#emc | Logs for 2009-10-26

Back
[01:14:46] <jepler> jepler has changed the topic to: Welcome! EMC (Enhanced Machine Controller) is a linux-based opensource CNC control. | Latest release: EMC 2.3.4 | http://www.linuxcnc.org | http://wiki.linuxcnc.org
[01:16:39] <Neo_The_User> jepler: where is the source to 2.3.4? source forge shows 2.3.3
[01:18:05] <jepler> Source tarballs are available from the
[01:18:05] <jepler> package repository:
[01:18:05] <jepler> http://linuxcnc.org/hardy/dists/hardy/emc2.3/source
[01:18:16] <jepler> our guy who updates sourceforge is in europe, so it will probably be updated tomorrow.
[01:18:27] <Neo_The_User> thanks!
[01:19:08] <MarkusBec> MarkusBec is now known as MarkusBec_away
[01:33:30] <Neo_The_User> do the patches in base/x86 add the RTAI layer or just in base/i386?
[01:37:40] <Neo_The_User> my C skills are terrible and i cant tell
[02:05:34] <a-l-p-h-a> how goes it everyone?
[02:06:12] <Neo_The_User> I'm good.
[02:06:39] <a-l-p-h-a> good
[02:11:02] <Neo_The_User> i think i just got rtai on arch. be right back
[02:14:10] <john_f_> Question: When the mode is MDI or Auto I want EMC to start and stop the spindel according to the M3 or M5 words. For some reason I can't find the output pins to do this.
[02:24:38] <john_f_> never mind I just figured it out. If you dont tell the controller a spindle speed it just leave it zero and it never turns on!
[02:26:09] <Neo_The_User> does anybody know if 3.2.2 of http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Debian_Lenny_Compile_RTAI are the only requirements for the kernel config?
[03:25:29] <MattyMatt> how do you get google to convert gram-centimetres to Nm?
[03:25:41] <cradek> MattyMatt: use units
[03:26:10] <cradek> well those aren't compatible...
[03:26:41] <cradek> ah you need to add a factor of gravity
[03:26:44] <cradek> You have: 1 gram cm gravity
[03:26:45] <cradek> You want: newton meter
[03:26:53] <cradek> * 9.80665e-05
[03:27:03] <klick0> /join #cam
[03:27:13] <klick0> oops
[03:27:34] <MattyMatt> so "gram-force" :)
[03:27:56] <cradek> I've not used google for that kind of thing - I always use units
[03:28:52] <MattyMatt> google is very good, when you've learned how it wants the units
[03:29:23] <MattyMatt> so actually it might well be "gf" for gram force
[03:30:00] <cradek> if it gives you the answer 9.8e-05 you know you're asking right...
[03:32:19] <MattyMatt> so 700 g-cm is 700 * 9.8e-5? so around 0.07 Nm that seems too low for a 0.5A motor
[03:33:07] <cradek> I don't have a gut feel for it, but I agree that's what the numbers say
[03:33:17] <cradek> 0.5A at what voltage?
[03:35:02] <MattyMatt> 5
[03:35:11] <cradek> that's a tiny tiny motor
[03:35:27] <MattyMatt> yep
[03:35:55] <cradek> converting to local units 700 g cm = 0.05 ft lb = can barely feel it
[03:36:06] <MattyMatt> just comparing what's on ebay
[03:36:11] <tom3p> gram force is almost an oxymoron to begin with
[03:36:15] <cradek> what's your application?
[03:36:26] <MattyMatt> 2ft wooden table
[03:36:49] <cradek> heh, your table would only point and laugh at that little motor
[03:37:02] <MattyMatt> yeah I got better from printers
[03:38:47] <MattyMatt> I find Nm most intuitive. I can imagine applying force through a lever without thoughts of bleeding fingers like oz-in gives me
[03:39:27] <cradek> I often use a torque wrench marked in ft-lb (only one you can get here) so I think of that
[03:40:04] <cradek> huh 1 ft lb (gravity) = 1.3 Nm
[03:40:08] <cradek> they are very similar
[03:41:14] <MattyMatt> there's an actual unit of torque isn't there? the dyne iirc
[03:41:43] <cradek> dyne = cm gram / s^2
[03:42:24] <MattyMatt> that's rotational inertia?
[03:42:52] <cradek> dyne = 1e-5 newton
[03:42:57] <MattyMatt> ah. gotta brush up on O level physics :)
[03:44:05] <L84Supper> one of my favs http://en.wikipedia.org/wiki/Furlongs_per_fortnight#Furlong_per_fortnight :)
[03:46:49] <MattyMatt> I'll have my table sliding today, so I can pull it back & fro with a spring balance to see how much basic torque I need to get moving
[03:47:33] <cradek> I bet it would be better to wrap a string around the screw and use a weight to turn it
[03:47:35] <MattyMatt> I hope it's less than 50g, because that's all my spring balance will do :)
[03:47:46] <L84Supper> MattyMatt : what are you using for bearings?
[03:47:47] <cradek> or your spring I guess
[03:48:40] <MattyMatt> not decided on the bearings. I got a set of roller skate ones but plan A was to ruin a bike wheel
[03:49:27] <MattyMatt> as I'm experimenting I might try both
[03:50:06] <MattyMatt> string+weight on the screw is good for that sort of comparison
[03:51:20] <L84Supper> MattyMatt : ball screw or lead screw?
[03:51:29] <MattyMatt> allscrew
[03:51:33] <MattyMatt> M8
[03:51:43] <cradek> bronze nut?
[03:52:08] <MattyMatt> gonna try plastic & brass
[03:52:44] <cradek> try to make it easy to replace them...
[03:52:51] <MattyMatt> indeed
[03:53:26] <MattyMatt> brass would be a split one, for tightening as it wears
[03:53:57] <cradek> acme rod would work a lot better and is not much more expensive
[03:54:06] <MattyMatt> plastic will be disposable. I'll have a slot for that to float in
[03:54:06] <cradek> delrin or bronze nuts
[03:54:43] <MattyMatt> I'll upgrade when it's working
[03:55:04] <Valen> 0.07NM isn't *that* teensy
[03:55:20] <Valen> a V8 jeep only has 400NM
[03:56:01] <MattyMatt> most of the cost so far has been the shelf sliders, but they were way cheaper than any polished bar
[03:56:15] <cradek> so you can move a car with 57,000 of those motors?
[03:56:42] <tom3p> disused filing cabinets? cheap sources for drawer sliders?
[03:56:44] <Valen> something like that ;->
[03:57:09] <Valen> I am thinking of making a cheapy
[03:57:22] <Valen> oh anybody got a source on cheap rotary encoders, for a motor style
[03:57:59] <cradek> us digital is as cheap as they get, as far as I know
[03:58:05] <MattyMatt> I've found 3 suppliers of equivalent sliders now. Screwfix are the cheapest, unless you order 10 pairs from ironmongerydirect.com
[03:58:44] <MattyMatt> I'm not using old ones, because these sliders go floppy after years of abuse
[03:59:55] <MattyMatt> squashing them in a workmate would probably revive them tho
[04:01:29] <MattyMatt> Valen, I saw a circuit which runs a tiny stepper in reverse as an encoder
[04:02:02] <cradek> those circuits lose position when going slow
[04:02:07] <cradek> no good for motion control
[04:02:45] <cradek> slightly frustrating but maybe adequate as a jog wheel
[04:03:03] <L84Supper> MattyMatt : do you have source of old inkjet printers? rotary encoders and linear optical inside many
[04:04:00] <MattyMatt> I'm only finding simple steppers in the ones I've butchered so far
[04:04:34] <MattyMatt> I've got a few old mice if I need rotary encoders
[04:05:34] <MattyMatt> I will mount a webcam over the vernier on my table, and try and teach the computer to read it
[04:06:53] <L84Supper> http://cgi.ebay.com/ENCAD-ENCODER-STRIP-MYLAR-NOVAJET-PRINTER-PARTS_W0QQitemZ290357589289QQcmdZViewItemQQptZLH_DefaultDomain_0?hash=item439aa93529
[04:08:06] <MattyMatt> and then how much for the sensor?
[04:08:27] <L84Supper> few $, led with photodiode
[04:09:01] <MattyMatt> how many rows?
[04:09:10] <L84Supper> how long do you need the strip to be?
[04:09:20] <MattyMatt> afaics, you'd need one for each bit in the Gray code
[04:10:15] <MattyMatt> or, use a cheap webcam to look at that strip :)
[04:10:20] <L84Supper> printers usually just count pulses and have limit swithces
[04:11:15] <L84Supper> direction is generally assumed by how the motor is powered
[04:11:18] <MattyMatt> that's all a cnc really needs too. absolute position sensing from power up is just a luxury
[04:12:58] <L84Supper> heh $10 http://cgi.ebay.com/ENCAD-ENCODER-SENSOR-NOVAJET-PRINTER-PARTS_W0QQitemZ300355229863QQcmdZViewItemQQptZLH_DefaultDomain_0?hash=item45ee9118a7
[04:14:00] <L84Supper> and they are a match
[04:14:23] <MattyMatt> neat
[04:14:52] <MattyMatt> although I have to say, if I had a 60" printer carrage I wouldn't be building a 2ft wood router
[04:15:43] <L84Supper> cut the strip to size
[04:16:06] <L84Supper> you'll have 2 spares as well
[04:16:16] <MattyMatt> i'd rather cut my wood to match that sensor, if I was using it :)
[04:16:51] <MattyMatt> go into the signwriting business
[04:17:21] <MattyMatt> which is one plan anyway, but on a smaller scale
[04:18:07] <Neo_The_User> I can very easily write a super in-depth HOWTO on getting emc2 working with RTAI in arch linux but building a single unpackable package woukd be quite difficult. might be a few days longer. i never use the package manager in archlinux so I'd need some time to read up as far as the syntax commands go.
[04:18:38] <Neo_The_User> would it be ok if i started working on a page in the wiki?
[04:21:26] <MattyMatt> that sounds like "no objections" to me
[04:22:49] <L84Supper> he'll write it up and someone with wiki access can add it
[04:34:14] <MattyMatt> on a setup like this http://cgi.ebay.co.uk/5-AXIS-CNC-DRIVER-BOARD-4-STEPPER-MOTOR-ROUTER-MILL-3-3_W0QQitemZ290361249854, is there anything like a standard connection layout for the driver to amp boards?
[05:09:58] <Neo_The_User> I am having a problem compiling emc. http://pastebin.ca/1643419
[05:12:42] <MattyMatt> I thought relative jumps were -128 to +126 on x86
[05:15:08] <MattyMatt> the error looks like a missing header
[05:16:53] <Neo_The_User> i DID compile rtai with --disable-leds because i kept getting some other error
[05:35:28] <MattyMatt> I just looked for rtai package on debian server. it only had rtai-doc
[05:36:16] <MattyMatt> better than nowt :) this is my docs-reading machine
[05:37:12] <Neo_The_User> I'm using rtai 3.7.1 directly from rtia.org
[05:37:16] <Neo_The_User> *rtai.org
[08:23:01] <MarkusBec_away> MarkusBec_away is now known as MarkusBec
[08:32:49] <pjm__> good morning
[08:41:24] <Valen> zup
[09:24:19] <MarkusBec> MarkusBec is now known as MarkusBec_away
[09:35:12] <MattyMatt> I can cut a straighter line with a 2 quid saw than with my table saw
[09:36:02] <MattyMatt> I just had to revise my table width downwards :)
[09:36:18] <MattyMatt> it'll be even stiffer now
[09:37:09] <archivist> stiffness really matters
[09:38:04] <MattyMatt> the box under the table will help a lot.
[09:38:33] <MattyMatt> it'll be around half the width of the table, and the full length
[09:40:04] <MattyMatt> maybe with a bottom on the box, to catch drips from the nut oiler :)
[09:41:00] <MattyMatt> maybe not, unless it's linseed oil
[09:44:22] <MattyMatt> clearances and mountings for 3 screws, as there's room and this is experimental
[09:45:43] <MattyMatt> I can compare them directly, and have 1, 2 or 3 symetrically
[09:47:21] <MattyMatt> or even have 2 screws with oil, one with grit and grind my own screws
[09:51:25] <MattyMatt> 2:1 gears and I can make 2.5mm acme from 1.25mm M8, then 5mm ballscrew from 2.5mm acme
[09:52:41] <MattyMatt> wooden screws :) right, where's that beech tree
[09:55:46] <MarkusBec_away> MarkusBec_away is now known as MarkusBec
[10:04:18] <MattyMatt> momma, I'm building you a barbeque on yo patio. you work the bellows and I'll er, sterilise this kebab spit
[10:05:44] <MattyMatt> machine, harden, temper, grind. that the best way of making a leadscrew from stock mild steel? is there a better common material?
[10:06:31] <archivist> buy a ballscrew
[10:07:12] <MattyMatt> I'm too cheap :) I do have a nice second hand one but 14.5" travel is disappointing
[10:07:39] <MattyMatt> I think I'll save that for the mk2 aluminium miller
[10:09:42] <MattyMatt> I'd use it for the Z, but I think I'd rather have acme on that so it doesn't droop if the motor loses power
[10:12:17] <MattyMatt> I'll see how it goes. plan A is simple M8 for everything
[10:13:20] <MattyMatt> large pitch screw demands high torque motors
[10:33:18] <The_Ball> I like how they made the router fold down, i'll make mine that way for sure when I get around to it; http://www.solsylva.com/cnc/25x25.shtml you always run out of space
[10:33:38] <The_Ball> but not out of wood, hehe
[10:43:42] <MattyMatt> yeah I was thinking of bolting the legs off a workmate on mine.
[10:44:39] <MattyMatt> they are £10 and I could use the vice for something else
[10:45:40] <MattyMatt> like my z motion. 2 acme screws about 10" long
[10:47:06] <MattyMatt> mine's going into a heavy pedestal at first, to discourage moving it and to stop the whole machine twisting
[11:35:52] <MrSunshine> http://www.brundin.biz/images/datasheets/4-axisdriverboard.htm <--- hmm, running that board (1.8A max) with 2A stepper motors will that work ? .. as there is no settings as it seems for max power? :/
[11:42:11] <MattyMatt> if you adjust the voltage so the steppers draw less than 1.8A
[11:42:38] <MattyMatt> it won't use the steppers to the full tho
[11:43:23] <MattyMatt> resistors in series will also reduce the current flowing, but at the loss of more power
[11:43:50] <aystarik> MattyMatt: please, don't spread FUD.
[11:44:23] <MattyMatt> FUD or just plain wrong? :)
[11:44:38] <aystarik> MrSunshine: yes, it will work. You will not have rated torque of the motor, but 1.8/2.0 of it.
[11:45:06] <aystarik> plain wrong might be better description :)
[11:45:16] <MrSunshine> MattyMatt, what do you mean with adjust the voltage ?
[11:45:37] <aystarik> MrSunshine: just forget
[11:45:45] <MrSunshine> as its a chopping driver it should only put in 1.8A at all times max right? :)
[11:45:54] <aystarik> yes
[11:46:11] <MattyMatt> I was thinking that if you lowered the voltage slightly, the current flowing for a certain impedence would fall
[11:46:38] <MattyMatt> ah I didn't know chopping drivers were self limiting like that
[11:47:23] <MrSunshine> MattyMatt, as they read off the current and start chopping when they hit the maximum current allowed
[11:47:25] <aystarik> MattyMatt: still thinking of L/R drivers?
[11:47:31] <MrSunshine> then its % of that for the microsteps
[11:48:14] <MrSunshine> i find it strange how the TA8435 chip wants 24V but that driver says it can take 36V max ?
[11:48:33] <Jymm> BEMF
[11:48:37] <aystarik> MrSunshine: it will still put full 1.8A as a sum in both phases
[11:49:03] <MrSunshine> that be true =)
[11:49:56] <MrSunshine> 26.4 max voltage
[11:50:24] <MrSunshine> i find that very strange ... but according to the store i buy it at i can run it on a 36V switched power supply if i adjust it down the 10% its down adjustable
[11:51:38] <MrSunshine> welll i guess i should get a new one if it burns then as he states that its posible and the instruction page says 12 - 36V :P
[11:53:23] <aystarik> max rating for ta8435h is 40V, you might be dealing with this one.
[11:54:55] <MrSunshine> ahh might be so =)
[11:55:06] <MrSunshine> as higher voltage == higher speed at less cost of torque right? :)
[11:55:27] <MrSunshine> This product is shortly being replaced with 3-4 Axis 12-30v 2.5A Driver board <-- dammit :P
[11:55:29] <MrSunshine> oh well :)
[11:55:34] <MrSunshine> i guess il have to buy another later then :P
[11:55:45] <MrSunshine> can have that one for my plamsa cutter later .. always good to have more driver =)
[11:55:49] <aystarik> get G540...
[11:56:14] <MrSunshine> aystarik, cant afford it =)
[11:56:35] <aystarik> for plasma???
[11:56:52] <MrSunshine> aystarik, huh? :)
[11:58:36] <aystarik> you mentioned plasma cutter, IMHO it's not a cheap table thingy...
[11:58:56] <MrSunshine> aystarik, hehe ... well i build the table, my sisters boyfriend has the plasma =)
[11:59:16] <MrSunshine> atm im converting a mini mill =)
[11:59:53] <aystarik> just remember: crap in -- crap out...
[12:00:35] <MattyMatt> not always. the first machines were built with crap
[12:03:01] <MattyMatt> my friend with the garage says he wants a plasma table, but then he says he wants a garage too
[12:03:33] <aystarik> with 2A steppers mini-mill may become underpowered. Lost steps is the result
[12:03:54] <aystarik> re: garage -- it's typical :)
[12:04:06] <MattyMatt> even cutting wood? I'm looking at 1A motors for mine to start with
[12:06:30] <aystarik> even moving along... you have dove tales and brass/steel metric screw... These do have friction
[12:08:33] <MattyMatt> I have some plastic gears from the donor printers. hopefully I can at least move the table, but I suppose I need proper motors & drivers myself
[12:08:40] <aystarik> Opti has catalog, there they state 2.75N/m torque on their _ballscrew_ analog of mini-mill on XY, and 4.4N/m on Z. With 1A steppers you are not anywhere close..
[12:09:49] <MattyMatt> ballscrew is 8:1 gearing disadvantage, but only 3 times more efficient friction wise
[12:10:10] <MattyMatt> iirc 90% for ballscrew vs 30% for V thread
[12:10:14] <aystarik> Oh, they also claim 12mm/s on ACME screws with such a torque
[12:11:03] <aystarik> and 24mm/s on ballscrew.
[12:11:32] <MattyMatt> speed isn't a concern right now for me. when the time/money equation kicks in I'll start saving for upgrades
[12:11:47] <aystarik> right, so you need 2.75*3 torque :)
[12:13:11] <aystarik> 12mm/s is 0.5in/s or 30in/min -- it's not huge. You might want to have _cutting_ speed in this range.
[12:15:14] <aystarik> gearing disadvantage of ballscrew -- could you elaborate on that? There do you get 8:1 figure?
[12:15:32] <MattyMatt> eventually, but this cheap drem-alike won't move that fast through anything
[12:15:50] <MattyMatt> 10mm pitch ballscrew vs 1.25mm pitch M8
[12:16:51] <aystarik> 12mm ballscrew is 4mm pitch regularly... 2mm is possible.
[12:17:37] <aystarik> and then you have stepper torque falling with increased RPM. Do you account for that?
[12:17:42] <MattyMatt> the actuator I got on ebay is 10mm. to steep for my first machine. I'll save that for mk2
[12:18:33] <MattyMatt> I would expect to have to run at the best torque speed, and live with the slowness
[12:19:23] <MattyMatt> some of the things I want to do with it can be done largely unattended. PCBs etc
[12:19:30] <aystarik> best torque speed of the stepper is 0 :)
[12:20:49] <MattyMatt> hmm, didn't know that. I thought the best would be once rolling resistance is overcome
[12:21:05] <MattyMatt> and I guess for the whole system that's still true
[12:22:25] <aystarik> http://www.kelinginc.net/KL23H284-35-4BT.pdf -- here is the typical curve... Maxed at 0RPM.
[12:22:38] <MattyMatt> ah well. I won't lose by trying progressively bigger ones. I need small steppers for other robots
[12:25:03] <aystarik> right, if you not account for time and resources :)
[12:25:16] <MattyMatt> but I suppose I'll be running sooner if I blow 150 quid on a 2A set
[12:25:59] <MattyMatt> I generously value my time at minimum wage right now
[12:29:42] <MattyMatt> I expected I'd need 2 motors & screws for my 2ft table anyway, with a belt between them to stop the table twisting if steps are missed
[12:30:10] <archivist> there should be NO missed steps
[12:30:19] <MattyMatt> but I didn't even get 2 decent motors from 2 printers so far
[12:30:53] <MattyMatt> should be, but I don't want my screws to bend when I push it all too far
[12:31:58] <MattyMatt> I will test this machine to destruction, but not on the first day due to a dodgy motor wire kthx
[12:50:00] <MattyMatt> what about a 2A driver with a 1A motor? is that when you need a current limiting resistor?
[12:50:42] <archivist> yes or driver is adjustable
[12:52:17] <MattyMatt> or 2 motors in parallel :)
[12:52:32] <archivist> I dont think so
[12:53:14] <MattyMatt> series rather, for failsafe
[12:53:37] <archivist> still 1 amp in series!
[12:54:00] <archivist> just do it right TM
[12:54:03] <MattyMatt> oh yeah, so parallel. what's wrong with that?
[12:54:12] <SWPadnos> it won't work
[12:54:19] <SWPadnos> other than that, it's fine
[12:56:07] <MattyMatt> but one 2A channel is so much cheaper than 2x 1A and I'm a cheapskate :(
[13:01:20] <MattyMatt> I should have made a toy machine out of a scanner with plastic ways . that would let me learn the software but I came and talked to you people and got big ideas about a machine that actually does work
[13:03:15] <MattyMatt> carving wood is good tho. I can turn that ability into cash for mk2
[13:34:59] <MrSunshine> aystarik, a person i usaly talk with has his running on those steppers =)
[13:36:07] <aystarik> good. hope it works for you as well.
[13:52:31] <MrSunshine> if it loses step i will use toothed gears with toothed belts =)
[13:52:39] <MrSunshine> to get it to a ratio about 3.333333 :P
[14:01:53] <MattyMatt> nice. I just bought 4x 1A 5V motors for £12
[14:02:18] <MattyMatt> that's 2 axes sorted at least :)
[14:02:54] <MattyMatt> 0.4 Nm 60 oz in. if one doesn't move my table, 2 will
[14:03:54] <MrSunshine> 0.4nm ... that was low :)
[14:03:57] <MrSunshine> :P
[14:04:02] <celeron55_> make a small one enough
[14:06:40] <MattyMatt> if the dremel's too heavy, I'll put a pen on the Z and call it a plotter
[14:07:04] <MattyMatt> it can draw where I need to chisel
[14:10:44] <celeron55_> i also have a machine mainly for just being able to play with emc
[14:11:26] <celeron55_> and other things
[14:12:09] <MattyMatt> I want proper machine tools eventually of course, but CNC isn't the main cost of it all. e.g. the price of a lathe is much more than the retrofit
[14:13:26] <cradek> machines are very cheap right now
[14:14:01] <celeron55_> yeah, software and electronics cost nothing compared to the machine if you don't have any contacts to help getting things cheaply etc
[14:14:08] <MattyMatt> accurate measuring gear is out of my reach even
[14:14:47] <MattyMatt> and I'm living at momma's, and working in the back bedroom, so no metalwork shop just yet
[14:15:18] <cradek> ah, different ideas of what "cheap" is then
[14:15:28] <celeron55_> space is a problem of course, too. the lack of it 8)
[14:15:40] <celeron55_> y
[14:16:02] <MattyMatt> it's a big house, the machine oil is a bigger problem than the size
[14:16:04] <celeron55_> (oops)
[14:16:16] <MattyMatt> sawdust can be hoovered
[14:18:36] <celeron55_> the other problem is that i'm not really interested in machining, just the controlling of machines
[14:19:25] <MattyMatt> it's a means to an end, unless you can afford an Asimo
[14:19:33] <celeron55_> though i often would need some machined parts (though not anything that you need cnc for)
[14:23:40] <MattyMatt> the artistic possibilities of CNC appeal to me, and also for making prototypes of complex injection molded forms
[14:24:46] <MattyMatt> and a new head for my robosapien :)
[14:25:55] <MattyMatt> although an RC car would give it better mobility
[14:53:42] <pjm__> yay, just updated to 2.3.4 + new hostmot firmware, no problems so far
[15:20:45] <L84Supper> is the 8.04 package of 2.3.4 in the repo yet?
[15:23:31] <pjm__> L84Supper i think so, u can just use the bog-standard update util to get 234 and hostmot firmware
[15:37:41] <Neo_The_User> I just realized I had PCI_LEGACY turned off in the kernel (Enable depreciated pci_find_*API) do I need that on? Because the error: implicit declaration of function ‘pci_find_device’ looks related.
[15:48:19] <Neo_The_User> If anybody wants to look over my config, its at http://neo-technical.wikispaces.com/file/view/rtai.config It is incredibly stripped down to only support my exact hardware so please do not use it as is.
[16:14:51] <Guest893> hey all
[16:15:35] <Guest893> I just installed emc2 on a computer hooked up to a masterwood project 317 that has a CN12 controller on it and am having a little trouble getting it to work. Anyone familiar with this machine?
[16:16:24] <Guest893> I have wiring diagrams but I'm afraid I dont really know enough (it doesnt help that they are in italian) to figure out the pinout.
[16:17:05] <archivist> we have an occasional italian user in here
[16:17:06] <Guest893> emc2 is also talking about a "Parallel port" connection, but my machine uses serial, is that a problem?
[16:17:20] <archivist> serial is no good
[16:18:01] <Guest893> no good? :(
[16:18:23] <archivist> not fast enough, the whole control is in the pc
[16:21:47] <Guest893> is there some way I can map a virtual parallel port to the serial port?
[16:23:30] <archivist> does not make sense, the serial on your box will not understand the type of data emc puts out, better to attach to the stepper drivers IN the box with a new interface
[16:23:51] <archivist> interface/breakout board
[16:24:36] <cradek> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Emc2HardwareDesign
[16:25:32] <cradek> this describes in general the kinds of hardware that work with emc. if your router takes serial commands it is a totally different control scheme. it is not just a matter of wiring different kinds of ports together.
[16:25:54] <cradek> if you want to control it with emc, maybe you can remove everything except the motors and their drivers, and interface there
[16:26:51] <archivist> which is how I just did a lathe
[16:26:59] <Guest893> oh I see :(
[16:27:18] <Guest893> I just read the 3.3 on that link
[16:28:25] <Guest893> does anyone know of a project that is compatable with this sort of interface? This is an old CNC machine (a decade now!).
[16:29:11] <archivist> I would get inside the interface to see what is there
[16:31:54] <archivist> take a pic and put is up somewhere
[16:33:46] <Neo_The_User> After I turned on PCI_LEGACY I got, WARNING: "fabs" [/home/neo1993/emc2/src/steptest.ko] undefined! WARNING: "ceil" [/home/neo1993/emc2/src/motmod.ko] undefined! WARNING: "__fixunsdfdi" [/home/neo1993/emc2/src/hostmot2.ko] undefined! cp *.ko ../rtlib/
[16:38:12] <Guest893> what interface? do you mean the PLC card?
[16:50:00] <tom3p> Guest893: your system ends with moving stuff and begins with commanding it. the moving stuff is connected to the commanding stuff by some interface ( some in between stuff).
[16:50:07] <tom3p>  you could rip off the old commanding stuff, and use the new emc commanding stuff, but you need to build & write the new in between stuff.
[16:50:08] <tom3p>  to do this requires looking at what the moving stuff needs.
[16:51:49] <archivist> if you open the box and find three stepper drivers and controller, you need to cut the wires between and feed there from the pc
[17:21:09] <Guest893> oic... eek
[17:34:44] <MrSunshine> how did emc start out ? ... just some simple app for running steppers or something? :)
[17:35:26] <skunkworks_> it started at nist.org (goverment project)
[17:35:50] <MrSunshine> humm ok ? :)
[17:36:26] <skunkworks_> http://www.linuxcnc.org/content/view/6/5/lang,en/
[17:38:06] <skunkworks_> so watch it - The man can take control of your machine at any moment.
[17:39:39] <MrSunshine> :P
[17:39:57] <MrSunshine> im from sweden .. im not very afraid of stuff like that .. its just paranoid americans that is like that :P
[17:40:55] <skunkworks_> * just some paranoid americans
[17:41:10] <MrSunshine> ok =)
[17:42:07] <skunkworks_> JT-Work: how is the THC going?
[17:43:17] <JT-Work> didn't get much chance to work on it this weekend but I have the 5i20 computer hooked up and that part is running good
[17:43:26] <skunkworks_> cool
[17:44:31] <JT-Work> I have all the parts and materials laid out to connect the tip voltage signal to the THC card
[17:45:18] <JT-Work> just have to figure out where to mount the Mesa THC card in the computer or the power supply/drives box
[17:55:02] <Jymm> MrSunshine: Just because we're paranoid, Does NOT mean the governmeant isn't out to get us.
[18:02:33] <tom3p> dont worry, there realy is a 'them', but they know whats best. muhahaha
[18:03:06] <Jymm> what's best for THEM that is, not US.
[18:25:04] <tom3p> the cli version of recordmydesktop can record emc ( the gui version hangs and blows up ).
[18:25:10] <tom3p> the cmd line can be intimidating and the output is ogg. but ogg can be converted.
[18:25:25] <tom3p>  recordmydesktop -fps 25 -windowid 0x4600017 -x 5 -y 49 -width 600 -height 500 --no-sound --overwrite -o emc01.ogg
[18:25:45] <tom3p> theres no sound so far, but the vid at 25fps is nice
[18:26:23] <tom3p> captures cursor
[18:27:04] <tom3p> use xwininfo to get the windo id
[18:29:57] <tom3p> ffmpeg -i mybrotherdarrel.ogg myotherbrotherdarrel.mpg will convert it
[18:32:32] <tom3p> ffmpeg -i blah.ogg blahblah.flv will convert it for youtube
[18:44:18] <MrSunshine> Jymm, hehe =)
[18:49:30] <mikeg> is it better to operate servo amplifiers in torque mode or velocity?
[18:51:20] <cradek> I think velocity is better but they become more equivalent if you have very high resolution encoders
[18:51:36] <cradek> if your amps want tachs, you have no choice of course
[18:52:10] <mikeg> my encoders are pretty high res..
[18:52:41] <mikeg> the amp can operate in either torque or velocity without feed back.
[18:52:57] <cradek> if your amps can't use tachs, you have no choice of course (heh)
[18:53:05] <mikeg> I was thinking about sending the encoder feedback to the PC and the servo amp
[18:53:24] <cradek> ac motors?
[18:53:29] <mikeg> BLDC
[18:53:40] <cradek> I am not familiar with no-feedback velocity
[18:53:47] <cradek> why not try both?
[18:53:55] <mikeg> i suppose I should
[18:54:04] <mikeg> they are in torque mode right now.
[18:54:17] <cradek> does it tune up nice?
[18:54:28] <mikeg> ...not really. that's the problem
[18:54:45] <mikeg> if I give it any D gain at all they get unstable
[18:54:57] <cradek> yep that is touchy in torque mode
[18:55:04] <cradek> super fine encoder feedback helps
[18:55:10] <cradek> what is your encoder scale?
[18:55:10] <mikeg> the tune isn't terrible. I mean it works, just think it could be better
[18:55:31] <mikeg> 400,000 counts/inch
[18:55:40] <SWPadnos> !
[18:55:44] <mikeg> heh
[18:55:44] <cradek> huh, seems like that should be plenty
[18:56:20] <SWPadnos> what's the resolution of your DAC output? (assuming you
[18:56:29] <SWPadnos> 're using analog control amps)
[18:57:13] <mikeg> DAC output?
[18:57:27] <mikeg> +/- 10 Volts
[18:57:37] <SWPadnos> ok, what hardware drives the amp?
[18:57:37] <mikeg> 5i20
[18:57:39] <SWPadnos> ok
[18:57:40] <Jymm> iPod
[18:57:40] <mikeg> 7i33
[18:57:52] <mikeg> in PDM mode
[18:58:10] <mikeg> PDM frequency is set somewhere in the high end of the band
[18:58:15] <SWPadnos> ok
[18:58:34] <cradek> setp hm2_5i20.1.pwmgen.pdm_frequency 6000000
[18:58:39] <cradek> setp hm2_5i20.1.pwmgen.00.output-type 3
[18:59:18] <mikeg> yeah, but I think i've got it set to more like 90 mhz
[18:59:28] <mikeg> hang on, lemme switch computers...
[18:59:41] <cradek> I think this is what pcw told me to use (a long time ago)
[19:00:29] <mikegg> back
[19:01:02] <mikegg> yeah, 90 Mhz
[19:02:06] <cradek> see page 7 of the 7i33 manual
[19:02:35] <cradek> (spoiler: it says to use 6MHz and >10MHz is right out)
[19:03:12] <mikegg> hmm
[19:03:17] <mikegg> indeed it does
[19:04:17] <MattyMatt> is this stuff commercial yet? http://www.newscientist.com/article/dn16102-material-slicker-than-teflon-discovered-by-accident.html
[19:04:43] <mikegg> http://www.linuxcnc.org/docs/devel/html/man/man9/hostmot2.9.html
[19:05:29] <mikegg> under the PWM generator section. "the 5i20 and 7i43 both have a 100 MHz clock, resulting in a 100 Mhz max PDM frequency"
[19:06:31] <cradek> sure but that does not mean the 7i33 likes it that fast
[19:07:04] <cradek> or am I missing the point about something here?
[19:07:23] <mikegg> well, i doubt it
[19:07:47] <MattyMatt> would you get 9 pulses from 10 @100Mhz if you ask for 90Mhz from that?
[19:08:31] <cradek> if you ask for an output 0.9, you get 9/10 pulses with each one being 1/100Mhz long
[19:08:54] <cradek> if you ask for 0.5, you get full 100Mhz on/off
[19:08:59] <cradek> the 7i33 probably does NOT like that
[19:09:44] <mikegg> maybe that's why they won't tune well
[19:09:52] <cradek> I think that's quite likely
[19:09:53] <mikegg> ..
[19:10:01] <cradek> it might end up really nonlinear
[19:12:52] <mikegg> okay, I was wrong earlier about no feedback velocity mode
[19:13:13] <mikegg> the motor's and drives have hall effect sensors
[19:13:43] <mikegg> but the manual warns that the hall effect sensors are not real good below 300 rpm
[19:14:01] <mikegg> so maybe I will try encoder velocity mode
[19:17:46] <mikegg> thanks guys.
[19:17:58] <cradek> hope you get it. report back!
[19:19:05] <mikegg> will do :)
[20:15:08] <mikegg> I switched the X-axis to velocity mode
[20:15:24] <mikegg> it looks better. still doesn't like D gain.
[20:20:47] <mikegg> and now I have two control loops to tune....
[20:28:54] <cradek> unless you have to use a soldering iron, I don't wanna hear it :-)
[20:39:44] <mikegg> if I had to use a soldering iron... I'd need to order some fuses
[20:42:51] <archivist_attic> gas axe, do it in style
[21:15:14] <skunkworks_> http://cgi.ebay.com/7-Ammco-Metal-Shaper-Shaping-Machine_W0QQitemZ270475701255QQcmdZViewItemQQptZLH_DefaultDomain_0?hash=item3ef99bac07
[21:17:35] <archivist_attic> expensive for a little one
[21:18:04] <archivist_attic> or is that they are cheaper and mor common over here
[21:18:26] <skunkworks_> no clue.. Thought it was cute :)
[21:18:33] <andypugh> Bah! I have broken my EMC2 installation
[21:19:22] <archivist_attic> hint restore your backup :)
[21:19:31] <andypugh> It was running fine as version 2.3.0, but I naively updated to 2.3.4 through synaptic, and now it is complaining that the kernel is the wrong name.
[21:20:14] <archivist_attic> see updating doc on the wiki
[21:21:35] <MarkusBec> MarkusBec is now known as MarkusBec_away
[21:21:52] <andypugh> Yes, looking at the Wiki page was my first error...
[21:22:18] <andypugh> I did what is says here: http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?UpdatingTo2.3
[21:22:43] <andypugh> Forgetting that I had done a binary install on a custom (SMP) kernel.
[21:23:00] <archivist_attic> ah.....
[21:23:28] <andypugh> I confess that I am rather surprised that a synaptic update breaks a working installation, especially over kernel naming.
[21:24:12] <andypugh> I am pretty sure EMC would work perfectly well if I could persuade it to try.
[21:24:23] <skunkworks_> one of these would make a nice cnc mill.. http://www.georgesbasement.com/PandWplaner.htm
[21:24:24] <MarkusBec_away> MarkusBec_away is now known as MarkusBec
[21:24:34] <archivist_attic> dunno my old debian got trashed in an update, but its still up and running
[21:25:24] <archivist_attic> hmm sweeping up around me time to burger orf
[21:25:30] <andypugh> Surely not every kernel that EMC2 is running on is called excactly the same name?
[21:26:14] <andypugh> skunkworks_: Most planers got converted to millers, and some are presumably still doing it.
[21:26:55] <skunkworks_> (we have a planer similar to that but it really needs a lot of tlc.
[21:27:04] <andypugh> Planers give a perfectly plane surface, but cutting things like gearbox cases they spend 90% of the time cutting air.
[21:27:29] <skunkworks_> it would be cool to find a small one to put in the corner of the garage... ;)
[21:28:05] <tarzan> nice planners indeed
[21:29:06] <andypugh> In fact, google for "planomill" or "plano mill" and you find lots of converted planers.
[21:29:34] <pjm__> my parport to 7i43 conversion is 90% complete, and it is truely excellant
[21:29:55] <pjm__> in fact it is almost like having a new machine!
[21:30:52] <andypugh> So, does anyone know how to persuade EMC to run on a kernel that isn't called exactly the right name?
[21:35:04] <andypugh> (It was running perfectly on a kernel called "2.6.24-16-rtai-smp" bit after synaptic update to v2.3.4 it is demanding a kernel called "2.6.24-16-rtai" and refusing to even try)
[21:36:52] <aystarik1> andypugh: it will need to install kernel modules, but they are different for smp and non-smp kernels. It means you can crash machine by forcing EMC to run on non-native kernel, if you are lucky...
[21:37:42] <andypugh> It was only an upgrade though.
[21:38:12] <andypugh> Surely not every install is running on _exactly_ that kernel, so demanding that one after upgrade is, well, surprising
[21:38:27] <aystarik1> then there should be 2.6.24-16-rtai kernel to install too.
[21:38:53] <andypugh> ?
[21:39:08] <aystarik1> No, every install of EMC matches its kernel.
[21:39:17] <andypugh> Clearly I should stick to Macs if that is Linux Logic
[21:39:52] <aystarik1> Good luck finding hard realtime Mac :)
[21:40:11] <andypugh> So, unless you have a liveCD install, you should never use Synaptic?
[21:41:10] <aystarik1> why? Just add proper (required) kernel and choose it on next reboot...
[21:41:48] <andypugh> But then I only use one kernel, and my latency is rubbish
[21:41:57] <andypugh> Sorry, not kernel, CPU
[21:42:54] <andypugh> It isn't a huge problem, I can rebuild it all from source.
[21:42:59] <aystarik1> you may wait for -smp version of EMC to appear. IIRC, it is maintained by someone else
[21:44:14] <andypugh> But I am very surprised that given a fully working installation of 2.3.0 (running on the SMP kernel) that a Synaptic Update breaks everyhting.
[21:45:13] <archivist> synaptic will be assuming normal noob users
[21:45:24] <aystarik1> IIRC, -smp is not native to linuxcnc.org repository, so they just restore status quo :)
[21:45:59] <andypugh> I like to think I am normal, and there is plenty of evidence for my noobitude in this discussion
[21:46:50] <archivist> noobs dont compile realtime smp
[21:46:58] <andypugh> It isn't the -smp bit that puzzles me though. What if my kernel was called "thelma"?
[21:47:28] <aystarik1> then your only choice would be to compile EMC
[21:47:33] <SWPadnos> andypugh, kernels are not mutually exclusive - you're allowed to have as many as you like installed at any given time
[21:48:06] <andypugh> Installed as in available from Grub, or actually running at the same time?
[21:48:08] <SWPadnos> unfortunately, when you install multiple kernels (specifically ours), the newly installed kernel is not set up to boot by default
[21:48:09] <aystarik1> and have as many matched EMC installs :)
[21:48:18] <SWPadnos> installed as in selectable via grub
[21:48:58] <SWPadnos> you can probably fix this problem by selecting the correct kernel at boot time, and/or changing grub to boot that one by default
[21:49:17] <andypugh> But the "correct" kernel sucks.
[21:49:41] <andypugh> (See, I have learned how to speak American)
[21:49:49] <SWPadnos> you must use the kernel EMC2 expects
[21:49:58] <SWPadnos> remember, EMC2 includes kernel modules
[21:50:34] <SWPadnos> you need to manually install the EMC2 package that goes with the kernel you want to use
[21:50:45] <SWPadnos> (as you probably manually installed the kernel itself)
[21:51:06] <andypugh> I suspect where I went wrong was to assume that the Package Manager would install a patch, not a whole new hard-coded binary
[21:51:25] <SWPadnos> most packages are binaries
[21:51:45] <SWPadnos> and there's clearly a difference between SMP and non-SMP kernel modules, which EMC2 contains
[21:51:57] <SWPadnos> so there's no way to just "patch" it sanely
[21:52:13] <andypugh> So, how are things managed in the case of other packages, which presumably can be installed on kernels with totally arbitrary names?
[21:52:53] <andypugh> I am asking purely out of curiousity at this point
[21:53:05] <SWPadnos> well, one way (which we've discussed for EMC2) is to have the EMC2 kernel modules in a separate package from the userspace programs
[21:53:20] <SWPadnos> since userspace largely doesn't care about which kernel is running
[21:53:34] <andypugh> Perhaps a more sensible question is: Will a git-pull and compile work with my installed kernel?
[21:53:49] <aystarik1> somewhere in your system there is a proper kernel package installed. EMC2 will install its modules besides it. And will not run until you boot with this kernel
[21:53:58] <SWPadnos> what you're asking for (or about) is the ability to install kernel modules for arbitrary kernels, and that's not going to happen in the near term
[21:54:17] <andypugh> I have the kernel modules installed.
[21:54:21] <SWPadnos> yes, you should be able to build EMC2 (packages even) and have them work with the present kernel
[21:54:22] <andypugh> They work fine.
[21:54:29] <SWPadnos> not the EMC2 kernel modules
[21:54:45] <SWPadnos> or they wouldn't be asking for a different kernel revisio
[21:54:47] <SWPadnos> n
[21:55:36] <andypugh> I hadn't anticipated that an update which involves no changes to kernel modules would contain a set of new kernel modules.
[21:55:55] <SWPadnos> I must not know what you're talking about then
[21:56:02] <SWPadnos> did you or did you not update EMC2?
[21:56:06] <andypugh> Yes
[21:56:13] <SWPadnos> EMC2 contains kernel modules
[21:56:26] <SWPadnos> therefore you must also install the correct kernel for this version of EMC2
[21:56:39] <andypugh> I had a fully working installation of 2.3.0, then clicked manage-synapitic package manager - update EMC2
[21:57:01] <SWPadnos> and you previously had the experimental SMP build of EMC2 and the kernel?
[21:57:09] <cradek> did you compile that 2.3.0 yourself?
[21:57:10] <andypugh> Yes.
[21:57:23] <SWPadnos> yes to ...
[21:57:52] <andypugh> No, I was running a 2.3.0 installed from the experimental binaries.
[21:58:10] <cradek> I don't know what that is
[21:58:19] <SWPadnos> the ones ehj built
[21:58:29] <SWPadnos> in /experimental on linuxcnc.org
[21:59:06] <SWPadnos> andypugh, you need to either build an update yourself, or wait for Eric to make another SMP one
[21:59:16] <andypugh> Indeed, these /deb files here: http://linuxcnc.org/experimental/hardy/smp/
[21:59:17] <alex_joni> if you install a 2.3.x package from another location
[21:59:29] <alex_joni> but still have the official repository in apt
[21:59:47] <alex_joni> any update will be offered and will grab the latest + deps (kernel if needed)
[21:59:48] <SWPadnos> the problem is that you installed a package called EMC2, then asked synaptic to look at the repositories for updated packages (which doesn't include the experimental directory you downloaded from)
[22:00:27] <andypugh> I also have a run-in-place 2.3.3 with cradek's threading patch.
[22:00:29] <alex_joni> so if you want to stay on experimental, then you need to remove the regular linuxcnc.org/hardy repository
[22:00:29] <SWPadnos> but in this case. it may not change the system to boot from the newly installed kernel
[22:00:39] <andypugh> OK, I think I am getting this now.
[22:01:28] <andypugh> So, I need to un-edit the Synaptic repositories so it never tries to update again.
[22:01:46] <andypugh> Then build from source?
[22:03:21] <andypugh> Would you like to guess if my 2.3.3 run-in-place that I built from source obtained through git is actually working correctly with my smp kernel and smp rtai modules? I am starting to worry that I made unwise assumptions there too
[22:03:41] <SWPadnos> there's no particular reason why it shouldn't work
[22:03:54] <alex_joni> the source you build runs with the kernel you have installed/booted
[22:03:57] <SWPadnos> if you built it while the SMP kernel was running
[22:04:17] <alex_joni> if you have more than one realtime kernel, then you can mix it up and it will stop working
[22:05:05] <andypugh> OK, that sounds easy enough then. I was beginning to wonder if there were changes in the sources required to run EMC on smp-patched RTAI
[22:07:13] <cradek> no source changes for smp
[22:07:55] <andypugh> Great, so I just do what I did to build the run-in-place EMC2 but without the --enable-run-in-place flag?
[22:08:37] <andypugh> Thanks chaps. Sorry about my noobishness
[22:09:33] <SWPadnos> andypugh, I'd suggest using --run-in-place
[22:09:50] <SWPadnos> there's no "uninstall" to clean up after a make install
[22:10:06] <SWPadnos> and make install will crap all over the installed packages
[22:10:22] <SWPadnos> (since the programs go in the same place whether they're locally built or package-managed)
[22:10:30] <andypugh> The installed packages that don't work any more you mean?
[22:10:37] <SWPadnos> the best way to do it is to make your own debian packages and install those
[22:10:50] <SWPadnos> yes, and may or may not be easily removable is crapped on ...
[22:10:54] <SWPadnos> s/is/if/
[22:11:15] <andypugh> I was rather hoping to get the desktop shortcuts to specific configs working again is all.
[22:11:29] <SWPadnos> sure - the self-built package is the best way
[22:11:37] <SWPadnos> there are some instructions on the wiki, I hope
[22:12:01] <SWPadnos> (I don't remember the specifics, but it's not much harder than building for run-in-place, and might evenbe easier)
[22:13:28] <andypugh> I have a combination machine you see, running 2 very different sets of configs (different GUI-bits, pendant does different things, axes have different names...)
[22:14:32] <alex_joni> debian/configure -a
[22:14:39] <alex_joni> cd debian/ && debuild
[22:14:49] <Jymm> hey SWPadnos, how ya doin?
[22:14:53] <SWPadnos> ok
[22:14:55] <SWPadnos> you?
[22:15:02] <Jymm> SWPadnos: SSDD
[22:15:03] <alex_joni> that assumes you have the proper kernel installed/booted and deps. installed
[22:15:07] <andypugh> But then I guess that is probably just a case of editing the shortcuts to point to the run-in-place executable rather than the broken one
[22:15:07] <SWPadnos> yep
[22:15:27] <SWPadnos> andypugh, not quite that easy, since you need the emc-environment script to be run as well
[22:15:27] <alex_joni> Jymm: it's not.. they just want you to think that it's a different day
[22:15:38] <SWPadnos> same seat, different drugs
[22:15:50] <Jymm> alex_joni: Are we one 96 hour days now? I didn't get the memo.
[22:16:04] <alex_joni> we're on the one day life
[22:16:11] <Jymm> SWPadnos: different Drugs? Share with the rest of us.
[22:16:28] <SWPadnos> sorry, all out :)
[22:16:37] <Jymm> alex_joni: Damn, that's worse than a butterflies life of 14 days.
[22:16:41] <andypugh> SWPadnos: Ah, avoiding that step every time was the main reason I was looking to do a not-run-in-place build.
[22:17:56] <L84Supper> EMC runs on ARCH now
[22:18:02] <andypugh> I guess the simplest thing to do is to revert back to the working smp 2.3.0 and then run the run-in-place if and when I want to do any threading.
[22:18:34] <cradek> you could build a 2.3.4 package against your smp kernel and then install it
[22:19:12] <cradek> well 2.3.4 does not have a threading change in it anyway, but it has other bugfixes
[22:19:19] <cradek> * cradek shrugs
[22:21:09] <andypugh> So, if I want the threading patch, I get the 2.3.4 sources from git, apply the threading patch, compile and then install?
[22:21:50] <cradek> yes you could do that
[22:22:07] <cradek> or use git-master (pre 2.4)
[22:22:14] <andypugh> I confess that my current patched version was created by blindly following instructions without really knowing what each stage was doing.
[22:25:23] <L84Supper> EMC 2.4.0 with 2.6.29.4 smp in ARCH
[22:26:03] <andypugh> ARCH?
[22:26:18] <cradek> there is no EMC 2.4.0 yet...
[22:26:26] <L84Supper> http://www.archlinux.org/
[22:26:45] <L84Supper> 2.4.0 from git master pre release
[22:26:56] <cradek> cool
[22:27:19] <L84Supper> he's writing up the howto
[22:36:18] <andypugh> OK, so, if I create a 2.3.4 .deb file which runs on the kernel and rtai on the /hardy/experimental/ directory, would it then be useful to put that .deb file in the experimental directory? Or would my .deb file only work with my computer?
[22:38:03] <andypugh> I suspect it is probably safer if all files in that directory are created by someone with a clue, on reflection.
[22:38:12] <cradek> heh
[22:38:42] <cradek> technically it could be shared with any others who are using the same kernel
[22:38:43] <L84Supper> will work on tweaking the scheduler in RTAI or look for whatever else may be done to optimize next
[22:40:40] <andypugh> The kernel/rtai .debs are in that directory already so I assume that a 2.3.4 .deb would be a match?
[22:41:28] <cradek> Do you know who did those packages?
[22:43:40] <andypugh> ncidentally, I had another initally rather puzzling problem today. I finally got the replacement driver for my A-axis, and installed it. I then found that when I jogged Y at full speed on the pendant (A playstation-style controller) that A also moves slowly. I eventually tracked it down to input.0.abs-z-position not being the right pin. for some reason that stick has an x, y and z, and the z is "left right + a bit of the up-down
[22:43:40] <andypugh> the other stick". changing to abs-y-position fixed it. I have no idea what the z is meant to be.
[22:44:05] <cradek> ugh
[22:44:18] <L84Supper> first try has latency about 2x from the 8.04 CD (16uS vs 7uS)
[22:44:53] <L84Supper> this is weird , we start glxgears and the latency drops to 1uS
[22:45:14] <andypugh> cradek: No idea. One of you lot, I assume. Who has write access to http://linuxcnc.org/experimental/hardy/smp/
[22:46:07] <cradek> all of us
[22:46:39] <andypugh> Which "us" is that?
[22:46:45] <cradek> the board members
[22:49:55] <andypugh> "ehj" is mentioned further up?
[22:51:20] <Neo_The_User> does anybody know why I'm getting negative nanoseconds in the /usr/realtime/testsuite/kern/latency/run test?
[22:51:50] <alex_joni> Neo_The_User: jitter can be both positive and negative
[22:52:00] <alex_joni> it's basicly the time since the deadline
[22:52:15] <Neo_The_User> the overalls aren't getting negative.
[22:52:17] <alex_joni> if the task runs faster than it should, then you have negative latencies
[22:52:27] <alex_joni> right.. overall is max(lat.)
[22:52:44] <L84Supper> any idea why starting glxgears would lower the latency?
[22:52:56] <alex_joni> L84Supper: that one beats me
[22:53:01] <L84Supper> wierd
[22:53:27] <alex_joni> L84Supper: you can put glxgears in startup :P
[22:53:34] <alex_joni> have it run on boot :D
[22:53:49] <L84Supper> hey whatever works :)
[22:54:04] <alex_joni> * alex_joni is off to bed
[22:54:06] <alex_joni> night all
[22:54:13] <L84Supper> gn
[23:18:09] <Neo_The_User> Here is my first draft for my archlinux RTAI + EMC2 guide :) http://neo-technical.wikispaces.com/emc2-arch
[23:18:58] <Neo_The_User> I will make arch packages that can be easily installed via: sudo pacman -U rtai-emc.tar.gz within the next week or so depending on how things go
[23:22:01] <aystarik1> did you try to disable CPU_IDLE?
[23:22:26] <Neo_The_User> * Neo_The_User checks his kernel config
[23:22:44] <Neo_The_User> CONFIG_ARCH_HAS_CPU_IDLE_WAIT=y
[23:23:07] <Neo_The_User> so if thats what you mean, then no ;)
[23:23:41] <aystarik1> you may get gain several us by doing so... :)
[23:23:59] <andypugh> I find this website strangely interesting, a bicycle derailleur collection. http://www.disraeligears.co.uk/Site/Home.html
[23:24:03] <Neo_The_User> Thank you!
[23:54:18] <Neo_The_User> I'm still around 16000 with ACPI and CPU_IDLE off
[23:54:55] <aystarik1> do you have C1E control in BIOS? Disable it there too.
[23:57:21] <Neo_The_User> I dont see C1E in my BIOS as a feature. I am googling and i cant find what it means. something to do with clock speed...
[23:57:39] <Neo_The_User> maybe its just labeled as something different
[23:59:20] <aystarik1> it sets CPU to low frequency/low voltage each time HALT is executed -- every idle cycle. Recovery is about 15us. -- exactly what you are seeing...
[23:59:45] <aystarik1> It should be under CPU features