#emc-devel | Logs for 2007-05-20

[00:19:59] <jmkasunich> cradek: still around?
[00:20:05] <cradek> yes
[00:20:26] <jmkasunich> are you looking at the 3000+ lb machines because you need/want a bigger machine?
[00:20:38] <jmkasunich> of because thats what you've seen available?
[00:21:23] <cradek> because I want to reasonably be able to cut metal, and aside from my toy mill, that's what I've used
[00:22:14] <cradek> I noticed smithy was going to demo a machine at fest - I'm anxious to see that
[00:22:19] <jmkasunich> have you considered adding a ballscrew conversion to either a used bport manual, or a new or used off-brand manual?
[00:22:23] <cradek> I assume it's their bench mill
[00:22:32] <jmkasunich> the $11,000 one?
[00:22:54] <cradek> I don't know their prices, but if you say so...
[00:23:25] <cradek> the manuals (without ballscrews) seem to cost about the same as the old cnc series 1 etc, so I figured starting with the screws on it would be much smarter
[00:23:51] <jmkasunich> this smithy: https://www.smithy.com/products.php?cid=11&scid=16&pid=7
[00:24:02] <jmkasunich> my bad, its only $10300
[00:24:19] <cradek> yep I bet that's what they will show
[00:25:24] <jmkasunich> you can get a new enco manual for $3200
[00:25:41] <jmkasunich> granted thats a lot more than the trashed out CNCs at HGR
[00:25:55] <jmkasunich> but its also not trashed out
[00:28:43] <cradek> unless I could get a bolt-on screw kit (if there is such a thing) I'm not confident I'd have the required equipment (or skill) to mount replacement screws
[00:28:59] <jmkasunich> http://www.microkinetics.com/convkit.htm
[00:29:23] <jmkasunich> there is also the "equipment/skill to repair trashedness" factor
[00:29:32] <jmkasunich> bottom of the page on that last link
[00:31:01] <cradek> do you think the chances of getting something untrashed enough to make parts for $2k at HGR is small?
[00:31:36] <cradek> I don't have the experience to know that really
[00:31:47] <jmkasunich> I think in your situation the chances are small
[00:32:09] <jmkasunich> if you could visit HGR every month or so, and be ready to buy when the right machine comes in, you could probably do good
[00:32:11] <cradek> the machine jon elson stuck new motors on - it looked like it had an average life, but I could see it making parts with not much more work
[00:32:22] <jmkasunich> might be a year or more till the right one comes in
[00:32:30] <cradek> yeah that's sure true
[00:33:54] <jmkasunich> I'd be willing to visit HGR to check out machines for you, but I'm not willing (or able) to make a proper evaluation of condition
[00:34:13] <jmkasunich> the person who is gonna do the work has to be the one who looks at the machine
[00:34:38] <jmkasunich> (I could serve as a first level screen, to tell you "that ones really trashed, you don't want it" )
[00:35:07] <cradek> I would not think of asking you to pick a machine and send it to me - that would be crazy
[00:35:12] <jmkasunich> right
[00:35:39] <jmkasunich> I could pick a machine and say "you should look at this one"
[00:35:58] <jmkasunich> its not exactly a hardship for me to visit HGR once a month ;-)
[00:36:02] <cradek> haha
[00:36:13] <jmkasunich> If I know you are in the market, I'll look more closely at mills
[00:37:33] <cradek> thank you
[00:38:25] <cradek> I'd happily pay ray or someone like him to help me pick a good candidate - but geography is a huge hindrance
[00:39:37] <jmkasunich> my experience with my 13" south bend lathe is a case in point
[00:40:01] <jmkasunich> they got three of them in at the same time, probably from the same plant
[00:40:03] <jmkasunich> same price
[00:40:18] <jmkasunich> one was trashed, missing lots of important bits
[00:40:21] <jmkasunich> one was so-so
[00:40:27] <jmkasunich> and one was pretty darned good
[00:40:56] <jmkasunich> although pretty darned good still meant it had several problems that made it unusable without repairs
[00:41:26] <jmkasunich> from the pics on the website, you couldn't tell which was which
[00:42:43] <cradek> yeah their photos and descriptions are about worth what you pay for them.
[00:43:19] <jmkasunich> they make their money by selling tons of stuff, not by getting top dollar per item
[00:43:38] <jmkasunich> their handling of the merchandise, and everything else, all reflects that
[00:43:59] <cradek> yeah
[00:44:00] <jmkasunich> I've seen many items that lost value after arriving at HGR because _they_ damaged the machine while moving it in their warehouse
[00:45:01] <jmkasunich> here is one of the things I had to fix on the lathe
[00:45:02] <jmkasunich> http://www.metalworking.com/dropbox/_2001_retired_files/tscrew1.jpg
[00:45:30] <jmkasunich> (not the fact that its cut in half, the fact that the threads were chewed by a rabid metal-rat
[00:46:38] <cradek> the right one is the tool post? what's the left?
[00:46:44] <jmkasunich> I can virtually guarantee you that an HGR grade machine (from anywhere) is gonna need repairs like that
[00:47:08] <jmkasunich> that is the tailstock screw
[00:47:24] <cradek> ohh
[00:47:29] <jmkasunich> story here: http://www.metalworking.com/dropbox/_2001_retired_files/tscrew.txt
[00:50:07] <cradek> should have switched tailstocks when they weren't looking?
[00:50:25] <jmkasunich> I don't recall the condition of the other tailstocks
[00:50:32] <jmkasunich> besides thats unethical
[00:50:55] <cradek> I suppose
[00:51:14] <jmkasunich> whoever wound up with the other two had enough problems as it is
[00:51:32] <jmkasunich> nice drill bit: http://www.metalworking.com/dropbox/_2001_retired_files/untwisted.jpg
[00:52:10] <SWPadnos> how in the hell does one do that?
[00:54:12] <cradek> I bet heat is involved
[00:54:24] <SWPadnos> I'm sure
[00:54:24] <jmkasunich> easy - buy a drill bit from china and then try to drill something that catches it
[00:54:41] <jmkasunich> I've seen china drills that were soft at room temperature
[00:54:43] <SWPadnos> but strangely, the edge still looks relatively sharp, and it's not broken
[00:54:54] <SWPadnos> I suppose if you use ductile iron or something
[00:54:59] <SWPadnos> (for the bit)
[00:55:01] <jmkasunich> its not broken because its shit steel or shit heat-treat or both
[00:57:17] <SWPadnos> http://www.hgrindustrialsurplus.com/search-products/product-detail.aspx?id=20-648-001
[00:58:16] <jmkasunich> I wonder what the spindle taper is?
[00:58:28] <jmkasunich> it doesn't look like the spindle nose on a manual bport
[00:58:33] <jmkasunich> 30taper maybe?
[00:58:38] <SWPadnos> dunno - looks like CAT 30-40 (if those exist)
[00:58:50] <SWPadnos> yeah - it's probabl not an R8 quill
[00:59:14] <SWPadnos> that actually looks pretty good from here
[01:02:33] <jmkasunich> nothing obviously broken
[01:02:43] <cradek> I think they have about three like that one
[01:02:51] <jmkasunich> I saw a Tree CNC mill that had the entire X motor drive busted off
[01:03:05] <jmkasunich> I mean, 1/4" thick cast iron bracket with ribs, broken
[01:03:38] <jmkasunich> this one: http://www.hgrindustrialsurplus.com/search-products/product-detail.aspx?id=30-319-007&searchtable=1&sortExpression=&SortASC=&pageSize=50&currentPageIndex=0
[01:03:49] <jmkasunich> note the visible pulley at the right end of the table
[01:03:55] <jmkasunich> it ain't supposed to be visible
[01:04:19] <cradek> ouch
[01:04:57] <jmkasunich> the series 1 that swp posted as the advantage of weighing a ton
[01:05:02] <jmkasunich> (instead of two tons)
[01:05:33] <jmkasunich> hmm, I bet its hard to check out a CNC machine
[01:06:08] <jmkasunich> one of the first things I do when looking at a machine is crank the screws, to feel for binding/grinding, to see how jammed up stuff is, and to check backlash
[01:06:18] <jmkasunich> hard to do when there is no access to the screws
[01:06:38] <SWPadnos> I'd just assume that the ballscrews will need replacing, and then be happily surprised if they don't
[01:07:08] <SWPadnos> that mill looks to be in reasonable shape, but the important things are there: the motor mounts
[01:07:10] <cradek> I think you guys are telling me this is an impossible task
[01:07:13] <jmkasunich> replacing the screws will cost half the price of the machine
[01:07:26] <SWPadnos> no - it's actually pretty easy to replace the ballscrews on a brigeport
[01:07:27] <SWPadnos> +d
[01:07:37] <SWPadnos> I've actually done that to mine
[01:07:46] <jmkasunich> how much did the screws cost?
[01:07:53] <jmkasunich> or can you get rolled ones that fit?
[01:07:55] <SWPadnos> I got good screws (precision ground, new old stock chinese) for $650
[01:08:05] <jmkasunich> ok, 1/3 the cost of the machine
[01:08:09] <SWPadnos> new, with yoke, X+Y
[01:08:33] <SWPadnos> sure, but motors cost me $350 (which is a very good deal for these)
[01:08:37] <jmkasunich> off the shelf somewhere, or as a result of much searching and some good luck?
[01:08:45] <SWPadnos> the mounts are a PITA to make, though I bought a Z mount for $300 or so
[01:08:53] <SWPadnos> eBay, Machine Tools Direct
[01:09:05] <SWPadnos> I'm not sure if they stil lhave any, that was ~3 years ago
[01:09:27] <SWPadnos> those were for a manual mill though, not CNC
[01:09:39] <jmkasunich> cradek: not an impossible task
[01:09:42] <jmkasunich> but not easy either
[01:10:25] <SWPadnos> the nice thing is, you already have a small CNC, so if you need to make a part, you can
[01:10:37] <jmkasunich> assuming the part in question fits on the small mill
[01:10:39] <SWPadnos> it's a little harder with only a manual machine that you're actively trying to modify ;)
[01:11:41] <jmkasunich> comes back to the time/money tradeoff
[01:11:54] <SWPadnos> yes
[01:12:13] <jmkasunich> the more time you can spend (not just working on the mill once you have it, but time spent searching/shopping), the more money you save
[01:12:22] <SWPadnos> though purchasing things for the larger machines seems to be waaaay more expensive, and there are very few plans around
[01:13:01] <jmkasunich> cradek: how tall is the ceiling in the space where you would put the mill
[01:13:18] <jmkasunich> * jmkasunich notices that the CNC bports are pretty tall
[01:13:28] <SWPadnos> they're around 7 feet
[01:13:40] <SWPadnos> plus some if you ever want to replace the drawbar ;)
[01:13:46] <cradek> I think I have 7' of door and a lot more height inside
[01:13:58] <cradek> height is not a problem except for the door
[01:14:00] <jmkasunich> the mills in the pics seem taller than that
[01:14:14] <SWPadnos> they could be, to the top of the motor wire on top
[01:14:17] <jmkasunich> (that = the 7' that swp mentioned)
[01:14:20] <cradek> ebay often has heights listed, and they tend to be about 7'
[01:14:30] <jmkasunich> ok
[01:14:49] <cradek> I should measure the door to make sure
[01:14:58] <jmkasunich> for manual bports, people often turn the head upside down when moving
[01:15:04] <jmkasunich> lowers the height and the CG
[01:15:11] <cradek> wouldn't surprise me if it's more like 8'
[01:15:11] <jmkasunich> don't think you can do that on the CNCs
[01:16:14] <SWPadnos> no, you can't
[01:16:42] <SWPadnos> you can remove the spindle motor/housing though
[01:16:54] <SWPadnos> but it's more of a pain that pivoting the head
[01:44:40] <cradek> does anyone remember if we changed/fixed the terrible tool table units thing?
[01:45:19] <cradek> g20 g43 h3 and g21 g43 h3 seem to give me the same position
[01:45:26] <cradek> but ... I don't remember changing it to do that
[01:45:41] <SWPadnos> I think it's in "the currently used units"
[01:45:55] <SWPadnos> or did you change it? hmmm
[01:45:59] <cradek> I know that's what the spec says, but did we fix/change it?
[01:46:23] <cradek> * cradek predicts that someone will tell me to UTSL
[01:46:29] <SWPadnos> heh
[01:46:44] <cradek> ... as if I will be able to tell
[01:50:38] <SWPadnos> I see a bunch of changes around the end of June last year, but all the comments are related to lathe tools and passing the extra data around
[01:53:24] <SWPadnos> hmm. apparently Mozilla sometimes shuts down entirely when you click the little X in one window ...
[01:53:48] <Unition> ahhh shit
[01:54:04] <SWPadnos> this is the EMC dev channel
[01:58:58] <Unition> bbl
[02:12:33] <cradek> that was interesting
[02:12:53] <SWPadnos> which one?
[02:13:04] <cradek> Unition
[02:13:37] <SWPadnos> heh - yeah
[02:13:50] <SWPadnos> I'm trying one more timewith rafa at the moment ...
[02:14:09] <cradek> jepler gave him detailed instructions last time he asked this - a few days ago
[02:14:33] <SWPadnos> yeah, though there seems to be a problem - he said "phase drive", and the line he pasted was for step/dir
[02:14:44] <cradek> I think it does use the current units as the diameter
[02:14:46] <cradek> that's so busted
[02:14:51] <SWPadnos> heh
[02:14:56] <cradek> well I doubt jepler made that mistake
[02:15:13] <cradek> you can't switch units with comp on
[02:15:29] <cradek> so it's impossible to cut (around) an inch part with a metric tool
[02:16:25] <SWPadnos> I also doubt jepler made that mistake ;)
[02:16:46] <cradek> you're both very brave
[02:16:54] <SWPadnos> heh
[02:29:41] <cradek> *very* brave
[02:29:46] <SWPadnos> hah
[02:32:59] <SWPadnos> hmmm - where are those connections? are they in core_stepper or standard_pinout?
[02:35:31] <jmkasunich> that connections to the parport are in stdpinout
[02:35:45] <jmkasunich> I'm not sure if the sig is connected to the stepgen in pinput or core
[02:36:03] <SWPadnos> ok - thanks
[02:39:30] <cradek> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?action=browse&diff=1&id=EMC_Fest_2007
[02:40:03] <jmkasunich> ;-)
[02:40:32] <SWPadnos> heh
[02:40:52] <cradek> to get it working, I made two entries for each tool, using different units
[02:41:03] <cradek> I have to remember to use different tool numbers based on whether I'm in g20/g21 mode
[02:41:11] <SWPadnos> I say put the units in the tool file, and then make it work as expected
[02:41:43] <SWPadnos> get rid of FMS (since nobody seems to know what it is), and put units in there instead ;)
[02:41:56] <cradek> I say always use the machine's units
[02:42:13] <SWPadnos> no - what if I have a 3mm endmill and a 3/8 endmill?
[02:42:32] <cradek> you wouldn't - but if you did, surely you can do the math
[02:42:41] <SWPadnos> it makes sense to default to machine units, but accept unit suffixes per tool
[02:42:49] <SWPadnos> why wouldn't I have metric and imperial tools?
[02:43:13] <cradek> sorry I was being glib
[02:43:18] <SWPadnos> ok - good
[02:43:32] <SWPadnos> sorry - just had a somewhat taxing IRC experience :)
[02:44:31] <cradek> you have my sympathy
[02:44:55] <SWPadnos> I wonder if it worked
[02:45:11] <cradek> no, he'll ask the same questions in a few more days
[02:45:23] <SWPadnos> ok. I'll ignore him then
[02:50:28] <cradek> I think the current implementation, which tries to do what the spec says, is broken
[02:50:40] <SWPadnos> I can agree with that
[02:50:46] <cradek> no no no
[02:50:49] <SWPadnos> even if it succeeds in doing what the spec says
[02:50:56] <cradek> I mean it doesn't even successfully do what the spec says
[02:50:59] <SWPadnos> heh
[03:01:12] <cradek> cool, lathe tool offsets work (as long as I don't switch units)
[03:01:24] <cradek> I've got four tools set up on my qctp
[03:01:31] <jmkasunich> slick
[03:01:32] <SWPadnos> cool
[03:01:42] <jmkasunich> you STILL owe us lathe pictures!
[03:02:39] <cradek> darn, I forgot to do the parting tool
[03:13:13] <jmkasunich> HE GOT IT!!!!!!
[03:13:17] <SWPadnos> YEAH!!!
[03:13:59] <jepler> HOORAY
[03:17:20] <jepler> 'night
[03:19:17] <jmkasunich> goodnight jeff
[03:22:52] <SWPadnos> hey - that's a good idea
[03:22:59] <SWPadnos> going to bed before 3 AM
[03:23:13] <jmkasunich> yeah
[03:23:19] <jmkasunich> I need to try that some day
[03:23:31] <SWPadnos> I'm thinking about trying that today :)
[03:24:51] <jmkasunich> only 36 minutes of today left
[03:25:03] <SWPadnos> I guess I'll get to bed then :)
[03:25:05] <SWPadnos> good night
[03:25:19] <jmkasunich> goodnight
[04:42:40] <jmkasunich> my turn, goodnight all
[11:59:38] <alex_joni> cradek: ping
[14:36:03] <cradek> alex_joni: hi
[14:36:10] <cradek> (I'm only here for a bit)
[14:36:12] <cradek> what's up?
[14:45:24] <cradek> darn, bbl
[17:21:46] <alex_joni> cradek: nothing nice ;) wanted to ask your oppinion on something
[17:22:53] <cradek> hi
[17:22:59] <alex_joni> hi there
[17:23:07] <alex_joni> I started looking at that task/interp/canon issue :)
[17:23:15] <alex_joni> about start-from-line
[17:23:16] <cradek> which?
[17:23:20] <cradek> oh right
[17:23:25] <alex_joni> bad vel/accels
[17:23:29] <cradek> ok
[17:23:41] <cradek> is it still broken? I fixed some stuff related to that (for probing)
[17:23:44] <alex_joni> the problem is that task runs the interp for all the previous lines (before start line)
[17:23:53] <cradek> ok
[17:24:01] <cradek> it definitely has to do that
[17:24:07] <alex_joni> that way canon gets updated with all the positions before
[17:24:19] <alex_joni> right, I agree
[17:24:33] <alex_joni> but when it starts running for real, canon has the wron endpoint
[17:25:15] <cradek> ok I understand
[17:25:30] <alex_joni> I'm not sure I know what the proper fix should be
[17:25:45] <alex_joni> actually I do .. but I don't want to go there ;)
[17:25:49] <cradek> how does task run canon, but have canon not issue the moves?
[17:26:00] <alex_joni> task runs interp
[17:26:05] <alex_joni> which calls the canon calls
[17:26:13] <alex_joni> then task calls a interp.clear()
[17:26:20] <alex_joni> which I bet clears the canon queue
[17:26:28] <cradek> ok
[17:26:32] <alex_joni> which otherwise would end up in taks again, and then to motion
[17:27:02] <cradek> could it be fixed with an ignored/cleared STRAIGHT_TRAVERSE(position where the tool really is) right before the run-from line?
[17:27:31] <alex_joni> * alex_joni thinks
[17:27:53] <alex_joni> I'm still trying to digest the code in emctaskmain.cc
[17:28:02] <alex_joni> (to see where that would fit in..)
[17:28:33] <alex_joni> grep for programStartLine if you like..
[17:34:52] <alex_joni> cradek: back in 15 minutes
[17:35:04] <cradek> ok
[17:37:39] <cradek> jepler: is there a simple way to make pluto_servo not load, if the device isn't there?
[17:56:23] <alex_joni> back
[17:59:43] <alex_joni> cradek: I'm not sure I know how to add that extra STRAIGHT_TRAVERSE
[17:59:56] <alex_joni> it should be in task.. but task only calls interp, not canon
[18:00:19] <cradek> I think you could just call it directly from task
[18:00:32] <cradek> task does make other canon calls, I think, like GET_POSITION
[18:01:25] <cradek> that was just an idea - don't forget, I don't have a clue whether that's the right answer
[18:01:27] <alex_joni> it does?
[18:01:47] <cradek> maybe I'm wrong :-)
[18:02:46] <alex_joni> I about as clueless as you :P
[18:06:30] <jepler> cradek: good question
[18:06:46] <jepler> cradek: when it doesn't load, does the 'communication-error' start incrementing?
[18:06:50] <jepler> (I think that's the pin name)
[18:06:54] <jepler> er, when it's not present, that is
[18:06:54] <cradek> yes
[18:07:49] <cradek> if dia > 1.0: # XXX current_tool can have any units
[18:07:49] <cradek> r = dia / (2. * 25.4)
[18:07:49] <cradek> else:
[18:07:49] <cradek> r = dia / 2.
[18:08:06] <cradek> * cradek cries
[18:08:42] <alex_joni> hah
[18:08:52] <cradek> we've got to fix the tool table - it's crazy
[18:09:05] <alex_joni> * alex_joni cries about task
[18:09:16] <cradek> and by fix I mean "change in an incompatible way, contrary to the spec"
[18:09:39] <cradek> anyone want to find a new project to work on?
[18:09:47] <alex_joni> what kind?
[18:10:36] <alex_joni> got anything with pink interfaces and flowers on the buttons?
[18:11:12] <cradek> I bet we could find something
[18:17:24] <jepler> cradek: try this: http://emergent.unpy.net/index.cgi-files/sandbox/pluto-hwdetect.patch
[18:17:30] <jepler> (the top chunk is probably unneeded)
[18:18:34] <cradek> trying
[18:20:17] <cradek> I get a kernel oops!
[18:20:32] <cradek> if the pluto is off
[18:20:39] <alex_joni> oops
[18:20:39] <alex_joni> :P
[18:20:42] <cradek> [5209807.900000] EIP: [<c01edb50>] vsnprintf+0x30/0x630 SS:ESP 0068:ce881a18
[18:21:30] <cradek> bufferlen in rtapi_print is plenty big enough...
[18:22:32] <jepler> oops!
[18:22:39] <jepler> wonder what I gone done wrong
[18:22:53] <cradek> I'm not spotting it...
[18:22:59] <jepler> looks like it needs to be rtapi_print_msg
[18:23:06] <jepler> void rtapi_print(const char *fmt, ...)
[18:23:06] <jepler> void rtapi_print_msg(int level, const char *fmt, ...)
[18:23:26] <cradek> ohh, duh
[18:23:32] <cradek> I looked right over the "level"
[18:24:04] <jepler> if it works please check it in
[18:24:54] <cradek> um, rebooting...
[18:28:48] <cradek> yes, it works
[18:29:21] <jepler> good
[18:34:19] <alex_joni> cradek: the 2.6.17-smp works really nice here
[18:35:02] <alex_joni> 11:12 < alex_joni> time make
[18:35:02] <alex_joni> 11:13 < alex_joni> real 1m42.367s
[18:35:02] <alex_joni> 11:13 < alex_joni> user 0m27.695s
[18:35:02] <alex_joni> 11:13 < alex_joni> sys 1m10.719s
[18:35:15] <alex_joni> that's on a VM running 2.6.17-smp
[18:41:56] <alex_joni> jepler: do you know without looking where current position is in stat?
[18:43:27] <jepler> alex_joni: no
[18:43:34] <alex_joni> found it..
[18:43:49] <alex_joni> emcStatus->motion.traj.actualPosition.tran.*
[19:09:24] <alex_joni> * alex_joni officially hates units
[19:27:36] <alex_joni> cradek: want to test a patch?
[19:32:28] <alex_joni> www.pastebin.ca/498926
[21:08:51] <alex_joni> cradek: I think it works as it should now.. the only thing I'm concearned is if I should update the current position after each interpreted line, or only once before the skipping ends
[21:18:26] <alex_joni> jepler: ping
[21:31:22] <jepler> alex_joni: seems like it works fine in all other cases
[21:32:10] <alex_joni> jepler: wanted to ask you somehting else though :)
[21:32:18] <alex_joni> did you get the email about the spanish translation?
[21:33:07] <jepler> no
[21:33:11] <jepler> I don't think we have one
[21:33:31] <alex_joni> ok, I'll tell him to start
[22:04:16] <alex_joni> jepler, cradek: that patch from above I did on the 2.1 branch
[22:04:27] <alex_joni> I merged it into trunk and it behaves completely differently
[22:04:49] <alex_joni> but it's too late now to start debugging it.. I'll look at it tomorrow or so