#emc | Logs for 2009-02-16

[01:01:35] <eric_unterhausen> I just discovered my bathroom wall is only 2x4 on edge thick
[01:18:54] <skunkworks> secret passage?
[01:19:03] <skunkworks> full of gold?
[01:19:30] <eric_unterhausen> they were building the house on the cheap and ran out of room for a real wall
[01:32:50] <shrdlu-> gerber doesn't have provision for 3d does it? Then why the hell does gcam a) have a 3d view and b) have gerber as its only file import function?
[01:39:14] <shrdlu-> oh, I see. to get previews of drill holes and stuff I guess
[02:02:38] <Guest834> anyone home
[02:07:56] <Guest834> Guest834 is now known as yawstick
[02:09:22] <shrdlu-> godamnit. ANOTHER cam program that does in fact not import what it claims to
[02:09:39] <shrdlu-> this one will import only adobe illustrator and pdf files
[02:09:40] <shrdlu-> wtf
[02:10:02] <shrdlu-> there is a sickness in this community
[02:23:31] <eric_unterhausen> OT, anyone got the sickness?
[02:29:23] <eric_unterhausen> also OT: I want a sandblasting cabinet that is somewhere in-between the $250 you get from HF, and the $3500 pro versions
[03:06:01] <bglackin> Also OT: I want the smithy someone adverted on craiglist for $100 last week
[03:25:05] <eric_unterhausen> bglackin: isn't that always the way it goes?
[03:30:29] <bglackin> Aye - Guy pulled the item off auction when he got slammed with email - put it back for $600 with tooling - was still a good deal - no cash toplay that day though
[03:31:25] <bglackin> Nite
[03:31:30] <eric_unterhausen> some people don't smarten up fast enough
[03:33:02] <bglackin> Lol - no kidding - from his photos - it was in great shape - normal hobbyist wear - not too clean - had me drooling for a day thinking about it
[03:34:05] <bglackin> I am happy though - I picked up an old hannond bench mounted belt sander/polisher in good conditions for cheap - perfect for finishing off the stuff I need to make around the house
[03:34:27] <bglackin> Don't work too hard tonight
[03:39:53] <fenn> eric_unterhausen: what are the features of a "pro" sandblasting cabinet?
[03:53:16] <eric_unterhausen> I guess they have thicker material, I'm just looking for larger ones for now
[03:53:49] <eric_unterhausen> need to fit a bike frame in there
[04:08:37] <eric_unterhausen> HF has a larger one, might work
[04:35:02] <beachsurfin> beachsurfin is now known as ibuffy
[05:07:23] <fenn> meh i'd just screw some plywood and 2x4's together
[05:07:47] <fenn> or do it outside in a kiddie pool
[05:36:49] <jmkasunich> got the robot 95% together - just need to customise some cables
[05:37:19] <jmkasunich> motors and encoders are working, and I have a hal config (plus a couple custom components) that will drive it from waypoint to waypoint
[05:37:46] <jmkasunich> tomorrow I gotta dive back into the vision code
[05:38:45] <toastydeath> can anyone explain why 1/(x+1)^3 integrates into 1/3(x+1)^2
[05:39:13] <toastydeath> i don't get why the (x+1)^3 doesn't make it more complicated than just x^3
[05:40:48] <jmkasunich> its been 25 years since I did that kind of calculus
[05:41:41] <toastydeath> lol
[05:41:48] <toastydeath> i'm thinking this is integration by parts?
[05:42:07] <toastydeath> because (3x+1)^3 does get more complicated, but not by much
[05:42:53] <tomp> jmkasunich: what kinda robot id you get?
[05:43:05] <cradek> because you are applying the chain rule, and derivate of x+1 is no more complex than derivate of x (both are 1)
[05:43:45] <cradek> oh, integrate
[05:43:49] <cradek> * cradek didn't read the question
[05:44:28] <toastydeath> i guess it's just u substitution
[05:44:31] <tomp> waypoint to waypoint, cool, teach mode
[05:48:07] <jmkasunich> tomp: not a robot arm
[05:48:13] <jmkasunich> its an engineering week program
[05:48:21] <jmkasunich> robot vehicle to navigate a maze
[05:56:04] <tomp> ah more inflection on 'waypoint' cool
[05:59:19] <tomp> what do you call the graphic legend on a rotary switch/pot? the lines and numbers?
[06:06:49] <tomp> bezel? what an ancient term, ancient sounding
[06:13:57] <eric_unterhausen> fenn: that's a good idea for a start
[06:15:48] <tomp> seems like 'legend plate' is more standard' term
[06:16:19] <fenn> bezel
[06:17:19] <fenn> or "dial plate"
[06:17:28] <fenn> or whatever you want
[06:19:47] <tomp> AB, cutlerhammer, klochner mueller, automationdirect,all went to 'legend plate', so i was able to find the stuff, thx
[07:36:16] <toastydeath> aha!
[07:36:37] <toastydeath> if the companies i have been finding for diamond turning are any indication, there should be plenty of job oppertunities if i get laid off
[09:18:27] <archivist> logger_emc: bookmark
[09:18:27] <archivist> Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2009-02-16.txt
[09:18:56] <beachsurfin> beachsurfin is now known as ibuffy
[09:29:27] <unas> hello all. please could anyone what's the difference between motions "X10" and "X+10"? cannot find anywhere in docs
[09:29:48] <unas> *could anyone _explain_
[09:30:46] <SWPadnos> there should be no difference, AFAIK
[09:31:08] <unas> i get different distance though. :-\i also thought so
[09:31:21] <SWPadnos> using EMC2?
[09:31:25] <unas> yep
[09:31:39] <unas> on 8.04 ubuntu
[09:31:50] <SWPadnos> did you use G90 or G91?
[09:33:07] <unas> relative
[09:33:14] <SWPadnos> so G91
[09:33:18] <unas> yes
[09:33:49] <SWPadnos> how much of a difference did you see, and how did you detect (measure) it?
[09:36:25] <unas> with slide gage. 130 mm with "X+10" and 120 with "X10"
[09:36:49] <SWPadnos> ?
[09:37:02] <SWPadnos> are you saying that the machine moved 120mm when you asked for 10?
[09:38:04] <unas> nono. X+10 is for example. that would be "X+120"
[09:38:13] <SWPadnos> phew :)
[09:38:18] <unas> ))
[09:39:37] <SWPadnos> unfortunately, I don't really know what else to ask, and since it's 04:40 AM here, I think I need to go to bed now :)
[09:40:19] <unas> ))
[09:40:31] <SWPadnos> alex_joni should be pretty close to your schedule, so feel free to hang around and ask him (or anyone) about it again.
[09:40:49] <unas> thank you !
[09:40:56] <SWPadnos> you're welcome
[09:44:39] <tomp> i just got an email from the emc email list saying i am being removed 'due to excessive bounces'
[09:44:55] <tomp> it'd be nice if there was some suggestion why that is
[09:44:55] <archivist> ew
[09:45:36] <tomp> this began when i reported not getting any ballots
[09:45:57] <tomp> so it gets better !
[09:46:38] <tomp> since i get aqll the other mail from cacamdroem and em dev and emc, i wonder why someone thinks i dont get stuff
[09:48:55] <unas> what is the smtp server?
[10:05:46] <tomp> its the same one that elivers all the other emc email, smtp.att.yahoo,com
[10:06:21] <tomp> add a 'd' whre appropriate (busted key on eeepc )
[10:29:11] <piasdom> g'mornin all
[10:29:42] <micges> good morning
[11:33:04] <unas> is there any kind of books or something about how to program g-codes besides that quick g-code reference?
[11:34:05] <BigJohnT> you might look at the g code generators on the wiki site
[11:37:08] <unas> learning by a example code you mean?
[11:38:34] <BigJohnT> sometimes that helps
[11:39:05] <BigJohnT> once you get past G0 and G1 many controls don't agree on how to implment the rest
[11:39:12] <unas> i saw them, yes. that helps but there're always appear not obvious things
[11:39:35] <unas> yep, i understand
[11:40:23] <unas> i'm looking for a decision for my question some hours ago
[11:40:29] <BigJohnT> what are you trying to do?
[11:40:48] <unas> what's the difference between motions G1 (e.g.) "X10" and "X+10"? cannot find anywhere in docs.
[11:41:18] <archivist> as SWPadnos said, should be none
[11:41:38] <unas> i write x+10 and move for 10 mm e.g. and then write x10 and move for 14 mm
[11:41:44] <unas> yes, yes
[11:41:59] <archivist> backlash ?
[11:42:06] <BigJohnT> the + should not change anything
[11:42:09] <unas> this is why i'm looking for a book or specifications
[11:42:26] <unas> should not, but there's a difference
[11:42:54] <BigJohnT> that is a machine or configuration issue
[11:43:14] <unas> hm. that really can be
[11:43:22] <unas> yes\
[11:43:33] <BigJohnT> are you in absolute mode?
[11:43:42] <unas> nope, relative
[11:44:46] <BigJohnT> version 2.2.8?
[11:45:55] <unas> 2.2.5
[11:47:49] <BigJohnT> hmmm works perfect for me here no difference
[11:48:21] <BigJohnT> the second move of 14mm is that measured or displayed in AXIS?
[11:48:30] <unas> measured
[11:49:04] <micges> unas: test this in MDI: G0 X10
[11:49:08] <micges> then G0 X10
[11:49:13] <micges> then G0 X-10
[11:49:26] <unas> yes, i will
[11:49:28] <micges> between then measure length
[11:49:35] <micges> them*
[11:49:36] <unas> i thought the same toght
[11:49:48] <BigJohnT> you might be loosing steps
[11:50:42] <BigJohnT> a test program is here in the docs http://www.linuxcnc.org/docview/html//common_Stepper_Diagnostics.html#r1_3
[11:51:01] <unas> loosing steps? but why?
[11:52:18] <BigJohnT> binding, accel too fast, voltage low on drive, could be many things
[11:52:54] <unas> agree
[11:53:19] <unas> but that strange anyway
[12:25:22] <archivist> losing steps is common until you understands the limitations of steppers
[12:31:22] <unas> i understand it. but in this case the difference is not so obviously understandable
[13:01:33] <unas> there is one more problem i have: at the beginning of the cycle i always lose 3mm. all next X's are okay. what can it be?
[13:01:55] <unas> Y's are always okay.
[13:02:18] <archivist> listen to it, does it make a poor sound
[13:04:25] <unas> no!
[13:05:02] <unas> in one of examples in emc2 i found an interesting phrase: start G0 lines with a .001 second pause to avoid motion blending problems.
[13:05:13] <unas> maybe that's the explanation?
[13:05:40] <archivist> unlikely
[13:07:28] <BigJohnT> that is only to prevent cutting corners unas if your accel is set real low
[13:08:33] <archivist> unas that would affect x and y x only points to a more obvious stepper and driver problem or mechanical
[13:17:51] <unas> thanks
[13:19:40] <unas> very familiar problem:
[13:19:41] <unas> http://osdir.com/ml/linux.distributions.emc.user/2007-04/msg00092.html
[13:20:58] <archivist> motion blending has had errors fixed since then
[13:26:21] <unas> yes. here i have 8.04 and at the work PC is weaker with emc2 2.1.4 works with this bug.
[13:51:50] <skunkworks_> -> I only tested this with a stepper setup and rotatekins kinematiks, but I think its a general problem
[13:53:31] <skunkworks_> Kins in emc at the moment have a problem where the calculation take place after the trajectory planner has calculated the path.. So the axis could be told to do things they cannot do. (individual axis could accel/move faster than the limits in the ini
[13:54:18] <skunkworks_> That is one of the things in the works axis/joint limits.
[13:57:54] <alex_joni> skunkworks_: yup, that's true
[14:11:39] <unas> "so we have to live with it" :) as they say
[14:16:49] <unas> in G4 (dwell) P is in seconds or milliseconds? or P in sec and E in msec?
[14:18:44] <BJT-Work> seconds
[14:19:21] <BJT-Work> http://www.linuxcnc.org/docview/html//gcode_main.html#sec:G4:-Dwell
[14:19:43] <BJT-Work> what E?
[14:20:19] <unas> in wikipedia this parameter. e.g. http://www.cncezpro.com/g04m.cfm
[14:20:30] <unas> here P500 is 0.5 sec
[14:21:03] <unas> http://en.wikipedia.org/wiki/G-code - here that E
[14:22:12] <BJT-Work> Oh, I thought you were asking about EMC
[14:23:55] <unas> well, actually about emc2. i thought gcodes are same for all...
[14:24:14] <BJT-Work> nope everyone is slightly different
[14:25:25] <unas> so in emc2 it will be in seconds.
[14:25:39] <BJT-Work> yes
[16:06:27] <JymmmEMC> JymmmEMC is now known as Jymmmmmm
[16:07:06] <Jymmmmmm> Jymmmmmm is now known as JymmmEMC
[16:55:35] <shrdlu-> hrm. I now get
[16:55:36] <shrdlu-> /usr/bin/emc: line 516: 6083 Segmentation fault $EMC2_BIN_DIR/$EMCSERVER -ini $INIFILE
[16:55:49] <shrdlu-> from 2.28 and 2.3 pre
[16:56:02] <shrdlu-> plus a bunch of other segfaults
[16:56:21] <shrdlu-> I love stuff like this
[16:56:28] <shrdlu-> go to bed, everything is working fine
[16:56:33] <shrdlu-> wake up, something broke
[17:00:34] <cradek> if yesterday it worked, but today everything seems broken, you probably have a hardware failure and should run memtest
[17:01:37] <shrdlu-> I think I may have done some weird stuff inbetween, I was getting some weird errors from gnome this morning too
[17:01:50] <shrdlu-> I think I'll just back up my config and reinstall everything
[17:01:58] <cradek> yes my advice only applies if you didn't do anything to mess it up
[17:02:06] <cradek> what did you do?
[17:02:49] <shrdlu-> I don't know, something to do with libc++ or something. Gnome wouldn't fire up properly
[17:03:00] <shrdlu-> I fixed it and it all came back
[17:03:10] <shrdlu-> but I've done several things like that over the last few days
[17:03:29] <cradek> ouch
[17:03:50] <shrdlu-> I think installing kde then brute uninstalling it was a very bad idea ;)
[17:04:22] <cradek> what is brute uninstalling?
[17:05:03] <shrdlu-> not being selective about what I removed. Like search for kde in synaptic and click select all, remove
[17:05:25] <shrdlu-> a lot of stuff went strange after that
[17:05:42] <shrdlu-> had to reinstall the ubuntu desktop etc
[17:06:01] <archivist> messing about like that will inflict pain
[17:13:41] <shrdlu-> right
[17:13:44] <shrdlu-> onwards
[18:47:38] <SWPadnos> oh, interesting.
[18:47:54] <SWPadnos> micges, is this a different problem from the other one you reported before?
[18:48:02] <SWPadnos> (with run from line and subroutines)
[18:48:41] <micges> yes
[18:49:03] <micges> last one every RFL executed from the same line and hang interpreter
[18:49:10] <SWPadnos> oh, ok
[18:49:20] <micges> after refine of gcode file this one shows up
[18:49:47] <SWPadnos> just to be sure, did you change the number of O1000 CALL lines to verify that it gets executed that number of times?
[18:50:30] <micges> no
[18:50:48] <micges> I've selected between 2nd and 3 call
[18:50:49] <SWPadnos> ok
[18:50:55] <micges> and it was 2 execution
[18:51:04] <micges> and so on
[18:51:13] <SWPadnos> ok
[18:51:44] <micges> luckly it didn't destroy my new machine :) (that bug was discovered in run time)
[18:52:23] <SWPadnos> heh
[19:53:58] <jensor> What would the g-code look like to pause the program manually change a tool, then touch-off for the new tool length, then continue? Or is this possible?
[19:54:19] <cradek> the gcode would not be special except it would have a safe entry move at the point where you want to resume
[19:54:38] <cradek> then you just abort the program, do the tool manipulation, and restart at the safe spot.
[19:55:17] <jensor> How do you start the program on a specific line?
[19:55:27] <jensor> I'm running axis
[19:55:28] <cradek> consider it just like a program that has two beginnings
[19:55:45] <SWPadnos> um, wouldn't TxxM6 be easier?
[19:55:57] <SWPadnos> (with hal_manualtoolchange)
[19:55:58] <cradek> highlight the line, select "set next line" on the menu before running (2.2), or right click and "run from here" (2.3)
[19:56:05] <cradek> SWPadnos: read the question again
[19:56:17] <SWPadnos> yes, I see the question
[19:56:34] <SWPadnos> TxxM6 plus tool-length-probe would be another solution, no?
[19:56:42] <cradek> yes that's a neat solution that I like
[19:57:31] <SWPadnos> you can do tool changes in a subroutine, so a tool change becomes O<toolchange> CALL [toolnumber]
[19:57:44] <SWPadnos> which does the TxxM6 / length probe
[19:57:52] <SWPadnos> (or however the actual call should be spelled)
[19:57:55] <cradek> yeah I'm sure you could put it in a subroutine
[19:58:14] <cradek> I didn't because my gcode was already autogenerated
[19:58:18] <SWPadnos> sure
[19:58:46] <jensor> I can highlite a line, but I don't see any options for selection in axis
[19:58:59] <cradek> what version is this?
[19:59:04] <jensor> 2.2
[19:59:13] <SWPadnos> 2.2.what?
[19:59:32] <jensor> 2.2.8
[19:59:37] <cradek> Machine/set next line
[19:59:40] <cradek> then, run
[19:59:41] <jensor> k
[20:06:18] <jensor> cradek, I highlite a line, do machine/set next line, then It still starts at the beginning
[20:08:15] <jensor> of the program
[20:08:35] <jensor> continue does nothing
[20:36:25] <topls64> does emc / hal have a provision for variable data input, like displaying voltage meaured from an a to d converter? I realize more harware would be invlolved.
[20:37:37] <skunkworks_> yes - you would have to create the hardware and hal componant to read it.
[20:37:55] <topls64> For sinker edm btw. I'd like to have z azis back off of it starts to short out. Or perhaps hardware that I can set a trip point on and single a limit switch hit?
[20:39:06] <topls64> I have't got that deep into hal yet, but intend to. Does hal support serial input by chance?
[20:39:18] <skunkworks_> emc cannot 'revearse' a gcode program... if that is what you are thinking..
[20:39:53] <skunkworks_> you could possible do something in hal..
[20:39:56] <topls64> I'd just like to use emc instead of making a whole AVR chip based thing
[20:40:40] <topls64> I'd be fine working with hal. Have to make a frontend too, right.
[20:41:00] <skunkworks_> at this point it is above my head ;)
[20:43:03] <alex_joni> topls64: we have some EDM support in emc2/HAL
[20:43:15] <alex_joni> does sinker use a spark voltage like wire edm?
[20:43:35] <alex_joni> we have an input to the motion controller called adaptive feed rate
[20:43:55] <alex_joni> basicly you connect a component you wrote which outputs 0..1 for the speed emc2 should go
[20:44:04] <alex_joni> if the voltage drops, you simply reduce the speed
[20:44:14] <topls64> still designing it and learning about the stuff. adaptive feedrate sounds like what i need
[20:45:57] <alex_joni> take it one step at a time, and you'll get there ;)
[20:46:03] <topls64> I'd like to monitor / adj voltage & current on the fly if possable, so I'll need some a/d d/a conversion stuff.
[20:46:15] <alex_joni> with proper hardware that's no problem
[20:46:23] <topls64> nice
[20:46:57] <topls64> can you point me where. The manual i suppose...
[20:47:43] <jensor> SWPadnos, I implemented hal_manualtoolchange. It stops the program and asks for the tool change, but I am still unable to touch off z to accomodate the new tool length
[20:51:01] <jensor> I'm just trying to figure out how to touch off when I put a different drill in the spindle to drill a second set of holes
[20:51:54] <jensor> Do I have to have a fixture to set up the drill in the collet - would that be the only way?
[20:52:51] <topls64> Not _sure_ but as I understand all tools have to be the same length (unless cpecifed in the gcode?)
[20:53:24] <topls64> hence setup fixture
[20:53:29] <jensor> Thats a bummer
[20:53:40] <topls64> not _sure_ though
[21:01:52] <cradek> jensor: drills are a pain.
[21:02:23] <cradek> I just crank the knee to the right place when changing drills
[21:02:28] <cradek> without that, ick
[21:02:55] <topls64> So, for a sinker edm with adaptive feedrate, I replace the usual encoder with my hardware, and prolly a custom hal component?
[21:03:45] <cradek> certainly all tools do NOT have to be the same length. that's what lengths in the tool table are for
[21:03:47] <alex_joni> topls64: depends on stepper or servo
[21:03:52] <topls64> stepper
[21:04:00] <alex_joni> but usually you don't take out encoders if you have them already
[21:04:10] <topls64> don't have them
[21:05:13] <topls64> thinking more just up/down linear motion, stops/adjusts when volt/amps thresholds are reached
[21:05:27] <topls64> mmm fuzzy logic
[21:06:51] <topls64> Will hal accept an 8 bit digital value as input? Like from an analog-digital converter chip?
[21:07:30] <alex_joni> depends how you want to feed it into the PC
[21:07:45] <topls64> can I use serial?
[21:07:58] <alex_joni> not very good for realtime
[21:08:05] <cradek> I think the motenc has adc
[21:08:17] <cradek> maybe a pico board does? I don't know for sure.
[21:08:25] <alex_joni> yeah, but pretty expensive :)
[21:08:35] <SWPadnos> you need the PPMC to get analog input
[21:08:42] <SWPadnos> (the card cage one)
[21:09:09] <SWPadnos> topls64, if you'd like to buy me an analog input card, I can write a driver for you :)
[21:09:25] <topls64> I can feed my AVR chip based hardware into LPT2
[21:09:46] <SWPadnos> (seriously though, there are some boards from measurement computing that could work, with a relatively simple HAL driver)
[21:10:06] <topls64> my buddy may have one
[21:10:23] <SWPadnos> as for 8 bits -> one value, yes. there's a weighted_sum component
[21:10:42] <topls64> I'd rather make the AVR hardware, as I can then add a nice graphic lcd and user interface
[21:10:42] <seb_kuzminsky> all the adc boards i've used have been slow, 40 KHz conversion rate or so
[21:10:42] <SWPadnos> defaults to powers of two for the inout bit weights
[21:10:58] <seb_kuzminsky> you either poll a "ready" bit in a register somewhere, or you wait for an end-of-conversion interrrupt
[21:11:08] <seb_kuzminsky> both seem somewhat hostile to the realtime context
[21:11:09] <cradek> a parport has 13 inputs
[21:11:13] <SWPadnos> the AVRs are in the 160 kHz range I think
[21:11:17] <alex_joni> seb_kuzminsky: adaptive-feed-rate is servo-thread dependent
[21:11:18] <SWPadnos> cradek, LPT2
[21:11:24] <alex_joni> so 40kHz is more than plenty
[21:12:56] <cradek> SWPadnos: yes I'm saying you could have more than 8 bits of adc resolution
[21:12:56] <SWPadnos> oh, ok
[21:12:56] <SWPadnos> true
[21:12:56] <seb_kuzminsky> alex_joni: not if you have to wait 25 us in the read function, called from the servo thread
[21:12:56] <cradek> wouldn't you have it sample constantly and just read the latest value?
[21:12:56] <SWPadnos> seb_kuzminsky, you make the ADC free-run, and read the latest
[21:12:56] <SWPadnos> I would :)
[21:12:57] <alex_joni> right
[21:12:57] <cradek> yeah, or what he says
[21:12:57] <alex_joni> what they say
[21:12:57] <alex_joni> (or try to)
[21:12:57] <SWPadnos> what we said
[21:12:59] <seb_kuzminsky> well said guys
[21:13:00] <topls64> yes, that Is how I've done it. freerunning
[21:14:00] <SWPadnos> if you'd like to spend $1000 or so, you could get a mesa card and one of my analog I/O boards
[21:14:09] <SWPadnos> but you probably don't :)
[21:14:13] <alex_joni> who'd want that :P
[21:14:20] <SWPadnos> well, there is one place ...
[21:14:29] <topls64> Its getting the value from the avr to the pc that bugs me - 8 bit or larger bus?
[21:14:36] <alex_joni> SWPadnos: stick it full of board if you still can :P
[21:14:40] <SWPadnos> heh
[21:14:47] <SWPadnos> topls64, you could use SPI instead
[21:14:51] <cradek> how many bits of adc in the avr?
[21:14:53] <alex_joni> SWPadnos: kidding, I'm sure there are applications where it's useful
[21:15:01] <cradek> 10? 11?
[21:15:14] <SWPadnos> they claim 10, but it's not very good
[21:15:19] <SWPadnos> (like all microcontrollers)
[21:15:21] <topls64> Yeah, spi. nice. Bits of adc depends on chip.
[21:15:33] <JymmmEMC> 10.5!
[21:15:38] <SWPadnos> even less
[21:15:42] <topls64> I do have some NICE adc chips, hires
[21:16:10] <JymmmEMC> SOLD! To the guy with a lathe on a trialer in the snow in his driveway!
[21:16:10] <SWPadnos> I was thinking SPI to the AVR, but if you don't need to use the AVR for other things, then a standalone ADC would also work
[21:16:39] <SWPadnos> just note that the function to read the ADC will take about 3 microseconds per bit on the PC
[21:16:45] <topls64> I want the avr for graphic lcd and basic user io
[21:16:55] <SWPadnos> (at least - I'm thinking CLK->1, read, CLK->0)
[21:17:30] <topls64> have the avr read the adc, pc reads avr. should be quick?
[21:17:51] <topls64> 8 bit bus faster, right.
[21:17:58] <cradek> 10 bit bus
[21:18:13] <cradek> I can't see any way of doing it on the cheap except 10 bit parallel
[21:18:23] <topls64> 8 bit bus, 16 bits data (2 reads)?
[21:18:39] <cradek> but I'm clueless about how you'd get edm current feedback into an avr in the first place
[21:18:50] <cradek> sounds like a very hard problem to me
[21:19:00] <SWPadnos> if you have other user I/O connected to the AVR, you will probably want to use some addressing on the parport and read specific registers
[21:19:07] <topls64> adc reads it off the line w/ appropriate sensor
[21:19:38] <topls64> SWP - i should be able to do that
[21:20:10] <SWPadnos> it sounds like much more of a PITA than it's worth though
[21:20:33] <jensor> SWPadnos, Thanks for the tip on hal_manualtoolchange. I wasn't aware of it.
[21:20:40] <SWPadnos> if you need to use the parallel port, the Mesa 7i43 gives you 48 I/O
[21:20:46] <topls64> I'm an electronics guy more than a machinist, so this is right up my alley
[21:20:47] <SWPadnos> jensor, you're welcomr
[21:20:51] <SWPadnos> e
[21:21:16] <topls64> I was looking at the mesas for work, maybe I'll have them get me one;)
[21:21:27] <SWPadnos> topls64, there's an SPI function that can be put in the mesa FPGA (though the driver doesn't yet support it)
[21:21:52] <SWPadnos> they make a few SPI connected I/O boards, with digital or analog I/O
[21:22:06] <topls64> at that point I wouldn't need the avr
[21:22:16] <SWPadnos> some are isolated, you'd have to take a look at the specs (7i64, 7i65, 7i66?)
[21:22:20] <SWPadnos> exactly
[21:22:26] <topls64> the mesa would have the io lines
[21:22:29] <SWPadnos> the 7i43 is cheap, $90 or so
[21:22:53] <topls64> could make a card with some old ass 8255 chips. Ha ha
[21:22:56] <SWPadnos> if you know anything about FPGAs, you can make your own "firmware"
[21:23:00] <SWPadnos> heh
[21:23:21] <topls64> my bro would know the fpga stuff. used to make laser trimming machines
[21:23:26] <SWPadnos> even if you don't (I did it ;) )
[21:23:46] <SWPadnos> that's not Les Watts is it? :)
[21:23:52] <topls64> kids in school (college) do it all the time
[21:24:00] <topls64> Bob golenia
[21:24:04] <SWPadnos> ah
[21:24:11] <topls64> chicago laser systems
[21:24:16] <topls64> long gone
[21:24:31] <SWPadnos> Les is a guy I haven't seen here in a long time, but he was working on some laser trimming designs a couple of years ago
[21:25:23] <SWPadnos> in any case, unless you value your time at $0/hour (as hobbyists often do :) ), designing a microcontroller board and writing the software to talk to it doesn't seem like a big win
[21:26:19] <topls64> cls went under like 15 years ago. Well, screw it. Lets buy a mesa.
[21:27:01] <SWPadnos> hmmm. I wonder if anyone has made a "poor man's A/D" in VHDL/Verilog
[21:27:17] <SWPadnos> one of the "time how long a cap takes to charge" circuits
[21:28:56] <topls64> mesa's webaddress is...?
[21:29:03] <seb_kuzminsky> mesanet.com
[21:29:08] <topls64> thzx
[21:29:40] <topls64> ooo pc/104 stuff ;)
[21:29:54] <seb_kuzminsky> pc/104-plus : even better :-)
[21:32:26] <topls64> lets do it wifi
[21:32:46] <topls64> run my mill from the beach
[21:33:36] <topls64> mill us up some martinni glasses
[21:36:32] <topls64> I'm eyeing the 7I66
[21:38:06] <topls64> I see it in the price list but not on the site
[21:43:56] <topls64> Time out. This sinker edm - how often would you say I have to poll the voltage/current to decide wether of not to advance/retract?
[21:44:18] <cradek> every servo cycle (millisecond or faster)
[21:44:29] <topls64> Im thinking avr and lpt port might be fine
[21:44:34] <topls64> ok
[21:44:44] <topls64> even for stepper
[21:57:19] <billykid2> hello
[22:00:01] <seb_kuzminsky> hi
[22:01:08] <topls64> is there any arduino components for hal?
[22:03:42] <seb_kuzminsky> drive-by greeting
[22:04:46] <topls64> hola
[22:09:58] <alex_joni> topls64: yes, but I don't think it's realtime
[22:10:19] <topls64> thx
[22:11:53] <alex_joni> topls64: http://axis.unpythonic.net/01198525592
[22:12:01] <alex_joni> http://axis.unpythonic.net/01198594294
[22:29:36] <topls64> Just confirming: adaptive feerate -> I can have this controlled by a pot or adc value with the correct hardware and hal comps
[22:30:10] <mshaver> Suppose a fellow had a mesa 5i20/hm2_pci all set up, happy, working OK, with TRUNK (2.3 BETA). Suppose that guy had to downgrade to 2.2.8. Suppose that guy did this, but noticed that, much to his surprise and dismay that there were no longer any stepgens or gpio pins. Suppose that the .BIT file is a custom made one, and that dmesg reveals "Inconsistent Module Descriptor for Stepgen", and "Version 1, expected 0". Can anyone say what
[22:32:44] <SWPadnos> "what"
[22:33:13] <mshaver> What do you mean, "what"?
[22:33:26] <SWPadnos> your question was cut off: "Version 1, expected 0". Can anyone say what
[22:33:46] <mshaver> Can anyone say what this all means?
[22:33:56] <mshaver> that was it....
[22:34:07] <alex_joni> probably that he shouldn't have downgraded?
[22:34:36] <mshaver> astute, but less than helpful
[22:35:13] <SWPadnos> if "he" is still using the bitfile from TRUNK, then this means that the stepgen is changed somewhat in the TRUNK bitfile (I think it is - there were some timing issues with the older one), and the driver version in 2.2.8 doesn't know how to deal with them
[22:35:36] <shrdlu-> I can see now way of cutting very small apertures from 100 micron mylar with vector cuts and still getting a square shape
[22:35:43] <shrdlu-> erm, no
[22:35:57] <mshaver> You see, we have this fancy new GUI, which is greatly desired to be installed on these machines by the powers that be, and this GUI won't work with TRUNK for some as yet unexplained reason.
[22:36:19] <SWPadnos> you can compile 2.2.8 yourself
[22:36:29] <SWPadnos> and manually backport the TRUNK hm2 driver
[22:36:40] <mshaver> SWPadnos: Actually, the bitfile was made special for me by Peter.
[22:37:01] <mshaver> hmmmmmm.......
[22:37:05] <SWPadnos> but if it was made with the later stepgens, then it's different
[22:37:35] <mshaver> my bitfile = ultra-super-good-stepgens
[22:37:42] <SWPadnos> then they're TRUNK stepgens :)
[22:37:55] <mshaver> yes
[22:38:08] <SWPadnos> you can probably just copy the TRUNK source in hal/drivers/mesa-hostmot2 over that from 2.2.8
[22:38:13] <SWPadnos> or use CVS magic to do it for you
[22:38:55] <SWPadnos> (check out -rv2_2_branch or -VERSION2_2_8 and then cvs -dPA (?) in the hostmot2 directory to clear the sticky tag)
[22:39:00] <SWPadnos> or maybe it's -dPC
[22:39:34] <SWPadnos> hmmm. you'll need to add more letters to what I typed ...
[22:39:56] <mshaver> I have 2.2.8 installed from apt-get. I wonder if I compile 2.2.8 separately, whether I can replace some binaries directly?
[22:40:12] <SWPadnos> could be
[22:40:37] <SWPadnos> you're probably better off figuring out why the GUI doesn't work on TRUNK, since TRUNK will be 2.3 in a couple months
[22:41:03] <mshaver> That's my argument, but I've been overrulled
[22:41:43] <seb_kuzminsky> mshaver: copy stepgen.c (and maybe ioport.c) from trunk to your 2.2.8 sandbox, recompile 2.2.8
[22:42:10] <SWPadnos> well, as long as people are OK with making their own packages, which are a hybrid between the TRUNK and v2.2 code, then everything's fine
[22:43:04] <SWPadnos> incidentally, if you compile your own hm2 driver, you're also compiliing the rest of emc2, so you might as well use a consistent set of binaries - no sense mixing and matching unless you need to
[22:43:30] <mshaver> OK, I'll try to make a 2.2.8-alien_human_hybrid, with a complete recompile!
[22:44:18] <mshaver> Is the "how to make a .deb" process written down anywhere?
[22:44:31] <seb_kuzminsky> just please be very clear with everyone that it's *not* 2.2.8, but some local hybrid monstrosity
[22:44:46] <seb_kuzminsky> mshaver: on the wiki somewhere i'm pretty sure
[22:45:02] <mshaver> I wonder how you change the displayed version text?
[22:45:06] <seb_kuzminsky> what os/distro are you on?
[22:45:24] <mshaver> u 8.04
[22:46:27] <seb_kuzminsky> hm a quick glance at the wiki doesnt deliver the goods
[22:46:48] <mshaver> * mshaver is rolling his chair over to the other computer to start this experiment
[22:46:52] <seb_kuzminsky> debian/configure -r && fakeroot debian/rules binary
[22:47:20] <mshaver> I've built a .deb once, just have to remember... Oh, Thanks!
[22:47:27] <seb_kuzminsky> but i agree with swp, port your gui to 2.3 :-P
[22:48:02] <mshaver> I wish. I can't really understand why it's so hard. What's so different in the GUI interface?
[22:49:02] <SWPadnos> mshaver, it looks like configure contains version information
[22:49:29] <SWPadnos> you can change the two strings PACKAGE_VERSION and PACKAGE_STRING to whatever you want
[22:49:40] <SWPadnos> please also change the email address for bug reports ;)
[22:49:44] <seb_kuzminsky> the build-time dependencies won't be managed for you if you build the .deb the way i said, but if you've successfully compiled emc2 trunk on this machine it shouldnt be a problem
[22:50:21] <SWPadnos> mshaver, what GUI interface?
[22:50:33] <SWPadnos> I don't know that there is one
[22:51:08] <SWPadnos> there's NML, and whatever widget set you're using, and whatever window manager, and .....
[22:52:27] <SWPadnos> oh. technically I think you're supposed to change configure.in and re-run automake or whatever to change the version number
[22:53:20] <mshaver> SWPadnos: eztrol
[22:53:43] <SWPadnos> no, what I meant was that there's no "GUI interface" to EMC2, per se
[22:54:36] <mshaver> What I meant was the interface between a GUI and the rest of emc. It can't have changed much.
[22:54:48] <SWPadnos> right, I was typing that in a much less succinct way :)
[22:55:04] <mshaver> Anyway, here goes....
[22:55:06] <SWPadnos> but there isn't one "GUI interface"
[22:55:34] <mshaver> true, but I almost wish there was
[22:55:40] <SWPadnos> there's NML for a lot of stuff, hal for some stuff, some miscellaneous helper functions (like inifind)
[22:56:01] <SWPadnos> yeah, it would probably be nice
[22:56:24] <seb_kuzminsky> mshaver: can you tell us how your gui fails on 2.3? maybe we can fix it
[22:57:05] <mshaver> it's written in qt, does that matter?
[22:57:16] <seb_kuzminsky> qt is a good thing to write guis in ;-)
[22:58:04] <mshaver> do you have a 5i20 system with 2.3?
[22:58:12] <seb_kuzminsky> yes
[22:58:16] <BigJohnT> I do
[22:58:35] <seb_kuzminsky> if by "system" you mean a pile of connected hardware on a workbench ;-)
[22:58:51] <mshaver> yes! that'll work
[22:59:00] <mshaver> hold on, let me dig something up
[22:59:13] <seb_kuzminsky> the 5i20 with 2.3 runs axis and turns servos and steppers
[22:59:19] <BigJohnT> if by "system" you mean a 5i20 card in the computer and wondering where my 50 pin flat ribbon cable is...
[22:59:30] <seb_kuzminsky> didnt you get cradek's fax?
[22:59:50] <BigJohnT> still waiting on it... he might have sent it to my machine shop
[23:00:15] <BigJohnT> I'm in the beer cave now
[23:00:32] <seb_kuzminsky> heh, no beer in the shop?
[23:00:48] <BigJohnT> oh yes, in the fridge there is plenty
[23:01:00] <BigJohnT> in case someone is thirsty
[23:02:01] <BigJohnT> my machine shop is close to town... I live on a dead end dirt road off of a dirt road...
[23:04:02] <seb_kuzminsky> back in a bit, real life intrudes...
[23:04:15] <mshaver> Ok
[23:04:32] <alex_joni> mshaver: you'll have to change VERSION (top emc2 dir)
[23:04:41] <alex_joni> change src/configure.in, rerun autoconf
[23:04:49] <alex_joni> (that will produce a new configure)
[23:04:57] <mshaver> Thanks!
[23:05:01] <alex_joni> then you build a package the regular way, and it'll get the new version
[23:05:34] <mshaver> I wonder if it's automatically put in the lower left corner of the splash screen?
[23:05:47] <alex_joni> I think so
[23:05:48] <mshaver> or if someone manually edits the image
[23:05:52] <mshaver> wow
[23:05:56] <alex_joni> no, the image hasn't been changed iirc
[23:06:18] <alex_joni> the image should be top dir, along with VERSION
[23:07:49] <mshaver> OK, I'll try the recompile now. When seb gets back, maybe he'll want to try out the gui (along with anyone else who wants to help)
[23:08:21] <alex_joni> any errors?
[23:08:31] <alex_joni> or things it doesn't do ?
[23:08:49] <mshaver> the gui?
[23:09:51] <shrdlu-> welp. thats another day wasted on this
[23:09:55] <shrdlu-> 7 in a row
[23:10:51] <shrdlu-> I wish I knew how to crack dongles
[23:11:49] <BigJohnT> I use a pecan press to crack dongles...
[23:12:48] <alex_joni> mshaver: yeah
[23:14:31] <shrdlu-> If my job was to cut paper stencils, I'd be flying
[23:14:46] <shrdlu-> the mylar is showing me no love in vector
[23:15:10] <BigJohnT> if it was easy everyone would be doing it :)
[23:15:39] <shrdlu-> I was doing it, very successfully
[23:16:23] <shrdlu-> then the usb protection for this sw died, and here I am
[23:16:29] <geo01005> Anybody know the maximum encoder frequency for the mesa 7i43 board and hm2?
[23:16:41] <SWPadnos> I think it's 1 or 4 MHz
[23:16:44] <shrdlu-> maybe I should just pay the $300 for the POS dongle
[23:17:22] <SWPadnos> shrdlu-, if you have a legal copy of the software, and the software does what you want, then getting support from the company that sold it to you seems appropriate
[23:17:50] <SWPadnos> if they charge $300 to replace the dongle, then that's annoying, but if it fixes your problem and gets you up and running again, it's probably worth it, no?
[23:17:51] <shrdlu-> they're in china
[23:18:10] <shrdlu-> they don't even have an english translation of their web site
[23:18:18] <alex_joni> shrdlu-: gotta start practising on your kanji
[23:18:27] <shrdlu-> quite
[23:18:29] <SWPadnos> if you want to do rasterized stuff like in that video you linked, then you'll need some custom software
[23:19:03] <shrdlu-> I wish I had known this several months ago
[23:19:11] <geo01005> SWPadnos-: Thanks.
[23:19:13] <SWPadnos> it's almost doable with existing components
[23:19:26] <SWPadnos> but it really depends on the resolution and speed you need
[23:19:59] <shrdlu-> what depends?
[23:20:28] <SWPadnos> deciding on the easiest way to do what you want depends on what you want
[23:21:52] <shrdlu-> well, there's no possibility of me writing a whole stepper driving system
[23:21:59] <SWPadnos> you don't need that
[23:22:21] <SWPadnos> what you're looking for is a printer, except instead of ink blobs, you want it to pulse a laser
[23:22:33] <shrdlu-> quite
[23:23:14] <SWPadnos> emc can deal with the axis motion, but what you need is a system that can load some representation of the line that's being traced and play it back as the motor moves back and forth
[23:23:27] <SWPadnos> there are ways of doing this - almost - already
[23:24:07] <SWPadnos> halstreamer is one - lets you feed a bunch of numbers from a file and have them "played out" in realtime, one after the other
[23:25:27] <shrdlu-> I'm weeks away from being able to implement something like that
[23:25:28] <SWPadnos> you'd need custom code to synchronize that with motion and generate the numbers in the first place
[23:25:35] <SWPadnos> sure, it's not trivial
[23:25:47] <mshaver> alex_join: yes, as I recall from a few days ago, I couldn't jog or turn the spindle on, I didn't try much else
[23:25:52] <SWPadnos> though it is interesting - do you normally start with a bitmap for that kind of work?
[23:25:59] <alex_joni> mshaver: I bet I have a hunch
[23:26:13] <SWPadnos> can't move without homing ...
[23:26:14] <alex_joni> hmm.. although jog should work
[23:26:17] <shrdlu-> I'm just despondent because I had hoped this vector cut might work with the PWM, after having failed with the gcode raster
[23:26:22] <alex_joni> SWPadnos: yeah, exactly
[23:26:23] <shrdlu-> but neither is going to work
[23:26:33] <SWPadnos> bummer
[23:26:41] <alex_joni> mshaver: in emc2.3 you need to home before you can use MDI or AUTO
[23:26:57] <alex_joni> there is an option NO_FORCE_HOMING which you can add to the ini
[23:27:01] <SWPadnos> unless you add [TRAJ]DON"TMAKEMEHOMEBEFOREMOVINF=1 :)
[23:27:08] <alex_joni> I bet that would make things better
[23:27:10] <SWPadnos> oh right, NO_FORCE_HOMING :)
[23:27:26] <SWPadnos> damn. out of coffee
[23:27:28] <alex_joni> (mini has the same issues, although it only spews some errors, but works in the end)
[23:28:00] <shrdlu-> I assume some most who cut plastic stencils use vector. But they probably have a decent laser with a tiny point, decent air assist etc
[23:28:01] <mshaver> alex_joni: this was more broken than that, I mean I couldn't jog at all
[23:28:17] <shrdlu-> my vector method was slow, but the results were 1st class
[23:28:34] <shrdlu-> I've taken several customers away from pro operations
[23:28:38] <shrdlu-> but now that's all fucked
[23:28:40] <shrdlu-> frankly
[23:29:09] <mshaver> * mshaver 's cvs check out of 2.2.8 is done!
[23:29:52] <shrdlu-> sorry, my raster method
[23:32:39] <PCW> geo01005: HostMot2 max encoder count rate for 7I43 is ~20 MHz (filter off)
[23:32:41] <PCW> and ~ 6 MHz (filter on) These values assume that the filter rate is the default 50 MHz
[23:32:43] <PCW> (it can be set for slower filter time constant but the driver does no give access to
[23:32:44] <PCW> the (global) encoder filter rate register yet
[23:36:45] <alex_joni> * alex_joni says good night all
[23:36:57] <shrdlu-> good night alex
[23:37:03] <SWPadnos> see ya alex_joni
[23:38:09] <mshaver> night alex!
[23:40:00] <geo01005> A couple of days ago I asked about analog input on the 7i43 card. I was planning on using an SPI ADC, still not sure on that one.
[23:40:41] <geo01005> I did some research on the logs, and it looks like Anders Wallin is using a voltage to frequency converter to measure some analog singals.
[23:40:48] <geo01005> Anybody else doing that?
[23:42:14] <geo01005> (AD654 reading the frequency on an up down encoder input)
[23:42:37] <SWPadnos> no idea
[23:42:38] <PCW> Its a good cheap way to get isolation...
[23:43:45] <geo01005> It dosn't seam quite as nice as an SPI ADC.
[23:44:16] <PCW> Also ability to trade resolution for conversion rate
[23:45:14] <fenn> fenn is now known as fenn_driving
[23:46:24] <geo01005> Well I'm still willing to try and contribute to the Hostmot2 driver. I had a look at the code, it may take me a while but I may be able to add support for SPI.
[23:46:37] <geo01005> The voltage to frequency sure seams easy though.
[23:49:18] <PCW> If you have multiple channels, SPI is nicer.
[23:49:20] <PCW> Were doing a experimental V-F A-D
[23:49:22] <PCW> for a THC A-D, the limited number of isolators
[23:49:23] <PCW> is nice, we should have < 5 PF of capacitance
[23:49:25] <PCW> from the input side to the digital/power side.
[23:51:57] <geo01005> PCW-: Is there somewhere that I can find more documentation than the man page on Hostmot2?
[23:52:41] <seb_kuzminsky> geo01005: the best is probably the hm2 register map
[23:53:20] <seb_kuzminsky> http://cvs.linuxcnc.org/cvs/emc2/src/hal/drivers/mesa-hostmot2/firmware/src/regmap?rev=1.4
[23:53:31] <seb_kuzminsky> this describes the interface that the hostmot2 firmware presents to the driver
[23:55:54] <mshaver> seb_kuzminsky: you up for trying this gui?
[23:56:12] <seb_kuzminsky> sure :-)
[23:56:19] <seb_kuzminsky> this is the one that doesnt work? :-)
[23:56:43] <mshaver> ok, first you need to enable backports and install qt4.4.0
[23:56:55] <mshaver> yes, doesn't work
[23:57:07] <seb_kuzminsky> actually first i need to go pick up the demon-spawn from preschool
[23:57:11] <seb_kuzminsky> can you email me the goods?
[23:57:28] <mshaver> sure! I forgot it's earlier there!
[23:57:34] <SWPadnos> I guess that makes you the demon ...
[23:57:38] <mshaver> e-mail on the way
[23:57:43] <SWPadnos> or maybe your wife. hmmm
[23:57:47] <seb_kuzminsky> SWPadnos: it's the wife
[23:57:58] <SWPadnos> ok. (bash.org, here we come) :)
[23:58:02] <seb_kuzminsky> ok talk to y'all later :-)
[23:58:33] <PCW> geo01005: if you want to try to add some SPI support to HostMot2, I can send you a bitfile
[23:58:35] <PCW> with the SPI hardware built in.
[23:58:36] <PCW> Hi Sebastian!
[23:58:38] <PCW> I need to update regmap as theres one new bit in the SPI channel descriptor
[23:58:39] <PCW> (the "dont-clear-frame-at-EOT" bit)