Back
[00:50:11] <Loetmichel> mornin'
[01:39:06] <WalterN> any good 3D CAD programs for linux around?
[01:40:49] <WalterN> Qcad... has that actually improved to be useable these days?
[01:41:39] <WalterN> now called librecad evidently
[02:14:50] <DJ9DJ> moin
[06:03:22] <jthornton> damm 250 spammers registered last night! WTF?
[06:29:32] <archivist> jthornton, has the forum been updated yet?
[06:29:49] <jthornton> not that I know of
[06:32:03] <archivist> if those that can dont nor give powers to others, seems a difficult situation not being resolved
[06:33:13] <jthornton> yup
[06:33:18] <archivist> I have experience fixing databases but no powers :(, I couls do a test upgrade here if I had the data
[06:34:22] <jthornton> yea, you need the Dream Host account
[06:34:53] <archivist> I need a backup of the data rather than the account
[06:35:25] <archivist> who keeps backups anyway
[06:35:33] <jthornton> dunno
[06:36:04] <jthornton> I don't have access to the DH account
[06:37:26] <jthornton> I'm sure alex_joni, cradek and SWPadnos have access not sure if anyone else does
[06:41:43] <r00t4rd3d> its stupid simple to update Joomla too
[06:41:50] <r00t4rd3d> couple clicks
[06:42:29] <archivist> but now its jump n versions so may be less simple
[06:42:31] <jthornton> nice to make a back up before clicking
[06:42:53] <r00t4rd3d> arch, nope they have a plugin, updates it from 1.0 to tht newest.
[06:43:37] <r00t4rd3d> http://extensions.joomla.org/extensions/core-enhancements/installers/9332
[06:43:40] <archivist> and plugins that write to its own directory is a well known security problem for any software
[06:43:52] <Thetawaves> is there a chance to move away from joomla?
[06:44:29] <archivist> for what definition of chance
[06:44:35] <Thetawaves> no wonder it broke, it's joomla
[06:44:39] <jthornton> right now the joomla version we use is 1.5.25 and current version is 3.0.1
[06:44:43] <Thetawaves> hehe :)
[06:45:07] <jthornton> Thetawaves, not without the keys
[06:45:49] <Thetawaves> is it a real root server?
[06:46:02] <Thetawaves> can you ssh in and restart mysql?
[06:46:04] <jthornton> what is that?
[06:46:31] <jthornton> I can't, I only have access to the front end
[06:46:36] <Thetawaves> if you can restart mysql, you can root it :-/
[06:47:15] <r00t4rd3d> committing a felony to update the forums seems...... dumb.
[06:47:39] <Thetawaves> why would it be a felony?
[06:47:48] <Thetawaves> it's not against the law to root your own shit
[06:48:16] <r00t4rd3d> well someone does own it techincally
[06:48:21] <r00t4rd3d> but no one here
[06:49:25] <Thetawaves> why don't you send that person an email :P
[06:50:21] <archivist> those people are lurking and their ears will be burning :)
[06:50:33] <r00t4rd3d> rehenry@fast-air.net or stevesng@NEWSGUY.COM
[06:51:23] <r00t4rd3d> someone who knows those guys should email them and give them the link to the Joomla plugin updater
[06:52:40] <r00t4rd3d> http://www.networksolutions.com/whois-search/linuxcnc.org
[06:52:50] <r00t4rd3d> their phone numbers are listed, call them JT
[06:54:09] <r00t4rd3d> "Hi this is BIG JOHN T could you updates our forums plz, kthnxbi."
[06:54:23] <jthornton> wow I have not seen Ray Henry on here in years
[06:55:00] <r00t4rd3d> so you friends with him?
[06:55:05] <r00t4rd3d> email his ass
[06:55:34] <jthornton> I've talked a couple of times on here to him
[07:01:29] <jthornton> dang the BP computer does not show up here, can you refresh the lan from ubuntu somehow?
[07:02:50] <jthornton> ah there it is
[07:05:04] <r00t4rd3d> I emailed them with the plugin link and the plugins makers instructions how to use it.
[07:07:47] <r00t4rd3d> I honestly cant believe someone on the BOD who has access cant spend a few minutes and update it.
[07:15:18] <r00t4rd3d> http://linuxcnc.org/index.php/english/component/kunena/?func=view&catid=36&id=25740#25740
[07:15:19] <r00t4rd3d> :D
[07:15:35] * r00t4rd3d waits....
[07:37:52] <r00t4rd3d> http://i.imgur.com/scWCN.gif
[07:48:23] <jthornton> r00t4rd3d, I don't think anyone with the keys reads the forum...
[07:49:40] <r00t4rd3d> alex does
[07:50:37] <jthornton> he's the man then
[07:51:25] <r00t4rd3d> Last Online: 2 months ago
[07:51:28] <r00t4rd3d> lol
[07:54:08] <jthornton> are you holding your breath?
[07:55:02] <r00t4rd3d> seems like epler has high access too
[07:56:05] <jthornton> last time online 6 months ago
[08:03:36] <jthornton> wow today sunny high of 85f and Saturday sunny and high of 55f
[08:14:51] <r00t4rd3d> have you seen the pictures of that high speed Canadian police chase?
[08:14:57] <r00t4rd3d> http://i.imgur.com/J8ack.gif
[08:41:57] <tjb1> Guys look at this "Freakin' huge" cnc router!!
http://hackaday.com/2011/09/04/freakin-huge-cnc-router/
[08:42:07] <tjb1> Another reason I am starting to hate HaD
[08:43:47] <archivist> some amateurs have no clue about size :)
[08:43:59] <r00t4rd3d> size matters
[08:44:51] <archivist> yup, seems I have a job for tomorrow doing a wide load escort job
[08:45:29] <tjb1_> What kind of load
[08:45:58] <archivist> a mobile home
[08:46:07] <r00t4rd3d> he is taking his GF to McDonalds.
[08:46:39] <archivist> gf...neva /me shall remain old free and single
[08:46:40] <tjb11> webchat isnt so hot
[08:46:58] <r00t4rd3d> chatting on a iphone sucks
[08:47:47] <tjb11> im about to switch to my iphone if webchat keeps dropping me
[08:49:32] <tjb1> And im on phone...
[08:51:08] <tjb1> But yeah wasnt that router huge?
[08:51:26] <tjb1> Makes r00t4rd3d look like a toy
[08:53:11] <tjb1_> Seriously?
[08:58:46] <tjb1> I should send my plasma to hack a day...."Freakin absolutely massive" plasma cutter
[09:00:32] <tjb1> But then my design would get ridiculed by all of the makerbot idiots that are now professionals in cnc
[09:00:59] <tjb1> Or the reprap ones
[09:01:16] <L84Supper> heh, exactly
[09:01:51] <L84Supper> yet they are incapable of any reliable design
[09:02:41] <tjb1> Yep...why are you using r&p on a 10' axis? Direct belt drive is much better
[09:02:55] <r00t4rd3d> Took some glamour shots of my box:
http://imgur.com/a/4biWi
[09:03:29] <L84Supper> nobody even makes a decent extruder, they all look like frankenstruders for school science projects
[09:03:30] <tjb1> Or why arent you using all-thread on your z axis, its much cheaper than acme
[09:03:50] <tjb1> Well the qu-bd extruder is nice
[09:04:48] <tjb1> r00t4rd3d: Thats sexy but I still think the knob is too big
[09:06:15] <L84Supper> freaking huge = 12" x 18", ginormous = 24" x 48", Freakin absolutely massive = 48" x 96"
[09:06:53] <L84Supper> pretty sure thats the SAE standard :)
[09:09:17] <tjb1> Yep ;)
[09:11:58] <tjb1> Or all the toolbags on HaD that think they are gonna strap a dremel to their 3d printer and mill stuff
[09:13:37] <tjb1> Im gonna route stuff with my little 1/8" belt drive
[09:15:52] <L84Supper> the general aim seems to be to design printers that are partially printable and the rest has to be cobbled together with materials only available at Ace hardware or similar
[09:17:05] <L84Supper> so most end up looking and working like erector sets or metal tinker toy creations
[09:17:33] <L84Supper> but you can't say that since it's become some sort of religion :)
[09:21:38] <tjb1> Yeah, one piece of crap prints parts to make another piece of crap
[09:21:57] <tjb1> Quality goes down every time lol
[09:23:17] <L84Supper> the only progress has been ARM m3 type all-in-one control boards, but they are $150-200 ea
[09:24:01] <tjb1> Only design I like are the ORD builds, hadron ord is the one I am going to build
[09:24:13] <L84Supper> that can be <$75
[09:25:22] <tjb1> Whats <75?
[09:25:30] <L84Supper> heh, Kelling started selling those
[09:26:15] <L84Supper> http://www.automationtechnologiesinc.com/products-page/3d-printer/3d-printer-mechanical-platform-no-controller
[09:26:30] <L84Supper> Kelling is Automation Technology now
[09:26:47] <L84Supper> he imports from China
[09:27:19] <tjb1> Only one not full of plastic crap
[09:27:49] <L84Supper> tjb1, the ARM M3 reprap control boards could be <$75, they are currently $150-200
[09:31:48] <Guest71224> hi guys! I'm trying to setup my PID parameters but it seems very difficult!! Anyone has a strategy for overcoming the problem? (Stepper motor with close loop -linear encoder-)
[09:33:03] <cradek> it's very hard to tune steppers with pid. they are very mismatched, since their torque goes down as you speed up. do you have any backlash in the system?
[09:33:56] <Guest71224> I have no significant backlash
[09:34:31] <jthornton> I've heard tell that someone did something like that and that the step count had to be double or more to the encoder count or something like that
[09:35:00] <Guest71224> I am working in velocity control, is it right? Or I have to switch to position control mode?
[09:35:11] <tjb1> jthornton: Your site is down i think
[09:35:23] <jthornton> IIRC they used velocity
[09:35:39] <jthornton> tjb1, it's been doing that this morning off and on
[09:35:55] <jthornton> back on now
[09:36:08] <Guest71224> Are there signifcant difference between velocity and position controls?
[09:36:53] <Guest71224> I have a lot of troubles with P I D FF0 FF1 etc...
[09:37:55] <cradek> I bet you will want to depend heavily on FF1=1
[09:38:33] <cradek> that and a bit of P? or I?
[09:38:35] <Guest71224> now ff1 is equal to "0"
[09:38:46] <cradek> you surely want to use FF1
[09:38:58] <pcw_home> Yes P = small FF1 = 1 nothing else
[09:39:19] <cradek> FF1 gives output proportional to input's derivative
[09:39:25] <tjb1> I still cant load it jthornton
[09:39:26] <pcw_home> This is guaranteed anythin else is simply wrong
[09:40:05] <jthornton> tjb1, not much I can do about it
[09:40:16] <pcw_home> stepgen with encoder feedback is a nearly perfect velocity mode servo (within .1 %)
[09:40:22] <tjb1> Just lettin ya know ;)
[09:40:36] <jthornton> I just checked and no problems
[09:40:44] <Guest71224> but if FF1 is linked to the derivative parameter, what can it does if D is set to 0?
[09:41:14] <cradek> D gives output proportional to derivative of ERROR. it is totally different from FF1.
[09:41:29] <pcw_home> are you sure you have the stepgens scale correct? (other wise the FF1 = 1 wont help)
[09:41:49] <Guest71224> my situation is this : P = 8 I = 0 D = 0 FF0 = 0 FF1 = 0 FF2 = 0
[09:42:08] <Guest71224> yes, the scale is correct
[09:42:50] <pcw_home> As I suggested before you shoul dget it to work almost perfectly with only FF1
[09:43:03] <pcw_home> if you cannot, the scale is wrong
[09:43:36] <tom3p> cad-schroer new demo 2D views to 3D
http://www.cad-schroer.com/Software/MEDUSA4/CADFreeware/1/1420/r/193/?pk_campaign=nd_08_eS
[09:43:39] <pcw_home> There are 2 scales involved
[09:43:43] <tom3p> also ver 5 free demo is out
[09:44:05] <Guest71224> the system seems to be slow and it stops if ferror and fminerror are setted to values near 1
[09:44:44] <pcw_home> are the PID output values limited?
[09:44:51] <Guest71224> yes, to 333
[09:45:08] <pcw_home> why?
[09:45:13] <cradek> what are the units of the pid output?
[09:45:19] <Guest71224> that is 20 meters per minutes, is it wrong?
[09:45:24] <pcw_home> mm/s
[09:46:04] <Guest71224> I think the unit of the control mode (mm/1')
[09:46:14] <Guest71224> yes yes
[09:46:15] <Guest71224> mm/s
[09:46:19] <Guest71224> sorry
[09:47:07] <pcw_home> what is the machine max velocity?
[09:47:34] <Guest71224> i try to use only FF1, just a second!! :)
[09:47:54] <Guest71224> below 20 m/1'
[09:49:33] <Guest71224> using only ff1 the system seems to have an error near 0.2mm
[09:50:54] <pcw_home> for how big a move?
[09:51:02] <Guest71224> 100mm
[09:51:27] <Guest71224> moving from 200 to 100 it makes 0.2 of error
[09:51:51] <cradek> sounds very promising to me
[09:52:30] <pcw_home> so it you add a tiny bit of P (say .1) what happens
[09:53:10] <Guest71224> so what kind of correction I have to do in your opinion?
[09:53:19] <Guest71224> ok
[09:53:31] <Guest71224> i try
[09:53:32] <Guest71224> :)
[09:54:58] <Guest71224> it makes about .1 of error and then it goes very slowly toward the exact measure
[09:55:56] <pcw_home> well now try P of 1
[09:56:07] <Guest71224> ok
[09:58:34] <Guest71224> 0.2 mm of error, this time the system go on the exact measure and then come back in order to compensate that error
[09:59:54] <pcw_home> not understanding, you mean it overshot?
[09:59:59] <Guest71224> P = 0.1 works well than P = 1
[10:01:08] <Jymmm> Is this 365pF EACH section?
http://www.tubesandmore.com/products/C-V365-X3
[10:01:09] <Guest71224> yes but it is not exact, the problem is that the original error (0.2 with P=0) has been reduced by setting P = 0.1 and than it rise again when I set P = 1
[10:01:42] <pcw_home> Thats very odd
[10:01:54] <Guest71224> so P = 0.1 seems to work well than P = 1
[10:02:38] <pcw_home> Are you sure you dont have some backlash or mechanical problem in the system?
[10:02:53] <Guest71224> yes, sure
[10:02:58] <Guest71224> :(
[10:04:02] <Guest71224> probably the error is still the same, the only difference is that with P = 1 the exact measure is achieved faster.
[10:04:30] <pcw_home> are you jogging anywhere near the maximum velocities?
[10:05:49] <skunkworks> I think a look at the ini and hal files might be in order...
[10:06:11] <Guest71224> yes, I'm working with the ini file!
[10:06:44] <pcw_home> Yes that would help Lots of people use this arrangement and its usually simple to tune
[10:07:17] <skunkworks> http://pastebin.com/ them
[10:07:19] <Guest71224> I have done an experiment, starting from 0 i have moved to 100 than 200, 100 and finally 0 again
[10:07:48] <Guest71224> 0->100 error 0.1
[10:08:01] <Guest71224> 100 ->200 error 0.3
[10:08:11] <Guest71224> 200 -> 100 error 0.1
[10:08:18] <Guest71224> 100 -> 0 error 0.3
[10:09:19] <cradek> do you have stepgen max accel and vel higher than the ini maxaccel and vel?
[10:09:56] <cradek> have you plotted ferror during the move, or do you just look at the end of the move? sharing a plot with us would tell us a lot more.
[10:09:57] <MattyMatt> jymmm generally yeah. those ganged ones are for superhets etc so you can tune 2 stages
[10:10:02] <Guest71224> it seems that the correction is applied only to the last part of the path, near the end
[10:10:08] <MattyMatt> or 3 in that case
[10:10:38] <Guest71224> how can I plot ferror??
[10:10:48] <pcw_home> halscope
[10:11:42] <Guest71224> halscope, sorry, I'm newbie :)
[10:11:54] <Guest71224> now I try!!
[10:13:19] <Guest71224> where i can find ferror? within pins, signals or parameters?
[10:14:19] <Jymmm> MattyMatt: ah, ok. ty
[10:16:03] <awallin> Guest71224: the following error is usually the difference between command and encoder inputs to a pid loop
[10:16:26] <awallin> Guest71224: I'm not sure the maximum allowed following error is available from the motion controller...
[10:16:41] <Guest71224> I've found it!
[10:16:44] <Guest71224> :)
[10:16:59] <Guest71224> the problem now is how to read the graph!
[10:17:00] <Guest71224> :)
[10:17:19] <Guest71224> what measure unit are these?
[10:17:30] <Guest71224> 20m/div?
[10:17:34] <awallin> do g1 or g0 moves back and forth and trigger the scope appropriately..
[10:17:43] <awallin> that's "milli"
[10:17:56] <awallin> so 0.020 mm/division of the scope
[10:17:57] <awallin> I think
[10:18:27] <awallin> Guest71224: I have some screenshots in the blog
http://www.anderswallin.net/?s=pid+tuning
[10:18:46] <Guest71224> okok
[10:18:47] <Guest71224> :)
[10:19:18] <JT-Shop-2> nice page awallin
[10:20:13] <awallin> JT-Shop-2: thanks, some case-studies in the linuxcnc docs would be nice. With different servo setups preferably (i.e. torque, speed mode etc)
[10:21:06] <JT-Shop-2> someone that knows what they are talking about would have to do that LOL
[10:22:14] <Guest71224> you are rigth! :D
[10:22:29] <JT-Shop> http://imagebin.org/233106
[10:22:45] <JT-Shop> http://imagebin.org/233107
[10:23:03] <JT-Shop> http://imagebin.org/233108
[10:23:39] <awallin> did you hit the machine or something where the mouse points to some noise?
[10:23:41] <JT-Shop> I prepared these starting with just P then using FF1 to null out the cruse and FF2 last to take care of over/under shoot
[10:23:58] <JT-Shop> which one?
[10:23:58] <cradek> awallin: that's the cursor
[10:24:23] <awallin> oh, ok :)
[10:24:56] <JT-Shop> in the last one I have ferror selected what does the f(0.54312) = 0.00015 mean?
[10:25:26] <awallin> with a mesa-card and pwm-amps I was never really able to tune out 'glitches' at the beginning and end of the constant-acceleration phases..
[10:26:08] <awallin> maybe that's the cursor reading?
[10:27:15] <awallin> JT-Shop: anything interesting to see if you zoom closer than 5m/div?
[10:28:10] <tom3p> is awallin's 1st video ok for all? my ffox 14.0.1 ubuntu canonical 0.1 wont display nor redirect to google site
[10:28:48] <awallin> tom3p: google-video may have ended operation altogether... can't remember
[10:28:59] <tom3p> haha thx!
[10:29:39] <awallin> here's a youtube version of that
http://www.youtube.com/watch?v=4Q7wrQY9OPc
[10:29:44] <awallin> I should update the blog post..
[10:31:17] <awallin> ok, blog post updated. For these kind of videos it would be nice to overlay the current G-code command on the video image..
[10:31:54] <tom3p> awallin, got it thx, is that motor brushed?
[10:32:19] <awallin> yes, iirc they were from dan mauch. came with us digital encoders fitted.
[10:32:50] <awallin> not so great pics over here
http://www.camtronics-cnc.com/servo-motors.asp
[10:33:33] <tom3p> http://www.youtube.com/watch?v=1UuHFS0cWxQ&list=UUj3obX1YCAqbhiR6ly5XshQ&index=5&feature=plcp
[10:34:19] <awallin> tom3p: yes, those are the ones. but testing without load isn't really useful for tuning.
[10:34:41] <JT-Shop> it shows my FF1 is a tiny bit high
[10:35:06] <JT-Shop> http://imagebin.org/233110
[10:35:32] <r00t4rd3d> JT-Shop,
http://pastebin.com/J1UNZffj
[10:36:03] <r00t4rd3d> now we know who to corner :D
[10:36:34] <JT-Shop> yea
[10:37:05] <awallin> is that periodic sawtooth related to the ballscrew pitch, or the motor 'cogging' ?
[10:37:17] <JT-Shop> dunno
[10:37:55] <JT-Shop> so yea the numbers f(0.54312) = 0.00015 change with the cursor but what are the numbers?
[10:38:36] <awallin> the first one is cursor position in time from trigger, the second one is the value of the current trace at the cursor
[10:38:52] <awallin> that would be logical anyway
[10:39:48] <eucalyptus> on ubuntu how to use emc powerpath ?
[10:39:50] <L84Supper> ~20hz sawtooth
[10:41:05] <awallin> eucalyptus: wrong emc?
[10:41:22] <JT-Shop> so is the value the ferror in user units?
[10:41:31] <cradek> eucalyptus: sorry, wrong channel: linuxcnc software is not related to emc.com's products
[10:42:27] <L84Supper> eucalyptus, maybe you're looking for
http://www.thegeekstuff.com/2009/09/how-to-install-and-configure-emc-powerpath-on-linux/
[10:42:45] <awallin> JT-Shop: check how high motor-pos-cmd is in relation to your g-code. if you find a factor of 25.4 you are using the wrong kind of g-code :)
[10:44:52] <JT-Shop> awallin: I don't understand
[10:45:59] <awallin> JT-Shop: you run these tests by doing "G1X50" for example. Then if your g-code and your scope are in the same units the red trace motor-pos-cmd should change 50 units (assuming we started at 0)
[10:46:17] <eucalyptus> thanks , this very cool ,
[10:48:32] <JT-Shop> just checking motor-pos-cmd vs joint-pos-cmd and joint is the same as G53
[11:02:23] <pcw_home> awallin: "with a mesa-card and pwm-amps I was never really able to tune out 'glitches' at the beginning and end of the constant-acceleration phases.."
[11:02:25] <pcw_home> This is basically unfixable without S curve trajectories
[11:04:33] <pcw_home> (infinite jerk tends to excite mechanical resonances as well)
[11:09:31] <pcw_home> With higher sample rates and FF2 you can get rid of some but the real problem is
[11:09:33] <pcw_home> force being a step function with trapezoidal trajectories
[11:10:41] <awallin> pcw_home: right. do you know anything about the araisrobo branch on github? they claimed s-curve velocities at some point?
[11:16:20] <L84Supper> awallin, have a link to that branch? is it off the main linuxcnc git repo?
[11:17:25] <awallin> https://github.com/araisrobo/linuxcnc
[11:17:38] <L84Supper> thanks
[11:17:54] <awallin> I haven't looked at that, just heard they were working on s-curve (jerk-limited) traj planning
[11:19:02] <L84Supper> awallin, I've been meaning to take a look into this but haven't gotten there yet
[11:26:47] <L84Supper> http://en.araisrobo.com/linuxcnc looks like they sell LinuxCNC with FPGA/USB Interface
[11:27:50] <L84Supper> LinuxCNC over USB to 7i43 with their own IO boards
[11:28:18] <r00t4rd3d> http://images.summitpost.org/original/185785.jpg
[11:31:08] <TekniQue> was that a bunny snagged by an eagle?
[11:32:21] <L84Supper> Headless LinuxCNC on the Beaglebone
http://en.araisrobo.com/a/araisrobo.com/en/linuxcnc/build-for-beagle
[11:36:17] <r00t4rd3d> http://www.thingiverse.com/thing:32839
[11:39:45] <tjb1> r00t4rd3d: thats awesome
[11:51:38] <L84Supper> if you're going to use LinuxCNC over USB what applications can settle for >100ms response times?
[11:55:40] <pcw_home> well you move motion, then eventually everything else to the remote device...
[11:55:58] <L84Supper> yeah :)
[11:56:13] <L84Supper> so the PC is just a UI
[11:56:17] <pcw_home> and you are back where youstarted
[11:57:34] <L84Supper> i just looked over the Arias tree for a few minutes
[11:59:29] <pcw_home> Although the set of motion thingies that need real time is finite and a good motion
[11:59:30] <pcw_home> RPC API definition and some switchable motion queues might make a remote trajectory generator feasible
[12:04:02] <L84Supper> not sure if they are doing that or are just using linuxCNC over USB to just control a single servo
[12:04:40] <pcw_home> But for a lot of thing (robotics is good example) having a lot of computing horsepower
[12:04:42] <pcw_home> in one place with real time control is ideal
[12:05:22] <pcw_home> they have a multi axis S curve trajectory generator (at least) in the 7I43
[12:05:25] <L84Supper> if forget how much room is left in the 7i43 FPGA for a soft micro without any external RAM
[12:06:08] <pcw_home> we have configs with 3-4 processors so its no real big deal
[12:07:12] <pcw_home> Spartan 6 is nicer in that respect (~twice as much blockram for same price chips)
[12:08:06] <L84Supper> we are just starting to look back into all this as well as using the GPU for trajectory planning and simulation
[12:08:25] <pcw_home> all of our USB configs have at least one processor (LBP protocol parser/low level FIFO interface)
[12:08:51] <pcw_home> (strangely enough there a HM2 USB config set for the 7I43)
[12:09:42] <pcw_home> but honestly USB is a dreadful interface for anything industrial
[12:14:28] <IchGuckLive> hi all
[12:14:59] <L84Supper> there is much to look back into over the next few months....
[12:17:38] <L84Supper> it will also be interesting to get feedback from machine tool manufacturers that have very little control expertise
[12:17:59] <L84Supper> they just slap on Fanuc, Siemens or Mitsubishi controllers
[12:19:41] <L84Supper> where does LinuxCNC fit in? is it just devs who customize?
[12:23:04] <L84Supper> will having Comedi and LinuxCNC working smoothly be popular?
[12:24:31] <L84Supper> HAL interface for COMEDI
[13:55:01] <JT-Shop> http://www.youtube.com/watch?v=BraEMAu5UkY&feature=youtu.be
[14:05:20] <rob_h> at least it can save doing the washing up i guess making a knife and fork
[14:32:43] <archivist> L84Supper, somebody else got the two working together iirc
[14:34:50] <L84Supper> archivist, yes, then andypugh had the idea to make a HAL interface for Comedi, it still needs to be written
[14:35:39] <andypugh> I still think it's probably a good idea. I am nowhere near starting.
[14:36:02] <L84Supper> http://wiki.linuxcnc.org/cgi-bin/wiki.pl?Debian_Etch_Compile_RTAI#Comedi_support_for_RTAI
[14:36:54] <L84Supper> andypugh, it's one more thing we wanted to take a look at in addition to the simulation and trajectory planner
[14:39:02] <L84Supper> we need all this for more complex 3d printers that combine more than one type of deposition tech
[14:40:46] <PCW> probably also want painting data via encoder count for that as well
[14:42:05] <PCW> or just path velocity if servo thread resolution is good enouigh
[14:42:44] <PCW> also extruders beneifit from lookahead
[14:43:36] <WalterN> oh... I was wondering if linuxCNC knows what direction its facing... eventually I might make an engraving tool and for cutting passes I would want the tool facing the direction its cutting...
[14:43:57] <WalterN> erm, not tool.. make an engraving machine
[14:43:58] <L84Supper> lots of room for tweaking
[14:44:02] <jdh> like a tangential knife
[14:47:29] <kirk_wallace> I want to set up a glade tab in axis for a lathe which needs to read the tool table to get tool orientation. Is there a set up out there that I can study?
[14:47:45] <andypugh> WalterN: There is a way to fudge it.
[14:48:42] <andypugh> WalterN: I wrote a kinematics module that kept a count of where it was last time through, and set the C axis accordingly.
[14:49:04] <andypugh> The difficulty is that there is no easy way to limit the speed of motion of the C axis.
[14:49:20] <andypugh> If it's a very fast axis then that might be OK
[14:49:50] <WalterN> oh, it might require the spindle to rotate faster than it actually can?
[14:51:14] <andypugh> Yes.
[14:51:32] <andypugh> Large f-error on that axis might be OK.
[14:51:44] <cradek> imagine you are going along, and then stop, and then go back the opposite way.
[14:52:14] <PCW> limit comp?
[14:52:49] <cradek> limiting it is easy - doing something smart and not breaking the tool at a sharp corner is harder if you do it in "postprocessing"
[14:53:20] <WalterN> well
[14:53:34] <WalterN> at work there is a machine that does what I described
[14:53:57] <andypugh> The optimum way to do it is with an input file filter.
[14:54:00] <WalterN> one of the main problems is that its a plugin for correl draw :-/
[14:54:33] <WalterN> but I kinda want something like that only without the fail for myself :P
[14:55:13] <cradek> yes doing it on the front end is the way to go
[14:55:24] <cradek> then you can lift and turn at a corner, or whatever you decide you need
[14:55:27] <PCW> I guess a filter could flag 180's an lift first
[14:55:28] <andypugh> It's not dreadfuly hard in principle, a single move in C to align with the start of each straight line, and a continuous coordinated move through each arc.
[14:55:51] <cradek> yes the entire motion is easily representable in gcode
[14:56:22] <cradek> and you get the benefit of synchronized moves, for instance a tiny arc will slow down if it needs to wait for C to keep up
[14:57:52] <cradek> but we've talked about this a hundred times, and people really want to try to do something worse instead
[14:59:15] <andypugh> I already did the "something worse" and have concluded that it isn't ideal :-)
[14:59:30] <cradek> :-)
[15:00:07] <andypugh> WalterN: You know that LinuxCNC can automatically filter any file that it opens before execution?
[15:00:57] <WalterN> I've never used linuxCNC or really looked at any documentation or anything
[15:14:58] <cradek> man what a way to kill an interesting conversation
[15:15:36] <WalterN> lol
[15:16:08] <WalterN> is there a bookmark folder somebody has for other people looking to build their own CNC machine?
[15:18:22] <r00t4rd3d> http://www.cnczone.com/forums/diy-cnc_router_table_machines/
[15:20:10] <WalterN> and I gotta go to work
[15:20:39] <WalterN> http://mylittlefacewhen.com/f/3262/
[16:12:00] <tjb1> Hello all
[16:19:32] <tjb1> andypugh: is this related to starting comp files?
http://linuxcnc.org/docs/html/gcode/m-code.html#_m100_to_m199_user_defined_commands_a_id_sec_m100_to_m199_a
[16:19:57] <tjb1> I save the comp as like M101 and then use one of those examples?
[16:22:49] <DJ9DJ> gn8
[16:27:23] <andypugh> tjb1: I think you may be confusing two rather different concepts.
[16:28:13] <tjb1> good possibility
[16:28:40] <andypugh> A comp file defines a HAL module, something like the parallel port driver, or the stepgen.
[16:29:15] <andypugh> A user-defined M-code is an executable file which is run only when called from the G-code
[16:29:54] <andypugh> For example, you could rename gedit to M101 and then M101 in the G-code would open gedit. (but that would be sittly)
[16:30:31] <andypugh> What do you want your custom M-code to do?
[16:31:48] <tjb1> Start the THC
[16:32:48] <andypugh> You mean "start" as in load, or "enable"?
[16:33:15] <andypugh> There are three stages to this.
[16:33:59] <andypugh> 1) One time only, compile and install the thcud.comp component:- sudo comp --install thcud.comp
[16:34:08] <tjb1> Ok
[16:34:12] <tjb1> Enable it
[16:34:40] <andypugh> 2) Edit the HAL file so that thcud is loaded (loadrt thcud) and add it to a thread (addf thcud servo-thread)
[16:35:10] <andypugh> You then need to set the thcud.enable pin.
[16:36:07] <andypugh> This could be done with a custom M-code (#!bin/bash // hamcmd setp thcud.enable 1 )
[16:36:54] <andypugh> Or simply in HAL, perhaps net enablethc halui.program-is-running=>thcud.enable
[16:37:43] <andypugh> Or with a G-doe digital output (M62 P0 and net thc-enable motion.digital-out-00 => thcud.enable)
[16:38:15] <andypugh> I need to turn ofd auto-cornet
[16:38:37] <andypugh> Oh, perhaps it _is_ off and I really did type "G-doe"!
[16:39:19] <tjb1> So I need to make a file named M101 and put "(#!bin/bash // hamcmd setp thcud.enable 1 )" in it and place the file in [Display] Program_Prefix in that ini file
[16:39:32] <tjb1> And then make m102 and thcud.enable 0
[16:40:02] <andypugh> Yes, that will work
[16:40:36] <andypugh> halcmd will work a lot better than hmcmd though (oh dear)
[16:40:47] <tjb1> Can you elaborate the [Display] Program_Prefix
[16:40:55] <andypugh> I can't even spell my miss-spellings correctly!
[16:41:27] <seb_kuzminsky> mmmm, hamcmd
[16:42:03] <andypugh> If you look in the INI file you will see a section [DISPLAY] and in that section is a PROGRAM_PREFIX which is where LinuxCNC looks for executable fles. Put the M101 file in there
[16:42:14] <tjb1> The file location
[16:42:19] <andypugh> hamcmd loadusr bacon
[16:42:24] <tjb1> Like C:/Usr/desktop/ etc...
[16:42:53] <andypugh> Yes, except this is Linux, so probably ~/linuxcnc/nc_files/
[16:44:08] <tjb1> so under PROGRAM_PREFIX I need to put "halcmd loadusr ~
[16:44:17] <tjb1> err...
[16:44:34] <tjb1> loadusr ~/linuxcnc/nc_files/M101
[16:46:20] <andypugh> No
[16:46:52] <andypugh> The #bin/bash etc goes in a text file in nc_files called M101, which needs to be executable.
[16:47:25] <tjb1> So what do I put under PROGRAM_PREFIX
[16:47:40] <andypugh> The halcmd loadusr part goes in the HAL file of your configuration (under linuxcnc/configs….)
[16:48:14] <andypugh> You don't put anything under program prefix. That is where you find out where linuxcnc is going to go looking for files
[16:50:03] <tjb1> Thanks andypugh
[16:50:12] <tjb1> One day I will learn most of this and teach someone else ;)
[16:50:58] <andypugh> Yeah, there is an optimum time to write it all down whilw you can remember why it was hard. I intended to, and missed. Now it all seems to entirely obvious….
[16:52:29] <tjb1> Im gonna go through JTs files and pull all the xml pins out of the thcud so I can understand it better
[16:52:37] <andypugh> seb_kuzminsky: I propose that the board of directors actually direct.
[16:53:28] <andypugh> (well, that might be going a little far)
[16:58:02] <tjb1> andypugh:
http://pastebin.com/p8Q0AMuR
[16:59:28] <andypugh> You actually want all the reads first, then the motion, then the writes in the addf section.
[16:59:44] <andypugh> The functions run in the order they are called.
[16:59:57] <tjb1> Thats pulled out of JTs
[17:00:10] <andypugh> (so, thc wants to be before motion)
[17:00:19] <andypugh> Yes, well, JT has it wrong :-)
[17:00:23] <tjb1> http://pastebin.com/Fz2dRzr9
[17:00:45] <tjb1> Do I need the servo-thread for stepper?
[17:00:45] <seb_kuzminsky> andypugh: what direction do you feel is lacking?
[17:00:58] <seb_kuzminsky> tjb1: yes
[17:01:37] <skunkworks_> north?
[17:01:43] <skunkworks_> ;)
[17:01:57] <andypugh> Though with software stepgen you want the parport writes on the base thread.
[17:02:51] <tjb1> Im so confused
[17:04:00] <tjb1> Why do I need the servo-thread?
[17:04:37] <skunkworks_> the servo thread is where all the motion is calculated
[17:04:46] <andypugh> That is the slow thread where all the hard sums are calculated
[17:04:51] <skunkworks_> (and then sum)
[17:05:03] <skunkworks_> *some
[17:07:27] <tjb1> So I need this -
http://pastebin.com/p8Q0AMuR placed into my HAL file
[17:09:37] <tjb1> Next question and I will go away for a while ;) - Do Ineed the encoder that JT is loading in his HAL file if I am not using the mesa cards?
[17:10:57] <PCW> no not for thcud
[17:11:18] <tjb1> The encoder is just for the mesa card then
[17:11:45] <PCW> only mesa and a THCAD
[17:11:48] <andypugh> tjb1: Do you have a Mesa card?
[17:12:05] <tjb1> nope
[17:12:23] <tjb1> Ill have an arduino measuring arc voltage and sending outputs
[17:12:23] <andypugh> Then the addf hm2….. stuff will fail horribly
[17:12:25] <Tom_itx> u should
[17:12:43] <PCW> if you get voltage some other way you dont need it (like a comparator bit for thcud)
[17:13:45] <tjb1> Ill have 2 inputs into the computer, torch up or down
[17:13:49] <tjb1> Thats it
[17:14:59] <PCW> The encoder counter is just used as a frequency counter (THCAD is just a well isolated and protected V-F converter)
[17:15:23] <andypugh> tjb1: Something like
http://pastebin.com/f24E4yn0 (but I guessed a lot, and the component might well be called thcud )
[17:16:17] <tjb1> So if the comp is thcud I need to change thc to thcud
[17:16:27] <andypugh> probably
[17:17:07] <andypugh> In fact certainly, because thc already exists as a HAL component and you will end up adding that instead.
[17:17:56] <tjb1> So that thc will = whatever I name the .comp file
[17:19:02] <andypugh> No, and this is a problem I have made for myself a few times
[17:19:28] <andypugh> the file name has to match the name in the "component …." line at the start of the comp file.
[17:19:30] <tjb1> Im currently having java pounded into my skull at this moment too :)
[17:19:54] <tjb1> component thcud "Torch Height Control Up/Down Input";
[17:20:06] <andypugh> If the component name and file name don't match then you get a bunch of not entirely self-explaining errors
[17:21:20] <tjb1> Does // work for comments in linuxcnc?
[17:21:52] <andypugh> That // was my way to put a line break in an IRC message
[17:22:15] <andypugh> The #! and halcmd lines need to be separate lines in the file
[17:22:34] <tjb1> What is comment in linuxcnc?
[17:22:40] <tjb1> I was just asking because // is comment in java
[17:22:43] <PCW> #
[17:22:51] <andypugh> The comment character is different depending on what language the segment is written in
[17:23:06] <andypugh> # is the comment in HAL and in bash.
[17:23:16] <andypugh> / is a comment in C and in .comp
[17:23:40] <andypugh> (there were two slashes there //// IRC ate one
[17:24:45] <andypugh> PCW: Do the PCI cards report a serial number? Should I consider using it...
[17:25:28] <PCW> No not ATM but they are "brandable" with a SN
[17:26:02] <andypugh> OK, I will not bother then. It would be of dubious utility anyway
[17:27:12] <PCW> well is has the same advantage/disadvantage that SNs have on sserial cards: great if you have more then one of the same
[17:27:35] <PCW> just more verbiage if you only have one
[17:29:00] <PCW> but with sserial I/O the need for 2 PCI cards is pretty rare nowadays
[17:29:44] <tjb1> Thanks everyone
[17:29:50] <tjb1> I have to drive home now…might be on in about 3 hours
[17:30:04] <PCW> long drive
[17:30:08] <tjb1> 109 miles home
[17:30:21] <PCW> 3 for me
[17:30:26] <tjb1> Im at school :)
[17:33:30] <uiu80> Hi guys! Anyone of you have had experience with linear encoder and stepper motors (closed loop)
[17:33:37] <uiu80> ?
[17:35:49] <uiu80> I have some trouble setting pid and ff parameters. I have heard that step+ encoder must be managed with p and ff1 only.. right?
[17:41:56] <uiu80> :(
[17:43:26] <JT-Shop> 9 minutes is not long to wait for an answer on the IRC
[17:46:24] <uiu80> Yes but i have to go out and i can not do other...
[17:47:22] <uiu80> So thanks a lot and goodnight
[17:47:22] <uiu80> :)
[17:52:51] <andypugh> Gah! Missed him.
[18:05:02] <JT-Shop> if you ain't got time to wait use the forum heh
[18:13:56] <JT-Shop> http://www.ebay.com/itm/ANTIQUE-1912-INSTANTANEOUS-WATER-HEATER-BRIDGEPORT-TOOL-/190505859465?pt=LH_DefaultDomain_0&hash=item2c5b08b989
[18:18:07] <andypugh> Is it bad that I want that?
[18:18:28] <Tom_itx> kinda
[18:19:05] <andypugh> I don't even own a Bridgeport.
[18:19:18] <Tom_itx> andypugh what's your arduino doing with lcnc? reading the hall sensors?
[18:19:25] <Tom_itx> and what arduino are you using?
[18:19:45] <JT-Shop> I thought you might get a kick out of that Andy
[18:19:47] <andypugh> I have a decemilieouavethinginitalian.
[18:20:12] <jdh> new one is 32bit, much faster, has 12bit ADC
[18:21:00] <andypugh> What it does is synthesise a sine wave to excite a resolver, then samples the return voltage synchronously to determine resolver angle.
[18:21:31] <andypugh> Is the clock speed higher? I am not short of CPU, just clock.
[18:22:42] <jdh> most arduinos have 16mhz crystals, but they can run at 20
[18:24:03] <jdh> new one (DUE) is 84MHz
[18:26:40] <jdh> I'll be there soon!
[18:26:44] <jdh> <urk>
[18:26:54] <andypugh> Ooh! Oooh! Oooh! 1Ms/S sounds marvellous
[18:28:00] <andypugh> And true D to A would make my resolver thingy much lower overhead too.
[18:28:14] <andypugh> Where's the "buy" butan?
[18:29:36] <r00t4rd3d> will probably take a week or so before you can actually buy one
[18:30:03] <andypugh> I can wait, I have had my Pi for a week and not even plugged it in yet
[18:30:19] <Tom_itx> jdh are those the xmegas?
[18:30:46] <jdh> had mine for 2 months. booted once
[18:30:56] <Tom_itx> the xmegas have dac i believe
[18:31:11] <Tom_itx> and pll clock
[18:32:39] <jdh> I have an ADC for Pi coming
[18:37:06] <andypugh> Mine might become a security camera controller, but I am also interested in the idea of linuxCNC on Pi
[18:37:24] <jdh> 15-20th
[18:38:33] <jdh> mine will monitor o2 and pressure and serve it over wifi
[18:39:16] <andypugh> A friend wants to monitor 8 voltages at very low sample rates at very low cost. Any ideas?
[18:40:01] <jdh> arduino + two mcp3424
[18:40:52] <jdh> or just the DUE if 12bits is enough. fixed vref
[19:07:08] <andypugh> I suspect he wants less hacking than that, but it might work.
[19:07:35] <andypugh> He wants to remote monitor a surprising number of batteries on a boat moored abroad.
[19:10:41] <PCW> battery monitoring is often rather tricky (no common gnd)
[19:11:42] <PCW> a bunch of 1 channel devices on a simple isolated bus might be a good way
[19:16:38] <andypugh> He mentioned 1-wire, but it seems that voltage sensors are scarce.
[19:16:50] <PCW> you can get 300 baud serial the the lousiest (and cheapest) of OPTOs
[19:17:00] <andypugh> On a boat there will be a common ground, I think.
[19:17:23] <L84Supper> last year was supposedly the time when smartphone dongles were supposed to take off, people we making all sorts of remote sensors
[19:17:45] <L84Supper> use a phone for remote monitoring
[19:17:49] <andypugh> I have the impression he was not wanting to design PCBs
[19:18:07] <andypugh> L84Supper: Links?
[19:18:18] <Valen> andypugh: what voltage is the system?
[19:18:29] <L84Supper> andypugh, from what I've seen they never became real
[19:18:47] <L84Supper> andypugh, is that the sort of thing that he needs?
[19:19:43] <andypugh> Probably 12V, possibly 24V, but large resistors exist for voltage division.
[19:20:35] <Valen> do keep in mind that method will put a constant drain on the cells and also reduce your accuracy
[19:21:08] <PCW> Big lead acid?
[19:21:13] <jdh> http://www.moxa.com/product/iologik-e2240.htm
[19:21:24] <jdh> I have a few of those at work. Pretty cheap
[19:22:05] <jdh> or not so cheap for that version.
[19:23:08] <PCW> just mentioning the isolation since often individual cells or batteries that are part of a series chain are monitored
[19:25:08] <L84Supper> http://www.amazon.com/USB-Multifunction-Module-Analog-Inputs/dp/B003DV56K2/ref=pd_sim_sbs_pc_4/176-9200346-5221340
[19:25:25] <jdh> http://store.moxa.com/a/product/iologik-w5340-series?id=M20090324006
[19:25:35] <jdh> that one and a mux.
[19:25:46] <PCW> bbl
[19:26:04] <L84Supper> http://www.portwell.com/products/detail_io.asp?CUSTCHAR1=USB-II-16
[19:26:28] <jdh> they have some almost-ladder stuff built in. You can have it phone home on condition
[19:26:40] <L84Supper> forget that one it's not analog IO
[19:27:20] <FinboySlick> L84Supper: Isn't most of their stuff terribly overpriced unless you buy thousands?
[19:27:21] <jdh> I've got some cheap NI analog usb things that work ok.
[19:28:16] <L84Supper> http://sine.ni.com/nips/cds/view/p/lang/en/nid/201986
[19:28:43] <jdh> mine are 14bit
[19:29:53] <L84Supper> http://numato.com/8-channel-usb-gpio-module $19.95
[19:31:45] <jdh> http://www.abelectronics.co.uk/products/3/Raspberry-Pi/7/ADC-Pi---Raspberry-Pi-Analogue-to-Digital-converter
[19:31:47] <L84Supper> FinboySlick, portwell?
[19:32:00] <jdh> I ordered one of those. They have an 18bit version also, but it's slow
[19:32:33] <FinboySlick> L84Supper: Yeah. I had a couple projects and while the hardware was spot-on, I couldn't justify the costs. I'm sure they're better on larger runs though.
[19:32:39] <L84Supper> good enough to monitor batteries
[19:33:30] <jdh> monitoring lead acid batteries for voltage doesn't tell you much about their health
[19:34:11] <L84Supper> FinboySlick, yeah, like most of the embedded vendors
[19:35:17] <L84Supper> FinboySlick, lots of low cost, decently made embedded boards from China are not sold outside of China
[19:37:27] <FinboySlick> Hehe, I have no love for china tonight... I'm working on getting that Z axis perpendicular again.
[19:37:51] <spamftw> Hey all. A few friends and I have converted over a mini mill from a Jaiman controller board to LinuxCNC controlled. Our current hurdle is that the bed is a bit uneven and would be very difficult to flatten. I'm curious if anyone has had experience/tips regarding mapping the bed of a mill
[19:41:18] <L84Supper> like using a laser micrometer mounted on the spindle?
[19:43:05] <jdh> probe the bed in a grid, save the heights, run a post-processor over your gcode to adjust the z heights
[19:43:31] <jdh> most of that has been done for pcb milling
[19:43:47] <Valen> put milling bit in mill, mill bed flat
[19:43:52] <Valen> be sure your right ;->
[19:44:00] <cradek> you have to consider what your goal is
[19:44:25] <jdh> always fix hardware with software.
[19:44:29] <spamftw> what we had looked at was not that high tech no. L84Supper. jdh, I think were on the same lines. unfortunatly Valen, the bed of the mill is slightly larger than the cuttable area
[19:44:33] <jdh> or is that never fix hardware with software.
[19:44:44] <cradek> flexible vs stout work seems like it would need to have different kinds of compensation
[19:44:54] <Valen> put a bigass face cutter in ;->
[19:45:15] <L84Supper> just be sure to be square first
[19:45:26] <cradek> a bogus mill can't mill itself flat
[19:45:54] <Valen> depends whats not flat
[19:46:18] <toastyde1th> "flat" generally refers to the load free path of the machine
[19:46:23] <cradek> imagine a bicycle wheel where your table is. you cut all around the wheel nice and even, but that doesn't make your table flat or allow you to cut flat things
[19:47:29] <L84Supper> so what I think you're saying is get a better mill that is flat and square
[19:47:49] <cradek> or measure your errors and know that they are your machine's tolerance
[19:48:04] <toastyde1th> either get a better mill or be prepared to jack the part around to cut flat
[19:48:05] <cradek> for certain jobs you can do better if you must, by thinking hard about your setup
[19:48:51] <spamftw> sorry, alot to take in? what are arguments against mapping the bed like jdh had mentioned?
[19:48:53] <cradek> but blindly cutting the top of the table or doing some other rash thing is just not going to help you do better work
[19:48:54] <toastyde1th> if you bolt clamps to the part, and then clamps to the table on top of shims, you can physically deform the part so that as the table flexes and axes move, the part is in the correct place
[19:49:08] <toastyde1th> by adding and removing shims
[19:49:12] <toastyde1th> this is done in grinding all the time
[19:49:53] <cradek> spamftw: ok say you map the table (say it's a sine wave) and "compensate" for that in software (by adding it to Z or something)
[19:49:54] <toastyde1th> spamftw, by mapping the bed, you know where the spindle is relative to the table ONLY for that particular table position
[19:49:57] <L84Supper> spamftw, like cradek said earlier, it depends on what you're actually trying to do
[19:50:08] <cradek> now set a nice granite surface plate on the table and face it off
[19:50:12] <toastyde1th> the rest of the table is actually in a physically different place from your map
[19:50:23] <Valen> it still depends on whats not flat
[19:50:48] <Valen> if the ways are wavy then sure, but if the table itself isn't flat and the ways are perfect then milling the bed flat is an option
[19:50:49] <cradek> now you have a surface plate that's as wavy as your table - what good does that do you?
[19:51:17] <cradek> Valen: yes understanding the machine and the work it needs to do is the first step, not cutting things
[19:51:36] <spamftw> Alright, I see what you guys are saying.
[19:54:51] <toastyde1th> and you really can set the machine up for small errors if you really care about flatness
[19:54:59] <L84Supper> play in the ways or spindle, table not flat, etc etc
[19:55:07] <spamftw> so if it happens that the bed is flat, but the spindle track is the issue? then mapping may be the best idea
[19:55:23] <toastyde1th> I used to have to cut very flat things and by using shims, could do way better than the machine's tolerance as long as the error was repeatable
[19:55:32] <toastyde1th> "spindle track?"
[19:56:18] <spamftw> Yes, sorry? the mill is setup like this
[19:56:19] <spamftw> http://image.ec21.com/image/szelian/oimg_GC04178517_CA04178657/SC2518_Mini_CNC_Router.jpg
[19:56:25] <toastyde1th> z axis
[19:56:43] <L84Supper> gantry mill
[19:56:56] <toastyde1th> you should be able to unbolt and align the head
[19:57:12] <toastyde1th> but tramming a mill is not trivial if you don't understand what has to be checked first
[19:57:58] <spamftw> as the head moves down as it rides to the left of the track (in the last 2 inches), so i'm not sure we can just align it at any one position
[19:58:32] <toastyde1th> that sucks and is something you can't really fix outside knowhow and a lot of elbow grease, or software
[19:58:48] <toastyde1th> for that particular error, software is the best bet since it's a very repeatable and consistent error
[19:59:27] <toastyde1th> if your machine is linear rails, those unbolt too
[20:00:00] <spamftw> got it. got a few more questions but wanted to take the chance to thank you guys for all your help and patience
[20:00:01] <toastyde1th> but unbolting the ways of a machine is a bit of a hail mary if you can get the error down to something you are okay with
[20:00:06] <toastyde1th> with software
[20:00:06] <Valen> tramming with a non flat bed would be challenging ;->
[20:00:31] <toastyde1th> not really
[20:00:44] <toastyde1th> same as tramming a head without unbolting a fixture
[20:00:59] <Valen> you need to be square to something, one generally assumes that the bed is square and lines up to that
[20:01:16] <spamftw> tramming is the process by which the z is mapped?
[20:01:23] <toastyde1th> bed isn't the greatest to begin with although for mostly accurate tramming it's quick and easy
[20:01:42] <Valen> tramming is making sure the mill head is square to the work your cutting
[20:01:43] <toastyde1th> most accurate is to set up a parallel or straightedge on the machine axis suspended 1/4 of the way in from the ends
[20:01:54] <andypugh> spamftw: If you _know_ that your bed is going to win over the work every time (for example a vacuum table and a PCB) then there is a kinematics module for LinuxCNC that models the surface in STL format and compensates.
[20:02:10] <toastyde1th> zero an indicator on one support, travel to the other, and zero the indicator by shimming or jacking the parallel up
[20:02:23] <Valen> so if you do a set of overlapping cuts it wont have a zig-zag pattern on the cut
[20:02:32] <spamftw> andypugh, any link you could point me to for more information on that one?
[20:02:37] <toastyde1th> then you have something quite prallel to the ways without worrying about bed condition or having a bigass fixture on the table
[20:02:38] <spamftw> or anything I should search for
[20:02:38] <andypugh> But in general the bed deforms the work and the work deforms the bed and the proporions of distortion vary.
[20:03:21] <L84Supper> spamftw, is that how the bed is mounted? moves in X on rails? and the spindle moves in Y and Z
[20:04:09] <spamftw> The bed moves in Y, the spindle travels on a rail in X, and the spindle moves on Z
[20:04:44] <andypugh> That's a slightly unfortunate combination.
[20:05:28] <spamftw> andy, in our case I'd say bed will win over work. we've got full size mills for the more hefty work? I think everyone is just excited at the chance at pcbs. If we need a vacuum table and to dedicate the machine to that, I'm sure that wouldn't be too big of a sell
[20:05:34] <andypugh> A knee-mill layout works fine as long as the bed is flat and parallel to the X ways.
[20:07:04] <cradek> knee mills tend to be part bicycle wheel
[20:07:32] <toastyde1th> agree with bike wheel
[20:07:58] <toastyde1th> there's a reason jig grinders are all gantry or C frame bed machines
[20:11:02] <spamftw> andypugh, was the module you refered to probe2stl
[20:11:05] <andypugh> The Deckel/Aciera/(too long on lathes.co.uk looking for the name) style sliding Y head mills are a lot less forgiving of way divergence, but luckily are all top-class machines
[20:11:29] <L84Supper> what do people do for vias and through holes when cnc routing a two layer PCB?
[20:11:45] <cradek> cry a bit and then try to avoid them
[20:12:05] <toastyde1th> womp womp wommmmp
[20:12:19] <cradek> sometimes solder them manually in a fit of desperation
[20:12:47] <andypugh> L84Supper: Rivlets?
[20:14:24] <spamftw> Alright, I think I'll start here
http://wiki.linuxcnc.org/cgi-bin/wiki.pl?ProbeKins. Thanks again for all the help all
[20:15:55] <andypugh> Yes, well found, that;s the one. Sorry, I was distracted
[20:16:16] <tjb1> Im home :)
[20:17:35] <L84Supper> I was just thinking about low tech solutions for the vias, cnc syringe with conductive ink, etc etc
[20:19:00] <L84Supper> conductive plugs + solder paste
[20:19:18] <L84Supper> the way they make them now
[20:22:51] <andypugh> http://uk.rs-online.com/web/p/pcb-processing/1603672/ is a rivlet (but discontinued, and spelt rivet)
[20:23:16] <r00t4rd3d> http://imgur.com/a/4XSZG
[20:23:32] <r00t4rd3d> I still need to stain the lettering, letting the first coat of oil dry though.
[20:23:35] <andypugh> http://wiki.answers.com/Q/What_is_a_rivlet
[20:24:30] <andypugh> "Jeffrey L Webster, dead, get over it"
[20:24:46] <r00t4rd3d> I made a coffin :/
[20:25:00] <r00t4rd3d> its for his ashes
[20:25:26] <L84Supper> there are better ways to do this but I was thinking about easy retrofit solutions for cnc pcb routers
[20:25:36] <andypugh> Hmmm, I think I just decided on the words to be scrawled in crayon on my cardboard box.
[20:26:26] <tjb1> What do those sell for r00t4rd3d ?
[20:26:50] <r00t4rd3d> that one is 125 cause its custom sized
[20:26:57] <tjb1> what size is it
[20:27:03] <r00t4rd3d> their is an engraving on the inside also
[20:27:31] <Tom_itx> L84Supper,
[20:27:34] <Tom_itx> http://au.element14.com/jsp/search/productdetail.jsp?sku=1143874
[20:27:36] <r00t4rd3d> the inside needed to be 8 3/4 x 6 3/4 x 4 3/4
[20:27:42] <tjb1> Oh that isnt that big
[20:27:48] <L84Supper> mill the board. drill the holes, and fill with paste and bake vs plate
[20:27:58] <Tom_itx> L84Supper ^^
[20:27:59] <r00t4rd3d> outside is -alot- bigger
[20:28:04] <tjb1> alot?
[20:28:06] <r00t4rd3d> like 12"
[20:28:07] <tjb1> >:(
[20:28:25] <r00t4rd3d> now that i know that word bugs you alot i will use it more often
[20:28:35] <r00t4rd3d> never show your colors
[20:28:38] <tjb1> That knob bugs me a lot too
[20:28:59] <r00t4rd3d> the lady came got it today and gave me my 75 bucks
[20:29:03] <andypugh> Tom_itx: Good find, those are the things.
[20:29:16] <Tom_itx> i've had that bookmarked a while
[20:29:27] <L84Supper> http://www.saturnelectronics.com/via-plugging.htm
[20:31:26] <L84Supper> http://www2.dupont.com/MCM/en_US/assets/downloads/techpaper/cb100_processing_paper.pdf Process for Plugging Low to High Aspect Ratio Through-Holes with Polymer Thick Film Conductive Ink in Production Volumes
[20:32:26] <andypugh> r00t4rd3d:
http://bit.ly/MWyPSG
[20:33:13] <r00t4rd3d> ?
[20:33:21] <andypugh> Errr, forget that, I have _no_ idea where that link came from
[20:34:12] <r00t4rd3d> what is the link you where trying to show me
[20:34:31] <r00t4rd3d> Andy -
http://pastebin.com/J1UNZffj
[20:34:49] <r00t4rd3d> I tried :(
[20:36:32] <andypugh> I can't figure out how to get the URL I wanted.
[20:37:15] <andypugh> My web browser is too clever
[20:37:34] <r00t4rd3d> thats a mac 4u
[20:37:53] <andypugh> Try
http://www.google.com/search?q=alot&hl=en&safe=off&biw=1806&bih=1096&prmd=imvns&tbm=isch&tbo=u&source=univ&sa=X&ei=EpWIUPysHoqa0QWw1oHgBQ&sqi=2&ved=0CDEQsAQ
[21:04:28] <tjb1> Anyone good with java take a look at this and tell me what I need to do to get it to work...
[21:04:29] <tjb1> http://pastebin.com/hjQKXkNH
[21:08:04] <jdh> style or syntax?
[21:09:13] <jdh> return if?
[21:21:33] <tjb1> Yeah fixed that now im having a new problem
[21:22:04] <tjb1> http://pastebin.com/QSpqMath
[21:22:14] <tjb1> That is the class which I believe is fixed now at the end
[21:22:33] <tjb1> But using this to test it -
http://pastebin.com/6QGn13x1 - returns 0
[21:25:08] <tjb1> Something wrong in those if statements :(
[21:27:19] <jdh> I see nothing calling shippingCharge()
[21:27:29] <tjb1> In which?
[21:28:05] <jdh> oh, in the println... nevemrind.
[21:28:43] <tjb1> Its not doing the if statements :(
[21:29:34] <jdh> I've done one java program in my life. Didn't much care for it.
[21:29:48] <jdh> but, it was pretty cool to have it running on my phone.
[21:31:47] <tjb1> I hate java...
[21:44:58] <pcw_home> I love java, sweet and hot, Whoops! Mr. Moto, I'm a coffee pot
[21:46:01] <jdh> java is jive
[21:54:04] <tjb1> Is there a way to watch a probe as a limit during a non probing move?
[22:31:25] <tjb1> Tom_itx:
[22:41:50] <tjb1> Anyone?
[22:42:58] <jdh> no.
[22:43:11] <tjb1> No what? :P
[22:45:31] <jdh> no, you can't.
[22:46:05] <tjb1> Im sure there is a way…its just more involved than anything I would like to do
[22:46:28] <jdh> if you have a probe and it goes off during a none probe move, it whines
[22:46:29] <FinboySlick> tjb1: Well, you could put it in series with your limit switches.
[22:46:44] <tjb1> FinboySlick: but then it would trip a limit when probing
[22:46:50] <jdh> if you did that and probed, you would trip
[22:47:07] <FinboySlick> I probably misunderstood what you meant then.
[22:47:27] <tjb1> If the switch is tripped when not probing, i want it to act as a limit
[22:47:44] <tjb1> I could make the g38 set an output to switch which input that sensor goes to
[22:48:07] <tjb1> g38 active, it sends that sensor to input 2, g38 inactive and that sensor goes to input1
[22:48:33] <jdh> I think it will stop if the probe trips during a non-probe move.
[22:48:39] <jdh> but, it might just whine
[22:48:46] <tjb1> No it just throws a little message
[22:49:12] <tjb1> I like the relay idea…dirty though
[22:50:19] <jdh> control.c:404: reportError(_("Probe tripped during non-probe MDI command."));
[22:50:27] <jdh> change that
[22:50:32] <tjb1> Where is that...
[22:50:49] <jdh> control.c
[22:51:00] <tjb1> gotta be more specific than that :)
[22:51:32] <tjb1> Or do you mean press control-c
[22:51:48] <tjb1> Well I guess that is copy on linux/windows...nevermind
[22:52:31] <jdh> it's in the source code. snag a copy, edit, rebuild everything
[22:53:08] <jdh> someday, I'll make a probe.
[22:53:12] <jdh> and all the other stuff
[22:54:13] <jdh> put in a throw switch. one way is probe, other is limit
[22:54:22] <jdh> s/throw/double throw/
[22:54:49] <tjb1> Ill just do a relay
[22:54:51] <tjb1> :)
[22:55:18] <tjb1> It probes every time it pierces so I dont want to sit there and switch it on and off all the time
[22:56:53] <jdh> wonder if you could just do it in HAL
[22:57:20] <jdh> probably with two ands
[22:58:08] <jdh> input and probing tied to probe input, input and notprobing tied to limit
[22:58:26] <jdh> and an mcode to setp probing
[23:00:43] <tjb1> can you use if statements in hal?
[23:01:56] <jdh> not that I know of, but that doesn't mean much.
[23:02:26] <jdh> you don't need to use an if if you just make the probe signal from an AND component
[23:03:18] <jdh> man and2
[23:03:33] <tjb1> and component
[23:03:34] <tjb1> ?
[23:03:45] <jdh> and2... type man and2
[23:03:54] <tjb1> im not on linux
[23:04:07] <tjb1> http://linuxcnc.org/docs/html/man/man9/and2.9.html
[23:04:10] <jdh> http://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=2&cad=rja&ved=0CCQQFjAB&url=http%3A%2F%2Flinuxcnc.org%2Fdocs%2Fhtml%2Fman%2Fman9%2Fand2.9.html&ei=XreIUO_WB4qA9gSBg4DYAg&usg=AFQjCNHA8yY9LSbT2eWnVH_LAR62SV2AFA
[23:04:15] <jdh> heh, or that.
[23:06:25] <tjb1> net torch-probe motion.probe-input <= parport.0.pin-11-in-not
[23:06:39] <tjb1> That is my probing input
[23:09:15] <tjb1> Is the N the pin name?
[23:09:41] <jdh> what N?
[23:09:55] <tjb1> and2.N.in0 bit in
[23:10:06] <jdh> oh, it is the instance of the and2
[23:10:31] <tjb1> Can you give example
[23:10:36] <jdh> if you loadrd and2 count=2, you will have a and2.0... and and2.1...
[23:11:26] <tjb1> count 3 would make 2.0, 2.1, 2.2
[23:11:37] <tjb1> then 2.3 would be the out bit?
[23:11:39] <jdh> and2.0, and2.1, and2.2
[23:12:00] <jdh> for the first one, the two inputs would go to and2.0.in0 and and2.0.in1
[23:12:11] <jdh> and when both became true, and2.0.out would be true
[23:12:37] <tjb1> Is this correct "and2.0.torch-probe"
[23:12:42] <jdh> no
[23:12:50] <tjb1> Of course not :)
[23:13:03] <jdh> it is just and2.0.[in0,in1,out]
[23:13:26] <tjb1> So where would I put the input pins?
[23:13:30] <tjb1> or name them?
[23:13:38] <jdh> you have to net your pport pin to .in0, then net your probing pin to and2.0.in1
[23:14:20] <jdh> then when you setp your probing pin, and the probe goes off, that and2.0.out becomes true
[23:14:35] <jdh> another and2 would do the same for the limit
[23:15:51] <tjb1> Can you give me an example of tying the probe to it?
[23:15:54] <jdh> no
[23:16:07] <jdh> but, you could ask someoen with a clue when they show up.
[23:16:08] <tjb1> :/
[23:17:41] <tjb1> Well until then im just going to activate an output before the probe and then deactivate it after the probe is done
[23:18:28] <jdh> extra hardware, but I could make that work!
[23:22:32] <tjb1> you know how to set that up?
