#emc-devel | Logs for 2009-03-12

[00:06:30] <CIA-2> EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/gcode/ (main.lyx tool_compensation.lyx): minor edits
[01:06:04] <CIA-2> EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/drivers/hostmot2.lyx: add info on dmesg printout
[02:24:24] <cradek_> cradek_ is now known as cradek
[12:21:31] <CIA-2> EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/gcode/images/comp-path-1.png: add image
[12:42:51] <CIA-2> EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/gcode/tool_compensation.lyx: fix bug in documents
[12:45:04] <BigJohnT> the image sucks but I'm out of time this morning :( so I'll work on it this evening
[14:10:11] <cradek> wonder if it was bigjohn who filed that one
[14:11:40] <jepler> * jepler wonders if BJT-Work is listening
[14:11:55] <SWPadnos> huh. isn't the actual behavior the expected behavior?
[14:11:58] <cradek> * cradek waves
[14:12:14] <BJT-Work> no, but he is reading
[14:12:18] <SWPadnos> since there's no corner to prevent the tool from following the full path
[14:12:34] <BJT-Work> no cradek it wasn't me
[14:12:36] <cradek> it was anon so I couldn't ask for clarifixation
[14:12:48] <cradek> ok, just curious
[14:12:56] <BJT-Work> I think he just didn't get what he expected
[14:13:03] <cradek> yeah
[14:13:14] <BJT-Work> I added a bit to the manual this morning to try and explain that
[14:13:24] <BJT-Work> did you see it?
[14:13:28] <cradek> thanks!
[14:13:32] <cradek> not yet
[14:13:46] <cradek> brb
[14:14:19] <BJT-Work> http://www.linuxcnc.org/docview/devel/html//gcode_tool_compensation.html#sec:Cutter-Radius-Compensation
[14:18:12] <cradek> I like the picture
[14:18:54] <cradek> that endpoint can go almost anywhere depending on the next move
[14:19:44] <SWPadnos> the image shows the last compensated segment following the complete programmed path, but I don't know how to point that out to someone who isn't looking for it
[14:20:37] <BJT-Work> I'm going to work that image over to make it more to the point and not show the lead in move
[14:21:12] <BJT-Work> what do you mean SWPadnos
[14:21:31] <SWPadnos> the bug report showed that the user didn't know what they would be getting
[14:21:49] <SWPadnos> the example code does one move, which is like the second segment in your diagram
[14:22:13] <SWPadnos> the tool moves the same length as the programmed path, because there's no "next segment" to avoid gouging
[14:22:46] <SWPadnos> (or at least it moves to a point which is offset perpendicularly from the endpoint of the programmed path)
[14:23:03] <cradek> in that case the tool moves to the endpoint where it would have been if it had been fully compensated all along (hard to explain)
[14:23:05] <SWPadnos> but any way I try to explain that would likely be confusing to someone who doesn't already know what I'm talking about :)
[14:23:11] <SWPadnos> yep
[14:23:13] <cradek> yes perpendicular from the endpoint along the programmed path
[14:23:36] <SWPadnos> which says nothing to most people :)
[14:23:43] <cradek> but really, the real problem is a program with only an entry move makes no sense - the endpoint is pretty much undefined
[14:23:57] <cradek> jepler pointed out that perhaps it should be an error case
[14:23:58] <BJT-Work> that's why I did the graphic :)
[14:24:29] <BJT-Work> cradek: that makes sense that a single move would be an error
[14:25:06] <SWPadnos> does M2 cancel cutter comp?
[14:25:09] <cradek> by the filer expecting the endpoint to be X0.25 I can tell he meant to turn left 90 degrees after entry. any other turn would put X at a different place. The problem is he didn't tell emc he wanted to turn left 90 degrees by programming that path!
[14:25:12] <SWPadnos> or %
[14:26:15] <cradek> SWPadnos: http://www.linuxcnc.org/docs/devel/html/gcode_main.html#sub:M0,-M1,-M2,
[14:26:26] <SWPadnos> :P
[14:26:46] <cradek> BJT-Work: would you fix that m2 section please - g92 is no longer unapplied at m2
[14:26:52] <jepler> oh wow -- I hadn't looked at bjt's cutter comp docs
[14:26:59] <jepler> I love the images
[14:27:06] <jepler> haven't been able to read the words yet, I'm fixated on them
[14:27:16] <BJT-Work> sure thing cradek
[14:31:25] <cradek> you can tell which ones he's rewritten and which ones he hasn't (1.1.3)
[14:31:53] <cradek> "1. If G43 Hn is programmed, A USE_TOOL_LENGTH_OFFSET(length) function call is made (where length is the value of the tool length offset entry in the tool table whose index is n), tool_length_offset is reset in the machine settings model, and the value of current_z in the model is adjusted."
[14:32:13] <cradek> (wtf?)
[14:32:28] <jepler> yeah blame me for the cut&paste job there, probably
[14:32:36] <jepler> "I don't know what all these words are, but I hear we should have documentation"
[14:32:42] <cradek> haha, it's terrible
[14:32:52] <jepler> anyway, just put some pictures on the same page and nobody'll notice the words
[14:33:08] <cradek> uh did you see the picture next to it?
[14:33:28] <cradek> I don't understand what it's trying to say either
[14:33:55] <BJT-Work> I was saving that part for last :)
[14:34:26] <cradek> :-)
[14:39:02] <cradek> is the gcode in right? I don't think those arcs under (Cut Profile) are valid and they can't make a square
[14:39:56] <BJT-Work> holy crap where is my G1!
[14:40:21] <cradek> also in it says G42 (Start Comp Left) but it's actually right
[14:40:30] <BJT-Work> or is it a G3S for square
[14:41:14] <BJT-Work> thanks for proof reading that
[14:43:25] <SWPadnos> G<dwim>
[15:04:01] <BJT-Work> WEE! I just printed a 24" x 48" drawing on my plotter
[15:06:02] <cradek> ??
[15:07:21] <BJT-Work> I had to define a custom sheet size to print it 1 to 1
[15:09:34] <cradek> ah, definitely an odd size
[15:10:13] <BJT-Work> yes, seems like I can print 24" wide (the width of the roll) and whatever long!
[15:10:57] <BJT-Work> we tape the print to the part and transfer all the holes :)
[15:11:06] <BJT-Work> makes it real easy
[20:20:26] <alex_joni> SWPadnos: how about one of these: http://www.ramsan.com/images/products/20.jpg
[20:21:05] <SWPadnos> flash/RAM disk?
[20:22:11] <alex_joni> SSD's with RAID5
[20:22:33] <alex_joni> http://www.ramsan.com/products/ramsan-20.htm
[20:22:42] <alex_joni> 450GB of them :D
[20:23:05] <alex_joni> only 15W, but 18k$ :/
[20:23:16] <cradek> wow, neato
[20:24:17] <alex_joni> wonder how many stepgens we could fit in those FPGA's :D
[20:24:54] <cradek> much cheaper: http://www.popamericana.com/!/16k.jpg
[20:25:07] <SWPadnos> more than needed for the number of GPIOs
[20:27:35] <cradek> Even if your present computer is not 4 MHz fast, this new Model 16KZ RAM equips you for the time when you'll need and want higher computer speed.
[20:27:39] <alex_joni> cradek: wow, cool.. where can I get one?
[20:27:40] <jepler> 15W? A 500GB 2.5" hard drive has read/write power dissipation of just 2.50W. http://www.wdc.com/en/products/products.asp?driveid=506&language=en
[20:28:02] <alex_joni> jepler: yeah, but you need at least 4 for RAID5
[20:28:12] <alex_joni> and those probably can't do 80k IOPS
[20:28:18] <jepler> no, probably not
[20:28:22] <cradek> alex_joni: I've got a machine with (I think) 4 of them...
[20:29:33] <jepler> alex_joni: I don't see the point of RAID5 when the chips are not socketed
[20:29:45] <jepler> what do you do when one goes bad, copy the array to a fresh card, then throw the old one out?
[20:29:56] <alex_joni> jepler: you can probably turn them off once they stop working
[20:30:22] <alex_joni> well.. at 18k$ I'd suspect they would replace defective ones
[20:34:51] <SWPadnos> pci hotplug should help with that
[20:35:30] <jepler> With 18k$ worth of 1TB hard drives you get a lot of bandwidth and a whole lell of a lot of storage. It'll use more than 15W, though.
[20:35:38] <jepler> s/lell/you know
[20:35:44] <SWPadnos> heh
[20:36:11] <SWPadnos> you could allocate $15k to power usage over the lifetime of the array, and still have a lot of 1TB drives
[20:38:21] <jepler> oh, no wonder the commodore 64's serial sucked so badly: it's fully bit-banged
[20:39:05] <cradek> wow, and it did 2400 baud iirc
[20:41:29] <jepler> $ units 1MHz/2400Hz; units 3GHz/40kHz
[20:41:29] <jepler> Definition: 416.66667
[20:41:29] <jepler> Definition: 75000
[20:41:58] <jepler> so in those days you could do interrupt-driven bit-bang in 400-odd CPU cycles; now it takes 75000 of them
[20:42:06] <cradek> funny
[20:42:25] <cradek> possibly a little more is going on meanwhile.
[20:42:44] <alex_joni> yeah, eyecandy
[20:42:54] <cradek> jellovision
[20:43:02] <alex_joni> wobbly windows
[20:44:03] <jepler> (for no easy-to-explain reason, I was reading a commodore 64 reference.. http://pastebin.ca/1359297 )
[20:45:35] <alex_joni> funny
[20:47:06] <cradek> ... #
[20:47:06] <cradek> ... for the highest baud rate, 2400
[20:47:06] <cradek> #
[20:47:16] <cradek> baud.
[20:47:21] <cradek> ack
[20:47:27] <cradek> wonder why pastebin causes that bad pasting
[20:47:39] <cradek> I always forget to click raw
[20:52:03] <jepler> http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=180225281541
[20:52:18] <jepler> Are you tired of being limited to 1200 baud on most Commodore 64 and 128 solutions? I have personally used Novaterm at 57600 baud without a hitch on a stock C64!
[20:53:07] <cradek> bizarre
[20:54:32] <jepler> "In April 2008, the first batch of 1541 Ultimates shipped, a project by the hobbyist VHDL-developer Gideon Zweijtzer. This is a cartridge that carries an Action Replay and Final Cartridge (whatever the user prefers) and a very compatible FPGA-emulated 1541 drive that is fed from a built-in SD-card slot (.d64, prg etc). The difference to other SD-based and .d64 mounting cartridges like the MMC64, Super Snapshot 2007 or MMC Replay is, that the 6502 t
[20:56:20] <alex_joni> haha, this is even nicer: http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=120378686150
[20:57:42] <cradek> I had one of these! It was the greatest thing ever! http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=180270550072
[20:58:16] <cradek> I had a daisy wheel printer that made my homework look like it was carefully and perfectly typed, not printed on some ugly 9-pin printer
[20:59:15] <alex_joni> heh
[20:59:29] <alex_joni> I always wondered abotu getting a line printer
[20:59:39] <alex_joni> but that would probably be too loud to have around
[21:00:31] <cradek> my parents would yell at me if I didn't close the door before printing something
[21:00:49] <cradek> it had sheet feed (!!) so sometimes I would start it and then leave the room :-)
[21:01:41] <SWPadnos> Diablo 630, baby!
[21:01:48] <SWPadnos> (never had one, but damn they were loud)
[21:04:23] <skunkworks_> hey - I had a daisy wheel printer for the commodore :)
[21:05:09] <alex_joni> still, I doubt it's as loud as a line printer
[21:05:19] <SWPadnos> CHUNK CHUNK CHUNK CHNK
[21:05:28] <SWPadnos> some of those were friggin fast
[21:05:31] <alex_joni> http://www.youtube.com/watch?v=U16GNsRYGbA
[21:05:38] <SWPadnos> faster than most modern laser printers actually
[21:05:41] <alex_joni> too bad there's no actual sound from the printing
[21:05:42] <skunkworks_> we had gotten a bunch of line printers from the local tech college - we ran a few test pages thru them - yes - loud
[21:05:56] <skunkworks_> (put them into test mode)
[21:05:59] <alex_joni> at least we have lpd now :D
[21:06:42] <SWPadnos> lpr is nice too
[21:06:52] <SWPadnos> with emphasis on the "r" part :)
[21:07:16] <SWPadnos> wow. airfares are really cheap right now
[21:07:46] <SWPadnos> I got an email about flights to Australia - $1000 per person (yes, that's round trip)
[21:08:41] <alex_joni> it you don't get "lost" on the way :P
[21:08:46] <SWPadnos> heh
[21:09:05] <SWPadnos> I'll make sure there's a soccer ball on board, just in case ;)
[21:12:49] <jepler> Ingrid was shopping for flights to Europe. Tickets? $200 round-trip ... plus $300 in taxes
[21:13:09] <SWPadnos> heh, yeah. they do get you that way
[21:13:25] <SWPadnos> though the $1003 price is inclusive of all taxes/facility fees
[21:13:45] <SWPadnos> (I usually search on the United site, since they always include all the fees in the proces they show)
[21:13:51] <SWPadnos> prices
[21:55:26] <alex_joni> good night all
[22:55:20] <jmkasunich> had these in college: http://www.youtube.com/watch?v=HMbyhYSbMG0
[22:58:21] <jepler> hi jmkasunich
[22:59:23] <jepler> it doesn't seem particularly fast. and I'm sure youtube isn't doing justice to the racket it made.