#emc | Logs for 2010-05-18

[01:05:45] <JT-Dev> logger_2; bookmark
[01:06:03] <JT-Dev> logger_2, bookmark
[01:06:03] <JT-Dev> Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/2010-05-18.txt
[01:06:11] <JT-Dev> finally I remembered
[01:07:38] <morfic> i gotta say, the jog during pause thread is interesting, if nothing else, it shows a lot of passion for the project
[01:09:57] <pfred1> I'm telling you change Abort to ||
[01:10:07] <pfred1> that way its whatever you want it to be
[01:10:40] <pfred1> or maybe X
[01:43:19] <jlmjvm> is image to gcode in 2.3.5
[01:45:24] <pfred1> jlmjvm that sounds nice
[02:44:46] <pcw_home> cragazendo
[02:44:55] <pcw_home> oops
[03:23:12] <morfic> pcw_home: hey, 5i22-1 + 1-2 7i33TA (based on lathe (2 axis + spindle) or mill (4axis + spindle) + 2 7i37TA + cables should cover it all motion and I/O if all drives (servos+spindle) understand a +-10V velocity signal, or did i miss something?
[03:36:18] <Valen> pfred1 you could install emc into ubuntuminimal without too much effort
[03:36:47] <Valen> though I don't really see the need, $100 gets you a SMP machine
[03:39:47] <pfred1> Valen My CNC machine does not need the features ubuntu provides
[03:40:02] <Valen> so dont use them?
[03:40:36] <pfred1> I'd prefer not spending weeks configuring an environment
[03:40:50] <pfred1> turning off everything if it is even posible to do
[03:42:28] <pfred1> Ubuntu is a great beginners distribution of linux in a lot of ways but by being that it hauls a lot of extra baggage to the table as far as I'm concerned
[03:43:21] <pcw_home> morfic: yes. probably overkill for most lathes however
[03:44:42] <Valen> turning off services takes a few minutes
[03:45:10] <Valen> and again they don't seem to be in the way so why bother?
[03:45:45] <Valen> If you want to do it the other way you wind up with gentoo
[03:45:59] <Valen> if your lucky, or linux from scratch if your not
[03:50:58] <Valen> ubuntu has good hardware support, and works out of the box on a wide range of machines, if you start stripping it down your going to start loosing that compatibility
[03:57:42] <pfred1> Valen and gaining performance
[03:58:25] <Valen> I'm getting 4000 latency, and ~900FPS in GLXgears, what performance can be improved?
[03:58:49] <pfred1> I'm getting 17,500 latency and about 500 on GLX gears
[04:00:04] <pfred1> so I suppose sometime down the road I'll do what I have to do
[04:00:32] <pfred1> which is to ditch Ubuntu for a simpler distribution
[04:02:04] <Valen> i'll bet you a coke it does nothing to improve the performance of your system
[04:23:03] <morfic> pcw_home: if that is overkill, what sane spec do you have in mind?
[04:40:41] <pcw_home> Just saying that a lathe may not need as much I/O so a 5I20, a 7I33TA and 1 or 2 7I37TAs may be sufficient
[06:10:06] <ries_> ries_ is now known as ries
[09:15:00] <MarkusBec_away> MarkusBec_away is now known as MarkusBec
[11:45:14] <JT-Dev> I have to make some 3mm standoffs for my PoKeys55
[12:09:20] <alex_joni> https://www.surplex.com/de/maschinen/ansicht,7,wandregal-mit-werkzeugen-und-aufnahmen-91438.html
[12:09:24] <alex_joni> tools anyone?
[12:24:32] <JT-Dev> how about I email you the money and you drop them off next time your in the neighborhood?
[12:41:26] <Dave911> JT-Dev: How long till you get it? I should order one today also ... I need some knobs and I think that will be cheaper than switches.
[12:47:08] <Dave911> The tooledit program in the Git master is still leaving zombie copies of the program in memory after a tool table edit session. Same with the 2.4 version. I tried to run some different versions that DGAR suggested but without success.
[12:47:10] <Dave911> I'm not setup yet to chase TCL problems, so I just wanted to let everyone know... I guess it isn't a big deal unless you do a lot of editing and then begin to run out of memory.
[12:49:23] <Dave911> JT-Dev: Do you see any reason why that Pokeys device might not work?
[13:32:51] <JT-Work> anyone use a PoKeys55 yet?
[14:36:12] <grommit> I put a dial indicator on my axis (specifically Y) and use MDI to command it small moves, say 0.01. I am seeing 0.006 in movement for a 0.01 command. The issue is likely in my hal or ini file but I am not sure where (SCALE?). My ini file is posted at: http://pastebin.org/248163 and my hal file is: http://pastebin.org/248190
[14:39:55] <eric_unterhausen> how far does the machine move if you command 1?
[14:43:39] <grommit> Hmm, let me check. I suspect it will move .6 but let me verify...
[14:51:08] <grommit> When commanded to to move from y0.0. to y1.0 it moved 1.005, it also moved back 1.005.
[14:52:57] <cradek> sounds like it might be working right (except for .005 or so of slop in the axis)
[14:54:08] <cradek> if you have a long gage block, wring one sticking off the end to make a step, then use a DTI to approach both *from the same direction*
[14:54:23] <cradek> use your longest block
[14:55:31] <grommit> cradek: was that for me? If so, I apparently don't speak greek ;-)
[14:56:35] <grommit> long gage block? DTI? approach what?
[15:00:57] <grommit> ok, wikipedia told me what a long gage block is, I don't have one....
[15:01:51] <grommit> and DTI (dial test indicator), I have one of these, and could borrow another...
[15:02:43] <grommit> still not sure what the rest means (approaching both from same direction)....
[15:09:02] <cradek> if your machine has backlash, you need to measure travel by going "left" toward two points a fixed distance apart
[15:09:39] <cradek> if you measure one point going "left" and the next going "right" you'll measure the distance plus the machine's backlash
[15:12:06] <grommit> Do I measure (going left towards two points) with 2 DTI?
[15:16:06] <cradek> no, that's why you need to make a step to measure
[15:17:37] <grommit> Do I leave the DTI in the same place, take a measurement, then remove (or add) a known block and take another measurement? I am not understanding the procedure...
[15:19:14] <renesis> bump the machine one direction a few inches to get the leadscrew against a known side of the nut
[15:19:49] <renesis> mark that location, keep going IN THE SAME DIRECTION to some other point
[15:20:55] <renesis> go back to the start, use some sort of indicator on a fixed part of the machine to judge that, the extra length the machine displays is backlash
[15:23:09] <grommit> AH,thanks, I get it now.
[15:38:06] <jimbo_> jimbo_ is now known as jimbo
[15:49:17] <jlmjvm> cradek:do you know where i can get a sample config for steppers with velocity and pid
[15:49:34] <cradek> no
[15:50:17] <cradek> start with a servo config and add stepgens, I guess
[15:51:22] <cpresser> jlmjvm: why dont you try stepconf?
[15:51:43] <cradek> stepconf gives you position mode
[15:51:56] <cradek> I assume jlmjvm wants to do velocity mode steppers with encoder feedback
[15:52:10] <jlmjvm> the gtom stepper with linear encoders is what i read about
[15:52:24] <jlmjvm> yes
[15:52:30] <cpresser> ah okay, i guess i get it by now. sorry, i wasnt sure about the 'pid'... english is not my native language :)
[15:53:44] <jlmjvm> btw 2.3.5 is looking good,luv the touchoff to tool table
[15:54:35] <cradek> if you're configuring a new machine you should probably be using 2.4.0
[15:59:17] <jlmjvm> 2.4 wouldnt work for me yesterday,just reloaded the os and got the 2.3.5 update
[16:00:16] <jlmjvm> i can load it again if you wanna take a look at the error
[16:06:10] <jlmjvm> when you load the os,should you get all of the updates,or just the emc updates?
[16:06:33] <cpresser> btw.. i had some trouble getting the realtime-system ready to run; in the process i created debian (lenny) packages for the kernel/emc2
[16:06:50] <cpresser> should i upload them somewhere to make them available for others?
[16:22:06] <elmo40> cpresser: I would appreciate it :)
[16:22:32] <elmo40> what is the kernel compiled with? is it thin and light, or 'normal' ?
[16:23:00] <cpresser> the kernel itself is 'light', but there are tons of modules :)
[16:23:09] <cpresser> almost every module is enabled
[16:24:24] <elmo40> still 32-bit, I suppose?
[16:24:31] <cpresser> yes
[16:25:05] <cpresser> its kernel 2.6.30. alsmost the same as the one from the emc2-wiki
[16:33:57] <Jymmm> Can anyone get to http://zircon.com/ ?
[16:35:15] <bill2or3> no.
[16:35:33] <mikegg> ditto
[16:36:27] <Jymmm> thanks
[16:43:09] <elmo40> even the google cached page is blank
[16:43:16] <elmo40> so they have been down for a while
[16:47:10] <cpresser> elmo40: http://ca.rstenpresser.de/~cpresser/tmp/emc2.5pre_lenny/
[16:56:23] <Jymmm> Cool, I get to pickup a HP/AGILENT 9872C PLOTTER this afternoon!
[16:57:20] <bill2or3> nice!
[16:57:26] <bill2or3> i <3 old plotters
[16:59:18] <Jymmm> Yeah, I;ll just rip it apart to get the led and throw the rest away.
[16:59:50] <bill2or3> :-(
[17:00:32] <Jymmm> Ok, what if I turn it into an engraver?
[17:00:55] <elmo40> good luck.
[17:01:04] <elmo40> need to bypass the HP boards
[17:01:12] <Jymmm> Not a problem
[17:01:18] <Jymmm> And not necesarily
[17:01:30] <elmo40> you could, though, just rip out the motors and use them to cut foam, or something
[17:01:58] <bill2or3> I dont think it has the strength to engrave, not built for it.
[17:02:03] <Jymmm> It uses HPGL so may not have to.
[17:02:35] <Jymmm> bill2or3: Yeah, bigger issue is keeping the swarf out of the rest of it
[17:02:51] <bill2or3> that too.
[17:03:08] <bill2or3> make a cnc sand-table, like at taomc.com
[17:04:04] <Jymmm> OH!!! I know.... Toss in a 350mW laser
[17:05:00] <telmnstr> which plotter?
[17:05:05] <telmnstr> I've got a few
[17:05:13] <telmnstr> our laser engraver that I'm converting to run on emc
[17:05:24] <telmnstr> was built out of a Roland DXY series plotter by a company called Meistergram
[17:05:38] <telmnstr> I've removed all the boards and replaced em with two stepper controllers
[17:09:12] <Jymmm> http://www.dealextreme.com/details.dx/sku.23560
[17:09:28] <Jymmm> Toss one of those in it!!!
[17:10:02] <elmo40> ok, that scanner page finally loaded.
[17:10:06] <elmo40> through google cache
[17:10:39] <elmo40> just a stud finder, no?
[17:11:01] <bill2or3> that wont do anything, really. also, DX wont ship >10mw lasers to the US anymore.
[17:11:15] <Jymmm> bill2or3: I know =) Muhahahahahahha
[17:11:29] <bill2or3> good thing I stocked up first. :-)
[17:11:39] <Jymmm> bill2or3: how many do you have?
[17:12:17] <elmo40> anyways... I picked up a Samsung 21.6" monitor from the side of the road. 5 capacitors are blown. I couldn't locate 820uF so I replaced them with 1000uF. it turns on, except the backlight is really dark. maybe the 1000uF is dampening too much? Or would using 50V caps compared the the OEM 25V models be the issue?
[17:12:32] <bill2or3> I just have pointers, 200mw red, 200mw green and some random 10&50mw's
[17:13:23] <celeron55> elmo40: sounds not very likely. maybe there's something else broken too?
[17:14:35] <bill2or3> dunno elmo, I'd put a meter/scope on the PSU output.
[17:15:31] <elmo40> I knew I should have picked up that $300 scope from kijiji...
[17:15:36] <celeron55> you could try buying an another backlight inverter
[17:15:54] <celeron55> they're cheap and you won't mess up the original electronics
[17:16:33] <bill2or3> I'd find the proper-spec'd caps before doing anything complicated.
[17:17:04] <elmo40> before nothing went on. not even the 'power' light. Now the light is on but the backlight is dim. It displays what it should just not bright enough. need to isolate that circuit. there were 4 820uF caps that I replaced with 1000uF caps. same 25V (tight space, 50V caps were too large)
[17:18:26] <bill2or3> I'd get some 820uF caps, try the simplest thing first & all that.
[17:18:31] <celeron55> too high voltage rating has no effect. and it's *very* unlikely 1000uF filter capacitor in place of a 820uF one will have any effect
[17:18:58] <celeron55> if the capacitors are of too low current capacity, that might cause problems
[17:19:18] <celeron55> just not extremely likely either
[17:20:43] <elmo40> bill2or3: I can't locate 820uF caps.
[17:20:45] <elmo40> that was my issue
[17:22:26] <bill2or3> http://shop.ebay.com/i.html?_nkw=820uF+25v+Capacitor
[17:22:35] <bill2or3> ^^lots
[17:23:26] <elmo40> I think the issue was 25V.
[17:23:30] <elmo40> why else would they blow?
[17:23:39] <elmo40> I am going to replace with 35V ones
[17:25:58] <elmo40> nothing in Canada... I hate shipping costs
[17:26:30] <bill2or3> a few years ago there was a plague low quality caps, we've lost 6 identical LCD's in the last year or so.
[17:26:47] <elmo40> they look cheap.
[17:26:53] <bill2or3> so it's probably not a design flaw.
[17:26:57] <bill2or3> what brand are they?
[17:26:58] <elmo40> CapXon brand.
[17:28:50] <bill2or3> google says that's a known-bad brand.
[17:31:15] <elmo40> I didn't make the monitor :P only repairing it
[17:31:20] <elmo40> I see them in MANY items.
[17:31:26] <elmo40> from mobos to power supplies!
[17:33:48] <celeron55> caps don't usually blow because of high voltage, they blow because they're cheaply manufactured
[17:33:53] <celeron55> that is, they just don't last long
[17:34:40] <bill2or3> What I'm saying is that you should replace the caps with identical caps. (except with a quality brand, not a known crappy brand.)
[17:35:25] <elmo40> bill2or3: and I said I can't locate 820uF caps.
[17:35:43] <celeron55> and what I am saying is that 1000uF caps shouldn't cause a problem when used in place of 820uF ones. 8)
[17:35:45] <bill2or3> did you look at that ebay link? there's plenty of 820uF caps there.
[17:35:45] <elmo40> 1000uF isn't all that far off.
[17:36:09] <elmo40> I don't shop ebay.
[17:36:31] <bill2or3> I agree, it's probably fine, but I would want to be *sure* it was fine, before I started looking elsewhere.
[17:36:52] <bill2or3> & replacing with 820uF caps is cheap, so I'd try that first.
[17:37:34] <eric_unterhausen> the only problems I see with replacing an 820 with a 1000uF is finding one that fits, has similar ESR, and has the correct voltage rating
[17:38:01] <elmo40> esr?
[17:38:10] <celeron55> i actually already mentioned ESR (though i didn't remember that abbreviation)
[17:38:20] <eric_unterhausen> series resistance
[17:38:36] <elmo40> as if these stores offer those kind of specs.
[17:38:42] <elmo40> only what is on the cap itself
[17:39:02] <elmo40> I tell ya, hobby stuff is on its way out! they want you to replace units, not components.
[17:39:21] <celeron55> a good store will give you the right ones if you say you're fixing a dead PSU 8)
[17:39:41] <bill2or3> the specs you care about are teh uF, the voltage, and the temp, match those and you'll be good.
[17:39:54] <elmo40> 2 out of 3 aint bad...
[17:39:55] <celeron55> and series resistance.
[17:39:55] <bill2or3> * bill2or3 recently replaced a bunch of dead lcd psu caps, btw.
[17:40:43] <gweepprefect> yeah, low ESR is important in many switching power supply designs
[17:41:03] <elmo40> I replaced the 330uF 25V with a 35V. replaced all 2 820uF 25V with 1000uF 25V and 2 820uF 25V with 1000uF 50V.
[17:41:37] <celeron55> a similar sized 1000uF 50V might have too high ESR
[17:41:57] <celeron55> (similar sized to a 820uF 25V)
[17:42:00] <elmo40> they just had to use non-nominal values...
[17:42:10] <elmo40> ok, I can change the 50V to 25V
[17:42:17] <elmo40> see if that changes anything
[17:49:27] <elmo40> I only have one left :/ need to pick up more.
[17:49:36] <elmo40> I will get the 25V 1000uF
[17:50:27] <elmo40> now, if only I was in Europe... http://picture.yatego.com/images/4a043db0c1c1c4.5/variant_820-35.jpg
[17:51:27] <elmo40> shit. no, I don't want them...
[17:51:30] <elmo40> 820 µF 35/44V CapX. GL
[17:51:34] <elmo40> CapXon brand!
[18:06:44] <Dave_911> dgarr: Jepler's patch and your patch seem to work fine for tooledit. I only get one zombie of tooledit now. NP.. So your patch writes the tool table and reloads it also?
[18:12:59] <dgarr> yes -- you can choose to "Write Tool File" (file only) or, conditionally if emc is running, "Reload Tool Table" (write file and reload tool table)
[18:31:03] <MattyMatt> my roundy mills and pointy bits arrived from France \o/ TC goodness in bijou form
[18:32:58] <MattyMatt> I could try some picture milling now, if my machine wasn't a production workhorse now
[18:33:01] <andypugh> MattyMatt: / MattyCNC http://cgi.ebay.co.uk/Joblot-of-8-9mm-CNC-Router-Rails-with-Slide-Block_W0QQitemZ250631746232QQcategoryZ98641QQcmdZViewItemQQ_trksidZp4012.m8QQ_trkparmsZalgo%3DMW%26its%3DC%26itu%3DUCC%26otn%3D15%26po%3DLCA%26ps%3D63%26clkid%3D8445667906858978578#ht_500wt_1130
[18:34:29] <MattyMatt> nice, but 275mm would shrink my X & Y
[18:34:41] <andypugh> But them end-to-end
[18:34:45] <MattyMatt> and I can't afford to spend that on my Z right now
[18:35:24] <andypugh> It's a charity shop, but seems to have some really odd stuff. They have about a brazillian stepper motors at the moment too.
[18:35:39] <andypugh> Another lot is 2 of the rails for £12 (at the moment)
[18:35:44] <MattyMatt> my Y is sorted anyway :) 2 pairs of those drawer slides on thick wood beams works a treat
[18:36:21] <MattyMatt> it's my X that needs it most right now
[18:36:35] <bill2or3> Mmmrails
[18:37:18] <MattyMatt> I've actually got a rail already, but I don't know how to identify what kind of slider I need
[18:37:33] <bill2or3> that's a trick. any identifying marks on the rail?
[18:37:42] <MattyMatt> nope
[18:38:02] <bill2or3> the rail is the cheap part anyway, it's probably cheaper to get a used rail+bearing off ebay, than to buy a new bearing for the rail you have.
[18:38:13] <MattyMatt> not on the visible surfaces. it's firmly attached to my ballscrew unit
[18:38:28] <bill2or3> ahh.
[18:38:47] <MattyMatt> yeah, I was going to run skate bearings in it
[18:39:07] <MattyMatt> or narrower ones, that tit nicely in the groove
[18:39:12] <MattyMatt> ^fit
[18:39:49] <MattyMatt> or maybe even make a recirc slide myself
[18:40:11] <MattyMatt> how hard can it be? :)
[18:40:48] <MattyMatt> that's a bit ambitious for now
[18:41:27] <MattyMatt> andypugh, are you considering bidding for either slide bundle?
[18:41:53] <bill2or3> hard.
[18:41:53] <andypugh> No, I have absolutely no need for them.
[18:42:41] <eric_unterhausen> I have given up on rails or carriages not accompanied by the other
[18:42:48] <MattyMatt> if I mimic the pressed steel preload technique of the drawer slides, accuracy wont be critical
[18:44:02] <MattyMatt> you can see this rail is well worn over a section anyway, it's probably why it was so cheap :)
[18:45:59] <eric_unterhausen> even brand new rails are often quite cheap on ebay
[18:46:33] <eric_unterhausen> I have some I'd give you if shipping didn't make it un-economic
[19:03:33] <MattyMatt> with the danger the customs will estimate their value :)
[19:05:11] <MattyMatt> looks like I'll be watching their stuff all day tomorrow. those Vextras look nice
[19:07:14] <MattyMatt> I'll be like a man at the bookies
[19:08:24] <MattyMatt> I refuse to snipe tho
[19:08:45] <MattyMatt> it's not sporting, IMO :)
[19:09:03] <eric_unterhausen> I can't afford not to snipe
[19:09:29] <eric_unterhausen> if nothing else, there are too many shill bidders
[19:10:22] <MattyMatt> with that many lots, I may get lucky
[19:10:53] <MattyMatt> I see 3 sets of rails, and about 20 lots of motors, all ending tomorrow
[19:11:15] <eric_unterhausen> could happen
[19:12:22] <MattyMatt> that's how I got the motors I got so cheap. I just lurked and punted low as various sets the same went for 3x what I ended up paying
[19:13:38] <MattyMatt> eventually the non-casual market had been satisfied :)
[19:14:10] <MattyMatt> momentarily
[19:19:21] <grommit> How does ferror relate to backlash, or does it? I measured my backlash today, by (according to renesis and cradek) setting a spot on an axis after moving in a given direction. Move in the same direction and mark that spot, then move back and difference is the backlash. I have x=0.003 y=0.005 and x is lt 0.0005. From time to time I get ferrors and those setting in my ini file are set to ferror=0.0015 and min_ferror=0.001 on X,Y and ferror=0.015 and min_fer
[19:19:57] <grommit> "x is lt" should be "z is lt"
[19:21:00] <andypugh> It rather depends on the feedback system.
[19:21:23] <grommit> I have steppers with 2048 ppr quadrature encoders on them
[19:21:29] <andypugh> In general the system can't see backlash at all, so there is no relationship between ferror and backlash
[19:21:36] <MattyMatt> aha 5 phase, that's why one lot is 9 bids
[19:22:29] <andypugh> The only time that backlash would be part of f-error would be if the position feedback were absolute at the carriage.
[19:23:16] <grommit> Hmm...
[19:24:03] <grommit> With encoders, lets say I am at 1.0 and I command it to move back to 0. But it only moves back to 0.097 (due to backlash), wouldn't that register as a ferror?
[19:24:51] <grommit> zero -> 1.0 rather (in my example)
[19:24:59] <andypugh> No, because the encoder doesn't _know_ it has only moved to 0.097. The encoder has moved to the right place, but the carriage hasn't. due to backlash
[19:25:17] <grommit> Oh, right.
[19:25:25] <grommit> of course.
[19:25:41] <grommit> so much for that theory!
[19:25:58] <grommit> I guess I just have my ferror and min_ferror limits too low...
[19:26:08] <frallzor> andypugh want to give me money? =/
[19:28:17] <jlmjvm> grommit:do your encoders have an index
[19:28:29] <grommit> Yes, they do
[19:29:14] <jlmjvm> have you been able to home to index?
[19:29:23] <grommit> No, how would I do that?
[19:30:41] <jlmjvm> i havent been able to do it yet
[19:32:47] <grommit> I don't see the index in my HAL info. I see index-invert, index-mask, index-mask-invert in parameters, and I see index-enable in Pins....
[19:33:11] <jlmjvm> index enable
[19:33:48] <jlmjvm> are you using a parport setup?
[19:34:57] <grommit> No, I have Mesa cards
[19:35:36] <jlmjvm> my config file prolly wont help you then
[19:37:15] <grommit> Hmm, I put a line in like this for each axis: net zindex-enable hm2_7i43.0.encoder.02.index-enable <=> axis.2.index-enable but when I look at Pin index-enable it is FALSE
[19:38:48] <jlmjvm> lemmee check mine
[19:39:59] <jlmjvm> mine is false also
[19:42:55] <grommit> In the integrators manual it says for the 5i20 (similar my 7i43): "(bit) m5i20.<board>.enc-<channel>-index-enable 2013 when true, and an index pulse appears on the encoder input, reset counter to zero and clear index-enable." So, I'm not sure how that would help me...?
[19:43:41] <grommit> oops gotta run. back later
[20:19:46] <andypugh> index-enable is an odd sort of pin. You can set it, then as soon as the encoder counter sees an index, it zeros itself and puts the pin back to false.
[20:20:56] <andypugh> http://linuxcnc.org/docs/html/config_ini_homing.html
[20:21:11] <andypugh> Explains that index homing is an ini-file option.
[20:21:38] <alex_joni> he left
[20:22:44] <andypugh> So he did. Oops.
[20:32:15] <jlmjvm> alex_joni:is homing to index possible with a stepper,encoder,parport setup?
[20:35:19] <alex_joni> it should be
[20:35:36] <alex_joni> if you run the stepper closed-loop
[20:35:53] <alex_joni> or you can and2 the homeswitch and index pulse to generate a very narow home switch
[20:39:09] <tom3p> wow, i just left HGR surplus, its just rows and rows of blocklong aisles of used machines.
[20:39:10] <tom3p> i left without buying what i came for, I never buy when i get that overwhelmed.
[20:41:00] <cradek> never been there before?
[20:42:02] <tom3p> no, its crazy
[20:42:18] <tom3p> good crazy
[20:43:15] <tom3p> i figger if i still want the stuff, i'll stop in after WIndsor Canada
[20:43:40] <cradek> what's there?
[20:44:50] <tom3p> Transformers for huge KVA, monitors hand pendants, mills lathers VMCs grinders motors big enuf for small towns....
[20:45:15] <cradek> no I meant in Windsor (I'll be near there soon)
[20:45:22] <DaViruz> they have a physical store?
[20:45:35] <tom3p> in Windsor? ask to go to the 'ballet' :)
[20:45:36] <DaViruz> i got a nice 0.75kW omron servo motor and drive from them on ebay
[20:45:58] <tom3p> the 'ballet' is windsor euphemism for the nudie bars
[20:46:20] <tom3p> and loads of mold companies in Windsor & OldCastle
[20:46:32] <cradek> ah, no thanks, there's very little more depressing than a nudie bar
[20:46:57] <tom3p> i just liked the euphemism
[20:47:40] <tom3p> its got a nice little itlaly, good vietnamese restaurants, & middle eastern food
[20:47:46] <tom3p> Italy
[20:50:27] <andypugh_garage> Scary, the cordless circular saw (with woodcutting blade) just whistled through a 3" x 1" ali bar
[20:55:11] <alex_joni> imagine what it does to your arm if you are not carefull
[20:59:05] <Jymmm> * Jymmm counts fingers... 12356789
[20:59:55] <Jymmm> Yep, all EIGHT are there!
[21:02:00] <Dave911> I know a guy that got his arm into a chop saw..ala wood chop saw. I have no idea how he did it. Got quite a ways through too! He tried to explain how he did it ... I couldn't see how he could have done it.. took a lot of surgery to fix that
[21:02:46] <Dave911> Table saws are quite effective at cutting alum also .. might want to wear a face shield though..
[21:04:30] <Jymmm> And no pockets on your shirt either, tend to fill up with swarf. Same goes with pant cuffs and shoes.
[21:04:49] <Jymmm> plus the whole "glitter head" factor.
[21:06:34] <andypugh_garage> TBH it is the battery that impresses me. it hasnt been charged since february
[21:06:49] <Jymmm> Ni-MH ?
[21:06:54] <andypugh_garage> Yes
[21:07:28] <andypugh_garage> Actually not. I just looked. Li-Ion
[21:08:31] <telmnstr> saw naked
[21:09:04] <andypugh_garage> What could possibl go wrong?
[21:09:40] <telmnstr> I guess you could get swarf in the fur
[21:10:37] <andypugh_garage> You might end up changing religion
[21:27:20] <tom3p> LiFePo4 (good batteries)
[21:31:53] <MarkusBec> MarkusBec is now known as MarkusBec_away
[22:16:20] <grommit> I was reading back in the archive for today, as I had to leave quickly. Thanks AndyPugh for respondig to the converstion about index-enable on encoders.
[22:17:12] <grommit> Also I add people talking about having all their fingers and it reminded me of the funniest (in a sick way) accident I remembering hearing about back in 2001 so I looked it up.
[22:17:14] <grommit> http://www.independent.co.uk/news/world/americas/construction-workers-suicide-bid-after-cutting-off-hand-704312.html
[22:17:42] <grommit> Moral of the story: If you're in pain, a nail gun won't help :-)
[22:17:58] <andypugh> I was about to research more info. I am pretty sure you will need to wire the encoder index (Z) channel into some pin on the axis.
[22:18:20] <grommit> in order to home to an index?
[22:19:24] <andypugh> Yes
[22:24:05] <SWPadnos> this is with Mesa hardware though, isn't it?
[22:24:11] <grommit> Yes
[22:24:17] <SWPadnos> stepgens in velocity mode and encoders
[22:24:24] <grommit> you got it :-)
[22:24:36] <SWPadnos> so you physically wire the Z wire to the Z input for that encoder counter
[22:24:37] <grommit> actually, no, stepgen in position and encoder
[22:24:55] <grommit> the velocity mode stepgen drives the spindle
[22:25:02] <SWPadnos> oh. stepgen in position mode screws things up quite a bit
[22:25:26] <SWPadnos> since there is no way to reset the stepgen internal position counter when the index is hit
[22:25:56] <SWPadnos> especially the one internal to the FPGA, which would probably require some extra connections inside the FPGA
[22:26:52] <grommit> Why would you need to reset the position counter?
[22:28:12] <SWPadnos> because the stepgen isn't guaranteed to be at zero when you hit home (if it were, you wouldn't need home switches), and when the encoder feedback resets from the index being seen, there will be a difference between where emc thinks the motor needs to be (0, since the feedback tells us that's where it is) and where the stepgen thinks it needs to be
[22:28:50] <SWPadnos> so the stepgen will try to move to the new commanded position (0), and will probably generate a following error, since emc will see the motion
[22:29:53] <grommit> Without fpga can one tell the stepgen to reset to 0?
[22:29:59] <SWPadnos> no
[22:30:15] <SWPadnos> the software stepgen also doesn't have that feature, but it is probably easier to add
[22:32:24] <grommit> Is pwc around? If so, Peter, is there wany way to reset the stepgenerator position on the 7i43/7i47 based on an encoder index?
[22:34:31] <andypugh> There is a way, you can add an offset on to the encoder output before passing it to the rest of the system.
[22:34:40] <SWPadnos> not really
[22:34:48] <andypugh> (A bit of a workaround, but doable)
[22:34:59] <skunkworks> is there a reason why you don't want to use velocity mode with pid?
[22:35:11] <SWPadnos> there will still be some error, since there's no way to latch the stepgen internal position when the index arrives
[22:35:26] <SWPadnos> so the delta will be calculated on a stale stepgen value
[22:35:59] <SWPadnos> skunkworks, they're steppers, so PID isn't the best thing going
[22:36:39] <grommit> There is no reason, necessarily, no not use velocity mode with pid, but I didn't set it up that way....
[22:36:40] <SWPadnos> limit2 or limit3 may be OK, but I wouldn't really suggest those at the moment, as I don't think anyone has actually used them for motion axes
[22:37:01] <grommit> what is limit2 and limit3?
[22:37:26] <SWPadnos> second-order and third-order limit functions in HAL
[22:38:02] <SWPadnos> essentially you would input a position into limit3, and it would apply velocity (first order) and acceleration (second order) limits on the change, to make a more smooth output
[22:38:48] <grommit> This would be used if I was in velocity mode with pid?
[22:39:18] <SWPadnos> no, it would be used instead of PID
[22:39:35] <grommit> but in velocity mode, or can it be used in position?
[22:39:45] <SWPadnos> oh, right. good point :)
[22:39:50] <SWPadnos> nevermind
[22:39:55] <SWPadnos> you'd need to use PID :)
[22:40:13] <SWPadnos> maybe I need more sugar or something
[22:40:35] <micges> coffee
[22:40:40] <andypugh> Does is matter that the stepgen can't be zeroed? The encoder can.
[22:41:13] <SWPadnos> andypugh, yes it matters, since a stepgen in position mode has an internal position value which it compares to the commanded value, to see if any steps need to be issued
[22:42:42] <PCW> There is firmware support for stegen index but no driver support
[22:42:50] <SWPadnos> I believe that the commanded value changes when the home position is found, since the encoder feedback changes
[22:43:00] <PCW> (stepgen)
[22:44:10] <grommit> The driver is an EMC piece of code
[22:44:15] <PCW> Note that like the encoder it does not really clear the position on index, just stores the stepcount at index
[22:44:16] <grommit> ?
[22:44:27] <andypugh> How about if you were to latch the stepgen counts into a sample-and-hold when is-homed went positive? I think stepgen velocity is zero at that point. Then you can add that value to the commanded position.
[22:44:51] <andypugh> However, I rather thought that homing to an index just used the index pulse as a super-accurate switch.
[22:45:05] <SWPadnos> as long as the driver can tell that the latched position is valuable, it should be able to figure out what to do
[22:45:24] <PCW> If you are slow enough thats fine, the hardware allows fulll spec capturing of step
[22:45:33] <PCW> (speed)
[22:45:36] <SWPadnos> I don't know how that would work with index-enable though, since you would now have two "devices" to enable index on
[22:46:18] <SWPadnos> and the RW pins are only really useful when there's one "answerer" connected
[22:46:19] <andypugh> Yeah, I was misunderstanding how it works, I was expecting to see a pin in the axis.N...... for the encoder index.
[22:46:40] <SWPadnos> no, that's a function of the hardware driver (which could be the software encoder counter)
[22:46:59] <SWPadnos> the motion controller tells the hardware that it's expecting an index, and to please notify once it's found
[22:47:01] <andypugh> Do we always see the index on the Z pin in HAL? Or is it possible for it to be missed at high speed?
[22:47:22] <SWPadnos> the Z pin doesn't exist in HAL when you're using a Mesa card
[22:47:28] <SWPadnos> since the hardware does the counting
[22:47:46] <andypugh> You can read it as a GPIO.
[22:47:50] <SWPadnos> (technically, I think you can read the pin value, so you could see the index in software as well)
[22:47:51] <SWPadnos> yes
[22:48:05] <SWPadnos> but since the stepgen is also in hardware, I think that's irrelevant in this case
[22:48:34] <SWPadnos> since the hardware is working at MHz speeds and the servo cycle is at kHz speeds
[22:48:36] <andypugh> So, potentially with a mess of toggles, edge detectors and and2 blocks you could probably create a very sharp home switch signal using the index.
[22:48:51] <SWPadnos> for very slow motion, yes
[22:49:06] <andypugh> The home latch move can be as slow as it takes.
[22:49:14] <SWPadnos> bbias
[22:49:18] <grommit> ok, i'll keep working on my hardware limit switch then :-(
[22:49:24] <PCW> Drive both stepgen and encoder index from same signal, and enable and clear both (bot some race condition exist)
[22:49:34] <PCW> (but)
[22:51:07] <andypugh> Yikes, I just realised I have to be up in 4.5 hours to catch a plane, and I was hoping for _some_ sleep.
[22:51:28] <grommit> define "some" :-)
[22:51:41] <PCW> Seems like for velocity mode steppe homing stepgen index is not required, only encoder index
[22:52:42] <grommit> I am in position mode currently, but are you saying that if I set it up with velocity mode and pid it would work? If the index pin is not available how does it work?
[22:53:40] <PCW> index pin is available on the encoder and I think that all you need
[22:54:26] <grommit> Are there any downsides to running in velocity mode with pid?
[22:56:58] <JT-Hardinge> If I have a 5v supply and want a pot to vary the output voltage from 0-5v what resistance do I need?
[22:59:33] <PCW> any value other that 0 or infinite
[23:00:06] <PCW> well depending on what reading it maybe 1 to 10K
[23:00:07] <JT-Hardinge> the input can range from 0v to 5.1v
[23:00:24] <JT-Hardinge> its one of those pokeys55 boards
[23:01:54] <PCW> Oh usually for A-Ds you want fairly low like 1K (or put .1 bypass from wiper to ground)
[23:02:17] <JT-Hardinge> http://www.poscope.com/uploads/files/PoKeys55%20protocol%20specification.pdf
[23:02:24] <JT-Hardinge> ok
[23:02:33] <SWPadnos> grommit, other than the fact that steppers don't work the way PID wants them to, there aren't any issues :)
[23:03:33] <grommit> that is a loaded answer!
[23:03:52] <grommit> Is there any (emc) documentation on it?
[23:07:14] <JT-Hardinge> ah I found something after rtfm again lol a 10k with +3.3 connected to 1, 2 to gnd, 3 to input pins
[23:07:33] <PCW> With a high enough uStep ratio I think the velocity mode stepper should work OK
[23:07:35] <PCW> but they are not ideal linear feedback devices due to the flat spot in torque: something like 1-cos(50 theta)
[23:08:05] <JT-Hardinge> * JT-Hardinge needs to disable scan mode when reading a new manual :)
[23:10:35] <PCW> 1-cos(50(theta_motorpos-theta_commandedpos))
[23:12:14] <SWPadnos> there's also the issue that they have no reserve torque - issuing faster/more steps when the motor falls behind doesn't make the motor catch up
[23:18:44] <JT-Hardinge> can a parallel port output pin both sink and source?
[23:22:08] <PCW> Well you have as much torque as you have bu t the consequences are different with a normal servo if you exceed max torque you will fall behind and get a FE
[23:22:09] <PCW> with a velocity mode stepper motor you will stall and get a FE (and have to stop and restart)
[23:25:54] <PCW> On a step motor in velocity mode, if the commanded position is within 90? electrical degrees of the rotor position you do have variable torque to apply
[23:25:56] <PCW> (subject to the flat spot)
[23:30:16] <grommit> So, if you don't exceed max torque then it sounds like there is quite a bit to be gained from velocity mode with encoders (closed loop, homing?). I am interpreting that wrong?
[23:31:24] <PCW> I think so, within its limits its a true closed loop system
[23:31:27] <JT-Hardinge> to cut an external left hand thread all that is needed is to have the start point on the left of the end Z? It seems that way in the sim anyway
[23:32:24] <PCW> parallel ports can sink and source but source current may be quite a bit smaller than sink (very small on pins that are open drain)
[23:35:52] <JT-Hardinge> thanks PCW
[23:38:14] <PCW> Usually better to design for sink (like driving OPTOs) This lack of symmetry probably has it roots in the distant past with RTL ot DTL
[23:38:28] <Valen> hey PCW I'm thinking of adding something from these guys product range for a rotary encoder on my axies http://www.austriamicrosystems.com/eng/Products/Magnetic-Encoders/Rotary-Encoders
[23:38:48] <Valen> looks like you can get 12 bit encoder for $4 ;->
[23:39:09] <eric_unterhausen> digikey has something like that, low resolution though
[23:39:18] <Valen> not particularly accurate but i am only after it for velocity feedback
[23:39:26] <PCW> Yes those are pretty neat (and absolute to boot)
[23:40:07] <PCW> though for now its easier to just use their quadrature output
[23:40:34] <PCW> beware swarf on the magnets...
[23:40:49] <Valen> yeah it'll need a boot of some description over it
[23:41:12] <eric_unterhausen> those are cool
[23:41:20] <eric_unterhausen> better than what I was thinking about
[23:42:15] <Valen> I'm yet to see magnets for them off the shelf though
[23:42:57] <PCW> They would be good for a low cost BLDC drive (absolute means no hall signals or bump start)
[23:43:53] <PCW> Doesn't AustriaMicrosystems have sample magnets?
[23:44:14] <Valen> I was planning on using them for BLDC drive too
[23:44:23] <Valen> I dont know I havent looked in detail there
[23:44:48] <eric_unterhausen> I managed to stupidly smoke all the bldc drives I felt comfortable with
[23:45:12] <eric_unterhausen> I'd feel better if there was large quantities of smoke
[23:45:21] <PCW> Well you should stop smoking...
[23:46:21] <eric_unterhausen> but it was all silly things like killing the serial ports instead
[23:47:16] <PCW> OH wondered how you smoked them, I would think the power elex would be pretty well protected
[23:47:39] <eric_unterhausen> I did manage to short one of the outputs when I moved the mill and a connection came loose
[23:47:51] <eric_unterhausen> physically moved the mill from one end of the room to the other
[23:48:14] <PCW> smoke or just OC fault?
[23:48:24] <eric_unterhausen> I'm not sure there is anything wrong with the drive, but it thinks there is
[23:50:50] <PCW> Depending on whether the drive has high side OC detection the IGBTs might survive a phase short to ground
[23:50:52] <PCW> but the transient may have killed something else
[23:51:38] <eric_unterhausen> I plugged in another controller and it seemed to work
[23:51:56] <eric_unterhausen> power board is separate
[23:53:21] <morfic> cradek: you live and breath interpreter code right? :)
[23:58:16] <skunkworks> alex is usually the one that take on the challeng... ;)
[23:59:17] <skunkworks> challenge
[23:59:27] <skunkworks> wow - is that right? doesn't look right