#emc | Logs for 2007-04-12

[00:00:57] <cradek> yeah that appears to be a bug
[00:01:16] <Jymmm> robin_szthinking about it
[00:01:49] <cradek> will you file it so we don't forget it please?
[00:01:55] <Jymmm> robin_sz The large one that does 1200x900mm 80W
[00:02:03] <Rugludallur> cradek: sure bugtrack on sourcforge ?
[00:02:07] <cradek> yes
[00:02:20] <Rugludallur> cradek: I'm messing around, might find some more
[00:02:25] <cradek> sure
[00:02:33] <cradek> I bet you will
[00:02:50] <cradek> the nontrivkins part of emc is not widely used, and the nontrivkins parts of AXIS are very new
[00:03:10] <Rugludallur> cradek: Never underestimate the abilities of a user who does not know what he is doing :D
[00:03:30] <cradek> that's sometimes an excellent test
[00:04:34] <Jymmm> robin_sz kinda "primative" (in a sense), but for us that would be a good thing - no propritary crap to deal with.
[00:04:49] <robin_sz> Jymmm, what you thinking of doing with it? plastic letters etc?
[00:06:57] <robin_sz> anyway ... we are so busy at the moment, both factories are full, ther was no room to put the truck inside over the weekend ... and guess what happened to it?
[00:07:43] <robin_sz> thats right ... the theiving gypsy bastards stole the aluminium sides to weigh in as scrap
[00:37:07] <ejholmgren_> ejholmgren_ is now known as ejholmgren
[01:36:10] <eric_u> anyone know of a library that can efficiently calculate the eigenvector associated with the largest eigenvalue of a sparse matrix?
[01:36:49] <tomp> eigen't think of one :)
[01:37:07] <eric_u> how special :)
[02:06:22] <Chris_sub_1> Hello. Any ubuntu gurus here?
[02:21:32] <cradek> Chris_sub_1: maybe, but there's also an #ubuntu irc channel with more of them
[02:37:09] <Chris_sub_1> cradek - Thanks. Got past the problem (unsupported CD-ROM).
[02:37:23] <cradek> yay
[02:37:26] <Chris_sub_1> :)
[02:38:04] <Chris_sub_1> I have a Hurco KM3 sitting in my shop longing for a new brain.
[02:38:14] <jmkasunich> brrrraaaaaiiiinnnnnnnsssss!
[02:39:06] <toastydeath> i started whistling "the yellow brick road"
[02:39:32] <cradek> darnit why don't I have a mill sitting here waiting for a new brain? what's the trick?
[02:40:22] <Chris_sub_1> Any opinions on the Mesa 5i20 & 7i33 for a servo machine?
[02:40:35] <jmkasunich> http://www.schlockmercenary.com/d/20061031.html
[02:41:15] <cradek> Chris_sub_1: that seems like the most popular thing people are using for EMC2 conversions
[02:42:33] <Chris_sub_1> The great thing is that as soon as I buy one and complete integration, they're bound to release a new and improved version. ;)
[02:42:34] <jmkasunich> just don't use abby's brain
[02:42:48] <cradek> Chris_sub_1: that doesn't matter one bit in my opinion
[02:43:26] <cradek> and the advantage of getting the current one is you're not the first to use it
[02:43:36] <Chris_sub_1> I'm an EE but my VHDL is weak. Will the 'canned' code likely get me running without much (firmware) hassle?
[02:43:45] <jmkasunich> yes
[02:43:47] <cradek> yes entirely
[02:43:53] <jmkasunich> there should be no need to do vhdl
[02:44:09] <jmkasunich> says /me, who is doing VHDL at this very moment ;-)
[02:44:15] <Chris_sub_1> I'm just starting to get up to speed...
[02:45:03] <tomp> http://abnormalbrain.bizland.com/index.html
[02:45:52] <Chris_sub_1> I've been working on the Hurco for just over 6 months, and I've got it running on single phase power with working amps, 1000 line encoders, limit switches and most other functions ready to roll.
[02:46:54] <Chris_sub_1> I also made a test pendant. It's now a 3800-pound etch-a-sketch. :)
[02:47:19] <jmkasunich> unrelated question: does "tar -xzvf foo.tar.gz" unzip, then untar, or does it do both at once?
[02:47:39] <cradek> together
[02:47:42] <jmkasunich> good
[02:47:49] <cradek> no intermediate file is made if that's what you mean
[02:48:07] <jmkasunich> (I have a 1G zipped tarball that expands to 2.4G, and about 3.8 G of space to unpack it in)
[02:48:44] <cradek> Chris_sub_1: sounds like an excellent start
[02:48:58] <cradek> Chris_sub_1: tool changer?
[02:49:06] <Chris_sub_1> That would be me. :)
[02:49:17] <Chris_sub_1> KS200 spindle
[02:49:21] <jmkasunich> where are you located?
[02:49:36] <Chris_sub_1> In MD just north of D.C.
[02:49:49] <jmkasunich> you should try to make it to the CNC workshop
[02:50:02] <Chris_sub_1> I'm active in CAMS and a good friend of Steve Stallings.
[02:50:17] <jmkasunich> are you Chris Daniel?
[02:50:28] <Chris_sub_1> No. He is Chris_sub_2.
[02:50:38] <jmkasunich> ;-)
[02:50:45] <Chris_sub_1> Chris Helgesen
[02:51:14] <jmkasunich> darned ubuntu - the updater installed a new kernel, even tho I have a RT one... and now its telling me to reboot
[02:51:38] <cradek> jmkasunich: you can uninstall kernels you don't use - then it will never update them
[02:51:39] <jmkasunich> * jmkasunich gets a sharp knife out and prepares for surgery on grub/menu.list
[02:52:10] <jmkasunich> it would be kind of handy to have a non-RT one just in case of emergencies
[02:52:38] <jmkasunich> the problem is that the updater doesn't realize thats not my default kernel
[02:53:19] <Chris_sub_1> Q: As of 5 mins. ago, I'm playing with the live CD on a PII-300. Is that likely too much of a dog to be useful for a servo system using the 5i20?
[02:53:31] <jmkasunich> for servo it should be fine
[02:53:46] <cradek> I even run my stepper mill on a PII-300
[02:53:55] <cradek> you need ram, 512 is best
[02:54:08] <jmkasunich> as long as the GUI stuff doesn't bog it down, the control should be good
[02:54:09] <cradek> 384 is ok, 256 might work but might not
[02:54:10] <Chris_sub_1> 384 at the moment.
[02:54:18] <cradek> that'll be ok I bet
[02:54:25] <jmkasunich> (newer gnomes and such need a lot more than EMC does)
[02:54:50] <Chris_sub_1> I plan to build a *real* machine for it, but I wanted to get up to speed before spending $.
[02:55:13] <jmkasunich> you might find a newer machine faster running code, but worse for latency
[02:55:30] <cradek> the best machine for EMC seems to be the fast PIIIs
[02:55:43] <Chris_sub_1> Noted. Thanks.
[02:55:44] <cradek> I have a PIII-800 on my lathe
[02:56:01] <cradek> fortunately that kind of hardware is virtually free
[02:56:52] <jmkasunich> hmm, I have linux-image-2.5.16-26, -27, and -28.... I thought the update was supposed to replace, not add?
[02:57:17] <cradek> they sometimes replace. they have to add if they do something like change the list of modules
[02:57:34] <cradek> just apt-get delete --purge the ones you don't want
[02:57:54] <jmkasunich> I should be able to "mark for complete removal" on the -26 and -27, right?
[02:58:06] <jmkasunich> (using synaptic, its convenient to see what is there that way)
[02:58:06] <cradek> if that's how you say "delete --purge", yes
[02:58:41] <jmkasunich> they have "mark for removal" and "mark for complete removal"
[02:58:43] <cradek> also nuke their correspond -restricted-modules-...-26 etc
[02:58:46] <jmkasunich> so I'm guessing yes
[02:58:52] <cradek> yep I bet so
[02:58:56] <cradek> corresponding
[02:59:19] <jmkasunich> it popped up to tell me about the restricted modules
[03:00:14] <jmkasunich> funny - all my kernels are 2.6.16-X, but linux-kernel-headers is at
[03:02:49] <cradek> that's ok
[03:05:33] <cradek> I had a reply to "d06" all typed (hook axis.N.amp-enable-out to stepgen.N.enable) but fortunately checked the source before sending it
[03:05:59] <jmkasunich> ?
[03:06:04] <cradek> looks like turning off enable doesn't turn off the outputs, it just inhibits stepping
[03:06:40] <SWPadnos> right - you can get another step at disable if you change the outputs
[03:06:55] <cradek> sure
[03:07:09] <jmkasunich> but if he's using a step type > 1, he might be driving motor windings directly
[03:07:12] <cradek> or, your Z axis might fall down
[03:07:13] <jmkasunich> and he wants to turn them off
[03:07:15] <cradek> right
[03:07:18] <cradek> I'm sure that's the idea
[03:07:38] <jmkasunich> and gates between the stepgen pins and the parport is the only thing that comes to mind
[03:07:40] <cradek> I don't see an easy way to do that
[03:07:51] <cradek> yeah, he needs a bunch of and gates
[03:07:53] <jmkasunich> gotta run them in the fast thread, fortunately they don't use FP
[03:08:26] <cradek> [HAL]SHUTDOWN does seem like something we might want
[03:08:46] <cradek> my choppers turn off because they're hooked to amp-enable-out
[03:09:35] <jmkasunich> to be honest, I'd probably try to do the shutdown in hardware, and use a single pin of the parport (like you did)
[03:09:52] <cradek> emc: SHUTDOWN=`$INIVAR -ini $INIFILE -var SHUTDOWN -sec HAL 2> /dev/null`
[03:09:56] <cradek> duh
[03:09:57] <cradek> we do have that
[03:10:36] <cradek> jmkasunich: clearly that's better. but with 3 motors and 12 parport outputs, not gonna happen
[03:10:50] <jmkasunich> so in there he can do "unlinkp parport.0.pin-XX-out ; setp parport.0.pin-XX-out 0"
[03:11:03] <cradek> yes I think so (untested)
[03:13:04] <cradek> are you answering?
[03:13:08] <jmkasunich> no
[03:13:22] <jmkasunich> you have (a) answer mostly written
[03:13:30] <cradek> no but I'll do it
[03:13:38] <jmkasunich> thanks
[03:20:49] <cradek> done
[03:21:24] <cradek> did you see the strange homing heavior with kins?
[03:21:38] <cradek> heavior?
[03:22:00] <cradek> guess I shorten words when it gets late
[03:22:16] <jmkasunich> it doesn't home in world you mean?
[03:22:42] <jmkasunich> I think thats a serious issue - because the homing code is in the joint controllers - its inherently a free mode (joint mode) thing
[03:23:03] <jmkasunich> homing in other than joint mode doesn't make sense
[03:23:05] <cradek> sure, it should give an error and drop the message
[03:23:08] <SWPadnos> it seems even worse that the home command got queued
[03:23:15] <cradek> currently it keeps it until later :-P
[03:23:20] <jmkasunich> that I don't get
[03:23:26] <cradek> silently - no error is shown
[03:23:28] <jmkasunich> * jmkasunich looks at code
[03:23:57] <cradek> I wouldn't have believed it, but I tried it myself
[03:25:49] <jmkasunich> /* need to be in free mode, enable on */
[03:25:49] <jmkasunich> /* this just sets the initial state, then the state machine in
[03:25:49] <jmkasunich> control.c does the rest */
[03:25:49] <jmkasunich> rtapi_print_msg(RTAPI_MSG_DBG, "HOME");
[03:25:49] <jmkasunich> rtapi_print_msg(RTAPI_MSG_DBG, " %d", joint_num);
[03:25:49] <jmkasunich> if (GET_MOTION_COORD_FLAG() || !GET_MOTION_ENABLE_FLAG()) {
[03:25:50] <jmkasunich> break;
[03:25:52] <jmkasunich> }
[03:26:16] <jmkasunich> I dunno if COORD_FLAG is true for teleop or not
[03:26:30] <jmkasunich> I despise those damn flags - a state variable would make far more sense
[03:27:12] <jmkasunich> that code is in "case EMCMOT_HOME:" in command.c
[03:27:37] <jmkasunich> I bet its getting past that test, setting the initial state, but the homing state machine doesn't run unless its in free mode
[03:27:45] <jmkasunich> so when the machine _does_ run, it starts homing
[03:29:17] <jmkasunich> in the actual homing code, I do use a state var
[03:29:17] <jmkasunich> if (emcmotStatus->motion_state != EMCMOT_MOTION_FREE) {
[03:29:17] <jmkasunich> /* can't home unless in free mode */
[03:29:17] <jmkasunich> return;
[03:29:17] <jmkasunich> }
[03:29:40] <jmkasunich> the state var is something I added, but I never found and removed all the places where those damn flags are used
[03:30:10] <jmkasunich> I guess the same test could be used in command.c
[03:30:19] <jmkasunich> want me to fix it?
[03:31:10] <cradek> at least in trunk, definitely
[03:31:30] <cradek> I think it should give an error
[03:31:36] <jmkasunich> you have a config that makes testing handy?
[03:31:40] <jmkasunich> I'll fix if you'll test
[03:31:41] <cradek> "must be in joint mode to home"
[03:32:14] <cradek> I sure will test tomorrow, but I am going to bed now
[03:32:54] <cradek> I think changing trivkins to rotatekins in sim is all I did
[03:33:18] <jmkasunich> I'm close to bedtime too - not willing to mess around with getting a config to test
[03:33:31] <cradek> ok maybe tomorrow or something
[03:33:36] <cradek> not like this is life and death
[03:33:40] <jmkasunich> the change won't take long though, I'll do that and then sleep
[03:34:06] <cradek> goodnight then
[03:34:12] <jmkasunich> goodnight
[03:34:12] <cradek> thanks
[03:34:23] <Chris_sub_1> <Thanks for help guys!> G'nite.
[03:35:27] <cradek> goodnight, other chris
[03:35:46] <cradek> keep us posted on the progress on your interesting machine
[03:36:10] <Chris_sub_1> I will *most definitely* be back. :)
[03:36:14] <jmkasunich> good
[03:36:38] <jmkasunich> if you can make it to the CNC workshop, you can see how stuff like toolchanger was done on the mazak
[03:36:54] <jmkasunich> and we'd be happy to talk about how to tackle your machine
[03:38:33] <Chris_sub_1> I've been reading quite a bit and saw the Mazak pics. I would love to make it to the workshop but I doubt that could happen this year.
[03:39:07] <cradek> quite a drive for you...
[03:39:24] <jmkasunich> hitch a ride with Steve Stallings
[03:39:26] <Chris_sub_1> Oh, I'd make Steve drive. ;)
[03:39:45] <cradek> yep, I know he and matt are driving down
[03:39:59] <jmkasunich> driving up
[03:40:03] <jmkasunich> (or over, actually)
[03:40:09] <cradek> pedant
[03:40:31] <jmkasunich> actually, its up in altitude, latitude, and longitude, so there!
[03:41:06] <CIA-19> 03jmkasunich 07TRUNK * 10emc2/src/emc/motion/command.c: better check of conditions before homing
[03:41:14] <jmkasunich> get testing!
[03:41:18] <cradek> bah
[03:41:25] <cradek> after some sleep I will
[03:41:41] <jmkasunich> get sleeping then!
[03:41:55] <Chris_sub_1> Sleep is overrated.
[03:42:03] <cradek> Chris_sub_1: yep, waste of time
[03:42:27] <jmkasunich> lack of it does catch up to you though
[03:43:16] <Chris_sub_1> It would be less of a problem if kids had snooze buttons.
[03:44:04] <jmkasunich> jobs don't have snooze buttons either
[03:44:18] <cradek> mine does, but it's best to not use it too much
[03:44:40] <cradek> why am I still here? goodnight!
[03:44:55] <jmkasunich> well, I have flex time - but to work on my preferred schedule, it would have to flex until noon ;-)
[03:45:12] <Chris_sub_1> Good night all.
[03:46:41] <tomp> any cheap tricks to getting a .25ohm sense resistor? ( i have some .15 ohms but they're huge 50watt alum housing units , was looking for something more reasonable )
[03:46:53] <SWPadnos> digikey?
[03:47:17] <tomp> :) i meant scrounged tonight
[03:47:23] <SWPadnos> oh, I see :)
[03:47:38] <tomp> like some trick of using a fet gate or....
[03:48:13] <SWPadnos> that sounds high, what range of currents are you sensing?
[03:48:22] <tomp> 2.4 amps
[03:48:49] <jmkasunich> 2.4 * 0.25 = 0.6V = 1.44 watts
[03:49:17] <tomp> limiting an L298 and clamping the enables when it hits ~2.5A ( right 1.5watts thanks )
[03:49:28] <jmkasunich> scrounged tonight makes it tough, seeing as how we don't know what's in your basement
[03:52:18] <tomp> lotsa stuff ;) , oh i do have some current shunts( but those are even bigger than the .15 ohms ) looks like order jameco/digitkey and wait... RS only goes down to 10ohm and that means a lot of pcs in par.
[03:53:04] <SWPadnos> just take a spool of some wired resistor and solder all the pins (on each side) together - you don't even have to remove the paper ;)
[03:53:19] <toastydeath> hahaha
[03:54:14] <jmkasunich> got some thin steel or nichrome wire?
[03:55:46] <tomp> jmkasunich: yep, different kinds... measuring now , thanks good idea
[04:19:51] <Guest622> anyone home?
[04:19:55] <Guest622> Guest622 is now known as reza
[04:20:42] <reza> reza is now known as guest123123
[04:20:51] <guest123123> guest123123 is now known as reza_a
[04:20:54] <tomp> about 2 feet of 28ga copper wire
[04:21:15] <reza_a> this client sucks
[04:21:20] <jmkasunich> tomp: that doesn't seem right
[04:21:26] <reza_a> question for someone - how hard would it be to intruduce feedback into the cnc application
[04:21:35] <reza_a> ala dro
[04:21:47] <SWPadnos> simple, that's how EMC is designed to work
[04:21:53] <jmkasunich> EMC's display is a DRO
[04:22:40] <SWPadnos> tomp, should be closer to 3 feet
[04:22:45] <tomp> hard to measure so little... need a precision current source
[04:22:55] <jmkasunich> or a wire chart
[04:22:55] <reza_a> EMC?
[04:22:55] <SWPadnos> or a 4-sire meter
[04:22:59] <SWPadnos> 4-wire
[04:23:10] <jmkasunich> reza_a: yeah, EMC
[04:23:19] <reza_a> what's emc stand for? (sorry, new to this)
[04:23:28] <jmkasunich> enhanced machine control
[04:23:30] <SWPadnos> reza_a, you are here through the client on linuxcnc.org, aren't you?
[04:23:34] <jmkasunich> this channel is "emc"
[04:24:15] <jmkasunich> tomp: 28 AWG is 66 milliohms per foot
[04:24:19] <reza_a> well, i'm getting a mini mill, and rather that spending a lot of $$ decreasing the backlash I thought that I could add feedback such that the software would compensate for the backlash
[04:24:20] <jmkasunich> http://www.interfacebus.com/Copper_Wire_AWG_SIze.html
[04:24:32] <reza_a> swpadnos - thanks, just noticed it at the top of the screen
[04:24:33] <jmkasunich> reza_a: not that simple
[04:24:40] <SWPadnos> this one says 78.941 mOhm/fthttp://xtronics.com/reference/wire_gauge-ampacity.htm
[04:24:47] <SWPadnos> this one says 78.941 mOhm/ft http://xtronics.com/reference/wire_gauge-ampacity.htm
[04:25:11] <tomp> jmkasunich: got one, and will just lengthen untill i get 0.6 V to drive base of 2n2222 with 4.7K p/u, output will clamp enables
[04:25:17] <reza_a> my thought was that it would query the dro's frequently, and update it's own location accordingly
[04:25:23] <jmkasunich> most minimill folks use steppers, not servos, and they don't use feedback
[04:25:31] <jmkasunich> EMC can use servos, and it can use feedback
[04:25:53] <reza_a> yeah, with a stepper, and it wouldn't be feedback to the stepper, but feedback to the cnc software
[04:25:58] <jmkasunich> but trying to control position using feedback from the table when there is feedback usually causes instability
[04:26:22] <jmkasunich> also, linear feedback devices (as opposed to rotary) are usually very expensive
[04:26:26] <reza_a> so lets say it changes direction, i twould do so slowly and query the actual location to compensate for backlash rather than assuming that 10 pulses translates to a certain distance...
[04:26:43] <jmkasunich> scales for a minimill might cost almost as much as the mill (several hundred dollars each...)
[04:27:01] <reza_a> nah, cheap chinese dro's, can get a set for under $200 (3)
[04:27:23] <reza_a> i dont mind hacking at the code myself, but i'm wondering how condusive it is to be modified to account for the feedback (not looked at the source yet)
[04:27:51] <SWPadnos> it's not the source code that would be the problem
[04:28:04] <jmkasunich> if the scales have quadrature output and the speed isn't too high, you don't need to change code to get the signals into the PC and count them
[04:28:09] <jmkasunich> the problem is doing the control loops
[04:28:27] <SWPadnos> cheap chinese scales (like those calipers with SPI output) are very slow, so they aren't great as feedback devices
[04:28:38] <jmkasunich> you can either believe us when we tell you that backlash inside a control loop is a bad thing, or you can go learn a lot more about motion control
[04:28:53] <reza_a> yeah, but you only really need to query it when backlash is an issue -- i.e. changing direction
[04:29:04] <reza_a> so you do that more slowly; take longer but should be accurate
[04:29:19] <jmkasunich> what you are describing is an idea, not an algorithm
[04:29:59] <reza_a> jmkasunich - true that, but if it's just a function of finding a proper control algorythm, then i'm not worried about that; my quetsion was the difficulty in integrating said algorythm into the cnc application
[04:30:15] <jmkasunich> you'd probably want to do it in HAL
[04:30:14] <SWPadnos> the idea is sound for some types of motion, but it isn't applicable to constant velocity around a circle, for example
[04:30:40] <jmkasunich> you can try all kinds of weird things with HAL, without writing any code
[04:31:08] <reza_a> when i posted this thought to the minimill list, i got the same type of feedback - very negative.. i figure i'll try it, if i fail, i learn something; if it suceeds, then i've figured out a way to do something that others can benefit from
[04:31:18] <reza_a> and i dont mind spending a couple hundred to find out
[04:31:29] <reza_a> btw, what's HAL?
[04:31:31] <jmkasunich> the position commands out of EMC, the feedback to EMC (for display purposes), the step generators, and the encoder signals are all available in HAL, and you can do lots of interesting things with PID loops and such
[04:31:54] <jmkasunich> its the system that EMC uses to connect I/O devices to the control
[04:32:05] <jmkasunich> read the HAL manual
[04:32:17] <reza_a> so i can write code that sits in HAL space to implement said control loop; excellent.
[04:32:21] <jmkasunich> go to the documentation page, its partway downthe page
[04:32:48] <jmkasunich> hal provides sort of a "scripting language" that lets you build complex systems from simpler blocks
[04:33:01] <jmkasunich> we have PID, summers, gain blocks, integrators, differentiators, etc
[04:33:07] <reza_a> thanks all for the help with that; i've got two other unrelated questions
[04:33:10] <jmkasunich> and you can write custom blocks if you want
[04:33:31] <reza_a> 1) why is linuxCNC not as popular as mach and this other one whos name i forget
[04:33:47] <reza_a> and 2)why do cnc applications need to run on real-time os's?
[04:34:03] <jmkasunich> 1) why is linux not as popuar as windoze?
[04:34:25] <SWPadnos> answer 1: because people are phobic of Linux, and may be remembering ho wdifficult it was (5 or 6 years ago) to get a working installation of EMC
[04:34:26] <toastydeath> 2) it is important for the OS not to hang while you are trying to cut something accurately
[04:34:37] <jmkasunich> 2) because its not good when your machine stops in the middle of a cut because the operating system decided it was time to check for email or deframent the disk or whatever
[04:34:56] <SWPadnos> answer 2: because of what they said ;)
[04:35:24] <tomp> reza_a: it's good to think & try new ideas, but before you invest a couple hundred, get a digital caliper in your hands, and move the jaw back and forth and consider that response time to your thought. you may continue, you may not, it's a cheap way to pre-test the idea.
[04:36:07] <eric_u> mach gets away with using windows which is not real time, because the machine stops if the control doesn't get any attention from the cpu
[04:36:16] <reza_a> i've got one already. the update rate is about 1Hz from what I've heard, which isn't a lot, but it might be enough..
[04:36:39] <jmkasunich> I can't speak for chinese DROs that are nothing more than fancy electronic dial calipers, but I stand by my statement that true linear encoders (usually glass scales) are hundreds of dollars per axis
[04:36:49] <reza_a> jmkasunich -with fast enough computers, is that still a problem?
[04:36:52] <eric_u> ebay
[04:36:56] <jmkasunich> EMC normally runs its position loop at 1000Hz
[04:37:25] <SWPadnos> reza_a, we've found that faster is not necessarily better for realtime performance
[04:38:01] <jmkasunich> on a moderately fast milling machine, disruptions of even a couple thousandths of a second would cause noises and bad surface finish
[04:38:05] <jmkasunich> and thats on a servo machine
[04:38:07] <SWPadnos> a late model P3 is usually very good for RT, whereas many multi-GHz P4 and other modern CPUs aren't so hot
[04:38:17] <reza_a> it also seems everyone uses the parallel port to interface w/ cnc equipment -- usb is prevalent and usb->digital output is super easy these days, why isn't it as popular?
[04:38:22] <jmkasunich> on a stepper machine, even 0.0001 seconds will probably cause lost steps
[04:38:32] <SWPadnos> USB can't be used for control with feedback
[04:38:39] <jmkasunich> same thing - usb has latency and realtime issues
[04:38:43] <ds3> USB sucks for RT feedback
[04:38:51] <ds3> USB is a Master/Slave polling system
[04:39:18] <SWPadnos> at least not without a very fast processor at the other end, and still having an RT OS on the PC (unless the controller does all the feedback / PID stuff)
[04:39:18] <reza_a> so, it's that you want the mill to have constant flow and not move inconsistently not that it wouldn't work
[04:39:27] <SWPadnos> it wouldn't work well
[04:39:43] <jmkasunich> losing steps = "not work" to me
[04:39:47] <SWPadnos> that question is more like "would it work for me to tow a house with a bicycle"
[04:39:49] <ds3> I suppose you could do it with an Isochronous end point
[04:40:07] <SWPadnos> the answer is no, ot wouldn't work, unless you really like exercise and don't mind moving only a few feet a day
[04:40:11] <jmkasunich> having a machine make thunking noises as it slams to a stop and begins moving again is "not work" to me
[04:40:24] <reza_a> hmm
[04:41:02] <SWPadnos> the other way of having it work is to drop the idea of feedback, and use steppers that are overkill for the application, so you're more sure of not losing steps
[04:41:10] <reza_a> the output of the application is a series of steps + direction with a duration between steps, right?
[04:41:14] <SWPadnos> then run in open-loop (like Mach does, and EMC can do)
[04:41:16] <jmkasunich> that doesn't solve the backlash issue
[04:41:22] <jmkasunich> reza_a: yes, discrete steps
[04:41:33] <SWPadnos> no, there's no good solution to backlash other than replacing the screws
[04:41:37] <reza_a> what's it mean to run in open-loop?
[04:41:39] <jmkasunich> but for a good machine, there are 10,000 or more steps per inch
[04:41:55] <jmkasunich> open loop = send steps to the motor and trust it to follow them
[04:41:56] <SWPadnos> open loop is like yelling at people to get their work done and assuming they listened
[04:42:06] <reza_a> ah, no feedback
[04:42:13] <jmkasunich> closed loop = measure the motor position, and adjust accordingly (every 1000th of a second)
[04:42:14] <reza_a> but i thought there was no feedback with steppers
[04:42:20] <SWPadnos> closed loop is checking on them and modifying your commands when you see they're not doing what you say
[04:42:21] <jmkasunich> there isn't
[04:42:26] <jmkasunich> steppers = open loop
[04:42:35] <reza_a> ah, servo = feedback
[04:42:42] <reza_a> but will not eliminate backlash
[04:42:49] <reza_a> it's becoming clear to me now
[04:42:53] <jmkasunich> you were discussing the idea of doing closed loop steppers
[04:43:13] <reza_a> yeah, i just didnt know how servo motors worked
[04:43:21] <jmkasunich> servos use feedback from the motor shaft
[04:43:23] <SWPadnos> steppers can't be run the same way as servos due to their torque curves. servos require feedback, steppers can be used with feedback, but it can't really be used to correct position, only to signal an error if the stepper stalls
[04:43:33] <reza_a> and the HAL is sent the step/direction/duration info to send to the motors?
[04:43:40] <jmkasunich> feedback from the table to correct for backlash is theoretically possible, but very very tricky to pull off
[04:43:45] <jmkasunich> not something for a beginner
[04:44:07] <jmkasunich> HAL is given a position from EMC (1000 times a second)
[04:44:19] <jmkasunich> and a HAL module uses that information to generate the steps
[04:44:44] <jmkasunich> for servos, HAL is given a position from EMC _and_ gets a position from the encoder on the motor
[04:44:57] <jmkasunich> then HAL uses a PID loop to decide what corrections to send to the motor
[04:45:03] <jmkasunich> (again, 1000 times a second)
[04:45:05] <reza_a> so, on a completly seperate train of thought, could you build a device (usb-based) that is sent the step/direction/duration info in bulk, it stores it locally at the stepper controller, and drivees it?
[04:45:26] <jmkasunich> you could, and some people have, but that concept doesn't play well with EMC
[04:45:32] <jmkasunich> and in general it has "issues"
[04:45:46] <jmkasunich> if it wasn't rapidly approaching 1am I'd go into more depth
[04:45:55] <reza_a> hmm, not that i'm going to go that route, but i'm just trying to understand why things are done the way they are done
[04:46:23] <reza_a> perhaps i'll catch you online later this week, thanks for the info so far
[04:46:29] <jmkasunich> goodnight
[04:48:25] <SWPadnos> bummer. I guess it is bedtime
[04:48:29] <SWPadnos> night all
[05:10:07] <tomp> "Here are a few stress testing routines proven to be good for determining the worst case latencies." http://www.captain.at/howto-linux-real-time-patch.php (in comparing kernel 2,6, and with xenomai and with rtai )
[05:10:37] <eric_u> that guy has some good stuff on his site
[06:51:54] <anonimasu> 5hm
[06:52:01] <anonimasu> somone talked about chineese scales..(cheap ones)
[06:52:04] <anonimasu> they do 60m/min
[06:52:37] <anonimasu> and max is at 120m/min..
[06:53:38] <anonimasu> jmkasunich/anyone
[06:53:55] <anonimasu> jmkasunich: wouldnt you be able to offset the position with the correction from the linear scales?
[06:54:23] <anonimasu> jmkasunich: if you substract the position of the linear scale to the current position
[06:54:25] <anonimasu> err with
[06:59:04] <ds3> 60m?
[06:59:20] <ds3> as in 60 meters per min?
[06:59:51] <anonimasu> yeah
[07:00:11] <anonimasu> i think 60m/min is for 1um scales..
[07:00:16] <anonimasu> 120 is for the 5um ones..
[07:01:14] <ds3> which ones are you talking about? the "glass" scales or the ones they also sell as a caliper?
[07:01:27] <anonimasu> the glass scales..
[07:01:54] <ds3> Ohhh those
[07:02:16] <anonimasu> yep
[07:02:43] <ds3> they seemed expensive; at least the discounted one from the Shumatech group
[07:03:10] <anonimasu> they are expensive but compared to a few $1000 for a heidenhain scale ;)
[07:03:24] <ds3> heh... it is all relative
[07:03:39] <ds3> wish the 60M/min figure was for the caliper style scales ;)
[07:03:59] <anonimasu> hehe
[07:04:22] <anonimasu> http://www.shumatech.com/support/chinese_scales.htm#Chinese%20Scale%20Protocol
[07:04:32] <ds3> seen that
[07:04:51] <anonimasu> im in the middle of mounting a dro on the lathe :
[07:04:55] <anonimasu> :)
[07:05:16] <anonimasu> got the scales mounted yesterday
[07:05:24] <anonimasu> just need brackets for the pickups
[07:28:33] <ds3> actually, if you want a cheap feed back system, what's wrong with the US Digital encoders?
[07:33:44] <anonimasu> because linear encoders are way better?
[07:34:03] <anonimasu> especially when you have any amount of backslash/pitch errors..
[07:43:25] <anonimasu> though you can probably build your own linear encoders..
[07:46:55] <ds3> US digital sells linear encoders
[07:49:26] <ds3> http://www.usdigital.com/products/em1-heds/
[07:51:31] <anonimasu> ds3: avaiability probably
[07:52:04] <ds3> oh
[07:53:27] <anonimasu> are thoose coplete units or just the linear strip/head?
[08:02:18] <ds3> just head + strip
[08:02:29] <ds3> and the strips are ordered to size
[08:07:26] <chr0n1c> what going on folks?
[08:08:36] <chr0n1c> my friend is looking for an mastercam v9 -> emc post? any pointers?
[08:08:51] <chr0n1c> i looked at the mastercam website and didn't see emc
[08:09:10] <lerneaen_hydra> I doubt you'll find a post on their website
[08:09:33] <lerneaen_hydra> If you're lucky someone else already has made one, if not it shouldn't be hard to make one yourself
[08:09:35] <chr0n1c> they have a few but nothing named emc
[08:10:05] <lerneaen_hydra> if you can find one called "standard ISO" or something like that then it should be relatively easy to modify for emc
[08:10:35] <chr0n1c> i tried a generic fanuc and it plunged my z into my mdf table
[08:10:37] <chr0n1c> woohoo
[08:10:50] <chr0n1c> it looked good on the display
[08:10:53] <chr0n1c> but oops
[08:11:00] <lerneaen_hydra> ouch
[08:11:11] <lerneaen_hydra> didn't you see that in the preview plot?
[08:11:11] <chr0n1c> i know a little g-code but i'm not proficient
[08:11:21] <chr0n1c> it didn't show up
[08:11:28] <chr0n1c> i thought i had everything seroed
[08:11:32] <chr0n1c> zeroed
[08:11:34] <anonimasu> chr0n1c: the fanuc one works..
[08:11:35] <anonimasu> kindof..
[08:11:44] <anonimasu> :)
[08:11:49] <chr0n1c> but! i did manage to engrave the axis demo cods!
[08:12:00] <lerneaen_hydra> afaik the only changes I had to do when making an edgecam post from a generic ISO post was changing the feed mode, inch/mm, and similar codes
[08:12:02] <chr0n1c> in plastic and into my mdf table for the first run
[08:12:16] <anonimasu> lerneaen_hydra: what kinf of post(machine) was it for?
[08:12:26] <lerneaen_hydra> std 3-axis mill
[08:12:36] <lerneaen_hydra> the lathe was also easy to do
[08:12:37] <chr0n1c> hmm...
[08:12:48] <chr0n1c> i am building a lathe next
[08:12:53] <lerneaen_hydra> except for threading, with non pure X/Z moves
[08:12:57] <chr0n1c> after i get this mill tweaked and running smooth
[08:13:06] <lerneaen_hydra> as most machines define pitch as movement in Z per rev
[08:13:19] <lerneaen_hydra> and not movement in sqrt(x^2+z^2)
[08:14:07] <chr0n1c> i can draw some i mean my friend can draw some amazing things in mastercam v9, he only wishes he had a post for emc ;)
[08:14:18] <anonimasu> the only problem with my mcam post is because it misses a g0 move after a drill..
[08:14:48] <anonimasu> rather omits it because emc should remember that the last command before the cycle was a g0
[08:15:05] <chr0n1c> can i borrow your post?
[08:15:08] <chr0n1c> :D
[08:15:24] <anonimasu> ol
[08:15:32] <anonimasu> it's just the fanuc one that's a bit modified..
[08:15:33] <chr0n1c> i looked at the file once to try and edit it for my mx2 machine
[08:15:39] <anonimasu> but sure..
[08:15:41] <chr0n1c> but i ended up just editing them by hand
[08:15:44] <anonimasu> got a mail?
[08:16:01] <anonimasu> if you pm me it
[08:16:01] <anonimasu> :)
[08:17:48] <chr0n1c> thank you
[08:18:48] <chr0n1c> i need to take pictures of this crazy machine i built
[08:18:53] <chr0n1c> it's pretty slick really
[08:19:25] <chr0n1c> precision linear slides... a cheap 3 axis stepper driver...
[08:19:50] <chr0n1c> my z axis motor cable is an old PS2 controller cord with the ends cut off
[08:20:11] <chr0n1c> the z and y cables are old cat 5 cables
[08:20:46] <chr0n1c> the motors aren't worthy of the slides but they do the trick for now
[08:21:41] <chr0n1c> my z axis motor couple is 3/16 fuel line from autozone and 4 wire ties (my homebrew 3/16 to 1/4 adapter)
[08:23:25] <lerneaen_hydra> ouch there's gotta be a lot of backlash in that
[08:23:36] <chr0n1c> not really
[08:23:56] <chr0n1c> it's only about 200 thousandths gap that it has to span
[08:24:23] <chr0n1c> it's double wwall auto fuel line
[08:24:31] <chr0n1c> they sell it by the foot
[08:25:53] <anonimasu> :)
[08:26:35] <chr0n1c> http://www.cnczone.com/forums/showthread.php?p=279964
[08:26:45] <chr0n1c> i just found that link
[08:26:53] <chr0n1c> for mcam and emc
[08:28:31] <anonimasu> yep
[09:59:36] <anonimasu> I just bought panel buttons
[10:07:40] <alex_joni> nice
[10:07:49] <anonimasu> 4.3eur each :/
[10:09:05] <anonimasu> now I need a panel mount monitor
[10:10:54] <anonimasu> http://www.elfa.se/images/highres/h1623.jpg
[10:44:41] <lerneaen_hydra> anonimasu: haha, I also have used those buttons before
[10:45:10] <lerneaen_hydra> err
[10:45:16] <lerneaen_hydra> I've have also
[10:45:20] <lerneaen_hydra> gah
[10:45:26] <lerneaen_hydra> I have also
[10:59:37] <anonimasu> how are they?
[11:01:03] <anonimasu> are they worth the cash?
[11:01:53] <anonimasu> lerneaen_hydra: I'm going to engrave them.. with X Y Z and such
[11:02:05] <lerneaen_hydra> they're OK i guess
[11:02:34] <anonimasu> did you find anything better?
[11:02:37] <lerneaen_hydra> they don't feel extremely robust, but they're quite alright as long as you dont hit them really hard
[11:03:04] <anonimasu> I guess they will do fine
[11:03:06] <lerneaen_hydra> I've got a hold of surplus industrial buttons (the kind that are found on PLC boxes and the like)
[11:03:23] <lerneaen_hydra> I don't think they'll like chips a lot
[11:03:53] <lerneaen_hydra> http://www.stephenjames.biz/images/ButtonBox.jpg <-- those kinds of buttons
[11:04:28] <anonimasu> lerneaen_hydra: well, that's good and all..
[11:04:32] <anonimasu> but not really useable..
[11:04:44] <lerneaen_hydra> what are they going to be attached to?
[11:04:47] <anonimasu> this is for a panel next to the monitor
[11:04:51] <lerneaen_hydra> oh
[11:04:52] <anonimasu> spindle/toolchanging/stuff
[11:04:56] <lerneaen_hydra> then those will be fine
[11:05:01] <anonimasu> I'm copying a heidenhain panel..
[11:05:03] <anonimasu> ;)
[11:05:06] <lerneaen_hydra> ah :p
[11:05:14] <anonimasu> the one at the mill at work..
[11:05:27] <lerneaen_hydra> they're a bit softer then heidenhain buttons
[11:05:32] <lerneaen_hydra> similar otherwise
[11:05:34] <anonimasu> yep
[11:06:00] <anonimasu> I wanted the low build height
[11:06:02] <anonimasu> :)
[11:06:11] <anonimasu> I think I'll handle spindle on in the plc..
[11:06:44] <anonimasu> so that the machine cant start the spindle entirely on it's own if I dont hit NC START on the panel that goes to the plc..
[11:06:48] <anonimasu> and to the pc :)
[11:07:37] <anonimasu> I trust the plc's I program more then I trust emc not to kill me..
[11:08:43] <lerneaen_hydra> is it a real hardware PLC?
[11:09:02] <anonimasu> yes
[11:09:16] <anonimasu> I have a spare off work.. or a few rather :)
[11:09:36] <lerneaen_hydra> nice
[11:09:43] <lerneaen_hydra> how much do they usually cost new?
[11:09:46] <lerneaen_hydra> 10k sek?
[11:09:53] <anonimasu> well, depends..
[11:10:02] <anonimasu> the setup we use for our machines are 200eur for the plc..
[11:10:07] <anonimasu> and 550eur for the io modules..
[11:10:23] <anonimasu> 250.. for the plc.. actually
[11:10:24] <lerneaen_hydra> does that include the PSU and case?
[11:10:26] <anonimasu> no
[11:10:33] <anonimasu> the io module is IP something..
[11:10:47] <lerneaen_hydra> so 10k or thereabouts for everything?
[11:10:50] <anonimasu> 20 pwm/analog in/out
[11:10:58] <lerneaen_hydra> nice :)
[11:12:02] <anonimasu> thoose modules are really sweet..
[11:12:19] <anonimasu> the plc's have serial port and can interface..
[11:12:54] <anonimasu> 130x110x25
[11:12:56] <anonimasu> mm
[11:13:50] <lerneaen_hydra> that thin?
[11:13:54] <lerneaen_hydra> 25mm?
[11:13:56] <anonimasu> yeah
[11:14:00] <anonimasu> that's just the cpu..
[13:39:44] <tomp> interesting mind-map like java app, uses cairo & java http://relations-rcp.sourceforge.net/download.html
[16:01:57] <Guest376> jepler: is that you?
[16:02:10] <Guest376> Guest376 is now known as skunkworks_
[16:02:26] <jepler> skunkworks_: yep
[16:02:32] <skunkworks_> back in the usa?
[16:02:43] <jepler> live from goettingen germany in a small internet "cafe"
[16:02:49] <skunkworks_> Nice
[16:02:59] <skunkworks_> How is the trip so far?
[16:04:20] <jepler> pretty good. some of the accomodations cost more than I'd like but the rest has been good
[16:04:41] <jepler> the trip's about half over now
[16:04:50] <jepler> how are you doing?
[16:05:10] <skunkworks_> Moved in :) thats about it.
[16:05:22] <skunkworks_> that is a nice trip if it is only half done so far
[16:06:03] <jepler> yeah -- we only go every few years, so we have to take more time
[16:06:27] <jepler> just got done with the least interesting (for me) part of the trip, visiting Ingrid's relatives in Dresden.
[16:06:43] <jepler> I like them just fine but few of them speak more than a few words of english, and I don't speak much german
[16:06:53] <jepler> and how is emc doing_
[16:06:56] <jepler> ?
[16:08:20] <skunkworks_> Doing good a there doesn't seem to be much activity on here.
[16:08:36] <jepler> I guess with me out of town ... :-P
[16:08:37] <skunkworks_> Doing good as there..
[16:08:40] <skunkworks_> could be ;)
[16:08:47] <jepler> everybod's loafing
[16:08:56] <skunkworks_> no one cracking the whip
[16:15:07] <SWPadnos> hey jepler, can yo usay hi to my sister while you're there? :)
[16:16:34] <jepler> SWPadnos: where is she, then?
[16:16:47] <SWPadnos> err - well that I don't know, but she lives in Gottingen
[16:16:54] <jepler> oh really
[16:17:22] <SWPadnos> yep. works at a hospital or physical rehabilitation clinic or something
[16:17:23] <jepler> what a coincidence
[16:17:28] <SWPadnos> heh
[16:17:49] <skunkworks_> jepler: did you rent a car?
[16:18:00] <SWPadnos> trains, baby
[16:18:07] <jepler> skunkworks_: no, we travel between cities on the train, and in cities by whatever means seems most appropriate
[16:18:28] <SWPadnos> U-bahn or S-bahn seem to work well
[16:19:06] <jepler> streetcars and buses too
[16:19:17] <SWPadnos> "busbahn" :)
[16:19:31] <SWPadnos> aren't the streetcars called S-bahn?
[16:19:37] <SWPadnos> (surface-bahn, for the most part)
[16:19:46] <jepler> SWPadnos: not in dresden
[16:19:54] <SWPadnos> oh, ok. never been there
[16:19:59] <jepler> I think berlin has separate s-bahn and street cars too though
[16:20:08] <SWPadnos> hmmm. never been there either ;)
[16:20:23] <jepler> SWPadnos: if you're serious about your sister, we have a handy so we could give her a call.
[16:20:32] <SWPadnos> now that would be funny
[16:20:55] <jepler> tough I don't have it right now and I don't seem to have anywhere to write it down either
[16:21:08] <SWPadnos> I need to see if she's there or here - she was in Galveston a couple of weeks ago (so was I) - I'm not sure she's gotten back to Germany yet
[16:21:10] <SWPadnos> heh
[16:21:15] <SWPadnos> IRC log ;)
[16:21:26] <jepler> but later I'll be away from the computer
[16:21:38] <jepler> it's just an internet "cafe", but not coffee in sight unfortunately
[16:21:41] <SWPadnos> bummer
[16:21:46] <jepler> oh well
[16:21:53] <SWPadnos> German coffee often takes getting used to, I found
[16:22:14] <jepler> 've found that if I order milchkaffe I get something I can drink
[16:22:19] <SWPadnos> heh
[16:22:22] <SWPadnos> yep
[16:22:28] <SWPadnos> or kaffe mit sahne
[16:22:38] <jepler> even better
[16:22:39] <a-l-p-h-a> 'milchkaffe'?
[16:22:40] <SWPadnos> oops: oder kaffe mit sahne
[16:22:49] <SWPadnos> coffee milk
[16:22:49] <a-l-p-h-a> kaffe, that's a really sweet sugary beer.
[16:22:56] <SWPadnos> not in Deutschland
[16:23:03] <a-l-p-h-a> "keyffe"?
[16:23:08] <jepler> a-l-p-h-a: think cafe latte
[16:23:20] <a-l-p-h-a> jepler, ooooh.
[16:23:28] <jepler> (no sugar by default)
[16:23:32] <SWPadnos> the first thing I learn when traveling is how to order coffee that may be close to the way I like it :)
[16:23:45] <a-l-p-h-a> dunno, had this beeer, that tasted really sugary.
[16:24:00] <a-l-p-h-a> SWPadnos, :)
[16:24:18] <a-l-p-h-a> you guys should try a "black tea latte", damn good at starbucks...
[16:24:19] <SWPadnos> so I can now ask for coffee in German, Spanish, French, and Polish (and of course English and American)
[16:24:22] <a-l-p-h-a> green tea latte, sucks.
[16:24:35] <jepler> SWPadnos: if you want, e-mail me the number and if I get back to my mail in time I'll call
[16:24:49] <SWPadnos> ok. that would be funny
[16:25:03] <jepler> anybody know what "MARIO." is talking about?
[16:25:19] <SWPadnos> a-l-p-h-a, try getting half green tea (iced tea) and half lemonade sometime - it's excellent
[16:25:19] <jepler> something about the documentation saying to never use arcs...?
[16:25:22] <a-l-p-h-a> SWPadnos, "ga-pheh" is Cantonese.
[16:25:32] <SWPadnos> I saw that (both copies), but have no idea what the heck he's referring to
[16:25:50] <a-l-p-h-a> deja vu...
[16:25:58] <SWPadnos> a-l-p-h-a, good - it almost sounds like "coffee" :)
[16:26:20] <a-l-p-h-a> SWPadnos, yup.
[16:26:31] <a-l-p-h-a> in Mandarin, I've got no clue. it's not in my vocab.
[16:26:36] <SWPadnos> mine either ;)
[16:27:04] <SWPadnos> this is how much Mandarin I know:
[16:27:07] <SWPadnos> (that was it)
[16:27:18] <a-l-p-h-a> "double double", "single single", "triple triple", "black", people know what I'm getting Tim Hortons. :)
[16:27:31] <SWPadnos> ewww - Tim Horton's
[16:27:35] <a-l-p-h-a> McDonald's coffee isn't actually bad these days... they changed their beans, and roasting, and it tastes good.
[16:27:37] <SWPadnos> what nasty coffee they have
[16:27:43] <a-l-p-h-a> SWPadnos, BLASPHAMY!
[16:27:47] <SWPadnos> oh, that's a surprise
[16:28:04] <a-l-p-h-a> Starbuck's coffee is too dark for my tastes.
[16:28:09] <a-l-p-h-a> their dark/strong blends.
[16:28:09] <SWPadnos> McD used to be pretty good, then they decided that scalding was more important than flavor, and I haven't tried any since
[16:28:16] <a-l-p-h-a> their light blends are more palletable (sp) to me.
[16:28:25] <SWPadnos> yeah, StarBucks is like McD - not great, but you know what you'll get
[16:28:31] <a-l-p-h-a> SWPadnos, after they got sued, it's back to good.
[16:28:41] <SWPadnos> yeah. if you try cutting it with more milk, it reduces the acidity
[16:28:58] <a-l-p-h-a> SWPadnos, I try to use creme...
[16:29:05] <a-l-p-h-a> damn it... haven't had one today.
[16:29:06] <a-l-p-h-a> bbiab.
[16:29:26] <SWPadnos> heh
[16:29:41] <SWPadnos> use cream for lightness and flavor, and some extra milk to cut acidity
[16:42:27] <jepler> nice talking to you guys .. I'l be back home before you know it
[16:42:45] <a-l-p-h-a> you'
[16:42:50] <a-l-p-h-a> re not home yet.
[16:42:54] <a-l-p-h-a> are you there yet?
[16:42:57] <a-l-p-h-a> now?
[16:42:58] <a-l-p-h-a> how about now?
[16:43:11] <jepler> you guys are all so funny
[16:43:18] <a-l-p-h-a> nah... just looking
[16:43:32] <a-l-p-h-a> I think I'm getting sick.
[16:43:37] <jepler> a-l-p-h-a: what I meant was, I'll be in my house before I announce it on irc, so I'll be home before you know it
[16:43:47] <SWPadnos> heh
[16:43:48] <SWPadnos> sneeeky
[16:43:52] <a-l-p-h-a> my sinuses are messed, and my ears feel like I've just gone up or down in elevation very very fast.
[16:43:52] <jepler> (unless I underestimate the surveillance you guys perform on me=
[16:43:53] <jepler> )
[16:44:01] <SWPadnos> cradek will be on the lookout
[16:44:14] <SWPadnos> damn. I wasn't supposed to say anything
[16:44:19] <a-l-p-h-a> jepler, are you wearing clothing with buttons?
[16:44:25] <a-l-p-h-a> we have gps button trackers.
[16:45:15] <jepler> a-l-p-h-a: in europe, everyone goes naked
[16:45:15] <jepler> no buttons
[16:45:19] <a-l-p-h-a> I created a pool, to gather money to replace all the buttons on your cltohes with small gps trackers... cost us a forture... but nothing but the best for our jepler.
[16:45:43] <a-l-p-h-a> doh... no ip to track you with either.
[16:45:51] <jepler> bye now
[16:45:52] <jepler> -quit
[16:45:53] <a-l-p-h-a> I was gonna visual trace route you. :)
[16:46:00] <a-l-p-h-a> it';s /quit
[16:46:00] <a-l-p-h-a> ;)
[16:52:21] <a-l-p-h-a> SWPadnos, what are you up to right now?
[16:52:33] <SWPadnos> thinking about designing a PC board
[16:52:49] <SWPadnos> and catching up on emails from the geckodrive and CCED lists
[16:53:26] <a-l-p-h-a> fun.
[16:53:32] <a-l-p-h-a> how about a little crafty project? http://www.instructables.com/id/EU0DY6QKNCEZ7BHEZ0/?ALLSTEPS
[16:53:38] <SWPadnos> depressing, actually
[16:54:01] <SWPadnos> there are so many places where I'd love to reply "well, with EMC2 and xxyyzz hardware, it would probably be a lot easier"
[16:54:10] <SWPadnos> but it's the gecko list, so I don't want to push other hardware
[16:54:12] <a-l-p-h-a> hehehe
[16:54:22] <SWPadnos> and nobody listens anyway when you mention Linux apps
[16:54:38] <a-l-p-h-a> oh... I wonder what's the progress on the 5i20, and step/dir.
[16:54:43] <a-l-p-h-a> is
[16:54:49] <SWPadnos> jmk has been working on it quite a bit
[16:55:00] <a-l-p-h-a> haven't heard a peep from him latley.
[16:55:04] <a-l-p-h-a> then again, I haven't been online.
[16:55:08] <SWPadnos> he has some pretty low resource usage stepgens designed, and several other components
[16:55:09] <a-l-p-h-a> been too busy with making money. :(
[16:55:15] <SWPadnos> well, that is a good thing
[16:55:26] <a-l-p-h-a> but I don't have enough time for myself.
[16:55:34] <a-l-p-h-a> gf takes up the free time.
[16:55:56] <a-l-p-h-a> I think it's her plan, so I can't date other girls.
[16:56:03] <SWPadnos> heh
[19:54:49] <Rugludallur> hmmm I just discovered something a bit odd about "home all", it turns out that if you define sequence but don't use 0 it's silently ignored
[19:55:38] <Rugludallur> so if you only have HOME_SEQUENCE=1 and HOME_SEQUENCE=2 "Home All" fails silently
[20:17:11] <alex_joni> Rugludallur: oh, that doesn't sound nice
[20:26:59] <Rugludallur> alex_joni: i'll bug it :D
[20:44:23] <alex_joni> Rugludallur: I've been looking over the code
[20:44:37] <alex_joni> I think it's a mistake to start at 1 with the sequences
[20:45:48] <SWPadnos> ideally, both home sequence and axis (joint) numbers would start at 1, since that's how non-conputer-people count :)
[20:45:57] <SWPadnos> s/n/m/
[20:46:43] <alex_joni> SWPadnos: and make them joints not axes in the ini
[20:46:49] <alex_joni> and, and, and..
[20:46:51] <SWPadnos> that would be nice too
[20:46:55] <alex_joni> but it would confuse the hell out of existing users
[20:46:55] <SWPadnos> and this other thing ...
[20:47:06] <SWPadnos> well, they're already experts, so ...
[20:47:07] <SWPadnos> ;)
[20:47:34] <SWPadnos> we can also implement the "ini version" idea, and actually use different names in later versions of the ini file
[20:48:01] <alex_joni> still makes it hard to upgrade
[20:48:21] <SWPadnos> no, since you leave the old style ini reading code alone
[20:48:41] <SWPadnos> so you can make no changes, or you can "upgrade" the ini
[20:50:47] <alex_joni> hmm
[20:53:36] <SWPadnos> actually, this may be easier with the new ini class
[20:53:44] <SWPadnos> easier to implement, that is
[20:53:57] <alex_joni> right.. just inherit it for the new one
[20:54:01] <SWPadnos> hmmm. though the "external" programs, such as GUIs don't use the new ini code, do they?
[20:54:10] <alex_joni> nope, they don't
[20:54:18] <alex_joni> ideally they shouldn't access the ini at all
[20:54:29] <alex_joni> and get it all from emc through NML
[20:54:33] <SWPadnos> that would be ideal
[20:54:36] <alex_joni> even table and var files and whatever
[20:55:02] <alex_joni> that way it's way easier for remote GUIs
[20:55:17] <SWPadnos> indeed
[20:55:37] <SWPadnos> I wonder if we should make a list of "pet peeves" on the wiki, for consideration during (or before) fest
[21:00:13] <feoc> i vote for emc's inability to tune its own PID loops to be on that list
[21:00:13] <feoc> lol
[21:00:38] <SWPadnos> nothing can tune its own PID loops, sorry ;)
[21:00:41] <alex_joni> feoc: that's actually doable
[21:00:50] <SWPadnos> at least nothing anyone I know can afford
[21:00:57] <alex_joni> but you need to code the results your after somehow
[21:01:02] <alex_joni> and that's the tricky part
[21:01:34] <SWPadnos> it's relatively easy to get something that'll tune a servo, if you give it reasonable limits
[21:02:05] <feoc> dunno if its possible
[21:02:26] <feoc> but if it could do it more itself would make life easyer for stupid people like me
[21:04:33] <feoc> didnt get around to doing any tuning on the mill today
[21:04:48] <anonimasu> hm
[21:04:53] <feoc> had to wire in a circuit for a powder coating oven in the end :(
[21:11:40] <alex_joni> good night all
[21:11:46] <Rugludallur> night alex
[21:12:55] <SWPadnos> sorry - got a phone call
[21:13:22] <SWPadnos> the problem with PID tuning is that you must give any automatic application very good information for it to do its job
[21:13:59] <feoc> i suppose
[21:14:09] <SWPadnos> it's also much easier for a person to "see" what a scope plot means - things like ringing and settling time are trivial fora person to see, but not so easy for a computer
[21:14:30] <feoc> not so trivial for me :(
[21:14:46] <SWPadnos> heh
[21:15:21] <feoc> ah well ill get there im sure
[21:15:20] <SWPadnos> I think there's a ferror-out output from the motion controller for each axis. if you scope that, you'll see ringing pretty clearly
[21:15:34] <feoc> ok
[21:16:33] <SWPadnos> I'm building the latest from CVS to check that - hold on one sec
[21:16:41] <feoc> is there a general sorta set of figures im to expect when tuned?
[21:16:47] <feoc> or can it be anything
[21:17:52] <SWPadnos> ok, the parameters to look at are axis.0.f-error (or axis.1, axis.2 ...)
[21:18:41] <SWPadnos> look at page 3 of this manual: http://www.geckodrive.com/photos/G320_REV-4_MANUAL.pdf
[21:19:02] <SWPadnos> the plots aren't perfect, but they're close to what you would expect from a plot of axis.n.f-error
[21:19:13] <SWPadnos> (possibly upside down though)
[21:20:41] <feoc> ok
[21:21:20] <SWPadnos> incidentally, his instructions are close to what you need to do for EMC as well. increase P until you see oscillation, then reduce it a bit and increase D until the settling time is acceptable
[21:21:51] <SWPadnos> then throw in a bit of I and you're all set
[21:22:08] <SWPadnos> until you want to add FF terms, to make things just that little bit better ;)
[21:22:37] <feoc> FF terms?
[21:23:11] <SWPadnos> feedforward terms. they let the PID loop use changes in the commanded position as well as errors between commanded and actual (feedback)
[21:23:39] <feoc> is it normal for the motors to whine ?
[21:23:40] <SWPadnos> they can safely be left at 0, but if you want a bit of extra performance, you can tweak those
[21:23:47] <SWPadnos> what kind of drives?
[21:24:08] <feoc> analog, velocity
[21:24:22] <feoc> the servomate controller
[21:24:35] <SWPadnos> well, I suppose it doesn't really matter - they're servos with encoder feedback
[21:24:48] <SWPadnos> they can whine a bit - they call it "singing"
[21:24:58] <feoc> oh ok
[21:25:26] <SWPadnos> it's due to the fact that the drive gets no feedback until the encoder moves one count, so it ping-pongs back and forth between two encoder edges (also called dithering)
[21:25:26] <feoc> as long as its not a prob with the servo control
[21:25:51] <SWPadnos> it could be, but some noise is normal. it should go down if the drive is properly tuned though
[21:25:56] <feoc> ok
[21:26:05] <SWPadnos> you need to tune the drive before you tune EMC ...
[21:26:08] <feoc> it does it even without the computer attached
[21:26:15] <feoc> ok
[21:27:13] <feoc> it seems ok
[21:27:19] <feoc> its just the singing
[21:27:59] <SWPadnos> you should look up the tuning procedure for the drive first. if that's not properly tuned, EMC won't be able to use the motor to its fullest potential
[21:28:16] <SWPadnos> regardless of how much of an expert on EMC tuning you become ;)
[21:29:05] <feoc> ok
[21:29:42] <feoc> iv not altered the X axis from when i got the mill and ripped out the old hurco computer
[21:30:14] <feoc> other than a minor adjustment on balance to stop it creeping slightly with emc attached
[21:31:09] <SWPadnos> hmmm. that should be an adjustment you make in EMC - there are offset settings for the DACS
[21:33:33] <feoc> it was only very minor
[21:34:06] <feoc> did it without emc loaded too iirc
[21:34:19] <SWPadnos> ok, that's something to correct in the drive then
[21:35:23] <feoc> yah was just a small tweak on a pot screw
[21:43:16] <feoc> glad i started on the mill first instead of the lathe
[21:43:26] <feoc> that thing is gonna be a huge job
[21:44:10] <feoc> hydraulics all over the place
[21:44:22] <SWPadnos> heh - that should be fun
[21:44:26] <SWPadnos> for some value of "fun"
[21:44:28] <feoc> yeah
[21:44:45] <feoc> has a lot of stuff on it
[21:44:48] <SWPadnos> have you seen this: http://www.expertune.com/tutor.html
[21:45:05] <feoc> iv not
[21:45:22] <SWPadnos> that has some nice graphs and explanations of the parts of PID
[21:49:06] <feoc> yeah
[21:54:59] <Rugludallur> hmm is it possible that postgui halfiles aren't able to solve INI variables ?
[21:55:26] <SWPadnos> they should have access to ini vars the same as any other HAL file (I think)
[21:55:28] <Rugludallur> s/solve/resolve/g
[21:56:04] <Rugludallur> SWPadnos: Looks like there might be a bug
[21:56:13] <SWPadnos> postgui or shutdown?
[21:56:23] <SWPadnos> I think axis is the only UI that runs the postgui file
[21:56:24] <feoc> gnight all
[21:56:29] <SWPadnos> see you feoc
[21:56:31] <Rugludallur> nite
[21:56:48] <Rugludallur> SWPadnos: I'm using Axis so ..., and the rest of the stuff from the postgui file is fine
[21:56:58] <SWPadnos> ok. one sec
[21:57:05] <Rugludallur> SWPadnos: and my variable pin assignments in hal seem to be fine to so ,, I suspect bug
[21:58:37] <SWPadnos> ok, it is a bug. axis doesn't pass along the ini file info when it runs the hal file
[22:00:41] <Rugludallur> SWPadnos: want me to report it ?
[22:01:01] <SWPadnos> I think I'll try fixing it, so don't worry about a bug report yet
[22:01:14] <Rugludallur> SWPadnos: ok :)
[22:08:59] <Rugludallur> hmm another thing I'm noticing is that there does not seem to be any way to set a initial/default value for pyvcp components
[22:12:12] <Rugludallur> and changes to spin buttons that are typed in don't get commited unless the button is incremented/decremented, enter or focus out does nothing
[22:21:36] <__eric__> Good afternoon everyone.
[22:22:12] <Rugludallur> hello
[22:23:08] <Rugludallur> What do you guys think would be the "correct" way to enable default/inital values for pyvcp components, allow them to be set in xml or as hal parameters ?
[22:23:30] <__eric__> I'm trying to run axis-remote, um, remotely and am receiving an error when I try to have it open a file:
[22:23:45] <__eric__> _tkinter.TclError: no application named "axis"
[22:24:40] <SWPadnos> Rugludallur, I think defaults should be in the XML file
[22:25:00] <SWPadnos> possibly loadable from the ini, in a way similar to the way HAL files do it
[22:25:24] <SWPadnos> __eric__, sorry - can't help with axis-remote
[22:25:48] <Rugludallur> SWPadnos: ok, that's the way it was with halvcp and I think I agree, although there might be instances where one might want to set default values with INI variables :D
[22:25:51] <__eric__> SWPadnos: ok, thanks.
[22:26:03] <Rugludallur> __eric__: neither can I
[22:26:21] <SWPadnos> __eric__, others may be able to help, but the main axis developer is on vacation fo r a week or two
[22:26:23] <__eric__> Anyone else know how I can have axis load a file if I'm ssh'ing in to the machine?
[22:26:51] <SWPadnos> is this with an installed emc or something compiled and run-in-place?
[22:26:52] <__eric__> SWPadnos: Where's he vacationing? ;)
[22:26:58] <SWPadnos> Europe
[22:27:40] <toastydeath> you know
[22:27:55] <toastydeath> in the vein that someone was talking about weeks ago, re: emc and an mp3 player on the machine
[22:28:01] <SWPadnos> me, me
[22:28:04] <toastydeath> you could rig a disco ball up to the spindle
[22:28:16] <toastydeath> party central.
[22:28:24] <SWPadnos> and with the proper sensors, you could do spindle-synced motion and have it move to the beat
[22:28:31] <toastydeath> boom tsh boom tsh
[22:28:40] <toastydeath> and have different lights on the table
[22:28:45] <Rugludallur> toastydeath/SWPadnos: and shoot people with laser beams
[22:28:46] <toastydeath> so you could change lighting schemes by moving that
[22:28:54] <toastydeath> we can already do that
[22:28:59] <toastydeath> we're looking for something NEW here
[22:29:00] <toastydeath> geez
[22:29:05] <toastydeath> lasers: officially passe
[22:29:23] <Rugludallur> toastydeath: somehow killing people with jets of water just does not sound as good
[22:29:30] <toastydeath> JETS OF PURE DEATH
[22:29:54] <toastydeath> this is very serious business.
[22:30:03] <Rugludallur> lol, waterpistol gets a whole new meaning at 500bar pressure
[22:32:23] <Rugludallur> to bad there isn't anyone here who could help him :(
[22:32:41] <toastydeath> also why are medium format digital cameras hellaciously expensive
[22:32:58] <toastydeath> i am going to have to start a "give me one free" campaign
[22:33:10] <toastydeath> wherein i lobby for a free medium format digital, for me
[22:34:25] <toastydeath> WHAT IS GOING ON.
[22:38:04] <SWPadnos> toastydeath, you can't have a medium format camera
[22:38:15] <SWPadnos> maybe film, but not digital
[22:38:39] <toastydeath> you can
[22:38:49] <SWPadnos> I know I can, but YOU can't ;)
[22:38:51] <toastydeath> oh =(
[22:38:53] <SWPadnos> heh
[22:39:07] <toastydeath> yeah, because i don't have 12000 for a digital back
[22:39:12] <SWPadnos> you may be able to get a Hasselblad H1 cheap, since the H3 (or so) is out now
[22:39:26] <toastydeath> i've seen the H1's go w/ digital back for around 8k?
[22:39:27] <SWPadnos> but cheap is still in the $10k+ range
[22:39:33] <toastydeath> used, mind you
[22:39:34] <SWPadnos> oh, that's not bad at all
[22:39:36] <SWPadnos> of course
[22:39:40] <toastydeath> yeah
[22:39:45] <SWPadnos> you could buy a Mercedes for a new one
[22:39:56] <toastydeath> the difference is I could make money with one
[22:39:57] <toastydeath> and not the other
[22:40:03] <SWPadnos> hmmm
[22:40:08] <Rugludallur> SWPadnos: who would give the ok to submit an "INIT" patch to pyvcp ?
[22:40:14] <SWPadnos> you could have a Mercedes taxi (like in Poland)
[22:40:23] <toastydeath> yeah, but then i'd have to drive
[22:40:30] <toastydeath> and that's a crappy job, even if it's in a decent car
[22:40:39] <SWPadnos> Rugludallur, probably one of the board members, like cradek, jmkasunich, or alex_joni
[22:41:04] <SWPadnos> true, with a hasselblad, you can take pictures of scantily clad women, which is much more fun
[22:41:18] <SWPadnos> and maybe even make money
[22:41:22] <Rugludallur> SWPadnos: kk I think I'll go ahead and make the changes, see if they will take them later
[22:41:26] <toastydeath> that's actually what models are starting to ask me for
[22:41:35] <toastydeath> which made me look into medium format stuff
[22:41:50] <toastydeath> but i will have to have a lot of that nonsense to earn enough to pay for one
[22:41:58] <toastydeath> ONE STEP AT A TIME
[22:42:08] <SWPadnos> Rugludallur, ok
[22:42:29] <SWPadnos> toastydeath, unless you're doing billboards, you can get by with a high end 35mm-format camera
[22:42:43] <toastydeath> yeah.
[22:42:59] <toastydeath> my d50 is doing well for now
[22:43:03] <SWPadnos> the new Canons are excellent (the EOS-1DS mark 2)
[22:43:11] <SWPadnos> I think that's their 16 MP unit
[22:43:14] <toastydeath> i am too poor to afford serious equipment
[22:43:25] <SWPadnos> Nikon probably has a higher resolution version of the D1/D2 line
[22:43:34] <toastydeath> the 1ds mk2 is a medium-format sensor
[22:43:39] <toastydeath> no nikon has that yet
[22:43:46] <SWPadnos> no, it's a full frame 35mm sensor
[22:43:50] <toastydeath> oh
[22:43:51] <lunix_> if you don't mind me asking what are cameras for? in the EMC context or simple off topic?
[22:43:51] <toastydeath> hmm
[22:43:55] <SWPadnos> off topic
[22:44:00] <lunix_> hehe
[22:44:02] <lunix_> :)
[22:44:20] <lunix_> i was trying to get an angle on this, lol
[22:44:24] <SWPadnos> in an EMC context, we'd be talking about much lower resolution and much higher frame rates
[22:44:26] <SWPadnos> heh
[22:44:29] <toastydeath> hahahah.
[22:44:43] <lunix_> SWPadnos: that was smarty
[22:44:56] <Rugludallur> http://www.youtube.com/watch?v=gSHToyGIxWE
[22:45:08] <toastydeath> i am turning down paid work
[22:45:13] <toastydeath> to build a portfolio
[22:45:17] <toastydeath> maybe manufacturing is not for me!
[22:45:40] <toastydeath> or then again maybe it is
[22:47:31] <SWPadnos> the photo world is a PITA, from my experience
[22:47:44] <toastydeath> yeah, that's what i hear
[22:47:58] <lunix_> PITA?
[22:48:01] <SWPadnos> I like to travel, but schlepping along 20 bags of equipment isn't the greatest method
[22:48:03] <toastydeath> pain in the ass
[22:48:16] <toastydeath> i've been doing nothing but studio type stuff
[22:48:30] <SWPadnos> that's a bit easier, but the clients are waaaay more picky
[22:48:42] <toastydeath> the ONLY thing that makes me even consider making any money
[22:48:45] <SWPadnos> and you need to know color management inside out for the high end stuff
[22:48:57] <toastydeath> is that model folks and other people have started propisitioning me
[22:48:59] <toastydeath> for money
[22:49:17] <toastydeath> on a very limited portfolio
[22:49:20] <SWPadnos> well, if you can get work, then I'd try it out
[22:49:39] <toastydeath> yeah, i'm going to feel it out and see what happens
[22:49:41] <SWPadnos> if you already have Nikon lenses, then the worst thing would be to buy a higher end body from them
[22:50:04] <toastydeath> yeah, exactly
[22:50:18] <toastydeath> and i'm not doing too shabby with my current camera, i'm more inclined to spend money on lighting
[22:50:42] <toastydeath> and lenses
[22:50:55] <toastydeath> i guess i will see what unfolds
[22:51:02] <SWPadnos> yah, lighting is very important - most people forget that because modern cameras are pretty good at getting good photos in crappy lighting conditions
[22:51:28] <toastydeath> hahah, i have noticed that
[22:51:32] <SWPadnos> hmm. D2xs - 12.4MPix
[22:51:40] <toastydeath> i usually spend 3x the time setting up lighting
[22:51:46] <toastydeath> versus actually taking pictures
[22:51:52] <SWPadnos> that's a pretty good ratio ;)
[22:51:59] <SWPadnos> unless you're doing school photos
[22:52:07] <toastydeath> hell no
[22:52:15] <Rugludallur> CNC lighting ?
[22:52:47] <SWPadnos> thought of that a long time ago. you can get something like it from martin (the LightComposer)
[22:52:51] <SWPadnos> and other companies
[22:52:58] <SWPadnos> that's Martin :)
[22:53:00] <toastydeath> it's more "i have to stop fiddling with lights or the model will wander off"
[22:53:05] <toastydeath> than "i am done messing with lights now"
[22:53:10] <toastydeath> that imposes the 3x limit
[22:53:38] <toastydeath> lightcomposer?
[22:53:46] <toastydeath> are you talking CN or photography
[22:54:02] <toastydeath> *CNC
[22:54:04] <SWPadnos> the lightcomposer is a computer-controlled lighting system
[22:54:12] <toastydeath> oh
[22:54:27] <SWPadnos> there are several robotic or semi-robotic systems for moving lights around
[22:54:34] <toastydeath> i prefer sneakernet
[22:55:02] <SWPadnos> my old company also makes radios that one day (when the manufacturers get off their duffs) will be able to remotely control power from individual packs, fed by a lightmeter with integrated radio
[22:55:25] <toastydeath> haha
[22:55:32] <toastydeath> ALSO: brushes my teeth?
[22:55:53] <SWPadnos> hmmm. nope, but it will also press the button on your camera ;)
[22:55:58] <toastydeath> hahahah
[22:56:08] <SWPadnos> http://www.pocketwizard.com
[22:56:28] <toastydeath> pocketwizards are so expensive
[22:56:36] <SWPadnos> well, they work. you get what you pay for
[22:56:38] <toastydeath> yeah
[22:56:51] <SWPadnos> plus I still own stock, so you increase my income ;)
[22:56:58] <toastydeath> hahahah
[22:57:03] <toastydeath> ty for the full disclosure
[22:57:08] <SWPadnos> any time
[22:57:21] <SWPadnos> I wrote the software in the original ones
[22:57:24] <toastydeath> i'd buy them if i had the spare cash
[22:57:29] <SWPadnos> co-founded the company, actually
[22:57:32] <toastydeath> wow
[22:59:55] <SWPadnos> hmmm. maybe I should get a D80 or thereabouts, I'm not sure I can still get film processed from my F100
[23:00:53] <toastydeath> it has pretty good reviews
[23:01:05] <SWPadnos> plus I already have some lenses
[23:01:19] <SWPadnos> but $1k is a bit too much for a hobby
[23:01:38] <toastydeath> ah
[23:01:59] <toastydeath> perhaps!
[23:02:20] <toastydeath> or perhaps it is just right!
[23:02:20] <SWPadnos> geez - Costo has a D80 kit with the 18-135 lens, plus 2x 2GB SD cards for $1249.99
[23:02:32] <toastydeath> hahahahah
[23:03:17] <toastydeath> brb
[23:03:17] <SWPadnos> interesting - SD is roughly $9/gig, regardless of size (1G through 8G)
[23:03:28] <SWPadnos> well, 1G are $11, but close enough
[23:05:17] <tomp> Rugludallu: vcp.widgets has default values in it already, and xml overloads them as you like.
[23:05:28] <tomp> eg vcp.widgets default " {"lower", "0.0", ATTRIB_FLOAT, offsetof(spin_button_data_t, lower) },
[23:05:30] <tomp> eg xml initialization <max_>5000</max_>"
[23:05:52] <tomp> Rugludallur:
[23:06:30] <SWPadnos> tomp, do you think it would be reasonably easy to add ini variable lookup to that?
[23:06:50] <Rugludallur> tomp: hmm I'm looking at it now, it has defaults harcoded but no way to set an initial value for a component
[23:07:06] <Rugludallur> tomp: but it's easy enough to add an init_ xml tag
[23:07:22] <tomp> SWPadnos: dunno, haven't tried any python file parsing so far
[23:07:42] <SWPadnos> there's already some ini parsing available, axis uses it
[23:08:28] <tomp> Rugludallur: why a special init_, just any attribute can be intitialized when the xml is read asw in the example
[23:08:52] <SWPadnos> is the "value" one of the things that can be initialized?
[23:09:10] <tomp> maybe the attributes for each widget are not clear
[23:09:11] <SWPadnos> oops - gotta run to the store. bbiab
[23:10:22] <Rugludallur> tomp: ok, if I have a spin button, and when I start pyvcp I want that spin button to have the value of 10, but min 1 and max 20 , what would the xml syntax look like ?
[23:10:36] <tomp> SWPadnos: ditto, the attributes are not clear, something i'm doing now with components... but componenet written as 'comps' are self documenting, and widgets are not, the attribute are in the code
[23:10:49] <tomp> Rugludallur: will work it up and pastebin it
[23:10:56] <Rugludallur> tomp: brilliant :D
[23:11:10] <Rugludallur> tomp: I think this isn't documented anywhere
[23:11:23] <Rugludallur> tomp: rephrase I don't think this is documented anywhere :P
[23:13:28] <tomp> so far upper, lower, value, dec_points are attributes seen in vcp_widgets.c ... lemme try some xml ( i got real free with changing the source here so will have to get virgin code outta the isolation directories ;)
[23:13:35] <a-l-p-h-a> SWPadnos, wtf is Don Imus?
[23:13:46] <Rugludallur> tomp: hehe ok :D
[23:26:01] <tomp> Rugludallur: looks like the widgets are borked, they allow an initialization, but it doesnt appear till it gets inc'd/dec'd, and then they wont change at all. will look, but this is not at the scripting level...
[23:28:01] <Rugludallur> tomp: what is the ideal solution, should I continue adding _init or ?
[23:32:38] <tomp> Rugludallur: i cant say, its not my code,
[23:32:47] <tomp> but if you add xml data types/tokens and keep them isolated,
[23:32:54] <tomp> the xml wont mind. the xml parser wont display them either,
[23:33:05] <tomp> will just ignore them. meanwhile you have your info ordered and
[23:33:13] <tomp> ready to go when it's reviewed by the developers.
[23:33:14] <tomp> I'd not put a lot of work into it untill theres some feedback.
[23:34:11] <tomp> meanwhile you might look at the code to see why this happens because in any event it's a bad reaction to good xml
[23:35:54] <Rugludallur> tomp: I'll add it for now but I can always throw it away, it's 20 minutes of work to add _init to all components
[23:37:22] <tomp> Rugludallur: i think it can be fixed ( and should be, the init value is not handled, that may be fixable in the instantiation code,but i'm still looking at the locked up spinner )
[23:38:49] <Rugludallur> tomp: the pyvcp_dial actually has init parameter
[23:39:14] <Rugludallur> tomp: in trunk , but that is something you added so ...
[23:40:38] <Rugludallur> tomp: hmm or is it just in the doc/help
[23:46:18] <tomp> Rugludallur: no,your correct, that is in trunk, in pyvcp_widgets.py , in __init__ (the instantiator) , 'self.out = init' good for comparison with the spinbtn widget....
[23:48:22] <tomp> Rugludallur: in the spinbtn, the value is forced to another term 'self.value = min_'
[23:48:28] <Rugludallur> tomp: yup
[23:49:05] <Rugludallur> tomp: it makes sense to restrict the value to be >min and <max but forcing it does not make sense
[23:51:31] <tomp> Rugludallur: in the __inti__ of the spin btn , a new term 'initval' needs to be added, defaulted to min_ (eg __init__ (.... initval=min_ ... ) ) and then it can be overidden or not as you like
[23:51:58] <Rugludallur> tomp: exactly
[23:52:25] <tomp> i can write it, but you'll need to get a developer to submit it
[23:53:08] <skunkworks> and 1/4 of the developers are on vacations ;)
[23:53:16] <skunkworks> give or take a few
[23:53:58] <tomp> i'd hate to have their jobs ( and their other jobs ) :)
[23:54:33] <Rugludallur> tomp: I can check with alex or cradek for permission to commit the code
[23:54:49] <Rugludallur> tomp: tomorrow or after I have tested
[23:55:13] <tomp> ok, will write & pb it for you
[23:55:23] <Rugludallur> tomp: brilliant, thanks :D