#emc | Logs for 2009-04-22

[00:33:39] <ds3> does this sound doable for EMC power tapping - modified battery powered drill to have a moderate resolution encoder (10-16 pulls per rev.) replace the chuck with some kind of tap holder. replace the reverse switch with a relay controlled by EMC?
[00:39:09] <jepler> ds3: 16 pulses per revolution is a lot less than anyone I know of doing rigid tapping
[00:42:57] <cradek> consider running a 3/8-16 tap. each turn of the tap requires a Z move of .0625. so at 16 pulses per revolution, each encoder pulse represents a Z move of .004 inches. when reversing it seems like the tap would end up being pulled out of or pushed into the thread at least .004 inches. this might or might not break the tap, but it doesn't seem like it would give a good thread.
[00:46:01] <cradek> would this be a special purpose machine just for tapping? I don't understand why you aren't using the machine spindle.
[00:52:10] <ds3> well, it is simple... I don't have a speed controller I can interface to easily
[00:52:27] <ds3> and I am not building a 1/3hp speed controller... but a drillis easily enough to deal with
[00:53:09] <ds3> the machine I want it on won't do well with a 3/8-16 tap anyways.... I think the biggest (if I am lucky) is a 1/4-20... I want it more for the stuff like 2-56 (with a forming tap)
[00:53:31] <ds3> jepler: how many PPR have people done it with?
[00:55:50] <ds3> (important detail - I have a ez430rf kit already... no AVRs on hand :D)
[00:58:54] <ds3> bah
[01:00:30] <skunkworks> I know someone using a 360ppr encoder http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Pjm
[01:02:53] <ds3> hmmm 360 hmm
[01:03:18] <ds3> that's 360 lines/marks I need to put on a disk
[01:03:34] <fenn> laser printer
[01:03:55] <fenn> * fenn curses his lack of a webserver
[01:04:32] <ds3> are they accurate enough? and how large of a disk are you recommending? (I was thinking of ~1inch disk and my printer is 300dpi)
[01:04:52] <fenn> kinkos has like 2000 dpi printers
[01:05:35] <fenn> 1 inch sounds pretty small
[01:05:36] <ds3> real 2000dpi or the enhancement technology 2000dpi?
[01:05:45] <fenn> i dont know
[01:06:09] <ds3> I am suspicious of any printers > 600dpi
[01:06:20] <fenn> i'm suspicious of any printers :)
[01:06:33] <ds3> =)
[01:29:51] <Martyn> Who else here uses a tormach?
[01:31:27] <Martyn> I'm seeing an issue that is different between using Mach and using EMC2... spindle RPM is off (slow) by about 10% when running emc2
[01:42:46] <fenn_> fenn_ is now known as fenn
[01:44:01] <skunkworks> Martyn: are you measuring rpm?
[01:44:31] <JymmmEMC> Crap, one of my routers went out
[01:46:26] <skunkworks> My guess.... Is you may need to tweek 'setp stepgen.4.position-scale 2.5' in your hal file...
[01:51:45] <dareposte> hi jmkasunich
[01:52:08] <dareposte> and hi all
[01:54:11] <geo01005_home> howdy.
[01:55:10] <dareposte> what did people do before e-bay??
[01:55:19] <dareposte> and machine tool auctions
[01:56:36] <Martyn> skunkworks : I hooked an optical encoder onto the spindle, and am using my fluke as a tachometer
[01:57:05] <Martyn> skunkworks : When I use Mach, it's accurate .. emc2 it's off by a consistent 10% or so
[01:57:07] <Martyn> which is confusing
[01:57:52] <Martyn> skunkworks : I think I may need to put a permanent encoder (maybe a magnetic one) onto the spindle, and change the way emc2 is reading it
[01:58:06] <Martyn> because cutting too slow is bad.
[01:58:38] <dareposte> for some reason i think this 9x20 lathe is a little more capable than my 9x20 lathe. http://cgi.ebay.com/CNC-Lathe-Okuma-LB-15-OSP-Chip-Tooling-Price-Reduced_W0QQitemZ330287644906QQcmdZViewItemQQptZBI_Lathes?hash=item330287644906&_trksid=p4011.c0.m14&_trkparms=72%3A1205%7C66%3A2%7C65%3A12%7C39%3A1%7C240%3A1318%7C301%3A1%7C293%3A1%7C294%3A50
[02:05:30] <Martyn> nice looking unit
[02:07:53] <dareposte> i'd like to have it
[02:08:01] <dareposte> it weighs 30x more than mine, but only costs 15x as much
[02:08:04] <dareposte> what a steal
[02:08:58] <dareposte> this one might suit me better though... http://cgi.ebay.com/MORI-SEIKI-SL-1_W0QQitemZ110380059838QQcmdZViewItemQQptZBI_Lathes?hash=item110380059838&_trksid=p4011.c0.m14&_trkparms=72%3A1205%7C66%3A2%7C65%3A12%7C39%3A1%7C240%3A1318%7C301%3A0%7C293%3A1%7C294%3A50
[02:14:27] <dareposte> * dareposte drools on the mori seiki
[02:17:49] <Martyn> That mori is in pretty piss poor condition
[02:17:56] <Martyn> $5k is not a fair price.
[02:19:12] <Martyn> Too bad you can't offer $3k :)
[02:19:26] <Martyn> oh, wait, you CAN!
[02:19:32] <Martyn> I just saw the "make offer" button
[02:27:54] <dareposte> yeah
[02:28:03] <dareposte> i'd offer 2800
[02:28:26] <dareposte> but i need to see more pics to even bother with the drive to inspect it in person
[02:30:55] <dareposte> too much sheet metal in the way to see anything useful.
[02:31:29] <Martyn> yep
[02:31:37] <Martyn> you should see the tormach I picked up
[02:31:51] <Martyn> complete, even have the 4th axis rotary ...
[02:31:57] <Martyn> less than $10k
[02:38:39] <dareposte> sweet
[02:38:55] <dareposte> i've been watching for something nice and solid for a few months now
[02:39:00] <dareposte> come real close a few times
[02:39:12] <dareposte> tormach is a fairly small machine right?
[02:40:25] <Martyn> yep
[02:40:28] <dareposte> i've kind of been leaning more and more towards one of those cnc swiss machines... so fast for production
[02:40:43] <Martyn> I repair the larger machines for a living. I'm on a 9 month hiatus
[02:41:13] <Martyn> so I bought the tormach, and I'm determined to learn Linux, use emc, and get a better foundation on open hardware and systems
[02:41:28] <dareposte> good plan
[02:41:38] <dareposte> i've seen some videos it looks like they are fairly solid for the size
[02:42:02] <Martyn> it's the only 'hobby' grade machine than can machine steel
[02:42:02] <dareposte> i almost got a manual mill that size, but wound up with a series 1 clone i'm converting
[02:42:12] <Martyn> in that way, it's as good as a bridgeport
[02:42:23] <dareposte> i've heard the zx-45 machines are solid for steel
[02:42:24] <Martyn> the tool is powerful, and you can do amazing things at low rpms
[02:43:04] <Martyn> duh
[02:43:13] <Martyn> the cx45 is a proper mill, geared drive
[02:43:22] <dareposte> isn't it a hobby class?
[02:43:37] <dareposte> the x2 is a gear drive too u know ;-)
[02:43:38] <Martyn> Sure, but it's not in the same class as the tormach when it comes to automated CNC
[02:44:08] <dareposte> yeah they are only manual, r8 collets i think
[02:44:30] <Martyn> so you have to servo it up, and then go through the whole process of calibrating emc2 to do anything with it
[02:44:37] <Martyn> as I found out, the tormach is practically plug and play
[02:44:56] <Martyn> I can flip between Mach and emc2 just by rebooting
[02:45:59] <dareposte> does the tormach come with ballscrews already in it? or servos and leads
[02:46:27] <Martyn> what do you think?
[02:46:28] <Martyn> *grin(
[02:46:37] <Martyn> It's hobby grade, but used as an accesory on pro lines.
[02:46:44] <dareposte> so screws then
[02:47:40] <Martyn> yep. That was a major decision on my dropping enough to buy a small car :)
[02:47:47] <dareposte> yeah that's where i'm at
[02:47:58] <dareposte> i really want a new motorcycle... or a used cnc turning center
[02:48:47] <dareposte> the price is a draw
[02:49:17] <dareposte> enjoyment is probably a draw...
[02:49:37] <dareposte> used cnc wins on depreciation i think..
[02:50:49] <dareposte> the tormach has a fancy repeatable tool holder doesnt it?
[02:51:01] <jmkasunich> repeatable yes
[02:51:02] <dareposte> i've seen references to the tooling it uses but never paid that much attention
[02:51:07] <jmkasunich> not particularly fancy
[02:51:24] <dareposte> repeatable is better than fancy i guess
[02:52:06] <jmkasunich> http://www.tormach.com/Flash/TTS_small.gif
[02:52:31] <jmkasunich> a 3/4" collet that lives in the spindle all the time, and a shoulder on the toolholders
[02:53:38] <dareposte> pretty fancy
[02:53:51] <dareposte> is it a special collet too?
[02:53:59] <jmkasunich> slightly
[02:54:09] <jmkasunich> the collet nose is a bit shorter than normal
[02:54:33] <jmkasunich> you don't want the tool to touch the nose of the collet, it should contact the nose of the spindle itself
[02:54:40] <dareposte> it looks like the tool holders wouldn't be all that difficult to make on a lathe...
[02:54:51] <jmkasunich> nope
[02:55:12] <jmkasunich> but simple endmill holders are pretty cheap from tormach, almost not worth the trouble
[02:55:36] <dareposte> probably ground from them too
[02:55:37] <jmkasunich> $23.50
[02:55:40] <dareposte> maybe hardened
[02:55:41] <jmkasunich> yep
[02:56:01] <jmkasunich> they want $92 for an ER-20 collet chuck
[02:56:20] <dareposte> wouldn't that sort of defeat the purpose of the special system?
[02:56:22] <jmkasunich> I got two $50 ones from Mari-Tool (see ebay, or google), and made the collars
[02:56:44] <jmkasunich> you put the tool in the collet chuck when you set up the job
[02:56:52] <dareposte> oh
[02:56:58] <Martyn> dareposte : It does indeed
[02:57:03] <Martyn> dareposte : But it's just a tool holder
[02:57:23] <dareposte> i can see if you were running a batch of parts and just needed one er chuck'ed tool it would be nice to have
[02:57:44] <jmkasunich> my TTS tooling: http://willepadnos.net/jmkasunich/mill-tooling-2012.jpg
[02:57:54] <Martyn> dareposte : I am going to work on a quick change holder once I get some parts in
[02:57:58] <jmkasunich> I've since added one of their 1.5" face mills
[02:58:05] <Martyn> I'm just going to modify the head from a Mitsubishi mill
[02:58:35] <dareposte> Martyn: did you consider bigger vmc's? you can get some full sized ones for pretty low cost these days, used of course
[02:59:07] <dareposte> space is an issue
[02:59:11] <Martyn> dareposte : I did, but since the point of this hiatus is to get a solid education on Linux and I wanted a mill I could get up and running quickly, I chose three mills
[02:59:15] <Martyn> I have a flashcut CNC
[02:59:18] <Martyn> The Tormach
[02:59:24] <Martyn> and a McWire reprap
[02:59:43] <Martyn> all three fit nicely in the garage
[02:59:50] <dareposte> so the goal is mostly educational then?
[02:59:52] <JymmmEMC> OK, router repaired and back up
[02:59:56] <Martyn> purely.
[03:00:07] <dareposte> makes sense then
[03:00:17] <Martyn> The faster I learn more python, linux, and such .. the faster I can work on an idea
[03:00:19] <dareposte> my goal is always to make boatloads of parts with as little work as possible
[03:00:37] <JymmmEMC> dareposte: isn't that called outsourcing
[03:00:42] <dareposte> lol
[03:00:44] <Martyn> (A friend of mine has been thinking of starting a company in the same vein as the shopbot)
[03:01:14] <Martyn> but instead of milling wood, he wants to build a linux-based CNC mill, and CNC lathe
[03:01:37] <Martyn> rapid prototyping has become more important than ever recently.
[03:01:56] <dareposte> hm well i got the cnc lathe part already, it's working out pretty well except for changing the *#(%@ tools every 5 seconds
[03:03:03] <JymmmEMC> dareposte: dude, outsource it = less work =)
[03:03:24] <dareposte> then i don't get the satisfaction of sitting back and watching it run
[03:03:40] <JymmmEMC> dareposte: Tell em to setup a IPCamera
[03:03:54] <dareposte> it's just not the same when u don't set it up yourself
[03:03:58] <dareposte> i'm sure you know
[03:04:11] <JymmmEMC> yeah, i know =)
[03:04:23] <dareposte> i'd love to have one of those screw machines that spits out a part every 2 seconds
[03:04:27] <Martyn> dareposte : how long does it take you to change the tool?
[03:04:36] <Martyn> dareposte : I can't imagine it's -that- much time
[03:04:43] <JymmmEMC> 2s
[03:04:45] <dareposte> Martyn: no it's not that long, but its annoying
[03:04:57] <Martyn> True, it's kind of hard to batch things in a lathe
[03:05:01] <Martyn> that's the magic of the mill.
[03:05:14] <Martyn> and why I bothered to get the 4th axis rotary
[03:05:25] <dareposte> I've been working on some gang tooling, but having a hard time figuring out a good strategy for it
[03:07:03] <dareposte> i can make most of what i'd need with an OD/facing + spot drill + drill + tap + chamfer
[03:07:30] <dareposte> but fitting 5 tools in there is a challenge, especially when the drills are so long
[03:07:44] <JymmmEMC> end drilling?
[03:07:48] <dareposte> yeah
[03:07:48] <cradek> I always use screw machine length drills
[03:08:06] <JymmmEMC> dareposte: gatlin gun
[03:08:18] <dareposte> cradek: those are better but they still sort of are longer than the turning and facing tools tend to be
[03:08:20] <cradek> if you get the split point 135s sometimes a spot drill isn't even needed if the drill is centered correctly
[03:08:31] <cradek> yeah, true
[03:08:57] <dareposte> i've seen one set up that used a boring bar for the od and facing tool
[03:09:40] <dareposte> and if the drill was mounted in the back t-slot and the boring bar in the front one, then it might fit with a short drill
[03:09:55] <cradek> I've done that - but be careful if using a screw-on chuck
[03:10:00] <dareposte> but still then the parting tool can't fit on the back side all that well
[03:10:19] <dareposte> my chuck is thread-on, but it has a clamping screw that is alarmingly small
[03:11:10] <ds3> for short parts, why not start with a mill and mount a chuck onto the spindle
[03:11:20] <ds3> then you can clamp all sorts of tools to the table
[03:11:35] <dareposte> sort of screws up the bar-feeder
[03:12:13] <ds3> you can still feed vertically
[03:12:34] <dareposte> in a mill?
[03:12:37] <ds3> equip the chuck with a controlled release
[03:12:40] <ds3> sure
[03:12:53] <ds3> line it over a suitable block and let gravity feed
[03:13:06] <dareposte> i don't think the spindle has a hole all the way through it
[03:13:06] <jmkasunich> most mills don't have a very big hole thru the spindle
[03:13:14] <jmkasunich> and there's usually a drawbar blocking it
[03:13:36] <dareposte> i usually get a 4' section of bar and then the last tool change is the parting tool, and just feed the bar till it touches and chuck it up and let it go again
[03:13:46] <ds3> get rid of the drawbar
[03:14:01] <ds3> this depends on the diameter of the barstock you want to deal with
[03:14:03] <dareposte> my mill has like a 1/2" hole where the drawbar goes maybe
[03:14:53] <dareposte> novel idea though if you are using part blanks it might work out pretty well
[03:16:02] <dareposte> i've seen some of those mini chucks that can be chucked in an R8 collet and hold up to a couple inches OD
[03:18:07] <ds3> what size barstock?
[03:18:14] <dareposte> 1"
[03:18:28] <dareposte> 3/4" usually actually
[03:18:54] <dareposte> sometimes some 5/8 hex or 1/2"
[03:19:10] <dareposte> mostly that's the machine's limitation though, i have a dinky one at the moment
[03:19:41] <dareposte> i'd like to be able to do parts up to 1.5" finished od on the one i'm hunting for
[03:20:40] <ds3> I think those import mills might pass 3/4"
[03:21:07] <ds3> Hmmm sounds expensive
[03:21:17] <dareposte> yeah it will be a bit pricey
[03:22:27] <dareposte> i keep wanting to get an old 16x40 and convert it, but the lack of a tool changer keeps killing those plans
[03:24:55] <dareposte> by the time i price out a tool changer to add to it, then it's already well into the price range of a normal turning center
[03:27:51] <dareposte> oh well
[04:21:41] <tomp> i wish theyd show some od relief of the drawbar-collet http://www.tormach.com/Flash/TTS_small.gif
[04:23:51] <tomp> actually they show quite a bit of interference (outside end edge of drawtube to end groove on toolholder shoulder )
[04:24:10] <jmkasunich> that's a drawing thing
[04:24:37] <tomp> an explain thing
[04:24:49] <jmkasunich> the collet nose is definitely smaller than the ID of the raised ring on the toolholder
[04:25:00] <jmkasunich> the OD of the toolholder is 1.5"
[04:25:16] <jmkasunich> guessing the ID is 1.3 ish
[04:25:22] <jmkasunich> R8 collet nose is 1.25
[04:25:30] <tomp> cool, i figgered it really would work, i just had to ignore what i saw
[04:25:39] <jmkasunich> if I didn't have a cat on my lap I'd measure the ID ;-)
[04:26:45] <tomp> so you got ATC on that shoptask? or a powe draw bar for now
[04:27:56] <jmkasunich> neither
[04:28:06] <jmkasunich> wrench on drawbar, hands on toolholder
[04:28:19] <jmkasunich> I can do about 30 seconds chip-to-chip when I'm on a roll
[04:29:56] <tomp> cool, nice holders
[04:30:07] <cradek> jmkasunich: those sure are good prices on angle plates, thanks
[04:30:23] <jmkasunich> you're welcome
[04:32:24] <cradek> now I play the game called "what else will I remember I need from enco right after placing the order"
[04:32:36] <jmkasunich> I hate that game
[04:33:04] <ds3> what's the current min. order for enco to qualify for free shipping?
[04:33:04] <cradek> free shipping alleviates it somewhat
[04:33:06] <toastyde1th> "the right shipping address"
[04:33:10] <jmkasunich> $25
[04:33:28] <ds3> Hmmm I am 1/4 of the way tehre with replacement drill bits
[04:33:45] <jmkasunich> cradek: unless first order was $26, and the item you forgot was $2
[04:33:58] <cradek> yeah
[04:34:34] <cradek> jmkasunich: that drill chuck I got is a current model, so I can get the real jacobs key for $4 (none of mine seem to fit right)
[04:34:52] <cradek> it was a good deal: new price is > $100
[04:35:11] <jmkasunich> what drill chuck?
[04:35:26] <cradek> oh maybe you didn't see it - a jacobs ball bearing chuck
[04:35:32] <jmkasunich> ah, nice
[04:35:41] <cradek> seems in perfect shape, but a little gummy
[04:36:16] <cradek> for the lathe - to replace a $5 chinese one I have on there now that doesn't really seem to grip anything
[04:36:48] <cradek> can't use an albrecht because I want to be able to run taps in it
[04:38:12] <jmkasunich> what size chuck?
[04:38:25] <jmkasunich> and does it already have the right arbor?
[04:38:30] <cradek> small to 3/8, JT2
[04:38:37] <cradek> no, but I had one
[04:38:49] <cradek> well, I had the makings of one
[04:39:16] <jmkasunich> what does the HNC want? straight shank?
[04:40:03] <cradek> yeah the boring tool holder is 5/8 so I cut off a 5/8 straight shank arbor and grind flats for the set screws
[04:40:37] <cradek> the two-bolt boring bar holder has a lip that sits over the edge of the turret so it's very strong that way
[04:42:07] <jmkasunich> dang
[04:42:28] <jmkasunich> I wish there was Free FEA software that you don't have to be an FEA expert to use
[04:42:55] <cradek> I doubt there's any, free or nonfree
[04:43:01] <jmkasunich> yeah
[04:45:34] <jmkasunich> I'm trying to calculate the deflection of an T, given that the ends of the crossarm are supported, and the load is applied sideways to the bottom of the upright
[04:45:46] <jmkasunich> upright and crossbar are both hollow rectangles
[04:47:24] <cradek> I don't have the slightest idea how to do that, but I know exactly who I'd ask (coworker)
[04:47:46] <jmkasunich> oh, thats right - you work for a company that specializes in that kind of thing
[04:48:02] <jmkasunich> I found this: http://www.aps.anl.gov/APS_Engineering_Support_Division/Mechanical_Operations_and_Maintenance/Calculators/ElasticBeam2.html
[04:48:19] <jmkasunich> break the problem into two parts
[04:48:31] <jmkasunich> the upright, treated as a cantilever loaded at the end
[04:48:43] <jmkasunich> and the crossarm, treated as supported at both ends with a moment in the middle
[04:49:09] <cradek> don't you nearly have case 64?
[04:49:12] <jmkasunich> but the calculator doesn't compute the angle at the middle in the latter case, only the angles at the ends
[04:49:28] <jmkasunich> yes, case 64 is the crossarm calc
[04:49:57] <cradek> oh linear deflection of the base of the T is what you want?
[04:50:24] <cradek> yeah I can see how it's "just" those two problems
[04:50:33] <jmkasunich> but I want to know the angle of deflection at the midpoint (because that angle times the length of the upright gives deflection at the end of the upright)
[04:50:57] <jmkasunich> actually its three, because the shape isn't really a T
[04:51:11] <jmkasunich> the crossarm is on one plane, and the upright is on another plane
[04:51:35] <jmkasunich> so the (short) segment that connects them is loaded in torsion by the moment, and contributes its own deflection
[04:51:45] <jmkasunich> that is a fairly straightforward calc tho
[04:53:36] <jmkasunich> I'm sure the sum of the three simplified calculations is not the right answer, but I hope its close
[04:54:05] <jmkasunich> FEA would give a much better answer though (including other deflections that I'm probably overlooking, such as out-of-plane ones)
[04:55:33] <tomp> http://www.caelinux.com/CMS/ open src linux based live cd with fea
[04:56:31] <jmkasunich> tomp: I've looked at Salome and Code Aster
[04:56:35] <jmkasunich> my head started spinning
[04:56:45] <jmkasunich> there is a really steep learning curve
[04:57:02] <jmkasunich> can't just say "here is my shape, here is my maternal, here are my forces, show me the deflection"
[04:58:03] <cradek> I see there are some tutorials - but yeah I bet you'd have to spend a couple full days on it
[04:58:28] <tomp> alternative: build a thing, and push on it ;)
[04:58:41] <cradek> heh
[04:59:33] <jmkasunich> the calcs are to tell me how big/heavy to build the thing
[05:01:05] <cradek> hm, it's going to feel unusual to go to work tomorrow. I wonder what awaits me.
[05:01:24] <cradek> also, maybe I should go to bed. goodnight folks.
[05:01:36] <jmkasunich> that sounds like a good plan
[05:05:39] <tomp> gnite
[06:18:20] <MrSunshine> the lower the inductance in a stepper the faster i can move it right =?
[06:18:45] <MrSunshine> so parallel would be better then serial connection for a unipolar motor to run it in bipolar mode?
[07:07:10] <pjm__> good morning
[09:06:12] <piasdom> g'mornin all
[09:16:04] <alex_joni> 'lo
[10:02:54] <tomp> g'nite :)
[12:30:45] <Valen> so what do people think is better
[12:30:57] <Valen> encoders on motors that are direct drive to ballscrews
[12:31:20] <Valen> or linear scale measurement doohickys that are accurate to 0.001mm
[12:31:28] <Valen> DRO type things
[12:31:45] <pjm__> Valen i spose it depends on what backlash u have in the ballnuts
[12:31:58] <Valen> 0 backlash ballnuts we will be getting
[12:32:06] <pjm__> certainly i think a lot of CNC's use encoders directly on the back of the motors
[12:32:21] <Valen> they do
[12:32:38] <pjm__> so thats probably a tried and tested method
[12:32:43] <Valen> what i'm thinking is "low" accuracy ballscrews are waaaay cheaper than high accuracy ones
[12:33:27] <Valen> and if i have the scale on the mill bed itself then that "low" accuracy matters very little
[12:33:58] <pjm__> wont the table wobble around when u take a cut, if there is slop in the ballnut assy's ?
[12:36:11] <Valen> no slop they just arent as accurate in terms of "turns per inch"
[12:36:23] <Valen> they are out by a few 0.00somethings of a mm
[12:36:33] <Valen> over 300mm or so
[12:36:46] <archivist> if you use scales then you need the ones with quadrature out so you know direction etc
[12:37:21] <Valen> yeah i was presuming that that would be the typical sort of scale a DRO would have
[12:43:10] <Valen> think it might be a cheap way of getting better accuracy?
[12:43:28] <Valen> oh yeah has anybody done a voice coil mill before? but a big one
[12:44:08] <Valen> well say 500 mm travel
[12:51:57] <MrSunshine> voice coil ?
[12:53:28] <MrSunshine> how high accuracy can you get with IR or something for distance measurement ?
[12:54:16] <archivist> inferometer angstroms
[12:57:10] <MrSunshine> 24bit ADC and a resistance wire that goes from the motor to a thingie on the table, so when it moves it wil englongate the wire, making the resitance greater .. voila :)
[12:57:28] <MrSunshine> then just read the value and you have a stepper with feedback ;)
[13:05:36] <cradek> Valen: both
[13:07:31] <cradek> Valen: http://jmkasunich.dyndns.org/cgi-bin/blosxom/shoptask/wichita-trip-02-20-08.html
[13:10:26] <jepler> yay, less than a month to fest in wichita
[13:10:32] <cradek> yay
[13:10:52] <pjm__> we need a cnc fest in the UK
[13:10:57] <archivist> yup
[13:11:16] <jepler> that's only allowed if you find a way to pay me to come
[13:11:46] <archivist> only if you pay me to come to wichita
[13:12:50] <archivist> pjm__, at midlands model engineer ex?
[13:13:12] <pjm__> it would be a good idea certainly
[13:14:10] <archivist> I was told there were some spare slots in the conference room /me not followed up though
[13:17:09] <Valen> cradek both?
[13:17:37] <cradek> Valen: see link
[13:17:44] <Valen> both what though?
[13:17:56] <cradek> haha, sorry
[13:18:06] <cradek> both encoder and scale
[13:18:27] <Valen> I didn't think EMC knew how to integrate the two?
[13:19:06] <MrSunshine> its bumbimagic!
[13:20:17] <Valen> was that you cradek?
[13:20:26] <Valen> any idea why it oscilated?
[13:20:53] <cradek> backlash in the screw
[13:21:03] <Valen> ahh yeah that'll kill it
[13:21:16] <Valen> but if you had 0 backlash could you see it working?
[13:21:57] <_justa_> 0 backlash ? unpossible!
[13:22:15] <Valen> 's called preload ;-P
[13:22:49] <Valen> where 0 means "really really rather small"
[13:22:52] <cradek> yeah I also question whether such a system exists
[13:23:02] <cradek> yes I've seen "really rather small" but not zero
[13:23:21] <cradek> you still have other effects, such as sticktion, that work against you just like backlash
[13:23:33] <Valen> where is the backlash introduced in a spring preload system?
[13:23:43] <Valen> yeah, but they should be really second order though
[13:23:53] <_justa_> 'when your encoder/etc precision is lower than the backlash amplitude' is really the situation you're describing
[13:24:18] <Valen> that's basically what i was talking about yes
[13:24:53] <Valen> IE you are probably going to see more "backlash" in the deflection of the table rather than actual movement
[13:24:54] <_justa_> Welll.. throw away LSB's in the datastream from the encoders until the oscillation stops ? ;)
[13:25:14] <Valen> problem is the motor starts spinning up then overshoots
[13:26:03] <Valen> oscilation would be defined by a relationship between motor power, inertia and backlash before reaching encoder movement
[13:26:49] <Valen> If the motor can build up enough momentum to cause a correction that drives the system back into the backlash then it will oscilate
[13:27:12] <Valen> So on to dreamland
[13:27:20] <Valen> I was thinking about a voice coil based mill
[13:27:41] <Valen> use some smegging large soilenoids to drive the table around
[13:28:06] <_justa_> True; but when it reaches the encoder, the encoder starts compensating. Ofcourse, fixing the overshoot/backlash in the first place is the right solution. But , if you've fixed everything you CAN fix on that side, decreasing encoder sensitivity (or integrate a smart dampening algorithm) might be as much as you can hope for.
[13:28:38] <Valen> decreasing the sensitivity wont actually help though
[13:30:50] <archivist> stability in feedback systems is well documented in control theory books
[13:31:09] <Valen> true but we don't know it so we can talk about it at length ;->
[13:31:28] <Valen> I still like my solenoid based milling machine
[13:31:32] <Valen> thats 0 backlash
[13:31:47] <Valen> and pretty good in terms of high speed traverse ;->
[13:31:53] <archivist> you still have mass ans resonance
[13:32:05] <Valen> didn't say it was perfect
[13:32:32] <Valen> the other problem it has is the inductance of the coil gives it "inertia" in force rather than velocity
[13:32:42] <Valen> so a sudden unloading will cause a "spring" effect
[13:33:43] <Valen> still it'd give a nice smooth cut ;->
[13:35:18] <Valen> hrm using scales opens up the posibility of using hydraulics for the servos, probably not as accurate though :-<
[13:57:12] <cradek> hydraulic servos used to be state of the art.
[13:57:42] <skunkworks> the worked pretty well
[13:57:48] <eric_unterhausen> hydraulics stink unless you need a lot of force
[13:58:12] <eric_unterhausen> if I have to touch hydraulic oil again, it will be too soon
[13:58:15] <cradek> yeah I bet there are reasons we have electric servos now. I have never worked on a hydraulic machine.
[13:58:54] <skunkworks> http://www.electronicsam.com/images/KandT/conversion/hyservo.JPG
[13:59:11] <archivist> there is a drybolic downstairs
[14:00:19] <archivist> http://www.archivist.info/cnc/works2008/P1010203.JPG
[14:00:34] <skunkworks> Those stalled at 75ft-lbs.
[14:00:37] <skunkworks> scary
[14:06:53] <_justa_> nice tub with oil beneath, there..
[14:26:59] <JymmmEMC> alex_joni: ping
[15:10:21] <Valen1> Valen1 is now known as Valen
[16:22:09] <roh> hi there
[16:22:49] <BJT-Work> hi
[16:25:09] <roh> i've got a syil X5/BF20 cnc mill with their controller (called superbox or sbox3) here and cannot get it to 'move'
[16:26:10] <cradek> any data about the controller online that you can point us to?
[16:26:15] <roh> we have set up 8.04 with rtai kernel and got the emergency stop working on the parport and we think we found the pinout, we just cannot get a yellow 'link' led to light up.
[16:26:26] <roh> cradek sadly no.. if i had i propably wouldnt ask
[16:26:53] <roh> i just hoped somebody knows their stuff already and just the config never found its way into the repo or so
[16:27:25] <roh> http://syil.com.cn/en/ is the manuf.
[16:28:43] <roh> they got some mach3 config xml files, but i couldnt find out any timings to use in stepconf
[16:29:00] <roh> can it be my controller needs some kind of 'general enable' ?
[16:29:13] <skunkworks> http://cnczone.com/forums/showthread.php?t=35576&highlight=syil+emc2
[16:29:21] <cradek> ok, if it works with mach3 it must be step/direction. that tells you something.
[16:29:35] <skunkworks> message 3 has pinout I think
[16:30:16] <cradek> skunkworks is good at finding things.
[16:31:34] <roh> http://www.cnczone.com/forums/showthread.php?t=47448 has the pinout.. i am fairly sure i got a correct one (compared it with the mach3-xml config)
[16:32:10] <roh> also the syil.com.cn has a .doc for download which contains the pinout and some photos of the controller.. just no proper info
[16:32:59] <cradek> what specifically are you trying to figure out right now? just the timings?
[16:33:13] <roh> inside the syil contoller box there is basically this kit: http://goods.ruten.com.tw/item/show?11061228859194 and 2 smpsu for 48V
[16:33:37] <roh> cradek the timings and how to 'enable' my controller (get the yellow led for 'link' lit)
[16:34:54] <cradek> sounds like the enable is the most critical - wonder what "signal 1" and "signal 2" are in that pinout list
[16:34:57] <roh> should the motor do anything when the timings are wrong?
[16:35:55] <eric_unterhausen> maybe not
[16:36:01] <roh> cradek pin 3 is flood (signal 1) and signal 2 (pin 9) seems not to do anything
[16:36:14] <eric_unterhausen> I think signal 1 and 2 are general purpose
[16:36:34] <cradek> pin3 says 'A Axis pulse' (step?) on this pinout
[16:37:00] <roh> jap. the boards are prepared for 4 axis
[16:37:12] <roh> just no stepper driver connected to the board
[16:37:18] <cradek> do you have any software that makes it work?
[16:37:37] <cradek> if so, it ought to be very simple to see what it does to enable
[16:37:42] <roh> take a look at the goods.ruten.com.tw link.. it has pictures of the pcbs/parts of the content
[16:38:23] <roh> cradek we have a win32 notebook with mach3. i just hoped there is something simpler than using mach3 and tapping the parport with an oscilloscope
[16:39:51] <cradek> can't extract any meaning from this page - can't read the driver chip numbers
[16:40:13] <eric_unterhausen> I'll get one of the chinese grad students to translate if that would help
[16:40:29] <cradek> if you can figure out the driver chips, you might find timing info for them online. but RC filters or optos will change the timings.
[16:40:43] <eric_unterhausen> does the relay switch on?
[16:40:45] <cradek> I'm pretty sure there are no optos, but may be RC filters on the inputs
[16:42:23] <cradek> I bet the enable is pin 9. all the other parport outputs are accounted for, and are not the enable
[16:42:53] <cradek> it might need to be low or high, or it might be a charge pump
[16:43:14] <eric_unterhausen> the mach config should verify that theory somehow
[16:43:26] <cradek> if your other software makes it enable, and you have a scope, you can have the information in 10 minutes
[16:43:42] <eric_unterhausen> or even a voltimeter
[16:43:43] <cradek> yes, or maybe you can decipher the mach configuration directly.
[16:43:44] <geo01005> Safety Charge pump?
[16:44:01] <eric_unterhausen> pastebin the mach config
[16:45:52] <geo01005> There is a jumper on the control board to disable the charge pump. I'll bet if you change it, it will start working.
[16:46:13] <jepler> ow reading that cnczone thread is depressing
[16:46:18] <cradek> so there is a charge pump?
[16:46:22] <roh> cradek i tried high and low and also chargepump...
[16:46:40] <jepler> not even the manufacturer representative can give useful information
[16:46:57] <geo01005> On page 57 of the manual it shows the charge pump jumper.
[16:47:16] <geo01005> It is located next to the X home terminals.
[16:47:26] <eric_unterhausen> jepler, which one, the general syil thread? or emc specific?
[16:47:28] <cradek> geo01005: where did you get this? I didn't see a link.
[16:47:38] <jepler> eric_unterhausen: this one: http://www.cnczone.com/forums/showthread.php?t=47448
[16:47:51] <geo01005> http://www.syil.com.cn/tem/BF20.doc
[16:48:09] <geo01005> page 57 when I open it with word 2007.
[16:48:31] <cradek> Works perfect!!!
[16:48:31] <cradek> And its runnign at 45kHz.... I wonder what it was that made the difference...
[16:48:33] <eric_unterhausen> how do you get word 2007 to run on linux? :)
[16:48:33] <cradek> haha
[16:48:56] <cradek> "I fixed it, I have no idea how" is so useful.
[16:49:33] <cradek> this says pin 9 is in fact a charge pump http://www.cnczone.com/forums/showpost.php?p=372234&postcount=40
[16:50:03] <jepler> controlling the spindle with a controller is one of the most fundamental and conveniet differences that separate our product from the commpetition.  Not only does it allow you to select the RPM of your  spindle but also the direction and on / off.
[16:50:13] <jepler> http://www.syilamerica.com/docs/The%20black%20magic%20of%20G.pdf
[16:50:46] <jepler> have you ever done anything .. high tech?
[16:51:49] <roh> jepler the bf20 doesnt have spindle control afaik
[16:52:13] <jepler> roh: I guess that product isn't separate from the competition, then
[16:52:15] <roh> atleast there is no connection between the controller and the spindle-control
[16:52:22] <roh> jepler 'it was cheap'
[16:53:23] <roh> cradek if i configure pin9 to chargepump in stepconf... which frequency will emc2 generate there or how to config it?
[16:53:39] <eric_unterhausen> there is something to be said for cheap
[16:54:09] <cradek> you can easily get the base rate or the servo rate -- with a tiny bit more work you can get any other rate you need
[16:54:09] <jepler> roh: half as fast as your top step speed, but at least 5kHz
[16:54:22] <geo01005> Much more to be said about low cost, high quality.
[16:54:54] <cradek> ``LINK: ... when the "RESET" button is unflickering, the LED will light.''
[16:55:37] <cradek> seems like this means the charge pump makes LINK light up
[16:57:16] <cradek> "13. Potemtiometer: adjust current which thought charge pump to make the charge pump work well."
[16:57:33] <cradek> maybe it does need a specific (unnamed) frequency
[16:58:11] <cradek> but geo01005 is right - with jumper 12 you can "open" or "close" the charge pump
[17:44:40] <roh> cradek if i understand that right, this link via chargepump is intended to make sure the motors dont do anything if there is no app running, correct?
[17:46:16] <archivist> yup yup will power down of the application crashes or is stopped
[17:47:28] <roh> ok.. then i have a reason figuring it out
[17:54:30] <roh> nah.. will build a subd25 male female with some taps to hook a dso onto mach3 and just see how the timings are
[18:34:34] <roh> hm. i configured pin9 as chargepump but on the dso i do not get a continous signal on it
[18:34:54] <roh> neither in stepconf wizard while testing an axis nor in emc2
[18:35:14] <cradek> hmmmmmm
[18:35:25] <cradek> can you pastebin the generated hal file please
[18:35:27] <roh> i am using emc2 2.3.0
[18:35:42] <roh> mompls.. need to usbstick them around.. the emc kernel sadly has no rt2500 wifi
[18:37:18] <cradek> may as well grab the whole config including the stepconf file then
[18:42:31] <roh> http://yamato.hyte.de/tmp/syil_bf20/
[18:44:35] <cradek> roh: sure looks ok at first glance
[18:44:49] <cradek> net estop-out charge-pump.enable iocontrol.0.user-enable-out
[18:45:02] <cradek> shoudl come on when you come out of estop, according to this
[18:48:07] <tj-j> tj-j is now known as clip9
[18:48:24] <roh> hm.. need a smoke... will check how mach3 really does things and then compare
[18:55:53] <jepler> alo try with external estop input disabled (set to unused) in stepconf.
[18:59:12] <roh> jepler estop works fine... after negating it
[19:03:03] <cradek> roh: you should be getting a charge pump output when not in estop. jepler is trying to help you debug that. the other thing you should do is examine the charge-pump.enable pin with halmeter and be sure that it turns on.
[19:08:33] <motioncontrol> good evening at all.alex please you can help me for python example for reload tab file'.I load the example in Idle and run it but have the error for import the emc module.i don't see the emc.py module.
[19:16:44] <jepler> motioncontrol: ". scripts/emc-environment" for the python emc module to be available.
[19:17:37] <jepler> it is lib/python/emc.so in run-in-place systems
[19:18:24] <motioncontrol> jepler emc.so is shared library?
[19:18:33] <motioncontrol> or python
[19:18:50] <motioncontrol> normaly python code terminal with .py
[19:18:50] <jepler> some python modules have the "so" extension.
[19:18:57] <motioncontrol> ok thanks
[19:19:04] <jepler> it is not python source code, it is object code
[19:19:20] <motioncontrol> now prove
[19:20:48] <jepler> $ python -c 'import emc; print "all is OK"'
[19:20:51] <jepler> ImportError: No module named emc
[19:20:59] <jepler> $ . scripts/emc-environment
[19:21:00] <roh> ok.. mach3 generates a 8.8kHz 50:50 signal on pin9
[19:21:04] <jepler> $ python -c 'import emc; print "all is OK"'
[19:21:04] <jepler> all is OK
[19:21:08] <roh> clean ttl levels
[19:24:13] <clip9> Hi everybody, I'm looking into buying a commercial CNC mill and i've been looking at this one: http://www.bzt-cnc.com/product_info.php/info/p269_BZT-PF750-P-Set-incl--Control-and-Ball-Bearing-Spindles-Motors-4.html And i'm wondering if i can use emc with it.
[19:24:43] <clip9> bzt recomends WinPC-NC though.
[19:25:04] <jepler> there is not enough information on that page to determine whether it will work with emc
[19:25:05] <cradek> quite possibly, but you should ask the manufacturer for real technical data about what signals it requires at the parallel port.
[19:25:43] <cradek> your best course of action before they have your money is to be clear that you require this information to purchase it.
[19:26:15] <cradek> just ask roh :-)
[19:26:26] <clip9> yeah. I've sent a mail asking for a real datasheet.
[19:26:37] <jepler> this document shows what information emc will require when doing the basic setup of the machine: http://linuxcnc.org/docs/html/config_stepconf.html
[19:26:48] <cradek> looks like a stepper machine -- one thing you could do is go through stepconf and make sure you know all the things it asks.
[19:26:51] <jepler> some of it is about your PC, but most of it is about the machine to be controlled
[19:27:02] <cradek> um, yeah, like jepler also says :-)
[19:27:38] <clip9> ok thanks. I'll get back to you ;)
[19:27:49] <cradek> great, good luck
[19:29:30] <roh> hm.. how do i get the hal to generate 8.8 khz?
[19:30:10] <cradek> roh: what are you getting now?
[19:30:38] <jepler> probably 5kHz
[19:30:48] <alex_joni> or nothing at all
[19:31:02] <cradek> 5 ~= 8.8
[19:32:42] <jepler> (I generated ini and hal from the earlier-linked stepconf file and after I forced estop-ext TRUE I get a 5kHz charge-pump signal according to halscope.)
[19:33:21] <jepler> (actually, I guess I get that even without forcing estop-ext TRUE)
[19:33:43] <roh> cradek constant '1' aka 5V
[19:34:02] <jepler> are you running emc, or running "test axis"?
[19:34:06] <roh> emc
[19:34:19] <cradek> what did you see when you examined charge-pump.enable with halmeter?
[19:34:25] <roh> in test axis i must note that the 'unused' outputs behave differently than in emc
[19:34:44] <billykid2> hello
[19:35:01] <roh> in test axis they honor invert to define the 'default' state even if marked 'unused' .. in emc they have the other logic state
[19:35:53] <jepler> In the menu, choose Machine > Hal meter. In the window that pops up, scroll down to "charge-pump.enable" and double-click it
[19:40:39] <roh> then i get a dialog which says its false
[19:40:44] <roh> how can i change it?
[19:40:56] <jepler> click the top left icon in AXIS. That's the estop button.
[19:42:15] <roh> does what it should, but not toggle that state or make it generate any frequency
[19:42:51] <jepler> hm really?
[19:43:02] <jepler> when I start, it is FALSE and when I click the estop icon it changes to TRUE
[19:44:00] <roh> the 'powerbutton' directly right from the emergency stop, topmost left
[19:45:00] <roh> enables manual control and the 'R' and 'T' controls
[19:46:48] <jepler> aha
[19:47:07] <jepler> your external estop input is always TRUE
[19:47:45] <jepler> or, at least, it's TRUE when you start emc, so emc starts in "machine off" mode, not "eop"
[19:47:47] <jepler> "estop"
[19:48:06] <jepler> even if you click estop, the user-enable-out never gets set TRUE so the charge pump never turns on
[19:50:58] <jepler> unlinkp charge-pump.enable
[19:50:58] <jepler> net estop-ext => charge-pump.enable
[19:51:03] <jepler> you could try these two lines in your custom.hal file
[19:51:33] <jepler> this would enable the charge pump whenever the external estop is OK. in that case, you'd never use the GUI estop button, always the machine's physical estop button
[19:52:14] <jepler> bbl
[19:54:27] <roh> mmmmh
[19:54:35] <roh> what is the problem linking it correctly?
[19:54:52] <roh> i havent fully entangled it.. but my estop is inverted also
[19:54:59] <roh> like most of the outputs it seems
[19:55:04] <roh> not sure about dir yet
[19:59:06] <BJT-Work> dir of HAL pins?
[20:10:32] <roh> uh.. hm.. got it 'unlocked' by locking it on all buttons and then unlocking
[20:10:35] <roh> weird
[20:14:05] <roh> aaaaaah. now i got it reproducible
[20:14:27] <roh> whenever i hit the estop in the ui the link goes off
[20:15:13] <roh> to enable it i need to press estop on the hw(!) then press estop in sw again and then release the hw estop. then i can 'power on' again and the link led goes on.
[20:15:20] <roh> _ouch_
[20:15:32] <alex_joni> roh: sounds like you need to put it on paper
[20:17:59] <roh> yeah. the stepper do 'something' ... thats progress :))
[20:24:55] <roh> * roh leaves it there for today.., enough exitement.. tomorrow i need to figure out how many of which steps my steppers do and how many mm the spindles are per turn
[20:25:12] <BJT-Work> sounds like your making progress
[20:25:19] <roh> .oO(and use the dso to get a idea of how the timing of the stepper is done there
[20:25:43] <archivist> put off slepp play and enjoy :)
[20:25:48] <archivist> sleep
[20:26:53] <roh> i secured the 'way to get the yellow led on' in our internal hackspace blog... we'll see.. eventually we come around making some stuff public (needs some custom trac-plugin)
[20:27:00] <alex_joni> roh: take small steps, and you'll get there
[20:31:18] <roh> thanks guys.. see you tomorrow ;)
[20:34:47] <BJT-Work> * BJT-Work takes the long way home... :)
[20:35:04] <archivist> bike a gain!
[20:35:14] <BJT-Work> yep :)
[20:35:34] <BJT-Work> see you guys later
[20:42:53] <motioncontrol> hello alex i hev the problem for the reloadtable.py i use the the idle python , and have the error for import the emc file.i have write jepler help but not ok
[21:05:00] <motioncontrol> ok fuction good , the proble is i don't have start emc and have error shmctl : invalid argument .
[21:05:21] <motioncontrol> after start emc is ok
[21:06:08] <motioncontrol> more thanks at jepler and alex for support and help
[21:15:43] <motioncontrol> good night at all
[21:50:12] <JymmmEMC> logger_emc: chump
[21:50:12] <JymmmEMC> I'm logging. I don't understand 'chump', JymmmEMC. Try /msg logger_emc help
[21:51:14] <JymmmEMC> logger_emc: chump A
[21:51:14] <JymmmEMC> A:See [http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2009-04-22#T21-51-14|discussion]
[23:01:37] <Goslowjimbo> Is there anyone working on getting EMC to support a SPI for the Mesa Electronics FPGA cards?
[23:03:56] <geo01005> I have done a little bit on that.
[23:04:28] <geo01005> Goslowjimbo: http://geo01005-ideas.blogspot.com/2009/04/extruder-tempuratue-control.html
[23:05:09] <Goslowjimbo> I'll check it out. Thanks geo01005.
[23:06:11] <geo01005> seb_kuzminsky said he might get a chance to add some formal support at the fest.
[23:09:53] <Goslowjimbo> Mesa has a card that would be ideal for my toolchanger. It speaks parallel, USB, or SPI.
[23:10:54] <Goslowjimbo> Ah- I belive that's the gentleman who also knows about watchdogs.