#linuxcnc | Logs for 2013-05-25

[00:20:08] <grandixximo> why if i use index only homing, the axis always returns to the original 0 position after reading the index?
[00:20:19] <grandixximo> shouldn't the index position be zero?
[00:24:48] <tjtr33> yesterday on irc CBJamo mentioned the MIT fablab machines, and one of their designs has a spindle made from off the shelf components.
[00:25:03] <tjtr33> they say up to 20Krpm.
[00:25:03] <tjtr33> i dug it out of the fablab-orynth ;) http://mtm.cba.mit.edu/toolheads/spindle/makeit.html
[00:38:11] <tjtr33> https://github.com/mkeeter/kokopelli some interesting cad/cam tools in python from mit
[02:15:49] <DJ9DJ> moin
[03:57:45] <grandixximo> can angular axis work from 0 to 360 and at each round go on counting from 0?
[04:04:17] <Loetmichel> mornin'
[04:06:23] <Loetmichel> ... from the bath tub ':-)
[04:06:28] <archivist> grandixximo, look at the wrapped mode
[04:14:43] <grandixximo> i don't find G91.2 and G90.2 in the g-code reference
[04:14:56] <grandixximo> don't need them?
[04:15:14] <grandixximo> or must i go with G91 to have wrapped functioning properly?
[04:38:53] <grandixximo> wrapped is not working properly
[04:40:55] <grandixximo> http://linuxcnc.org/index.php/english/forum/38-general-linuxcnc-questions/17804-4th-axis-rotation-angle-limit-problem?limitstart=0
[04:40:59] <grandixximo> third post
[04:43:11] <grandixximo> i cannot go under -0.001 with wrapped active
[04:43:39] <grandixximo> even if my limit is -999999
[04:43:58] <grandixximo> doesn't make sense if it's wrapped forward why not backwards?
[05:00:11] <archivist> grandixximo, add bugs, I think here are some problems with the way wrapped has been implemented
[05:00:34] <grandixximo> no i think it isn't working at all
[05:00:55] <grandixximo> in the documentation there is nothing present, only in the wiki
[05:01:29] <grandixximo> and if i do WRAPPED = 1 i should not have any limit, but instead the soft limits are still there
[05:01:54] <archivist> I have moaned and seen others have problems with wrapped
[05:01:59] <grandixximo> doesn anything change from WRAPPED=1 to WRAPPED = 1?
[05:03:51] <grandixximo> i'll try without space
[05:03:55] <grandixximo> eheh
[05:11:14] <grandixximo> i found it, it should be WRAPPED_ROTARY = 1
[05:11:19] <grandixximo> not just WRAPPED
[05:29:30] <grandixximo> Now axis is wrapped, only it's not taking the shortest paths, it does what it likes, in G90 +20 and -20 are the same, if i go to 0 then 90 then 0 again, it will do a total spin plus 1/4 instead of coming back 1/4
[05:29:53] <grandixximo> 0 to 90 goes 1/4
[05:30:09] <grandixximo> 90 to 0 5/4
[05:30:20] <grandixximo> strange behavior if you ask me
[05:30:34] <grandixximo> if i have time i'll report a bug see what the developers think
[05:30:45] <grandixximo> afk
[05:40:00] <archivist> I do not use wrapped because of its documented behaviour
[05:40:38] <archivist> I would rather calculate and tell the axis where to go
[05:44:30] <grandixximo> actually i understand now
[05:44:45] <grandixximo> if i go from 90 to -0 then i get 1/4
[05:45:41] <grandixximo> because the sign decides the direction you are going, it's actually pretty neat, i'll have to see for a postprocessor just as neat
[05:45:58] <grandixximo> thank you archivist always pointing in the right direction
[05:46:07] <grandixximo> giong for supper have a nice day
[05:46:09] <grandixximo> bye
[09:43:23] <IchGuckLive> hi all !
[09:43:31] <IchGuckLive> B)
[09:49:45] <IchGuckLive> chopper791: ?
[10:01:34] <JT-Shop> see you guys Sunday!
[10:06:58] <Tom_itx> where ya off to JT-Shop?
[10:07:22] <Tom_itx> another smokey mountain adventure?
[10:17:59] <IchGuckLive> take a weppen with you to get the dinner meal meat !
[10:18:43] <IchGuckLive> or hilbilly style first overdrive animal is taken to the kitchen O.O
[10:18:47] <JT-Shop> Tom_itx: St. Louis for a short git away
[10:21:56] <IchGuckLive> off for BBQ
[12:08:17] <IchGuckLive> hi all
[12:19:56] <IchGuckLive> not many talking today
[12:25:53] <DJ9DJ> hi there
[12:27:59] <IchGuckLive> B)
[12:28:13] <IchGuckLive> not many action today
[12:59:00] <IchGuckLive> have a nice suturday evening
[13:34:13] <skroon> hi
[13:38:02] <micges> hi
[13:47:08] <skroon> when I buy a manual mill, besides servo's what else do I need to buy in order to drive it via LinuxCNC? e.g. where do I connect the servo's to?
[13:51:46] <pcw_home> If you can (and you have room) you would be better off starting with a CNC mill perhaps with a bad control
[13:52:31] <skroon> pcw_home: what exactly does that mean? Don't most CNC mills already have software for it?
[13:53:25] <pcw_home> There is a lot of older CNC hardware around (mostly big) that has old, bad or flaky electronics
[13:55:37] <pcw_home> but good or at least repairable mechanics (note: these are not desktop machines)
[13:59:11] <skroon> pcw_home: right so buy a old CNC machine, and just hookup a pcb to it with linuxcnc on it?
[13:59:39] <pcw_home> well something like that
[14:00:46] <skroon> is only means that I need to be sure that the CNC machine I buy support a parallel port and understand gcode already?
[14:01:16] <skroon> or doesn't it need to understand gcode, and does linuxcnc directly drive the motors?
[14:02:53] <pcw_home> Most older machines will use analog servos so LinuxCNC needs some interface hardware for those
[14:03:06] <FinboySlick> skroon: A machine such as what pcw_home refers to will not likely use a parallel port. It will have its own set of electronics. You would need to replace some of the electronics to be able to drive it with linuxCNC
[14:05:37] <pcw_home> If the machine is small enough to use step motors, you can use parallel ports
[14:05:39] <pcw_home> but note theres often an inverse relationship in mill size and used price
[14:05:41] <pcw_home> (not much hobby market for machines that wont fit in the garage)
[14:07:31] <skroon> FinboySlick: right, I would thought so already indeed, is there a list of hardware that linuxcnc prefers to talk to directly?
[14:09:18] <pcw_home> Now to see if this new spark coil fixes my weed-whacker or I spent $114. 00 for nothing...
[14:09:20] <pcw_home> bbl
[14:09:21] <FinboySlick> skroon: One of the simplest is a parallel port. Then there are Mesa cards, which are pcw_home's area of expertise. It can probably talk to a bunch of other hardware with some work too. gpio, rs232 rs485, etc.
[14:11:14] <FinboySlick> skroon: http://www.mesanet.com/fpgacardinfo.html are popular computer-to-machine interfaces.
[14:11:40] <FinboySlick> They perform a lot better than a parallel port and linuxcnc knows how to talk to them pretty much out of the box.
[14:12:44] <skroon> cool
[14:12:45] <skroon> thanks
[14:13:27] <FinboySlick> They also make driver cards which will interface with servos or steppers.
[14:14:13] <skroon> what kind of CNC do you have?
[14:17:01] <FinboySlick> I'm just a hobbyist so bought my mill ready for parallel port and everything. I just bought a mesa card to drive it better than with parallel. http://www.syilamerica.com/machine_x5speed.php is the mill but I wouldn't recommend it to anyone unless they're able to test the accuracy prior to paying.
[14:17:08] <FinboySlick> They seem to have poor quality control.
[14:18:15] <FinboySlick> It would be okay if it was built right, but for the kind of quality you get, it's overpriced.
[14:19:23] <skroon> I always feel terrible if something like that happens :-)
[14:19:57] <FinboySlick> skroon: What size of mill are you looking for?
[14:20:37] <FinboySlick> skroon: You have to think about a lot of things when buying a mill besides the mill itself.
[14:20:50] <FinboySlick> Not least of which is where you're going to put it.
[14:21:13] <skroon> FinboySlick: I currently have a super tiny one (Roland iModela) and would like to start learning a bit more about bigger ones, not sure yet which size
[14:21:55] <skroon> the thing is the iModela is usb controlled, so unfortunately I can't use linuxcnc, but I would like my next mill to be able to drive with linuxcnc
[14:22:23] <FinboySlick> skroon: They get heavy very fast. Mine is 800 pounds and it turns out that the floor I put it on is not rigid enough. It bounces when it moves and that gives me poor precision.
[14:22:44] <skroon> FinboySlick: yikes, indeed really good point
[14:23:15] <FinboySlick> skroon: You can convert your imodela to interface with linuxcnc relatively easily.
[14:23:24] <FinboySlick> skroon: I assume that it uses steppers?
[14:23:49] <skroon> FinboySlick: really? yeah it uses steppers
[14:24:56] <FinboySlick> do you know if the drivers are separate from the usb electronics?
[14:25:12] <FinboySlick> (the motor drivers)
[14:25:43] <FinboySlick> eg: is it just one board, or do you have separate electronics for each motor?
[14:25:45] <skroon> FinboySlick: good question, I should check out the controller board cover
[14:26:17] <FinboySlick> If it has existing drivers, all you need is an interface board that would connect to a parallel port.
[14:26:21] <skroon> the wires go to this controller section, I haven't yet opened up this controller part though, but I can indeed do that
[14:26:26] <FinboySlick> Or a mesa card if you want to go fancy.
[14:26:50] <FinboySlick> Either way, you can also get stepper drivers.
[14:27:09] <skroon> and connect the stepper drivers to the motors right?
[14:27:41] <FinboySlick> Yes. You'll also need something to interface your limit switches to, and to drive the spindle.
[14:28:11] <FinboySlick> But overall, it's a lot cheaper than buying a big heavy mill and trying to convert it.
[14:28:24] <FinboySlick> It'll also give you the knowledge you need to tackle a bigger mill later on.
[14:28:30] <skroon> right, that's also something I was thinking about, I saw pretty big secondhand cnc mills
[14:29:05] <skroon> does anyone know if a CNC Punch is also controller with gcodes, and does linuxcnc also know how to drive those?
[14:29:36] <FinboySlick> skroon: linuxcnc can drive pretty much anything up to 9 axis, I think.
[14:30:15] <FinboySlick> skroon: All it does is convert g-code to movement.
[14:31:41] <FinboySlick> It can control those assembly line robot arms, hexapod mills, etc.
[14:32:04] <FinboySlick> Of course, that requires a bit more knowledge than a basic cartesian mill.
[14:32:11] <skroon> assembly line robot's those are cool things :-) always wondered how they we're controller/programmed
[14:32:29] <skroon> so they are mostly gcode as well?
[14:32:45] <skroon> instead of cartesian what instead?
[14:33:05] <FinboySlick> skroon: I think that the gcode is mostly abstracted away by the controllers nowadays.
[14:33:22] <FinboySlick> For commercial robot arms, I mean.
[14:33:34] <FinboySlick> Basically, you 'teach' them how to move.
[14:33:47] <skroon> so they record like a macro?
[14:34:55] <FinboySlick> skroon: Yes. Obviously there's quite a few ways to program them but manufacturers are trying to make it easer.
[14:36:29] <FinboySlick> A robot arm would be a good example of a non-cartesian system. LinuxCNC can compute the inverse kinematics and translate robot movements to cartesian coordinates. That way, you could put a spindle on the head of a robot arm and have it mill with gcode.
[14:37:20] <skroon> ah right "inverse kinematics" i've heard from that before, looked at wikipedia and quickly closed it before going nuts :_)
[14:37:48] <FinboySlick> http://www.youtube.com/watch?v=9olpU79H0o8
[14:37:48] <Tecan> (9olpU79H0o8) "Arm for welding robot with LinuxCNC" by "viestartss" is "Tech" - Length: 0:02:05
[14:39:20] <skroon> really cool :-)
[14:44:15] <andypugh> skroon: The page you want is: http://wiki.linuxcnc.org/cgi-bin/wiki.pl?LinuxCNC_Supported_Hardware
[14:48:45] <andypugh> The best plan is probably to find oput what motors + drives you can find cheaply, or are fitted in the machine you buy, and then shoose the interface electronics to suit.
[14:49:39] <andypugh> (For example my milling machine is using brushless motors with resolver feedback. You probably wouldn't choose that system, but the 800W servos were £50 for the set of three.)
[16:41:51] <DJ9DJ> gn8
[17:07:04] <gene78> I have that 1HP motor turning the 7x12's spindle, but the geardown I used was almost too much
[17:07:34] <gene78> low gear max is about 340 revs, high backgear is a hair over 800.
[17:08:18] <andypugh> How high can you go?
[17:09:45] <gene78> Thats it. Part of the problem is that the C41 I used, is all tapped out at about 8.8 volts. So I'm only getting about 2/3rds of the range
[17:10:22] <andypugh> 800 rpm sounds about 4x too slow for a 7" lathe.
[17:10:33] <gene78> I need to see if I can adjust its feedback as it does have 11.85 volts on the high side rail.
[17:11:13] <andypugh> Can you adjust the sensitivity of the motor drive?
[17:11:18] <gene78> It is Andy, but in practice I have never actually cut at e3ven that high a speed.
[17:12:53] <gene78> It has one pot, doesn't seem to do a lot. My 7812 driving the C41 from about 38.5 volts is smoking but the thermal says is about 160F and holding.
[17:13:01] <andypugh> You don't use CSS mode?
[17:14:05] <gene78> Whazzat?
[17:14:18] <gene78> Oh, no, notr yet.
[17:14:41] <gene78> That is on the agenda howqever.
[17:15:16] <andypugh> It makes sense to. Then you just need to set the surface speed for your material and the software does the rest.
[17:15:24] <gene78> But that can't be used while threading, as I found earlier.
[17:15:34] <andypugh> It should be a route to better surface finishes.
[17:15:57] <gene78> I'm sure.
[17:16:11] <andypugh> I am pretty sure that my threading routine uses CSS.
[17:16:39] <gene78> I've also replace the 4" 4 jaw with a 5", and at 800 revs its spinning cosmolene all over the place. :)
[17:17:14] <andypugh> I love that smell, it smells of new toys.
[17:17:37] <gene78> If the spindle speed is changed, the root of the thread sifts sideways Andy, due to the way it locks up after finding the index.
[17:18:03] <gene78> He he, but not on my glasses :(
[17:18:46] <andypugh> Have you seen my lathe GUI? A set of pre-programmed macros. You just adjust the numbers and press "Go". http://www.linuxcnc.org/index.php/english/forum/41-guis/26550-lathe-macros
[17:21:13] <gene78> now I have, sweet!
[17:22:41] <gene78> humm, no brakes yet, 10 seconds to a dead stop, not bad considering the weight swinging.
[17:23:08] <andypugh> A few little niggles to work out, and I ought to add toolpost-drilling and parting tabs.
[17:23:38] <andypugh> But it should work for anyone, except that the threading screen makes metric assumptions.
[17:23:54] <pcw_home> Thats interesting that the thread start point changes, you would think the plunge speed would be keyed to spindle speed
[17:24:15] <andypugh> (trivial G-code changes required to work in TPI and suitable thread-rounding)
[17:24:34] <andypugh> I haven't seen that effect, I have to say.
[17:26:12] <gene78> i don't think it noticeable for a 10% change while cutting a thread
[17:28:00] <gene78> but it you start slow to see if the limits are about right, then crank it up 10x to finish the job. it will be pretty well trashed, shifting the start point a major fraction of a turn
[17:29:57] <gene78> Its about ~30~ time here, but I still need tp put my toolbox back together, is decided to be self locking with the weight of the 7x12 sitting on it :(((
[17:35:10] <pcw_home> I bought a $4.99 Harbor Freight DVM thinking its got to be good for something
[17:35:12] <pcw_home> I was wrong
[17:36:56] <cpresser> define 'good for something'
[17:37:13] <pcw_home> measuring voltage maybe
[17:37:30] <cpresser> it doesnt even do that?
[17:37:39] <pcw_home> actually it does measure voltage maybe
[17:38:37] <pcw_home> sometimes its right and sometimes it reads ~1/10
[17:38:39] <pcw_home> and sometimes 0
[17:38:50] <cpresser> i wouldn count on it to be accurate, but it will tell you if there is voltage or not :=
[17:39:10] <pcw_home> not this one
[17:54:56] <andypugh> My DVM lies when the battery is flat. To be fair, it does warn when the battery is flat.
[18:53:19] <andypugh> Has this giy actually edited his machining video to suit the soundtrack? http://www.youtube.com/watch?v=O1CFIAPPB2o
[18:53:20] <Tecan> (O1CFIAPPB2o) "Milling gear rack" by "OriginMultislayer" is "Tech" - Length: 0:03:57
[18:53:47] <andypugh> archivist: Any idea how racks are cut commercially?
[19:28:35] <Tom_itx> andypugh i wonder what angle that cutter is
[19:28:48] <Tom_itx> looks similar to one i've got for cutting casting relief
[19:29:10] <andypugh> I would guess 20 or 14.5 degrees.
[19:29:26] <andypugh> Casting relief is 1.5 or so?
[19:29:35] <Tom_itx> i don't recall what mine is but i think it's 7 or 15
[19:29:48] <pcw_home> Dont racks need an involute curve?
[19:30:22] <pcw_home> or is that all done in the spur gear
[19:30:23] <Tom_itx> depends on what kind of rack
[19:30:35] <Tom_itx> if a flat gear rides on it, no
[19:31:47] <andypugh> Involute racks are flat-sided.
[19:32:18] <andypugh> In fact, it is the involute rack that defines the involute curve of the gear.
[19:32:42] <andypugh> Involute gear hobs are flat-sided too.
[19:34:13] <andypugh> Cycloidal racks have curved flanks.
[23:50:27] <ssi> SWEET
[23:50:30] <ssi> managed to get a program on the fpga