#linuxcnc-devel | Logs for 2015-02-05

[02:07:43] <KGB-linuxcnc> 03Sebastian Kuzminsky 05seb/master/po4a 6af3919 06linuxcnc 10docs/src/config/lathe_config.txt 10docs/src/config/lathe_config_fr.txt docs: make lathe_config{,_fr}.txt comparable * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=6af3919
[02:07:43] <KGB-linuxcnc> 03Sebastian Kuzminsky 05seb/master/po4a 63eb6f1 06linuxcnc 10debian/control.in build-depend on po4a, for translated docs * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=63eb6f1
[02:07:43] <KGB-linuxcnc> 03Sebastian Kuzminsky 05seb/master/po4a a3a76e8 06linuxcnc 10docs/src/Submakefile docs: build translated asciidoc files from master files + po * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=a3a76e8
[02:07:46] <KGB-linuxcnc> 03Sebastian Kuzminsky 05seb/master/po4a eff7946 06linuxcnc 10docs/src/Submakefile docs: supply asciidoc lang attribute from Submakefile * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=eff7946
[02:07:49] <KGB-linuxcnc> 03Sebastian Kuzminsky 05seb/master/po4a 62e9397 06linuxcnc 10docs/README docs: add info on using po4a to README * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=62e9397
[02:07:53] <KGB-linuxcnc> 03Sebastian Kuzminsky 05seb/master/po4a 88564a0 06linuxcnc 03docs/src/config/lathe_config.se.po docs: begin swedish translation of config/lathe_config * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=88564a0
[02:07:57] <KGB-linuxcnc> 03Sebastian Kuzminsky 05seb/master/po4a fa2c9b7 06linuxcnc 10docs/src/config/lathe_config.se.po docs: do some swedish translations for lathe_config.txt * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=fa2c9b7
[02:08:01] <KGB-linuxcnc> 03Sebastian Kuzminsky 05seb/master/po4a cfef4c8 06linuxcnc 03docs/src/config/lathe_config.fr.po docs: begin french translation of lathe_config.txt * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=cfef4c8
[02:08:06] <KGB-linuxcnc> 03Sebastian Kuzminsky 05seb/master/po4a 044f025 06linuxcnc 10docs/src/config/lathe_config.fr.po docs: clear the "fuzzy" tags in in the french .po * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=044f025
[02:08:10] <KGB-linuxcnc> 03Sebastian Kuzminsky 05seb/master/po4a 6a7856a 06linuxcnc 04docs/src/config/lathe_config_fr.txt lathe_config_fr.txt is no longer tracked by git * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=6a7856a
[03:35:40] <BuildCNCMaster> Hi Got a Question on Rotary Feed
[03:36:11] <BuildCNCMaster> Isent it posible to USE a G94.1 with a D as Diameter Specification
[03:36:30] <BuildCNCMaster> to Calculate the Rotary Stand alone feedrate in Deg/min
[03:36:48] <BuildCNCMaster> as a ABC Move only is made
[03:37:32] <BuildCNCMaster> so G1 C180 will move at F/ (D*Pi ) *360
[03:39:00] <BuildCNCMaster> most post processors in Cam dont Calculate Extra F in Deg/min
[03:41:32] <BuildCNCMaster> coudt be implemented in the next 2.8 version
[03:44:06] <BuildCNCMaster> so it changes from UU/min into Deg/min on a Rotation only move on given Diameter Calculation
[03:45:28] <BuildCNCMaster> Thanks
[08:38:32] <cradek> that's an interesting idea
[09:18:53] <mozmck> PCW thanks for the sample configs
[09:20:54] <cradek> although I guess the postprocessor could easily calculate that feed, or write an expression that calculates it
[09:22:27] <mozmck> pcw: pcw_home: what does the PID do for steppers?
[09:23:33] <pcw_home> All linuxcnc stepgesn are closed loop in the sense that they are digital oscillators with count feedback
[09:24:23] <pcw_home> the stepgen has a built in loop in the driver (when run in position mode)
[09:25:53] <mozmck> so how does the PID work? you don't move the motor back do you?
[09:26:30] <pcw_home> but... this built in loop is not very robust with respect to servo thread jitter,
[09:26:32] <pcw_home> so its better to use the PID component wheich hase more knobs to tweak
[09:27:08] <mozmck> oh, I think I see a little.
[09:27:26] <pcw_home> no the feedback is only from the stepgen position count
[09:27:33] <mozmck> ok
[09:28:06] <pcw_home> so effectively you are using the stepgen hardware as a velocity mode servo
[09:28:20] <mozmck> another thing I don't understand is what all needs changes in a bit file, and what can be done from the ini and hal files.
[09:28:32] <mozmck> I need to figure out how a velocity mode servo works.
[09:29:58] <mozmck> From what I have heard so far, stepgens are hardwired in the bit file (that's the FPGA firmware right?) to certain pins.
[09:30:19] <pcw_home> changing pinouts of anything other than GPIO (stepgens, encoders etc) require new bit files
[09:30:20] <pcw_home> changing numbers and types of modules requires a new bitfile
[09:30:39] <mozmck> modules?
[09:31:01] <pcw_home> stepgen,uart,SPI,encoder etc
[09:31:35] <pcw_home> all I/O can be GPIO on all bitfiles
[09:31:44] <mozmck> so what does the num_stepgens setting do?
[09:32:05] <mozmck> just tell how many of the predefined modules to use?
[09:32:23] <pcw_home> yes (and unused module pins revert to GPIO)
[09:32:44] <mozmck> Ok, so for GPIO, I presume you can change from output to input in HAL?
[09:33:02] <pcw_home> the hardware can do this on a pin by pin basis but the driver takes a simpler view
[09:33:46] <pcw_home> Yes GPIO defaults to input but can be chaned to output or open drain
[09:34:10] <mozmck> ok.
[09:34:21] <pcw_home> also output bits can be inverted
[09:34:52] <pcw_home> not much use for GPIO but useful for things like stegen/PWM etc
[09:35:25] <mozmck> So would it be hard to make the HAL driver be able to tell stepgens which pin to use for output?
[09:36:01] <pcw_home> the pins a stepgen uses are fixed by the bitfile
[09:36:52] <mozmck> yes, but I'm talking about a hypothetical change - thinking out loud :)
[09:37:13] <pcw_home> its possible but quite large
[09:37:32] <mozmck> a large change? I guess it would be.
[09:38:07] <pcw_home> (requires a crossbar switch to do in a general way)
[09:38:12] <seb_kuzminsky> linuxcnc-build: force build --branch=2.6 1401.rip-wheezy-rtai-i386
[09:38:13] <linuxcnc-build> build #765 forced
[09:38:13] <linuxcnc-build> I'll give a shout when the build finishes
[09:38:53] <seb_kuzminsky> ok, that should fix the "git auto-gc causes build failures" problem
[09:39:01] <mozmck> crossbar switch - another term for me to look up :)
[09:39:05] <pcw_home> Normally you would want a bitfile to match the I/O hardware
[09:40:18] <pcw_home> mesaflash can print out the current bitfiles pinout
[09:40:19] <mozmck> My thought was that if it could be made a little more generic, the hal configuration would match the I/O hardware, and hal files are much easier for mere mortals to edit.
[09:40:32] <mozmck> oh, I should try that!
[09:41:17] <pcw_home> There really are not that many parallel port stepgen pinouts
[09:41:28] <mozmck> I got linuxcnc talking to my 7i92 last night, but I don't know if the bitfile in it will work.
[09:42:29] <pcw_home> there are about 8 or so 7I92 bitfiles in the dist now
[09:42:31] <mozmck> hmm, that's probably right. Our hardwre wants the step signals in the same pins as the G540, so I think that bitfile will work for me.
[09:43:27] <pcw_home> mesaflash --device 7i92 --addr --readhmid
[09:43:29] <pcw_home> will print the pinout
[09:43:44] <cradek> once you get above 4 axes it gets really nonstandard, but for 4 or fewer axes there are only really the two choices
[09:43:52] <cradek> really
[09:44:07] <mozmck> ok, except I'm using the address
[09:44:21] <mozmck> I loaded the etherhm2.xise in webpack 14.7, and I got about 18 files missing, which I assume is not a problem?
[09:44:38] <pcw_home> OK (thats a better choice anyway)
[09:44:54] <pcw_home> probably not an issue
[09:45:32] <pcw_home> did you see tom_itx's little tutorial?
[09:45:42] <mozmck> Looks like it is set up for the 7i80. What all do I need to change to play with 7i92 files? I presume I need to change the 7i80db.ucf to the 7i92.ucf
[09:46:06] <pcw_home> that and the chip type
[09:46:13] <mozmck> I saw it, but I'm not sure it tells me enough.
[09:46:38] <pcw_home> and select the pinout file and card type in the top level file
[09:47:46] <mozmck> in the file? I assumed you do that in the project manager on the left.
[09:49:00] <pcw_home> those are selected in the top level file
[09:50:10] <mozmck> I have never actually used an FPGA or the xilinx tools yet, so I may need to do some reading.
[09:50:14] <pcw_home> (I use a batch file that replaces the @CARD and @PIN tokens to build all)
[09:51:27] <mozmck> If I un-comment the line "use work.i92_x9card.all; will it automatically add the 7i92.ucf file to the project list in the hierarchy?
[09:51:46] <mozmck> batch file - that sounds like windows :)
[09:51:56] <pcw_home> No that part needs to be done in the gui
[09:52:28] <pcw_home> you can do all this from the command line
[09:52:47] <pcw_home> (which avoids all the mousing around)
[09:53:29] <mozmck> yes, but then I have to know the commands to type, which involves mousing around, or more typing.
[09:53:57] <pcw_home> or just mesaflash the 7i92_g540x2D config if it will do :-)
[09:54:39] <pcw_home> when you get it to compile, the gui leaves the command log
[09:54:51] <mozmck> that's too easy! but probably what I should do. I would like to learn FPGA, but should probably wait.
[09:55:02] <KGB-linuxcnc> 03Sebastian Kuzminsky 052.7 62ae9cd 06linuxcnc Merge remote-tracking branch 'origin/2.6' into 2.7 * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=62ae9cd
[09:56:09] <pcw_home> If you want to change a pinout/module setup you edit a pin_xxx.vhd file
[09:59:40] <linuxcnc-build> build #765 of 1401.rip-wheezy-rtai-i386 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1401.rip-wheezy-rtai-i386/builds/765
[10:05:07] <mozmck> pcw_home: I just tried to write the 7i92_G540x2D.bit file and it told me "Error: wrong bitfile destination device: 6slx9tqg144, should be xc6slx9
[10:05:56] <pcw_home> Hmm latest mesaflash?
[10:06:16] <mozmck> this is from the 7i92.zip file I downloaded yesterday - I used the command "mesaflash --device 7i92 --addr --write 7i92_G540x2D.bit
[10:06:28] <mozmck> mesaflash 3.1.0
[10:07:09] <mozmck> built Nov 24 2014
[10:08:43] <pcw_home> I would update
[10:08:45] <mozmck> It's the one in linuxcnc master I guess.
[10:09:31] <mozmck> how?
[10:10:03] <pcw_home> maybe fetch from https://github.com/micges/mesaflash and build
[10:10:29] <pcw_home> I think there was FPGA name error
[10:10:58] <micges> yeah name error was fixed in nov or dec
[10:11:29] <pcw_home> I havent noticed since I use an ancient copy thar doesnt even check
[10:11:53] <pcw_home> (I tell it a 7I92 is a 7I80)
[10:12:25] <pcw_home> maybe not reccommended if you dont have JTAG :-)
[10:12:29] <mozmck> Is there a version in linuxcnc master? I'm assuming that's how I got it.
[10:12:54] <micges> bbl
[10:13:25] <mozmck> heh, I have a cheapo JTAG adapter, but it's for ARM chips - will that work?
[10:13:38] <pcw_home> probably not
[10:13:39] <mozmck> and OpenOCD
[10:14:02] <mozmck> oh, I do have a xilinx parallel cable iv
[10:14:13] <pcw_home> well maybe if openOCD knows how to progran xilinx
[10:14:26] <pcw_home> parallel cable will work
[10:15:04] <mozmck> I don't know if i can get it to work on this version of linux and all - it seemed like a bit of a pain looking at the docs.
[10:16:05] <pcw_home> yeah that may be
[10:17:37] <pcw_home> normally JTAG shoud not be needed, but is hasny for development where you may make bitfiles that dont work
[10:17:45] <pcw_home> handy
[10:18:41] <pcw_home> normal stuff like changing I/O pinouts is safe (cant make a config you cant access)
[10:19:00] <mozmck> yeah, I'm sure. For my ARM cortex stuff I wrote a bootloader that I generally just have to load once, then update after that through my usb->rs485 hub.
[10:19:20] <mozmck> But sometimes I make a bug bad enough I have to reload the bootloader.
[10:19:56] <pcw_home> bad things are wrong ucf file, busting something in the Ethernet ROM etc
[10:20:20] <cradek> pcw_home: should the mesaflash packages on linuxcnc.org be updated? I hate when people have to build from source and then never again get the automatic updates.
[10:21:17] <cradek> mesaflash:
[10:21:17] <cradek> Installed: 3.1.0
[10:22:41] <mozmck> That's the one I have, and I just figured out it is a separate package.
[10:22:49] <mozmck> from linuxcnc that is/
[10:22:57] <cradek> yeah it appears to be the latest
[10:23:07] <cradek> if it needs to be updated we should do it
[10:23:47] <mozmck> pcw_home: what is the mesaflash3 utility in the 7i92 zip file? it does not report a version, but the mesaflash in the same directory is 3.0.0
[10:24:00] <pcw_home> I think the latest changes (other than enhancements) only affect the 7I92
[10:24:51] <pcw_home> Yeah thats old, I should probabl just remove it
[10:30:03] <mozmck> ok, the latest from github gives me the same error
[10:30:17] <mozmck> version 3.2.0-pre
[10:33:34] <pcw_home> ahh so it has not been fixed
[10:35:09] <pcw_home> freeby.mesanet.com/mesaflash.exe should work under windows
[10:35:27] <mozmck> I fixed it :)
[10:35:46] <mozmck> line 291 in eth_boards.c
[10:37:16] <pcw_home> ahh rather truncated chip name
[10:38:43] <mozmck> and with an extra xc at the beginning
[10:38:47] <pcw_home> you can ues the reload option if you get tired of pulling the power plug
[10:39:18] <mozmck> I see that, probably will do if I do much more of that.
[10:39:24] <mozmck> no reset button?
[10:39:36] <pcw_home> nope
[10:40:12] <mozmck> those are handy - until you bump it in the middle of a big job I guess...
[10:40:53] <pcw_home> could wire one to a TP
[10:41:11] <pcw_home> (those are currently debug outputs)
[14:04:07] <mozmck> pcw_home: how can I get a "charge pump" signal of about 12 khz out on pin 17?
[14:05:45] <mozmck> I see that pin 16 has a stepgen on it, and that is the charge pump pin on the G540, so do you normally use a stepgen for charge pump?
[14:05:54] <mozmck> PCW ^^
[14:23:30] <cradek> a stepgen in velocity mode would sure do it
[14:30:30] <mozmck> What is exactly is velocity mode?
[14:30:41] <PCW> Yes in velocity mode and either set the pulse width to what you want or select quadrature mode
[14:30:55] <mozmck> If you are controlling velocity, how do you do positioning?
[14:30:55] <PCW> man hostmot2
[14:31:27] <PCW> like a velocity mode servo
[14:32:24] <mozmck> I'll have to look that up.
[14:32:58] <PCW> if you imagine that every 1 ms you set the velocity to the required velocity, other than timebase errors and quantization errors the position would be correct
[14:33:28] <mozmck> I see, that makes sense.
[14:33:31] <PCW> so that what you do, the PID loop takes care of the residual errors
[14:34:20] <mozmck> so is there a way to get a charge pump signal on pin 17 without making a new bit file?
[14:34:29] <PCW> (FPGA clock <> PC clock, frequency quantization etc)
[14:34:49] <mozmck> I guess I could load different bit files and see what they look like using mesaflash
[14:35:29] <PCW> yeah (Actually I should do that and make the xml file but have been too busy)
[14:36:04] <PCW> you can look at the corresponding 5i25 .pin files
[14:36:55] <PCW> and I can make a bitfile if you send me a desired pinout
[14:36:55] <mozmck> what is the xml file?
[14:37:03] <PCW> for pncconf
[14:37:37] <mozmck> Let me do some looking and see what I can figure out - thanks!
[14:40:48] <PCW> if its close to an existing pinout, only takes me a minute or so
[14:47:31] <mozmck> ok
[15:49:18] <micges_> mozmck: fix name to 6slx9tq144 works with --write to 7i92?
[15:49:53] <mozmck> 6slx9tqg144 works, yes
[15:50:07] <mozmck> (you missed the g)
[15:50:39] <micges> PCW: you missed the g in 7i92 doc
[15:50:52] <mozmck> very difficult change - do you want a pull request ;)
[15:51:15] <micges> yes please ;)
[15:52:04] <mozmck> haha!
[15:52:05] <micges> I think I didn't update my 7i92 (which I can't find on desk) so I didn't catch that typo
[15:54:23] <skunkworks> I lost a small motherboard on my desk...
[15:54:57] <mozmck> I guess your mind is not empty
[15:55:48] <mozmck> PCW: I tried compiling a bit file, but webpack doesn't like my license. I think I got the license for an earlier version.
[15:57:44] <PCW> I think you have to register to get a working webpack license
[15:59:42] <mozmck> I know I did that once, but not with the latest download. I'm working on a pinout for you
[16:12:53] <micges> I found my 7i92!
[16:15:18] <PCW> I should make it bigger
[16:16:47] <PCW> hmm maybe I should make the bitfile ID number 0x7192 or whatever
[16:19:21] <mozmck> hah, I went to the xilinx site and got another license and it built the file this time.
[16:52:13] <micges> PCW: I need to clean it up once in a while ;)