Back
[00:30:57] <Jymmm> WORLD ENDS IN MINUTES, EEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkkk
[00:34:23] <L84Supper> depending on your timezone
[00:35:04] <Jymmm> Yeah Mayans had time zones... DUH
[00:36:48] <Jymmm> L84Supper: YOU DIE FIRST! ;)
[00:37:21] <L84Supper> I've been dead for years
[00:38:36] <Jymmm> That's one hell of an internet connection
[01:10:00] <capricorn_1> world still OK in CA, USA
[01:16:02] <r00t4rd3d> https://www.youtube.com/watch?v=Z0GFRcFm-aY
[01:26:32] <Jymmm> Heh, I have that song on cd =)
[01:26:47] <Jymmm> It was part of a movie soundtrack
[02:10:39] <DJ9DJ> moin
[02:20:50] <Loetmichel> mornin'
[02:22:00] <Valen> heres a question, RTai is nifty and all, and made mucho sense on single core machines, quad cores are cheap these days, I wonder if "latency" could be improved by isocpuing 3 of the cores and just have a simple wait loop on the timers
[02:23:38] <archivist> Valen, are you watching the dev mailing list
[02:27:27] <Valen> no, did i invent something too late again?
[02:27:46] <archivist> possibly
[02:28:37] <Valen> archivist: is the list online somewhere?
[02:28:52] <Valen> I was mainly thinking I want to run my servo loops really fast for mesa hardware
[02:29:07] <archivist> yes I was just trying to find a starting point
[02:30:17] <archivist> sourceforge has a terrible interface to the mail archive
[02:30:26] <archivist> http://sourceforge.net/mailarchive/forum.php?forum_name=emc-developers&max_rows=25&style=ultimate&viewmonth=201211
[02:31:12] <Valen> what should i be looking for in there?
[02:31:13] <archivist> last three or four months, look at what mhaberler has been trying
[02:31:27] <archivist> stuff in user mode
[02:31:48] <mhaberler> ?
[02:33:04] <Loetmichel> sooo, fits... ans half as loud as with the 40mm fan ... ;-) ->
http://www.cyrom.org/palbum/main.php?g2_itemId=13829
[02:33:16] <archivist> mhaberler, I have a feeling some of your experiments may prove very fruitful
[02:33:44] <mhaberler> I have the feeling they might prove a bit support-intensive ;)
[02:33:52] <Valen> I was thinking about replacing RTai with a multi core cpu, isocpu and a spinlock
[02:34:21] <archivist> I am watching and keeping quiet :)
[02:34:42] <mhaberler> so how do you assure your core is scheduled properly
[02:34:56] <Valen> isocpu, nothing else runs on it
[02:35:28] <archivist> considering a hardware interrupt did something elsewhere on another core
[02:35:35] <mhaberler> well fine, but you still have to deal with a capricious scheduler underneath
[02:35:41] <Valen> I know practically nothing about the underlying gotchas, I was just thinking times have changed somewhat since RTai came about
[02:36:13] <Valen> but if its isocpued then the linux scheduler should leave that core totally alone is my understanding?
[02:36:29] <mhaberler> there is one trick in the bag which hasnt been tried in LinuxCNC but 'said to work' at least per xenomai mailing list
[02:36:48] <mhaberler> 'leaving alone' doesnt necessarily translate into 'schedule in time'
[02:37:07] <Valen> Thats what I'm not understanding
[02:37:17] <Valen> http://www.linuxtopia.org/online_books/linux_kernel/kernel_configuration/re46.html
[02:37:30] <Valen> Remove the specified CPUs, as defined by the cpu_number values, from the general kernel SMP balancing and scheduler algroithms.
[02:37:57] <mhaberler> well go ahead, do that and plot a delay spectrum from such a isolcpus thread
[02:38:20] <Valen> this is where it goes over my head ;->
[02:38:55] <mhaberler> nah, a simple test is worth more than musings
[02:39:18] <Valen> simple test is still over my head ;->
[02:39:23] * Valen is a python weenie
[02:39:29] <Valen> real time means less than half a second ;->
[02:39:40] <mhaberler> right… pedestrian motion
[02:39:48] <Valen> or getting 40FPS in a game is good enough
[02:39:58] <mhaberler> what you could try:
[02:40:30] <Valen> hmm, my mobile phone is getting kicked from linuxcnc saying "invite only #"
[02:40:32] <mhaberler> read the shackleford/proctor paper on jitter - the summary gives a hint
[02:40:52] <mhaberler> the idea is basically this - trade CPU cycles against jitetr
[02:40:56] <mhaberler> jitter
[02:41:03] <Valen00> Works now
[02:41:16] <mhaberler> schedule the thread a bit early, and busy-wait until the release point has come
[02:41:38] <Valen> That always seemed obvious to me I just kinda presumed it would be done that way
[02:41:46] <mhaberler> it would be quite simple to do in threads/hal/rtapi/motion
[02:41:57] <Valen> but again i spose different era's had different requirements
[02:42:03] <mhaberler> no, it relies on OS timers for the release point
[02:42:40] <Valen> it doesn't use timed interrupts?
[02:42:52] <Valen> as in tell the CPU when you want to be woken up?
[02:42:54] <archivist> seeing the release point is where dma and interrupt holding the bus get in the way
[02:42:55] <mhaberler> interrupts are pretty much unused in Linuxcnc
[02:43:17] <Valen> I figured you would just spin on the HPET timer or something along those lines
[02:43:20] <mhaberler> it's a polling affair except for the thread functions, which use OS timers
[02:43:43] <mhaberler> well right, the 'schedule early' trick would in effect do that
[02:43:50] <mhaberler> outline: add a early=x (ns) param to threads and motion
[02:44:11] <Valen> this one I understand ;->
[02:44:13] <mhaberler> set timer to period-early and pass that down to hal/rtapi threads
[02:44:28] <mhaberler> loop after timer release until exact period
[02:44:36] <Valen> actually you don't even need to call early
[02:44:41] <Valen> perhaps
[02:45:00] <Valen> call "on time" then busywait to some defined "late" interval
[02:45:08] <Valen> might reduce the code changes
[02:45:14] <mhaberler> what would that buy you?
[02:45:41] <mhaberler> same effort but delay incurred imo
[02:46:06] <Valen> is it a delay though?
[02:46:31] <Valen> hmm I spose it could be
[02:46:39] <mhaberler> what you want is isochronous behaviour, not exactly late by some fixed amount of time
[02:46:57] <Valen> you shouldn't be late by a fixed amount of time
[02:47:04] <Valen> i'm saying leave all the trigger stuff as it is
[02:47:23] <Valen> but then busywait on the hpet or whatever until you are at a defined point
[02:47:30] <mhaberler> let us stand back for a moment: the more interesting question IMO is: why is this so terribly important?
[02:48:13] <Valen> My actual goal was to run the servo thread in the 10's of khz range to improve the performance of voltage controlled devices
[02:48:17] <Valen> (think mesa hardware)
[02:48:30] <Valen> also for higher speed things
[02:49:04] <mhaberler> you sure are aware of the sampling theorem - how fast is the underlying hardware?
[02:49:19] <Valen> "fast enough" ;->
[02:50:12] <Valen> I figured just having one process running the servo thread with busy waits in the "down time" would nerf latency issues and let me use a high speed processor without jitter causing weird timing errors in the PID loops
[02:50:17] <mhaberler> well if the hw has a say a -3db cutoff at 100Hz 50kHz servo period are a very interesting academic experiment to verify that it doesnt matter
[02:50:30] <Valen> by 10's I meant literally 10-20khz
[02:50:57] <mhaberler> let me refocus this a bit
[02:51:32] <Valen> If you look at some of the laser machines in raster mode their controllers must be running fairly quick (although I imagine they cheat and vary the laser to track the position rather than controlling the laser position well)
[02:51:49] <mhaberler> for my part I care exactly zilch about 'weird timings', I care about path quality - and that we have no measure for
[02:52:24] <Valen> What i meant by weird timing was running a PID loop with a varying time base is generally regarded as a "bad thing" yes?
[02:52:48] <mhaberler> I understand the PID loop is one part where timing really matters, yes
[02:53:00] <mhaberler> most of tp could care less
[02:53:08] <Valen> that's really the only part I'm looking at
[02:53:15] <mhaberler> so pid loops
[02:53:17] <archivist> knowing the time and the error means you can make a better error calculation
[02:53:41] <mhaberler> (oh, anybody looking into the ferror problem btw?)
[02:53:56] <mhaberler> ferror being calculated on last, not current commanded pos?
[02:53:57] <Valen> with the mesa hardware its voltage controlled, so some of the nice things you get from having an external controller turning that into torque go away
[02:54:26] <Valen> My understanding is that many of the off the shelf "controllers" run PID loops in the khz range
[02:54:35] <Valen> (newer ones)
[02:55:11] <mhaberler> right, but that's pushing the limits of rtapi/hal a bit and you tradeoff performance against config flexibility
[02:55:27] <Valen> bandwidth reduction, I've heated my dinner twice now lol going to TV and food and IRC ;->
[02:55:32] <mhaberler> there's an interesting paper on that issue, need to dig that out
[02:56:07] <mhaberler> about numerical stability, servo thread speed requirements etc
[02:57:11] <mhaberler> anyway what I was trying to say is: I wish we had a means to gauge path quality in a spatial and temporal sense, rather than relying on some artifical, semi-related thread timing window detection
[02:59:32] <Valen> We use linear scales on our mill
[02:59:46] <Valen> What I'd like is to record the error during a run
[02:59:52] <Valen> then correct for it in the next run
[03:00:22] <mhaberler> 'the error' - pid error?
[03:03:58] <mhaberler> not sure this will hunt.. you recorder error of a stochastic process, and add that in as a correction in the next stochastic process?
[03:04:44] <mhaberler> I guess you are assuming error behaviour is 1:1 reproducible and that might not be the case
[03:31:06] <Valen> the difference between commanded and actual positions
[03:31:25] <Valen> with the glass scales we measure the actual tool position
[03:31:35] <Valen> well pretty close
[03:31:56] <Valen> so errors due to there being a lump of iron in the way could be taken out
[03:34:43] <archivist> can you properly separate out systematic errors though
[03:36:32] <archivist> and errors due to a heavy cut should not be applied to a light finishing cut
[03:54:12] <mhaberler> maybe you should validate your theory about ability to correct errors by plotting piderror over several runs and see if there is any similarity in the error curves - I guess you will see mostly uncorrelated noise
[03:59:48] <archivist> one would have to average the noise out to see mechanical friction and cutting load etc
[04:02:51] <Valen> a low pass filter would be required thats for sure
[04:03:03] <Jymmm> http://www.timeanddate.com/countdown/maya
[04:03:05] <Valen> The question is how would one save the information
[04:03:47] <archivist> you probably need a moving average anyway
[04:04:25] <Valen> something like that
[04:04:42] <Valen> Is there a way to dump piderror to a file?
[04:12:59] <mhaberler> halsampler
[08:58:34] <jthornton> a little progress today...
http://imagebin.org/240218
[08:59:56] * archivist spots a mistook on the helical dive
[09:29:17] <jdh> http://wilmington.craigslist.org/tls/3491857640.html
[09:32:35] <TekniQue> heh I've never seen a toolchanger on the tailstock of a manual lathe before
[09:32:46] <archivist> rewiring the internals of a motor 440 3ph to 220 3ph is easy enough
[09:33:06] <jdh> isn't it just a turret?
[09:33:17] <archivist> or capstan
[09:33:27] <jdh> there is no 3ph residential here of any voltage.
[09:33:58] <jdh> would a phase converter and the old motor be any better than a new single phase?
[09:34:02] <archivist> nor here hence single phase to 3ph inverter/vfd
[09:34:27] <cradek> TekniQue: that's called a turret lathe
[09:34:28] <jdh> or chdaper.
[09:34:32] <jdh> or cheaper?
[09:34:41] <TekniQue> I see
[09:34:54] <jdh> I'd go buy it if I could sell my chinese 9x20 quickly
[09:35:03] <cradek> jdh: you'd want a 2hp single phase vfd and 2hp 3ph motor
[09:35:15] <cradek> then run it on your normal single phase 240
[09:35:42] <archivist> rewire the motor internally then you dont need to get a new motor
[09:35:56] <cradek> depends how big it is.
[09:36:24] <jdh> we have one at work that looks physically the same, 480 though
[09:36:27] <cradek> if it's an 8hp motor you simply can't run it on your household power (but it's probably not very big)
[09:39:00] <jdh> looks like they have small motors
[09:39:07] <AR_> k
[09:39:16] <AR_> ordered chinese sbr16 rails for new mini machine
[09:39:17] <AR_> lol
[09:39:36] <AR_> now i need to decide between aluminum extrusions or steel tube for frame
[09:39:46] <cradek> yes I would be surprised if that lathe was more than 2hp.
[09:40:32] <cradek> in case it's not obvious, that is not an engine lathe: you can't cut threads with it
[09:42:00] <archivist> capstan v turret
http://www.ustudy.in/node/7743
[09:43:04] <AR_> ok here is what i will have in stock from previous halfassed purchases
[09:43:08] <L84Supper> so engine lathes are not for making engines, the same as band saws are not for cutting bands
[09:43:32] <AR_> two 350mm long supported shaft rails with 2 bearing blocks each
[09:44:05] <AR_> two 28 inch long IKO LWL12 profile rails with 2 bearing blocks each
[09:44:19] <AR_> do you think i can make an x/y table with that
[09:44:26] <AR_> with aluminum extrusions/plate
[09:44:29] <AR_> ?
[09:44:59] <AR_> i'm thinking something like
http://www.cnczone.com/forums/vertical_mill_lathe_project_log/32609-80_20_mill_build-4.html
[09:45:01] <AR_> but a bit smaller
[09:46:32] <archivist> anything is possible but bars can be flexible if there is not enough support or stiffness
[09:47:13] <AR_> which bars do you mean
[09:49:08] <AR_> i think if i cut my IKO rails down to like 18" for X axis
[09:49:17] <AR_> use 350mm rails for Y
[09:49:26] <AR_> it should be fairly stiff
[09:49:58] <AR_> even if i used just AL extrusions
[09:50:19] <archivist> you say shaft rails do you mean round bar unsupported except the ends
[09:50:21] <AR_> but i'm contemplating AL plate and/or steel tube frame
[09:50:30] <AR_> oh, i mean supported
[09:51:33] <AR_> i was thinking maybe steel tube frame filled with sand
[09:52:11] <AR_> this would be a little machine
[09:52:46] <AR_> quite literally a desktop machine size
[09:52:52] <archivist> draw it up, insufficient data to guess
[09:56:05] <r00t4rd3d> http://i.imgur.com/b1HRk.jpg
[09:56:31] <jdh> yeah, lack of threading is not great, but I already have lack of threading.
[09:56:45] <r00t4rd3d> lacker
[09:57:50] <jdh> so, I have an aluminum compressor head. On top is a steel intake valve body, with a bad valve under it. I cannot get the damned valve cover/body/whatever off.
[09:57:53] <archivist> jdh but if adding a leadscrew and cnc it becomes more usable
[09:58:23] <jdh> archivist: the one we have at work that looks like that just clamps the saddle on to the ways wherever you want.
[09:58:35] <archivist> heat and a sharp tap with a hammer
[09:59:01] <jdh> I froze it, then hit the aluminum body with a torch (not too much I hope), no luck.
[09:59:31] <jdh> http://tinyurl.com/cjgr7hy
[09:59:34] <jdh> that's the head
[10:00:04] <archivist> is it screwed in
[10:00:17] <jdh> the valve body fits a 32mm socket. The center allen is a jacking bolt to hold the valve components against the head
[10:00:31] <jdh> it is screwed in, with a crush gasket between them.
[10:00:42] <r00t4rd3d> http://imgur.com/a/5toGa
[10:00:47] <jdh> teh big ugly rusty part is the thing that needs to come off
[10:00:47] <r00t4rd3d> lol
[10:01:54] <archivist> will be tight, the sharp tap with a hammer to compress the washer and to undo
[10:03:10] <jdh> the jacking bolt won't budge either
[10:04:02] <archivist> I also hammer side ways all the way round the bolt/screwed item where possible
[10:04:17] <archivist> some times with some gusto
[10:04:46] <archivist> getting a higher success rate these days
[10:04:56] <jdh> I'll try that. And maybe a longer bar
[10:05:21] <jdh> someone suggested transmission fluid and acetone 50/50 as penetrating oil.
[10:05:32] <archivist> getting enough shocks to the corroded threads
[10:06:12] <jdh> I tried 30 mins in an ultrasonic cleaner too. It is clean anyway
[10:06:58] <jdh> I can only get the torch on about 50% of the mount due to the cooling fins
[10:07:01] <archivist> one thing I have been doing last year was "metal cleaning" for a scrap yard where I was separating all sorts of rubbish
[10:07:43] <archivist> hammer and parallel punches were the main tools
[10:08:04] <jdh> the bottom part is metal/metal sealing surface so I don't want to damage or warp it with heat.
[10:08:34] <jdh> ~280bar compressor
[10:09:04] <archivist> tapping the valve down will compress the washer and separate the thread corrosion
[10:09:46] <jdh> and the body is finned for cooling. Can't tap where the washer is compressed
[10:10:23] <jdh> seems like a really poor design. Guess if it had been removed in the last 20 years, it might come off easier.
[10:10:25] <archivist> at that pressure I expect the torque to loosen to be very high
[10:11:17] <jdh> guess I could stick it on the mill and cut the whole thing out if I have to.
[10:11:17] <pcw_home> heavy walled tube to reach between fins?
[10:11:39] <jdh> the fins are in plane with the load
[10:11:51] <jdh> (the fins on the valve body, not the head)
[10:16:35] <archivist> another thing is get a pneumatic spanner on the socket see if it still argues
[10:16:49] <pcw_home> How hot did you heat the head?
[10:18:09] <archivist> inch drive with 500 ft lbs but holding the head still will be a problem
[10:18:36] <pcw_home> seem like aluminums higher coefficient of expansion has to win at some point (hopefully before the AL melts)
[11:05:34] <tjb1> Hurray for craigslist, 0 legitmate emails/texts, 7 scammers
[11:07:58] <Jymmm> Then don't put your phone number out there in plain text =)
[11:08:38] <tjb1> I used google voice number :)
[11:08:45] <Jymmm> so?
[11:08:53] <tjb1> Its not my real number...
[11:09:08] <Jymmm> again, so?
[11:09:30] <tjb1> SO it doesnt affect me if scammers keep texting me because I can just shut that off?
[11:09:34] <Jymmm> dont put number out there in plain text
[11:09:57] <Jymmm> DONT = 555-1212
[11:10:00] <tjb1> They still text it
[11:10:03] <tjb1> yes
[11:10:05] <Jymmm> DO = 5 FIVE 5 - 1212
[11:10:07] <tjb1> five5five
[11:10:14] <Jymmm> etc
[11:10:16] <tjb1> one2onetwo
[11:10:30] <tjb1> They still text that
[11:10:39] <Jymmm> mix it up
[11:11:04] <tjb1> Then the idiots you want to sell it to won't be able to contact you
[11:11:31] <Jymmm> I never have any issues.
[11:12:51] <tjb1> I knew that last one was a scam…"Do you still have "insert add title here"?"
[11:14:16] <Jymmm> Just have them email you their NAME and NUMBER, and go from there.
[11:14:55] <tjb1> I had one offer me $125 for a $200 item
[11:15:53] <tjb1> I told him 175 and he was like no I have to drive! Well sucks to be you doesn't it?
[11:17:04] <Jymmm> there are cheap bastards everywhere
[11:18:52] <archivist> sometimes circumstances make one negotiate
[11:19:45] <Jymmm> I have bought and sold a OT of things on CL over the years. All have pretty much been painless. Some have even been less than 30s transactions.
[11:19:51] <Jymmm> s/OT/LOT/
[11:25:25] <Jymmm> Other than the resolution, this doens't look too shabby...
http://www.frys.com/product/7345714?site=sa:adpages%20page:P1_FRI%20date:122112
[11:26:33] <AR_> ok here is my general idea
[11:26:35] <AR_> http://i.snag.gy/PTEXE.jpg
[11:26:51] <AR_> bottom rails are supported shaft type
[11:27:10] <AR_> the ones on top are 12mm profile rail
[11:27:27] <AR_> bearings mounted to plate, rails move on top with table
[11:28:27] <tjb1> Why supported shaft and rail?
[11:28:45] <AR_> because that is what i have :P
[11:28:59] <tjb1> Brave man drawing solids in Mastercam
[11:29:20] <AR_> :P
[11:30:27] <AR_> do you think i should shorten the 28" rails for stability?
[11:31:27] <awallin_> put the Y-rails much farther apart..
[11:33:22] <AR_> farther is better with the expense of losing travel
[11:33:35] <AR_> however i suppose with my long rails it will be sufficient to go farther
[11:33:43] <AR_> but then i will need wider base
[11:33:47] <AR_> hmm
[11:34:13] <tjb1> Its all comrpomises
[11:34:20] <AR_> yes
[11:34:33] <archivist> but you can modify later
[11:34:53] <Jymmm> AR_: I think you need additioanl support on the riser
[11:35:14] <archivist> and bas
[11:35:17] <archivist> base
[11:35:25] <AR_> oh i plan on not just bolting it lol
[11:35:34] <Jymmm> _| ---> _/
[11:35:53] <AR_> i want something like this, but smaller
[11:35:54] <AR_> http://www.cnczone.com/forums/vertical_mill_lathe_project_log/32609-80_20_mill_build-4.html
[11:36:30] <AR_> his base and vertical riser is 3060 aluminum extrusion with threaded rod the length of it
[11:36:30] <archivist> if the back and base are too thin they twist under cutting load
[11:36:36] <Jymmm> that has the added support I speak of, it's in black
[11:36:40] <AR_> yes
[11:36:55] <awallin_> here's the general idea :)
http://www.hurco.com/en-us/machine-tools/machining-centers/5-axis-vertical/PublishingImages/VMX60SR-frame.jpg
[11:37:12] <AR_> :P
[11:37:15] <Jymmm> awallin_: WTF
[11:37:27] <Jymmm> awallin_: Where the beer/coffee dispenser?
[11:37:32] <archivist> note the column is basically square
[11:37:48] <awallin_> that one is not for making coffee..
[11:38:09] <Jymmm> awallin_: Sure it is, it has stepper motors
[11:38:18] <AR_> i'm going to cnc my coffee maker brb
[11:40:18] <archivist> you need a slave for the coffee making
[11:41:06] <Jymmm> AR_: Nah, they'll just spit in it
[11:41:15] <Jymmm> archivist: Nah, they'll just spit in it
[11:41:43] <AR_> rofl
[12:01:09] <IchGuckLive> hi all B)
[12:01:30] <andypugh> Hai
[12:07:08] <tjb1> ello
[12:15:36] <IchGuckLive> is the world still standing where you are ?
[12:16:15] <IchGuckLive> Newsealand and bunch of asia is tomorrow so i guess we are in good mute
[12:19:36] <DJ9DJ> phew :D
[12:19:38] <DJ9DJ> hi IchGuckLive
[12:20:43] <IchGuckLive> O.O
[12:26:15] <tjb1> Is this what happens when you get older? I want to buy an old gameboy and play pokemon now...
[12:26:57] <jthornton> skunkworks, see the gremlin today?
[12:27:25] <skunkworks> no - link?
[12:27:32] <IchGuckLive> i play on C64
[12:27:44] <jthornton> http://imagebin.org/240218
[12:28:43] <tjb1> Holy hell, pokemon red itself is almost $30...
[12:29:04] <IchGuckLive> jthornton: is this the new G2 G3 word or a G12/13
[12:29:16] <skunkworks> jthornton: nice work!
[12:29:38] <jthornton> IchGuckLive, ?
[12:30:06] <IchGuckLive> in 2.6 there is a R and Z for this in G2 G3
[12:30:16] <IchGuckLive> or is it a cycle implemented
[12:31:23] <jthornton> I don't know much about master
[12:32:08] <jthornton> skunkworks, I'm getting closer to understanding all I don't know about gremlin... just a couple more nuts to crack
[12:32:12] <IchGuckLive> http://www.linuxcnc.org/docs/2.5/html/gcode/gcode.html#_g2_g3_arc_feed_a_id_sec_g2_g3_arc_a
[12:32:38] <IchGuckLive> see the note on G17
[12:32:46] <IchGuckLive> Z helix
[12:33:32] <Jymmm> AWESOME
http://www.wimp.com/boatramp/
[12:33:39] <jthornton> that's 2.5 and yes Z and P describe a helix
[12:34:24] <IchGuckLive> Jymmm: the girl in the first boat is O.O
[12:34:29] <jthornton> I don't see a note for G17
[12:34:57] <Jymmm> IchGuckLive: Fuck the girl, the rainbow boat/trailer/van
[12:35:46] <IchGuckLive> if you are a freak
[12:35:47] <jdh> no idea how hot I got the head. I got a longer breaker bar and succeeded in rounding off the flats on the valve carrier.
[12:42:08] * jthornton goes to town to get a couple of drill bits
[12:42:36] <jdh> going up to mt pilot?
[12:42:48] <tjb1> Id be scared to back something that heavy that far into the far…those dock areas are usually pretty slippy
[12:42:54] <tjb1> that far into the water...
[12:45:10] <jdh> in my world, boats go in saltwater.
[12:47:09] <archivist> the truck was probably front wheel drive
[13:12:57] <jdh> any code generators that will generate a hex bolt head kind of thing?
[13:13:34] <IchGuckLive> 5 or 6 edges
[13:13:49] <jdh> isn't 6 required to be hex?
[13:14:00] <Connor> jdh: ROFL. :)
[13:14:05] <archivist> hex....how does five fit in that
[13:14:32] <jdh> must be some metric thing
[13:15:04] <archivist> you should see a hex turning lathe though
[13:17:47] <IchGuckLive> do you need a Drawing or a G-code
[13:17:58] <jdh> either.
[13:18:13] <Connor> jdh: Watcha making ? :)
[13:18:35] <jdh> I rounded off my 32mm valve body
[13:18:40] <jdh> so, I want to make it 30mm
[13:18:54] <archivist> http://www.cnczone.com/forums/general_metalwork_discussion/54494-cutting_hex_cnc_lathe_live.html
[13:19:01] <Connor> valve ? Scuba ?
[13:19:14] <IchGuckLive> there are so many dwg out
[13:19:29] <jdh> scuba compressor valve
http://tinyurl.com/btu5yqt
[13:20:05] <IchGuckLive> jdh: what cad do you got
[13:20:12] <jdh> draftsight
[13:20:21] <jdh> or autocad if I wait long enough for it to load
[13:20:39] <IchGuckLive> http://www.cadcorner.ca/cadblocks/Mechanical/Fasteners/Bolts%20Hex%20Head.dwg
[13:21:03] <IchGuckLive> solidworks has this funktion
[13:22:25] <IchGuckLive> jdh: what about heekscad this also got the poligone funktion with one klick polygone and G-code
[13:23:43] <jdh> no SW, no patience for heeks
[13:24:37] <IchGuckLive> il doo a tutorial video on this
[13:27:18] <IchGuckLive> jdh:
http://mechmo.de/hex.png
[13:27:26] <jdh> guess a little thought on my part would have done wonders.
[13:28:10] <IchGuckLive> ok then till tomorrow
[13:29:00] <IchGuckLive> By and keep up to the Earth as it is supost to go down today the 240+ time
[13:50:26] <tjb1> Well somebody is working at Gecko, my drive got delivered atleast
[13:50:51] <p0st4L> :)
[14:00:13] <tjb1> I could be walking tomorrow for my 2 associates ;)
[14:00:40] <tjb1> https://dl.dropbox.com/u/3090951/IMG_20121221_144059.jpg
[14:03:59] <jdh> it was 70 and sunny here yesterday.
[14:10:09] <tjb1> 32 here now
[14:11:04] <AR_> it's been flurrying here all day
[14:11:14] <AR_> in eastern PA
[14:12:24] <tjb1> Im in western PA
[14:13:55] <jdh> only 50 here today, and blowing. Below freezing tonight. 2nd time this season I think.
[14:40:39] <tjb1> My facebook is full of winners… "120 past tractor trailers on snow covered roads. Kinda makes me miss winters. Time of the year that separates the people who think they can drive and the people who can."
[14:42:20] <tjb1> Natural selection at its finest ^
[15:44:34] <tjb1> Heh
https://dl.dropbox.com/u/3090951/Screen%20Shot%202012-12-21%20at%204.31.32%20PM.png
[15:44:47] <tjb1> cheap aircraft grade aluminum...
[16:30:55] <kwallace1> I just hooked up a ShuttleXpress but the udev rule doesn't seem to change the permissions to /dev/hidraw0. I used this link:
http://linuxcnc.org/docs/html/man/man1/shuttlexpress.1.html
[16:32:18] <kwallace1> Changing permission manually using "sudo chmod +r /dev/hidraw0" does work but my guess is that isn't the best way to do this.
[16:32:30] <AR_> horrible implementation
[16:32:34] <AR_> please buy a new computer
[16:34:07] <kwallace1> Oops, I meant ShuttleXpress USB MPG joggy thingy.
[16:35:28] <kwallace1> http://retail.contourdesign.com/?/products/22
[16:36:29] <jdh> kwallace1: I hooked mine up, used the sample hal stuff, it worked as-is
[16:38:41] <kwallace1> My search didn't turn up much, so I figured this isn't usually a problem, I must be special.
[16:39:12] <jdh> that's what they say. I'm sure they only mean good things by that.
[16:39:52] <jdh> mine cost like $38 from adorama
[16:40:58] <jdh> my dev is +r, but I don't recall changing it. Not likely I would remember though.
[16:42:44] <kwallace1> Oops, I double checked that I created the rules.d file and I don't see it. I'll try again. brb
[16:43:40] <jdh> what file?
[16:46:11] <jdh> and, I got my valve cover off!. I milled it down to a 27mm hex and it came right off.
http://tinyurl.com/bt828er
[16:47:06] <kwallace1> /etc/udev/rules.d/99-shuttlexpress.rules
[16:48:14] <cradek> kwallace: pastebin your shuttlexpress.rules?
[16:48:25] <cradek> oh
[16:48:28] <jdh> yep
[16:48:34] <archivist> jdh, the cutting vibration may have helped too
[16:49:39] <archivist> jdh, but the flange now has less support so may have sprung back a bit to reduce the grip
[16:49:47] <jdh> probably. It wasn't a crush gasket after all, there is a large rubber seal
[16:49:49] <DJ9DJ> gn8
[16:51:21] <jdh> If I can get the set screw out of there, I could reuse the body. Replacement version looks like:
http://www.compressor-parts.com/cart/images/NP-2201-0030A.jpg
[16:51:29] <kwallace1> Yup, that was it, did a "save as" but to my home, doh. Is there a way to reload udev rules without rebooting?
[16:52:46] <cradek> service udev restart
[16:53:13] <kwallace1> Thanks.
[16:53:32] <cradek> jdh: what is that thing?
[16:53:56] <kwallace1> jdh: Do you mean the bit with the acorn nut?
[16:53:56] <jdh> cradek: it's the discharge valve for the 2nd stage of a high pressure compressor
[16:54:05] <cradek> ahh
[16:57:37] <jdh> hardest part was guessing the proper orientation for the G10 L2 P1 R<something close>
[17:08:05] <kwallace1> jdh: I think the screw goes out the wide end of the body. You'll need to get the acorn nut off.
[18:45:29] <r00t4rd3d> http://i.imgur.com/tmFLG.jpg
[19:20:37] <andypugh> Scrolling back: I have machined hexes on my lathe _without_ live tooling. :-)
[19:21:31] <andypugh> Just posted a minor update on my Harrison build to the 'Zone. Can't believe it is so long since the last one.
[19:25:20] <andypugh> http://www.cnczone.com/forums/showthread.php?p=1212353#post1212335
[19:25:33] <dr00bie> andypugh: did you get a chance to take a look at pycam yet? I am still not understanding how to get meaningful gcode from it...
[19:26:16] <andypugh> I am nowhere near needing CAM at the moment. Tonight, for example, was all MDI.
[19:27:22] <andypugh> (Lots of M3S1000G98G81X10Y20R0.1Z-10 type stuff drilling holes and boring registers)
[19:28:08] <andypugh> However I _did_ use polar coordinates for the first time. I am hooked, it's brilliant for bolt circles.
[19:28:17] <jdh> mdi typos are the cause of all my vise dings
[19:29:28] <andypugh> If you don't know, as well as using XY in the current plane you can use @ and ^. @ is radius from X0 Y0 and ^ is angle (with 0 degrees along X).
[19:29:47] <andypugh> So, for a 6-bolt circle of 25mm radius.
[19:29:51] <PCW> wheres the MDI preview when you need it (with 3d vise models)
[19:30:03] <andypugh> G0 @25 ^0
[19:30:19] <andypugh> G98G81..... (hole cycle)
[19:30:32] <andypugh> G0 ^60 G98......
[19:30:41] <andypugh> G0 ^120 G98......
[19:30:49] <andypugh> and so on.
[19:31:25] <cradek> you can just mdi ^60 and ^120 to repeat the cycle at the new angle
[19:31:28] <PCW> well if you are dealing with circles, Cartesian coordinates are awkward
[19:32:43] <andypugh> cradek: I thought I probably could, but I had the time, and wasn't that confident.
[19:32:58] <cradek> drill cycles are so cool that way
[19:33:32] <cradek> actually cycles are have repeats - and you can use polar - so you can do all the holes on your circle with one command
[19:33:58] <cradek> g91 ^60 g8x ... l6
[19:34:16] <andypugh> Brilliant.
[19:34:32] <andypugh> And I bet nearly 2 people ever have :-)
[19:34:35] <cradek> (from memory, haha)
[19:34:47] <cradek> I sure have used it that way, so >= 1
[19:35:35] <andypugh> I probably will next time. But this was only 3 holes.
[19:36:23] <cradek> I wonder how many people have actually used our polar coordinates
[19:36:32] <andypugh> There is a bolt-circle generator in the Wiki, perhaps it ought to be mentioned that it is for the Weak!
[19:36:38] <cradek> haha
[19:37:04] <cradek> polar is tricky. g0 x0 y0; g0 @1 => ???
[19:37:17] <andypugh> Well, there is one more user as of today. I knew they existed, but hadn't until tonight.
[19:37:21] <cradek> I probably spent more time designing and writing it than anybody has ever saved using it
[19:38:13] <andypugh> Is it unique to LinuxCNC?
[19:38:37] <cradek> lots of other controls have various ways of doing polar, but of course ours is the best implementation :-)
[19:38:48] <cradek> (that's just a fancy way of saying yes)
[19:42:40] <andypugh> What other characters can we add to our G-code? We could be the APL of G-code :-)
[19:43:14] <cradek> there are NOT very many left if you restrict yourself to 7 bit ascii, and you sure as hell should
[19:43:37] <jdh> unicode
[19:43:42] <cradek> ` & {} \ |
[19:43:48] <cradek> ' "
[19:43:57] <cradek> or so?
[19:44:21] <andypugh> ? is free. Could launch a dialog for numeric entry. I only said that for fun, but actualy it's not a stupid idea,
[19:44:46] <jdh> I have often wanted to be able to input a Z value
[19:44:48] <cradek> stupid is in the eye of the beholder
[19:46:27] <andypugh> So how do you behold it?
[19:47:45] <andypugh> Spotting a key on the keyboard that isn't used in G-code isn't the most sensible way to come up with the idea for a new feature, I have to admit.
[19:48:21] <andypugh> You would rather expect to start from the other end.
[19:48:58] <AR_> i thought about making a higher level gcode
[19:49:17] <AR_> mov x2.450 y-2.00
[19:49:21] <cradek> not sure how I behold it. it's not a particularly flexible way to get parameters from the user. ngcgui and touchy macros both already do a much less bad job.
[19:49:25] <AR_> arc
[19:49:28] <AR_> etc
[19:50:04] <cradek> every ? found on the line should cut the feed rate in half
[19:50:12] <cradek> g0 y0???
[19:50:12] <AR_> lol
[19:50:16] <AR_> why
[19:50:20] <cradek> that gives you 1/8 of rapid feed
[19:50:35] <cradek> it's how you mark the level of uncertainty in the gcode, obviously
[19:50:36] <AR_> how about just another parameter
[19:50:52] <jdh> like f?
[19:50:57] <AR_> ^
[19:50:58] <cradek> snrk
[19:51:21] <cradek> my way is more readable
[19:51:31] <AR_> lol
[19:52:36] <andypugh> AR we already use ^, are you suggesting we overload it like Q?
[20:27:50] <AR_> no andypugh
[20:27:52] <AR_> omg
[20:49:50] <tjb1> ello.
[23:45:48] <abetusk> anyone have a good gerber to gcode link?
[23:59:40] <tjb1> mmmmm, 4G -
http://www.speedtest.net/result/2387667661.png
