#linuxcnc | Logs for 2012-11-23

[01:00:30] <Loetmichel> mornin'
[02:01:17] <DJ9DJ> moin
[06:58:15] <r00t4rd3d> http://i.imgur.com/DYfo4.jpg
[07:04:56] <mazafaka> r00t4rd3d: this way, Mother Earth says it's okay to watch onto it.
[10:02:45] <IchGuckLive> Hi all
[10:06:23] <mazafaka> hi
[10:07:25] <pcw_home> Crazy shopping day in US, good time to stay far away from malls
[10:10:21] <IchGuckLive> black friday also here in the US bases Area around Europe
[10:11:05] <IchGuckLive> factory outlat with around 40k parking places packed to the burst
[10:11:18] <pcw_home> Our newspaper was 3/4 advertising
[10:11:43] <IchGuckLive> thats good for the business and Tax
[10:12:08] <IchGuckLive> 2 Turkeys may have survived wars in the News today
[10:14:53] <pcw_home> we have wild turkeys here (non native, established by fish and game commission on the 1960s)
[10:14:55] <pcw_home> they can become quite a nuisance
[10:15:38] <IchGuckLive> what is the hunting fee
[10:16:07] <pcw_home> Dont know
[10:16:39] <ReadError> these turkeys here
[10:16:47] <ReadError> they stand 5' from the interstate
[10:16:50] <ReadError> eating stuff
[10:17:33] <IchGuckLive> ReadError: so they are filled already O.O
[10:17:39] <pcw_home> They are native in the east/midwest but not west coast
[10:23:00] <IchGuckLive> im off by til later
[10:28:22] <awallin> pcw_home: around? I wanted to ask about the latest PCI-E FPGA card. Does that have a linuxcnc driver? How about the SPI extension-board?
[10:32:02] <mazafaka> guys give me some good article for noob about the differences between steppers and servos (e.g. DC brushless)
[10:32:21] <mazafaka> there was an article but I have no link by now
[10:34:17] <pcw_home> awallin yes the PCIE card (6I25) is supported
[10:34:44] <pcw_home> Theres no SPI breakout but thats just mechanical
[10:36:03] <awallin> pcw_home: I was thinking of having some 24-bit ADCs for temperature measurement over SPI. by SPI-extension I meant 7I46
[10:36:43] <awallin> If I were to build my own cards with SPI ADC and DAC chips, would it be hard or easy to get the signals into HAL ?
[10:38:35] <tjb1> I recommend not posting a question about laser on cnczone unless you want "frank wang" and the "gweike" crew to spam you
[10:38:50] <awallin> pcw_home: oh, actually I meant the 6I68. that is the latest development?
[10:52:36] <pcw_home> 6I68 has been supported for along time, the latest and lower cost card (the 6I25) is supported by the 5I25 driver
[10:54:06] <pcw_home> There's a HostMot2 SPI driver but because SPI devices are so different you basically need a component for setup for each chip type
[10:57:51] <awallin> pcw_home: ok thanks.
[10:58:28] <awallin> pcw_home: also I wanted to ask about counting pulses with the FPGAs, for example the 3X20. what is the fastest it can count?
[10:58:55] <pcw_home> Theres an example component for the 7I65
[10:59:23] <pcw_home> The quadrature counter?
[11:00:02] <pcw_home> the FPGA itself can count maybe a few hundred MHz
[11:00:13] <awallin> the pulses would be TTL pulses, short pulses where we just want to count how many rising edges we see
[11:00:27] <awallin> the pulse-train would not be regular, pretty random in fact
[11:02:15] <pcw_home> the encoder counter has a limit because is has a digital filter (3 clocks minimum with filter 'off') on all inputs
[11:02:17] <pcw_home> so with a3X20 (50 MHz clocklow) this is 60 nS minimum recognizable pulse
[11:02:25] <awallin> the pulses come from a photomultiplier. one of the has a 10ns wide pulse according to the spec
[11:02:43] <awallin> pulse height 2-2.2V
[11:02:57] <pcw_home> OK so that doable, but requires a different counter type
[11:04:02] <awallin> we would count the number of pulses into a register during some time-interval, say 1ms or 10ms, and then read+reset the register over PCI/HAL
[11:04:08] <pcw_home> how many channels do you need
[11:04:19] <ReadError> hey guys
[11:04:25] <ReadError> http://www.youtube.com/watch?feature=player_detailpage&v=MwZCuTlSKeY#t=21s ;)
[11:04:28] <ReadError> that is so awesome
[11:04:31] <awallin> for the pulse-counting just one (at first at least)
[11:06:06] <awallin> pcw_home: for fast pulse-counting is it possible to raise the clock with PLL for some parts of the fpga?
[11:06:26] <pcw_home> Yes, I would probably just use a newer Spartan 6 card like a 5I25 (ClockHigh of 200 MHz)
[11:06:28] <pcw_home> and add a simple 32 bit counter
[11:06:39] <awallin> ReadError: wouldn't it be even greater if it was made with open-source cam + machine-controller
[11:07:16] <pcw_home> that would resolve down to 5 ns without needing a FPGA hardware clock pin
[11:08:24] <awallin> pcw_home: what about the 3x20, it has a larger spartan3? how fast can that go?
[11:09:19] <pcw_home> Its default ClockHigh is 100 MHz but could probably go higher
[11:09:40] <pcw_home> (you donr need a big FPGA for this kind of thing)
[11:09:45] <pcw_home> dont
[11:10:23] <awallin> right. well I will have to think some more about these things and convince people at work about the open-source way :)
[11:41:12] <Loetmichel> re @ home
[12:08:27] <IchGuckLive> Hi all B)
[12:47:59] <skunkworks> http://www.practicalmachinist.com/vb/bridgeport-hardinge-mills-lathes/add-z-ez-trak-2-axis-255376/
[12:50:29] <IchGuckLive> nice thing
[12:56:49] <tjb1> Cant believe people go to practicalasshats :P
[12:57:41] <IchGuckLive> some find it very usevull maybe
[12:58:22] <IchGuckLive> in india the mashines are also blessed by a priest
[12:58:28] <IchGuckLive> in Hawai 2
[12:58:57] <tjb1> I stopped going there are milacron or whoever the admin is started banning people because they didnt list their town in the location
[13:00:10] <IchGuckLive> yea that personilisation spreds worldwide facebook Virus
[13:00:23] <L84Supper> it's a slippery slope, first they "require" your town to be listed on a forum, next they implant RFID cards in you at birth or invade Poland
[13:01:07] <IchGuckLive> its the "Gläserne Mench " caled over here
[13:01:19] <IchGuckLive> Orvel 1984
[13:01:57] <IchGuckLive> glass man
[13:02:58] <IchGuckLive> what is on the attempt to bring linuxcnc to 12.04
[13:03:11] <IchGuckLive> is thee a timeline or still realtime issues
[13:03:36] <awallin> rtai might never happen for new kernels I think
[13:03:47] <L84Supper> RTAI is in the works
[13:03:50] <awallin> so it's xenomai or rt_preempt or something else
[13:04:31] <IchGuckLive> ubuntu studio works on realtime so it might be solved somewhere
[13:04:47] <L84Supper> RTAI has only one main dev, we won't be able to help him out until Jan-Feb
[13:05:27] <L84Supper> if xenomai gets pushed upstream, so will RTAI
[13:05:28] <IchGuckLive> ah ok so Next year it is the plan to be up to date
[13:05:48] <awallin> it seems mah's work will make linuxcnc build for a number of rt-kernels. that would give more choice
[13:05:49] <L84Supper> IchGuckLive: not sure about the official word from the LinuxCNC devs
[13:06:15] <IchGuckLive> wait and see
[13:06:15] <L84Supper> I just know that I will need it for some projects next year
[13:06:39] <L84Supper> we will share whatever we work on with RTAI
[13:06:53] <IchGuckLive> im in the way to build up a new education room with 26 routers inside
[13:08:00] <IchGuckLive> we will see
[13:08:09] <IchGuckLive> as time moves on
[13:08:41] <IchGuckLive> Question where is a good store in the USA to get Pulleys XL 037 15 teeth
[13:08:57] <L84Supper> people are also looking at RTAI for ARM and linuxCNC
[13:09:40] <L84Supper> so it will get pushed into 3.x sooner or later
[13:11:19] <jdh> Ich: sdp-si.com
[13:11:28] <IchGuckLive> :D
[13:16:08] <andypugh> Ooh! 1mm pitch toothed belts!
[13:16:12] <IchGuckLive> jdh: price is more then here in europ with tax
[13:17:22] <IchGuckLive> http://maedler.de/Product/1643/1616/1013/1015.aspx
[13:18:42] <IchGuckLive> as i need 150 pcs its 2,9Eur + tax here
[13:30:29] <IchGuckLive> BY
[13:33:04] <tjb1> First thing I do to new computers…re-install the OS, gets rid of bloatware :)
[14:00:26] <ReadError> always
[14:06:28] <JT-Shop> http://www.bbc.co.uk/news/technology-20395212
[14:10:59] * JT-Shop goes to try and finger out why my offsets go crazy
[14:15:39] <JT-Shop> I see it now the post doesn't do a G43
[14:17:35] <tjb1> Do they still offer the option to buy a computer without all the crap on it?
[14:18:32] <JT-Shop> hmm it should work after the tool change it puts G43 H1 Z0.1
[14:18:53] <tjb1> You broke it
[14:49:20] <JT-Shop> Z scale being off might explain it
[15:03:06] <mevon> hello the PID values are set in the .ini file right? by default, if none are defined in the ini, I and D are set to 0?
[15:03:46] <JT-Shop> they can be set in either the ini or the hal it just depends on what is in your hal file
[15:05:36] <mevon> can I see to what it it set in HAL config?
[15:06:04] <mevon> it is* set
[15:06:06] <JT-Shop> can you rephrase the question?
[15:06:27] <mevon> HAL configurations
[15:06:35] <JT-Shop> http://linuxcnc.org/docs/html/config/ini_config.html#_custom_sections_and_variables_a_id_sub_custom_variables_a
[15:06:47] <mevon> does it show the value of P, I and D?
[15:06:54] <JT-Shop> does that answer your question?
[15:07:07] <JT-Shop> what is it?
[15:07:54] <mevon> integrators?
[15:08:01] <JT-Shop> yes
[15:08:08] <mevon> where do I set the values
[15:08:24] <mevon> in the ini section in the link you pposted?
[15:08:55] <JT-Shop> the values can be in either the ini or the hal file, it depends on how they are written
[15:08:58] <mevon> ive read this page many times I cannot say I understand every part of it
[15:09:04] <mevon> ok
[15:09:10] <awallin> if your HAL file has things like [AXIS_0]P then that value comes from the INI
[15:09:29] <mevon> ok I get that thks
[15:09:36] <JT-Shop> if it is used in the hal file
[15:10:02] <JT-Shop> your hal file has to have something like setp stepgen.0.position-scale [AXIS_0]SCALE
[15:10:15] <mevon> my last question was can I use the HAL configurations interface to see the values they are set by default
[15:10:21] <JT-Shop> to use[AXIS_0] SCALE = 16000
[15:10:32] <mevon> yep
[15:10:37] <JT-Shop> ah yes the show hal configuration
[15:10:46] <JT-Shop> or the tune one
[15:10:52] <awallin> use halmeter or halshow ?
[15:11:01] <mevon> position-scale is P
[15:11:04] <mevon> ?
[15:11:26] <mevon> thing is Im trying to reduce the overshoot
[15:12:00] <mevon> when jogging manually, holding the jog down give me a accumulated error
[15:12:12] <mevon> and after a while it gets too large
[15:12:21] <mevon> thus the f-errors I get
[15:13:02] <mevon> I would like to change the values of the Differential part
[15:13:43] <JT-Shop> http://gnipsel.com/linuxcnc/tuning/servo.html
[15:13:58] <JT-Shop> velocity drive?
[15:14:21] <mevon> thats gold man
[15:14:48] <mevon> yes exactly that subject
[15:14:55] <mevon> thanks a lot Ill read it
[15:15:11] <awallin> mevon: some plots in my blog http://www.anderswallin.net/2008/04/x-axis-test/ (mesa card + pwm servodrives)
[15:17:28] <mevon> nice thank you
[15:18:34] <JT-Shop> mevon: if you open up Machine > Calibration you can adjust any values that are in the ini file and open up halscope to see how you did
[15:22:17] <mevon> will do
[15:25:39] <mevon> I get some weird error with that JT-Shop I will investigate more later, just worried it my incomprehension that is not working here...
[15:28:36] <mevon> think its because my servo are position driven, not velocity or voltage..
[15:30:39] <awallin> you would need to pastebin your HAL+INI and take some screenshits of the pid-error using halscope. otherwise it's very hard to say anything...
[15:30:50] <awallin> screenshot... haha :)
[15:32:36] <JT-Shop> position driven? like step and direction?
[15:48:11] <mevon> JT-Shop, no more like go to x4.500000 y3.200000 etc.
[15:48:44] <JT-Shop> how do you send that to the drive?
[15:48:52] <mevon> usb arduino
[15:49:11] <mevon> and i update the pos every 1ms
[15:49:38] <awallin> not really a standard linuxcnc way that...
[15:49:47] <mevon> ikr
[15:49:51] <JT-Shop> have fun, that is not in my guesspertice
[15:50:26] <mevon> lol well its just that insteed of setting the joints velocity, i use pos-cmd
[15:51:33] <mevon> and receive pos-fb while moving
[15:59:08] <andypugh> But _whebn_ do you get the feedback? Have you any idea of the timing?
[16:05:26] <andypugh> Are you aware that LinuxCNC doesn't send endpoints? It sends a new position request every 1mS.
[16:05:55] <andypugh> And all the velocity and accel data is encoded in how far those points are from each other.
[16:14:32] <tjb1> I have to find out why my wait for input isnt working…I believe I have the input inverted
[16:17:51] <r00t4rd3d> your going to be dead by the time that machine is ready to cut stuff
[16:18:30] <tjb1> it already cuts stuff
[16:19:01] <r00t4rd3d> what? your finger while bolting on more steel?
[16:19:35] <tjb1> Android make your brain fuzzy?
[16:21:13] <tjb1> Make your aluminum side plates yet?
[16:32:57] <DJ9DJ> gn8
[16:58:01] <Xabster> my friend and i are making our own CNC router XYZ thingy
[16:58:17] <Xabster> we're talking about what kind of sensors we need at the ends of the axis
[16:58:25] <JT-Shop> my post processor generates code like this:
[16:58:33] <JT-Shop> T1 M06
[16:58:39] <JT-Shop> G00 G90 X0.3215 Y-0.363 S1800 M03
[16:58:43] <Xabster> to 1) make it emergency stop and 2) make sure that the Z is stopped in time to not ruin the machine itself
[16:58:45] <JT-Shop> G43 H1 Z0.1
[16:59:08] <JT-Shop> any danger in have a move before the G43?
[16:59:29] <JT-Shop> micro switchs work fine
[16:59:34] <JT-Shop> switches
[17:00:00] <Xabster> microswitches? what kind?
[17:00:06] <Xabster> we found 1/10 millimeter switches
[17:00:24] <Xabster> 1/10 of a millimeter could be cut off from the underlying board on the CNC
[17:00:36] <Xabster> if the Z is impresice
[17:00:40] <Xabster> unprecise
[17:00:41] <Xabster> whatever
[17:00:56] <Xabster> or potentially it would not cut clear in whatever object it was cutting in
[17:03:02] <Xabster> do i make sense?
[17:03:38] <JT-Shop> http://www.geocities.ws/killer_siller/2zzvvlcontroller/micro.switch.jpg
[17:03:56] <Xabster> yep, we looked at them
[17:04:02] <JT-Shop> not exactly clear to me what your saying
[17:04:13] <Xabster> but 1/10 millimeter might mean that it will cut downwards into whatever is below my object
[17:04:16] <Xabster> that i'm working on
[17:04:25] <Xabster> and taking 1/10 of a millimeter off the machine itself
[17:04:33] <Xabster> when cutting completely through something
[17:05:25] <tjb1> Shouldnt be JT-Shop
[17:05:49] <tjb1> Should activate G43 at a safe position though, but good that you have the spindle on before you activate it
[17:09:09] <JT-Shop> I think it is bad coding as it relies on the Z being in a safe position for the XY move before the G43 is applied
[17:11:01] <tjb1> Let me see if I have a mastercam generated code here
[17:17:00] <tjb1> Here you go - http://pastebin.com/nSi8YTyh
[17:18:17] <tjb1> Its always pulling from tool change position in that code
[17:24:55] <JT-Shop> that's what has me confused as I don't have a tool change position... yet
[17:28:14] <tjb1> Cant you just code a "G28 G91 Z0.0" in your post? after tool changes or before it calls the G43?
[17:30:31] <JT-Shop> I just need to add TOOL_CHANGE_POSITION = 0 0 2 to the EMCIO section
[17:31:09] <JT-Shop> actually this looks better TOOL_CHANGE_QUILL_UP = 1
[17:32:27] <tjb1> Im not sure if G28 works the same with linuxcnc anyway so you might want to check that if you decide to use it
[17:35:29] <JT-Shop> G28/30 is go to a predefined position
[17:42:22] <tjb1> Once I get this laptop up and running, ill finally make a post for my plasma
[18:45:08] <r00t4rd3d> http://i.imgur.com/5xYOr.jpeg
[18:46:52] <r00t4rd3d> only in West Virginia
[18:49:58] <jdh> I'm not sure a knife would be enough.
[19:08:07] <Xabster> she's foxy, she wouldn't even need a knife if she just smiled more
[19:08:28] <andypugh> She doesn't look very fast on her feet, and she had her pants down at the time. Running would work.
[19:09:23] <andypugh> <trying to get back to topic> Was it a CNC knife?
[19:10:32] <andypugh> I was sent a spam about these earlier: http://cubify.com/cube/ I wonder if they are any good. The samples look suspiciously good, and pricing is Ultimaker-comparable.
[19:18:57] <andypugh> Hmm, will these come in handy, I wonder? http://www.ebay.co.uk/itm/13-Taps-4-5-to-7-MODEL-ENGINEERING-CLOCK-WATCH-MAKERS-/330831010697?pt=LH_DefaultDomain_3&hash=item4d07109b89
[19:22:54] <tjb1> I am going to outbid you.
[19:22:57] <PCW> not going to take much of a mistake to snap that .7 mm one
[19:23:13] <andypugh> I just want to see it :-)
[19:26:52] <andypugh> I have used M1.2 screws (holding a ceramic striker onto a piezo element).
[19:27:20] <andypugh> That was a fairly expensive thing. It was gold-plated too :-)
[19:29:16] <andypugh> (The special part of this: http://www.nordson.com/en-us/divisions/dage/products/bondtesters/Pages/4000HSHighSpeedBondtester.aspx )
[19:31:48] <PCW> 2-56 is the smallest I've used but Ive lost a lot of smaller ones :-)
[19:33:05] <andypugh> PCW: Have you ever seen this sort of thing? http://www.nordson.com/en-us/divisions/dage/products/ExampleProductFamily/Documents/CT%20brochure11%20press%20quality.pdf
[19:33:34] <andypugh> The pictures are in-situ images of BGA and bond wires.
[19:33:47] <tjb1> Smallest I have done so far is M3…that was pretty iffy for me.
[19:34:34] <PCW> No but some of the cards we do are x-rated for BGA soldering inspection (not CT though)
[19:34:51] <PCW> xrayed :-)
[19:34:55] <andypugh> x-rated? Yikes!
[19:35:29] <alex4nder> ~
[19:35:30] <alex4nder> `
[19:35:42] <alex4nder> er
[19:36:04] <andypugh> I helped design the X-ray tubes. Very interesting machines from a motion control POV. If I had known then what I know now I don't think they would have sacked me.
[19:37:22] <andypugh> mevon: How is it going? Ready to give up yet?
[19:37:50] <mevon> andypugh, no its getting better and better :P
[19:38:24] <andypugh> If it works properly then it will turn 10 years of LinuxCNC recieved-wisdom on its head.
[19:38:45] <andypugh> How long do you buffer for?
[19:39:18] <mevon> i can change the setpoint on the fly
[19:40:10] <mevon> i can buffer usb msg for as large as the arduino serial object can
[19:40:43] <mevon> and I update the position every ms
[19:41:25] <mevon> was trying to get a measure of that time but I kinda forgat to shutdown the com before reprogramming
[19:41:51] <mevon> and that is not something you really want to do with the arduino software
[19:43:03] <andypugh> Does this use ordinary USB, or some sort of realtime-USB?
[19:43:12] <mevon> andypugh, since I use motmod to control the servo, where would be the best place to tweak the integrators settings?
[19:43:26] <mevon> andypugh, its a python script
[19:43:36] <mevon> so I would say pyserial
[19:43:44] <andypugh> About 4 layers up outside the realtime code
[19:45:22] <mevon> so are every other modules running in emc that are python based
[19:45:49] <mevon> wonder if HAL is python
[19:46:02] <andypugh> _if_ the Arduino intercepted the trajectory planner queue then it ought to work. I _think_ this is how AriasRobo use the USB-connected 7i43 to do this: https://www.youtube.com/watch?v=BraEMAu5UkY (which is clearly working just fine).
[19:46:38] <andypugh> HAL is C. All realtime code is kernel code, and all kernel code is C.
[19:47:15] <mevon> so is the python machine
[19:49:32] <mevon> Ive been coding for at least 15 years and I never seen something better then Python
[19:51:02] <Jymmm> GO TURTLE!!! GO TURTLE!!! GO TURTLE!!!
[19:51:18] <andypugh> I've been coding for 33+ years (yikes!) but all the C-code I ever wrote is in the LinuxCNC Git repo.
[19:51:23] <PCW> You can certainly use a buffered approach (as Arais does) but you pretty much lose HAL
[19:53:39] <mevon> i think c code will always rule
[19:55:02] <andypugh> There are reasons that there are so many languages, they all do different things.
[19:55:55] <andypugh> Matlab is brilliant for processing numerical data. I wrote an STL slicer (STL to Z-slices) in about 10 lines of code.
[19:56:38] <andypugh> But you couldn't write a device-driver in it. I tried using Matlab to toggle parport bits once. 6Hz.
[19:56:54] <andypugh> VBA in Excel could do 10kHz.
[19:57:26] <andypugh> (except when windows decided to do something else, which killed that approach)
[19:57:41] <mevon> try python one day
[19:57:53] <mevon> maybe it ll change your mind?
[19:58:01] <andypugh> I have tried Python.
[19:58:03] <mevon> i hate vba
[19:58:06] <mevon> :P
[19:58:22] <mevon> you didnt like it?
[19:58:54] <andypugh> The only programming language I dislike is C :-)
[19:59:50] <andypugh> I know why it is like it is, but really, strings would be so handy.
[19:59:55] <mevon> can you get realtime with java?
[20:00:43] <alex4nder> some time is realer than others
[20:01:26] <cpresser> mevon: depends on the VM implementation. and your understanding of reattime :)
[20:01:56] <cpresser> (as in timing requirements)
[20:03:23] <andypugh> The programming languages I know, in order of learning them: Sinclair BASIC, Z80 machine code, Forth, BBC Basic, Fortran, QBASIC, LabView, MATLAB, LabViewRT(the only time I have been paid to code since I was 14 in Sinclair BASIC), VBA, C, Python.
[20:04:35] <andypugh> With VBA I can do things to Excel you would not believe (including replacing a $10,000 per year package in 100 engine dynos)
[20:06:51] <mevon> stay away from Delphi
[20:08:42] <andypugh> VBA is surprisingly powerful, as it gives you full control of the Office object model. If you want to learn OO programming I can't think of a better way. The IDE does completion, listing the properties and methods, and you have absolutely no choice about using OO as all you have to play with is an immense OO structure. And it's interpreted to you can breakpoint and modify etc. I am not saying it is a great language, but
[20:08:42] <andypugh> a place to leatn OO? Probably the best.
[20:09:16] <t12> scientists do hilarious things in excel
[20:09:20] <t12> WITHOUT vba
[20:09:37] <t12> lik ODE solvers
[20:09:54] <t12> that use that goofy excel iteration thing
[20:10:36] <andypugh> Some loon did a 3D render of a rotating cube entirely in cell formulas.
[20:11:23] <t12> http://excel.acdc.com/
[20:11:38] <t12> http://www.youtube.com/watch?v=h9_YkXHCkgA for not the actual spreadsheet
[20:15:07] <FinboySlick> http://www.youtube.com/watch?v=j8Nz5l0qiD0 <http://www.youtube.com/watch?v=j8Nz5l0qiD0&amp;feature=plcp> and http://www.youtube.com/watch?v=yu6drtokNOA <http://www.youtube.com/watch?v=yu6drtokNOA&amp;feature=plcp> for anyone wanting to see my crappy chinese mill doing actual work.
[20:16:13] <andypugh> What sold VBA to me were not great reasons. 1) I can distribute a spreadsheet in my corporate environment without anyone raising an eyebrow. 2) It can contain 20,000 lines of code (there is a 120 page manual for this "macro"). 3) I can import any DLL from any other installed software. This is the main point, as Excel acts as a familiar interface for the users to somerthing that behind the scenes controls a major applicat
[20:16:13] <andypugh> on the PC which interfaces with the vehicle ECU, and also controls 3 other PCs in the test cell (by three different interfaces) to control the dyno, the test equipment, and the room. (My macro can turn the lights out :-). Yes, I could have done it in a different language, outside Excel. it would have been better. But I would not have been allowed.
[20:20:55] <andypugh> The future arrived while I was drunk: http://www.guardian.co.uk/artanddesign/architecture-design-blog/2012/nov/23/print-yourself-3d-photo-booth
[20:23:38] <andypugh> FinboySlick: Nice. Needs EMC2 and LinuxCNC tags. But you win, your machine works, mine is still a project. (A project that skipped machines)
[20:24:10] <FinboySlick> andypugh: Hehe, I'll let you have mine for a year if you straighten it up ;)
[20:24:20] <FinboySlick> andypugh: And I'll add the tags right away.
[20:24:44] <andypugh> Have you considered straightening it in software?
[20:26:28] <andypugh> What made the toolpath? it looks like a good one?
[20:29:15] <FinboySlick> andypugh: HSMWorks made the toolpath, and I don't think software will help much for tram-like adjustments. I guess it could compensate for the uneven ways in X and Y.
[20:30:14] <FinboySlick> Uneven ways on the Z column means that the head shifts around an unknown radius depending on its position.
[20:30:18] <andypugh> Somewhere there is a kinematics for a twisted gantry.
[20:31:01] <andypugh> If you make that a _known_ radius then a custom kins can fix it.
[20:31:03] <FinboySlick> andypugh: I want to take it all appart and rebuild it right, hoping most of the errors are assembly errors.
[20:31:49] <FinboySlick> andypugh: It's a Z variable radius, but yeah. Wouldn't fix the un-adjustable tramming though.
[20:32:34] <andypugh> It would. If you can parameterise the errors into equations, it can be done.
[20:32:39] <FinboySlick> andypugh: Bottom line, I want it straight and I'll aim for that first. If it were, it would actually be a very very nice little mill.
[20:33:12] <andypugh> Robots and hexapods work, and they are _lots_ worse than your not-quite-cartesian machine.
[20:34:09] <FinboySlick> andypugh: software correction is nice for the actual cutting, but when I'm setting up the parts, all my references are off.
[20:34:33] <andypugh> Yes, fix the hardware, but bear in mind that anything curved or irreperably skewed is probably fixable in software if the hardware is too borked.
[20:34:55] <andypugh> I am not sure that is true
[20:35:18] <FinboySlick> andypugh: simple example for that part. The reference is a through hole so I can machine both sides.
[20:36:06] <andypugh> Consider the situation if your machine was some sort of crazy thing like a SCARA on the platform of a hexapod, being waved around by a Huge PUMA robot.
[20:36:39] <andypugh> As far as LinuxCNC is concerned it is a simple machine that moves in XYZ.
[20:36:53] <FinboySlick> 1- The hole doesn't go straight down (Z column not straight). 2- the hole isn't round (head not trammed), 3- I'm not able to indicate the hole position because my dial indicator is longer than the drill bit.
[20:37:12] <andypugh> Once you are in World mode it will move and jog in perfect XYZ space.
[20:38:47] <andypugh> JT-Shop: cradek Video http://www.youtube.com/watch?v=vVgc8ksstyg&feature=endscreen
[20:38:58] <FinboySlick> andypugh: With inaccurate tramming, moving in perfect XYZ involves knowing the tool length.
[20:39:13] <andypugh> Yes? and?
[20:39:30] <FinboySlick> I'd have to input a tool length for my dial indicator :P
[20:39:42] <andypugh> And?
[20:40:31] <FinboySlick> Sounds terribly unpractical... And hard to get repeatability.
[20:41:06] <FinboySlick> And it still won't prevent drill bits from going in 'sideways' instead of straight down.
[20:41:54] <andypugh> Yes, actually, it will :-)
[20:42:21] <FinboySlick> Yeah, I take that last one back.
[20:42:35] <FinboySlick> But it does really really really mean things to my OCD.
[20:43:19] <andypugh> It's not the _ideal_ solution, but bear it in mind as a option.
[20:44:21] <andypugh> Evil christmas present for OCD friends. Those day-of-the-week socks. Argh!
[20:45:13] <andypugh> (Better still, a lot of grey ones marked "left" and black ones marked "right" :-)
[20:45:51] <FinboySlick> But socks do end up left-right as they adapt to your feet.
[20:46:02] <FinboySlick> the stretching pattern is different.
[20:48:10] <andypugh> The point is that given that set there is no consistent solution.
[20:50:57] <andypugh> And I am not sure about the stretching. Have you noticed that if there is a hole in your sock at the big toe, if you swap feet the hole is still at the big toe? How is that possible?
[20:51:28] <FinboySlick> It is not! Heretic!
[20:54:14] <andypugh> Anyway, I need to be up in 3 hours!
[20:54:16] <andypugh> Night all
[20:54:26] <FinboySlick> Goodnight :)
[21:19:39] <FinboySlick> qasw.Élkjdfwplkuytfdrsedrghjnl,.é
[21:19:40] <FinboySlick> <lkhgxzxcbm
[21:19:47] <FinboySlick> woops
[21:29:09] <r00t4rd3d> is that your password?
[21:39:49] <FinboySlick> r00t4rd3d: Yup. root access to my vacuum cleaner.