#emc | Logs for 2010-12-22

[07:45:10] <ChanServ> [#emc] "This is the #emc channel - talk related to the Enhanced Machine Controller and general machining. Website: http://www.linuxcnc.org/, wiki at http://wiki.linuxcnc.org/"
[07:57:50] <Howard> psha: Hi man how are you doing today ?
[07:59:44] <psha> fine
[08:00:10] <Howard> Are you snowed in ?
[08:00:38] <psha> yea, yesterday here was very fine weather :)
[08:00:42] <psha> windy and showy :)
[08:02:50] <Howard> London Heathrow only has 1 runway open and Frankfurt is diverting all air traffic to other places.
[08:04:40] <psha> brother said that in edinburgh trains are going only to glasgo :)
[08:05:54] <Howard> What is the temperature where you are now ?
[08:06:24] <psha> -5 - -7 i think
[08:06:49] <psha> pretty warm
[08:07:40] <Howard> OK that's a bit warmer and you can at least get to work.
[08:10:11] <psha> heh, temperature is not important :)
[08:14:15] <Howard> Hate to brag but we are at +25c sea temp 22C and a swim is great.
[08:14:36] <psha> )))
[08:14:51] <psha> yea, that weather is even worse for work )
[08:15:51] <Howard> True that happens as the Guy's are often found surfing :0)
[08:47:40] <oem> A recent video of my EMC Index Werke Screw Machine. http://www.facebook.com/profile.php?id=508489044#!/video/video.php?v=177377668951083
[08:54:03] <awallin> hrm, requires vb login...
[08:54:07] <awallin> fb
[08:56:15] <oem> oh lame
[08:56:23] <oem> I dont have it anywhere else right now either
[08:56:31] <oem> Ill hook it up next time i come around
[08:57:28] <awallin> this kind of monster?? http://www.orca-inc.com/Index-Werke-DG-12-Automatic-Single-Spindle-Screw-Machine.aspx
[09:05:56] <oem> VERY similar
[09:06:09] <oem> the particular one I have is a B60
[09:08:34] <oem> awallin what rig are you using?
[09:09:55] <awallin> oem: many balls in the air... this is the mill we've worked a long time with http://www.youtube.com/jmi80
[09:10:22] <awallin> here's an unfinished lathe project... http://www.anderswallin.net/2009/11/jhq-lathe/
[09:10:46] <awallin> and just recently som reprapping http://www.anderswallin.net/2010/12/3d-printer-day-2/
[09:13:04] <awallin> bbl
[09:13:23] <oem> nice one
[09:13:25] <oem> okay
[09:13:31] <oem> Ill catch you next time im in
[09:13:35] <oem> later man keep up the good work
[12:15:00] <atmega> awallin: you have a BF20?
[12:32:49] <nullie> Should EMC2 run from live cd? It gives me local apic error
[12:37:00] <JT-Shop> yes it should run
[12:46:59] <archivist> apic..what are you running it on?
[12:50:40] <jthornton> what's an apic?
[12:52:00] <archivist> iirc its the power control
[12:52:23] <archivist> so its very active on a laptop
[12:52:30] <jthornton> ah ok
[12:52:45] <jthornton> concrete :) http://i47.photobucket.com/albums/f163/johnplctech/Shop%20Addition/P1010050.jpg
[13:00:55] <awallin_> new workshop floor ?
[13:05:41] <jthornton> yes
[13:05:57] <jthornton> just need walls and a roof now :)
[13:32:11] <archivist> and toys to fill
[13:52:02] <JT-Shop> I have the toys now that is the reason I'm adding on I can't park a tricycle in the garage :/
[13:53:45] <archivist> I just know space gets filled....add more concrete before the walls get built :)
[13:56:09] <JT-Shop> LOL, I'd have to add some flat ground first
[14:39:33] <alan78_> hi
[14:39:33] <the_wench> hello alan78_, you have a question?
[14:40:22] <alan78_> alan78_ is now known as Alan78
[14:41:02] <Alan78> anybody have a good stepper driver schematic who works fine with emc2?
[14:43:43] <jthornton> a wiring schematic?
[14:44:53] <MattyMatt> all stepper drivers work fine if they have Step and Direction inputs
[14:45:49] <MattyMatt> choose your chip then use the example circuit in the datasheet, is a workable plan
[14:47:05] <Alan78> it's needed programe chip for my steppers? or is this a generic driver?
[14:47:37] <MattyMatt> most generic drivers have step & dir inputs these days
[14:48:08] <Alan78> aja
[14:48:26] <Alan78> at this moment i have an stepper from laser print
[14:48:39] <Alan78> 7,5º
[14:49:10] <Alan78> you think will be simple to make it work?
[14:49:55] <MattyMatt> some older circuits have 4 inputs - one for each coil, and the software needs to generate the steps in a more complicated way. I don't know if emc2 can do that easily
[14:51:23] <Alan78> thats the problem
[14:51:51] <jthornton> stepgen has a variety of stepping types http://www.linuxcnc.org/docview/html/man/man9/stepgen.9.html
[14:51:52] <Alan78> i don't know if emc2 can send correct pulses to stepper
[14:52:02] <Alan78> by soft
[14:52:52] <atmega> doesn't seem like a job for emc
[14:52:55] <jthornton> and read http://www.linuxcnc.org/docview/html/hal_rtcomps.html#sec:Stepgen
[14:54:37] <MattyMatt> ah neat :) those stepgen options make it easy
[14:56:13] <MattyMatt> personally I'm most interested in the 5 phase options. full step/dir drivers for 5ph motors are expensive but I can make 5 hbridges
[14:57:45] <MattyMatt> I've been sorting through my junkpile recently tho, and I've got quite a lot of TTL, so I may have enough to make a step/dir
[14:57:56] <atmega> that's a lot of pulses (for a 3 axis machine)
[14:58:09] <atmega> and a lot of h-bridges
[14:58:46] <MattyMatt> yep, there's an Allegro chip with 2x 2A hbridges in tho. small and neat
[15:00:38] <MattyMatt> I'll probably improve the leadscrew on my slow Y first, and then my incentive to go 5ph on the motors will diminish
[15:00:45] <Alan78> ummm
[15:01:15] <Alan78> not easy for a newbie jeje
[15:01:23] <MattyMatt> 5mm/s = 12ipm :p
[15:02:13] <MattyMatt> Alan78 yeah for a newb I suggest sth like http://www.pololu.com/catalog/product/1202
[15:02:43] <Alan78> directly with simple controler, i would control a 7,5º 24v stepper? or need always programe with stepgen?
[15:02:53] <Alan78> i will see it
[15:03:31] <Alan78> i can make my own pcbs and asembly components
[15:03:33] <MattyMatt> stepgen is used to configure your machine with a simple driver too
[15:03:40] <Alan78> that's no problem
[15:03:50] <MattyMatt> ah cool, can you do SMT?
[15:03:56] <Alan78> problem it's find a good schematic to mount jejeje
[15:04:33] <Alan78> simple pcb yes, smd no
[15:05:55] <MattyMatt> in that case L297+L298 is popular, but there are more modern single chip ones
[15:07:25] <MattyMatt> how big will your machine be? will you buy better motors later on? 7.5deg steppers from printers are usually tiny and weak
[15:08:43] <Alan78> yes, small machine
[15:09:06] <Alan78> i wont start with simple machine and learn to work with emc2
[15:09:58] <Alan78> look this schematic
[15:10:00] <Alan78> http://www.roboticsguy.com/images/misc/L297-L298-stepper-motor-controller-schematic.png
[15:10:23] <Alan78> could serve?
[15:10:46] <atmega> you can get a 3-axis chinese tb6560 board for $50 shipped
[15:11:32] <MattyMatt> that's what I did :) except I got 4 axis so I had one spare
[15:12:36] <elmo40> always a good idea to have room for expansion :)
[15:12:55] <atmega> or for frying one axis when the wire breaks
[15:12:56] <MattyMatt> I use all 4 axis now. I put a 4th motor in my vice and mill teeth on gears
[15:13:29] <atmega> heh, cheap rotary?
[15:14:18] <MattyMatt> as cheap as it gets :)
[15:14:53] <MattyMatt> I actually paid ~$3 for the motor tho, so not quite free
[15:16:22] <Alan78> my stepper: http://www.mitsumi.co.jp/latest/Catalog/compo/motor/m49sp2k_e.html
[15:16:30] <MattyMatt> I'm making a more sophisticated one now. belt drive at first to get 1800 steps/rev, and with that I'll mill a wormgear and get 36k steps
[15:16:43] <Alan78> i only need broke another two laser printers jejeje
[15:17:05] <atmega> what do you use to generate the code for the gears?
[15:18:07] <MattyMatt> hand written with a bit of help from blender
[15:19:36] <MattyMatt> I draw the gear in blender, then draw the toolpaths for one tooth, export that as gcode then wrap it in a loop by hand
[15:20:10] <Connor> awallin: I had a question about your setup for printing.. Did you setup M codes to control the temp?
[15:23:47] <MattyMatt> I'll do something more sophisticated with that blender script and blender's animation ability, but mention of awallin reminds me I should be moving towards HeeksCAD instead :)
[15:24:20] <MattyMatt> voxel milling sim = nice
[15:24:26] <atmega> has it had any updates in the last month or two?
[15:24:32] <atmega> (heeks)
[15:24:49] <MattyMatt> #cam is very quiet recently
[15:25:08] <atmega> what is a voxel?
[15:25:20] <MattyMatt> 3d equiv of a pixel
[15:25:40] <MattyMatt> volume element
[15:30:05] <MattyMatt> http://www.youtube.com/watch?v=17mpgQ4KVLg
[15:31:34] <MattyMatt> this one shows the point of it better http://www.youtube.com/watch?v=9sctWgd_5rs&feature=related
[15:34:33] <MattyMatt> of course, voxels in blender would be nice too
[15:42:13] <MattyMatt> MattyMatt is now known as MattyMob
[15:51:23] <nullie> Hello. Manual says that motherboard should not have integrated video. How this affects EMC performance
[15:51:24] <nullie> ?
[15:52:20] <jthornton> some work fine
[15:53:02] <jthornton> like the D510MO has integrated video and it is a very good board you just have to test your latency
[15:54:01] <Jymmm> ?
[15:54:18] <Jymmm> oh,nm
[15:56:06] <Jymmm> nullie: Most on-board video sorta take shortcuts in performance versus a dedicated video card, this usually causes high latency.
[15:56:31] <nullie> so d410 should work fine too?
[15:56:47] <Jymmm> nullie: But, what you can do is disable the onboard video, and install a video card to get around that issue.
[15:57:05] <nullie> well, actually I don't want to use videocard
[15:57:14] <nullie> I'd like to have headless setup if that's possible
[15:58:27] <Jymmm> Do you have a non-computer controlled E-STOP circuit?
[15:58:34] <nullie> no
[15:58:58] <Jymmm> Then I wouldn't recommend headless without one.
[15:59:32] <nullie> I guess I can add emergency button trough geckodrive
[16:00:16] <Jymmm> Well, estop is not just motors, but spindle too.
[16:00:46] <atmega> I would suggest a non-compuer controlled e-stop even non-headless
[16:01:25] <Jymmm> atmega: very true, and use a safety relay as well
[16:03:32] <nullie> you mean just cut off all power?
[16:03:50] <Jymmm> Not necessarily.
[16:04:23] <Jymmm> If your setup has brakes, estop would engage them, etc.
[16:04:33] <nullie> oh
[16:06:54] <Jymmm> And do NOT use standard relays in a latching configuration, there is a flaw in those designs where the estop can fail.
[16:11:31] <Jymmm> http://search.ebay.com/270682393863 and the specs for it http://www.sti.com/relays/SR103AM/index.htm
[16:12:23] <Jymmm> as example
[16:12:46] <Connor> What is that for?
[16:13:13] <Jymmm> Its a safety relay for use in an estop circuit.
[16:13:30] <Connor> and, what flaw are you talking about with a latching relay?
[16:14:07] <Jymmm> Connor: hold the reset button and press the estop button and see what happens.
[16:14:37] <Connor> I don't have one. I don't even have my estop hooked up yet.
[16:16:41] <JT-Shop> Connor: did you get a chance to look at the delete button?
[16:16:57] <Jymmm> In EU, you are required to have TWO safety relays in your estop circuit.
[16:17:02] <Connor> Not yet. Christmas Rush has had me slammed.
[16:17:17] <Connor> JT-Shop: Let me take a look at it again.
[16:17:40] <Jymmm> JT-Shop: Go lathe me a xmas tree!
[16:18:21] <Connor> JT-Shop: Thanks for the spacers BTW, They work great! I've removed any perceivable backlash out of my X axis.
[16:18:30] <JT-Shop> cool
[16:18:44] <Connor> and can now tighten the jam nut down without binding the screw.. still a bit harder to turn than I would expect.. but.. it works.
[16:18:47] <JT-Shop> Jymmm: a miniture one
[16:18:57] <Jymmm> JT-Shop: out of 8" round al stock would work =)
[16:19:24] <JT-Shop> biggest collet I have is 1.625
[16:19:49] <Jymmm> JT-Shop: Heh, that might be cute, but competing with china too.
[16:20:18] <psha> hi
[16:20:18] <the_wench> hello psha, you have a question?
[16:20:21] <psha> hello the_wench
[16:21:12] <grommit> Hi psha
[16:21:20] <Connor> JT-Shop: Okay, the reason your delete doesn't work.. is their isn't a function to handle the delete process after you click the delete. :)
[16:21:29] <psha> grommit: hi )
[16:21:41] <grommit> you have been busy
[16:21:51] <grommit> I am trying to get to time to read about the glade stuff
[16:22:11] <psha> you are welcome :)
[16:22:16] <grommit> :-)
[16:22:17] <psha> now it's pretty stable and i hope working
[16:22:33] <psha> and you can use it in place of pyvcp panel :)
[16:22:56] <grommit> Yeah, it looks nice. I plan to replace my current panel with it when I can get time to play with it
[16:22:59] <psha> unfortunately i've no screenshots comparing same panels in pyvcp and gladevcp
[16:23:17] <grommit> well, when I get to it, I'll send you some
[16:23:27] <grommit> my pyvcp panel isn
[16:23:32] <psha> but with gladevcp it's more easy to compose widgets as you like with proper alignment and sizes
[16:23:37] <grommit> isn't very impressive but...
[16:23:45] <grommit> cool
[16:24:12] <grommit> I do remember being restricted and not liking some of the results...
[16:24:30] <psha> now you are free in placing things as you like :)
[16:24:39] <grommit> excellent
[16:26:07] <skunkworks> wow - I am going to have to play with glade now. Darn it!
[16:26:35] <grommit> yeah, get right on that. You've spent enough time on that monster ;-)
[16:26:46] <skunkworks> bblheh
[16:26:48] <skunkworks> heh
[16:28:16] <psha> if i won't be bored with it maybe next series will have different actions like estop/on/off/etc... so it will be possible to use them in vcp without need of halui
[16:29:07] <grommit> psha: do you run a machine with EMC and if so, what kind of machine?
[16:30:53] <psha> no machine
[16:31:11] <psha> for me now it's some for of charity ;)
[16:31:31] <grommit> Why EMC? vs other software out there?
[16:31:56] <psha> it's very interesting to work on emc and i feel that it deserved spending my 'hobby' time on it :)
[16:32:14] <grommit> very cool. I (for one) appreciate it!
[16:32:46] <psha> i'm still waiting for signed contract for some emc related work but chances are not hight...
[16:33:04] <psha> so i'm fixing it here or there while waiting ;)
[16:33:32] <grommit> i see.
[17:31:19] <JT-Shop> Connor: well that makes sense
[18:07:03] <psha> Bonny: hi
[18:07:06] <Bonny> hello... Just generic C question.
[18:07:10] <Bonny> Hello psha..
[18:07:48] <Bonny> I don't know how to proper cast this to work usbMsgPtr[0]=rq->wValue;
[18:08:49] <Bonny> UsbMsgPtr is uchar rq is pointer
[18:09:07] <Bonny> and wValue is word parameter
[18:09:30] <andypugh> Does anyone know what encoder.N.position-latched and it's associated pins is intended for?
[18:09:32] <Bonny> and I just wan't low 8 bits to get in usbMsgPtr
[18:11:14] <andypugh> If usbMsgPtr is a Uchar then it only has 8 bits?
[18:11:40] <psha> *((uchar *) ptr) is first 8 bits of data pointed by ptr
[18:11:41] <andypugh> (I missed anything you said before 3 lines ago)
[18:11:55] <Bonny> Yes.. And I want just 8 bits from rq->wValue
[18:12:11] <psha> so replace ptr with req->wValue
[18:12:18] <andypugh> psha: _First_ 8 bits, but you want to be very sure about your endian-ness.
[18:12:45] <psha> low bits are (uchar) (0xff & *ptr)
[18:12:59] <Bonny> rq->wValue is structure and wValue is not 1'st entry
[18:13:03] <psha> i assume *ptr is pointing to something integral
[18:13:29] <psha> pastebing rq definitionsomewhere
[18:13:38] <andypugh> Aye, I think mask and cast is probably safest.
[18:14:15] <andypugh> (But I know less than almost everyone else here about C)
[18:14:36] <Bonny> http://pastebin.ca/2026436
[18:14:41] <psha> hm, it seem that my spacebar don't want to work all time
[18:16:01] <Bonny> and structure here
[18:16:02] <Bonny> typedef struct usbRequest{ uchar bmRequestType; uchar bRequest; usbWord_t wValue; usbWord_t wIndex; usbWord_t wLength; }usbRequest_t;
[18:16:27] <psha> low bits of wValue are (0xff & rq->wValue)
[18:16:30] <Bonny> http://pastebin.ca/2026441
[18:16:42] <psha> high bits - 0xff & (rq->wValue >> 8)
[18:17:17] <Bonny> I got compiler error
[18:17:18] <Bonny> main.c:70: error: invalid operands to binary & (have 'int' and 'usbWord_t')
[18:17:46] <psha> what is usbWord_t? unsigned short?
[18:17:52] <psha> then 0xffu i think
[18:18:03] <Bonny> http://pastebin.ca/2026441
[18:18:38] <psha> btw what's library?
[18:19:08] <Bonny> .... it's AVR C
[18:19:37] <psha> http://vusb.wikidot.com/driver-api
[18:19:41] <psha> this one?
[18:19:43] <Bonny> I had some program in C and want to modify. In Assembler is simple but C is still so stranger for me
[18:20:08] <Bonny> jep just VUSB
[18:21:30] <andypugh> sprintf("%i", sizeof(usbWord_t)) :-)
[18:21:49] <psha> heh, man
[18:21:53] <Bonny> should be 8
[18:21:57] <psha> have you seen what is usbWord_t?
[18:22:06] <psha> typedef union usbWord{ unsigned word; uchar bytes[2];
[18:22:06] <psha> }usbWord_t;
[18:22:13] <Bonny> and I just wonder of byte 4
[18:23:23] <andypugh> It's a union, so doesn't that mean you can access the bytes directly?
[18:23:30] <psha> so it's either 0xff & rq->wValue.word
[18:23:46] <psha> or just rq->wValue.bytes[0] (it's not low but first byte)
[18:23:52] <andypugh> (I have never actually _used_ a union, so they are mysterious to me)
[18:24:10] <psha> andypugh: it's a way to make different 'views' of same data
[18:24:32] <andypugh> Aye, I get that part, but I am hazy on the details.
[18:24:34] <psha> for example this union states that you'd either access data as single 16bit value or as two byte values in array
[18:24:35] <Bonny> hmm seems that just simple usbMsgPtr[0]=data[3]; works
[18:24:57] <Bonny> as usbRequest_t *rq = (void *)data;
[18:25:08] <psha> Bonny: you'd better tatk rq->wValue.bytes[0] then
[18:25:24] <psha> if you need first byte of wValue
[18:25:42] <psha> at least you'll see what's happening instead of misterious constant '3'
[18:26:25] <andypugh> I suppose you can pick out any byte you want with indexing, but if anyone ever changes the typedefs it will break in a fun way.
[18:26:29] <psha> Bonny: i suspect that in avr 'unsigned' designates 16 bit value?
[18:26:56] <Bonny> I think 8 if not declared 16
[18:27:03] <psha> andypugh: it's data layout of USB control message :) it won't change :)
[18:27:57] <psha> Bonny: i've looked around in that code and it's definitely 16bit
[18:28:08] <psha> #if USB_CFG_LONG_TRANSFERS /* if more than 254 bytes transfer size required */
[18:28:11] <psha> # define usbMsgLen_t unsigned
[18:28:41] <psha> so usbWord_t is a 16bit structure with access either as single value or as pair of 8bit values
[18:28:50] <Bonny> That's worked and endians is ok usbMsgPtr[0]=rq->wValue.bytes[0];
[18:29:30] <psha> then avr is little endian arch
[18:30:03] <Bonny> seems... I'm just happy as now work
[18:32:13] <psha> just curious - what are you doing now with avr and usb?:)
[18:32:24] <Bonny> psha do you do something with VUSB?
[18:32:48] <Bonny> IR receiver...
[18:33:29] <Bonny> I spend 2 weeks to lye to windows to pretend computer to be IrDA printer.
[18:34:16] <Bonny> and find that's not possible. in Win the IrLPT is hardwired to be out of computer and can't be in.
[18:35:05] <Bonny> As I have that already done in old computer with IrDA on COM port just need to make new interface..
[18:43:47] <psha> heh, great :)
[18:46:27] <Bonny> Just need that... And don't want to buy two dongled at 20Eur each,,..
[19:44:06] <Sansiba> hi @ll
[19:56:09] <atmega> hi
[19:56:09] <the_wench> hello atmega, you have a question?
[19:56:14] <atmega> no, bitch.
[19:58:13] <andypugh> :-)
[19:58:16] <dimas_> :)
[19:59:09] <andypugh> Sansiba: the-wench is a logging bot, if you were wondering
[20:01:18] <Sansiba> andypugh: good to know^^
[20:02:43] <elmo40> hi
[20:02:44] <the_wench> hello elmo40, you have a question?
[20:02:46] <elmo40> help
[20:02:50] <elmo40> log
[20:02:56] <elmo40> !log
[20:03:55] <elmo40> anyways... I used Bodhi Linux and installed EMC onto it. runs SO MUCH faster then the default 10.04 .iso image available.
[20:04:09] <elmo40> gnome is a resource hit.
[20:17:42] <i_tarzan> Bodhi is debian or rpm?
[20:20:14] <psha> elmo40: replace debian with dwm and you'll get same speed boost on ubuntu :)
[20:20:19] <elmo40> ubuntu 10.04 based
[20:21:04] <psha> oops s/debian/gnome/
[20:21:05] <psha> :)
[20:23:08] <elmo40> psha: this is why I chose Bodhi. It uses e17
[20:26:03] <mhaberler__> mhaberler__ is now known as mhaberler_
[20:33:14] <elmo40> http://www.resplendence.com/latencymon
[20:33:28] <elmo40> for all you EMC/Mach3 users...
[21:12:40] <andypugh> 1000uS? Ewww!
[21:14:19] <L84Supper> that is what USB timing would be like.... if we supported it
[21:15:03] <psha> andypugh: you are not surprised with 500ms pagefault time? :)
[21:15:41] <psha> i thought windows has something like mlock
[21:16:30] <andypugh> There was an RTAI USB extension. Somebody keen and talented should look at it. psha?
[21:17:24] <psha> it's dangerious to allow me touching kernel )
[21:28:33] <elmo40> didn't that USB extension die long ago?
[21:29:43] <psha> there are surprisingly many usb rt implementations
[21:29:48] <psha> i've found at least 3
[21:30:05] <psha> usb20rt, usb4rt and rtusb :)
[21:30:11] <elmo40> must be something else I was thinking about.
[21:30:36] <psha> but i think neigher of them is alive
[21:30:44] <elmo40> :P
[21:30:47] <psha> usb4rt was last touched in 2008
[21:30:59] <psha> usb20rt in 2006
[21:31:12] <psha> rtusb was born dead
[21:32:23] <andypugh> Indeed, A set of dead projects.
[21:33:16] <andypugh> The point is that they were largely implementations looking for an application. EMC2 is an application looking for an implementation.
[21:33:43] <psha> andypugh: do _you_ have application that needs rtusb? :)
[21:35:57] <andypugh> Need, no. But There is a real gap in the EMC2 interface capability where analogue inputs (including digital, numerical values, such as absolute encoders) are concerned.
[21:38:10] <psha> unfortunately i've no real hardware so even if i'll try to do something on this front it's impossible to test it :(
[21:39:06] <dimas_> someone should gift psha some hardware
[21:45:09] <psha> andypugh: you a crafty provocateur!
[21:46:28] <andypugh> Yeah, I was thinking of writing an SPI comp this evening, but again, no hardware is a bit of an issue, and I have enough to be going on with. (Though in a bit of a hiatus waiting for a Mesa firmware)
[21:47:09] <psha> andypugh: i first have to finish actions in gladevcp :)
[21:47:51] <andypugh> Look at it this way, now you can finish Gladevcp without worrying about what to do next :-)
[21:49:16] <psha> andypugh: after i'll finish gladevcp i'll leave my current job and go to google :-P
[21:50:34] <psha> so let's hope i'll finish gladevcp as late as possible :)
[21:55:14] <elmo40> psha: do you want an absolute encoder to test with?
[21:57:43] <Sansiba> I have a question about linear scales and encoder on an servo. When I use a linear scale than I nead a encoder on the motor too? The next question is, what hardware i need that I can run the previously described system?
[22:04:33] <awallin> Sansiba: the mesa pci cards are popular (for reading encoders and driving servo amps)
[22:05:40] <awallin> I'm not sure if the linear scales are usually used in closed-loop while machining, but at least they could be used for measuring screw-comp, and for causing an e-stop if something goes wrong
[22:06:18] <psha> elmo40: not now :) if worm placed in my mind by insidious andy will eat at least half of my brains - then probably i'll need some hw :)
[22:10:11] <Sansiba> in commercial controllers like Siemens I know it so there is an inverter of the Encode-signal receiving and processing the control signal of the linear scale
[22:11:34] <awallin> not sure what you mean by inverter?
[22:11:36] <Sansiba> is this with the mesa pci cards possible?
[22:11:56] <awallin> you can receive two encoder signals, sure. but what do you want to do with them?
[22:12:16] <Sansiba> mom
[22:13:12] <Sansiba> awallin: not two encoders one encoder that i know the motor position and one linear scale to have the exact position
[22:14:36] <awallin> is the linear scale output an absolute encoder?
[22:14:42] <Sansiba> awallin: http://www.cnc-ready.at/components/com_virtuemart/shop_image/product/1.5kW_Digital_AC_4bebe17674043.png hear the motor and the inverter i think
[22:15:09] <awallin> ok, does that inverter also take linear-scale input ?
[22:15:17] <Sansiba> no
[22:15:22] <Sansiba> only the motor encoder
[22:15:27] <awallin> right.
[22:16:48] <Sansiba> i thin a servo is dont work without an encoder, it don't know it own position
[22:16:55] <awallin> you could probably drive that inverter with step/dir from emc2. and take the linear-scale reading into emc2.
[22:17:10] <Sansiba> ah
[22:17:18] <Sansiba> thats the way
[22:17:18] <L84Supper> open loop servo... is that like hot ice?
[22:18:27] <awallin> but I don't know if you can use the linear-scale reading in a pid loop, or only for screw-comp after a calibration run (+e-stop if following-error gets large)
[22:18:45] <Sansiba> i think this guy use this way: guy
[22:18:56] <Sansiba> http://www.cnc-ready.at/index.php?option=com_kunena&Itemid=0&func=view&catid=6&id=3&lang=de
[22:19:30] <L84Supper> awallin: doesn't it depend on how much slop there is in the system beteen the motor shaft and whatever the linear encoder is mounted on
[22:19:52] <L84Supper> plus the amount of play could be non-linear
[22:20:16] <awallin> L84Supper: yes, a sloppy coupling would make for a pretty wild and oscillating pid-loop. not good.
[22:21:44] <awallin> I could imagine using the linear-scale as the primary encoder for the servo-motor. only if things get "wild" would you fall back on the rotary encoder. but the rotary encoder is probably needed for commutation if it's an ac motor.
[22:21:56] <L84Supper> Sansiba: everything you've mentioned for you systems sounds like it could be controlled by EMC, it will come down to some tuning and how well you design your mechanisms to get your tight tolerance
[22:22:35] <awallin> Sansiba: if you can find out what, if anything, those fancy seamens/fanuc controllers do with both the rotary encoder signal and the linear-scale signal, then it's possible to discuss if it's doable in emc2
[22:22:48] <awallin> siemens :)
[22:24:31] <L84Supper> +/- 1/2 the diameter of the atoms that your device is made of will call for some temperature control and regular calibration just due to metal fatigue
[22:24:38] <andypugh> I think you could possibly use an I-term linked to the linear scales to correct the servo loop.
[22:25:35] <andypugh> But then you have to question if you trust the ballscrews or the linear encoders more? There is no inherent reason to consider the scales superior.
[22:26:14] <andypugh> Arguably if the ballscrews are the same material as the machine and the workpiece you have automatic temperature compensation.
[22:27:44] <awallin> 3-5um is what the 20-50keur/dollar machines usually quote as repeatability
[22:27:51] <L84Supper> yeah, if you can build everything from the same alloy so the CTE is the same, but good luck
[22:29:36] <Sansiba> i will trust the linear scale
[22:30:29] <Sansiba> the ballscrew Manufacturing accuracy is limited
[22:31:27] <Sansiba> a big heap of trouble...
[22:34:00] <awallin> so use linear motors ?
[22:34:42] <Sansiba> I see problems with the waste heat
[22:34:49] <cradek> can you find a linear scale with the same CTE as your work material?
[22:35:30] <Sansiba> not in the high resolution cradek
[22:36:36] <andypugh> Aye, the ballscrew is going to be closer in CTE than the linear scale.
[22:37:09] <L84Supper> do away with the linear scale and use a laser interferometer
[22:37:21] <andypugh> Linear scale manufacturing accuracy is limited too (and consider how they make the scales, and what with..)
[22:38:20] <andypugh> Interferometry is the way to go, and Polytec have a neat twist on that using polarisation to get the resolution sub-wavelength,
[22:39:35] <Sansiba> L84Supper: the interferometer is to expensive and have problems with the Atmospheric disturbances
[22:39:45] <awallin> L84Supper: in high-end laser experiments you then either fill the optics-box with helium, or if you crazy enough remove the air (vacuum), so the laser beam isn't affected by the varying refractive index of air-currents etc...
[22:41:36] <L84Supper> 50nm pore in anodized aluminum, then fabricate a pin-diode at the edge, use DNA or other long chain molecules and pull them through the pore and count each molecule.... simple
[22:41:57] <L84Supper> bbl
[22:42:37] <Sansiba> andypugh: Polytec = http://www.polytec-group.com ?
[22:43:09] <awallin> http://polytec.de/eu/
[22:43:19] <andypugh> This makes ineresting reading, especially the cam-comepnsation: http://www.lathes.co.uk/sip/
[22:43:29] <Sansiba> oh cool at home ^^
[22:43:45] <andypugh> And awallin has the right link
[22:43:57] <HomeBrew> hey
[22:44:30] <andypugh> Sandiba I also have a very good friend at Mahr in Gottingen who might have useful info.
[22:44:59] <HomeBrew> I got some old printer step motors (bipolar) and was wondering if anyone knew of a good schematic for using them with my parallel port for linux cnc?
[22:46:20] <andypugh> The Polytec equipment is very impressive, though I did spend about a month of my life decoding their binary file format to decide that, actually, their software was written by an imbecile and they really were discarding the imaginary part of their fourier transform. I think they have learned that lesson now.
[22:46:52] <andypugh> HomeBrew: What voltage/power, what is your budget, and how well do you want them to work?
[22:47:23] <HomeBrew> 24V - want them to work fairly well.
[22:47:42] <HomeBrew> going for cheap
[22:48:01] <awallin> this page has both buy and diy stepper drives http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?EMC2_Supported_Hardware
[22:48:23] <HomeBrew> yeah I am looking for diy
[22:49:00] <HomeBrew> is there a h-bridge schematic of this anywhere that you know about?
[22:49:54] <andypugh> Any idea of current?
[22:50:13] <HomeBrew> will check. 1 sec
[22:50:35] <HomeBrew> 550ma
[22:50:40] <andypugh> Steppers don't, actually, need an H bridge. More important is the step sequence
[22:50:42] <awallin> the L297/8 based simple ones are... simple
[22:51:10] <andypugh> Ah! In that case you can actually use the absolute minimal config, EMC2 and a ULN2003.
[22:51:39] <awallin> that might require 4 pins per motor?
[22:51:57] <andypugh> http://www.mycncuk.com/forums/showthread.php/2039-Unipolar-stepper-motor-with-ULN2003
[22:52:58] <andypugh> Aye, it's 4 pins per motor and demands unipolar motors. I didn't say it was a _good_ solution. In fact it's a rubbish one, but costs 50p.
[22:53:06] <HomeBrew> the uln2003 will not work with bipolar I thought?
[22:53:21] <HomeBrew> yeah,.
[22:53:54] <awallin> http://pminmo.com/PMinMOwiki/index.php5?title=L297-8
[22:55:34] <HomeBrew> interesting link
[22:55:36] <dimas_> L297 is what I was suggested and it worked
[22:56:40] <dimas_> i bought bare pcb's from pminmo
[22:56:49] <HomeBrew> ok, will do some research on that. Is there a schematic that you know of for hooking it up to the motor?
[22:57:27] <HomeBrew> like are there a lot more components or can the chip be used as is?
[22:58:14] <awallin> you just find out with a multimeter where the motor coils are and hook them up like in the schematic...
[22:58:17] <HomeBrew> the schematic posted by awallin looks like a lot of parts. But it may be the way I put it togather
[22:58:24] <dimas_> stepper motor drivers and BOB actually
[22:59:14] <HomeBrew> ok, cool, thank you so much for all your help guys. :)
[22:59:32] <HomeBrew> going to go read the web site and get some parts. :)
[22:59:46] <Sansiba> It's time for me to go to bed
[22:59:50] <HomeBrew> night
[23:00:12] <Sansiba> Thanks for your help
[23:00:26] <psha> hey, it's really night!
[23:00:44] <Sansiba> n8@ll
[23:01:11] <psha> i'm next in order
[23:01:17] <mhaberler__> mhaberler__ is now known as mhaberler
[23:01:19] <psha> * psha preparing for sleep
[23:03:08] <andypugh> Ah, I got distracted. I am very much of the opinion that you will struggle to beat the eBay prices for the TA6250 (or whatever the actual code is) prices with a home-build.
[23:04:10] <LawrenceG> if you like pics, have a look at http://www.romanblack.com/lini.htm
[23:04:38] <psha> LawrenceG: you are talking as a spamer :D
[23:05:19] <LawrenceG> wrong type of pics!.... and I hang out on #cam as well... I should know better!
[23:06:12] <psha> maybe spammers are sending pics of drive controllers too! who knows :)
[23:12:00] <psha> good nigth europe, and good day to others :)
[23:13:31] <andypugh> Linear seems a bit old-school for CNC, but I can see the attraction for telescope positioning.