#linuxcnc | Logs for 2012-11-12

Back
[00:00:31] <fenn> i thought you were the cnc lapping pro
[00:00:45] <kwallace1> This is closer to my budget: http://www.mirror-o-matic.com/
[00:01:19] <toastydeath> me?
[00:01:21] <toastydeath> no way
[00:01:26] <toastydeath> i know nothing about cnc lapping
[00:01:51] <toastydeath> i have heard some lectures and read some papers on deterministic polishing, but that was a long ass time ago
[00:01:55] <t12> i scored a 3axis computerized microscope stage recently
[00:02:08] <t12> i wonder if it has enough torque to machine tiny things
[00:02:43] <toastydeath> i know a pretty good amount about ultraprecision machining and how that proceeds, and a good amount about lapping in general
[00:02:47] <toastydeath> but nothing about cnc lapping
[00:35:15] <Loetmichel> mornin'
[02:02:39] <DJ9DJ> moin
[09:18:02] <r00t4rd3d> Seems like someone was tickling elmo.
[09:29:16] <Himanshu__> Hello Everyone, Is there any QT GUI Binding Example for LinuxCNC??
[09:30:24] <Himanshu__> is it easy to write GUI using QT4?? i am developing Small Machine operating utility and AXIS is really heavy for that so i want to write my own GUI.. i am Good at C++... Axis is in python and i dont want to learn new language for small gui
[09:46:56] <cpresser> Himanshu__: i am not sure if there is an QT-Binding. but you should take a look into other linux-cnc UIs. There is also 'emcsh' as a text-mode client. they all interface to linuxcnc via NML.
[09:47:50] <Himanshu__> that means if i can use NML in QT than i can write GUI right??
[09:48:08] <cpresser> yes. http://www.linuxcnc.org/docs/html/code/NML_Messages.html
[09:54:43] <jthornton> there is a qt demo on the forum
[09:57:13] <Himanshu__> jthornton:can you send me the link?
[10:04:18] <jthornton> it's in the GUI part of the forum
[10:17:10] <skunkworks> OT anyone got lirc working with mce style usb remotes? From everything I have been reading - lirc is broken in 12.04
[10:17:39] <spiderdi1on> What part do the fpga cards like the 5i20 actually play in the milling process, how are they actually utilised? I'm thinking of getting one to remove potential problems with jitter but I want to get a firm grasp in my head about how it actually achieves this.
[10:28:12] <cpresser> skunkworks: lirc is broken for most kernel-version :/
[10:28:51] <cpresser> but you might use "inputlircd" or "eventlircd"
[10:29:00] <cpresser> which are both userspace.
[10:30:17] <skunkworks> cpresser: thanks! Do you have any links?
[10:31:10] <skunkworks> spiderdijon: the mesa cards move the heavy liftiing out of the computer. (PWM, Step generation, encoder counting...)
[10:35:16] <cpresser> skunkworks: nope sorry. only google.
[10:35:50] <cpresser> unfortunately both of them have few documentation; i know some howtos, but all in german. sorry
[10:35:59] <skunkworks> Sheh
[10:36:01] <skunkworks> heh
[10:36:15] <cpresser> basically eventlircd replaces lircd.
[10:52:09] <rizo> hello, Just one short HAL question... can i somehow connect parallel port pins (bit value) to integer input. Or can i set some bits of integer value to be constant and connect the bit variable to other integer bits?
[10:53:09] <cpresser> rizo: you might use a mux, controlled by the parport-pin
[10:56:42] <spiderdijon> skunkworks: so does it act as if it is "replaying" the PWM, stepgen etc or is the FPGA actively at runtime compiling the gcode?
[11:01:23] <rizo> cpresser, this would be an option, but mux8/16 requires float input. I would prefer integer.
[11:01:55] <cpresser> rizo: build a "mux8i" comp. its just copy&paste coding
[11:02:22] <rizo> will do that, thank you
[11:54:46] <IchGuckLive> hi All B)
[11:55:14] <JT-Shop> hi
[12:00:04] <JT-Shop> tjb1:
[12:02:18] <IchGuckLive> not here plasma burneing
[12:02:24] <IchGuckLive> tjb
[12:06:18] <IchGuckLive> how mutch RAM do you folks got on your ->Intel D525MW mini-ITX
[12:06:34] <JT-Shop> all
[12:06:51] <jdh> 2gb in one, 4gb in the other.
[12:06:57] <jdh> 2 seems to be more than enough.
[12:07:00] <mhaberler> the 4GB which I just ripped out of my macbook
[12:07:23] <IchGuckLive> ok
[12:07:39] <mhaberler> echo like in empty gothic dome
[12:10:37] <IchGuckLive> i see there are now 2 mini itx atom D525 available one from intel and one from Gigabyte
[12:15:23] <mazafaka> What if you cut with the depth like 2 mm each time, and sharpen the tip of the end mill bit when it is needed, will you be satisfied with this tactics? Or you would prefer to program some pass for fining?
[12:15:46] <mazafaka> http://www.youtube.com/watch?v=UnKjIdlGXgQ&feature=plcp
[12:15:57] <IchGuckLive> mazafaka: material
[12:16:22] <IchGuckLive> it depends on shape and material
[12:17:13] <IchGuckLive> mashine also if not able to cut 10mm+ depth so no factor
[12:17:59] <mazafaka> IchGuckLive: steel, and maachine lacks rigidity if the end mill isn't sharp
[12:18:35] <IchGuckLive> if your mashine can hold climb 10mm go for it
[12:18:52] <mazafaka> we made a puncheon this way, I think what if we finally organise a cheap workflow using HSS end mill bits.
[12:19:24] <IchGuckLive> there are many cheep CO8 cutters around
[12:19:25] <mazafaka> No, 10mm for cutting height is too much.
[12:19:32] <IchGuckLive> you might be impressed
[12:19:56] <mazafaka> IchGuckLive: it is not me who is buying the stuff
[12:20:38] <mazafaka> I cut 2mm-deep passes, and the width of the cut equals to the diameter of the mill
[12:20:39] <IchGuckLive> 2mm seams very littel if it is a straigt shape to be milled
[12:21:10] <IchGuckLive> did you try 10 depth 3mm wide with a 10mm cutter
[12:21:15] <mazafaka> but the diameter of the mill bit is 3/4 inches
[12:21:23] <IchGuckLive> it makes better chips
[12:21:25] <mazafaka> IchGuckLive: i tried
[12:21:45] <IchGuckLive> the side is always better on standard steel
[12:21:49] <mazafaka> but... we have to take new end mill bit each 1.5 hours
[12:22:17] <mazafaka> it's tricky on a non-rigid machine
[12:22:31] <IchGuckLive> ok then go your way
[12:23:00] <syyl_ws> my trick on a non rigid machine is a roughing cutter :P
[12:23:56] <mazafaka> the way you say i did some work on stainless steel where the tolerances were quite big. And then had to mix up something for a precise work. Now I need to know if it's common or not
[12:25:53] <IchGuckLive> i use CRC always so with one Rouph and on finish tool you are good
[12:26:14] <IchGuckLive> if the finish tool showes marks then you got a new rouphing tool
[12:26:31] <mazafaka> So, it's more common to add a fining pass using new and sharp end mill bit?
[12:26:47] <syyl_ws> jep
[12:26:53] <mazafaka> uhu!
[12:26:55] <IchGuckLive> yes
[12:26:56] <syyl_ws> or go always full depth
[12:26:58] <mazafaka> now i know!
[12:27:00] <syyl_ws> with the cutter
[12:27:07] <syyl_ws> and take shallow cuts to the side
[12:27:15] <syyl_ws> so it will wear even
[12:27:27] <IchGuckLive> i keep 0.2mm for the finish path in depth and wide
[12:27:28] <syyl_ws> thats a strategy i use very often
[12:29:21] <mazafaka> IchGuckLive: 0.2 mm in wide and full height, or how?
[12:29:35] <IchGuckLive> yes up to 500mm height
[12:29:39] <IchGuckLive> 50.
[12:30:02] <mazafaka> yeah, why not... if you have end mill bits... :)
[12:30:05] <IchGuckLive> stainless steel
[12:30:29] <IchGuckLive> my max is 225 mm in cutting height
[12:30:40] <IchGuckLive> FOAM O.O
[12:30:54] <mazafaka> I had 2 old already short 16mm end mills for 3 puncheons and 3 matrices
[12:31:26] <IchGuckLive> you need to calculate the tools into your price
[12:31:35] <IchGuckLive> special tools are full price
[12:31:47] <mazafaka> i know, but it's a plant.
[12:31:58] <IchGuckLive> even if
[12:32:08] <mazafaka> right now i'm going to another plant
[12:32:35] <IchGuckLive> but if you got your own sharpener then maybe heads first is better for you
[12:32:57] <IchGuckLive> but miling time also counts
[12:33:26] <IchGuckLive> mazafaka: where are you from USA Europ
[12:33:40] <mazafaka> yeah, sharpening of the tiny end bits takes time... I'm in Russia
[12:33:46] <psha> IchGuckLive: ru
[12:34:23] <mazafaka> I probably will work in formerly secret underground shop, I don't know...
[12:34:38] <IchGuckLive> russia then tool price shoudt be very low
[12:35:17] <mazafaka> We used HSS, and many delicious thingies on WFL.
[12:35:28] <IchGuckLive> polen tool manufactures sell 10mm co8 for less then 5 Eur
[12:35:51] <IchGuckLive> Ukrain do so 2
[12:36:01] <mazafaka> heh, i will know it now.
[12:39:57] <IchGuckLive> with a small 4 Axis cutter sharpener you can make money
[12:40:54] <IchGuckLive> worlds famos KAINDL sharpener are developing now on the mach4 scale
[12:42:17] <IchGuckLive> on youtube B) http://www.youtube.com/watch?v=TvzqGeNudo8
[12:42:23] <IchGuckLive> proto type
[12:43:06] <mazafaka> but does it create the back angle?
[12:43:33] <IchGuckLive> its NC controled so on your choice yes
[12:43:35] <mazafaka> we had one driven manually, and created back surface manually
[12:44:05] <mazafaka> how nice
[12:44:24] <IchGuckLive> i think they will fit in the main tool cataloges
[12:44:36] <IchGuckLive> so give it a number and you are done
[12:46:20] <rizo> what is the easiest way to test the my custom hal component? I would like to change the input values and monitor the output.
[12:46:50] <IchGuckLive> mazafaka: http://www.youtube.com/watch?v=d868lekecJw MANuel
[12:47:41] <IchGuckLive> rizo: python output to terminal
[12:50:37] <rizo> thank you, will try that
[12:50:41] <rizo> bye
[12:51:34] <mazafaka> yeah, sharpening become easy and interesting using these devices
[12:52:21] <IchGuckLive> Kaindl is cheep
[12:54:04] <awallin> does that reduce the diameter of the endmill each time it is sharpened?
[12:54:28] <IchGuckLive> side grinding yes
[12:54:34] <IchGuckLive> fromt no
[12:54:39] <IchGuckLive> front
[13:07:51] <Caanon_work> where would I go to learn about the mathematical algorithms behind slicing and geode generation?
[13:08:41] <IchGuckLive> heekscad
[13:09:42] <IchGuckLive> for G-code generation or on howto do it your own way
[13:10:15] <IchGuckLive> Caanon_work: ?
[13:10:43] <Caanon_work> on your own
[13:10:51] <IchGuckLive> awallin is a good place for opencamlib
[13:11:13] <IchGuckLive> do you go for 2D or more like 3D
[13:11:33] <IchGuckLive> engraving or freeform
[13:11:37] <Caanon_work> as in, mesh-plane intersection, using djikstra's algorithm for perimeter generation, etc
[13:11:47] <Caanon_work> hm, thanks
[13:11:57] <Caanon_work> I'll check into awallin and opencamlib
[13:12:18] <IchGuckLive> there is also reprap a good thing for mesh plane issues
[13:12:22] <Caanon_work> IchGuckLive: I'm thinking more along the lines of what slic3r does or skein forge does
[13:12:30] <awallin> opencamlib is for 3-d milling with real milling tools. that is different from a pure "z-slice" where you do not care for the shape of the cutter
[13:12:45] <IchGuckLive> Caanon_work: there is the Master
[13:13:08] <andypugh> Caanon_work: AFAIK there is no need for any complex algorithms.
[13:15:19] <andypugh> The STL is a set of triangular faces. You search for all triangles which span the slicing plane. Then work out where each line crosses the plane and add those to the path as points, then make the two points from each triangle into a line, then assemble the lines into a path.
[13:16:22] <andypugh> Caanon_work: Do you speak Matlab?
[13:17:24] <IchGuckLive> Caanon_work: check this out its in a free trial and will fit all your needs http://www.deskproto.com/products/newinv6.htm
[13:17:51] <archivist> he wants the algorithms!
[13:17:58] <IchGuckLive> ok
[13:18:19] <andypugh> Lines 30 to 65 here do all that is needed to slice an STL into an outline: http://pastebin.com/g1aD2hEb
[13:18:52] <IchGuckLive> http://www.google.de/url?sa=t&rct=j&q=stl%20to%20g-code%20algorithm&source=web&cd=1&cad=rja&ved=0CCEQFjAA&url=http%3A%2F%2Fweb.firat.edu.tr%2Fiats%2Fcd%2Fsubjects%2FManufacturing%2FMTE-31.pdf&ei=skihUOy-DYjKswbX84CADw&usg=AFQjCNHXWg4eylEZQBQYAVhauPa6p1Ofmg
[13:19:13] <andypugh> (That's the "Pugh Algorithm" :-)
[13:19:36] <IchGuckLive> slice and dice is also a good info
[13:20:17] <archivist> I like a set of books called Graphics Gems although a bit old now
[13:20:42] <Caanon_work> andypugh: yup, sure do
[13:21:01] <Caanon_work> thanks
[13:21:23] <Caanon_work> yeah, I'm more looking into the plane-mesh intersection and the optimal path generation
[13:21:55] <IchGuckLive> in the pdf i posted the link are all infos
[13:22:52] <IchGuckLive> W. Ma, P. He, “An adaptive slicing and selective hatching strategy for
[13:22:54] <IchGuckLive> layered manufacturing”, Journal of Materials Processing Technology,
[13:22:55] <IchGuckLive> 89–90, 191–197, 1999
[13:23:21] <Caanon_work> perfect :D
[13:23:23] <Caanon_work> thanks
[13:24:46] <IchGuckLive> in the 0.22 version of heekscnc there are 4 Reprap post processors and a new slice python strategie so you might get a good info there 2
[13:25:38] <Caanon_work> cool, I'll look into it
[13:29:16] <IchGuckLive> deskproto is a good and fust CAM for this kind of use also, give it a tryle
[13:29:25] <IchGuckLive> ok im off By
[13:42:07] <alex_joni> Newton's third law of emotion: For every male action, there is a female overreaction!
[13:46:45] <Jymmm> alex_joni++
[13:49:07] <Jymmm> alex_joni: My better half's response: "only problem with that is getting male action "
[13:56:58] <tjb1> JT-Shop:
[13:57:52] <gmouer> PCW around? got a funny tuning servos w a 7i77 board, strange waveform on halscope
[14:02:44] <gmouer> seeing approx a 30hz noise imposed on halscope on the velocity, analogout and ferror lines, no idea what is causing it
[14:03:17] <cradek> share a plot with us
[14:03:41] <gmouer> oh man, i am new to linux, how do you do a screen capture?
[14:04:01] <Tom_itx> lcnc has a plot builtin
[14:04:13] <Tom_itx> may or may not work for that
[14:04:40] <gmouer> is it shift print=screen or something similar?
[14:04:49] <Tom_itx> shift or ctrl
[14:04:50] <Tom_itx> one
[14:04:51] <archivist> accessories has a screenshot utility
[14:04:54] <Tom_itx> or the other
[14:04:56] <Jymmm> CTRL+PRN_SCRN
[14:05:09] <cradek> applications / accessories / take screenshot
[14:05:26] <gmouer> ok, let me give it a whirl.
[14:05:27] <cradek> yes it may also be bound to a keyboard shortcut
[14:05:38] <Caanon_work> another question: I'm blundering my way through SolidWorks. I've used 3D modeling programs before, but never really in depth. Are there any good tutorials for SolidWorks?
[14:05:46] <gmouer> I will try the point and shoot method first
[14:06:07] <gmouer> brb (if I am lucky)
[14:06:27] <cradek> Caanon_work: google gives 3,210,000 results when I search for solidworks tutorial
[14:06:47] <tjb1> Caanon_work: youtube
[14:06:57] <tjb1> and playing with it
[14:07:02] <Jymmm> cradek: ut, how many of those are viagra ads?
[14:07:06] <Jymmm> s/ut/but/
[14:08:20] <archivist> Caanon_work, I found blundering through with the occasional reading of the help system got me up to speed mainly
[14:08:30] <alex_joni> 3d solid viagra?
[14:08:39] <Tom_itx> mastercard card will come with builtin keyboard and display soon
[14:08:53] <Jymmm> alex_joni: isn't that the point?
[14:08:55] <Tom_L> http://www2.electronicproducts.com/New_Mastercard_features_LCD_screen_includes_keyboard-article-fajb_credit_card_keyboard_nov2012-html.aspx
[14:13:00] <gmouer> ok, got the screenshot, how do I post it here?
[14:13:16] <Tom_itx> imagebin
[14:13:19] <Tom_itx> .ca
[14:13:24] <gmouer> thx
[14:14:06] <gmouer> my luck, imagebin website is down LOL
[14:14:18] <gmouer> be back soon though LOL
[14:14:33] <Tom_itx> try .org
[14:15:06] <alex_joni> or .com
[14:15:18] <Jymmm> or .gov
[14:16:46] <gmouer> .org works, but it won't take a .png file from the screenshot utility
[14:17:11] <Jymmm> LOL Got Gas??? http://newyork.cbslocal.com/2012/11/06/sandy-gas-woes-continue-people-trading-gas-for-sex-on-craigslist/
[14:19:36] <gmouer> got it http://imagebin.org/235592
[14:20:35] <gmouer> about a half a volt noise on my velocity and analogout signals that is at 50 ipm, it goes to 1.5v at 100 ipm
[14:20:59] <gmouer> amplitude follows the speed commanded
[14:21:11] <cradek> are you in mm or inch?
[14:21:17] <gmouer> inch
[14:21:31] <gmouer> noise is around 27-30hz rate
[14:21:45] <cradek> tell us about your setup
[14:22:14] <JT-Shop> tjb1:
[14:22:22] <tjb1> JT-Shop: Yes?
[14:22:27] <gmouer> 7i77 mesa, servo dynamics drive, brushed servo, vel loop in drive, position loop in emc
[14:22:36] <JT-Shop> you get your homing sorted out?
[14:22:58] <gmouer> It tunes pretty normal but that noise shows up on the following error also
[14:23:41] <tjb1> cncbasher posted something but I cant try it until wednesday
[14:24:01] <gmouer> I am stumped what the cause of that noise is? tried filter caps on the dc power in to the 7i77 board, no effect
[14:24:05] <JT-Shop> are you trying to home more than one axis at a time with all sharing the same input?
[14:24:21] <cradek> I think you're jumping to the conclusion that it's noise
[14:24:47] <cradek> you could have oscillation in the velocity loop, or in the position loop, or you could have a mechanical resonance
[14:24:55] <andypugh> gmouer: PCW is of the opinion (I think, not wanting to put words into his mouth) that you will get better results tuning to minimise the PID error term than the f-error.
[14:24:56] <tjb1> No, one at a time
[14:25:00] <gmouer> that could well be !! Linuxcnc is new to me I am used to nice clean curve lines on servo displays
[14:25:16] <cradek> how did you tune the velocity loop?
[14:25:52] <gmouer> first tuned the velocity loop open loop, by the manual for the drive, that seems to behave normal, (no noise)
[14:26:12] <cradek> do you hear this disturbance in the machine motion? what does it sound like?
[14:26:13] <gmouer> but it was a low speed where the noise is much lower
[14:26:23] <gmouer> no, the machine is smooth
[14:26:25] <JT-Shop> bbl
[14:27:07] <gmouer> I have searched around and not seen this on other halscope plots I found
[14:27:08] <cradek> that's hard to believe with velocity varying 0.3ish inch/sec
[14:27:38] <cradek> it always sounds smooth, even when making this plot you shared?
[14:28:33] <gmouer> yes, smooth during that plot, my only concern is it showing in the following error, wouldn't know it was happening if not for halscope
[14:29:02] <cradek> you didn't show following error on the plot...
[14:29:30] <cradek> I have a hard time understanding how it can sound smooth. is it a heavy/light/big/small machine?
[14:29:42] <gmouer> yes, I left the following error out for now, I have things detuned attempting to see if I could impact that "noise" which I could not
[14:30:33] <gmouer> It is a omniturn lathe, I would class it bewteen small and medium, its a 5c collet type machine, linear rails,
[14:31:00] <cradek> ok, so you're not moving much mass
[14:31:13] <cradek> how about if you put your hand on it while it's moving? can you feel the disturbance?
[14:31:25] <gmouer> right, very little mass, maybe 10-20 lbs tops
[14:31:50] <pcw_home> If you set P,I,D to 0 do you still get the noise?
[14:33:00] <gmouer> pcw: I tuned P way down to a min that would move the motor, it lowers the noise but does not eliminate it
[14:33:21] <gmouer> My I and D are at 0, vel loop in the drive, pos loop in linuxcnc
[14:33:34] <pcw_home> the motor should mainly be moved by FF1
[14:34:24] <pcw_home> so you should get approximate motion (and some drift) with PID = 0
[14:34:29] <gmouer> I am a novice at tuning, but yes, FF1 has a substancial effect in lowering the following error
[14:35:06] <pcw_home> FF1 is the principle term in a velocity mode loop
[14:35:28] <SORAO> salve
[14:35:34] <gmouer> well, I had no idea of that, I thought it was P
[14:35:43] <pcw_home> and really should be adjusted first (with maybe a very small bit of P)
[14:35:47] <gmouer> I have 2.93 FF1 and 300 for P
[14:36:09] <pcw_home> inch scale?
[14:36:16] <gmouer> yes, inch
[14:36:41] <gmouer> I had adjusted ff1 to center the following error above and below zero
[14:37:13] <pcw_home> does setting P to 0 eliminate the noise?
[14:37:18] <gmouer> I get a good following error at low speeds, but it goes to hell fast as speed increases
[14:37:27] <gmouer> I will try P=0 right now
[14:37:35] <gmouer> machine is next to me
[14:37:42] <cradek> until you've verified your velocity loop doesn't oscillate, you can't tune the position loop
[14:37:50] <pcw_home> (warning it will drift)
[14:38:32] <cradek> oh I see what pcw is doing: FF1 only will let you move it at various speeds
[14:38:42] <pcw_home> Thats why I am suggesting "opening the PID loop" by setting P,I.D to 0
[14:38:47] <cradek> gotcha
[14:39:41] <gmouer> no, with p=0 the noise is there but very slightly less in amplitude
[14:39:59] <gmouer> FF1 still set at about 3.0
[14:40:05] <cradek> then you can stop trying to tune the position loop...
[14:40:52] <cradek> you might have a noisy tach, or a periodic mechanical binding, or a badly tuned amp
[14:41:14] <gmouer> so right now, the guess is the noise is within the velocity loop?
[14:41:14] <cradek> but can you feel it when you put your hand on the axis?
[14:42:18] <cradek> if it really does sound and feel smooth, you might have bogus position feedback
[14:42:22] <pcw_home> Yeah, could also be some kind or input connection/ground loop error (assuming you are misreading 60Hz as 30Hz)
[14:42:26] <gmouer> mechanical binding for a second..... there are 2 1 inch dia springs that retract the way covers, quite a bit of tension on those springs, more as you go towards Z0
[14:42:27] <cradek> no guessing, still troubleshooting
[14:43:15] <gmouer> the amplitude of that noise varys directly with speed, it reached 1.5 v p-p at 100 ipm
[14:43:33] <gmouer> (on the 10V analog control line)
[14:43:33] <cradek> frequency always the same?
[14:43:48] <gmouer> yea, around 27-30hz
[14:44:01] <cradek> you've seen that on a real wired scope?
[14:44:08] <gmouer> no, on halscope
[14:44:21] <cradek> you said "the analog control line"
[14:44:43] <cradek> you meant the commanded dac value?
[14:44:45] <gmouer> yes, I lied, analogout2 signal
[14:45:13] <cradek> with only FF1 there is no reason you should see any variation on that commanded dac value. are you sure??
[14:45:29] <gmouer> let me recheck
[14:46:25] <gmouer> yes, its on hm2....analog2 signal, about .7V p-p
[14:46:51] <cradek> but that makes no sense. you are making a mistake somewhere.
[14:47:02] <cradek> while jogging a constant speed, right?
[14:47:13] <gmouer> yes, a g01 command at 50 ipm
[14:47:20] <gmouer> one inch move
[14:47:38] <gmouer> noise is present during entire move, also on longer moves
[14:47:41] <cradek> you are maybe not changing the right pid values?
[14:47:47] <pcw_home> Yeah that should just be a trapezoidal velocity signal
[14:48:36] <gmouer> its the right pid values, if I crank P up a lot it goes into oscillation
[14:49:00] <cradek> did you zero out I, D, etc? everything but FF1?
[14:49:16] <gmouer> I and D have been at zero all along
[14:49:46] <gmouer> I does not have any great effect, D is not liked at all, very bad things happen
[14:50:19] <pcw_home> D is normally not needed for a velocity loop (the drive does 'D')
[14:50:26] <gmouer> pcw: what should be a trapezoidal velocity signal? where should that be seen?
[14:50:29] <cradek> you are making a mistake somewhere but I am not sure what it could be :-/
[14:51:12] <pcw_home> 'I' should not be tuned normally until everything else is very close
[14:51:13] <gmouer> I have tuned other servos but seen clean plots without all that noise
[14:52:03] <gmouer> pcw, yes, tuning the pos loop it behaves like you would expect, its just that noise translates into following error
[14:52:35] <pcw_home> Are you saying that you see 30 Hz noise on the analog out as viewed in halscope?
[14:53:00] <pcw_home> (with only FF1 in the PID loop)
[14:53:02] <gmouer> as a note, I played with the tach and gain on the drive pcb last night, no effect on that noise
[14:53:39] <gmouer> yes, 20hz noise on analogout2, as shown on the halscope plot
[14:53:48] <gmouer> opps, 30hz noise
[14:54:13] <gmouer> nothing, adjustment wise, seems to effect that noise
[14:54:24] <gmouer> other than speed
[14:54:37] <gmouer> the noise is proportional to speed
[14:54:40] <pcw_home> OK so thats all internal, not related to 7I77, analog.tachometer,
[14:54:48] <pcw_home> etc
[14:55:07] <asdfasd> gmouer some things may cause that kind of noise, like crack in glass encoder for example
[14:55:26] <cradek> no, not with P,I,D=0
[14:55:38] <cradek> there is no feedback in the equation
[14:55:54] <cradek> that's why this doesn't make sense yet
[14:56:01] <pcw_home> can you edit the hal file and verify that the PID values are 0? this really make no sense
[14:56:05] <asdfasd> I see
[14:56:17] <pcw_home> makes
[14:56:55] <alex_joni> or use halmeter to look at the actual values
[14:57:06] <cradek> are you positive you are watching the dac output value, not the feedback velocity?
[14:57:08] <gmouer> it was setup initially with pncconf, the PID values in hal originate in the INI file, I can change them there if you want, I have been using the calibration window so far
[14:58:05] <alex_joni> calibration window.. eww ;)
[14:58:14] <gmouer> watching on the halscope display hm2......analogout2 and hm2 encoder2 velocity (f error off for the moment)
[14:58:47] <gmouer> want me set pid to zeros in the ini file and reboot?
[14:59:17] <cradek> you could check them with halmeter (or halcmd show) like alex suggested
[14:59:34] <cradek> when in doubt, troubleshoot, don't shoot in the dark
[15:00:05] <cradek> something is not working how you expect - we should find out what
[15:00:11] <gmouer> ok, I will check them with halmeter brb
[15:00:34] <alex_joni> and remember, a reboot is seldom the answer
[15:01:04] <cradek> restarting linuxcnc would be enough after editing the ini.
[15:02:21] <syyl> using the cnc mill to shape some markings...
[15:02:23] <syyl> http://www.youtube.com/watch?v=E8PVtRCPhSM&feature=youtu.be
[15:02:50] <syyl> came out pretty good
[15:02:50] <syyl> https://dl.dropbox.com/u/24396704/Saupe%20Kopf/IMG_1001.JPG
[15:03:19] <cradek> c
[15:03:41] <Connor> Man is that painfully slow...
[15:03:51] <gmouer> ha! Pgain=300 on halmeter for Z, shows as zero on calibrate screen
[15:04:08] <pcw_home> calibrate is fussy
[15:04:12] <syyl> just for the video that slow
[15:04:18] <syyl> later i went with 1000mm/min
[15:04:25] <cradek> ok, something is confusing the calibrate app
[15:04:39] <gmouer> I will change P to zero in the ini file and restart linuxcnc
[15:04:41] <cradek> what linuxcnc version is this?
[15:05:29] <cradek> syyl: awesome
[15:05:48] <gmouer> 2.6 pre, master
[15:06:02] <cradek> oh dear
[15:06:58] <cradek> currently, master is MUCH more untested/untrusted than usual
[15:07:14] <cradek> do you have a strong reason to use it?
[15:08:33] <pcw_home> Thats neat syyl
[15:08:45] <gmouer> some of the pncconf fixes are there along with remap fixes for front/rear tool remap
[15:09:05] <cradek> oh ok
[15:09:27] <gmouer> I was under 2.5 originally and seen the same servo behavior
[15:09:57] <gmouer> I put it off till later, later is now, all is done except for servo tweaking
[15:10:22] <cradek> I was wondering whether the calibrator was broken in 2.5
[15:10:39] <cradek> most likely it's an incompatibility with pncconf-generated configs :-/
[15:10:40] <gmouer> it appears that with P=0 I get a following error quickly, I have Ferror set to 1/4 inch while tuning
[15:11:15] <cradek> that's not surprising. but is it smooth?
[15:11:16] <pcw_home> Sure but does it oscillate?
[15:12:10] <gmouer> can't really tell, it only moves 1/4 inch, I will turn the f error up and look at the halscope plot
[15:12:39] <cradek> I hope you verified your limit switches already
[15:12:59] <pcw_home> Its should not be terribly bad if you have FF1 right
[15:13:12] <cradek> ... and your travel to them is unobstructed
[15:16:02] <gmouer> encoder02 velocity has the noise, analogout2 is extremely clean
[15:18:39] <pcw_home> so tach/drive/encoder/analog interface issue
[15:18:57] <gmouer> ok, I am following you
[15:19:22] <pcw_home> can you feel the vibration?
[15:20:19] <gmouer> that is hard to say, those big springs under the slide that hold the way covers make noises and do "twang" things
[15:20:30] <gmouer> still, I would say it moves pretty smooth
[15:21:21] <gmouer> its something you hear more than feel. I tore the whole thing apart being concerned only to find it was those springs making the noises, not the linear rails or ballscrew
[15:21:57] <gmouer> they are those telescoping metal way covers
[15:22:11] <pcw_home> to isolate interface vs tach/drive/encoder issues
[15:22:13] <pcw_home> you can try driving it either way with a battery
[15:22:14] <pcw_home> (are you really really sure its 30 Hz and not 60 Hz?)
[15:23:09] <gmouer> yea, I was hoping for 60hz, a open shield or such, the halscope peroid calcs out to 27-30hz area
[15:24:05] <gmouer> roughly 33 ms
[15:24:16] <pcw_home> well sure sounds like drive/tach issue
[15:24:37] <gmouer> with your troubleshooting help, it certainly points that way
[15:25:35] <gmouer> I am thinking to unhook the servo coupling to the ballscrew, and use a battery for the control voltage and see what it looks like then
[15:25:49] <pcw_home> you can also use the 7I77 DAC to generate a step velocity signal for tuning the velocity loop
[15:26:52] <pcw_home> (well the PID loop with just FF1 will do that as well if accell is turned up high)
[15:27:01] <gmouer> this lathe only has 5" travel in the Z so it would get hairy without a position loop
[15:27:16] <andypugh> Be aware that with the motor unconnected to the rest of the machine it will be harder to tune.
[15:27:58] <gmouer> andy: yes I have seen that before, but a final tune is not the goal, only finding the source of that noise/oscillation
[15:28:59] <gmouer> pcw: the accel is pretty mild, set at 15 right now, inch units
[15:29:04] <gmouer> its a linear rail machine
[15:30:17] <pcw_home> Just saying ts a easy way to tune the velocity loop by just changing some parameters
[15:31:03] <gmouer> oh, ok, for tuning, I thought you were talking a high accel
[15:31:36] <gmouer> "Will do that" speaking in ref to the noise
[15:31:44] <pcw_home> (If accell is set high, a jog will output what loots like a step in velocity to the drive)
[15:31:46] <pcw_home> so you can tune the drives response for no overshoot/oscillation
[15:31:47] <gmouer> I understand now
[15:32:18] <pcw_home> just a lazy way to generate a step
[15:32:25] <gmouer> gottcha, a 7i77 reversable, variable freq, battery emulator
[15:34:16] <gmouer> well, fellows, I guess our work is done for now, I have to troubleshoot in the drive/tach area searching for the cause of that oscillation
[15:34:54] <gmouer> thanks much guys, I knew you guys would get me on the right track to finding the cause
[15:35:10] <gmouer> plus, I wanted to confirm that it was not normal
[15:37:31] <pcw_home> no, un-responsive oscillations are not the norm (but anythings possible on a unknown machine)
[15:47:49] <cradek> if it sounds and feels smooth, it still might be bogus position feedback.
[15:48:30] <cradek> a real scope on the real tach might be useful (use differential mode on the scope, don't short out your tach circuit)
[15:51:11] <gmouer> yes, crakek, I do have a real scope, at least I now have ideas of where to look
[15:52:18] <Aero-Tec> need some help
[15:52:35] <Aero-Tec> I zero all tool to tool one
[15:53:11] <Aero-Tec> I need to add a tool to the tool table that was not in Mach
[15:53:32] <Aero-Tec> well it was but some how things got screwed up
[15:53:48] <Aero-Tec> not there now
[15:54:28] <Aero-Tec> in mach there was a way to touch off tools and load the offset into the tool table
[15:54:50] <Aero-Tec> have not found anything like that in EMC
[15:55:10] <Aero-Tec> so what is the normal way to load offsets into EMC?
[15:55:27] <JT-Shop> the touch off button
[15:56:14] <Aero-Tec> would I do a tool change to tool 1 as it is the zero for all others and write down the off set numbers?
[15:56:40] <Aero-Tec> so the touch off will load info into the tool table?
[15:56:53] <JT-Shop> yes
[15:57:04] <Aero-Tec> thought touch off was to zero the machine offsets
[15:57:22] <JT-Shop> does that too
[15:57:30] <JT-Shop> http://gnipsel.com/linuxcnc/g-code/gen05.html
[15:58:30] <Aero-Tec> thanks for the link
[15:59:03] <JT-Shop> np
[15:59:15] * JT-Shop goes to walk the dog
[16:08:26] <tjb1> JT-Shop:
[16:12:37] <Aero-Tec> he is walking his dog
[16:23:21] <DJ9DJ> good night
[16:26:53] <JT-Shop> tjb1:
[16:29:40] <djheinz> does anyone know a software that can export .plt ???
[16:30:32] <Loetmichel> djheinz: OS?
[16:31:38] <djheinz> linux, win, or osx...
[16:32:15] <andypugh> What _is_ .plt?
[16:32:38] <djheinz> its a HP, plotter file format...
[16:33:05] <Loetmichel> HPGL
[16:33:12] <andypugh> Yes, I thought it might be HPGL
[16:33:17] <Loetmichel> djdelorie: corel draw can
[16:33:24] <Loetmichel> sorry
[16:33:33] <djheinz> it can?
[16:34:02] <Jymmm> Yep
[16:34:04] <Loetmichel> djheinz: corel draw, BoCNC, and any program that can drive a HP 75** plotter ;-)
[16:35:14] <andypugh> You will probably find that print-to-file with the printer set to pen plotter will work. Or Use Graphic Converter (Mac) http://www.lemkesoft.com to convert other things into it.
[16:35:40] <t12> djheinz: coreldraw
[16:35:50] <t12> er
[16:35:52] <t12> i coulda read that heh
[16:36:11] <andypugh> djheinz: Inkscape doe HPGL
[16:36:20] <t12> you can also pretty trivially write out hpgl sans-curves via javascript in illustrator
[16:36:46] <t12> theres also a few windows printer drivers
[16:36:51] <t12> that will output hpgl
[16:36:52] <andypugh> That code I linked earlier converts STL to HPGL :-)
[16:37:07] <t12> and apparently some of the older designjet drivers can too
[16:37:16] <t12> though i've never really got that to work
[16:37:29] <t12> my expirence is that theres alot of dialects of hpgl also that some devices like/dontlike
[16:38:15] <andypugh> I used it to slice STL into sticker paper layers to make: https://picasaweb.google.com/lh/photo/tjht3D69CSp25i9tYwD7dtMTjNZETYmyPJy0liipFm0?feat=directlink using my Mac and a pen-plotter.
[16:39:36] <andypugh> But, of all the suggestions, I reckon Inkscape is the most free and cross-platform, and it can export G-code too with the right plug-in.
[16:40:48] <Loetmichel> t12: thats why corel draw for example uses only "PU" "pD" and "ST"
[16:42:23] <djheinz> hmm
[16:43:14] <Loetmichel> ... has the drawback of some facettes on circles, if the adjustments are not fine enough in the exporter
[16:43:53] <Loetmichel> http://www.cyrom.org/palbum/main.php?g2_itemId=12527
[16:44:03] <djheinz> is HPGL the SAME as PLT?
[16:44:14] <Loetmichel> it is
[16:45:06] <djheinz> so it works to just rename?
[16:46:37] <Loetmichel> if the interpreter can read the special hpgl: yes
[16:46:57] <Loetmichel> like t12 saidf: hpgl ist necessarily like hpgl
[16:47:21] <Loetmichel> there are nearly as much variants outside as it had plotters in the old days.
[16:49:51] <djheinz> Ok... i bought an OLD Cipher CNC machine... and it uses some STONE-AGE software... and it eats plt files..... i plan to use another controller on it later, so i can use linuxcnc on it...
[16:50:35] <toastydeath> what's the control
[16:50:44] <toastydeath> do you have pictures>
[16:51:01] <Loetmichel> djheinz: has the machine steppers or servos?
[16:51:43] <djheinz> steppers..
[16:51:49] <Loetmichel> if steppers its normally no big problem to identify the step/dir signals going into the drivers and feed them from a PC LPT ;-)
[16:52:05] <Loetmichel> s/normally /usually
[16:52:07] <djheinz> http://www.finn.no/finn/torget/tilsalgs/viewimage?finnkode=38040918 -- pictures!
[16:52:20] <toastydeath> oh wow
[16:52:31] <djheinz> hehe..
[16:53:07] <toastydeath> i was like "what milling machine takes .plt"
[16:53:13] <toastydeath> mystery solved
[16:53:29] <Loetmichel> hrhr
[16:53:37] <djheinz> lol
[16:53:51] <Loetmichel> it IS a milling machine... it has rotating engraving bits ;-)
[16:54:00] <djheinz> it works... and i allready have jobs on it...
[16:54:17] <toastydeath> engraving machine < router < milling machine
[16:54:53] <djheinz> it can do all of the things.... its a solid piece of machine..
[16:55:16] <toastydeath> i meant that the labels are general size classes
[16:55:34] <djheinz> ok..
[16:56:24] <Loetmichel> toastydeath: so what do you call my little machine?
[16:56:46] <toastydeath> Loetmichel, what's your machine
[16:56:55] <Loetmichel> http://www.cyrom.org/palbum/main.php?g2_itemId=12569
[16:57:05] <Loetmichel> http://www.cyrom.org/palbum/main.php?g2_itemId=12578
[16:57:08] <toastydeath> benchtop mill
[16:58:50] <Loetmichel> for size reference: the table is about 15" wide and 5" deep
[16:59:36] <toastydeath> bench mill.
[16:59:36] <Loetmichel> it has a movement of 200*110*110mm (roughly 8" by 5" by 5")
[17:01:06] <toastydeath> it's not really an issue of "is this a router or a mill"
[17:01:12] <toastydeath> because of the machine design
[17:01:22] <t12> loet: yeah
[17:01:31] <t12> loet: have you found that coreldraw botches some images?
[17:01:39] <t12> i find that with crosshatch fills sometimes it just wont fill some things
[17:01:51] <t12> or will leave half of the fills missing on something, or will run forever or crash
[17:08:18] <Loetmichel> t12; my corel is version 8
[17:08:34] <Loetmichel> so i dont know whats up with the newer ones
[17:08:43] <t12> maybe i should use an old one heh
[17:08:49] <t12> i also never got color selection to work
[17:08:53] <Loetmichel> maybe
[17:08:54] <t12> weirdly i actually use it for pen plotting
[17:08:57] <Loetmichel> mine runs fine
[17:15:34] <toastydeath> i thought i had more pictures of the machines i used to run
[17:15:35] <toastydeath> https://sphotos-b.xx.fbcdn.net/hphotos-ash4/3331_171286785082_6548606_n.jpg
[17:15:51] <toastydeath> some collets, in order: ER-32, R8, and CAT-50
[17:15:52] <toastydeath> https://sphotos-b.xx.fbcdn.net/hphotos-ash4/4664_190865730082_7721615_n.jpg
[17:16:34] <Loetmichel> toastydeath: i made a bigger one for my ex boss:
[17:16:48] <Loetmichel> http://www.cyrom.org/palbum/main.php?g2_itemId=4935
[17:17:10] <Loetmichel> 1500mm* 1020mm*165mm movement ;-)
[17:17:25] <toastydeath> nice
[17:17:57] <Loetmichel> mostly for glass fibre plastic/carbon fibre plastic, some plywood and some aluminium sheets
[17:18:00] <Aero-Tec> why would the Gcode quick reference show a error?
[17:19:08] <toastydeath> here's a big ass endmill
[17:19:09] <toastydeath> http://i2.photobucket.com/albums/y5/toastydeath/DSC_4255.jpg
[17:19:12] <Aero-Tec> also in the instruction for setting tool table with touch off it says to
[17:19:14] <Aero-Tec> Verify that Machine > Touch off to fixture is selected.
[17:19:17] <Loetmichel> http://www.cyrom.org/hajo/09_01_09.avi <- That was a day at the old company
[17:19:52] <Aero-Tec> there is no Touch off to fixture that I can find
[17:19:56] <Loetmichel> $me is the guy in the blach shirt
[17:20:03] <Loetmichel> ... the red sweater was my boss
[17:23:21] <Aero-Tec> JT-Shop: you around?
[17:23:29] <Aero-Tec> back from walking the dog?
[17:23:30] <JT-Shop> yup
[17:23:34] <Aero-Tec> cool
[17:24:07] <Aero-Tec> I had asked some questions and was hoping for a answer
[17:24:18] <Aero-Tec> if you can help that would be great
[17:24:48] <JT-Shop> depends on the question if I can help or not :)
[17:25:02] <Aero-Tec> why would the Gcode quick reference show a error?
[17:25:14] <Aero-Tec> also in the instruction for setting tool table with touch off it says to
[17:25:22] <Aero-Tec> Verify that Machine > Touch off to fixture is selected.
[17:25:30] <Aero-Tec> there is no Touch off to fixture that I can find
[17:26:00] <JT-Shop> it is in the machine menu at the bottom
[17:26:31] <Aero-Tec> there is a menu at the bottom?
[17:26:31] <JT-Shop> are you using the Axis GUI?
[17:26:38] <Aero-Tec> yes
[17:26:55] <Aero-Tec> I had not seen it
[17:26:59] <JT-Shop> look
[17:27:01] <Aero-Tec> will look
[17:27:04] <Aero-Tec> lol
[17:27:06] <Aero-Tec> will do
[17:27:54] <Aero-Tec> also why would the Gcode quick reference bring up a error message?
[17:28:06] <JT-Shop> what error?
[17:28:25] <Aero-Tec> will have to look at it
[17:28:29] <Aero-Tec> BRB
[17:32:11] <Aero-Tec> no such file or dir
[17:32:37] <Aero-Tec> also I can not find a menu at the bottom of axis
[17:32:58] <Aero-Tec> there are info boxes
[17:33:20] <Aero-Tec> but nothing that looks like a menu
[17:34:05] <Aero-Tec> JT-Shop: the error is no such file or dir
[17:34:10] <JT-Shop> the Machine Menu item
[17:34:48] <JT-Shop> the gcode quick referenc uses some smoke and mirror magic that does not always work, just use the user manual
[17:35:19] <Aero-Tec> the top has File Machine View Help
[17:35:32] <Aero-Tec> ok
[17:35:34] <Aero-Tec> will do
[17:35:56] <Aero-Tec> is not the machine menu at the top?
[17:36:01] <Aero-Tec> it look like it is
[17:36:05] <JT-Shop> yep Machine is the winner
[17:36:06] <Aero-Tec> looks
[17:36:15] <JT-Shop> click on it and look at the last item
[17:36:54] <Aero-Tec> it may be the winner but has nothing that is Touch off to fixture
[17:37:10] <Aero-Tec> I did meny time and there was no Touch off to fixture
[17:37:25] <JT-Shop> what version of LinuxCNC are you using
[17:37:44] <Aero-Tec> the new one I think
[17:37:50] <Aero-Tec> can look
[17:38:09] <JT-Shop> it is there then
[17:38:10] <Aero-Tec> 10. something if mem serves
[17:38:30] <Aero-Tec> will check the version number
[17:38:33] <JT-Shop> "Tool touch off to fixture" is the last thing
[17:44:46] <Aero-Tec> JT-Shop: looks like it is 8.04
[17:44:55] <Aero-Tec> axis 2.4.6
[17:45:07] <Aero-Tec> Gnome 2.22.2
[17:46:03] <JT-Shop> 2.5.1 is the latest version and you can install that on 8.04
[17:47:05] <Aero-Tec> how does one install it?
[17:47:05] <Aero-Tec> I have done the updates
[17:47:06] <Aero-Tec> would it not have been updateded with the auto updates?
[17:47:27] <Aero-Tec> guess not as it is not updated
[17:48:23] <Aero-Tec> should I update the system to the latest version as well?
[17:48:40] <Aero-Tec> if so is that hard to do?
[17:49:21] <JT-Shop> LinuxCNC will not update from 2.4 to 2.5 automagiclly
[17:49:45] <Aero-Tec> sorry for being such a noob at all this
[17:49:51] <JT-Shop> http://wiki.linuxcnc.org/cgi-bin/wiki.pl?UpdatingTo2.5
[17:50:11] <JT-Shop> everyone is a nubee when they start it is expected
[17:51:01] <andypugh> Upgrading from 8.04 to 10.04 runs the risk of making your machine run worse, so just upgrade LinuxCNC nor Ubuntu unless there are things in 10.04 that you really want.
[17:51:01] <Aero-Tec> thanks
[17:51:19] <Aero-Tec> my computer is a intel P4
[17:51:36] <andypugh> Stick with 8.05 then.
[17:51:39] <JT-Shop> better stay with 8.04 on a P4
[17:51:41] <andypugh> 8.04 even
[17:51:46] <JT-Shop> hey that rimes
[17:55:24] <andypugh> Well, there is something I didn't know. Native in-the-code units of LinuxCNC are mm.
[18:00:08] <JT-Shop> well there you go, all this time you assumed EMC was imperal
[18:00:51] <cradek> except for in AXIS, insanely enough
[18:02:22] <JT-Shop> LOL
[18:02:42] <jp1> axis = imperial touchy = jedi
[18:02:45] <andypugh> Yes, that does seem a little crazy. But then Axis is crazy. It's almost a pity it works so well.
[18:03:28] <cradek> agreed.
[18:03:34] <andypugh> Tcl is number 71 in a popularity list which has APL as 31...
[18:03:56] <cradek> yes but which popularity list is the most popular?
[18:04:34] <cradek> reminds me of seb's suggestion that we make a wiki page listing the most out-of-date wiki pages
[18:04:51] <andypugh> Depends if you ask John Ousterhout or anyone else :-)
[18:05:03] <JT-Shop> if that page gets out of date does it make the list
[18:05:22] <SWPadnos> ?PARSE ERROR
[18:06:08] <cradek> my ex who worked in insurance IT said their actuaries were still using APL, ca 200x. it might be the best tool for some jobs...
[18:06:17] <cradek> ?REDO FROM START
[18:07:14] <jp1> anyone know if the 7i77 manual is up to date? seem to remember an issue with the spindle terminals
[18:07:15] <SWPadnos> I wonder what the difference would be between these "most popular" lists: most programs using a language, most people wanting to use a language, most programs still running which were written in a language
[18:07:31] <cradek> SWPadnos: big, would be my guess
[18:07:45] <SWPadnos> yeah, I was thinking that
[18:08:09] <cradek> SWPadnos: those lists are generally about what language geeks find cool right now
[18:08:33] <SWPadnos> sure. the revision history of such a list would also be interesting :)
[18:08:34] <JT-Shop> jp1: the online manual is up to date
[18:08:37] <cradek> SWPadnos: another useful list would be what languages to learn to support yourself in a useful way
[18:08:55] <SWPadnos> "which languages will be the most popular in the future" ...
[18:09:20] <cradek> that's just a list that helps you troll
[18:09:59] <andypugh> Well, this was the list in question, I think it is based on sourceforge projects: http://lang-index.sourceforge.net
[18:10:04] <jp1> jt-shop: thanks
[18:10:08] <cradek> a language geek at work has been recently raving about "native code" i.e. compiled languages, because they use less battery power on mobile devices
[18:10:17] <SWPadnos> heh
[18:10:20] <JT-Shop> np
[18:10:34] <cradek> shortly before (of course): CPU is free - use interpreted languages
[18:10:58] <cradek> all of a sudden, oh hell, efficiency matters
[18:11:27] <SWPadnos> that would be an interesting benchmark - power usage to run XYZ algorithms or whatnot, per language
[18:11:38] <andypugh> APL and Matlab are together, and both are great for vectorised data.
[18:12:00] <cradek> SWPadnos: probable result: use C
[18:12:35] <SWPadnos> probably true
[18:13:43] <andypugh> If you were looking earlier when I posted an STL slicer, the key to that is the two lines of code that build a list of all facets through the slicing plane.
[18:13:43] <Tom_itx> reprap guys are using slic3r currently
[18:13:43] <SWPadnos> depends on whether the benchmark is complex enough that a compiler or some objecty environment could end up more efficient than a person thinking about the problem
[18:13:47] <andypugh> Firstly those with one node > z, plus those with 2 nodes >z. That would be rather more code in a non-vectorised language
[18:14:42] <L84Supper> skeinforge also still works
[18:15:21] <L84Supper> there is also work on using voxels and the GPU to slice in real time on the fly
[18:15:22] <andypugh> Err, I was discussing programming languages, I have no actual need to slice STL files :-)
[18:15:22] <jp1> repsnapper wored well too!
[18:15:32] <L84Supper> sorry
[18:15:40] <jp1> ditto
[18:15:44] <SWPadnos> andypugh, but what happens under the hood? sure, you wrote 2 lines of code, but did the library your code uses make 10 passes over the data or something else silly
[18:15:44] <Tom_itx> andypugh did you hand laminate those layers?
[18:16:01] <Tom_itx> in the handle pic
[18:16:07] <SWPadnos> andypugh, yeah, I had to think about the standard template library vs. the 3D model file format :)
[18:16:35] <andypugh> SWPadnos: Oh, yes, it's horribly inefficient under the hood. It's best thought of as a very high level data manipulation scripting language. But it definitely has uses.
[18:17:27] <andypugh> Tom_itx: Yes, though the plotter cut two little holes in each piece, so I only had to slot them on to pegs and rub them down/
[18:17:44] <Tom_itx> wondered how you aligned them
[18:18:31] <andypugh> Even then it was spectacularly dull. One of those jobs that has you clenching your legs and feeling a bit sick. Or is that just me?
[18:21:08] <t12> theres an even rougher compiler problem
[18:21:22] <t12> can you optimize code for opecode and io power efficency
[18:21:26] <t12> opcode even
[18:22:19] <jp1> andypugh: have you seen the ic haus interpolator ic's
[18:22:33] <andypugh> No.
[18:22:52] <andypugh> What do they do? (Or, more specifically, what do they interpolate?)
[18:23:02] <jp1> was thinking of trying them out on some scales
[18:23:16] <jp1> interpolate sin cos
[18:23:28] <andypugh> Ah, OK. I might have seen them then.
[18:23:38] <andypugh> Slightly spendy?
[18:25:19] <jp1> yeah but not terible
[18:25:29] <andypugh> AH, Actually I was thinking of http://uk.rs-online.com/web/p/resolver-to-digital-converters/7094709/
[18:26:30] <jp1> think there like 15-30 for the ic depending on the type
[18:26:56] <andypugh> (To be fair, there are £20 variants with less resolution, that one is a bit special)
[18:29:18] <jp1> not bad to use as a converter even if you just want quad output
[18:30:23] <andypugh> jp1: You could probably use my Arduino code (you can throw away the sine-wave excitation generator)
[18:30:48] <jp1> yeah im playing with that now
[18:31:06] <andypugh> Though then all you are left with is a fast integer arctan and quadrature emulation, and you can probably do that yourself.
[18:31:46] <jp1> just thought the ic would be faster than the arduino
[18:31:51] <andypugh> I confess I gave up on it, 1mS update rate seemed a bit slow.
[18:32:08] <andypugh> I bought the Mesa 7i49.
[18:33:31] <Tom_itx> what's it do?
[18:33:45] <jp1> resolver card
[18:33:50] <Tom_itx> oh
[18:33:58] <Tom_itx> to replace your arduino?
[18:34:08] <andypugh> (The Arduino stopped working, I scoped the excitation signal and it had gone very funny, and I couldn't be bothered to debug it. Probably a bad capacitor, might have been a 3.3V one)
[18:36:20] <andypugh> SWPadnos: Do you know what saicannon is for? It seems to be a partial duplicate of emccannon but with surprisingly little common code.
[18:44:23] <JT-Shop> http://www.youtube.com/watch?v=FQMbXvn2RNI
[18:45:52] <andypugh> cradek: What happens if three code files share a header with a function declaration in it, then all files contain different definitions of the function? Specifically, what controls which function gets executed by a call from a 4th file?
[18:47:22] <kwallace> Isn't SAI = Stand Alone Interpreter?
[18:49:11] <andypugh> Ah, yes, that rings a bell.
[18:50:16] <kwallace> I recall there being a section on the SAI in the NIST/EMC1 documentation.
[18:54:37] <kwallace> http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.57.6516&rep=rep1&type=ps
[18:57:05] <andypugh> OK, that is what it is. So, it looks like saicannon, emccannon and gcodemodule all do similar but different things at the same hierarchy level. The digging continues...
[18:58:57] <kwallace> ftp://ftp.isd.mel.nist.gov/pub/emc/
[19:12:04] <JT-Shop> http://www.secretlifeofmachines.com/
[19:17:12] <andypugh> Right, time to sleep.
[19:17:15] <andypugh> Night all
[19:17:20] <JT-Shop> night
[20:44:59] <abetusk> does anyone do PCB milling? What is your clearance?
[20:45:36] <abetusk> also, has anyone made their own pick and place machine?
[20:47:52] <mutilator> clearance?
[20:47:58] <Tom_itx> i've done laser toner transfer but no milling to speak of
[20:48:13] <Tom_itx> mutilator, between traces
[20:55:48] <mutilator> oh
[20:56:03] <mutilator> i've only done 47 mils on mine and only twice as a test
[20:56:12] <abetusk> yeah, as in the minimum track you can take out with your endmill
[20:56:24] <abetusk> I was hoping to get down to around 10 mil...
[20:56:28] <mutilator> i have some smaller buts i want to try but i have to test speeds
[20:56:36] <mutilator> i imagine it'd have to cut really slow
[20:56:43] <mutilator> bits*
[20:56:47] <abetusk> how slow?
[20:57:02] <mutilator> i dont know i havnt tested it
[20:57:08] <mutilator> just based on them being so fragile
[20:57:23] <abetusk> I have some v-bits, they seem pretty sturdy...
[20:57:53] <mutilator> oh yea i have some flat endmill flutes
[21:00:15] <abetusk> speaking of which, I can only find ones that are .1 mm at the end, do you know of any source to get sharper ones?
[21:01:24] <mutilator> i just bought a nice one from cmt
[21:01:25] <Tom_itx> you need high rpm
[21:01:44] <abetusk> cmt?
[21:01:49] <abetusk> and what rpm would you recommend?
[21:01:56] <mutilator> http://www.amazon.com/CMT-858-001-11-Cutting-Diameter-16-Inch/dp/B000P4NSYG
[21:02:17] <mutilator> nice and big
[21:02:18] <Tom_itx> abetusk figure your chip load and tool diameter
[21:02:23] <Tom_itx> the numbers don't lie
[21:02:26] <Tom_itx> http://tom-itx.dyndns.org:81/~webpage/etching/mills1.jpg
[21:02:34] <Tom_itx> i tried but my spindle is just too slow
[21:03:42] <Tom_itx> i'd suspect around 10k rpm
[21:03:43] <abetusk> mutilator, what is the end width?
[21:04:02] <mutilator> i cant measure it..
[21:04:14] <mutilator> it goes to a point
[21:04:17] <Tom_itx> use carbide too as FR4 is quite abrasive
[21:04:31] <t12> these desktop centrifuges appear to have some really nice little 3ph motors in them
[21:04:39] <Tom_itx> and copper is crap to mill anyway
[21:04:40] <t12> 13.5krpm, 170w, 115v
[21:05:07] <abetusk> Tom_itx, I have an 8k rpm spindle...
[21:05:18] <mutilator> i just tried engraving a old copper block from a heatsink yesterday
[21:05:25] <mutilator> needs some tweaking to speeds for sure heh
[21:06:11] <mutilator> glad it was pretty thick, i can just mill off the top layer and start over
[21:13:48] <abetusk> mutilator, I'm really hesitant to just buy a $20 bit without knowing what the end width is. If you try it, report back
[21:14:48] <mutilator> yea i dont think i'll use a v to cut it
[21:14:59] <mutilator> i'll use my flat end mills
[21:55:22] <mutilator> abetusk?
[21:55:35] <mutilator> https://dl.dropbox.com/u/24715738/IMG_20121112_224222.jpg
[21:56:01] <mutilator> my tablet cant focus that close
[21:56:18] <mutilator> but it comes down to as much of a point as you can get..
[21:56:35] <pfred1> the linuxcnc website is just the wiki now?
[21:57:11] <abetusk> mutilator, hmm, so .1mm or .2mm at least? It's really hard to differentiate with the naked eye when it's that small
[21:57:51] <mutilator> it looks like half the width of the pin but i dunno how deep you need to go
[21:57:56] <mutilator> so you'd have to calc the width
[21:58:13] <mutilator> whats 1oz cupper like 1.5mil
[21:58:46] <abetusk> 1.34 mil
[21:58:53] <abetusk> about
[21:59:12] <pfred1> ought oh I smell a PCB miller
[22:02:10] <tjb1> What is the advantage of placing things on the servo thread vs the base thread?
[22:03:02] <tjb1> or disadvantage…or reason
[22:07:17] <Tom_itx> with mesa cards i'm told to use the servo thread
[22:07:31] <Tom_itx> my config doesn't define a base thread
[22:08:37] <tjb1> I could put all servo items on the base thread and it wouldnt matter and vise versa if the speed was the same?
[22:09:59] <Tom_itx> i'm not sure
[22:10:07] <Tom_itx> i just did what i was recomended to do
[22:16:58] <kwallace> The base thread is for software signal generation, such as step pulses. The servo thread is for reading the machine status and calculating the number pulses to put out.
[22:18:32] <Tom_itx> well in the case of mesa cards the pulses are generated on board i think
[22:19:33] <kwallace> That's why the base thread isn't needed.
[22:19:39] <Tom_itx> yeah
[22:20:47] <kwallace> Also, functions that need floating point calculations can't be done in the base thread, but only in the slower servo thread.
[22:22:50] <pfred1> the Swiss really do make nice stuff don't they? https://www.youtube.com/watch?v=S8qkaTsr2_o
[22:44:30] <Tom_itx> yup
[22:52:49] <pfred1> when I hit the lottery I'm getting one of those P&P machines just so I can watch it.
[22:53:08] <Tom_itx> i didn't click
[22:53:28] <Tom_itx> i just know swiss machines are good
[22:53:49] <pfred1> it looks like the video is speeded up but it isn't
[22:54:00] <Tom_itx> oh, pick n place..
[22:54:08] <Tom_itx> yeah some of those are pretty impressive
[22:54:32] <pfred1> that one is a Swiss one and they're calling it ultra high speed
[22:55:06] <pfred1> needless to say it is slamming some parts into boards
[22:55:52] <Tom_itx> i wonder if it detects if a part slips in transit
[22:55:56] <Tom_itx> i would suppose so
[22:56:14] <Tom_itx> you see the good parts of it but i bet sometimes they crash
[22:56:25] <Tom_itx> just wonder how the crash recovery works
[22:57:25] <pfred1> I don't know something tells me it runs like a Swiss watch though :)
[23:01:12] <theos> :)
[23:22:51] <Aero-Tec> I upgraded EMC
[23:23:20] <Aero-Tec> now have my max jitter real high again
[23:23:42] <pfred1> SMI interrupt?
[23:23:56] <Aero-Tec> any one of the gurus here?
[23:24:03] <Aero-Tec> I would guess that
[23:24:10] <Aero-Tec> that is what fixed it last time
[23:24:36] <Aero-Tec> I checked the file and the SMI fix is there
[23:24:48] <Aero-Tec> unless it is using a new file
[23:25:17] <Aero-Tec> in the home dir is where the file is
[23:25:33] <Aero-Tec> is there a new place for it?
[23:26:43] <Aero-Tec> rtapi.conf file
[23:27:05] <Aero-Tec> looks to still have the SMI update in it
[23:29:12] <Aero-Tec> looks like the one I looked at may be a backup copy I made
[23:29:35] <Aero-Tec> will look where the instructions say to look
[23:30:45] <Aero-Tec> also I did a test set up of a simple 5 axis cnc for testing out the new update
[23:31:07] <Aero-Tec> it brings up what looks to be a model of the CNC machine
[23:31:18] <Aero-Tec> is that normal?
[23:31:30] <Aero-Tec> 2.4 did not do that
[23:47:55] <Aero-Tec> calling emc guru
[23:48:24] <Aero-Tec> the rtapi.conf will not allow me to edit it
[23:49:11] <Aero-Tec> the file belongs to root, I can load it into editor but can not save it
[23:49:23] <Aero-Tec> any help on how to fix this?