**** BEGIN LOGGING AT Thu Apr 26 03:00:00 2012 Apr 26 03:05:26 oh man i wish that PyBBIO library did interrupts Apr 26 07:46:18 hi Im new to the Beaglebone and Cloud9. I have a fresh image and wanted to run the demo blinkled.js. When I attempt to run it I get an Error: Cannot find module '/var/lib/cloud9' anyone else seen this? Apr 26 08:21:57 hmm, exodus? Apr 26 08:25:45 I know, its really quiet today Apr 26 08:29:46 <_av500_> gm Apr 26 08:36:25 * koen reads up on hwmon Apr 26 08:37:54 you don't need it to interact with other drivers, do you? Apr 26 08:38:03 do you want to add the adcs to it? Apr 26 08:38:33 I'm exploring ways to make the battery cape more "intelligent" Apr 26 08:38:38 e.g. adding a INA219 Apr 26 08:39:03 there's a thread on the lm-sensors list about changing hwmon sysfs layout Apr 26 08:39:13 triggered by someone writing a ina219 driver Apr 26 08:39:58 what I actually want is to have such a cape hook into the power subsystem (or whatever makes userspace recognize it as a batter) Apr 26 08:40:15 so I start with reading up on hwmon :) Apr 26 08:41:01 aholler: but you're right, I do want to 'export' AIN8 to a power related subsystem to mark it as "current measurement" Apr 26 08:41:37 and I found out I'm completely ignorant on how that stuff is done in the kernel and userspace :) Apr 26 08:42:13 less Documentation/power/00-INDEX Apr 26 08:43:15 right Apr 26 08:43:24 aholler: I have implemented this already: http://dominion.thruhere.net/koen/angstrom/beaglebone/0001-beaglebone-connect-batterycape-GPIO-to-gpio-charger.patch Apr 26 08:44:22 * koen <3 gpiolib Apr 26 08:44:22 eh, no dt Apr 26 08:44:43 the DT capable kernel for bone lacks mmc support Apr 26 08:44:47 so it's useless for me Apr 26 09:18:00 <_tasslehoff_> koen: I get "PowerVR device not found" from /usr/bin/ES3.0/pvr2d_test. Can that be the heisenbug at work? Apr 26 09:23:31 _tasslehoff_: I get the same Apr 26 09:26:27 _tasslehoff_: TBH I'm note sure if that is supposed to work at all, since there's no 2d hardware in the pvr :) Apr 26 09:27:03 _tasslehoff_: I just got of the phone with _roger_ and we hope to schedule some sgx work next week if time permits Apr 26 09:30:27 <_tasslehoff_> koen: ok. how can I check if I have the same issue you do? I'm trying to run the demos and get this http://pastebin.com/Y9xHnHJv. No X server running, but the demos don't need that? Apr 26 09:56:57 _tasslehoff_: check powervr.ini Apr 26 09:57:10 if that says front or flip then you have the same issue Apr 26 09:57:19 if that says X11, change it to FRONT or FLIP Apr 26 09:58:48 <_av500_> not FLOP Apr 26 09:59:22 -ELOP Apr 26 09:59:28 <_tasslehoff_> koen: same issue then (WindowSystem=libpvrPVR2D_FRONTWSEGL.so.1) Apr 26 10:12:43 _tasslehoff_: use ${PV} instead of 1.3.8 in the SRC_URI, remove the PR = r0 and move the srcuri hashes between SRC_URI and S Apr 26 10:21:37 <_tasslehoff_> koen: will do. Apr 26 10:22:24 _tasslehoff_: on, and install -m 0755 instead of cp in do_install Apr 26 10:27:05 <_tasslehoff_> koen: ok. did "on," mean anything? Apr 26 10:29:29 * _tasslehoff_ needs to find a good intro to git send-email Apr 26 10:55:19 _tasslehoff_: tried the man page? Apr 26 11:02:55 <_tasslehoff_> mru: yep. my send-email didn't heed the --subject prefix I fed it, but when I prefix in format-patch all is well Apr 26 11:03:04 <_tasslehoff_> --subject-prefix Apr 26 11:03:25 send-email mostly accepts format-patch options Apr 26 11:03:59 Hi - I am now on ubuntu and beaglebone, i2cdetect seems to detect all device But when I try to open i2c from c++ program, open() function returns < 0 (ERROR)... Apr 26 11:04:28 Has anybody some clue what could be wrong? Same code works perfectly on angstrom Apr 26 11:09:33 mru can you help ? Apr 26 11:09:51 no Apr 26 11:10:12 that was pretty .. uhmmm... Apr 26 11:13:46 sorry, I don't know how i2c works Apr 26 11:37:59 I was positively surprised when I saw ubuntu 11 on beagle includes drivers for rtl8192cu Apr 26 11:38:27 Butt it looks it behaves as AccesPoint Apr 26 11:39:26 and "sudo ifconfig wlan0 down" just takes and takes too long... Apr 26 11:52:25 * mdp tries to accept that A5 bone still has flaky phy support Apr 26 12:42:47 hi, can anyone tell me why this is happening? /bin/sh: can't access tty; job control turned off Apr 26 12:44:39 I canot bring i2c to 400 kHz - I tried with: i2c_bus=2,400 - hows that ? From dmesg: omap_i2c omap_i2c.3: bus 3 rev2.4.0 at 100 kHz Apr 26 12:45:07 av500: Helo are you there ? Apr 26 12:45:41 talking smtp now, are we? Apr 26 12:46:11 :-) Apr 26 12:46:40 lol Apr 26 12:46:46 zhivko: 250 Apr 26 12:54:16 Ok, I have files with no file modes whatsoever, is this a normal thing? Apr 26 12:54:23 I'm starting to question myself Apr 26 12:55:07 sorry, we can't help with existential conundrums here Apr 26 12:56:20 heh Apr 26 12:56:55 42 Apr 26 12:57:25 easy ;) Apr 26 12:58:54 well, it seems my entire rootfs was corrupt Apr 26 12:58:56 I have doubt Apr 26 12:59:05 so it explains the days of issues that i've been having Apr 26 13:00:48 av500 : what 250 ? Apr 26 13:01:44 So they give us drystones and than BB runs i2c on 25kHz - great ;) Apr 26 13:03:35 then change it Apr 26 13:04:54 and yes, i2c_bus=2,400 should do that Apr 26 13:05:00 add water to make it wet stones ? Apr 26 13:05:13 zhivko: bus 2 or 3? Apr 26 13:05:31 bus 3 Apr 26 13:05:39 av500: bus 3 Apr 26 13:06:05 av500: I already change it and I am saying it is not working! Apr 26 13:06:36 av500: there is no effect. I use ubuntu got from RobertNelson prebuilt image Apr 26 13:07:44 av500: I keep getting: [ 1.469635] omap_i2c omap_i2c.3: bus 3 rev2.4.0 at 100 kHz Apr 26 13:08:07 av500: an then I check on oscilloscope and it's 25kHz - GREAT ! Apr 26 13:08:09 and you are root? Apr 26 13:11:19 av500: PING ? Apr 26 13:14:20 av500: have I miss something :) Apr 26 13:17:05 av500: PING ? Apr 26 13:17:26 yes my dear? Apr 26 13:49:29 av500: you think I am making something wrong with that uEnv.txt or it should work? has anybody reported running i2c on 400kHz ? Apr 26 13:49:51 pastebin a full boot log Apr 26 13:51:02 koen: "The SCREEN Phenomenon" Apr 26 13:51:05 creepy Apr 26 14:28:26 hi, why do you have to use a usb hub between a usb device and the BB? Apr 26 14:28:55 power Apr 26 14:31:54 rory: you don't have to Apr 26 14:32:09 I have plugged things directly into the BeagleBone. Works fine Apr 26 14:32:19 hi all, to change muxing of the omap, have I just to type commands "MUX_VAL(CP(MCBSP3_DX), (IEN | PTD | DIS | M4)) /*GPIO_140*/\" decribbed here? http://elinux.org/BeagleBoardPinMux Apr 26 14:32:26 As long as you are powering it with the full power input, not the microusb Apr 26 14:33:05 what if I am powering it with the microusb what kind of effects would that cause? Apr 26 14:33:09 lpi: don't confuse source code with commands Apr 26 14:33:27 ok so I have to modify the source code of Uboot and recompile it, right? Apr 26 14:34:08 the kernel will set the muxing again, so changing it in u-boot won't help Apr 26 14:35:37 So it's more simple to use omap_mux function of the kernel? Apr 26 14:35:38 I think the microusb input does not provide enough milliamps. The power-management chip slows down the CPU and may disable power to the main USB port - I am not sure about that Apr 26 14:37:06 well I am powering it with microusb and have a usb device plugged in (Arduino with a LCD attached to that) and it all works ^^, but the communication from the BB to the Arduino is extremely slow, I was wondering if a lack of power could be hte cause Apr 26 14:40:10 no Apr 26 14:40:21 aholler: me? Apr 26 14:40:25 no Apr 26 14:40:45 no ? Apr 26 14:41:02 lpi: I can't decide whats more simple for you Apr 26 14:41:54 aholler: Anyway, I have either to recompile Uboot or the kernel, isn't it? Apr 26 14:42:31 lpi: should I quote myself? Apr 26 14:43:29 I know I don't always write correct english, but most if it should be understandable Apr 26 14:43:38 barely Apr 26 14:43:57 fuzzy helps Apr 26 14:45:19 so a lack of power will change the mux ;) Apr 26 14:46:20 I understood, but I just want to know mux my pin. The wiki doesn't mention kernel compilation Apr 26 14:47:22 + how to Apr 26 14:47:24 a wiki isn't the answer to live, the universe and everything Apr 26 14:47:29 how can I find out why the serial communication is so slow (how to fix it)? Apr 26 14:47:36 define slow Apr 26 14:47:40 s/live/life/ Apr 26 14:47:47 ok so what can I do if I'm beginner? Apr 26 14:48:05 no wiki, no irc? just mailing list then? Apr 26 14:49:35 lpi: what you want to do? Apr 26 14:50:09 some thing just aren't easy. I don't know how to change pin muxing (by hand waving or similiar easy stuff). Apr 26 14:50:41 Barely basic: just to blink a LED. The problem is that it doesn't work with /sys/class/gpio technic. So I guess the muxing isn't good, and I'm trying to change it Apr 26 14:51:10 you can pin mux in two ways through u-boot or kernel Apr 26 14:51:50 I know that, but aholler said it useless to change it with uboot because when the kernel is booting, it override muxing Apr 26 14:52:12 thats not totally true, i think Apr 26 14:52:24 depend which pins Apr 26 14:52:24 Well I set the baudrate to 9600, but I'm lucky to send 1 integer per second Apr 26 14:52:29 anyway, the best solution is to use kernel module, right? Apr 26 14:52:54 Witn this method, I can change every pins on the expansion bay Apr 26 14:54:09 i needed to pin mux to get spi working and i had success modifying u-boot source Apr 26 14:54:33 rory: sorry, dont understand Apr 26 14:54:37 9600 is 9600 Apr 26 14:54:49 if the other side is slow to respond, thats anothert issue Apr 26 14:55:29 siegen: my final goal is to use SPI in fact. So how do you say which pin aren't modified by the kernel? Apr 26 14:55:59 -say + know Apr 26 14:56:19 but i think kernel modifies pins if CONFIG_OMAP_MUX is set it up in kernel config Apr 26 14:57:59 Ok, I haven't any "omap_mux" in my /sys/class folder, so the module you're talking about is not installed? Apr 26 15:00:55 in the kernel config in the same section where you can set up that option you can enable some debugging about pin mux which will printk info about which pinmux is done by kernel Apr 26 15:11:22 siegen: thank you very much Apr 26 15:11:24 Hello everybody! I am having trouble making changes in the driver modules delivered to my board. I can change the defconfig file and recompile, and I do see the changes reflected in the defconfig in the build tree, but deploy does not seem to see any changes and the 'new' kernel has the same drivers as before. What am I doing wrong? Apr 26 15:11:56 /sys/kernel/debug/omap_mux is essential reading Apr 26 15:17:45 * XorA wonders why pmount is so difficult for people, or maybe there is a new ubuntu branded keyboard with sudo key Apr 26 15:17:49 XorA, except the omap-isms in the mode output can be disorienting if they expect it to match am335x padconf regs Apr 26 15:18:43 mdp: the mode is the most essential information and thats a number 0-7 :_D Apr 26 15:19:01 depends on what you're doing :) Apr 26 15:19:48 the omap specific output parse is useless Apr 26 15:19:58 damn those ti people Apr 26 15:21:22 * XorA is outnumbered by them Apr 26 15:22:19 now I am trying -cf deploy Apr 26 15:25:58 'bitbake -f virtual/kernel -c deploy' still returns 'Attempted 393 tasks of which 393 didn't need to be rerun and all succeeded.' Apr 26 15:27:00 that bitbake stuff is just usefull to install a preconfigured kernel. for everything else git clone a kernel. Apr 26 15:27:21 So even though 'compile' sees work to do, and the work seems related to what i did, deploy does not. Apr 26 15:28:22 why not use -f -c compile then? Apr 26 15:28:45 followed by -c deploy Apr 26 15:29:35 I did use -f on the compile but i do not know if it was really needed. Apr 26 15:38:09 your answer went over my head. I used git clone to get the Angstrom distribution of course but you are saying that i should what, clone this and then stop using bitbake? Apr 26 15:45:04 clone the kernel and just don't use bitbake. Apr 26 15:46:27 most of the linux world does it so Apr 26 15:46:48 s/so/this way/ Apr 26 15:49:19 at least people which want to modify the kernel or the kernel-comfig Apr 26 15:49:51 the others use ubuntu :) Apr 26 15:50:18 ubuntu has no interest in the kernel Apr 26 15:50:27 you are talking about going back to the base kernel and starting from there, abandoning Angstrom and Openembedded? Apr 26 15:51:02 av500, ubuntu users usually dont ... Apr 26 15:51:06 I have the solution to my issues with the kernel Apr 26 15:51:38 sjbtk: no, angstrom has somewhere the patches and maybe even a git-repo. it's just just well hidden to not destroy our running gag #1 Apr 26 15:51:42 basically, the kernel built by Angstrom is not working on some beaglebones Apr 26 15:52:02 I took the sources from koen kooi and now it is working ... Apr 26 15:54:56 hey guys how to clone this kernel ? https://github.com/koenkooi/linux/tree/beaglebone-3.2 Apr 26 15:56:57 I did https://github.com/koenkooi/linux/tree/beaglebone-3.2 Apr 26 15:57:01 did you try clicking on the "Git Read-Only" button near the top left ofthe page? Apr 26 15:57:18 then put git clone in front of the link it gives Apr 26 15:57:55 but getting error as fatal: https://github.com/koenkooi/linux/tree/beaglebone-3.2/info/refs not found: did you run git update-server-info on the server? Apr 26 15:57:55 anybody please help me Apr 26 15:58:21 hack the server Apr 26 15:58:24 jeshwanth... I just answered you Apr 26 16:04:40 rory, av500: I notified slow i2c running on 35 kHz instead on 400 Mhz ? Apr 26 16:06:15 meant 25kHz instead of 400kHz :) Apr 26 16:08:18 zhivko_: you can specify the speed of each i2c bus when you bring it up... in the board file Apr 26 16:09:13 thurbad: Yes I tried to do that but in dmesg it keeps telling me it runs on 100kHz Apr 26 16:09:44 you still have not pasted a full boot log Apr 26 16:11:25 hi Apr 26 16:11:58 the i2c thats wired to DVI slot is fixed at 100Khz I think Apr 26 16:16:13 avr500: you mean me? Apr 26 16:16:52 XorA: the pico dlp gets upset with >100kHz Apr 26 16:17:14 koen: I read the comments so I know this :-D Apr 26 16:17:28 :) Apr 26 16:17:37 also sending i2c fast over 10m of cable does not work Apr 26 16:19:56 zhivko_: yes Apr 26 16:20:03 at least setting: i2c_bus=2,400 in uEnv.txt should work ? isn't it true ? Apr 26 16:20:28 av500: you mean to paste whole dmesg here ? Apr 26 16:20:53 I'm afraid some bot will kick me out beacause of spam :) Apr 26 16:21:08 ubuntu@omap:~$ dmesg | grep i2c [ 0.099131] omap_i2c.1: alias fck already exists [ 0.110812] omap_i2c omap_i2c.1: bus 1 rev2.4.0 at 100 kHz [ 1.489265] omap_i2c.3: alias fck already exists [ 1.494429] omap_i2c omap_i2c.3: bus 3 rev2.4.0 at 100 kHz [ 2.035838] i2c /dev entries driver Apr 26 16:21:59 dpaste.org Apr 26 16:22:00 xorA: that's i2c on bus 0 I think? yes ? Apr 26 16:22:06 full log Apr 26 16:22:13 zhivko_: my memory aint that good Apr 26 16:22:31 zhivko_: of course not Apr 26 16:23:08 I can send yyou dmesg on private message - ok ? Apr 26 16:23:24 no Apr 26 16:23:29 [18:21:59] dpaste.org Apr 26 16:23:34 or pastebin Apr 26 16:24:24 since you seem to be running ubuntu... sudo apt-get install pastebinit; dmesg|pastebinit Apr 26 16:24:45 and post the url this returns Apr 26 16:24:52 http://dpaste.org/nOD1b/ Apr 26 16:25:17 zhivko_: read lin 18 Apr 26 16:25:47 (after enabling wordwrap) Apr 26 16:26:11 you mean: Built 1 zonelists in Zone order, mobility grouping on. Total pages: 65024 Apr 26 16:26:16 no 19 Apr 26 16:26:47 18 without wordwrap Apr 26 16:26:54 you mean: Kernel command line: console=ttyO0,115200n8 vram=12MB omapfb.mode=: omapdss.def_disp= root=/dev/mmcblk0p2 ro rootfstype=ext4 rootwait fixrtc ip=none Apr 26 16:27:43 yes Apr 26 16:27:50 So this means what? It will not take into account uEnv.txt ? Apr 26 16:28:10 paste your uEnv.txt Apr 26 16:28:27 not here Apr 26 16:28:45 aholler: do you have the latest version of your led patch online, or should I just get it from patchwork somewhere? Apr 26 16:29:16 http://dpaste.org/BbUDj/ Apr 26 16:29:23 koen: v4 is on lkml Apr 26 16:29:41 damn it, I need to bother b7500af1 Apr 26 16:29:45 that disables the heartbeat on panic too+ Apr 26 16:29:46 ah Apr 26 16:30:00 b7500af1, free the codes! Apr 26 16:30:59 let my codez go! Apr 26 16:31:07 zhivko_: use optargs=i2c_bus=2,400 Apr 26 16:31:29 zhivko_: see mmcargs=... Apr 26 16:31:44 that is what ends up in the kernel cmdline Apr 26 16:32:28 Crofton|work, soon enough.. give me another couple months. Apr 26 16:33:01 it is ok, you can use a public repo Apr 26 16:33:06 so no, -f is not needed by bitbake compile to see my changes to defconfig and run do_compileconfigs. except for the deploy, everything works as koen says. clean, update defconfig, compile, deploy. only deploy seems to 'fail'g, Apr 26 16:33:08 heh, nah, I should be done in a week or so. I think. (I always under estimate how long it will take.) Apr 26 16:33:08 we will not make fun of you :) Apr 26 16:33:29 aholler: so i need to update line with mmcargs ? Apr 26 16:33:35 aholler: like that: mmcargs=setenv bootargs console=${console} ${optargs} vram=${vram} omapfb.mode=${defaultdisplay}:${dvimode} omapdss.def_disp=${defaultdisplay} root=${mmcroot} rootfstype=${mmcrootfstype} ${device_args} i2c_bus=2,400 Apr 26 16:33:38 we are just concerned you will get busy with new adventures and not quit "finish" Apr 26 16:33:44 zhivko_: no Apr 26 16:34:22 zhivko_: read that line and think about what $(optargs) there is used for Apr 26 16:35:26 aholler: so this should be ok than: optargs=i2c_bus=2,400 mmcargs=setenv bootargs console=${console} ${optargs} vram=${vram} omapfb.mode=${defaultdisplay}:${dvimode} omapdss.def_disp=${defaultdisplay} root=${mmcroot} rootfstype=${mmcrootfstype} ${device_args} Apr 26 16:38:05 http://26-26-54.hardwarebug.org/30 Apr 26 16:49:19 so an empty optargs (optargs=) in uEnv.txt could help some people ;) Apr 26 16:54:53 i wonder why you have all the other stuff, like mmcargs, in uEnv.txt Apr 26 16:55:36 zhivko_: and an empty optargs would not help, you haven read mmcargs too ;) Apr 26 17:16:21 <_av500_> reading is overrated, amazon sells steel now Apr 26 17:16:39 metal! Apr 26 17:17:18 <_av500_> mru: dont trigger the sandtroll Apr 26 17:18:44 they have sold only plastic before? Apr 26 17:20:33 <_av500_> and paper Apr 26 17:20:48 no rock or scissors? Apr 26 17:21:32 * djlewis looks on amazon for Jack Daniels black label :) Apr 26 17:22:08 JD is the plastic of whisky Apr 26 17:22:25 <_av500_> mru: americans.... Apr 26 17:23:29 i like that "its been in the wooden barrel a long time", after taste :) Apr 26 17:23:43 define a long time Apr 26 17:23:56 even though it's the budweiser of whiskies…there's still that insane demand from yuropeens to consume it Apr 26 17:24:11 chavs... Apr 26 17:27:40 Hey guys, In beaglebone if I want to access pwm driver I have to read the files again and again, this will take more processor time right ? Apr 26 17:29:30 reading files consumes cpu cycles, yes Apr 26 17:30:48 ya but is there any other way to overcome this ? Apr 26 17:31:35 because I am writing an user space application for pwm in this it always reads analog input and writes to pwm . Apr 26 17:34:04 http://cheezburger.com/6154773504?siteId=75 Apr 26 17:42:49 Crofton|work: incredibly accurate, of course Apr 26 17:43:41 yes Apr 26 17:43:43 ask koen Apr 26 17:43:48 hrm, I left him on hold in the level 2 customer support call center for too long Apr 26 17:43:56 and canada is North Texas? Apr 26 17:44:05 * mdp closes the support ticket Apr 26 17:46:19 * koen only visits TX and cA Apr 26 17:46:34 and trying to get approval for the NY OSHW thing Apr 26 17:46:39 so yes, accurate :) Apr 26 17:47:28 I do resent the fact that it places me in TX…boo Apr 26 17:49:40 koen, I have an eye on that one too. Apr 26 17:49:54 Texas has some fine distinctions - I'm in West TX apparently. Apr 26 17:50:15 so CA? Apr 26 17:50:41 Or East CA according to the map... Apr 26 17:51:18 I know some Oregonians who'd resent being lumped into CA. Apr 26 17:53:20 emeb, I am highly pleased the CA and NY are segregated Apr 26 17:53:47 mdp: But there's only 1 state between them! Apr 26 17:54:06 that's the "real america" state I think :) Apr 26 17:56:16 As lauded in this classic: http://en.wikipedia.org/wiki/Beers,_Steers,_and_Queers Apr 26 17:58:43 yes, never forget that line from FMJ Apr 26 18:01:29 Crofton|work: lol Apr 26 18:01:43 emeb: css fail on wikipedia Apr 26 18:02:35 mru: canada is like north Texas but with better beer Apr 26 18:02:53 mru: works for me. you must have bad internets. Apr 26 18:03:27 emeb: don't you get a huge, empty space under the 'original album' heading? Apr 26 18:03:53 mru: nope - track listing. Apr 26 18:04:01 yes, _then_ the track listing Apr 26 18:04:21 the 'professional ratings' table in the right margin is too wide and pushes the track table down Apr 26 18:04:59 mru: looks like about 1 char height space to me. Maybe browser dependent? I'm using Chrome. Apr 26 18:05:14 browser depedent is fail Apr 26 18:05:21 nod Apr 26 18:08:20 Anyone recommend a debugger that supports ARM SWD? Apr 26 18:08:30 Only one I know is Blackmagic Probe. Apr 26 18:09:38 Doing some Cortex M? work lately and the ST Discovery boards + OSS drivers are failsauce. Apr 26 18:10:58 http://sourceforge.net/apps/trac/libswd/ Apr 26 18:13:03 thx Apr 26 18:13:13 so it looks like openocd already supports it Apr 26 18:13:36 hard to say. I've seen conflicting info about openocd + swd. Apr 26 18:16:05 http://www.olimex.com/dev/ARM/JTAG/ARM-JTAG-SWD/ARM-JTAG-SWD-schematic.pdf Apr 26 18:17:03 doesn't look like swd has less pins ;) Apr 26 18:17:51 actually it does. Only 3 pins really needed. clk, bi-dir data & gnd. Apr 26 18:18:31 that's 1 more wire Apr 26 18:18:45 like spy-bi-wire Apr 26 18:19:12 ds2: one more wire than what? Apr 26 18:20:07 spy-bi-wire Apr 26 18:20:26 take that back Apr 26 18:20:28 ds2, I think he was comparing to jtag Apr 26 18:20:29 you counted ground Apr 26 18:20:36 ds2, right :) Apr 26 18:21:21 ground is always implied Apr 26 18:21:27 maybe gnd doesn't count. hard to be successful w/o it though. Apr 26 18:21:31 otherwise, SPI becomes 4 wires and I2C becomes 3 wires ;) Apr 26 18:21:56 emeb, it makes for better marketing Apr 26 18:22:06 though a bit decieving Apr 26 18:22:10 emeb: by that line of reasoning, you should also include power Apr 26 18:22:15 good thing marketers don't do actual design. Apr 26 18:22:16 its "assumed" Apr 26 18:22:27 like the ez430's connector is 4 pin for the most basic one Apr 26 18:22:42 they need wireless h/w debug…even 1 wire is too many :) Apr 26 18:22:56 "using only the power of my mind..." Apr 26 18:23:03 even that is deceiving...its really 2 wires. Apr 26 18:23:05 emeb, s/marketers/markeTEERs/ Apr 26 18:23:38 I wonder if RFids have a RF debug interface ;) Apr 26 18:23:44 ka6sox: yep, very fast and loose in those product glossies Apr 26 18:24:27 mdp: I hear they make a tasty candybar. Apr 26 18:25:22 emeb, filled with a mystery substance known only as, "nougat" :P Apr 26 18:25:29 imagine people remotely debugging your RFid creditcard ;) Apr 26 18:25:48 ds2: imagine people remotely debugging your pacemaker or insulin pump. Apr 26 18:26:05 that's being done ;) Apr 26 18:26:06 * mdp adjusts tinfoil hat Apr 26 18:26:11 emeb: watch karen sandlers "unchain my heart" talk Apr 26 18:26:48 emeb: first documented attacks on pacemakers were published last year Apr 26 18:27:31 but pace makers have a power source Apr 26 18:27:32 koen: shades of David Lynch/Dune/Harkonnen heartplugs. Apr 26 18:27:37 credit cards don't Apr 26 18:28:15 * emeb buys tinfoil wallet Apr 26 18:28:42 now consider public transportation ;) Apr 26 18:29:21 Is the presence of ehrpwm in the sysfs tree dependendant on a kernel config? (they seem to come and go very other version) Apr 26 18:30:08 Or Orwell... Apr 26 18:32:27 My favorite Orwellian moment is when amazon deleted 1984 from all the kindles Apr 26 18:32:49 so priceless Apr 26 18:34:35 Meta FTW. Apr 26 18:51:01 jsabeaudry: what exactly is your observance with ehrpwm coming and going? Apr 26 18:52:03 maybe it is also applying PWM to the kernel versions Apr 26 18:54:38 jsabeaudry: in your EDMA travels did you ever try the am335x EDMA test app by chance? I can't recall if you had mentioned such a thing Apr 26 18:59:26 Hello! I've a question on Y-cable. I see that reference manual instructs to use OTG port on BB as a primary source for power in the section where it describes HS USB2.0 OTG Port. It also mentions about Y-cable for additional power. Will it work if I connect one end of Y-cable to BB, other end to USB port on system and remaining end to a USB female to female adapter to run USB 2.0 hub? Apr 26 19:02:28 <_av500_> orated: nope Apr 26 19:04:11 av500: 'coz both the USB A port in the Y-cable are only for power, no data? Apr 26 19:09:18 aholler: just for the record - to raise speed of i2c to 400kHz modify uEnv.txt like: i2cspeed=3,400 mmcargs=setenv bootargs console=${console} i2c_bus=${i2cspeed} ${optargs} vram=${vram} omapfb.mode=${defaultdisplay}:${dvimode} omapdss.def_disp=${defaultdisplay} root=${mmcroot} rootfstype=${mmcrootfstype} ${device_args} Apr 26 19:10:22 no Apr 26 19:11:36 put "optargs=i2cspeed=3,400" before the line with mmcargs Apr 26 19:12:07 and cat /proc/cmdline to verify Apr 26 19:14:22 jsabeaudry, fwiw, I think koen mention that one PSP release forgot the ehrpwm driver or something like that..and he had to add it in to his tree…probably should use his tree if you aren't already Apr 26 19:17:29 <_av500_> orated: you plug both Y ends into the PC Apr 26 19:17:35 <_av500_> and single end into BB Apr 26 19:19:03 av500: Yes, I know the connections. I just wanted to confirm the reason why the setup I'm trying won't work... is that because the Type A ones in the Y cable are solely for power and not data? Apr 26 19:25:23 _av500_, what is the magic google to get your partition tester Apr 26 19:28:38 got it Apr 26 19:29:24 http://www.bbc.co.uk/news/technology-17853523 why is this made headlines ive no diea Apr 26 19:29:25 *idea Apr 26 19:29:54 using beaglebone..... can someone accompany me for some meaningless flaming? Apr 26 19:31:56 mdp, all my DMA stuff is working now, but no I did not see the am335x edma test app, although I have seen one for another chip Apr 26 19:32:23 mdp, yes I'm on his tree, the r10d branch, previously I was on the r6 branch Apr 26 19:32:59 ok, thx Apr 26 19:33:50 I had the test app fail on the r6 branch today…locked up tight Apr 26 19:34:00 the edma one? Apr 26 19:34:41 it is the same one from the ti814x page, does many iterations of simple chained mem2mem copies Apr 26 19:34:44 arcanescu: i think it's the same guy from #beagle that burned out one of his BBs by putting 5V on gpio Apr 26 19:35:41 jay6981: are you joking ? Apr 26 19:35:47 5v on gpio ? Apr 26 19:35:55 jsabeaudry: I was trying it because my co-worker is trying to debug his crypto driver that uses dma…and he said the test driver worked at least Apr 26 19:36:18 jsabeaudry, so I'm amused to see it die for me…I recalled you had things working Apr 26 19:36:48 jay6981: I dont like the concept of the project.... and its in the news..... :/ # Apr 26 19:37:14 no, i think it's the same guy from this channel Apr 26 19:37:36 mdp, I do get occasionnal lockups but very infrequent im not sure they are related, but even the heartbeat dies Apr 26 19:37:40 it made it on the front page of reddit too… he must have pretty good PR to make bbc Apr 26 19:38:03 do you knwo his nick so i can tell him personally that thing sucks Apr 26 19:38:18 arcanescu: lol Apr 26 19:38:26 it has a case around it Apr 26 19:38:36 or against it? Apr 26 19:38:40 +1 for the case Apr 26 19:38:43 mdp: why lol :(.... i mean "all my DMA stuff is working now, but no I did not see the am335x edma test app" this is more intresting Apr 26 19:39:14 is edma supposed to be hard? Apr 26 19:39:20 yes Apr 26 19:39:26 looks like a bog standard dma to me Apr 26 19:39:30 seen one, seen 'em all Apr 26 19:39:31 is typing a sentence hard ? and getting it printed? Apr 26 19:39:43 arcanescu: MattRichardson Apr 26 19:40:17 It might be another issue to get the usb port to work on ur custom kernel to drive the printer or network... but yea ... other than that..... its not even a case Apr 26 19:40:24 mru, indeed it's similar to other dma engines Apr 26 19:40:49 except this one has poor driver support.. Apr 26 19:41:05 who needs drivers when there are mmio registers? Apr 26 19:41:31 just mmap the control regs to userspace and off you go Apr 26 19:41:32 jay6981: Ty ... ill hunt him down... Apr 26 19:41:35 another starterware user Apr 26 19:41:46 haha Apr 26 19:42:06 mru, map them to userspace to use in a kernel driver? ok Apr 26 19:44:41 oh, if you're already in the kernel you don't even need to do that Apr 26 19:44:52 hehe Apr 26 19:44:53 just pray nobody else is trying to use it at the same time... Apr 26 19:44:59 bingo Apr 26 19:45:02 arcanescu: sounds like you are envious. I find that idea to use the mechanical turk not bad. It's just a bit pricey Apr 26 19:45:16 do go on…tell me more about why you want to bang on the edma directly in your kernel driver Apr 26 19:45:53 oh shit…you just broke the spi driver! :) Apr 26 19:46:37 now we remember why there are frameworks for these things Apr 26 19:47:04 frameworks are for people with weak discipline Apr 26 19:47:11 hehe Apr 26 19:47:27 * panto feels an urge to bitbang something Apr 26 19:47:34 plus, anything that's worthwhile doing can be done by bitbanging gpio Apr 26 19:47:53 panto, you just wanted to use that word…you've been lurking...waiting Apr 26 19:48:07 yes... yes... yesssss... Apr 26 19:48:27 use the pru to bitbang usb Apr 26 19:48:41 panto, "Sparta!" Apr 26 19:49:05 aholler: easier than getting musb to work for sure Apr 26 19:49:14 hi, i've some questions: the SPI port can trigger a interruption? Apr 26 19:49:43 aholler, *yawn*….can you bitbang descriptions of photos from humans with that approach? Apr 26 19:50:26 depends on the humans Apr 26 19:50:39 energy: 2.1 eV Apr 26 19:50:43 spin: clockwise Apr 26 19:50:51 and the CPU how fast can do math operations, like multiplications and sums with integers? Apr 26 19:51:14 aholler: not at all envious.... mechanical turk.... i mean listen to yourself you are paying someone to basically translate an image.... wat 2$ 1 $ i wouldnt pay for that.... Apr 26 19:51:42 I don't pray, I just don't care. Apr 26 19:51:47 id rather go like hey wait a minute ill go to a room where i would have the image i uploaded myself translate that and send it back to myself Apr 26 19:51:48 arcanescu: it's groundbreaking :) I only laughed cause you want to stalk him and tell him it sucks :) Apr 26 19:52:01 journalists are writing all kind of stupid stuff Apr 26 19:52:02 mdp: :D spot on :P Apr 26 19:52:26 yes it is groundbreaking... literally Apr 26 19:53:13 ok, I'm outta here, cya Apr 26 19:58:32 arcanescu: you've read that and even spammed it here. so the journalist won. Apr 26 19:59:15 and we can await more such groundbraking news Apr 26 19:59:59 aholler: well atleast its giving a message to all other people doesnt matter howworthwhile your stuff is if you know the right people your'e famous Apr 26 20:00:14 incase anyone didnt know Apr 26 20:00:25 *_o Apr 26 20:00:32 so he knows you? Apr 26 20:00:51 ofcouse not Apr 26 20:01:22 but you made him famous here Apr 26 20:02:54 or you both get paid by amazon ;) Apr 26 20:03:01 shills! Apr 26 20:03:01 :] Apr 26 20:03:17 well im sorry about that that was un intentional Apr 26 20:03:59 keep up the good work..perhaps he will drive 500k+ sales of am3358 and I can feed my chilluns Apr 26 20:05:14 wait, who's famous? Apr 26 20:06:23 Russ, you can be if you like Apr 26 20:06:45 I missed out on whatever arcanescu was babbling about Apr 26 20:06:46 write an article about your project and send it to arcanescu Apr 26 20:07:15 Russ: not babbling Apr 26 20:07:19 fact1 Apr 26 20:07:27 If I had a project that was worth a news story, I'd send it as an elc-e proposal Apr 26 20:07:37 3:29 Apr 26 20:07:37 arcanescu Apr 26 20:07:37 http://www.bbc.co.uk/news/technology-17853523 why is this made headlines ive no diea Apr 26 20:08:38 right up there with a scrolling led board that has your twitter feed Apr 26 20:08:43 Russ, I have enough doubt about even elc-e for mine worthiness much less MSM :) Apr 26 20:08:50 which piece of technology is revolutionary Apr 26 20:09:13 I need to send them my msp430 with the reed switch monitoring the garage door….revolutionary! Apr 26 20:09:34 maybe I could hook it to mechanical turk's api so somebody can call me when I leave it open Apr 26 20:10:31 mdp, you need an angle Apr 26 20:10:56 mdp, with a camera, that way, if its just you working in the garage they wouldn't send anything Apr 26 20:12:09 the next big thing, telepresence seeing eye dogs Apr 26 20:12:32 Russ, submit the BeaconBoard scripts you did Apr 26 20:12:50 you could setup a huge center with seeing eye dogs that would be tasked out to telepresence bots as needed Apr 26 20:12:55 Russ, maybe I could pay them for a written description of the mess in the garage Apr 26 20:13:07 mdp, I don't think that's 50 minutes worth Apr 26 20:13:22 "It appears that the owner did not clean up after his last 27 projects in the garage" Apr 26 20:14:06 "There are exposed nails, seems to be a hazard" Apr 26 20:15:01 Russ, I like that idea…Sheldon-style telepresence devices :) Apr 26 20:19:25 oh dear god, what a horrible ad choice Apr 26 20:19:29 http://www.nydailynews.com/life-style/health/swiss-women-dies-giving-water-food-thought-live-sunlight-article-1.1067359 Apr 26 20:19:42 'WANT TO LOSE WEIGHT? DO THE MATH' Apr 26 20:20:05 oh geez, that's a link to a story not an ad Apr 26 20:20:58 a story that yells Apr 26 20:21:28 oh Russ, I actually received a rev c xM today Apr 26 20:21:43 I was so excited except I have no use for it now Apr 26 20:21:56 maybe 4-6 weeks ago Apr 26 20:25:09 finally Apr 26 20:25:40 http://www.adafruit.com/products/801 Apr 26 20:27:06 mdp, is your msp430 sensor wireless? Apr 26 20:28:14 no, but it on my silly projects roadmap to upgrade it Apr 26 20:29:30 I have an msp430 usb stick thing laying around, so I ordered an rf transmitter receiver pair from sparkfun Apr 26 20:29:37 cool! Apr 26 20:30:09 if it works well, I'm make a wireless remote sensor board with an msp430 Apr 26 20:30:39 ti does sell them Apr 26 20:30:52 yup, and they are very low power Apr 26 20:31:00 I started with the cc1100 eval stuff to learn a bit Apr 26 20:31:17 i mean wireless msp-stuff. isn't that what that clock uses? Apr 26 20:31:33 but I'm looking to deploy my various sensor nodes using the $10 cc100 xcvrs from circuit specialists Apr 26 20:31:37 wireless msp and the clock? Apr 26 20:32:13 er, cc1100 Apr 26 20:32:26 ez430 Apr 26 20:32:51 but what clock? Apr 26 20:33:00 chronos Apr 26 20:33:22 ah Apr 26 20:33:35 mdp, what is the expected range on the cc1100? Apr 26 20:35:30 I think you can get close to 50m or so with a decent rf design indoors Apr 26 20:36:18 my intention was to aggregate a few of these on to a beagle* and maybe run a contraption like Syntro…but that's a bit overkill Apr 26 20:36:35 I do want to centralize some notification things though Apr 26 20:36:35 these things are claiming 500ft range given ideal conditions Apr 26 20:36:37 http://www.sparkfun.com/products/10534 Apr 26 20:36:48 but they are one way and don't have fancy framing features or anything like that Apr 26 20:37:17 that's in a 500 ft long Tempest test chamber :) Apr 26 20:37:23 heh Apr 26 20:37:27 I'll experiment Apr 26 20:37:50 yeah, all you can do is experiment Apr 26 20:37:59 the also have something similar to the cc1100 Apr 26 20:38:00 http://www.sparkfun.com/products/10154 Apr 26 20:38:08 rfm22b Apr 26 20:38:17 a chip antenna like that makes me "have doubt" Apr 26 20:38:30 it can take an external antenna Apr 26 20:38:50 the reason to consider cc1100 is to be able to go mesh too Apr 26 20:39:23 let all those little low power nodes walk your sensor data back to home base Apr 26 20:39:47 but then you need to synchronize sleep/wake cycles Apr 26 20:40:13 yes, need to have wake events Apr 26 20:41:09 that's more on the vision side of things for me…today I just need to go back to the aggregator directly Apr 26 20:42:28 let me know how those work…the price is nice Apr 26 20:42:59 I've tested them in the 1 inch apart configuration so far Apr 26 20:43:14 ship it!! Apr 26 20:43:20 ok, gtg Apr 26 21:01:44 gone an hour and too much beagletalk to catch up on. Apr 26 21:03:46 summary: you just need to check out the camera Apr 26 21:08:12 hi, has someone ever converted the BBxM OrCAD design to Altium Designer? Apr 26 21:23:48 I thaught I raised speed of i2c to 400kHz, but than I checked i2cdetect with time - and found out it's same result! Apr 26 21:24:13 so i2c is still running on 25kHz :( Apr 26 21:24:37 aholler: does it sound familiar ? Apr 26 21:25:16 I don't know anything about i2detect Apr 26 21:25:42 what about i2cdetect? Apr 26 21:25:44 in dmesg it says it's on 400kHz, but I measured with osciloscope and it's 35kHz - who can help ? Apr 26 21:25:56 sorry 25kHz Apr 26 21:26:25 sounds bitbanged Apr 26 21:26:35 you sure you are measuring the clock time and not the byte cycle time? Apr 26 21:26:48 (25kHz is 1/16th of 400kHz) Apr 26 21:26:48 aholler: how do you mean it ? Apr 26 21:26:51 if the i2c speed is set at post fs boot wont the kernel override that? Apr 26 21:27:08 if it is hard coded in... Apr 26 21:27:36 I also builded kernel - but I don't know where to look for - in which code... Apr 26 21:28:47 djlewis: that would be great - they mention you can set it in uEnv.txt ?? Apr 26 21:29:09 seems i2c used to default to 100k Apr 26 21:30:11 zhivko: perhaps so, I havent used uEnv.txt. I have set it in u-boot but that would be trampled on whatever the fs sets it to. Apr 26 21:31:18 somehow 25k seems "just not right" Apr 26 21:31:43 zhivko, do you have a screenshot of your scope? Apr 26 21:31:48 How can I override this - or where to change it in kernel if I wanna build it? Apr 26 21:31:52 russ: yes Apr 26 21:32:06 can you post it to imgurl or something? Apr 26 21:33:23 https://picasaweb.google.com/110230689089207649183/Beaglebone25kHzI2c#5735782372143198978 Apr 26 21:33:52 cool, another rigol :) Apr 26 21:34:03 let's get to the point :) Apr 26 21:34:27 can you zoom in so you can see a single transaction more clearly? Apr 26 21:34:28 So where is master koen to consult me ;) Apr 26 21:34:39 they dont do so good a job freezing fast logic transitions. Apr 26 21:34:39 and I'm assuming this is clock, not data, right? Apr 26 21:35:19 jsabeaudry: *sigh* our sample code on the wiki is an epic fail as is the TRM since it shows 4 TCs in the block diagram Apr 26 21:35:54 yes it's CLK Apr 26 21:36:34 jsabeaudry, the am335x edma driver wiki page links to the sample code written for am387x/ti814x…a part that has 4 TCs whereas am335x only has 3 TCs… Apr 26 21:36:42 you wanna data also? isn't it CLK enough to see speed od clk ? Apr 26 21:36:58 jsabeaudry, trivial fix but sloppy docs :( Apr 26 21:37:03 that screen shot is looking like 400kHz to me Apr 26 21:37:13 each division is 20uS Apr 26 21:37:33 it looks like there are 8 clock cycles per division Apr 26 21:37:45 so 2.5uS per clock cycle Apr 26 21:38:19 djlewis: prpplague's hantek hacking has made me have second thoughts about a rigol Apr 26 21:38:38 mdp: i havent seen that Apr 26 21:38:47 Russ: time will show improvement from 100kHZ to 400kHZ don't you think? it should be 4 times faster - but it didn't Apr 26 21:39:13 ok, first of all, how are you measuring 25kHz? Apr 26 21:39:43 djlewis: http://elinux.org/Das_Oszi_Protocol Apr 26 21:40:08 djlewis: or really, http://elinux.org/Das_Oszi Apr 26 21:40:43 zhivko: you can use the cursor to measure, similiar to what I've done here: http://ahsoftware.de/ds1307_does_somewhat_work.gif Apr 26 21:40:45 do you think lowest clock of both busses overrides ? Apr 26 21:41:02 I don't think you've reviewed the i2c protocol Apr 26 21:41:11 how are you measuring 25kHz? Apr 26 21:41:22 is this data and clk ? Apr 26 21:41:44 you told me it was running at 25kHz, then posted a screenshot from a scope that looks like its running at 400kHz Apr 26 21:41:50 I'm curious where you got 25kHz Apr 26 21:44:48 * Russ wonders where realtors keep finding 'hardwood laminate flooring' and 'laminate hardwood flooring' Apr 26 21:45:13 "never ever ever confuse marketing with the truth" Apr 26 21:46:22 Russ: the hard wood occurs as they fantasize about screwing the customers with such wordings Apr 26 21:46:23 mdp: the guts of the hantek's look so much ike those of the rigols Apr 26 21:46:40 Russ I will produce picture of CLK and DATA Apr 26 21:48:10 and position the cursors Apr 26 21:49:07 then you don't need to do the math by yourself Apr 26 21:49:30 djlewis: yeah, that was the general assessment of people that dissected these units (or any of the other ones that are identical to the hantek)…that they were as good of quality as a rigol…which got me thinking about the openness aspect as a plus Apr 26 21:50:15 djlewis, I've been holding off getting a decent hobbyist grade DSO like a rigol until I really need it..as my old analog still works Apr 26 21:50:40 but hackable/extendable is always an attraction :) Apr 26 21:51:29 https://picasaweb.google.com/110230689089207649183/Beaglebone25kHzI2c Apr 26 21:51:54 zhivko, where are you getting the number, '25kHz' Apr 26 21:51:55 mdp: you get a lcd, case and an arm for 300€ Apr 26 21:51:58 ;) Apr 26 21:52:08 LOL Apr 26 21:52:12 So now you have data as well - and 3 pictures -do you see them all 3? Apr 26 21:52:16 including tons of buttons Apr 26 21:52:27 yes Apr 26 21:52:29 zhivko, where are you getting the number, '25kHz' Apr 26 21:53:52 aholler, it sounds really bad when you conveniently omit the 1GS/s ADCs and the amps :) Apr 26 21:54:06 russ - if you are checking CLK signal it went 4 times up and down in 10us Apr 26 21:54:14 mdp: I rarely have a use for that, I'm a softie ;) Apr 26 21:54:35 I also rarely use more than a logic analyzer Apr 26 21:54:47 for the dayjob Apr 26 21:54:53 zhivko, show me your math Apr 26 21:56:05 zhivko: move the cursors to two spikes of the clock Apr 26 21:56:19 aholler, he doesn't need to do that Apr 26 21:56:37 he just needs to show me how he got from 4 cycles / 10 us = 25kHz Apr 26 21:56:39 but then the screen would show it. Apr 26 21:57:07 so its 1/0.0000025 Hz isn't it ? Apr 26 21:58:30 :) Apr 26 21:58:56 I'd like to think that you just discovered that 4 cycles / 10 us = 400kHz Apr 26 21:59:48 you mean 100kHZ * 4? Apr 26 21:59:48 but why then the time is same when measuring i2cdetect ? Apr 26 21:59:58 Russ, are we ready to move to lesson 2? differential equations Apr 26 22:00:12 zhivko, what did you calculate that it would change by? Apr 26 22:00:49 my math shows that moving up the i2c clock to 400kHz would have a very small effect on your timing since a very small percentage of time is the actual time that data is clocking in/out Apr 26 22:01:00 * prpplague is so over the ds1307 Apr 26 22:01:06 or are you just guessing and throwing stuff at the wall Apr 26 22:01:23 the ds1307 has so many drawbacks when you consider using it with a low power device Apr 26 22:01:38 russ_ Apr 26 22:01:55 prpplague: it could make ti rich ;) Apr 26 22:02:02 prpplague, ew...it does leap year and stuff Apr 26 22:02:14 russ: i woud expect i2cdetect to be 4 times faster on 4 times faster i2c bus Apr 26 22:02:25 zhivko, then you haven't thought about the problem Apr 26 22:02:52 aholler: hehe Apr 26 22:02:56 http://www.microchip.com/wwwproducts/Devices.aspx?dDocName=en550280 Apr 26 22:02:58 the i2c clock speed only effects the rate at which data is being clocked in and out Apr 26 22:03:01 much better solution Apr 26 22:03:09 with lots of extra features Apr 26 22:03:12 zhivko, how many bits are you clocking/ Apr 26 22:03:15 and super cheap Apr 26 22:04:12 (8 bits addr + 16 bits data + 1 stop + 1 start) * 3 axis * 3 sensors? Apr 26 22:04:27 *yawn* Apr 26 22:04:35 234 bits Apr 26 22:04:51 russ: that math falls apart depending on how the driver is written Apr 26 22:05:02 <--- been up to eyeballs with sensors :/ Apr 26 22:05:09 ds2, I'm just talking about the portion of the time that data is clocked in/out Apr 26 22:05:26 ds2, to calculate the time difference for running at 100kHz vs 400kHz Apr 26 22:05:39 Russ: that is a start but that usually turns out to be such a small portion of measured time Apr 26 22:06:03 aholler: ds1307 is around $1.83 is 100qty, the MCP79410 is just $0.79 is same qty Apr 26 22:06:12 aholler: and MCP79410 requires no level shifters Apr 26 22:06:24 aholler: plus you get eeprom Apr 26 22:06:25 prpplague: I know, I just used because I have that breakout here. Apr 26 22:06:33 aholler: ahh Apr 26 22:06:34 dallas has a 1.8V RTC Apr 26 22:07:00 ds2: with eeprom? Apr 26 22:07:09 prpplague: no, but with xtal Apr 26 22:07:17 hehe Apr 26 22:07:20 can't find my notes... looked it up last week Apr 26 22:07:25 so at 234 bits at 100kHz = 2.34ms Apr 26 22:07:35 eeprom don't matter for what I was looking for. internal xtal was more important Apr 26 22:07:39 at 400kHz, 0.585ms Apr 26 22:07:40 prpplague: my next thingy will be that one: http://www.watterott.com/en/MOD-RTC?x30a3c=3554d7b9b9580a231c79685d0bbe67fd Apr 26 22:08:01 zhivko, is it running 1.755ms faster? Apr 26 22:08:20 aholler: don't forget to order a few items here - http://www.watterott.com/de/Tin-Can-Tools?x2f390=4c7e5c20c3ad69d2ee19ae6ebfd0b089 Apr 26 22:08:27 prpplague: you wouldn't happen to know of a RTC with on board battery backup, on board xtal, and 1.8V I/O, would you? Apr 26 22:08:45 ds2: nope Apr 26 22:08:58 ds2: i never really looked for one with onboard xtal Apr 26 22:09:08 prpplague: why? Apr 26 22:09:11 you probably want something with on die mems Apr 26 22:09:17 if it exists Apr 26 22:09:32 Russ: mems? Apr 26 22:09:36 for an RTC? Apr 26 22:09:39 ds2: we use the xtal on a bunch of projects so we have close to about 100K in stock Apr 26 22:09:44 http://www.maxim-ic.com/datasheet/index.mvp/id/6861/t/al Apr 26 22:10:00 russ: I read 6 bytes from each device each time it is called...which is once every 1ms Apr 26 22:10:05 prpplague: ah... I was thinking more of mfg costs... more parts == more assembly costs Apr 26 22:10:08 Russ: got my purple package today Apr 26 22:10:13 prpplague, cool Apr 26 22:10:23 inventory is secondary Apr 26 22:10:26 zhivko, don't forget start bits, stop bits, and address Apr 26 22:10:37 ds2: yea, in this case the inventory offsets the assembly costs Apr 26 22:11:02 prpplague: *nod* except now you need to keep reels of those 20pf caps around too Apr 26 22:11:07 and board space for them Apr 26 22:11:08 ds2: plus the mcp part is so cheap, and takes the place of a number of components Apr 26 22:11:23 ] Apr 26 22:11:25 russ so it's 30bytes max each ms Apr 26 22:11:33 ds2: that pcf8563 is usable with 1.8v Apr 26 22:11:47 zhivko: you probally don't want to be poking the sensors at 1000KHz Apr 26 22:11:49 Russ: thanks i'll test and talk to rusty Apr 26 22:12:05 prpplague, I realized after I sent them that I forget esd protection diodes Apr 26 22:12:49 aholler: that's a hella lot of board space Apr 26 22:13:08 zhivko, Add in the delay between writing a command and reading back the reply, I don't think you can fit that amount of transfer no matter what you do Apr 26 22:13:24 got that + xtal + caps Apr 26 22:13:48 use the onboard fifo's Apr 26 22:13:56 also, you are reading 6 bytes from each device, do you also need a command and register address byte along with the address? Apr 26 22:14:01 and poke at it with a burst read at a slower rate Apr 26 22:14:14 1000KHz polled sample rate is going to be painful to get right Apr 26 22:14:18 Russ: we are running same code! in mbed without seeing any problems there - but values on beagle are drifting. Apr 26 22:14:29 mbed? Apr 26 22:14:33 We are reading from IMU Apr 26 22:14:38 imu? Apr 26 22:14:52 Russ: yea ? Apr 26 22:15:02 yea? Apr 26 22:15:30 zhivko: why are you insisting on banging your head on the wall? Apr 26 22:15:36 use the hw fifos Apr 26 22:16:45 the DS3232M is pretty awesome, too bad it isn't 1.8v Apr 26 22:16:54 ds2 - so you think beagle is not meant to be used to read data from i2c device (so fast) or what? I need subtitles :) Apr 26 22:17:22 zhivko: I am saying, use the hw that is on the sensors. Apr 26 22:17:41 only off by around 1 second every 2 days Apr 26 22:18:11 zhivko: the beagle can do that just fine. I have working drivers on the board. there is no reason to chase what you are chasing down Apr 26 22:19:09 ds2: just wanna get where this discrepancies come out comparing mbed and beagle Apr 26 22:19:24 ds2: what do you think I am chasing down ? Apr 26 22:19:28 Russ: whoa... you really do mean MEMs Apr 26 22:19:44 zhivko: per sample latency at 1KHz Apr 26 22:20:15 what drifts? Apr 26 22:20:28 ds2, we live in the future Apr 26 22:20:41 what's an mbed? Apr 26 22:20:56 russ: google? Apr 26 22:21:21 Russ: I been mostly working with mics, gyros, accels, etc Apr 26 22:21:31 zhivko, you are using bone, right? Apr 26 22:21:32 the mbed runs linux? Apr 26 22:21:42 Russ: yes Apr 26 22:21:47 aholler: no Apr 26 22:21:58 the bone has two microcontrollers on it call pru's Apr 26 22:22:24 zhivko: so how are you able to use the same code on mbed an the bone? Apr 26 22:22:34 your other option is to instrument the i2c driver in the kernel to analyze what is going on with the timing Apr 26 22:22:40 Russ, stuff Boris is crazy about Apr 26 22:22:55 mdp, can Boris get some official looking documentation out for them? Apr 26 22:23:10 So when we read 3 sensors 30 bytes of date each ms, on mbed and we leave imu still - there is almost no drift in angle values (3d angle is what imu measures) Apr 26 22:24:08 imu? Apr 26 22:24:12 you are using the Invense chip, right? Apr 26 22:24:17 or was that someone else? Apr 26 22:24:25 If we then run same code (now on 400kHz i2c bus), the values of angles are drifting Apr 26 22:24:31 ds2: yes Apr 26 22:24:34 Russ, Boris is only capable of hitting +1, he's a dog Apr 26 22:24:38 "drifting"? Apr 26 22:24:38 then use their software Apr 26 22:24:49 they have it all written Apr 26 22:24:54 IMU = Inertial measurement unit Apr 26 22:25:24 why not just slow down the sample rate then Apr 26 22:25:36 drifting= values are sliding - they are not still... :) Apr 26 22:25:47 how is that a latency problem Apr 26 22:26:10 are you building something dangerous that the powers to be would not like? Apr 26 22:26:58 I would bet something like a quadcopter Apr 26 22:27:11 russ: wanna build some stabilizing system. Apr 26 22:27:36 russ: and must run fast Apr 26 22:28:37 if you want to use the in-kernel i2c driver and it isn't meeting your timing expectations, you'll need to instrument it Apr 26 22:28:55 otherwise, you can use the PRUs Apr 26 22:31:33 ross: mbed has an API that does RAW io and that is only difference - code is same - beagle has linux overhead :) Apr 26 22:32:25 Russ: PRUs ? Apr 26 22:32:32 the microcontrollers on the bone Apr 26 22:33:03 zhivko, i2c is typically not something where you need 1kHz update times Apr 26 22:33:39 zhivko, if there needs to be a delay for 1uS, the driver will typically just put a sleep instead of a busywait, which will probably be a 30 some ms delay Apr 26 22:33:41 Russ: that's why the chip vendors provide fifos Apr 26 22:33:56 ds2, well you should tell him to check out the hardware fifos Apr 26 22:34:02 :p Apr 26 22:34:11 Russ: I have said about 5 times already. Apr 26 22:34:12 only difference is execution time since the sensors are going to produce the same values on the beagle, the processor must pass what it sees up through the Kernel to the driver the driver must write to a file and applications reads it Apr 26 22:34:20 clearly, this person don't want things to work. Apr 26 22:34:34 zhivko, have you even looked at the i2c section of the TRM or the driver written for it? Apr 26 22:34:54 i2c is not complicated and the drivers are not that complex Apr 26 22:35:48 Russ: no - sory but if it will solve it I am more that willing to :) Apr 26 22:36:16 but it doesn't have subtitles Apr 26 22:36:33 uh oh Apr 26 22:36:41 and sometimes it's incorrect too Apr 26 22:37:02 zhivko, you really need to look at what is going on in the driver, or if you are more comfortable with microcontrollers, use the PRUs, they are probably a smart thing to have on a system like your anyway Apr 26 22:37:31 the whole point of those Invense chips is the hardware handles these problems Apr 26 22:39:47 ds2: - which sample should I check ? itg3205 doesn'have any special functions I think - I mean you can read those 3 *2 bytes and that's it - correct me if I'm wrong please Apr 26 22:39:58 maybe try starterware instead Apr 26 22:40:00 zhivko: look at the datasheet Apr 26 22:40:13 yea I'm loooking :) Apr 26 22:40:23 and if you are doing a IMU, you should be really using the MPU3050 part Apr 26 22:40:29 it should be a pin compatible setup Apr 26 22:41:19 the MPU3050 has hw to sync samples and do other magic so you don't have to be exactly on the spot with poking at the chip Apr 26 22:41:47 ah, that's what he means by drift Apr 26 22:42:11 or go with the MPU6050 with more parts integrated Apr 26 22:42:12 ds2: OK you are talking now about another piece of hardware: MPU3050 instead of ITG3205 ;) Apr 26 22:42:28 hehe :) or by helicopter ;) Apr 26 22:42:42 zhivko: you are talking about IMU's so, one would think you are using the hw that is meant for that Apr 26 22:43:15 otherwise, why bother with the Invense part? there are nicer gyros around Apr 26 22:43:27 I bought all 3 sensors in breakout board for 30$ on ebay ;) Apr 26 22:43:45 blah... use decent hw Apr 26 22:43:56 it is probally $30 on ebay cuz someone f'ed up Apr 26 22:44:02 you could also use multiple i2c busses Apr 26 22:44:06 ds2: OK give me some cost effective solution ;) obviously itg3205 has problem with beagle ;) Apr 26 22:44:20 regardless what you are doing, you cannot truely sync up things on all 9 axis Apr 26 22:44:28 it doesn't matter beagle or not. Apr 26 22:45:04 talk to invense... maybe a MPU6050 + a recommended magnetometer Apr 26 22:45:15 plenty of options available Apr 26 22:45:24 what do you estimate the cost of this combination ? Apr 26 22:45:40 depends on volume obviously Apr 26 22:46:03 Russ: multiple i2c-s on beagle ? Apr 26 22:46:34 i2c is the least of your problems Apr 26 22:46:50 which magnetometer and accelerometer are you trying to use? Apr 26 22:47:13 zhivko: you are tempting me to dig out my old nintendo gyroscope and play. Apr 26 22:47:32 adxl345 hmc5883L and itg3200 Apr 26 22:48:20 is that an analog part or a digital out part from ADI? Apr 26 22:48:38 IIRC, the HMC5883 requires you to request a conversion Apr 26 22:48:51 so you'll need to sync up the sampling yourself... Apr 26 22:48:52 djlewis: it's cool idea - but I currently lack some drystones :) kernel overhead is killing me I mean this mbed - runs on Apr 26 22:50:18 zhivko: i havent seen the mbed code. does it average each read? Apr 26 22:50:29 ds2: We have same hardware, same code , only beaglebone replaces mbed - and we have instability in readings ;) Apr 26 22:51:03 this is getting annoying Apr 26 22:51:09 it is not the bone Apr 26 22:51:19 the data just looks less like garbage Apr 26 22:51:52 code is on http://code.google.com/p/uav-helicoptor-ece4007/ Apr 26 22:52:58 * djlewis wonders it the bone is giving more detailed readings? Apr 26 22:53:10 which are correct for that moment in time.. Apr 26 22:54:06 my bet is that the i2c-omap.c driver is written to minimize cpu usage rather than latency Apr 26 22:54:08 how do you mean more detailed ? Apr 26 22:54:31 there is nothing wrong with the i2c-omap.c driver Apr 26 22:54:54 zhivko: looking at: http://aeroquad.com/showthread.php?5191-beaglebone-and-itg3205-problem-%28gyro-values-unreasonable%29 Apr 26 22:55:07 it looks like you pick it up at the last 5 lines Apr 26 22:55:21 w/o doing much work, a 80-150Hz sample rate is doable... to go faster, one needs to tune a few things Apr 26 22:55:30 +1 Apr 26 22:55:31 djlewis: ok I solve that already ;) Apr 26 22:55:35 ds2, he's looking for 1kHz Apr 26 22:56:19 Russ: I know... I have those gyros and mags working easily at the slower rate. 1KHz is possible but you'll get junk back unless you get the hw to cooperate (hence the MPU3050 suggestion) Apr 26 22:56:30 Russ: the delays is not coming from I2C Apr 26 22:56:36 ah Apr 26 22:57:16 Russ: what's why I was saying your numbers is a start but that's not the bottle neck....to get down to 1KHz, you do need to up the bus speed but that is secondary Apr 26 22:57:52 IIRC, the mag and gyro might support the high speed I2C bus (higher then 400KHz)...but this is going from what I recall Apr 26 22:59:20 I think 400kHz is max for gyros and mag Apr 26 23:00:20 also there are two i2c busses brought out Apr 26 23:00:39 so one bus to one set of sensors, the other to another Apr 26 23:00:52 of if you end up using the pru, one pru per bus Apr 26 23:01:13 * djlewis is heading home :) Apr 26 23:01:15 russ: I have breakoutboard - only one i2c bus Apr 26 23:01:24 *shrug* possible... don't have the public datasheets on hand Apr 26 23:03:01 ds2: what could cause this drifting of data on beagle - do you have some idea ? I see you worked with similar projects in past Apr 26 23:03:43 zhivko: define "drifting" Apr 26 23:04:01 maybe I should submit an elc-e proposal for PRUs Apr 26 23:04:10 sample's not matching up btwn the different sensors or? Apr 26 23:04:18 s/sample's/samples/ Apr 26 23:04:48 ds2: no it looks like some error is acumulating and angle is not holding betwin 1 degree when imu is still. Apr 26 23:05:08 ds2: but same algoryth on mbed... I don't have to repeat :) Apr 26 23:05:23 you probally have problems with your math Apr 26 23:05:24 ds2, are you also confused why the problem is being presented as being a latency issue and not a read issue? Apr 26 23:05:40 zhivko: other obligations prevents me from saying more. sorry. Apr 26 23:06:20 Russ: I think I know what is being described. Apr 26 23:06:33 dead reckoning and it is not integrating to zero when it is not moving, right? Apr 26 23:10:06 Russ: have you done anything specific with the PRU? Apr 26 23:26:23 sorry for such a noob question, but after running bitbake via "MACHINE=beaglebone bitbake systemd-image", where is the .gz or xz image? or is there some other command to make those? Apr 26 23:28:11 I did some i2c bechmark on beagle and I am seeing that for 10 READS it takes 0s 4ms Apr 26 23:28:42 that is LOOONG time Apr 26 23:31:20 russ: when the bus you are running on delivers 1 byte in 2.5us it takes over 100 times longer to deliver the byte to the application than it takes the i2c bus to deliver it Apr 26 23:32:20 ds2: do you know how to solve this dead reckoning? Apr 26 23:40:53 ds2, not yet, but if I submitted a proposal on it, I'd learn about it or have nothing to present Apr 26 23:41:29 zhivko, like I said, if you want to figure out why, you'll have to do some digging Apr 26 23:43:17 you mean rewrite i2c_driver ? Apr 26 23:48:44 zhivko: yep. Apr 26 23:49:15 ds2: can you go a little bit more in detail? pm maybe ? Apr 26 23:49:44 have obligations that prevent that Apr 26 23:49:55 there are plenty of guys in that area solving that problem Apr 26 23:51:09 ds2: I checked mpu-9150 at: http://video.designworldonline.com/channel/b8c3046087cf4ac39357b3f212cf98cd - and I am start thinking that maybe it's better to buy more integrated solution. Apr 26 23:51:48 "Available Q2 2012" Apr 26 23:52:15 most of the MPU line have hardware inside to help and they have sw to help with all this Apr 26 23:52:39 other vendors have sw to solve this on other chip in general Apr 26 23:54:27 Russ: that would be a great session Apr 26 23:56:00 Russ: you got someone to cover travel and lodging for ELC-E? Apr 26 23:58:21 it'd have to be TI Apr 26 23:58:40 ds2 - is there any integrated IMU available already ? Apr 27 00:00:08 zhivko: Full 9 axis in a single chip... not that I know of... 6 axis, there are plenty... the MPU series will look like an integrated IMU to SW Apr 27 00:19:56 does anyone know of any problems with calloc on bb? Apr 27 00:27:47 niro: what problem do you think you're seeing? Apr 27 00:28:40 mru: empty pint glass.... thats the problem i am seeing Apr 27 00:30:36 mmm beer Apr 27 00:35:12 prpplague: you need quart glasses ! Apr 27 00:35:20 or the yard glasses Apr 27 00:37:23 the yard house? Apr 27 00:38:22 litre glasses **** ENDING LOGGING AT Fri Apr 27 03:00:00 2012