbleep> OK.. And you want it sync'd once? twice a day, right?
[02:17:37] <asdfqwega> But what I want to do is to be able to checkout emc2 at least once a day
[02:17:52] <asdfqwega> and either checkin or import it into my local
[02:18:27] <Phydbleep> Run a local 'control' set and make dos to a different set.
[02:18:44] <Phydbleep> ???
[02:18:52] <Phydbleep> s/dos/changes
[02:19:06] <asdfqwega> Nyce typo ;)
[02:19:30] <Phydbleep> Interference from an rf source I think.
[02:19:39] <asdfqwega> I'm just starting, and I don't want to commit stuff to SF right now
[02:20:19] <Phydbleep> Run 2 local copies.. 1 control set , 1 mod set.
[02:20:26] <asdfqwega> Mostly, I'm trying to figure out branches/tags
[02:21:10] <Phydbleep> OK.. asdfqwega Linux box?
[02:21:21] <asdfqwega> Of course!
[02:21:30] <asdfqwega> All my boxes are linux
[02:21:45] <A-L-P-H-A> units
[02:22:33] <Phydbleep> Konqueror has a CVS view that makes it easy to figure out tags.
[02:24:03] <asdfqwega> How do you use that?
[02:26:48] <Phydbleep> Crap, It's gone..
[02:27:26] <Phydbleep> asdfqwega: gcvs ?
[02:27:42] <SWP_Away> or you can get cervisia, which is a KDE app for managing CVS sandboxes
[02:28:32] <Phydbleep> Cervisia, That was it.. konqueror had a cervisia plugin for a while.
[02:33:18] <Jymmm> asdfqwega Hey I know someoen that can help you with that CVS and said he wouldn't mind.
[02:54:10] <A-L-P-H-A> weyland, did I say the 609 loctite retaining compound works mint? just need to wait another 18hrs for it completely cure.
[02:54:37] <weyland> Hey, 18hrs?
[02:54:40] <weyland> lol
[02:54:56] <weyland> Are you *sure* a press fit wouldn't be better? LOL
[02:55:55] <A-L-P-H-A> who registered his nick and ghosted him?
[02:55:57] <weyland_> hmph, what the heck was that
[02:56:05] <A-L-P-H-A> I think it's Jymm.
[02:56:33] <weyland_> ?
[02:56:49] <Phydbleep> * Phydbleep sends a wheelie-bin chasing Jymmm.
[02:56:58] <A-L-P-H-A> weyland_, freenode has a nickserv service. which someone registered your nick, to play with you.
[02:57:10] <weyland_> Ah
[02:57:19] <A-L-P-H-A> so if you use weyland, someone can type "/ns ghost weyland <password>" and it'll disconnect you
[02:57:28] <weyland_> <shrug> I guess...
[02:58:11] <weyland_> I'm secure in my identity... (:>)
[02:59:50] <weyland_> funny thing is - I saw that message come up about registering your "nick" and tried to, but it kept erroring out
[02:59:57] <weyland_> about a week ago
[03:00:07] <A-L-P-H-A> oh
[03:00:15] <A-L-P-H-A> Jymmm, give weyland the password.
[03:00:25] <weyland_> hahaha
[03:00:30] <weyland_> Jymmm did it?
[03:00:43] <A-L-P-H-A> he mentioned he's steal it... so I'm blaming him.
[03:00:57] <A-L-P-H-A> he's=he's
[03:00:58] <A-L-P-H-A> he's=he'd
[03:01:38] <weyland_> lemme go get my spear....
[03:07:57] <cradek> that's funny, I thought we were adults
[03:08:17] <weyland_> I know *I'm* certainly not...
[03:08:22] <weyland_> ...well I try not to be
[03:08:33] <weyland_> :)
[03:08:48] <cradek> in other news, I guess June 3rd is donut day
[03:08:52] <weyland_> life's too short
[03:10:24] <weyland_> I thought every day was donut day?!?!
[03:12:45] <A-L-P-H-A> NO! Everyday is coffee day
[03:12:51] <cradek> that's the truth
[03:13:16] <weyland_> WEll for me, it's donut AND coffee day
[03:15:58] <weyland_> weyland_ is now known as weyland
[03:18:02] <weyland> WTF?
[03:21:53] <SWP_Away> mmmmmmmm - Donuts!
[03:30:00] <A-L-P-H-A> weyland, that was you 'leaving' no one else.
[03:30:28] <weyland> think I got it figured
[05:03:58] <A-L-P-H-A> http://www.homestarrunner.com/sbemail58.html
[05:03:59] <A-L-P-H-A> hahaha
[05:07:52] <A-L-P-H-A> A physicist, an engineer, and a mathematician are staying in a hotel in separate rooms. A fire breaks out in the physicist's bathroom. The physicist wakes up, sees the fire, does some calculations on his calculator, fills a cup of water, and throws it at the base of the fire putting it out while getting the rest of the bathroom hardly wet at all, and then goes back to sleep.
[05:07:53] <A-L-P-H-A> A fire breaks out in the engineer's bathroom later that night. The engineer wakes up, sees the fire, runs into the hallway and brings the firehose into the bathroom and lets the stream go full blast. After a minute or so, the fire is out, and the bathroom is soaking wet with water dripping everywhere, but the fire is out and the engineer goes back to bed.
[05:07:53] <A-L-P-H-A> A fire breaks out in the mathematician's room. The mathematician wakes up and sees the fire, does some lengthy calculations on paper, lights a match and drops it in a glass of water, says "It can be done", and goes back to bed.
[06:54:40] <A-L-P-H-A> logger_aj, bookmark
[06:54:40] <A-L-P-H-A> See http://193.226.12.129/irc/irc.freenode.net:6667/emc/2005-06-01#T06-54-40
[11:22:46] <jacky^> morning
[11:23:11] <ValarQ> g'day
[11:23:29] <jacky^> hi :-)
[11:24:22] <ValarQ> oh mein got
[11:24:43] <ValarQ> i got a new book today, "The Practice of Programming"
[11:25:34] <ValarQ> and some of the bad examples is really scary
[11:26:19] <ValarQ> stuff like "#define FALSE 1" almost makes me cry
[13:19:23] <rayh> Morning dan.
[13:33:12] <rayh> looks like my chasing the tool table through the interpreter broke things.
[13:33:25] <SWP_Away> SWP_Away is now known as SWPadnos
[13:34:02] <ValarQ> hello Padnos
[13:34:16] <SWPadnos> Hi there
[14:29:19] <anonimasu> hello everyone
[14:29:36] <SWPadnos> I'm sorry, but everyone isn't here
[14:29:38] <SWPadnos> hi
[14:29:40] <SWPadnos> :)
[14:30:04] <anonimasu> ValarQ: oh, if I remember it right thoose were not defined by standard before
[14:30:08] <anonimasu> ValarQ: old habits die hard
[14:30:13] <anonimasu> * anonimasu is finally off work
[14:30:37] <SWPadnos> but the conventions is to use 1 for TRUE and 0 for FALSE
[14:30:47] <anonimasu> ah, didnt read..
[14:30:48] <anonimasu> yeah
[14:30:49] <anonimasu> ;)
[14:30:52] <anonimasu> that's horrid..
[14:30:55] <SWPadnos> heh
[14:31:01] <anonimasu> but still that might be why it's defined in the first place
[14:31:27] <SWPadnos> it's almost as bad as (in EMC), #defining things like:
[14:31:33] <SWPadnos> #define SET_TO =
[14:31:40] <SWPadnos> #define IS ==
[14:31:47] <SWPadnos> #define IS_NOT !=
[14:31:55] <anonimasu> the person that did that define should be stabbed in the eye.
[14:31:57] <SWPadnos> #define AND &&
[14:32:02] <SWPadnos> etc.
[14:32:23] <SWPadnos> it's as though the code was meant to be ported to Pascal or something
[14:32:23] <anonimasu> f-ugly
[14:32:33] <anonimasu> turbocnc?
[14:32:39] <SWPadnos> emc
[14:32:45] <SWPadnos> in rs274ngc.hh
[14:32:47] <anonimasu> yeah, but with the porting ;)
[14:32:58] <SWPadnos> ah - could be
[14:33:13] <anonimasu> oh, but turbocnc has a freaky syntax...
[14:33:15] <anonimasu> ;)
[14:33:17] <SWPadnos> turbocnc compiled with turbo Pascal
[14:33:21] <anonimasu> yep
[14:36:20] <anonimasu> I got some more prototyping to do for work..
[14:36:25] <anonimasu> going to make a mold for a panel
[14:38:30] <rayh> Initially the interpreter was written for about 12 different systems.
[14:38:52] <rayh> So we ought not to poke Dr. Kramer to hard.
[14:38:55] <SWPadnos> did some of them not have a C compiler?
[14:39:32] <rayh> alpha dec ce w3.1 and the list goes way on.
[14:40:20] <anonimasu> rayh: I guess he can keep he's eye then ;)
[14:40:36] <anonimasu> if he _knew_ what he was doing it's nothing wrong with it..
[14:40:36] <anonimasu> heh
[14:40:52] <rayh> Just leave him one of 'em so he can finish up the work on the STEP interpreter.
[14:41:05] <anonimasu> yeah
[14:41:13] <anonimasu> what's good about step
[14:41:15] <SWPadnos> we'll leave both until he's done, then we can decide (based on that) what to do ;)
[14:41:19] <anonimasu> I read a bit about it :)
[14:41:20] <anonimasu> yeah
[14:41:38] <rayh> * rayh knows nothing about STEP.
[14:42:18] <rayh> or ANSI C
[14:42:51] <rayh> My latest c programming book is an autographed K&R first edition.
[14:44:10] <SWPadnos> d00d - u Rock!
[14:45:32] <rayh> If only I could read!
[14:52:49] <anonimasu> aheh
[14:52:52] <anonimasu> *yawns*
[14:59:02] <ValarQ> SWPadnos: is there a "#define SET_TO =" in the emc source?!?!
[14:59:30] <SWPadnos> yes (emc1 - it was changed to = in emc2)
[14:59:42] <SWPadnos> src/emc/rs274ngc/rs274ngc.hh
[14:59:55] <ValarQ> that is quite terrible
[15:00:07] <SWPadnos> wprse than #define FALSE 1, IMO
[15:00:23] <SWPadnos> worse
[15:12:07] <SWPadnos> rayh: do you have a minute to talk about gearchanging (and general IO vs. Motion) issues
[15:13:31] <rayh> Certainly.
[15:13:48] <SWPadnos> cool
[15:14:11] <rayh> * rayh is in a fight to the death with a motion card and emc2 so needs a break.
[15:14:23] <SWPadnos> heh
[15:14:35] <SWPadnos> I'm not sure this will qualify as a break ;)
[15:14:58] <SWPadnos> I think the overall paradigm under which EMC was designed may be wrong for the kinds of machines we're talking about
[15:15:09] <SWPadnos> not too far wrong, but still a little off
[15:15:33] <rayh> no fight there.
[15:15:48] <SWPadnos> I'm not sure what the original motivation for separating IO and motion was
[15:15:59] <rayh> we only got the code for half the split backplane used for the K&T
[15:16:03] <SWPadnos> was it to reduce the amount of RT code?
[15:16:14] <SWPadnos> ?
[15:16:42] <rayh> On that split backplane motion was one half IO the other.
[15:16:54] <rayh> ,IO
[15:17:00] <SWPadnos> ah - so there were two "computers", one doing IO and the other motion
[15:17:11] <rayh> There were two processor modules.
[15:17:25] <rayh> NML handled the task of communication between.
[15:17:48] <SWPadnos> ok - thinking ...
[15:17:58] <Jymmm> * Jymmm smells the smoke
[15:18:04] <rayh> thinking is good.
[15:18:07] <SWPadnos> nope - I just changed oil
[15:18:21] <Jymmm> SWPadnos must be me then =)
[15:18:27] <rayh> I thought the prevailing wind in NE was east northeast
[15:18:27] <SWPadnos> well - there are two options as I see it:
[15:18:48] <SWPadnos> 1) define any task that requires both motion and IO as non-realtime
[15:19:08] <SWPadnos> 2) provide a mechanism for IO and motion to communicate in RT
[15:19:31] <rayh> The second cuts out the task as the locus of control
[15:19:45] <SWPadnos> let's push that question till later
[15:19:47] <rayh> Unless task has it's own rt proc.
[15:19:57] <rayh> k
[15:20:16] <SWPadnos> I'd like to thikn of what the machine needs, then see how well the current system does it, and then decide what to throw away :)
[15:21:01] <SWPadnos> if there needs to be a single point of control, then it has to be RT, because there are RT tasks that sometimes need control
[15:21:32] <SWPadnos> (homing, for example, should be RT)
[15:21:34] <rayh> I don't tend to think of wait or inhibit as necessary a control thing.
[15:22:36] <rayh> Is it sufficient to think of the task controller, when it encounters a home command,
[15:23:03] <rayh> to send motion to rt and also set up a home pin watch?
[15:23:35] <SWPadnos> yes, but how long does it take to receive the home sense, and stop motion?
[15:23:39] <rayh> I understand that it is easier to just write all that into a hal mod and just call it.
[15:24:02] <rayh> Next task cycle.
[15:24:21] <SWPadnos> which is unknown
[15:24:34] <rayh> +-
[15:24:44] <SWPadnos> the task cycle in user space is only guaranteed to have a minimum delay, there is no maximum
[15:24:50] <SWPadnos> (though it shuold be pretty good)
[15:24:59] <SWPadnos> never -, only +
[15:25:06] <rayh> That is why the rt homing and probing routine stores the value when the condition is met.
[15:25:51] <SWPadnos> OK - so there are clever workarounds (hacks) to get around a design that may be fundamentally flawed for the task at hand
[15:26:12] <rayh> You are permitted to think that way.
[15:26:19] <SWPadnos> ty :)
[15:26:33] <rayh> I'm not required to...
[15:27:02] <les> hmm
[15:27:05] <SWPadnos> the reason I look at it this way is because any new "thing" requires changes to the core code
[15:27:05] <les> hi guys
[15:27:09] <SWPadnos> hi
[15:27:32] <SWPadnos> that removes the idea of machine logic being captured in something other than C
[15:28:51] <rayh> One thing that machine tool builders tend to do is store a lot of stuff in parameters.
[15:29:17] <SWPadnos> ok
[15:29:26] <rayh> For example a bool might say whether to look for a operator door close or not.
[15:30:05] <rayh> That model looks a lot like a lookup table rather than c.
[15:30:18] <SWPadnos> OK - I can start looking at it as a well-defined set of options for doing certain machine logic :)
[15:30:26] <rayh> But I've got to admit that there will be a lot of switch or if commands in the c
[15:30:53] <SWPadnos> and every new method requires a change to the core code (and might break something else)
[15:31:07] <rayh> What bothers me and I suspect you as well is that much of these things are done so damned easy in HAL
[15:31:13] <SWPadnos> yes
[15:31:32] <SWPadnos> and the main objection may not be important
[15:31:40] <SWPadnos> (that of a single point of control)
[15:31:55] <rayh> It's almost like what we really need to do is make what looks like a logic engine
[15:32:09] <rayh> bur really tunes hal parms and signals
[15:32:34] <rayh> but
[15:33:02] <SWPadnos> regardless of the implementation, I think the fundamental issue is that there are different places where motion and /or IO need to be controlled, depending on what operation is being performed by the machine
[15:33:32] <SWPadnos> look at it this way - there are different "modes" that a machine operates in:
[15:33:36] <SWPadnos> 1) machining a part
[15:33:39] <SWPadnos> 2) homing
[15:33:43] <SWPadnos> 3) changing tools
[15:33:52] <rayh> That same logic engine can supply task with what it would need to command motion.
[15:34:07] <rayh> okay.
[15:34:13] <SWPadnos> only if there is no task that needs RT motion and IO
[15:34:20] <SWPadnos> (other than machining)
[15:34:48] <SWPadnos> consider threading as well
[15:35:04] <SWPadnos> that's an IO -> motion thing
[15:35:25] <rayh> The only way that rt can do IO motion is if it has both in the same proc
[15:35:49] <SWPadnos> HAL solves that, I think
[15:36:19] <rayh> I consider threading to be a coordinated motion issue more than I consider it to be IO.
[15:36:36] <SWPadnos> the IO part is sensing the spindle rotation
[15:36:44] <rayh> The positions of all, (including spindles) needs to be known to motion.
[15:36:54] <SWPadnos> true
[15:37:05] <rayh> If a spindle is an axis then that is not true. I don't think.
[15:37:36] <SWPadnos> with an encoder, it would be known
[15:37:46] <rayh> The bastardized bridgeport spindle could be thought of as is a special case
[15:37:53] <SWPadnos> but that's one hell of a rotary axis :)
[15:38:11] <rayh> All of the big boys do exactly that.
[15:38:53] <rayh> I asked a guy at K&T about the ability of the test bed to do rigid taping
[15:39:11] <rayh> He didn't think it was in there or that there was a spindle encoder.
[15:39:22] <rayh> Some of the 800 did and some didn't.
[15:39:38] <SWPadnos> hmmm - all you need is a tach sensor on a 3-phase motor
[15:39:49] <SWPadnos> (index only)
[15:40:17] <SWPadnos> the speed is constant intil the motor is overloaded, at shich time it stops due to the overloads popping
[15:40:20] <SWPadnos> which time
[15:40:20] <rayh> Yes but you get one hell of a slip when a 1 inch tap hits the hole.
[15:41:40] <SWPadnos> indeed
[15:41:46] <SWPadnos> I certainly do :)
[15:41:57] <rayh> I favor full spindle axis control and then back off from that to a Sherline.
[15:42:29] <SWPadnos> sure - plan for the most complex case, and pare it down for easier ones
[15:42:55] <rayh> Again I believe that HAL is the easiest way to accomplish that.
[15:43:36] <rayh> for a single processor machine.
[15:43:43] <SWPadnos> I guess the thing that keeps happening (when discussing design) is that we end up getting bogged down in details of current implementations or currently known tasks
[15:44:34] <rayh> Many of the other PC based motion systems start from the premise of a 3 axis cartesian machine.
[15:44:43] <rayh> and then begin to gob stuff on.
[15:44:49] <SWPadnos> yep - and an integrated GUI as well
[15:45:21] <rayh> So true
[15:45:35] <SWPadnos> luckily it's up to the CAM program to translate from a CAD model to an n-axis machine
[15:45:52] <SWPadnos> the machine controller doesn't care what the axes *do*
[15:46:22] <SWPadnos> (unless it's translating from cartesian to stewart platform or something)
[15:46:33] <rayh> I would not want to even begin to think about how an integrated drafting front end might work.
[15:46:57] <SWPadnos> heh - use Rhino CAD to do NURBS modeling, and pass the model directly to the TP :)
[15:47:07] <rayh> The kinematic notion of world view vs joint space is good.
[15:47:11] <rayh> I think.
[15:47:17] <SWPadnos> I agree
[15:47:41] <SWPadnos> I don't know if STEP supports non-catesian coordinates
[15:47:47] <rayh> Dan Falck has been thinking about using Rhino for a lot of years.
[15:47:48] <SWPadnos> cartesian
[15:48:06] <SWPadnos> I don't like it because it's non-parametric (as far as I can tell)
[15:48:24] <rayh> I think that is one of it's claims to fame.
[15:48:31] <SWPadnos> parametric?
[15:48:39] <rayh> non
[15:48:42] <SWPadnos> ah
[15:48:55] <SWPadnos> I prefer the ability to edit geometry :)
[15:49:20] <rayh> I do to.
[15:49:40] <SWPadnos> I like CadMax, though others hate it (and it's Windows only)
[15:49:54] <rayh> Even grabbing a nurb in the middle and dragging scares me.
[15:50:01] <SWPadnos> $295 for true parametric solid modeling
[15:50:26] <rayh> and 200 for the xp to run it on
[15:50:35] <SWPadnos> Win2K, already paid for :)
[15:50:43] <SWPadnos> I may try it on WINE at some point
[15:50:43] <rayh> and 100 for the extra memory for the pile of sw
[15:50:57] <SWPadnos> I've only got 1.5GB in this machine :D
[15:51:17] <ValarQ> that is enough :)
[15:51:18] <rayh> and 200 for the 400gig drive to store the result on.
[15:51:19] <Jymmm> SWPadnos That BETTER be HDD
[15:51:31] <SWPadnos> 120G HD, 1.5G RAM
[15:51:39] <rayh> sorry that was off topic.
[15:51:41] <Jymmm> SWPadnos <--- Bastard!
[15:51:45] <SWPadnos> heh
[15:51:48] <Jymmm> =)
[15:51:51] <ValarQ> * ValarQ got 0.5TB HD and 1GB RAM
[15:52:03] <SWPadnos> but it's PC133 SDRAM
[15:52:20] <SWPadnos> so it takes *forever* to boot
[15:52:20] <Jymmm> * Jymmm is STILL looking for a mobo that supports 32GB ram
[15:52:20] <rayh> and bob was talking yesterday about running synergy on a amd64
[15:52:20] <ValarQ> 333Mhz here ;)
[15:52:52] <SWPadnos> tyan
[15:52:52] <SWPadnos> K8WE, I think
[15:52:52] <Jymmm> rayh dont they still have soem issues?
[15:52:52] <rayh> So SWPadnos were are we on the IO and Motion
[15:52:52] <Jymmm> AMD64's I mean
[15:53:56] <SWPadnos> Not sure - again, there are a couple of ways to go forward
[15:53:56] <rayh> According to the competition, yes.
[15:53:56] <Jymmm> heh
[15:53:59] <rayh> but then they try to hide two loops through as 64
[15:54:03] <SWPadnos> AMD64 is the best bang for the buck, and the best bang overall, in most cases
[15:54:47] <rayh> Unless you step outside the intel box.
[15:54:55] <SWPadnos> http://www.tyan.com/products/html/thunderk8qsdpro.html
[15:55:03] <rayh> then my kid swears by quad g5
[15:55:27] <Jymmm> lol
[15:55:28] <SWPadnos> G5 is great for media processing -AltiVec is far superior to SSE
[15:55:37] <rayh> Keeps the house nice and warm.
[15:55:39] <SWPadnos> don't know about general purpose stuf though
[15:55:42] <SWPadnos> yeah
[15:56:11] <rayh> The two ways to go forward SWPadnos
[15:56:13] <SWPadnos> so - on motion/IO
[15:56:15] <SWPadnos> :)
[15:56:54] <SWPadnos> first, I assume that there will be "new" things that people want to accomplish - ie, that as developers, we can't foresee all possible uses of the software or a machine
[15:57:12] <ValarQ> do i have to have rtai math support?
[15:57:16] <rayh> right.
[15:57:31] <SWPadnos> so, there will need to be a clear decision made as to how those new things will be done
[15:57:46] <rayh> Generalized motion control generalized IO control is one possibility
[15:57:55] <rayh> If we save the dual or more processor notions with separation we
[15:58:10] <rayh> must allow for a single locus of control.
[15:58:23] <rayh> but distributed similar tasks.
[15:58:39] <SWPadnos> NML is non-realtime, correct?
[15:58:55] <rayh> There was an attempt but no it's not as is.
[15:59:11] <rayh> You have paper mills up your way.
[15:59:23] <SWPadnos> OK, so by definition any task that requires coordination between different locations is non-RT
[15:59:35] <SWPadnos> yes - Ticonderoga is a couple of hours from here
[15:59:41] <rayh> 1000' foot long machine that must act as a single device.
[16:00:00] <rayh> several hundred motors, several dozen processors.
[16:00:00] <SWPadnos> yep - printing presses as well (Heidelberg-Harris machines up your way)
[16:00:14] <SWPadnos> (how close to Baraboo are you?)
[16:00:16] <rayh> A few printers
[16:00:37] <rayh> 3 hours drive if the police aren't looking.
[16:00:59] <SWPadnos> OK - but any task that needs to communicate over that distance either does it with wires or it's non-RT
[16:01:35] <ValarQ> * ValarQ still can't get emc to start
[16:01:36] <rayh> Even between processsors it's non or quasi rt
[16:02:09] <SWPadnos> sure - we can say that any machine task that requires coordination vbetween modules (local or remote) can not be RT
[16:02:24] <SWPadnos> but - is that OK?
[16:02:39] <rayh> CAN gets round that by triggering every device from the same clock
[16:02:45] <SWPadnos> is it a problem that should be solved, or is it an acceptable constraint on future machines?
[16:03:16] <SWPadnos> not CAN - ModBus maybe?
[16:03:21] <SWPadnos> or Profibus
[16:03:24] <rayh> Ok
[16:03:56] <SWPadnos> anyway, is it a problem to restrict things in that way?
[16:04:17] <rayh> To my way of thinking we loose little by thinking non rt for connected processes.
[16:04:24] <SWPadnos> OK
[16:04:42] <rayh> If IO is on one box and motion another.
[16:05:00] <rayh> Then IO or the supervisor needs to tell the other to move.
[16:05:17] <SWPadnos> well - the idea behind NML is that you don't need to change the code when you change the topology of the control system
[16:05:34] <rayh> Right.
[16:05:55] <SWPadnos> so we shouldn't look at it as "if they're separate, A, but if they're on the same machine, then B"
[16:06:01] <SWPadnos> it's all A
[16:06:20] <rayh> But B is so easy with HAL
[16:06:25] <SWPadnos> I agree
[16:06:32] <JJB1> Hello
[16:06:36] <JJB1> I'm new to this chat
[16:06:42] <SWPadnos> but you can't run the same process on two machines
[16:06:43] <JJB1> i'm using EMC for the UF PKM
[16:06:45] <rayh> Hi
[16:06:50] <JJB1> I don't know if anybody has heard of it
[16:07:01] <SWPadnos> nope - not me
[16:07:11] <rayh> * rayh either
[16:07:20] <JJB1> it's a stewart platform used at UF
[16:07:41] <SWPadnos> University of Florida?
[16:07:45] <JJB1> so we use EMC to control it
[16:07:46] <JJB1> yeah
[16:07:59] <rayh> Great.
[16:08:12] <JJB1> so the problem is that the machine is on, as the manuals indicate
[16:08:25] <JJB1> but the EMC GUI doesn't seem to recognize it's on
[16:08:34] <JJB1> does anybody know anything about this?
[16:08:38] <rayh> tkemc?
[16:08:47] <JJB1> xemc
[16:09:02] <rayh> You'd have to use yemc
[16:09:09] <JJB1> yeah, it's yemc
[16:09:10] <JJB1> sorry
[16:09:14] <rayh> in order to have access to each joint.
[16:09:24] <JJB1> okay
[16:09:41] <JJB1> the thing is this
[16:09:46] <rayh> Okay the machine is under power
[16:09:53] <JJB1> yeah, the machine is powered up
[16:10:00] <JJB1> adn the ESTOP is on reset
[16:10:31] <JJB1> but I can't get the machine on command to work
[16:10:38] <rayh> servo drives
[16:10:51] <JJB1> would that be a problem?
[16:11:00] <JJB1> servo drives?
[16:11:15] <rayh> servo might drift between estop reset and machine on.
[16:11:29] <JJB1> oh
[16:11:45] <rayh> unless you have the bias set just exactly right on the drive amps.
[16:11:59] <JJB1> oh, i see
[16:12:05] <rayh> You should try running one time with tkemc.
[16:12:14] <JJB1> where can i obtain?
[16:12:18] <rayh> and look at the color of the display numbers.
[16:12:44] <SWPadnos> wouldn't servo drift cause following errors? (and therefore an indication that this is why the machine stopped)
[16:12:44] <rayh> If the numbers are red, then one or more joints are sitting on a limit.
[16:13:05] <rayh> Over time or distance that would be true.
[16:13:22] <rayh> Do you have limit switches on the joints?
[16:13:44] <JJB1> yeah, i have limit switches
[16:13:53] <JJB1> also, where can i find tkemc?
[16:14:20] <rayh> You can start it from the ini file DISPLAY = tkemc
[16:14:26] <SWPadnos> it's included with the EMC source
[16:14:30] <JJB1> oh ok
[16:14:43] <rayh> Since you have defined 6 axes in the ini it will show all of them
[16:15:08] <JJB1> k
[16:15:09] <JJB1> let me try
[16:15:17] <rayh> You can also switch from joint to world mode on the main screen.
[16:15:48] <rayh> SWPadnos: I think that you got to the heart of the issue.
[16:15:59] <SWPadnos> thank you :)
[16:16:07] <rayh> Do we write it for distributed or single proc.
[16:16:13] <SWPadnos> or both
[16:16:22] <rayh> Certainly.
[16:16:45] <SWPadnos> luckily, with HAL you can write modules that work on a single PC
[16:17:00] <rayh> If we write it for distributed, it will work on a single.
[16:17:04] <SWPadnos> right
[16:17:19] <rayh> Not as efficiently as one dedicated.
[16:17:34] <rayh> Unless we are able to do some really fancy config work.
[16:17:46] <SWPadnos> right - and HAL isn't as efficient as a monolithic configured machine either, but it's probably better
[16:18:18] <rayh> Seems like it from what I've seen the last few days.
[16:18:40] <SWPadnos> the "no RT between processes" rule makes sense - you don't know how long ethernet will take to get a packet somewhere (or serial, for that matter)
[16:19:07] <SWPadnos> however, with HAL and a suitable piece of hardware, it may be possible to have RT communication between machines
[16:19:24] <rayh> I could see that.
[16:19:52] <rayh> udp and ping timing
[16:20:05] <SWPadnos> one thing that happens with HAL (as I see it) is that you group various modules together, and they effectively become one process
[16:20:24] <rayh> Yes.
[16:20:29] <SWPadnos> so you're dynamically putting together a simgle process context
[16:20:51] <SWPadnos> if we do a little with HAL to make it more like a PLC, then some RT communication can be accomplished between modules
[16:21:06] <SWPadnos> I was just thinking about it this morning
[16:21:54] <rayh> * rayh is all eyes and ??'s
[16:22:17] <JJB1> okay rayh
[16:22:21] <JJB1> so now this is the deal
[16:22:24] <SWPadnos> if we add a "read" and a "write" function to each module, then the HAL can call all read functions first, then the "process" functions, then the output functions
[16:22:25] <JJB1> i use tkemc
[16:22:39] <JJB1> nothing is red
[16:22:45] <SWPadnos> actually, only hardware drivers need that
[16:22:48] <JJB1> everything look fine
[16:22:53] <rayh> lettters are yellow?
[16:23:05] <JJB1> yeah
[16:23:12] <JJB1> but now i try ESTOP off
[16:23:14] <JJB1> and it doesn't
[16:25:30] <rayh> You using bridgeporttask and bridgeportio or minimilltask and minimillio
[16:27:44] <JJB1> no
[16:30:03] <rayh> Look in the ini and see which are not commented out.
[16:32:20] <fenn> PKM huh
[16:33:14] <JJB1> okay, the bridgeporttask was comented the other wasnt
[16:33:45] <JJB1> should i comment?
[16:34:40] <rayh> Minimill should allow you to come out of estop without doing much of anything external.
[16:35:49] <rayh> Where did you get your emc install from. A bdi? If yes what number?
[16:36:01] <JJB1> okay, minimilltask is not commented, should i leave uncommented? and then brdigeporttask, should i comment?
[16:36:19] <JJB1> i didn't install, it was previosly install, and there is not a lot of documentation
[16:36:36] <rayh> Ah. The story of our lives.
[16:36:40] <SWPadnos> heh
[16:36:45] <JJB1> hehe
[16:37:05] <JJB1> and only one professor knows about it and she doesn't know enough details
[16:38:03] <rayh> Is there a little wrench like icon near the left side down on the bottom panel
[16:38:03] <JJB1> like a gear you mean?
[16:38:03] <cradek> try typing 'uname -a' at a command prompt
[16:38:25] <rayh> good thought cradek
[16:39:19] <Jymmm> cradek : Geeze... quit making sense, you'll scare the noobies away!
[16:39:43] <JJB1> Linux samm2.me.ufl.edu 2.0.36 #1 Fri Jul 20 15:09:23 EDT 2001 i686 unknown
[16:40:14] <cradek> that's an old one!
[16:40:37] <JJB1> yup
[16:40:37] <rayh> 2.12 I'd bet
[16:40:42] <Jymmm> Would him running LIVE hurt anything?
[16:40:43] <SWPadnos> woo - I haven't seen a 2.0.x kernel in about a decade!
[16:40:53] <rayh> How fast is the box it's running on?
[16:40:57] <cradek> pretty sure my rtlinux system says -rtl in the uname
[16:40:58] <SWPadnos> (at least it seems like a decade)
[16:41:17] <rayh> New ones give the kernel version as well
[16:41:34] <cradek> SWPadnos: by reading the screen I see that it's only 4 years old
[16:41:48] <SWPadnos> indeed it is - I noticed that too
[16:42:08] <JJB1> okay guys, so what should i do? take out the gun and shoot myself?
[16:42:13] <rayh> But we don't know how old the disk was when loaded.
[16:42:27] <Jymmm> JJB1 s/gun/tequilia/
[16:42:30] <cradek> pretty sure that's the date the kernel was compiled.
[16:42:30] <rayh> Nah. Lemme talk to the prof.
[16:42:46] <JJB1> she's not around now
[16:42:58] <JJB1> should i set a meeting so that we can discuss the situation?
[16:43:25] <cradek> [chris@max chris]$ rtl-config --rtlVersion
[16:43:25] <cradek> 3.2-pre2
[16:43:27] <rayh> That wold be a good plan.
[16:43:38] <cradek> JJB1: try running rtl-config
[16:43:38] <JJB1> okay, r u usually online?
[16:44:02] <rayh> Email would be the better way to get a time set up.
[16:44:09] <fenn> i would like to talk to her too.. gloria wiens right?
[16:45:21] <rayh> The problem with running live from cd would be configuration and saving it from one run to the next.
[16:45:29] <JJB1> hey, i can't run rtl-config
[16:45:43] <JJB1> yeah
[16:45:44] <SWPadnos> do you need to be root?
[16:45:58] <JJB1> do u guys want 2 meet through e-mail?
[16:46:19] <JJB1> okay, look at this e-mail me ur e-mails 2 bestard1@ufl.edu
[16:46:27] <JJB1> and i'll talk 2 her after lunch
[16:46:47] <rayh> bdi 2.18 shows 03 as the date.
[16:47:05] <rayh> This may be a real live 2.04
[16:47:40] <rayh> We will need to preserve the kinematics files before we do any reinstall.
[16:47:52] <rayh> rehenry@up.net
[16:48:06] <fenn> fenn at sdf.lonestar.org
[16:48:18] <rayh> I'll be out for a while this afternoon. 2:30-5:00 your time.
[16:48:43] <JJB1> okay
[16:48:46] <JJB1> i'll e-mail u guys
[16:48:49] <JJB1> thanks a lot!
[16:48:57] <rayh> You bet.
[16:51:01] <rayh> Question. Shall we do this conversation right here or on another forum?
[16:51:30] <fenn> heh too late
[16:51:50] <fenn> it would be good to try to stay on topic though
[16:52:08] <rayh> and that is a bit of a problem here.
[16:52:23] <fenn> i want to pick her brain, and get her to send me some papers she's written
[16:52:28] <SWPadnos> so - how about those Mets?
[16:52:39] <rayh> Should not take to long.
[16:52:45] <anonimasu> lol :D
[16:52:59] <rayh> fenn: She got one of those ph whatevers after her name.
[16:53:06] <SWPadnos> Doktorrrr
[16:53:12] <Jymmm> if it's done here, at least you'll have a log of the conversation for future ref.
[16:53:24] <SWPadnos> BS - we all know what that means
[16:53:30] <SWPadnos> MS = More of the Same
[16:53:31] <anonimasu> * anonimasu hides in a corner and shuts up
[16:53:39] <SWPadnos> PhD = Piled Higher and Deeper
[16:53:44] <anonimasu> ;)
[16:53:55] <Jymmm> got shovel?
[16:54:00] <rayh> Hey I can pile it as deep as anyone.
[16:54:18] <rayh> We all know that goes without the need to say it.
[16:54:23] <SWPadnos> heh
[16:54:41] <SWPadnos> so - multiple cources of control...
[16:54:44] <SWPadnos> sources
[16:55:10] <rayh> We do need to take the notion of supervisor very seriously however.
[16:55:20] <SWPadnos> yes
[16:55:37] <rayh> I suspect that is why TASK was put were it is.
[16:55:41] <SWPadnos> but the supervisor does hand off instructions to subordinates, and then lets them get to work
[16:55:56] <rayh> So we need a dynamic TASK planner.
[16:56:09] <SWPadnos> and it waits for one instruction to finish before sending the next
[16:56:12] <fenn> dynamic = ?
[16:56:13] <rayh> more dyanmic than we hve.
[16:56:42] <rayh> Able to adjust to the characteristics of the machine it is running.
[16:57:07] <rayh> NML has the advantage of specifying the need to wait or not.
[16:57:07] <fenn> like, to compensate for a dull tool ?
[16:57:35] <SWPadnos> so if one motion command is "change tool", then it will only happen after other motion has stopped, and no new motion will be done until it has finished
[16:57:44] <rayh> Sensing dull tool is a real science these days.
[16:57:51] <SWPadnos> maybe we just need to look at some of the machine logic as motion commands
[16:58:05] <SWPadnos> (that happen to have other things that go along)
[16:58:33] <anonimasu> rayh: I had a look at a paper about it, it's not that advanced
[16:58:54] <anonimasu> rayh: not on the siemens machines atleast
[16:59:07] <Jymmm> s/dynamic TASK planner/Peon Planner/ =)
[16:59:12] <rayh> This could work if we make the TASK aware that it is commanding motion outside of the motion stack.
[16:59:22] <anonimasu> but it requires setup for the material you cut..
[17:00:01] <fenn> anonimasu, it just compares current consumption vs feedrate to a calibrated value right?
[17:00:12] <rayh> One guy I talked to at NIST was defining tool wear by the machine casting harmonics.
[17:00:51] <anonimasu> fenn: yeah somthing like it, they just average the current consumption and calculate it and compare to a table..
[17:00:58] <SWPadnos> just put it on the motion stack
[17:01:01] <anonimasu> for the specific tool..
[17:01:21] <fenn> right, much easier than trying to calculate what it "ought" to be
[17:01:48] <SWPadnos> actually, there's a lot done with detecting bad bearings by doing spectral audio analysis on their sound
[17:01:52] <rayh> SWPadnos: If we could open up the motion stack and insert stuff.
[17:02:32] <SWPadnos> a toolchange (and many other motion and io tasks) must be done in order - motion has to complete, then the change, then more motion
[17:02:41] <rayh> Dynamic motion stack?
[17:02:55] <SWPadnos> so it's reasonable to have a single command stack that includes toolchanges and joint motions
[17:03:00] <rayh> Pass that one by FredP
[17:03:18] <SWPadnos> consider a hierarchy of controllers
[17:03:20] <anonimasu> SWPadnos: not a bad idea
[17:03:40] <SWPadnos> at the top is the GUI, which tells the interpreter to get a file and cut the profile
[17:04:01] <SWPadnos> next is the interpreter, which sends lots of motion commands down the line
[17:04:25] <fenn> (and tool change commands)
[17:04:39] <SWPadnos> next is the machine controller (task / iotask), which also deals with things like turning on coolant and stuff
[17:04:47] <SWPadnos> fenn: yes - and other things as well
[17:05:02] <anonimasu> brb dinner
[17:05:14] <SWPadnos> at each stage, there may be more or fewer commands that get passed down to lower level modules
[17:05:44] <SWPadnos> one command may be serviced at a high level, or be broken into multiple smaller commands for one or more lower level modules
[17:06:04] <rayh> The clue here may be "interp stack" vs "task stack"
[17:06:38] <SWPadnos> yes
[17:07:01] <SWPadnos> but the split may need to be at a lower level than it is now
[17:07:04] <rayh> The task stack could be left intact.
[17:07:15] <rayh> and task uses a pointer to next.
[17:07:30] <rayh> In terms of both motion and io.
[17:07:43] <SWPadnos> or just look at a single level of the hierarchy having multiple modules, and each may be able to add items to lower level module command queues
[17:08:16] <rayh> But then how would IO pass a motion command back to interp stack
[17:09:04] <SWPadnos> make it one stack with a "director" that passes commands to the appropriate module
[17:09:37] <SWPadnos> is interp "interpolator" or "interpreter" in that last sentence?
[17:09:48] <rayh> Interpreter.
[17:10:05] <SWPadnos> the interpreter doesn't need to get the command
[17:10:08] <rayh> It creates a stack several hundred deep as it reads ahead.
[17:10:28] <rayh> So some way, we need to be able to insert stuff.
[17:10:34] <SWPadnos> and those commands are filtered through a subordinate, which would normally pass them on unaltered
[17:10:51] <SWPadnos> if it needs to do extra stuff, it just sends its own conmmands for a little while
[17:11:01] <rayh> Task is the process that takes each item and executes it.
[17:11:06] <SWPadnos> then goes on passing through commands from the higher level
[17:11:38] <rayh> This list is a fifo.
[17:11:47] <SWPadnos> in a HAL-like world, task would be a "manager"
[17:12:09] <SWPadnos> it would have an input command pin, and at least one output command pin
[17:12:25] <SWPadnos> hold on - thinking again
[17:12:49] <SWPadnos> right - OK - got it, I think
[17:13:06] <rayh> * rayh waits breathless
[17:13:20] <SWPadnos> HALtask sends commands out via its output pin
[17:13:40] <SWPadnos> if you have a machine that has a toolchanger, then that output pin is connected to the toolchange module
[17:13:55] <SWPadnos> (this is what we were discussing a few days ago)
[17:14:10] <SWPadnos> the toolchanger can do IO via various HALPins
[17:14:21] <SWPadnos> and it also filters all motion commands
[17:14:36] <SWPadnos> filter meaning that it passes them on unaltered, unless it's doing a toolchange
[17:14:49] <rayh> Nah that makes the "les" problem that much bigger.
[17:15:06] <SWPadnos> which it never does unless the higher level controller has asked it to
[17:15:32] <SWPadnos> so it has I/O and a "HalMotion" output pin
[17:15:50] <rayh> That's closer.
[17:15:54] <SWPadnos> (which John will hate, since it can't be atomic :) )
[17:16:02] <rayh> right.
[17:16:21] <SWPadnos> we may be able to get around that with pointers
[17:16:43] <SWPadnos> you can string together as many of these "filtering" modules as you want - they have no effect unless commanded to do something
[17:16:53] <SWPadnos> (by a higher level controller)
[17:17:09] <SWPadnos> a single controller, since you can't have more than one writer to a HAL pin
[17:17:58] <SWPadnos> so instead of a "wide" hierarchy, with multiple sources of control, it's a "tall" hiererchy, with a single chain of filters
[17:18:22] <rayh> So the motions needed to change a tool would be contained in the change tool item.
[17:18:37] <SWPadnos> yes, that would be a semi-custom HAL toolchange module
[17:18:45] <rayh> That change tool is routed to a change tool module
[17:18:50] <SWPadnos> possibly parameterized, as you had mentioned the "big boys" doing
[17:19:11] <rayh> Okay
[17:19:30] <SWPadnos> the toolchange logic in task would be one possibility, using the same parameters
[17:19:58] <SWPadnos> a "set I/O bit and wait for return bit" module would be used for machines with an external PLC
[17:20:02] <SWPadnos> etc.
[17:20:10] <rayh> So the task module does one level of Machine logic.
[17:20:28] <SWPadnos> yes - it asks for a logic function to be performed
[17:20:48] <SWPadnos> other than that, no - it's done by a HAL (or HAL-like) module
[17:20:57] <rayh> Or even a "ladder" or sequence of logics.
[17:21:44] <SWPadnos> sure - you plug in whatever controller you want
[17:21:44] <SWPadnos> it sits patiently forwarding motion commands until it's asked to do something
[17:21:44] <SWPadnos> then it adds its own commands to the output command stream
[17:22:16] <SWPadnos> and then goes back to forwarding commands in its input stream to its output stream
[17:22:24] <rayh> It would need to be aware of how those additional motion commands
[17:22:35] <rayh> affect the next task in the stream.
[17:22:57] <rayh> But I can see something like that being possible.
[17:23:07] <SWPadnos> and you can chain several of these modules, since they remain dormant until a specific command activates them
[17:23:22] <SWPadnos> there should be no effect on downstream modules
[17:23:46] <SWPadnos> (side effect, I mean)
[17:23:55] <SWPadnos> consider any single command:
[17:24:07] <SWPadnos> it starts at the interpreter (for instance)
[17:24:17] <SWPadnos> take a toolchange command
[17:24:29] <SWPadnos> this gets passed to the first "handler" in the chain
[17:24:42] <SWPadnos> it says "I'm a homing handler - this means nothing to me"
[17:25:04] <SWPadnos> the next one looks at it and says "I'm the tollchanger - this is for me"
[17:25:16] <SWPadnos> it then issues motion and IO commands down the chain
[17:25:46] <SWPadnos> the next handler says "I'm the spindle handler - these motion and IO commands mean nothing to me" - so it passes them on unaltered
[17:26:34] <SWPadnos> each module will only respond to commands meant for it, and any commands issued by it will be meaningless (or intended for) other modules further down the chain
[17:26:41] <SWPadnos> make sense?
[17:26:54] <rayh> Or in the case of spindle orient it says, hey wait I've gotta do something here.
[17:27:02] <SWPadnos> sure
[17:27:10] <les> scuse me...
[17:27:17] <les> is till in stutgart?
[17:27:27] <rayh> yes
[17:27:32] <les> ah
[17:27:33] <SWPadnos> the only issues I can see are performance, and ordering
[17:27:42] <fenn> you gotta think very carefully about how they are ordered
[17:27:46] <les> I think one of his students mailed me about TPs
[17:27:50] <fenn> performance is negligible
[17:28:08] <SWPadnos> you need to be sure that any module is above any other module that it may need to send commands to
[17:28:27] <SWPadnos> we'd have to think about circular control paths, but I think that's not a hard problem
[17:28:36] <SWPadnos> (there probably aren't any)
[17:29:03] <rayh> phone brb
[17:29:07] <fenn> make a list of all the high-level commands you would want to do, and see how it looks
[17:29:45] <fenn> ooh this could be the start of something good
[17:29:53] <SWPadnos> yep
[17:30:15] <SWPadnos> or at least a continuation of something good :)
[17:30:55] <SWPadnos> les: I talked to my mother about the TP issue - she's interested in the math aspects
[17:31:01] <SWPadnos> "sounds fun" she said
[17:31:06] <SWPadnos> what w weirdo :)
[17:31:08] <SWPadnos> a
[17:33:07] <les> SWP: cool
[17:33:48] <SWPadnos> are you familiar with a book on splines by DeBoor
[17:33:59] <les> Oh the about student in stutgart working on the tp...I am trying to get him on this list.
[17:34:07] <les> His Name is Bai
[17:34:27] <les> not familiar with the Deboor book
[17:35:02] <SWPadnos> OK - she had used it in one of her classes (on interpolation or something)
[17:36:16] <les> Bai was looking again at segmentqueue and says the structure SEGMENT is not initialised
[17:36:45] <SWPadnos> oops
[17:36:46] <les> I told him to check thr header file but no joy
[17:37:02] <SWPadnos> initialization doesn't happen in a header
[17:37:16] <anonimasu> it's just the definition of the struct..
[17:37:27] <les> Heh I learned that...have not written in c in years
[17:37:29] <les> heh
[17:39:42] <SWPadnos> though I suppose it is possible with C++ constructors
[17:40:23] <les> I am checking SQ again right now
[17:40:30] <fenn> depends on what you mean by initialization
[17:40:59] <SWPadnos> you can have inline member functions defined in the class declaration
[17:41:22] <SWPadnos> or member functions declared inline in the class definition
[17:41:46] <SWPadnos> (rephrasing)
[17:41:56] <fenn> er, isn't that two different things?
[17:42:17] <SWPadnos> I think so - that's why I rephrased :)
[17:42:36] <fenn> dont you always have to declare member functions in the header file?
[17:43:11] <SWPadnos> you only have to have the prototypes, but it's also possible to put the function definition (implementation) in the header
[17:43:33] <fenn> oh, i never did it that way before
[17:44:12] <SWPadnos> I'm not sure it's considered good practice, but it's possible
[17:44:30] <SWPadnos> for something like a simple get/set function, it can be useful
[17:44:47] <fenn> i'd rather have all the "real" code in one place
[17:45:08] <fenn> why do you need header files anyway? i forget
[17:45:21] <Jymmm> <STDIO.H>
[17:45:23] <SWPadnos> true, but you do double the code size for a member function like int GetColor() {return Color};
[17:45:36] <SWPadnos> source code size, that is
[17:45:52] <fenn> Jymmm, including a header file includes the code, ie #include <stdio.h> also includes stdio.c
[17:46:02] <SWPadnos> nope
[17:46:18] <fenn> no?
[17:46:22] <fenn> well, usually
[17:46:31] <SWPadnos> stdio.o needs to be linked into the final object to have the functions in the program
[17:47:04] <SWPadnos> the header only includes whatever functions and variables are in it
[17:47:10] <fenn> okay i dont know what i'm talking about :)
[17:47:18] <SWPadnos> no problem ;)
[17:47:33] <fenn> need to read up on how linking and compiling works
[17:47:39] <SWPadnos> that's why a lot of variables are declared extern
[17:47:58] <SWPadnos> otherwise there would be a copy allocated for each .c file that includes the header
[17:48:03] <SWPadnos> and you'll get a link failure
[17:48:52] <rayh> fenn: http://www.powells.com/usedbooks/Engineering.4.html
[17:49:48] <fenn> rayh? engineering?
[17:49:51] <SWPadnos> don't think that's the right page :)
[17:50:16] <fenn> oh "software engineering" bah
[17:50:55] <rayh> nah gloria weins
[17:51:14] <SWPadnos> oh - only $78.00
[17:51:28] <fenn> heh
[17:52:07] <fenn> i'm trying to get rid of paper books
[17:52:19] <fenn> got it down to a few little piles
[17:53:05] <fenn> thanks for looking
[17:54:32] <rayh> Impressive person, this.
[17:55:34] <SWPadnos> ooohhh - automatic deburring and finishing methods - hmmmmm
[17:55:37] <fenn> indeed
[17:55:50] <rayh> logger_aj: where is log
[17:55:50] <rayh> I'm logging. I don't understand 'where is log', rayh. Try /msg logger_aj help
[17:56:03] <fenn> logger_aj, bookmark
[17:56:03] <fenn> See http://193.226.12.129/irc/irc.freenode.net:6667/emc/2005-06-01#T17-56-03
[17:56:11] <rayh> thanks.
[17:56:18] <Jymmm> who own logger_aj ?
[17:56:23] <fenn> alex
[17:56:25] <SWPadnos> Alex Joni
[17:56:33] <SWPadnos> aj - get it? ;)
[17:56:35] <Jymmm> ah, he musta changed the mime type
[17:56:57] <SWPadnos> mime type of what?
[17:57:26] <Jymmm> the txt logs.... they are not coming across as TEXT/HTML
[17:57:41] <SWPadnos> hmmm
[17:57:57] <SWPadnos> they were text/plain, I thought
[17:58:06] <Jymmm> either one.
[17:58:20] <Jymmm> either that or he set it up for download instead
[17:58:23] <fenn> a .txt file is a .txt file.. the problem is probably with your browser
[17:58:39] <SWPadnos> no - the type is passed by the server
[17:58:46] <Jymmm> fenn nope, the server can be setup differently
[17:58:55] <fenn> well blow me down!
[17:59:10] <SWPadnos> it is text/plain now - hmmm
[17:59:15] <Jymmm> fenn I can have movie.txt but setup the mime type as movie/avi
[17:59:26] <fenn> how do you do that?
[17:59:35] <Jymmm> fenn mime type in apache
[17:59:43] <Jymmm> httpd.conf
[18:00:11] <Jymmm> or your script application you can send alternate headers to alter the mime type as well
[18:00:45] <Jymmm> Some ppl do that so instead of viewing a PDF online, it's sent to be downloaded instead.
[18:01:42] <SWPadnos> I wish the rdf files had style information embedded (or linked)
[18:01:52] <SWPadnos> I get a big structural view in Mozilla
[18:01:57] <les> hmm was looking for a struct SEGMENT{ and saw nothing in segmentqueue.c
[18:02:22] <SWPadnos> BDI?
[18:02:38] <Jymmm> SWPadnos you can setup Moz to have RDF's use an alternate viewer if you like
[18:02:57] <SWPadnos> haven't done that - what viewer do you recommend?
[18:03:28] <Jymmm> SWPadnos Let me ask someone in the know
[18:03:34] <les> there is a typedef in segmentqueue.h though
[18:03:46] <SWPadnos> les: what version of emc?
[18:03:57] <les> old...
[18:04:15] <les> a year old but SQ was not changed
[18:05:36] <les> typedef struct _seg{stuff} SEGMENT;
[18:07:00] <SWPadnos> yep - what were you looking for in the SEGMENT struct?
[18:07:58] <les> Bai was just trying to find where it was initialized
[18:08:04] <SWPadnos> ah
[18:08:10] <les> looking at my c book...
[18:08:30] <les> It would seem the typedef I saw would do that...
[18:08:44] <les> but it is in the header file...
[18:08:50] <les> not sure though
[18:09:28] <SWPadnos> typedefs don't initialize data
[18:09:32] <les> Hopefully he will get on here so he can talk to you ral programmers.
[18:10:07] <les> typedef only defines them?
[18:10:29] <fenn> * fenn looks around for the real programmers.
[18:10:38] <les> haha
[18:11:11] <SWPadnos> look at sqInitQueue at line 1100 of segmentqueue.c
[18:11:22] <les> k
[18:12:59] <SWPadnos> that's part of the initialization - I'd assume that any module that calls a function in segmentqueue.c has initialized anything it passes in
[18:13:16] <les> don't find it...(no line# on this text editor)
[18:13:44] <SWPadnos> there may be a couple of instances where something is uninitialized though - I didn't look very carefully
[18:13:53] <SWPadnos> search for sqInitQueue
[18:14:07] <les> I did
[18:14:14] <les> let me check spelling
[18:14:18] <SWPadnos> in segmentqueue.c?
[18:14:22] <SWPadnos> and capitalization
[18:14:49] <SWPadnos> this is copied/pasted: sqInitQueue
[18:14:51] <les> seems ok...right file
[18:14:53] <les> ??
[18:15:39] <SWPadnos> I'm looking at the emc1 source (pulled from CVS a few days ago) -they could be different
[18:15:55] <les> found it ...searched wrong way
[18:16:02] <SWPadnos> heh
[18:17:15] <SWPadnos> there may be some variables that have "delayed initialization" - it's a speed improvement if certain code won't get executed
[18:17:24] <les> ok
[18:17:33] <les> it's beyond me...
[18:18:46] <SWPadnos> heh - stick to the complex math ;)
[18:19:04] <Jymmm> very complex math
[18:21:52] <les> ha ok
[18:22:42] <Jymmm> logger_aj, bookmark
[18:22:42] <Jymmm> See http://193.226.12.129/irc/irc.freenode.net:6667/emc/2005-06-01#T18-22-42
[18:25:03] <les> well better get out to the shop...damn cpu fan in the cnc box is making noises
[18:25:14] <SWPadnos> heh
[18:25:19] <les> seems no one makes a fan that lasts more than about a year
[18:25:22] <SWPadnos> how can you tell with the machine running?
[18:25:30] <Jymmm> les stuff it with a blanket, no more noise (or cnc for that matter)
[18:25:45] <les> Servos are dead quiet if the spindle is off
[18:25:47] <SWPadnos> get a ball bearing fan, oil it from time to time
[18:26:07] <SWPadnos> I'd bet that the oned from PC Power and Cooling last a while
[18:26:10] <SWPadnos> ones
[18:26:42] <les> I used a cpu fan ball bearing for my encoder product
[18:26:51] <les> about $.25 each
[18:26:58] <les> quan 10,000
[18:27:07] <les> which fits in a shoebox
[18:27:27] <SWPadnos> the balls themselves, or the whole bearing?
[18:27:40] <les> the whole bearing...abec-1
[18:27:42] <SWPadnos> (is it actually a separate piece?)
[18:27:45] <SWPadnos> wow
[18:27:51] <fenn> you spent $2500 on bearings?
[18:28:06] <les> ITW
[18:28:14] <les> I designed it for them
[18:28:28] <fenn> oh
[18:28:29] <Jymmm> It's amazing what is inside a cpu fan if you ever take one apart.
[18:28:32] <les> and yeah...they have about 100,000 units/yr sold
[18:29:22] <les> one difference....I specify a 100% grease fill
[18:29:41] <les> normally that would churn and overheat
[18:30:05] <rayh> be back later
[18:30:27] <les> but must use it in these because in some cases the balls don'teven travel as much as one ball space
[18:30:38] <les> so grease would get starved
[18:31:07] <Jymmm> WTH? A CD-ROm is included with a porter cable router?!
[18:31:16] <les> ?
[18:31:29] <SWPadnos> so you can attach it to the spindle and see it explode
[18:31:35] <les> haw
[18:31:37] <fenn> swp that's what i was gonna say!
[18:31:40] <Jymmm> last item in the list http://www.porter-cable.com/index.asp?e=547&p=4857
[18:31:41] <SWPadnos> heh
[18:31:58] <Jymmm> hey cdroms spin at like 54K rpm
[18:32:02] <fenn> actually exploding cdroms are pretty scary
[18:32:14] <SWPadnos> doesn't every red-checkered-flannel-shirt-wearing workshop guy have a computer?
[18:32:26] <SWPadnos> no - CDs explode at around 20K
[18:32:29] <Jymmm> you mean overall's dont ya?
[18:32:32] <anonimasu> :)
[18:32:55] <SWPadnos> nope - do you remember the Sears Craftsman tools that had digital readouts on them?
[18:33:11] <SWPadnos> (like a table saw with a digital depth of cut readout)
[18:33:41] <SWPadnos> they couldn't sell them,and wondered why
[18:33:48] <les> Just buy the motor...no stand...saves money
[18:34:04] <SWPadnos> so they did market research, and found that their customers don't like "newfangled gadgets"
[18:34:14] <Jymmm> lol
[18:34:16] <SWPadnos> and they wear red checkered flannel shirts
[18:34:43] <SWPadnos> so they got rid of the lasers and stuff (and they're just coming back now)
[18:34:55] <anonimasu> hm, brb
[18:35:10] <fenn> lasers are only useful for contractor type stuff.. not accurate enough
[18:35:25] <fenn> the spot is too big
[18:35:27] <Jymmm> LOL lasers my ass. I just bought a circular saw wiht a laser from sears. the laser was off and no way to calibrate it
[18:35:34] <SWPadnos> easier to align a cut on a radial arm or slide compound miter saw though
[18:35:37] <les> I am planning to put dros on much of the woodworking stuff now that they are cheap
[18:36:02] <fenn> and you have to turn the saw on to get the line, right?
[18:36:12] <SWPadnos> probably
[18:36:25] <Jymmm> fenn no, seperate button
[18:36:29] <SWPadnos> I don't have one (a laser), so I don't know
[18:37:12] <Jymmm> but they say to not leave it on after your done cutting, but I kept forgetting to turn it off
[18:37:41] <Jymmm> and the picture of the laser beam on the box lies.... it's REALLY tiny and short too
[18:37:44] <les> The fuzzy laser lines I have seen are no use to me
[18:38:51] <SWPadnos> maybe you need glasses? ;)
[18:38:57] <les> heh
[18:39:41] <les> I will admit that they are good on a sawmill
[18:40:05] <SWPadnos> I'd love one for my DW708, but I can get pretty accurate with it now
[18:40:24] <les> chop saw?
[18:40:37] <SWPadnos> yep - 12" slide compund miter
[18:40:41] <SWPadnos> compound
[18:41:27] <les> I may have to pick one up....have to cut lots and lots of cherry to specific lengths for the calls
[18:41:39] <les> sometimes use a chain saw!
[18:41:47] <SWPadnos> it's a great saw
[18:42:05] <les> I will have to check it out
[18:42:12] <SWPadnos> 90 degree cuts up to 4x10 or so
[18:42:19] <les> many models I see seem flimsy
[18:42:56] <SWPadnos> this doesn't seem that way, but I'm coming at it from the other end of the spectrum :)
[18:43:27] <les> Yeah industrial use is a different ball game
[18:43:40] <SWPadnos> this is a contractor saw
[18:43:56] <les> Well that could see some heavy use
[18:44:10] <fenn> i got a HF 10" compound slide to use for cutting aluminum stock
[18:44:11] <SWPadnos> 15A motor, 12" blade, 12" crosscut capacity, 51/4 inch thickness
[18:44:30] <les> hmm
[18:44:35] <les> let me hunt that up
[18:44:48] <fenn> plan on adding a feedscrew to control cut feed
[18:44:49] <SWPadnos> swings left and right in both roll and yaw
[18:45:24] <SWPadnos> http://www.dewalt.com/us/products/tool_detail.asp?productID=4741
[18:46:12] <les> good review on epinions for what it's worth
[18:46:26] <SWPadnos> good review on #emc as well :)
[18:47:09] <les> uses thopson type linear bearingd huh?
[18:47:12] <SWPadnos> I've rebuilt a country porch with it
[18:47:17] <les> thompson
[18:47:24] <fenn> yeah i really like my roommate's dewalt 12" non-sliding chop saw
[18:47:34] <SWPadnos> two sides of the house, about 8 feet deep, and 20 or so feet per side
[18:47:54] <SWPadnos> did both the deck (and supports) and the entire roof (other than plywood) on that saw
[18:47:56] <les> street price for the 708?
[18:48:01] <SWPadnos> $550-600
[18:48:08] <fenn> yow
[18:48:18] <SWPadnos> Home Depot or Costco both have it in that price range
[18:48:18] <les> heh
[18:49:39] <les> wonder how square it can make a cut
[18:49:54] <SWPadnos> not as good as a CNC, but it's pretty good
[18:50:20] <SWPadnos> it's adjustable as well (bevel and miter)
[18:50:51] <les> The shaper, table saws, planer etc usually do about +/- .005" or so
[18:51:10] <SWPadnos> .005 over what distance?
[18:51:38] <Phydbleep> SWPadnos: Whatever happened to using a handsaw and mitre-box?
[18:51:46] <SWPadnos> that's so 1800's
[18:51:50] <fenn> only problem with those dewalt saws is clamping little parts
[18:51:56] <les> Well like ripping....it's usually that good or better over the whole plank if it is joined straight
[18:52:04] <SWPadnos> any big saw will have that problem, I think
[18:52:05] <Jymmm> I'd say 1700's
[18:52:13] <les> same with the shaper
[18:52:17] <Phydbleep> SWPadnos: Yeah, But it's quiet and I can remodel the bahroom without waking the wife and kid. :)
[18:52:26] <les> but...they have constant power feed
[18:52:30] <SWPadnos> wife and kid - that's so 1600's
[18:53:02] <fenn> SWP you mean 1950's
[18:53:10] <SWPadnos> no - 1590s
[18:53:17] <fenn> 1600s was more like family + extended family
[18:53:22] <SWPadnos> or 0159's
[18:53:31] <Jymmm> 1590 BC
[18:53:39] <SWPadnos> 9510 BC
[18:53:47] <fenn> bah
[18:53:55] <fenn> fenn is now known as fenn_afk
[18:54:01] <SWPadnos> family units have been around for a lot longer than power saws :)
[18:54:16] <Phydbleep> * Phydbleep goes back to 24,000BC looking for a functional vimiana..
[18:54:31] <les> heh
[18:54:34] <SWPadnos> drop by when you get back
[18:54:48] <les> back to the shop for me
[18:55:26] <Phydbleep> 26,000 years.. That's longer than anyones ever been gone before.. :)
[18:55:50] <SWPadnos> longer than I can remember
[18:56:21] <Phydbleep> SWPadnos: See there.. If you keep doing that you get short term memory something..
[18:56:32] <SWPadnos> who are you?
[18:56:33] <Jymmm> SWPadnos Yeah, you aren't you in the twilight zone much of the time? I'm sure they have some time machine there.
[18:56:34] <Phydbleep> Hehehe.. :)
[18:56:54] <SWPadnos> just call me Rod
[18:56:54] <Phydbleep> SWPadnos: Who me?
[18:57:00] <SWPadnos> where's my Reel?
[18:57:12] <SWPadnos> a rod = 16.5 feet you know
[18:57:14] <Jymmm> SWPadnos in your fly!
[18:57:26] <SWPadnos> all tied up
[18:58:55] <SWPadnos> well - I should go and do mundane, house-related crap
[18:58:58] <SWPadnos> see ya all later
[18:59:06] <SWPadnos> SWPadnos is now known as SWP_Away
[19:04:15] <Jymmm> horseshit?!
[19:05:27] <Jymmm> Jymmm is now known as MrAsshole
[19:05:32] <MrAsshole> MrAsshole is now known as Jymmm
[19:06:44] <Jymmm> I'm looking at buying a router, but I dont get the auto-release thing "Auto-release collet system ends "frozen" bits"
[19:10:34] <anonimasu> * anonimasu yawns
[19:13:14] <anonimasu> :)
[19:23:53] <Phydbleep> Jymmm: It means you dont have to remove the nut and fight to get the collet and bit back apart.. It probably has a tapper-bar into the top of the spinder.
[19:24:06] <Phydbleep> spindle
[19:24:44] <Jymmm> Hmmm, they say it comes with two wrenches
[19:24:50] <Jymmm> (if that means anything)
[19:25:18] <Phydbleep> Prolly that it has no spindle lock button or 1 is a spare.
[19:26:35] <Phydbleep> asdfqwega! :)
[19:26:50] <Jymmm> no mention of a spindle lock, but I liked that feature
[19:37:55] <Jymmm> Sears has a kit for $99, but it's only 1/4" collet
[19:52:22] <anonimasu> ok
[19:58:35] <Jymmm> anonimasu ?
[19:59:01] <anonimasu> nice :)
[19:59:10] <Jymmm> anonimasu : I bought a sears 2hp router, but it only has a 1/4" collect on it
[19:59:28] <Jymmm> I can still return it, and buy something a lil better
[19:59:28] <anonimasu> ok
[19:59:36] <anonimasu> hm, wont 1/4 do?
[20:00:23] <Jymmm> Yes, but I would liek to buy/make a 2-3" extension for it and 1/2" to 1/4" seems a whole lot better for that
[20:00:45] <Phydbleep> Jymmm: I'd go for something like A-L-P-H-A is doing with the heavy bearings and bolt the router to the top.
[20:01:34] <Jymmm> Phydbleep I'm looking and mounting it to this http://k2cnc.com/KT-2514_detail.asp
[20:01:45] <Jymmm> s/and/at/
[20:01:56] <Jymmm> oh, asdfqwega is on satellite
[20:02:54] <asdfqwega> Yes, and satellite internet really bites the bag
[20:02:59] <Jymmm> I already have this for it --> http://www.xylotex.com/3AxSysKit.htm
[20:03:03] <Jymmm> asdfqwega the lag?
[20:03:20] <asdfqwega> Almost, but not quite, better than nothing
[20:03:38] <Jymmm> asdfqwega whos the signal strength?
[20:03:41] <Jymmm> how's
[20:03:43] <Phydbleep> Jymmm: Bleah! I was thinking more like http://www.lathes.co.uk/globe/
[20:03:57] <asdfqwega> For a tcp/ip connection, packets have to make a 30K mile round trip...twice
[20:04:25] <Jymmm> asdfqwega oh joy
[20:04:37] <Jymmm> asdfqwega too bad they dont support ATM/SONET
[20:05:31] <asdfqwega> And for how much it costs, I'm looking into putting up a tower to get a line-of-sight wifi shot to the nearest town
[20:06:02] <Jymmm> asdfqwega Heh, well if you want todo microwave instead of wifi, let me know
[20:06:29] <Jymmm> range is apx 30KM
[20:06:32] <Jymmm> LOS
[20:06:52] <asdfqwega> Jymmm...I'd do wifi because of the licensing and low cost
[20:07:22] <Jymmm> asdfqwega apx distance?
[20:07:49] <asdfqwega> 8-11 miles
[20:07:59] <asdfqwega> I'm on top of hill
[20:08:20] <Jymmm> asdfqwega do you know someone on the other end?
[20:08:32] <asdfqwega> Or I'd do the Sonja!
[20:08:47] <Jymmm> nfc on sonja
[20:08:49] <Phydbleep> asdfqwega: Use 802.11g cards and a yagi or dish on each end.
[20:08:54] <asdfqwega> There is someone, but it'd take some convincing
[20:09:07] <Jymmm> asdfqwega forget that
[20:09:12] <asdfqwega> Phydbleep: Already know that
[20:09:26] <asdfqwega> Yagi - no. Dish - yes
[20:09:50] <Jymmm> asdfqwega A or G ?
[20:10:16] <Jymmm> A is a lot emtier
[20:10:19] <Jymmm> emptier
[20:10:23] <asdfqwega> Either one, it will be a broader pipe than the endpoint connection
[20:10:48] <Phydbleep> asdfqwega: How many others within LOS could you drag in and micro-link? You'll already have 75% of it done once the tower is up.
[20:10:49] <Jymmm> asdfqwega can you get your hands on a 2+ ft dish?
[20:11:26] <asdfqwega> Heh... I already have two old Direcway dishes laying around :P
[20:11:36] <Jymmm> lol
[20:12:00] <asdfqwega> Once I make the link, I'd have a third available <g>
[20:12:01] <Phydbleep> asdfqwega: Elipticals?
[20:12:03] <Jymmm> I dont think a diretway LNB is gonna work =)
[20:12:30] <Phydbleep> asdfqwega: I mean offset lnb.
[20:12:36] <asdfqwega> Phydbleep: Parabolic, elliptical in shape
[20:12:46] <Phydbleep> asdfqwega: Ow.. :\
[20:13:13] <asdfqwega> Jymmm: Don't use the LNB...replace with a wifi antenna
[20:13:37] <asdfqwega> Already did that with one...it's what I'm using right now to connect between house and shop
[20:13:55] <Phydbleep> asdfqwega: You'd be better off with a 7 6cm center-fed.
[20:14:05] <Phydbleep> 76cm
[20:14:11] <Jymmm> I think your biggest problem is gonna be Tx power more than anything else.
[20:14:14] <asdfqwega> Then, turn it upside down so you can actually aim it low enough
[20:14:34] <asdfqwega> No, problem is with gain
[20:15:00] <Phydbleep> asdfqwega: guy two counties over with a noisy shaver?
[20:15:01] <Jymmm> asdfqwega you planning on staying legal?
[20:15:03] <asdfqwega> FCC regulations permit certain ratios of power/gain
[20:15:40] <Phydbleep> EIRP.
[20:16:04] <asdfqwega> If I wasn't planning on a LOS shot into the middle of a town, I wouldn't bother with legal
[20:16:52] <asdfqwega> Since I am, it would be best if my signal didn't overload and smoke all other wifi equipment in the area ;)
[20:17:05] <Phydbleep> asdfqwega: So how'd you figure LOS? Ladder or trampoline?
[20:17:05] <Jymmm> but 1 W EIRP with such a bandwidth isnt' very much over 10miles
[20:17:53] <asdfqwega> Then, there's always trying out Sonja
[20:18:01] <Jymmm> sinja?
[20:18:21] <Jymmm> heh, you near any freeways?
[20:18:54] <asdfqwega> lemme find the link
[20:25:58] <asdfqwega> Ah...misremembered it
[20:26:11] <asdfqwega> Ronja optical ethernet
[20:26:32] <Jymmm> asdfqwega clouds/rain/fog
[20:26:34] <asdfqwega> http://ronja.twibright.com
[20:26:53] <asdfqwega> Big whoopee...weather already causes outages
[20:27:16] <Jymmm> wasnt' aware it was THAT big of an issue on satcom
[20:28:42] <Jymmm> asdfqwega Heh, feel like digging a 10mile trench to lay down soem glass?
[20:28:45] <asdfqwega> Weather can affect satcom at two points - either here, or at the NetOp Center
[20:29:24] <anonimasu> hm, you could do laser communication
[20:29:42] <Jymmm> with glass or ronja?
[20:29:50] <asdfqwega> Well, laser is a bit harder - power/cost/efficiency
[20:30:12] <Jymmm> if you can lay glass, I KNOW where to get the lasers
[20:30:12] <asdfqwega> I could more easily scale up the LED transmitter, or even make it an array
[20:31:13] <Jymmm> if you want to do LOS with a laser, you wouldn't need much more than 5W I'd suspect
[20:31:26] <anonimasu> yep
[20:31:36] <Jymmm> if even that, maybe 2W
[20:31:48] <asdfqwega> 5W would cost quite a bit
[20:32:04] <cradek> back in the bad old days, I did load balancing over three phone lines a few times.
[20:32:05] <asdfqwega> And modulation would be a whole engineering project by itself
[20:32:08] <Jymmm> 5W CUTTING would be, but you dont need that,
[20:32:09] <cradek> it actually worked pretty well.
[20:33:38] <asdfqwega> I can more easily get the optics/LEDs for 1-5 watt - and get it in my choice of spectra
[20:33:40] <Jymmm> asdfqwega http://www.n1bug.net/tech/laser/alc_wa6ejo.html
[20:34:57] <asdfqwega> Jymmm, I already know about low-speed optical communcation
[20:35:02] <Jymmm> asdfqwega I think I'd make it a serial link - KISS
[20:35:07] <asdfqwega> Ronja is about 10Mbit full duplex
[20:35:15] <Jymmm> ah
[20:35:36] <Jymmm> asdfqwega that link was for the resouces in it. MWK
[20:37:31] <Jymmm> Here we go.... The system will be capable of transmitting 1.2 Gigibits/second over a distance of 1800 kilometers.
[20:38:26] <Jymmm> asdfqwega go get a job here so you can get employe discount http://www.mrv.com/products/line/terescope.php
[20:38:30] <asdfqwega> What percentage of the GNP does it cost, and how much of a vacuum does it need to operate at that distance
[20:38:57] <Jymmm> laser communications for a lunar base http://www.asi.org/adb/04/03/02/laser-com.html
[20:39:08] <asdfqwega> Zigactly
[20:40:49] <Jymmm> 2w diode lasers aren't that expensive
[20:41:41] <Jymmm> I think the optics will be worse.
[20:42:28] <asdfqwega> Still, wifi is a better choice for an initial attempt
[20:43:42] <Jymmm> Yeah, I'd go 802.11A personally
[20:45:18] <Jymmm> The only thing you'll really be compeating with is radar guns and those automatic door openers at the grocery stores
[20:48:39] <alex_joni> g'evening
[20:48:47] <Jymmm> howdy alex_joni
[20:48:59] <alex_joni> hey Jymmm
[20:49:18] <Jymmm> alex_joni question.... with logger_aj, did you chang ethe mime type?
[20:49:32] <alex_joni> nope.. why?
[20:50:05] <Jymmm> I used to be able to read the txt logsin my browser, now it tries to find a viewer or dl them
[20:50:12] <alex_joni> firefox?
[20:50:15] <Jymmm> Moz
[20:50:19] <alex_joni> same sht
[20:50:21] <alex_joni> ;)
[20:50:30] <alex_joni> did the same thing for me too
[20:50:42] <alex_joni> but I told it to use firefox as the external ap
[20:50:44] <alex_joni> and it works
[20:51:06] <Jymmm> let me look at the headers....
[20:51:11] <Jymmm> logger_aj, bookmark
[20:51:11] <Jymmm> See http://193.226.12.129/irc/irc.freenode.net:6667/emc/2005-06-01#T20-51-11
[20:53:39] <Jymmm> totally weird
[20:54:03] <alex_joni> it's a .txt file
[21:00:10] <alex_joni> anyways.. long day today
[21:00:17] <alex_joni> was at a customer
[21:00:23] <alex_joni> just got home (23:50)
[21:00:29] <alex_joni> going to bed..
[21:00:35] <alex_joni> bye Jymmm & all.
[21:03:38] <Jymmm> G'Night (belated)
[21:34:59] <fenn_afk> boy the ronja looks pretty primitive
[21:49:26] <SWP_Away> SWP_Away is now known as SWPadnos
[21:49:43] <rayh> Hey SWPadnos What did I miss.
[21:50:06] <SWPadnos> don't know - I did some errands shortly after you left :)
[21:50:11] <SWPadnos> just got back
[21:50:18] <rayh> Ah Okay.
[21:50:40] <rayh> Ben thinking about that more capable task manager and that would work. I think.
[21:50:53] <SWPadnos> cool - I think it bears further discussion, for sure
[21:51:20] <SWPadnos> I think I'll have a chance to write the gearchanger HAL module this week as well
[21:51:24] <rayh> The interpreter does not need to know all the details of the motions added in by task
[21:51:52] <SWPadnos> right - it only has to know where the tool will be after the function is completed (much the same as a linear move)
[21:52:07] <rayh> It would only need to know where it will be when its next block is pulled off the stack.
[21:52:15] <SWPadnos> actually, where it is, and how fast it's moving :)
[21:52:20] <rayh> You beat me.
[21:52:23] <SWPadnos> heh
[21:52:34] <SWPadnos> just finishing the pot of coffee :)
[21:52:38] <rayh> No not even that.
[21:52:50] <rayh> Just position like it does now.
[21:53:09] <SWPadnos> true - the TP sits underneath, and that will be keeping tack of velocity
[21:53:17] <rayh> If it encounters a tool change, it should know where it will be when that change is finished.
[21:53:49] <rayh> the interp stack is a deck of cards.
[21:54:01] <SWPadnos> so, the only issue would be getting to the toolchange position safely
[21:54:13] <SWPadnos> (ie, without crashing the tool into the workpiece or fixtures
[21:54:14] <SWPadnos> )
[21:54:42] <SWPadnos> but that would have to be done by the CAM system anyway
[21:54:53] <rayh> Yes and this may be different for each tool
[21:57:43] <SWPadnos> right
[21:58:06] <SWPadnos> I'll take a shot at making a diagram of the layout of these new modules
[22:24:06] <fenn_afk> fenn_afk is now known as fenn
[22:24:25] <fenn> how do the pros make sure they don't crash into part fixtures when doing toolchanges?
[22:26:37] <fenn> apparently surfcam verify does it
[22:37:20] <SWPadnos> hiya Robin
[22:37:32] <robin_z> hi
[22:37:53] <fenn> woah. HSK shanks have a radial cavity for locating an information microchip. This memory cell carries information about tooling nomenclature, preset dimensions, and wear level. During the tool change cycle, data are exchanged between the microchip and the CNC control.
[22:38:11] <SWPadnos> cool
[22:38:13] <robin_z> wow
[22:38:46] <robin_z> all the swarf is barcoded too, so it can be tracked to the recycling process
[22:38:51] <robin_z> every piece/
[22:38:52] <SWPadnos> heh
[22:40:24] <anonimasu> heh
[22:41:10] <anonimasu> have anyone tried running a rotary table with emc?
[22:44:26] <SWPadnos> not me, though I will someday
[22:44:40] <anonimasu> hm
[22:44:58] <anonimasu> need to machine a scrwe..
[22:45:28] <anonimasu> with a very special thread..
[22:45:33] <SWPadnos> I've heard that there are issues with feedrate calcu;ations, but I don't know in what situations
[22:45:43] <SWPadnos> calculations
[22:46:13] <anonimasu> :(
[22:49:14] <fenn> eh, what are you asking?
[22:49:35] <fenn> odd pitch screw, machined on a lathe?
[22:50:14] <fenn> use rigid tapping cycle, set pitch to your pitch
[22:50:15] <anonimasu> no..
[22:50:20] <anonimasu> machined on a mill.
[22:50:52] <fenn> use helical interpolation and a single-point threadmill
[22:51:28] <anonimasu> that wont do it
[22:51:44] <fenn> why not? why are you using a mill to machine a screw anyway?
[22:52:08] <anonimasu> well, because it a special case..
[22:52:17] <SWPadnos> then do it by hand :)
[22:52:25] <SWPadnos> dremel to the rescue
[22:52:35] <fenn> heh make a thread chaser
[22:52:55] <anonimasu> well, rotary axis and a balled endmill.
[22:53:13] <anonimasu> the screw is non unfiorm in threads.,
[22:53:36] <fenn> reminds me of trying to figure out how to machine a ballscrew on a hexapod
[22:53:45] <SWPadnos> out of curiosity, how does it screw (on,in) to something if the threads aren't uniform?
[22:53:58] <SWPadnos> aret hey tapered as well?
[22:54:02] <SWPadnos> are they
[22:55:33] <anonimasu> oh it's the profile that isnt uniform either.. ;)
[22:56:47] <anonimasu> night
[22:56:51] <fenn> good luck
[22:57:01] <robin_z> ooh, how i *hate* people who make up their own network cables.
[22:57:14] <fenn> * fenn thumbs his nose at robin_z
[22:57:21] <robin_z> why are these idiots let loose with wirecutters eh?
[22:57:38] <anonimasu> fenn: if emc can machine it if I draw it it'll be ok
[22:58:04] <robin_z> fenn: when they do it right, its ok, when they do it wrong, well, its a hanging offence.
[22:58:19] <fenn> both very good points
[22:58:26] <anonimasu> robin_z: why ;)trynig to run 100mbit on a bad cable?
[22:58:56] <robin_z> anonimasu: yeah, OK at 10, bad at 100, inspecting the colour code shows they did it wrong
[22:59:10] <anonimasu> hehe
[22:59:12] <anonimasu> * anonimasu nods
[22:59:24] <robin_z> its 1:1 as it should be, but pairs in the wrong order
[22:59:34] <anonimasu> http://www.zeatz.co.uk/PHL.htm
[22:59:35] <anonimasu> sweet
[22:59:36] <anonimasu> :D
[22:59:40] <anonimasu> night
[23:01:22] <SWPadnos> 1,2,3,6 - that's all I need to know
[23:05:52] <robin_z> ummm
[23:05:59] <robin_z> nealry all you need to know
[23:06:08] <robin_z> 1&2, 3&6
[23:06:14] <SWPadnos> well - I suppose 4,5,7,and 8, plus the color order are good too :)
[23:06:37] <robin_z> thats you must use a pair on 1&2, and a pair on 3&6
[23:07:01] <SWPadnos> yes, orange then green (blue is 4+5 and brown is 7+8)
[23:07:18] <robin_z> mmm, nealry :)
[23:07:21] <cradek> orange-white orange green-white blue blue-white green brown-white brown
[23:07:45] <robin_z> pairs are blue+blue/white
[23:08:11] <SWPadnos> right - orange is pair 3 (pins 1+2). green is pair 2 (pins 3+6), blue is pair 1 (4+5), and brown is pair 4 (7+8)
[23:08:11] <robin_z> blou - orange -green - brown - slate
[23:08:19] <robin_z> thats the order, each with a white
[23:08:37] <SWPadnos> and the solid/white pattern is constant, though the colors change
[23:08:45] <robin_z> then with a yellow, then a black ..
[23:09:15] <robin_z> anyway ...
[23:09:22] <cradek> * cradek can make ethernet cables in his sleep
[23:09:43] <SWPadnos> so yes, those who don't know how to make cables should be shot for deploying cables they've made
[23:09:47] <robin_z> anyway this idiot had run it pair-pair-pair-pair across the plug
[23:10:05] <SWPadnos> correct for silver satin, but not for UTP (or STP)
[23:10:05] <Phydbleep> * Phydbleep makes a note to leave cat-5 parts where cradek can find them in his sleep.
[23:10:18] <SWPadnos> just do the cutting before you go to bed, please
[23:11:00] <Phydbleep> And line up a wholesale I can flog them too for a tidy profit.. Free labor! Woohoo. :)
[23:11:16] <robin_z> robin_z is now known as robin_sz
[23:12:26] <robin_sz> heh, non. there are machines that do it *soooooo* fast
[23:13:20] <robin_sz> somewhere .. I have a wire stripping machine .. you put on a reel, set the length and the amount of strip on each end and enter how many pieces you want
[23:13:36] <robin_sz> its awesomely quick ... liek 200 pieces a minute
[23:14:02] <lilo> [Global Notice] Hi all. There's a news article up on http://freenode.net/news.shtml on the status of the PDPC fundraiser. Highlights: pie chart is now up to date again; new "byte" and "nibble" donation levels. Fiscal year begins July 1. Please take a look. Have a great night, and thank you for using freenode! :)
[23:14:14] <robin_sz> hmm ... maybe 100
[23:14:21] <robin_sz> but bloody quick anyway
[23:14:27] <Phydbleep> robin_sz: What? It doesn't have the marker to label them every 3" at the same time?
[23:14:36] <robin_sz> heh no.
[23:14:57] <robin_sz> although I do have one of theose dot-matrix printheads on another machine ...
[23:15:07] <robin_sz> that was expensive .. 8K i think
[23:15:13] <robin_sz> and that ink ....
[23:15:25] <robin_sz> ick. dries in less than a second,
[23:15:31] <Phydbleep> * Phydbleep can tell robin_sz has never worked on things that had to meet FAA spec with it.
[23:15:35] <robin_sz> if you get it on your hands, its there for days
[23:15:58] <robin_sz> nope. never done any aero stuff
[23:16:15] <robin_sz> glass fibre covered wire and crimps huh?
[23:16:47] <fenn> ick.. 1/4 of PDPC donations go to UNICEF?
[23:16:50] <SWPadnos> Kapton
[23:17:43] <Phydbleep> That reminds me.. I need 5 gallons of anti-static prop-wash and a pair of muffler bearings...
[23:17:52] <robin_sz> ooh, kapton ... ex - pen seive
[23:19:05] <robin_sz> I do occasionally wire stuff on a plane, but its on a 'permit' so not needing inspection .. I think thjey call it 'experimental' in the US
[23:21:40] <fenn> US aviation regs are insane, even for experimental planes
[23:22:03] <robin_sz> oh.
[23:22:11] <robin_sz> over here its pretty simple
[23:22:27] <fenn> you build it, you fly it, try not to crash on top of anyone?
[23:22:29] <robin_sz> you promise not to fly over cities, they promise to clear the mess up when you crash
[23:22:33] <robin_sz> yeah
[23:22:41] <fenn> that sounds reasonable to me
[23:22:55] <robin_sz> I fly a Emaraude with a mate
[23:23:01] <robin_sz> Emeraude
[23:23:09] <robin_sz> late. spelling gone west
[23:23:43] <Phydbleep> * Phydbleep thinks 99% of it should be common sense, but since 99% the of the population has no common sense maybe it's a good thing..
[23:24:03] <robin_sz> nice little wood-and-rag taildragger
[23:24:28] <Phydbleep> robin_sz: You're dating a hippie chick?
[23:24:44] <robin_sz> heh, I wish
[23:24:50] <fenn> heh.. the emeraude : http://www.capcomespace.net/dossiers/espace_europeen/ariane/espace_francais/emeraude VE121 02.jpg
[23:28:00] <robin_sz> heh
[23:28:23] <robin_sz> http://www.gsazz.co.uk/Plane_Images/CP23501_1.JPG
[23:29:05] <fenn> nice.. i like the racing stripes on that one
[23:29:15] <fenn> or whatever you call it for airplanes
[23:29:36] <robin_sz> ours is slightly different in paint, but you get the idea
[23:29:54] <robin_sz> and built in 1956 if I remember right :)
[23:51:46] <SWPadnos> SWPadnos is now known as SWP_Away
[23:52:17] <Jymmm> yo