**** BEGIN LOGGING AT Thu Mar 12 02:59:58 2015 Mar 12 03:31:26 woot, directly accessing the ADC works Mar 12 03:32:11 (via /dev/mem) Mar 12 03:33:33 now, time to hook that up to edma and see if the adc can really deliver 1.6 Msps Mar 12 03:45:37 What ENOB are you shooting at err for? Mar 12 03:47:27 * zmatt stares at that sentence but keeps getting "parse error" as result Mar 12 04:43:43 aaand then suddenly the adc's FSM got stuck o.O **** BEGIN LOGGING AT Thu Mar 12 05:03:11 2015 Mar 12 05:03:56 actually, I'll be back on in the morning. I tried to post in the forums on BeagleBoard.org, but it would not let me (I registered last year) Mar 12 05:05:02 looking for some PRU initialization expertise. I have everything working well, c compiler, prussdrv, uio_pruss. the issue is initializing R30 before the pins are mux'd Mar 12 05:05:50 there is no info coming out of TI other that one tiny mention in one document that says it should be done. Mar 12 05:06:28 I don't have any experience with the tools, but iirc you can read/write the core registers when they are halted Mar 12 05:07:40 the documentation is somewhat sketchy about whether R30 is accessible from outside the PRU. it seems like its mapped to a debug register. so I am hacking in that general direction. Mar 12 05:08:20 but it needs to happen before the pins are muxed, so that may mean in u-boot. which is tough to debug.... (for us mortals) Mar 12 05:09:53 you can of course leave the pins initially muxed to gpio and change them to pru later on Mar 12 05:10:34 that was another idea. is there a cheat-sheet for poking the pinmux from the user application? Mar 12 05:10:59 the TRM has all the info, but its pretty buried sometimes. Mar 12 05:11:44 well my jbang project ( https://github.com/dutchanddutch/jbang ) happens to poke at pinmux registers from userspace (though not to actually change pinmux, but it's easy enough to adapt) Mar 12 05:11:59 worst case, I can load a PRU program that resets the pin states, then switch the pinmux. (this is backup-plan #2) Mar 12 05:12:45 interesting project. Mar 12 05:12:48 you can find mux options by cross-referencing the TRM and datasheet, or more conveniently my spreadsheet -> https://docs.google.com/spreadsheets/d/1CK5c-Cs8G1RtzGo-J3VJsD9m5K-fp06AncgeYWsdjSU/view Mar 12 05:13:26 the "BBB" tab also has a filter view that shows just the pins accessible on the BBB (menu Data -> Filter views -> CAPE) Mar 12 05:13:38 the TI pinmux tool can create some of this info as well Mar 12 05:13:41 the BBB2 tab shows similar info Mar 12 05:14:38 it seems odd that this situation is not discussed commonly, especially considering the linuxCNC project and other PRU based things. Mar 12 05:15:26 although the PRU seems to be considered black magic, and the TI tools are klunky at best. Mar 12 05:15:31 like I said, I haven't looked at the PRU tools yet Mar 12 05:15:40 unfinished, undocumented, unsupported..... Mar 12 05:16:04 I'd normally expect them to be supposed to take care of such a thing, since pinmux is normally done after the peripheral is initialized Mar 12 05:16:20 and setting R30 would be part of that initialization Mar 12 05:16:28 yeah, me too. Mar 12 05:16:49 thus my surprise when there is nothing in the documentation. Mar 12 05:17:12 its mentioned in PRUSS Software Mitigation Guide for AM335x Mar 12 05:17:57 Note the R30 and R31 registers should be initialized before releasing the PRU from HALT. Also, to avoid driving uninitialized values on GP pins configured as OUTPUTs, the register write of this initialization must complete soon after reset, but before setting the SoC level Pin Mux to OUTPUT values from PRU"" Mar 12 05:18:28 well, I'm not too impressed by the quality of TI's software Mar 12 05:18:42 that is the ONLY mention from TI, and for that matter anyone else (beagleboard.org, E2E, etc.) Mar 12 05:18:57 other that this ONE post: Mar 12 05:19:25 http://e2e.ti.com/support/arm/sitara_arm/f/791/p/379070/1335516 Mar 12 05:20:16 one user comes up with a solution (I have not gotten it to work yet). TI replies with a virtual "whatever" Mar 12 05:20:55 hmm, ok with the ADC running at 800 Ksps (167 ns sample time) the inputs definitely need to be driven with lower impedance than 10 Kohm Mar 12 05:22:19 I have not found conclusive info on whether the device tree is loaded by u-boot, but initializing R30 within u-boot should do the job. Mar 12 05:23:03 device tree is loaded into memory by u-boot, and some last minute patch-ups are done, but other than that u-boot doesn't interpret device tree afaik Mar 12 05:23:50 initializing R30 within u-boot means you'll first have to take PRUSS out of reset though Mar 12 05:24:06 relative to your ADC comment, we are seeing lots of jitter on our ADC reads, perhaps its related to the sample time, thats next weeks bug-chase. Mar 12 05:24:59 yeah, thats what the user that posted on E2E did. I have not gotten it to work, but I am headed in that direction. Mar 12 05:25:23 // PRU initialization code writel(0x2, 0x44e00c00); // PRU reset writel(0x0, 0x44e00c00); writel(0x2, 0x44e000e8); // PRU enable writel(0x2, 0x44e00c00); // PRU reset writel(0x0, 0x44e00c00); writel(0x0, 0x4a322478); // PRU0 r30 clear (output low) writel(0x0, 0x4a324478); // PRU1 r30 clear (output low) Mar 12 05:25:45 not sure if there are formatting tags on here Mar 12 05:26:06 I'm seeing a fair bit of noise too, but I've only just gotten it to work and haven't looked much at it yet Mar 12 05:26:16 when I compile that into u-boot the board hangs. Mar 12 05:26:54 I'm still a bit concerned with the fact that the ADC's FSM locked up once (requiring system reset), but I haven't been able to reproduce it Mar 12 05:27:25 we are seeing +/- 5% on our ADC reads. Mar 12 05:27:37 yeah, thats worrysome... Mar 12 05:27:51 I may have been doing something inappropriate though Mar 12 05:28:35 I now make sure I cleanly stop the ADC (disable all steps, wait for the FSM to go idle) before disabling it and changing any config Mar 12 05:29:36 yeah, I have locked up the CAN controller by doing things out of order (I'm pretty sure I know what I did) Mar 12 05:30:07 I feel very isolated from the hardware with this embedded linux stuff. I'd rather poke the registers myself.... Mar 12 05:30:11 overall I'm happy I actually got the ADC working with EDMA, all from userspace Mar 12 05:30:30 oh I ditched the ADC kernel driver Mar 12 05:30:37 thats pretty cool. Mar 12 05:31:13 fortunately the EDMA kernel driver always reserves the channel and 1 associated slot for each DMA declared in device tree, even if the peripheral is disabled Mar 12 05:31:14 yeah we have a touchscreen, so the driver is combined. we don't have time to fix it right. Mar 12 05:31:55 I have all 8 channels on continuous sampling, with EDMA copying the data into a fixed buffer in ocmc ram Mar 12 05:32:12 so I can just read the buffer to get the latest data Mar 12 05:32:15 no irqs to deal with Mar 12 05:32:45 any filtering or averaging going on? I'm afraid I will have to resort to that to fix the noise. Mar 12 05:33:01 haven't tried the averaging function yet Mar 12 05:33:09 but presumably that will help Mar 12 05:33:51 FSM of ADC stuck? Mar 12 05:34:14 ds2: yup, a permanent 0x20 Mar 12 05:34:22 even when disabled in control register Mar 12 05:34:28 the noise we are seeing is really strange, noise as bad as the ADC is reporting would be easily visible on my crummy o-scope. scope showed clean voltages. Mar 12 05:35:35 TurboBob: I wonder whether the kernel driver configures the steps to use vref+/vref- as reference inputs or vss/vdd Mar 12 05:35:36 zmatt: wow... never seen that before... Mar 12 05:35:51 the ADC has absolutely horrible noise Mar 12 05:36:06 ds2: the weird thing is, I don't think I "did" anything when it occurred Mar 12 05:36:09 w/o filtering, you're pressed to get 7 useable bits Mar 12 05:36:23 ds2: as in, it was running fine on EDMA Mar 12 05:36:40 thats about what we have been seeing, +/- 5% Mar 12 05:36:46 zmatt: oh you are DMA'ing too? is that done by the latest driver? Mar 12 05:36:47 I hooked up a wire to channel 0 (but hadn't connected the other end yet to anything) Mar 12 05:37:02 TurboBob: what is the analog range of your signal? Mar 12 05:37:10 ds2: read up, about 7 minutes ago :) Mar 12 05:37:48 the raw/real-world range depends on teh channel. we are keeping it within the 1.8v at the chip Mar 12 05:38:36 ds2: anyhow, I polled the buffer again, but got no edma completion bit set after that, so I tried to reinitialize the adc, but no go.. FSM remained stuck at the first step Mar 12 05:38:46 zmatt: the jbang project? Mar 12 05:39:06 ds2: overall I'm happy I actually got the ADC working with EDMA, all from userspace oh I ditched the ADC kernel driver Mar 12 05:39:53 zmatt: oh that... I have a mod'ed kernel driver... I have seen an odd case (always start up) where I get no IRQ's Mar 12 05:40:31 TurboBob: try keeping it in the 0.9V range... for some reason if I limit my signals to 0.9V, it comes out cleaner as if it is stuck in some pseudo diff mode Mar 12 05:40:43 ds2: a thought that occurred to me is the possibility that when I connected the wire I introduced some ESD and glitched up the peripheral Mar 12 05:40:55 but it doesn't seem very likely to me Mar 12 05:41:19 zmatt: have you verified there is no funny EDMA side effect with user space? Mar 12 05:41:49 well, the hardware is "done" so I will have to live with it.... (fix it with software) Mar 12 05:41:59 ds2: shouldn't be... the kernel driver is used to EDMA being used by "third parties" as well, and I'm only using resources which it has reserved for the adc anyway Mar 12 05:42:00 the low ADC channels supposely have some errata relating to charge leakage from the TSC options even though TSC is not used Mar 12 05:42:44 the adc is "world"-----------<22k>-----+------- Mar 12 05:43:01 22k... that's rather high impedance Mar 12 05:43:02 ------12k------GND Mar 12 05:43:24 with a .1uf at the input Mar 12 05:43:43 zmatt: I really wonder if your EDMA problem is related to my occasional lack of interrupt Mar 12 05:43:59 didn't want to load down the signal source, no room for an op-amp buffer. Mar 12 05:44:03 ds2: well it definitely wasn't EDMA... the ADC itself was stuck hard Mar 12 05:44:18 zmatt: using the reset bit fixes it? Mar 12 05:44:19 TurboBob: maybe put a cap on the adc-side? Mar 12 05:44:27 .1uf is there Mar 12 05:44:32 TurboBob: what data rate do you need? Mar 12 05:44:33 ds2: examine its sysconfig register, find it has none Mar 12 05:44:47 s.l.o.w. Mar 12 05:44:56 averaging REALLY helps Mar 12 05:45:01 and we do have a TSC Mar 12 05:45:22 yeah, going to spin a thread at .5khz and do a bunch averaging. Mar 12 05:45:26 TurboBob: what channel are you using? Mar 12 05:45:42 all the ones the TSC will let us use...... Mar 12 05:46:07 ds2: I don't understand why the kernel driver doesn't use DMA though.... the ADC even has a separate data port on the L3 for that purpose Mar 12 05:46:57 here is teh device tree for the TSC/ADC Mar 12 05:47:00 tscadc { compatible = "ti,am3359-tscadc"; reg = <0x44e0d000 0x1000>; interrupt-parent = <&intc>; interrupts = <16>; ti,hwmods = "adc_tsc"; status = "okay"; tsc { compatible = "ti,am3359-tsc"; ti,wires = <4>; ti,x-plate-resistance = <200>; ti,coordinate-readouts = <5>; ti,wire-config = <0x00 0x11 0x22 0x33>; }; adc { compatible = "ti,am3359-adc"; ti,adc-channels = <4 5 6 Mar 12 05:47:23 I guess that means we are using channels 4-7 Mar 12 05:47:39 zmatt: someone needing to write it? and manage it between 2 drivers? Mar 12 05:47:53 TurboBob: what channel is it wired to? Mar 12 05:47:57 (I didn't configure that part of this, I'm just the guy that has to fix it) Mar 12 05:48:09 this is a SW thing Mar 12 05:48:20 ds2: yeah, plus the kernel EDMA driver doesn't look very appealing either Mar 12 05:48:46 zmatt: yeah... didn't have time to fix it myself Mar 12 05:50:00 zmatt: i wonder if there is a reset in the PRCM block (still reading TRM) Mar 12 05:50:12 I just realized this is also the first time I'm actually using my edma header, rather than just using it as a repository for my notes on edma Mar 12 05:50:26 and apparently I didn't screw it up \o/ Mar 12 05:50:44 ds2: prcm has none Mar 12 05:50:54 you can try the target agent's reset bit Mar 12 05:51:14 the analogs are connected to AIN4 - AIN7 Mar 12 05:51:14 but the one time I tried that it turned out it wasn't connected to the peripheral Mar 12 05:52:58 (you can find docs on the target agents in the interconnect chapter of course... ha ha right except there's nothing there for the am335x, but see the interconnect chapter of the dm814x for example, or an omap4) Mar 12 05:53:42 damn it... i am used tothe full blown PRCM Mar 12 05:54:12 am335x's PRCM looks like whoeever did its layout was drunk Mar 12 05:54:25 you are so kind Mar 12 05:55:34 I was used to the dm814x Mar 12 05:55:56 I'm still traumatized from subsequently examining the am335x prcm Mar 12 05:57:15 like, *everything* about it is inconsistent, both internally and with the generic layout found on other omap4/5-generation SoCs Mar 12 05:57:27 the reset control bits don't even line up with their fucking status bits Mar 12 06:00:28 but peripherals normally don't have an individual reset in PRCM, only big subsystems Mar 12 06:00:29 if you really want trauma, install winCE on the beagle like i just did Mar 12 06:00:34 lol Mar 12 06:01:14 i do suspect some drinking was involved in the beagle, 0xBABE7175 in the pru for instance :) Mar 12 06:01:17 ds2: however I really don't understand why there's no reset bit in its sysconfig like any normal peripheral has Mar 12 06:01:44 the am335x looks to me like it was done rather in a hurry or something Mar 12 06:02:51 it also has a ton of ghost modules... where they backspaced some peripheral but not its clkctrl register or target agent Mar 12 06:04:23 zmatt: maybe it is undocumented/broken? Mar 12 06:05:12 I'm fairly certain the am335x doesn't have PCIe Mar 12 06:05:23 also given that all pins are accounted for Mar 12 06:05:33 you decapped it and verified? Mar 12 06:05:44 maybe it does but the pins are not bonded out Mar 12 06:06:12 except the am335x (for whatever reason) has padconf registers for literally _everything_ Mar 12 06:06:42 sounds like unchecked copy-paste from $somewhere Mar 12 06:06:45 even the power-on reset pin has one Mar 12 06:06:49 yes, centaurus Mar 12 06:06:59 subarctic is very clearly centaurus-derived Mar 12 06:07:30 (except for the prcm) Mar 12 06:07:39 with the investment to do a chip, it could be there are 3-4 different chips that share the same die but with minor tweaks Mar 12 06:08:07 sure, subarctic is not just am335x but also dra61x / dra60x (one or both, not sure) Mar 12 06:08:10 i'm headed off to bed, I'm going to leave this 'puter logged in to IRC, I'll read it all in the AM (well, later in the AM) Mar 12 06:08:31 that's why there's MLB and such Mar 12 06:08:46 MLB? Major League Baseball? Mar 12 06:08:53 media local bus Mar 12 06:09:13 zmatt: decap it and make pictures of each of the chips to confirm Mar 12 06:09:48 I actually don't *entirely* exclude the possibility the PCIe subsystem might actually be there Mar 12 06:10:47 since it has a clkctrl, target agent, L3 firewall, and even the registers of the PCIe APLL are present in the control module (but don't respond to writes) Mar 12 06:11:03 is the jbang project that contains the EDMA code? Mar 12 06:11:18 there's no edma code in jbang Mar 12 06:11:26 oh Mar 12 06:11:35 is your edma/adc code somewhere? Mar 12 06:11:40 yes, my hd Mar 12 06:11:46 oh ok n/m Mar 12 06:12:30 I don't exclude the possibility I may eventually release it as example, but that would also need to be OK'd by my employer :) Mar 12 06:12:49 *nod* Mar 12 06:13:08 the other question would be - is it a GPLv2 compatible license (for kernel) Mar 12 06:13:27 you'll never get any of my code accepted in the kernel Mar 12 06:13:42 it's C++14 Mar 12 06:14:08 (see the jbang project, it is representative of my current code style) Mar 12 06:14:19 it isn't the coding style Mar 12 06:14:56 was looking for a working register setting for using EDMA on the ADC and the PRU Mar 12 06:15:19 TRM every now and then has intersting surprises Mar 12 06:15:40 the ADC is very straightforward for DMA Mar 12 06:15:50 you just configure a DMA threshold on a FIFO and enable its DMA bit Mar 12 06:15:58 it works exactly as described w/o quirks? Mar 12 06:16:05 afaict Mar 12 06:16:45 sounds just like the interrupt mode Mar 12 06:17:49 yup, only difference is that the signal goes to edma (and is pulsed rather than level) Mar 12 06:18:49 I've configured the threshold equal to the number of steps, and edma with a really simple static job that slurps the fifo in one burst and spits it onto a fixed buffer Mar 12 06:23:45 in general, I've found the adc to be really simple to configure Mar 12 06:25:26 but peripherals can definitely hide "entertaining" surprises now and then Mar 12 06:26:12 like I discovered the hard way that the uart's FIFO is incapable of accepting data at the rate the L4 interconnect can deliver it Mar 12 06:27:12 there are many interesting surprises Mar 12 06:27:33 Just taking an informal poll---- what OS is everyone running on their BBB? Mar 12 06:28:31 ds2: the i2c controller is fun too Mar 12 06:29:02 personal favorite is the McSPI Mar 12 06:29:13 mcspi seemed relatively sane so far Mar 12 06:29:22 but I haven't really played with it yet Mar 12 06:29:31 until you try to use something more then basic features Mar 12 06:29:34 i2c otoh is broken as shit Mar 12 06:30:09 my advice to anyone using the i2c controller would be to keep it in reset until you need to perform a transaction, and put it back in reset afterwards Mar 12 06:30:16 (and forget about multimaster) Mar 12 06:30:32 single master i2c seems to work okay Mar 12 06:30:56 if you look at the linux driver, you'll find that it actually resets the controller if it encounters any error Mar 12 06:31:15 getting a nack from a slave will lock up the FSM already Mar 12 06:31:35 i have enough problem elsewhere that I don't peak inside another driver unless needed :D Mar 12 06:31:39 hehe Mar 12 06:32:20 i like system to do what I need Mar 12 06:32:45 a single pulse low on SDA will also lock up the i2c controller... you won't see anything in its status regs, bus-busy will be cleared, but if you try to start a transaction simply nothing happens Mar 12 06:33:00 the start bit remains set, the FSM continues to pick its nose Mar 12 06:33:36 I been hot swapping the I2C lines and it seems fine Mar 12 06:34:11 Has anyone here mounted an lvm2 pv volume in debian? Mar 12 06:34:23 yeah a mere glitch shouldn't cause it, there's still the input filter Mar 12 06:34:39 I meant a pulse long enough to qualify as a start condition Mar 12 06:35:02 maybe it is the reset you are pointing out but I have seen it recovery just fine from a poorly behaving slave (in this case, a MCU stopped in the middle of a xfer) Mar 12 06:35:28 if it times out - reset, if it gets a NAK - reset Mar 12 06:36:27 besides, it can't be worse then musb Mar 12 06:36:28 ;) Mar 12 06:36:52 musb looks.... interesting Mar 12 06:37:29 I've never seen so many test, diagnostic, tweak, tune, twiddle registers in one peripheral Mar 12 06:37:46 really makes me feel they have a lot of confidence in that thing Mar 12 06:38:37 hehehe Mar 12 06:39:16 it actually took a moment before I realized a big list of registers in the TRM doesn't include the actual usb core Mar 12 06:39:16 the test modes you can even use to bitbang usb... Mar 12 06:39:28 it's just dma, phy, and integration stuff Mar 12 06:39:52 tbr: I'd still consider that moderately normal Mar 12 06:40:13 why would you do such a thing? because Nokia hardwired the bus for device operations on the N900, while it was capable of host and people wanted to use host, but the FSM would reset if you tried to put it in host mode. Mar 12 06:40:23 but a "Disable the VBUS debouncer circuit fix" bit otoh.... Mar 12 06:42:37 "Flags if the RX data packet has PID error. NOT IMPLEMENTED YET" Mar 12 06:45:40 everybody loves musb... Mar 12 06:46:50 <3 Mar 12 06:48:29 to be fair, a big part of the problem is of course USB itself Mar 12 07:17:46 ok, u-boot's mass-storage gadget is definitely a _lot_ slower than the kernel's **** BEGIN LOGGING AT Thu Mar 12 07:25:16 2015 Mar 12 08:18:32 Greetings. How can I erase emmc on BBB? Mar 12 08:18:58 for what purpose? Mar 12 08:19:08 you could boot from µSD and then dd /dev/zero over the emmc device Mar 12 08:19:23 but it wouldn't /really/ erase everything Mar 12 08:19:28 due to wear levelling etc Mar 12 08:19:48 would need "nation state capabilities" to recover it though ;) Mar 12 08:20:26 Is anyway to erase everything? Mar 12 08:21:02 what is your specific reason you want to do that though? Mar 12 08:21:13 that migth help to point you towards the right thing Mar 12 08:21:57 Well I don't like that latest debian flashed on emmc, I want to downgrade to Ubuntu Mar 12 08:22:24 But don't want to hold boot button all the time as well :) Mar 12 08:22:26 so install Ubuntu Mar 12 08:22:32 no need to erase emms Mar 12 08:22:41 then all you need to delete is MLO Mar 12 08:22:43 on the emmc Mar 12 08:22:50 how? Mar 12 08:25:15 either boot into debian Mar 12 08:25:30 or boot from sd and mount the emmc fat partition Mar 12 08:25:31 and dlete it Mar 12 08:27:02 Ok so I need to ssh to debian, mount 90 Mb FAT16 partition and delete content if it, right? Mar 12 08:27:29 yes Mar 12 08:27:31 MLO should be enough Mar 12 08:27:34 Ok Isee Mar 12 08:27:48 And last question if you don't mind Mar 12 08:28:30 ask, dont ask to ask Mar 12 08:28:56 Can I install additional software to emmc flashed OR I need to build it SD first? Mar 12 08:30:33 Why I'm asking this: I've installed Synaptic and GParted for LXDE, Neither of them launches. Mar 12 08:34:48 you installed them using apt-get? Mar 12 08:35:24 sure Mar 12 08:35:40 using sudo Mar 12 08:35:56 then they should work. if they don't, usual linux trouble-shooting applies. Mar 12 08:40:25 ok will play around Ubuntu now. Thanks for advising Mar 12 09:34:21 Hi All. I’m trying to build kernel 3.14 for android. Anyone knows which config option I should enable for USB Host support? Mar 12 09:50:16 michaelgobbers, dont do it android's da devil Mar 12 09:50:53 why??? Mar 12 09:51:14 It’s a requirement for my project so no way around it. Mar 12 09:52:08 don't feed the troll Mar 12 09:52:50 what troll? Mar 12 09:53:21 michaelgobbers: you'll need several of those, but if there's a defconfig for bbb it should have it enabled Mar 12 09:53:42 michaelgobbers: also make sure you have the android patches applied in that tree Mar 12 09:54:14 there is one. However, it seems that the kernel upgrade that I’m did stopped it from working Mar 12 09:54:52 any particular reason you're not using whatever TI is using for their android thing on BBB/Sitara? Mar 12 09:55:16 I started with the BBBAndroid project Mar 12 09:55:45 I think this build alows you to actually access the GPIO pins Mar 12 09:56:05 not sure if the build you are refering to is doing that Mar 12 09:56:45 I have no idea about android on bbb. my advice is generic Mar 12 09:56:56 michaelgobbers, ignore the troll Mar 12 09:58:02 hmm ok. Could you provide me with a link to the TI projects? Mar 12 09:58:41 I'd be googling myself now. I remember seeing that they had their usual butchered releases somewhere. Mar 12 09:59:42 http://processors.wiki.ti.com/index.php/TI-Android-JB-4.2.2-DevKit-4.1.1_DeveloperGuide Mar 12 09:59:51 yeah that 4.2 Mar 12 09:59:54 I need 4.4 Mar 12 10:00:01 ha, good luck with that Mar 12 10:00:04 that probably the reason :) Mar 12 10:00:14 Thanks anyway Mar 12 10:00:58 not sure what to suggest. if rowboat or something else is still around, maybe that. Mar 12 10:02:20 I would guess no real android drivers for 4.4? Mar 12 10:02:29 at least for display Mar 12 10:03:10 looking at google for "ti bbb android kitkat" there seem to be a few things. Might be missing the graphics acceleration though? Mar 12 10:03:46 that one even claims sgx working: http://www.2net.co.uk/tutorial/android-4.4-beaglebone-update Mar 12 10:04:23 he's using the ancient 3.2 kernel though Mar 12 10:04:35 The BBBAndroid project is working somewhat. but it was missing bluetooth le. I tried adding it but to support the le mode I had to upgrade the kernel. Mar 12 10:05:16 so 3.2 kernel isn’t sufficient Mar 12 10:08:25 lol, some coworkers testing new hardware... but I suspect it still has some minor issues Mar 12 10:08:44 judging by the loud bang after power-on and the plume of smoke that rose from the PCB Mar 12 10:09:02 heh Mar 12 10:09:08 +1 for magic smoke Mar 12 10:11:12 zmatt: sell them magic-smomke-refill packs Mar 12 10:11:47 ble/smart will be tricky under android in general Mar 12 10:20:38 Yeah it is tricky… I had to replace the bluedroid stack in android with the bluez stack already. Mar 12 10:24:46 michaelgobbers hm looks like you need to patch the kernel stuff on rowboat for 3.14 Mar 12 10:24:58 but I personal believe it will not work in the end Mar 12 10:25:08 because of the binaries Mar 12 10:26:48 hmm Ill take a look at it Mar 12 13:12:34 Hi all Mar 12 13:12:59 Just loading an 8GB SD with the latest Ubuntu 14.04 image for the first time. Mar 12 13:13:13 How long should I expect to wait on the Beagleboard XM for the device to start? Mar 12 13:13:37 Some instructions online suggest that it could take 5 mins to start Mar 12 13:13:52 But it's taken longer, almost 10 so far. Mar 12 13:14:30 The SD light is flashing in a particular pattern rather than the sort of random flash I would expect from disk writes / reads. Mar 12 13:15:26 But something is going on, as the screen times out but turns itself back on if I hit a cursor key. Mar 12 13:15:59 Any help would be extremely gratefully received. Mar 12 13:43:04 Hi! Can one assign multiple triggers to one LED? Mar 12 13:43:51 yegorich does this make sense? Mar 12 13:44:17 what happens when both events at the same time? Mar 12 13:44:27 how usefull is the output? Mar 12 13:46:43 woglinde: I have a CAN controller and socketcan exports "can0-tx" and "can0-rx", but I only have one LED, that shows date being transmitted/recevied Mar 12 13:46:58 interesting, flashing exactly the same image using BBBlfs to two fresh bbbs that happened to differ in the eMMC part used... Mar 12 13:47:03 kingston: 1572864000 bytes (1.6 GB) copied, 236.263 s, 6.7 MB/s Mar 12 13:47:13 micron: 1572864000 bytes (1.6 GB) copied, 607.657 s, 2.6 MB/s Mar 12 13:47:27 woglinde: we always have two LEDs: red - for error, green - for data Mar 12 13:49:21 zmatt welcome to the world auf mmc Mar 12 13:52:02 woglinde: I'm surprised that both parts seem to be used pretty much at random, as if there's little difference Mar 12 13:53:07 (both numbers are much less than the spec claims you should be able to achieve though) Mar 12 13:53:22 hahahahahaha specs Mar 12 13:53:29 and mmc/sd Mar 12 13:53:30 lol Mar 12 13:53:55 well I did manage to them, with some effort (both marked "confidential") Mar 12 13:54:01 *to find them Mar 12 13:55:16 micron claims seq write >= 6.6 MB/s, kingston claims seq write >= 12 MB/s Mar 12 13:56:27 (there are more differences, kingston is eMMC v4.51 vs micron v4.41, kingston has larger boot partitions (2 MB instead of 1 MB) but smaller user partition (3688 MB instead of 3744 MB) Mar 12 13:57:11 intersting! Mar 12 14:00:51 micron specs 90 iops for random write, kingston doesn't spec random access performance Mar 12 14:01:45 (i.e. small writes will demolish performance, but that's not really news) Mar 12 14:15:42 We need to buy Beaglebone and Beaglebone black Mar 12 14:15:51 How Long it taskes to deliver to uae Mar 12 14:17:23 Amjad ask your distributor? Mar 12 14:18:02 actually I know nobody here, who is selling the boards directly Mar 12 14:19:53 I use actually again Mar 12 14:47:03 hi Mar 12 14:47:45 I am unable to see the changes made in .config when i flash it using tftp Mar 12 14:48:08 please can some one help me with this? Mar 12 14:49:38 flash? tftp? Mar 12 14:49:50 yes Mar 12 14:50:14 configure your kernel to provide config.gz via /proc Mar 12 14:51:28 git clone git://github.com/beagleboard/kernel.git cd kernel git checkout 3.12 ./patch.sh cp configs/beaglebone kernel/arch/arm/configs/beaglebone_defconfig wget http://arago-project.org/git/projects/?p=am33x-cm3.git\;a=blob_plain\;f=bin/am335x-pm-firmware.bin\;hb=HEAD -O kernel/firmware/am335x-pm-firmware.bin cd kernel make ARCH=arm CROSS_COMPILE=arm-linux-gnueabi- beaglebone_defconfig -j4 make ARCH=arm CROSS_COMPILE=arm-linux-gnueabi- Mar 12 14:51:48 I have followed above mentioned process Mar 12 14:52:04 http://wiki.beyondlogic.org/index.php?title=BeagleBoneBlack_Building_Kernel Mar 12 14:52:14 this is the link Mar 12 14:52:25 i want to enable a driver module Mar 12 14:53:36 I made the changes and flashed the image as mentioned in the link Mar 12 14:53:56 but I am unable to see any debug messaged from my kernel module Mar 12 14:54:06 i dont even think its installed Mar 12 14:57:25 I am struggling with the same since a week Mar 12 14:57:27 :( Mar 12 15:02:49 than install the modul Mar 12 15:03:19 flashing only the kernel image will not install any modules Mar 12 15:54:13 Okay, so I have a distributed network of display screens (could be lcd or 7-seg). It's all connected by cat5, with spans <= 300 ft. Data is controlled by a centralized bbb. Do you all think that the easiest solution is to just network it all and communicate via tcp/ip stack to a pic18 (or similiar) controlling a display screen? Or is there something I'm missing using the rs485 standard? Mar 12 15:56:13 why tcp/ip? Mar 12 15:56:22 just make a very simple serial protocol Mar 12 16:02:18 I guess my thinking is that the distances of the lines would be an issue, needing line drivers and termination...and using a switch would handle all of that (I'm not being paid very much for this, so I'm more interested in saving me time). Mar 12 16:03:30 As it's not really an area I'm too familiar with, can I ask...why not tcp/ip? Mar 12 16:05:21 well, if you want to save time, just use ethernet Mar 12 16:05:29 and a PC at each end Mar 12 16:05:30 solved Mar 12 16:05:36 or a BBB at each nide Mar 12 16:05:37 node Mar 12 16:13:14 that's exactley what i was thinking. I like the pic18 option as it would be cheaper than additional bbbs. Ideally, I'd like to keep the switch under $200 and the display module at each node under $30...hence the pic18, an ethernet port, and a cheap lcd. Mar 12 16:16:30 endoped: just go with tcp/ip. i've done something very similar before and it worked just fine Mar 12 16:17:03 endoped: calculate price vs work time Mar 12 16:17:09 still does, actually Mar 12 16:22:13 hello, I'm chatting for the first time from my beaglebone black \o/ Mar 12 16:23:09 hi Mar 12 16:23:49 dmelani: thank you, I like confirmation of what I'm considering. KotH: yeah, that's probably a good idea. It may not be worth it, time-wise, to build my own boards out of cheaper parts. Mar 12 16:34:57 fg Mar 12 16:35:05 bah. wrong terminal Mar 12 19:01:10 got the u-boot fix to init the R30's working. the lines in that E2E post worked except that it locks up unless there is a printf in the middle. perhaps compiler optimization or something. Mar 12 20:23:46 hello Mar 12 20:24:11 Where do I get external LED's for beagle? Mar 12 20:25:40 usm: not sure but (sorry to be that guy) Google it hehe Mar 12 20:26:14 and also randomly PMing people isn't usually welcome :P Mar 12 20:26:39 Can anyone answer a question related to eQEP's maximum speed? Mar 12 20:30:17 Can anyone answer a question related to eQEP's maximum speed? Mar 12 22:27:59 The Art of electronics is similar to Halbleiter-Schaltungstechnik (German Edition) you mentioned for jamesaxl yesterday (well save it is in English instead of German) Reminds me of the discussion on Quantum electronics I had 3 days ago (ahem). Anyhow I see your point about RTL in any case. Getting a YACC lex based parser working for VHDL is a bit of a pain because of VHDL ADA inheritance as the array operator is the Mar 12 22:27:59 same as the function and expression closure operator (duh). I ramble, Mar 12 22:29:00 GenTooMan: could you send it to me ? Mar 12 22:30:04 This is what I am refering too (http://www.amazon.com/Art-Electronics-Paul-Horowitz/dp/0521809266/ref=sr_1_1?s=books&ie=UTF8&qid=1426199370&sr=1-1&keywords=The+art+of+electronics) both are good books. Mar 12 22:31:04 He is write in that likely the english translation for the book he mentioned isn't that great (it's a tradition to make good books harder to understand when translated I swear) Mar 12 22:32:59 GenTooMan: bcs i have not found it , i torrent files Mar 12 22:34:30 buy it it's worth every cent of it, some of the books I have are 80 years old and they are STILL good. It's an investment, besides you want a physical book, or you will have a very difficult time with it. Mar 12 22:35:25 GenTooMan: i want to buy it, my country does not allow Amazon, i should move to France to o that Mar 12 22:36:20 ahh see if your local book chains can get it for you? ISBN-13: 978-0521809269 Mar 12 22:37:04 Has KotH suggested if you can learn german a lot of very good technical books are written in it. Mar 12 22:37:11 GenTooMan: i am going to try anyway Mar 12 22:37:56 Nothing worth while is easy, just remember that :D Mar 12 22:38:30 GenTooMan: of coorse and i can spend a lot of money for scince Mar 12 22:46:38 well in the US most people do not know how to learn, learn to learn which is done by reading then listening then understanding Mar 12 22:48:53 GenTooMan and trying Mar 12 22:55:11 Doing anything is better than 1:10 people doing something useful is better than 1:100 doing something great puts you as 1:10000 however you have to start by doing something :D Mar 12 22:56:33 GenTooMan: Lean to learn and teach Mar 12 23:41:50 Hi. I'm trying to flash a custom image on the eMMC of the BBB by following the steps in the Sitara Linux Program the eMMC on Beaglebone Black page. The page is not clear as what the AM335x is trying to do when booting from usb (holding the user button with no sd card inserted). I understand everything as soon as u-boot gets involved. But what does the bootloader in the AM335x's ROM exactly? Does it try to Mar 12 23:41:56 get the SPL+u-boot from tftp via dhcp? BTW, I'm on Fedora 21 (and the guide suggest ubuntu 12.04LTS for setting up dhcp/tftp). Mar 12 23:43:59 Oh and I hook up my FTDI cable and monitoring on the UART0... there is not much going on aside from outputting CCCC... Mar 12 23:53:06 Hmmm 26.1.8 in TRM... Mar 13 00:39:18 Is it possible to remove the power connector from BBB and hard wire power to it? Mar 13 00:40:32 I want to power it from my car battery using a step down thing I found on the Internet that would provide the 5 amps Mar 13 00:41:06 everything is possibel Mar 13 00:41:10 but why would you ? Mar 13 00:41:18 it seems very overkill Mar 13 00:42:26 And you're right, it maybe I'm just afraid of that connector failing. I'm using it for an amateur radio purpose and it won't be moved but I figured a hardwire would be a clean set up Mar 13 00:43:44 And it is easier to hardwire it instead of finding a 12 V power source with the right connector on the end Mar 13 00:44:05 make a lead Mar 13 00:44:11 or chop one in two :) Mar 13 00:44:43 I've just been cannabalising a bunch of PSUs that are no use to me Mar 13 00:45:05 I thought about cutting the cord and attaching Anderson power pole to it. One side it would be the 5.5 MM connector I need and the other side would have a red and black power pole on it Mar 13 00:45:19 Basically a quick disconnect without having to remove the connector Mar 13 00:45:40 in fact .. I made a lead up to power a beagle from a 5V switchin supply I stole from somewhere else :D .. its on a 'ups' supply with a lead-acid backup battery :D Mar 13 00:46:11 runs from 7-1x volts Mar 13 00:46:42 had to cut a track on a step-up that was onboard though lol Mar 13 00:47:45 Can I post links in here? Mar 13 00:47:57 yup Mar 13 00:50:11 ok time to git .. bbl :) Mar 13 00:52:51 http://m.eachbuyer.com/dc-dc-converter-12v-step-down-to-15w-5v-3a-power-supply-p14004.html?currency=USD&utm_source=google&utm_medium=CSE&utm_campaign=%5BPLA%5DAmerica(EN)&gclid=CjwKEAjw8oSoBRCss4qDrP3Y7yUSJACKumZ9HFwa3uyrC4Pv3Om5LCTOt5rb7nn2pRw-T1Qz8n8-phoC_Jnw_wcB&from=site Mar 13 00:54:09 http://m.eachbuyer.com/12v-to-5v-usb-dc-dc-buck-converter-car-power-supply-regulator-p174591.html?currency=USD&utm_source=google&utm_medium=CSE&utm_campaign=%5BPLA%5DAmerica(EN)&gclid=CjwKEAjw8oSoBRCss4qDrP3Y7yUSJACKumZ91YOCV6nOPO6Mq4gVGjOWdQeuzsNkTUzJUaqpq5byXRoCd3Lw_wcB&from=site Mar 13 00:54:51 I was going to use one of those two items. I have a wireless to ethernet adapter that requires USB power as well so that USB one would work good for me I think I can cut off one USB port and hardwire it directly to the board Mar 13 00:58:22 My phone had to reload I think I missed probably any comments anybody made Mar 13 01:02:15 Not many people chatting Mar 13 01:13:36 to save or not to save an old dead project from 2007 from google code...... Mar 13 01:23:24 hello, I want to know how to access my beaglebone black from the internet. Mar 13 01:23:56 ifmx4ever, ever done dynamic dns & forward porting before? Mar 13 01:24:41 no. I am willing to follow instructions if there is a URL that has information on it. Mar 13 01:26:32 google: "dynamic dns beaglebone black" lots of people have done it.. ;) Mar 13 01:27:07 it's really easy , if your wifi/ethernet router supports one of the dynamic dns compaines.. Mar 13 01:27:27 ok thanks. appreciate your help. Actually was not able to put in the correct search terms in google. Mar 13 01:28:41 the hard part is... are you double-nat'ed and what ports do your isp block.. Mar 13 01:32:22 i am not sure about "what ports your isp block". And I do not believe I am double-nat'ed. Noob to BBB and networking. but I will read up and see what fits my need. Thanks again. Mar 13 02:11:56 hello **** ENDING LOGGING AT Fri Mar 13 02:59:59 2015