#emc | Logs for 2010-07-18

Back
[00:04:44] <theorb> theorb is now known as theorbtwo
[00:44:28] <andypugh> Watching a kernel build, I just saw all the comedi drivers scrolling past. Somebody really ought to integrate them into HAL.
[01:15:34] <Jymmm> comedi?
[01:16:53] <andypugh> Open-source drivers for a lot of data acquisition cards.
[01:17:02] <Jymmm> ah
[01:18:33] <andypugh> Could be a nice way to get analogue data into EMC2, or generic digital IO
[01:21:11] <Jymmm> anaWHAT?
[01:22:06] <Jymmm> Did you mean analog? (silly brit)
[02:02:21] <lepton> I've got a gantry machine with two steppers (one on either end). I'm setting up homing, and Axis wants me to only home in Joint mode. Unfortunately, joint mode doesn't drive the two X motors together. Is there a way to either home in axis mode, or tie X1 and X2 together in joint mode homing?
[02:07:33] <andypugh> Do you have a homing sequence set?
[02:07:59] <andypugh> I think if you give them both the same homing sequence number then they will home together
[02:19:24] <lepton> Yeap, it's already set like that
[02:19:37] <lepton> Unfortunatelty that' snot helping
[03:18:50] <madsci44> so dija get your kernel built andypugh?
[03:27:24] <morfic_> morfic_ is now known as morfic
[04:16:10] <KimK> cradek: On initial glance (will test it later), it looks like we can get 99 tools using gedit. We only needed 63 or 67 or so, whatever number of holes John drilled in the tool stand. BTW, we are also (on 2.4.2) seeing that error message "parport 4 not supported, continuing anyway". And so I guess we too are "continuing anyway". John says hi.
[06:24:29] <zeeshan> does anyone have the link to the digi key encoder that people use with servos?
[07:09:07] <Jymmm> you mean US Digital?
[07:11:00] <zeeshan> im not sure
[07:11:06] <zeeshan> apparently its cheap.
[07:13:16] <Jymmm> http://usdigital.com/products/encoders/
[07:23:47] <foxtrot> sigh
[07:23:52] <foxtrot> i wish i had a better mill
[07:33:48] <archivist> make what you have better
[07:34:30] <archivist> lean on it see where it bends, add stiffening, rinse repeat
[07:58:37] <foxtrot> archivist, http://epicstuff.50webs.com/cnc/
[07:58:53] <foxtrot> i didnt do it right so my machine is inferior
[07:59:05] <foxtrot> i need a moving gantry, fixed stationary bed
[08:00:03] <archivist> nah you need to learn about stiffness
[08:00:26] <foxtrot> its pretty stiff
[08:00:47] <archivist> I can see from the pics its flexible
[08:01:30] <foxtrot> it shakes a good bit
[08:01:39] <foxtrot> its not working atm
[08:01:53] <foxtrot> im not getting enough out of the nema 17 60oz/in steppers im using
[08:02:09] <archivist> there is a reason mill columns are a box section
[08:04:09] <archivist> slow down on your cuts to reduce vibration etc
[08:04:36] <archivist> never never ever climb mill with a flexible machine
[08:05:16] <foxtrot> climb?
[08:05:21] <foxtrot> z?
[08:07:54] <archivist> http://www.quadrantepp.com/default.aspx?pageid=201
[08:09:24] <archivist> easy to remember to do on a manual machine, make sure gcode you write is the same, cam software expects a solid machine that can climb mill
[08:13:17] <archivist> fix a dial indicator on the bench, probe on part of your machine, press with your little finger on the machine, should be no movement seen on the dial indicator
[08:13:28] <archivist> add stiffness to fix
[08:13:59] <archivist> note you need a stiff bench as thats gonna bend too
[08:33:21] <foxtrot> ive been thinking about that
[08:33:30] <foxtrot> whats the best way to get a stable, perfectly level table
[08:37:40] <archivist> the bench I built my cnc on was 4"x3" legs and front to back with 3"x2" (3off) X direction but the floor was flexible there and the level I used sensed us walking around
[08:40:11] <archivist> when I first made mine the column was too flexible http://www.archivist.info/cnc/works2008/P1010198.JPG
[10:06:30] <MattyMatt> my machine is still on the floor. I can't afford legs for it yet :)
[10:07:59] <MattyMatt> and 2 of these cheap workmates would give halfway decent fold down legs tho
[10:14:09] <MattyMatt> proper vibration-free feet are expensize, but fridges & washing machines have simple screw down feet
[10:32:17] <MattyMatt> I wish I had a dovetail cutter. I've never seen one with a shank small enough for my spindle tho
[10:41:27] <MattyMatt> foxtrot. 60oz in is all I have from my steppers. it's plenty as long as aI don't try to go too fast
[10:43:58] <MattyMatt> mm dial indicator. my machine would rip the needle off the end it's that rough atm :)
[11:21:10] <fragalot> anyone here with experience in PCB engraving?
[11:22:05] <fragalot> trying to figure out which tool tip angle is best suited for this.. 10 15 20 25 30 40 deg etc..
[11:22:22] <fragalot> I'm thinking of 30deg 0.1mm V-bits but i'm not sure if that's the best all-rounder
[11:24:06] <andypugh> I always imagined that PCBs would be done with cylindrical cutters
[11:25:57] <andypugh> To be honest, next time I want a PCB I think I will let somebody else make it. $40 for 10 of?
[11:25:58] <andypugh> http://www.seeedstudio.com/depot/open-source-service-prototyping-service-c-64_12.html
[11:26:13] <andypugh> (Or $20 for small ones)
[11:26:27] <fragalot> I find that rather expensive :/
[11:26:46] <fragalot> obviously the cheaper solutino was to buy a CNC machine to mill them myself >.>
[11:26:47] <fragalot> xD
[11:27:25] <andypugh> Well, making your own is a much shorter development cycle.
[11:27:40] <fragalot> esp. since I only do one-off's
[11:28:13] <fragalot> andypugh: also - reason I'm using V-bits is because I can't find any end-mills that go down to 0.3mm or smaller :P
[11:28:16] <andypugh> I am likely to need 4, but I can imageine seling-on the extras
[11:30:26] <andypugh> For the very narrow gaps you will probably find that it has to be a V-bit, but I think you might need others for clearing areas.
[11:30:42] <fragalot> oh yeah i've got a few 1mm end mills for that
[11:31:09] <fragalot> (and 2mm, 3mm etc up to 6mm)
[11:31:24] <fragalot> shame though that none of them have a depth collet so I need to touch-off every time :/
[11:32:05] <fragalot> (Speaking of which - is there a way to do that when EMC asks you to change a tool without having to stop the program, and remove the first part of the program to make it continue where it left off?)
[11:33:43] <fragalot> * fragalot orders 10 25deg 0.1mm bits
[11:34:03] <fragalot> should be good enough to last until I finally get 'round to fixing the carbide bit sharpener >.>
[11:36:30] <fragalot> also - I have several pyramid engraving bits, and I'm curious as to what material those are best suited for.. they seem to cause a lot of burrs in aluminium and copper :/
[11:42:57] <andypugh> fragalot: No, touch-off during tool change is a contentious issue. :-)
[11:43:13] <fragalot> figured as much :P
[11:43:27] <andypugh> Not that anyone disagrees with the concept, but it would take a lot of work.
[11:43:58] <fragalot> well if I were able to pause EMC, touch off, and then continue where I paused it'd be fine
[11:44:11] <jthornton> so would a lot of other people
[11:44:17] <andypugh> Perhaps you could put in a tool-probe and an M-code after each tool change, to automeasure?
[11:44:38] <fragalot> but I can't seem to figure out how, so I have to stop it, touch off, remove code up to point where I change, start program
[11:45:13] <jthornton> are you changing tool size or for a worn tool?
[11:45:24] <fragalot> andypugh: that would work if I had proper homing switches :P
[11:45:42] <fragalot> jthornton: tool size, changing drills or changing engraving mills
[11:46:21] <fragalot> worn tools beeing replaced would need to be re-measued anyway because none of my tools have depth collets on them so they're never the same length.
[11:46:29] <andypugh> fragalot: No, you don't need homing switches. You have a spring-loaded conductive sensor at a known position on the table, probe down into it, set the offset and carry on, all in G-Code
[11:46:58] <fragalot> andypugh: known position on the sable -> machine has to be homed the same way to be able to figure out the offset.
[11:47:01] <jthornton> for that I would split my program up for each tool
[11:47:01] <andypugh> Only any good for predictable tool changes
[11:47:06] <fragalot> table*
[11:47:21] <fragalot> jthornton: that's what I'm going to do in the future now
[11:47:31] <jthornton> yes but works for predictable tool changes
[11:48:00] <andypugh> No, it doesn't matter if you homed, it still works, the tool probe runs in the program you are running
[11:48:41] <andypugh> You would have something like M6T4M101 to auto-probe
[11:48:41] <fragalot> andypugh: if my homing point isn't the same, how does it know where the tool probe is at then?
[11:48:44] <jthornton> fragalot: can you bottom out your tools in your collet holder?
[11:48:54] <fragalot> jthornton: sadly not :(
[11:49:00] <andypugh> Ah, now you put it that way..
[11:49:05] <jthornton> just a thought
[11:49:25] <andypugh> Add depth rings?
[11:49:37] <fragalot> dno where I could find those things
[11:51:01] <jthornton> can you put something in your collet holder to bottom out your tools on?
[11:51:31] <fragalot> don't know what would be a good idea to put in there as it's spinning at 65000rpm :/
[11:51:52] <jthornton> oh, that is fast
[11:52:29] <fragalot> anyways -> bottoming out won't be a very good solution for long since some of the same tools I have have different lengths due to beeing resharpened. :P
[11:52:51] <fragalot> jthornton: also - do you have any idea what carbide pyramid bits are good for? :? I haven't found a practical application for them yet
[11:53:30] <jthornton> I load all my tools in my mill prior to a job and set the Z offset for each one if you could bottom out you could do the same
[11:53:50] <jthornton> no idea what a pyramid bit looks like
[11:54:11] <fragalot> carbide rod with a 3-sided pyramid tip at the end
[11:54:32] <fragalot> it's supposed to be for engraving aswell and came with my machine
[11:55:01] <fragalot> jthornton: I don't have a tool changer on mine (it's a desktop sized machine) so can't pre-load Z offsets (wouldn't be asking if I could :P)
[11:55:18] <andypugh> My dad has some of those. We have no idea either.
[11:55:19] <jthornton> I don't have a tool changer on my mill either
[11:55:35] <fragalot> jthornton: ah I see what you mean now
[11:55:40] <jthornton> ah ok
[11:56:03] <jthornton> lol I'm the tool changer for my mill
[11:56:15] <andypugh> fragalot: You could potentially put a depth-pin in the bottom of the collet so that you could bottom-out the bits
[11:56:17] <jthornton> it uses quick switch collets
[11:56:46] <jthornton> or holders
[11:56:53] <fragalot> jthornton: that counts as a tool changer imo :P
[11:57:29] <jthornton> no, I have to stand in front of the mill and change tools but the Z is pre set for each one
[11:57:44] <fragalot> I know
[11:58:10] <fragalot> meh i'll just sstick with the method of loading several programs for each tool
[11:58:18] <fragalot> s/several/one/
[12:01:33] <andypugh> Guesses on the minimum safe cylinder wall thickness for a pneumatic cylinder?
[12:01:57] <jthornton> fragalot: that does simplify the process
[12:02:11] <jthornton> how much pressure? what material?
[12:02:35] <andypugh> Does it matter? :-)
[12:02:42] <andypugh> 120psi Aluminium
[12:03:13] <andypugh> 6082 T6
[12:04:16] <fragalot> 7 inches should do the trick :P
[12:04:20] <jthornton> most small pneumatic cylinders I've seen made from aluminum have about 0.080" to 0.100" or so wall thickness
[12:05:04] <jthornton> yea I think 8" would be safe...
[12:05:10] <fragalot> iirc engine blocks are only 0.125"
[12:05:25] <andypugh> OK, so perhaps I can go a bit lower than my current 4mm (0.16)
[12:05:37] <fragalot> 170*
[12:06:10] <jthornton> what does your cylinder do?
[12:06:20] <andypugh> Drawbar release
[12:06:46] <andypugh> So, it is at head height, about 3" from my eyes...
[12:07:31] <jthornton> tie rod type of cylinder?
[12:07:42] <andypugh> No, a sort of cup
[12:07:54] <andypugh> Slightly an odd design, actually.
[12:08:46] <jthornton> do you have a photo of it?
[12:09:46] <andypugh> It doesn't exist yet
[12:09:51] <fragalot> xD
[12:11:28] <fragalot> andypugh: if you're making it yourself, use the engineer's default answer: "it's not strong enough yet."
[12:11:47] <andypugh> http://imagebin.ca/view/yPr__7Mi.html
[12:12:19] <fragalot> shiny
[12:13:20] <andypugh> If I am honest, I bought slightly smaller material than I should have, so I have a little less piston area than I was hoping for
[12:14:04] <jthornton> can you make it a double piston?
[12:14:45] <andypugh> I am looking at that possibility, but it is a fair bit more height, work, and seals.
[12:15:05] <jthornton> yes it is triple the work at the least
[12:16:12] <jthornton> we use some that have a build in air channel for the second -nth piston for extension but only the first piston is use to retract and the rest just vent on retract
[12:18:32] <andypugh> Well, my calculations suggest that 100psi on a 3mm wall thickness 76mm ID cylinder is 10MPa. The material has a 250MPa yield strength, so I have decided I am no longer worried
[12:21:41] <fragalot> lol
[12:21:54] <fragalot> <3 how you use psi and mm
[12:31:24] <fragalot> haha
[12:31:51] <fragalot> firefox 4.0b1 go to "Themes" apparently the Default 4.0b1 isn't compatible with firefox 4.0b1
[12:31:55] <fragalot> :3
[12:39:40] <gangsta> hi
[12:40:59] <gangsta> is there anyone here who could help me with pwm/pdm spindle control?
[12:41:49] <jthornton> depends on what gang your in :)
[12:41:59] <gangsta> lol
[12:42:17] <jthornton> just ask the question
[12:43:34] <gangsta> I need 0-10v analogue out for my VFD, I cannot afford a fancy card with analogue out. Can I just run the pwm through a low pas filter and voltage doubler?
[12:44:07] <gangsta> or is it more complicated than that?
[12:48:40] <jthornton> are you trying to do threading or something fancy like that?
[12:49:11] <gangsta> not yet, mabey later
[12:49:50] <gangsta> I have a spindle index signal, but threading is not a priority just now
[12:51:03] <jthornton> your using the parallel port to control your machine?
[12:51:17] <gangsta> yup
[12:52:14] <jthornton> I've not heard of anyone doing that but that don't mean it won't work... have you searched the wiki site yet?
[12:53:37] <gangsta> yeah, not much on there regarding it, and pretty much nothing about pdm. infact pdm is only gets a few lines in the integrators manual
[12:54:07] <gangsta> tis is the circuit im thinking of - http://www.picbasic.co.uk/forum/attachment.php?attachmentid=40&d=1077243747
[12:57:01] <jthornton> you might post that schematic on the forum or the mailing list... Peter from Mesa visits the forum a lot and is very helpful with this kind of thing
[12:58:54] <gangsta> excellent, i shall do that now
[12:59:05] <gangsta> thanks
[13:07:26] <moop> anyone know about inverse kinematics?
[13:07:35] <moop> for puma type robots?
[13:10:11] <moop> how do i set linkage length?
[13:28:27] <andypugh> Darn, I missed gangsta watching the MotoGP. I have a schematic and everything :-(
[13:29:22] <andypugh> moop: Does the Puma kinematics create parameters? If so, they go in there.
[13:36:40] <andypugh> moop: I just checked the source. Pumakins should create 4 pins (pumakins.A2, pumakins.A3, pumakins.D3, pumakins.D4) which you need to "setp" to the correct values in your HAL file.
[13:40:25] <moop> I am probably being dense, but i also have no experience or training in robotics or kinematics
[13:47:21] <moop> but the emc docs dont say what the pumakins pins represent?
[13:48:28] <moop> and the puma sample config dont even use the A2 A3 D3 D4 pins?
[13:49:42] <moop> I have only spent an hour researching, maybe i have not been searching for the correct things?
[13:51:05] <moop> I found : http://www.linuxcnc.org/component/option,com_kunena/Itemid,20/func,view/catid,10/id,1332/lang,english/
[13:51:53] <moop> but that shows no solution??
[13:52:52] <andypugh> The parameters default to the correct measurements for a Puma 560, it seems
[13:53:13] <moop> I have had a quick look at the src/emc/kinematics/pumakins.c source file but that has just confused me more??
[13:53:42] <moop> A2 is the lowest arm?
[13:54:38] <andypugh> I have no idea at all. :-)
[13:55:05] <andypugh> I know nothing about robots (or kinematics, really)
[13:55:44] <andypugh> I think scarakins is less friendly than pumakins
[13:57:08] <moop> I get the feeling this is going to take me a long time to workout
[13:57:16] <andypugh> I guess that D and A are the Devanit-Hartenberg parameters
[13:57:38] <andypugh> http://en.wikipedia.org/wiki/Denavit-Hartenberg_Parameters
[14:01:36] <andypugh> Yes, watching the video, that all makes perfect sense (it describes parallel axes towards the end, and you do just need D and A
[14:23:40] <moop> A is the angle D is the distance?
[14:24:04] <moop> at joints 2,3, and 4?
[14:24:53] <moop> I think i need to do more research as i think i still dont get it
[14:26:08] <moop> andypugh: thanks for the advice, it did help, but i think i still dont have enough background to know what im doing yet
[14:26:33] <andypugh> Watch the video on the Wiki page
[14:27:03] <moop> the video talks about alpha and r the radius. not A and D?
[14:27:13] <andypugh> They are both distances. D is the dislacement from one joint to the next in the Z direction of the joint, and A (normally called r) is the distance between the axes
[14:27:42] <andypugh> I assume that the design of a Puma means that some of the displacements are always zero, so they have no pins.
[14:28:33] <andypugh> In a Puma the axes are parallel, so there is no theta or alpha
[14:29:26] <moop> i can see roughly what the video is getting at
[14:30:09] <moop> in my robot there is no diplacement in the z direction
[14:30:58] <moop> so I only need the A distance (the length of the arm link)?
[14:33:19] <moop> i think its almost sinking in to my dull brain cells
[14:33:30] <moop> cheers again for advice andy
[14:34:08] <moop> I go and do some more configuration testing and experiments
[15:19:45] <andypugh> Do I need Texlive to compile a kernel?
[15:20:53] <Jymmm> i doubt it
[15:21:02] <andypugh> It's pretty huge
[15:21:37] <andypugh> I have an 8GB ssd on the machine, and I keep running out of space at the end of the compile
[15:22:03] <moop> I have no idea, but when i tried compiling emc apt said i needed lyx and some other doc utils but i did not bother apt-getting them and it still compiled OK
[15:22:15] <Jymmm> SSD or CF?
[15:22:34] <andypugh> Not sure.
[15:22:48] <andypugh> It's a little thing that plugs into the SATA port..
[15:23:04] <Jymmm> size of a hdd or smaller?
[15:23:26] <andypugh> Smaller. Much Smaller.
[15:23:37] <moop> probanly a compact flash?
[15:23:39] <Jymmm> then it's not SSD,
[15:23:40] <madsci44> If its SATA would have to be SSD then no?
[15:23:51] <andypugh> When I say it plufs into the SATA port, I am meaning without a cable.
[15:23:53] <Jymmm> madsci44: no
[15:24:43] <madsci44> only CF's I have seen are all IDE - i guess there must be adapters now or something
[15:24:54] <moop> http://www.addonics.com/products/flash_memory_reader/adsahdcf.asp
[15:24:56] <andypugh> t calls itself "DOM"
[15:25:13] <Jymmm> Disk On Module
[15:25:20] <madsci44> kewl
[15:25:48] <andypugh> http://www.kingspec.com/solid-state-disk-products/series-dom.htm
[15:25:50] <Jymmm> how much did you pay for it?
[15:26:15] <moop> all my technology is at least 5 years old, I dont even know what sata is
[15:26:25] <andypugh> £25
[15:26:49] <Jymmm> andypugh: It's not SSD, just sorta kinda close
[15:27:45] <fragalot> * fragalot considers buying Vecttric cut2D
[15:27:47] <fragalot> actually
[15:28:01] <fragalot> i'd buy it if I could get the trial version to run on my workstation instead of only on my Eee >.>
[15:28:53] <atmega> buy me a copy while you are at it.
[15:29:43] <fragalot> I need to be sure it 'l actually work on my workstation before I spend 210 euro on software though
[15:30:00] <Jymmm> andypugh: If you look at the menu on the left of that link, you'll see SSD's
[15:30:19] <andypugh> What is the difference in practical terms?
[15:30:29] <archivist> none
[15:30:56] <archivist> the box and screw holes
[15:31:43] <madsci44> SSD's (unless this has also changed since I last looked) are faster, way more expensive
[15:32:18] <madsci44> I think they tend to have a higher write-cycle life as well
[15:33:46] <archivist> only by spreading the writes over the area
[15:34:26] <madsci44> im not sure if thats the only reason - many modern UDMA CF cards have wear levelling, some dont
[15:35:01] <archivist> I still dont trust them a lot
[15:35:43] <madsci44> yeah - I use CF's for emc boxes - most var and temp stuff gets set up on ramdisk to reduce writes
[15:36:50] <Jymmm> Never use them if you need security
[15:36:52] <archivist> I got burnt in the early days using flash as eprom replacement on a development job
[15:37:15] <madsci44> security? in what sense do you mean?
[15:37:42] <madsci44> They are great if you have a console that might get bumped and jarred alot
[15:38:03] <Jymmm> madsci44 I said security, not reliability
[15:38:48] <madsci44> ah... I use mine on a udma/IDE adapter - similar to the ones in that link (and now im going to start using the sata versions now that I know about it :-)
[15:38:48] <moop> I got some ims im483 stepper drives from ebay that had flash burnout probelms
[15:39:00] <madsci44> so i dont see what the diff would be over a hard drive
[15:39:25] <moop> most flash only have 100k read cycles?
[15:39:57] <Jymmm> moop: They have increased that now, but still something to be aware of.
[15:40:02] <madsci44> write cycles - and often 10K
[15:40:23] <madsci44> read is unlimited except retention time - which is like 10 years
[15:40:46] <moop> i think they also have a max number of reads?
[15:40:51] <Jymmm> Heh, I've never seen a hdd last 10y
[15:41:04] <madsci44> i have
[15:41:06] <Jymmm> moop: nope, read is unlimited,
[15:41:13] <madsci44> but i wouldnt expect that from newer ones :)
[15:41:23] <moop> i have a 15year old hd, but i have not used it for 5 years
[15:41:41] <Jymmm> If you have an operational 10year hdd, it's time to upgrade the whole system
[15:41:55] <moop> i hope its still working, it has my porn collection on it
[15:42:15] <moop> maybe i should back it up?
[15:42:39] <madsci44> lol
[15:42:42] <Jymmm> Yes, I have a working 486SX25 with 30MB hdd. Doesn't mean I use it
[15:43:29] <moop> i was using a 486dx up till about 5 years ago
[15:43:39] <madsci44> i have an old dual p2 system with a seagate 10gig SCSI cheetah on it - when they first came out around 1999, it has been a server for ages, it still runs
[15:43:51] <madsci44> spend probably 95% of its life running 14/7
[15:43:52] <Jymmm> O_o
[15:43:54] <madsci44> er 24/7
[15:44:28] <madsci44> when i first got the cheetah - it was the early 10krpm drive, i thought that wont last long - it runs so hot - but its still going
[15:44:32] <Jymmm> Get rid of that crap - quit being a packrat
[15:44:58] <madsci44> heeh
[15:45:16] <madsci44> its my only SCSI based system so its still handy for stuff
[15:45:20] <Jymmm> Donate it all to some school
[15:45:55] <Jymmm> madsci44: keep one scsi card and cable, and toss the rest.
[15:45:56] <jthornton> Jymmm: did you call me?
[15:46:09] <Jymmm> jthornton: On the phone, no.
[15:46:12] <moop> my epia home server has been running 24/7 since 2003, but its had a replacement mb and hd
[15:47:59] <Jymmm> jthornton: Did you need me to?
[15:48:02] <jthornton> a packrat :P
[15:58:14] <Jymmm> * Jymmm calls jthornton @ 1-800-packrat
[15:59:32] <madsci44> oh no theres a hotline now?
[16:00:07] <archivist> * archivist looks at the pile and thinks 3-800-packrat
[16:00:08] <Jymmm> Yeah, call that number and they will send over a pod so you can store MORE stuff that you don't have the room for
[16:00:34] <Jymmm> http://www.1800packrat.com/
[16:00:42] <madsci44> pretty soon it will be cheaper than disposal heh
[16:00:55] <Jymmm> not at $3000 it won't
[16:01:36] <Jymmm> they can be more expensive that movers
[16:02:14] <Jymmm> Just get yourself a crane and some cargo containers, at least you can stack them
[16:04:30] <madsci44> i got lots of space
[16:06:33] <Jymmm> ds2: http://sfbay.craigslist.org/sby/tls/1849406411.html
[16:06:48] <madsci44> i wanted to start a CNC/machinery/gadget club here - i have the perfect area to host it in, lots of parking, etc but cant find people
[16:08:20] <Jymmm> You mean a Hacker Dojo?
[16:08:57] <madsci44> kinda but not exactly that - along the lines of a LUG - like an EMC2UG
[16:08:58] <madsci44> heh
[16:11:04] <Jymmm> LOL, at first I thought it was a PERSON on the rocks, not a bottle of water... http://sfbay.craigslist.org/sby/grd/1848702650.html
[16:11:13] <madsci44> the computer stuff is pretty easy to collaborate over the web - but where theres machinery/mechanics/etc - i think people can benefit way more from local interaction - one guy has a lathe other guy has a plasma table - people can make parts for each other etc
[16:11:59] <Jymmm> madsci44: http://techshop.ws/
[16:12:26] <madsci44> yeah i have seen the techshops
[16:12:36] <madsci44> (online at least)
[16:13:27] <madsci44> im in Ontario, Canada, you can barely buy nuts and bolts at the hardware store anymore here
[16:13:45] <madsci44> deeply yuppified society
[16:14:09] <Jymmm> * Jymmm hands madsci44 putine
[16:14:38] <Jymmm> err poutine
[16:15:07] <madsci44> mm
[16:15:12] <madsci44> ew
[16:15:32] <madsci44> needs more salt
[16:16:37] <Jymmm> top it with rock salt for that crunchy taste
[16:16:57] <Jymmm> goes great with the squeeky cheese
[16:21:29] <madsci44> you dont need it if the fries are well-done enough
[16:21:53] <Jymmm> I wouldn't know
[16:25:54] <madsci44> like there was someone here a couple weeks ago trying out a machine he made with pipe fittings and threaded rod, i have piles of acme rod kicking around and would take me like 30 minutes to whip up some decent drive screws for him, but - cant upload that
[16:26:47] <Jymmm> madsci44: FedEx SAME DAY service.
[16:27:02] <fragalot> my machine uses threaded rod, seems to work well enough for my needs
[16:27:46] <madsci44> yeah not saying it doesnt - just the idea that people could benefit from localised groups/clubs with this stuff is all im saying
[16:29:15] <fragalot> *nod*
[16:29:24] <fragalot> it's perticularly hard to find parts here in belgium
[16:29:25] <madsci44> not everyone is working as a job shop that can justify fedex, or techshop, or factory supply houses for diy/experimental stuff
[16:30:10] <Jymmm> then start one and meet once a month.
[16:30:25] <fragalot> Jymmm: don't know anyone that's got something here
[16:30:50] <Jymmm> Then start a virtual group and go from there.
[16:31:02] <madsci44> yeah
[16:34:55] <lepton> Hey everybody. I'm still trying to figure out a problem with homing the X axis of my gantry machine (which has two steppers on the axis, controlled through gantrykins)
[16:35:02] <lepton> In World mode, everything moves around find
[16:35:04] <lepton> fine*
[16:35:30] <lepton> in Joint mode, the motors turn independetly. Since I have to be in joint mode to home, when I hit "home the x axis," only one motor moves
[16:35:32] <lepton> which is very bad
[16:36:14] <lepton> In my ini file I have both X1 and X2 set to home_sequence = 0
[16:37:16] <madsci44> how many axis total?
[16:37:21] <madsci44> or joints i guess
[16:37:37] <lepton> 3 axes, XYZ, and 4 joints, X1, Y, Z, X2
[16:37:55] <madsci44> do the other two also have a specification for homing_sequence?
[16:38:13] <lepton> yeap, they're 1 and 2, respectivly
[16:38:26] <lepton> Actually, when I hit "home all" in axis, nothing happens
[16:38:49] <lepton> If I hit "home x" or y or z, the joint correspond to that axis moves (so y and z are great, but x only moves x1)
[16:39:43] <madsci44> thats written as "HOME_SEQUENCE = 0" right?
[16:40:04] <lepton> yes (I'm being sloppy in my irc typing :p)
[16:40:13] <madsci44> yeah just making sure
[16:40:48] <madsci44> when you home them separately do they all home properly as intended?
[16:41:14] <lepton> Yes, I've gone through the homing on y probably a dozen times without issue
[16:45:18] <madsci44> not sure what other prerequisites there could be? HOME_OFFSET maybe? youve prolly already read http://www.linuxcnc.org/docview/html//config_ini_homing.html
[16:46:29] <lepton> Yeah, I was referencing that pretty heavily last night
[16:46:47] <lepton> I can throw my ini file up on pastebin, that's probably quicker than trying to explain it
[16:46:48] <madsci44> are x1 and x2 treated as one - like dual drive on a gantry sorta?
[16:47:26] <lepton> http://pastebin.com/C7F1baCK
[16:47:35] <madsci44> yeah - maybe can spot an error that way if one exists
[16:47:59] <lepton> Using gantrykins they're seperately controlled (seperate stepgens, etc), but they're linked together (they move in unison)
[16:48:42] <lepton> I didn't pastebin my hal or classic ladder (which is only used to control my estop and start/reset buttons) files
[16:49:39] <lepton> A few of the constants in the ini are still just placeholders (maxes and mins, etc)
[16:49:57] <lepton> Hopefully I'll be tuning all of that this afternoon (once I move past this nagging homing issue)
[16:54:34] <madsci44> final_vel=0.000 ?
[16:56:29] <lepton> That might be fixed on the machine, actually (the version I pasted up might not include a few tweaks I did on the machine, which isn't on a network)
[16:56:31] <lepton> I'll check
[16:57:05] <madsci44> Z has no latch vel and stuff also
[16:57:28] <lepton> On the machine's computer, the final_vel lines are commented out for all axes
[16:57:33] <madsci44> I would try maybe making the home offset slightly - of the final position
[16:58:16] <madsci44> just to be sure the condition that the previous axis must be at the home offset position as specified in the documentation
[16:58:50] <madsci44> if the final vel is 0 maybe its not considering itself having gone to the offset pos somehow?
[16:59:36] <lepton> Hmm, that's an interesting thought
[16:59:36] <madsci44> or just halting the process, since at 0 velocity whatever the last move is would take infinity
[16:59:44] <lepton> exactly
[16:59:53] <lepton> I'll try more sensible values now
[17:09:36] <lepton> So I changed final vel to something non-zero, and gave everything an a home position of 1, and a home offset of -0.25
[17:09:58] <lepton> Unfortunatelty, no change in my problem (X1 and X2 still don't move together in joint homing, only X1 moves)
[17:11:45] <madsci44> hm - i have never tried "linking" in that manner so could be something there - in my setups i just cheat and run the same ouptut to drive both
[17:12:18] <micges> hi lepton
[17:12:21] <madsci44> well i did try once but didnt work
[17:12:36] <lepton> Hi micges
[17:12:42] <micges> can you shortly describe what's wrong with your homing?
[17:13:14] <lepton> It took me a few tries to get gantrykins working well, but in motion I'm very happy with it
[17:13:20] <lepton> As far as homing goes...
[17:13:50] <lepton> I've got a 4 joint (X1, Y, Z, X2), 3 axis (XYZ) machine, using gantrykins to link x1 and x2
[17:14:07] <lepton> I can home the Y axis happily with my current configuration
[17:14:40] <lepton> However, in joint mode, X1 and X2 are not linked (they're only linked in World mode). When I try to "Home X" in joint mode, only the X1 motor moves
[17:14:42] <lepton> which is quite bad
[17:14:51] <lepton> When I try to "home all" in joint mode, nothing happens
[17:15:02] <lepton> When I "home y" in joint mode, everything goes fine
[17:15:58] <madsci44> in the file you pasted there is no serch_vel or latch_vel for Z - not sure what that would do
[17:16:29] <micges> in current emc design you must home all joints in gantry machines to home X1 and X2 properly
[17:17:02] <micges> you can only home single y and z
[17:17:26] <lepton> That makes sense, but why does nothing happen when I hit "home all"?
[17:17:46] <micges> pastebin.com your ini file
[17:18:00] <lepton> madsci44: I've updated that pastebin link with my up to date file (updated in the last 15 mineuts)
[17:18:09] <lepton> http://pastebin.com/uYhhbUfS
[17:18:10] <madsci44> oh ok
[17:18:11] <lepton> ^New link
[17:18:31] <madsci44> would x1 and x2 have to be last in the sequence then micges ?
[17:18:36] <lepton> The description I wrote for micges applies to the current config / the current pastebin
[17:18:49] <micges> moment
[17:19:15] <micges> lepton: home sequence is counted from 0
[17:19:29] <micges> that's the issue
[17:20:02] <lepton> Oh crap! I have home_sequence listed twice for Y, too
[17:20:21] <lepton> as both 0 and 2 [head slap]
[17:20:30] <micges> madsci44: X1 and X2 must only have same number
[17:20:59] <lepton> What I SHOULD do home in the order Z, Y, X
[17:21:11] <lepton> So that Z can be raised all the way up for when Y and X are homing
[17:21:38] <lepton> but I cooked my limit switch on the Z this weekend by accidently reversing it's polarity
[17:21:43] <lepton> So I have to fake it for the moment
[17:22:00] <micges> so Z=0 Y=1, X1=X2=2
[17:22:02] <madsci44> make Z the home sequence 0 and have the others with 1 and 2
[17:22:37] <micges> ( home_sequence of course)
[17:22:40] <lepton> yeap
[17:22:42] <lepton> I'm trying this now
[17:23:17] <lepton> GOT IT!
[17:23:20] <lepton> Hurray
[17:23:31] <lepton> Thanks, everyone!!
[17:23:36] <micges> welcome
[17:24:32] <lepton> This IRC channel has been a major help in getting this system up and going
[17:24:50] <lepton> Now I've just gotta get modbus going for our Spindle's VFD, and I'm pretty much done :)
[17:25:14] <lepton> Fortunately I've had really good experiences with classic ladder so far, so I think we're gonna get that one alright
[17:25:37] <lepton> I'll try to do a write up on it if / when we do, since there isn't currently a lot of documentation about modbus through classic ladder
[17:28:22] <jthornton> lepton: anything you figure out that needs to be added to the docs just let me know
[17:30:33] <lepton> Will do, jthornton :)
[17:31:13] <jthornton> I'm also JT-.... when I'm at other computers
[17:31:36] <Jymmm> ...and just a PITA all other times.
[17:31:44] <jthornton> :P
[17:31:44] <Jymmm> ;)
[17:36:59] <mikegg> anyone else have an issue where all the calibration parameters are on the first tab, and the other two tune tabs are empty?
[17:38:44] <micges> mikegg: if you run emc in console, do you have any errors when you run emccalibration ?
[17:42:28] <mikegg> it just spits this out. http://pastebin.com/zAt8fBV4
[17:42:34] <mikegg> doesn't look like any errors
[17:43:48] <moop> has anyone looked creating scripts to define hal machine configs?
[17:44:39] <moop> e.g. is anyone working on something like stepconf wizard but for machine axis configurations
[17:44:59] <awallin> can the hal_joystick driver read the position/buttons/wheel of a second mouse connected to emc2 ?
[17:45:35] <jthornton> isn't hal_joystick depreciated?
[17:45:45] <micges> awallin: hal_input has that ability
[17:46:14] <awallin> micges: so if you connect two mice to your machine then one mouse can talk to hal via hal_input ?
[17:46:22] <micges> hal_joystick is depreciated
[17:46:51] <moop> seems to me a script could be writen that ask how many dof your machine has, then ask a few questions about each axis and its joints and then create a hal config automatically?
[17:46:54] <micges> awallin: it should work
[17:47:11] <awallin> I just read about hal_joystick on the wiki. Maybe that wiki page should be modified.
[17:47:35] <jthornton> I thought I put something on all the wiki pages saying that
[17:47:43] <jthornton> what page is it?
[17:48:06] <awallin> joypad-page (maybe I just wasn't reading carefully...)
[17:49:04] <micges> hal_joystick was removed in emc 2.4.1
[17:49:35] <awallin> ok. just someone asking me on another forum how to use a mouse-wheel as a jog wheel...
[17:49:51] <micges> mikegg: pastebin.com your ini and hal files please
[17:50:16] <micges> mikegg: but first: what emc version you're using?
[17:51:15] <micges> there was bug in emcalib fixed in 2.4.2
[18:07:41] <mikegg> http://pastebin.com/bERKbtKP http://pastebin.com/WbAgrmhk
[18:07:47] <mikegg> using 2.4.2
[18:08:47] <mikegg> I think it wasn't a problem with with 2.3.5
[18:12:42] <mikegg> screen shot -> http://imagebin.ca/view/hFfrN3X.html
[18:12:51] <mikegg> the other two tabs are empty
[18:25:15] <JT-Hardinge> mikegg: calibration shows up only Tune 0 and 1 & 2 tabs are empty on my Hardinge (only X & Z are configured)
[18:26:04] <Fox_M|afk> Fox_M|afk is now known as Fox_Muldr
[18:26:14] <micges> JT-Hardinge: I think it should show all used axes
[18:27:38] <JT-Hardinge> 2.4pre all three axis show up on my plasma but on 2.4.2 on my plasma only the first tab has any tune info
[18:28:24] <JT-Hardinge> so something between 2.4.pre and now might have borked it
[18:28:44] <micges> JT-Hardinge: can you check 2.4.1 on plasma?
[18:29:09] <JT-Hardinge> by rolling back or something like that?
[18:30:26] <micges> JT-Hardinge: you have plasma emc from packages?
[18:31:59] <micges> JT-Hardinge: better don't try it, I'll try check it here
[18:32:00] <JT-Hardinge> 2.4.2 was installed by synaptic
[18:32:21] <JT-Hardinge> 2.4.pre was a leftover from something I forget now
[18:44:07] <gangsta> Hi, is there anyone here familiar with electronics, who can help me connect my vfd to Parallel port using pwm through a low pass filter and voltage doubler?
[18:47:55] <awallin> it might be a good idea to have optical isolation btw. pc and vfd
[18:48:47] <gangsta> yeah, i too think that would be a good idea, i have a few opto couplers kicking about here :)
[18:55:32] <micges> JT-Hardinge: yep there is surely bug in emccalib.tcl but I don't know tcl enough to fix it
[18:56:22] <micges> mikegg: it seems that you found a bug
[18:56:43] <micges> mikegg: thanks for report
[18:57:03] <mikegg> wow, really?
[18:57:28] <mikegg> cool.
[19:00:50] <mikegg> should I submit a bug report?
[19:01:59] <lepton> So I'm without a home/limit switch on my Z axis for a little while, since we toasted ours and have to have a replacement shipped in
[19:02:12] <lepton> In the mean time, I want to do a "home all" without homing the Z (I'll do that manually)
[19:02:34] <lepton> If I just remove the home related lines from the Z axis in the ini file, will "home all" skip the Z? Is that a reasonable thing to do?
[19:05:17] <cradek> set Z with zero homing velocities, and still home it
[19:05:33] <cradek> you can't do anything if it's not homed
[19:09:10] <lepton> My temporary solution would be to drive it to a home position, call it homed, and then set local coordinates using a probe plunge / classic ladder
[19:09:33] <lepton> I did order limit switches for it, I just reverse polarity'd two of 'em :/
[19:51:11] <madsci44> does anyone know if there is any advantage or disadvantage to using gantrykins over trivkins for a straight simple XYZ setup - which doesnt need the joint mappings?
[19:51:23] <madsci44> or vice versa for that matter
[19:55:46] <micges> madsci44: there is no difference
[19:57:09] <madsci44> cool thanks
[19:58:36] <madsci44> only diff i have found myself is gantrykins shows joint numbers in the preview when the manual tab is selected, and axis letters when the mdi tab is selected, trivkins nicely just shows the axis letters in either
[20:00:11] <micges> yes
[20:01:14] <micges> gantrykins is 'both' type of kins, so it can operate in joint mode (joints numbers) and world mode (axis names)
[20:01:27] <micges> trivkins is only world mode
[20:03:50] <micges> but there is no functional difference
[20:06:07] <madsci44> ok
[20:06:15] <alex_chally> for some reason the name gantrykins makes me lulz
[20:06:27] <alex_chally> sounds like a name for a cat
[20:07:23] <madsci44> in the case of a gantry im not totally sure i understand the difference between the two modes, except where perhaps being in joint mode is what allows the independance to home a pair of linked axis?
[20:07:55] <madsci44> hah thats a good idea for a cat name - sounds british :-)
[20:09:39] <madsci44> but once its homed - idonno
[20:16:27] <andypugh> I see gangsta came back, with the question I have a ready-made answer for, and I missed him for the second time...
[20:17:04] <archivist> andypugh, do !later nick message text
[20:17:05] <micges> madsci44: joint mode is when you can operate each motor individually
[20:17:11] <andypugh> gantrykins is so that the two axes can home individually (and so auto-square the gantry)
[20:18:10] <micges> madsci44: if you map two joints to one axis in gantrykins, it will have effect on world mode
[20:18:15] <andypugh> !later gangsta I have a circuit that does what you want. Opto-isolated and floating too. You might not need the "floating" part.
[20:18:17] <madsci44> right that part i understand
[20:18:26] <andypugh> That worked well.
[20:18:32] <madsci44> er.. understand regarding andypugh's comment
[20:20:12] <moop> anyone know if load order is important for the derivative module?
[20:20:32] <andypugh> It also means that the two motors can run in in opposite directions. Or for a really eccentric setup, have different leadscrew pitches.
[20:21:00] <andypugh> moop: The functions execute in the order they are added to the thread. It can occasionally matter.
[20:21:56] <moop> what I mean is: is it important to addf ddt.2 before i put addf ddt.3?
[20:22:42] <moop> i suppose it must be as it seems to be giving me problems?
[20:22:45] <andypugh> No, you should add them in the order you want them to execute I think, regardless of their index number
[20:22:53] <cradek> that depends on what you are doing
[20:23:13] <cradek> if you are doing signal -> ddt.1 -> ddt.2 to get a second derivative, you want ddt.1 to execute before ddt.2
[20:23:51] <moop> but if there is no link between the two it should not matter
[20:26:18] <moop> I wonder if the problem may be elsewhere as there seems to be another error now even though I have not changed the hal config?
[20:26:52] <andypugh> The mozmck 10.4 kernel building instructions use a git pull of the kernel from kernel.ubuntu.com. That ends up being 4.5GB, which means there is not enough space on the drive to finish the compile and it fails. (luckily it only takes 3 hours for this to happen). Is there any likelyhood that a vanilla kernel patching excercise will take up any less space?
[20:27:29] <moop> hehehehehe i love linux
[20:29:41] <cradek> andypugh: but I thought you learned your lesson already!
[20:29:48] <moop> there is also a strange bug when loading axis with my hal: it locks with a spinning wheel cursor until i move the axis window by dragging the titlebar
[20:30:15] <andypugh> cradek: I learn lessons every day, and nearly always forget them again.
[20:30:25] <madsci44> me 3
[20:30:37] <andypugh> The lesson in question would be to just run my shiny new board single-core?
[20:30:57] <cradek> mozmck's kernel does smp - does it not work for you?
[20:30:59] <madsci44> wow 4gb
[20:31:10] <moop> my lessons is never try to do anything, it is always harder than you think and will most likly end in failure
[20:31:13] <andypugh> Which kernel?
[20:31:20] <cradek> linuxcnc.org/mozmck/
[20:31:32] <madsci44> my source tree for 2.6.34 is only 848 mb - hah - "only"
[20:31:37] <andypugh> I was trying to get a jump on Lucid, I could only find Karmic Mozmck packages
[20:32:52] <andypugh> Ah, yes. I did spend a long time looking through the mailing list and IRC archives for that link, you know.
[20:33:25] <andypugh> And then decided I was deluding myself and was thinking of http://www.linuxcnc.org/experimental/
[20:41:34] <madsci44> i take it andypugh your trying to make a compact installation?
[20:42:04] <andypugh> It needs to fit on an 8GB drive.
[20:42:20] <madsci44> most of mine will fit on a 1 gb drive
[20:42:31] <madsci44> including gui with axis
[20:42:56] <andypugh> So the standard installation (at 3GB) is fine, but the build files when trying to compile a kernel overwhelm the drive
[20:43:24] <madsci44> but then in the run version i would not include the full build environment, and only the kernel headers
[20:43:47] <madsci44> ah i see
[20:44:36] <dgarr> i saw mention of need for scrollbars on tooledit, for test: http://www.panix.com/~dgarrett/stuff/tooledit.tcl
[20:44:47] <madsci44> yeah ive never tried a kernel compile in the smaller space
[20:45:42] <madsci44> i usually use a hard drive for building and then a minimal install on flash which gets the results
[20:47:14] <madsci44> you could maybe also try using an NFS share from another machine for your kernel tree, then after just install the headers
[20:47:31] <madsci44> would be slower compile i guess
[20:52:35] <andypugh> I would need to be a fair bit cleverer than I am to use any of those options.
[20:52:59] <madsci44> im pretty dense and they have worked for me so far - heeh
[20:53:25] <madsci44> time involved i guess figuring out each thing
[20:58:38] <madsci44> maybe build the kernel on another machine and just install the running kernel and header packages on the target?
[21:09:12] <jthornton> mikegg: there is a bug in emccalib.tcl and if your running a git checkout I can tell you how to fix it if you need to
[21:11:16] <andypugh> madsci44: Now I have found the debs in www.linuxcnc.org/mozmck/ it was all entirely trivial.
[21:12:25] <jthornton> andypugh: the only problem I had was trying to get rtai to run on this quad core... on my dual core the mozmck runs great
[21:12:49] <andypugh> My Atom is only dual, and is already up and running.
[21:20:18] <Jymmm> sadly dual quad octo cores dont speed up much, just lets you run more junk
[21:22:35] <alex_chally> these dudes turned a large 9 axis robot arm into a catapult
[21:22:35] <alex_chally> http://www.manapotions.com/robopult.html
[21:22:39] <alex_chally> freaking awesome
[21:22:45] <alex_chally> I have to find a way to play with one of those things
[21:25:30] <andypugh> My new cnc computer is called "mill" which means whenever I ssh into it it is andpugh@mill. Which always seems to need to be read in a strong yorkshire accent and preceeded by "Eeeh! There's trouble"
[21:28:13] <Fox_Muldr> Fox_Muldr is now known as Fox_M|afk
[21:31:31] <lepton> Hello all, I'm currently working on a scaling problem with our machine. We've gone over our steps -> inch math several times, and are confident that our SCALE values in the ini file are correct
[21:31:56] <lepton> However, we're moving approximately 3.07 inches inch real life per in in EMC commands / Gcode
[21:33:38] <lepton> It does appear to be consistent, that value of 3.07 is a result of averaging multiple measurements (they were all very close to that)
[21:37:10] <Jymmm> what is your scale value?
[21:38:00] <JT-Hardinge> what is the commanded length of the move?
[21:38:04] <lepton> 1527.887453483
[21:38:17] <lepton> Which corresponds to steps per unit (unit = inch)
[21:38:37] <Jymmm> That dont sound right
[21:38:39] <alex_joni> lepton: describe setup
[21:38:45] <skunkworks> what is your hardware? is this a leadscrew machine or rack?
[21:38:50] <alex_joni> motor steps, microstepping, etc..
[21:38:51] <lepton> If I give it a move 1 inch command, I measure it at ~3.068 inches
[21:39:05] <Jymmm> lepton: What is your leadscrew?
[21:39:09] <lepton> It's a microstepping stepper system with rack and pinion
[21:39:12] <skunkworks> ah
[21:39:16] <JT-Hardinge> then your scale is not right
[21:39:19] <lepton> Oh?
[21:39:41] <alex_joni> how many steps / turn has the motor?
[21:40:24] <lepton> It's at 0.05 degrees per pulse
[21:40:43] <lepton> and then there's a gear box after the motor shafter, before the pinon, with a 7.2:1 ratio
[21:40:50] <alex_joni> hang on
[21:41:01] <alex_joni> how did you get the 0.05 degree / pulse value?
[21:41:10] <Jymmm> * Jymmm hands alex_joni the slide rule
[21:41:21] <skunkworks> scale should be.. 498.00764455117340286831812255541?
[21:41:24] <skunkworks> ;)
[21:41:35] <alex_joni> skunkworks: you don't know the pinion size yet
[21:41:47] <Jymmm> alex_joni: (42)
[21:41:59] <alex_joni> but a motor that does 0.05 degrees / pulse has 7200 pulses/rev
[21:42:08] <alex_joni> I'm not sure I ever heard of such a motor
[21:42:11] <Jymmm> yeow
[21:42:20] <skunkworks> I should be listening... (always wanted to know how to calculate a rack system)
[21:42:32] <alex_chally> aren't most steppers 1.8 degrees per full step?
[21:42:34] <lepton> The pinon has 30 teeth and a pitch of 20
[21:42:39] <Jymmm> skunkworks: 36-24-36
[21:42:47] <alex_joni> lepton: lets stay with the motor first
[21:42:58] <JT-Hardinge> and with that gear box it takes 51840 pulses to move the output shaft on rev
[21:43:22] <alex_joni> JT-Hardinge: I still doubt that motor has 7200 pulses/rev
[21:43:52] <JT-Hardinge> so do I alex_joni
[21:43:54] <lepton> I'm basing this off the manufacturer spec sheet, it's an oriental motor ASM98AA-T7.2
[21:44:53] <lepton> It does explicity say 0.05 degree / step
[21:45:52] <alex_joni> I think you are mistankingly reading the datasheet
[21:46:09] <JT-Hardinge> lepton: do you have their drive too?
[21:46:15] <lepton> Yeap
[21:46:33] <lepton> I'm wondering if the issue is confusion about when their gear box comes into play
[21:46:41] <lepton> The stepper has a gear box internally
[21:46:47] <lepton> with a 7.2:1 ratio
[21:46:51] <alex_joni> lepton: the datasheet I'm reading has a "Resolution" field
[21:46:59] <alex_joni> http://site.motadistribution.com/Brochures/orientalmotor/OMUSA_2007_ASSeries_Catalog.pdf
[21:47:25] <alex_joni> but it says somehting like: Resolution 1000P/R ... 0.05 degree/pulse
[21:47:39] <lepton> I'm checking out this: http://catalog.orientalmotor.com/plp/itemdetail.aspx?cid=1002&categoryname=all-categories&productname=all-categories-alphastep-closed-loop-step-motors&itemname=asm98aa-t7-2&&plpver=11&origin=keyword&filter=&by=prod&PDF=T
[21:47:46] <lepton> and yes, I'm confused by that same resolution field
[21:47:48] <alex_joni> I'm pretty sure that means at 1000 pulses/rev you get about 0.05 degrees max error at each step
[21:48:14] <Jymmm> ratio: 7.2
[21:48:17] <lepton> hmmmm, interesting. I interpreted that as 0.05 degree / pulse, rather than error
[21:48:32] <alex_joni> the 1000P/R = pulses/revolution
[21:48:38] <alex_joni> but it also says it's configurable
[21:48:45] <lepton> Yeah, it is
[21:48:53] <lepton> Via the driver (which we do have)
[21:48:55] <alex_joni> 500P/R, to 10000P/R
[21:49:05] <alex_joni> then you should check what value it's set at
[21:49:24] <alex_joni> the 0.05 degree/step is certainly not how far the motor moves for each step
[21:49:28] <lepton> Indeed. I'm going to check that (again) and redo my math to see what happens
[21:49:39] <JT-Hardinge> can you take the motor off and figure out by trial and error what scale gives you one rev then you can go from there and figure in your rack gearing
[21:49:44] <lepton> Yeah, the 0.05 degree/step defiently doesn't seem right at this point :)
[21:49:54] <archivist> and they quote .05 for 500 p/r! someone is math challenged
[21:50:22] <alex_joni> archivist: they mean that at 500 p/r you get max 0.05 degree errors on each step
[21:50:27] <archivist> sorry 100 but still !!
[21:50:30] <archivist> 1000
[21:50:35] <alex_joni> whatever ;)
[21:50:46] <alex_joni> so it travels 360/1000 +/- 0.05
[21:50:53] <JT-Hardinge> * JT-Hardinge goes back to cleaning up the shop
[21:51:28] <alex_joni> see page C-39
[21:51:37] <alex_joni> Resolution select switch:
[21:51:41] <archivist> I agree its 360/steps angle per step
[21:51:47] <alex_joni> 1000 x 1 -> 1000 Pulses (0.36 degree/step)
[21:52:46] <alex_joni> lepton: so how do you have X1, X10 set?
[21:52:47] <archivist> actually that should be accuracy not resolution they are using the wrong word
[21:53:07] <alex_joni> archivist: indeed they are
[21:53:31] <alex_joni> lepton: there are 4 small dip switches on the drives
[21:53:40] <alex_joni> the first one is by default on the left (1000)
[21:53:46] <alex_joni> and the second one too (x1)
[21:53:48] <lepton> Yeap, I'm all to the left
[21:53:52] <lepton> So 1000, 1p
[21:54:01] <lepton> x1
[21:54:12] <alex_joni> ok, so you have 1000 pulses / rev
[21:54:24] <alex_joni> how is the 1P/2P set up?
[21:54:29] <lepton> 1P
[21:54:35] <alex_joni> ok, so step/dir
[21:54:39] <lepton> Yeah
[21:54:43] <lepton> I've had motion going for a while :)
[21:54:46] <lepton> Just with bad scaling
[21:55:03] <alex_joni> ok, so we have 1000 steps/rev
[21:55:03] <lepton> Haven't actually cut anything yet, I took a big detour into classic ladder land
[21:55:08] <alex_joni> next is the gearbox
[21:55:14] <alex_joni> you said it's ..
[21:55:30] <alex_joni> 7.2:1 ?
[21:55:56] <lepton> Yes
[21:56:08] <alex_joni> so 7.2 turns of the motor -> 1 turn of the pinion?
[21:56:14] <lepton> Yes
[21:56:24] <alex_joni> 7200 steps -> 1 turn of the pinion
[21:56:38] <alex_joni> ok, next is the rack/pinion
[21:57:29] <alex_joni> pinion has 30 teeth ?
[21:57:35] <lepton> Ye
[21:57:38] <lepton> yes*
[21:57:43] <alex_joni> you said pitch is 20?
[21:57:47] <lepton> Yes
[21:57:47] <alex_joni> 20 mm ?
[21:57:50] <alex_joni> 20 " ?
[21:57:51] <lepton> inches
[21:58:07] <archivist> 20 per inch teeth?
[21:58:09] <andypugh> 0.05 degrees * 7.2 is 0.36...
[21:58:21] <alex_joni> 20 teeth / inch ?
[21:58:31] <andypugh> I think the 0.05 is degrees per step on the output shaft
[21:58:44] <alex_joni> andypugh: could be ;)
[21:59:20] <alex_joni> lepton: 20 teeth / inch ?
[21:59:36] <lepton> Well, on the rack, I measure 6.3514579 teeth per inch
[21:59:40] <JT-Hardinge> which makes sense if you have 1000 pulses per rev then a 7.2:1 gear box
[21:59:48] <andypugh> My Atom is up and running GLX gears (beautifully) and giving me 10,000 worst case latency
[21:59:55] <alex_joni> lepton: on the rack is what we need to know
[22:00:10] <archivist> I think perhaps his DP is 20
[22:00:25] <lepton> Definetly measuring 6.3514579 teeth per inch on the rack, after averaging several measurements
[22:00:35] <andypugh> rack tooth spacing on DP gears will be factors of pi
[22:01:20] <alex_joni> right
[22:01:23] <andypugh> I think you are measuring 6.283185307179586 inaccuratley
[22:01:24] <madsci44> 4 mm spacing per tooth?
[22:01:47] <lepton> Yeah, since that's just barely off of a factor of pi, I agree
[22:02:19] <alex_joni> lepton: can you measure aprox. how far one turn of the pinion gets you?
[22:02:22] <lepton> So it's probably 6.2831853 teeth per inch
[22:02:34] <alex_joni> it should be 30/6.2831853 "
[22:02:46] <alex_chally> why don't you put two pins in the rack teeth, measure across them, subtrac the radius of both pins?
[22:03:06] <cradek> you guys are smart. I didn't notice 3.07 is near pi.
[22:03:41] <lepton> I'll trying measuring that, alex_chally
[22:03:44] <Jymmm> cradek: And there are no blackberries either! Such a rip off
[22:03:45] <alex_joni> well.. it wasn't at first
[22:04:12] <alex_chally> lepton, just make sure you mic over the pins to confirm their size and that they have a good enough surface finish :D
[22:04:30] <lepton> It's a bit tricky, the rack and pinon is quite greasy
[22:04:43] <lepton> We've been running this machine a lot this year, before I started the EMC2 conversion
[22:04:47] <alex_joni> lepton: anyways, you should have a scale (if our guesses are ok) of: 7200steps/rev = 4.77464" -> 1507.964
[22:04:50] <lepton> but I'll do my bess
[22:05:02] <andypugh> What's the specification of the rack? Is it 20 DP, 0.2mod, or some other measure?
[22:05:17] <lepton> Gosh I wish I had actual specs on it
[22:05:28] <alex_joni> measure lots of teeth
[22:05:29] <lepton> but I came with the machine, so thus far I've only been able to manually measure :/
[22:05:45] <alex_joni> diameter of the pinion ?
[22:05:53] <lepton> Measuring lots is how I came up with the numbers I was posting earlier (measuring about 40 teeth, after which I run out of caliper length)
[22:06:00] <andypugh> Travel for 10 pinion revs measured with a tape measure will get you close
[22:06:03] <lepton> Major diameter of the pinion?
[22:06:13] <alex_joni> yeah
[22:06:23] <lepton> As in, the biggest diameter I could possible measure? :p
[22:06:27] <madsci44> 25.4/6.3514579 (his measurement) = 3.999081849 wouldn't metric pitch =4mm make sense?
[22:06:34] <andypugh> Major diameter is no real help, you need the pitch circle diameter, and it is easier to measure the rack
[22:06:40] <Jymmm> Can't you just make a mark at both, rotate one revolution, the measure the distance traveled?
[22:06:59] <Jymmm> Heck, just makr the grease for that matter
[22:07:00] <alex_joni> Jymmm: that's what andypugh suggested, but for 10 turns
[22:07:07] <alex_joni> to average it
[22:07:11] <lepton> ...yes, though I'll need to pull the pinion off the motor shaft
[22:07:11] <alex_joni> madsci44: could be
[22:07:19] <lepton> Since I can't manually spin the motor (strong magnets)
[22:07:23] <andypugh> madsci44: It would make sense, and 4mm pitch is possible, but would be unusual.
[22:07:27] <alex_joni> lepton: but you got CNC ;)
[22:07:32] <alex_joni> set scale to 7200
[22:07:38] <alex_joni> then 1" -> 1 rev
[22:07:53] <alex_joni> g01x10F10
[22:07:56] <alex_joni> measure
[22:08:10] <andypugh> Normally gears are defined as mod (mm diameter per tooth) or DP (teeth per inch of diameter)
[22:09:04] <lepton> It looks like it is a 4mm spacing of the rack teeth
[22:09:10] <Jymmm> alex_joni: Ok, I missed andypugh saying that. (I knew he was good for something)
[22:09:31] <JT-Hardinge> I'm sure glad I didn't have a calculator when I set up my plasma rack... tape measure, dial indicator a few minutes of testing and done
[22:09:48] <alex_joni> lepton: then one turn of the pinion gets you 120 mm ?
[22:10:01] <andypugh> circular pitch gears do exist, so 4mm is possible.
[22:10:09] <Jymmm> JT-Hardinge: Is that a slide rule in your pocket, or and you just happy to see us?
[22:10:29] <alex_chally> i was the only person in my mathclass to raise their hand when the teacher asked if anyone had used a slide rule before
[22:10:41] <alex_chally> I blew his mind right off his face when i then pulled it out
[22:10:49] <Jymmm> lol
[22:10:50] <alex_chally> and did our test with it :D
[22:11:03] <alex_chally> * alex_chally might be a nerd
[22:11:10] <Jymmm> ya think?
[22:11:24] <alex_chally> hey, it is just trig
[22:11:33] <Jymmm> I have a mini one here, have no clue how to use it though
[22:11:38] <alex_chally> and besides, there were a bunch of logs on that test
[22:11:54] <alex_chally> and doing logs on a sliderule is way faster then doing it on a calculator
[22:12:29] <Jymmm> Actually, I should have said... I have a mini one here, for sale. Have no clue how to use it.
[22:12:29] <andypugh> Sliderules are logs, aren't they?
[22:12:40] <alex_chally> andypugh, yeah
[22:12:51] <alex_joni> hmm, if it's 4mm / teeth, then I get 1524 SCALE, which is pretty close to what lepton already had
[22:12:57] <alex_chally> they use logs to make multiplication and division into simple addition and subtraction
[22:13:16] <alex_chally> wikipedia has a good description, and I think a link to a flash slide rule to play with
[22:13:36] <alex_chally> honestly I would use it more but I have a 6", which is only good for about 3 sigfigs
[22:13:49] <alex_chally> which just is not enough for the shop
[22:14:21] <andypugh> alex_chally: I am happy if I get stuff right to 2 sig figs.
[22:14:26] <alex_chally> and has a tendancy to get the last digit only close in math class
[22:15:01] <andypugh> If its closer to 10mm than 11 or 9, it will do :-)
[22:15:13] <alex_chally> andypugh, I like to calculate to at least 1 more digit of accuracy then I need to maintain, and when working in inches that can be 5 digits..
[22:15:54] <andypugh> I assume that the number of digits doesn't depend on the units. Though where the decimal point happens does.
[22:16:00] <alex_chally> example, I tend to use the full value of 1/64 when working in those measurements
[22:16:05] <alex_chally> hence 1 / 64 = 0.015625
[22:16:08] <alex_chally> and lots of digits
[22:17:59] <lepton> Sorry, my connection dropped out there^
[22:18:06] <alex_joni> lepton: wb
[22:18:27] <lepton> I just changed my scale to 7200, but unfortunately a move of 10 would be more than I have range of motion (at that scaling!)
[22:18:41] <lepton> So I did a g01y1f1, instead
[22:18:42] <alex_joni> how about a move of 1?
[22:18:48] <lepton> Which moved ~12.375 inches
[22:18:49] <alex_joni> does that give you one full rev?
[22:18:53] <lepton> Measured very rough
[22:18:55] <alex_joni> of the pinion?
[22:19:00] <lepton> That had to have been a good bit more than a rev of the pinion
[22:19:35] <alex_joni> if your rack has 4mm spacing, and the pinion has 30 teeth-> 1 rev = 120mm
[22:19:50] <alex_joni> 1 rev = ~4.72"
[22:20:04] <alex_joni> so the 12.375" should be about 3 revs
[22:20:36] <alex_joni> so something is still screwy with the motor/gear
[22:22:09] <alex_joni> lepton: can you observe the pinion while moving?
[22:22:46] <lepton> Unfortuntaely, not really
[22:22:51] <lepton> I can barely see part of it
[22:23:17] <alex_joni> try painting one tooth white
[22:23:39] <alex_joni> the part that you can see
[22:23:43] <alex_joni> then set scale to 1
[22:23:54] <alex_joni> and start jogging (slowly) in one direction
[22:24:01] <alex_joni> and see when it does a full rotation
[22:27:36] <lepton> Unfortunatly the drivers won't move when I set the scale to 1... I'll need to change my max vel in the ini file
[22:27:40] <lepton> How about a scale of 10, instead
[22:29:11] <madsci44> 508
[22:29:23] <lepton> 508?
[22:29:31] <madsci44> sorry typed wrong window
[22:29:59] <alex_joni> lepton: sure, 100 or even 1000
[22:29:59] <lepton> :)
[22:30:13] <alex_joni> but start with some lower value
[22:32:06] <lepton> We're going to run this test again, but I just did a scale of 100, and went from 5.6059 to 29.2452 in one rev
[22:33:03] <skunkworks> * skunkworks still thinks it is going to be 500
[22:33:07] <archivist> one rev is a constant distance
[22:34:36] <JT-Hardinge> anyone know what key word I might search for in a tcl file for column width???
[22:35:08] <alex_joni> what tcl file? what column?
[22:35:33] <lepton> skunkworks: How'd you figure that out?
[22:35:39] <lepton> 500 appears to be correct
[22:35:47] <alex_joni> he devided by 3 ;)
[22:36:02] <JT-Hardinge> ngcgui.tcl
[22:36:13] <lepton> But that would suggest that 500 isn't the *exact* value
[22:36:16] <lepton> Just an approximate
[22:36:16] <alex_joni> where's that?
[22:36:49] <alex_joni> 00:41 < skunkworks> scale should be.. 498.00764455117340286831812255541?
[22:36:50] <JT-Hardinge> http://www.panix.com/~dgarrett/ngcgui/README
[22:37:14] <alex_joni> JT-Hardinge: oic
[22:38:33] <JT-Hardinge> alex_joni: any clue on what to look for to change a column size?
[22:39:52] <alex_joni> depends what the column is
[22:40:16] <alex_joni> usually it's a control of some sort, and you search for the control type online to see if it has some size parameters
[22:40:35] <alex_joni> but it can also be inside a grid, and you need to change the grid size
[22:40:43] <skunkworks> actually - I divided by 3.07 and multiplied by 1. (been a long day)
[22:41:07] <JT-Hardinge> I'm trying to widen the 4th column http://www.linuxcnc.org/component/option,com_kunena/Itemid,20/func,view/catid,40/id,3408/lang,english/
[22:41:23] <lepton> Gotcha. We're trying to find an analytical solution to make us feel better (and be confident in our accuracy)
[22:41:38] <alex_joni> JT-Hardinge: does it change size when you resize the window?
[22:41:40] <archivist> lepton, is your rack a dp pitch or metric 20dp should be 4.71" per rev
[22:42:10] <JT-Hardinge> LOL I've not tried that
[22:42:38] <JT-Hardinge> it only allows vertical sizing
[22:42:46] <alex_joni> change that first
[22:43:19] <JT-Hardinge> it still only allows vertical sizing
[22:43:36] <alex_joni> there should be a WMRESIZE or similar
[22:43:42] <alex_joni> for the whole canvas/window
[22:44:15] <JT-Hardinge> don't seem to be
[22:44:36] <JT-Hardinge> I'm finding stuff like set wC [frame $wL.create -bd 2 -relief sunken] ;# create frame
[22:46:11] <moop> has anyone made any hal configs with strange kinematics, my application has 2 motors that move a grab about the x axis when running the same directions and about the z axis then running in opposite directions....
[22:46:38] <alex_joni> JT-Hardinge: vert {wm resizable .ngcgui 0 1}
[22:47:00] <JT-Hardinge> this seems to be the section where the form or grid is created http://www.pastebin.ca/1903241
[22:47:19] <alex_joni> JT-Hardinge: change that to 1 1
[22:48:30] <andypugh> moop: I have experimented with kinematics
[22:48:47] <JT-Hardinge> the columns don't expand just the container
[22:49:29] <andypugh> moop: Have you seen this rather clever bit of string-kins? http://www.cnc-hotwire.de/
[22:52:14] <alex_joni> JT-Hardinge: then it sounds like it's fixed width
[22:52:26] <alex_joni> try finding out how the column is called
[22:52:34] <alex_joni> and see when it gets packed, it gets a width
[22:52:47] <JT-Hardinge> looking now
[22:53:19] <JT-Hardinge> tcl is tough to read
[22:53:21] <JT-Hardinge> ok
[23:01:11] <JT-Hardinge> easier to just shorten my descriptions :/
[23:04:31] <alex_joni> JT-Hardinge: it should be around '# use entry since it can be expanded by user to see overfill '
[23:04:36] <moop> http://82.152.150.86/ppj/pegasus/robot4.jpg
[23:05:52] <alex_joni> JT-Hardinge: priv msg
[23:06:39] <alex_joni> moop: what's the real question?
[23:06:42] <alex_joni> about kins?
[23:06:55] <celeron55> that looks interesting 8)
[23:06:59] <moop> that hotwire is no help andy, http://82.152.150.86/ppj/pegasus/robot4.jpg shows the robot arm i want to control. The grabber has bevel gears that rotate it about one axis when they rotate the same direction and another axis when they run opposite directions
[23:07:43] <moop> I want to know if this is implemented anywhere in emc samples
[23:07:56] <alex_joni> not as such
[23:08:04] <alex_joni> but similar (in theory) there is..
[23:08:09] <alex_joni> for example on a hexapod
[23:08:27] <alex_joni> http://synthetica.eng.uci.edu/~curtis/images/project2_1.jpg
[23:08:35] <alex_joni> you still have compont movements
[23:09:13] <alex_joni> in your case you would have: axis_r = (j1 - j2) * ...
[23:09:33] <alex_joni> axis_s = j1+j2 * ...
[23:09:37] <alex_joni> or something like that
[23:09:46] <alex_joni> just need to figure out the math, the emc2 part is easy
[23:09:58] <moop> i would have to write some c code to do the kinematics
[23:10:05] <alex_joni> of course
[23:14:49] <andypugh> If the kinetics is parallel then you have to write C. I think that genserkins can do any general serial kinematics (though figuring it out is a task in itself)
[23:17:25] <alex_joni> this isn't quite a serial kin
[23:18:28] <andypugh> I can't get the picture to download.
[23:20:27] <andypugh> Time to log
[23:39:46] <JT-Hardinge> say goodnight Gracie