#emc | Logs for 2007-11-03

[00:36:56] <jmkasunich__> jmkasunich__ is now known as jmkasunich
[00:46:44] <dmes1> dhc-8 Q-400 saga continues.... stray 'o' ring in the damping orifice... prevented downlock by hydraulic lock-up.... SAS maintenance will be ripped a new one i'm sure
[00:48:46] <dmes1> who want to tho sharsies on an relatively excellent aircraft??? i could get a better deal on 3-4
[02:57:14] <dmes1> high all anyone home>
[08:45:11] <skinnypuppy34> zas
[08:46:38] <anonimasu> skinnypuppy34: are you running without current limit?
[08:47:11] <skinnypuppy34> zas that was the cat on the keyboard.
[08:47:27] <skinnypuppy34> No I'm set at 3 A
[08:47:36] <anonimasu> ok
[08:47:37] <anonimasu> good :)
[08:47:39] <skinnypuppy34> I fixed some of my probs
[08:47:46] <anonimasu> geckos _BLOW_ when you dont have that
[08:47:50] <anonimasu> and you run your motors for too long
[08:48:04] <anonimasu> ^_^
[08:48:41] <skinnypuppy34> I got my rapid up to 90, reduced accel
[08:48:48] <anonimasu> ok
[08:49:36] <skinnypuppy34> I had mentioned the motors stalling earlier, apparently the trigger was bumping into the backlash too hard
[08:52:03] <skinnypuppy34> Thanks again anonimaus , 5am nap time :o)
[08:59:05] <anonimasu> laters
[08:59:05] <anonimasu> :)
[10:48:29] <alex_joni> heh, that's a cute name "anonimaus"
[10:50:31] <Vq^> :)
[10:55:51] <alex_joni> haha http://www.mediablog.cz/story.php?id=5099
[10:55:51] <alex_joni> stupid dog
[10:56:57] <Vq^> :/
[10:58:47] <Vq^> alex_joni: my latest toy: http://arda.no-ip.org/crapmiller.jpeg
[11:03:15] <Guest877> Hi, I have some questions that i hope someone can help me answer. 1. Is it possible to use EMC to controll a stepper motor connected directly to the printer port (using transistors to amplify the signals). 2. is there any way to calculate make EMC calculate the total velocity of a 2 axis machine eg. sqrt(velX^2+velY^2) and use the velocity to give signals to another pin (for example if i want 100 pulses / inch)
[11:05:05] <alex_joni> explain 2
[11:05:30] <alex_joni> 1 is possible, you just need to use a different stepgen step type
[11:05:31] <Vq^> nr 1 is certainly possible
[11:06:03] <alex_joni> 2 is also possible, but I don't understand what you want to be the output on the pin
[11:06:34] <alex_joni> Guest877: for 1. look at this: http://www.linuxcnc.org/docview/2.2/html//hal_rtcomps.html#sub:Stepgen-Step-Types
[11:06:50] <alex_joni> depending on your motor you probably want step type 5 I think
[11:07:48] <Guest877> I will use it to control a laser, and I want to get the number of laser pulses per inch to be constant even if the speed change.
[11:08:04] <alex_joni> Guest877: then you need to calculate the speed
[11:08:13] <alex_joni> the sim/ configs have hooks to do that
[11:09:24] <Guest877> Thanks, I'll try to find out how :), just wanted to know if it was possible before wasting alot of time.
[11:09:33] <alex_joni> http://cvs.linuxcnc.org/cvs/emc2/configs/common/core_sim.hal?rev=1.13;content-type=text%2Fplain
[11:09:52] <alex_joni> net XYvel hypot.0.out
[11:10:04] <alex_joni> you can then define anothe stepgen (laser-power)
[11:10:16] <alex_joni> and connect the XYvel to it
[11:10:30] <alex_joni> although now that I think of it freqgen is probably a better choice
[11:10:39] <alex_joni> (something like a velocity-mode stepgen)
[11:11:03] <Guest877> thanks again :) now i just have to start building.
[11:11:08] <alex_joni> again sorry.. freqgen is the old name.. it's called pwmgen now :)
[11:11:10] <alex_joni> http://www.linuxcnc.org/docview/2.2/html//hal_rtcomps.html#sec:PWMgen
[11:14:54] <fenn> Vq^: what are you going to do with all that Z travel?
[11:18:03] <Vq^> fenn: it should be possible to use it as a pluck robot for smaller machines
[11:19:19] <Vq^> it's quite long for plasma cutting thought :)
[11:39:22] <Vq^> btw, has anyone here used emc for plasmacutting?
[11:44:24] <fenn> not anyone here right now
[12:09:14] <alex_joni> Vq^: Dallur has
[12:09:19] <alex_joni> www.rugludallur.com
[12:37:26] <Vq^> that wiki-page looked interesting...
[12:44:09] <alex_joni> which one?
[12:45:28] <Vq^> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Torch_Height_Control
[12:46:52] <anonimasu> well, if you build anything talk to dallur..
[12:46:54] <anonimasu> he's had lots of problems on the awy
[12:46:55] <anonimasu> way
[12:47:03] <anonimasu> on the other hand I ran my plasma without problems..
[12:47:07] <anonimasu> without thc..
[12:47:17] <anonimasu> though I swapped it out for oxy because I need to cut really thick stuff
[12:55:41] <alex_joni> how thick?
[12:58:41] <alex_joni> haha http://blog.dreamhost.com/2007/08/10/the-internet-is-not-for-people/
[13:02:46] <jepler> alex_joni: freqgen is now rolled into stepgen with ctrl_type=1; pwmgen is something else
[13:20:47] <awallin_emc> make: Warning: File `depends/rtapi/rtai_ulapi.d' has modification time 3.1e+08 s in the future
[13:22:10] <awallin_emc> hmm. maybe my date is set wrong (1998 Jan 1)
[13:25:37] <awallin_emc> emc/task/emctaskmain.cc 2619: can't initialize motion
[13:25:51] <awallin_emc> when running the latest TRUNK...
[13:32:31] <anonimasu> alex_joni: 40mm
[13:40:09] <alex_joni> awallin_emc: sounds like a bogus compile
[13:40:17] <alex_joni> I suggest make clean, sudo ntpdate, and make
[13:40:29] <alex_joni> anonimasu: that's still ok for plasma :P
[13:40:50] <alex_joni> although a bit expensive
[13:41:24] <awallin_emc> alex_joni: yeah, I'm doing make clean && make now
[13:42:51] <alex_joni> jepler: you're right.. I was still a bit asleep at the time :D
[13:43:13] <alex_joni> but for his purpose I think pwmgen might be better than stepgen
[13:50:22] <awallin_emc> seems my config files are out of date:
[13:50:25] <awallin_emc> emc/task/emctask.cc 312: interp_error: Coordinate system index parameter 5220 out of range
[13:50:25] <awallin_emc> Coordinate system index parameter 5220 out of range
[13:50:25] <awallin_emc> emc/task/emctaskmain.cc 2629: can't initialize interpreter
[13:51:56] <jepler> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?UpdatingConfigurationsForDevelopmentVersions#var_file_changes
[13:52:00] <jepler> The .var file has been extended to require coordinate system offsets for U, V and W axes. Copy a fresh .var file. An incorrect var file leads to emc printing 'emc/task/emctask.cc 312: interp_error: Coordinate system index parameter 5220 out of range' at startup.
[13:53:19] <awallin_emc> thanks, will try
[13:55:33] <awallin_emc> is linkpp going to disappear soon?
[13:56:37] <alex_joni> awallin_emc: yes
[13:57:03] <jepler> it will be in the 2.2.x series but probably not 2.3.
[13:57:32] <awallin_emc>
[13:57:33] <awallin_emc>
[13:59:44] <awallin_emc> ok
[14:30:56] <JymmmEMC> JymmmEMC is now known as Red70sShow
[14:31:23] <Red70sShow> Red70sShow is now known as Jymmm
[16:46:31] <awallin_emc> anyone remember if it's possible to reverse the polarity of the PWM output from the m5i20?
[16:48:55] <alex_joni> doesn't the pin have an invert param?
[16:50:00] <awallin_emc> looking...
[16:53:40] <awallin_emc> there's an inverted pin for normal IO, but I can't find anything on the DAC/PWM
[16:55:37] <awallin_emc> maybe negative gain?
[16:55:42] <jepler> use the dac -scale and -gain parameters ?
[16:56:08] <jepler> that will probably have the effect of inverting the "dir" pin of that dac
[16:56:15] <jepler> which may or may not be what you want..
[16:56:38] <awallin_emc> trying...
[16:58:08] <awallin_emc> hmm, so my motor stands still whenever I input -10 or +10 as the DAC command
[16:58:21] <awallin_emc> setting a negative gain did not seem to work
[16:58:38] <jepler> and a small value or zero sets it moving?
[16:59:47] <awallin_emc> at zero it jumps violently and then shuts down. maybe the servodrive goes into overcurrent
[17:00:49] <awallin_emc> these are Jon Elsons servodrives
[17:00:58] <awallin_emc> they have optoisolators for the PWM input
[17:01:07] <awallin_emc> so could I just reverse the leads on the PWM
[17:01:32] <awallin_emc> now I just automatically put GND to the '-' input and PWM to the '+' input, but maybe it works the other way around too?
[17:02:57] <awallin_emc> I will have to try that.
[17:03:21] <jepler> if it's an optoisolator, seems like that has the effect of inverting
[17:04:41] <awallin_emc> yep, jon describes it as an optocoupler http://pico-systems.com/pwm_pins.html
[17:06:33] <jepler> hm, it won't do anything if you put GND to the + input, since PWM doesn't swing below GND
[17:07:44] <jepler> wouldn't you put VCC to + and PWM to - then?
[17:08:46] <jepler> depending on jumper settings on the 5i20 board, either +3.3V or +5V is available at pin 49 of each connector
[17:09:09] <awallin_emc> yes, thats more logical, since we want the LED to conduct
[17:09:32] <awallin_emc> I have a separate -5V supply... will just take some time to wire
[17:10:05] <awallin_emc> I guess the DIR input doesn't really matter, we can take care of polarity with encoder scale
[17:10:28] <awallin_emc> +5V supply I meant
[17:12:32] <jepler> i don't think that drawing 3mA per signal will present a problem
[17:21:46] <awallin_emc> now let's see...
[17:23:03] <awallin_emc> yeah! it works. now I'll wire the other two axes the same
[17:23:19] <awallin_emc> then I think I'll do a DAC vs. RPM test for all three motors
[17:26:07] <awallin_emc> has anyone done something like a spectrum analyzer in HAL?
[17:26:25] <awallin_emc> would be nice to drive the motors with a sine-wave command and look at the mechanical bandwidth
[17:26:42] <awallin_emc> will be totally different now without load and then when the motors are on the machine ofcourse....
[17:46:09] <awallin_emc> what's the easiest way to do an RPM measurement from an encoder counter value? ddt block probably?
[17:46:32] <alex_joni> yup
[17:46:53] <alex_joni> but encoder counter usually has velocity estimation
[17:47:03] <alex_joni> (not sure about the 5i20 one..)
[17:47:24] <awallin_emc> don't think the 5i20 has one (very un-canonical driver...)
[17:52:56] <jmkasunich_> ddt block is easiest, but it will suffer from quantization noise
[17:53:18] <jmkasunich_> run it in the slowest possible thread for best results, and/or run the output of the ddt through a low pass filter block
[17:53:26] <jmkasunich_> (I think we have one of those)
[17:59:59] <Roguish> how 'bout using the index on the encoder? only 1 per revolution.
[18:00:17] <jmkasunich_> that will make it worse, not better
[18:00:46] <Roguish> ?
[18:01:05] <jmkasunich_> if you have a 5i20 or other hardware counter, you want to use the highest possible PPR and run the ddt block slowly to mimise the "noise" on the speed signal
[18:01:17] <Roguish> ok
[18:01:42] <Roguish> smearing the noise out then
[18:01:48] <jmkasunich_> if you are counting in software with the most recent version of the HAL encoder module then you can use it's velocity output, which will be less noisy than ddt'ing the raw counts
[18:03:03] <jmkasunich_> example: 1000 rpm, 360 counts per rev, = 360,000 counts/minute = 6000 counts per second
[18:03:28] <jmkasunich_> with a ddt running every 1/1000 of a second, the result will be 6 counts
[18:03:48] <jmkasunich_> if you change from 1000 RPM to 1100, it will be 6.6 counts, so you'll get 6 sometimes and 7 sometimes
[18:04:32] <jmkasunich_> if you use a 6000 count per rev encoder, that gives you 6,000,000 counts per minute = 100,000 counts per second, = 100 counts every 1/1000 second
[18:04:44] <jmkasunich_> change to 1100 RPM and you get 110 counts every 1/100 second
[18:04:52] <awallin_emc> ok, I will put it in a slow thread, and then scale the output to RPM
[18:05:00] <awallin_emc> ddt outputs the change between ddt.N calls, right?
[18:05:07] <jmkasunich_> right
[18:05:20] <jmkasunich_> but it scales it so it is in units per second
[18:05:42] <jmkasunich_> in my first example, the frequency is 6000 counts per second, so the ddt output would be 6000
[18:05:49] <awallin_emc> ah, ok. what is the ddt.time parameter?
[18:06:21] <jmkasunich_> .time parameters are how many clock cycles the function took the last time you ran it
[18:06:43] <jmkasunich_> more for debugging or performance analysis, not system configuration
[18:07:11] <awallin_emc> ok. I'll put it in a 1s thread or something and try.
[18:07:23] <jmkasunich_> back to the example - at 1100 RPM, you have 6600 counts per second, but a ddt running at 1/1000 second will output either 6000 or 7000
[18:07:49] <jmkasunich_> this is for an RPM meter or something?
[18:07:50] <awallin_emc> anyone worked with Jon Elsons PWM servodrives? I remember there was a special bootup sequence for charging some caps, but right now I'm getting a little erratic behaviour
[18:08:10] <jmkasunich_> you using his board to generate the PWM?
[18:08:39] <jmkasunich_> the ppmc driver generates the bootup pulses, other pwm generator drivers don't
[18:08:40] <awallin_emc> nope, a m5i20
[18:08:59] <jmkasunich_> I don't remember the details of the bootup pulses, and don't have his pwm amps
[18:09:16] <awallin_emc> ok, I'll ask him by email
[18:17:57] <jepler> http://www.bunniestudios.com/blog/?p=139 -- huh, I had never heard of a "16-point program" when it came to chinese counterfitting
[18:18:25] <jepler> "I don't think those typos are just because the person who copied this doesn't know how to type. I was talking to a friend in Hong Kong, and he said that knock-offs like this are regulated by what's called a "16-point program". The manufacturer must demonstrate 16 differences between his product and the original: this can include things like changing "Apple" to "Appls", or in another case, "Motorola" to "Mokorola". Each misspelling counts as one differen
[18:19:44] <SWPadnos> ce
[18:19:46] <SWPadnos> :)
[18:22:33] <jepler> Each misspelling counts as one difference. So as long as the manufacturer can point to a total of 16 differences, it is okay to sell."
[18:34:57] <awallin_emc> now I did a nice DAC vs. RPM plot for my motor. anything else I could try while the motor is still on the table and not on the machine?
[18:35:21] <awallin_emc> I guess I could try PID control now too. Are there any nice interactive PID tuning pyvcp panels out there??
[18:36:06] <SWPadnos> I think you should make one
[18:36:22] <SWPadnos> with sliders and analog-like meters and stuff
[18:38:11] <awallin_emc> right :) do you think the command should come from a pyvcp panel or could that come from EMC?
[18:38:28] <awallin_emc> do most people use step functions or a short linear move when tuning
[18:39:49] <anonimasu> awallin_emc: well, on commercial machines you sue step functions..
[18:40:07] <anonimasu> use
[18:40:12] <awallin_emc> something like this: if I'm running EMC with a servo config, does EMC by default wire the PID parameters to values set in the ini file? could I not do that and have the PID gains set by a pyvcp panel?
[18:40:57] <SWPadnos> I think the PID parameters aren't pins, so it's not trivial/possible to use pyvcp to tune them
[18:41:31] <SWPadnos> you can tune live though - EMC sets the parameters to the ini settings, but it doesn't force them to stay that way
[18:41:36] <awallin_emc> SWPadnos: I think I tried to change that, but maybe someone reversed my commit?:)
[18:41:56] <SWPadnos> ok, it could be. I know siggen was changed to all pins, dunno about PID
[18:42:42] <awallin_emc> ok, long day, need to clean up and go home - probably get to closed loop PID next week. bye.
[18:43:06] <SWPadnos> they are still params
[18:43:09] <SWPadnos> see you
[18:43:14] <SWPadnos> I'm out to do some chores as well
[19:33:57] <tomp> cool, i'm on a windows box but i find opera has an irc client :)
[19:43:59] <alex_joni> it does?
[19:45:29] <alex_joni> oh, it does.. never tried it :D
[20:13:48] <tomp> i've been looking at electric bike hub motors. they come with dc chopping controllers , like 36V/600W is common. maybe they can be used for spindle controls.
[21:57:16] <alex_joni> http://imagebin.org/11479
[21:58:04] <Hugomatic> Hi guys, I'm looking for an easy way to convert vectors (svg) to gcode (something like truetype-tracer?). It has to run on my emc machine (Ubuntu 6.06). Any ideas?
[22:01:53] <Ziegler> yup
[22:01:59] <Ziegler> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Cam
[22:03:55] <Hugomatic> Thanks Ziegler
[22:04:02] <Ziegler> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?AptProgrammingForEMC
[22:04:07] <Ziegler> one more
[22:04:35] <Ziegler> might give you a leg up
[22:18:14] <Hugomatic> Ziegler, I've tried ps2gc and it worked OK (except for a few errors with weird gcodes, missing feed, and some font outlines were rotated 90 degres). Thanks again.
[22:29:36] <SkinnyPuppy34> g1 x-.05y-.05