#linuxcnc | Logs for 2013-05-06

Back
[00:07:33] <ssi> my C is perilously rusty
[00:47:23] <ssi> mhaberler: just the man I wanted to see!
[00:48:37] <mhaberler> what's up?
[00:48:59] <ssi> I started writing a skeleton for the hal driver, and I'm having a lot of trouble with the shared memory and registering pins with hal
[00:49:17] <mhaberler> ok, which trouble?
[00:49:29] <ssi> I have a port_data_t struct, and it contains an array of hal_bit_t pins
[00:49:35] <mhaberler> ok
[00:49:50] <ssi> hal_pin_bit_newf wants a hal_bit_t**
[00:49:58] <ssi> and I've had to jockey code around a bit to give it that
[00:50:04] <mhaberler> yes
[00:50:04] <ssi> and even in a way that I think should work, I still get:
[00:50:11] <ssi> HAL: ERROR: pin_new(bb_gpio.0.in-45) called with already-initialized memory
[00:50:58] <mhaberler> didnt you allocate the blockof memory where the pin pointers live with hal_malloc() ? (just lift the code from hal_parport.c)
[00:51:05] <ssi> yeah, I did
[00:51:25] <ssi> in fact, now I'm hal_malloc'ing the port_data structure, and it contains an array of bit pointers,
[00:51:34] <ssi> and I'm hal_mallocing each bit pointer individually
[00:51:44] <mhaberler> can you push it so I can reproduce?
[00:51:47] <ssi> sure
[00:52:45] <ssi> aw crap
[00:52:49] <ssi> I gave you the imcmahon user
[00:52:54] <ssi> but on the BBB I'm logged in as "linuxcnc"
[00:52:59] <ssi> wait I can fix that in .ssh config
[00:54:12] <ssi> ...maybe
[00:55:34] <ssi> ugh
[00:55:41] <ssi> I'm getting: fatal: remote error: access denied or repository not exported: /emc2-dev.git
[00:55:44] <mhaberler> just create a imcmahon user and push with that, probably easier
[00:55:48] <ssi> and it doesn't give me any clues as to the user it's using
[00:55:51] <ssi> ok I'll try it
[00:55:55] <mhaberler> oh, the git push uri
[00:56:11] <ssi> linuxcnc@arm:~/linuxcnc-dev$ git push --verbose
[00:56:11] <ssi> Pushing to git://git.mah.priv.at/emc2-dev.git
[00:56:13] <mhaberler> edit .git/config in the emc2 repo you're pushing
[00:56:24] <ssi> ok..
[00:57:08] <ssi> that did it
[00:57:19] <ssi> take a look at src/hal/drivers/hal_bb_gpio.c
[00:57:27] <mhaberler> ok
[00:57:42] <ssi> and like I said, I've monkeyed around with the structures a bit
[00:59:01] <ssi> OH
[00:59:03] <mhaberler> dont worry, but need to head over to the office, no bb here; it will be 15mins or so
[00:59:10] <ssi> my structure needs to have the pointers to the pin data
[00:59:13] <ssi> but hal will create the actual memory
[00:59:16] <ssi> I bet that's it
[00:59:32] <mhaberler> yeah, its very confusing initially with pins
[00:59:55] <ssi> yeah I think that did it
[01:00:15] <ssi> Loaded HAL Components:
[01:00:16] <ssi> ID Type Name PID State
[01:00:16] <ssi> 32770 RT hal_bb_gpio ready
[01:00:16] <ssi> 8478 User halcmd8478 8478 ready
[01:00:16] <ssi> Component Pins:
[01:00:18] <ssi> Owner Type Dir Value Name
[01:00:20] <ssi> 32770 bit IN FALSE bb_gpio.0.in-00
[01:00:23] <ssi> 32770 bit IN FALSE bb_gpio.0.in-01
[01:00:24] <mhaberler> bingo
[01:00:25] <ssi> 32770 bit IN FALSE bb_gpio.0.in-02
[01:00:28] <ssi> yeah
[01:00:30] <ssi> sweet
[01:00:54] <ssi> ok you go to the office, I'm going to BED
[01:00:58] <ssi> that's a good note to end on
[01:00:59] <mhaberler> I thought about collisions a bit (say pru)
[01:01:03] <ssi> yea?
[01:01:04] <mhaberler> I have a simple solution
[01:02:02] <mhaberler> we use a shared bitmap which is in initialized data and an exported symbol (actually two, the bitmap and size)
[01:02:06] <ssi> haha I have to fix my user info in git on the bbb
[01:02:22] <mhaberler> it has bits set to 1 for system-type pins (eg etc)
[01:02:53] <mhaberler> every hal driver needs to check the pins it uses against the bitmap, and may do so only if bit for that pin is 0
[01:03:11] <mhaberler> if she succeeds, set bit of that pin to 1
[01:03:19] <mhaberler> I'll do a prototype
[01:03:20] <ssi> that makes sense
[01:03:54] <mhaberler> however we'll switch your code to a development branch, say bb-hal-gpio
[01:04:41] <ssi> that would make me happy
[01:04:43] <mhaberler> jeesh, I really need to port that bitmap library
[01:05:00] <ssi> I dunno if you can fix the usernames on those last three commits or not
[01:05:09] <ssi> if not, I guess Joe Chipcutter will have a spot in the credits ;)
[01:05:25] <mhaberler> we'll divine the author
[01:05:29] <ssi> hahah
[01:06:34] <mhaberler> anyway, switch branches, push again, and I'll reset the arm335x-hal-pru-module branch to where it was so we dont trample in Charles' garden
[01:06:39] <ssi> ok
[01:07:12] <mhaberler> all these nice swamp flowers ;)
[01:07:30] <ssi> walk me through this so I do it right
[01:07:36] <ssi> git branching isn't my strong suit :/
[01:07:44] <mhaberler> git branch bb-hal-gpio
[01:07:48] <mhaberler> git checkout bb-hal-gpio
[01:08:09] <mhaberler> git push <repo> bb-hal-gpio (repo is likely 'origin')
[01:08:35] <mhaberler> it will come, it is a bit steep to start with, yes
[01:08:55] <mhaberler> perfect, worked fine
[01:09:14] <ssi> ok
[01:09:21] <ssi> so this branch already has the commits I made to the other?
[01:09:45] <ssi> yeah I guess it does
[01:10:11] <mhaberler> not anymore, I just pruned them
[01:10:24] <ssi> pruned them from bb-hal-gpio as well?
[01:10:28] <mhaberler> no
[01:10:31] <ssi> ok
[01:10:39] <mhaberler> you have your private sandbox now ;)
[01:10:47] <ssi> love it
[01:11:17] <mhaberler> dont worry about what's being pushed now, that can be cleaned up eventually before merging
[01:11:18] <ssi> so does it make sense that the pin names in hal reflect the physical pins on the device?
[01:11:35] <ssi> and if they can't be claimed, they simply won't exist?
[01:11:36] <mhaberler> I think so, confusion reduction
[01:11:56] <ssi> I was helping a guy in #beagle earlier who wanted to deal with GPIO in C
[01:12:02] <mhaberler> I would think the driver should either give a warning or fail
[01:12:03] <ssi> and he was going crazy cause nothing would work, including my code
[01:12:07] <mhaberler> aja
[01:12:11] <mhaberler> pinmux..
[01:12:12] <ssi> and it turns out he thought GPIO1_28 was P8, pin 28
[01:12:26] <ssi> so yes, there's a lot of room for confusion
[01:12:30] <mhaberler> yeah, thats standard
[01:12:41] <mhaberler> no, use header pin names, that's the way to go
[01:12:53] <mhaberler> port pins.. reverse engineering hell
[01:12:56] <ssi> yep
[01:13:12] <ssi> and it's terrible, because there's header pins, port pins, package ball pins
[01:13:20] <ssi> the mode0 name for a given pin
[01:13:24] <ssi> on and on
[01:13:50] <mhaberler> right… but old Joe most likely figures header pins and gets stuck with any other naming scheme
[01:14:13] <ssi> so the other thing I'm not sure about is if pin direction should be defined at driver load time or runtime a la mesa
[01:14:31] <mhaberler> let me see the mesa manpage
[01:15:15] <ssi> hostmot2
[01:15:17] <ssi> gpio section
[01:15:18] <ssi> is_output
[01:15:45] <ssi> I imagine he's registering the pins in hal as IO
[01:16:01] <ssi> and then just treating them differently in read_port/write_port depending on the parameter setting
[01:16:34] <mhaberler> maybe you want to read the section in the manual about canonical pin naming
[01:16:50] <ssi> which manual is that
[01:16:56] <mhaberler> let me see
[01:16:58] <ssi> hal manual I assume
[01:17:21] <mhaberler> http://www.linuxcnc.org/docs/html/hal/canonical-devices.html
[01:18:38] <mhaberler> personally I dont see much use for runtime direction changing, and it might be hairy
[01:18:43] <ssi> ok
[01:19:02] <ssi> so then configuration bitmasks
[01:19:05] <ssi> input mask, output mask
[01:19:53] <mhaberler> yes, maybe variations for pullup. pulldown, floating
[01:20:13] <ssi> that's where my original suggestion came from
[01:20:21] <ssi> charstring, -, O, I, U, D
[01:20:34] <ssi> for "not connected", output, floating, pullup, pulldown
[01:20:45] <mhaberler> sorry, please dont - this string sausage will be the perfect error trap
[01:20:48] <ssi> right
[01:21:07] <ssi> I just am having a hard time finding the middle ground between that and andy's runtime parameter suggestion
[01:21:24] <mhaberler> make sure you break the rules in time ;)
[01:21:44] <ssi> what about a mask for input, a mask for output, and then have pullups be runtime configured?
[01:22:08] <mhaberler> kiss...
[01:22:09] <ssi> (I haven't been successful in making a pin anything but a pullup so far, so maybe that's moot)
[01:22:50] <mhaberler> (philosophical) re runtime changes in hal: hal supports too much of this to start with
[01:23:14] <mhaberler> we eventually need a clear HAL session concept where a HAL setup gets committed and frozen
[01:23:38] <mhaberler> so components dont have to chase possible halcmd changes on the fly, which can be hairy
[01:23:52] <ssi> makes sense,
[01:23:59] <ssi> but from a pure configuration language perspective
[01:24:01] <mhaberler> in some cases that translates into runtime overhead
[01:24:21] <ssi> foo.pin-01.pullup = TRUE on fifteen lines is easier to deal with than a mask string
[01:25:03] <mhaberler> fifteen lines.. dont get it; the config must be a loadrt hal_gpio param(s)
[01:25:12] <mhaberler> oh I see
[01:25:19] <mhaberler> yeah sure
[01:25:41] <mhaberler> lots of useless pins - swamps namespace
[01:25:57] <ssi> another alternative might be to take a "firmware" approach
[01:26:16] <ssi> where there's a file that defines pinmuxing, and filename is passed in on config line
[01:27:17] <mhaberler> we currently dont use that and not sure if it makes sense to create an extra external file dependency if halcmd proper suffices
[01:27:30] <ssi> ok
[01:28:13] <ssi> again, drawing ideas from the mesa drivers... they sort of work that way with fpga firmwares
[01:28:19] <mhaberler> I have been looking into a flex scanner for the config strings; just use the simplest method for now to get it to work and we'll polish later
[01:28:24] <mhaberler> right
[01:28:30] <ssi> although I don't know the specifics of how hm2 configures pins based on the firmware you choose
[01:28:58] <ssi> yea for right now I'm going to do either a pair of mask strings or a pair of comma separated pin number strings
[01:29:01] <mhaberler> I think the firmware and the firmware params define the pin usage
[01:29:02] <mhaberler> and its
[01:29:02] <ssi> the latter might be better initially
[01:29:10] <mhaberler> read back from the setup info
[01:29:17] <ssi> cause I think we're more likely to have fewer than half of the pins on a header in use
[01:29:20] <ssi> at least as GPIO
[01:29:21] <mhaberler> that case we dont really have
[01:29:52] <ssi> so something like input=1,2,3,4,10,15 output=5,6,8,15 might be better than input=XXXX------X----X etc
[01:30:07] <mhaberler> absolutely, much less error potential
[01:30:23] <mhaberler> no more column counting like in COBOL or FORTRAN ;)
[01:30:26] <ssi> :)
[01:31:01] <ssi> what do you think about adding 100 or 200 to the pin number for P8/P9?
[01:31:08] <ssi> so P8, pin 1 would be 101, P9 pin 1 would be 201
[01:32:02] <mhaberler> use it for now; I'll implant a decent scanner eventually which can more easily work with board-specific pin naming
[01:32:07] <ssi> ok
[01:32:12] <mhaberler> so eventually we'll have input=p8-2 etc
[01:32:32] <ssi> that works
[01:32:40] <ssi> and I'll just run with all inputs are pullup for now
[01:32:45] <ssi> since I can't seem to make it do otherwise anyway
[01:32:48] <mhaberler> that was due anyway.. everybody starts up massaging strings by char inspection ca 1969 style..
[01:33:12] <mhaberler> that'd be fine
[01:33:35] <ssi> ok well, I'll get some sleep, and hopefully tomorrow I'll be blinking leds in halcmd :)
[01:33:40] <mhaberler> sure
[01:33:44] <ssi> thanks for the help
[01:33:54] <mhaberler> btw whats the case with the onboard leds
[01:34:03] <ssi> what about them?
[01:34:05] <ssi> they're on GPIOs
[01:34:09] <mhaberler> you detect those and disable kernel fiddling with same?
[01:34:09] <ssi> but I'm not sure if they're shared with pins
[01:34:10] <ssi> I think not
[01:34:19] <mhaberler> ok, random detail
[01:34:32] <ssi> well, what I did for now is build a kernel with the gpio_leds driver modularized
[01:34:42] <mhaberler> aja, I remember
[01:34:43] <ssi> I don't yet know how to disable the leds driver
[01:35:01] <mhaberler> blacklist in /etc/modules or somesuch
[01:35:09] <ssi> oh I don't mean the module
[01:35:14] <ssi> I easily can not load the module
[01:35:24] <ssi> I'm not sure if there's a way you can tell the driver to release the gpios via sysfs or someting
[01:35:42] <mhaberler> you mean stop kernel wiggling the leds.. there is a way, I just dont remember
[01:35:44] <ssi> yeah
[01:35:50] <ssi> I'll look into it, but it's not a high priority
[01:36:02] <ssi> also, do we still think that the way to tell the kernel we're using pins is sysfs?
[01:36:12] <ssi> or is it feasible to call gpio_request directly?
[01:36:20] <mhaberler> best bet: eewiki - Robert Nelson; bet #2 - #beagle (which is a tad unprofessional if I may say so..)
[01:36:53] <mhaberler> gpio_request() is a userland function, or kernel? (loosing track...)
[01:36:59] <ssi> I think it's kernelspace
[01:37:05] <mhaberler> then no - sysfs
[01:37:08] <ssi> ok
[01:37:14] <mhaberler> we dont want kernel code if we can avoid it
[01:37:34] <ssi> what's the penalty?
[01:37:50] <mhaberler> sysfs walking at startup
[01:38:22] <ssi> I mean what is the downside to having kernel code in our drivers?
[01:38:32] <mhaberler> well to start with, it's harder to debug, and the whole point of the userland threads effort was to get rid of kernel code dependencies
[01:38:41] <ssi> gotcha
[01:38:45] <ssi> oh that raises another question
[01:39:00] <ssi> is it straightforward to debug these drivers loading with gdb or something?
[01:39:05] <ssi> I wasn't really sure how to go about it
[01:39:06] <mhaberler> absolutely
[01:39:21] <mhaberler> but trapping the call isnt totally obvious, you do it like so
[01:39:36] <mhaberler> start realtime, this will give a pid for rtapi_app
[01:39:52] <mhaberler> now 'gdb -p <pid of rtapi_app>'
[01:39:59] <mhaberler> b rtapi_app_main
[01:40:01] <mhaberler> continue
[01:40:08] <ssi> start realtime... halrun will do that?
[01:40:21] <mhaberler> then loadrt the driver, peng, breakpoint in rtapi_app_main
[01:40:27] <mhaberler> no, shellscript
[01:40:34] <mhaberler> 'realtime start'
[01:40:49] <ssi> ok
[01:41:00] <ssi> returns with no output,
[01:41:02] <mhaberler> need to check what's needed in this branch; it could be it isnt yet
[01:41:05] <ssi> and I don't see a process either rtapi or realtime
[01:41:11] <ssi> oh I'm still on vanilla kernel
[01:41:14] <ssi> maybe that screws me up?
[01:41:17] <mhaberler> doesnt matter
[01:41:34] <mhaberler> no rtapi_app might start only after first module load in that branch
[01:41:41] <mhaberler> so halcmd load or2 or so
[01:41:47] <mhaberler> then you'd have the process
[01:42:04] <ssi> root 9167 1.0 0.2 33888 1276 ? Ss 06:28 0:00 /home/linuxcnc/linuxcnc-dev/bin/rtapi_app load or2
[01:42:04] <mhaberler> then attach gdb
[01:42:07] <ssi> that one?
[01:42:09] <mhaberler> yes
[01:42:14] <mhaberler> gdb -p 9167
[01:42:24] <mhaberler> b rtapi_app_main
[01:42:33] <mhaberler> coninue
[01:42:44] <mhaberler> in a different window do 'halcmd loadrt hal_gpio'
[01:42:52] <ssi> yep got it
[01:42:53] <mhaberler> that should break you in the module init routine
[01:43:32] <mhaberler> (that used to be kernel-only with the rtai flavor.. extra penalty, printk and all)
[01:44:14] <ssi> are the x86 builds movin away from rtai to xenomai as well?
[01:44:44] <mhaberler> no, dont think so - we'll have x86 builds which run on all of them (rtai, xenomai, rtpreempt)
[01:44:48] <ssi> gotcha
[01:45:34] <mhaberler> there's work in progress to support all rtos flavors with a single linuxcnc build so no more reinstall needed, just boot different kernel and go
[01:46:18] <ssi> looks like what you get if you just hit the website and download first available image is rtia
[01:46:23] <ssi> that's what's on the machine I just built
[01:46:31] <mhaberler> yes
[01:46:50] <mhaberler> it used to be the only game in town and in terms of pre-built packages it still is
[01:49:05] <mhaberler> hihi, joe@chips.com ;)
[01:49:27] <ssi> yep :)
[01:50:48] <ssi> ok bedtime.. gnite :)
[01:51:26] <mhaberler> hey.. re hostmot2..
[01:51:32] <mhaberler> that's a bug, not a feature
[01:52:04] <mhaberler> you need to set things up post-load since it cant be done now with the loadrt args
[01:52:12] <mhaberler> good night
[02:11:12] <DJ9DJ> moin
[02:20:05] <Loetmichel> moin
[05:50:10] <andypugh> I am semi-accidentally subscribed to a thread on CNCZone. Until I started reading it, I hadn't heard of CAM-Soft. I am surprised how accepting this user seems to be of a strange quirk it appears to have: http://www.cnczone.com/forums/camsoft_products/178451-linux_operating_system-2.html#post1265490
[05:56:54] <Loetmichel> as long as it finishes the job: whats to worry about?
[05:57:33] <Loetmichel> some time in the future i would buy a faster PC if i were ih his place, but it simms to work anyway? so why bother?
[05:57:39] <Loetmichel> seems
[07:13:57] <andypugh> Loetmichel: Well, it sounds like in that case it didn't finish the job, it stopped the Y move early then randomly ran off in the X direction.
[07:15:10] <jthornton> doesn't sound like he had the e stop wired properly
[07:16:49] <andypugh> Eh? He seems to be saying that e-stop was the only way to stop the unprogrammed movement.
[07:17:09] <andypugh> Anyway, whatever, it's a CAMSoft problem, not ours.
[07:17:39] <jthornton> I thought he said it didn't stop the movement... I must not be fully awake
[07:44:00] <Loetmichel> andypugh: If i read correctly that was an other pal, not the OP ;-)
[09:29:13] <R2E4> anyone get msea boards woking with Sanyo Denka drives on a hitachi seili vm40?
[09:30:58] <andypugh> Are the drives + / - 10V?
[09:31:07] <R2E4> yes
[09:31:56] <L84Supper> treat them like any other +/- 10V servo drive
[09:33:07] <andypugh> Should be simple then.
[09:34:24] <andypugh> How likely is it that the distance from the top of my table to the leadscrew is 3 11/32" ? Do you Imperialists use such not-round numbers?
[09:34:26] <R2E4> I figured that portion would work. The issues would be everything else..... The logic, there is alot of connections between the seicos BS controller and the drives.....
[09:35:03] <andypugh> Well, you only asked about the drives...
[09:35:07] <jdh> andypugh: who designed it?
[09:35:22] <andypugh> Harrison, Heckmondwike in about 1970.
[09:35:26] <R2E4> hehe..... yeah....
[09:35:57] <jthornton> andypugh, very likely
[09:36:32] <andypugh> My measurements give 84.87mm. So it _could_ be 85mm.
[09:39:57] <jthornton> is the rest of the machine metric?
[09:41:30] <jthornton> I have had machinery with imperial, metric, and whitworth threads on it from Taiwan
[09:42:26] <L84Supper> they use everything, not like the USA where it's mostly imperial or the EU where it's all metric
[09:44:23] <jdh> how about a machine with whitworth threads and lucas electronics
[09:46:00] <TekniQue> and a rover A series engine?
[09:46:40] <Valen> we have a lathe that has a belt in it with metric tooth size but imperial spacing
[09:46:54] <jdh> maybe a drip lube system... not designed, it just happens to drip.
[10:00:38] <andypugh> This machine does appear to have some metric dimensions, but has a little plaque with a picture of a bolt and the legend "UNC"
[10:01:13] <TekniQue> A funny artifact of the american snowmobiles back in the days
[10:01:38] <TekniQue> both Arctic Cat and Polaris used japanese engines in their machines
[10:01:46] <TekniQue> so every bolt on the engines would be metric
[10:02:04] <TekniQue> while the chassis would be imperial
[10:02:32] <andypugh> Morris trucks of the 1920s were like that. Metric engines, imperial chassis. Odd, as everything was made by Morris.
[10:10:37] <Tom_itx> lucky we've made it as far as we have ehh?
[10:13:00] <L84Supper> we've had imperial margarine since the 60's
[10:13:28] <pcw_home> Fit for a King
[10:13:52] <JT-Shop> ssi: I am now
[10:31:21] <ssi> what was I gonna ask you
[10:32:06] <ssi> oh right, the mach3 conf
[10:32:10] <ssi> but it's at home, and I'm at work now
[10:32:12] <ssi> so carry on :D
[10:37:33] <JT-Shop> ok, just email it to me
[10:38:12] <ssi> ok thanks
[10:56:38] <ssi> I have this weird hangup about invoicing people
[10:56:50] <ssi> I always feel like they're gonna balk at the number and/or not pay me
[11:00:34] <chopper791> test
[11:01:28] <ssi> 5x5
[11:02:06] <L84Supper> ssi: look at it this way, if they balk or act like it's too expensive you really didn't want that project anyway
[11:02:15] <ssi> yeah but the project is done :)
[11:03:00] <L84Supper> time to start quoting ahead or you are going to get people that don't want to pay for whatever reason
[11:03:17] <ssi> I don't think it'll be a problem, because I know what they were quoted just to attempt to repair their old control, and my number came in within 5% of that number
[11:03:22] <ssi> I just get anxiety about it
[11:07:27] <chopper791> Have a couple questions about the thcud.comp file. The first question- There are the following pins in the comp file, pin in float z_pos_in "Z Motor Position Command in from axis.n.motor-pos-cmd";
[11:07:27] <chopper791> // Output Pins
[11:07:27] <chopper791> pin out float z_pos_out "Z Motor Position Command Out";
[11:07:27] <chopper791> pin out float z_fb_out "Z Position Feedback to Axis";
[11:07:27] <chopper791> Do those need to be in the HAL file for them to work?
[11:09:01] <JT-Shop> yep
[11:13:28] <chopper791> Ok- So I need to place them in the Z axis(Joint 2) section to allow axis.2.pos-cmd to go to z_pos_in then, so on and so on?
[11:15:12] <chopper791> I thought I needed to do that, but when I do I get an error when launching stating that pin z_pos_in does not exist.
[11:26:44] <ssi> quiet in here today
[11:26:47] <ssi> a little... too quiet
[11:27:07] <fomox> anyone here in the essen area in germany?
[11:29:26] <JT-Shop> I'm stumped, Z keeps incrementing by the depth of cut each time I reload the subroutine
[11:31:32] <L84Supper> fomox: hang around a bit, they should be around
[11:32:11] <fomox> L84Supper: Nice
[11:33:19] <L84Supper> fomox: have questions or recruiting for a cnc army? :)
[11:33:39] <fomox> On a vacation there and would love to see some DIY cnc in action :P
[11:33:49] <fomox> and just chat, maby they know about hackerspaces etc.
[11:35:55] <chopper791> Ok-I keep receiving errors about thcud.z_pos_in does not exist. Here is my HAL file so maybe some light can be shed on this. http://pastebin.com/LaKyMLwY
[11:36:13] <chopper791> I have tried this with out the thcud also and keeps coming up with the same error.
[11:37:21] <chopper791> Any suggestions?
[11:38:52] <ssi> is there a component number?
[11:38:57] <ssi> thcud.0.z_pos_in?
[11:40:42] <ssi> guess not
[11:40:54] <chopper791> The comp is vague when it comes to this section. There is nothing in the comp file that I seen that would indicate a component number
[11:42:24] <JT-Shop> probley a spelling error
[11:42:33] <pcw_home> halcmd show pin | grep thcud should be illuminating
[11:43:59] <chopper791> pcw- I tried that and " could not open shared memery apprears
[11:44:10] <chopper791> memory
[11:44:40] <chopper791> jt- yeah my spelling has some errors sometimes.
[11:44:52] <JT-Shop> comment out any offending pins till it loads then use Show Hal Configuration to see the names of the pins
[11:45:40] <pcw_home> thats probably because linuxcnc was not running
[11:46:00] <chopper791> JT & pcw - that worked.
[11:46:45] <chopper791> sthe pins show it a z-pos-in and not z_pos_in as shown in the comp
[11:46:47] <pcw_home> you can re-direct the output of halcmd show pin to a file as a donor file when editing the hal file
[11:47:18] <JT-Shop> that is comp magic working on you
[11:49:20] <JT-Shop> also in a terminal when LinuxCNC is not running you can do halrun, loadrt thcud, show pin, the comma indicating press the enter key
[11:50:03] <chopper791> got ya
[11:50:29] <chopper791> So I ned to change the underscores to dash and then it should be fine I would guess?
[11:51:29] <chopper791> or maybe not as all of the pins in the comp are under scores and not dashes, but the terminal shows dashes.
[11:51:52] <chopper791> What is this comp magic you speak of JT?
[11:52:18] <chopper791> The ability to change dashes to underscores or was I not readignt he sarcasm correctly
[11:52:18] <pcw_home> copy paste the results of show pin and you never need to type a name again
[11:52:36] <JT-Shop> the magic of converting _ to - when the component is built
[11:53:50] <chopper791> pcw- I am not following you on that
[11:53:57] <JT-Shop> now tell me why G0 Z-0.006 moves to -0.006 * the number of times I try and run my subroutine
[11:54:37] <chopper791> I can copy and paste them but to where?
[11:54:45] <pcw_home> if you load all the components and then do a a halcmd showpin > pins
[11:55:24] <pcw_home> you can open the pins file while you edit your hal file and copy/paste from it onto the hal file
[11:56:10] <pcw_home> so there is no question about naming
[11:56:10] <jdh> jt: any g92? writing #5223?
[11:57:25] <ssi> pcw_home: is sserial based on SPI?
[11:58:37] <pcw_home> No its async serial
[11:59:28] <ssi> ok... and what does LBP stand for?
[11:59:53] <pcw_home> Little Binary Protocol
[11:59:56] <ssi> :D
[12:01:25] <pcw_home> its just a wrapper so we can send our normal register read/write commands over a serial/usb/Ethernet link
[12:01:36] <ssi> gotcha
[12:01:51] <ssi> I'm trying to divine how the 7i77 works to figure out what it'd take to run it on the beagle
[12:02:26] <ssi> I guess the 25 pin just carries several channels of sserial?
[12:02:33] <ssi> it seems like none of them are just straing input/output pins
[12:03:12] <pcw_home> Probably tough but possible
[12:03:14] <pcw_home> the beagle has multiple serial ports that can run up to 3.698 MB
[12:03:16] <pcw_home> and be driven by the PRU
[12:03:55] <pcw_home> well 3.6something MB
[12:05:49] <ssi> how many ports does the 7i77 use?
[12:06:03] <JT-Shop> jdh: no G92 let me look at #5223
[12:06:04] <pcw_home> The PRU should be able to run effectively the same code that runs on the FPGAs processor
[12:06:05] <pcw_home> (this code hides the protocol timing and other complexities and presents a register interface to the HAL driver)
[12:06:43] <pcw_home> 3 channels (analog, digio, expansion)
[12:06:55] <ssi> aha
[12:07:37] <JT-Shop> the only variable I'm using is #5410
[12:07:54] <JT-Shop> I just did G0 Z0 and it went to -0.024
[12:10:28] <pcw_home> only pain is the the Sitaras UARTs are glorified 8250s so not flexible baud rate wise
[12:10:30] <pcw_home> and dont have the ability to read the number of chars in the buffer, so the code will be
[12:10:30] <frallzor> ahoyhoy
[12:10:31] <pcw_home> a bit awkward (you get spoiled being able to make the hardware match the task)
[12:10:50] <ssi> haha
[12:11:06] <ssi> well I have a good start on my GPIO driver
[12:12:52] <pcw_home> Good. It would be nice if the cape EEPROM said not only that the cape
[12:12:54] <pcw_home> reserved this or that pin but whether to allow the kernel driver to control them or not
[12:14:07] <JT-Shop> just for S&G I did a G53 G0 X0 Z0 then G92.1 then G0Z0 and it went to Z-0.024
[12:14:48] <frallzor> http://imageshack.us/a/img812/7607/img4922j.jpg Not really CNCed, but used my machine manually for it =P
[12:15:35] <ssi> pcw_home: yeah I've been rolling that around in my mind... if we build capes that are tailored to specific tasks, it could have metadata in the eeprom that the driver could inspect to configure the signals
[12:18:21] <generic_nick|2> yuck, rain.
[12:18:40] <IchGuckLive> hi all b)
[12:18:50] <ssi> generic_nick|2: yeah tell me about it
[12:19:01] <ssi> my basement has a half inch of water in it :/
[12:19:02] <generic_nick|2> pcw_home: figured out how to put the drive in velocity mode, works much better now, thanks.
[12:19:09] <pcw_home> Not sure what level of control is needed (setting up pin mux and GPIO)
[12:19:11] <pcw_home> but in any case if you wished to get along with other capes that likely use kernel drivers
[12:19:12] <pcw_home> you need something to say 'dont touch, we will use our own driver'
[12:19:25] <ssi> right
[12:19:28] <chopper791> I switched the _ to - in the HAL file and the errors are different now. The new error is 'thcud.z-pos-out' must not be the same as a pin. I would have to guess that I have this signal name in the wrong location or would it be caused from something else?
[12:19:35] <ssi> mhaberler and I have been talking about how to do that a lot lately
[12:19:39] <ssi> it's not an easy problem to solve
[12:20:09] <pcw_home> No, makes me like FPGAs more and more :-)
[12:20:10] <generic_nick|2> i had to uncheck a box in the software that allowed more options (like putting it in velocity mode)
[12:20:35] <pcw_home> Yeah velocity mode is usually much easier to tune
[12:20:40] <generic_nick|2> i have a tiny bit of p in linuxcnc and thats it
[12:20:48] <generic_nick|2> like .05
[12:21:21] <pcw_home> velocity mode uses FF1 also for best performance
[12:21:32] <generic_nick|2> couldnt add any, it made it worse
[12:21:41] <andypugh> chopper791: You can't have a signal that has the same name as a pin. The first thing after the "net" command is always a signal name. You have wither forgotten to make up a signal name for the net, or you are trying to create a signal with the same name as a pin.
[12:21:49] <pcw_home> then you are not adding the right amount
[12:22:17] <generic_nick|2> i tried as little as .005 in ff1, it ferrored pretty quickly
[12:22:38] <generic_nick|2> it has the least amount of ferror with no ff1
[12:23:12] <pcw_home> I doubt if thats possible
[12:23:25] <chopper791> andy- I have thcud.z-pos-out going
[12:23:29] <chopper791> net zpos-cmd thcud.z-pos-in <= axis.2.motor-pos-cmd
[12:23:30] <chopper791> net thcud.z-pos-out => stepgen.2.position-cmd
[12:23:30] <chopper791> #net axis.2.motor-pos-fb <= thcud.z-fb-out
[12:23:48] <generic_nick|2> i checked it on the scope
[12:23:55] <ssi> net zpos-actual thcud.z-pos-out => stepgen.2.position-cmd
[12:23:58] <ssi> give it a signal name
[12:24:09] <generic_nick|2> but regardless, it's probably not perfect but it works good enough lol
[12:24:24] <IchGuckLive> chopper791: what thc is it
[12:24:28] <JT-Shop> somehow the G10 L10 P1 Z0.000 line is changing the G54 offset even though it is not ran
[12:24:30] <generic_nick|2> at least it has holding torque now!
[12:24:40] <pcw_home> if you have too little FF1 you will lag on a long fast move, with too much, you will lead
[12:24:45] <JT-Shop> opps G10 L20 P1 Z0.000
[12:25:19] <JT-Shop> generic_nick|2: now that your in velocity mode have you seen my velocity mode tuning tutorial
[12:25:20] <pcw_home> since a P of.05 works your scaling is such that FF1 may be very small
[12:25:50] <generic_nick|2> JT-Shop: is it on linuxcnc.org?
[12:26:27] <IchGuckLive> chopper791: http://pastebin.com/fp8F6jdA
[12:26:46] <chopper791> ssi- that worked. The net zpos-actual that you suggested was just a random name so allow the creation of the pin that allowed zpos and the thcud.zpos to work?
[12:27:24] <generic_nick|2> pcw_home: ill play with it some more when i have some time. the ferror was fairly even on the scope, no surges or anything. just a little ripple, but parellel to the baseline
[12:28:10] <JT-Shop> generic_nick|2: it's on my web site gnipsel dot com
[12:28:18] <generic_nick|2> im guessing the ripple is from slight uneven belt tension. my pulley has a tad of runout
[12:28:23] <JT-Shop> in the linuxcnc area
[12:28:45] <chopper791> ich- thanks, but the file contains the info for using hostmot2
[12:28:46] <pcw_home> maybe that or torque ripple in the drive
[12:29:06] <generic_nick|2> found it, thanks jt
[12:29:20] <generic_nick|2> could be pcw_home
[12:29:36] <generic_nick|2> i may take the belt off to check and see if it smooths out
[12:29:40] <ssi> chopper791: that's correct
[12:30:45] <pcw_home> you can think of the first name as a wirename
[12:30:46] <pcw_home> net wirename pin_a pin_b pin_c
[12:31:49] <generic_nick|2> i have to wire up the home switch so i can run the machine. other than that i think im ok with running it for now. when i am bored i will work on it some more.
[12:31:50] <andypugh> I like to think as the signal as a wire colour. So it hints at, but does not define, the signal :-)
[12:32:15] <JT-Shop> I should add thcud to LinuxCNC so it gets a man page
[12:32:22] <generic_nick|2> i gotta gfigure out what i want to do with the control pannel. i have a new one i made with a touch screen and mpg built in
[12:32:43] <generic_nick|2> i was thinking of putting it on a pedistal beside the machine
[12:33:12] <generic_nick|2> since it's a knee mill, there are no real good places to put it on the mill without it getting sprayed with coolant
[12:33:25] <generic_nick|2> the touch screen doesnt work well when wet
[12:33:33] <IchGuckLive> chopper791: did you aply a servo tread to the thcud
[12:33:42] <JT-Shop> http://gnipsel.com/shop/bp1/bp1-01.xhtml
[12:33:48] <L84Supper> IchGuckLive: are there any hackerspaces in your area?
[12:34:03] <chopper791> ich-I have not since the thcud file did not show that was needed
[12:34:21] <IchGuckLive> so no interpreter feedback
[12:34:25] <generic_nick|2> JT-Shop: thats where i have mine now and it's covered in coolant and chips
[12:34:35] <IchGuckLive> (loadrt thcud) and add it to a thread (addf thcud servo-thread) chopper791
[12:34:44] <generic_nick|2> i run mass production on it with flood coolant so it makes a bit of a mess
[12:35:08] <JT-Shop> how do you manage to use coolant on a knee mill without making a huge mess in your shop?
[12:35:19] <generic_nick|2> i buit a box around the table
[12:35:27] <JT-Shop> got a photo?
[12:35:39] <generic_nick|2> not on this pc
[12:35:44] <generic_nick|2> ill snap one soon.
[12:35:48] <JT-Shop> ok
[12:36:10] <JT-Shop> you could build a lean to over the MPG
[12:36:12] <chopper791> ich- why is the servo thread needed?
[12:36:16] <L84Supper> http://www.jhchoppers.com/photobucket/cnczone/coolant-0.JPG
[12:36:19] <generic_nick|2> it does make a little bit of a mess in the shop
[12:36:51] <IchGuckLive> chopper791: to get the numbers updated to avoid the feedback fold
[12:36:52] <L84Supper> http://www.jhchoppers.com/photobucket/pm/coolant4.jpg
[12:36:59] <IchGuckLive> the mashine is moving in XY
[12:37:00] <JT-Shop> that's a nice one, everytime I try and build something it gets in the way so I gave up
[12:37:29] <generic_nick|2> JT-Shop: you can kinda see it here. http://farm9.staticflickr.com/8559/8707933356_0416be6dac_b.jpg
[12:37:49] <JT-Shop> L84Supper: did you convert the Anilam to LinuxCNC
[12:38:03] <chopper791> ich-I got ya. Is the pastebin file you sent. Is that your file?
[12:38:08] <JT-Shop> oh yea I remember that photo
[12:38:24] <JT-Shop> you can't work in my shop LOL
[12:38:36] <generic_nick|2> lol
[12:38:37] <L84Supper> JT-Shop: not mine, just posting examples of knee mills with tanks
[12:39:02] <IchGuckLive> chopper791: is your Z axis reversed
[12:39:21] <generic_nick|2> JT-Shop: i use the crap out of that mill. looks like that after about a day
[12:39:41] <IchGuckLive> that may fail in the thcud Comp function
[12:41:27] <generic_nick|2> im going to buy a vmc shortly
[12:41:35] <generic_nick|2> as soon as i move into my new shop
[12:41:42] <ssi> I want to get a vmc soon
[12:41:52] <JT-Shop> nice, the new shop has proper 3-phase power?
[12:41:55] <andypugh> We have some guards on the BP at work that are actually almost OK
[12:42:03] <generic_nick|2> JT-Shop: yes
[12:42:13] <JT-Shop> very nice indeed
[12:42:17] <generic_nick|2> and huge roll up doors
[12:42:36] <chopper791> ich-no it runs in the rpoper directions when I jog
[12:42:47] <chopper791> proper
[12:42:48] <Connor> I wan to enclose my G0704 at some point..
[12:42:52] <ssi> I'm thinking about moving my machines to one of my hangars at the airport
[12:42:52] <generic_nick|2> ok im off to do work.
[12:42:55] <generic_nick|2> adios
[12:42:58] <IchGuckLive> chopper791: there is a breket fail in the mach3 versions that are out did you looked at the comp function to correct this
[12:43:11] <ssi> I have an end hangar that has a half-hangar office attached, and it's the end of the building with the power box
[12:43:15] <ssi> there's 3ph in that box
[12:43:15] <Connor> brb
[12:43:24] <andypugh> The ones I mean are a bit like this: http://www.silvaflame.com/products/miller-adjustable-table-guard/
[12:43:49] <andypugh> Except that they hinge at the corners, pivot at the back, and slide at the front.
[12:43:52] <chopper791> ich-I have looked at anything in Mach3
[12:44:01] <tjb1> postaL:
[12:44:04] <chopper791> *not i mean
[12:44:57] <andypugh> So as you pull the handles apart at the front the front panels slide apart, and swing out. No, There is no way I can describe it in words. And I can't find a picture. I will just give up :-)
[12:45:16] <IchGuckLive> chopper791: did you also set the acelleration setp stepgen.2.maxvel [AXIS_3]MAX_ACCELERATION
[12:45:48] <JT-Shop> you will just have to take a photo tomorrow
[12:45:58] <chopper791> ich-no
[12:46:00] <andypugh> Ah, I think I found it, this one looks very similar: http://www.silvaflame.com/products/fully-enclosed-table-guard/
[12:46:16] <chopper791> Ich-I am not using axis_3 for anything
[12:46:47] <IchGuckLive> AXIS_2 sorry
[12:46:55] <JT-Shop> nice looking setup
[12:47:14] <chopper791> ich-no I have not
[12:47:18] <Connor> running flood with that will still be messy as hell.
[12:47:44] <jdh> I cut some stuff with hand-spray coolant this w/e, it was a mess
[12:48:27] <jdh> big end mills are much messier than small end mills.
[12:49:09] <jdh> so, when the Z plunges for each iteration of this profile, it leaves a scallop in the work. Is this likely due to tool flex or machine rigidity or what?
[13:14:48] <Tom_itx> probably so
[13:14:54] <Tom_itx> or it's not a center cutting endmill
[13:15:40] <IchGuckLive> Loetmichel: http://www.wetter.com/wetter_aktuell/niederschlagsradar/deutschland/
[13:17:56] <Tom_itx> jdh, what's your plunge feed?
[13:18:07] <Tom_itx> it should be at least half your regular feedrate
[13:18:27] <jdh> what's the correlation?
[13:19:47] <Tom_itx> it's just a good idea
[13:20:04] <jdh> but, it's not feeding when it's plunging?
[13:20:22] <jdh> plunge was set to 5ipm, feed to 15ipm, but I was only running about half that.
[13:20:30] <Tom_itx> my cam puts out a separate defineable plunge feed aside from the regular feedrate
[13:20:55] <Tom_itx> tool too long?
[13:21:44] <jdh> normal HSS 1/4" 2-flute end mill
[13:22:20] <Tom_itx> is it dull?
[13:22:51] <jdh> no. I've noticed it with other tooling also, also happens with my router, mainly 1/8" carbide in plastic
[13:22:56] <Tom_itx> i switched to carbide on my sherline since i don't have flood coolant on it
[13:23:02] <Tom_itx> seems to do better for me
[13:23:17] <Tom_itx> your z may have flex in it then
[13:23:24] <jdh> my spindle maxes at 3000rpm, carbide doesn't seem beneficial there for 6061
[13:23:29] <Tom_itx> or x and y aren't rigid enough
[13:24:02] <Tom_itx> you can always drill a pilot hole to plunge in
[13:27:29] <jdh> I assume if I put the pilot hole in the same place, it would have the same issue
[13:33:36] <Tom_itx> why?
[13:33:46] <Tom_itx> the endmill isn't really made for drilling
[13:34:15] <Tom_itx> if you drill a pilot hole and put the endmill down in it then move in x or y it shouldn't walk on you
[13:38:49] <jdh> just curious. I sometimes make the plunge position someplace that won't matter, but don't usually bother.
[13:44:51] <archivist> slot drill is the better form for plunge
[13:52:33] <Loetmichel> or move into the workpiece in a helix. works also
[13:55:15] <Loetmichel> like this: http://www.youtube.com/watch?v=UEd4LCJ3uWk (caution. LOUD)
[13:55:16] <Tecan> (UEd4LCJ3uWk) "milldrillcycle15.avi" by "Loetmichel" is "People" - Length: 0:01:13
[13:55:50] <Loetmichel> http://www.youtube.com/watch?v=2wS_URQsyTU <- same in small
[13:55:52] <Tecan> (2wS_URQsyTU) "milldrillcycle.avi" by "Loetmichel" is "People" - Length: 0:00:46
[14:02:30] <ssi> hm... I want to hook a siggen's square output up to a bit signal
[14:02:37] <ssi> but siggen.x.square is a float
[14:02:40] <ssi> how can I "cast" in hal?
[14:04:47] <jdh> (boolean) siggen.x.squre
[14:04:54] <ssi> is it that easy?
[14:05:03] <jdh> doubt it, sounds good though doesn't it.
[14:05:06] <ssi> haha yeah
[14:06:37] <ssi> linuxcnc.org needs better hosting :/
[14:07:25] <cradek> ssi: use the not or chargepump components to make a toggling bit
[14:07:59] <mhaberler> ssi is seriously into the blinkenleds business
[14:08:25] <mhaberler> what's next, linuxcnc-driven throwies ;-?
[14:09:13] <ssi> hahah
[14:10:15] <ssi> I Think all cnc machines should be driven by blinkenleds that drive optos
[14:10:16] <ssi> :D
[14:14:25] <frallzor> http://imageshack.us/a/img221/8673/img4920at.jpg http://imageshack.us/a/img191/1433/img4923jj.jpg these will create some sweet sweet loving!
[14:16:09] <PetefromTn> afternoon fellas..
[14:19:18] <ssi> cradek: I don't see how not helps me, and I can't find a man page for the chargepump component
[14:19:32] <ssi> and everything on linuxcnc.org is taking tens of minutes to load :(
[14:19:57] <JT-Shop> http://www.linuxcnc.org/docs/html/man/man9/charge_pump.9.html
[14:20:02] <andypugh> ssi: conv_float_bit would be what you needed, but doesn't seem to exist. comp will threshold a float to a bit, or you might find that the "clock" bit output from siggen is what you want
[14:20:21] <ssi> ooh clock will fix me
[14:20:23] <ssi> thanks andy
[14:22:24] <r00t4rd3d> frallzor, how did you finish that sign?
[14:23:48] <frallzor> uhm
[14:23:58] <frallzor> dont know the english term =P
[14:24:13] <PetefromTn> how about fanfreakintastic?
[14:24:19] <r00t4rd3d> did you paint it all then sand it down/
[14:24:21] <ssi> except clock seems to be always false
[14:24:26] <frallzor> some kind of paint/varnish that makes it hate water =P
[14:24:35] <frallzor> sanded first
[14:24:40] <r00t4rd3d> no i meant how did you color your letters
[14:24:43] <frallzor> ah
[14:24:52] <frallzor> paint kind of sloppy
[14:24:57] <frallzor> then just sand
[14:25:03] <r00t4rd3d> ok
[14:25:04] <ssi> and charge_pump is the component, but it's way too fast
[14:25:13] <frallzor> and then oil with chinese oil
[14:25:23] <r00t4rd3d> chinese oil?
[14:25:35] <frallzor> combination of special herbs, oil from the tung tree and some cat
[14:25:41] <r00t4rd3d> wtf
[14:25:48] <frallzor> Im just guessing now
[14:25:55] <frallzor> the chinese put cat in everything
[14:26:03] <jdh> moo shu cat
[14:26:07] <PetefromTn> LOL cat.... Tung oil is really nice.
[14:26:22] <jdh> you squeeze cat tungs for oil?
[14:26:24] <r00t4rd3d> i mix tung and boiled linseed
[14:26:25] <jdh> how many does it take?
[14:26:36] <r00t4rd3d> splash of poly
[14:26:38] <frallzor> about a dozen
[14:27:01] <frallzor> the left overs make a nice stew though
[14:27:09] <frallzor> so nothing goes to waste
[14:27:11] <r00t4rd3d> is that for a business or your house?
[14:27:24] <frallzor> for a cousins "mansion" =P
[14:27:55] <frallzor> or it´s more like "manor"
[14:28:19] <frallzor> old farm so he calls its "the name of the place" manor =)
[14:28:35] <r00t4rd3d> can i have the dxf of the plate?
[14:28:56] <PetefromTn> Just updated my Photobucket account on my Cincinatti Arrow 500 retrofit. FLOOD COOLANT IS WORKING!! LIKE A BOSS!! http://s150.photobucket.com/user/matospeter/media/Cincinatti%20Arrow%20500%20Retrofit/IMG_20130506_144359_zps21d1ea10.jpg.html?sort=3&o=0
[14:29:10] <frallzor> I rarely share things I one-off =)
[14:29:20] <frallzor> but its simple to make a copy
[14:30:20] <ssi> PetefromTn: beautiful :D
[14:30:30] <frallzor> rectangle with internal radiuses, and 2 ellipses
[14:30:35] <frallzor> 1 top, 1 bottom
[14:30:37] <frallzor> then just trim
[14:31:12] <ssi> god I didn't expect hooking a siggen to an LED to be this damn hard
[14:31:43] <cradek> waah
[14:31:46] <PetefromTn> ssi: Thanks man. Honestly that picture really does it no justice. That is some serious flow coming out of there I was kinda worried I would not have adequate flow for chip evac with the factory pump but it surprised me. I may make a nice manifold to add some nozzles to bump up the pattern later but for now this is real nice.
[14:32:41] <cpresser_> a usb2dmx interface which uses the same protocol is also planned. i got the prototypes up an working.
[14:32:44] <cpresser_> 21:15 < cpresser_> the schematics for the interface will be free. its a low-cost interface (about 10$ for DIY-builders)
[14:32:48] <cpresser_> 21:15 < cpresser_> any chance of getting the driver-plugin into upstream?
[14:32:51] <cpresser_> sorry, copy-paste error :/
[14:33:02] <ssi> PetefromTn: excellent
[14:33:32] <r00t4rd3d> frallzor, http://i.imgur.com/Q5YLoit.jpg
[14:34:00] <ssi> doh I know what I'm doing wrong
[14:34:47] <frallzor> r00t4rd3d yup
[14:35:07] <frallzor> if you want it more like mine make the ellipse larger and change shape =)
[14:35:32] <r00t4rd3d> was a 2 minutes copy :)
[14:35:45] <PetefromTn> Glad I had the forethought to put a bucket under the coolant nozzled when I first turned on the pump because some NASTY water came out first then it cleaned right up LOL
[14:36:25] <ssi> what are you running for coolant?
[14:36:27] <JT-Shop> pretty blue...I hate cleaning out coolant tanks :(
[14:36:35] <andypugh> I do wish PyCAM had more rapidity.
[14:36:51] <JT-Shop> does it work?
[14:37:04] <andypugh> PyCAM? Yes, the results are fine.
[14:37:09] <PetefromTn> I am using TRim Sol E206 I bought from Enco and it is pretty nice stuff so far.
[14:37:17] <JT-Shop> for 3-d?
[14:37:31] <ssi> gotcha
[14:37:40] <ssi> I ended up running straight oil in my hnc
[14:37:45] <ssi> it was expensive, but it doesn't grow anything :)
[14:37:51] <ssi> and no chance of corrosion
[14:38:05] <andypugh> JT-Shop: Yes, it did this one: http://www.youtube.com/watch?v=ZSLB-ue4CTI
[14:38:06] <Tecan> (ZSLB-ue4CTI) "Foundry Pattern Machining on my Mini-Mill" by "andy pugh" is "Tech" - Length: 0:03:34
[14:38:18] <PetefromTn> JT-Shop: yeah it is pretty blue and the coolant troughs are basically brand new now so as long as I keep it clean it should be all good LOL
[14:39:07] <PetefromTn> I considered straight oil but my pal Joe has the same machine and uses it in a hobby environment mostly and has had excellent results and no funky smells for a LONG TIME on the same batch.
[14:39:18] <andypugh> I found the place to set G64 now too, so it should be smoother on this job.
[14:39:23] <ssi> PetefromTn: I put ten gallons of mobilmet in it
[14:39:23] <PetefromTn> This stuff is pretty damn thick LOL
[14:39:25] <ssi> was about $150
[14:39:40] <JT-Shop> andypugh: pretty cool
[14:39:45] <PetefromTn> This was only about $75.00 or so with free shipping coupon
[14:40:08] <andypugh> I wanted to use Castrol Clearedge, but couldn't find it in less that 20l quantities.
[14:40:19] <frallzor> andypugh foamed urethane?
[14:40:26] <andypugh> Yes.
[14:40:35] <frallzor> I love it =)
[14:40:41] <frallzor> have few blocks of it =P
[14:40:50] <andypugh> I found this in the skip at work :-)
[14:41:07] <r00t4rd3d> http://i.imgur.com/DahzflM.jpg
[14:41:08] <r00t4rd3d> there damit
[14:41:10] <frallzor> tool grade and that apricot light stuff
[14:41:35] <PetefromTn> Now I can hopefully start making some test chips and simple projects to get a feel of the control and test my Sheetcam post processor output.
[14:42:50] <PetefromTn> One of the first actual parts I need to make will be a mount for the spindle motor encoder. Should be fun.
[14:42:54] <andypugh> PetefromTn: Consider mounting a sacrificial sheet of Alu on top of the bed until you are sure :-)
[14:43:39] <PetefromTn> Yeah man actually I wil be using a piece of 3/4 aluminum plate set on top of 123's for most stuff. This is not my first CNC but I get your point for sure...
[14:44:08] <JT-Shop> your using sheetcam for your mill?
[14:44:12] <Tom_itx> andypugh you should add a cover over your screw
[14:44:42] <andypugh> I have never managed to think of a way to cover it
[14:44:50] <PetefromTn> JT-Shop: Yup have for years now. Works great... Not 3d but I have made all sorts of stuff with Sheetcam.
[14:45:08] <JT-Shop> I use it for plasma but have never used it for a mill
[14:45:25] <Tom_itx> wonder about some tiny acordion air duct material
[14:45:30] <PetefromTn> It even has a new V carve router setup
[14:45:35] <Tom_itx> think under dash of car
[14:45:47] <JT-Shop> fill with chips quick
[14:46:00] <JT-Shop> flat spring coil covers
[14:46:17] <Tom_itx> yeah if you can find em small enough
[14:46:19] <andypugh> If I was doing the conversion again, or was still interested in that machine, then I would shift the motor further in the headstock direction to make room for a spiral spring cover. But at the moment if I fitted one of those I wouldn't be able to machine up to the chuck in lathe mode.
[14:46:38] <JT-Shop> that sucks
[14:47:01] <Tom_itx> got my mill holders today
[14:47:08] <andypugh> The wipers on the screw seem to have done a good job so far.
[14:48:39] <Tom_itx> ahh that's the same machine you turned those bowls on
[14:52:54] <PetefromTn> How many of you fellas are using pycam seriously? Is it generating good toolpaths for you?
[14:53:24] <andypugh> To be honest, it is far too slow on the useful contouring processes to be any real use.
[14:53:50] <PetefromTn> how do you mean slow?
[14:54:03] <ssi> I might try it for plasma cutting when I get the machine up
[14:54:09] <andypugh> Like, overnight to create the toolpath.
[14:54:18] <PetefromTn> a 3d program for plasma?
[14:54:18] <ssi> http://www.youtube.com/watch?v=Jfi0b5cUNVw
[14:54:19] <Tecan> (Jfi0b5cUNVw) "Linuxcnc Hal driver for beaglebone black" by "imcmahon" is "Tech" - Length: 0:00:08
[14:54:23] <ssi> blinkenleds :D
[14:54:50] <andypugh> Drop-cutter and push-cutter are fine, but the contouring algorithms are impractical.
[14:55:38] <PetefromTn> honestly I don't have much 3d Experience at all what is drop cutter and push cutter?
[14:56:08] <PetefromTn> Almost everything I have made has been 2.5D
[14:56:29] <PetefromTn> ssi: Nice man...
[14:59:17] <generic_nick|2> sweet, homing and whatnot is working now on my 4th axis.
[14:59:29] <generic_nick|2> it's good to go, other than putting it on the mill
[14:59:43] <PetefromTn> generic_nick|2: nice man....PICS!!
[15:00:05] <generic_nick|2> ill try and get some once i clean up my disaster zone lol
[15:00:57] <PetefromTn> understand that...
[15:01:17] <Tom_itx> andypugh what file formats does it accept?
[15:02:16] <Tom_itx> overnight to generate or run the toolpath?
[15:06:11] <PetefromTn> Just Downloaded pycam....gonna play a bit.
[15:07:38] <GammaX-Shop> I have the proper accell and vel but for mach... in mms lol
[15:08:02] <GammaX-Shop> 2000mm per minute
[15:08:20] <GammaX-Shop> accel is 344 mm's/sec/sec ???
[15:10:34] <ReadError> you just said a bad word
[15:10:40] <GammaX-Shop> I know I know...
[15:11:16] <GammaX-Shop> m not using it and actually getting Precision matthews to convert there cnc systems to linuxcnc... just gotta convert there values over and test...
[15:12:55] <ReadError> accel is always unit/s^s
[15:14:04] <GammaX-Shop> ReadError: well what value would I put in lol I did 30 and it seems to work good.
[15:14:16] <FinboySlick> ReadError: Just for the sake of accuracy: unit/s^2 No?
[15:15:12] <ReadError> huh?
[15:15:17] <ReadError> ah yea
[15:15:25] <FinboySlick> Unless you meant *
[15:16:20] <ReadError> yea typo'd
[15:16:24] <ReadError> seconds squared
[15:16:45] <r00t4rd3d> frallzor, do you know what painters tape is?
[15:16:49] <frallzor> yup
[15:16:52] <r00t4rd3d> that blue stuff to mask with
[15:17:04] <frallzor> blue, white etc, comes in many shapes =)
[15:17:12] <r00t4rd3d> if you cover your board with that before you engrave your letters makes coloring them much easier
[15:17:17] <andypugh> Tom_itx: STL for 3D, DXF and SVG for contours.
[15:17:26] <frallzor> I´d say... no
[15:17:35] <r00t4rd3d> its how i do it
[15:18:00] <frallzor> cant see how it will stay when machining through it
[15:18:09] <frallzor> not really strong stuff
[15:18:11] <r00t4rd3d> it sticks
[15:18:26] <r00t4rd3d> i wrap it around the edges
[15:19:24] <r00t4rd3d> then all you have to do is a light sanding to clean up the edges
[15:19:58] <frallzor> I dont mind, allways sand anyway for a nice finnish =)
[15:20:25] <r00t4rd3d> sanding for finish and sanding to remove excess paint is two different things...
[15:21:09] <frallzor> same work and nice finnish and paint removed
[15:21:21] <r00t4rd3d> sucks when you sand out the detail of things trying to remove paint/stain
[15:21:51] <frallzor> cant see how one can remove details in a v-carve unless its extremely shallow
[15:23:23] <GammaX-Shop> im losing steps on my 7 and I think its cause of this stupid delron coupler.
[15:23:30] <GammaX-Shop> No one should EVER used these stupid things
[15:24:26] <r00t4rd3d> dumpstercnc
[15:24:42] <r00t4rd3d> nice couplers
[15:25:26] <r00t4rd3d> http://dumpstercnc.com/
[15:27:09] <r00t4rd3d> the threaded couplers are sweet if you dont have machined ends
[15:27:11] <GammaX-Shop> r00t4rd3d: yeah those!
[15:27:45] <GammaX-Shop> i wouldnt buy them unless you have a saig
[15:28:08] <r00t4rd3d> saig?
[15:28:17] <GammaX-Shop> i meant the x2 type mils
[15:28:41] <r00t4rd3d> i have them on all my axis
[15:29:00] <GammaX-Shop> I had one break ok me... the others just dont hold
[15:29:37] <r00t4rd3d> http://www.cncrouterparts.com/dumpstercnc-acme-assembly-p-36.html
[15:29:48] <r00t4rd3d> i got one of those coming in the mail
[15:29:53] <r00t4rd3d> gonna check that out
[15:30:02] <andypugh> JHybrid of a Sieg and a Taig?
[15:31:44] <ReadError> r00t4rd3d, remember my router motor problems?
[15:31:48] <ReadError> i took the brushes out
[15:31:54] <ReadError> and blew a bunch of air through it
[15:31:56] <ReadError> good as new
[15:32:00] <r00t4rd3d> lol
[15:32:14] <ReadError> i think it just got a bunch of dust and crap inside it
[15:32:18] <ReadError> brushes where fine though
[15:32:56] <Tom_itx> the dust probably bound the brushes
[15:33:09] <r00t4rd3d> i vacuum my router out after using it
[15:33:57] <r00t4rd3d> put the hose on the air intake and cup your hands around it
[15:34:07] <r00t4rd3d> like a good girl
[15:36:58] <r00t4rd3d> i got another box of cncrp parts coming this week
[15:38:40] <r00t4rd3d> everyone of my axis is getting a bearing block on each end, shaft adapters, thrust bearings on each side of the blocks
[15:40:13] <GammaX-Shop> my stepper drivers lowest setting for micro stepping is 2.... why do I keep the stepconfig at .5?
[15:41:07] <r00t4rd3d> cause thats what it equals too I think
[15:41:18] <GammaX-Shop> r00t4rd3d: i bought a pneumatic 5 c closer! Woooo 190 bucks shipped. and I traded a chinese 3 phase motor for a 32 pc 5c collet set and a 50pc hold down set!
[15:41:31] <GammaX-Shop> and the guy is paying shipping both ways!
[15:41:53] <r00t4rd3d> i dont know what any of that stuff is for
[15:42:29] <GammaX-Shop> lathe
[15:42:36] <r00t4rd3d> oh
[15:43:59] <GammaX-Shop> anyone use modbus on vfd currently?
[15:48:36] <frallzor> tried, never got the hang of it =)
[15:54:03] <JT-Shop> yep
[15:57:59] <ssi> sweet, got my second beaglebone in
[16:05:55] <ssi> this is a cute little setup
[16:05:56] <ssi> https://pbs.twimg.com/media/BJnEBPECEAASqq4.jpg:large
[16:09:27] <ReadError> https://dl.dropboxusercontent.com/u/25091878/Photo%20May%2006%2C%2012%2014%2024%20PM.jpg
[16:09:30] <ReadError> i finishes that today
[16:10:12] <ssi> what is it, quadrotor driver?
[16:10:57] <ssi> http://store.openpilot.org/openpilot-hardware/1-coptercontrol.html
[16:11:00] <ssi> that guy I guess
[16:11:02] <GammaX-Shop> jthornton: you using a rs485 adapter on serial port?
[16:11:07] <generic_nick|2> http://wiki.linuxcnc.org/cgi-bin/wiki.pl?WrappedRotaryAxes i put the [AXIS_3] WRAPPED = 1 in my ini, but then i cant use the calibration tool in axis
[16:11:19] <GammaX-Shop> JT-Shop: read above porfavor ^^^ :D
[16:11:55] <generic_nick|2> the linuxcnc page doesnt say where it should go in the ini file. should i put it down where the parameters for that axis is, or up in the traj section, or what?
[16:12:24] <ReadError> yea
[16:12:26] <ReadError> clone...
[16:12:46] <cradek> generic_nick|2: that wiki page says Spec (draft). you should look in the real docs instead.
[16:13:21] <frallzor> http://imageshack.us/a/img801/4599/img46812.jpg mah old baby =)
[16:13:29] <andypugh> generic_nick|2: Try this page: http://www.linuxcnc.org/docs/html/config/ini_config.html
[16:13:35] <cradek> http://www.linuxcnc.org/docs/html/config/ini_config.html#_axis_lt_num_gt_section_a_id_sub_axis_section_a
[16:13:45] <cradek> yep I was just finding it too, thanks andypugh
[16:13:59] <generic_nick|2> thanks cradek
[16:14:01] <andypugh> You had a more exact link.
[16:14:07] <ReadError> frallzor, fail fail
[16:14:10] <ReadError> is that a naza?
[16:14:15] <frallzor> yup
[16:14:26] <ReadError> baby jesus just cried a little
[16:14:44] <frallzor> naza rocks
[16:14:58] <GammaX-Shop> 6 pound 8 ounche tiny newborn, baby infant jesus?
[16:15:16] <ReadError> frallzor, yea if you want to hover in 1 spot maybe
[16:16:12] <frallzor> I assume you never tried one with gps =)
[16:16:27] <ReadError> frallzor, I do FPV and acro
[16:16:41] <ReadError> a naza+gps is useless
[16:16:51] <ssi> JT-Shop: mach3 file emailed to you
[16:16:58] <ssi> they're nasty inside
[16:17:05] <frallzor> ReadError not if you use it for video like I did =)
[16:17:06] <andypugh> This may be a page that hasn't been updated for a while, it still references EMC2, and Axis is listed as a separate package: http://www.sm-ee.co.uk/resources/cadcam.htm
[16:17:11] <GammaX-Shop> ssi: you use mach 3?
[16:17:20] <ssi> I did once upon a time
[16:17:21] <ReadError> frallzor, im sure i can do better with a 20$ open source controller
[16:17:50] <GammaX-Shop> can you make a config file for it?
[16:18:04] <ssi> I don't have windows running anywhere in my house now, thankfully
[16:18:05] <GammaX-Shop> so you can send it to other people and them have the same settings?
[16:18:26] <frallzor> ReadError prove it then =)
[16:18:36] <ssi> ReadError: have any experience with controlling CP helis with autopilot hardware?
[16:18:45] <frallzor> words are useless unless proven
[16:18:54] <r00t4rd3d> still wet sign i made today http://i.imgur.com/CgkSy7H.jpg
[16:19:06] <generic_nick|2> yay, that worked great. now it wraps around from 360 to zero
[16:19:09] <r00t4rd3d> easy sell now that every douche bag in america has a harley
[16:19:39] <ssi> r00t4rd3d: haha definitely
[16:20:05] <ReadError> ssi, yea theres a few that will do flybarless helis
[16:20:06] <skunkworks> r00t4rd3d: be careful about copy write...
[16:20:14] <r00t4rd3d> fuck the police
[16:20:18] <ssi> lol
[16:20:34] <ssi> "fuck the police... I prefer Sting's solo work"
[16:20:44] <frallzor> r00t4rd3d dont worry about the police
[16:20:57] <frallzor> lawyers are way worse =)
[16:21:13] <ReadError> r00t4rd3d, you seen the southpark ep about choppers?
[16:21:21] <r00t4rd3d> no fag, why?
[16:21:25] <ReadError> lol
[16:23:26] <r00t4rd3d> ill put that sign on craigslist for for $35 bucks when it dries and I bet some asshat in a helmet will be in my driveway picking it up within 24 hours
[16:24:39] <generic_nick|2> hmmm i wonder if i can fit enough stations on my rotary table to make 3 complete parts in one run.
[16:24:40] <ssi> seems like it'd cost more than $35 to make if you value your time :P
[16:24:58] <generic_nick|2> that would be nice, id only have to load it every 30-45 minutes or so
[16:25:05] <ReadError> r00t4rd3d breaks equipment so he gets paid to sit at home
[16:25:07] <frallzor> r00t4rd3d which version of aspire did you purchase?
[16:25:12] <ReadError> like a real pro
[16:25:40] <r00t4rd3d> the wood was free, had the stain and poly and the design was effortless, cut time was 40 minutes
[16:25:55] <generic_nick|2> 40 minutes??
[16:25:59] <generic_nick|2> desktop mill?
[16:26:10] <r00t4rd3d> 17x25 cut area
[16:26:27] <r00t4rd3d> more then desktop i guess
[16:26:47] <r00t4rd3d> my machine is slow
[16:27:03] <generic_nick|2> took me less than 15 to make this, and i have to trim a lot of fat off the program still. http://farm9.staticflickr.com/8395/8704736685_63028c81dd_c.jpg
[16:27:29] <DJ9DJ> gn8
[16:27:56] <ReadError> in aluminum generic_nick|2 ?
[16:28:01] <generic_nick|2> yes
[16:28:08] <r00t4rd3d> i played video games while i monitored the machine too so :)
[16:28:29] <Gigs-> I'm looking at woodruff cutters on MSC... how do you tell the maximum depth of cut?
[16:28:42] <generic_nick|2> lol, well i drank beer so i guess we're even
[16:29:10] <r00t4rd3d> im rebuilding my machine and my goal is 600ipm
[16:29:16] <r00t4rd3d> instead of 18 i get now
[16:29:21] <generic_nick|2> thats quick
[16:29:31] <generic_nick|2> im at about 300ipm i think
[16:29:45] <generic_nick|2> which is fairly decent for a big knee mill
[16:30:05] <ssi> yeah that's pretty quick
[16:30:11] <ssi> i was running that atrump machine at 120ipm
[16:30:19] <ssi> and the X axis shakes the machine at that speed
[16:30:23] <ssi> I think it's not leveled quite right
[16:30:49] <generic_nick|2> my mill is about 7000lbs so not much shakes it lol
[16:31:09] <r00t4rd3d> ReadErrors gf could
[16:31:14] <generic_nick|2> ha
[16:31:18] <r00t4rd3d> :D
[16:31:29] <generic_nick|2> r00t4rd3d: i didnt know your sister weighed that much
[16:31:38] <r00t4rd3d> yeah shes fat
[16:32:28] <r00t4rd3d> she is super pissed they took the bones out of kfc too
[16:32:38] <JT-Shop> ssi ok
[16:32:54] <ssi> JT-Shop: ok!
[16:33:47] <generic_nick|2> im waiting for some sealant to dry on my coolant enclosure on the mill. my buddy gave me some toyota stuff to try, said it's the cat's meow. we'll see if it can handle water soluable coolant.
[16:34:08] <generic_nick|2> that stuff eats through regular rtv sealant
[16:37:13] <r00t4rd3d> high temp silicone
[16:38:47] <r00t4rd3d> I use permatex on lots of gas applications
[16:39:44] <generic_nick|2> tried it, doesnt work
[16:40:03] <generic_nick|2> coolant is the most destructive crap there is for rubber
[16:40:20] <r00t4rd3d> that and your sisters over bite
[16:40:33] <generic_nick|2> har
[16:40:59] <r00t4rd3d> :F
[16:54:13] <ssi> I hate keyboards with keys in weird places
[16:55:27] <ssi> https://pbs.twimg.com/media/BJnPcUuCYAI76eN.jpg:large
[16:55:31] <ssi> they call that a "compact" keyboard
[16:55:33] <Tom_itx> i guess this will have to be a 'don't do that' button cause i can't seem to figure out the logic to get it the way i want it
[16:55:55] <ssi> it's MAYBE an inch narrower than my "full size" keyboard, but it's completely unusable because all the keys are crammed together and nothing is where my fingers expect it to be
[17:03:26] <r00t4rd3d> rdy for craigslist :D http://i.imgur.com/VJhs464.jpg
[17:03:51] <ssi> nice
[17:04:43] <GammaX-Shop> how do you get into the hal command line area
[17:04:54] <ssi> halrun
[17:04:54] <GammaX-Shop> thanks
[17:06:23] <GammaX-Shop> gs2 aint working. classicladder here I come!!!
[17:13:03] <asdfasd> ssi: I have smilar but logitech, and I love it, now Im typing without looking at the keyboard :)
[17:13:53] <asdfasd> it is good for typing but on the machine I use standard, easy to find the arrouws
[17:16:22] <PetefromTn> whats GS2
[17:16:29] <GammaX-Shop> vfd
[17:17:08] <GammaX-Shop> http://wiki.linuxcnc.org/cgi-bin/wiki.pl?VFD_Modbus under sj200 is that a driver for vfd?
[17:53:39] <PetefromTn> These babies look REALLY sweet...I am sure they are WAY outta my price range tho...SOB...http://www.youtube.com/watch?feature=endscreen&v=OpaNy0xM7mo&NR=1
[17:53:42] <Tecan> (OpaNy0xM7mo) "Jergens-Bock Brand Machine Vises" by "Jergensinc" is "Tech" - Length: 0:03:23
[17:55:46] <Tom_itx> kurt are the shizzle
[17:55:47] <Tom_itx> that looks like a kurt
[17:56:10] <Tom_itx> with softjaws
[17:56:16] <ssi> I have a kurt double vise on my big manual mill
[17:56:26] <ssi> I intend to try a glacern on my g0704
[17:56:40] <ssi> I hope the kurt hasn't spoiled me :)
[17:59:46] <Tom_itx> the tooling plate should have a full grid of tapped holes though
[17:59:56] <Tom_itx> and an alternating grid of reamed holes
[18:01:30] <PetefromTn> it is not a tooling plate really more a fixture locator.... YOu mount your tooling plate to that plate apparenlty.
[18:01:43] <PetefromTn> http://www.youtube.com/watch?v=ZZifefyTgRM
[18:01:47] <Tecan> (ZZifefyTgRM) "Jergens Ball Lock Quick Change Fixturing System vs. Traditional Vise" by "Jergensinc" is "People" - Length: 0:03:43
[18:01:57] <Tom_itx> well call it what you want
[18:02:20] <Tom_itx> they make 'plugs' you can put in those tapped holes for all sorts of holddowns
[18:02:45] <PetefromTn> basically yeah...
[18:03:02] <PetefromTn> Nice looking stuff. I am sure it is equally nicely priced.
[18:03:07] <Tom_itx> i forget the brand
[18:03:07] <Tom_itx> yup
[18:03:31] <Tom_itx> my friend has those at his shop and i must say it makes setups alot quicker
[18:04:36] <PetefromTn> apparently the six inch vise alone is $1500.00 +
[18:04:44] <Tom_itx> could be
[18:04:55] <Tom_itx> the average 6" single is around $6-800
[18:05:02] <Tom_itx> you can get them on sale usually
[18:05:17] <Tom_itx> since they're so common
[18:05:49] <Tom_itx> he had a couple dual but mostly 6 & 8" single kurt
[18:06:05] <PetefromTn> average where? I need to pickup a vise for the VMC now that it is running but I am a poor bastard now LOL
[18:06:19] <Tom_itx> see what travers has
[18:06:31] <PetefromTn> never bought from travers tool
[18:06:50] <Tom_itx> or mts
[18:07:33] <Tom_itx> http://www.travers.com/311677-61-421-172?Category=UserSearch=kurt%20vise
[18:08:07] <andypugh> I have a spare 8" Kurt double. When I say "spare" I mean "going rusty on the floor as I will never use it"
[18:08:21] * Tom_itx gasps
[18:08:33] <andypugh> Partly because it has no jaws. (Which is quite a drawback)
[18:08:38] <Tom_itx> andypugh, pack it in your carry on :)
[18:08:42] <PetefromTn> yeah but you are in jolly old england and it would cost more to ship probably than it costs to buy here LOL
[18:08:53] <ssi> jaws are available
[18:08:56] <andypugh> I can barely lift it.
[18:08:58] <ssi> tehy're not terribly cheap, but they're out there
[18:09:00] <Tom_itx> i bet
[18:09:02] <ssi> but yea a double 8 is a BIG vise
[18:09:09] <Tom_itx> the 8" single are heavy enough
[18:09:10] <ssi> mine's a double 4, and it's a good size for my 9x42
[18:09:12] <PetefromTn> That would be a sweet vise tho...
[18:09:28] <PetefromTn> I bet it weighs 200 lbs..
[18:09:40] <andypugh> Yes, quite. I can buy a much more suitable vice for my purposes for a lot less than a set of jaws for the Kurt.
[18:10:07] <PetefromTn> Hell I usually make my own sacrificial jaws for most everything anyways.
[18:10:57] <PetefromTn> never even seen an 8" double vise, how long is it?
[18:11:04] <Tom_itx> yup
[18:11:29] <andypugh> About 2'6"?
[18:11:35] <ssi> holy cow
[18:11:35] <ssi> hahah
[18:11:49] <andypugh> It looked smaller on the eBay advert.
[18:11:54] <ssi> hahaha
[18:12:07] <Tom_itx> http://www.americastoolcrib.com/cgi/CGP2SRIM?PMITEM=HT80D688P&PARTPG=CGP2LMXE&PAMENU=&PAHDID=000000040278704&PARDID=559178412220989
[18:12:09] <PetefromTn> HOLY SMOKES> 30" is crazy...
[18:12:09] <Tom_itx> a bit cheaper
[18:13:10] <PetefromTn> Probably gonna hafta buy some cheapo vise just to get started and then once I get some paying jobs in here upgrade....unfortunately.
[18:13:44] <ssi> PetefromTn: this is what I'd get
[18:13:44] <ssi> http://www.glacern.com/gpv_615
[18:13:54] <ssi> whoa they went up
[18:13:59] <ssi> I'd buy the kurt for that money nm
[18:14:00] <ssi> ;p
[18:14:19] <ssi> I wish they made a 4" with the flat sides like that
[18:16:04] <andypugh> Looks like Abwood are no longer in the vice business.
[18:17:06] <generic_nick|2> dang it! i lost all my work offsets AGAIN!
[18:17:10] <PetefromTn> actually that glacern is not terrible and reasonable for a better quality vise...
[18:17:13] <Tom_itx> http://www.americastoolcrib.com/cgi/CGP2SRIM?PMITEM=HT80D40P&PARTPG=CGP2LMXE&PAMENU=&PAHDID=000000040278704&PARDID=559178412220989
[18:17:17] <Tom_itx> 4" width
[18:17:41] <generic_nick|2> anyone know why my work offsets dissapear if my machine shuts off due to a power loss?
[18:17:42] <PetefromTn> I am really gonna have to go cheaper than that to right now. My damn AC unit at home here is smoked so gotta ante up for that baby first....
[18:17:59] <generic_nick|2> this has happened at least 4 times so far.
[18:18:25] <Tom_itx> http://www.americastoolcrib.com/cgi/CGP2SRIM?PMITEM=HT71ATW40S&PARTPG=CGP2LMXE&PAMENU=&PAHDID=000000040278704&PARDID=559178412220989
[18:18:30] <ssi> PetefromTn: gmt 4" is $300
[18:18:34] <Tom_itx> smaller 4"
[18:18:54] <Tom_itx> import
[18:19:56] <generic_nick|2> anyone?
[18:20:19] <generic_nick|2> really sucks having to touch off all my fixtures everytime this happens
[18:21:29] <andypugh> generic_nick|2: The offsets are saved at shutdown. If you don't get a proper shutdown, then they don't get saved.
[18:22:02] <Tom_itx> no way to back up the file?
[18:22:11] <andypugh> Can I say that the problem is your power supply, not LinuxCNC?
[18:22:43] <generic_nick|2> andypugh: not the power supply when the whole house looses power
[18:22:45] <andypugh> But, if you do all the touch-off, then quit LinuxCNC, then re-open, that shpiuld save the offsets
[18:23:15] <andypugh> generic_nick|2: I think you might want to consider a small UPS, just for the PC.
[18:23:17] <generic_nick|2> andypugh: my offsets havent changed in years, why doesnt it just not delete the last ones it saved?
[18:23:29] <generic_nick|2> it changes it all to zero
[18:23:43] <andypugh> Ah, OK, if that is what is happening, I am at a loss to explain it.
[18:24:55] <generic_nick|2> yea, they all go to home position
[18:24:59] <andypugh> A PC-sized UPS is pretty cheap.. http://www.amazon.com/CyberPower-CP350SLG-Standby-350VA-Compact/dp/B004OR0V2C/ref=sr_1_7?ie=UTF8&qid=1367881826&sr=8-7&keywords=UPS
[18:26:10] <PetefromTn> yeah import...
[18:26:18] <generic_nick|2> the work offsets are saved to the var file, right?
[18:26:25] <andypugh> I don't know how long that would give you to exit and shutdown though.
[18:26:26] <PetefromTn> a 4" vise is not gonna cut it for me...
[18:27:05] <generic_nick|2> vices are cheap if you buy them used, i got a few kurt vices for $50-100
[18:27:26] <andypugh> I lack headroom in my machine, so I am looking at this sort of thing: http://www.ebay.co.uk/itm/2-pc-LIMITLESS-MILLING-VICE-WITH-6-JAWS-/370658211083?pt=LH_DefaultDomain_3&hash=item564cf3750b
[18:28:20] <generic_nick|2> gotta imagine a fair amount of deflection with that
[18:29:53] <andypugh> And many opportunities to gouge the table.
[18:33:22] <andypugh> This is what that pattern came out like (the one that PyCAM did. This is after about 10 minutes with a file wrapped in sandpaper, but then that would be expected with any software. https://plus.google.com/photos/108164504656404380542/albums/5747722155741347649/5875009568208252162?banner=pwa
[18:40:05] <PetefromTn> Andy...I thought you said Pycam was slow LOL
[18:41:19] <PetefromTn> andypugh: Oh yeah and whatsabout that probe in your pics?
[18:43:09] <andypugh> PetefromTn: The toolpath generation is really slow, with some of the algorithms. The Algorithms that are fast are slow to machine, too. Lots of travel to safe-Z and rapid to the opposite corner of the work, lots of cutting air. But it gets the job done.
[18:43:34] <andypugh> The probe is just a probe I got from eBay.
[18:44:07] <andypugh> I was paranoid about starting the spindle with it mounted, so I decised to make a "Magsafe" connector for it. Wjich turned out pretty well.
[18:47:57] <JT-Shop> one of those magnetic connectors?
[18:53:52] <andypugh> Yes.
[18:54:54] <PetefromTn> nice man....
[18:55:01] <andypugh> A couple of ring-shaped 8mm OD magnets from eBay, and a bit of brass. https://plus.google.com/photos/108164504656404380542/albums/5747722155741347649/5866812284499794850?banner=pwa
[18:55:10] <andypugh> And plastic, and a spring.
[18:56:27] <andypugh> The fixed contact has insulation between the magnet and the inner. The other one doesn't, but the pin is quite long and well-guided to can't short.
[18:57:12] <JT-Shop> very cool
[19:01:37] <andypugh> I maybe ought to make some drawings and document it somewhere. (I "winged it" when I made it, no drawings or even G-code was used). But it is nicer than this one: http://techcrunch.com/2009/06/05/diy-make-your-own-magsafe-connector/
[19:02:41] <JT-Shop> that's Rube Goldberg work there
[19:05:04] <andypugh> Yeah. I can imagine doing that, I can't imagine bragging about it.
[19:12:50] <andypugh> Bah! I should know better by now! It is not a great plan to guess a dimension when building a 3D model of something which you don't actually start to make until weeks later, because you will forget that you guessed a dimension...
[19:13:35] <andypugh> (Yes, the pattern I made today, (not the one in the photo, the next one) is completely wrong...
[19:14:08] <andypugh> It came out nicely though....
[19:15:28] <generic_nick|2> old bicycle tubes make great bellows for piano hinges on the front door of my mill
[19:15:45] <generic_nick|2> no more coolant leak.
[19:22:53] <generic_nick|2> andypugh: that sucks. ive done that before
[19:30:17] <PetefromTn> andypugh: Hey man just about how long did that part for instance take to code up and how long did the actual program run on your machine?
[19:34:57] <Valen> :-<
[19:42:36] <andypugh> PetefromTn: Well, the part I made today was http://imagebin.org/256731
[19:43:22] <andypugh> And it was pretty much all I got done today (bank holiday, no work) but then I wasn't full-on at it and spent a fair bit of time trying to replace the chain on my bike.
[19:43:57] <PetefromTn> braggart LOL
[19:44:58] <PetefromTn> what did you use to draw that part?
[19:45:16] <andypugh> The CAD has been going on for weeks while I work out other things round it. But the actual CAM was probably a couple of hours (of which half was waiting for PyCAM to produce the toolpaths) and total machining time was 4 hours or so (you can take big cuts with the foam)
[19:45:25] <andypugh> Autodesk Inventor.
[19:46:16] <andypugh> But I get a free home license as a side-effect of my job. It's pretty expensive to buy.
[19:47:09] <PetefromTn> Oh so that was foam then? Got a picture of that part? How does it work in almuminum?
[19:50:30] <andypugh> No pictures of the part, and it's getting late here, so I am rather more likely to go to bed than the garage.
[19:51:17] <andypugh> It would work perfectly well in Alumnium, if my milling machine was up to the task. It would take a whole lot longer, though.
[19:53:26] <PetefromTn> My deal is Honestly I don't do much 3d at all but I WOULD like the ability so a FREE program is highly desirable until I can ante up for something more commercial in nature.
[19:54:38] <tjtr33> any suggestion for a replacement? has to be dc brushed. http://imagebin.org/256734
[20:01:41] <generic_nick|2> hmm got quoted about 400 to move my mill and lathe across town.
[20:12:06] <skunkworks> through what?
[20:20:52] <andypugh> PetefromTn: Yes, that is pretty much what PyCAM is doing for me. I generally use 2.5D.
[20:21:56] <Connor> What did I miss? Had a few issues with my ISP
[20:21:58] <PetefromTn> It is a shame it is limited as you say but apparently you are making 3d parts successfully so how bad could it be right...
[20:22:05] <andypugh> I am also hopeful that pyCAM can be improved. However, there are several other things out there, that are also free. I just haven't tried them yet.
[20:22:23] <PetefromTn> Oh man you really missed out on some amazing stuff sorry LOL
[20:22:49] <L84Supper> ... and thats how you turn lead into Gold
[20:23:03] <PetefromTn> yes exactly....Can't wait to try it.
[20:23:48] <PetefromTn> any of you guys using that FSwizard?
[20:24:23] <andypugh> ?
[20:24:25] <PetefromTn> or rather HSMadvisor now...
[20:24:37] <L84Supper> more people seem to be interested in adding 4-5 axis support to PyCam than any other open CAM tool
[20:25:42] <PetefromTn> you using it?
[20:26:01] <andypugh> I would have thought making it produce a toolpath in less than a day would be a better enhancement. (The push-cutter and drop-cutter grid algorithms work fine, but do spend an awful lot of time cutting air)
[20:26:28] <andypugh> The contouring algorithms ought to be better, and give a better finish, but are unusably slow.
[20:27:28] <PetefromTn> does it do waterline and linear etc 3d paths?
[20:27:29] <generic_nick|2> wow, mastercam does it in seconds
[20:27:43] <PetefromTn> how much did you pay for mastercam?
[20:27:59] <L84Supper> we figured speeding up the toolpath generation would come with refactoring PyCam
[20:28:18] <L84Supper> it just needs a makeover
[20:28:31] <andypugh> The Waterline algorithms are the slow ones. I suspect it spends too long considering the tool geometry even in cases where the tool chape is very simple.
[20:29:12] <PetefromTn> aw that sucks...
[20:29:41] <L84Supper> people have been talking about using voxels and the GPU
[20:29:41] <andypugh> L84Supper: When I learn Python, and have infinite spare time, looking at the toolpath algorithms in PyCAM is on my list...
[20:30:15] <andypugh> Yes, I did find myself thinking about voxels while watching the machine do its thing.
[20:30:29] <L84Supper> lots of talk right now and interest, maybe things will get going over the summer
[20:31:08] <L84Supper> having a couple funded devs to work on it will also help
[20:31:16] <FinboySlick> I think Minecraft is responsible for all the voxel hype ;)
[20:31:16] <andypugh> One idea that occurred to me was a simple G-code postprocessor, that scans through G-code and subtracts voxels, then trims any path that is only cutting air.
[20:31:38] <L84Supper> voxelpress is another
[20:32:19] <andypugh> I was thinking about voxels in 1998. I worked in a medical imaging research group.
[20:32:33] <L84Supper> yeah, it's not new
[20:33:04] <L84Supper> image processing and MRI
[20:33:04] <andypugh> My thinking was that building an STL of the part-machined shape was likely to be wasteful.
[20:33:38] <andypugh> Not that Voxels are even slightly efficient.
[20:33:54] <L84Supper> it would be nice to come to some consensus on the approach
[20:34:28] <L84Supper> http://voxelpress.org/rasterization_demo.html
[20:34:32] <tjtr33> reading that paper on jitter made me think, someone should invite Mssrs Proctor & Shackleford to the Witchita get-together
[20:35:11] <andypugh> I need to work out how the current algorithms work. I did have some discussion with sumpfralle on the subject.
[20:36:33] <andypugh> For a cylindrical cutter creating a waterline from STL is very fast and easy. It is when you add a tool radius that it gets hard. And if you have to worry about undercuts, then it gets harder again.
[20:37:29] <L84Supper> after looking over Slic3r, skeinforge, PyCam and Heeks there needs to be some serious effort into making useful CAM tools that aren't >$10K
[20:38:26] <L84Supper> CAM for cutting as well as additive manufacturing
[20:38:50] <andypugh> Anyway, time I was gone.
[20:38:59] <L84Supper> rather that CAM for cutting and another camp working on 3dp
[20:39:12] <tjtr33> L84Supper, did you see the Russian youtube toolChain tutorial on FreeCad->Heeks->Linuxcnc?
[20:39:37] <L84Supper> no, I'll have to check it out
[20:40:33] <L84Supper> the real missing piece is CAM for 3DP, even mastercam and NX CAM don't have anything for additive yet and Slic3r is a toy
[20:42:13] <L84Supper> http://www.youtube.com/watch?v=NP9YZwHbBE4 Freecad to Heekscnc to LinuxCNC Tutorial sounds Polish
[20:42:26] <Tecan> (NP9YZwHbBE4) "Freecad to Heekscnc to LinuxCNC Tutorial" by "Gab Trzyde" is "Tech" - Length: 0:07:57
[20:42:40] <tjtr33> yes it does, dunno where i got the russian idea
[20:49:13] <tjtr33> and i didnt see why he'd deck the top with a path bounded by the shape, or how he'd hold the thing while he ran a perimeter cut
[20:49:32] <tjtr33> but its 'exemplary' i 'spose
[21:32:21] <ssi> hrm
[21:42:39] <eric_unterhausen> got my beaglebone black today, someone make a bob interface for me plz
[21:51:54] <ssi> aoeuaoeu
[21:52:11] <ssi> eric_unterhausen: http://www.youtube.com/watch?v=Jfi0b5cUNVw
[21:52:19] <ssi> getting there :)
[21:58:46] <L84Supper> bob interface?
[22:01:55] <ssi> breakout board
[22:02:03] <ssi> a bbbbb if you will
[22:02:12] <ssi> first one will be pretty simple
[22:02:20] <ssi> a bare bones beaglebone black breakout board
[22:03:40] <L84Supper> most of the GPIO is shared
[22:04:18] <ssi> ALL of it CAN be shared
[22:04:18] <ssi> none of it has to be
[22:04:18] <L84Supper> so 0.1" header to what? screw connectors?
[22:04:45] <ssi> most folks probably will want some analog support circuitry
[22:04:59] <ssi> it's all 3v3 levels and fragile pins
[22:07:23] <L84Supper> yikes, they put the headers on top of the board again
[22:07:32] <ssi> they did
[22:08:01] <pcw_home> Yeah so you have to mount it upside down
[22:08:15] <ssi> wouldn't be that hard to R&R them
[22:08:19] <ssi> well
[22:08:25] <ssi> remove them and install new ones
[22:09:25] <L84Supper> whats the am3358 selling for?
[22:09:55] <Tom_itx> i started one of those for the STM32F4
[22:10:04] <ssi> the number $2.50 sticks out in my mind
[22:11:58] <pcw_home> 1K $22.52 or so for 1Ghz
[22:12:26] <ssi> off by an order of magnitude, dang
[22:12:42] <L84Supper> ~$14 in China
[22:12:58] <ssi> L84Supper: for the ti part? or a clone
[22:13:06] <L84Supper> TI
[22:18:13] <L84Supper> http://circuitco.com/support/index.php?title=BeagleBoneBlack#LATEST_PRODUCTION_FILES_.28A5A.29
[22:19:03] <L84Supper> it would only take a few days to fix this and bump it to 1GB DDR3
[22:19:23] <ssi> fix what?
[22:20:55] <L84Supper> better board layout and 1GB of RAM
[22:21:24] <ssi> do it
[22:21:39] <ssi> what would you change about the layout besides putting the headers on bottom?
[22:23:18] <L84Supper> depends on who it's for
[22:23:48] <eric_unterhausen> how does it hurt that the headers are upside-down?
[22:24:44] <L84Supper> cheapskates just want it for as low a price as possible, they don't care if it has handlebars and a saddle mounted in the center
[22:25:11] <ssi> not an invalid strategy
[22:25:38] <ssi> and suits my needs too... if you make the thing $200, there's little point in working toward replacing $200 PCs
[22:26:14] <eric_unterhausen> depends on how capable the thing ends up being
[22:26:37] <eric_unterhausen> if it works really well, it would probably be worth investing some $ in it
[22:27:00] <ssi> it already works pretty well, he's just grousing because someone else is making the cash :)
[22:28:08] <L84Supper> SODIMM connector or PC-Card type connector on one edge vs connectors everywhere
[22:28:31] <pcw_home> I doubt if anyone is making much cash on the Beaglebone (except TI getting developer mindshare for cheap)
[22:28:51] <L84Supper> it's the 512MB that gets me, for a few $$ more it could have 1GB
[22:28:51] <ssi> that'd certainly make it nicer for fitting into a system, but not great for experimenting
[22:29:17] <ssi> pcw_home: I'm sure you're right about that... especially if the soc is half the retail price
[22:29:33] <ssi> although I've also heard rumblings that TI is subsidizing some of it
[22:29:37] <ssi> exactly for the mindshare reason
[22:29:50] <L84Supper> the cpu could be $6
[22:29:54] <pcw_home> Yep cheap advertising
[22:30:07] <eric_unterhausen> funny thing is, people were whining about the price
[22:30:22] <pcw_home> Not $6 if you or I buy it
[22:30:36] <eric_unterhausen> "olimex blah blah"
[22:30:42] <ssi> eric_unterhausen: it's hard to justify spending much more than $50 on a dev board if you're harry hacker with no clue what to do with it
[22:30:57] <L84Supper> yeah, why i don't bother anymore, the RPi proved that open source doesn't really matter, just price
[22:31:19] <ssi> rpi is a joke
[22:31:42] <pcw_home> I wonder how munkie is doing with his DMA stepgen on RPI
[22:31:44] <ssi> TI is handling this much better than BCM did the rpi
[22:32:01] <eric_unterhausen> raspi is exactly the right price for people to buy it, put it on their desk and later throw it away
[22:32:13] <ssi> eric_unterhausen: heh you got that right
[22:32:31] <ssi> although it's "buy it, put it on their desk, try to find docs on how to do something with it other than write python, throw it away"
[22:32:31] <pcw_home> BCM is shitty to deal with (TI used to be the same)
[22:33:35] <ssi> I had one simple little thing I wanted to do with the rpi
[22:33:36] <ssi> I2S out to a dac
[22:33:45] <ssi> and it's not possible without airwiring to traces on the board
[22:33:51] <eric_unterhausen> BCM is targeting high-volume customers and doesn't care about the rest of us
[22:34:08] <ssi> and even then, you have to write drivers for an undocumented device
[22:34:09] <ssi> or rather, NDA documented device
[22:34:37] <ssi> eric_unterhausen: that's the typical corporate strategy, but then they have no business being involved in a hacker project like the rpi
[22:34:45] <ssi> and I guess that's not really bcm's fault, it's the rpi foundation's fault
[22:34:48] <eric_unterhausen> it's just amazing to me what is under nda nowadays
[22:34:54] <ssi> but they have these lofty academic goals about education and crap
[22:35:03] <L84Supper> the am3359 is ~$14 out the backdoor, 2x the A10
[22:35:18] <eric_unterhausen> bcm makes some neat stuff, don't understand the stupidness about nda
[22:35:26] <pcw_home> Only nifty thing about the BCM chip is the stacked RAM
[22:35:56] <L84Supper> BCM evolved out of military silicon
[22:36:07] <L84Supper> they still have that mentality
[22:36:16] <L84Supper> everything is under NDA
[22:46:51] <L84Supper> i.MX6 Solo 1GHz: ~$11 @ 10k+ i.MX6 Dual 1GHz: ~$23 @ 1k+ i.MX6 Dual lite: ~$16 @ 1k+
[22:48:10] <L84Supper> http://pastebin.ca/2375279 more i.mx6 pricing
[22:48:49] <L84Supper> the quad core sells for around the same price as the am3359
[22:52:59] <ssi> L84Supper: I backed the goofy mx6 + avr project on kickstarter
[22:53:23] <L84Supper> have alink?
[22:53:37] <ssi> http://www.kickstarter.com/projects/435742530/udoo-android-linux-arduino-in-a-tiny-single-board
[22:54:22] <L84Supper> I saw that
[22:54:33] <ssi> was thinking it could be interesting to try writing drivers to run a RAMPS shield on it
[22:54:36] <ssi> since it's pin compatible
[22:55:18] <zultron> Hey L84Supper, I've been hacking on PyCAM, right now just to fix the broken task settings persistence.
[22:55:28] <zultron> If you start working on the TPGs, let me know.
[22:56:15] <L84Supper> zultron: people have been mostly complaining about the slow path generation
[22:56:27] <zultron> I was looking at integrating some of Anders Wallin's work, and it turns out that someone's already started doing some of that maybe; there's an opencamlib branch in SF git.
[22:56:33] <zultron> Yeah, it's very slow.
[22:57:09] <zultron> Still no TPGs for 5-axis, but I've been looking at some journal articles. Just musing for now, not sure whether that'll turn into actual effort or not.
[22:57:50] <zultron> I'd also like to integrate cut simulation; Anders has a library for that, too.
[22:59:32] <L84Supper> ssi: whats the link between the imx6 and the sam3?
[22:59:42] <ssi> I think it's just a uart
[23:02:43] <L84Supper> I was surprised by the popularity of the arduinos, I could never even stand the name :)
[23:03:40] <eric_unterhausen> hard to understand?
[23:03:58] <ssi> L84Supper: I used to be seriously into PIC programming, and the arduino stuff used to irk me too
[23:04:03] <ssi> but I got the fuck over it... they're convenient
[23:05:19] <L84Supper> $100 for the dual core board?
[23:05:22] <ssi> yea
[23:05:43] <L84Supper> $129 for the quad
[23:09:13] <L84Supper> SATA only on the 4-core version
[23:12:10] <ssi> oh there's sata? neat
[23:12:43] <eric_unterhausen> kickstarter has a lot of silly projects
[23:12:51] <ssi> yeah
[23:12:57] <ssi> I should know, one of them's mine :)
[23:13:07] <eric_unterhausen> amazing how much money is on there now
[23:13:20] <eric_unterhausen> I guess for the price of a raspi, you can be a backer
[23:13:53] <eric_unterhausen> with the advantage of no messy desk
[23:14:16] <ssi> the messy desk is the best part
[23:14:17] <L84Supper> heh, imx6 + RPi on the same board
[23:14:37] <L84Supper> with socket for arduino
[23:14:57] <eric_unterhausen> if socket == usb port, I'm in
[23:15:55] <L84Supper> if people would actually cooperate the SOC on SODIMM or similar with standard breakout boards
[23:16:51] <L84Supper> but OEM's don't want that
[23:17:05] <ssi> http://www.kickstarter.com/projects/crystalfontz/cfa-10036-open-hackable-linux-arm-embedded-gpio-mo
[23:17:07] <ssi> people don't want that
[23:17:08] <ssi> I backed that one
[23:17:46] <L84Supper> i.mx28
[23:18:02] <eric_unterhausen> i remember that
[23:18:37] <eric_unterhausen> the sodimm thing always seems to involve $$$ for some reason
[23:18:40] <L84Supper> but a a10, a31, imx6, and similar 4-8 core arm's as they evolve
[23:19:30] <L84Supper> I tried to work with the EOMA-68 stuff but the ringleader was too difficult to work with
[23:19:36] <ssi> well like I said earlier, it's just not super hackerfriendly
[23:19:59] <ssi> at least with the headers on the beagle, I can airwire to it, or build a breadboard, or whatever
[23:20:09] <ssi> sodimm requires spinning a board and sourcing connectors at the very least
[23:20:19] <L84Supper> the beaglebone could be a $30 sodimm module
[23:20:48] <ssi> L84Supper: I could see a use for having that if it's otherwise exactly compatible
[23:20:57] <ssi> then designing a nice machine control board with a sodimm connector on it
[23:21:11] <L84Supper> if the sodimm could be standardized
[23:21:40] <L84Supper> and then have several IO boards
[23:22:57] <L84Supper> laptop, tablet, kitchen sink, machine controller, etc etc
[23:23:34] <pcw_home> some I/O can be standardised betweens SOCs (USB,serial,I2C, SPI, MDIO,MMI.GMII.GPIO,GPMC etc)
[23:23:42] <L84Supper> reusing the PC-card form factor isn't a bad idea either
[23:24:34] <pcw_home> I like SODIMMs better though
[23:25:48] <ssi> yeah probably more convenient for embedded boards
[23:26:10] <L84Supper> http://rhombus-tech.net/evaluated_cpus/
[23:26:45] <L84Supper> it was pretty easy to standardize most of the IO
[23:30:05] <L84Supper> only a few ARM soc vendors would actually sell parts at <1M pcs
[23:30:35] <L84Supper> many of those soc's never made it past the dev board stage and into any production
[23:31:11] <L84Supper> even fewer had open Linux support, many had Linux support only under NDA
[23:33:37] <ssi> what were you evaluating them for?
[23:34:20] <L84Supper> making an open standard soc card form factor
[23:35:01] <L84Supper> making the cpu cards and also the IO boards
[23:38:32] <L84Supper> but most OEM's don't want anything swap-able or upgrade-able, it should die right after the warranty is up and be worthless after
[23:39:01] <fiesh> I'm sorry if this is not really concerning linuxcnc directly, but I figured this might be a good place to ask since people are knowledgable and open. I'm trying to predict the runtime of certain G-code on certain machins (actually controlled by a Siemens Sinumerik), but I'm not sure how this is handled: if I have say a G02 command of certain length and radius, and the speed and absolute value of the acceleration are limited on this
[23:39:06] <fiesh> "segment", plus there are limitations on the acceleration (and jerk, but that wouldn't really matter) and speed of the individual axes, what is the trajectory? Is it somehow well-defined, does it depend on the controller, or on the people who programmed it? Can I somehow predict it?
[23:40:20] <ssi> it should be able to be known by the machine controller
[23:40:39] <fiesh> (Of course this question applies to G01 command to, but I suppose it's safe to assume the optimization problem on those is actually solved correctly, since it's mathematically simple. But on non-straight curves it's different, and I don't want to solve it numerically optimal if that's not how it's done anyway)
[23:40:39] <ssi> it's specific to the acceleration and velocity limits of the machine, which are likely unknown by the programmer
[23:41:08] <fiesh> I actually do know those limits, and will leave them adjustable in my implementation anyway
[23:41:46] <ssi> if you're trying to implement that prediction yourself, and you know the accel and velocity of each segment, I don't see why you couldn't predict it
[23:41:55] <ssi> just remember to account for decel at the end of moves if it will exist
[23:42:25] <fiesh> yes, but the underlying optimization problem is not easy, in fact to solve it correctly you seem to have to take into account the whole NC-program at once
[23:42:26] <ssi> (some machines, like plasma for instance, don't decel, and round corners instead)
[23:42:39] <fiesh> so I was wondering if that's actually what the controller does, seems quite involved a solution
[23:43:04] <eric_unterhausen> not linuxcnc, which is a matter of annoying discussion
[23:43:34] <fiesh> ok, can somebody point me into how it's handled in linuxcnc?
[23:43:40] <eric_unterhausen> linuxcnc can stop at the end of every step, so lookahead is strictly limited
[23:43:54] <fiesh> "can" or "must"?
[23:43:57] <ssi> eric_unterhausen: I was afraid of that
[23:44:02] <ssi> "must be able to"
[23:44:06] <fiesh> oh
[23:44:11] <fiesh> and it always does?
[23:44:46] <eric_unterhausen> no, but it certainly affects the speed of operation
[23:45:00] <ssi> and QUALITY of operation for some applications
[23:45:11] <fiesh> is there a paper online that describes linuxcnc's behavior?
[23:45:18] <eric_unterhausen> I think if we could figure out how to change that it probably would be good
[23:45:18] <ssi> plasma cutting with lots of small segments is very much adversely affected by exact-path planning
[23:45:39] <eric_unterhausen> but it is a very complex issue
[23:46:18] <eric_unterhausen> better than a paper, there is source :)
[23:46:22] <fiesh> in general, the only solution seems to do it numerically, and yes, it can get pretty involved
[23:46:42] <fiesh> hehe, do you have a link? I'd really appreciate that
[23:46:43] <eric_unterhausen> well, people want "run from line"
[23:46:52] <fiesh> "run from line"?
[23:46:57] <ssi> I've never had good luck with run from line
[23:47:52] <eric_unterhausen> I don't know if that is the only issue, as I mentioned before, I find the path planning discussions to be annoying
[23:48:03] <ssi> haha
[23:48:14] <fiesh> oh I'm sorry, I didn't know this was a longstanding issue
[23:48:42] <eric_unterhausen> commercial controls all use some form of multi-line lookahead from what I can tell
[23:48:50] <fiesh> but I felt it fairly complex and wondered if actual controllers really solve it optimally, seems not then I suppose
[23:49:00] <eric_unterhausen> which introduces non-deterministic behavior in many cases
[23:49:08] <fiesh> it does? how so?
[23:49:37] <eric_unterhausen> if you are optimizing over multiple lines, then where are you if you stop?
[23:49:56] <fiesh> oh, well if you really want to optimize, you should consider the whole trajectory at once, of the full program
[23:50:27] <eric_unterhausen> that's not how machines are used apparently
[23:50:46] <ssi> there are historical reasons for that
[23:51:07] <ssi> machines until relatively recently couldn't even hold the full program in ram at once
[23:51:13] <fiesh> I suppose in earlier times, hardware was much too limited to even get close to solving it properly
[23:51:16] <fiesh> yes
[23:51:25] <ssi> I have a machine in the basement that ran off tape
[23:51:32] <ssi> PUNCH tape
[23:51:33] <ssi> D:
[23:51:39] <eric_unterhausen> love that punch tape
[23:51:50] <eric_unterhausen> edit with a razor blade
[23:52:14] <fiesh> heh I suppose it doesn't solve the full variatonal problem at once ;)
[23:52:16] <ssi> but even today, programs can be hundreds of thousands of lines
[23:52:24] <eric_unterhausen> linuxcnc is essentially a fancy drip feed system
[23:53:04] <fiesh> ssi: which would be a tiny problem for a modern computer
[23:53:45] <fiesh> well I don't want to clog up people's backlogs with unpleasant path planning talk if it's not liked, thank you for your help ssi and eric_unterhausen!
[23:54:23] <eric_unterhausen> you could look in the mail archives for lookahead and trajectory planning
[23:55:15] <fiesh> oh that's a good idea
[23:56:49] <fiesh> uhm... is there no search possibility for the mail archives?
[23:58:07] <fiesh> ah the forums have hits
[23:58:35] <Loetmichel> mornin'