**** BEGIN LOGGING AT Wed Mar 15 03:00:03 2017 Mar 15 05:22:58 Hi guys, has anyone worked with the MPU-9250? I'm not getting any reading changes on the magnetometer while the gyroscope and accelerometer are working fine. Any advice on this? Mar 15 06:02:51 Hello , I have running firmware and it is working fine. in between i am unable to communicate with beaglebone. I am unable to do ssh or either i am not getting data on debug console. Anyone have idea that what could be the reason? Help appreciated Mar 15 06:07:34 Hi KK, what are you trying to run on the beaglebone? Mar 15 06:08:02 i mean do you have adevice connected to it or? Mar 15 06:28:14 yes there are few cameras connected which is recording streams. Mar 15 06:29:06 some how in between it stops recording and i think it stops all the services so i can not communicate to beaglebone Mar 15 06:29:24 usb connected? Mar 15 06:30:13 No external USB device connected Mar 15 06:30:29 so how are the cameras connected? Mar 15 06:31:23 we are using external board to control cameras using POE switch Mar 15 06:31:52 beaglebone is just processing camera recording Mar 15 06:32:06 reading and writing bin files Mar 15 06:32:08 so that is coming in via what? the network? Mar 15 06:32:17 yes Mar 15 06:32:59 do you have the debug console connected before that happens? Mar 15 06:34:08 no Mar 15 06:34:52 then I'd suggest to do that, make sure the kernel console is turned on for that port and catch possible output leading up to the device becoming unresponsive Mar 15 06:34:58 after that happened i dont have access to debug console Mar 15 06:35:44 that's why I said *before* Mar 15 06:35:58 ok got it Mar 15 06:36:08 thanks i will try it Mar 15 06:36:13 also what is the behaviour of the LEDs? Mar 15 06:37:04 device is pinging but unable to ssh device Mar 15 06:37:30 device is on other location so i have to check Mar 15 06:37:42 about LED Mar 15 06:38:07 focus on collecting UART output leading up to the event Mar 15 06:38:18 also which kernel version are you running on the board? Mar 15 06:38:45 2.6.32 Mar 15 06:41:59 i noticed that kern.log also dont have any information about it Mar 15 06:42:24 previously i collected last kern.log info Mar 15 06:53:32 what for the mercy of the flying spaghetti monster! Mar 15 06:53:42 that's like stone age Mar 15 06:54:30 nobody will be able to help you. Mar 15 06:54:51 I'd strongly suggest to move to an up to date image and kernel Mar 15 06:54:55 *STRONGLY* Mar 15 06:55:24 or at least something from the last ... 5 years or so. Mar 15 07:01:58 Ok thanks Mar 15 08:10:04 Hi AV500 Mar 15 08:11:18 AV500: i'm having a problem with reading the data of the magnemoter on the MPU-9250. Can you assist? Mar 15 08:12:29 Hi. We are trying to run WinCE on BeagleBone Black using this instruction https://beaglebonebsp.codeplex.com/documentation Mar 15 08:12:59 And WinCE didn't launch Mar 15 08:14:11 We also tried launch linux. And it has been launched. Mar 15 08:24:10 Yartuning: sorry, I think few people here use WinCE Mar 15 08:24:39 also that info is 3 years old Mar 15 08:25:02 Aryan: I know little about the MPU-9250 Mar 15 08:31:46 @av500: ok then Mar 15 08:32:24 can Anybody advise on the MPU-9250? Mar 15 08:32:28 but then, you could describe your problem Mar 15 08:32:35 just stating "it does not work" wont help much Mar 15 08:32:42 we dont see you desak Mar 15 08:32:44 desk Mar 15 08:32:48 nor your code Mar 15 08:38:06 Av500: i have my MPU connected to the beaglebone, interfacing with the Debian 8.6. I can see the data changes of the Gyroscope and accelerometer but not of the magnetometer. Mar 15 08:38:36 Av500: the magnetomer data are showing zero all the time Mar 15 08:39:22 I haven't written any code yet. first wanted to access the data. Mar 15 08:50:24 if you have no code, how are you reading the data? Mar 15 08:55:01 ...Pass-Through mode is also used to access the AK8963 magnetometer directly from the host. In this configuration the slave address for the AK8963 is 0X0C or 12 decimal.... Mar 15 08:57:46 Av500: using the linux command i2cdump Mar 15 08:58:14 did you read the data sheet? Mar 15 08:58:22 the magneto is not on the same bus Mar 15 08:58:26 but on the aux bus Mar 15 08:58:54 Av500: i can see the device at that adress, but the issue is about the data Mar 15 08:58:58 i didnt read further, but it might be you need to poke some registers Mar 15 08:59:16 well, maybe you need to configure stuff Mar 15 08:59:29 the data stays 0x00 Mar 15 08:59:33 well, maybe you need to configure stuff Mar 15 08:59:55 ok, because I have ben trying to do that in vain Mar 15 09:00:42 well, see the data sheet Mar 15 09:00:49 the magneto has a control register Mar 15 09:00:54 with several modes Mar 15 09:01:00 one of them is power down Mar 15 09:01:41 and its the default state ofter reset Mar 15 09:01:49 so 0x00 is not a surprise Mar 15 09:07:24 Av500: ok will try to change the modes and see what happens Mar 15 09:14:23 #exactsteps Mar 15 09:35:16 Ok. But maybe anybody knows where can i find help with WinCE? Mar 15 09:42:31 microsoft? Mar 15 10:02:07 Av500: none of those mode induce a change Mar 15 10:33:31 Hi All! Mar 15 10:33:47 any help how to setup wifi with BBB Wireless? Mar 15 11:47:34 Aryan: you're testing this using i2cget ? Mar 15 13:22:05 Does anyone here have makefile in g++ Mar 15 13:23:36 I have a problem with makefile in Blacklib library , my beagle only have g++ compiler can you help me to sole out pls Mar 15 14:15:25 Are there any .img files available for Beaglebone Debian Jessie and a Mainline kernel? Mar 15 14:17:36 PLEASE **** BEGIN LOGGING AT Wed Mar 15 16:22:12 2017 Mar 15 16:26:44 Please confirm ECCN for BB Blue Mar 15 17:47:33 hello! can anybody guess typicall latencies of changing PWM duty_cycle on a BBB? Mar 15 17:48:43 https://hackaday.com/2013/12/07/speeding-up-beaglebone-black-gpio-a-thousand-times/ this suggests that it is probably smaller than 1ms. Mar 15 17:52:21 for my application case it might be not enough. how can i disable the file tree overlay, so that i can use it by directly writing to memory registers? Mar 15 18:02:09 Does anyone know how to boot Jessie version BBB to a console? I tried "systemctl set-default multi-user.target", but apparently it didn't work Mar 15 18:52:59 cash: there is a headless image Mar 15 18:53:25 cash: that should work actually Mar 15 18:54:59 nextloop: I actually use PWM for generating a ws2812 signal :) Mar 15 18:55:27 though I'll admit that was not completely trivial (and in retrospect I really should have used PRUSS instead) Mar 15 18:56:36 so that means 800 kHz frequency, duty cycle changed each period Mar 15 18:57:16 but I cheated, I setup dma to update the duty cycle reliably ;) Mar 15 19:04:33 zmatt: so i don't do it with the sysfs overlay? Mar 15 19:04:51 did you disable the file tree overlay? so it does not interfere with DMA? Mar 15 19:14:30 Hello everyone Mar 15 19:15:57 how much faster is the ADC port of beaglebone black? Mar 15 19:20:27 a lot! Mar 15 19:23:34 faster... than? Mar 15 19:24:06 nextloop: dma is a complicated subject, you're unlikely to need it unless you're doing something as crazy as I did Mar 15 19:25:37 nextloop: depending on how the kernel driver is written, you may be able to leave it in place and then just overwrite the duty-cycle register via /dev/mem or UIO Mar 15 19:25:45 that has the benefit that the kernel performs all initialization for you Mar 15 19:27:19 oooh Mar 15 19:28:15 or, you can use UIO instead of the normal kernel driver... if you setup device tree right that has the benefit that the kernel will automagically ensure the necessary clock signals are enabled when you open the device Mar 15 19:28:45 or you can do it yourself entirely via /dev/mem... I don't recommend this Mar 15 19:29:00 (especially in case of eHRPWM) Mar 15 19:29:44 as for how fast that is, I was curious myself so I'm doing a quick benchmark Mar 15 19:40:42 oh, brb Mar 15 19:41:47 sorry, was afk unfortunately. as for benchmark, zmatt, did you have a look at my link. they talk about factor 1000 between sysfs and DMA. Mar 15 19:42:23 i need it for pwm output for hexacopter motors Mar 15 19:43:01 I doubt they compared against DMA Mar 15 19:43:54 several ms just for updating the pwm duty_cycle might be a problem. as i need to read sensors and to calculations too Mar 15 19:44:40 ./dev/mem = DMA? zmatt? Mar 15 19:44:44 no Mar 15 19:45:22 although I can understand the confusion :) Mar 15 19:45:41 since one might very well say that via /dev/mem one is using "direct memory access" Mar 15 19:45:49 but DMA already means something very specific Mar 15 19:46:12 that's when the transfer is performed by hardware rather than software Mar 15 19:46:26 (e.g. based on some trigger) Mar 15 19:47:38 okay. zmatt. then they compare it to /dev/mem ;-) Mar 15 19:47:44 http://chiragnagpal.com/examples.html Mar 15 19:47:58 and https://stackoverflow.com/questions/13124271/driving-beaglebone-gpio-through-dev-mem Mar 15 19:48:18 hold on, having a tiny fight with gcc since some stuff changed apparently and broke a clever hack I did a while ago Mar 15 19:51:47 the "ping time" to pwmss is about 190 ns Mar 15 19:54:49 if you use a standard kernel that's what any read/write of a pwmss register via /dev/mem or UIO would cost Mar 15 19:55:51 1, 2, or 4 bytes. an 8-byte load/store costs 20 ns extra Mar 15 19:58:55 that's fast enough Mar 15 19:59:02 going over sysfs probably isn't Mar 15 20:00:01 ah, funny, the duty cycle register of the eHRPWM module inside PWMSS is a bit slower to access than the version register of PWMSS which I used for the previous tests Mar 15 20:00:48 200 ns instead of 190 ns .... I've actually only measured *read*, but write should be basically the same on standard kernels Mar 15 20:01:34 on my patched kernel writes cost 60ns if you rapid-fire them Mar 15 20:02:19 but they get queued, so a write basically just costs 1 cycle as long as you don't fill up the queue :) Mar 15 20:05:14 zmatt: you know IIO? Mar 15 20:19:29 well I know that it exists :) but haven't spent much attention on it yet Mar 15 20:45:07 ayjay_t: i know Robert had been using up to dec. 2016 Mar 15 20:49:03 a composite device isnt really the same as emulating more than one device (i.e. acting like a hub) Mar 15 21:32:06 Does anyone know the io voltage tolerance on the beaglebone blue? I can't find any info in the specs (I assume 3.3V since it's an A8) Mar 15 21:56:02 psidhu`: the core has nothing to do with the I/O supply Mar 15 21:56:45 I don't really know about the blue specific Mar 15 21:57:09 I don't really know much about the blue specific, but all beaglebones so far have used 3.3V for digital I/O and 1.8V for the analog inputs Mar 15 21:58:30 if I understand correctly though the Blue also has motor drivers, which of course will have separate specs Mar 15 22:00:41 zmatt: you're right, core's don't. But i've just noticed over the past that ARM cores have digital io's at 3.3V (so the companies that use those core set the limit). Mar 15 22:01:46 sorry but it doesn't work that way... TI put in special effort to still support 3.3V in their SoCs, but the chips really don't like it Mar 15 22:02:01 the omaps, including the cortex-a8 based omap3, had 1.8V/1.2V (selectable) I/Os Mar 15 22:02:14 *have Mar 15 22:04:11 the cortex-a8 on the AM335x runs on 1.1V-1.35V repending on clock frequency, the rest of the core logic runs at 1.1V Mar 15 22:04:53 sure, I see what you're saying, but I can pick out any number of chips from various companies, and they'll support 3.3 for certain arm lines. I'm just telling you what i've seen in the field and in my experience, nothing more. Other's will support only 1.8 and others may support 5 Mar 15 22:05:06 "certain arm lines" is very vague Mar 15 22:05:24 A6-A8 usually, M0's i've seen vary Mar 15 22:05:30 it's a generalization for a reason, lol Mar 15 22:06:56 3.3V support still seems popular yes, except for mobile devices Mar 15 22:07:30 yeah, that seems right. Mar 15 22:08:19 tricky part is that afaik in the IC process used by the am335x, the abs max voltage across a single transistor is 2V :-) Mar 15 22:09:14 yeah, smaller you go, the less v you want Mar 15 22:09:26 too much crosstalk / other issues I think otherwise Mar 15 22:09:48 note that the AM335x actually groups its digital I/O into 6 voltage domains, and when you design a board you can choose for each voltage domain whether to power it at 3.3V or 1.8V Mar 15 22:10:05 Yeah, that's a fairly common thing to do Mar 15 22:10:09 crosstalk? no I mean physical damage Mar 15 22:10:18 Oh, even worse Mar 15 22:13:09 of course there are clever tricks, with cascaded transistors and such Mar 15 22:13:28 intermediate bias supplies Mar 15 22:14:32 note that the schematics and such of the blue are in github already Mar 15 22:14:40 i'm glad I never got into CMOS design for a living, haha. Mar 15 22:14:42 so if you want to check specs of components you can Mar 15 22:14:49 Oh, I hadn't noticed that, I'll check it out. Thanks Mar 15 22:14:55 https://github.com/beagleboard/beaglebone-blue Mar 15 22:16:57 looks like their digital stuff is 3.3 Mar 15 22:20:15 yay I see they managed to make a proper schematic export now, with real text Mar 15 22:20:45 instead of the horrible vector-drawn fonts you still see far too often used by EAGLE schematics Mar 15 22:20:49 i've never used a beagle product before; did they have issues before? Mar 15 22:20:57 EAGLE, the schematic program used Mar 15 22:20:59 ahh, yeah Mar 15 22:21:12 eagle can be difficult to make a nice looking schematic Mar 15 22:21:32 it's not just appearance, with the vector font you can't search for anything and can't copy-paste anything Mar 15 22:21:47 yeah Mar 15 22:21:49 that's true Mar 15 22:22:48 and yes the OSD335x uses 3.3V for all digital I/O, no choice Mar 15 22:27:31 ugh I really don't like these kinds of "puzzle-adventure" schematics... where everything is connected using named nets instead of a line, so if you try to follow a signal flow you need to search where the hell it went this time :P Mar 15 22:28:28 zmatt: you no fun .. :P Mar 15 22:43:21 not really clear why they need that S-8261 in addition to the BQ29209 Mar 15 22:48:47 apparently just for voltage monitoring, since VSS and VM being tied together renders its current monitoring non-functional Mar 15 23:21:59 can i make my own version of Beaglbone? Mar 15 23:22:48 is there sommeone? Mar 15 23:23:25 yes and yes Mar 15 23:23:58 greate Mar 15 23:24:48 can you tell me the max charging curent for the 2s lipo on the bbblue Mar 15 23:25:30 ? Mar 15 23:32:13 I have no idea, check the documentation in the github repo maybe? Mar 15 23:32:31 or if necessary the schematic Mar 16 02:43:30 hello friends! **** ENDING LOGGING AT Thu Mar 16 03:00:01 2017