#emc | Logs for 2009-05-04

[00:33:21] <roh> *sigh* .. a laser i also wish for ;)
[01:00:38] <Goslowjimbo> Someone mentioned a EMC workshop in the recent past, but I can't find any reference to the time and place.
[01:05:02] <roh> hm.. i am thinking on adding more switches on this mill here for (eventually encoders), more switches (home) and eventually seperated limits, and thus i need more io. whats the recommended way to do this? add a second serial port via pci-card?
[01:05:17] <roh> s/serial/parallel/
[01:07:14] <roh> ah.. and in the end i need some for the reprap-extruder head ;)
[01:10:35] <Goslowjimbo> How many in's and out's do you need? I'm certainly not in a position to recommend for the EMC group, but I like the Mesa Electronics boards for much expansion.
[01:13:16] <roh> dunno exactly yet. currently i just have 3 end/home-switches (only one switch per axis) and no feedback from the steppers
[01:13:39] <roh> so emc doesnt know what one does 'by hand'
[01:14:39] <roh> i would like to add encoders on the 3 axis (means 2 inputs per axis) and eventually one switch per axis for homing.
[01:15:34] <roh> the reprap extruder needs 2 lines per motor/heater, so its 4 outputs and one sensor (temperature of the head)
[01:16:23] <JymmmEMC> Goslowjimbo: http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?EMC_Fest_2009
[01:16:38] <Goslowjimbo> Thanks
[01:16:42] <JymmmEMC> np
[01:16:43] <Optic> is the reprap useful?
[01:23:03] <JymmmEMC> Google Directions to fest #14 of 20: Slight right at I-40 E (signs for Needles/I-40) Passing through Arizona, New Mexico, Texas Entering Oklahoma - 1,215 mi
[01:30:08] <roh> Optic as far as i can see.. we havent completed the extruder yet.
[01:31:57] <roh> Optic but i know people who have one and it seems to be worth all troubles
[01:32:04] <Optic> excellent
[01:32:09] <Optic> it can make real stuff?
[01:43:40] <roh> as far as i can see yes
[01:44:23] <roh> limited material and precision of course, but one can always make it better and more precise of course ;)
[01:45:05] <roh> uuuh.. i think i just found a interesting sensor vendor.. just quite expensive http://www.waycon.de/
[01:58:11] <cradek> JymmmEMC: 1215 sounds like too many miles.
[01:58:39] <JymmmEMC> cradek: No, that's only step 14 of 20, it's actually 1700 miles
[01:58:46] <cradek> ouch.
[01:59:26] <JymmmEMC> cradek: Yeah, since everyone would be together, thought I'd bring out my POS and see if we could get it working.
[01:59:51] <cradek> sounds interesting
[01:59:57] <cradek> are you coming then?
[02:02:38] <cradek> google says from LA to wichita KS is 1500
[02:02:43] <cradek> = 21 hours
[02:02:57] <JymmmEMC> cradek: 1700 miles x 2 @ 15MPG @ $2.25/gal == ~$510 in gas alone, unloaded.
[02:02:58] <cradek> two full days driving each way - it's possible but not much fun
[02:03:25] <JymmmEMC> cradek: But I'm in NorCal, need to drive down to LA then up to Vegas and across.
[02:03:41] <cradek> ah, thought you were in LA for some reason.
[02:05:01] <JymmmEMC> Oooops, not vegas, but still.
[02:05:22] <cradek> * cradek shudders at the thought of driving a car that gets 15 mpg highway
[02:05:22] <JymmmEMC> SF to LA is 400 miles about 7 hours
[02:05:35] <JymmmEMC> almost 4200 lbs
[02:05:39] <JymmmEMC> SUV
[02:06:21] <cradek> I definitely win with this location: 275 miles from home.
[02:06:38] <JymmmEMC> And it's EXACTLY in the middle of the country too
[02:06:46] <cradek> yeah, pretty close
[02:06:48] <cradek> it's a good location
[02:07:24] <JymmmEMC> http://maps.google.com/?q=2100+S.+West+Street+,Wichita,+KS+67213+&ie=UTF8&ll=37.655558,-97.649231&spn=55.104421,82.880859&t=h&z=4
[02:07:33] <JymmmEMC> Looks in the middle to me =)
[02:08:30] <JymmmEMC> Also, bad dates... my gf goes in for surgery at that time
[02:09:56] <cradek> heck. hope it goes well for her.
[02:11:41] <roh> how much does a 'normal' encoder usually cost?
[02:13:07] <JymmmEMC> cradek: Yeah, only been waiting 2.5 years
[02:13:50] <cradek> roh: that's like asking how long a stick is. they go from $50ish on up to $huge
[02:14:29] <cradek> you could browse usdigital.com for a good idea of what kinds are available and they have their prices online.
[02:14:44] <Roguish> hey all. will ubuntu work on an atom processor? 8.04 + emc2
[02:17:04] <Optic> yes
[02:18:18] <roh> cradek well.. what do you guys use? i found stuff which is up to 4keuro per axis
[02:19:18] <cradek> in every case when I've done a retrofit, the encoders came on the machine or motors I used, or were bought from usdigital in an appropriate form for mounting in the necessary place.
[02:19:43] <cradek> usdigital has a lot of shapes for different mounting schemes.
[02:20:10] <cradek> often if you have to fabricate a little something anyway, you can use their "kit" encoders for only a few dozen dollars each
[02:20:54] <roh> besides 'more is better' .. how much resolution makes sense when one has 200 steps per rpm and 10microsteps? 2000? 4000?
[02:21:37] <cradek> "steps per rpm" makes no sense, and microsteps means you have stepper motors which don't need encoders anyway
[02:21:46] <cradek> so I'm at a loss to answer your question :-/
[02:22:13] <cradek> oh I bet you mean per rotation
[02:22:25] <cradek> but still, why are you putting encoders on steppers?
[02:22:30] <SWPadnos> roh, there are some from DigiKey that are $30, and Mariss (of geckodrive fame) likes them
[02:22:37] <SWPadnos> others have also had good luck with them
[02:22:50] <SWPadnos> http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=102-1307-ND
[02:23:08] <SWPadnos> programmable resolution even, 16 settings from 48 to 2048 CPR
[02:23:43] <cradek> cool - programmable and with index
[02:23:47] <SWPadnos> yeah
[02:24:06] <cradek> 30000 rpm!?
[02:24:19] <SWPadnos> I don't know how well they last in an industrial envrionment, but Mariss has been using them for a while for drive developmnet
[02:25:04] <SWPadnos> and that's apparently cycles per rev, not counts. you get 192-8192 counts
[02:28:26] <cradek> they can do 100/400 - maybe I'll use them for the jogwheels I'll make someday.
[02:28:36] <SWPadnos> heh
[02:28:47] <SWPadnos> could be inexpensive enough to make it worthwhile, but I'm not sure
[02:32:31] <cradek> I sincerely doubt it
[02:43:14] <Roguish> linux install question: error '/bin/modprobe/ abmornal exit'
[02:43:52] <Roguish> can't seem to get past it. brand new disk, brand new atom processor board. new everything.
[02:50:12] <toastydeath> ffff
[03:57:55] <roh> cradek the idea was to use the index for a more precise homing and the encoder in general to have a feedback of the real position, in case one moves the axis by hand
[03:58:21] <roh> the bf20 doesnt replace the handwheels with steppers, the steppers just move them further out
[03:58:47] <Roguish> ok, guys. is the emc iso 8.04 or 8.04.1 ???
[04:05:24] <topls64> Had lots of learning/fun with the mill the last five days ;)
[05:05:05] <Roguish> ok, 8.04.2 works on the Intel D945GCLF. 8.04 does not install !!!!
[09:06:20] <piasdom> g'mornin all
[09:58:24] <MrSunshine> 1uS step speed =)
[09:58:26] <MrSunshine> yeey
[11:42:34] <micges> good morning
[12:38:13] <steve_stallings> steve_stallings is now known as steves_logging
[13:01:29] <steves_logging> steves_logging is now known as steve_stallings
[13:03:48] <Optic> hi
[13:41:44] <piasdom> what's the difference between rt and rtai ?
[14:44:28] <steve_stallings> rt is "soft" realtime, good enough for many things, possibly even medium performance servo systems, but not good enough for driving steppers; rtai is "hard" realtime
[14:44:45] <steve_stallings> EMC does not yet have an implementation that runs on rt
[14:46:35] <steve_stallings> rtai is usually reliable down to less than 20 uS intervals, rt is more typically several hundred uS
[14:53:30] <piasdom> steve_stallings; thanks..been wordering about that
[16:52:31] <steve_stallings> steve_stallings is now known as steves_logging
[17:16:51] <MrSunshine> dammit i cant get my stepper to run smooth at low speeds :/
[17:43:44] <motioncontrol> good evening.i want compile yemc , but i have the problem is possible a help?
[17:50:27] <MrSunshine> with a higher iductance stepper motor the pwm cycle needed to hold a voltage in the coils should decrease right ?
[17:52:17] <cradek> lower inductance is better for steppers, because you want to minimize the time taken for the current to change.
[17:52:42] <MrSunshine> cradek, ahh yes so inductance has to do with hax step rate also then right? :)
[17:52:57] <MrSunshine> can only get my big steppers to about 2100uS step rate, while the small one could do all the way down to 1uS
[17:53:24] <cradek> yes very much so
[17:53:51] <MrSunshine> but the bigger stepper works alot smoother at lower speeds then the little one :/
[17:54:10] <MrSunshine> or not, it shakes like hell also :P
[17:54:15] <cradek> a big stepper will have much more low speed and holding power. a small stepper will go faster
[17:54:28] <cradek> he sure you are half stepping or micro stepping. do not ever full step.
[17:54:44] <MrSunshine> i do high torque half stepping atm if everything is as it should
[17:54:55] <MrSunshine> cradek, im writing my own controller using an AVR and a L298 :P
[17:55:25] <cradek> why not use L297?
[17:55:53] <MrSunshine> cradek, cause i want to learn how it all works =)
[17:56:18] <MrSunshine> and L297 do not use high torque half stepping and does not do anything more then half stepping :P
[17:56:27] <MrSunshine> it uses the usual, full on, half on, full on, half on
[17:56:47] <cradek> what is high torque half stepping?
[17:56:56] <MrSunshine> cradek, both coils always on at some degree
[17:57:21] <MrSunshine> 100%, 40% - 40%, 100% etc insted of 100%, 100% - 0%, 100%
[17:57:31] <cradek> oh, that sounds like microstepping
[17:57:32] <MrSunshine> so much more uniform strenght in the steps then the usual half stepping
[17:57:39] <cradek> I see
[17:57:54] <MrSunshine> cradek, ye microstepping does the same thing but alot more current values :)
[17:58:49] <MrSunshine> seems like my pwm code is working tho, the motor does not run hot at 12V like it did with my old only full step drivers :P
[18:01:42] <jepler> cradek: it's simple to see that with only 0% and 100% as winding currrents, the motor power is lower for 1-winding-on vs 2-windings-on
[18:02:19] <cradek> seems like that depends on how your chopper works
[18:02:50] <cradek> won't it keep the current the same, no matter how many windings it's going through?
[18:03:01] <jepler> l298 uses one sense resistor for each winding
[18:03:23] <cradek> oh ok, then it would make a big difference.
[18:03:43] <cradek> a ballast resistor unipolar setup would be another story
[18:03:57] <MrSunshine> ye, im reading that into the ADc .. i put PWM on 100% until the winding hits a given value, then i turn the PWM to pwmForMaxCurrent * say 0.4 to get the pwm cycle, and i set that, and wait for the next step to occur
[18:04:30] <cradek> interesting, how fast is the adc?
[18:04:47] <MrSunshine> cradek, dunno realy :) its an AVR adc so .. 100khz - 200khz or so
[18:05:07] <MrSunshine> By default, the successive approximation circuitry requires an input clock frequency between 50
[18:05:07] <MrSunshine> kHz and 200 kHz to get maximum resolution
[18:05:22] <MrSunshine> or that might not be the speed :P
[18:05:31] <jepler> one conversion takes multiple adc clocks
[18:05:56] <MrSunshine> yeap, 13
[18:06:54] <jepler> 15kHz, or smaller than the top step rate on a lot of stepper systems (about 600rpm at 1/8 stepping)
[18:08:56] <MrSunshine> ohh set the prescaler of the ADC to 128 insted of 64 and got alot faster step times
[18:09:29] <MrSunshine> all the way down to 1250uS insted of about 2kuS
[18:14:53] <MrSunshine> but it still feels very slow :)
[18:15:07] <MrSunshine> looking at the preconfigured stuff in emc, they got a billion faster step times then i can get :P
[18:15:12] <MrSunshine> like gecko ... lightning fast
[18:15:51] <cradek> I love microcontrollers, but they're not very good for stuff like this
[18:16:08] <cradek> also quadrature counting
[18:16:23] <MrSunshine> cradek,feels like its a tad to slow :)
[18:16:23] <JymmmEMC> not good for _________ ?
[18:17:09] <cradek> JymmmEMC: reading step/direction, reading quadrature, sensing analog stuff and responding very fast (running a chopper)
[18:17:50] <JymmmEMC> cradek: Really? I would think far better than paraport
[18:18:15] <cradek> yeah but that's a pretty low bar
[18:18:40] <MrSunshine> im not realy expecting that much of this project more then learning how it all works =))
[18:18:47] <cradek> encoders spin much much faster than a parport or avr can poll them
[18:19:05] <cradek> that's why fpgas are universally used for the task
[18:19:30] <JymmmEMC> some AVR's are 10MHz, how can encoders go faster than that?
[18:19:36] <cradek> MrSunshine: I understand that for sure - learning the limitations of the design is an important part.
[18:20:03] <cradek> 10MHz clock cycle, sure, but how fast can you read and decode quadrature and then do something smart with it?
[18:20:07] <JymmmEMC> While it's the CLK speed, that's 10 Million cycles per second.
[18:20:21] <JymmmEMC> and encoder puts out that much?
[18:21:03] <MrSunshine> JymmmEMC, running mine at 10mhz now, aprently lost my 20mhz crystal
[18:21:07] <MrSunshine> going 20 as soon as i find one
[18:21:08] <JymmmEMC> lets say the overhead is 5x, leaving 2 Million/s
[18:21:59] <MrSunshine> but i can say one thing, my drivers performance kicks the other drivers i have asses a long way already :)
[18:22:00] <cradek> you're going to read and decode quadrature and do something with the result in 5 instructions?
[18:22:22] <cradek> not even John K could write it to work that fast :-)
[18:22:22] <MrSunshine> wtf is a quadrature? :)
[18:22:58] <JymmmEMC> cradek: iirc, AVR can do 2 instructions per cycle.
[18:23:47] <JymmmEMC> MrSunshine a methid of encoding
[18:23:50] <JymmmEMC> method
[18:25:56] <MrSunshine> time to hook this up to my 24V supply soon :)
[18:26:02] <MrSunshine> its not even close to gettin ghot at 12V :)
[18:26:29] <JymmmEMC> MrSunshine http://prototalk.net/forums/showthread.php?t=78
[18:29:53] <MrSunshine> ohh looks like my halfstepping doesnt work when i do it real slow :P
[18:29:56] <MrSunshine> step by step that is :)
[18:30:34] <MrSunshine> ohh wellt he duty cycles might not be a linear representation of the coil current :P
[18:31:30] <JymmmEMC> what are you mucking with?
[18:32:07] <MrSunshine> i use pwm to set the current going throught the coils .. like a chopper but using pwm insted, but using just maxPwmCycle * 0.4 to get a 40% coil current doesnt seem to work very good :P
[18:33:17] <MrSunshine> seems it goes to 0 in current
[18:43:45] <motioncontrol> good evening.i want compile yemc , but i have the problem is possible a help?
[18:49:08] <JymmmEMC> cradek: http://www.lsicsi.com/pdfs/Data_Sheets/LS7083_LS7084.pdf
[19:07:01] <cradek> motioncontrol: http://acronyms.thefreedictionary.com/DATAJA
[19:28:24] <motioncontrol> thanks for help , my english is school version
[19:29:53] <motioncontrol> but i have the problem with compiling the yemc.cc file.the terminal write: make:failed to remake makefile 'Makefile'. please hel me ?
[19:34:20] <motioncontrol> i thing the problem is dependents the posemath.h and emcpos.h. is possible?
[19:34:48] <alex_joni> motioncontrol: maybe it's best if you say what you did so far?
[19:35:05] <alex_joni> simply replace xemc.cc with yemc.cc ? or something else?
[19:36:12] <motioncontrol> thanks alex i have replease the xemc with yemc but not compile , have error.
[19:39:51] <motioncontrol> alex i want control with emc 5 axis , with gui similar at xem.xemc control only 3 axis , but yemc 6 axis, but i have the problem for compile.you can help me please?
[19:44:05] <alex_joni> did you edit yemc.cc ?
[19:44:24] <motioncontrol> yes i can edit
[19:44:57] <motioncontrol> i have open in editor yemc
[19:45:01] <alex_joni> I think the remake makefile is because the dependencies for xemc are now wrong
[19:45:06] <alex_joni> did you try a make clean?
[19:45:54] <motioncontrol> yes i have write make clean and after make , but compile stop not finish and not have the error.
[19:46:48] <motioncontrol> simply exir at compile and blink the cursor terminal
[19:46:59] <motioncontrol> simply exit
[19:48:13] <motioncontrol> alex you can prove the compile yemc.cc on your system?
[19:48:21] <alex_joni> I am starting now
[19:48:31] <motioncontrol> thanks
[20:03:34] <motioncontrol> alex the compile is ok or have exit and no complete compile?
[20:08:02] <alex_joni> lots and lots of errors
[20:08:12] <alex_joni> there was the problem with the remake makefile
[20:08:22] <alex_joni> but I fixed that by adding a couple includes
[20:08:27] <alex_joni> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?FailedToRemakeMakefile
[20:09:09] <alex_joni> 112 error lines now
[20:10:33] <motioncontrol> it's a problem.you thing the error code or dependencies?
[20:12:13] <alex_joni> 94 errors
[20:12:23] <alex_joni> LOTS of things have changed from emc1 to emc2
[20:13:36] <motioncontrol> yes i thing it the problem are the change emc1 at emc2. alex if possible correct the error is very good.
[20:14:11] <motioncontrol> i don't have the know very good the c++
[20:21:12] <alex_joni> 48 errors
[20:21:45] <motioncontrol> thanks for your patience alex
[20:22:26] <motioncontrol> you are great
[20:26:05] <alex_joni> 19 now
[20:28:27] <motioncontrol> ok alex
[20:31:30] <alex_joni> make
[20:31:30] <alex_joni> Depending emc/usr_intf/yemc.cc
[20:31:30] <alex_joni> make: Entering directory `/home/pubuntu/emc2-test/src'
[20:31:30] <alex_joni> Compiling emc/usr_intf/yemc.cc
[20:31:30] <alex_joni> Linking yemc
[20:34:04] <alex_joni> motioncontrol: it's compiling, but not quite working
[20:35:08] <motioncontrol> alex excuse whath is the step for compile?
[20:35:58] <motioncontrol> you have the code the yem.cc correct?
[20:36:17] <motioncontrol> you can attach in pastebin
[20:36:27] <alex_joni> http://eneas.juve.ro/~juve/emc/yemc.cc
[20:36:48] <alex_joni> most of the changes I did (commenting out things), I used //FIXME-AJ:
[20:36:52] <alex_joni> or /* FIXME-AJ:
[20:37:00] <alex_joni> should be easy for you to see what I commented out
[20:37:59] <motioncontrol> ok thanks alex now compile
[20:43:28] <alex_joni> motioncontrol: http://imagebin.org/47855
[20:43:52] <motioncontrol> ok alex the compile and the start the interface is ok.very thanks i are debt at you
[20:44:22] <alex_joni> motioncontrol: you need to remove the logging buttons, those don't do anything
[20:47:05] <motioncontrol> i don't understand alex. i don't remove the button.
[20:47:22] <alex_joni> then forget it.. :)
[20:49:17] <motioncontrol> alex thanks for your help . i want retrofit the maschine with 5 axis at my friends and i have proposed the use emc
[20:49:33] <motioncontrol> and not siemens 840 d
[20:49:55] <alex_joni> if it works ok.. why not :)
[20:50:14] <motioncontrol> the drive are analog and i use 2 card m5520
[20:51:15] <motioncontrol> alex the next mouth i go in RO.i stay at Timiscioara.You stay at Timiscioara?
[20:51:50] <alex_joni> not sure, I have a trip to germany planned
[20:53:08] <motioncontrol> i have my collega meccanico is Romania and work alla day at me.Can i speek you an my friend traslate if possible when you are free?
[20:53:36] <alex_joni> I'll see how free I'll be
[20:54:43] <motioncontrol> ok alex.the next week i write you, if possible i and my friends go in Ro.Thanks for all your help is very utils.
[20:55:07] <alex_joni> sure, bye
[20:55:20] <motioncontrol> good night at all.
[20:57:26] <alex_joni> good night all
[23:03:25] <LawrenceG> skunkworks, It lives!!! no smoke leaked out so far.... running a 24vdc/20a motor
[23:03:50] <LawrenceG> motor is getting warm, but the drive is cool
[23:09:52] <jepler> drat, I broke my .01" endmill. and I don't have a spare :( (past) time to order more, I guess
[23:10:10] <jepler> worst is, I have no idea why it broke. I looked over and saw that I wasn't milling any copper away..
[23:37:27] <jepler> in related news, I should learn to take others' advice more readily. having finally faced my table flat where I put the copper clad, I got very good results from a "vee" bit for trace isolation milling
[23:54:12] <eric_unterhausen> somebody come over to my house and help me get my mill running