Back
[01:45:44] <mrsun> centerdrills ambitions in life is to snap off in the holes.. specialy little ones =)
[01:46:35] <archivist> do not force them, get your chuck concentric
[01:47:12] <ssi> are you actually centerdrilling?
[01:47:14] <ssi> or are you spot drilling
[01:47:50] <mrsun> hehe was something doubleboost on youtube said =)
[01:48:18] <mrsun> found it quite funny as ive found the same thing, but mostly due to slop in the quill of the lathe =)
[01:48:39] <ssi> :)
[01:48:45] <ssi> too many people spot drill with centerdrills
[01:48:54] <ssi> it's my ambition in life to stamp that practice out :D
[01:48:59] <mrsun> ssi, i do that in the milling machine
[01:49:16] <mrsun> as i do not have any spotdrills =)
[01:49:17] <ssi> bad mrsun, bad
[01:49:23] <ssi> spotdrills are cheap :)
[01:49:26] <ssi> and they don't snap off!
[01:49:28] <ssi> and they start better holes
[01:49:48] <ssi> ok now that I've thouroughly chastised someone
[01:49:50] <ssi> I'm going to bed!
[01:50:01] <mrsun> hah
[01:50:14] <mrsun> like yo are going to get any sleep now that you are all wound up :P
[01:50:19] <ssi> :'(
[02:01:26] <DJ9DJ> moin
[05:39:58] <RyanS> how is your CNC scratch build going Valen?
[05:41:02] <mrsun> 1 2 3 a block 4 a block rock!
[05:41:03] <Valen> slowly
[05:41:12] <frallzor> booooh mrsun
[05:41:15] <mrsun> frallzor, haha
[05:41:31] <frallzor> block you
[05:42:13] <frallzor> weey package to pick up with seals inside, here I come vacuum-experiments
[05:43:11] <RyanS> I have a project like that except it's been going slowly for four years
[05:44:07] <Valen> doing other stuff around it at the moment
[05:44:18] <Valen> getting 3 phase power to the shed, extractor fans and the like
[05:44:57] <RyanS> jeezus what sort of beast are you building
[05:45:31] <Valen> the extractor is the thing
[05:45:34] <Valen> stupid powerfull
[05:45:49] <Valen> we got our current mill up to 5A on the spindle ;->
[05:45:52] <Valen> 240v
[05:47:09] <RyanS> Are you building to some available plans for the new machine or rolling your own?
[05:47:10] <Valen> thats actual draw
[05:47:14] <Valen> rolling our own
[05:56:59] <RyanS> I read and conflicting reports. Some making it out to be easy to convert a mill using a kit, but then I read about the process and see that there is a lot of work involved.
[05:57:44] <RyanS> Not many options for turnkey stuff here
[05:58:51] <jthornton> convert a manual mill?
[05:59:04] <Valen> you can convert
[05:59:12] <Valen> but unless the manual mill is good to start with
[05:59:28] <Valen> you now have a cnc piece of crap rather than a manual piece of crap
[06:00:48] <RyanS> if Money wasn't an issue the choice is easy - get a vmc
[06:01:46] <RyanS> no a multitasking centre
[06:02:13] <RyanS> You had the HF45 yeah?
[06:11:47] <Valen> yeah
[06:14:47] <RyanS> no good?
[06:15:44] <Valen> its chineese
[06:15:50] <Valen> and "cheap"
[06:17:36] <RyanS> Depends on if your building space shuttles :p idk people seem to be able to build amazing things with cheap machines
[06:18:07] <Valen> probably moreso with manual machines
[06:18:20] <Valen> as you can fiddle to compensate for problems
[06:21:49] <RyanS> true. thinking of getting into model engineering, design in cad & use cnc, but yeah the cost thing
[08:08:59] <skunkworks> http://www.youtube.com/watch?v=BSF1F2dOui4
[08:09:00] <TombCat> (BSF1F2dOui4) "RPi LinuxCNC usb glitch" by "Kinsa Manka" is "Tech" - Length: 0:02:40
[08:09:29] <skunkworks> I thought it was decided that the rpi was a bit too underpowered..
[09:09:55] <pcw_home> I think the RPI is probably fine if its not running a fancy GUI
[09:36:47] <mozmck> anyone here know anything about a Pratt & Whitney Turnmate lathe? There is one near me in working condition with 6" chuck for $1500
[09:49:04] <skunkworks> mozmck, are you able to come to the fest?
[10:05:55] <mozmck> skunkworks: looks like not :(
[10:06:07] <mozmck> too much to do.
[10:09:22] <skunkworks> aww
[10:10:32] <ssi> always too much to do
[10:13:32] <archivist> I cannot come to play either
[10:18:25] <cpresser> nice machine for a retrofit:
http://cgi.ebay.de/ws/eBayISAPI.dll?ViewItem&item=321138279870
[10:18:36] * cpresser is trying to get this one :)
[10:19:44] <ReadError_> H&K ?
[10:19:44] <ReadError_> wow
[10:20:09] <pcw_home> on wheels!
[10:20:41] <cpresser> i guess i will strip the wheels :D
[10:20:41] <ssi> neat
[10:20:50] <ssi> 2 spindles?
[10:20:52] <ssi> one high speed?
[10:21:04] <cpresser> toolchanger
[10:21:12] <ssi> oh
[10:21:20] <ssi> oh ok I see it now
[10:21:26] <ssi> in the wee picture it looked like a second spindle :P
[10:32:58] <Loetmichel> cpresser: a little small but grat for hobby work i would assume ;-9
[10:33:06] <Loetmichel> great
[11:12:54] <archivist> Loetmichel, for some work small is just right, I wish it was nearer me and affordable by me
[11:16:43] <archivist> it would be perfect for my little gears except for lack of height for the other two axes
[11:16:59] <pcw_home> Looks better than hobby grade...
[11:17:31] <archivist> justrightforme grade
[11:18:27] <pcw_home> pretty small but shipping would not be cheap
[11:54:36] <gene78> Got a halscope prob guys, it shows encoder.0.velocity as 0 microvolts, while turnin 600 rpms
[11:56:18] <jdh> you measure your speed in uV?
[11:57:04] <gene78> actually, its turning about 750 but I can set that with pid.0.FF0, no, but the scope is flat lined even at microvolt gain
[11:58:09] <gene78> i can look other places in the chain, like scale.1.in
[11:58:54] <gene78> dead there too
[12:02:05] <pcw_home> is the encoder counting?
[12:04:15] <cbjamo> pcw_home: How do i put that bitfile you sent me on the fpga?
[12:04:18] <mhaberler> cpresser: red mal mit Norbert Schechner <norbert@schechner.info>, der hat mehrere von denen und vielleicht eine übrig mit zubehör
[12:05:09] <pcw_home> Can opener?
[12:06:02] <pcw_home> you copy it to /lib/firmware/hm2/5i22/ (as root)
[12:07:14] <cbjamo> 5i22 or 5i22-1?
[12:07:30] <pcw_home> and then invoke it in the config line of the hal file
[12:07:34] <pcw_home> 5i22
[12:08:06] <cbjamo> and where am I going to want to put the .pin and .xml files you sent?
[12:08:25] <pcw_home> loadrt hm2_pci config="firmware=hm2/5i22/svsttp6_6_12 ...
[12:08:42] <pcw_home> they can go there as well
[12:09:15] <pcw_home> the .pin file is just for your use (its just a text pinout file)
[12:09:27] <cbjamo> ok
[12:09:49] <pcw_home> I recommend reading it before attaching daughtercards
[12:10:42] <cbjamo> Ok, but, isn't there only one way to attach them?
[12:10:46] <IchGuckLive> hi all B)
[12:11:21] <IchGuckLive> sunny and warm here in germany today
[12:12:40] <cbjamo> Shouldn't the hall sensors be returning to the 5i22?
[12:13:08] <IchGuckLive> pcw_home: hi isent there a way you can ad a ppa to get all the newest files in one shot to the linuxcnc distro
[12:15:10] <pcw_home> Yes the Hall signals (if you have them) cand be read by the 7I39
[12:15:55] <cbjamo> right, but they arn't being sent back to the mainboard in that .pin file.
[12:16:00] <pcw_home> IchGuckLive: you mean all latest bitfiles?
[12:16:26] <pcw_home> there are no specific Hall pins, they are GPIO
[12:17:02] <cbjamo> Ah, ok
[12:19:34] <IchGuckLive> pcw_home: and also the firmware
[12:20:21] <IchGuckLive> in one dep packet
[12:34:38] <cbjamo> pcw_home: I want num_pwmgens=0 and num_3pwmgens=3, correct?
[12:36:06] <pcw_home> num_3pwmgens=6 num_encoders=6
[12:36:20] <cbjamo> ok
[12:36:49] <pcw_home> there are no pwmgens so no need to specify 0
[12:37:14] <pcw_home> num_stepgens = however many you want
[12:37:56] <pcw_home> (if unspecified, the will all be enabled which is very unlikely to be what you want)
[12:38:35] <cbjamo> Yup, I only need 4. will it use channels 0,1,2,3 by default?
[12:42:20] <pcw_home> No, you need to specify: num_stepgens =4
[12:42:31] <pcw_home> otherwise all 12 will be used
[12:42:48] <cbjamo> Right, i set num_stepgens=4, but what channels will it use?
[12:43:02] <pcw_home> 0 through 3
[12:44:02] <pcw_home> actually the hardware can enable any set or subset of modules but the driver only allows enabling 0 .. N-1
[12:46:14] <cbjamo> I ran pnnconf to get an example file setup and then changed the firmware and modules, what else am going to need to set. I have
[13:00:38] <cbjamo> does anyone know what the error "can not find-sec MOT -var MOT -num 1 means?
[13:01:04] <andypugh> It means nothing, ignore it.
[13:01:44] <andypugh> Somebody probably ought to do something about those messages.
[13:02:16] <cbjamo> how about "-sec IO var IO -num 1" or "-sec LINUXCNC -car NML_FILE -num 1" or "
[13:02:23] <andypugh> The same.
[13:02:31] <andypugh> There are 4 of them I think
[13:02:51] <cbjamo> hmm, yeah, one is the same as the LINUXCNC one, except is says EMC
[13:03:18] <cbjamo> anyway, they are the only errors I'm getting, and linuxcnc is crashing.
[13:04:36] <cbjamo> there is more in dmesg, but no error messages, just status info, as far as I can see.
[13:04:39] <andypugh> Crashing or refusing to run? There is a difference. OK, it amounts to the same thing from your perspective.
[13:04:53] <andypugh> put the dmesg at www.pastebin.com
[13:05:27] <cbjamo> Good point, it opens the config panel, I choose my config then it sits for a sec and closes.
[13:06:59] <pcw_home> do a 'sudo dmesg -c' before you post the log
[13:08:53] <pcw_home> (and before you launch linuxcnc)
[13:09:15] <andypugh> If he does that we miss the "no lapic" message..
[13:09:58] <pcw_home> AFAIK he had things running beyond this point before
[13:10:56] <cbjamo> yes and no, I had the sim running before, not a real setup.
[13:11:10] <cbjamo> I have both the trimmed dmesg and a full one
[13:11:40] <pcw_home> does the latency test run?
[13:11:53] <cbjamo> trimmed: pastebin.com/hvDfwqvX
[13:12:01] <cbjamo> yes, the latency test works fine
[13:12:44] <andypugh> In that case it isn't the lapic
[13:12:56] <cbjamo> Ok
[13:13:11] <pcw_home> didn't find bitfile
[13:13:12] <andypugh> [ 4467.431296] hm2/hm2_5i22.0: firmware hm2/5i22/svsttp6_6_12 not found
[13:13:13] <andypugh> [ 4467.431306] hm2/hm2_5i22.0: install the package containing the firmware.
[13:13:14] <andypugh> [ 4467.431313] hm2_5i22.0: board fails HM2 registration
[13:13:33] <andypugh> sudo apt-get install linuxcnc-firmware-all
[13:13:51] <cbjamo> this is a custom bitfile, it won't be in the dep
[13:14:02] <cbjamo> path must be off
[13:14:03] <pcw_home> you left off the .bit
[13:18:13] <IchGuckLive> andypugh: sudo apt-get install hostmot2-firmware-all B)
[13:18:43] <andypugh> Is it? Shows how long it is since I did it.
[13:18:55] <IchGuckLive> O.O
[13:19:14] <IchGuckLive> therefor i alsk every week we need a ppa fr this
[13:19:22] <IchGuckLive> as it is changing
[13:20:36] <IchGuckLive> im figting with the same problem o nthe 5i25 7i76 somme times on updating cnc mashines
[13:20:55] <IchGuckLive> as from 2.4 -> 2.5 - master 2.6.pre
[13:21:05] <IchGuckLive> getting lost of the hostmots
[13:21:47] <pcw_home> there should be no issue with the 5I25/7I76, the included firmware is fine
[13:22:10] <cbjamo> fixed that, and changed permissions, still doesn't work. new dmesg: www.pastebin.com/ZagLYs6P
[13:23:18] <pcw_home> well it did work, but you probably have an error in your hal file
[13:23:39] <cbjamo> very likely, how do I go about tracking them down?
[13:23:54] <andypugh> If you start linuxcnc from the command line you will probably see a "can't find pin" message
[13:24:09] <pcw_home> hal error are almost always reported with line number
[13:24:16] <pcw_home> errors
[13:26:36] <cbjamo> "Printer.hal:10: parameter or pin 'hm2_5I22.0.pwmgen.pwm_frequency' not found"
[13:27:06] <pcw_home> well thats true, there are no PWM gens
[13:27:51] <cbjamo> right, so I change it to hm2)5I22.0.3pwmgen....
[13:28:26] <pcw_home> And i notice this is a hacked 7I39 config with stepgen outputs on the sense lines, that was a mistake, but it should be harmless
[13:29:42] <pcw_home> dont know if pncconf knows anything about 3 phase pwmgens
[13:29:49] <cbjamo> Yeah, I thought the stepgens up there were a bit odd, but figured it was related to the 7i39 board in some way.
[13:30:39] <cbjamo> well, what do I need to do then?
[13:30:47] <pcw_home> No, its just that I copy,pasted from a custom config, I will fix it
[13:31:22] <cbjamo> ok
[13:31:41] <pcw_home> if pncconf doesn't know how to make a 3 phase config you will need to cobble it together yourself
[13:32:15] <pcw_home> (the stepgens will cause no harm unless you want to use the 7I39 sense pins as inputs)
[13:32:35] <cbjamo> Ok, will I be better off starting where I am or from scrach?
[13:33:08] <cbjamo> My motors have hall feedback, and I want to use it.
[13:33:40] <pcw_home> if you have a servo config, its a starting point. maybe andy has some sample 3pwmgen boilerplate
[13:34:19] <cbjamo> that would be great
[13:36:32] <pcw_home> also man hostmot2 for signal names
[13:36:33] <pcw_home> (or list pins and parameters with halcmd)
[13:36:37] <andypugh> http://pastebin.com/Ac94d9eY
[13:37:48] <andypugh> (Which seems to lack the rather important "loadrt bldc" and "addf bldc" bits...
[13:43:31] <Jooa> Hi! Is there way to jog two axis, at the same time, with just one keyboard shortcut?
[13:44:18] <cradek> you can push both jog keys
[13:45:54] <Jooa> but i want them to jog same amount
[13:46:24] <cradek> what are you trying to do?
[13:48:34] <cbjamo> andypugh: what thread to I want the blcd function to be on?
[13:48:53] <andypugh> servo
[13:48:58] <cbjamo> thanks
[13:52:29] <andypugh> You probably want h,h,h for three motors with hall sensors
[13:54:16] <cbjamo> I actually have 6, not 3. And it already has 3
[13:54:31] <cbjamo> two per axis
[13:56:21] <cbjamo> and in the addf area it has "addf bldc.N servo-thread"
[13:56:53] <cbjamo> am I going to want 0-5?
[13:57:20] <cbjamo> and am I going to need a pid for each one?
[13:57:31] <pcw_home> yep
[13:58:30] <pcw_home> loadrt pid num_chan=6
[13:58:46] <cbjamo> is the N in "addf bldc.N" just a lable? Can I name them x1, x2, etc.
[14:00:08] <cbjamo> this config has "loadrt pid names=pid.x,pid.y,pid.z"
[14:00:36] <pcw_home> ok so you can extend that with 3 more names
[14:00:42] <cbjamo> ok
[14:00:48] <cbjamo> and the name can be anything?
[14:01:21] <pcw_home> Yes
[14:02:07] <andypugh> You can probably name the bldc components. But perhaps leave "blcd" in the name so you don't mistake them for pids..
[14:02:55] <cbjamo> ok, will bldc.x1 and suchlike work?
[14:03:08] <Jooa> cradek: i have a xyuv hotwire cutter and sometimes it would be handy to fix x-u and y-v movements
[14:04:19] <andypugh> cbjamo: Should do. You still need to declare the config types.
[14:04:30] <cbjamo> ok
[14:04:46] <cbjamo> so it'll be
[14:05:40] <cbjamo> "loadrt blcd names=bldc.x1,blcd.x2, etc cfg=h,h,h,h,h,h"
[14:08:29] <andypugh> Yes, that ought to work.
[14:08:58] <cbjamo> and I need to double everything up in the axis configs, yes?
[15:15:48] <Matous> Why is everyone so silent?
[15:17:35] <cbjamo> What is the parameter bldc.N.pattern?
[15:19:52] <Matous> does this help?
http://www.linuxcnc.org/docs/html/man/man9/bldc.9.html
[15:20:29] <cbjamo> yup, thanks
[15:31:02] <cbjamo> what happens when you put a net command in a hal file without a <=, => or <=>?
[15:32:42] <JT-Shop> cbjamo:
http://www.linuxcnc.org/docs/html/hal/basic_hal.html
[15:32:49] <micges> cbjamo: these are only for better reading
[15:34:06] <cbjamo> I see, thanks
[15:40:29] <Matous> Has anyone got any good ideas on how to probe for tool length using one of the input switches?
[15:44:21] <archivist> there is an example on your hard disk
http://www.linuxcnc.org/docs/2.4/html/examples_gcode.html
[15:55:09] <Matous> aah thanks a lot! I thought there has to be a simpler solution than the ones I tied... Guess searched for bad keywords or something.
[15:59:25] <andypugh> I just collected my new lathe. It's lovely. In fact I only bought it because it is lovely, I have absolutely no idea what I will do with it.
[15:59:41] <andypugh> It isn't really CNC-convertible.
[16:00:43] <ds3> what kind of lathe is it?
[16:02:19] <andypugh> http://www.lathes.co.uk/rivett/page2.html
[16:03:19] <ds3> u got the armstrong tool holder still on it?
[16:03:37] <andypugh> If that's the cylindrical one, then yes.
[16:03:49] <ds3> the 'rocker' one
[16:04:07] <andypugh> No, it's the cylindrical adjustable-height one.
[16:04:57] <andypugh> I might have to make an oak cabinet to hide the CNC parts :-)
[16:06:25] <andypugh> Making an oak cabinet is not inconceivable, I made this:
https://picasaweb.google.com/lh/photo/RXISmcDuivLRVKO9-hnzqdMTjNZETYmyPJy0liipFm0?feat=directlink
[16:06:29] <archivist> you need a manual lathe for one off quickies
[16:06:33] <andypugh> Makes flood-coolant a bit tricky thought.
[16:07:25] <ds3> brush on oil coolant
[16:08:37] <andypugh> Right, first things first, I need to unload it from the car.
[16:08:57] <ds3> get out the gantry crane?
[16:09:02] <andypugh> Actually, zeroeth things first, I need to clear the bench so there is somewhere to put it down.
[16:09:18] <archivist> borrow a slave/volunteer
[16:10:14] <andypugh> It's full of stupid/cool features.
[16:10:31] <andypugh> Have a look at the compound/top slide. How does that work?
[16:10:36] <ds3> what's the size of yours? (swing/length)
[16:10:47] <fragalot_> archivist: or a cnc lathe with a 512 tool changer :P
[16:11:18] <andypugh> only 56 tools allowed in LinuxCNC
[16:11:26] <archivist> fragalot_, bug there is no programming time with manual
[16:11:27] <fragalot_> aw bum there goes that then
[16:11:29] <archivist> but
[16:11:44] <fragalot_> archivist: good CAM software should help there
[16:11:46] <archivist> that tool limit is dumb
[16:11:52] <andypugh> archivist: There is no programming time with mt CNC.
[16:12:12] <fragalot_> if you can afford a 512 tool cnc lathe, you can afford CAM software too
[16:12:12] <fragalot_> :P
[16:12:19] <archivist> my inside rear of skull CAM works a manual lathe just fine
[16:13:21] <fragalot_> unless you've got a turrent lathe, you still need to change tools & touch off again for the new tool though
[16:14:06] <andypugh> archivist: Have you seen
http://www.linuxcnc.org/index.php/english/forum/41-guis/26550-lathe-macros ?
[16:14:17] <andypugh> One screen for each operation. No programming.
[16:15:47] <cbjamo> any idea what this error is? "emc/task/taskintf.cc 614: Error on axis 0, command number 317"
[16:15:50] <ds3> andypugh: that sounds just like the Denford edu lathes
[16:16:03] <andypugh> Must be a Brighouse thing :-)
[16:16:03] <archivist> I saw the screen a few weeks ago, but thats all
[16:16:17] <fragalot_> andypugh: very neat, crashing the tool has never been easier :D
[16:16:24] <DJ9DJ> gn8
[16:17:11] <andypugh> (ds3 I am from the same small town as Denford)
[16:17:30] <ds3> andypugh: oohhh.. are there a lot of old denford iron around?
[16:17:34] <archivist> and I haz a Denford
[16:20:12] <archivist> at least one starturn is modernised
http://www.collection.archivist.info/archive/DJCPD/PD/2009/2009_09_10_Starturn_cnc_lathe/P1010038.JPG
[16:28:03] <cbjamo> andypugh: did you see my response in that other channel?
[16:29:50] <cbjamo> is there any way to run the open loop test from pnnconf without running pnnconf?
[16:35:49] <JT-Shop> open loop test?
[16:36:27] <cbjamo> open loop servo test
[16:48:41] <JT-Shop> how can you run a servo in an open loop?
[16:48:48] <fragalot_> JT-Shop: you run it as a dc motor
[16:49:00] <JT-Shop> ah ok
[16:49:00] <ssi> then it's not a servo :)
[16:49:02] <fragalot_> and it just goes 'round and 'round and 'round :P
[16:49:05] <JT-Shop> had me confused
[16:49:25] <cbjamo> yeah, not my name, thats what its called in pnnconf/
[16:50:13] <cbjamo> I have brushless motors, so I'm just trying to test the driver. I'll worry about feedback when I know I can make them spin.
[16:50:19] <JT-Shop> I would assume you have to have pncconf running to do anything like that
[16:50:33] <JT-Shop> 0-10v input?
[16:50:58] <cbjamo> mesa 7i39, so no
[16:51:48] <cbjamo> it uses pwm.
[16:56:15] <Tom_itx> 96° F Partly Cloudy
[16:56:49] <Tom_itx> pcw_home what temp are the mesa boards rated for?
[16:57:26] <Tom_itx> i think that may have contributed to my errors yesterday. either the pc or other electronics may have been too warm
[16:59:07] <JT-Shop> -C version is 0C to 70C
[16:59:38] <Tom_itx> it did a bit better when i took the atom pc out instead of the old clunker
[16:59:53] <Tom_itx> but i still got a few random errors
[17:00:14] <Tom_itx> ie, stopping in the middle of a toolpath and asking for the next tool
[17:04:33] <JT-Shop> yuck
[17:04:59] <Tom_itx> took all evening to get a file halfway run
[17:05:25] <JT-Shop> sounds like my VMC tripping the drive all the time
[17:05:35] <Tom_itx> i'll wait til it's cooler and try again or get an ac
[17:05:51] <ssi> http://www.prototechnical.com/~imcmahon/spi32bit-4.png
[17:05:52] <ssi> weeee
[17:05:57] <Tom_itx> yeah i figure mine is a heat issue
[17:06:07] <Tom_itx> it worked fine before
[17:46:35] <Aero-Tec> I want to call a sub
[17:46:45] <Aero-Tec> but my sub call is to long
[17:47:26] <Aero-Tec> the number of variables and there names are to long
[17:47:57] <JT-Shop> too long?
[17:48:29] <JT-Shop> gives you and error?
[17:48:35] <Aero-Tec> is there any sting function that I could assign the variable line to and then call?
[17:48:38] <Aero-Tec> yes
[17:48:41] <Aero-Tec> to long
[17:49:21] <JT-Shop> not that I know of
[17:49:22] <Aero-Tec> o<cut_circle_rotory_table> call [#<tooldia>][#<FeedRateAaxis>][#<inside>][#<rpm>][#<feedrate>][#<inc-cutDepth>][#<fCutDepth>][#<zsafe>][#<zstart>][#<z-man-auto>][#<cen-x>][#<cen-y>][#<DiaCircle>][#<RadCircle>][#<StartDeg>][#<NumberOfDeg>][#<StartX>][#<StartY>]
[17:50:38] <Aero-Tec> cutting a arc using a rotary table flat on the table
[17:51:25] <Aero-Tec> have to much back lash to just cut a circle using X and Y
[17:51:33] <JT-Shop> I wonder what the line lenght limit is?
[17:52:54] <JT-Shop> Aero-Tec:
http://www.linuxcnc.org/docs/html/gcode/overview.html#_format_of_a_line
[17:54:04] <JT-Shop> your 4 characters too many
[17:54:18] <Aero-Tec> lol
[17:56:24] <Tom_itx> abbreviate a couple vars
[17:56:29] <Tom_itx> or can you
[17:57:35] <Aero-Tec> yes I can
[17:58:11] <Aero-Tec> would be cool if there was a continue on next line
[17:58:27] <Tom_itx> there is for some languages
[17:58:33] <Tom_itx> dunno about subs
[17:58:50] <Aero-Tec> so one can add several 256 lines together to make one line
[18:00:03] <Aero-Tec> it could be used to make code more readable
[18:00:26] <Aero-Tec> o<cut_circle_rotory_table> call
[18:00:34] <Aero-Tec> [#<tooldia>]
[18:00:44] <Aero-Tec> [#<FeedRateAaxis>]
[18:00:54] <Aero-Tec> and so on like that
[18:01:17] <CaptHindsight> BBB's came in today, why would anyone still use the RPi?
[18:02:06] <Aero-Tec> one would need a symbol like ~ or | or something
[18:02:44] <jdh> pi is cheaper, has composite out
[18:02:50] <andypugh> Got distracted by the new toy. It really is the most beautifully over-made thing.
[18:02:53] <Aero-Tec> do the call and have a list of what is passed in a more readable format
[18:03:06] <jdh> did you buy the lathe?
[18:03:16] <andypugh> I didn't mean to.
[18:03:22] <jdh> heh, cool.
[18:04:11] <Tom_itx> did the devil make you do it?
[18:04:13] <jdh> my wife and kids are gone until saturday. My goal is to be able to get to my jet 9x20 I've never used.
[18:04:50] <andypugh> The tailstock has a dovetail to slide across for slight tapers. It really looks like they assembled it, then machined all over, then polished it.
[18:05:51] <andypugh> The guy I bought it from appears to be a loony. He had about 6 lathes and sundry other tools in his garage, all running from line-shafting.
[18:06:11] <andypugh> I don't think anyone needs 6 lathes in their garage.
[18:06:56] <jdh> how many do you have so far?
[18:07:18] <andypugh> Only two. I need a bigger garage.
[18:07:41] <jdh> I'd trade my 2.5 for one that sucked less!
[18:08:56] <Aero-Tec> if one is doing production I can see a use for multi lathes
[18:09:24] <andypugh> Lets see, he had a
http://www.lathes.co.uk/pittler/index.html and a
http://www.lathes.co.uk/boley/page3.html and a
http://www.lathes.co.uk/holbrookB8/ And those are just the ones I recognised.
[18:09:31] <jdh> or one could just be the lathe equivalent of 'the crazy cat lady down the road"
[18:09:52] <Aero-Tec> lol
[18:09:56] <Aero-Tec> could be
[18:10:07] <Aero-Tec> 2 lathes and 2 mills so far
[18:10:34] <Aero-Tec> 1 plasma cutter 4 welders
[18:10:57] <Aero-Tec> not counting the gas torch
[18:10:59] <ssi> I've got three lathes, 2 mills
[18:11:12] <ssi> cnc plasma table, surface grinder, 2 bandsaws, 2 welders (3 counting torch)
[18:11:35] <ssi> and two small homebuilt desktop mills
[18:11:39] <ssi> and four 3d printers
[18:11:40] <Aero-Tec> got a band saw
[18:12:05] <Aero-Tec> one cutoff saw
[18:12:16] <Aero-Tec> oops forgot the spot welder
[18:12:27] <jdh> http://charleston.craigslist.org/tls/3863415635.html
[18:12:54] <ssi> jdh: what kind of machine is that?
[18:13:12] <jdh> no clue
[18:13:17] <ssi> 5k for both?
[18:13:22] <jdh> looks like it
[18:13:24] <ssi> sounds like a reasonable deal if you can move them
[18:13:31] <jdh> and power them
[18:13:38] <ssi> secondary details :)
[18:13:42] <jdh> I can do neither (nor house them)
[18:13:44] <ssi> they're in a house though
[18:13:56] <ssi> I might call him
[18:14:04] <jdh> not to far
[18:14:07] <ssi> nope
[18:14:08] <CaptHindsight> heh, he purchased the house with the machines already in the workshop
[18:14:13] <ssi> I could put them in my hangar
[18:14:16] <ssi> I've got 3ph there
[18:15:00] <Aero-Tec> looks like 5K for both
[18:15:20] <Aero-Tec> getting them moved looks to be lots more money
[18:15:21] <ds3> ssi: u have a hanger?
[18:15:27] <ssi> I have two :P
[18:15:31] <ds3> where?
[18:15:32] <Aero-Tec> buying them is the cheap part
[18:15:37] <ssi> south of atlanta
[18:15:51] <ds3> small civ. aviation airports?
[18:15:55] <ssi> yes
[18:16:04] <ds3> nice
[18:16:09] <ds3> how expensive is a hanger per month?
[18:16:40] <Aero-Tec> can one buy a hanger?
[18:17:17] <ds3> hand some friends talk about renting hanger space for a shop
[18:17:34] <ssi> sure
[18:17:36] <ssi> ds3: varies with the airport
[18:17:39] <ssi> my airport, they'r about $300/mo
[18:17:45] <ssi> by "my airport", I mean the one seventy miles from me that I have hangars at
[18:17:55] <ds3> $300 include power?
[18:17:56] <ssi> because the one a mile from my house has a 12 year waiting list for hangars and they're $1200/mo
[18:18:00] <ssi> yep
[18:18:10] <ssi> they typically don't have enough power to run a shop in
[18:18:18] <ssi> but I happen to have the hangar on the end where the 3ph comes in
[18:18:23] <ds3> ohhhh
[18:18:33] <ds3> $300 per month sound sweet
[18:18:50] <CaptHindsight> 3 phase 220 or 221?
[18:18:53] <ssi> I pay $400 for that end hangar cause it's bigger
[18:18:57] <ssi> 3ph 220
[18:19:04] <jdh> whatever it takes
[18:19:13] <Aero-Tec> but if one runs a shop machine shop in it the price will go up
[18:19:15] <ds3> can you abuse power by welding stuff and running a small foundry?
[18:19:17] <ssi> Aero-Tec: yes
[18:19:29] <ssi> ds3: it's one of those things... it'll be fine til someone notices
[18:19:31] <ssi> then it won't be fine anymore
[18:19:36] <ds3> ah
[18:20:02] <Aero-Tec> back to work
[18:20:11] <Aero-Tec> rewrite the code
[18:29:57] <Aero-Tec> is #<td> and #<TD> the same thing in linux CNC?
[18:31:39] <Aero-Tec> thunder storm rolling in here, my power will often go off during a storm, so if I suddenly drop out you will know why
[18:33:33] <Aero-Tec> so are names case sensitive in linux CNC?
[18:34:10] <JT-Shop> variable names?
[18:34:30] <JT-Shop> or should I say named parameters?
[18:35:00] <JT-Shop> http://www.linuxcnc.org/docs/html/gcode/overview.html#sub:Named-Parameters
[18:37:00] <Aero-Tec> cool, thanks a mill
[18:37:14] <Aero-Tec> so not case sensitive
[18:37:45] <Aero-Tec> storm is really getting loud
[18:38:02] <ssi> hdl is hard
[18:38:07] <JT-Shop> no, they get converted to lower case
[18:38:09] <ssi> just letting you all know :(
[18:38:32] <Jymmm> JT-Shop: vars dude, vars
[18:39:39] <Aero-Tec> speaking of witch, what is the best way to protect your CNC and also to best setup to deal with power bumps and outages during a storm?
[18:40:00] <JT-Shop> I turn them off
[18:40:13] <Jymmm> unplug them
[18:40:31] <JT-Shop> and what Jymmm said
[18:40:32] <Aero-Tec> I was thinking a UPS for both the CPU and the servos/steppers drivers
[18:40:56] <Jymmm> Aero-Tec: A UPS will NOT protect you from a storm
[18:41:26] <Jymmm> Aero-Tec: An expensive power conditoner might.
[18:42:02] <Aero-Tec> we have been getting a large number of power bumps here lately
[18:42:13] <Jymmm> Aero-Tec: Most UPS do not deal with inductive laods very well.
[18:43:02] <Aero-Tec> I do not need to produce during a power outage, but would be cool to have the machine start up from where it left off in mid cut
[18:43:56] <Aero-Tec> so large transformers for the steppers and servos would not work well with a UPS?
[18:44:49] <Aero-Tec> have a 1500 watt supply for mill and a 3000 watt one for the lathe
[18:45:27] <Aero-Tec> for the steppers for the mill and servos for the lathe
[18:46:13] <Aero-Tec> do not care if spindle winds down as long as the feed rate follows
[18:46:35] <Aero-Tec> and same with it coming on
[18:46:58] <Aero-Tec> if I was doing a heavy cut if maybe a problem on startup
[18:47:31] <Aero-Tec> soft start may fix that
[18:47:39] <Aero-Tec> any ideas?
[18:48:45] <Aero-Tec> one of the things I love about linux CNC is how it will stop mid cut when spindle is turned off and start up again with the spindle turned on
[18:48:55] <Aero-Tec> Mach would not do that
[18:49:30] <Aero-Tec> it took a read of spindle start of cut and that was assumed to be the speed for the full cut
[18:49:58] <Aero-Tec> sucked
[18:50:44] <JT-Shop> well guys see you all next week
[18:51:03] <Aero-Tec> have fun
[18:51:09] <JT-Shop> thanks
[18:51:30] <Jymmm> Aero-Tec: You will have to check your specific UPS to be sure. But no consumer grade UPS does well with inductive loads that I've ever seen. Industrial UPS is a different story.
[18:51:39] <Jymmm> JT-Shop: You slutting off to Ohio?
[18:51:50] <JT-Shop> Maggie Valley NC
[18:52:10] <Jymmm> JT-Shop: Same diff, just be glad you're not going to Misery
[18:52:27] <Jymmm> ;)
[18:52:30] <JT-Shop> nope not going to the left coast
[18:52:39] <Jymmm> JT-Shop: Have Fun!!!
[18:53:05] <JT-Shop> we will try our best to have as much fun as two old farts can
[18:53:41] <Jymmm> JT-Shop: Could always get/find a PYT to spice things up a bit =)
[18:54:19] <Jymmm> JT-Shop: Then all three of you can throw a party =)
[18:57:09] * JT-Shop puts on his chef hat
[19:47:39] <andypugh> CNC metal-tickling…
http://www.youtube.com/watch?v=3TRdvPtvLOY
[19:47:40] <TombCat> (3TRdvPtvLOY) "AutoTurn Lathe Controller Turns a Ball Handle" by "John Demme" is "Tech" - Length: 0:03:47
[19:56:25] <Tom_itx> he may have nicked the live spindle around 2:13
[20:18:31] <pcw_home> Tom_itx: normal Mesa commercial temperature range is 0 to 70C, -I versions are -40 to +85C
[20:18:32] <pcw_home> A 7I43 should be fine at even 90C
[20:19:34] <Tom_itx> i suspect the pc got a bit warm. i had fans on all of it but still...
[20:20:26] <pcw_home> PCs usually have narrower operating ranges
[20:21:46] <pcw_home> (mainly because of hot chips but also because they tend to be pushed closer to the edge for competitive reasons)
[20:24:02] <ssi> pcw_home: hdl is hard :(
[20:24:09] <pcw_home> (looking at the Atmel A5 ARM chips, pretty nice, 536 MHz, floating point and 200 mW full blast)
[20:24:23] <pcw_home> hdl is not really software...
[20:24:32] <ssi> yeah I know
[20:24:35] <ssi> and it's not what I'm used to :)
[20:27:11] <pcw_home> its just a strong hint to the compiler of what hardware you want
[20:27:25] <ssi> I'm building this spi slave in verilog, and the example I'm working from uses three bit shift registers to detect edges of SCK so it can do all its work in the fpga's clock domain
[20:27:55] <ssi> and it works fine on the transmit side, but I'm having trouble with the receive side because the falling SCK edge signal is delayed by over a full SCK period
[20:28:09] <ssi> sorry by transmit I mean on the MOSI side (slave rx)
[20:28:47] <pcw_home> pipeline the data as well
[20:29:13] <ssi> well my concern is because I believe the master is gonna clock in data from the slave on the sck edges...
[20:29:41] <ssi> so like, if I'm transmitting a 32 bit word, and my MISO side is running a full sck period behind, I'm only gonna get the 31 top bits back
[20:29:50] <ssi> because when it gets to the LSB, there's no SCK edges left
[20:29:54] <ssi> does that make sense?
[20:30:35] <Tom_itx> slave usually lags master doesn't it?
[20:30:41] <ssi> I dunno, does it?
[20:30:45] <ssi> maybe that's how it's supposed to work
[20:30:46] <Tom_itx> don't you usually have to send a dummy byte to flush it
[20:30:59] <pcw_home> if you are crossing clock domains you will have to oversample by some reasonable amount
[20:31:05] <ssi> I'm oversampling 2x
[20:31:08] <ssi> maybe that's not enough
[20:31:22] <ssi> well in my testbench anyway
[20:31:36] <ssi> I'm not running on real hardware yet, I just set up a testbench to fake the spi signals to the best of my knowledge
[20:31:46] <ssi> and I'm running that one with 50mhz fpga clock and 25mhz sck
[20:31:52] <ssi> which is faster than I'll actually run on real hardware
[20:31:57] <ssi> maybe I should slow it down to 10mhz and see how that goes
[20:32:06] <ssi> 10mhz is closer to what I will likely run
[20:32:24] <pcw_home> 2-1 will be a lot harder
[20:32:50] <pcw_home> bu there no reason not to run your HSclock at 100 MHz or so
[20:33:23] <ssi> the clock I have on the fpga board is 50mhz
[20:33:46] <pcw_home> But you have DCMs
[20:33:51] <ssi> true
[20:33:54] <ssi> I don't know much about it :)
[20:34:02] <ssi> I'm fully on the path of least resistance right now
[20:34:02] <ssi> heheh
[20:34:13] <pcw_home> 5i25 code has 33Mhz, 100 MHz and 200 MHz clocks
[20:34:35] <pcw_home> 33 for PCI, 100 for processors and 200 for PWM
[20:34:53] <ssi> 33mhz comes from the pci bus, and then you have what, a 100MHz oscillator onboard?
[20:35:04] <pcw_home> 50 MHz
[20:35:06] <ssi> gotcha
[20:36:03] <ssi> well I'm still in simulation, so practically speaking I can have whatever clock I want :)
[20:36:04] <eric_unterhausen> pcw_home, is there a future for external pcie?
[20:36:08] <pcw_home> 100 mhz is awkward for CMOS singe ended OSCs
[20:36:52] <pcw_home> Sure, for example the imx6 has gen2 pcie host or endpoint
[20:39:02] <pcw_home> so if you want something with > gig Ethernet throughput and ~2 usec latency to an external box...
[20:39:32] <ssi> I'm pretty sure there's no lag in spi, at least at the word level
[20:39:50] <ssi> when sck rises, slave clocks in MOSI, and master clocks in MISO
[20:39:51] <ssi> at the same time
[20:39:58] <pcw_home> External PCIE is also subsumed in Thunderbolt
[20:39:59] <ssi> both MSBs at the same time
[20:40:22] <ssi> pcw_home: oversampling at a higher rate seems to help though
[20:40:25] <ssi> 4x isn't quite enough
[20:40:26] <ssi> but close
[20:40:56] <eric_unterhausen> I look at the freefall of the desktop and wonder what my next mesanet purchase is going to entail
[20:41:25] <ssi> 5x seems to do the trick
[20:41:35] <ssi> I'm running 100MHz primary clock and 20MHz SCK
[20:42:38] <Valen> eric_unterhausen: with any luck it'll be an arm core with built in FPGA ;->
[20:43:27] <pcw_home> Not so sure the desktop is going away as fast as some think.
[20:43:28] <pcw_home> Some poobah said that they are more a victim of Moores law
[20:43:30] <pcw_home> (what people have is good enough)
[20:44:07] <Valen> thats what I reckon pcw_home
[20:44:11] <eric_unterhausen> I agree with the moore's law said we should be able to buy a better comp than we can
[20:44:23] <eric_unterhausen> wow, that was garbled :)
[20:44:25] <Valen> computers are fast enough now that they wait for us more than we wait for them
[20:44:37] <eric_unterhausen> I wait for my comp plenty
[20:44:43] <eric_unterhausen> you guys aren't working hard enough
[20:44:47] <Valen> computing is becoming a commodity, like water or power
[20:45:09] <eric_unterhausen> of course, mathworks
[20:45:23] <RyanS> Although surely you can't die a lack of computing?
[20:45:30] <ssi> try it some time :)
[20:45:32] <RyanS> Think I could ...
[20:45:46] <RyanS> Can you die of mental distress?
[20:45:50] <Valen> realisticaly, in an office any desktop has enough power to run the entire office
[20:45:57] <eric_unterhausen> I think that power management is making all computers run at the same speed
[20:46:26] <eric_unterhausen> if you buy a really powerful computer, they throttle it back because it's using too much power
[20:47:05] <ssi> eric_unterhausen: I haven't had that problem :P
[20:47:46] <RyanS> I reckon the stock market is using a single 386 for everything... That's why it keeps crashing :P
[20:48:30] <eric_unterhausen> people tell me their computer is too slow fairly often. I usually find some program causing it
[20:48:47] <eric_unterhausen> one person had a program that kept them from using facebook more than 2 hours a day
[20:48:59] <RyanS> lol
[20:49:03] <eric_unterhausen> it was using multiple cores at 100%
[20:49:20] <pcw_home> doing what?
[20:49:28] <eric_unterhausen> checking facebook, apparently
[20:49:45] <ssi> there we go
[20:49:51] <ssi> pcw_home: thanks, I'm in better shape now :D
[20:50:15] <RyanS> Perhaps it is lazy programming, too much processing powe and no incentive to economise ?
[20:50:32] <Valen> java will fix having too much computer power
[20:50:35] <eric_unterhausen> hp printer drivers will use a core at 100% if you have a windows firewall setting blocking the port that they use
[20:50:36] <Valen> as does "the cloud"
[20:51:06] <pcw_home> Thats lovely
[20:51:06] <eric_unterhausen> matlab is an interpreter written in java
[20:51:09] <Valen> "my computer is slow" == "the internet is fuxed"
[20:51:42] <eric_unterhausen> it will reduce any computer you have to its knees
[20:51:59] <Valen> and generally "my computer is slow" == "my browser have 40000 toolbars"
[20:52:02] <ssi> ttps://github.com/panicsteve/cloud-to-butt
[20:52:07] <ssi> https://github.com/panicsteve/cloud-to-butt even
[20:52:14] <RyanS> urghh the 'cloud'. What was wrong with "wide area network" it sufficed for a long time
[20:52:28] <Valen> that'd be a neat addition to firefox, remove the ability to add toolbars and "helpers"
[20:52:49] <Valen> firefox does have a neat "restore to factory settings" button these days
[20:53:14] <eric_unterhausen> that's pretty cool. I have greasemonkey installed but it scares me a little
[20:53:26] <Valen> creates a new profile, copies over bookmarks and passwords and wipes everything else
[20:53:47] <RyanS> Should I be expecting Firefox to utilise 2gb ram if I constantly have 7 tabs open
[20:53:58] <Valen> probably ;->
[20:54:36] <RyanS> Eventually it crashes isn't that what they call a memory leak....
[20:54:43] <Valen> could be
[20:54:52] <Valen> caching stuff in ram is common these days
[21:01:50] <RyanS> I was trying to find some information about that laser and camera measurement technique you were talking about the other day. Not sure what to google, umm 'diy noncontact measurement perhaps'?
[21:02:15] <PetefromTn> Evening folks..
[21:03:25] <Valen> i dunno, I was just going to do it lol
[21:04:16] <RyanS> You must have read about it somewhere :)
[21:04:42] <RyanS> Maybe it comes naturally to techies...
[21:05:37] <Valen> i don't recall reading about it
[21:05:43] <Valen> it just seemed obvious lol
[21:06:04] <Valen> what is a first principle way of measuring straightness
[21:06:10] <Valen> well light is straight
[21:06:21] <Valen> how can I use that to make something straight
[21:06:47] <Valen> fire a laser at a bare CCD sensor
[21:07:11] <RyanS> yeah but don't you need some serious precision optical hardware to do that?
[21:07:18] <Valen> then i had a look around the interwebs at commercial way servicing companies, and they seem to do it in a similar way
[21:07:47] <Valen> shouldn't do, CCD is going to be a few mm to a side (at most), put a dot in the middle of said CCD
[21:07:54] <Valen> (say a 1.3mpix CCD)
[21:08:26] <Valen> then find the centre of the dot, so you are going to have a couple of thousand "edges" to work from
[21:09:07] <Valen> in that few mm you are going to get 1280x1024 pixels
[21:09:32] <Valen> say the sensor is 5mm, each pixel is .004mm in size
[21:11:55] <RyanS> right so you bring up the image on-screen and find that the dot is located at 640, 512, well you've hit the centre
[21:13:56] <Valen> the thing is you can get better resolution than that because you don't just use the one pixel, you use a few hundred pixels, you find the centre of a big dot
[21:14:11] <Valen> so all those edge pixels "add" precision if you will
[21:15:11] <Valen> Getting a defined dot like that would be quite challenging, so odds are it'll be averaging some sort of noisy image probably with a bit of speckle, unless you do use some decent optics on the laser output.
[21:16:23] <jdh> measure twice before ordering!
[21:16:49] <Jymmm> As always.... Cut twice, measure once =)
[21:18:26] <RyanS> Okay so then let's say you want to get a linear rail straight you put the laser at one and direct towards the ccd, then if the dot is not visible in the image you adjust the rail until you have whatever image indicates you have the correct alignment of the rail?
[21:18:29] <jdh> I'm making a boat ladder. I way oversized the aluminum needed. I think I way oversized the tubing
[21:19:48] <RyanS> this is where someone says "measuring is for sissys"?
[21:20:15] <Valen> jdh: you are just being generous with your engineering margins
[21:20:32] <Valen> my process was to put the laser at the far end and get it centered
[21:20:39] <Valen> then do the same at the near end
[21:20:45] <Valen> then work along it
[21:20:57] <Valen> move the laser so near and far ends line up
[21:22:48] <RyanS> so you are using an epoxy granite bed, with commercial linear rails?
[21:23:23] <Valen> for the "good" one
[21:23:31] <Valen> the "router" will be on steel box section
[21:26:09] <jdh> Valen: more than generous, I dont' think I'll be able to lift it out of the water
[21:26:35] <RyanS> so your measurement instrument sits on the bed, centre the laser and then use that to intersect the rail with the laser beam to check parallelism of the rail relative to the bed?
[21:26:57] <Valen> lol
[21:27:44] <Valen> RyanS: laser generally goes off the side of the table, I was going to put the receiver on the carriage itself to get a reference
[21:29:50] <RyanS> laser beam perpendicular relative to the carriage? or parrallel
[21:30:02] <Valen> parallel to the rail
[21:30:06] <Valen> runs along next to it
[21:33:11] <Valen> http://www.measurementandmachine.com/html/linear.html is one way
[21:34:05] <RyanS> Okay so you traverse the carriage to one end hypothetically your dot is at pixel 200, but once you traverse it to the other end, you find the one.is at pixel 900 - okay you know it's a fair bit off. So this is synonymous with tramming a vice with a -dial test indicator?
[21:34:20] <Valen> i guess
[21:34:29] <Valen> so you nudge that end of the rail over
[21:34:46] <Valen> then move down the rail to each of its bolt down points and move them to be at pixel 200
[21:35:03] <RyanS> ah ah it sounds pretty simple in theory
[21:35:06] <ssi> I think it's about there!
[21:36:00] <RyanS> And how much better accuracy does this give you than using a micron dial test indicator perhaps?
[21:36:21] <Valen> you need to have a reference for that
[21:36:44] <RyanS> The edge of the bed?
[21:36:52] <Valen> and the bed is straight because?
[21:38:13] <Valen> the old school way of doing this kind of thing is a precision surface plate, and super fancy interferometic jobbies
[21:40:15] <RyanS> so it's not necessarily more accurate than that method it's just a pretty accurate DIY method?
[21:41:08] <Valen> I'd say without actually having done it, it should be more accurate as it operates from first principles.
[21:41:22] <Valen> its a DIY way of performing the "best" method of alignment
[21:42:01] <Valen> note though that the "good" ones also then do a laser interferometic measurement of the screw, to compensate for screw error
[21:42:04] <Valen> mine won't do that
[21:42:12] <TombCat> http://i.imgur.com/dhjbcBX.jpg
[21:42:38] <Valen> you need a good laser to get coherence over that distance and you also need to know the wavelength very precisley, also not a strong point of a diode laser
[21:43:12] <RyanS> I guess you could expect accuracy somewhere between the traditional surface plate method and laser alignment with expensive commercial equipment?
[21:43:46] <Valen> for the alignment I think it should be on par with commercial
[21:43:54] <Valen> I don't see them doing anything different
[21:44:11] <Valen> I just wont get the screw compensation
[21:46:18] <RyanS> hmm what sort of laser, solid-state?
[21:46:37] <Valen> I was just going to use a not crap laser pointer ;->
[21:46:42] <Valen> and a $5 webcam
[21:48:46] <frankt> I'm 4 hours into a 3D print using linuxcnc, and the machine has crawled to an almost complete stop. My job is still running, but it executes 1 line of gcode, then pauses for 2 seconds. Axis isn't showing the preview. The DRO reads all 0. I don't know what line I'm on to restart. Anyone seen this before? Can I recover?
[21:50:39] <Valen> that is odd
[21:50:50] <Valen> I'd suggest running top and seeing whats happening
[21:50:57] <Valen> it sounds something like CPU overheating
[21:51:55] <RyanS> http://www.ebay.com.au/itm/TUF-5-Pointer-Vector-Laser-/190833624405?pt=AU_Surveying&hash=item2c6e920555
[22:03:29] <RyanS> hmmm
http://www.lasercomponents.com/de-en/news/precision-lasers-for-the-alignment-of-machines/
[22:05:09] <jdh> how do you align the laser? or is that what you were discussing?
[22:06:40] <Valen> thats just talking about the alignment of the laser inside with the module housing, that doesn't matter for my setup
[22:06:44] <Valen> the light itself cant bend
[22:06:57] <Valen> well unless you have had *way* too much ice cream
[22:08:45] <toastyde1th> could also use an autocollimator but that's probably out of budget; you can get them used for 400-600
[22:09:08] <RyanS> So you want the oem laser module you can just as well stick it to anything?
[22:09:34] <toastyde1th> http://www.ebay.com/itm/Hilger-Watts-Autocollimator-with-Case-TAI-6-/271214791339?pt=LH_DefaultDomain_0&hash=item3f25a94aab
[22:09:36] <Valen> the laser just sits on a tripod on the floor
[22:09:43] <Valen> I was just going to use a laser pointer
[22:09:45] <Valen> $2
[22:10:55] <RyanS> That might not be 'not crap' :P
[22:11:15] <Valen> put it through a few bits of card to despekle it at least somewhat
[22:11:31] <RyanS> What the hell is that thing
[22:11:50] <RyanS> Looks like a gun sight
[22:12:29] <RyanS> Or a doctor who prop
[22:13:17] <toastyde1th> autocollimator, it's a very sensitive measure of angle used to check machine axes and surface plates
[22:16:46] <Tom_itx> Valen, take a couple pics of your setup when you do it
[22:19:01] <Valen> Tom_itx: don't hold your breath, I'm notoriously slow lol
[22:22:59] <toastyde1th> also rather than despeckle, you might frequency filter it
[22:23:05] <toastyde1th> since you'll have time data
[22:23:29] <toastyde1th> that way you won't be throwing out any potential displacement information
[22:25:27] <Valen> displacement is too hard to do with a diode laser
[22:25:37] <Valen> they arent coherent over very far
[22:25:38] <toastyde1th> i meant laterally
[22:25:50] <Valen> what do you mean?
[22:25:56] <toastyde1th> you're trying to align a way, right?
[22:26:02] <Valen> yes
[22:26:36] <toastyde1th> so the ccd is essentially showing you the small displacement on the plane normal to the beam
[22:26:45] <Valen> yes
[22:26:53] <toastyde1th> that's what i was referring to
[22:27:11] <Valen> I'm not seeing how a variation in frequency can cause the light to bend?
[22:27:36] <toastyde1th> no no, i meant frequency of the pixel oscillation
[22:28:21] <Valen> why would they be oscillating?
[22:28:24] <toastyde1th> my personal concern (if i were doing this) would be that any sort of filtering per frame woudl throw out actual information
[22:28:57] <eric_unterhausen> someone at work tried to order 20" of 80/20 and got 200 feet instead
[22:29:06] <Valen> bonus
[22:29:25] <toastyde1th> but if you look at how each pixel varies in time, any high frequency change can be eliminated
[22:29:48] <Valen> yeah, i figured I'd average a few thousand frames
[22:29:49] <toastyde1th> that way you preserve the very fine spatial information in the frames
[22:29:56] <eric_unterhausen> it came in a 20' long box, the machinist called the guy and asked what he wanted and he said 4x5" long pieces
[22:30:49] <toastyde1th> i think straight averaging would also throw information away because it will occlude the same information
[22:31:00] <Valen> what information?
[22:31:01] <eric_unterhausen> there is a lot of information about measurement of linear stages in the ASPE journal
[22:31:14] <Valen> I'm trying to find the centre of a dot
[22:31:17] <toastyde1th> any pixel luminosity that has a high correlation to the ccd's position
[22:31:35] <toastyde1th> i'd argue you're not, but finding the center of a dot is one way of attempting to find it
[22:32:09] <Valen> i'd argue its the simplest way, as anything outside the central "dot" isn't colimated and as such is going to be subject to change over distance
[22:32:20] <Valen> keeping in mind the distance being measured is around a meter and a half
[22:33:15] <toastyde1th> sure, i'm not disagreeing that it's simple, i'm just thinking out loud of other ways to possibly improve on what is probably going to be a nice method for machine alignment
[22:35:06] <toastyde1th> not necessarily finding the center of a dot, but finding out how the ccd is moving in space using any information available
[22:36:13] <toastyde1th> say i look at every pixel, and try to figure out how it changed compared to its neighbors and then average those
[22:36:29] <toastyde1th> i.e. the pixel to the left became brighter, to the right is dimmer
[22:36:57] <toastyde1th> come up with some metrization based on change in luminosity
[22:37:16] <Valen> when I say "dot" i'm implying its going to be several hundred pixels across
[22:37:19] <toastyde1th> yep
[22:37:35] <Valen> you can get "super accuracy" by measuing the pixels it partially crosses
[22:37:45] <toastyde1th> you can get a measurement, yes
[22:37:53] <Valen> but random noise outside that is just that random noise
[22:38:07] <Valen> have you played with a laser pointer and a card ?
[22:38:13] <toastyde1th> except you don't HAVE a dot, you have function that averages into a dot.
[22:38:31] <toastyde1th> my suspicion is that i can beat the accuracy of that average
[22:38:41] <Valen> good luck with that ;->
[22:38:58] <Valen> seriously, if you come up with something I'm all ears
[22:39:07] <toastyde1th> that's what i'm trying to describe, but i don't have a machine to play with
[22:39:46] <Valen> but for measuring position offset from centre, i don't believe there are any better options other than doing it multiple times
[22:40:07] <Valen> you can do fancypants wave front stuff to measure angle and the like
[22:40:09] <toastyde1th> if you improve the single-pass accuracy, you will improve the multi-pass accuracy
[22:40:16] <toastyde1th> i'm not talking about any interferometric effects
[22:40:32] <pcw_home> http://proceedings.spiedigitallibrary.org/proceeding.aspx?articleid=791948
[22:41:11] <toastyde1th> my method would not be center-estimating at all
[22:41:16] <RyanS> If you're doing a screen grab of one frame why would the next frame be any different assuming there is no movement
[22:41:44] <Valen> RyanS: there are several noise sources, particularly with cheap lasers
[22:41:45] <toastyde1th> RyanS, changes in air current change the refraction
[22:41:51] <pcw_home> Its a known problem with known solutions
[22:42:20] <Valen> over the distance of a meter in still air the effects of air will be pretty trivial
[22:42:22] <toastyde1th> that's not a response, that's just acknowledging that "this problem is a studied one"
[22:42:43] <Valen> and the only way around it is adaptive optics
[22:42:45] <toastyde1th> we don't come up with one way of doing things then go home
[22:43:14] <RyanS> We have several DSLRS at home full frame as well, that work better?
[22:43:21] <Valen> I'm just not seeing how you are going to use random dots on the CCD that have no correlation with position to help you determine position
[22:43:36] <Valen> RyanS: probably not actually, the smaller the sensor, the more dense the pixels
[22:43:50] <toastyde1th> differential measurement; start the laser at some point, call it zero.
[22:43:52] <Valen> they would help if you ran the laser through an optic and blew it up to the size of the sensor
[22:44:04] <RyanS> You want a smaller sensor?
[22:44:21] <Valen> the sensor should be about the size of the laser dot
[22:44:35] <Valen> so the dot fills up say 2/3rds of the sensor field
[22:45:14] <RyanS> But then you don't have as much space to move?
[22:45:23] <toastyde1th> run a high pass filter across the time-series data, to eliminate vibration
[22:45:36] <Valen> the "movement" is error
[22:46:04] <toastyde1th> then, instead of doing ANY averaging, compare each frame to its predecessor and successor
[22:46:29] <toastyde1th> look at the luminosity change and record its amount and direction.
[22:47:14] <Valen> if you do that you are loosing your central reference
[22:47:15] <toastyde1th> you now have a vector field; you can take the avearge of THIS and get the average movement in the plane
[22:47:24] <toastyde1th> why do you care about central reference?
[22:47:40] <Valen> because if you do it your way every frame you will introduce error
[22:47:58] <toastyde1th> what error?
[22:48:00] <Valen> you have no datum to measure from, you are stacking your measurement errors
[22:48:06] <RyanS> Okay if it takes up two thirds of the sensor and you are out of alignmen. dot may quickly end up out of frame? I don't understand why more resolution is a bad thing
[22:48:14] <toastyde1th> right, welcome to high precision measurement
[22:48:23] <toastyde1th> even way interferometers do it this way
[22:48:36] <Valen> you say "i moved .1mm +- .01mm to the right this frame"
[22:48:40] <toastyde1th> yep
[22:48:40] <Valen> then you repeat that every time
[22:48:43] <toastyde1th> yep
[22:48:47] <Valen> the +-.01mm adds up
[22:48:58] <toastyde1th> yep
[22:48:58] <Valen> so after 1000 measurements you are +- 10 mm
[22:49:12] <toastyde1th> except the numbers, in reality, are much tinier.
[22:49:15] <Valen> which is pretty much useless
[22:49:26] <Valen> actually interferometry doesn't do that
[22:49:28] <Valen> by definition
[22:49:49] <Valen> you have a coherent laser beam which has X wavelengths per meter
[22:49:49] <toastyde1th> i don't know what kind of alignment interfereometer YOU have
[22:50:00] <toastyde1th> but we aren't talking about distance from the origin
[22:50:14] <toastyde1th> and peak counting doesn't work here
[22:50:37] <Valen> you then *count* the number of wavelengths you pass, and your error is as a set % of the wavelength
[22:50:38] <toastyde1th> also, differential levels AND autocollimators work this way
[22:50:49] <Valen> if you are talking about angle measurement the same thing applies
[22:51:16] <Valen> they are "counting" the number of wavelengths difference from the centre point to their measurement edge
[22:51:30] <Valen> not the absolute number over the length sure
[22:51:52] <Valen> but the number from the centre of the target spot to the edge, thats how you get the newtons rings
[22:52:19] <toastyde1th> i think you are confusing how that data is used, sorry.
[22:52:29] <RyanS> .... this is why I am an arts student
[22:53:32] <toastyde1th> and, insofar as I am capable of understanding, my method of measurement is consistent with way measurement insofar as i have had limited practice, and seen it practiced professionally
[22:53:53] <Valen> have you seen a laser way alignment?
[22:53:59] <toastyde1th> i've seen several types of it
[22:54:04] <toastyde1th> not just one
[22:55:14] <RyanS> If you have two rails how are you going to check for parallelism between the two rails not just rail to bed
[22:55:16] <Valen> my contention is that random noise and other crap will mean measuring drift rather than an absolute position will lead to much greater error over the course of a measurement
[22:55:37] <Valen> RyanS: the bed, is basically irrelevant, the rails are what needs to be right'
[22:55:38] <toastyde1th> and the most accurate laser alignment tools use differential measurements and the error stacks as you go down the way. there ARE, and always have been, absolute way measurement methods, and they are not as accurate as the differential methods for distances under about 10 meters
[22:57:17] <toastyde1th> although, that does give me a futher idea, if you did a similar thing but compared it to the baseline image
[22:57:31] <Valen> my understanding was the "most accurate" was interferometry?
[22:57:37] <RyanS> do they anticipate that the error is consistent or random? If it's consistent it is probably easier to compensate for
[22:57:50] <toastyde1th> there's a whole cloud of methods that constitute interferometry when it comes to checking machines
[22:58:14] <Valen> yes, but they all come down to counting wavelengths or fractions thereof
[22:58:27] <toastyde1th> yep, the difference is what those measurements represent.
[22:58:27] <Valen> by definition
[22:58:37] <Valen> right, those are all absolute measurements
[22:58:56] <toastyde1th> no, they're really not.
[22:59:22] <Valen> how do you have a non absolute interfrometic measurement?
[22:59:27] <toastyde1th> so, one popular method is to stick a prisim on the axis that responds to tilt.
[22:59:36] <toastyde1th> I don't remember the details of the mirror/prisim configuration
[23:00:11] <toastyde1th> as you move the way, you no change in peak count as the thing moves down the way.
[23:00:34] <toastyde1th> as the mirror tilts up and down, you start to get a count.
[23:00:45] <toastyde1th> (or left/right)
[23:00:46] <Valen> right, you are getting an absolute measure of the tilt
[23:00:49] <toastyde1th> yep
[23:01:17] <toastyde1th> and when you use that data, you say, "ah, since the machine moved 5 inches, I have X degrees of tilt over this span"
[23:01:21] <RyanS> So you stick the camera next to the laser Toyota and the mirror on the carriage?
[23:01:22] <Valen> now there are several differences between that and what you were proposing
[23:01:22] <toastyde1th> therefore the error is Y
[23:01:34] <toastyde1th> i UNDERSTAND this, valen
[23:01:44] <toastyde1th> "this" being your objection
[23:01:52] <toastyde1th> not interferometry as a whole
[23:02:13] <Valen> firstly that is taking an absolute measure of the "tilt" at the point in the movement, which is fine, and your error is going to be expressed as a fraction of a wavelength of light
[23:02:18] <Valen> which is verry friggin small
[23:02:31] <RyanS> Laser pointer (I use voice dictation it thought I said Toyota)
[23:02:45] <toastyde1th> okay dude, I'm going to just back out of this because you're explaining things to me that I already understand
[23:02:55] <Valen> now in the laser shining at a camera scenario your error is measured in pixels, which are *way* bigger
[23:02:55] <toastyde1th> so i'm not expressing myself clearly enough to continue
[23:03:22] <toastyde1th> and that you say it's expressed in pixels shows that you don't understand the method i proposed, because we ain't counting pixels
[23:03:46] <toastyde1th> with sufficient resolution, you could do this with TWO pixels
[23:03:53] <toastyde1th> resolution in luminosity, that is
[23:03:58] <Valen> right
[23:04:03] <Valen> so its pixels * resolution
[23:04:05] <toastyde1th> but i'm done, sorry.
[23:04:11] <Valen> which is best case 256 levels
[23:04:33] <Valen> end result is your error is going to be much larger than a fraction of 640nm
[23:05:00] <Valen> and the biggest difference, is the interferometric measurements don't stack up
[23:05:07] <Valen> they are measured to a constant reference
[23:05:09] <pcw_home> One paper shows about 1/20 pixel uncertainty using centroid calculation, and about 1/2 frame spot size
[23:05:32] <pcw_home> diode laser
[23:05:33] <Valen> your method stacks error terms
[23:05:40] <Valen> thats pretty good pcw_home
[23:05:50] <RyanS> Hasn't anyone done this on cnc zone?
[23:05:58] <pcw_home> Yeah I thought so (100 frame averaging)
[23:10:04] <toastyde1th> Valen, i have to clarify this because it's bugging the shit out of me: tilt of the interferometer is absolute, but that isn't what anyone cares about
[23:10:14] <Valen> i get that
[23:10:30] <RyanS> https://en.wikipedia.org/wiki/Interferometry which category should I read? :P
[23:10:36] <toastyde1th> because aligning a laser is not easy
[23:10:51] <toastyde1th> and using a differential measure removes THAT error
[23:11:00] <Valen> not for interferometry at least lol ;->
[23:11:37] <toastyde1th> i.e. take the cumulative error term, because the absolute error term can't be quantified as easily
[23:11:59] <Valen> the point i was making, is that the *measurements* you make with your tilt system don't have stacked error (presuming you don't loose count)
[23:12:10] <Valen> unless you don't count as you traverse?
[23:12:12] <toastyde1th> yep, but nobody CARES about that
[23:12:20] <toastyde1th> that's not what's being actually measured
[23:12:43] <Valen> well even given that, the measurement of tilt at any point does not depend on the measurement at a previous point
[23:13:04] <Valen> the offset you calculate from those measurements does
[23:13:05] <toastyde1th> yep, but you're still in the realm of measurements that don't tell us anything
[23:13:12] <toastyde1th> right!
[23:13:16] <toastyde1th> which is what we are actually measuring
[23:13:25] <toastyde1th> which is cumulative.
[23:14:06] <Valen> which is a flaw in the system
[23:14:15] <toastyde1th> right!
[23:14:28] <toastyde1th> because the absolute error is vast and cannot be quantified
[23:14:42] <Valen> which is not the case for a CCD system
[23:14:54] <Valen> because you can directly measure the position at each point
[23:15:00] <toastyde1th> "Is that .005 per foot an error in the laser's alignment, or is it misalignment of the way?"
[23:15:10] <Valen> the laser is perfect
[23:15:14] <toastyde1th> nope, because you're measuring your laser's alignment with the machine.
[23:15:15] <Valen> it is by definition
[23:15:21] <toastyde1th> the LASER is perfectly straight.
[23:15:25] <Valen> yes, you must align the laser to the machine
[23:15:26] <toastyde1th> where you've pointed it is arbitrary.
[23:15:36] <Valen> or know the offset
[23:15:44] <toastyde1th> and now you have a massive error term because you've got no idea where the laser is compared to the machine.
[23:15:50] <Valen> actually i do
[23:16:04] <Valen> i know exactly where it is
[23:16:12] <toastyde1th> okay, say you've aligned your first axis.
[23:16:19] <toastyde1th> It's perfectly straight along your laser.
[23:16:20] <toastyde1th> Align the second axis.
[23:16:30] <Valen> ahh thats where you get tricky ;->
[23:16:52] <toastyde1th> Bingo.
[23:17:20] <Valen> i did have that solved, now I cant remember what it was
[23:18:12] <Valen> first up there are 2 things to considder
[23:18:30] <Valen> firstly if the axis is straight, which we can solve with the same laser system
[23:18:46] <Valen> the second is either getting it perpendicular to the first or at least measuring its offset
[23:19:43] <Valen> I *believe* i was just going to fire mah lazor into a precision mirror and depend on that as its cheap and good enough for me
[23:21:12] <toastyde1th> or maybe get a 2-4-6 block and machine a housing for your laser pointer
[23:21:19] <toastyde1th> and put the laser on the table instead of the ccd
[23:21:44] <Valen> well what you are trying to measure is an angle
[23:21:50] <RyanS> That's what I don't get why doesn't it matter if the rails are not parallel to the bed only to each other on the Y, what about X and Z what is their relationship to the bed?
[23:21:59] <toastyde1th> who cares about the bed
[23:22:05] <toastyde1th> you can machine that on the tool if it REALLY matters
[23:22:07] <Valen> the first principles method of that is a collimator
[23:22:25] <Valen> or some other sort of newtons rings arrangement
[23:22:26] <toastyde1th> also the bed's a good reference, but it's not the best - soft jaws cut ON the machine tend to be the most accurate
[23:23:03] <toastyde1th> soft jaws with clamping force set by torque wrench
[23:23:04] <Valen> the thing is RyanS you clamp the job to the bed somehow, and whenever you do that its going to clamp up in the wrong spot
[23:23:36] <Valen> once its all clamped up, you are then able to move the tool accurately around the job
[23:23:42] <Valen> until you release the clamp
[23:23:52] <Valen> at which point you are screwed, basically
[23:24:22] <toastyde1th> the only way i've ever found to eliminate springback is to not really clamp the part
[23:24:38] <toastyde1th> even on a mag chuck you have to shim the part up or it will spring
[23:25:30] <RyanS> hmm, that's what I mean isn't it vitally important that the axis rails are parallel to the bed as well as to each other?
[23:25:31] <toastyde1th> on a good import HMC we were able to get flatness under .0001" over 15" that way
[23:25:51] <toastyde1th> not much under, but under
[23:26:06] <toastyde1th> ryan, no
[23:26:09] <toastyde1th> the bed's irrelevant
[23:26:27] <RyanS> The base of the machine?
[23:26:34] <Valen> the one we wanted to make wasn't going to have a "bed
[23:26:45] <Valen> just a pair of stacked rotary tables
[23:26:46] <toastyde1th> put your vice on the bed, make two aluminum blocks that bolt to the jaw bolts of your vice
[23:26:59] <ssi> I'm curious about these things
[23:26:59] <ssi> http://www.igus.com/wpck/default.aspx?Pagename=drylin_zlw&C=US&L=en
[23:27:09] <ssi> I emailed my sales rep about prices earlier
[23:27:11] <toastyde1th> put a thin-ish block of aluminum in the vice near the top of the jaws
[23:27:22] <ssi> might be good for a little laser machine
[23:27:25] <toastyde1th> clamp the vice down, cut a lip into the block.
[23:27:31] <toastyde1th> release the pressure, reclamp as light as you can
[23:27:45] <toastyde1th> then skim the lip you just cut for like .005" off every surface
[23:27:47] <RyanS> Don't you need one 'master reference point' for the entire machine? ie the base/bed
[23:27:52] <toastyde1th> no?
[23:27:58] <toastyde1th> the ways tell the machine where to go
[23:27:59] <Valen> the master reference point is the ways
[23:28:19] <Valen> often you will machine the bed to make it true to the ways
[23:28:32] <Valen> but that only really helps for "low" accuracy stuff
[23:29:13] <RyanS> the X and Z axis will reference the y-axis rails?
[23:29:21] <Valen> yes
[23:29:44] <Valen> they all need to be perpendicular to each other
[23:30:17] <Valen> once you have that, you can then use the tool and cut the bed
[23:31:36] <toastyde1th> most of the time the bearing surfaces on the bed will be machined on via way grinder, and then you fixture the bed to a grinder and grind it
[23:31:41] <toastyde1th> (commercially)
[23:31:57] <toastyde1th> (if you use an aluminum/wood table it's obviously much simpler to do yourself)
[23:32:06] <toastyde1th> s/bed/table
[23:34:19] <RyanS> the columnn where it's bolted to the bed you could align by sight with a machinist square? And then if you use rails you have to shim/position bolt them with precise alignment using measurement instruments?
[23:34:30] <RyanS> I don't know why that is underlined
[23:34:50] <toastyde1th> the column is generally ground independently and then the mounting surfaces are scraped
[23:34:57] <toastyde1th> i.e. where it bolts to the rest of the machine
[23:35:32] <toastyde1th> that way you fit it up once, get your error and do a little trig, and figure out how you have to scrape down the mounting surface to make it normal to the table
[23:35:57] <toastyde1th> (same way headstocks are aligned on lathes, actually)
[23:36:00] <RyanS> You can't really scrape (or at least in a DIY setting its going to take years to practice)?
[23:36:11] <toastyde1th> nah, it's not as hard as people make it out to be
[23:36:22] <toastyde1th> it's just a) time consuming, and b) steep learning curve
[23:36:25] <Valen> the linear rails don't have scraping or grinding
[23:36:49] <toastyde1th> (also note i'm not talking about scraping any way surface, but a mounting gasket for the column)
[23:36:50] <Valen> different kettle of fish to cast iron ways
[23:37:02] <toastyde1th> (not a guiding surface, just an alignment surface)
[23:37:18] <Valen> anyway I'm off to do stuff
[23:37:22] <toastyde1th> doing something like that is MUCH easier than trying to scrape actual ways
[23:37:26] <toastyde1th> bai2u
[23:37:30] <RyanS> these good enough?
http://stores.ebay.com.au/CNC-AND-CUPCAKE-WORLD/Linear-Carriages-/_i.html?_fsub=601751919&_sid=657086089&_trksid=p4634.c0.m322
[23:37:44] <toastyde1th> linear ways have to be aligned too
[23:37:46] <Valen> heh thats what we are using ;->
[23:37:55] <toastyde1th> that's the thing we've been talking about
[23:38:01] <Valen> aligned, but no metal removal ;->
[23:38:05] <toastyde1th> ya
[23:38:12] <Valen> if you stuff it up its a screwdriver job to fix ;->
[23:38:31] <toastyde1th> although mfgs scrape or grind the machine base where the rails will go
[23:38:35] <RyanS> CNC and cupcake world?
[23:38:35] <toastyde1th> as though they were way surfaces
[23:38:41] <Valen> those are some decent prices for the rails
[23:38:54] <RyanS> Are you getting any cupcakes to go with that?
[23:39:17] <Valen> we were going to put an bed of water consistency epoxy down and let it self level to give a decent mounting surface
[23:39:45] <Valen> tossing up putting a hunk of ground flat bar on it then mounting the rails to that
[23:40:09] <RyanS> That's how they make enormous 'surface plates' for motor racing
[23:40:17] <RyanS> Put the entire car on its
[23:40:37] <Valen> with the right additives it can be "ok"
[23:41:22] <RyanS> Why don't you just get an okay granite surface plate?
[23:41:43] <Valen> because they cost a craptone and weigh more
[23:42:00] <Valen> also our router is 1.2x600 working area
[23:42:29] <RyanS> I was thinking for a minimill
[23:42:43] <Valen> still heavy and kinda fragile
[23:43:47] <RyanS> http://www.carbatec.com.au/granite-surface-plate_c20499
[23:44:05] <RyanS> Or maybe a cast-iron surface plate
[23:45:19] <RyanS> I think I would use cast iron because it is easy to drill and tap
[23:47:50] <RyanS> http://www.ebay.com.au/itm/Surface-Plate-Cast-Iron-/261228456404?pt=LH_DefaultDomain_15&hash=item3cd26de9d4
[23:48:05] <RyanS> in Sydney!
[23:58:54] <Xfriend> is possible to usr arduino with linuxcnc??
