#emc | Logs for 2008-06-23

[04:12:59] <AzRockCrawler> Chris cradek are you there?
[04:20:50] <cradek> uh-oh
[04:21:02] <AzRockCrawler> no, no uh oh
[04:21:03] <AzRockCrawler> :)
[04:21:16] <AzRockCrawler> I was reading your website
[04:21:38] <AzRockCrawler> saw your entry about "Floating Point Irritant"
[04:22:03] <AzRockCrawler> There are many...mine was always this one...
[04:22:17] <AzRockCrawler> float d = 0.0;
[04:22:22] <AzRockCrawler> d += 0.01;
[04:22:27] <cradek> heh, I got your email
[04:22:28] <AzRockCrawler> d -= 0.01;
[04:22:59] <AzRockCrawler> then print results
[04:23:23] <AzRockCrawler> it will generally be a negative number in most C/C++ compilers.
[04:23:39] <AzRockCrawler> Did I send and email?
[04:23:47] <AzRockCrawler> * AzRockCrawler scratches head
[04:24:05] <cradek> well if not you, someone else did the other day
[04:24:12] <AzRockCrawler> I see.
[04:24:16] <cradek> it's hard to tell when people don't use their actual names
[04:24:27] <AzRockCrawler> Good ole binary.
[04:24:34] <cradek> so I was guessing from the context only
[04:24:38] <AzRockCrawler> I see.
[04:24:52] <AzRockCrawler> Barry here.
[04:25:19] <cradek> the emailer suggested that d+=0.01f; ... would fix it. s/he was right.
[04:25:29] <AzRockCrawler> Yes.
[04:25:50] <AzRockCrawler> But it is not intuitive IMO.
[04:26:08] <cradek> no, probably not.
[04:26:35] <AzRockCrawler> I was going tosay
[04:26:47] <AzRockCrawler> that 'double' generally resolves those issues too, rather than using floats.
[04:27:14] <AzRockCrawler> But it has its own issues.
[04:27:54] <AzRockCrawler> anyway...the joys of binary.
[08:21:45] <micges> morning
[09:14:00] <micges> Is it possible to have different acc applied to G0 and G1 ?
[11:22:17] <[1]a-l-p-h-a> [1]a-l-p-h-a is now known as a-l-p-h-a
[11:25:45] <Sweeper_> Sweeper_ is now known as Sweeper
[12:28:16] <skunkworks> alex_joni: Pingh
[12:28:21] <skunkworks> pingh?
[14:33:57] <xyzee> Can some one tell me how to do a tool change. I had assumed I could pause the programm, change the tool, home the z axis and resume, but not so. Help please!
[14:35:46] <renesis_> yeah emc people
[14:35:50] <renesis_> why cant i do that?
[14:36:15] <renesis_> so i have to use seperate files for each tool
[14:36:24] <renesis_> unless they have collars
[14:36:56] <renesis_> not all of us have big expensive tool changers with pre indexed tools
[14:38:49] <archivist> some g code then, move to sensible places, gcode to change tool, click dialog box to continue, I have no tool changer
[14:40:05] <renesis_> no if you have to reset the Z
[14:40:14] <renesis_> to an unknown
[14:40:31] <renesis_> like during the toolchange, im pretty sure you cant do that in a cycle
[14:40:51] <renesis_> youd have to stop it and select a start line and restart cycle
[14:41:10] <renesis_> and i dont like the select start line feedback in axis, like im never sure if it actually did what i wanted
[14:42:01] <xyzee> I have basic set up, my file has gcode for tool change, I get the tool change dialog, but I cannot zero the tool in z axis
[14:42:33] <BigJohnT> you need to zero the tools before running the gcode
[14:42:44] <renesis_> should have a 'next line to execute' area in status bar, maybe like [ 345 / 12343 ] for next line/total lines
[14:42:48] <BigJohnT> and put that into your tool table
[14:43:01] <renesis_> no we cant
[14:43:05] <renesis_> spring collets, hai
[14:43:29] <BigJohnT> then use a z probe move
[14:43:48] <renesis_> and that will reset z in cycle?
[14:43:52] <renesis_> i need a probe?
[14:44:01] <renesis_> wtf is a prob my machine is open loop
[14:44:02] <BigJohnT> yes and no
[14:44:50] <renesis_> if you could reset Z during a tool change, it would make issue go away
[14:45:02] <xyzee> Is it possible to stop the machine and then start it again from a certain line in the code?
[14:45:10] <renesis_> i cant index tools
[14:45:11] <renesis_> yes
[14:45:19] <renesis_> sec ill check machine
[14:45:38] <BigJohnT> just break the code up for each tool...
[14:45:38] <renesis_> oh wow its off
[14:45:53] <renesis_> yeah thanks
[14:46:23] <xyzee> sec ill check machine does what?
[14:46:25] <renesis_> but its machine > set next line, i think
[14:46:42] <renesis_> i was gonna see what the menu option was, but my machine is off
[14:46:57] <renesis_> but you have to highlight the line in the bottom pane
[14:47:08] <renesis_> in the gcode, and then do machine -> set next line
[14:47:17] <renesis_> and then start cycle and itll start from there
[14:47:41] <renesis_> i kinda wish it had more visual feedback, line a program counter
[14:50:18] <xyzee> Thanks. I think I,ve got it
[14:51:54] <cradek> I agree with BigJohnT that the safest thing to do is split the program up, one run per tool
[14:52:28] <cradek> a much better scheme, if you can pull it off, is to use a tool height sensor
[14:55:09] <renesis_> thats what i do
[14:55:39] <renesis_> the split files
[14:56:57] <renesis_> more time, more annoying, more room for error loading multiple files instead of just one and reading which tool to load
[14:57:14] <BigJohnT> renesis_: that is the safe way if your machine is not set up for tool changes
[14:57:31] <renesis_> well thats kinda the only way right now
[14:57:56] <cradek> I agree it's annoying and error-prone
[14:58:23] <cradek> have you looked into the tormach tooling? it's affordable and a good upgrade for collet CNC machines
[14:58:34] <renesis_> like with the tiny drills, i can do one file and read the tool and load it, goes very fast
[14:58:49] <renesis_> they have collars, tho, and its just drills
[14:59:32] <renesis_> even that + touch off would be alot faster than finding files and kinda verifying its right and touching off
[14:59:45] <renesis_> and tormach stuff is neat
[15:00:37] <renesis_> but yeah kinda pricey compared to $0 and some text editing to make collets work
[15:01:29] <renesis_> i have to call my failob and see if they fixed the lan
[15:01:41] <cradek> yeah I like $0 too. depends what the time/annoyance is worth
[15:01:53] <renesis_> yeh
[15:02:02] <renesis_> tormachs lathe thing looks neat
[15:02:16] <cradek> hm the tool setter probe I saw at workshop is $800.
[15:02:37] <renesis_> yeh thats like half the price of my mill almost
[15:03:42] <cradek> I think I could make something suitable for my needs in a day or two's work. it wouldn't need to be super precise.
[15:03:44] <archivist> annoyance is bugging me, I need to automate measuring a mounted cutter eccentricity/centerline and outside dia
[15:03:48] <BigJohnT> you can make one with a microswitch and a couple of parts
[15:03:56] <cradek> BigJohnT: my thoughts too
[15:04:15] <BigJohnT> kinda like the reverse of my floating head on my plasma cutter
[15:04:20] <cradek> BigJohnT: I'm thinking a dowel pin with the end ground flat, in a simple reamed hole
[15:04:27] <cradek> ends
[15:04:41] <cradek> then a simple microswitch underneath somehow
[15:04:43] <renesis_> neat, job is having issues
[15:04:44] <BigJohnT> just turn down the center for the roller of the micro switch
[15:04:46] <renesis_> wanna know why?
[15:04:55] <archivist> Im contemplating a laser line and opto and following the profile
[15:04:56] <cradek> shouldn't be too hard to get .001 repeatability
[15:04:58] <renesis_> cuz they think its the ISP, even tho their LAN drives are down
[15:05:00] <renesis_> failjob
[15:05:10] <renesis_> microswitches are that good?
[15:05:17] <BigJohnT> yep
[15:05:29] <cradek> I've been very surprised, but yes they seem to be
[15:05:43] <renesis_> hmm, neat
[15:05:48] <renesis_> renesis_ is now known as renesis
[15:05:59] <BigJohnT> cradek: let me see if I can find a picture of my touchoff switch on my plasma
[15:06:03] <cradek> don't use a springy lever of course - press on the switch plunger
[15:06:17] <BigJohnT> or the roller one
[15:07:08] <renesis> i wonder ow accurate the cherry switches in the arcade buttons i have are
[15:07:35] <BigJohnT> the vertical rod on the left is my Z home and limit with a roller microswitch mounted to the flat plate. http://s47.photobucket.com/albums/f163/johnplctech/?action=view&current=zassb.jpg
[15:07:42] <renesis> they have the plane buttons
[15:08:06] <BigJohnT> much better picture http://s47.photobucket.com/albums/f163/johnplctech/Plasma%20Cutter/?action=view&current=Plasma012.jpg
[15:09:24] <cradek> that's quite the thing, wow.
[15:09:37] <renesis> yeah, pretty
[15:10:24] <BigJohnT> The lower switch is my touch off switch
[15:10:25] <renesis> so wait i put a switch on my table and write some code to make it touch off on the switch?
[15:10:38] <cradek> yes
[15:10:39] <BigJohnT> yep
[15:10:44] <renesis> hmm, neat
[15:11:03] <cradek> http://cvs.linuxcnc.org/cvs/emc2/nc_files/tool-length-probe.ngc?rev=1.4
[15:12:10] <renesis> ty
[15:12:14] <BigJohnT> cool, using G30 to go to the switch position
[15:12:56] <renesis> where does (print, happen?
[15:13:15] <cradek> the console or a popup window, I forget which
[15:13:16] <renesis> pop up dialog or terminal or the lower pane or what
[15:13:25] <cradek> (debug, does the other thing)
[15:13:31] <renesis> hmm, okay ill try next time
[15:13:57] <renesis> i wonder where i put the arcade buttons
[15:20:04] <renesis> hmm
[15:20:22] <renesis> Max Movement Differential
[15:20:29] <renesis> .012
[15:20:50] <archivist> hysteresis
[15:21:10] <cradek> doesn't matter
[15:21:17] <archivist> in anod out swtching points, choose one only
[15:21:20] <renesis> the other one is Operating Point
[15:21:35] <renesis> 0.543/.598
[15:21:47] <renesis> like .055 diff
[15:22:41] <renesis> i think arcade button cherry switches might be fail for NC, heh
[15:23:03] <renesis> ill stick in a vice and try it like 20 times i guess
[15:23:23] <archivist> smaller ones have a better spec for cnc
[15:25:01] <renesis> vise
[15:25:28] <renesis> i should just buy a handful, i prob wanna do limit/home switches too
[19:34:45] <chr0n1c> yo machine heads!
[19:37:52] <anonimasu> hello
[19:43:30] <BigJohnT> * BigJohnT 's head is hurting from all the incoming info
[19:46:43] <chr0n1c> * chr0n1c started a PIC programming tutorial this morning
[19:47:34] <skunkworks> * skunkworks is trying to wrap his head around 6 axis.
[19:50:08] <anonimasu> lol
[19:50:17] <anonimasu> got videos from fest online yet?
[19:51:16] <chr0n1c> is curious about the videos also
[19:51:50] <anonimasu> I want to see the 6 axis stuff :)
[19:52:42] <skunkworks> No - not that I see. That would be cradek, jepler or jmkasunich
[19:53:12] <skunkworks> We got a good deal on a puma 560 robot arm - (just the arm and servos - no control)
[19:53:49] <skunkworks> Emc has that exact config. So... I was playing around with the sample.
[19:54:35] <chr0n1c> * chr0n1c needs a robot arm
[19:54:43] <anonimasu> :)
[19:55:35] <skunkworks> Another project... ;)
[21:47:39] <micges> gcode properties doesn't work on trunk
[21:57:48] <micges> when I touch off file 3d_chips.ngc I have error: emc/task/emctask.cc 312: interp_error: Bad character 'c' used
[21:57:49] <micges> Bad character 'c' used
[22:00:04] <jepler> I can reproduce the first problem but not the second.
[22:00:44] <micges> jepler: moment
[22:00:45] <jepler> (the first is a consequence of improvements made on trunk and I know in general what the problem is)
[22:01:15] <micges> jepler: ok
[22:08:40] <micges> jepler: hold on I am try to rereproduce second one
[22:18:32] <micges> jepler: this is the same kind of problem that running from line bug from about 4 months ago
[22:18:52] <micges> something is uninitialized and return bogus values
[22:19:07] <micges> can't find what it is
[22:19:32] <micges> in mean time I saw error by opengl: (invalid operation)
[22:23:26] <micges> ok I got it
[22:23:41] <micges> run 9 axis config
[22:24:11] <micges> and select joint 8 and then touch off --> above error
[22:24:51] <jepler> oh -- so the actual bug is that you can "touch off" in joints mode
[22:25:28] <micges> yes
[22:25:45] <micges> I selected 8 joint and started touch off
[22:26:38] <micges> I cant touch off xyz only
[22:27:16] <micges> ok I search third bug
[22:35:16] <dmess> hi all
[22:35:35] <micges> thanks for FAST response time. I'm impressed
[22:35:42] <chr0n1c> yo dmess
[22:35:44] <micges> hi dmess
[22:36:01] <dmess> any thing new??
[22:36:08] <jepler> micges: I am not sure if my latest checkin actually fixed the problem you saw in touch off -- I never saw the error myself
[22:36:48] <micges> Ill check it
[22:40:39] <dmess> http://www.stemmewest.com/pages/S10_E.pdf im still jonesn' for this baby....
[22:41:26] <jepler> bbl
[22:44:34] <ednspace> Can anyone outline how to set up a probe to zero the z-axis when the bit touches a plate closing the switch?
[23:11:47] <dmess> look at the touch probe for the lathe set-up... i do believe i seen that somewhere
[23:30:35] <ednspace> I did see a couple of short tutorials, just wondering how I can hook this into axis
[23:30:51] <ednspace> right now I have an input pin hooked to z axis home
[23:31:17] <ednspace> and when I hit the home button, in axis it sends the axis down
[23:31:41] <ednspace> I am hopeing when the switch throws, it will stop and set Z to 0
[23:31:52] <ednspace> really I need a slight offset to adjust for the plate
[23:33:26] <BigJohnT> you use a G38 probe move and the switch needs to be hooked up to the probe input
[23:34:26] <BigJohnT> same as #2 on this page http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Plasma
[23:35:04] <ednspace> so if I put my work on the machine table, then just go to mdi mode in axis and type in G38
[23:35:45] <BigJohnT> depends on how your switch is wired and if your going to or away there are 4 choices
[23:36:07] <BigJohnT> I use G38.2 http://linuxcnc.org/docs/2.2/html/gcode_main.html#sub:G38.2:-Straight-Probe
[23:37:03] <BigJohnT> you also have to tell it what axis and what direction and how far to go before giving up
[23:37:22] <BigJohnT> for example G38.2 Z-2
[23:38:00] <BigJohnT> and you have to have a feed rate set at the time you use the G38
[23:39:38] <ednspace> Ok I think I got that
[23:39:46] <ednspace> still working on wireing up the switch
[23:40:06] <ednspace> whats the difference between home and touch off?
[23:40:13] <BigJohnT> make sure you "connect" it to the motion probe input
[23:40:28] <ednspace> I think I can do that with stepconf
[23:40:35] <ednspace> let me see
[23:41:48] <BigJohnT> yep u can
[23:42:09] <ednspace> ok just connected it to probe in
[23:42:21] <ednspace> this is actually nicer then the way I was picturing
[23:42:33] <BigJohnT> cool
[23:42:46] <ednspace> because now I think I can get it to rezero
[23:42:49] <ednspace> after a tool change
[23:43:04] <ednspace> well possibly
[23:43:08] <BigJohnT> yes, you can
[23:43:20] <BigJohnT> I do it with my plasma torch before each cut
[23:43:29] <ednspace> how do you account for offset from the plate
[23:43:38] <ednspace> or surface of the switch
[23:43:41] <BigJohnT> I have a floating head
[23:44:03] <ednspace> yes my bits are at ground
[23:44:09] <ednspace> so is the table I think
[23:44:15] <ednspace> so I need a third element
[23:45:32] <BigJohnT> look in your nc_files folder for a file called tool lenght probe.ngc
[23:45:45] <BigJohnT> it outlines what you need to do
[23:45:53] <BigJohnT> after you get things hooked up
[23:49:44] <dmess> you the MAN Big J
[23:50:02] <BigJohnT> why is that dmess?
[23:52:23] <ednspace> Ok thanks for the help, I will check it out
[23:52:42] <BigJohnT> np
[23:53:48] <ednspace> One other question
[23:54:04] <BigJohnT> your out of questions for the day LOL
[23:54:14] <ednspace> is there a way to get say 0 - 400 hz square wave for spindle control
[23:54:21] <ednspace> oops
[23:54:29] <ednspace> instead of pwm
[23:54:41] <BigJohnT> that is beyond my guesspertise
[23:54:49] <ednspace> yes me too!
[23:55:05] <ednspace> ah well, I can stick with the touch probe for now
[23:55:16] <chr0n1c> square wave is pwm?
[23:55:26] <ednspace> I have enough torn apart to keep me busy for a couple of days
[23:55:34] <BigJohnT> keep asking there are plenty that can answer you just have to catch them on line
[23:55:43] <ednspace> well, I need to vary frequency instead of pulse width
[23:55:53] <chr0n1c> ohh, ok..
[23:56:08] <ednspace> I have an old controller card for the spindle
[23:56:29] <ednspace> it seems to work fine if I hook it to a square wave generator and vary frequency
[23:56:48] <ednspace> but the pwm gives me a limited range of speed
[23:57:01] <ednspace> it does change, just not much control
[23:57:28] <BigJohnT> * BigJohnT needs to go upstairs and get control of the remote as law and order is almost over and time for some MASH
[23:57:43] <chr0n1c> i dunno, above my knowledge of emc2 there...
[23:57:59] <BigJohnT> talk to you guys later