#linuxcnc | Logs for 2013-11-18

[00:18:04] <somenewguy> is there a conclusive way to check if a particular M# or G# is already used?
[00:18:25] <somenewguy> as in if I want to do somethign custom, is there a quick way of making sure it is not already a functino defined or in dev?
[00:18:45] <somenewguy> alternatly, is there a number range reserved for custom use?
[00:23:38] <Jymmm> http://linuxcnc.org/docs/html/gcode/m-code.html
[00:23:59] <Jymmm> http://linuxcnc.org/docs/html/gcode.html
[00:24:08] <Jymmm> http://linuxcnc.org/docs/html/gcode/o-code.html
[00:24:52] <Jymmm> http://gnipsel.com/linuxcnc/g-code/index.html
[00:25:27] <uw> MOAR LINKS
[00:26:13] <Jymmm> google em, I did
[00:26:49] <Jymmm> https://www.google.com/search?q=linuxcnc+free-sex
[00:27:06] <Jymmm> https://www.google.com/search?q=linuxcnc+g+codes
[00:27:39] <Jymmm> somenewguy: http://www.linuxcnc.org/docs/devel/html/gcode/m-code.html#sec:M100-to-M199
[00:29:23] <somenewguy> ah thank you for the last one
[00:29:57] <somenewguy> I have been googling them, however as linuxcnc grows etc you never know what new functions may have sprung into existance and have yet to be properly documented
[00:30:20] <somenewguy> I have learend the hard way in many an opensource project that documentation is not something ppl get terrribly excited about ,and don't always keep up to date
[00:32:22] <somenewguy> but the M100-199 thing will save me from passing random M numbers into the MDI to make sure nothing bites back before I use it lol
[01:47:34] <Jymmm> archivist: archivist_herron Your electro etching suggestion was bullshit... THERE AINT ANYTHING LEFT! LOL
[01:48:56] <Jymmm> archivist: Well, except gooey pieces of paint that I used as a test resist.
[02:05:04] <kengu> http://wiki.linuxcnc.org/cgi-bin/wiki.pl?Rastering_With_A_Laser is there something that is more recent than 4 years or is this still the best(ish) solution
[02:06:07] <archivist> Jymmm, you did it wrong do not blame me
[02:06:43] <Jymmm> archivist: I'm still gonna blame ya! I just never expected it to work so damn well
[02:07:08] <Jymmm> archivist: and only 5V too
[02:07:31] <archivist> one checks the process....you obviously .....
[02:08:00] <Jymmm> I waited till the amperage dropped to check
[02:09:24] <Jymmm> kengu: best solution for what?
[02:10:00] <Jymmm> archivist: Thank you for suggesting the electro etching, I appreciate it!!!
[02:10:53] <Jymmm> Now I just gotta figure out some kind of current limiter
[02:11:39] <Jymmm> Hmmm, maybe as simple as less salt might do the trick.
[02:12:37] <_DJ_> moin
[02:15:11] <archivist> Jymmm, quality id related to amps per unit area, so do some thinking(googling) on that
[02:18:17] <Jymmm> archivist: Thanks. Tonight was just a general test; couple of piece of scrap 1"x3" SS 0.010" spray painted, then scribed. I had to go thru 3 PS before I found one that didn't shut down (think it saw a short)
[02:27:56] <RyanS> $500 for a somewhat used drill press, (new $800) http://www.appliedmachinery.com.au/uploads/stock/images/Machtech-BD35-24419a.jpg although the base has been painted. Which leads me to believe it was owned by an idiot
[05:15:58] <kengu> y axis of plasma moving (: need to wire up some more axis
[05:27:19] <MacGalempsy> pics?
[05:27:57] <archivist> or it didnt happen :)
[05:28:17] <kengu> you have seen the pics
[05:30:03] <kengu> https://www.dropbox.com/s/zpm78idalzfjibq/20130616_002.jpg
[05:30:24] <archivist> note this is one of the older IRC jokes :) http://www.collection.archivist.info/archive/DJCPD/PD/2009/2009_08_13_adjustable_spanner/P1010269.JPG
[05:31:20] <kengu> archivist: yes. i know.
[05:31:47] <archivist> actually I dont think I had seen that one before
[05:32:23] <kengu> you have seen these,https://www.dropbox.com/s/tiynelh9oys7xik/20130904_002.jpg
[05:33:20] <archivist> have now
[05:40:00] <kengu> there will be some major reorganizing needed when all the machines start to work
[05:46:45] <MacGalempsy> that looks like a pretty heavyduty setup. What kind of parts are you going to cut?
[05:57:17] <kengu> i have no idea. I am just trying to get the plasma moving (and cutting)
[05:57:51] <kengu> uuh. x axis moving. i need to figure out timing settings
[05:59:38] <MacGalempsy> u are running linux cnc?
[06:00:04] <kengu> yes
[06:02:10] <kengu> 6i25+7i76
[06:02:17] <kengu> one of those, http://www.bsjd.com/en/productshowinfo.asp?id=858
[06:02:32] <kengu> and the other one is dq3722m
[06:15:24] <MacGalempsy> nice. im running the 5i25 and 7i77, and just got the axis tuned, and wired in the last of the relays.
[06:15:34] <MacGalempsy> now just need to program them into HAL
[06:16:13] <MacGalempsy> that is the servo amplifier?
[06:16:23] <MacGalempsy> stepper, my bad
[06:16:59] <MacGalempsy> need to go out to the garage and get back to work. will log in from out there
[08:48:41] <Vinci89> hello there, short question, I seem to have trouble registering at the linuxcnc forum? may it be that the activation is kinda broken?
[08:51:04] <archivist> it can be very fussy it has aggressive anti spam measures
[08:51:44] <Vinci89> great, so who do I have to annoy in order to get an activation mail?
[08:52:22] <archivist> jthornton_, may fix it for you
[08:52:41] <Vinci89> thanks
[08:52:43] <archivist> or ask any questions in here too
[08:54:33] <Vinci89> no questions yet, its still installing =)
[09:05:05] <ju-emb> somebody here that runs a 3D printer with the beaglebone?
[09:08:22] <archivist> ju-emb, you can find someone on the mailing list that does
[09:11:04] <ju-emb> I now, Charles Steinkuehler has contributed some to the machine Kit
[09:11:07] <ju-emb> that I took and did some improvements
[09:12:35] <archivist> best to feed back to him :)
[09:14:33] <ju-emb> ok, I was just asking here, perhaps there is some folks here too who's playing around with the bb
[09:36:59] <somenewguy> does anyone else have experience with the qpocket.ngc sub routine that ships with linuxcnc?
[09:37:10] <somenewguy> whenever I try to use it I get a "unkown m code" error
[09:37:18] <cradek> on what M code?
[09:38:45] <somenewguy> it doesn't say
[09:39:13] <cradek> does it tell you the line number of the error?
[09:39:41] <somenewguy> no, I thought it had in the past
[09:39:49] <somenewguy> but all I get is the little pop-up box w/ a big red X
[09:39:55] <archivist> a basic bug in error messages is not showing what the code was
[09:39:58] <cradek> at run time or load time?
[09:40:03] <somenewguy> "unknown m code used"
[09:40:05] <somenewguy> at run time
[09:40:19] <cradek> are you using the AXIS gui?
[09:40:21] <somenewguy> Iyes
[09:40:32] <cradek> odd
[09:40:42] <somenewguy> and qpocket.ngc is from the set of demo subroutines
[09:40:44] <cradek> any error should be reported at load time
[09:41:02] <somenewguy> I tried copy/pasting the sub into the top of my file to see if it wasn't seeing it for some reason, but that didn't fix it
[09:42:29] <cradek> the only M codes I see in there are m2 and m3
[09:42:57] <cradek> how are you calling qpocket?
[09:43:53] <somenewguy> o<qpocket> call [12][2900][2][25][.004][0.02][0.1][0][0.5][-1.5][0.5][-0.5][5.6181][-0.5][5.6181][-1.5][1][0][0][0][0.35][0.002][1][0]
[09:44:21] <cradek> oh my
[09:44:28] <somenewguy> this all feels super dejavu to me
[09:44:34] <cradek> I mean, do you have a gcode file that calls it?
[09:44:40] <cradek> or are you typing that in mdi?
[09:44:52] <ju-emb> what's the lcnc community preferred pastebin for pictures?
[09:45:19] <cradek> I've seen people use imagebin.org, but really it doesn't matter
[09:45:30] <archivist> imagebin I can see
[09:45:31] <cradek> the ones that add the least amount of bs to your image are best
[09:45:48] <somenewguy> haha
[09:45:53] <somenewguy> yes I am calling it from a gcode file
[09:46:08] <somenewguy> but that plus my pre and post amble is all that is in the file
[09:46:09] <cradek> some try to avoid you getting a direct link, etc, and those are best avoided
[09:46:18] <cradek> somenewguy: pastebin that file please
[09:46:39] <somenewguy> I am quickly checking that my pre-amble isn't messed up, I am only just sarting to use ngcgui
[09:46:52] <cradek> that's what I was going to check too :-)
[09:48:11] <somenewguy> http://pastebin.com/9ZGmsCdU
[09:50:25] <cradek> hm, it has dos line endings. did the pastebin add those, or did your file get infected by a windows machine?
[09:51:40] <somenewguy> linux machine, gedit
[09:52:01] <somenewguy> unless it exsits in the ngcgui tools, I didn't do it
[09:53:03] <somenewguy> arggg
[09:53:15] <somenewguy> I konw i saw this error like a week ago on a sub I had written
[09:53:24] <somenewguy> but I don't remember the solution lol
[09:53:42] <somenewguy> I think I had mis-formatted a variable or something, syntax or summtin
[09:55:42] <cradek> after finding the 7 (!) subsubfiles it needs to run, I'm getting "zero radius arc"
[09:56:57] <cradek> maybe it eventually calls M110 in spin.ngc
[09:57:15] <cradek> which runs halcmd
[09:57:25] <cradek> you probably don't have your M path set up to find that
[09:57:44] <cradek> are you sure your ngcgui setup is correct? (unfortunately I know almost nothing about how ngcgui is used)
[09:57:58] <somenewguy> haha I was able to use a default config pretty much plug and play
[09:58:02] <somenewguy> lemme double check my path
[10:08:57] <somenewguy> dman
[10:09:10] <somenewguy> add explicit path to the ngcgui_lib folder and still no fix
[10:09:28] <cradek> but it's the M code path you're missing
[10:09:44] <somenewguy> where does the m code live?
[10:09:50] <cradek> look at the working ini file
[10:11:27] <somenewguy> oh duh, I see the folder
[10:11:31] <somenewguy> lets see if this helps...
[10:12:54] <somenewguy> THANK YoU
[10:13:19] <somenewguy> I am ready ot see what comes out now
[10:14:02] <cradek> I wonder if AXIS fails to report line numbers in deeply nested subroutines, or if it fails to report M1xx problems
[10:14:06] <cradek> I bet it's the M1xx
[10:14:23] <cradek> because it doesn't know which of those are valid -- only task does (since it searches the directories for them)
[10:14:34] <cradek> that's irritating
[10:15:25] <jmasseo> archivist: Ironically, i acquired a wind up clock for $3 at a garage sale saturday. I thought that was funny after talking to you.
[10:15:44] <jmasseo> ehh ironic not the right word
[10:15:47] <jmasseo> coincedentallyu
[10:16:09] <archivist> some clocks would be over priced at $3
[10:16:16] <jmasseo> this might be one of them
[10:16:17] <jmasseo> it's a regulator
[10:16:33] <jmasseo> paper face
[10:16:37] <jmasseo> wall clock w/ pendulum
[10:16:54] <archivist> not a real regulator then
[10:17:05] <archivist> vienna wall clock
[10:17:16] <jmasseo> it says 'regulator' on the glass
[10:17:17] <cradek> it says collectible, er, regulator, right on it
[10:17:24] <jmasseo> right
[10:17:27] <cradek> :-)
[10:17:28] <jmasseo> it was $3 at a tag sale.
[10:17:32] <jmasseo> i've seen them before
[10:17:35] <jmasseo> i don't think it's anything special
[10:17:43] <jmasseo> i just wanted to strip and study a wind up clock
[10:17:50] <jmasseo> because clockwork is awesome
[10:17:59] <archivist> some of those are......hard to describe nicely
[10:18:03] <jmasseo> and i can always solder the bits onto some silver and sell it as steampunk :)
[10:18:23] <jmasseo> assuming they are real brass
[10:18:25] <jmasseo> and that's not a lie too :)
[10:18:59] <archivist> later ones are really cheap east european rippoff stamped out dross
[10:19:12] <jmasseo> =(
[10:19:18] <jmasseo> at least it was only $3
[10:19:49] <archivist> and then there are the korian copies of the crap ones
[10:19:51] <jmasseo> i don't remember what the name on the face was
[10:19:59] <jmasseo> this guy had some older crap for sale
[10:19:59] <cradek> if it runs and shows the time and chimes pleasantly, it's $3 well spent
[10:20:05] <jmasseo> he had a huge console 8-track/record player.
[10:20:12] <cradek> I had to pay $10 for mine that does that
[10:20:20] <jmasseo> eh
[10:20:27] <jmasseo> i don't want a wind up clock for practical reasons
[10:20:36] <jmasseo> i'm not going to wind it
[10:20:36] <jmasseo> heh
[10:20:43] <jmasseo> i just like clockwork
[10:21:21] <jmasseo> wind up toys, wind up music boxes, etc.
[10:21:26] <jmasseo> anything with a spring and gears fascinates me.
[10:21:43] <cradek> the one on my wall is monthly wind. I'd probably get tired of it if it was daily or weekly.
[10:21:44] <Jymmm> Model T ?
[10:21:54] <archivist> more fun to get into something like http://www.collection.archivist.info/searchv13.php?searchstr=crane+clock
[10:21:55] <jmasseo> i had an automatic watch once.
[10:22:04] <jmasseo> i'd like to buy a movement to take apart at some point
[10:22:20] <cradek> I have on on my arm
[10:22:21] <jmasseo> nice
[10:22:23] <narogon> anybody has used maxon servos????
[10:22:28] <jmasseo> so it's a clock and a music box :)
[10:22:34] <archivist> cradek, erm the 30 day ones are far eastern usually :)
[10:23:31] <cradek> jmasseo: seiko has new and very usable automatics for around $50
[10:23:43] <jmasseo> i think automatic watchs are practical.
[10:23:56] <jmasseo> same with those seiko digital automatics
[10:23:58] <cradek> actually they are quite nice watches, shocking for the price
[10:23:59] <jmasseo> ecodrive or whatever
[10:24:12] <archivist> jmasseo, be careful let the mainspring down first
[10:24:14] <jmasseo> i had a rolex for a while. it's not worth the dollars.
[10:24:18] <jmasseo> archivist: right.
[10:25:23] <archivist> that crane clock iirc did have a barrel go while I was working on it
[10:28:28] <archivist> when you break something you get to make a replacement or repair the part
[10:31:12] <narogon> anybody has used ic2????
[10:31:54] <archivist> ic2 sounds like a part number on a pcb
[10:32:36] <narogon> i have those drivers
[10:32:36] <narogon> http://www.robot-electronics.co.uk/htm/md22tech.htm
[10:32:53] <narogon> maybe i can use them with emc2
[10:33:59] <archivist> with the analog input yes
[10:34:42] <narogon> i found 3 maxon motor in the laboratory, but i haven't datasheet
[10:34:46] <narogon> hehehe
[10:34:51] <narogon> with those drivers
[10:37:31] <archivist> try something like pwm -> filter-> analogue in (0v - 2.5v - 5v Analog)
[10:41:50] <somenewguy> cradek, thanks! I would assume since a sub routine doesn't really have a line number, it just doesn't try
[10:42:07] <somenewguy> like when multiple files get concatonated and THEN inspected, how do you report the line of the error
[10:43:27] <JT_Shop> most editors show what line your on
[10:43:51] <narogon> archivist could i get a good control with it????
[10:44:54] <somenewguy> i just had a problem with using qpocket.ngc where it couldn't find M110, but since that was called from a subrouting within a subroutien it took quite a wihle to find
[10:45:00] <somenewguy> it doesn't even tell you what M code isn't a real one
[10:45:34] <archivist> narogon, with feedback yes
[10:46:03] <archivist> narogon, do your motors have encoders fitted
[10:47:13] <JT_Shop> that's the main reason I don't use "helper" subroutines
[10:48:22] <narogon> yes
[10:48:30] <narogon> they have encoders
[10:49:02] <narogon> http://es.farnell.com/jsp/search/productdetail.jsp?sku=1654885&CMP=KNC-GES-FES-GEN-SKU-G30&mckv=sw3Cpo4YP|pcrid|9219593229|plid||kword|heds-5540%23a11|
[10:51:47] <kwallace2> Error messages usually print the 'filename:line' so one could open the file and count down from the top the number of lines to the offending code.
[11:05:22] <kwallace2> narogon: The motor drive you linked above uses a PIC to convert different forms of input into bridge signals. The motor control characteristics will be dependent on the software in the PIC. The PIC's control loop rate or resolution may be an issue, or may not be four-quadrant which might be needed for a CNC application. The board seems to be meant as a DC brushed motor controller for robot locomotion.
[11:06:42] <narogon> i've found the data sheet of the motors
[11:07:51] <archivist> the analogue in should make the drive into a plain servo with linuxcnc methinks, worth a try
[11:09:18] <narogon> but i think maybe they are so weak
[11:09:22] <narogon> only 40w
[11:09:27] <narogon> 0.15 Nm
[11:10:08] <archivist> with gearing and a smaller application you can experiment
[11:10:36] <seb_kuzminsky> wow, 20% off and free shipping from enco today, codes: VIP20 and VIPUPS
[11:11:39] <cradek> I got that too - they must be desperate
[11:11:44] <cradek> if only I needed something
[11:12:12] <cradek> with no minimum order!
[11:12:14] <cradek> wow
[11:12:19] <narogon> achivist i need them to a 2x1m plasma table
[11:12:55] <seb_kuzminsky> the local hackspace has been talking about buying a drill index and a tap & die set for the metal shop, maybe now's the time
[11:12:58] <narogon> maybe with gearing
[11:15:31] <cradek> http://www.use-enco.com/CGI/INLMKD?SIITEM001=317-9916&SIQTY001=1&SICOUNT=1
[11:15:54] <cradek> these are also on sale. they are worth every penny.
[11:16:10] <archivist> narogon, bit small for that
[11:16:22] <jmasseo> machine taps?
[11:18:11] <seb_kuzminsky> cradek: that link takes me to an empty shopping cart
[11:18:29] <seb_kuzminsky> enco's urls and use of cookies is kind of borked
[11:18:38] <cradek> model number 317-9916
[11:19:10] <cradek> so you can get it for $70, wow
[11:19:25] <cradek> I sure wouldn't want to share them with anyone, though.
[11:19:28] <narogon> achivist
[11:19:29] <narogon> http://www.maxonmotor.es/medias/sys_master/8807089504286/13_362_EN.pdf
[11:19:32] <narogon> the 813
[11:19:56] <kwallace2> narogon: My understanding of plasma is that one needs enough axis speed for the cutting process and 2 x 1m is not small. I would be thinking in the 400Watt or higher range.
[11:20:21] <narogon> i have other servos
[11:20:35] <narogon> but i can't control them without baldor software
[11:20:53] <narogon> because the drivers only have ethernet powerlink and bus-can i/o
[11:23:33] <cradek> seb_kuzminsky: I also have a couple of 300-0050, which you can actually drill holes with if you're careful - but I use the good split-points when I need the hole to be a particular size.
[11:25:46] <seb_kuzminsky> wow, that second one is a pretty complete set
[11:26:07] <seb_kuzminsky> might be good for a shared shop
[11:26:46] <kwallace2> narogon: If this where my project, I would be looking at a Mesa 6i25 for fast PWM out and encoder input, Pico PWM DC amps, and Keling or eBay treadmill 90V DC brushed motors.
[11:29:35] <seb_kuzminsky> cradek: it's strange that the same set in screw machine length costs 3x as much (302-2400)
[11:30:16] <cradek> seb_kuzminsky: yes it has all the sizes - they're adequate drills - I recommend you also buy good split points in the tap drill sizes you use (if you're me, you stock up on #7s)
[11:30:37] <cradek> I agree that's weird - they're just not made in the same quantity I guess
[11:36:17] <cradek> > Most bits are not straight although actual sizes are close. Even the case came apart after a couple openings.
[11:36:20] <cradek> hm
[11:36:27] <cradek> maybe they're worse since I bought them
[11:41:31] <kwallace2> narogon: Oops, 400 Watts might be a bit much, maybe 200 Watts.
[11:41:59] <seb_kuzminsky> hooray for user reviews
[11:43:17] <seb_kuzminsky> guess you get what you pay for
[11:43:34] <seb_kuzminsky> i dont mind buying dull drill bits, as long as they're the right diameter and straight
[12:04:27] <IchGuckLive> hi all
[12:06:44] <IchGuckLive> someone effected by the storms in the USA
[12:07:43] <jdh> what storms
[12:07:58] <IchGuckLive> 80tornados cnn news
[12:08:30] <jdh> heh... bummer
[12:11:26] <CaptHindsight> IchGuckLive: lost power for 2 seconds yesterday. also shoes got wet
[12:12:43] <CaptHindsight> lots of trees down and we're >100 miles from the worst of it
[12:20:57] <IchGuckLive> all ok thats good
[12:23:10] <CaptHindsight> was 20 C and raining sideways all day, today is sunny and back to winter 3C
[12:24:18] <IchGuckLive> i did a 20km walk today nice cold here in germany no sun
[12:28:04] <kengu> what kind of MPM should I have for plywood router. there are all those details for metals but with wood it is something that simple there isn't that much of it online
[12:29:33] <CaptHindsight> http://en.wikipedia.org/wiki/MPM ?
[12:31:31] <IchGuckLive> kengu: depends on your mill Chees or pro your bit your spindel and how toff is the wood
[12:31:56] <kengu> CaptHindsight: SFM for the imperialists
[12:31:58] <IchGuckLive> plywood may go on a 2 teeth tool as fast as your mill on 10mm depth
[12:32:14] <kengu> only one tooth
[12:32:44] <IchGuckLive> not good it may harme your spindel
[12:34:04] <CaptHindsight> IchGuckLive: which spindles do you like to use for aluminum. copper, PCB's and plastics?
[12:34:22] <kengu> that is what they sold me at a tool experts
[12:35:12] <kengu> IchGuckLive: http://www.rinaldi.it/modules_cms/PrCatPage.php?cc=33&pc=4
[12:35:50] <IchGuckLive> CaptHindsight: if the mashine is big enopg go for a hyanyang 1,1kw
[12:35:59] <IchGuckLive> kengu: http://www.youtube.com/watch?v=R2_OM7okHrA
[12:42:50] <CaptHindsight> IchGuckLive: this supplier? http://huanyangelectrical.en.ec21.com/Motor_Spindle--6878300_6879130.html
[12:45:34] <CaptHindsight> they use a: Germany imported SKF6002 bearing.
[12:46:31] <kengu> IchGuckLive: so you are just saying that it is something like max feed..
[12:48:09] <IchGuckLive> kengu: if the spindel stands yes above 7m/min
[12:50:07] <IchGuckLive> kengu: http://www.ebay.com/itm/6tlg-Satz-SP-Langlochbohrer-EDESSO-Z2-S13-RECHTS-o-LINKS-Langlochfraser-SLB-1-R-/
[12:50:19] <jdh> I'm thinking about buying a chinese 40w laser with Moshicrap. I talked myself out of it last time, but can't remember why.
[12:51:01] <IchGuckLive> http://www.ebay.com/itm/6tlg-Satz-SP-Langlochbohrer-EDESSO-Z2-S13-RECHTS-o-LINKS-Langlochfraser-SLB-1-R-/151015808410 sory this is the tool that works fine
[12:51:29] <IchGuckLive> jdh: maybe wasted money
[12:52:18] <IchGuckLive> kengu: this plywood tools are Right or left keep atenchin wen interested
[12:52:44] <kengu> yeah. that is what it states there
[12:53:12] <IchGuckLive> the short ones are some cheeper
[12:53:28] <IchGuckLive> i use them for Foam as well as wood
[12:53:49] <jdh> yeah, I just can't justify $2k for one that sucks less. Seems to be nothing between $600 40w crap and $2000 not-so-crap
[12:54:04] <kengu> jdh: i have the same feeling
[12:54:24] <kengu> i just cannot buy the crap @$600 when there should be something less crap for $1000 but just is not
[12:54:27] <IchGuckLive> and its alays a risk to burn up your shop
[12:54:38] <kengu> i have not yet done that
[12:55:25] <kengu> i might have tried, https://www.dropbox.com/s/h68ngmno41f9lmd/20130815_009.jpg
[12:57:02] <jdh> that one looks pricey
[12:57:31] <kengu> yeah
[12:58:02] <kengu> it was old crap in china and thus cheap(ish). now with linuxcnc it is fine.
[12:58:30] <kengu> https://www.dropbox.com/s/oethgp3scfartg1/20130616_004.jpg
[12:58:42] <kengu> same machine
[12:59:07] <kengu> actually before linuxcnc.
[12:59:51] <jdh> Professional Series CO2 20x12 Laser with RetinaEngrave 3D Ethernet
[12:59:57] <kengu> this is the router we are sort of working on at the moment, https://www.dropbox.com/s/vtqo4gz3pndi7tj/20130616_003.jpg
[13:00:04] <jdh> 'Retina Engrave' doesn't sound like a good thing.
[13:00:28] <kengu> and the plasma that I am working on getting linuxcnc is the red one on the left
[13:00:49] <kengu> jdh: you do not feel like having some engraving in your retina?-)
[13:01:31] <jdh> not really.
[13:01:53] <kengu> the plasma, https://www.dropbox.com/s/zpm78idalzfjibq/20130616_002.jpg in the pic with router thus router moved to next floor from there
[13:02:07] <kengu> jdh: yeah. some laser goggles maybe to prevent retina engraving
[13:08:11] <sono> laser goggles are tuned to specific laser frequencies.
[13:10:23] <kengu> i know. that is why i do not have those
[13:11:10] <jdh> I have some YAG ones
[13:16:30] <Loetmichel> re @ home
[13:29:04] <IchGuckLive> Loetmichel: i orderd a tablet
[13:29:13] <IchGuckLive> 37Euros
[13:30:15] <IchGuckLive> im off BYE for today
[13:50:13] <sirdancealot> is there a reason against doing pcb z-probing entirely in python, spitting out a height map and then preprocessing a gcode file with it?
[13:51:13] <sirdancealot> im getting tired of the half assed g-codes and hal components
[13:51:43] <sirdancealot> or is there something sane out there?
[13:51:53] <andypugh> I can't see any reason it wouldn't work. There was an experimental kinematic created to do the same thing, you might want to look at the way that code worked (converting the probed heights to a triangular mesh)
[13:52:15] <sirdancealot> yeah the probekins
[13:53:07] <Jymmm> pcb z probing? How many heights does a pcb have???
[13:53:38] <sirdancealot> btw can kinematics be stacked? or does that limit the functionality of that module to cartesian(?) machines?
[13:56:28] <sirdancealot> im using some python gui utility that preprocesses the gcode to probe before milling. If you screw up once, you have to find a fresh place and do it all over again
[13:56:48] <sirdancealot> then there is https://github.com/cnc-club/linuxcnc-engraving-comp , with typos or bugs(i dunno) in the readme
[13:57:03] <sirdancealot> cant even make it work with my limited gcode skills
[13:58:09] <sirdancealot> i guess i could reuse the interpolation and visualization from probekins
[14:07:07] <cradek> sirdancealot: a much less unelegant software solution to that problem is http://phk.freebsd.dk/CncPcb/
[14:07:37] <cradek> (although I prefer to use a reasonable machine and double-sided tape to make the board flat)
[14:09:07] <Jymmm> two part epoxy?
[14:09:42] <sirdancealot> somebody rewrote that in python cradek , and thats what im using now. its not very reasonable
[14:10:32] <cradek> I imagine reasonable is in the eye of the beholder
[14:10:57] <Jymmm> cradek: Quit being so unreasonable!
[14:11:49] <sirdancealot> yeah its not nice when just setting things out and making mistakes, since the probing code is inserted at the beginning, and has to be always rerun(on an unmilled surface)
[14:12:23] <cradek> but each board is different, so you always DO need to run it
[14:12:35] <sirdancealot> not about each board
[14:12:51] <Jymmm> Wait, how hard is it to to mount your PCB flat?
[14:12:54] <cradek> ok I must not understand what problem you're trying to solve
[14:13:10] <Jymmm> assuming your table top is flat/parallel to your z ?
[14:13:55] <sirdancealot> i dunno Jymmm , i just assumed that z-probing might be a better way than bothering to make things flat
[14:15:04] <Jymmm> kinda sounds like a half-assed solution to me. If ya start with crap setup, you can't go too much better from there.
[14:15:10] <sirdancealot> cradek, well imagine im a newb, got one 20x30mm board, and make various cuts in it trying to make everything work. i screw up and stop jobs often etc
[14:15:13] <pcw_home> how about some casters that push the board down
[14:15:34] <Jymmm> French cleat
[14:15:43] <Jymmm> edge clamp
[14:16:25] <sirdancealot> now after trying out clamping, two sided tape sounds like a better option to try
[14:16:33] <pcw_home> a triad of spring loaded casters around the spindle would do
[14:16:40] <cradek> clamping the edges doesn't do anything to make the board flat
[14:16:55] <cradek> casters are an interesting idea
[14:17:10] <pcw_home> or vacuum table
[14:17:38] <cradek> I hear commercial engravers ride on the top surface to set the depth, but that seems like any burrs would mess it up
[14:17:50] <cradek> maybe ultra high speed spindles don't leave any burrs
[14:18:21] <pcw_home> I think thats the case
[14:19:09] <cradek> I think mine's about 10k and it should be more like 30k
[14:19:30] <cradek> although I do get really great results...
[14:20:14] <pcw_home> PCB drills are 200K or more not sure how fast they run the edge routers
[14:20:28] <cradek> yikes
[14:20:44] <Jymmm> What about this?
[14:20:47] <Jymmm> ----------/ \--------------------------------
[14:20:47] <Jymmm> | / \ 1/4" material cut on an angle
[14:20:47] <Jymmm> --------/ ============= PCB =================== \-----------------------------
[14:20:48] <Jymmm> ----------------------------------WORK SURFACE--------------------------------
[14:21:42] <sirdancealot> thats a clamp
[14:21:49] <sirdancealot> the middle of the pcb bows up
[14:22:01] <Jymmm> how hard is it being clamed?
[14:22:09] <Jymmm> clamped*
[14:22:36] <sirdancealot> z-probing is just cooler, allows you to do things on things that inherently arent flat, etc:)
[14:22:50] <andypugh> cradek: Keep an eye out for one of these antiques: http://www.lathes.co.uk/precise/
[14:23:52] <Jymmm> sirdancealot: You have a pic of this bowed pcb???
[14:25:44] <sirdancealot> Jymmm, umm. you mean the one i understuffed with a bunch of folded paper so the bowing would at least be firm?:) yeah i do
[14:27:02] <MacGalempsy> afternoon guys
[14:27:03] <sirdancealot> but otherwise its more like 0.2mm movement that it allows in the middle when you press it down
[14:27:17] <sirdancealot> afternoon
[14:28:46] <sirdancealot> ill keep playing with the software if even profi engravers have to solve this
[14:35:02] <MacGalempsy> been tuning on the servos for a few days and am coming up pretty solid with 0.0003 error at about 8in/min. the machine specs say 0.0002, do you think just being used and 20 years old, that this is still fair?
[14:35:36] <andypugh> I suspect that the original spec was not at 8in/min
[14:36:03] <MacGalempsy> on a finishing pass, what kind of feed rates should I be thinking about?
[14:36:09] <MacGalempsy> for alum
[14:37:25] <andypugh> I really don't have any idea about feed rates, I am in the middle of recalibrating away from a rubber Mini-Mill where 15mm/min was pushing it.
[14:37:44] <pcw_home> You may be able to do a little better, how did you get your drive to work? Do you actually have tachometers?
[14:38:15] <MacGalempsy> the odd thing is even if I set the feedrate to the 200in/min it is supposed to be capable, it will only go up to 60in/min
[14:39:02] <MacGalempsy> there are no tachs, the tachs are pluged into the input
[14:39:38] <pcw_home> I can't parse there are no tachs, the tachs
[14:39:39] <ReadError> i made a little jig for pcb
[14:41:13] <ReadError> https://dl.dropboxusercontent.com/u/25091878/Photo%20Jul%2012%2C%204%2052%2025%20PM.jpg
[14:41:14] <pcw_home> What are the tach inputs an the drives connected to?
[14:41:22] <pcw_home> on the drives
[14:43:17] <MacGalempsy> the drive stated it was driven by the difference between the command and the feedback, with the ideal value being zero. so I bipassed the loop by putting the tach feedbacks into the commands going to motor control
[14:43:51] <MacGalempsy> it runs great on all three axis
[14:47:27] <pcw_home> That probably is why you cannot get to 200 IPM
[14:48:16] <pcw_home> Tach inputs are usually not +-10V but +-20V to +-60V
[14:48:50] <pcw_home> (without changing input sensitivity)
[14:53:44] <cradek> or more
[14:55:48] <pcw_home> But its hard to tell. If the original control synthesized the tach
[14:55:49] <pcw_home> feedback from the encoder signals, the tach input might be +-10
[14:55:51] <pcw_home> (and then I dont know why you cant get beyond 60 IPM)
[14:57:14] <pcw_home> You can check easily by measuring the analog control voltage at 60 IPM.
[14:57:16] <pcw_home> if its around 10V then the problem is that you are using the wrong drive input
[14:58:59] <MacGalempsy> im cool with 60in/min. maybe in the future, after the entire machine is up and running, the topic may resurface
[14:59:14] <MacGalempsy> too much other stuff to get up and going ;)
[15:02:14] <pcw_home> to use the proper (reference) input you will have to reverse the feedback
[15:02:16] <pcw_home> (either swap the motor wires or reverse the 7I77 analog output in HAL)
[15:02:17] <pcw_home> Do not use negative PID values
[15:05:46] <MacGalempsy> yes. it required plugging the feedback into the references. then reversing the motor polarity
[15:06:56] <MacGalempsy> no negative PID values, I was under the impression that I and D are not usually used in velocity modes, but why wouldnt someone want to try and dial those in too?
[15:08:13] <pcw_home> You are not running in velocity mode unless you have velocity feedback
[15:09:51] <MacGalempsy> so you think the accuracy would improve if I figured out how to program the encoders to generate the feedback signals?
[15:09:59] <pcw_home> Theoretically negative PID values are OK but the current PID component has a bug with negative integral terms
[15:10:07] <sirdancealot> for anyones reference - https://github.com/koo5/linuxcnc-engraving-comp - fixed the gcode there...one down, just 500321 obstacles to go!
[15:10:12] <MacGalempsy> I mean, through the 7i77
[15:10:18] <pcw_home> (and if one is negative, all will be negative)
[15:12:33] <pcw_home> I would carefully check how your drives are connected, they may be in torque mode.
[15:12:35] <pcw_home> and if you dont have velocity feedback they must be in torque mode for decent performance
[15:12:36] <pcw_home> (that is a velocity mode drive with missing feedback will not perform very well at all)
[15:15:19] <MacGalempsy> well, I can change the resistors on the drives to run in torque mode
[15:16:18] <MacGalempsy> when you say wont perform well, what exactly wont work well? what do you recommend to collect the feedback data?
[15:31:15] <pcw_home> A velocity mode drive without feedback will have a double integration
[15:31:16] <pcw_home> in its control path which will make tuning difficult and sub-optimum
[15:34:23] <pcw_home> I suspect (but do not know for sure) that at 8" per minute and no load,
[15:34:24] <pcw_home> the indicated (via encoder feedback) error should be quite a bit less than .0003"
[15:35:32] <MacGalempsy> .00003
[15:36:13] <pcw_home> OK thats better (how many encoder counts does that represent?)
[15:37:02] <pcw_home> you did say 0.0003 above
[15:37:26] <MacGalempsy> my mistake
[15:37:35] <MacGalempsy> the encoder is 50000
[15:37:41] <MacGalempsy> 2500 counts per turn
[15:37:49] <MacGalempsy> and its 5 turns per inch
[15:38:34] <MacGalempsy> so one click is .00002
[15:38:50] <pcw_home> so is the error 0.0003 or 0.00003?
[15:40:03] <andypugh> 0.3 tenths. I would be ecstatic with that :-)
[15:41:11] <MacGalempsy> let me go run it again. and double check.
[15:41:37] <pcw_home> well thats 15 counts which is not great tuning for a 10000/count/turn encoder moving at slow speeds and no load
[15:42:30] <pcw_home> I would expect a well tuned system to keep within a few counts under those conditions
[15:43:46] <pcw_home> maybe with some bigger spikes at the begin/end of acceleration
[15:49:26] <kengu> what is this "ok to move" signal in plasma that is not wired in my setup at the moment
[15:49:58] <kengu> (well, i am not that far in the connecting, but it wasn't connected in the previous version either)
[15:51:32] <kengu> something avhc-related that is not implemented in my setup (yet?)
[16:08:54] <MacGalempsy> http://www.flickr.com/photos/58181938@N03/10933060086/
[16:09:33] <MacGalempsy> it is interesting how the baseline is 0.0002
[16:10:03] <MacGalempsy> but there is an initial spike in the data
[16:12:31] <MacGalempsy> so the initial spike sets the ferror at 2 encoder clicks
[16:27:09] <_DJ_> gn8
[17:55:26] * JT_Shop has messed up enough this evening and retires to the couch
[18:01:14] <kengu> good choice
[18:18:42] <Jymmm> JT_Shop: Next time... PUT THE SEAT DOWN!!!
[18:27:46] <MacGalempsy> anyone got a moment to look at that link I posted earlier to maybe give some motor tuning advice?
[18:40:21] <andypugh> Try some D
[18:41:51] <andypugh> But a very small amount. You are looking at micro-inches per ms, so start with a very small mumber.
[18:42:40] <andypugh> Like, start at .000001 then 0.00001 and watch how it goes.
[18:43:00] <MacGalempsy> that is to kill the initial spike, right?
[18:43:18] <andypugh> it's to damp the whole response
[18:44:16] <andypugh> If it makes things worse, try a negative number, then I will try to remember what misconfiguration meant that I needed that :-)
[18:45:10] <andypugh> It is instructional to play with a bare motor with only D-gain.
[18:45:14] <MacGalempsy> andypugh: have you had any luck with the autotune PID?
[18:45:57] <andypugh> I have never tried it, but I have the impression that it only ever worked for the guy who submitted it. I might be wrong.
[18:46:11] <MacGalempsy> hah. ok
[18:46:17] <andypugh> It's a great idea.
[18:48:01] <andypugh> I tune loops for a job, but nasty ones where the normal tools don't work.
[18:49:00] <andypugh> I have been working on this oil pump controller for 18 months.
[18:49:05] <MacGalempsy> cool. I tuned the PID on my heated bed on the cnc gluegun, but this takes the cake. the ferror seemed to get better
[18:49:13] <MacGalempsy> oh, dang.
[18:50:06] <MacGalempsy> need to figure out how to export the PID data from halscope because there are a few free PID tuning softwares that could help
[18:50:25] <andypugh> My P gain is the product of two 16x16 maps, the D gain the same. The Igain is the product of 3 maps. I am not sure that this enough to cover the complexity of the system response.
[18:52:36] <andypugh> So, in theory, I have 192 P numbes, 192 D numbers and 288 I numbers to get right. In practice I flat-map a lot of it.
[18:55:56] <andypugh> But given that I control a valve that controls the flow of oil that goes into a spring-loaded chamber that controls the stroke of a pump that pumps the oil that I started with, and the viscosity and hence pressure and flow of the oil are a strong function of tempreature and pressure, it is not a problem ameanable to analysis (especially as large areas of the parameter space are off limits as they destroy the engine)
[18:58:32] <andypugh> If you decide that your system response is very non-linear you do have the option of making your P, I or D gains dependent on other system parameters, I created the lincurve HAL component specifically for that scenario. But it makes life a lot easier if you stick with 3 numbers.
[19:00:03] <MacGalempsy> andypugh: do you know if it is possible to make the 7i77 put put more than 10V for the TACH signals?
[19:00:54] <andypugh> I don't think that they can. In fact i don't know where they get 10V from.
[19:01:40] <MacGalempsy> ha. yeah 10v for the drive signal
[19:02:08] <andypugh> You might be able to do something clever with an Op-Amp. Or ou might be better being clever still and describing the problem to PCW
[19:02:27] <MacGalempsy> 60inches per minute should be plenty
[19:03:19] <andypugh> Sounds a bit slow, but you don't spend that much time in rapids. It's a fast cut.
[19:03:24] <MacGalempsy> we chatted this afternoon a little. the only thing I could think of was sending the ref at a lower voltage, say 5v and let the tach send 10
[19:03:57] <MacGalempsy> the main goal is to get it all running and finetune the speed later
[19:04:48] <MacGalempsy> the thought bipassing the tach loop would induce a lot of error, but from that chart you looked at, you could see the following error is pretty low at 8in/min
[19:05:18] <MacGalempsy> figured that would be nice for a finishing cut
[19:08:58] <andypugh> I am not a perfectionist. If I was you I would be making partd
[19:09:06] <andypugh> (err, parts)
[19:10:36] <andypugh> And _now_ I know I have a Y-axis backlash probelm that I need to worry about: https://picasaweb.google.com/lh/photo/_G0r_v_xAMLbDRznCRhP-tMTjNZETYmyPJy0liipFm0?feat=directlink
[19:11:56] <andypugh> (I do actually have a larger, preloaded, leadscrew, I am just not sure I can make it fit)
[19:12:32] <Jymmm> Hey andypugh
[19:13:24] <MacGalempsy> you couldnt tune that out?
[19:14:15] <andypugh> I am pretty sure it is mechanical.
[19:15:04] <andypugh> I might actually mount an eccentric in the head and deliberately wobble the table to work out what moves relative to what
[19:28:51] <PCW> MacGalempsy: if you want to use free 7I77 channels for Tachometer simulation you will need to set the drives
[19:28:52] <PCW> tachometer input gain to be the same (10V full scale) as the reference input. Pretty sure this is set via a resistor
[19:28:54] <PCW> on that plug in module (If your drive is like the PDF)
[19:29:31] <MacGalempsy> it is like the one on the PDF
[19:30:13] <MacGalempsy> the guy on the phone said the only reason the number is different is because of the custom order, which had some special resistors done
[19:31:08] <MacGalempsy> not that the resistors themselves are special, but they did tuning with the reisitors
[19:35:09] <MacGalempsy> kkhttp://www.flickr.com/photos/58181938@N03/10933060086/
[19:35:22] <andypugh> One of these with CNC would be fun: http://www.lathes.co.uk/rindis/index.html
[19:35:24] <MacGalempsy> sorry
[19:35:55] <MacGalempsy> simple and sweet
[19:37:06] <PCW> Though as Andy said you might just try some D term (It really should be identical to feeding the velocity into the tach input)
[19:37:42] <PCW> A higher sample rate may be need to get as much D term as you want
[19:37:49] <PCW> needed
[19:38:32] <MacGalempsy> ok. the wife is doing laundry at the moment, so I'll have to take my turn on the electricity
[19:39:10] <MacGalempsy> PCW: earlier, you thought the error may be higher without the tach signal. do you think it could get much better than .0004?
[19:39:34] <PCW> Do you have any D?
[19:39:39] <MacGalempsy> not at the moment
[19:39:45] <MacGalempsy> only P, FF1 and FF2
[19:40:01] <andypugh> Can you measure .0004?
[19:40:01] <MacGalempsy> was going through John's tuning tutorial
[19:40:07] <PCW> Well with no Tach feedback D will be needed
[19:40:21] <MacGalempsy> the encoder resolution is .0002
[19:40:30] <MacGalempsy> well, encoder and ballscrew
[19:40:47] <andypugh> Yes, but my Calipers are .0005
[19:41:32] <PCW> what ballscrew pitch, gearing, and encoder counts?
[19:41:41] <andypugh> If this is a lathe, then I would want 10x better, and be using micrometers.
[19:41:58] <MacGalempsy> oops. it is .00002, so I should try to do the feedback thing
[19:42:28] <PCW> Yeah I thought it would be higher
[19:42:31] <MacGalempsy> 5 turns per inch, 2500c/t encoder, and no gear reducer
[19:42:59] <PCW> yeah so 50000 counts/inch
[19:43:03] <MacGalempsy> any thoughts on HAL code?
[19:43:18] <andypugh> For fun, work out how much your machine moves in a cold draught.
[19:43:19] <MacGalempsy> or known example to make this work?
[19:44:05] <MacGalempsy> when I turned it on this morning cold, it had a hard time with my PID values. how long do these things need to warm up before it can be considered steady state?
[19:44:35] <andypugh> Until the remperatures stop chanigng
[19:44:41] <PCW> I would just try D first (since on your drive, the tach input and reference input are just summed this can just as well be done in the PID comp)
[19:45:01] <MacGalempsy> ok. I will try that and see what happens
[19:45:33] <andypugh> That's the best plan. It might help, or you might be near to machine limits.
[19:46:07] <PCW> pump up D until its too buzzy to stand and then back off till its livable
[19:47:10] <PCW> (you can get more at a higher sample rate)
[19:47:12] <PCW> then you should be able to increase P
[19:47:18] <MacGalempsy> the other night, I had it up to 2340 and it was doing well, but the next day, I could barely get 1800 until it was 100% hot
[19:47:48] <andypugh> (and possibly experiment with negative D, if that helps I will be forced to remind myself what misconfiguration made that useful)
[19:47:49] <PCW> Not enough (or any) D
[19:47:51] <MacGalempsy> with the 2340 the table buzzed
[19:48:57] <andypugh> Yes. As the machine warms up it will get more responsive (mainly oil friction).
[19:49:00] <MacGalempsy> maybe it would be good to put together a warmup program to run the machine around a while after first turning it on
[19:49:31] <MacGalempsy> so should I warm it up 100% then do the PID tuning?
[19:49:33] <andypugh> But you may be able to get more P with a hint of D
[19:51:04] <PCW> I think its likely now that theres no D (either from external feedback or the PID)
[19:51:06] <PCW> which will rather severley limit P
[19:52:03] <andypugh> I would say tune for the cold machine. Though this will depend on the work you are doing. Really high accuracy work is done in temperature-controlled roooms, and they use ballscrews and slides with coolant channels running at controlled temperatures.
[19:52:56] <MacGalempsy> no coolant lines here
[19:53:38] <MacGalempsy> was hoping to have the stand done by now, but my uncle went on vacation. the nerve, right?
[19:59:51] <andypugh> It is possible to make the PID terms temperature dependent, but it adds a lot of compelxity.
[20:00:19] <andypugh> Anyway, time to sleep
[20:01:06] <MacGalempsy> going to work on D
[20:01:13] <MacGalempsy> DDDDDS NUTS!