#emc | Logs for 2010-05-09

[00:01:14] <andypugh> There is a chapter of NML messages in http://www.linuxcnc.org/docs/EMC2_Developer_Manual.pdf
[00:01:40] <andypugh> Not that I am sure that helps much.
[00:02:05] <Dave911> You are right Andy ... I totally blanked on that .... I'll look at that... thanks ..
[00:07:40] <DaViruz> http://daviruz.meeep.net/gallery2/v/eye-fi-dump/IMG_1796.JPG.html?g2_imageViewsIndex=1
[00:07:44] <DaViruz> aaalmost done
[00:08:05] <DaViruz> stupid three week lead time on the key switch for the jog enable
[00:09:15] <jthornton> how are you doing the feed and spindle feed override?
[00:09:36] <DaViruz> i'm not sure yet
[00:09:49] <DaViruz> i need some kind of analog to digital hardware to get the signal to HAL
[00:10:23] <DaViruz> maybe a usb joystick interface
[00:10:27] <jthornton> I have a patch for halui that does that actually at least for feed override
[00:10:39] <andypugh> It has been done with bit-bashing and pyserial
[00:10:48] <DaViruz> oh.
[00:10:50] <jthornton> does your switch output a binary?
[00:11:10] <andypugh> But we really do need a proper SPI or I2C driver
[00:11:11] <DaViruz> they are potentiometers (at the moment at least)
[00:11:22] <jthornton> oh
[00:11:53] <andypugh> You could gut a gamepad, wire the pots in place of the joysticks, and use hal_input.
[00:12:09] <DaViruz> that's pretty much what i'm thinking
[00:12:29] <andypugh> Or voltage to frequency and track it with an encoder.
[00:12:34] <DaViruz> perhaps a http://www.poscope.com/product.php?pid=3
[00:13:33] <DaViruz> a ADC daughter bord for the mesa cards would be nice
[00:14:10] <jthornton> they have that
[00:14:25] <jthornton> 0-10v or 0-300v
[00:14:29] <andypugh> Better idea than that. Use an Arduino. It's trivial to make one of those output 6 quadrature pulse trains tracking the 6 analogue inputs.
[00:14:36] <DaViruz> yeah but i don't like them
[00:14:37] <DaViruz> :-)
[00:14:45] <jthornton> them who?
[00:15:22] <DaViruz> the mesa voltage-to-frequency converters
[00:15:33] <jthornton> ah ok
[00:17:16] <jthornton> nice panel 16mm buttons?
[00:17:46] <DaViruz> yeah
[00:17:56] <andypugh> If you want to buy an Arduino I can knock you up some code to make it convert voltages to quadrature trains very easily. (this is me volunteering)
[00:18:17] <DaViruz> i'm quite proficient in avr programming if i may say so myself :-)
[00:18:21] <DaViruz> but thanks for the offer
[00:19:02] <DaViruz> jthornton: even the e-stop sits in a 16mm hole, pretty neat little thing
[00:19:13] <jthornton> I have some of those rectangle 16mm's on my Hardinge so they looked kinda familiar
[00:19:51] <DaViruz> we have a bunch of those exact buttons on a machine at work, i always liked the look and feel of them so i got a bunch of them for this
[00:20:09] <jthornton> what don't you like about the Mesa THC card?
[00:20:23] <andypugh> OK, then I will just describe how I have it for my Reslover interface. Basically you can use the index-enable to reset both sides to zero, then count up to the current value, then it just needs to track.
[00:20:25] <DaViruz> i had to take it apart to find the model numbers.. :-P
[00:20:43] <DaViruz> jthornton: they are a bit overkill
[00:20:51] <DaViruz> and i would need two at $69 each
[00:21:04] <jthornton> * jthornton likes overkill when dealing with 300volts
[00:21:19] <jthornton> for you it would be overkill
[00:21:31] <DaViruz> yeah
[00:21:45] <DaViruz> they are probably superb for their intended purpose
[00:21:57] <jthornton> too bad they don't have a card to use pots as inputs
[00:22:09] <jthornton> yea, works nice for my plasma
[00:22:42] <jthornton> ADC never seems to be low cost for some reason
[00:23:53] <andypugh> $20 for an Arduino is pretty cheap for 6 channels and arbitrary procesing/protocol.
[00:24:05] <andypugh> Though it does take work.
[00:24:34] <andypugh> Sorry, I seem to be turning into an Arduino fanboi.
[00:24:45] <jthornton> lol
[00:25:24] <jthornton> how does the Arduino interface with EMC?
[00:26:03] <andypugh> It's a lot like the time I first did some metal-spinning. I suddenly knew how to make lots more things I saw around the place. The Arduino is the same. "Central heating boiler controller dead? No problem, just replace from scratch with something you understand"
[00:26:52] <DaViruz> PLC on the cheap
[00:27:01] <andypugh> Simplest is probably for it to act like a quadrature encoder. You can track the volatage up and down into a quadrature counter.
[00:27:27] <DaViruz> i've built a few small automation systems with just AVRs and custom boards, but the arduino does indeed simplify these sort of things
[00:27:33] <cradek> halui also has increase/decrease inputs for those overrides I think
[00:28:30] <andypugh> Yeah, it isn't any kind of magic, other than a simple IDE and all the right libraries with a board that just plugs into USB.
[00:28:30] <cradek> (but if I wanted this, I'd just put the smarts in halui)
[00:29:02] <DaViruz> i'm not very fond of the IDE though
[00:29:24] <DaViruz> maybe it's just the terminology that gets to me
[00:29:32] <andypugh> We need to beat Seb with (small) sticks until he writes the SPI drivers for Hostmot.
[00:30:06] <DaViruz> that would be a pretty neat way to, say, read my potentiometers
[00:30:49] <DaViruz> maybe i could use a mesa digital-to-analog converter
[00:31:11] <DaViruz> and a comparator that sends digital increase/decrease back to hal until the values match
[00:31:50] <andypugh> For that application software PWM would be fine.
[00:32:07] <andypugh> Same idea, no hardware.
[00:32:15] <DaViruz> i was kidding :/
[00:32:30] <andypugh> I'm not.
[00:32:35] <DaViruz> small servos on the back of the knobs perhaps? :)
[00:33:08] <DaViruz> starting the machines and seeing the override knobs center on 100% automatically
[00:33:41] <andypugh> No, what you said. Pot to Arduino (or similar). increase-decrease lines and a PWM feedback voltage would work perfectly.
[00:34:04] <DaViruz> yeah probably
[00:34:31] <DaViruz> although it could be done with just an opamp in comparator mode
[00:34:38] <andypugh> But do you have to use pots? there is a reason that most people use quadrature pulse generators that just go round and round and round.
[00:34:48] <DaViruz> with a little hysteresis for stability
[00:35:12] <andypugh> I believe you can even use little stepper motor as the encoders.
[00:35:27] <DaViruz> quadrature pulse generators forget where they are if you reboot the thing
[00:35:34] <DaViruz> pots are absolute
[00:35:41] <andypugh> Yes, that is the main advantage
[00:35:53] <DaViruz> maybe two absolute encoders?
[00:36:07] <DaViruz> would eat io lines though
[00:36:16] <andypugh> No, let the software remember. Make the knobs open-loop
[00:37:52] <andypugh> The advantage I mentioned there was that MPGs forget. That means that you don't need any feedback, they work open loop. (Also if they get knocked when the cleaner is dusting, it isn't an issue)
[00:38:29] <cradek> the reason emc wants you to use encoders is because there can be several controls that don't fight (including the one in the gui)
[00:38:54] <andypugh> Well, not truly open-loop, the loop is closed by the operator "yes, that's what I wanted"
[00:39:09] <DaViruz> i haven't actually looked at how hal wants the override signals
[00:39:20] <cradek> quadrature from a knob
[00:39:26] <DaViruz> oh
[00:39:36] <cradek> (or increase/decrease buttons)
[00:39:45] <DaViruz> that renders my front panel scale useless
[00:39:50] <DaViruz> pretty much?
[00:39:51] <cradek> yes
[00:40:09] <DaViruz> this makes me sad
[00:40:30] <andypugh> Not unsurmountably so, but quadrature does work rather better.
[00:41:43] <andypugh> But the panel is so lovely, it is important to make it work.
[00:41:54] <DaViruz> yes, i think so too :-)
[00:43:24] <andypugh> I really do like it a lot.
[00:43:59] <andypugh> Of course, I guess you CNC-ed it, so making it again isn't a vast amount of work?
[00:44:22] <DaViruz> i had a friend make it for me, i don't have a machine that can manage the size
[00:45:36] <andypugh> Simple solution... Switch to encoders with nice big spinner knobs that hide the 0-200 scale?
[00:45:38] <jthornton> if you can get your input into a number you can use the patch I have for halui that overrides the sliders
[00:46:09] <DaViruz> oh
[00:47:21] <DaViruz> i'm guessing a usb joystick interface would be easiest and cheapest
[00:47:37] <jthornton> I use it with my binary feed override knob on the Hardinge
[00:51:26] <andypugh> DaViruz: You can multiplex a lot of your other controls through the joystick if you do that, and save a ton of IO.
[00:56:31] <DaViruz> yeah, i believe the 72 io provided by the mesa card will be plenty though
[00:57:50] <DaViruz> although i distinctly remember saying words to that effect when i once upgraded from a 120MB harddrive to a 200MB one
[01:13:55] <Dave911> andypugh: I saw your patch for the LEDs ... which leds are those ?
[01:14:12] <andypugh> The ones on the Mesa boards
[01:14:31] <Dave911> Are there ones next to the I/O on the 7i37 and so?
[01:15:19] <Dave911> That is the 16 in / 8 out isolated I/O board
[01:16:29] <andypugh> It only drives the ones on the FPGA board itself, as far as I know. In fact I only know for a fact that it works the 8 on a 7i43 as that is all I have
[01:17:48] <Dave911> OK.. I looked closely at the 7i37 and I don't see any on there. I think there are some on the 7i47 though that do not light ... There are some on the 5i20 but I haven't quite figured out what they are for..
[01:18:40] <andypugh> It ought to drive the 5i20 ones.
[01:18:56] <andypugh> In fact, please apply the patch and check it for me.
[01:19:11] <Dave911> I see CR1-8 on the 5i20 I think...
[01:19:34] <Dave911> Does your changes create hal pins for each of those lights?
[01:19:43] <andypugh> You should get 8 new HAL pins: hm2_5i20.led.CR01 etc
[01:20:40] <Dave911> OK.. I'll do that in just a few minutes ... and let you know what I find. It would be nice to be able to use those for diagnostics etc
[01:39:27] <encee> so far, i beat up the pc is never goes above 6800ns in latency-test
[01:39:49] <andypugh> That's really rather good.
[01:39:57] <encee> so intel d510 board + 2x1gb seems pretty win
[01:39:59] <encee> ya
[01:40:06] <encee> er
[01:40:14] <encee> i meant 5811ns
[01:40:22] <encee> never goes above 6us
[01:40:51] <andypugh> That's as good as I have heard of, I think
[01:40:51] <encee> i have some video error tho
[01:40:53] <MattyMatt> DaViruz: is the screen gonna be a touchy or DSiXL?
[01:41:19] <encee> there are 330 atoms with similar times on the wiki
[01:42:11] <encee> mine is a little better, which makes sense
[01:57:59] <andypugh> Goodnight chaps (and chapesses, if there are any.)
[01:58:42] <andypugh> Just one thing: enceem, is that an SMP kernel?
[02:00:43] <andypugh> The Intel Desktop Board D510MO is designed to unleash the power of the Intel Atom processor D510 which supports the revolutionary two-chip layout
[02:01:05] <andypugh> Eh? I remeber when putting the CPU onto one chip was revolutionary.
[02:18:04] <skunkworks> http://electronicsam.com/images/KandT/conversion/computermesa.JPG
[02:18:20] <skunkworks> got it to load the second mesa board and figured out how all the i/o is mapped
[02:18:43] <skunkworks> http://electronicsam.com/images/KandT/conversion/mostio.JPG
[02:19:02] <skunkworks> boy there are still a lot of things to hook up ;)
[02:25:00] <ries> hey Guys, I implemented auto Z from http://wiki.linuxcnc.org/emcinfo.pl?ClassicLadderExamples and this works fine but two problems. Sometimes the z axis seeks upwards instead of downwards, and I have no idea wy.. second, sometimes when my touchplate move a tiny bit (this is woodworking) I get a error of probe touched on MDI and my z doesn't move backup.
[02:26:41] <DaViruz> MattyMatt: touchy. you can control emc with a DSi XL?
[02:27:54] <ries> Ah... on two occasions I also got that it cannot do the probe because junction 2 would be below the limit (or something) only restarting axis would help to solve that... Any body an idea where to search for answers?
[02:29:00] <Dave911> encee: Are you getting that i810 error with the D510 board?
[02:30:02] <Dave911> I believe that error is related to Linux's process of figuring out what monitor it is hooked to
[02:32:22] <Dave911> Some are saying this is a fix for that problem: http://ubuntuforums.org/showthread.php?p=8302145#post8302145
[02:33:28] <Dave911> I ran into the issue when hooking a older monitor to a Intel Atom 330 board ... and turning it on.. Linux got really confused.
[04:31:57] <MattyMatt> DaViruz: it needs a bit of porting :) I was thinking more about running emc on the DSi
[04:32:49] <MattyMatt> but an AXIS clone would be cool enough
[04:33:39] <MattyMatt> 2nd screen mounted somewhere random on the machine with the 3d view
[04:34:59] <MattyMatt> free wifi in the pendant that way
[04:35:25] <MattyMatt> free as in "DSi aren't much more expensive than a touchy"
[04:35:57] <MattyMatt> get a Lite with a broken hinge on ebay :)
[04:36:58] <MattyMatt> the faster cpu in the DSi might make the difference in running the whole emc tho
[04:41:41] <MattyMatt> porting a gtk prog to DS might be instructional for me
[04:43:24] <MattyMatt> I'm monologging
[05:04:21] <skunkworks> heh
[05:04:49] <skunkworks> 'you caught me monologging!' (incredables.)
[05:41:18] <upgrdman> i've been looking into build my own primitive cnc router or mill. looking on youtube, it seems many people just have their threaded rods supported by bearings that a press-fit into some wood. A nut on each side of the bearing keeps the threaded rod in place.... but wouldnt that press-fit bearing just pop out when the machine is under operating loads?
[09:27:37] <MarkusBec_away> MarkusBec_away is now known as MarkusBec
[09:31:02] <piasdom> g'mornin all
[10:24:53] <micges> upgrdman: maybe those machines allows small operation load?
[12:08:30] <ries> hey Guys, I implemented auto Z from http://wiki.linuxcnc.org/emcinfo.pl?ClassicLadderExamples and this works fine but two problems. Sometimes the z axis seeks upwards instead of downwards, and I have no idea wy.. second, sometimes when my touchplate move a tiny bit (this is woodworking) I get a error of probe touched on MDI and my z doesn't move backup.
[12:12:11] <ries_> ries_ is now known as ries
[12:12:26] <JT-Dev> you might have an offset in effect when you try the second time so it is still trying to go to Z-3 but Z-3 is a different place after the offset
[12:14:07] <JT-Dev> I assume your talking about the one button touch off on that page?
[12:16:05] <ries> JT-Dev: yes... Hmmm I think that makes sense (somehow)
[12:16:39] <ries> so it always try to move towards zero, right?
[12:16:48] <JT-Dev> it can be very confusing at first then somewhat confusing after you know more :)
[12:17:15] <JT-Dev> it always tries to move to where you tell to move if it is Z0 then yes
[12:17:17] <ries> and with zero I mean not the machine 0, but the zero of the last touchoff, relative...
[12:17:40] <JT-Dev> yes, AFAIK
[12:17:56] <ries> Ahh ok... not a big deal...
[12:18:10] <ries> what about my debounce problem during retract??
[12:18:16] <JT-Dev> dunno if you can have a G53 on that line but if you can it will always go to the machine corrd with G53
[12:19:05] <ries> I don't have a G53, just saying 'move 20mm above work piece 0
[12:24:04] <ries> for my bounving problem I think I will run that pin through CL and as soon as it wan't to retract I will disable the probe pin
[12:24:38] <JT-Dev> yea, g53 won't work with a probe move
[12:24:51] <JT-Dev> which g38 are you using?
[12:34:25] <JT-Dev> http://www.linuxcnc.org/docview/html//gcode_main.html#sub:G38.2:-Straight-Probe
[12:37:43] <ries> I am using G38.2 Z-10 F200
[12:38:56] <JT-Dev> G38.3 does not give out an error message
[12:41:49] <ries> If you talk about the debounce error, that is when G0 Z20 is programmed
[12:42:00] <ries> so, going down is fine, but going back up in MDI not
[12:45:15] <JT-Dev> does a slower speed help that?
[12:47:47] <ries> I think that would bounce even more because the probe has more thime to contact my touch plate.... but I can try
[12:48:15] <JT-Dev> could be the acceleration causing the bounce
[12:48:45] <JT-Dev> can you put a stronger spring in your touchoff thingy
[12:49:36] <ries> it's currently just a router bit + touchoff, much liek what is shown here : http://wiki.linuxcnc.org/uploads/touch_off.jpg
[12:49:54] <ries> but I will make a more heavy touch plate I think... mine is currently somewhat small
[12:50:34] <JT-Dev> ok, so the plate is bouncing up... I understand now
[12:52:36] <ries> JT-Dev: yeaaa properly a tiny bit... because it's wood, and wood gives a tiny bit...
[12:53:01] <ries> I think it solves my questions.... appart from that when it worked (8 out of 10) it was working great
[13:47:10] <JT-Dev> * JT-Dev wonders why lathe mode in Axis defaults to G17
[14:10:00] <elmo40> that is odd.
[14:10:11] <elmo40> it isn't g18?
[14:10:45] <SWPadnos> axis is the display program, and changes its display setup in lathe mode
[14:10:53] <SWPadnos> emc is the machine controller, and it has no lathe mode
[14:10:59] <SWPadnos> so it still defaults to G17
[14:11:46] <SWPadnos> I think you can add G18 to RS274NGC_STARTUP_CODES if you want
[14:14:31] <piasdom> where are the RS274NGC_STARTUP_CODES ?
[14:14:47] <SWPadnos> it's an ini file key
[14:15:02] <SWPadnos> it may be CODE, not CODES though, I don't recall
[14:15:12] <piasdom> thanks
[14:15:17] <SWPadnos> sure
[14:15:19] <SWPadnos> bbl
[14:30:22] <JT-Dev> I didn't think of the difference between the display and the machine controller
[14:30:38] <JT-Dev> not on top of my game this morning with this head cold
[14:32:07] <JT-Dev> seems like the lathe.ini would have G18 in the startup codes in the sample configs
[14:42:09] <cradek> if you find that works, I agree
[14:43:20] <cradek> I'm suspicious that M2 will still put it back to G17 though, and if so, the user still needs to have G17 in the gcode, and putting it in the ini would make the problem worse because G18 in gcode would be needed inconsistently
[14:43:25] <JT-Dev> seems to work here
[14:43:39] <JT-Dev> ah didn't think of that
[14:44:25] <cradek> er, I meant 'user needs to have G18 in the gcode'
[14:46:31] <JT-Dev> and indeed M2 does restore it to normal mill plane G17...
[14:47:00] <cradek> in that case, I think it's best not to put it in the ini
[14:47:10] <JT-Dev> yup
[14:47:51] <JT-Dev> I just need to train my post processor fingers to put it in every time lol
[14:48:09] <cradek> ha
[14:48:28] <sealive> hi to all
[14:49:11] <sealive> isent there in the docu the part where i can find how to put my own ngc to the start
[14:50:56] <JT-Dev> in the Integrator manual in the config section
[14:51:15] <sealive> thanks
[14:51:33] <JT-Dev> beware of M2 resetting some things...
[14:53:24] <sealive> i use M30
[14:56:04] <sealive> i think this ist the section http://www.linuxcnc.org/docview/html//config_ini_config.html#sub:[DISPLAY]-section
[14:56:53] <JT-Dev> http://www.linuxcnc.org/docview/html//config_ini_config.html#sub:[RS274NGC]-section
[14:57:04] <sealive> OPEN_FILE = /full/path/to/file.ngc
[14:57:21] <JT-Dev> oh the splash file yes
[14:59:31] <sealive> i have mostly 10-50 parts of eatch milling
[14:59:57] <sealive> so it woudt be nice to come up with this programm as it starts
[15:44:22] <frallzor> hello chiiildren
[15:44:30] <sealive> hi
[15:45:06] <MattyMatt> it's vader abraham. sing us a smurf song
[15:46:03] <frallzor> * frallzor throws a pony at MattyMatt
[16:02:13] <piasdom> i don't have a "RS274NGC_STARTUP_CODE" section in my ini.. should there be?
[17:37:05] <Jymmm> Morning
[17:38:01] <piasdom> g'mornin
[17:42:58] <sealive> god evening from germany
[18:08:51] <frallzor> first cut in the fastened table, woho!
[18:09:11] <sealive> :D
[18:09:21] <frallzor> (intended cut)
[18:09:28] <frallzor> in yer face! :P
[18:57:33] <isssy> hi all
[19:07:56] <jepler> jepler has changed the topic to: Welcome! EMC (Enhanced Machine Controller) is a linux-based opensource CNC control. | Latest release: EMC 2.4.0 | http://www.linuxcnc.org | http://wiki.linuxcnc.org
[19:15:30] <frallzor> why is emc bitching about Y-axis limits being too small for the work when I set them like 5x longer than the value in the gcode?
[19:16:11] <frallzor> want to mill 1650mm, it complained, changed to 10000mm and -10000mm
[19:16:14] <frallzor> still complains
[19:17:23] <JT-Dev> do you have some offsets in effect that might be stacked up?
[19:17:34] <frallzor> nope
[19:17:47] <frallzor> just homed
[19:17:54] <MattyMatt> I was getting that when I had the sign of the axis wrong
[19:18:09] <ries> frallzor: it just corssed my mind...
[19:18:13] <ries> tru a G21 (set mm)
[19:18:15] <ries> try..
[19:18:30] <ries> I have one or two occasions when 'some' code set the machine to inches
[19:18:39] <ries> you should see that in your display though
[19:19:19] <frallzor> its all mm
[19:19:21] <MattyMatt> I was also getting problems until I learned to touch off properly, with jobs with lots of y<0
[19:20:06] <frallzor> well it shouldnt matter if the length of the cut is 1650mm and i set limits to 10000mm and -10000mm lots of spare room to fail in
[19:21:46] <fenn> gratz on the big 2.4
[19:23:59] <frallzor> ill just blame artcam until I tried the same code in another software =)
[19:28:45] <JT-Dev> frallzor: is the code small enough to pastebin?
[19:31:36] <frallzor> yup
[19:31:41] <frallzor> believe so
[19:31:58] <frallzor> http://pastebin.com/2Et5UdzE
[19:34:19] <frallzor> just a simple gode for leveling a table
[19:40:15] <JT-Dev> your machine has 106" of x and 65" of y travel?
[19:40:28] <frallzor> yup
[19:40:37] <frallzor> and a little more
[19:40:37] <JT-Dev> from the center out?
[19:40:52] <frallzor> from 0,0
[19:41:22] <frallzor> say bottom left corner
[19:41:40] <frallzor> ^Y X>
[19:42:19] <JT-Dev> did you home to the bottom left corner?
[19:42:23] <frallzor> yup
[19:42:27] <frallzor> of course
[19:42:39] <JT-Dev> just checking
[19:44:17] <frallzor> im betting its an artcam issue
[19:44:42] <JT-Dev> how is that?
[19:45:05] <JT-Dev> the code looks fine on my sim
[19:45:26] <JT-Dev> I just need to open up the limits to run it
[19:45:55] <frallzor> can one force emc to ignore limits?
[19:46:25] <JT-Dev> I'm not sure
[19:46:42] <JT-Dev> on your back plot does it show the paths inside your limits?
[19:47:03] <frallzor> its outside I think
[19:47:18] <frallzor> but all limits are set to 10000mm and -10000mm
[19:48:29] <frallzor> but It does seem like the old limit of 1630mm is stuck somehow when I think about it
[19:49:44] <frallzor> i edited the stepconf manually, could that be the reason?
[19:50:36] <JT-Dev> did you run stepconf after?
[19:50:40] <JT-Dev> then yes
[19:50:53] <frallzor> didnt run it after
[19:51:05] <SWPadnos> wait, you edited the stepconf xml file?
[19:51:08] <frallzor> yes
[19:51:20] <frallzor> useless? :P
[19:51:21] <SWPadnos> then you have to re-run stepconf to get the changes to propagate to the ini/HAL files
[19:51:28] <SWPadnos> no, but you do need to re-generate the EMC2 config
[19:51:39] <frallzor> ah I see
[19:51:39] <SWPadnos> the xml is input to stepconf, the ini/hal files are the output
[19:51:58] <frallzor> then I guess I have to backup my hal-files that I edited
[19:52:12] <JT-Dev> would have been easier to just run stepconf again and pick your config as edit a config
[19:52:28] <frallzor> but that still kills settings in the files you add manually
[19:52:34] <JT-Dev> or just change your ini file and be on manual edit from now on
[19:52:47] <frallzor> you can do that?
[19:52:53] <JT-Dev> yep
[19:53:09] <SWPadnos> stepconf can be used once to get a base, then you can edit the files as much as you want
[19:53:25] <JT-Dev> just don't run stepconf again on that config
[19:53:29] <frallzor> how to make it manual?
[19:53:31] <SWPadnos> you can also save the edited files, and use programs like diff and patch to maintain your changes to them
[19:53:53] <SWPadnos> so you can make some changes in stepconf, and "merge" your own edits into the generated files
[19:55:51] <JT-Dev> just edit the ini and hal files with gedit
[19:56:29] <SWPadnos> right. then (a) save the files and (b) save a diff between the stepgen-generated ones and the edited ones (ie, a patch with your changes)
[19:56:33] <frallzor> which file stores limits and suck?
[19:56:39] <frallzor> *such
[19:56:45] <frallzor> hal or ini?
[19:56:53] <SWPadnos> then, when you blow away your changes by re-running stepconf for some reason, you can more or less patch the newly created files with your chagnes
[19:56:56] <SWPadnos> changes
[19:57:01] <SWPadnos> ini
[19:57:03] <JT-Dev> the ini file
[19:57:32] <frallzor> well this is a rare occasion =) just need to change my limits by 30mm so I can live with some hassle
[19:57:36] <JT-Dev> http://www.linuxcnc.org/docview/html//config_ini_config.html#sub:INI-Detail
[19:59:15] <frallzor> ty =)
[20:01:36] <JT-Dev> Wee! the first arc Quadrant right in 5 tries at hacking my arc buddy to arc Lathe G18 Buddy
[20:02:53] <JT-Dev> wow two right out of 6
[20:04:20] <JT-Dev> last one was hosed LOL
[20:36:29] <krushia_> krushia_ is now known as krushia
[21:02:04] <skunkworks> did anyone see this? http://www.youtube.com/watch?v=wxwgzj4KmCE
[21:03:28] <cradek> neat
[21:03:30] <cradek> looks fast
[21:03:36] <skunkworks> yes
[21:05:42] <JT-Dev> mixed tabs and spaces can be a royal PITA
[21:05:48] <JT-Dev> in python
[21:29:07] <MattyMatt> handy for your battery powered lathe -> http://www.robotshop.com/dimension-engineering-sabertooth-2x25-6.html
[22:03:24] <JT-Hardinge> in a M100 bash file I need to setp xx 1, pause 1/2 second or so the setp xx 0 then exit0 how can I do the pause
[22:03:43] <JT-Hardinge> I put sleep 1 s in there but it blew right past that I think
[22:08:31] <morfic> try with a longer sleep time to be sure?
[22:08:50] <JT-Hardinge> ok
[22:22:31] <MattyMatt> arrg. I'm gonna encode G41 and G42 in the object name for now
[22:22:55] <MattyMatt> fire and forget
[22:45:49] <JT-Hardinge> the most amazing thing about EMC is it does "exactly" what you tell it to do... even if that is not what you wanted it to do :)
[22:50:44] <morfic> JT-Hardinge: sounds like emc is a "real cnc" then :P
[22:57:28] <ries_> ries_ is now known as ries
[23:00:47] <JT-Hardinge> real as it gets!
[23:08:28] <morfic> JT-Hardinge: it's one of the most used lines newguy: "man, it just crashed, i don't know 'what the machine did'" Me: "No, the machine does exactly what we tell it to, if it crashed, we messed up" (of course that's not true if it's some fadal prototype running on win2k embedded, then all bets are off)
[23:11:54] <JT-Hardinge> lol
[23:17:20] <morfic> oh btw, looks like we bought a Mori SL35, but i know nothing more than that
[23:17:41] <andypugh> Does anyone know of an opto-coupler with a 200kHz frequency cut-off, ie one that will accurately pass-through a 20kHz PWM signal?
[23:18:17] <andypugh> The ones I have turn the PWM into something rather more curvaceous/
[23:20:05] <renesis> its prob the transistor side the emitter should be fast enough
[23:20:47] <andypugh> I think I am seeing the difference between £2 optos and £0.2 ones...
[23:21:48] <renesis> yeah i think current transfer is the big $$$ spec, prob has a lot to do with the speed
[23:22:39] <renesis> current through the emitter side has a lot to do with it
[23:23:00] <renesis> 20khz isnt that fast
[23:24:08] <andypugh> The 20kHz gets through as 20kHz, but it ceases to be recognizably a PWM
[23:24:13] <renesis> cheap ones can prob do it with alot of input current (>10mA) if you find the right load
[23:24:22] <renesis> yeah itll take the edges off
[23:24:43] <renesis> prob why they get used a lot in audio compressor feedback
[23:25:03] <andypugh> The ones I have are 5uS rise/fall.
[23:25:11] <renesis> oh
[23:25:26] <andypugh> If I spend more money I can get 6nS ones.
[23:25:34] <renesis> how much more?
[23:26:20] <renesis> 5us can maybe do if you have an input thats really consistent about logic switching levels
[23:26:31] <andypugh> £3.67 each rather than £0.16 each.
[23:26:35] <renesis> yeah
[23:26:50] <renesis> how much can you drive the led?
[23:27:02] <renesis> this from a parport? prob a few mA or something?
[23:27:07] <renesis> thats prob the big problem
[23:27:16] <andypugh> It's from an Arduino at the moment.
[23:27:29] <andypugh> I am tempted just to wire direct.
[23:27:31] <renesis> yeah dont do more than a few mA
[23:27:42] <andypugh> But there is 300V DC on the board the signals are going to.
[23:27:45] <renesis> its close to max spec for the avr
[23:28:01] <renesis> how many channels?
[23:28:11] <andypugh> 6
[23:28:23] <renesis> yeah you cant load up the port like that
[23:28:45] <renesis> you can get away with one or two at 10mA, but not 6, and i wouldnt call it sane
[23:28:54] <andypugh> I have 1k resistors on each opto.
[23:28:57] <renesis> just get some 2n4401 or something
[23:29:14] <renesis> yeah so like 3mA
[23:29:29] <renesis> thats safe but thats why the opto output is so curvy
[23:29:46] <renesis> the leds in the opto are like 50mA max or something
[23:29:56] <renesis> theyre like 2x normal little leds
[23:31:00] <renesis> but yeah get some logic fets or 2n4401 and drive the led at like 25mA or half of whatever the absolute max is
[23:31:04] <renesis> prob gets a lot better
[23:31:57] <andypugh> I have just remembered, I am not driving directly from the AVR, I have a 74-244 in there so they can share a common PWM clock.
[23:32:30] <renesis> if this is on breadboard keep the wires from the arduino to the transistor to the opto led short, any extra capacitance is gonna slow switching of the led and curve the output
[23:32:55] <renesis> yeah same deal, those chips are usually like 10ns or something
[23:33:07] <renesis> thats a tristate or something?
[23:33:24] <renesis> anyway same deal, driving direct you cant source so much current
[23:33:48] <andypugh> Tristate buffer, 500nS switching time.
[23:33:51] <renesis> you need a buffer so you can put like 20-40mA into the opto, i bet that works
[23:34:14] <renesis> yeah i dont think thats a big deal except the current output limits
[23:34:33] <andypugh> The PWM out of that is fine on the scope, nice and square. As is the input to the optos. I am pretty sure that it's the opto itself causing the problem.
[23:34:53] <renesis> its because of the current though
[23:35:09] <renesis> the switching time isnt a problem
[23:35:12] <andypugh> I am wondering if I can just skip the optos. It isn't like the Arduino was expensive.
[23:35:17] <renesis> just need to turn the led more on
[23:35:27] <renesis> naw dont
[23:35:35] <renesis> just need a $.05 transistor
[23:36:43] <andypugh> Well, 6 of them, and board space.
[23:36:54] <renesis> sot23!
[23:37:12] <renesis> http://darkertechnologies.com/image/ph2dmm__pcb_20100509.png
[23:37:16] <renesis> \o/
[23:37:25] <pfred1> the optos I'm using I've found I'm best off driving the detectors pretty hard not the led side
[23:37:49] <renesis> yeah it has a lot to do with the load on the output
[23:38:04] <andypugh> The datasheet says that mine work a lot better with a 100R load than a 1k
[23:38:26] <pfred1> but who knows maybe I've mine wired all wrong?
[23:38:40] <andypugh> Page 5, http://docs-europe.origin.electrocomponents.com/webdocs/0c67/0900766b80c67606.pdf
[23:38:48] <renesis> its npn and led wire em how you want, heh
[23:38:56] <pfred1> the traces still look godo to me they do what i want now
[23:39:23] <renesis> yeah 50mA input
[23:39:46] <renesis> 50mA output
[23:39:49] <pfred1> I just use regular 330 ohm dropping resistors at 5 volts on the input side its fine
[23:40:01] <pfred1> which translates to something like 12 ma
[23:40:09] <renesis> yeah thats 10mA or so, thats better than 3mA
[23:40:40] <pfred1> but I noticed if I didn't pull the out pretty damned low I'd get saggy waveforms
[23:40:53] <pfred1> or i didn't satisfy TTL high and low or something
[23:41:11] <andypugh> Looks like I need to order some 100R bussed resistors then.
[23:41:18] <renesis> yeah i always end up messing with the load on the output
[23:41:43] <renesis> you dont have resistors lying around you can test with?
[23:41:50] <pfred1> andypugh I'm using toshiba TLP 521 optos now they're not bad
[23:41:59] <pfred1> andypugh they're pretty fast
[23:42:13] <renesis> haha @ 2v rise/fall test condition
[23:42:17] <renesis> i wonder if thats normal
[23:42:26] <pfred1> some optos are crazy slow
[23:42:54] <renesis> isolation voltage and led efficiency prob why theyre so diff
[23:43:45] <andypugh> pfred1: Mine are only a little slower than the toshiba ones.
[23:43:48] <pfred1> and give crappy rises and falls like you end up with more of a ramp or sine wave out of them than a TTLish signal
[23:44:20] <andypugh> What I was missing was that 100R load is normal, not 1k.
[23:44:39] <pfred1> these toshibas I don't even have to run through a schmitt trigger to clean up though i still am
[23:45:19] <pfred1> well I'm down to 220 on my clock for the detector
[23:45:29] <pfred1> I still run 330 on the direction line
[23:45:48] <pfred1> but the clock line I wanted a bit more drive on
[23:46:30] <pfred1> I think theoretically i could run down to 120 ohms and still be below max current
[23:46:52] <pfred1> but i don't like ot push semis to their absolute limits if i cna help it
[23:47:40] <pfred1> andypugh i did spend some time and measuring to get all of my opto stuffs all sorted out
[23:48:10] <andypugh> I have found a nice 4-channel, logic output, 30nS one. But it is £58
[23:48:45] <pfred1> right now my max clock is 45 us so I don't need that sort of speed
[23:49:38] <pfred1> but in a lot of applications optos being slow is a factor
[23:50:29] <andypugh> Unfortunately all the fast ones seem to be 8-pin, and I don't have the room.
[23:50:37] <pfred1> I can't get EMC to give me any faster a clock than 45 us
[23:50:53] <pfred1> yeah mine are dulie 8 pinners
[23:51:04] <pfred1> but I need step and dir so it works out
[23:52:44] <andypugh> I would have thought that there would be an 8 - way logic-output device out there for data bus use.
[23:53:49] <pfred1> andypugh nah they just have a whole row of optos if they need them
[23:54:14] <pfred1> they usually just use the onesies all next to each other in sockets
[23:54:37] <pfred1> making a great wall of optos in effect
[23:55:10] <pfred1> I guess that comesi n handy if you blow one and need to change it too
[23:55:30] <pfred1> the way I wire mine I end up with a signal inversion
[23:55:44] <pfred1> something you may want to consider
[23:55:45] <andypugh> I have found a dual 45nS one for £2.62... But it's out of stock
[23:56:16] <pfred1> not so sure how you even could wire one not to get an inverted signal out of it
[23:57:22] <pfred1> but i run it through a schmitt inverter anyways so its all good
[23:58:01] <pfred1> this is how i typically use optos on my motor drivers: http://img695.imageshack.us/img695/5841/tb6560ahqa.png
[23:58:54] <pfred1> that blocks the noise
[23:59:56] <pfred1> "It Works" (TM)