#emc | Logs for 2007-02-03

[00:00:06] <crepincdotcom> how does that work? theres no feedback from my mill to the computer
[00:00:28] <anonimasu> you tell it your position
[00:00:41] <crepincdotcom> ahh
[00:02:22] <jepler> crepincdotcom: you put the axis in a position you know, then tell axis what that position is
[00:03:10] <crepincdotcom> ahh ok. what is a good method for knowing the position? i know that sounds stupid, but i was trying to come up with a way to get the mill in "home" position yesterday and I realized I had no method to do so accuratly
[00:03:41] <jepler> you can use a feeler gauge
[00:03:49] <jepler> you know accurately how thick it is
[00:04:13] <jepler> jog down to just above the material, and find where the gauge no longer passes between the tip of your tool and the material
[00:04:18] <jepler> then enter the thickness of the gauge in the touch off dialog
[00:04:33] <crepincdotcom> ok, ill work on picking up a set of those
[00:05:20] <jepler> you want to use "incremental jog" mode, so that you move the axis by a fixed amount for each press of the key
[00:05:52] <cradek> many people use a piece of paper, or alum foil, if they don't need it to be so exact
[00:06:10] <crepincdotcom> so when it doesnt slide anymore, then I know its close
[00:06:13] <cradek> if you have good calipers, you can measure the foil thickness
[00:07:56] <crepincdotcom> i consider them to be fairly good, theyre 1/1000"
[00:09:23] <crepincdotcom> oh by the way: I picked up a 2.5 gallon (==tiny) shop vac from Home Depot yesterday since I'm running this mill in my dorm room on my bed.
[00:09:41] <crepincdotcom> I highly recommend it, this vac sucks REALLY well... in a good way
[00:11:30] <cradek> I have a small one that's nice for chips too - it's dark red, I don't know the brand
[00:11:51] <crepincdotcom> i was very suprised actually
[00:13:03] <cradek> I have to run, bbl
[01:07:14] <tomp> jepler: fill & expand dont work like I expect, it's not about understanding pack, i just re-cvsd to make sure it wasnt any local changes http://imagebin.org/7138
[01:15:36] <tomp> jepler: you get the same result for fill='x' fill='y' fill='both', i made some chgs that allow it to fill x when told x or both, but never got both to be ... both ( the chg is add in **kw to the __init__ of hbox & vbox
[01:16:16] <tomp> something stops y from filling
[01:19:34] <tomp> jepler: here's with the **kw passed http://imagebin.org/7139
[01:29:55] <crepincdotcom> * crepincdotcom is very happy :-)
[01:30:06] <crepincdotcom> truetype-trace works amazingly
[01:30:11] <crepincdotcom> i just milled my first thing
[01:30:25] <crepincdotcom> there is no noticable backlash
[01:30:47] <crepincdotcom> and my ghetto-rigged shop vac left my bed aluminium chip-free
[01:32:44] <crepincdotcom> cradek and jepler: thanks again for helping me get the few issues sorted out
[02:28:48] <jepler> crepincdotcom: sounds like a good start
[02:29:32] <jepler> crepincdotcom: any chance of a photo?
[02:41:58] <jepler> * jepler wanders off again
[02:44:21] <Jymmmm> * Jymmmm puts a tracking device on jepler's ankle... irc arrest =)
[02:45:21] <jmkasunich> heh - jepler can IRC from anywhere
[02:46:16] <cradek> it's true
[02:46:17] <Jymmmm> jmkasunich Nah, the other way around... just say his name and he gets a gentle 5KV shock =)
[02:46:56] <crepincdotcom> jepler: as soon as i can find a camera to borrow Ill take lots of pics
[02:47:50] <cradek> crepincdotcom: what are the leadscrews? I'm surprised you don't see backlash
[02:48:32] <crepincdotcom> cradek: theyre 1mm/rev... i mean the Mill itself was very accurate, I just added the CNC portion
[02:48:41] <crepincdotcom> but yeah I couldn't find any evidence of flat spots on circles
[02:48:51] <crepincdotcom> I'll do some larger ones this weekend and see if I notice anything
[02:49:00] <jmkasunich> its most likely to show up on very small circles
[02:49:12] <jmkasunich> and even those only show it if its severe
[02:49:32] <tomp> the old emc include ballbar tests
[02:49:33] <cradek> I'm looking forward to seeing your photos
[02:49:45] <cradek> glad the truetype-tracer works for you
[02:50:09] <crepincdotcom> yah it was great. I was supprised that it just used a bunch of tiny g1's as opposed to some curves
[02:51:02] <cradek> truetype uses splines, but gcode doesn't support them ... but emc does a nice job of blending together the g1s into the curves they approximate
[02:51:25] <crepincdotcom> yeah I noticed. It really does a great job of approximating the curves.
[02:51:39] <cradek> one of these days jepler and I will add spline support to our gcode interp... we talk about it every once in a while
[02:52:04] <cradek> but don't quote me on that :-)
[02:52:12] <crepincdotcom> is a g2/3 considered a type of spline, or would you have to make your own Gcode for that
[02:52:31] <cradek> g2/3 is an arc or helix which is not a spline
[02:52:47] <crepincdotcom> so how would you go about dealing with splines in the interpretter?
[02:52:47] <jmkasunich> "one of these days jepler and I will add spline support to our gcode interp - Chris Radek, 2-Feb-2007".... added to my quotes file ;-)
[02:52:55] <crepincdotcom> haha
[02:52:58] <cradek> jmkasunich: :-P
[02:53:13] <crepincdotcom> He's gonna hold you to that ;-)
[02:53:18] <cradek> crepincdotcom: that's one of the hard parts. rs724d may have spline support, I'd look there first
[02:53:25] <cradek> 274
[02:53:32] <crepincdotcom> is that the NC code spec?
[02:54:01] <cradek> my understanding is that rs274d is the ancestor of most gcode specs
[02:54:09] <crepincdotcom> ahh ok
[02:54:20] <cradek> EMC2's spec is rs274ngc plus some things added
[02:54:33] <crepincdotcom> "some things added" heh
[02:54:35] <cradek> I think rs274d is in machinery's handbook (but mine is several years old)
[02:54:49] <crepincdotcom> I have that in PDF form, assigned for a class
[02:54:54] <crepincdotcom> I'll read that section i guess
[02:55:18] <cradek> probably more useful to read the ngc spec, which is directly relevant to emc
[02:55:28] <crepincdotcom> ha
[02:55:30] <jmkasunich> today I was using Pads (a pcb layout tool) to generate photoplot files.... the photoplot fomats are RS-274G and RS-274X
[02:55:32] <crepincdotcom> anyway, time to leave work
[02:55:36] <crepincdotcom> talk to you later, good night
[02:55:40] <cradek> goodnight
[02:55:46] <jmkasunich> goodnight crepincdotcom
[02:57:14] <cradek> jmkasunich: if you're keeping track, add this one too: 19:30:06 < crepincdo> truetype-trace works amazingly
[02:57:49] <jmkasunich> ok ;-)
[02:58:20] <tomp> i dont see a standard tho pmac & machinemate used g5 to define the spline & g6 to interpolate it. not finding any examples, any my machinemate license expired
[02:58:55] <cradek> tomp: it would be spectacular if you have the ability to figure out how splines are typically specified, and could make a wiki page about it
[02:59:23] <cradek> tomp: I simply don't have access to that kind of information/documentation
[03:00:15] <tomp> i'll look for stuff and put it on a new page. wish everybody had svg capable browsers.... great for beziers anyway
[03:00:17] <cradek> but I thought g5 was a parabola (or ellipse??) in rs274d
[03:00:48] <tomp> these were not standard, just 2 controls that had the ability
[03:00:54] <cradek> I thought the spline gcode had a dot in it (Gxx.x)
[03:01:06] <cradek> ok
[03:01:25] <cradek> this may be another of those times we would make up our own scheme and people could program to it if they want
[03:01:27] <jmkasunich> the great thing about standard is that there are so many of them
[03:01:52] <cradek> no cam software is going to generate splines unless an emc2 user makes it
[03:02:12] <Jymmmm> cradek is an enc2 user =)
[03:02:20] <Jymmmm> emc2 too =)
[03:02:54] <Jymmmm> Picked up some micro fiber cloths $2.50 for 3pk
[03:03:36] <CIA-8> 03cradek 07v2_1_branch * 10emc2/nc_files/comp311.ngc: example Fig.3 from handbook section B.3.1.1 - radius compensation entry move
[03:04:02] <cradek> * cradek hopes nobody notices that's not a bugfix
[03:05:45] <jmkasunich> it fixes a documentation bug
[03:05:48] <jmkasunich> (lack therof)
[03:05:58] <tomp> did the change explain why the little arc was needed? i liked your explanation this morning
[03:05:59] <cradek> that's a good way to look at it
[03:06:28] <cradek> tomp: what do you mean? I didn't change anything except I typed in the sample program
[03:06:54] <tomp> lemme go look, i didnt see yet
[03:12:54] <tomp> cradek: i see, you added the nc file so people can see the comp get 'turned on' as it entered the contour. i was talking about the tiny extra arc that occurs, and is not programmed, seen in figure3, discussed by erDiZz, yourself & myself. I thought the docs might admit that it was there & was neccesary to prevent gouging as you explained.
[03:15:03] <cradek> I don't think of that first arc as special - in comp mode, you get two moves for nearly every one you program, and the extra one is an arc
[03:15:25] <cradek> an arc is always used to go around a corner while keeping the tool in contact with it
[03:16:13] <tomp> ok, i always ran in exact stop mode, ( edm, slow, ... )
[03:16:21] <cradek> you can see that in the "interpreter does it this way" (rounded corners) picture
[03:16:43] <cradek> this has nothing to do with blending - it's how it turns corners in radius comp mode
[03:18:00] <tomp> yes, just 2 different usages, i also made all programs tangential, so there was no need for blending... i'd never program a square to get a rounded rectangle, just not the way id do it
[03:18:25] <cradek> argh that's not the point
[03:18:34] <cradek> if you want to cut a square, you program a square in radius comp mode
[03:18:46] <cradek> the interpreter will cut it with 4 lines and 4 arcs (around the corners)
[03:19:51] <cradek> the point of radius comp mode is you program the shape you want, and let the interpreter figure out the exact tool path
[03:20:44] <tomp> i understand that , i'd never use radius comp, i'd program the exact path... it's a valid use that you dont use..
[03:21:35] <cradek> sure, I do that all the time too
[03:21:48] <cradek> sorry, I guess I've lost track of what we're talking about
[03:22:16] <tomp> no problem, i might spin a tool someday and use it :)
[03:22:45] <cradek> you could use it in edm to compensate for kerf too
[03:23:19] <erDiZz> need sleep... bye!
[03:23:29] <tomp> what kind of splines were you considering?
[03:24:29] <cradek> probably cubic (two control points)
[03:25:01] <tomp> ok, thats popular with lotsa apps ( esp layout , esp svg :)
[03:25:48] <cradek> and postscript I think
[03:27:12] <tomp> what about extending the interpreter to call a python program with the data, then, when finished, returning to gcode. this segway might have lotsa uses
[03:27:58] <cradek> that's already done, except it can be any program (using AXIS filters)
[03:28:36] <cradek> since splines (currently) can only be approximated with gcode (lines/arcs) we have to extend gcode itself
[03:31:49] <Twingy> k, fixed a couple bugs, gonna post another release shortly
[03:32:15] <owad> Turns out the I burnt up the diode rectifier in my power supply. Not sure how attaching two resistors could do that…
[03:32:30] <tomp> is this code for bezier interpolation useful ? http://ibiblio.org/e-notes/Splines/Bezier.htm
[03:32:35] <jmkasunich> how big of a diode was it?
[03:33:09] <jmkasunich> tomp: the actual spline interpolation isn't the hard part - the formulas are available in 100 places
[03:33:10] <owad> 1000V 35A rectifier
[03:33:24] <jmkasunich> the hard part is finding a way to specify the spline in the g-code
[03:33:25] <cradek> tomp: no web browser right now...
[03:33:44] <jmkasunich> (ideally, a way that is compatible with they way other people do it, if they do)
[03:33:55] <jmkasunich> principle of least astonishment and all that
[03:34:03] <cradek> and transforming the spline to be constant-velocity (jepler has found some scholarly paper about this)
[03:34:03] <tomp> lol
[03:34:17] <jmkasunich> owad: that bridge should be pretty hard to kill
[03:34:44] <cradek> g[something] control point 1 in XYZABC
[03:34:49] <cradek> g[something] control point 2 in XYZABC
[03:34:59] <cradek> g[something else] target in XYZABC
[03:35:11] <owad> it's got a burnt smell to it and the two diodes on the negative side are open in both directions
[03:35:33] <jmkasunich> well... that certainly sounds daid
[03:35:33] <cradek> I think this will be the only possible way - there are NOT enough letters to specify three entire coordinates in one line
[03:36:29] <cradek> or maybe the control points should be relative to the initial point, like arcs
[03:36:35] <cradek> but not enough letters again
[03:36:42] <tomp> why 1 line?
[03:37:04] <jmkasunich> one line would be simpler
[03:37:26] <jmkasunich> you wouldn't have to worry about cases where the guy programs the target point before the control points, for instance
[03:37:42] <owad> it was working find until I hooked two 1k 1W resistors (in series) up to the caps. They started sparking as soon as I turned it on, so I killed the power. Removed the resistors and tested them. They were fine. Tested the power supply with them off, and it started humming. Removed the caps and the bridge, and it stopped. Then I noticed the smell from the bridge.
[03:37:43] <jmkasunich> the line stand alone, and either is or is not valid as such
[03:37:57] <cradek> that would be nice but I think it's impossible
[03:38:15] <jmkasunich> the resistors were across the caps from plus to minus?
[03:38:18] <cradek> we could do conic with the arc scheme (IJK)
[03:38:36] <cradek> but i think cubic just has too many points for one line
[03:38:41] <owad> right
[03:38:45] <jmkasunich> cradek: I know, I was answering tomp's question
[03:38:48] <tomp> 3 lines of data then a trigger line would avoid 'not enuf letters', but yes would require a test to see everything was ready to fire
[03:38:53] <cradek> oh sorry
[03:39:08] <cradek> you just give an error if it's wrong, like anything else
[03:39:18] <jmkasunich> owad: do you have a voltmeter or an ohmeter?
[03:39:28] <owad> yes, a multimeter with both
[03:39:46] <jmkasunich> and you measured the resistors, you are sure they are 1K?
[03:40:21] <owad> definitely. Just bought them brand new at radio shack. Labeled 1K and tested 1K. 2K in series
[03:40:25] <jmkasunich> ok
[03:40:39] <jmkasunich> how many volts did you have on the cap before you attached the resistor?
[03:40:39] <owad> in fact
[03:41:21] <owad> before I soldered the two together, I tested it with just one resistor, and that worked fine, 1W was a bit low for 1K, so I wanted to hook the pair up
[03:41:26] <owad> 34v
[03:41:37] <jmkasunich> ok
[03:41:54] <owad> I also looked all over for a short, but couldn't find a thing
[03:42:08] <jmkasunich> if a 1K load didn't hurt it, there is no way a 2K load did
[03:42:15] <jmkasunich> so there had to be something else going on
[03:42:27] <jmkasunich> how did you connect the resistors?
[03:42:33] <owad> solder
[03:42:47] <jmkasunich> with the unit powered down I assume?
[03:42:53] <owad> not my normal solder, actually…
[03:42:57] <owad> yes
[03:43:04] <jmkasunich> not normal solder?
[03:43:19] <jmkasunich> abby-normal solder?
[03:43:35] <owad> not _my_ normal solder. This stuff came with an on old soldering gun, and was handy, since I was in the garage
[03:43:54] <jmkasunich> not acid core I hope
[03:44:06] <jmkasunich> althought that wouldn't cause immediate failure
[03:44:15] <jmkasunich> were you using a soldering iron or the gun?
[03:44:25] <owad> I can't imagine I'd have any (and I know I've used this solder before once or twice for other stuff, and didn't have a problem), but that's what I'm half wondering
[03:44:24] <owad> iron
[03:44:27] <jmkasunich> ok
[03:44:33] <jmkasunich> how big is your cap?
[03:44:38] <jmkasunich> (uF, volts)
[03:44:51] <owad> two 6300uF caps
[03:44:56] <jmkasunich> and how big is your transformer (VA if you know, estimated weight if not)
[03:45:10] <owad> 24V 10A
[03:45:19] <jmkasunich> whats the rated voltage on the caps?
[03:45:44] <owad> I think it's 40
[03:46:16] <jmkasunich> 40 isn't a common cap rating (but anything is possible I guess)
[03:46:20] <jmkasunich> 35 and 50 are more common
[03:46:31] <jmkasunich> anyway - 34V on a 40V cap is pushing things
[03:46:42] <owad> these caps are old, btw. I either pulled them out of some power supplies I was throwing away a few years ago, or got them used with some surplus lots
[03:46:57] <jmkasunich> then the voltage is really pushing things
[03:47:15] <jmkasunich> when it failed, did it do it as soon as you turned it on?
[03:47:35] <jmkasunich> duh, yes
[03:47:38] <jmkasunich> "I hooked two 1k 1W resistors (in series) up to the caps. They started sparking as soon as I turned it on"
[03:47:39] <owad> as soon as I turned it on, it started sparking. I turned it back off within a second or two
[03:47:56] <owad> actually, it may have been a second or so until it started sparking
[03:48:00] <jmkasunich> what was sparking (or were you too busy reaching for the switch to be sure)
[03:48:11] <owad> it was right at the resistor
[03:48:25] <owad> I was standing half way across the room at the switch :-)
[03:48:26] <jmkasunich> you don't have a photo of this thing do you?
[03:48:38] <owad> no
[03:49:23] <owad> I'm ready to ditch the caps, though, and need a new diode rectifier, so I'm practically starting from scratch.
[03:49:25] <jmkasunich> its very unlikely that the resistors were actually what was sparking - overheated resistors smoke, not spark
[03:49:33] <jmkasunich> unless you _really_ overheat them
[03:49:40] <owad> well, the resistors were right on top of the cap
[03:49:47] <jmkasunich> and for a 1K 1W, that would take 100V or more
[03:49:52] <owad> so it could have just as easily been the cap
[03:50:11] <jmkasunich> is this a can type cap with screw terminals
[03:50:15] <jmkasunich> or PC board mount?
[03:50:21] <owad> can type
[03:50:35] <jmkasunich> two of them, right
[03:50:36] <owad> right
[03:50:39] <jmkasunich> wired in parallel?
[03:50:43] <owad> yes
[03:50:50] <jmkasunich> polarity right?
[03:51:01] <jmkasunich> (probably, otherwise it wouldnt have lasted that long)
[03:51:15] <owad> Yes. It was three, but I blew the other one the other day, when I hooked that one up backwards. (just occurred to me I don't think you were here for that conversation)
[03:51:38] <jmkasunich> I think I missed that
[03:52:08] <jmkasunich> do the caps have vents? usually its a little rubber plug of sorts, sometimes between the terminals on top of the can
[03:52:21] <owad> yes
[03:52:34] <owad> there's was no rubber plug, though
[03:52:42] <owad> it has sort of a washer-looking thing
[03:52:43] <jmkasunich> just a hole?
[03:52:53] <owad> right, with the washer-looking deal inside
[03:52:57] <jmkasunich> hmm
[03:53:02] <owad> and then solid beneath
[03:53:03] <jmkasunich> hard to visualize
[03:53:10] <jmkasunich> any juice around the hole?
[03:53:17] <owad> no
[03:53:35] <jmkasunich> when electrolytic caps fail, they usually vent juice or steam or both
[03:53:43] <owad> (and they seemed to be working fine and holding a charge, yesterday)
[03:53:53] <owad> the one that blew yesterday just vented steam
[03:54:11] <owad> but it still looks the same as it did before
[03:54:14] <jmkasunich> so it didn't actually "blow" as in blow up
[03:54:19] <owad> right
[03:54:20] <owad> sorry
[03:54:34] <jmkasunich> if the vent doesn't work they blow up
[03:54:53] <owad> so the one that vented might still be usable?
[03:55:07] <jmkasunich> smaller ones (PC board mount) sometimes just have a weak spot in the can instead of a real vent - the can splits open at the weak spot
[03:55:14] <jmkasunich> unlikely
[03:55:44] <jmkasunich> the vent is a safety thing - the cap fails, the insides heat up, and the vent prevents an explosion - but its already failed electrically
[03:55:44] <tomp> jmkasunich: can he 'charge' it with the vmeter, and reverse leads to 'discharge' it? as a test
[03:55:59] <jmkasunich> tomp: not with the voltmeter, but the ohmeter might work
[03:56:03] <owad> I think I'm going to ditch them all
[03:56:28] <owad> I still have a 6800uF cap that I think is rated for 50v
[03:56:48] <owad> and I have a 4200uF that's rated for 40v. is that cutting it too close?
[03:57:04] <jmkasunich> yes
[03:57:14] <owad> what's the minimum you'd recommend?
[03:57:18] <jmkasunich> I would never use less than 50V for a 35V supply
[03:57:32] <jmkasunich> at work, for low voltage electrolytics, we like to derate 2:1
[03:57:48] <jmkasunich> 24V supply, 50V caps, 5V supply, either 10 or 16V caps
[03:58:13] <jmkasunich> don't do many 35V supplies, but if I did I'd probably be looking for a 75V cap
[03:58:22] <jmkasunich> unless I really had nasty size or cost issues
[03:58:22] <K`zan> jmkasunich: Having the internals of an electrolitic shot at my face I agree with 2:1 big time :-).
[03:58:26] <owad> ok, I'll probably end up ordering a couple then. Any place you recommend?
[03:58:33] <K`zan> 2x
[03:58:37] <K`zan> minimum!
[03:58:53] <jmkasunich> K'zan: not the ones I sent you?
[03:59:02] <K`zan> * K`zan was made a real coward about eletrolitics over that experience :-).
[03:59:14] <K`zan> jmkasunich: No, but those are not cheap caps either :-).
[03:59:27] <jmkasunich> phew - I was worried for a minute
[03:59:35] <K`zan> jmkasunich: You did see my thanks about that I hope :-)!
[03:59:39] <jmkasunich> yes
[03:59:41] <jmkasunich> you're welcome
[03:59:50] <tomp> owad: try digikey or mouser, maybe jameco
[03:59:54] <K`zan> IIRC about a +7V increase...
[04:00:06] <jmkasunich> owad: tomp's advice is probably good
[04:00:37] <jmkasunich> my most common source is the dumpster at work, so I don't know who is the best seller
[04:01:00] <jmkasunich> be prepared to be overwhelmed with choices at digikey and probably the other places
[04:01:18] <jmkasunich> digikey lets you sort on just about anything _except_ price (the bastards)
[04:01:32] <K`zan> jmkasunich: over that issue, I requested the printed catalog from digikey.
[04:01:40] <owad> if it's got the right specs and it's new, can I go wrong no matter where I get it?
[04:01:48] <K`zan> Still confusing, but somewhat less so.
[04:01:58] <K`zan> owad: Shouldn't
[04:03:40] <jmkasunich> Digikey: 6800uF, 50 or 63V: "You have selected 42 items, spanning 2 pages."
[04:03:44] <owad> Can anybody remind me of the equation for calculating require capacitance? IIRC, it was a guys' name
[04:03:53] <jmkasunich> Mariss from gecko
[04:03:59] <owad> yes. Thanks
[04:04:09] <jmkasunich> I think its 80000*I/V or something like that
[04:04:24] <jmkasunich> what is your load current?
[04:04:31] <K`zan> jmkasunich: under the category of "Too Damn Odd", when that cap went off, everyone there said it *should* have hit me in the face, the mark on the wall was in direct line with my head. Wrote that one off to the schrapnel going into another dimension briefly...
[04:05:05] <jmkasunich> either that or you got scared into another dimension briefly
[04:05:17] <K`zan> jmkasunich: That is MUCH more likely :-)
[04:06:16] <jmkasunich> ouch - 6800uF 50V screw terminal caps are $17-20
[04:06:20] <owad> two 260 oz/in motors; 2.8A per phase
[04:06:26] <K`zan> Never used a 12V cap on a 12V supply again...
[04:06:32] <jmkasunich> ok, so you need about 4A
[04:06:52] <jmkasunich> about 9400uF
[04:06:53] <owad> I was only using the 6800uF caps because that's what I had. IIRC, I needed about 20,000uF total
[04:07:06] <owad> or I could be misremembering :-)
[04:07:21] <jmkasunich> 80000*4A / 34V = 9411
[04:07:34] <jmkasunich> maybe you were using 2x the motor current and 24V
[04:07:57] <owad> How do you get 4A?
[04:08:29] <jmkasunich> power supply current usually tops out at about 2/3 of the motor current, I called that 2A per motor (actually 1.86, but close ehough)
[04:08:37] <jmkasunich> two motors at 2A each is 4A
[04:09:10] <owad> you only need to calculate for 1 phase?
[04:09:34] <jmkasunich> right - you never have full current in both at the same time
[04:09:47] <tomp> maybe look at 'snap ins' instead of screw leads, solder to 'em, 10kuf 63v 5.50$ jameco
[04:09:58] <owad> so that 6800 uF 50v gets me most of the way there, then
[04:10:30] <jmkasunich> tomp: right - screw terminal ones cost twice as much as leaded ones in this size range
[04:10:40] <jmkasunich> three time as much actually
[04:11:49] <jmkasunich> heh, quantity discounts are huge
[04:12:01] <tomp> the big trick is who actually has it and who doesnt rape you on shipping orders < 25$
[04:12:16] <jmkasunich> digikey runs around $6-7 for each of the several kinds of snap-in 10000uF 63V caps they have
[04:12:29] <jmkasunich> except the one with a 200 piece minimum, they are $2.85
[04:13:53] <owad> any nice way to mount those?
[04:14:25] <K`zan> I do miss the goodies out of the dumpster at work...
[04:14:37] <tomp> upside down, use a standard cap clamp, legs up ( and smilin )
[04:14:54] <K`zan> Just don't lean over on it :)
[04:15:17] <K`zan> * K`zan is a cap coward :-)
[04:16:11] <jmkasunich> tomp: cap clamp probably costs more than the cap
[04:16:18] <jmkasunich> I dunno how small they make them
[04:16:46] <tomp> tie wrap to a bolt covered with shrink wrap
[04:17:01] <jmkasunich> yeah
[04:17:09] <jmkasunich> 10,000uF 63V is 30mm dia x 50mm long
[04:17:10] <tomp> tie wraps the duct tape of electronics
[04:17:13] <Jymmmm> U-Bolt =)
[04:17:13] <owad> How much leeway do i have on that 9400uF figure?
[04:17:38] <jmkasunich> quite a bit
[04:17:47] <jmkasunich> If I was buying, I'd buy 10000
[04:17:54] <owad> Can I get away with 6800uF?
[04:17:58] <Jymmmm> You can always go over the rate uF, correct?
[04:18:10] <jmkasunich> if I was scrounging, I'd use anything between 8000 and 20000
[04:18:17] <jmkasunich> 6800 might be pushing things a little
[04:18:38] <jmkasunich> if I could scrounge two 6800's, I'd parallel them and call it good
[04:18:51] <jmkasunich> but if you're gonna buy new, buy the right thing
[04:19:08] <jmkasunich> I bet you'll find there isn't that much differnce in price between 6800 and 10000
[04:19:10] <owad> I think I only have one of those. I may have something else to hook up in parellel, though
[04:19:45] <Jymmmm> jmkasunich: PS caps, if you need (lets say) 10,000uF can you use 20,000?
[04:20:06] <jmkasunich> Jymmm: there are limits
[04:20:20] <Jymmmm> jmkasunich less than double?
[04:20:25] <jmkasunich> depending on the situation 2x the recommended value could have issues, or it might be fine
[04:20:34] <Jymmmm> 1.5 x ?
[04:20:39] <jmkasunich> 1.5x I wouldn't hesitate at all
[04:20:56] <Jymmmm> k, good to know. I always liek over-engineering things
[04:21:17] <jmkasunich> in owad's case, where he as a 10A transformer and a 4A load, he can use bigger caps if he wants
[04:21:24] <jmkasunich> as long as he has a beefy diode
[04:21:33] <Jymmmm> bridge?
[04:21:35] <jmkasunich> yeah
[04:21:48] <jmkasunich> the issue with really big caps is this:
[04:21:49] <owad> I couldn't get away with a 50,000 uF cap, could I?
[04:21:59] <Jymmmm> those are cheap enough, but buying 10Kuf caps then needing 20Kuf later on is a waster
[04:22:19] <jmkasunich> when the diodes conduct, they charge the cap to the peak voltage of the sinewave
[04:22:37] <jmkasunich> then the sinewave goes back down, the diodes turn off, and the caps supply the load
[04:22:48] <jmkasunich> with a moderate cap, it discharges a few volts during that time
[04:23:19] <jmkasunich> the next time the sinewave comes back up (1/120th of a second later), the cap has dropped a couple volts, so the diode turns on a couple volts before the peak
[04:23:28] <jmkasunich> and the cap gets recharged over a period of time
[04:23:48] <jmkasunich> if you use a fscking HUGE cap, the voltage barely drops during the time that the sinewave is low
[04:24:06] <jmkasunich> so when it comes back, the diodes don't start conducting until almost the very peak of the sinewave
[04:24:14] <jmkasunich> then you have to recharge the cap all at once
[04:24:16] <jmkasunich> with very high peak current
[04:24:24] <jmkasunich> hard on diodes and transformers
[04:24:29] <jmkasunich> * jmkasunich shuts up now
[04:24:43] <Jymmmm> jmkasunich is THAT why I shouldn't use dimmers?
[04:24:59] <Jymmmm> jmkasunich VERY GOOD explaination btw.
[04:25:08] <Jymmmm> ty
[04:25:16] <jmkasunich> yw
[04:26:01] <Jymmmm> Heh, you gotta loves sears... every part of things they make/brand has a parts list and is available to order from theri website.
[04:26:51] <Jymmmm> My "Craftsman Professional Router" is really a Bosch Router rebranded, and I can order a replacement cord from the website =)
[04:27:46] <Jymmmm> or should I be thanking K-Mart now?
[04:27:59] <Jymmmm> since KMart owns Sears
[04:28:44] <Jymmmm> jmkasunich how much are 120VAC speed controls going for on avg?
[04:29:47] <jmkasunich> no clue
[04:30:17] <Jymmmm> k, the one for my router (8k to 30K rpm) is $25
[04:38:04] <Twingy> just posted new gcam
[04:39:38] <tomp> Twingy: thanks, will try soon
[04:40:44] <Twingy> k
[04:40:52] <Twingy> next release will be stable
[04:41:09] <Twingy> project will split into development and stable releases
[04:42:16] <tomp> jepler: please look at http://imagebin.org/7138 and http://imagebin.org/7139, there something screwy in the packing, visible with very simple xml test
[04:42:37] <tomp> gnite all, fallin asleep
[04:43:20] <jmkasunich> Twingy: jeff is almost certainly asleep
[04:43:26] <jmkasunich> but he'll probably read back tomorrow
[04:44:05] <Twingy> next release will probly be 2-3 weeks unless something critical comes up
[10:41:45] <robin_sz> bah .. who understands this MRTG thing?
[10:55:45] <alex_joni> * alex_joni does a bit
[10:56:31] <alex_joni> but I totally lack the understanding of rrdtool
[12:01:28] <A-L-P-H-A> * A-L-P-H-A is just confuzzled.
[12:08:34] <alex_joni> that happens
[13:00:07] <slundell> is Dallur here?
[13:49:40] <lerneaen_hydra> 'lo all
[14:31:00] <alex_joni> hi LH
[14:32:53] <lerneaen_hydra> alex_joni; anything fun happening?
[14:35:06] <alex_joni> lerneaen_hydra: not much
[14:35:16] <alex_joni> working on some webpage translations
[14:35:19] <alex_joni> puma kins
[14:35:21] <alex_joni> etc
[14:35:30] <lerneaen_hydra> nice
[14:35:36] <alex_joni> yeah.. kinda
[14:39:32] <lerneaen_hydra> becuase translation is oh-so-fun ;)
[14:42:37] <alex_joni> and I'm trying to compile some older ARM linux
[14:42:49] <alex_joni> strangely called mizi
[14:45:01] <lerneaen_hydra> fun fun
[14:45:14] <alex_joni> not at all
[14:45:19] <alex_joni> it's got docs in korean
[14:45:25] <lerneaen_hydra> O_o
[14:45:25] <alex_joni> and some in chinese
[14:45:33] <lerneaen_hydra> you're not working on a gp2x by any chance?
[14:47:12] <alex_joni> smdk2410
[14:47:19] <alex_joni> if that rings any bell
[14:47:24] <lerneaen_hydra> hmm, nope
[14:47:30] <lerneaen_hydra> that would have been too spooky otherwise
[14:47:57] <alex_joni> s3c2410x
[14:48:04] <alex_joni> some samsung ARM9
[14:48:27] <lerneaen_hydra> ah, ok
[14:49:14] <alex_joni> the stuff I'm working with is cca. 2003-2004
[14:49:21] <alex_joni> couldn't find anything newer
[14:49:28] <lerneaen_hydra> oh, I see
[14:49:47] <lerneaen_hydra> btw, random question, which file controls which files in /etc/init.d are started on bootup?
[14:50:01] <alex_joni> there are some symlinks
[14:50:16] <lerneaen_hydra> where?
[14:50:22] <alex_joni> in /etc/rcX.d/
[14:50:26] <alex_joni> where X is the runlevel
[14:50:45] <alex_joni> the symlinks are usually called Sxy<name>
[14:50:49] <alex_joni> for starting a service
[14:50:55] <lerneaen_hydra> ah, ok
[14:51:05] <alex_joni> the xy is a number, and shows in what order they start
[14:51:14] <alex_joni> and Kxy<name> is for stopping
[14:51:31] <alex_joni> (this is assuming you are running init)
[14:51:38] <lerneaen_hydra> yeah
[14:51:45] <alex_joni> the latest Ubuntu doesn't anymore
[14:51:54] <lerneaen_hydra> was it ubuntu that bastardised the init.d stuff?
[14:51:55] <lerneaen_hydra> oh
[14:52:03] <alex_joni> they use something called jumpstart? or something like that
[14:52:03] <lerneaen_hydra> by latest do you mean edgy?
[14:52:08] <lerneaen_hydra> hmm
[14:52:12] <alex_joni> I think Feisty
[14:52:16] <lerneaen_hydra> ah, ok
[14:52:21] <lerneaen_hydra> then I've still got init.d
[14:52:24] <alex_joni> Edgy has it as an option afaik
[14:52:33] <alex_joni> it's supposed to make starting up lots faster
[14:52:40] <alex_joni> as it has a way to run things in paralel
[14:52:45] <alex_joni> unlike init
[14:52:54] <lerneaen_hydra> if I rename a link from SXY to old-SXY
[14:53:07] <lerneaen_hydra> will it not be run then or is that a bad idea?
[14:53:15] <alex_joni> /opt/host/armv4l/bin/armv4l-unknown-linux-gcc -D__KERNEL__ -I/home/cdan/linuette/target/box/kernel/include -Wall -Wstrict-prototypes -Wno-trigraphs -Os -mapcs -fno-strict-aliasing -fno-common -fno-common -pipe -mapcs-32 -march=armv4 -mtune=arm9tdmi -mshort-load-bytes -msoft-float -DKBUILD_BASENAME=clnt -c -o clnt.o clnt.c
[14:53:33] <alex_joni> lerneaen_hydra: you can do that.. not sure what happens..
[14:53:42] <alex_joni> simply removing the symlink should do the trick
[14:53:45] <lerneaen_hydra> what's the best way to remove it
[14:53:46] <lerneaen_hydra> ah, ok
[14:54:06] <alex_joni> armv4l-unknown-linux-gcc <- LOL
[14:54:21] <lerneaen_hydra> haha
[14:55:25] <alex_joni> strange way to name a compiler
[14:58:25] <lerneaen_hydra> btw, do you know how /dev/random generates its pseudorandom numbers?
[14:58:36] <alex_joni> randomly?
[14:59:16] <robin_sz> meh
[14:59:16] <lerneaen_hydra> I've heard stuff about it running out of entropy, making it sound like it samples stuff from some input device
[14:59:33] <lerneaen_hydra> if you read too much data from it too quickly
[14:59:44] <lerneaen_hydra> and hence urandom if randomness isn't as crucial
[14:59:53] <robin_sz> that only applies to desktop systems
[14:59:58] <robin_sz> not servers
[15:00:10] <alex_joni> the serious systems have a HW random chip onboard
[15:00:16] <jepler> lerneaen_hydra: it gathers "entropy" from various sources into a "pool". When you request random bytes, it uses the "pool" and a cryptographic hash function to give you random bytes. If you read from /dev/random, it reduces the estimate of entropy available, but if you read from /dev/urandom it doesn't.
[15:00:17] <robin_sz> servers have extra entropy installed at manufacture
[15:00:47] <lerneaen_hydra> jepler; yeah, exactly, where does a standard desktop system get that random seed data /entropy from?
[15:00:52] <robin_sz> like what alex_joni said ..
[15:01:02] <jepler> lerneaen_hydra: in basic desktop systems, most entropy data comes from the timing of external events like network packets, keypresses, etc
[15:01:04] <robin_sz> keyboard presses, disk data ..
[15:01:13] <robin_sz> basically, anyting and everything!
[15:01:18] <lerneaen_hydra> hmm, I see
[15:01:55] <robin_sz> some servers have a hardware noise generator feeding a counter ...
[15:02:21] <lerneaen_hydra> yeah
[15:02:48] <robin_sz> practically, though, its jst not a problem
[15:02:55] <jepler> http://www.cs.berkeley.edu/~daw/rnd/linux-rand seems to describe the method in some more detail
[15:03:03] <robin_sz> and if you are paranoid, you can always use Yarrow
[15:03:05] <lerneaen_hydra> how much data can you usually grab from random?
[15:03:14] <lerneaen_hydra> enough for almost all applications?
[15:03:22] <alex_joni> lerneaen_hydra: same amount as for /dev/zero
[15:03:23] <robin_sz> more than enough
[15:04:27] <robin_sz> http://www.schneier.com/yarrow.html
[15:04:48] <jepler> lerneaen_hydra: most applications would only need tens of bytes of random data, to use as a seed for the application's own PRNG
[15:05:03] <lerneaen_hydra> yeah
[15:05:09] <lerneaen_hydra> that's what I was thinking
[15:05:44] <robin_sz> most prng algps are crp anyway
[15:05:48] <robin_sz> algos
[15:06:09] <lerneaen_hydra> radioactive sample and sensor is t3h shit
[15:09:56] <robin_sz> yeah, but bulky
[15:10:12] <robin_sz> due to the large bax they come in and the cat
[15:10:17] <robin_sz> box
[15:10:30] <jepler> plus you only get one bit of data, and someone has to open the box to get it
[15:10:43] <robin_sz> and then the cat sies ...
[15:10:44] <robin_sz> dies
[15:10:55] <jepler> not every time
[15:11:01] <robin_sz> 50:50
[15:11:01] <jepler> on average you get two bits of data per cat
[15:11:06] <robin_sz> true
[15:11:14] <robin_sz> 4 cats to a byte
[15:11:14] <jepler> but the relevant figure is bits per litter, if I'm not mistaken
[15:11:24] <robin_sz> * robin_sz nods
[15:11:45] <robin_sz> much over 1kb of data and disposal of cats becomes a problem
[15:12:34] <jepler> The "PARC" in Xerox PARC stood for "parallell array of radioactive cats".
[15:12:51] <robin_sz> I suspect the government data centres where they generate lots of random numbers (for use in tax bills for example) muts have some sort of automated conveyor system for dealing with the large numbers of cats required
[15:13:34] <lerneaen_hydra> well, the total waste isn't all that great
[15:13:38] <lerneaen_hydra> just recycle the cats
[15:13:47] <robin_sz> fair point
[15:13:56] <alex_joni> feed them to other cats
[15:14:07] <robin_sz> umm
[15:14:14] <lerneaen_hydra> and add some new raw material to keep entropy happy, as the cats are not 100% efficient, they do generate heat after all
[15:14:19] <alex_joni> then they only need a tiny bit of fresh radioactivity
[15:14:23] <lerneaen_hydra> soyulent green!
[15:14:28] <robin_sz> would that not lead to a general increase in the cyanide levels as time went by?
[15:14:46] <alex_joni> robin_sz: gotta feed them some radioactive dogs once in a while
[15:14:49] <lerneaen_hydra> not if you filter it out
[15:15:13] <robin_sz> hmm ... the more you get into it, the more troublesome this random number generation is
[15:16:09] <robin_sz> "ad finally, to complete your security audit, can I check how you generate secure random numbers"
[15:16:25] <robin_sz> "we think we have a radioactive cat in that box over there"
[15:16:28] <robin_sz> "think?"
[15:16:38] <robin_sz> "well, yeah .. sorta maybe"
[15:16:46] <robin_sz> "can I see it?"
[15:17:00] <robin_sz> "well, no .. not really, that would runi the entropy"
[15:26:06] <steve_stallings> steve_stallings is now known as steves_logging
[16:00:28] <SWPadnos_> SWPadnos_ is now known as SWPadnos
[16:07:30] <lerman_> lerman_ is now known as lerman
[16:17:55] <alex_joni> anyone knows how to fit a linux inside 1MB ?
[16:18:14] <alex_joni> without the kernel and bootloader I mean
[16:18:51] <cradek> rm everything except /bin/init
[16:19:13] <cradek> then replace /bin/init with ... ?
[16:19:18] <alex_joni> how about libstdc++ ..
[16:19:27] <alex_joni> is that used for normal executables/
[16:19:29] <alex_joni> ?
[16:19:37] <alex_joni> like ls, dir, rm, etc
[16:19:57] <cradek> no, since those are not c++
[16:20:02] <cradek> use ldd to be sure what they use
[16:20:16] <alex_joni> I can't use ldd.. this is a crosscompile for arm
[16:20:24] <alex_joni> maybe I can find the ldd for that
[16:21:28] <alex_joni> hmm.. better, now only 156KB short
[16:21:57] <alex_joni> if ldd reports not a dynamic executable, that means I don't need any libs.. right?
[16:22:04] <jepler> yes
[16:22:04] <cradek> right
[16:22:10] <lerman> Busybox (I think that is the name) has a shell and minimal utlities.
[16:22:43] <alex_joni> lerman: busybox is included
[16:22:55] <jepler> $ objdump -p /bin/sh | grep NEEDED
[16:23:00] <cradek> what are you working on?
[16:23:09] <alex_joni> Everything: 1096 kilobytes
[16:23:35] <alex_joni> cradek: helping a friend compile some older 2.4.18 for an arm board
[16:23:52] <alex_joni> http://www.mizi.com/developer/s3c2410x/index.html
[16:26:36] <CIA-8> 03jepler 07TRUNK * 10emc2/docs/man/man9/.cvsignore: new generated files
[16:29:02] <CIA-8> 03alex_joni 07TRUNK * 10emc2/src/emc/kinematics/ (pumakins.c pumakins.h): kinemtics file for PUMA typed robots
[16:29:37] <CIA-8> 03alex_joni 07TRUNK * 10emc2/src/Makefile: compile puma kins
[16:32:57] <CIA-8> 03alex_joni 07TRUNK * 10emc2/configs/puma/ (README puma.ini puma.tbl puma.var puma_sim_6.hal): sample sim config for puma robots
[16:33:13] <alex_joni> awallin: are you around?
[16:35:26] <awallin> alex_joni: yeah, I'm grumbling about WordPress defects (http://trac.wordpress.org/ticket/3330) and (http://trac.wordpress.org/ticket/3744)
[16:35:41] <alex_joni> awallin: care to help me with a pyvcp panel ?
[16:36:00] <awallin> alex_joni: I can try :) (don't ask to ask. ask.)
[16:36:43] <awallin> although I'm not that familiar with the latest and greatest stuff from jepler and tomp
[16:37:19] <tomp> alex_joni: i'll help if i can
[16:37:22] <alex_joni> do you know of any widget that looks like a carspeed indicator?
[16:37:36] <awallin> that would be meter
[16:37:39] <alex_joni> I want to simulate 6 rotary joints
[16:38:06] <tomp> alex_joni: with bar_charts?
[16:38:12] <awallin> but the way meter is written now they don't "rotate" 360 degrees
[16:38:59] <awallin> http://www.linuxcnc.org/docs/html/hal/vcp/pyvcp_meter.png
[16:39:24] <awallin> it would be easy to allow the needle to rotate 360deg if that is what you want
[16:40:41] <awallin> alex_joni: write an OpenGL widget that draws your robot in 3D!
[16:44:27] <tomp> alex_joni: ok, you can look at draw_ticks in the meter ( in a loop from 0 to 11 ) which is never 360 deg, and the draw_ticks code in the dial widget dial which is always 360 degree
[16:44:50] <tomp> i'll look at a rotary degree display later, running out of terminals now :)
[16:45:11] <awallin> "compass" ?
[16:45:31] <tomp> do you need modulo 360? compass is good idea, esp when i get lost :)
[16:47:07] <tomp> alex_joni: need modulo 360? do puma joints rotate more than 1 rev?
[16:47:47] <awallin> I bet it get's a bit tricky with cables etc if you rotate more than 1rev
[16:48:08] <tomp> btw: there was a nice java puma simulator somewhere
[16:48:30] <tomp> bbl
[16:48:51] <fenn> alex_joni: look at how tomsrtbt is constructed
[16:48:59] <fenn> if you have enough ram to expand it that is
[16:50:32] <alex_joni> tomp: they shouldn't move than 360
[16:50:41] <tomp> k
[16:50:43] <alex_joni> fenn: less ram than that
[16:50:50] <alex_joni> tomp: they shouldn't move more than 360
[16:50:55] <tomp> kk
[16:51:12] <CIA-8> 03alex_joni 07TRUNK * 10emc2/src/Makefile: copy emc.nml to puma config,
[16:51:17] <CIA-8> 03alex_joni 07TRUNK * 10emc2/src/emc/kinematics/pumakins.c: copy emc.nml to puma config,
[16:54:14] <awallin> bbl
[16:55:27] <robin_sz> meep?
[16:57:33] <alex_joni> tomp: I can set the extents to [AXIS_0]MIN_LIMIT and [AXIS_0]MAX_LIMIT
[16:58:41] <alex_joni> once that will work of course :)
[16:59:06] <alex_joni> yay.. I removed tar and now it's 972kB :D
[16:59:27] <tomp> ok, those are the min & max of the meter, if degrees, the face could act that way ( like show 90 to 270 or whatever part of a circle you ask for )
[16:59:42] <alex_joni> yeah, they should be degrees
[17:20:55] <tomp> alex_joni: awallin: if i change range_ to (math.pi-.000001) the stock meter widget works like this... http://imagebin.org/7149 i can make range_ programmable via xml and default it to the std 2.5... any good?
[17:22:46] <tomp> lemme know where you want 0 degrees to be
[17:25:49] <owad> Any danger in soldering to the terminals of a snap-in resistor?
[17:27:25] <tomp> owad:solder to the thing it snaps into
[17:27:49] <owad> What would that be?
[17:29:04] <tomp> i gathered from yourquestion, this resistor is designed to snap into something
[17:29:40] <owad> oh shoot
[17:29:44] <owad> I meant capacitor
[17:29:46] <owad> sorry
[17:30:13] <tomp> oh, what wee spoke of yesterday: answer is yes, its ok
[17:30:46] <owad> found a handful on eBay that look cheap, so I'm thinking about going with those
[17:31:27] <tomp> they snap into holes on pcb AND get soldered, the snap is to hold em steady during wave solder
[17:32:38] <owad> oh ok. I've even soldered some of those, actually. (didn't know what the style was called)
[17:35:43] <tomp> the snap is kind of scissors spring that snaps into a hole. if you are in chicagoland you can goto allied near downtown or Tri-State Electronic Corp. 200 W Northwest Highway Mt. Prospect
[17:37:53] <owad> I'm in PA. Thinking about these: http://cgi.ebay.com/11000uf-80vdc-Cornell-Dubilier-Capacitors-NEW-L-K_W0QQitemZ130075613125QQihZ003QQcategoryZ111603QQssPageNameZWDVWQQrdZ1QQcmdZViewItem
[17:39:10] <tomp> owad: good deal at <20$
[17:39:29] <cradek> wonder if 9350 is the date code
[17:39:37] <cradek> if so they're a little on the old side
[17:40:19] <cradek> (15 yrs unused is pretty old for an electrolytic)
[17:40:44] <cradek> more like 13 I guess
[17:40:59] <owad> What would the '50' part be?
[17:41:05] <cradek> week
[17:43:42] <cradek> what voltage do you intend to use?
[17:43:59] <owad> 35v
[17:44:22] <tomp> 35V 2 amp stepper supply
[17:44:23] <cradek> oh it's probably ok then
[17:44:34] <cradek> still, look away the first time you turn it on :-)
[17:44:49] <owad> oh, I stand on the other side of the room! :-)
[17:45:18] <cradek> good enough
[17:45:30] <tomp> jepler: awallin: plz look at http://imagebin.org/7138 and http://imagebin.org/7139
[17:45:36] <cradek> bbl
[17:46:00] <lerneaen_hydra> testing
[17:47:19] <owad> googling, it looks to me like you're reading the date code correctly
[17:48:00] <owad> The other contender is this: http://cgi.ebay.com/4-pcs-10000uF-63V-Rubycon-ALUMINUM-ELECTROLYTIC-CAPACIT_W0QQitemZ230084456954QQihZ013QQcategoryZ116999QQcmdZViewItem
[17:48:14] <owad> but that's got to ship from asia
[17:49:48] <ejholmgren> nice cig
[17:51:04] <notranc> Hi guys, comment on caps; I worked with computer restoration team on PDP-1. We had a bunch of 40+ years old caps. We used cap restoration method with supplying voltage from 0 to whatever and limited current (less than 1mA?, can't remember exactly). None of the large caps failed and they were all close to original specs! Computer is working without any power (or other) problems since then. For what's worth.
[17:53:45] <owad> Hi notrane. Thanks for the tip. What PDP-1 were you restoring?
[17:54:15] <owad> Computer History Museum?
[17:57:51] <robin_sz> notranc, thats pretty common too among vintage radio restorers, hang the caps off a 10K resistor and some volts for a week or so
[17:58:02] <robin_sz> notranc, it reforms the dielectric
[17:59:26] <notranc> Agree, I'm not sure if that applies to newer caps with "modern chemicals". Old caps are likely a bit more (?) poisonous. In either case, you don't want to vent them :-)
[18:00:49] <robin_sz> shrug
[18:01:18] <robin_sz> but yeah, letting the goo out is not going to help them work
[18:01:31] <jepler> tomp: try this
[18:01:31] <jepler> +++ vcpparse.py3 Feb 2007 18:01:22 -0000
[18:01:37] <jepler> if container==pyvcp0:
[18:01:37] <jepler> -widget.pack(side='top', fill=BOTH)
[18:01:37] <jepler> +widget.pack(side='top', fill=BOTH, expand=YES)
[18:02:07] <alex_joni> jepler: is Ctrl-Home supposed to work in AXIS?
[18:02:16] <tomp> jepler: great will try now!
[18:02:18] <jepler> alex_joni: yes
[18:02:22] <alex_joni> it doesn't work here :(
[18:02:32] <notranc> owad <--- http://www.pdp-1.org/ That PDP-1 is at http://www.computerhistory.org/. We have regular demo over there.
[18:03:07] <jepler> alex_joni: I just tested it on TRUNK with sim/axis
[18:03:24] <alex_joni> this starts in joint mode..
[18:03:28] <alex_joni> maybe that matters?
[18:03:31] <jepler> btfoom
[18:03:36] <jepler> I've only tested on a trivial machine
[18:03:42] <alex_joni> can you try TRUNK puma config?
[18:03:45] <alex_joni> it's also a sim
[18:03:46] <owad> notranc, follow the ClassicCmp mailing list?
[18:03:46] <alex_joni> :)
[18:04:46] <notranc> No, I spend too much time on old computers, including my 4 years old PC :-)
[18:05:40] <owad> I subscribe, but I'm currently 6,654 emails behind.
[18:06:45] <notranc> No wonder, there must be a lot to say about old (vintage) computers.
[18:06:48] <jepler> alex_joni: after adding HOME_SEQUENCE=0 to all the [AXIS_n] blocks in puma.ini, pressing the Home All button sends the joints to <0,0,0,0,1,0>
[18:07:14] <jepler> alex_joni: switching to cartesian coordinates, I get X=350, Y=70 Z=-400, R=180 P=57.296, W=0
[18:07:16] <alex_joni> oh, so it doesn't work without HOME_SEQUENCE.. didn't know that
[18:07:25] <jepler> alex_joni: without HOME_SEQUENCE, AXIS believes the joints must be homed independently
[18:07:26] <alex_joni> jepler: switch back to joints, and jog :)
[18:07:41] <alex_joni> jepler: bet you'll like the circles
[18:08:09] <jepler> I suppose you lost that patch i sent you to try, for fixing jog speed of joints 3-5
[18:08:23] <alex_joni> yeah, but I made units mm/deg
[18:08:29] <alex_joni> so it shouldn't matter
[18:08:54] <alex_joni> mm/deg have the advantage they are both 1
[18:11:02] <alex_joni> awallin: there's a bug for <bar>, or maybe I don't know how to use it
[18:11:08] <alex_joni> I set max to 180, mnin to -180
[18:11:20] <alex_joni> when it starts up with 0 it doesn't colour half the bar..
[18:12:17] <jepler> alex_joni: uh oh, the joint positions have all gone to nan
[18:12:33] <alex_joni> now you keep all parts
[18:13:16] <alex_joni> jepler: slowly jog the other way around
[18:13:22] <alex_joni> if you remember what joint you were jogging
[18:14:17] <alex_joni> jepler: something is not working right, but I can't figure it out..
[18:14:51] <alex_joni> (when I home joint 0, and jog joint 0 from 0 to 6.3 it does a full circle on the preview..)
[18:14:59] <alex_joni> feels like some scaling is off in the kins
[18:15:29] <tomp> alex_joni the code for bar sez... # FIXME specifying negative startval or endval does not work, plz look at http://imagebin.org/7149 in meantime, i will look at the neg min problem after FlowDesigner finally builds :)
[18:15:33] <alex_joni> 6.3 very much sounds like 2*PI
[18:18:02] <tomp> alex_joni: the suggestion just posted worked with -180 to +180 also
[18:18:16] <alex_joni> what suggestion?
[18:18:24] <tomp> alex_joni the code for bar sez... # FIXME specifying negative startval or endval does not work, plz look at http://imagebin.org/7149 in meantime, i will look at the neg min problem after FlowDesigner finally builds :)
[18:18:53] <tomp> and before that: alex_joni: awallin: if i change range_ to (math.pi-.000001) the stock meter widget works like this... http://imagebin.org/7149 i can make range_ programmable via xml and default it to the std 2.5... any good?
[18:19:33] <alex_joni> tomp: you need to slow down on me.. I didn't understand half of what you said
[18:20:32] <tomp> alex_joni: the standard meter widget can work like http://imagebin.org/7149 with a minor change
[18:20:41] <alex_joni> change to the source?
[18:20:47] <tomp> yes, just 1 line
[18:21:15] <alex_joni> tomp: no need to change the source.. I just chagned s/bar/meter/
[18:21:20] <alex_joni> and it looks good
[18:21:24] <tomp> k
[18:27:41] <CIA-8> 03alex_joni 07TRUNK * 10emc2/configs/puma/ (puma.xml puma_postgui.hal puma.ini): pyvcp joint preview
[18:42:54] <tomp> jepler: re changes to vcpparse.py, great! yes, it was the parent would allow it, great thanks! ( i gots BIG buttons now )
[18:51:43] <awallin> tomp: are you committing these changes into cvs ? you probably should get cvs access since you seem to be keen on improving pyvcp
[18:52:42] <tomp> awallin: i just make suggestions, like last one isnt needed, so it'd just f up cvs
[18:56:34] <awallin> tomp: ok, but there is a real chance that some of your work is lost when I or jepler or someone else doesn't have the time to look it through and possibly commit it. random pastebin:s are kind of hard to keep track of...
[18:58:22] <tomp> awallin: i asked to become a dev and the jist was i had to be sponsored and evaluated (not a quote, but what i took it to mean ) , years ago, i thought i was a dev, but apparently not. this is no place for this discussion, thanks tho
[19:01:41] <awallin> alex_joni: are you preparing an update to a makefile, with a cvs update from a few minutes ago I get "no rule to make target configs/puma/emc.nml"
[19:02:15] <alex_joni> awallin: I commited that already
[19:02:21] <alex_joni> but I think you updated inside src/
[19:02:34] <alex_joni> not all emc2/ and thus you don't have the new configs dir
[19:02:38] <awallin> that might be the case...
[19:02:52] <awallin> trying again
[19:04:40] <alex_joni> logger_emc: bookmark
[19:04:40] <alex_joni> Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2007-02-03.txt
[19:07:30] <alex_joni> * alex_joni can't stop laughing at this: http://www.youtube.com/watch?v=wplyCogv3to
[19:08:13] <CIA-8> 03jepler 07TRUNK * 10emc2/lib/python/vcpparse.py: make the child of pyvcp0 able to fill the window
[19:09:15] <jepler> it's too bad "tkinter3000" and "aggdraw" aren't packaged in ubuntu
[19:09:35] <jepler> with them you can do nice antialiased gizmos in tkinter apps: http://emergent.unpy.net/index.cgi-files/sandbox/pyvcp-agg.png
[19:10:46] <tomp> http://www.tkzinc.org/tkzinc/index.php
[19:11:59] <jepler> that doesn't seem to be packaged for ubuntu either
[19:13:09] <tomp> svg is the trick to get pretty, scalable & antialiasing
[19:17:22] <jepler> I'm not sure I agree. there are plenty of ways to specify curves and lines -- "the trick" is finding the renderer, not the format
[19:20:15] <jepler> and xml is a terrible format to use inside of a single program for almost any purpose whatever
[19:27:04] <jepler> unfortunately, that one dial seems to take 10% of the CPU on this machine
[19:27:07] <jepler> not high performance
[19:41:34] <tomp> my system went from 5% at idle to 10% while cranking 1 of 4 displayed dials, is that bad?
[19:42:02] <tomp> changing the size of a window cost more
[19:45:26] <awallin> here's an improvement to pyvcp_bar so it can handle negative numbers: http://imagebin.org/7151
[19:45:45] <awallin> alex_joni: that's what you wanted?
[19:46:26] <awallin> the bar always starts at 0.0, and goes to whatever value is set. if it's negative the bar extends to the left, if it's positive it grows to the right
[19:47:20] <jepler> alex_joni: I updated and get dials with the puma cfg, but they don't seem to be very useful
[19:47:36] <jepler> halcmd suggests the joint values are all <1, so the needles don't appear to move from 0
[19:50:19] <CIA-8> 03awallin 07TRUNK * 10emc2/lib/python/pyvcp_widgets.py:
[19:50:19] <CIA-8> improve bar widget so it handles negative numbers better
[19:50:19] <CIA-8> comment out percentage stuff from bar, not needed anymore
[19:50:19] <CIA-8> comment out importing some modules twice (redundant?)
[19:55:14] <alex_joni> jepler: there is something wrong with the joints & kins
[19:55:26] <alex_joni> the output for joints is 0..6.3 for a full turn
[19:55:34] <lerneaen_hydra> anybody here that knows java?
[19:55:41] <alex_joni> so I suspect it's a probelm of converting from 2*Pi to 360 degrees
[19:56:07] <awallin> lerneaen_hydra: I did take a class some years ago...
[19:56:25] <lerneaen_hydra> awallin; feeling helpful? ;)
[19:57:15] <awallin> lerneaen_hydra: I bet I won't be able to anwer if you have some hard Q
[19:57:40] <lerneaen_hydra> awallin; I've only had java for two weeks now, so it's probably (hopefully) simple
[19:58:33] <lerneaen_hydra> my compiler doesn't like row 8 at all, http://www.pastebin.ca/338956, it says it can't find the symbol and points to the instance of "new"
[20:01:47] <CIA-8> 03jepler 07TRUNK * 10emc2/src/hal/user_comps/pyvcp.py: make -c optional
[20:04:21] <awallin> lerneaen_hydra: sorry, I can't see anything wrong with that... long time since I did anything in java
[20:05:00] <lerneaen_hydra> ok, thanks anyway
[20:05:10] <lerneaen_hydra> were there any new-coder things that you saw?
[20:05:12] <awallin> jepler, alex_joni: has there been any more talk about including HAL commands in XML, possibly so that pyvcp xml files would be more standalone, i.e. don't require a separate .hal or halrun file
[20:05:19] <lerneaen_hydra> like this is a common thing that people do that's not good
[20:06:08] <jepler> awallin: not that I'm aware of. in fact, I'm against having two different languages (halcmd and xml) for specifying how to load and connect HAL components.
[20:06:10] <awallin> lerneaen_hydra: looks ok to me, but I'm a beginner like you
[20:06:28] <lerneaen_hydra> ok, thanks anyway
[20:06:36] <jmkasunich> hi guys
[20:06:46] <lerneaen_hydra> hey jmk
[20:06:48] <awallin> jepler: what about wrapping halcmd statements in XML?
[20:06:54] <alex_joni> lerneaen_hydra: not sure if java doesn't expect the class you instantiate to be declared before the other one
[20:07:08] <lerneaen_hydra> alex_joni; huh?
[20:07:39] <awallin> lerneaen_hydra: move the whole of class Bird {} to before ass7 ?
[20:07:56] <jepler> awallin: python's os.system, os.popen, or subprocess module should make that doable
[20:08:08] <lerneaen_hydra> oh, that's not the problem, I've done other stuff where the reference to the class is after the code
[20:08:14] <lerneaen_hydra> err, the class is after the reference
[20:08:42] <awallin> lerneaen_hydra: public class Bird have you tried that?
[20:09:05] <jepler> lerneaen_hydra: my compiler gives this error:
[20:09:05] <jepler> ass7.java:8: error: Can't find constructor 'Bird(Ljava/lang/String;Ljava/lang/String;ILjava/lang/String;Ljava/lang/String;)' in type 'Bird'.
[20:09:16] <jepler> lerneaen_hydra: make sure the argument types all match the constructor you give later on
[20:09:36] <lerneaen_hydra> jepler; oh, that's not at all the same error I got O_o
[20:09:41] <jmkasunich> does adding a level of nesting in xml mean a lot of work?
[20:09:45] <jmkasunich> I was thinking of something like this:
[20:10:16] <jmkasunich> <halpin>"button-sample"<connect_to>"some-signal"</connect_to></halpin>
[20:10:31] <lerneaen_hydra> awallin; it only needs to be public class bird if it's in a seperate file
[20:10:36] <awallin> jepler: OK, that might be interesting for some applications? a few days ago rayh ans SWP seemed to think that a migration towards one single (big!) configuration file (possibly XML) would be desirable
[20:10:38] <jmkasunich> obviously the connect_to part would be optional
[20:11:18] <lerneaen_hydra> jepler; ah, right
[20:11:20] <lerneaen_hydra> I got it
[20:11:21] <awallin> jmkasunich: traversing the xml tree is not a problem, we just need to settle on a syntax/stucture how it's done.
[20:11:32] <lerneaen_hydra> bird() wants a boolean while I've entered a string
[20:12:03] <awallin> jmkasunich: one problem is that XML is very good at describing trees, but not cyclic graphs, or net's where things connect arbitrarily to eachother
[20:12:06] <jmkasunich> is what I wrote legit xml, or does the presence of <connect_to> mean that the pin name now needs to be wrapped in tags?
[20:13:25] <jmkasunich> hmm, I just realized... <halpin>some-name</halpin> assumes that the widget is only going to export one pin
[20:14:11] <awallin> jmkasunich: if there are many, the widget will create them: some-name.out1 some-name.in1 etc (you don't have control over naming of out1 in1 etc)
[20:14:42] <awallin> jmkasunich: XML attributes are allowed <halpin connect="mysig">"mypin"</halpin>
[20:14:50] <jmkasunich> so <halpin> is really the prefix?
[20:15:09] <awallin> jmkasunich: prefix in what context/meaning?
[20:15:32] <jmkasunich> mux2.0.sel1 is a typical pin name
[20:15:54] <awallin> ah, yes I understand. it works like a prefix. but for some widgets there is only one pin, so it would be named comp.prefix
[20:15:59] <jmkasunich> mux2 is the prefix (in this case, the block type) "0" is the instance ID, and "sel1" is the pin name
[20:16:25] <awallin> pyvcp is not entirely consistent with this, but it can be made to be like this
[20:16:46] <jmkasunich> I'm not sure that we want to push pyvcp to match the existing model
[20:17:11] <awallin> something like: hal_component.pyvcp_widget.<widget_instance>.pin_name
[20:17:24] <awallin> where <widget_instance> is an integer
[20:17:30] <jmkasunich> huh? thats quite a handfull
[20:17:46] <awallin> mypanel.button.0.in
[20:18:06] <jmkasunich> ok, so hal_component is specified by the user in the xml
[20:18:17] <jmkasunich> pyvcp-widget is automatic
[20:18:21] <jmkasunich> instance number is automatic
[20:18:35] <jmkasunich> and pin-name comes from the halpin? or is that automatic?
[20:18:43] <awallin> hal_component comes from the command line: pyvcp -c mycomp myfile.xml
[20:18:51] <jmkasunich> ok
[20:19:10] <CIA-8> 03jepler 07TRUNK * 10emc2/src/hal/halmodule.cc: allow the prefix for new pins to be retreived
[20:19:58] <awallin> right now pyvcp-widget is automatic if the user does not specify anything in <halpin>, but get's replaced by anything the user might specify. modifying the example above: mypanel.coolant-on
[20:20:11] <jmkasunich> I think for the next generation of hal, the "mux2.0" part of "mux2.0.sel1" should be (optionally) user configurable
[20:20:33] <awallin> if <halpin> is nonempty then we forget all about numbers and the "in"
[20:21:14] <jmkasunich> are the pin naming rules written down in some doc that I should be reading instead of asking these questions?
[20:21:48] <awallin> jmkasunich: it's very much just thrown together right now with no formal naming conventions
[20:22:12] <jmkasunich> hal itself is only slightly better
[20:22:36] <jmkasunich> if we're gonna work out a consistent system, we should work them out together, not independently
[20:23:17] <jmkasunich> the next generation of HAL (maybe in emc 2.2) will seperate the "load" and "create instance" actions from each other
[20:23:34] <jmkasunich> right now, when you do "loadrt mux2 count=2" it loads the module and creates two muxes
[20:23:48] <jmkasunich> called mux2.0 and mux2.1, each with pins in0, in1, sel
[20:23:58] <awallin> jmkasunich: yes. I'm sure pyvcp will be fairly easy to modify to a reasonable naming convention that you or someone else will write down
[20:24:04] <jmkasunich> the next version, "loadrt mux2" will only load, not create
[20:24:14] <jmkasunich> and "newinst" will create an instance
[20:24:43] <jmkasunich> "newinst mux2 foo" will create a mux2 instance called "foo" instead of "mux2.0", but the pin name portion won't change
[20:24:52] <jmkasunich> foo.in0, foo.in1, foo.sel
[20:25:12] <jmkasunich> if you don't specify a name "newinst mux2", then you'll get the old style name
[20:25:54] <jmkasunich> right now all of this is just conjecture, not decided yet
[20:26:07] <jmkasunich> I want to discuss how/if it maps to pyvcp pin naming
[20:26:37] <awallin> will it be possble to load and/or create instances in separate files and in separate places. right now I remember there is a limitation of only one loadrt call for each component
[20:26:41] <jmkasunich> in pyvcp, you have the component name, the widget name, and the pin name(s)
[20:27:00] <jmkasunich> right - the whole newinst thing is to fix that limitation
[20:27:01] <awallin> jmkasunich: yes, that's the basic idea I think
[20:27:14] <jmkasunich> you can loadrt, then newinst, then much later do another newinst
[20:27:28] <awallin> jmkasunich: that's a good improvement
[20:27:51] <jmkasunich> today, if you "loadrt mux2 count=2" and later want a third one, you need to unload (which deletes the first two) then reload with count=3, and reconnect the first two
[20:28:42] <jmkasunich> back to pyvcp: you have the component name, the widget name, and the pin name(s)
[20:28:48] <awallin> yes
[20:28:56] <jmkasunich> its not clear to me which one "halpin" sets
[20:29:00] <jmkasunich> pin name, or widget name
[20:29:29] <jmkasunich> if pin name ( which halPIN ) implies, then which pin (for widgets that have more than one)?
[20:29:40] <awallin> right now, it depends on the widget. obviously if the widget has two or more pins, you must only set widget-name to what the user specified
[20:30:13] <awallin> if the widget only has one pin, you can remove one level of naming and only have component_name.userspec_name (no pin name)
[20:30:42] <jmkasunich> earilier you said: XML attributes are allowed <halpin connect="mysig">"mypin"</halpin
[20:30:42] <awallin> but I don't know if that's logical
[20:31:07] <awallin> jmkasunich: yes, that's for general XML, not neccessarily related to pyvcp
[20:31:38] <jmkasunich> today, you do <scale><orient>horizontal</orient><halpin>foo</halpin></scale>
[20:31:53] <awallin> yes
[20:32:02] <jmkasunich> why not <scale>orient=horizontal halpin=foo</scale>
[20:32:11] <jmkasunich> I that just considered bad xml form?
[20:32:36] <awallin> that is not easy at all to parse with the built-in xml parser, I would get the string in the middle into my program and have to do the parsing myself
[20:32:40] <jmkasunich> I guess it implies that orient and halpin are "leaves" on the tree, nothing under them
[20:32:58] <jepler> actually I recently added support for: <scale orient="horizontal" halpin="foo"/>
[20:32:58] <jmkasunich> oh, the parser doesn't handle attributes?
[20:33:26] <awallin> jmkasunich: it handles attributes (what jepler just wrote), but not text strings (what you wrote)
[20:33:47] <jepler> * jepler disappears again
[20:33:47] <jmkasunich> oh, attributes are _inside_ the <>
[20:33:58] <awallin> yes
[20:34:35] <jmkasunich> so what jepler wrote is a scale tag, with attributes inside, and since there are no children, the closing /> replaces ></scale>
[20:35:27] <awallin> yes.
[20:35:47] <jmkasunich> I guess I have issues with halpin
[20:36:06] <jmkasunich> as written, it only makes sense for widgets with a single pin
[20:36:36] <jmkasunich> which included all of the original widgets
[20:36:44] <awallin> most of them
[20:36:48] <jmkasunich> are there any multi-pin widgets yet>?
[20:37:10] <awallin> jsut for fun, I did both an int and a float output from Scale
[20:37:14] <awallin> just
[20:37:44] <jmkasunich> so its not clear which pin would be modified by halpin
[20:37:50] <jmkasunich> (for that widget)
[20:38:17] <awallin> if you sepcify halpin="my_scale" I think they are named comp.my_scale.i and comp.my_scale.f
[20:38:41] <jmkasunich> I see - you add another level to the name
[20:39:17] <jmkasunich> IMO, that usage should not be called halpin - you are basically changing the name of the widget itself
[20:39:31] <jmkasunich> I consider the last part of the full name to be the pin name
[20:39:54] <jmkasunich> just like the last part of /usr/share/doc/foo/bar.doc is the filename, and the rest is path
[20:39:53] <awallin> yes, but you also remove one or two, if halpin is not defined they are named comp.scale.n.i and comp.scale.n.f
[20:40:19] <awallin> doc is the filename ?
[20:40:23] <jmkasunich> exacttly - the default widget name is "scale.n", and halpin changes the widget name
[20:40:28] <jmkasunich> bar.doc
[20:40:59] <jmkasunich> (for linux files, / is the separator, not . )
[20:42:23] <awallin> so you would rather have the user-specified name replace the instance count n: comp.scale.feed-override.out-int
[20:42:28] <jmkasunich> I think <scale> stuff </scale> should create "compname.scale.N.pin-name(s)", and <scale name=foo> stuff </scale> should create "compname.foo.pin-name(s)"
[20:42:43] <jmkasunich> I'd replace scale and N
[20:43:13] <jmkasunich> the N only exists because there might be multiple scales
[20:43:28] <awallin> yep, that's about how it's done right now, it's just that pin-name(s) is sometines omitted when there is only one pin
[20:44:15] <jmkasunich> ah - looking at it that way makes more sense
[20:44:43] <jmkasunich> it never occurred to me, because except for VCP components, a comp with only one pin is quite rare
[20:44:58] <jmkasunich> most comps process a signal in some way, so they have in and out
[20:45:47] <awallin> for vcp comps the other end is the graphical widget
[20:45:51] <jmkasunich> right
[20:46:13] <jmkasunich> the same would apply to hardware drivers - the other end is the hardware
[20:46:27] <jmkasunich> but most hardware has multiple channels, so still multiple pins
[20:47:09] <jmkasunich> * jmkasunich realizes that many of the existing pin numbering conventions are not quite right
[20:47:23] <awallin> so far there hasn't been requests for many multi-pin vcp widgets
[20:47:49] <jmkasunich> parport.0.pin-01-in and parport.0.pin-01-in-not should actually be: parport.0.pin-01.in and parport.0.pin-01.in-not
[20:48:13] <jmkasunich> parport.0 is the comp, pin-01 is the "widget", and "in" and "in-not" are the pin names
[20:51:15] <anonimasu> hm
[20:53:25] <jmkasunich> getting back to pyvcp: I now think the the _behavior_ of halpin is fine - but its name is wrong - its actually a widget name, not a pin name
[20:54:11] <jmkasunich> I think the actual pin name should never be changed (ok, maybe _never_ is a bit strong)
[20:54:22] <jmkasunich> but think about the canonical devices
[20:54:53] <jmkasunich> an encoder device should always have the same pins, regardless of whether its a mesa encoder, a motenc encoder, and stg encoder, or a software based encoder
[20:55:13] <jmkasunich> and you just change the first part to switch from one to the other
[20:56:08] <jmkasunich> being able to set the instance name, or some aliasing capability, would let you switch things around easier
[20:56:31] <jmkasunich> (I'm not rambling about hal in general, not pyvcp_
[20:56:41] <awallin> just seems a bit redundant to always append the pin-name even if there is only one: halpin="" (not defined): mypanel.button.0.out halpin="coolant": mypanel.coolant.out
[20:56:55] <jmkasunich> alias x-encoder m5i20.enc.1
[20:57:10] <jmkasunich> linksp x-pos x-encoder.postiion
[20:57:19] <jmkasunich> linksp x-index-en x-encoder.index-enable
[20:58:09] <jmkasunich> the "drop pin name if there is only one" option seems ok to me
[20:58:37] <jmkasunich> but we should think of it as "dropping the pin name and using the widget name only", not "changing the pin name to the widget name"
[21:00:23] <robin_sz> * robin_sz upates his prediction calendar
[21:00:38] <robin_sz> eraser ....
[21:00:46] <jmkasunich> what are you predicting?
[21:00:51] <robin_sz> pencil ...
[21:01:19] <robin_sz> the day vcp and hal replace the other GUI's :)
[21:01:26] <robin_sz> I was just moving the date a bit closer
[21:01:36] <jmkasunich> when heck freezes over
[21:01:56] <anonimasu> hehe
[21:02:04] <alex_joni> we need a 3D pyvcp opengl widget
[21:02:09] <robin_sz> * robin_sz shrugs
[21:02:14] <jmkasunich> no we don't
[21:02:22] <alex_joni> jmkasunich: I'm kidding obviously
[21:02:52] <robin_sz> theres been a lot of widget activity recently
[21:03:00] <jmkasunich> actually, I have wanted hal->opengl for a while now, but not for g-code preview or other EMC GUI stuff
[21:03:33] <jmkasunich> I want to be able to build a crude 3d model of a machine out of boxes and cylinders and cones and such, and connect it inplace of real mtoros
[21:03:36] <jmkasunich> motors
[21:03:50] <alex_joni> jmkasunich: I need that too
[21:03:53] <jmkasunich> for visualizeing hexapods and pumas and such
[21:03:55] <alex_joni> I want to debug the puma kins..
[21:03:55] <robin_sz> for simulation?
[21:04:25] <jmkasunich> robin_sz: yes
[21:04:38] <anonimasu> nice
[21:05:03] <robin_sz> I guess in some situations, it could be useful for predicting stuff ...
[21:05:08] <jmkasunich> we're never gonna get non-trivial kin machines working _really_ well until multiple developers have such machines
[21:05:17] <jmkasunich> and we're never gonna have real machines like that
[21:05:25] <robin_sz> liek a SCARA type bot with some sort of "no go" area
[21:05:36] <anonimasu> jmkasunich: how about a _kit_ machine design that anyone can mill
[21:05:41] <anonimasu> with steppers and stuff..
[21:05:59] <anonimasu> ~rant
[21:06:06] <jmkasunich> anonimasu: that is still a massive effort, both to designa and build
[21:06:20] <anonimasu> jmkasunich: hm, you dont need gobs of torque and stuff..
[21:06:25] <jmkasunich> and when you are done, you still only have ONE kind of non-trivial kins machine
[21:06:25] <alex_joni> jmkasunich: I'd be interested in a sim
[21:06:33] <robin_sz> sort of a DIY sherline?
[21:06:35] <anonimasu> yeah
[21:06:37] <jmkasunich> a simulator would give you many kinds for the same effort
[21:07:11] <robin_sz> anonimasu, the trouble with the sherline is it needs castings
[21:07:22] <anonimasu> robin_sz: yeah, but most people here can mill 1cm alu stuff..
[21:07:27] <jmkasunich> actually, sherline uses extrusions, but the point is still valid
[21:07:40] <robin_sz> 'k
[21:08:16] <jmkasunich> anonimasu: assume the design work for the kit mill is done, and the opengl sim framework is also done
[21:08:28] <robin_sz> to be doable, you'd ideally want to open-source the plans and arrange for various suppliers in various countries to produce kits of part
[21:08:28] <robin_sz> s
[21:08:39] <jmkasunich> I bet I given the framework, I can have my hexapod working before you finish 5% fo the parts for your hexapod
[21:08:41] <anonimasu> robin_sz: well, stepper nema XX are standard..
[21:08:57] <robin_sz> then you are paying for machine time on a fair rate ... not "we think we can get away with this much" ...
[21:09:17] <anonimasu> that's like saying I can bet if im given the drawings..
[21:09:31] <jmkasunich> let's take the discussion about hal nameing (and its affect on pyvcp) to -devel
[21:09:44] <jmkasunich> so kit machines can stay here, and we won't collide with each other
[21:10:10] <robin_sz> jmkasunich, strangely enough I saw that ABB have a hexapod in production
[21:10:18] <anonimasu> jmkasunich: what's up with you, you talk like some emo kid today..
[21:10:24] <anonimasu> it was just a suggestion nothing more.
[21:10:39] <jmkasunich> nothing wrong with that
[21:10:49] <jmkasunich> and it _would_ be a good project
[21:10:53] <anonimasu> 22:06 < jmkasunich> we're never gonna get non-trivial kin machines working _really_ well until multiple developers have such machines
[21:10:56] <anonimasu> regarding that..
[21:11:15] <jmkasunich> but it doesn't meet my needs (or alex's) for simulating a non-trivial kins machine
[21:11:44] <jmkasunich> I don't have space, time, or money to build a real non-trivial kins machine, even if there was a kit available tomorrow
[21:12:00] <awallin> has anyone seen any real cutting done with the hexapods that only use strings for the struts? or are those just demo-toys?
[21:12:04] <jmkasunich> and I want to try hexapod, scara, puma, and others
[21:12:13] <jmkasunich> I can't have all those machines sitting around gathering dust
[21:12:28] <ottos> good day gents..
[21:12:44] <jmkasunich> awallin: cutting no (that I'm aware of)
[21:12:53] <jmkasunich> welding and material handling, yes
[21:12:54] <anonimasu> hm I've seen drilling..
[21:12:56] <awallin> jmkasunich: alex_joni: have you done some work on a simulator for non-triv kins?
[21:13:01] <robin_sz> http://www.abb.com/product/seitp327/02ea8c165394b8c8c1256efc0028eb74.aspx
[21:13:11] <robin_sz> not exactly a hexapod, more a tripod I guess
[21:13:30] <anonimasu> http://www.cuttingedgecnc.com/Movies/ffmm.wmv
[21:14:03] <jmkasunich> I've seen a video of that ABB machine (or one like it) sorting hot-dogs
[21:14:40] <ottos> sorry to bother I'm getting a dependancy error while installing 2.1 on the 5.10 Ubuntu system any workarounds..?
[21:14:48] <jmkasunich> the machine in the ffmm video - I want one!
[21:14:56] <jmkasunich> gimme! now!
[21:15:02] <jmkasunich> (that is just cool)
[21:15:20] <jmkasunich> ottos: we'll try to help, need details
[21:15:34] <jmkasunich> installing packages, or from source?
[21:15:37] <robin_sz> bah, wmv
[21:16:33] <ottos> installed the generic ubuntu 5.1 and when executing install script getting python 2.4 dependency error..
[21:16:58] <jmkasunich> can you paste the exact error message? either here if its short, or at pastebin.ca if its long
[21:17:35] <ottos> one sec..,
[21:18:40] <alex_joni> awallin: not yet
[21:19:00] <alex_joni> jmkasunich: flex-picker?
[21:19:14] <awallin> alex_joni: do you have the math for puma/hexapod readily available ?
[21:19:28] <alex_joni> both are commited to TRUNK
[21:19:41] <anonimasu> now that's something you could display with vtk easily..
[21:19:41] <alex_joni> but I can't understand much from them
[21:19:42] <awallin> oh, ok
[21:19:49] <awallin> ;)
[21:20:02] <jmkasunich> XML attributes are allowed <halpin connect="mysig">"mypin"</halpin
[21:20:06] <jmkasunich> oops
[21:20:28] <jmkasunich> http://www.youtube.com/watch?v=ADLMXMxLRy8
[21:20:48] <jmkasunich> pastries, not hot dogs, but thats the flexpicker in action
[21:21:25] <crepincdotcom> ok I've been looking around for the past few minutes.... does the EMC Ubuntu system really not include sshd? Theres no startup script for it
[21:21:33] <anonimasu> no
[21:21:42] <crepincdotcom> :-(
[21:21:45] <anonimasu> if I remember it right you need to install it
[21:21:51] <anonimasu> apt-get install ssh-server
[21:21:59] <crepincdotcom> never in my life have a met a linux distro that didnt include it.... ok then
[21:21:59] <anonimasu> i _think_ that's right..
[21:22:01] <ottos> sorry don't have java installed on that machine but in large all ok up to following packages have unmet dependencies , python2.4-numarray
[21:22:10] <crepincdotcom> thanks anonimasu I'll work on it
[21:22:11] <jmkasunich> ubuntu is meant to be a desktop linux for non-geeks
[21:22:21] <anonimasu> otherwise you can find it with aptitude or the xwin installer
[21:22:25] <anonimasu> also, it's a space issue right?
[21:22:28] <jmkasunich> it doesn't even include a compiler by default
[21:22:38] <jmkasunich> but all the good stuff is available as packages
[21:22:47] <anonimasu> for the livecd I mean
[21:23:04] <crepincdotcom> iv never used this apt-get thing
[21:23:14] <anonimasu> crepincdotcom: You can do it graphically somehow but I've forgotten how
[21:23:14] <anonimasu> :D
[21:23:14] <crepincdotcom> i tryed apt-get install xchat but it couldn't find the package
[21:23:18] <jmkasunich> synaptic is the gui version
[21:23:24] <crepincdotcom> ewwww graphics :-p
[21:23:27] <jmkasunich> synaptic has a nice search tho
[21:23:34] <crepincdotcom> oh does it?
[21:23:38] <crepincdotcom> * crepincdotcom goes to find
[21:23:39] <jmkasunich> search for packages by words in the name or description
[21:23:52] <jmkasunich> I use apt-get when I know exactly what package I want
[21:23:53] <anonimasu> jmkasunich: doing apt-get install thing* and breaking.. it before it installs stuff is a pain :D
[21:24:02] <jmkasunich> synaptic when I just have a general idea
[21:25:06] <jmkasunich> crepincdotcom: don't have to go far to find synaptic
[21:25:16] <crepincdotcom> heh i love how the load average on the box sits at 2.54. i guess thats due to the real-time things
[21:25:25] <crepincdotcom> yeah i got synaptic open
[21:26:09] <jmkasunich> ottos: sorry, I haven't forgotten you
[21:26:21] <jmkasunich> its just that I'm not very familiar with the python stuff
[21:26:22] <alex_joni> apt-cache search foo (searches just like synaptic)
[21:26:46] <crepincdotcom> oh sweet
[21:26:46] <crepincdotcom> thanks
[21:28:45] <anonimasu> hm, what's duraluminium number wise?
[21:29:56] <jmkasunich> a robot with a fish for a brain: http://www.youtube.com/watch?v=ADLMXMxLRy8
[21:31:09] <ottos> <jmkasunich> is the msg of any use..?
[21:33:13] <jmkasunich> it means you don't have the package python-2.4 numarray installed
[21:33:24] <jmkasunich> I'm not sure why it didn't automatically try to install that
[21:33:33] <jmkasunich> are you at the ubuntu machine right now?
[21:33:41] <ottos> no at my ws..
[21:33:52] <ottos> 5 meters away..
[21:33:52] <jmkasunich> is the ubuntu box on the net?
[21:33:54] <ottos> yes.
[21:34:06] <jmkasunich> can you go there and open a chat client?
[21:34:19] <ottos> can try.. what are we looking for.,.?
[21:34:34] <jmkasunich> I'll walk you thru it
[21:34:50] <jmkasunich> thats why I want you at the computer and with an IRC client open
[21:35:47] <ottos> ok.. let me try.. my screen is an old lcd 640x480 at that cnc machine so it's kinda pain ...
[21:35:52] <crepincdotcom> this is the strangest ircd I'v ever met...
[21:37:21] <jmkasunich> ottos: another possibilty is to ssh into the ubunto box from your workstation (is the ws running linux?)
[21:38:22] <ottos> let me try..
[21:40:06] <ottos> firewall on... i'm just going o take notes and run back and forth.. if you don't mind..
[21:40:06] <awallin> good night all
[21:40:14] <jmkasunich> ok
[21:40:22] <jmkasunich> on the ubuntu box:
[21:40:35] <ottos> ok..
[21:40:37] <jmkasunich> open a shell, and do "apt-cache search numarray"
[21:40:47] <jmkasunich> on my box that gets multiple results
[21:40:52] <alex_joni> jmkasunich: check the repositories he has
[21:41:14] <jmkasunich> alex_joni how can I tell which repo my box got the package from?
[21:41:36] <robin_sz> I dont think you can ...
[21:41:55] <robin_sz> once its installed, it could have coem from a repo, or a dpkg -i
[21:42:02] <jmkasunich> then its kinda hard for me to tell him if he has the right repos
[21:42:31] <alex_joni> http://packages.ubuntu.com/breezy/python/
[21:42:36] <robin_sz> look at /etc/apt/sources.list
[21:43:02] <ottos> ok execute without any feedback or errors..
[21:43:08] <jmkasunich> that will tell me which repos I have selected, not which on actuallyhas that package
[21:43:19] <jmkasunich> ottos: what results did it print? nothing?
[21:43:21] <alex_joni> hmm.. ther is no python2.4-numarray it seems
[21:43:30] <alex_joni> only python2.3-numarray
[21:43:45] <jmkasunich> I have 2.4 here (dapper)
[21:43:53] <jmkasunich> maybe not available for breezy?
[21:43:56] <ottos> I got that error msg from a 5.1 box installing the 2.1 EMC2
[21:44:12] <alex_joni> jmkasunich: sorry.. I'm tired
[21:44:15] <alex_joni> there is one..
[21:44:21] <jmkasunich> 5.1 is actually 5.10, which is breezy
[21:44:35] <alex_joni> ottos: open a shell and try apt-cache search python2.4-numarray
[21:44:50] <alex_joni> if that fails, search /etc/apt/sources.list for any lines with universe
[21:44:59] <alex_joni> and make sure they are not commented out
[21:45:09] <alex_joni> then apt-get update, and try the apt-cache search again
[21:48:52] <ottos> what do you mean with universe
[21:52:16] <ottos> ok the update was sucessfull...
[21:55:55] <alex_joni> ottos: there are certain ubuntu repositories
[21:56:04] <alex_joni> standard, universe, multiverse
[21:56:24] <alex_joni> universe and multiverse are comunity maintained (not official packages, but close enough)
[21:59:00] <ottos> ok.. so what are my next steps...
[21:59:23] <jmkasunich> cool: http://www.youtube.com/watch?v=mpBG-nSRcrQ
[21:59:43] <ottos> alex.. trying to get the 2.05 or 2.0.1 to work as beafore I've installed the 6.10 version..
[21:59:58] <alex_joni> ottos: 2.1.0 should work too
[22:00:08] <alex_joni> what have you done so far?
[22:00:37] <ottos> installed the base adn d/l the install .sh in which case doesn't workk...
[22:00:46] <alex_joni> try it again now
[22:00:56] <alex_joni> jmkasunich: something nice: http://www.youtube.com/watch?v=wplyCogv3to
[22:01:29] <ottos> the script?
[22:02:29] <alex_joni> ottos: yes
[22:02:32] <jmkasunich> alex_joni seen that one
[22:02:43] <alex_joni> jmkasunich: reminded me of the guys in here
[22:02:50] <alex_joni> * alex_joni included
[22:03:14] <ottos> still the python prob..
[22:03:35] <jmkasunich> ottos: did you change anything in your sources.list?
[22:04:06] <ottos> no just uncommented the universe you've refered to..
[22:04:31] <alex_joni> ottos: sudo apt-get update
[22:04:38] <jmkasunich> then the answer is "yes, I uncommented the universe" that is a change ;-)
[22:08:24] <ottos> ok done..
[22:08:32] <ottos> installed packages..
[22:09:02] <alex_joni> ottos: is emc2.1.0 installed?
[22:09:23] <alex_joni> anyone knows of a decimal-> hex converter?
[22:09:36] <ottos> no..still getting the pyhon err..
[22:09:54] <ottos> do I uncomment all the lines in sources.list..?
[22:09:54] <alex_joni> ottos: sudo apt-get install python2.4-numarray
[22:12:44] <ottos> package does not exist but is refered to by other package.. ?!?!
[22:12:54] <ottos> won't install..
[22:13:00] <alex_joni> ottos: sudo apt-get install python-numarray
[22:13:52] <alex_joni> ottos: somehow I suspect you didn't uncomment the good line
[22:13:59] <alex_joni> does it start with deb ?
[22:14:02] <alex_joni> or deb-src ?
[22:14:30] <alex_joni> ottos: could you please pastebin your sources.list ?
[22:14:41] <jepler> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Installing_EMC2#Simulator_on_Ubuntu_6_10_Edgy_Eft
[22:14:59] <jepler> for whatever reason, python2.4-numarray is not a package in 6.10.
[22:16:00] <jmkasunich> jepler: ottos is running 5.10 (breezy)
[22:16:02] <alex_joni> hmm.. might be the same for breezy?
[22:16:21] <jmkasunich> * jmkasunich checks on a VM
[22:16:29] <jepler> oh I thought someone said 6.10.
[22:16:40] <jepler> 15:59:45 <ottos> alex.. trying to get the 2.05 or 2.0.1 to work as beafore I've installed the 6.10 version..
[22:16:41] <alex_joni> 6.10 is edgy
[22:16:58] <alex_joni> jepler: hope that was a typo
[22:17:16] <alex_joni> jepler: you installed a fresh edgy and emc2 inside a VM in 14 minutes?
[22:17:24] <jmkasunich> very first thing he said: "sorry to bother I'm getting a dependancy error while installing 2.1 on the 5.10 Ubuntu system any workarounds..?"
[22:17:37] <jmkasunich> one or the other is wrong, which is it?
[22:17:59] <jmkasunich> ottos: are you working with ubunti 5.10 or 6.10?
[22:18:05] <ottos> sude one sec..ubuntu 5.10
[22:18:19] <jmkasunich> ok
[22:18:28] <jepler> alex_joni: no, I did that a few days ago
[22:18:40] <alex_joni> jepler: I'm relieved :D
[22:18:57] <ottos> ok got the list.. where do you want me to post it?
[22:19:00] <alex_joni> jyi, I got vivi to boot on an ARM9 :)
[22:19:06] <jepler> http://pastebin.ca
[22:19:08] <alex_joni> ottos: www.pastebin.ca
[22:20:36] <ottos> ok..
[22:20:50] <jepler> I just verified that the correct package name for ubuntu 5.10 is python2.4-numarray.
[22:21:45] <ottos> post number 339168
[22:21:56] <jepler> python2.4-numarray is in main, so it is *not* an issue of enabling multiverse or universe
[22:22:15] <jmk-vm03> yeah, just figured that out ;-)
[22:22:37] <jepler> OK the problem is that the regular network repositories are not enabled
[22:22:41] <jmk-vm03> jmkasunich@VM03-breezy-rt:~$ apt-cache search python2.4-numarray
[22:22:41] <jmk-vm03> python2.4-numarray - An array processing package modelled after Python-Numeric
[22:22:42] <alex_joni> rest of the file looks good too
[22:22:44] <jepler> this happens when you install from CD without the network available
[22:22:53] <jepler> "## Uncomment the following two lines to fetch updated software from the network"
[22:22:59] <jepler> you need to do this, then run sudo apt-get update
[22:23:02] <jepler> ottos: ^^^
[22:23:23] <ottos> the install was installed without the net.. installed it afterwards..
[22:23:36] <alex_joni> jepler: but there's another 'main' further down..
[22:24:07] <jepler> alex_joni: that's breezy-backports, not breezy
[22:24:09] <jmkasunich> there is>?
[22:24:19] <jmkasunich> I just see breezy-backports main
[22:24:19] <ottos> so do I reinstlal with net..?
[22:24:26] <jmkasunich> no
[22:24:30] <jepler> ottos: there is no need to reinstall
[22:24:32] <jmkasunich> you do what jelper said
[22:24:43] <jmkasunich> edit sources.list (you will need to use sudo)
[22:24:47] <ottos> ok..
[22:24:53] <jmkasunich> uncomment lines 5 and 6
[22:25:07] <jepler> might as well uncomment lines 10 and 11
[22:25:13] <jepler> as well
[22:25:15] <jmkasunich> line 4 says exactly what you need to do:
[22:25:29] <jmkasunich> ## Uncomment the following two lines to fetch updated software from the network
[22:25:44] <jepler> "uncomment" means to remove the leading "#" from the line
[22:26:16] <ottos> ok.. done..
[22:27:12] <jepler> sudo apt-get update, then you should be able to install python2.4-numarray
[22:27:56] <jmk-vm03> he can probably just run the install script again
[22:28:15] <jmk-vm03> it does an update, and them installs emc2 which should bring in numarray
[22:28:19] <jepler> yes that should work too
[22:31:30] <ottos> ok when executed the script still getting the error , when apt-get install python2.4-numarray no packages available..I'm i missing something.. ?!
[22:31:54] <alex_joni> did you save the changes?
[22:32:05] <ottos> yes.. let me redoit..
[22:32:07] <alex_joni> (edit the file as root/sudo ?)
[22:33:44] <ottos> yes..jsut rechecked all seems to be ok.. 8(
[22:35:39] <ottos> getting helpless...
[22:36:11] <alex_joni> did you sudo apt-get update ?
[22:36:23] <alex_joni> pastebin that output please
[22:36:40] <alex_joni> ottos: are you sure your machine is able to reach the network?
[22:36:48] <ottos> yes let me paste it..
[22:37:15] <ottos> yes it does the update and get an ip from the dhcp..
[22:39:18] <jmk-vm03> here is what I see when I install numarray on ubuntu 5.10:
[22:39:19] <jmk-vm03> http://www.pastebin.ca/339187
[22:48:59] <ottos> sorry alex mu usb just died.. I mean I should not be that compl;icated.. ?
[22:49:43] <alex_joni> ottos: you can upload form the linux box just as easy
[22:54:06] <jmkasunich> ottos: none of this is complicated, assuming you are connected to the network when you do the install
[22:54:26] <jmkasunich> if you are not, then you get out of "click and watch" and into "you need to do some hacking"
[22:54:53] <jmkasunich> its still not that hard, IF you are connected to the net
[22:55:03] <jmkasunich> edit sources.list, and run a couple of apt commands
[22:57:09] <ottos> ok alex post 339220
[22:58:12] <jmkasunich> ottos: why don't you just paste the URL instead of the number?
[22:59:46] <ottos> http://www.pastebin.ca/339220
[23:01:04] <jmkasunich> ottos: please try "sudo apt-get install python2.4-numarray" and pastebin the results
[23:05:19] <ottos> <jmkasunich> if you don't mind I'll continue tomorow.. let's just say It's been a long day..
[23:06:06] <ottos> gents thatnk you for all support. and the newbie questions..
[23:07:57] <ottos> cheers good night..!
[23:07:56] <alex_joni> ottos: ok
[23:09:56] <jmkasunich> sigh
[23:14:07] <crepincdotcom> jepler: around?
[23:14:46] <crepincdotcom> anyone here used jepler and cradek's scripts for eagle --> nc-code?
[23:18:11] <cradek> yes
[23:20:05] <crepincdotcom> so i nabbed the files from CVS, i just threw them all in Eagles "ulp" dir
[23:20:25] <crepincdotcom> what do i need to do to make eagle recognize that the script and its configs exist?
[23:21:33] <cradek> just run it
[23:21:38] <cradek> there's a button to run a ulp
[23:21:47] <crepincdotcom> * crepincdotcom looks for that
[23:22:20] <crepincdotcom> aha
[23:22:22] <crepincdotcom> there we go
[23:22:29] <crepincdotcom> ok thanks, sorry for the stupid question