#linuxcnc | Logs for 2013-06-14

[00:00:02] <roycroft> indeed
[00:00:09] <toastyde2th> even cnc is about feel
[00:00:23] <roycroft> but if you can't feel it yourself, at least you can smell it
[00:00:27] <roycroft> and that is somewhat telling
[00:00:30] <toastyde2th> (brb)
[00:00:54] <RyanS> I'm wondering if you get visual and sound cues tho?
[00:01:05] <toastyde2th> i always argue that cnc is very very close to manual machining - 70% of the same tasks exist
[00:01:13] <roycroft> yes, but really nothing like the feel of cutting
[00:01:51] <toastyde2th> putting a vice on the table, making sure it's clean, understanding how hard to hit it
[00:02:08] <RyanS> If you do it by CNC you have no idea about feel however? I'm talking about physically
[00:02:34] <RyanS> Because visually I can learn that myself and see if it looks like the cutting is going ok
[00:02:45] <roycroft> if you get a mill drill you'll be able to feel the vibration of the machine
[00:02:58] <toastyde2th> yeah, it's mostly sound + vibration
[00:03:04] <toastyde2th> a machine can sound like shit but the vibration is fine
[00:03:09] <toastyde2th> or it will sound FINE but you know the mill's about to break
[00:03:25] <toastyde2th> most of the feel is in the setup
[00:03:51] <toastyde2th> it takes a LOT of practice to know when a surface is clean
[00:04:25] <toastyde2th> or when a part just isn't sitting quite right in the vice
[00:04:41] <RyanS> so if I heard a vibration sound or whatever that the person could easily tell me whether they can feel vibration?
[00:04:53] <toastyde2th> they will feel vibration, it's what kind of vibration
[00:05:06] <toastyde2th> a good cut has a huge amount of vibration to it
[00:05:27] <toastyde2th> very smooth, bassy vibration
[00:05:52] <RyanS> Okay but if you do it in CNC you have nothing else except visual and audio cues?
[00:06:03] <toastyde2th> no?
[00:06:08] <toastyde2th> i always put my hand on the enclosure
[00:06:57] <toastyde2th> the only way to figure out if it will work for you, unfortuantely, is to try it
[00:07:14] <toastyde2th> because i use a huge amount of physical senses when i set a machine up and very little/almost no visual information
[00:07:34] <toastyde2th> but that's just because it's how i do it - you might be completely successful and just rely on different cues
[00:08:31] <toastyde2th> for instance, whoever does your setup will have to become a legit setup person - a machinist in their own right
[00:08:46] <toastyde2th> you can become an excellent programmer and never touch the setup
[00:09:06] <RyanS> I guess I could prop myself against the machine stand (my wheelchair is not terribly rigid) I can tell what sort of vibration....
[00:09:28] <toastyde2th> it comes through the floor too
[00:10:38] <toastyde2th> but i'd say 45% of the job is setup, 45% is programming, and 10% is running the dumb thing
[00:10:56] <toastyde2th> and "running the dumb thing" becomes more involved the harder the part is to make
[00:11:40] <RyanS> yeh, I went to a industrial show and sat next to a Mazak (I think it was an i-300) and because the place had a car park underneath you could feel the whole vibrating
[00:12:00] <RyanS> Even the laser cutters were shaking all over the place
[00:12:24] <toastyde2th> haha yeah, i loved doing hard cuts when i was at votech
[00:12:31] <toastyde2th> because they had their cnc gear in this fancy glass room
[00:12:34] <toastyde2th> as showpieces
[00:12:41] <toastyde2th> and nobody ever really loaded them down
[00:13:03] <toastyde2th> you could make classes come out of their rooms with some of the facemills they had
[00:13:18] <toastyde2th> huge roar
[00:13:49] <RyanS> on a factory floor with a really thick slab I guess the vibration would be dampened a fair bit
[00:13:57] <toastyde2th> nah, cement sends it forever
[00:14:10] <RyanS> Rather than on top of a car park?
[00:14:26] <toastyde2th> oh, no, not saying it was as loud as a car park
[00:15:02] <toastyde2th> just that it carried the sound all over the building, so that in the front office when our biggest machines were doing roughing the walls and floor of the room would sing a little
[00:15:54] <toastyde2th> (brb)
[00:25:37] <RyanS> What sort of machine/lathe/mill do you have at home?
[00:52:53] <CaptHindsight> http://www.pmdx.com/k9/K9-first-photo.JPG BBB with K9 breakout board
[00:57:42] <ssi> saw that
[00:57:44] <ssi> I want one
[00:58:27] <CaptHindsight> trying to find any schematics, don't see any servo support yet
[00:58:36] <ssi> https://pbs.twimg.com/media/BL-CmZjCAAAUfGu.jpg:large
[00:58:39] <ssi> that's what I'm working on
[00:59:39] <CaptHindsight> can't see the part number
[00:59:54] <ssi> http://ebrombaugh.studionebula.com/embedded/bcc/index.html
[01:00:33] <CaptHindsight> Spartan 3A FPGA with 200k gates
[01:01:13] <ssi> yep
[01:01:22] <CaptHindsight> what real time version are they using for the BBB?
[01:01:32] <CaptHindsight> rt-preempt, xenomai?
[01:01:34] <ssi> there's a newly released xenomai 3.8
[01:01:36] <ssi> but I haven't tried it yet
[01:02:21] <CaptHindsight> looks like ARM support is ahead of x86
[01:02:49] <CaptHindsight> support for RTAI is getting pushed to 3.8
[01:03:22] <CaptHindsight> and we're still battling with RT-Preempt for 3.8 for non debian kernels
[01:04:33] <CaptHindsight> ssi: are your IO pin-outs compatible with anything by Mesa?
[01:04:40] <ssi> not yet, but that's my goal
[01:04:42] <ssi> one of my goals anyway
[01:05:17] <CaptHindsight> pcw was working on an FPGA board for the cubieboard
[01:06:07] <CaptHindsight> ssi: whats the ultimate plan for your FPGA board?
[01:06:28] <ssi> well right now I'm using it as a testbed for hardware bitcoin mining
[01:06:44] <ssi> but I also want to find a way to make hm2 drivers for it and put a cut down 5i25 firmware in it
[01:06:50] <ssi> and use the 28 pin header as a superport for a 7i77
[01:06:51] <CaptHindsight> ssi: talk to Finboy
[01:07:02] <ssi> I don't know finboy
[01:07:25] <CaptHindsight> he's usually in here ~12 hrs from now
[01:07:44] <ssi> I'm making progress: http://www.prototechnical.com/~imcmahon/goodhash.png
[01:08:17] <CaptHindsight> we looked at fitting an entire hash in one cycle in one fpga
[01:08:31] <ssi> it's doable
[01:08:36] <CaptHindsight> it fits but an ASIC is abetter solution
[01:08:39] <ssi> I'm discussing with my friend on it right now
[01:08:44] <ssi> he's actually ni the process of taping out an asic
[01:08:47] <ssi> which is why we're doing this
[01:08:55] <ssi> I'm doing this exercise to get the host-side protocols worked out
[01:08:59] <ssi> gonna use BBB as a controller
[01:09:06] <ssi> plus it's a fun learning exercise
[01:09:22] <ssi> but in the short term he's got an fpga design that'll put three fully unrolled pipes in an artix 200T
[01:09:22] <CaptHindsight> but it seems like 1-2 years ago was the time to get into it
[01:09:27] <ssi> yeah you're right about that
[01:09:30] <ssi> c'est la vie
[01:10:22] <CaptHindsight> I was considering the BBBish design with 1GB DDR3 and an FPGA on the same board
[01:10:51] <ssi> easier just to make an fpga cape
[01:10:58] <CaptHindsight> pass the files around China, hijinks ensue
[01:11:32] <CaptHindsight> just got my BBB's in
[01:11:50] <CaptHindsight> just concerned about the 512MB of ram
[01:11:53] <ssi> I've got two blacks and a white
[01:12:21] <CaptHindsight> playing with the zedboards as well now
[01:15:53] <CaptHindsight> the ZYNQ isn't popular in China yet
[01:16:31] <CaptHindsight> same for the AM335x
[01:21:37] <CaptHindsight> XC3S200AVQ100 are ~$6
[01:22:34] <ssi> I paid more than that from digikey
[01:22:37] <ssi> like $11 or $12 I think
[01:22:42] <ssi> maybe $6 in qty
[01:22:52] <CaptHindsight> China pricing
[01:23:01] <ssi> right k
[01:23:16] <CaptHindsight> usually out the backdoor
[01:24:24] <CaptHindsight> so the BOM for that board could be $12 assembled
[01:29:58] <Tecan> http://static.electro-tech-online.com/imgcache/9838-_penny_heatsink.jpg
[01:30:40] <CaptHindsight> Linuxcnc is only up to 3.5.7 kernels and SSE is broken as well
[01:31:09] <CaptHindsight> ssi: what kernel are you currently using with the BBB and Linuxcnc?
[01:31:28] <ssi> 3.8.13, but I'm not really actively doing linuxcnc stuff
[01:31:34] <ssi> I haven't tried michael's xenomai build yet
[01:32:14] <CaptHindsight> so you're not using any working kernel yet
[01:32:25] <ssi> define "working"?
[01:32:49] <CaptHindsight> a kernel that works with Linuxcnc
[01:33:02] <CaptHindsight> with real time support
[01:33:13] <ssi> no, I don't have that personally
[01:33:21] <ssi> but I think michael does
[01:33:39] <CaptHindsight> have a link to try?
[01:34:11] <ssi> not offhand
[01:34:14] <ssi> I know he posted it to the dev list
[01:34:51] <pfred1> LinuxCNC works with 2 realtime kernels now doesn't it?
[01:35:01] <pfred1> RTAI and what is the name of the other thing xenomai?
[01:35:06] <CaptHindsight> on x86 or ARM?
[01:35:24] <CaptHindsight> RT-PREEMPT, Xenomai and RTAI
[01:35:31] <pfred1> I was close
[01:35:37] <pfred1> I knew it started with an X
[01:35:45] * pfred1 uses RTAI
[01:36:44] <pfred1> beyond the name I don't know anything about Xenomai
[01:37:55] <CaptHindsight> the most recent real time kernel that works with Linuxcnc is v2.6
[01:38:01] <pfred1> I have patches some kernels to run RTAI though
[01:38:13] <pfred1> nah RTAI works on 3.8 now
[01:38:17] <pfred1> kind of
[01:38:57] <pfred1> but I think I built it against 2.6.38?
[01:39:03] <pfred1> which was new enough for me
[01:39:11] <CaptHindsight> ah ok
[01:39:27] <pfred1> I have to be honest the 3.X linux kernels kind of scare me a little
[01:39:31] <pfred1> I've never run one
[01:39:50] <pfred1> although I hear they just changed the major number because of the 25th anniversary
[01:40:02] <pfred1> not because of any major code change
[01:40:50] <pfred1> still to me at least changing the major number seems like a big step so I just haven't taken it yet
[01:42:22] <pfred1> I've been running 2.X Linux kernels since forever
[01:43:07] <pfred1> I can't be sure today but I think the first Linux kernel I ever ran was like 2.0.11 or something
[01:43:42] <pfred1> I can remember by the time 2.0.28 came out I felt like an old hand at Linux
[01:43:59] <pfred1> even though back then i guess I really wasn't
[01:44:05] <CaptHindsight> we are trying to clean up the debianization of Linuxcnc and kernels
[01:44:20] <pfred1> what does that mean?
[01:44:35] <pfred1> one thing i hate about Debian is them stripping out the binary globs
[01:44:41] <pfred1> it is such a pain
[01:45:01] <CaptHindsight> so Linuxcnc can play with kernels from kernel.org git
[01:45:03] <pfred1> I just go get a generic kernel off of kernel.org build that run it and toss the stuff Debian ships
[01:45:31] <CaptHindsight> RTAI just posted a 3.8 patch
[01:45:40] <pfred1> yeah I know I'm on their mailing list
[01:46:11] <pfred1> Paulo helped me out on my last build
[01:46:34] <CaptHindsight> and were battling with page faults with Linuxcnc and RT-preempt kernel passed 3.0
[01:46:52] <pfred1> something about me running multicore i needed to select a kernel option in order for some symbols to be correct
[01:47:00] <CaptHindsight> seems debian has several kernel patches for the page fault problems
[01:47:23] <pfred1> well Debian is far and away the best Linux distro going today
[01:47:33] <pfred1> nothing else even comes close
[01:47:38] <ReadErro-> mint is nice
[01:47:42] <pfred1> pfft
[01:47:45] <pfred1> mint is trash
[01:47:56] <pfred1> the QA there is no comparison
[01:48:14] <pfred1> I have a laptop running mint it sucks
[01:48:18] <ReadErro-> for out of the box and "just works", it just works
[01:48:24] <pfred1> but I'm too lazy to install Debian on it
[01:49:04] <pfred1> I've had Mint freeze up and need to be cold reset
[01:50:05] <pfred1> I ruined the uptime on this box because i just hooked another monitor to it the other day
[01:50:20] <pfred1> but my box upstairs last time I checked it was at 90 days
[01:50:44] <CaptHindsight> http://bb-lcnc.blogspot.com/ fresh off the ML
[01:50:46] <pfred1> I lose power a lot so getting really good uptime is challenging
[01:51:02] <CaptHindsight> http://www.youtube.com/watch?v=AK_OYoNOBDQ&feature=youtu.be
[01:51:03] <Tecan> (AK_OYoNOBDQ) "BeagleBone Black + BeBoPr + LinuxCNC = machinekit" by "Charles Steinkuehler" is "Tech" - Length: 0:12:02
[01:51:23] <pfred1> what kind of latency can the BB do?
[01:51:28] <CaptHindsight> BeagleBone White and a 3.2 kernel
[01:52:26] <CaptHindsight> the AM335x has two embedded microcontrollers to hand real time interrupts
[01:53:27] <pfred1> I wish this box could maintain these numbers http://i.imgur.com/oPOsH0N.png
[01:53:52] <pfred1> not bad for a $5 PC though
[01:54:46] <pfred1> well $12 with a keyboard, mouse and monitor
[01:55:51] <pfred1> that is what I love about PCs they're dirt cheap
[01:57:27] <CaptHindsight> I've been asking about BBB and cubieboard latency tests but I haven't seen any yet
[01:59:09] <CaptHindsight> http://circuitco.com/support/index.php?title=BeBoPr_Cape
[02:00:38] <CaptHindsight> only stepper support :( good for glue gun printers
[02:01:00] <CaptHindsight> up to 62.5 kHz steprate
[02:01:43] <pfred1> CaptHindsight stepper motors are only good for 3D printers?
[02:02:19] <CaptHindsight> are you asking me or telling me :)
[02:02:41] <pfred1> CaptHindsight I'm asking you to clairfy your previous statement
[02:03:00] <CaptHindsight> are you serious?
[02:03:14] <pfred1> CaptHindsight I'm trying to determine if you're serious
[02:04:49] <CaptHindsight> ah only a stepper interface. no drivers on the board
[02:06:54] <CaptHindsight> interesting am3359's are ~$20
[02:08:55] <archivist> that cape only has a bianry and no source
[02:09:49] <CaptHindsight> http://circuitco.com/support/index.php?title=BeBoPr_Cape#Documentations
[02:10:05] <CaptHindsight> on an open systems reference manual
[02:10:44] <CaptHindsight> not a big problem to make a board like that anyway
[02:11:27] <CaptHindsight> LCC version of the am3359 is $13
[02:14:01] <CaptHindsight> am3359 + FPGA + 1GB DDR3 + flash + glue +passives ~$40
[02:17:25] <CaptHindsight> it could be a popular board for cheap in China if it could use the Allwinner A31 or Mediatek quad core
[02:18:46] <CaptHindsight> $30 BOM
[02:22:50] <DJ9DJ> moin
[02:40:20] <BlackSonOfAfrica> Hello LinuxCNC, i want to buy a chinese laser cutter. any advice?
[02:48:43] <Loetmichel> mornin'
[02:51:30] <CaptHindsight> BlackSonOfAfrica: that question must come up 10 times a day
[02:53:09] <archivist> since when
[02:54:12] <archivist> proof that the question is not that common http://emclog.archivist.info/
[02:54:54] <archivist> there are a few laser related questions a month
[02:55:07] <CaptHindsight> define few!
[02:56:13] <archivist> your nick is new in here, I can trawl my logs
[02:56:37] <CaptHindsight> my other nick is busy right now
[02:57:19] <CaptHindsight> BlackSonOfAfrica: depends on your budget, working area and wattage
[02:58:19] <CaptHindsight> I haven't heard of any bad ones, just occasional problems with QC (broken tubes from shipment and pinched wires))
[02:59:36] <ssi> I know someone who bought a chinese machine and ended up putting a us made tube in it
[02:59:44] <ssi> because the chinese one was practically unusable
[02:59:48] <CaptHindsight> ask Jymmm, he's been happy with his
[03:01:40] <CaptHindsight> http://www.automationtechnologiesinc.com/products-page/laser-engraving might be another option
[03:02:15] <archivist> there are 54 laser related lines in my log in the last month, some obviously relate to the same query so the real count is much lower
[03:02:47] <CaptHindsight> http://www.ebay.com/itm/Vytec-Laser-Cutter-240-Watts-/190845851641?pt=LH_DefaultDomain_0&hash=item2c6f4c97f9
[03:03:10] <CaptHindsight> I was tempted to get this one ^^
[03:04:26] <CaptHindsight> 10 times a day, 54 times a month, it's a figure of speech in the US
[03:10:19] <CaptHindsight> every CNC machine vendor at Amerimold 2013 makes their machines in China, but the all qualified their response with "but it's to our specs"
[03:12:18] <CaptHindsight> even AgieCharmilles http://www.us.gfac.com/products/index.cfm
[03:16:59] <RyanS> Is there any reason to use a dead centre in a tail stock when live centres have less friction?
[03:28:47] <archivist_herron> price and accuracy
[03:29:34] <archivist_herron> there is no play on a fixed as long as adjusted correctly
[03:30:28] <archivist_herron> yes it wears out and becomes loose so the user needs to know what he is doing
[04:00:35] <pfred1> can't dead centers be shaped half cut out too?
[04:00:49] <pfred1> so you can face off the end of the piece
[04:09:55] <archivist_herron> yup and you can make specials like internal ones
[05:58:20] <mrsun> isnt that called a half center? :)
[05:58:23] <mrsun> what he is talking about :P
[06:44:43] <pimperle> ac ja on
[06:44:50] <pimperle> sry
[10:22:59] <willburrrr2003> Good Morning Everyone :)
[10:23:13] <IchGuckLive> hi all B)
[10:24:17] <willburrrr2003> After finally getting my classicladder to communicate Modbus RTU to my external PLC lastnight, I had to come to work and print out all the addressing that I will need to access in the External PLC. IDEC was nice enough to give all the data in a wonderful spreadsheet
[10:25:14] <willburrrr2003> The plan this weekend is to get spindle indes pulse, and spindle encoder pulse working
[10:25:39] <willburrrr2003> that was supposed to read Index pulse :p
[11:58:33] <andypugh> This is mad, I am haunting eBay looking for exactly the right oak sideboard to make a lathe stand...
[11:59:09] <cradek> andypugh: how about an old hideaway sewing machine? it'd have nice drawers and everything.
[11:59:20] <andypugh> It needs to be a fair size.
[11:59:29] <cradek> ah yeah those are tiny.
[11:59:56] <cradek> hmm, maybe I oughta get one of those for my watch lathe
[12:00:14] <andypugh> Mine is an overgrown watch-lathe
[12:00:36] <andypugh> I need something a bit like this: http://www.lathes.co.uk/rivett/img14.gif
[12:01:48] <cradek> wednesday night I dragged mine out and fixed a damaged balance pivot (was 16, ended up about 14.5)
[12:02:00] <andypugh> I took the saddle feed clutch apart (for a look) yesterday. It's exquisite but contains some tiny little parts that are bound to wear out. Maybe in the next 200 years.
[12:03:14] <cradek> would be nice, but evil, to retrofit my 1857 english fusee with incabloc settings... :-)
[12:04:08] <andypugh> This style of base would be easier: http://www.lathes.co.uk/rivett/img16.gif but not nearly so retro.
[12:04:15] <cradek> I don't know if they ever made them that big. I don't think I've ever seen them in a big pocket watch.
[12:04:46] <cradek> the shield around the back and sides is a nice idea
[12:04:53] <archivist> that is a naughty idea
[12:04:53] <andypugh> Just make our own. Start with a ruby mine.
[12:05:15] <cradek> archivist: I know I know. but still.
[12:05:35] <archivist> get george daniels book, shows how to make jewels
[12:05:42] <cradek> oh I have it
[12:05:55] <andypugh> archivist: Don't suppose you have any spare triple-belt pulleys lying about?
[12:06:03] <archivist> nope
[12:07:02] <archivist> fitting jewels is a delicate job :)
[12:07:18] <andypugh> My sister does it all the time.
[12:08:10] <andypugh> I like to caption this "cute welder" https://picasaweb.google.com/doltonsjewellers/Workshop?noredirect=1#5032587453469365410
[12:10:05] <archivist> I was talking watch pivot jewels :)
[12:12:51] <andypugh> Right, off to pick up the changewheels for the Rivett that I decided I didn't need (still don't) but then decided shouldn't be separated from their lathe.
[12:13:24] <IchGuckLive> h all B)
[12:13:41] <cradek> archivist: I have trouble getting them set straight and at the right depth so the oil bubble stays where I want it
[12:14:07] <archivist> dont you have a jewel press?
[12:14:29] <Jymmm> sledge hammer
[12:15:21] <archivist> or a staking tool and accessories
[12:16:07] <ktchk> IchGuckLive: Hi I read your button set G54, nice. but is there a way to use video to look at a cross hair and set G54?
[12:20:53] <IchGuckLive> ktchk: you can do what ever you want
[12:21:05] <IchGuckLive> just use PY PIL
[12:21:14] <IchGuckLive> to find your hair
[12:21:39] <IchGuckLive> and then set the coordinates
[12:22:24] <ktchk> anyone did it before?
[12:23:08] <IchGuckLive> i do this on my airfoil
[12:24:56] <ktchk> nice, the best is to have a tangent knife to go with
[12:26:36] <IchGuckLive> in the simple g-code programms is the airfoilgenerator
[12:32:29] <ssi> so I got my hdl design all finished last night, works perfectly in simulation, and it's a thing of beautiy
[12:32:32] <ssi> but it doesn't fit in the device :(
[12:32:57] <CaptHindsight> you can never have too many gates
[12:33:20] <ssi> this is true
[12:33:29] <ssi> and the device I'm using is one of the smallest you can get
[12:33:31] <ssi> but still
[12:33:33] <ssi> how disappointing
[12:33:42] <IchGuckLive> ktchk: this "getpixel((x,y))[0] == 0" is what your looking for
[12:33:59] <CaptHindsight> how about hand packing/schematic entry?
[12:34:14] <IchGuckLive> ktchk: on a black and wite image you may found only 255 and 0 numbers
[12:34:33] <IchGuckLive> ktchk: line 241 in the script
[12:34:39] <ssi> CaptHindsight: it's too big by over 100%
[12:34:47] <ssi> I don't think I'm gonna achieve it with hand packing :P
[12:34:52] <ssi> especialy since I'm not skilled there
[12:39:01] <ktchk> IchGuckLive: Thanks I am looking at it
[12:41:15] <ktchk> IchGuckLive: Is the inkscape tangent knife still working?
[12:41:38] <IchGuckLive> i dont know
[12:42:33] <ktchk> any tangent knife cam for linux?
[12:45:37] <IchGuckLive> the script reads in the Airfoil dat file with 0-1 coordinate points produces a EPS file for drawing on canvas
[12:45:37] <IchGuckLive> for XY UV airfoil it looks at the position of the line at given coordinate on the other plane to synchronize the G-code
[12:45:38] <IchGuckLive> at given point edges of etchother SO i always got the same linequantity in XY UV
[12:45:40] <IchGuckLive>
[12:55:31] <ktchk> The ttt is more interesting
[12:56:12] <IchGuckLive> there is a ttf2cxf available to get more fonts
[12:56:57] <IchGuckLive> short script brings in all your fonts from OS to Qcad folder
[12:57:33] <ktchk> utf8 kanji
[12:58:23] <IchGuckLive> works
[12:59:31] <ktchk> librecad same?
[12:59:37] <IchGuckLive> yes
[13:00:24] <ktchk> I am using inkscape to inport kanji and then heekscnc to make g-code
[13:01:21] <IchGuckLive> also nice i woudt scale at 400px so best resolt on dxf out at inkscape
[13:01:43] <IchGuckLive> doing that 100 times a day for the plasma g-code
[13:02:25] <ktchk> some time need bigbluesaw to make it faster
[13:02:54] <IchGuckLive> Gimppng -> incscape bitmap vectorising svg -> dxf out -> qcad,librecad dxf -> sheetcam -> g-code
[13:03:05] <ssi> ktchk: I know the guy that runs bigbluesaw
[13:03:18] <ssi> :P
[13:03:22] <ssi> but now, lunch!
[13:03:35] <ktchk> the plugin for inkscape is goof
[13:03:38] <ktchk> good
[13:04:20] <IchGuckLive> http://www.schweden-feuer.de/images/Feuersaeulen/Feu_hell/chinese.jpg
[13:04:31] <ktchk> I have to go to bad now bye
[13:04:38] <IchGuckLive> BY
[13:05:09] <ktchk> japanese
[13:12:57] <IchGuckLive> ssi: still here
[13:15:08] <cradek> archivist: experimental waltham 16s with incabloc: http://www.watertownwatchandclock.com/prodimages/exp03mov.jpg
[13:15:35] <cradek> looks like unusual flat/free sprung too
[13:16:47] <archivist> hmm I have a couple of Waltham docs here
[13:21:45] <IchGuckLive> Ah the big blue script gives me better and less lines on the svg
[13:21:48] <IchGuckLive> nice
[13:22:23] <IchGuckLive> but sheetcam reduces also by deg folowing
[13:26:46] <IchGuckLive> im off by
[13:34:39] <Tom_itx> mouser has BBB now
[13:35:06] <archivist> cradek, just found my parts catalogue from 1948 with the 16 A in it in two forms
[13:35:41] <cradek> can you tell if they had incabloc?
[13:37:53] <cradek> (this is a bit OT isn't it)
[13:38:21] <cradek> huh the hamilton 992B I'm wearing is also 1948
[13:38:30] <cradek> american watches were pretty darn good by then
[13:40:01] <archivist> just amoment will scan 4 pages
[13:46:18] <CaptHindsight> I just got my BBB's from Newark, they had 20
[13:47:30] <CaptHindsight> http://www.newark.com/jsp/search/productdetail.jsp?sku=65W6016&COM=superwidget-link_OpenSourceHardware
[14:07:22] <archivist> cradek, box went into swap slowed me down http://www.collection.archivist.info/archive/PC/PC24_Waltham_1948_watch_and_clock_material/
[14:09:46] <ssi> Tom_itx: they have them in stock?
[14:11:34] <cradek> p74/4691-M shows a solid setting
[14:11:57] <cradek> actually they all do
[14:12:09] <cradek> that must be an experimental model he has for sale - interesting
[14:12:58] <cradek> p75 shows an obvious regulator and overcoil too
[14:13:00] <archivist> the serial number is close to the start number so could well be
[14:15:36] <archivist> mice got at this document so it ended up catalogued twice, I need to fix that
[14:19:21] <Jymmm> Can anyone think of a cheap waterproof panel mount connector? (2 conductor, 120VAC@0.1A Max)
[14:20:30] <Jymmm> panel or bulkhead
[15:08:26] <skunkworks> https://www.youtube.com/watch?v=AK_OYoNOBDQ
[15:11:06] <skunkworks> https://www.youtube.com/watch?v=7WHqX0HhXzc
[16:13:34] <jdh> anyone know if a BBB should work with a micro-hdmi->vga adapter?
[16:21:30] <Tom_itx> you asked that already
[16:21:58] <Tom_itx> <jdh> is it microhdmi? or mini?
[16:22:00] <Tom_itx> <andypugh> Micro
[16:23:28] <jdh> that's not the same question thouhg.
[16:23:56] <jdh> but, it appears the answer is: yes, with some externally powered adapters.
[16:42:03] <andypugh> Que?
[16:42:55] <PCW> Que pasa
[16:44:43] <jdh> que sa dilla
[16:46:37] <PCW> que sera sera
[16:47:00] <DJ9DJ> gn8
[16:50:22] <CaptHindsight> PCW: yes, not much IO on that BBB fpga shield/cape/expansion/IO board. I just got some BBB's haven't decided if I'm still going to try and work with the Allwinner quad core SOC's
[16:51:03] <CaptHindsight> PCW: how did your FPGA board for the cubieboard work out?
[16:56:32] <Nick001-Shop> <PCW> After I get all the stepper lines moved for the 7i53S card, Am I still driving the steppers with the gecko stepper drives?
[17:08:33] <PCW> I wont have the cubieboard card for the fest but soon after
[17:09:39] <PCW> Nick001-Shop not sure I understand the question...
[17:11:02] <PCW> I guess Cubie has a A31 board in the works
[17:11:40] <PCW> and the pin compatible A20 cubie should be available "real Soon Now"
[17:13:06] <micges> PCW: are you going to the emc fest?
[17:13:41] <PCW> Yes for the weekend
[17:14:33] <micges> cool
[17:15:33] <PCW> 6-21,22,23
[17:15:49] <Nick001-Shop> <PCW> Trying to run steppers in closed loop and the 7i52S card has outputs. from the 7i2S card - how an I connecting to the steppers? Right m=now In using gecko drivers connected to the parport
[17:16:36] <PCW> Now they need to connect to the step/dir outputs on the 7I52S
[17:19:30] <andypugh> I gave up trying to buy 8-40 ASME thread screws when I discovered that just about the only mention on the whole internet is on a consolidated table of all threads on one web site. My web site.
[17:19:39] <andypugh> So tonight I have been making them.
[17:20:19] <Tom_itx> 8 is 36 threads per my chart
[17:20:23] <andypugh> I only need 2 of that size, and as they also need to be oversize head, fillister form, slotted.
[17:20:36] <Tom_itx> ah hah!
[17:20:52] <Tom_itx> the chart in my office _DOES_ show an 8-40 thread
[17:21:08] <Tom_itx> so somewhere out there is an 8-40 screw
[17:21:40] <Tom_itx> i bet you could find em next week
[17:22:14] <Tom_itx> although i bet it is an odd size
[17:23:03] <Tom_itx> 8 actually comes in 4 thread pitches
[17:23:21] <Tom_itx> 30 32 36 and 40
[17:25:25] <Jymmm> LinuxCNC on Pi http://www.youtube.com/watch?v=WLIo7TA82xE
[17:25:26] <Tecan> (WLIo7TA82xE) "LinuxCNC on Raspberry Pi" by "Kinsa Manka" is "Tech" - Length: 0:04:35
[17:25:47] <Jymmm> with an android tablet as the iface
[17:25:54] <Jymmm> via wifi
[17:35:17] <andypugh> Tom_itx: Yes, now get me one with a slot, fillister form and a 7/16 wide head :-)
[17:36:14] <Tom_itx> the surplus dude has alot of odd screws here
[17:36:35] <Tom_itx> but he's been tossing alot of stuff out lately
[17:38:27] <andypugh> He should toss it onto eBay.
[17:40:05] <andypugh> The other thing is that perhaps 8-40 exists in places where it was historically used, but when it was a current standard (ie, when the lathe was made) we in the UK were using Whitworth.
[17:43:05] <cradek> I wouldn't be too sure of finding 8-40 in the US without some kind of special order. even 5-40 is hard to find at the hardware store.
[17:45:48] <andypugh> Anyway, I don't need it now. I made the 2 off 8-40 screws, and tomorrow I may make the 10-24 and 8-32 screws that are used each about an inch away from the 8-40. Would you be offended if I suggested that the US had rather more screw sizes than strictly necessary>
[17:46:13] <ds3> in what way?
[17:46:37] <cradek> I would only cringe because I'd figure you were trolling the channel :-)
[17:49:17] <andypugh> I am actually serious. The Metric system for screws basically goes M1.6, M2, M2.2, M2.5, M3, M4, M5, M6, M8, M10, M12, M16, M20. How many US screw sizes exist in that interval? (ignoring pitch variants)
[17:49:54] <ds3> well... I have only seen even screw sizes
[17:50:02] <ds3> so that means they are moving in steps of 0.026
[17:50:24] <ds3> up to about 0.25 then it is just fractional
[17:50:59] <ds3> so there are a few more steps
[17:51:19] <Nick001-Shop> <PCW> I see a P2 TX3A /TX3a TX3B /TX3B connection and mention of /TX3BENA which I cant find a connection for. No step/dir outputs on the 7I52S
[17:51:49] <Nick001-Shop> that are marked anyway
[17:53:13] <gene78> yeah 10 & 12, in either thread, 24 or 32, are special order, and broadcast equipment racks use them in 5 lb boxes. Irritating
[17:54:22] <Tom_itx> andypugh, we have coarse and fine thread starting with numbered screws going up to fractional sizes
[17:55:36] <Tom_itx> whitworth was just ... odd
[17:56:24] <andypugh> Eh? Whitworth is the best thread ever, designed on scientific principles to be ideal.
[17:56:37] <Tom_itx> so why did it go away?
[17:56:48] <andypugh> I am not really sure.
[17:57:30] <andypugh> The only odd thing was the 55 degree thread angle, which was chosen to give the best balance between tension and self-locking.
[17:58:07] <andypugh> 60 degrees is easier to _draw_ but that really has no relevance to anything related to making the thread.
[17:59:32] <Tom_itx> so the engineers were just being lazy
[18:01:59] <andypugh> Yes. Curiously BA is much steeper again, 47.5 degrees. Possibly to make the mathematical relations of the other thread proportions work out: http://en.wikipedia.org/wiki/British_Association_screw_threads
[18:03:10] <andypugh> BSP is a 55 degree pipe thread, and is the international standard now: http://en.wikipedia.org/wiki/British_Standard_Pipe
[18:04:37] <jdh> I ran into something that used a 5-40 a few months ago at work. Had to order a box since there were none to be found on the plant site.
[18:08:39] <PCW> Nick001-Shop what the name of your firmware?
[18:09:22] <andypugh> http://www.ebay.co.uk/itm/171055508047 Looks quite a lot like a lathe underdrive cabinet, doesn't it?
[18:10:24] <Nick001-Shop> SVST6_6_7I52S.bit is the file you sent me and I put it into the ini file
[18:11:50] <Nick001-Shop> CONFIG="firmware=hm2/5i20/SVST6_6_7I52S.BIT num_encoders=3 num_stepgens=3"
[18:18:47] <CaptHindsight> anyone ever work with the Kaman SMU-9000-15N Position Sensors? http://www.ebay.com/itm/390445838383?ssPageName=STRK:MEWAX:IT&_trksid=p3984.m1438.l2649
[18:20:30] <CaptHindsight> http://www.kamansensors.com/pdf_files/Kaman_SMU-9000_data_sheet_web.pdf resolution to 1 angstrom
[18:21:29] <andypugh> 0.1% of full scale sounds less impressive.
[18:22:23] <andypugh> It's basically an analogue proximity sensor.
[18:22:29] <CaptHindsight> yeah
[18:23:07] <CaptHindsight> I see them on older wafer rotary stages
[18:23:56] <CaptHindsight> Danaher stopped making that stage and I can't find the old specs
[18:25:02] <CaptHindsight> http://www.dovermotion.com/SpecificProductFamily_Rotary_Tables the new ones are really nice
[18:26:45] <andypugh> http://www.polytec.com/us/products/vibration-sensors/single-point-vibrometers/ can measure distances if 1% of the wavelength of the beam, over several meters, and at kHz frequencies. Not cheap.
[18:31:07] <PCW> Nick001-Shop:
[18:31:08] <PCW> TX0A = DIR0
[18:31:10] <PCW> TX0B = STEP0
[18:31:12] <PCW> TX1A = DIR1
[18:31:13] <PCW> TX1B = STEP1
[18:31:15] <PCW> etc etc
[18:32:39] <andypugh> This looks great, I wish I knew what was going on :-) https://www.youtube.com/watch?v=I6ZxQrjfAE0&feature=player_embedded#!
[18:34:47] <PCW> also for 7I52S GPIO bits 0 and 23 must be set as outputs and set to 0
[18:48:48] <CaptHindsight> how come all the nice granite vibration isolation tables are all 2K miles away from me on ebay?
[18:48:51] <Nick001-Shop> What are the line commands to do GPIO bits 0 and 23 must be set as outputs and set to 0?
[19:04:46] <Tom_itx> andypugh, is he setting up a part to probe for verification?
[19:04:51] <PCW> setp blah.blah.blah.gpio00.is_output true
[19:05:17] <Tom_itx> like a cmm
[19:07:15] <Nick001-Shop> Thanks for the info. Hopefully I'll get to it this weekend. Got buried in work so had to put this to the side
[19:09:32] <PCW> for 7I52S (or any daughtecard) pinouts you need to look at the .pin file that goes with the firmware and the daughtercard manual (thats what I did)
[19:13:03] <andypugh> Tom_itx: I think he is probing a tool for diameter. Note how it gets bigger on screen at the end.
[19:13:48] <andypugh> They have done a lot of work on the graphics. It makes Axis look antique.
[19:21:19] <Nick001-Shop> Is that pin file loaded by default or to I have to put in in the dir by hand?
[19:26:30] <PCW> the pin file is just a text file that comes with the firmware
[19:26:38] <andypugh> Pin file is for humans. XML file is for Pncconf. All that Hostmot2 looks at is the actual .bit file.
[19:27:26] <Nick001-Shop> I'll pick this up at the house computer in about 1/2 hr
[19:27:53] <andypugh> I am not at all sure why Pncconf can't parse pin files. I know why it can't trial-load BIT files. That's because I am late on my promises.
[19:28:43] <PCW> freeby.mesanet.com/SVST6_6_7I52S.PIN
[19:29:23] <PCW> you can also get the same info from the kernel log (dmesg)
[19:29:34] <PCW> (after linuxCNC runs)
[19:30:01] <andypugh> I hear you will be in Wichita, Pete?
[19:30:39] <PCW> Yes Friday to Monday
[19:31:37] <andypugh> Be nice to see you again. You are the only other EMC-ista I have met. I have bizarre visions of a Pico-Mesa Death_Match :-)
[19:32:37] <PCW> I doubt there will be fisticuffs
[19:33:41] <andypugh> Top trumps? "I have PCI" ... "Oh"
[19:35:31] <andypugh> I get the impression that the Pico stuff is good, and easy, but hasnt really moved on that much. It seems ideal for that Indian guy, mainly because I know he can buy it, it will work, and it won't be my fault, or your fault, if it doesn't.
[19:36:51] <PCW> I would like to implement the Fanuc serial protocol encoder(and PWM drive) at some point but I dont have the hardware
[19:36:53] <Tom_itx> ok i did a memtest 3x over on the pc and no errors
[19:37:19] <Tom_itx> brought the mill back in where it's cool to try again
[19:39:05] <andypugh> I keep looking for a chap Fanuc motor to test that bldc really does work with them. Though a guy on CNCzone seems to have made it work.
[19:41:11] <PCW> I think it might be wort making a daughtercard with the differential drive, SPI A-D for current sense and correct connector if I could verify it on the hardware
[19:41:22] <PCW> s/wort/worth/
[19:42:25] <PCW> Tom_itx: try a simpler config maybe theres a HAL tangle
[19:42:49] <Tom_itx> i removed all the postgui stuff
[19:42:55] <Tom_itx> in the ini
[19:43:41] <PCW> I would start really simple (like the hm2 sample config, edited to use your bitfile)
[19:44:12] <Tom_itx> this was working fine before i added a couple things to the postgui like cycle timer
[19:44:24] <Tom_itx> i've run quite a few parts with it
[19:46:38] <Tom_itx> i gotta make a boot thumbdrive so i can run memtest on the old pc now
[19:46:47] <Tom_itx> then i'll do some testing with the atom pc
[19:46:51] <PCW> Something changed that caused the problem and interpreter out of sequence sounds like a task/motion/ something haywire
[19:47:44] <PCW> if you have the same error on both machines its unlikely bad memory (and bad memory is likely to cause crashes)
[19:48:08] <PCW> try a minimal known hal file
[19:48:17] <andypugh> Tom_itx: Properly minimal test. halrun // loadrt hostmot2 // loadrt hm2_pci // show pin.
[19:48:44] <Tom_itx> what's show pin?
[19:49:15] <PCW> bbl Dinner!
[19:49:17] <andypugh> (for a 5i25, you need to specify firmware for others, unless you loaded it earlier)
[19:49:37] <Tom_itx> this is a 7i43 7i47 using a 7i47 bitfile
[19:49:37] <andypugh> show pin is a command to list all the pins
[19:49:57] <PCW> the problem looks to be higher up (not low level I/O)
[19:50:29] <PCW> thats why I think its not hardware related
[19:51:17] <andypugh> show param / show funct / show thread are also useful (and can be inserted in your HAL file). I tend to put in a "show pin" after the "loadrt" section to have a list of actual pin names to compare error messages to.
[19:51:57] <PCW> (unless there a bug in the hm2 driver that causes breakage elsewhere in some odd circumstances but I think this is not likely)
[19:52:04] <PCW> bbl
[19:56:49] <andypugh> Great quote from a friend "I managed to get myself a job that was meant to be Postdoctoral by cashing in my three masters' degrees in the academic equivalent of swapping 3 houses for a hotel in Monopoly"
[19:57:21] <jdh> I'm thinking about using G43 on an outside profile and telling it the tool is 0.020" bigger than it really is, then cutting it again with the correct tool size for a finishing pass. Will this actually work?
[19:58:17] <Tom_itx> unless you have radaii that are smaller than the tool radius plus the .020
[19:58:32] <Tom_itx> then it will bitch
[19:58:59] <andypugh> But yes, that will work.
[19:59:45] <jdh> there are some inside corners. I guess that might trigger that error
[19:59:52] <Tom_itx> you can't just offset the tool .020 in the cad file?
[19:59:53] <andypugh> Ah, err, lathe or mill?
[19:59:58] <jdh> mill
[20:00:19] <andypugh> It will only have any effect at all in G41/G42 mode.
[20:00:21] <Tom_itx> then the path will be corrected
[20:00:34] <jdh> yeah, I guess offsetting and having two different cuts would be better
[20:00:52] <Tom_itx> i always add a bit before the finish pass in the tool path
[20:01:23] <Tom_itx> cut to depth taking however many steps you need then do a final pass at full depth
[20:02:28] <jdh> initial cut (in 0.5" 6061) with a 0.25" end mill, then a finish pass with 0.125 end mill to make the internal radius smaller
[20:05:27] <Tom_itx> it'll really squeal in the corners
[20:05:39] <jdh> or snap
[20:05:47] <Tom_itx> not if you go slow
[20:06:33] <Tom_itx> or make a 'to depth' cut wiht the .125
[20:06:41] <Tom_itx> in the corners then the final pass
[20:06:52] <Tom_itx> takes longer
[20:07:22] <Tom_itx> just make a 'U' lead in / out in each corner then do the whole profile final pass
[20:07:47] <Tom_itx> probably not practical for production but...
[20:08:25] <Tom_itx> how deep is the final cut?
[20:08:52] <jdh> it is .5" 6061
[20:09:08] <Tom_itx> a .125 em with a .5 flute may be hard to come by
[20:09:30] <jdh> I have 1, and only 1
[20:09:40] <Tom_itx> with a full cut at .5"?
[20:09:41] <jdh> has 1"
[20:09:48] <Tom_itx> good luck
[20:10:02] <jdh> heh, probably flex and/or snap at the bottom
[20:10:32] <Tom_itx> take the corners at 1/2 feed
[20:10:32] <jdh> I'll just do the whole thing with 1/4" endmill and clean up the radius by hand
[20:11:03] <Tom_itx> another thought would be to pre drill the corners with a drill
[20:11:17] <Tom_itx> .125 drill
[20:11:22] <Tom_itx> then do the clean up pass
[20:11:32] <jdh> that's what I was goign to do
[20:11:38] <Tom_itx> there's always more than one way to skin a cat
[20:12:05] <Tom_itx> that might work then, just run that long skinny cutter slow
[20:12:57] <Tom_itx> you sure don't want any flex in your z axis
[20:13:08] <jdh> http://www.artichoke.org/bracket.jpg
[20:14:31] <Tom_itx> you're gonna get chatter with that cutter anyway so why not do the cleanup pass with the .250 em then clean out the corners with the .125
[20:14:51] <Tom_itx> with just a short cutter path
[20:15:10] <jdh> I'd have to code it by hand
[20:15:20] <jdh> or at least extract it by hand
[20:15:22] <Tom_itx> do your lead in / out at like a 20 deg angle or such
[20:15:49] <Tom_itx> that will help blend the cut
[20:16:21] <jdh> yeah, or just clamp it in the vise and use the keyboard
[20:16:43] <Tom_itx> you'll need a .125 radius with a straight cut on each side to get the .250 radius clean then 2 lead in / out moves
[20:18:18] <Tom_itx> you better move your .125 holes though
[20:18:28] <Tom_itx> or they will be undercut
[20:18:41] <Tom_itx> you have the center on the finish cut line
[20:19:01] <Tom_itx> move them in .06250 in x and y
[20:20:08] <Tom_itx> unless that was intentional
[20:20:23] <Tom_itx> sometimes that is done in sheet to avoid tearing
[20:20:50] <jdh> the holes are centered on the finished corner
[20:21:07] <jdh> that is before the finish cut with .125
[20:21:32] <Tom_itx> it doesn't appear so
[20:21:36] <Connor> Question. What do you do when you want to use a end mill with not set screw flat in a set screw holder? Grinding a flat acceptable ??
[20:21:45] <Tom_itx> sure
[20:23:05] <Tom_itx> jdh, if the finished part is to have .125 radius'd corners i think you have those holes too far into the part
[20:23:17] <Tom_itx> or you're removing more than .020 on the finish pass
[20:24:35] <jdh> that was generated with a single finished pass with 0.25. If I recut with a .125, those would disappear
[20:24:53] <Tom_itx> k
[20:24:57] <jdh> this isn't real CAM, it's cut2d... it can only generate cuts for closed vectors
[20:24:57] <Tom_itx> just trying to avoid scrap
[20:25:36] <jdh> I got some plastic the same size for a test cut. In retrospect, the plastic cost almost as much as the aluminum
[20:25:48] <Tom_itx> try wood
[20:25:57] <Tom_itx> or wax
[20:26:09] <jdh> I have no wax, don't like cutting wood, it's messy
[20:27:08] <jdh> http://www.artichoke.org/plate.jpg
[20:27:18] <Tom_itx> lately i'll run one in air and if it looks close i'll run it
[20:28:36] <Tom_itx> until this silly machine started acting up
[20:31:35] <Tom_itx> do those parts mate?
[20:31:39] <Tom_itx> in the slots
[20:34:58] <jdh> in theory.
[20:35:06] <jdh> after I fix those corners also
[20:35:43] <jdh> then welded
[20:35:52] <Tom_itx> yeah i figured
[20:36:37] <jdh> I did some test runs with a 90 degree v-bit. It puts a nice groove for penetration/bead
[20:46:17] <andypugh> Night all
[20:56:54] <Tecan> anyone here use emc to reprap ?
[20:57:27] <Tecan> im debating on using my printerport driver or teacup
[20:57:43] <Tecan> should probably go with the flow on this one
[21:47:07] <atom1> ok so i ran the memtest and that turned out fine now a latency test gives Servo thread max interval: 1013797 Max Jitter: 18637 Base thread max interval 41982 Max Jitter: 19768
[21:48:02] <cradek> sounds fine
[21:48:04] <atom1> so will the standard SERVO_PERIOD = 1000000 be ok with those numbers?
[21:48:14] <atom1> no base period since it's mesa cards
[21:48:40] <atom1> i'm just trying to iron out these errors i'm getting
[21:48:49] <atom1> kinda starting from scratch again
[21:49:03] <jdh> got any old hal/ini copies?
[21:49:38] <atom1> i was just gonna basically start over
[21:49:48] <cradek> yeah it'll be totally fine
[21:49:50] <atom1> i have my current configs for all the pinouts etc
[21:49:54] <cradek> what errors? I didn't read back.
[21:50:00] <atom1> it's really odd
[21:50:17] <atom1> it will stall out during the middle of a cut and sometimes ask for another tool
[21:50:33] <atom1> doesn't always generate an error
[21:50:36] <cradek> oh I remember hearing about that
[21:50:52] <cradek> have you checked dmesg after it happens?
[21:50:56] <atom1> so i brought it inside where it's cool thinking it could have been a heat problem
[21:51:05] <atom1> i posted dmesg
[21:51:34] <atom1> http://tom-itx.dyndns.org:81/~webpage/cnc/cnc_files/dmesg.txt
[21:51:37] <cradek> after the error happened?
[21:51:40] <atom1> not the most recent but one of them
[21:51:51] <atom1> i believe that one was
[21:52:33] <atom1> also did 'tail -f /var/log/syslog and nothing showed up really
[21:52:34] <cradek> this is three linuxcnc startups, one with debugging on
[21:52:46] <atom1> how do i clear dmesg?
[21:52:54] <cradek> sudo dmesg -c
[21:52:58] <atom1> i can run it again
[21:53:14] <cradek> you always get the weird behavior?
[21:53:26] <atom1> no it ran fine for quite a while
[21:53:38] <atom1> i've cut quite a few parts with no problems
[21:53:39] <jdh> but it is repeatable now?
[21:53:42] <cradek> I'm interested in the dmesg right after you get the weird behavior
[21:53:53] <atom1> started a new file and now it's acting up
[21:54:00] <atom1> i also posted the gcode
[21:54:30] <atom1> http://tom-itx.dyndns.org:81/~webpage/cnc/cnc_files/Cube.txt
[21:54:47] <atom1> it always happens after the drill and during Tool 5 cuts
[21:55:18] <atom1> it's as if you pushed the pause button
[21:55:36] <cradek> what line does it stop on?
[21:55:43] <atom1> then if you try to continue it acts like it resets the file and asks for the first tool again
[21:55:48] <atom1> it's random where it stops
[21:56:10] <atom1> i got a bit further into the file this evening before it stopped
[21:56:57] <cradek> what are you doing with debounce and toggle2nist?
[21:57:13] <atom1> debounce is for the limits
[21:57:27] <atom1> toggle2nist i believe is part of the pendant code
[21:57:27] <cradek> oh I see the link to your config
[21:57:43] <atom1> http://tom-itx.dyndns.org:81/~webpage/cnc/configs/sherline/
[21:58:04] <atom1> i did comment out the postgui.hal and display.xml in the .ini file and it didn't seem to matter
[21:58:37] <atom1> that post is probably before i commented those out
[21:58:41] <cradek> my_jog.hal has complicated stuff hooked to halui pause/resume
[21:58:53] <atom1> it does
[21:59:24] <cradek> are you sure that's not just triggering it? you could watch halui.program.pause with halscope's trigger function
[21:59:39] <cradek> you could also turn on TASK ISSUE debugging and look for program pause messages
[22:00:06] <cradek> or you could unhook all your pendant stuff
[22:00:25] <atom1> DEBUG = 0x7FFFFFFF
[22:00:28] <atom1> right now
[22:00:29] <cradek> the most likely reason you're seeing a pause is that something is pausing it
[22:00:47] <jdh> $200 looks pretty sw33t
[22:00:49] <atom1> i could remove the pendant code entirely
[22:01:03] <cradek> narrowing the problem space is always a smart approach
[22:01:22] <atom1> but it had been working for quite a while
[22:01:34] <atom1> i started adding stuff to the axis screen like cycle times etc
[22:01:41] <atom1> then things seemed to go south
[22:02:06] <cradek> if you had your config in revision control, you could revert your changes and retest
[22:02:16] <atom1> does it matter if all your loadrt lines are together?
[22:02:32] <cradek> their order can matter
[22:03:16] <atom1> they may not all be together in the file is why i asked
[22:04:14] <atom1> there's some in the main hal and some in the my_jog hal as well
[22:08:06] <eric_unterhausen> 3d printing looks like it is exploding
[22:08:11] <eric_unterhausen> which is good, I suppose
[22:08:22] <jdh> lots of hype
[22:08:35] <eric_unterhausen> I mean the diy stuff
[22:08:46] <eric_unterhausen> the commercial side is a bit of a clown show
[22:09:06] <jdh> a lot of the DIY crowd are clownlike also
[22:10:06] <jdh> local professor guy built one, did the 'anyone can print a gun' thing story for local press
[22:10:25] <eric_unterhausen> ok, plz to be smacking him upside the head for all of us
[22:10:59] <eric_unterhausen> I swear some colleges teach people to hype things
[22:11:23] <jdh> maybe it's required for grants and stuff
[22:11:44] <atom1> ok sitting idle i'm getting several errors
[22:11:52] <jdh> what kind of errors
[22:12:10] <atom1> can't do that (EMC_TASK_PLAN_RUN) in auto mode
[22:12:36] <jdh> weird halui pins?
[22:12:38] <atom1> can't do that (EMC_TRAJ_SET_TELEOP_ENABLE) in auto mode with the interperter idle
[22:12:49] <atom1> for a couple
[22:13:20] <atom1> some say the task plan one in auto mode an others in manual mode
[22:13:26] <atom1> and i haven't moved anything
[22:13:55] <jdh> any pause/resume/on/off pins?
[22:14:32] <atom1> monitoring the halui.program.is-paused .pause and .is-running
[22:14:46] <atom1> .is-paused is yellow the other 2 are red
[22:14:55] <atom1> but .is-paused was red
[22:15:54] <atom1> i'm also using some direct inputs to the extra 50 pin header on the 7i43 using some pullup resistors
[22:16:31] <atom1> i'd have to check exactly which lines they are and if they pull up or down, i don't recall off hand
[22:16:45] <atom1> i wonder if those resistors may not be strong enough
[22:16:54] <atom1> iirc i used 330 ohm
[22:17:32] <atom1> ok it keeps switching the .is-paused pin, now it's red
[22:17:39] <atom1> and also generated another error
[22:17:59] <atom1> can't do that (EMC_TASK_PLAN_RUN) in manual mode
[22:18:11] <jdh> seems like is-paused is just a symptom, same as the error message. Do you have any pins tied to modes?
[22:18:12] <skunkworks> noise
[22:18:54] <atom1> just a sec i'll post a screen shot
[22:19:31] <cradek> you should monitor with halscope's triggering. nothing else will catch noise.
[22:19:56] <atom1> i wasn't sure how to set that up so i tried this first
[22:19:58] <cradek> you pretty clearly just have a noise problem don't you? I mean you used debounce already for some stuff as a workaround.
[22:20:15] <cradek> your pendant stuff is randomly flailing on all the halui inputs
[22:20:53] <atom1> i used debounce on the limit switches
[22:20:58] <atom1> afik that was it
[22:20:58] <cradek> why?
[22:21:21] <atom1> http://tom-itx.dyndns.org:81/~webpage/cnc/cnc_files/screenshot.png
[22:21:38] <atom1> because they were bouncing when the machine was on
[22:21:47] <atom1> i did watch that on halscope
[22:22:36] <cradek> do you understand that this is probably another symptom of the same problem?
[22:22:51] <atom1> i'm kinda leaning toward the pullups on the io
[22:22:53] <cradek> you should work on fixing the problem in hardware, not software
[22:22:54] <atom1> yes
[22:23:12] <atom1> i rewired the whole machine using shielded wire on all the io and motors
[22:23:20] <atom1> it helped alot
[22:23:25] <jdh> I have one flakey home/limit switch that I debounce. It's just not quite as flakey now.
[22:23:29] <cradek> yes maybe you need pullups or something
[22:23:46] <cradek> there may be pullup-related settings in the mesa stuff, not sure
[22:23:48] <atom1> i did have to invert the logic on those directly tied to the 7i43
[22:24:04] <atom1> are they 3.3v or 5v logic at that point?
[22:24:17] <cradek> but you've sure got a hardware problem, don't try to fix it in software
[22:24:33] <atom1> i'm not disagreeing with that
[22:24:41] <cradek> I don't know, sorry, check the 7i43 manual
[22:24:57] <atom1> i just wasn't sure what direction to head but i think i have an idea now
[22:25:14] <atom1> i'll disable the pendant and see what happens
[22:25:17] <cradek> great
[22:25:28] <atom1> but probably not tonight
[22:25:47] <jdh> I got a 7i37ta for my other port. Makes everything simple.
[22:25:53] <atom1> it had been working ok for a while though so i really didn't suspect it
[22:26:09] <atom1> i got a 7i47 on the first port
[22:26:41] <atom1> the 2nd port is direct wired
[22:26:58] <atom1> and some of the pendant switches are on it i'm sure
[22:27:08] <atom1> i'll dig out my wiring chart tomorrow
[22:28:04] <atom1> jdh, the 7i37 is just a buffered io isn't it?
[22:28:34] <atom1> i wonder if the buffers are schmitt triggers
[22:29:08] <jdh> more than just buffered
[22:29:37] <jdh> http://www.mesanet.com/aiodaughter.html
[22:30:14] <jdh> it will drive anything reasonable, high or low and read anything reasonable.
[22:30:32] <atom1> how much was it?
[22:30:49] <jdh> $80 maybe. About the same as the 7i47
[22:33:26] <atom1> yeah, looks like the regular one is 69 and the screw terminals are 79
[22:34:15] <atom1> does that require a bitfile for it?
[22:34:57] <atom1> i wonder how you'd combine the 7i47 and 7i37 bitfiles if it isn't
[22:35:12] <jdh> mine does both
[22:35:28] <atom1> both what?
[22:35:43] <jdh> 7i47 & 7i37ta (on a 7i43)
[22:36:00] <atom1> oh so you're running basically the same setup
[22:36:16] <atom1> what bitfile do you use for it?
[22:36:21] <jdh> ignoring the part about mine working,sure :)
[22:37:01] <atom1> i had all that xilinx stuff loaded once but i don't think it is now
[22:37:20] <jdh> can't see it from here. It is one of the stock ones
[22:37:41] <atom1> i think i had to get mine from mesa
[22:37:56] <atom1> i don't think it came by default
[22:40:46] <jdh> loadrt hm2_7i43 config="firmware=hm2/7i43-2/SVST2_4_7I47.BIT
[22:41:05] <atom1> that looks the same
[22:41:07] <jdh> maybe I did get it from mesa
[22:41:42] <atom1> firmware=hm2/7i43-4/SVST2_4_7I47B.BIT
[22:42:01] <jdh> I have the little one
[22:42:08] <atom1> mine is B
[22:42:09] <atom1> yes
[22:42:16] <atom1> 'big'
[22:42:40] <atom1> probably didn't need it
[22:44:57] <atom1> loadrt hostmot2
[22:45:00] <atom1> is that right?
[22:45:15] <atom1> i don't have hm2_7i43 that i see
[22:46:00] <atom1> or is that the same?
[22:46:48] <jdh> my install has hm2_7i43 and hm2_pci
[22:47:09] <atom1> you have a pci card?
[22:47:13] <jdh> no
[22:47:24] <atom1> i wonder what the difference is
[22:47:27] <jdh> my linuxcnc install. My hal config has the line I pasted above.
[22:47:48] <atom1> my hal has the line i pasted
[22:48:36] <atom1> running again with no pendant code loaded
[22:51:17] <grandixximo> I'm looking for a low latency ITX motherboard
[22:51:25] <grandixximo> the Intel D945GCLF2 it's hard to find these days
[22:51:42] <grandixximo> i was wondering if there is any new hardware with low latency
[22:52:14] <grandixximo> and does DCP latency have anything to do with the real-time latency of linuxcnc?
[22:52:40] <grandixximo> DPC latency sorry
[22:53:10] <atom1> well it ran clear through that time
[22:53:34] <atom1> i bet it's the pullups on my pendant io going to the 7i43 2nd port
[22:53:53] <atom1> maybe i should get one of those 7i37ta cards
[23:54:25] <RyanS> hmm a 'fitter and turner' does that mean they don't let them loose on a milling machine....