#emc | Logs for 2010-04-06

[00:00:07] <PCW> sounds like a car crash
[00:00:52] <jthornton> after all that plug switching the wife's laptop still has internet :)
[00:00:56] <SWPadnos> plymouth is the bootup hardware detection thingy, isn't it?
[00:01:00] <mozmck2> heh, it could cause one I guess. I think it may be crashing my rtai kernel.
[00:01:20] <mozmck2> I believe it's just the pretty boot thingy.
[00:01:25] <SWPadnos> oh hmm
[00:01:33] <SWPadnos> add nosplash to the boot command line :)
[00:02:25] <mozmck2> Oh, I can get it to boot by adding nomodeset - but after my latest upgrade that doesn't work and I have to add noapic.
[00:03:04] <mozmck2> I removed 'splash' and 'quiet' as well.
[00:03:18] <SWPadnos> huh
[00:04:37] <mozmck2> they're putting all this work into getting a pretty bootup and making thing more unstable as far as I can tell. Looks like lots of folks are having plymouth related problems with the standard ubuntu kernels as well...
[00:05:01] <Valen> working on pretty bootup seems to be going about things backwards lol
[00:05:12] <Valen> my PC boots in ~8 seconds anyway
[00:05:33] <Valen> it doesn't need more work on looking pretty during that period
[00:05:38] <pfred1> I don't see PCs boot all that often really
[00:05:50] <Valen> I turn mine off at night
[00:06:01] <mozmck2> 10.04 does seem to boot fast and that's one of their major pushes this release - boot speed.
[00:06:10] <SWPadnos> when you carry a laptop traveling for business, you see a lot more bootups
[00:06:12] <pfred1> Valen do you turn your refrigerator off at nite too?
[00:06:20] <mozmck2> haha
[00:06:34] <tom3p> Dave911 did you find anyone working on the position closed loop PID with velocity mode stepgen? I just emailed gtom asking if he had done so
[00:06:34] <Valen> no, but I need my food to stay cold, I don't need my electrons to stay warm
[00:06:38] <SWPadnos> and in that case, it's actually good to have something that looks nicer than a bunch of text scrolling by
[00:06:57] <pfred1> Valen Linux actually does cron jobs at nite
[00:06:57] <Valen> usplash is pretty enough ;->
[00:07:05] <SWPadnos> (especially when you work with people in the film or photo industries)
[00:07:07] <mozmck2> some folks I read though said that they gained 15 to 30 seconds by uninstalling or disabling plymouth
[00:07:19] <Valen> yes, and i leave my servers on at night
[00:07:22] <jthornton> at least I can make a good pizza sauce even if I know squat about networks
[00:07:30] <SWPadnos> jthornton, prove it!
[00:07:34] <Valen> my desktop however does nothing overnight other than waste power
[00:07:44] <Valen> and power costs money
[00:07:47] <jthornton> gnipsel.com food recipes
[00:07:54] <pfred1> Valen it doesn't appear to be doing much more than that to me now
[00:07:55] <SWPadnos> oy vey
[00:08:14] <SWPadnos> geez guys, let's be civil, shall we
[00:08:26] <Valen> I think i'll leave that alonw
[00:08:27] <Valen> alone
[00:08:30] <SWPadnos> thank you
[00:09:37] <Valen> so SWPadnos galavanting about the country side much?
[00:10:01] <jthornton> go to Tasty Pizza Sauce
[00:10:14] <SWPadnos> not the past few weeks, but I'm headed to Las Vegas next week, New York the week after, and then off to a week long Caribbean cruise
[00:10:23] <Valen> sounds tough
[00:10:23] <SWPadnos> so there is a little travel in my near future :)
[00:10:27] <SWPadnos> heh
[00:12:41] <Valen> I take it your not working on the cruise?
[00:13:54] <SWPadnos> no
[00:14:12] <SWPadnos> well, I may do a little software development, but not much
[00:14:36] <Valen> lots of fun sun and women then ;->
[00:14:58] <SWPadnos> yes
[00:29:00] <morfic> SWPadnos: is the anything I/O pci stuff from mesa with the appropriate daughter board likely the best choice for a drive/servo combo? (it appears to me the daughter boards provide the actual +-10V)
[00:29:30] <Valen> depends how you want to drive stuff
[00:29:46] <Valen> mesa have amps that plug directly into the anything io boards
[00:29:49] <SWPadnos> yes, the daughter board converts the PWM + direction output that the FPGA makes into the +/- 10V that the drives expect
[00:30:06] <SWPadnos> Valen, probably not for the size motors he's got
[00:30:11] <Valen> but if you want to use external amps then you can use the +-10v things
[00:30:19] <Valen> morfic: what you pushing?
[00:30:26] <SWPadnos> BP or larger CNC machines, probably using 150V/10-40A motors
[00:30:39] <Valen> yeah thats getting serious
[00:30:50] <Valen> although they do have some larger drivers too I saw
[00:31:10] <SWPadnos> actually, I think I'd be surprised if any of the machines are as small as a BP, from what I understand of the shop :)
[00:31:36] <Valen> oh yeah thats right morphic is doing the big boy toys ;->
[00:31:49] <SWPadnos> yep
[00:32:34] <Valen> I'm a little suprised mesa dont have some larger drivers really
[00:33:00] <morfic> SWPadnos: i am one "yes" closer to understanding things now, at the beginning of trying to understand anything i am way to ADD (i should get this officially diagnosed some day), good thing is, once i do have enough questions answered, i should be dangerous
[00:33:11] <SWPadnos> excellent!
[00:33:14] <SWPadnos> we love danger
[00:33:16] <morfic> Valen: the servos do have drives/amps so i just need to talk to them
[00:33:37] <Valen> yeah then the +-10v boards are probably the way to go
[00:33:37] <morfic> and what dialect of klingon they speak i check out in the shop tomorrow i hope
[00:33:54] <morfic> like making sure they DO speak +-10V
[00:34:31] <morfic> SWPadnos: +-10V would be an analog drive/servo, not digital, right? (since we talked about percentages earlier)
[00:34:54] <SWPadnos> for the most part, the servo is guaranteed to be analog
[00:35:04] <SWPadnos> the drive will have an analog output as well
[00:35:22] <SWPadnos> but yes, that's an analog drive, since it takes an analog command input as well
[00:35:48] <SWPadnos> (motors are analog devices, so it's kind of a question of where the digital / analog changeover happens)
[00:35:59] <morfic> i am trying to filter out info we got from the windows based control guys, one called them analog, then his sales guy says he can not deal with our controls since our drives are digital, might explain why i am so majorly confused
[00:36:12] <SWPadnos> ?
[00:36:13] <morfic> SWPadnos: yes, i do rever to the drive as digital/analog then
[00:36:14] <SWPadnos> indeed
[00:36:21] <morfic> refer*
[00:36:46] <SWPadnos> there are drives that take serial commands as input, those are often called smart drives, and could probably be called digital drives as well
[00:37:11] <morfic> i wish Hans Platt with his awesome german accent had helped us more, not the texas sales guy who contradicted everything everyone else said
[00:37:14] <SWPadnos> since they're all analog on the motor end, people most likely call them by whatever the input is :)
[00:37:21] <SWPadnos> heh
[00:37:30] <morfic> serial, ahh
[00:37:31] <SWPadnos> which Fanuc is this again?
[00:37:32] <morfic> sec
[00:37:45] <morfic> 0-T on the lathe, 16M on the mill
[00:38:21] <SWPadnos> ok. that tells me nothing ;)
[00:38:36] <morfic> spindle servo number i got referred to something serial, going to bring up that page again, sec
[00:38:53] <SWPadnos> ok
[00:39:08] <SWPadnos> there are some drives that have a serial interface, as well as other control interfaces
[00:39:14] <morfic> can i paste 3 lines? (there is usually some pre-pastebin limit of lines)
[00:39:34] <pfred1> 3's OK
[00:39:35] <SWPadnos> I have some Yaskawas like that - you can tune and set up via serial, and also monitor current or other things
[00:39:37] <SWPadnos> sure
[00:39:48] <pfred1> 4's OK 5 is pushing it
[00:40:07] <morfic> Spindle servo - A06B-6064-H322#H550 No ES5205393 A
[00:40:07] <morfic> Servo Amplifier - A06B-6058-H005 No F3Y 00389-B
[00:40:07] <morfic> Servo Amplifier - A06B-6058-H006 No F3Y 00462-B
[00:40:57] <morfic> http://fanuc.cnc-electronics.com/drive/serial-spindle-a06b6064/h322.html
[00:41:26] <Valen> mmm electronics with the covers off
[00:41:30] <morfic> above info is all i got via email when i asked for info from work.
[00:42:09] <Valen> morfic got a picture of the machine?
[00:42:16] <morfic> oh nice, zimbra let's me get to the attached jpg i sent
[00:42:34] <morfic> yes, was gonna show what i got on that one email anyway :)
[00:43:31] <morfic> oh wow, forgot our wireless has such amazing upload :( be a moment
[00:43:37] <Valen> lol
[00:44:39] <morfic> boss should upgrade his microwave deal some day
[00:44:56] <morfic> hey did JT-Hardinge ever get his lan straightened out?
[00:45:04] <Valen> he's going to run 2 cables
[00:45:22] <jthornton> ok, I have one problem with plan C it is too far from the antenna to the router...
[00:45:26] <Valen> so internet goes into router then into hub/perhaps something new
[00:45:41] <jthornton> this is JT-Hardinge
[00:45:42] <Valen> whats the power injector to the antenna?
[00:45:58] <jthornton> the cat5 cable provides the power
[00:46:00] <Valen> usually its a box with 2 network jacks on it?
[00:46:08] <Valen> yes but how does it get power in
[00:46:16] <pfred1> jthornton I think that is how reprap does it
[00:46:18] <jthornton> a wall wart
[00:46:29] <Valen> so you have a power over ethernet hub?
[00:46:43] <Valen> as in you plug the power into the hub and that into the antenna directly
[00:46:59] <tom3p> morfic Hans Platt from PowerAutomation? ( german cnc company )
[00:47:06] <Valen> theres no little black box in the line from the hub to the antenna?
[00:47:06] <pfred1> jthornton reprap just uses RJ-45s as cheap connectors
[00:47:31] <jthornton> the wireless thing has a wall wart and has an in and out rj45 connector
[00:47:34] <morfic> tom3p: that one
[00:47:44] <jthornton> and one plug for the power
[00:47:54] <Valen> yeah just plug the cable into the out of that wall wart
[00:48:04] <morfic> Valen: http://www.zerorealm.org/12LC/3-12-10_001.jpg http://www.zerorealm.org/12LC/3-12-10_002.jpg http://www.zerorealm.org/12LC/3-12-10_003.jpg http://www.zerorealm.org/12LC/3-12-10_004.jpg http://www.zerorealm.org/12LC/3-12-10_005.jpg
[00:48:31] <tom3p> wow, i worked with him in broken arrow, he was just making the move to take over leading that company, can you clue me into contacting him?
[00:48:53] <jthornton> so get a female jack to connect to the cat5e down to the router?
[00:49:08] <Valen> hang on I'm lost
[00:49:11] <morfic> tom3p: know him? power automation, camsoft, soft servo and MDSI/OpenCNC (meanwhile out of the run) are windows based solutions we look at, emc2 would be the going against them
[00:49:12] <tom3p> btw his is NOT windows based, thats just the front end, the cnc is hardware running linux
[00:49:30] <Valen> I'm thinking this, you have antenna > cat 5 > box > cat 5 > hub
[00:49:51] <jthornton> yes
[00:50:02] <skunkworks> so - is there and more elegent way to convert bits (switches) in ladder to a number? The one way that comes to mind is to have a variable assignment for each switch and then add them togather to get the number - so the first switch would set %w0 to 1 if it was activated - switch two would set %w1 to 2 if it is activated - switch 3 would set %w2 to 4 if it is activated and so on (15 total) and then add them togath
[00:50:07] <morfic> tom3p: i never got this much info, i do not know how much his distributor/integrator knows about this, it has certainly never been mentioned
[00:50:13] <pfred1> morfic these pictures claim they have errors in them?
[00:50:19] <Valen> so you should just need to change box > cat 5> hub to box > cat5 > router
[00:50:30] <morfic> tom3p: i will see if i can get his email address, i did not deal with power automation, just was called in on a phone conference
[00:50:37] <Valen> pfred1 they are working for me, just slow
[00:50:59] <pfred1> morfic use imageshack
[00:50:59] <Valen> morfic: damn thats a tool holder on that thing lol
[00:51:06] <jthornton> ok, one of those double female rj45 things
[00:51:13] <morfic> pfred1: what errors? there was no 404, i actually checked for once
[00:51:20] <skunkworks> it would be great if you could actually change the bits of a intager...
[00:51:22] <Valen> yeah, or just make the cable long enough
[00:51:31] <pfred1> morfic must just be my browser being brainded it is loading
[00:51:32] <tom3p> morfic http://www.powerautomation.de/index.php?id=268 'sink edm' :)
[00:51:53] <morfic> pfred1: i didn't know the resolution they were sent in, else i had upped them to something faster
[00:52:14] <Valen> I love the cable out the back going back into the box
[00:52:16] <jthornton> the cable from the antenna to the antenna device can only be 300' so I can't make the cable longer
[00:52:34] <Valen> yeah i meant the cable from the router to the hub/box
[00:52:43] <morfic> Valen: large tool holder? we have a 1.25" square turret, that machine has the guts to take .5" DOC on cobalt
[00:53:04] <pfred1> morfic I donno what it is I'm looking at here but I'd love to strip it for parts!
[00:53:11] <morfic> haha
[00:53:31] <Valen> I reckon I could sell your swarf as stock for parts morfic ;->
[00:54:16] <pfred1> morfic when I strip boards I don't play around i have a 5 pound solder pot i hold them over
[00:54:21] <morfic> Valen: if we know how well this all works for us, boss will either buy machine sans control, or yank the fanuc control and ebay it when we buy the new machine
[00:55:09] <pfred1> morfic I bought it after I aquired 2 old mini-computers and hand desoldering was no longer an option
[00:55:34] <morfic> this lathe would be retrofit #1, mill would be #2, if those two work, do rest
[00:56:09] <Valen> whats wrong with the old ones out of curiosity?
[00:57:03] <morfic> any little thing you want enabled, fanuc bends you over and uses no vaseline when they charge you, and when stuff breaks it's the arms and the legs, so no longer bending over needed
[00:57:24] <Valen> ahh
[00:57:37] <Valen> so your keeping the fanuc amps?
[00:57:46] <morfic> ideally, yes
[00:58:15] <pfred1> morfic look at the size of that transformer!
[00:58:30] <morfic> at the same time whatever we do get should be flexible enough to replace them with ither drive/servo combo easy
[00:58:55] <Valen> yeah EMC is good for that
[00:59:00] <morfic> that little one on the left in 005?
[00:59:34] <tom3p> morfic: fwiw: the 'texas salesman' lead you wrong, the PA control runs from a browser ( any ) and communicates to the hardware with OS dependant libraries ( like EMC can be remoted ), and Hans is a good guy
[00:59:36] <morfic> man i wish i was allowed to take my slr to take good pictures of that stuff, so much shaken, not stirred stuff
[00:59:38] <Valen> I just hope you don't need to use ladder or any of that stuff to get running
[01:00:00] <pfred1> morfic how old is this stuff? I see 7 segment LED displays on a board
[01:00:08] <tom3p> running from a browser means you write operator panels in HTML & javascript :)
[01:00:44] <morfic> 1995ish for most machines, without knowing that particular machine's exact age
[01:00:47] <jthornton> ladder is easier than networks LOL
[01:01:06] <Valen> jthornton: only for you ;->
[01:01:16] <morfic> tom3p: i will ask PA about that, or maybe get in contact with the other texas PA guy
[01:01:20] <jthornton> :)
[01:01:23] <Valen> with luck we should have an air cylinder to balance out Z soon
[01:01:30] <Valen> should make it a bit smoother
[01:01:49] <tom3p> Hans was from texas at that time, moved to Wi to take over US ops, then .... lost track
[01:02:38] <morfic> tom3p: i just know he was on the road, he was our first contact and he got us in contact with the texas distributor/integrator
[01:03:46] <morfic> Valen: hydraulics on X to balance the turret on the bigger lathe.....since you seem to like big stuff
[01:04:15] <Valen> bah why aren't you just using hydraulic servos then ;-P
[01:04:17] <morfic> tom3p: and Hans gave quite a bit of detail and the distributor does not seem to mesh up with what Hans said on the phone
[01:04:45] <morfic> Valen: ask Daewoo? :P
[01:05:21] <morfic> today was extremly useful, and this channel and all people participating extremly helpful, thanks all
[01:05:45] <morfic> i;m gonna look closer at the drivers tomorrow, see what mesa bits and pieces it might take
[01:05:45] <Valen> have fun
[01:05:59] <Valen> pinouts are fun ;->
[01:06:10] <morfic> April 21st will show if i get to have fun or not.
[01:06:12] <Valen> I have a feeling it will be hard to get them out of fanuc
[01:06:17] <Valen> whats then?
[01:06:25] <jthornton> dmesg is your friend with mesa
[01:06:38] <morfic> deadline to pick between the windows based controls and/or emc
[01:06:41] <morfic> bbiab
[01:06:42] <Valen> ahh
[01:07:03] <Valen> I kinda wondered why mesa doesn't work with mach
[01:07:34] <Valen> I know there are no drivers or whatever, but it seemed odd that mesa wouldn't make them
[01:07:38] <jthornton> I guess no drivers
[01:07:46] <Valen> lol snap
[01:08:08] <jthornton> both typing the same thing at the same time LOL
[01:08:46] <pfred1> eventually Windows will just be another footnote in the computer biz much like CP/M is today
[01:09:36] <skunkworks> someone is working on a mach driver for mesa... But it is vaporware at the moment.
[01:10:23] <Valen> they using hostmot2 or the mesa motion controller?
[01:10:26] <jthornton> * jthornton can't figure out why you would pay for mack when EMC2 is better
[01:10:31] <pfred1> skunkworks the way someone was talking about Mach here last nite it sounded like Mach is vaporware lately
[01:10:33] <Valen> if they use mesa motion it might have a chance
[01:14:18] <pfred1> Eagle Lite strikes again! http://img62.imageshack.us/img62/6751/hsoptogs.png
[01:14:49] <Valen> getting serious
[01:15:04] <Valen> still a few air wires
[01:15:29] <pfred1> oh i never plan on etching this board I'll drill it onto some phenolic and hand solder it all up
[01:15:32] <skunkworks> mesa motion firmware
[01:15:48] <pfred1> its just a drilling guide template
[01:16:50] <skunkworks> pfred1: isn't the free version great? http://www.electronicsam.com/images/KandT/servostart/schem/latestcurrentlimit/latestboard.png
[01:16:57] <pfred1> and when I hand solder it i use the eyeball tool in Eagle to show me where the wires all have to go too
[01:17:04] <Valen> I'm getting into kicad
[01:17:30] <Valen> parts library isnt so good but other than that its not so bad
[01:17:30] <pfred1> skunkworks honestly i htink it is because of the free version of Eagle that all the FOSS schematic layout programs suck so badly
[01:17:43] <Valen> pretty clever on their part really
[01:17:58] <pfred1> I tried kicad and geda neither worked on my system
[01:20:22] <pfred1> skunkworks really I find the 3x5 limit in the lite version makes me make better boards than I would if i had unlimited area to work with
[01:21:16] <pfred1> skunkworks does your version of Eagle ever smear scrolling the library?
[01:21:43] <Valen> mine used to, it also wouldn't work with compiz, being transparent all the time without some workaround
[01:22:17] <jthornton> time for me to call it a day thanks for the help
[01:22:17] <tom3p> does emc2's PID loop need to have position & posn feedback as inputs? couldnt it have a pot and a voltage as input ( just work on the diff between 2 floats, whatyouwant & whatyougot?)
[01:22:22] <pfred1> on high zoom even work windows can smear for me too but i see that less
[01:22:54] <tom3p> goodnite george
[01:22:55] <pfred1> tom3p is your computer analog?
[01:23:01] <tom3p> my process is
[01:23:24] <pfred1> tom3p then i guess you'd need a DA
[01:23:39] <pfred1> or actually an AD
[01:24:23] <SWPadnos> tom3p, the PID in EMC2 is just float in -> float out, like any self-respecting PID should be
[01:24:27] <pfred1> tom3p soundcards are AD/DAs
[01:24:33] <tom3p> nah, the question was if the thing really knew anything about the data it processes, it doesnt. wether the data is from a pot or a DAC or a SWAG doesnt matter
[01:24:49] <tom3p> SWPadnos, thx, thought so
[01:30:50] <Dave911> Mach already has a Galil plug in but it can't do threading.... so what's the point?
[01:31:21] <tom3p> really? Galil can thread, sounds crippling
[01:31:28] <Dave911> Plus the Galil cards are not inexpensive
[01:31:43] <SWPadnos> well, that's the critical problem with Mach drivers - they have to do everything
[01:31:54] <Dave911> Yes, the Galil itself can thread, but the Mach plug in was never finished..
[01:32:14] <SWPadnos> that's why the smoothstepper does some things but not others, and the G100 did some things (but never got threading right), etc.
[01:32:20] <Dave911> Seems like a lot of the Mach stuff gets to the 95% complete mark and it stops for various reasons..
[01:32:23] <tom3p> ah, ya know the original ancient emc was a shell to a pmac
[01:32:31] <SWPadnos> yes
[01:32:48] <Dave911> I didn't know that.. that is very interesting...
[01:33:04] <SWPadnos> PMAC motion controller to run the PID and motion algos
[01:33:25] <tom3p> i got the code somewhere, always dreaded looking into that pmac address register mess
[01:33:32] <SWPadnos> that's one reason why the motion controller and IO controller are separate, and why NML was used for communications between them
[01:34:08] <SWPadnos> and the user interface ran on a separate CPU card, which is one reason why the UI talks to the task controller via NML :)
[01:34:09] <Dave911> Oftentimes a plug in goes just so far and then something needs to be fixed internally in Mach to make it fully functional and there it dies.. you can't do everything in a Mach plug in sometimes without some help ...
[01:34:24] <SWPadnos> it was originally a distributed computing system :)
[01:34:30] <tom3p> yeh
[01:34:55] <Dave911> The pmac cards were very advanced at the time ..
[01:35:56] <SWPadnos> they (Delta Tau) still have nice stugg
[01:35:59] <SWPadnos> stuff
[01:36:20] <SWPadnos> they were going to release a Linux based motion controller a few years ago, but I don't know that they ever did
[01:37:05] <Dave911> Yes they do. I haven't used much of the new stuff, I went through training on the pmac VME cards. I still have the 4 inch binders of stuff around here someplace. That card was simply overhwhelming.
[01:37:26] <tom3p> cant see how vel mode stepgen can have position limits, going for a walk :) bbl ( yes Dr. Dimitri is great )
[01:37:57] <SWPadnos> it doesn't, except possibly for 32-bit number limits
[01:38:34] <Valen> that'd be a long way
[01:39:03] <SWPadnos> for software stepgen, yes, but then again a year is a lot of seconds
[01:39:07] <tom3p> crap its raining
[01:39:10] <SWPadnos> (about 10^7*pi)
[01:39:28] <SWPadnos> or pi * 10 ^7
[01:39:30] <pfred1> tom3p what do you want to live in a desert?
[01:39:42] <tom3p> no i meant like posn mode goes to a float number limit. I want a numeric limit but to drive there in velocity mode
[01:40:33] <tom3p> s/limit/target
[01:41:00] <SWPadnos> oh. use a limit, limit2, or limit3 block. that'll take care of accel and vel limits too
[01:42:55] <tom3p> "limit2 - Limit the output signal to fall between min and max" uh the signal would be a velocity, the limit would be a position... apples & pranges
[01:43:07] <SWPadnos> oh hmm.
[01:43:14] <tom3p> s/o/p
[01:43:24] <SWPadnos> or s/p/o/ :)
[01:43:33] <tom3p> yeh , not a vi guy
[01:43:34] <SWPadnos> apples and pringles (same difference ;) )
[01:44:05] <pfred1> I don't have any Pringles I do have an apple though
[01:44:37] <tom3p> i run into not being able to use floats in the fast thread too, so i used 'units' and hoped to get the mechanics to be a nice value like 1 step = 1um
[01:44:39] <SWPadnos> if you have two, you could eat one and then compare apples and cores
[01:45:05] <SWPadnos> it should be possible to use floats in the base thread
[01:45:34] <SWPadnos> I don't know if motion will create that thread with float support though, so you may have to do some shenanigans to get a FP-using fast thread
[01:46:29] <tom3p> yes, i was hoping to evaluate as fast as possible, maybe its not necessary to be in servo thread ( thats the faster one right? i dont use those names )
[01:46:38] <morfic> SWPadnos: i guess the biggest thing i had is what hardware goes between drives and pc/emc, since most software based solutions just talk about "boards", but never say what the board really is (our goal is the least possible amount of proprietary hardware), and for EMC i think i have a much better idea now what this hardware is
[01:46:51] <SWPadnos> base is usually the fast "step generator" thread, and servo is the 1ms thread
[01:47:12] <SWPadnos> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?EMC2_Supported_Hardware
[01:47:17] <SWPadnos> I don't know if that's complete
[01:47:17] <tom3p> morfic, as far as dollar per i/o point, emc2 is WAY cheaper than anything you'll find
[01:47:41] <pfred1> those mesa bords for what they are are quite a reasonable deal
[01:47:52] <SWPadnos> I believe that EMC2 also supports more hardware, from more vendors, than any other CNC controller
[01:48:01] <morfic> tom3p: it is
[01:48:05] <SWPadnos> definitely more than any sub-$5000 controller
[01:48:21] <tom3p> these cnc companies lock you into thier hardware/expanders/breakouts and your pockets go empty
[01:48:34] <morfic> SWPadnos: indeed, i think it comes down to who programs on payroll things that emc2 currently doesn't do
[01:48:36] <pfred1> tom3p everybody loves money!
[01:49:01] <SWPadnos> morfic, one interesting thing about EMC2, which may not be explicitly stated anywhere, is that you can use more than one hardware interface, by different companies, at the same time
[01:49:18] <morfic> tom3p: idea was "pay once, use many", but now it turns into a very fanuc like "well, then you need this option and that option and probably this too"
[01:49:34] <SWPadnos> so you could use a Mesa card for some things, and also use a Pico systems card for other things, on the same machine
[01:49:56] <morfic> SWPadnos: drive 3 axis with a mesa board and 2 from another board?
[01:50:02] <tom3p> ah the days of trading fanuc codes... need barrel interpolation?
[01:50:03] <SWPadnos> heh. look at AjaxCNC or Centroid for some really long lists of "extra options"
[01:50:07] <SWPadnos> morfic, sure
[01:50:16] <morfic> SWPadnos: i see i type too slow, you already answered my question
[01:50:46] <morfic> centroid is now turnkey machines, i only saw centroid on an old bridgeport 2 jobs ago
[01:51:01] <SWPadnos> you could also run e,g, a slow rotary indexer with the parallel port, while the servos and spindle are run by a Mesa card, etc.
[01:51:21] <SWPadnos> there's another one that charges extra if you want the interpreter to support the S word
[01:51:32] <SWPadnos> yes, it's extra if you want to be able to set spindle speed
[01:51:49] <morfic> SWPadnos: so if machine goes down and mesa is out of stock, i could replace it with another card, reconfigure emc and go, and if i run out of IO i could get 2x a smaller one configure and go and just make sure i have my bigger interface back for later?
[01:52:02] <SWPadnos> yep
[01:52:29] <morfic> good, one point where emc2 surely kills the others
[01:52:31] <SWPadnos> and, if you set things up right (for instance, making the conversion to engineering units as soon as possible), you don't even have to change any tuning parameters
[01:53:10] <morfic> engineering units for what?
[01:53:23] <SWPadnos> if you set the scaling on the PWM/analog outputs so that they represent e.g. inches/minute of speed (for a velocity servo drive), then you can change interface hardware, and all you have to do is set the output scale correctly and you're good to go
[01:53:36] <SWPadnos> encoder feedback in inches, for example
[01:53:40] <SWPadnos> instead of counts
[01:53:50] <morfic> ahh
[01:54:10] <SWPadnos> the hardware drivers all have scaling parameters, so the rest of the system never sees anything but "inches"
[01:54:45] <SWPadnos> so if a 1000 line encoder dies, and all you have on hand are 2000 line encoders, just stick one on there and change one number in the ini file, and start machining again
[01:54:50] <morfic> get away from counts, but set what actual distances it moves
[01:55:31] <SWPadnos> that also makes other things make sense, for example acceleration is then in inches/second^2 instead of some weird units nobody can figure out
[01:55:55] <tom3p> WUNCFO's
[01:56:08] <SWPadnos> there are debugging tools that let you see exactly what's happening inside EMC2 and the hardware/realtime layer - a scope and multimeter basically
[01:56:19] <SWPadnos> it's easier when the numbers you see actually mean something :)
[01:56:29] <morfic> the mesa boards that together with daughter boards drive servos, also do the input/output for the ladder stuff? limit switches/toolchanger counts/* ?
[01:56:36] <pfred1> http://www.youtube.com/watch?v=mk7VWcuVOf0
[01:57:04] <SWPadnos> Mesa has different daughtercards for generic I/O
[01:57:29] <SWPadnos> any I/O point on the hardware can be connected to any ladder I/O, or any other internal signal
[01:57:42] <morfic> right, one daughter board for drives, one for generic i/o
[01:57:45] <SWPadnos> (subject to limitations like you can only connect a bit to a bit)
[01:57:47] <SWPadnos> yep
[01:57:56] <SWPadnos> the servo board also has encoder inputs
[01:58:10] <SWPadnos> which don't have to be used with the analog outputs, incidentally
[01:58:44] <morfic> which is the most versatile any I/O board then that works with emc2?
[01:58:46] <SWPadnos> you can use the fourth analog output to run the spindle, and use the encoder for a handwheel if you like (assuming you don't need spindle feedback)
[01:59:03] <SWPadnos> probably the 5i22, since it has 4 connectors
[01:59:32] <morfic> i am going overboard on the $ figure for wednesday, i think i can afford it, seeing some of the other totals
[01:59:45] <SWPadnos> they're pretty similar though, the $200 5i20 or $220 5i23 are just fine too
[02:00:11] <SWPadnos> the 5i22 is $500 or so, for the high end one (with the high gate count FPGA)
[02:00:29] <SWPadnos> oh no, $429
[02:01:11] <morfic> 1M or 1.5M gates?
[02:01:15] <SWPadnos> but for that price, you can buy 2 5i20, or close to 2 5i23, and get 6 I/O connectors (192 I/O points, if they were all used for bits)
[02:01:20] <SWPadnos> 1.5M
[02:01:23] <SWPadnos> 1M is $369
[02:02:10] <morfic> no i mean what would i need, seeing how 5i20 is what 200K? 1M seems huge
[02:02:17] <SWPadnos> it is
[02:02:39] <SWPadnos> can't answer what you need, since we don't know what you want to hook up yet :)
[02:02:45] <morfic> so 5i22-1M would be plenty,
[02:02:51] <SWPadnos> but I can tell you that the 5i20 can run 12 servos, if you want to use that config
[02:03:25] <morfic> in that config, is there all IO used up by the servos?
[02:03:43] <SWPadnos> yes
[02:03:57] <tom3p> got estimate of i/o points needed?
[02:04:09] <SWPadnos> you get up to 4 servos on a connector, or 24 digital I/Os, or whatever the attached daughtercard gives you
[02:04:35] <pfred1> SWPadnos they're making one of these: http://www.boxofficeprophets.com/images2/wild091108.jpg
[02:04:52] <morfic> 7i33 is a quad board, this is where the 12 servos comes from? 3 daughter cards on one 5i20?
[02:04:55] <SWPadnos> heh
[02:04:59] <SWPadnos> yes
[02:05:13] <SWPadnos> have you looked at the Mesa website?
[02:05:24] <morfic> i am flipping between that and here right now
[02:05:34] <SWPadnos> heh, ok
[02:05:37] <morfic> just you talk more my language
[02:06:03] <morfic> when do i need the gate count to be high?
[02:06:04] <SWPadnos> yeah, there's not a ton of EMC2 specific info on their wbsite
[02:06:20] <morfic> like one 5i22 vs 2 5i20 ?
[02:06:22] <SWPadnos> if you want the FPGA to do advanced stuff
[02:06:33] <morfic> define advanced stuff
[02:06:35] <SWPadnos> which isn't really necessary at this point
[02:06:58] <SWPadnos> well, Mesa makes a motion processor they call SoftDMC (not supported by EMC2), which uses some gates
[02:07:28] <pfred1> SWPadnos did you see this? http://www.youtube.com/watch?v=quN37YskoaM&feature=fvw
[02:07:35] <SWPadnos> it's also possible to make big FPGA configs that have loads of stuff you'd end up turning off, like having 24 step generators and 24 PWM outputs and a dozen encoder inputs, etc.
[02:07:48] <SWPadnos> yep. it's pretty cool
[02:08:00] <pfred1> SWPadnos I still can't get over it
[02:08:26] <SWPadnos> what might be cool, but I don't think anyone is working on it, would be to have at least part of classicladder, or a reduced logic-only PLC, running in the FPGA
[02:08:52] <SWPadnos> the basic deal is that you don't need the gates now, but it's possible that they'll come in handy later on
[02:09:08] <morfic> pfred1: ohh, i never clicked that link, thought the 4 axis hand was cool
[02:09:37] <pfred1> morfic I had no idea people were doing anything quite like this until i saw this video
[02:09:45] <SWPadnos> I think I would recommend spending the extra $30 on the 5i23 though, since it's a bigger/faster FPGA, and it's a later family which is still supported by the free Xilinx webpack software
[02:09:48] <morfic> SWPadnos: i could put things that run on the pc into the fpga ? (again, put very simplified)
[02:10:01] <SWPadnos> if you know how to program FPGAs, yes
[02:10:07] <SWPadnos> it might not get done for you ...
[02:10:23] <morfic> never touched fgpas...knowingly
[02:11:22] <morfic> pfred1: the left upper foot seems to slide about, but it does not seem to lose position o_O
[02:11:48] <pfred1> morfic its absolutely insane
[02:12:32] <morfic> 8i20 is an amp in case we ever replace the stuff daewoo put in there and 2.2KW is enough for our servos (wuld love to say it isn't but i would also love to be wrong)
[02:12:36] <morfic> SWPadnos: ^
[02:12:58] <SWPadnos> morfic, it's not simple to put PC software into an FPGA
[02:13:20] <ds3> given a big enough FPGA, it might not be too hard
[02:13:22] <ds3> ;)
[02:13:52] <SWPadnos> so it's possible to put some functions that are now done by the PC software into the FPGA (like the PWM and step generators, as well as encoder counters), only some things really lend themselves to it, or provide an improvement by being in the FPGA
[02:14:19] <ds3> or make a SoC with on board flash and put that on a FPGA
[02:14:24] <SWPadnos> 2.2KW may be enough, but I think your servos may be AC, not DC
[02:14:24] <morfic> SWPadnos: but like you said ladder stuff done in there would be a use (if it gets actually done or not)
[02:14:40] <SWPadnos> it would be nice, and it wouldn't fit in a 5i20
[02:14:57] <SWPadnos> but as I said, I don't know if anyone is even thinking about working on it
[02:15:18] <pfred1> SWPadnos if you off loaded the step generation would EMC just put out timed high to low signals then or something?
[02:15:30] <morfic> SWPadnos: that would be if they had to go/are dead and we look at what to do, or when the current drives die, might be cheaper to buy a new drive/servo combo than one of them replaced with what we have
[02:15:31] <SWPadnos> you'd get a 5i23 or 5i22 to "future-proof" your investment, for the most part
[02:15:42] <tom3p> working with whats available already is good, dont get your tit in a wringer over a cool idea
[02:15:53] <SWPadnos> heh
[02:15:57] <morfic> SWPadnos: yeah, i won't bring up the future proofing
[02:16:17] <SWPadnos> pfred1, the Mesa cards get a step frequency, and output that
[02:16:36] <pfred1> morfic looking at the industry's historic track record I can't see why one can ignore the future with tech
[02:16:42] <SWPadnos> yeah, Fanuc replacement parts are pretty expensuve
[02:16:48] <SWPadnos> expensive
[02:17:03] <morfic> tom3p: nice way of putting it, i am more interested in what we CAN do, than tell my "takes no bullsh#t" boss what *COULD* be
[02:17:14] <tom3p> it must be the lack of dead fanucs to rape for parts
[02:17:31] <tom3p> heh
[02:17:37] <tom3p> i kmow hat guy
[02:17:40] <morfic> pfred1: not ignoring, but not saying this *will* come when it is a *maybe* at best
[02:17:43] <tom3p> know that guy
[02:17:45] <pfred1> morfic oh i can guarantee yo uthat companies will go under and or charge crazy prices for "obselete" components eventually
[02:17:53] <SWPadnos> the 5i23 is a no-brainer to me - double the gates for $30
[02:18:14] <SWPadnos> the 5i22 is more of a "what if" kind of thing, at twice the price
[02:18:17] <pfred1> morfic and "obselete" may come as soon as a year with some stuff!
[02:18:30] <morfic> tom3p: he already told us to tell whoever we email with (on the windows side of things) "my boss is an asshole and he wants answers, fast" to speed things up, he is good at the good cop/bad cop game
[02:19:31] <morfic> 5I23 it is, $229 is nothing compared to what it goes against
[02:20:07] <tom3p> ( get a spare, tell the boss everyhthing was doine redundant at 1/2 the cost :)
[02:21:57] <morfic> my hope is we keep spares
[02:22:10] <morfic> i need a free cd, i want to boot up the livecd and play with sim at work
[02:22:33] <morfic> run it on office machines of various spec and see what latencies i get
[02:22:39] <atmega> I think I ordered a pair of 7i43's and a pair of terminal boards last week, but no confirmation.
[02:22:47] <morfic> should help me get the computer the right size
[02:23:27] <morfic> i guess getting good feedback/response/service from mesa would be a good start too
[02:23:39] <atmega> their web site blows
[02:24:06] <tom3p> they're in here a lot & very helpful. check the old logs
[02:24:15] <pfred1> morfic low latency seems to not have as much to do with CPU speed as it does with other system overheads always
[02:25:14] <pfred1> atmega I just got my Digi-Key confirmation of shipping email :)
[02:26:11] <pfred1> ha and its already in the tracking system too!
[02:26:31] <atmega> the digikey min. order is kind of annoying
[02:26:53] <pfred1> whats it like $25 ?
[02:27:05] <atmega> I think $10
[02:27:20] <atmega> but, last time I just needed one 10k thermistor
[02:27:32] <pfred1> I mean come on you can't find $10 of parts you could use?
[02:27:48] <pfred1> I blow more than $10 for lunch
[02:27:54] <atmega> I did, but...
[02:28:19] <pfred1> I just buy parts to fry when i order I figure I probably will initially
[02:28:19] <atmega> I think I paid for 2nd day delivery too
[02:28:43] <atmega> like $24 total for a $0.95 thermistor
[02:28:54] <pfred1> keeps the pressure down prototyping knowing i have twice as many as I really need
[02:29:03] <morfic> pfred1: i am looking at integrated gfx vs dedicated gfx, vs qualities of north and south bridge chips
[02:29:24] <pfred1> morfic integrated gfx and emc2 is usually a big no-no
[02:29:29] <morfic> we had all dells, until they hired so quick, they picked up machines from best buy, should give me an interesting spread of specs
[02:29:48] <morfic> pfred1: i would like to show my boss how big a no no :)
[02:30:01] <morfic> dedicated gfx cards are still cheap
[02:30:24] <pfred1> I recall reading a $6 card helped in one case
[02:30:44] <atmega> if you have a real machine and use it to generate real income, buy a real computer
[02:31:07] <SWPadnos> I can recommend a Core2 Duo machine from Acrosser
[02:31:23] <SWPadnos> they're $800 or so as I have them configured, but they do quite nicely for EMC2
[02:31:29] <pfred1> atmega emc2 latency numbers cannot really be guaranteed by just throwing money at a problem at least not right off
[02:31:33] <tom3p> yes, there are some tested motherboards that you can get , known latencies
[02:31:56] <SWPadnos> the small case also has a PCIe slot and a PCI slot, so you can add a nice video card if you have to, in addition to one Mesa card
[02:32:13] <morfic> http://www.acrosser.com/ them?
[02:32:18] <SWPadnos> yep
[02:32:26] <SWPadnos> AR-ES0892 I think
[02:32:39] <pfred1> tom3p I've seen that chart and there is no guarantee anyone posting all of that information knew what they were doing etc.
[02:32:53] <SWPadnos> that is the limitation though - there's one PCI slot, so no adding another Mesa card to that one
[02:33:05] <atmega> it's not just throwing money at it... buying a known decent setup for say $800 is a whole lot cheaper than trying to reuse a 'free' computer and me having to spend 4 hours screwing with it.
[02:33:29] <morfic> atmega: i will make sure it's not a "free computer"
[02:34:05] <pfred1> morfic when you go PC shopping bring the emc2 live CD with you
[02:34:26] <pfred1> benchmark it on the spot
[02:34:46] <morfic> SWPadnos: the overall speed of the computer has much or little effect on max speed machine can cut at with emc2? (where speed X is between 0 and the hardware's current Vmax)
[02:34:49] <atmega> need a bootable usb drive for latency checks
[02:35:08] <pfred1> atmega why is that?
[02:35:16] <SWPadnos> it's funny. I had the same thought, but for an Embedded Systems Conference - trouble is none of the little embedded boxes have CD drives :)
[02:35:23] <morfic> reminds me, i think my thumb drive didn't look so good when i pulled my keys out earlier
[02:35:23] <SWPadnos> little
[02:35:59] <SWPadnos> since you're planning on using hardware PWM generation, you don't even need very good latency numbers
[02:36:25] <SWPadnos> you could get by with about 100 us of latency, if you plan to use the default 1 ms servo cycle
[02:36:54] <SWPadnos> that's another advantage of using something like the Mesa or Pico products, instead of the parallel port :)
[02:37:27] <morfic> [157413.893287] sdd: sdd1 <-- thumb drive still lives, just barely though
[02:37:42] <pfred1> I noticed when i scoped my parallel port with emc2 generating step pulses here the signal was low then just had high spikes in it is that normal or do I have something misconfigured on that system?
[02:38:18] <pfred1> like it looked rather like this ______|________|________|________
[02:38:32] <SWPadnos> that's correct
[02:38:37] <pfred1> yeah it did work
[02:38:48] <SWPadnos> if you need negative step pulses, you need to invert the step output
[02:38:54] <pfred1> but I just like my clocks more squarish in stuff i make is all
[02:39:03] <SWPadnos> the step pulse width is the steplen setting
[02:39:10] <SWPadnos> it's not a clock
[02:39:12] <pfred1> ah
[02:39:30] <pfred1> SWPadnos it seems like a clock signal to me
[02:39:48] <SWPadnos> heh, sort of, but the phase is unimportant - it's only the active edge that matters
[02:39:59] <SWPadnos> so duty cycle is irrelevant, as long as all the timing needs are met
[02:40:37] <pfred1> I've noticed my motors run a tad better with even clocks is all they run OK on these pulses too
[02:41:04] <SWPadnos> strange
[02:41:43] <pfred1> it is subjective on my part just listening to them but emc2 runs them well enough
[02:42:17] <pfred1> when i built them I just used a 555 timer to generate step pulses
[02:42:44] <pfred1> and with that of course i was able to set the pulses square
[02:43:13] <tom3p> acrosser had to be taiwan despite all the round eyes on the website
[02:43:39] <pfred1> tom3p taiwan is the good stuff anymore
[02:44:00] <tom3p> i believe so, wasnt knocking it, work there a lot
[02:44:03] <pfred1> heck I can remember when jamanese meant chinese quality
[02:44:21] <pfred1> back like in the early 70s
[02:44:26] <morfic> i hope i remember how to use cdrecord, been so long
[02:44:44] <tom3p> linux box?
[02:44:50] <atmega> wow, you must be *OLD*
[02:45:04] <pfred1> come on is 45 that old?
[02:45:18] <atmega> 45? and you can use a computer?
[02:45:24] <morfic> tom3p: me? yeah seems it's burning :)
[02:45:42] <morfic> is 37.5 yrs too old?
[02:45:46] <atmega> actually, I have you beat by a year or so :)
[02:45:48] <pfred1> atmega well they weren't around when I was a kid but I built my first one out of chips
[02:45:59] <tom3p> morfic good, wouldnt be hard to get some higher level tools, but if its cooking, just wait
[02:46:32] <morfic> tom3p: i didn't want to wait for k3b or so to compile, dolphin unlike nautilus did not offer anything in right click menu
[02:46:37] <SWPadnos> pfred1, the 555 would have not only square pulses, but also infinite timing resolution
[02:46:54] <morfic> tom3p: i don't mind CLI, just rusty, not burning many isos is problem
[02:46:56] <SWPadnos> emc2 can only change the inter-step time by a full base period
[02:47:18] <tom3p> early home computers were from the USA ( my cosma elf, my 8080, ...) hmmm we gave it away again :(
[02:47:19] <pfred1> SWPadnos 555 duty cycle is adjustable
[02:47:32] <SWPadnos> yes, infinitely
[02:47:37] <SWPadnos> just like the frequency
[02:48:02] <SWPadnos> so you get a smoother pulse train than EMC2 can generate - there's no RT latency to deal with using a 555
[02:48:06] <pfred1> SWPadnos well I've just noticed any motor drivers i build sort of like them even is all
[02:48:14] <pfred1> SWPadnos they like 50%
[02:48:34] <SWPadnos> have you tried setting a 555 circuit to 10-20%?
[02:48:39] <pfred1> sure
[02:48:40] <SWPadnos> it should sound about the same
[02:48:43] <morfic> SWPadnos: just because the price seems so low, the 7I33 for 4 servos at $79 is it, no several benjamin daughterboard??
[02:48:48] <SWPadnos> oh, well then I'm back to "that's odd"
[02:48:59] <SWPadnos> $79 is it
[02:49:00] <tom3p> 555 no smoother than mesa, tho, should be nailed ( i think )
[02:49:09] <SWPadnos> tom3p, correct
[02:49:10] <morfic> opencnc guys blew initial quite up to $50K, so the current price tag is low, camsoft is around $20K
[02:49:22] <SWPadnos> a little smoother, but not enough that a slow stepper motor would care
[02:49:39] <SWPadnos> oh. then I'll sell you a tested PC with Mesa cards :)
[02:49:48] <SWPadnos> I'm sure I can still beat $20k, even with EMC2 installed
[02:50:02] <tom3p> things to embed inside the anything i/o card,, the on& offtime generator for EDM :)
[02:50:08] <SWPadnos> heh
[02:50:14] <SWPadnos> and THC, what the heck
[02:50:21] <tom3p> yah
[02:50:35] <tom3p> wait, gona think on that one
[02:50:52] <morfic> SWPadnos: i think this is where bosses idea of "then we get it programmed" is coming from
[02:50:53] <SWPadnos> no need for floating point - ADC read counts
[02:51:00] <SWPadnos> OK by me
[02:52:19] <morfic> are you the only active dev in here btw? and cradek's host says board of directors, not sure how much dev that means
[02:52:35] <morfic> going to change xchat to show all hostmasks real quick
[02:52:44] <pfred1> I can't believe UPS ground is $10.88 on a .7 pound package!
[02:53:02] <morfic> brb
[02:53:10] <tom3p> use USPS as often as possible
[02:53:38] <pfred1> tom3p most places do not give it as an option
[02:53:39] <Valen> morfic: I'd use the intel atom boards
[02:53:55] <Valen> latency is pretty good, mostly solid state and $100 or so
[02:53:56] <SWPadnos> oh hmm. mine should say that too. one sec
[02:54:07] <Valen> we and a few others are using them and they seem pretty good
[02:54:08] <mozmck2> I think cradek does plenty of dev. He and jepler wrote AXIS...
[02:54:08] <SWPadnos> weird
[02:54:24] <SWPadnos> yes, I'm one of the developers because I have contributed code ever
[02:54:43] <SWPadnos> I haven't done much lately, and cradek and jepler (and jmkasunich and others) have done waaaaaay more overall too
[02:55:02] <mozmck2> heh, I even did a tiny bit!
[02:55:04] <SWPadnos> all those three, plus me and alex_joni are the board
[02:55:05] <tom3p> elvis has left the building (morfic is gone)
[02:55:10] <SWPadnos> yep - mozmck too :)
[02:55:27] <Valen> I saw his brb, not an exit?
[02:55:35] <SWPadnos> me too
[02:55:46] <tom3p> ok, elvis has closed his eyes
[02:55:59] <Valen> I figured he'd get it when he gets back
[02:56:15] <Valen> RAID 1, SSD's for disk ;->
[02:56:46] <pfred1> I better get on the stick over here designing a board for these driver chips before they arrive
[02:57:22] <tom3p> i think mouser has USPS, and soemtimes when i compare, they get the bill over digikey
[02:57:30] <pfred1> I am stoked for them they seem like the best out there for the price
[02:57:33] <Valen> pfred1 yaknow sparkfun have some offshoot that'll make boards in onsies, as i recall the rates were fairly cheap
[02:58:06] <grommit> Has anyone attended the CNC Workshop in the past? I am curious what EMC Labs is about...
[02:58:08] <SWPadnos> there's Olimex as well
[02:58:12] <pfred1> Valen I know but until I've prototyped a circuit and know it works I don't see commiting to board art as very intelligent
[02:58:13] <SWPadnos> (or maybe that's the one)
[02:58:24] <morfic> tom3p: i never really leave
[02:58:28] <Valen> depends on how you value your time vs your money
[02:58:30] <morfic> i don't reboot the sheeva much
[02:58:31] <SWPadnos> grommit, yes, I've been to the last few CNC workshops
[02:58:35] <cradek> a lot of us here have gone to several of them
[02:58:59] <pfred1> Valen time and money have nothing to do with it i just think it is stupid to commit to board art before all parts and values are knowns
[02:59:04] <cradek> this year's is at a new location and none of us really know what it's going to be like
[02:59:15] <morfic> three directors and one dev i see
[02:59:35] <cradek> I'm much more dev than director, fwiw
[02:59:43] <SWPadnos> heh
[02:59:48] <cradek> I only direct when I absolutely have to
[02:59:51] <SWPadnos> and I seem to be the other way around :)
[02:59:53] <Valen> I'm usually fairly sure what i'm making is going to work before I actually make it
[02:59:53] <grommit> Have you heard of EMC Labs (as one of the seminars)?
[03:00:02] <cradek> not by that name
[03:00:23] <cradek> what have you heard? we've had some classes and stuff.
[03:00:25] <SWPadnos> grommit, I bet that's just what they're calling the "hands-on" class they're planning
[03:00:36] <pfred1> Valen what I make usually works right off too but I like htings to work as well as possible before i am done with them
[03:00:46] <morfic> cradek, only reason i ask is since *IF* my boss does decide to go for emc, he would like to pay actual project devs to implement (what we consider to be) missing bits
[03:00:46] <grommit> ok, that is kind of what I was hoping for...
[03:00:50] <SWPadnos> as far as I know, there's going to be one "intro to EMC2" class, taught on two days
[03:01:02] <SWPadnos> (the same class - a repeat AFAIK)
[03:01:16] <pfred1> Valen and that often times means a few changes here and there
[03:01:24] <grommit> I need a little more then intro, I need an advanced class for dummies ;-)
[03:02:14] <pfred1> morfic any programmer in a storm ...
[03:02:35] <cradek> morfic: I've done some work like that. Drop me an email if you'd like to talk about it. But that's not the only way (or the most common way) things get added.
[03:03:13] <pfred1> cradek you devs aren't holding back any secret stash of code now are you?
[03:03:20] <SWPadnos> no comment
[03:03:22] <morfic> cradek: your email is?
[03:03:37] <cradek> pfred1: don't be silly
[03:03:41] <grommit> There's a secret stash of code in the development branch :-)
[03:03:47] <cradek> morfic: chris@timeguy.com
[03:03:51] <pfred1> cradek why not? I am silly by nature :)
[03:04:04] <SWPadnos> there's a secret stash of code in my head, but it's so secret I don't know what it is
[03:04:16] <mozmck2> I've got a secret stash, but it doesn't do anything yet :)
[03:04:23] <Valen> morphic yeah the Intel D945GCLF2 mini-ITX works well for us
[03:04:25] <SWPadnos> I wonder why my cloak isn't working
[03:04:26] <grommit> In fact, I was pleasantly surprised that there is the beginnings of a configuration gui for Mesa cards.
[03:04:26] <morfic> need to head to bed now, need to start my work week a day early tomorrow
[03:04:31] <Valen> ~4000 latency out of the box
[03:04:32] <grommit> I kind of even worked....
[03:04:43] <tom3p> if i'd tell myself, i'd have to kill me
[03:04:46] <morfic> SWPadnos: wanna share your email too, better to have two people who are interested than one
[03:04:58] <SWPadnos> heh. sure. my business email is thoth@sover.net
[03:05:45] <morfic> thanks all, 'night
[03:05:49] <Valen> cya
[03:05:51] <tom3p> gnite
[03:06:02] <L84Supper> AMD released the coreboot patch for 7xx chipsets so SMI free mainboards using AMD 7xx and 8xx chipsets are around the corner
[03:06:03] <pfred1> SWPadnos it looks like it is working here
[03:06:24] <SWPadnos> great. what's working?
[03:06:29] <cradek> haha
[03:06:32] <pfred1> SWPadnos your cloak
[03:06:35] <SWPadnos> oh
[03:06:46] <SWPadnos> maybe my client knows too much :)
[03:06:52] <pfred1> SWPadnos 23:05 -!- SWPadnos [~Me@emc/developer/SWPadnos]
[03:07:13] <mozmck2> I like your real name: No project too simple to make too complex
[03:07:16] <SWPadnos> strange. that should be board-of-directors, like cradek
[03:07:20] <SWPadnos> heh
[03:07:30] <SWPadnos> been my motto for years :)
[03:08:13] <pfred1> I wish i could remember the movie where Don Johnson utters the classic line, "Nothing starts out complicated, it just gets that way."
[03:08:32] <L84Supper> has anyone ever used the TwinCAT windows automation software by Beckhoff?
[03:08:33] <SWPadnos> well, he's only been in Miami Vice, so it must be that :)
[03:08:35] <mozmck2> I have a collection of one-liners and one goes something like this: For every complex problem there is an answer that is obvious, simple, and wrong.
[03:08:47] <SWPadnos> yep, that's a good one too
[03:08:48] <pfred1> SWPadnos nope he has been in a few movies
[03:08:59] <SWPadnos> huh. who'd a thought?
[03:09:09] <pfred1> one of them is one of my favorites but not because he's in it
[03:09:36] <pfred1> I happen to like the author of that story
[03:09:54] <SWPadnos> oh, :I forgot he was in "A Boy And His Dog"
[03:10:03] <pfred1> thats the one
[03:10:10] <SWPadnos> yep, that's a good one
[03:10:10] <pfred1> but he didn't say that line in that movie
[03:10:13] <mozmck2> anyone else getting a connection refused trying to use the RTAI cvs?
[03:10:14] <Valen> has anybody tried emc on a coreboot machine?
[03:10:24] <mozmck2> what is coreboot?
[03:10:29] <pfred1> I've read most of everything Ellison has ever wrote
[03:10:41] <SWPadnos> coreboot is what became of the LinuxBIOS project
[03:10:56] <pfred1> he has a wry sense of humor that agrees with me
[03:10:56] <SWPadnos> very fast booting to Linux
[03:10:59] <SWPadnos> heh
[03:11:10] <SWPadnos> and he did some of the best Star Trek epsiodes too
[03:11:19] <L84Supper> yes coreboot is what LinuxBIOS evolved into
[03:11:26] <pfred1> he's an SOB but a great euthor
[03:11:27] <Valen> I'm starting to like AMD more
[03:11:47] <L84Supper> AMD has been pretty open source friendly
[03:11:53] <pfred1> Valen get 'em while you can Intel is burying them!
[03:11:54] <Valen> I really prefer intel and nvidia hardware, but AMD seems to be becoming way more open source latley
[03:12:21] <L84Supper> nvidia dropped all of its open source driver support
[03:12:32] <L84Supper> binary only for nvidia now
[03:12:48] <mozmck2> I've had better latency results with AMD so far, and they are sure cheaper for the speed it seems.
[03:12:59] <SWPadnos> too bad the SuperMicro H8DCE isn't supported
[03:13:03] <Valen> I'm all intel/nvida in everything I do at the moment but if AMD can come up with something like an atom and core2 series in terms of performance per watt I'd be interested
[03:13:06] <L84Supper> oh and dual core ARM boards with PCIe are due out later this summer
[03:13:06] <pfred1> L84Supper you realize that Nvidia can't do anything about that though right?
[03:13:37] <SWPadnos> the overall cost is usually less, because the CPU (for a given speed) is around the same price (usually a little lower), and the motherboard is generally a lot less
[03:14:01] <L84Supper> pfred1: yeah, so it will be reverse engineered open drivers for nvidia
[03:15:00] <pfred1> L84Supper to be #1 you have to sometimes trad on paths you'd rather not, but do anyways to get where you're going to
[03:15:04] <pfred1> tread even
[03:15:10] <mozmck2> SWPadnos: ever do any D programming?
[03:15:17] <SWPadnos> nope
[03:15:31] <L84Supper> quad core ARM's are due later this year, 1-2GHz per core, the current dual outperform the Atom's
[03:15:51] <pfred1> Intel has 4 generations waiting in the can
[03:15:57] <pfred1> last i heard
[03:16:09] <pfred1> AMD is that far behind
[03:16:37] <mozmck2> looks like D is getting a little more popular. I keep running into it more.
[03:16:42] <L84Supper> AMD released the 8 and 12 core Opterons last week, we're using them in the next >1petaflop supercomputer
[03:17:04] <pfred1> yeah but they can't afford to build new fabs like Intel can
[03:17:21] <SWPadnos> they've gone fabless, so that doesn't really matter much
[03:17:37] <L84Supper> it uses more of the HD59xx gpu's than Opterons to get >1 petaflop
[03:17:38] <pfred1> SWPadnos it does when yo uwant to break new tech
[03:17:41] <SWPadnos> (the CPU department contracts with the manufacturing department, or someone else, to make the chips)
[03:17:48] <SWPadnos> yes, that's true
[03:18:06] <pfred1> SWPadnos and intel has it waiting in the wings but won't release because they really don't need to
[03:18:18] <SWPadnos> I don't believe that actually
[03:18:27] <pfred1> like Intel cares what you believe
[03:18:31] <SWPadnos> (not saying you're lying, it's just implausible)
[03:18:33] <Valen> intel would like nothing more than to wipe AMD off the planet
[03:18:53] <SWPadnos> I don't much care what Intel believes either actually
[03:18:54] <pfred1> SWPadnos what is so implausible about not wanting to drop 2 billion if you don't have to?
[03:19:02] <L84Supper> Intel is even using TSMC now
[03:19:14] <SWPadnos> well, it takes a moderate sized team years to develop a CPU these days
[03:19:33] <SWPadnos> so having 4 generations "in the can" means that they would have already invested billions in the design
[03:19:34] <pfred1> SWPadnos and it takes over 2 billion USD to start up a new fab plant too
[03:19:49] <SWPadnos> that's assuming that you need a new fab for a new CPU, which isn't always the case
[03:19:58] <pfred1> no new tech
[03:19:59] <SWPadnos> that's only necessary when you go to a new process
[03:20:20] <tom3p> ooh hailstorm
[03:20:29] <SWPadnos> cool!
[03:20:50] <tom3p> time to make sharbat-el whatever iranian ice cream
[03:21:09] <KimK> Hi all. Is there an Ubuntu app for viewing the .rdf file IRC logs of #emc? Google and Synaptic failed me on this. I see that RDF is like XML, but it seems like this is intended to be viewed as "formatted", and not with just another XML editor. Any suggestions?
[03:21:28] <cradek> I don't know what the heck those files are
[03:21:32] <SWPadnos> I'd view the text files :)
[03:21:33] <tom3p> richtext
[03:21:46] <pfred1> like old write files?
[03:21:58] <L84Supper> I was down at the CME earlier today, SMI kills the clusters they use for trading, just like SMI interferes with EMC
[03:21:58] <pfred1> KimK have you tried Open Office?
[03:22:02] <cradek> yeah there's two for each day: the text file, and the screwy one you don't want
[03:22:02] <SWPadnos> it's not quite rich text, because it has tags like "user", not "color"
[03:22:29] <SWPadnos> L84Supper, what's the C stand for?
[03:22:35] <SWPadnos> Chicago?
[03:22:39] <L84Supper> SMI needs to die but Intel, AMI and Phoenix still have a death grip on it
[03:22:51] <L84Supper> yeah, Chicago mercantile Exchange
[03:23:14] <L84Supper> SMI actually holds up trading
[03:23:23] <SWPadnos> SMI is required for all Intel CPUs since the P4, or you're operating the chip out of spec (that's how some of the thermal throttling is handled)
[03:23:45] <tom3p> like the guys who have seats down there cant afford a better computer
[03:23:52] <SWPadnos> heh
[03:24:16] <L84Supper> heh, this is for the actual trading in the CME vs the traders
[03:25:05] <L84Supper> they have a sea of 1U x86 servers, they used to use HP lockstep dual core MIPS
[03:26:15] <L84Supper> I'm not sure how they handle parity errors now, maybe after 2 decades of trading derivatives it just doesn't matter
[03:26:32] <SWPadnos> heh
[03:26:41] <SWPadnos> there's no parity in that system, so why bother checking
[03:26:58] <KimK> pfred1: I have now. OOo opens it like an XML editor. But maybe I'm missing a formatting file (something similar to a .css file, maybe?) that the poster just assumes that I already have?
[03:27:05] <pfred1> they farm it out to Enron
[03:27:10] <L84Supper> they probably fired the regulators anyway :)
[03:27:30] <SWPadnos> there were regulators?
[03:27:37] <SWPadnos> huh. who knew?
[03:27:51] <pfred1> SWPadnos sure the regulators rode with Billy The Kidd
[03:28:13] <SWPadnos> oh. so their bones were doing the actual regulatory oversight. that explains a lot.
[03:28:16] <L84Supper> we'll they do check your ID and have a metal detector on the way in
[03:28:49] <pfred1> L84Supper those day traders can never bee too careful around them!
[03:28:49] <L84Supper> guns are passe anyway
[03:28:57] <Valen> KimK tried it in gedit?
[03:28:59] <KimK> * KimK would like to open the "screwy one you don't want", LOL
[03:29:40] <tom3p> KimK 800+ rdf tools http://www.mkbergman.com/new-version-sweet-tools-sem-web/
[03:30:26] <tom3p> or http://rhodonite.angelite.nl/
[03:30:29] <pfred1> I thought someone made plastic guns for metal detector environments?
[03:30:55] <L84Supper> SWPadnos: does Atom use SMI?
[03:31:03] <SWPadnos> L84Supper, I don't know
[03:31:12] <SWPadnos> it's sort of core2, so probably
[03:31:35] <L84Supper> those Atom mini-itx had really low jitter
[03:32:18] <SWPadnos> yes, they're quite nice little boards
[03:32:31] <L84Supper> $65 is hard to beat
[03:32:33] <SWPadnos> only one PCI slot though
[03:32:54] <tom3p> glock
[03:32:56] <L84Supper> the dual core ARM will have 2 PCIe x4
[03:32:59] <Valen> only need one for a mesa card ;->
[03:33:23] <SWPadnos> for one mesa card, yeah :)
[03:33:38] <pfred1> L84Supper free is a tough act to follow too though
[03:33:53] <grommit> I am using the Intel D945GCLF2 mini-ITX (just getting started) and it works great
[03:34:01] <SWPadnos> yep, that's the one
[03:34:03] <mozmck> L84Supper: where is some info on that dual core ARM? who's making it?
[03:34:06] <grommit> I have my Mesa 7i43 connected vi EPP
[03:34:26] <L84Supper> moznck: they use Marvell ARM soc's
[03:35:14] <L84Supper> mozmck sorry
[03:35:14] <Valen> I wonder if the new atom boards with the non crap northbridge will be any good
[03:35:20] <KimK> Thanks for your advice, that helps. Oops, supper. If anyone thinks of anything else, please write it. I'll be back a little later. Thanks all!
[03:35:56] <L84Supper> any news on RTAI for ARM?
[03:37:19] <L84Supper> some of the Jetway min-itx using VIA would have extra irq lines so that you could use a 2 PCI slot riser, do any of the Atom boards do the same?
[03:38:02] <mozmck> I think rtai works fine on some ARM processors, but there are a lot of different ones.
[03:39:06] <grommit> jitter (on D945GCLF2 ) is less then 14,000 (if running lots of apps), usually < 9000 if just running EMC
[03:39:39] <Valen> grommit: if you use the SMP you get ~4000 or so
[03:40:08] <grommit> Yeah, I wasn't able to build the kernel correctly for some reason and didn't spend too much time with it
[03:40:31] <Valen> you don't need to build the kernel, somebody made a pre-packaged one that worked fine
[03:40:38] <Valen> all you needed to do was compile emc
[03:40:50] <grommit> I didn't see that (apparently)
[03:41:00] <Valen> lemme see
[03:42:02] <Valen> its somewhere in the wiki
[03:42:03] <SWPadnos> KimK, RDF is meant to be a machine readable format. It's like RSS (in this case, I think), which you would never want to look at yourself. I don't know if there's any point to looking at that file, since there is no information there that isn't also in the text file
[03:42:05] <grommit> TBH, since I am using the Mesa for stepgen and encoders I don't really care all *that* much about latency. But having smp working would be nice just for general perf.
[03:42:06] <L84Supper> 3D support for pre ATI HD GPU's is a mess with Ubuntu 10.04, the Ubuntu devs don't really get DRM yet
[03:42:28] <SWPadnos> that said, you can look at it with Kate (the KDE text editor), which will let you collapse various XML elements
[03:42:37] <L84Supper> has anyone started on EMC2 for 10.04 yet?
[03:42:49] <SWPadnos> L84Supper, mozmck has been working on it
[03:42:58] <SWPadnos> kernels and CD
[03:43:09] <Valen> SWPadnos: SMP by default?
[03:43:12] <SWPadnos> (thanks mozmck)
[03:43:12] <mozmck> got a kernel compiling right now.
[03:43:15] <SWPadnos> dunno
[03:43:22] <mozmck> I'm shooting for that
[03:43:27] <L84Supper> mozmck: if you need any help I have someone here with time
[03:43:41] <Valen> I'll beta test on an atom
[03:44:11] <mozmck> SMP that is. I've had pretty good success running SMP generic kernels for karmic on single and dual core computers.
[03:44:17] <tom3p> there is no point in having rdf files alongside the text files ( except to piss off the mouse challenged file chooser )
[03:44:42] <Valen> I figured it was some kind of rich document format, or some such
[03:45:18] <SWPadnos> that's probably true
[03:45:25] <mozmck> L84Supper: thanks. The hardest part sometimes is just figuring out how to do everything.
[03:45:26] <SWPadnos> though they can be used like RSS feeds
[03:46:15] <L84Supper> moxmck: esp with ubuntu with their undocumented configs at times
[03:46:45] <L84Supper> they even modified Grub2!
[03:47:25] <mozmck> yep. I think I finally have the basics down of making kernel packages the ubuntu way. Some of the information on the web is wrong. finally got some guys on ubuntu-kernel to tell me the right way.
[03:47:28] <Valen> grub2 was hard enough
[03:47:41] <mozmck> didn't know that. grub2 is somewhat of a pain.
[03:48:13] <Valen> I'm yet to see an upside to grub 2, the UUID thing I could see a reason for, and now I acutally use it
[03:48:22] <mozmck> I have a feeling plymouth is going to cause problems, and apparently you won't be able to uninstall it.
[03:49:33] <Valen> again, not really seeing the need for it, I'd rather they worked on just making boot faster than trying to make it prettier
[03:49:47] <mozmck> me too.
[03:50:19] <mozmck> but if they keep on the way they are going there won't be anything to boot to! Less is more...
[03:50:19] <Valen> still open source rarely works that way
[03:50:24] <Valen> lol
[03:50:55] <Valen> my boot time is like 8 seconds or so now, the machine spends about the same time in bios fartaresing around
[03:52:07] <L84Supper> we were building kernels for 10.04 over the weekend, the stock kernels are pre-ATI DRM fixes
[03:52:29] <L84Supper> so modesetting almost works most of the time
[03:55:12] <Guest843> hello cradek
[03:55:19] <Guest843> you still here?
[03:55:53] <mozmck> hmm, modesetting works fine on the stock 10.04 kernel on my machine with intel graphics, but not after I patched it for rtai.
[03:56:22] <mozmck> I built from the latest lucid git.
[03:57:44] <mozmck> Just figured out that if you add to the version string in the changelog you can't use a dash: like -rtai or you get errors compiling hcd.c
[03:58:10] <mozmck> I added .rtai instead and it's compiling fine now. (at least I think that's the problem!)
[04:00:58] <Guest843> cradek did a mill that could do rigid tapping with a 3ph spindle motor
[04:01:15] <Guest843> does anyone know what is needed to be able to do that?
[04:01:52] <Valen> balls of steel
[04:01:56] <Valen> or lots of taps ;->
[04:02:22] <Guest843> one would need some kind of spindle encoder, was wandering what kind of encoder one cound use
[04:02:24] <Guest843> lol
[04:02:38] <Guest843> he has a youtude vid of it
[04:03:07] <Guest843> wanted to talk to him about it
[04:10:41] <grommit> I plan to use a simple hall effect sensor on my spindle to do tapping. I haven't figured out how I am going to mount it yet, or the magnets but....
[04:11:09] <Valen> I was under the impression you needed a decent number of counts to do tapping
[04:11:58] <SWPadnos> you need an encoder, minimum resolution is probably 100 lines (400 counts) per revolution
[04:12:02] <SWPadnos> it must have index
[04:12:16] <SWPadnos> the more counts per rev, the better
[04:12:22] <grommit> Why such high res?
[04:12:32] <Valen> there are some high speed magnetic rotary absolute position sensors floating about
[04:12:43] <Valen> i believe they even have quadrature outputs
[04:12:53] <SWPadnos> rigid tapping isn't the same as threading on a lathe (which can be done OK with only the index)
[04:13:00] <grommit> Seems like 2-4 readings /rev would suffice.
[04:13:14] <Valen> well if you only have one pulse per rev and you try to reverse the tap at 300RPM
[04:13:14] <grommit> and braking on the spindle...
[04:13:17] <SWPadnos> not so great when you reverse the tap though
[04:13:28] <Guest843> here is his sample of rigid tapping using EMC http://www.youtube.com/watch?v=9HLKXeWqTF0
[04:13:35] <SWPadnos> you have to know how far it overtraveled, you can't assume that the spindle stops when you tell it to
[04:14:19] <SWPadnos> that's also why you need a real encoder - the encoder can tell EMC when the spindle reverses, whereas a hall sensor or three can't
[04:14:51] <SWPadnos> (well, three could, but you'd have 1/3 revolution when you don't know what's going on))
[04:15:08] <Valen> you could use hall sensors and a bunch of magnets to make a poor mans encoder
[04:15:21] <Valen> basically a quadrature sensor
[04:15:28] <Valen> messy way of doing it though
[04:15:39] <Valen> still, you could retrofit it easy enough
[04:15:40] <SWPadnos> as long as it can tell you direction, and the resolution is good enough for the finest screw you want to tap, that's fine
[04:15:45] <ds3> for that, couldn't a linear scale work? (i.e a cheap digital caliper)
[04:15:59] <Valen> they dont work in a rotary fashion
[04:15:59] <Guest843> what ever encoder you use it will have to be able to handle the high rpms of the spindle
[04:16:06] <SWPadnos> Guest843, yes
[04:16:12] <ds3> Valen: no, use that to determine the final stopping position
[04:16:26] <SWPadnos> ds3, no, you move Z in relation to the spindle rotation, so a linear scale is of no use
[04:16:27] <Valen> how?
[04:16:27] <ds3> and keep the index pulse from the spindle
[04:16:31] <grommit> I assume you tap at a relatively slow speed, no?
[04:16:36] <SWPadnos> and I don't know if they work if you bend them around the spindle :)
[04:16:53] <tom3p> calipers dont output data now, they output sometime later
[04:16:53] <ds3> SWPadnos: no no.. you said you can't tell where it stopped, well a linear encoderwould tell you that
[04:16:56] <Valen> we use linear scales on our linear axies ;->
[04:16:57] <SWPadnos> I think the video Guest843 linked to was at 400 RPM, doing 1/4-20 threads
[04:17:08] <Valen> he means stopped in terms of spindle rotation
[04:17:20] <SWPadnos> ds3, but you'd be measuring the thing you want to control, not the thing it's supposed to be synched to
[04:17:28] <ds3> tap down, count index pulse, stop spindle, read scale, count back pulse
[04:17:45] <ds3> SWPadnos: it'd justbe a secondary measurement to see how much you overshot
[04:17:50] <Valen> you need to move Z down at the right rate to do the tap
[04:17:59] <Valen> otherwise you will just spin the tap like a drill
[04:18:15] <ds3> Z movement and spindle rotation is sync'ed for taopping
[04:18:17] <renesis> k guys im going to buy thos foxxcon atom board and but some small ubuntu on it and then you guys are gonna help me turn it into emc
[04:18:21] <renesis> right? ok good
[04:18:23] <SWPadnos> the problem here is that EMC2 is measuring spindle rotation and moving Z. if you stop Z while the spindle is turning, the tap breaks (or the workpiece does) - it doesn't pull the quill down for you
[04:18:29] <Valen> renesis lol
[04:18:34] <ds3> ah I see
[04:19:11] <renesis> valen: intel atom board is like fastest shit on wiki latency table
[04:19:15] <Valen> hard part with that that I had was the livecd wouldn't boot on my atom
[04:19:23] <renesis> oh
[04:19:24] <renesis> fuck
[04:19:28] <Valen> 2nd fastest, I have well had the asus PC-DL delux
[04:19:31] <SWPadnos> I guess you could freewheel the Z motor, but that seems like an icky and possibly dangerous solution
[04:19:35] <renesis> whatever whats amd
[04:19:36] <Valen> that got hit by lightning
[04:19:45] <Valen> now I have one of the atoms
[04:20:01] <renesis> hey can i do smp and realtime?
[04:20:03] <renesis> or what
[04:20:07] <Valen> thats what I'm doing
[04:20:14] <Valen> on that dual core atom board
[04:20:14] <SWPadnos> it's possible, but harder to build kernels for that
[04:20:15] <ds3> have a clutch to switch to a stepper to drive it for tapping only?
[04:20:19] <renesis> so it works or youre not machining anything
[04:20:22] <grommit> you can do realtime on one core, everything else on the other
[04:20:24] <renesis> okay
[04:20:37] <Valen> ds3: or just use a spindle encoder
[04:20:38] <SWPadnos> no, that's the opposite of the effect I was looking for :)
[04:20:48] <Valen> actually I'm looking at milling threads now
[04:20:56] <SWPadnos> letting the motor go might allow the tap to "pull" the Z as the spindle slows
[04:20:59] <renesis> neat
[04:21:07] <renesis> valen: what board?
[04:21:07] <SWPadnos> then you measure it with the Z encoder/scale
[04:21:07] <Guest843> you would need a good encoder for knowing what speed the spindle is going so you can track the speed of the z axis in sink with the spindle speed slowing down to a stop and then as it speeds up in the reverce direction
[04:21:08] <ds3> Valen: isn't there a max speed encoders can run at?
[04:21:11] <Valen> it really looks like the way to do stuff, and not bother with needing spindle encoders
[04:21:25] <Valen> renesis the intel one in the table
[04:21:31] <renesis> oh
[04:21:32] <Valen> I get ~4000 or so latency
[04:21:35] <renesis> fuck newegg just ran out
[04:21:38] <SWPadnos> that's why I said you need at least 100 cycles (400 counts), and the more the merrier
[04:21:39] <renesis> i almost bought one
[04:21:43] <Valen> lol
[04:21:49] <Valen> there should be more around the place
[04:21:56] <renesis> which atom?
[04:21:58] <SWPadnos> that's better than 1 count per degree, which should be OK for most taps
[04:22:00] <renesis> was d510?
[04:22:01] <Valen> I use mesa hardware though
[04:22:04] <grommit> what looks like the way to do it without spindle encoder?
[04:22:05] <renesis> oh
[04:22:12] <renesis> im doing stepdir on parport
[04:22:23] <SWPadnos> grommit, there is no way, if you're talking about rigid tapping
[04:22:24] <Valen> I think thats one of the new ones, I don't know if anybody has tried one of those yet
[04:22:35] <SWPadnos> you can always buy a tapping head if you want to do it that way
[04:22:52] <Valen> grommit, its not rigid tapping, its milling the thread itself, only really works on larger >M8 threads
[04:23:07] <grommit> oh
[04:23:08] <Valen> cut the actual spiral of the thread with a high speed cutter
[04:23:35] <SWPadnos> oh, threadmilling would work too, but only for larger holes
[04:23:46] <grommit> unfortunately I don't know if I can mount anything more then a hall effect type on my spindle :-(
[04:23:50] <ds3> let's all threadmiill a 0-80 hole ;)
[04:24:15] <SWPadnos> you could use it like a lathe in that case, but no rigid tapping
[04:24:54] <SWPadnos> you only need the index pulse for lathe-style threading, because the spindle never needs to reverse, and it's assumed that it has reasonably consistent speed
[04:24:55] <tom3p> its no problem to run the encoder off a belt for this application is there?
[04:24:55] <renesis> http://www.newegg.com/Product/Product.aspx?Item=N82E16813186184
[04:25:05] <SWPadnos> if the spindle sucks, you need more feedback to track it
[04:25:07] <renesis> thats the foxxcon 945gc atom 330
[04:25:07] <Valen> should be ok, make sure theres no lash
[04:25:14] <grommit> I tries the d510 MB, and I think it would work fine. Unfortunately, I had another problem and returned it for the other one before figuring out the problem was with my monitor.
[04:25:23] <Valen> dunno I havent used that one
[04:25:24] <SWPadnos> tom3p, as long as it's 1:1 drive with no slip, that's fine
[04:25:29] <grommit> If the 510 would work it would be nicer as there is no fan on it
[04:25:32] <ds3> SWPadnos: for a blindhole, you'd still need morethen an index, I'd guess
[04:25:33] <renesis> jetway has one thats big
[04:25:57] <Valen> I used the intel one
[04:26:03] <SWPadnos> ds3, maybe, maybe not. you certainly can't "peck tap" with a tapping head
[04:26:08] <renesis> http://www.newegg.com/Product/Product.aspx?Item=N82E16813131396
[04:26:09] <renesis> hmm
[04:26:25] <ds3> *nod*
[04:26:42] <renesis> asus prob has it all hacked up for hardware monitoring and quick booting and weird hacks
[04:27:07] <Valen> renesis http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?JakeAndRussells is ours
[04:27:20] <renesis> k ty
[04:27:26] <tom3p> the man page for hostmot2 describes 'counts' as "Feedback position in counts (number of steps)" but what is it in velocity mode? nanosecs per linear unit of measure? (or no meaning/use)
[04:27:31] <grommit> This is the one I got: http://www.mini-box.com/Intel-D945GCLF2D-Mini-ITX-Motherboard
[04:27:32] <Valen> D945GCLF2 is the atom board we have used
[04:27:43] <renesis> grommit: they ran out
[04:27:47] <renesis> in the last week
[04:27:53] <renesis> im either going to get a foxxcon or asus
[04:27:57] <renesis> or get both and race them
[04:27:58] <grommit> They have the 510: http://www.mini-box.com/D510MO-mini-ITX-Intel
[04:28:01] <Valen> why not the new intel?
[04:28:04] <renesis> and then make one a htpc
[04:28:18] <renesis> because i know this works
[04:28:30] <renesis> but if i buy two i might try the 510
[04:28:30] <Valen> if you want a htpc you want a zotac ion one with the onboard intel 9300 so you can do vdpau
[04:28:37] <grommit> oh come on, live on the edge!
[04:28:45] <Valen> or something else with the 9300
[04:28:57] <renesis> yeah youre like 3rd person says this
[04:29:02] <grommit> :-)
[04:29:19] <grommit> everyone justs wants you to be the guinea pig
[04:29:25] <renesis> i know
[04:29:27] <renesis> =(
[04:29:29] <Valen> it'll probably be ok ;->
[04:29:36] <Valen> no guarantees though
[04:29:51] <grommit> I agree with both those sentiments
[04:29:51] <renesis> well no guarantees buying a foxconn or asus
[04:29:56] <Valen> ey look at it this way, the mbo costs the same as an 8mm carbide end mill, which you'll snap anyway
[04:30:06] <renesis> i have a taig
[04:30:12] <renesis> i dont pay close to that for endmills
[04:30:14] <renesis> maybe a 10 pack
[04:30:21] <Valen> hss or carbide?
[04:30:36] <renesis> carbide, usually ebay
[04:30:42] <Valen> thats cheap
[04:30:46] <renesis> american company
[04:30:54] <Valen> http://www.newegg.com/Product/Product.aspx?Item=N82E16813131635
[04:30:58] <Valen> thats what you want ;->
[04:31:02] <renesis> its like 1/4" and under
[04:31:03] <Valen> wait no thats what I want
[04:31:05] <Valen> ahh
[04:31:31] <renesis> still good deals tho, uncoated but its not like it matters on my machine
[04:31:51] <renesis> i wont use HSS, it wears down and locks steppers and my shit is open loop
[04:32:16] <renesis> the carbide keeps cutting and cycle finished the finish is just all fucked
[04:32:42] <renesis> *finishes
[04:33:31] <renesis> heh @ atom board remote as cnc controller interface
[04:34:36] <Valen> I want an atom board with hardware virtualisation
[04:34:45] <renesis> http://www.newegg.com/Product/Product.aspx?Item=N82E16813186194&cm_re=d510-_-13-186-194-_-Product
[04:35:02] <Valen> why that one?
[04:35:09] <renesis> thats a d510
[04:35:22] <Valen> is there not an intel one? thats closer to a known working one
[04:35:27] <renesis> because there isnt one on the latency table
[04:35:40] <Valen> is there one of those?
[04:35:48] <renesis> oh wait there was it was passive cooled
[04:35:55] <renesis> big black sink why didnt it pop up...
[04:36:06] <Valen> they still need a fan ;->
[04:36:19] <renesis> http://www.newegg.com/Product/Product.aspx?Item=N82E16813121399
[04:36:36] <renesis> well the case is going to have 120mm fans
[04:36:43] <Valen> thats probably enough
[04:36:56] <Valen> thats where I would be leaning for the EMC machine
[04:36:58] <renesis> those little fans suck
[04:37:27] <Valen> stick 2gb of ram in it
[04:37:29] <renesis> did you use a sata drive?
[04:37:31] <renesis> yeah
[04:37:34] <Valen> ywah
[04:37:44] <Valen> lol 260gb one
[04:37:55] <renesis> i just got two 500gb sata drives in the mail =\
[04:38:03] <Valen> oh keep in mind the intel ones are VGA output not DVI
[04:38:13] <renesis> that would be neat for an htpc tho
[04:38:13] <Valen> thats my only complaint really with it
[04:38:24] <renesis> i dont care
[04:38:35] <renesis> damn this is going to get expensive
[04:38:38] <Valen> atoms are a bit weak for HTPC, thats why we use the onboard 9300's
[04:38:50] <renesis> i want to buy new drives too, the xylotex blew the z now im using the a for it
[04:39:26] <renesis> yeah i might get something bigger
[04:39:31] <Valen> MESA has some of those ;->
[04:39:38] <Valen> if you want to go servo
[04:39:40] <renesis> i prob not want to buy two after i see the price tag for this one
[04:39:46] <renesis> i dont
[04:39:52] <Valen> actually they are pretty reasonable
[04:40:12] <renesis> no i mean whole system after drives and ram and case
[04:40:33] <Valen> bah cases are for pussies ;->
[04:41:08] <Jymmm> Yeah! Mount it on the wall like art!
[04:41:26] <renesis> i had tons of mobo on my wall at my last place =\
[04:41:34] <Valen> good wall
[04:46:55] <Jymmm> FUNCTIONAL on the wall
[04:47:02] <Jymmm> actually working
[04:47:13] <Jymmm> not target practice
[04:48:29] <Valen> I meant to have thousands of kg of crap hanging off it it'd have to be a good wall ;->
[04:48:49] <Valen> though I am thinking of framing my TV computer now Jymmm
[04:48:58] <Valen> I wonder what the WAF will be like on that one
[04:49:40] <Valen> renesis you know you dont need much in the way of disk right
[04:49:51] <Valen> you could use like a 2gb USB stick or something
[04:49:54] <Jymmm> Valen: Hope you like being celbet
[04:49:54] <renesis> http://www.newegg.com/Product/Product.aspx?Item=N82E16822136553
[04:50:02] <renesis> heh
[04:50:21] <Valen> way overkill for EMC
[04:50:32] <renesis> whatever its cool anyway!
[04:50:49] <renesis> i prob get a 5400 rpm samsung
[04:50:56] <Valen> http://www.newegg.com/Product/Product.aspx?Item=N82E16820139149 is cool
[04:51:10] <renesis> too much
[04:51:31] <Valen> meant to be $75
[04:52:01] <Valen> i meant instead of that 10Krpm drive you were looking at
[04:53:54] <renesis> o
[04:53:58] <renesis> http://www.newegg.com/Product/Product.aspx?Item=N82E16822148231
[04:54:25] <renesis> or a samsung same price
[04:54:55] <Valen> thats more like it ;->
[04:55:18] <renesis> yeah now its just an old vs new loyalty thing
[04:55:19] <renesis> =\
[04:56:04] <renesis> i have two new samsung drives already i get the seagate
[04:56:54] <renesis> http://www.newegg.com/Product/Product.aspx?Item=N82E16820226111
[04:57:48] <Valen> sure you want a dual channel kit?
[04:57:59] <Valen> the atom boards are usually only a single ram slot
[04:58:10] <renesis> http://www.newegg.com/Product/ImageGallery.aspx?CurImage=11-112-227-TS&ISList=11-112-227-S01%2c11-112-227-S02%2c11-112-227-S03%2c11-112-227-S04%2c11-112-227-S05%2c11-112-227-S06%2c11-112-227-S07%2c11-112-227-S08%2c11-112-227-S09%2c11-112-227-S10&S7ImageFlag=1&Item=N82E16811112227&Depa=0&WaterMark=1&Description=LIAN%20LI%20PC-Q07%20Black%20Aluminum%20Mini-ITX%20Tower%20Computer%20Case
[04:58:22] <renesis> the d510 was two, dual channel
[04:58:26] <renesis> (i was like yay)
[04:58:43] <renesis> cute lian-li^
[04:59:19] <Valen> weird
[05:00:14] <renesis> i got one of the older boxy cases, i <3 it
[05:00:45] <renesis> except for the front port door, headphones only go in right if you hold the little door level while sliding the jack in
[05:00:50] <mozmck2> finally, emc is running with realtime on 10.04 here!
[05:01:12] <Valen> mozmck2 awesome
[05:01:16] <renesis> but i like how this thing got two ports in front
[05:01:21] <renesis> keyboard and mouse ftw
[05:01:51] <renesis> i can totally hack a 120mm fan onto that side vent
[05:02:54] <renesis> standard ATX PS/2 power supply units.
[05:03:07] <renesis> neat im getting it
[05:04:08] <KimK> SWPadnos: (re: earlier discussion of rdf files) Yes, I agree with you completely. Looking at an rdf file is like looking at HTML code instead of looking at a webpage. So I thought I'd ask what rdf viewer everyone was using that made it so important to have those "screwy" files posted there as well as the txt files. And to see what they were seeing. But I guess few are using them?
[05:07:11] <KimK> In case anyone missed the earlier discussion: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emc/
[05:10:55] <KimK> Ha, I guess that's some kind of recursive reference. I meant the link just to show the list of files (including rdf files). But I guess it also points (or will point soon) to the discussion we had earlier. Oh well, nevermind.
[05:23:42] <tom3p> cool www.linuxcnc.org is in Polish now "Strona główna "
[05:59:45] <Dave911> Anyone know why touchy wasn't put in 2.3.5 release version?
[06:03:19] <Dave911> mozmck2: Wow, that took some time didn't it..
[06:03:45] <Dave911> The D510 is a good board. It is like the Intel dual core 330 MiniItx board but faster.
[06:58:48] <alex_joni> Dave911: 2.3.x releases are bugfix releases
[06:58:56] <alex_joni> new stuff goes into 2.4.
[08:24:19] <H_MrSun> hmm, milling rubber, does that work ? :)
[08:25:01] <archivist> I know grinding does, often frozen to do it
[08:25:33] <H_MrSun> hmm
[08:25:39] <H_MrSun> hard to keep stuff frozen on the mill :/
[08:27:31] <archivist> I worked at a company making printers, we had fun getting the roller company to do what we needed
[08:33:12] <Valen> depends on the rubber
[08:33:20] <Valen> we lathe urithanes somewhat often
[08:36:19] <MrSunshine__> its rubber for a car, the one that is from where the fuel comes from and goes into the engine block
[08:36:27] <MrSunshine__> hell ive got total blackout on every word today
[08:36:37] <MrSunshine__> aparently cant get new ones for this car :/
[08:58:43] <toastydeath> if you spray rubber with liquid nitrogen you can turn it
[08:58:48] <toastydeath> i'm sure milling works the same way
[09:01:23] <pjm> my dad used to turn rubber wheels for a extrusion machine, they used to freeze them over night, and turn whilst very cold. the surface finish wasnt pretty but apparently it was good enough
[09:09:55] <Valen> MrSunshine, got a picture of what your trying to make?
[09:10:01] <Valen> bbl
[09:10:52] <piasdom> g'mornin all
[12:05:06] <Valen> zup
[12:53:16] <Valen> http://cgi.ebay.com/GEARMOTOR-HARMONIC-DRIVE-SYSTEMS-100-1-RATIO_W0QQitemZ280487059939QQt_1165#ht_711wt_1165
[12:53:25] <Valen> $100 shipping to get it here
[13:26:34] <nickoe> hi
[13:27:25] <nickoe> why ain't there a amd64 source of emc2_2.3.5 on http://linuxcnc.org/hardy/dists/hardy/emc2.3/source/ ???
[14:02:55] <cradek> I wonder what nickoe meant
[14:03:17] <skunkworks_> Good morning chris
[14:03:18] <SWPadnos> yeah. just noticed that
[14:03:27] <skunkworks_> and SWPadnos
[14:03:28] <cradek> hi
[14:03:34] <SWPadnos> maybe he thinks the source needs to be different for AMD64 ...
[14:03:37] <SWPadnos> hi skunkworks
[14:03:51] <cradek> yeah maybe so.
[14:05:25] <skunkworks_> cradek: did you see this? so - is there and more elegent way to convert bits (switches) in ladder to a number? The one way that comes to mind is to have a variable assignment for each switch and then add them togather to get the number - so the first switch would set %w0 to 1 if it was activated - switch two would set %w1 to 2 if it is activated - switch 3 would set %w2 to 4 if it is activated and so on (15 total) and then ad
[14:05:29] <skunkworks_> togather
[14:06:37] <cradek> how about -| |----[%W0=%W0|1]- etc? I think there is bitwise or.
[14:06:59] <skunkworks_> oh - sexy
[14:07:11] <cradek> assuming that's how you "spell" it
[14:07:14] <skunkworks_> I will try it.
[14:08:26] <skunkworks_> or is !
[14:08:33] <skunkworks_> iirc
[14:08:50] <SWPadnos> that would be strange (but still possible :) )
[14:08:53] <skunkworks_> nope - |
[14:26:12] <cradek> I like how classicladder can easily be expressed in ascii
[14:28:21] <Dave911> I "installed, not a run in place 2.3.5... works fine Then I did a GIT clone etc, and compiled the master. Now I can't get a 5i20 config to run but I can get a regular parallel port stepper config to run???? I get a lot of NML errors writing to blah blah blah ...
[14:28:23] <Dave911> I can have an "installation and a run in place" on the same machine .. right? Any ideas?
[14:28:51] <Dave911> I meant to say I did a full installation of 2.3.5, not a run in place ....
[14:29:15] <Dave911> I compiled the master and installed it as run in place
[14:29:24] <cradek> Dave911: maybe the path to your firmwares is not right in the rip's config you are trying to run
[14:29:44] <cradek> sometimes it's hard to identify the real error...
[14:29:53] <cradek> if you can't find it, pastebin the entire output
[14:30:08] <Dave911> OK, that sounds about right ..... where would the path for the firmware be set?
[14:30:30] <cradek> in the hal file, where it is loaded
[14:30:30] <Dave911> So I can have an installation and a run in place at the same time.... right?
[14:30:33] <cradek> sure
[14:31:20] <Dave911> OK, I'll check the path. It seems to be particular to the 5i20 so firmware makes sense.... I'll check... thanks!
[14:31:33] <cradek> welcome
[14:37:05] <skunkworks_> cradek: thanks a bunch - looks like that should work.
[14:37:37] <skunkworks_> and you had the syntax right
[14:39:35] <Dave911> The mesa firmware I need is not in the emc2-dev git clone directory .... perhaps I needed another compile switch to compile the firmware also?? I did find the mesa firmware in the lib directory from the original installation..
[14:45:37] <SWPadnos> there's a separate git repository of firmware - it's not with the emc2 source any more
[14:46:16] <Dave911> OK, I missed that... how does that work?
[14:47:06] <SWPadnos> well, you check it out just like the emc2 repo, only you use hostmost2-firmware.git instead of emc2.git in the checkout line
[14:47:46] <SWPadnos> I'm not sure you actually need the new firmware though - the old firmware should work
[14:48:28] <SWPadnos> which would mean that you should be able to use the same firmware location as the installed emc2 (/lib/firmware/hm2/whatever)
[14:50:17] <Dave911> OK, cool I can do that.... You are probably right, but I knew that I was doing something wrong with the firmware... now I know why... I'll do the git clone on the hostmot2-firmware.git.... just so I know that works... on some other installations I might not have any "old" firmware to reference...
[14:50:31] <Dave911> Thanks!
[14:50:34] <SWPadnos> right-o
[14:53:10] <Dave911> I'm curious.... why wasn't Touchy included in 2.3.5? Anyone know?
[14:53:23] <SWPadnos> too new?
[14:53:30] <SWPadnos> (I don't know for sure)
[14:54:40] <Dave911> I thought that a number of people had beat on it pretty hard. Isn't Stuart using it also on some machines?? I thought that cradek would have slid it in to 2.3.5.... ;-)
[14:55:07] <Dave911> No biggie... just curious...
[15:00:50] <cradek> it will be in 2.4.0. It depends on some things that are not in 2.3.
[15:01:21] <cradek> by all means build the 2.4 git branch if you want to try it out
[15:04:57] <cradek> good morning seb. is your garage ready?
[15:06:07] <skunkworks_> cradek: what is going in its place?
[15:06:14] <skunkworks_> cradek: what is going in its place?
[15:06:22] <skunkworks_> heh
[15:06:58] <cradek> skunkworks_: it will be empty space for a while, I think. I also finally junked the old lathe control. it turns out we had 1.5 tons of scrap metal on the property...
[15:07:12] <skunkworks_> Nice
[15:07:22] <cradek> some day in the distant future we may have garage space for two cars again. you never know.
[15:07:41] <skunkworks_> My parents are doing the same thing. Getting rid of stuff they will never use. (steel and copper is pretty decent right now)
[15:08:07] <cradek> I tihnk we got almost 10c/lb for steel scrap
[15:08:48] <skunkworks_> brb
[15:13:43] <JT-Work> my general plan is to put away, throw away or box up 2 items/tools/things for every 1 I take out until the Hardinge explosion has been cleaned up
[15:19:57] <skunkworks_> heh - I have issues with that. All my project have tools strewn all over the place.
[15:20:25] <cradek> I need a box of a thousand #2 philips screwdrivers
[15:20:47] <cradek> or is it #3 - you know, the one you can never find because you've left them everywhere
[15:20:54] <JT-Work> all the same length too
[15:20:56] <skunkworks_> #2
[15:21:04] <cradek> that other one that's too small is too small. don't even try to use it.
[15:21:32] <cradek> yes it will sort of fit in the head of the screw. no it won't unscrew it. it'll just mess it up.
[15:21:37] <cradek> and the bigger one won't fit at all
[15:21:44] <skunkworks_> I get the vermont american bits that fit in the drill... then you can buy 10 at a time. :)
[15:22:53] <archivist> just face it the tools will be it the last box you look in
[15:23:11] <cradek> no, they're slightly to the right of whatever I was working on last
[15:23:28] <skunkworks_> http://www.amazon.com/Vermont-American-0102323-Insert-Phillips/dp/B000PIBM0Y
[15:24:21] <skunkworks_> I really like those - they grip really well.
[15:25:46] <skunkworks_> *and I have stock in the company... ;) j/k