#emc-devel | Logs for 2008-07-09

[00:13:59] <SWPadnos_> SWPadnos_ is now known as SWPadnos
[02:43:49] <jmkasunich> ok, this is weird
[02:44:01] <jmkasunich> I've managed to create two hal signals with the same name
[02:44:14] <SWPadnos> cool!
[02:44:21] <SWPadnos> or is your font terrible?
[02:44:31] <jmkasunich> oops, no I haven't
[02:45:02] <jmkasunich> one of those "look at it 5 times and still not realize that 'spd' and 'speed' aren;t the same name
[02:46:01] <jmkasunich> 'net spindle-spd-1ppr encoder-foo mux2.foo' and much later in the hal file 'net spindle-speed-1ppr sampler.0.pin.0'
[02:46:18] <jmkasunich> should have used linksp for the later one - it will error out if you spell the signal name wrong
[02:46:36] <SWPadnos> heh
[02:46:57] <jmkasunich> getting ready to make a recording of 1ppr threading (if I stop making stupid mistakes)
[02:47:02] <SWPadnos> heh
[02:49:03] <jmkasunich> sampler.0.enable is handy
[02:49:16] <jmkasunich> starts sampling at the beginning of the first threading pass, stops when the spindle stops
[02:50:30] <SWPadnos> cool
[02:55:16] <jmkasunich> 64 seconds times 1mS times 5 samples plus timestamp = 3.5meg of data
[02:55:57] <jmkasunich> I hope open-office can handle 64K lines (for plotting)
[02:56:07] <SWPadnos> just barely
[02:56:25] <SWPadnos> that's the total number of rows available
[02:56:31] <jmkasunich> right
[02:56:46] <jmkasunich> stupid computer
[02:56:58] <jmkasunich> I tried to open "run3.log" in calc, and it invoked writer
[02:57:24] <SWPadnos> heh
[02:57:34] <SWPadnos> is it comma delimited?
[02:57:44] <jmkasunich> no, spaces
[02:58:01] <SWPadnos> hmm
[02:58:06] <jmkasunich> I'm used to excel's text import dialogs, where you can specify the delimiter, etc
[02:58:21] <jmkasunich> I can use sed to put commas in the file
[02:58:25] <SWPadnos> OO has the same thing, but may treat "log" as a clue that it's a text file
[02:58:33] <SWPadnos> rename it to *.asc
[02:59:01] <SWPadnos> or txt, paradoxically
[02:59:09] <jmkasunich> asc didn't work
[02:59:43] <jmkasunich> stupid stupid programs
[03:00:59] <jmkasunich> sed s/ /,/g and calling it .csv worked
[03:01:18] <jmkasunich> actually, probably calling it .csv would have worked
[03:01:29] <jmkasunich> since "worked" merely means "invoked the dialog"
[03:01:36] <SWPadnos> probably. did you try .txt?
[03:01:45] <jmkasunich> yes
[03:02:05] <SWPadnos> strange. they're listed in my "open" dialog when I select "spreadsheets" as the type
[03:02:44] <jmkasunich> renaming the original file to .csv gets me the dialog
[03:03:02] <jmkasunich> and the dialog has "space" in the delimiter list, so the sed was unneeded
[03:03:15] <SWPadnos> it's too bad there isn't an "open as" function
[03:03:24] <SWPadnos> so it thinks it has to guess for you
[03:05:39] <jmkasunich> guessing is one thing
[03:06:00] <jmkasunich> starting a completely different application when it guesses wrong (or even if it guessed right) is stupid
[03:06:15] <SWPadnos> it's not a completely different application
[03:06:21] <SWPadnos> it's a different UI for the same app
[03:06:24] <jmkasunich> if I'm in a spreadsheet program, either open the document as a fscking spreadsheet, or don't open it at all
[03:10:39] <jmkasunich> gah - there has to be a way to say "paste into the 64000 cells below this one" that is faster than selecting them all
[03:11:22] <SWPadnos> sadly, I haven't found it
[03:11:37] <SWPadnos> you can control-downarrow to get to the bottom fast (holding shift will select on the way)
[03:11:49] <jmkasunich> ah, that helps
[03:12:05] <jmkasunich> I just selected the entire column (click on letter) and pasted it everywhere
[03:12:15] <jmkasunich> I'll delete the ones at the end later
[03:12:17] <SWPadnos> it moves to the next "edge" between empty and non-empty cells
[03:12:23] <SWPadnos> ah, that works too
[03:12:41] <jmkasunich> fortunately I don't have 30,000 filled cells ;-)
[03:12:59] <SWPadnos> heh
[03:15:11] <jmkasunich> the maximum deviation between the interpolated spindle position and the 500 count/rev spindle position was 0.0389 revs
[03:15:24] <jmkasunich> and the min was -0.0346 revs
[03:16:09] <jmkasunich> that included some moderately heavy threading passes, and two passes where I jammed a piece of wood rather firmly against the motor pulley to induce a speed change
[03:17:07] <jmkasunich> scrolling thru the data, the vast majority of the samples have less than 0.005 error
[03:19:32] <SWPadnos> revolutions?
[03:19:57] <jmkasunich> yes
[03:21:01] <jmkasunich> making graphs in OO is a pain
[03:21:12] <SWPadnos> he
[03:21:14] <SWPadnos> h
[03:23:14] <jmkasunich> as in "simply doesn't work"
[03:23:21] <jmkasunich> I have a chart, but no data in it
[03:24:03] <jmkasunich> bah, I may have to use excel
[03:24:30] <SWPadnos> is it still thinking or is there just no data?
[03:25:03] <SWPadnos> did you log the Z position as well as spindle position?
[03:28:44] <jmkasunich> yes
[03:28:49] <jmkasunich> I dunno
[03:29:03] <jmkasunich> I tried it in excel, easily opened the file, and mostly made the chart
[03:29:16] <jmkasunich> then it bitched about having more than 32K points per series
[03:29:47] <jmkasunich> thats a stupid limitation - if you can handle 64K rows, then you should be able to handle 64K points
[03:31:02] <SWPadnos> you should be able to select the columns, click the pie dingus, and change the series cell ranges pretty easily
[03:32:29] <jmkasunich> right now I'm looking at the first 32K points
[03:32:44] <jmkasunich> I can see the pass where I dragged on the pulley
[03:33:23] <jmkasunich> using the 1ppr speed data, I can see a small amount of slowing during threading passes, and more slowing when I dragged it
[03:34:10] <jmkasunich> the high ppr speed data has lots of quantization noise, you can't see the speed variations during threading, and can only deduce the existance, not the magnitude, of the drop when I dragged it
[03:35:48] <jmkasunich> disregarding the spot where I dragged the pulley, the error pretty much never exceeds +/- 0.01 revs, and is mostly under +/-0.005 revs
[03:36:11] <cradek> that sounds great
[03:38:10] <jmkasunich> I should convert the speed numbers into rpm
[03:43:02] <jmkasunich> speed droop due to cutting was about 3 rpm
[03:43:19] <jmkasunich> speed droop due to me dragging on it was about 10 rpm, in about 1 second
[03:45:57] <jmkasunich> thats odd - run3.ods is only 605 bytes
[03:46:15] <jmkasunich> hard to believe, for a spreadsheet with 420000 cells
[03:46:31] <SWPadnos> well, it is a gzipped file ;)
[03:46:41] <jmkasunich> the data isn't that compressible
[03:47:33] <jmkasunich> I'm gonna put the .csv (raw data) and the .xls on my webpage, if you want to take a look at it
[03:48:07] <SWPadnos> maybe tomorrow. I'm headed for bed tonight
[03:48:14] <SWPadnos> and Germany tomorrow ;)
[03:48:15] <jmkasunich> goodnight
[03:48:17] <SWPadnos> see you
[03:50:24] <jmkasunich> http://jmkasunich.com/pics/run3.csv
[03:50:28] <jmkasunich> http://jmkasunich.com/pics/run3.xls
[03:50:59] <jmkasunich> column order in the raw data: sample number, spd-normal, spd-1ppr, pos-normal, pos-1ppr, z-position
[03:51:14] <jmkasunich> speeds are in revs/sec, spindle pos in revs, z pos in inches
[03:51:58] <jmkasunich> goodnight
[12:06:26] <CIA-34> EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/hal/components.lyx: This looks a lot better
[12:10:23] <skunkworks> BigJohnT: You're so busy :) Great work.
[12:48:55] <BigJohnT> thanks skunkworks
[12:52:56] <BigJohnT> are components that are depreceated ever dropped?
[12:55:50] <alex_joni> BigJohnT: eventually :D
[12:55:58] <alex_joni> I think comp got dropped
[12:56:27] <BigJohnT> so any that still show up should be left out of the 2.3 manual?
[12:56:38] <alex_joni> right
[12:57:11] <alex_joni> another step we talked about is changing all components' parameters to pins
[12:57:38] <BigJohnT> for 2.3?
[12:58:46] <BigJohnT> * BigJohnT is off to my machine shop ttul
[13:42:59] <skunkworks> so - 1 ppr threading now could be tested in trunk?
[19:51:53] <alex_joni> BigJohnT: think you can fit something into the devel manual?
[19:52:02] <alex_joni> I always planned to .. but never got around doing it
[19:52:26] <BigJohnT> sure there is plenty of room
[19:52:57] <alex_joni> http://eneas.juve.ro/~juve/emc/nml/NML%20messages%20usage%20desc.csv
[19:53:03] <alex_joni> it needs serious massaging
[19:53:49] <BigJohnT> is it a comma delimited file:
[19:53:51] <BigJohnT> ?
[19:54:05] <alex_joni> yeah, you can load it with Calc from OpenOffice :)
[19:54:24] <alex_joni> not sure how to make it fit into lyx though.. maybe as a table
[19:54:45] <BigJohnT> ok, I'll look at it at home
[19:55:01] <BigJohnT> nothing but windoz boxes at my machine shop
[19:55:11] <alex_joni> Excel opens it just fine
[19:55:17] <alex_joni> (one of the reasons it's .csv)
[19:55:33] <BigJohnT> yep, looking at it now
[19:55:53] <alex_joni> it's a bit old .. 1-2 years, but should mostly be up-to-date
[19:58:11] <BigJohnT> * BigJohnT has it spread out over two monitors and it's 4' wide
[20:00:12] <alex_joni> wonder if it's readable/intelligible
[20:00:22] <BigJohnT> kinda :)
[20:00:39] <BigJohnT> a list of NML messages?
[20:00:42] <alex_joni> yeah
[20:00:51] <alex_joni> and what each is used for/where it comes from
[20:01:07] <alex_joni> where it goes to, etc
[20:04:12] <BigJohnT> I noticed that 3 HAL components were marked as depreciated and about 3 had no description in the man page this morning
[20:04:56] <alex_joni> how rude of them
[20:06:14] <BigJohnT> just when you need cradek to kick them...
[20:06:47] <BigJohnT> I'm just going to leave them out of the manual for now
[20:07:08] <BigJohnT> oh, what is the difference between axis.1 and axis.9?
[20:07:10] <alex_joni> kick who? :D
[20:07:35] <cradek> * cradek kicks alex_joni
[20:07:36] <BigJohnT> who didn't put a description in the man page :)
[20:07:53] <skunkworks> that is one hell of a long leg.
[20:08:19] <alex_joni> at least you didn't say one heck of a big ass
[20:08:42] <skunkworks> heh ;)
[20:08:52] <alex_joni> cradek: it was supposed to be them, not me
[20:09:36] <BigJohnT> is this going to escalate into a one legged butt kicking contest?
[20:10:19] <alex_joni> * alex_joni kicks BigJohnT with one leg strapped behind my back
[20:10:49] <BigJohnT> you always pick on the easy targets :)
[20:11:11] <BigJohnT> * BigJohnT washes the lunch dish
[20:15:27] <alex_joni> hmm.. we had a page with funny irc moments
[20:15:51] <alex_joni> ahh.. here it is: http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?IRCquotes
[20:17:36] <BigJohnT> that's funny bunny
[21:06:55] <CIA-34> EMC: 03jepler 07TRUNK * 10emc2/src/Makefile.modinc.in: without these flags, compiling modules on 64bit machines may fail when fp functions are used
[22:31:14] <CIA-34> EMC: 03bigjohnt 07TRUNK * 10emc2/src/hal/components/deadzone.comp: Add a description
[22:47:59] <CIA-34> EMC: 03bigjohnt 07TRUNK * 10emc2/docs/src/hal/components.lyx: added description