**** BEGIN LOGGING AT Sat May 02 02:59:58 2015 May 02 03:40:24 * zmatt just realized an excellent solution to the battery power issue when one doesn't need a lot of power from the 3.3V rail... May 02 03:40:46 desoldering the 3v3b regulator and tying 3v3b to 3v3a :P May 02 03:40:59 ooo go zmatt :D May 02 03:42:03 the pmic can supply 400 mA... I'm pretty sure that's actually plenty in many cases May 02 03:42:28 I'll say May 02 03:43:30 especially since only peripherals and I/O are powered by it, not any core part of the am335x May 02 03:45:48 (the core, mpu, and ddr supplies are "limited" to 1.2A each) May 02 03:45:49 hmm well.. note any issues .. but seems odd they've added an extra regulator? but I suppose it decouples it from the pmic in case there would be issues, or you want peripherals powered when the pmic shuts down .. maybe May 02 03:46:08 except neither is allowable May 02 03:46:38 decoupling from the pmic is bad: it actually detects overcurrent and will initiate a properly sequenced shutdown May 02 03:47:14 powering peripherals while the pmic shuts down is doubleplus badness May 02 03:48:21 yeh, good point :) May 02 03:48:27 did you read my latest big forum post on this topic? May 02 03:49:06 where I documented the impact of making the mistake of leaving the serial cable connected at shutdown May 02 03:49:19 (especially on battery power, to lesser degree on DC power) May 02 03:49:46 no - I'd be interested to get updates.. will hae to add to my rss? May 02 03:49:54 ahem .. url .. I'll sort the rss :D May 02 03:49:55 lol May 02 03:49:56 you can subscribe to the topic May 02 03:50:07 to get emails May 02 03:50:15 right .. yeah where to? May 02 03:50:32 https://groups.google.com/d/msg/beagleboard/7sxPePT7wkM/V1Ft-xxh0agJ May 02 03:50:37 that's my latest post May 02 03:50:49 ugh google :/ ok sec .. ;p May 02 03:51:05 well it's better than what you get via the beaglebone website :P May 02 03:51:30 this is my original post, in case you've missed the whole thread -> https://groups.google.com/d/msg/beagleboard/7sxPePT7wkM/3vFMPydR20IJ\ May 02 03:51:39 whoops, without the \ May 02 03:51:47 hehe no prob May 02 03:51:51 (why did they put that so close to the enter key? :P ) May 02 03:51:53 has Gerald seen the thread? May 02 03:52:07 I got a # by mine :P May 02 03:52:26 unles the keyboard map is wrong or I'm drunk-typing :D May 02 03:52:57 dunno, I'd hope so, he participated in the thread earlier (read: in 2012/2013) May 02 03:53:32 cool May 02 03:53:43 * veremit finally installs nodejs on his beagle .. ! May 02 03:53:48 i.e. the problem with using the 3v3a as enable was already known with the beaglebone white May 02 03:53:58 right May 02 03:54:03 yet rev A6A of the BBB reverted to using the 3v3a as enable May 02 03:54:04 :/ May 02 03:54:15 well . I think there's still scope for further valid board revisions May 02 03:54:34 if it can be justified May 02 03:54:52 ideally 3v3b should track 3v3a voltage level May 02 03:55:09 surely there must be regulators which take a reference voltage as input? May 02 03:56:25 keeping 3v3a and 3v3b within 0.3V of each other is the only sure way of avoiding problems May 02 03:57:26 (this also takes advantage of the more accurate voltage calibration of the 3v3a, and its runtime adjustability for fine-tuning) May 02 03:57:54 hmm not personally met a trackign regulator yet .. but its possible I'da thought .. some form of 'trim' input maybe .. May 02 03:58:31 yea looks like it .. what's the one on board? May 02 03:58:38 for 3v3? May 02 03:58:59 U4 May 02 03:59:04 bottom side, near DC plug May 02 03:59:13 chip id/part no pls? :) May 02 03:59:17 oh May 02 03:59:18 eh May 02 03:59:25 mine's .. over there *waves* lol I need to get another May 02 03:59:33 tl5209 May 02 03:59:39 actually .. i should wire up my BanPi May 02 03:59:46 oo ti part no doubt May 02 04:01:20 they do exist... voltage-tracking LDOs May 02 04:01:24 but not very common it seems May 02 04:01:48 but e.g. TPS7B4253-Q1 May 02 04:05:54 are there pinout compatible LDOs with a higher turn on threshold? May 02 04:06:48 it can't supply as much current as the TL5209, but that could be compensated by moving all on-board peripherals to the 3v3a .. I don't understand what they're doing on the 3v3b anyway :P May 02 04:07:07 I am surprise gerald hasn't said a peep May 02 04:07:11 ds2: that's not really a solution May 02 04:07:25 ds2: significant current flow can occur once 3v3a drops 0.3V below 3v3b May 02 04:08:07 merely pulling the serial console rxd high caused this -> https://lh3.googleusercontent.com/-C9Uq48C-wwc/VUOT3kcXUXI/AAAAAAAAAC8/c6xC9VFFW5A/s1600/off-bat-serial.png May 02 04:08:24 oh the serial path May 02 04:08:28 uhuh May 02 04:08:44 drives from 3v3b straight into a cpu pin, very nice May 02 04:09:13 replace that gate he has in there with a analog MOSFET switch then May 02 04:09:23 then drive the switch from like PMIC good May 02 04:09:37 it should have been supplied by the 3v3a to begin with May 02 04:09:39 not 3v3b May 02 04:09:56 i do wonder if there are bugs with this PMIC May 02 04:09:57 but a CAPE can easily end up doing the same thing May 02 04:10:14 think the am335x webpage mentions there are 3-4 PMIC options May 02 04:10:16 yeah I'd flag it up to Gerald .. I think he's open to constructive criticism?! May 02 04:10:24 apart from some questionable design choices, it seems to be doing its job as advertised May 02 04:10:26 don't think the pmic is the problem ds2 ;P May 02 04:10:50 zmatt .. yeah push for a tracking 3v3 reg .. there are quite a few options .. just not all are tI .. lol. May 02 04:11:00 ds2: there are 4 variants, different programming of the voltages and sequencer May 02 04:11:12 more then that May 02 04:11:20 there should be one option w/o the battery charger May 02 04:11:32 eh, no May 02 04:11:45 you're free to use it with or without battery May 02 04:11:52 (though it's clearly more design for use with) May 02 04:11:54 *designed May 02 04:12:08 no no May 02 04:12:23 the battery charger gets replace with something else. it supports battery but not charging May 02 04:12:32 useful if you want to do proper USB charging May 02 04:12:38 ds2: no idea what you're talking about May 02 04:12:46 ds2: but TI has lots of PMICs May 02 04:12:48 let me see if I can find the page May 02 04:13:19 the am335x evm actually uses something other than a tps65217 iirc May 02 04:14:29 look at the PMIC chart: http://processors.wiki.ti.com/index.php/AM335x_Hardware_Design_Guide May 02 04:14:42 veremit: I'm checking the specd consumption of on-board peripherals to see if there's any reason to supply them from 3v3b rather than 3v3a May 02 04:14:48 there is a WLED option or a 5V boost option May 02 04:15:43 zmatt .. I would imagine not .. 0.5a is a beefy regulator for embedded May 02 04:15:44 ds2: only the 65217 is really am335x-specific, the 65910 is a very fancy generic pmic May 02 04:15:54 probably cheaper ;P May 02 04:15:54 veremit: 0.4 May 02 04:16:03 zmatt .. thats the 5209 May 02 04:16:07 veremit: ah May 02 04:16:15 so 0.4 should be fine .. surely?! May 02 04:17:12 veremit: yeah so the 3v3a has 400 mA available, if that's enough for the am335x IOs and on-board peripherals (and I have trouble believing it wouldn't be) then the regulator could be entirely for the CAPEs (just like it was on the BBW) May 02 04:17:40 I'm wondering if there's a reason for the a/b split though, it already existed on the BBW where two of the PMIC LDOs were at 3.3V May 02 04:17:54 maybe the external regulator is to designed to act more like a fuse May 02 04:18:06 but then why are the internal peripherals on it? May 02 04:18:11 a cape with a short cant fry the PMIC May 02 04:18:14 ds2.. that was kinda my thinking .. but the decoupling has its issues .. May 02 04:18:19 you can't fry the PMIC anyway May 02 04:18:29 it closely monitors its outputs May 02 04:18:30 well thats the whole point of the pmic lol May 02 04:18:34 that a mighty strong statement :D May 02 04:18:44 ds2: indeed it is! May 02 04:19:04 you (probably) can't fry the PMIC by overcurrent on one of its LDOs, unless the LDO is configured as load-switch instead May 02 04:19:25 in that case it permits current-limiting without shutdown May 02 04:19:29 which does risk overheating May 02 04:19:37 but it's not configured like that anyhow May 02 04:19:45 (except in the A-variant) May 02 04:21:12 ds2: for comparison btw, the TPS65910 has 3 buck converters, 1 boost converter, 8 LDOs, integrated RTC with its own LDO, two I²C interfaces (one dedicated to smartreflex) May 02 04:21:30 ds2: it's a bit... overweight :P May 02 04:21:49 (it's designed for OMAPs) May 02 04:22:02 yea that'd suit the x15 .. ;) May 02 04:23:30 the boost converter is really handy May 02 04:23:41 I'd have prefered they used that instead May 02 04:23:55 veremit: it may also make sense to move ethernet over to 1.8V ... it has its own IO supplies (VDDHV5 iirc) May 02 04:24:24 i much rather delete the ethernet all together May 02 04:24:35 wth May 02 04:24:48 NOOOOOO May 02 04:24:50 that thing is useless May 02 04:24:58 what are you on? :P May 02 04:24:59 waste of power and space May 02 04:25:03 its got REAL ethernet .. not like that piece of shit pi May 02 04:25:06 that May 02 04:25:13 *ouch* May 02 04:25:16 cough* May 02 04:25:30 it's more a bummer that they used a 100Mbit PHY instead of gigabit May 02 04:25:37 if you want a stm32 .. buy a friggin stm32 :p or a m4/etc etc etc May 02 04:25:39 much rather they bring out RMII or RGMII to a header May 02 04:25:43 zmatt .. its an old chip :P May 02 04:25:57 still better than <48mbs shared with a usb hub lol May 02 04:26:06 veremit: what is, the PHY? May 02 04:26:20 not a fan of ethernet May 02 04:26:26 zmatt .. I would ssuspect most of the am335x processor lol May 02 04:26:32 ds2: what would you use?? May 02 04:26:41 wifi?! *spew* lol :p May 02 04:26:48 veremit: for most of my use cases, I have no use for ethernet May 02 04:26:50 veremit: ehh, the am335x has two gigabit ethernet ports with integrated switch May 02 04:26:52 nor wifi May 02 04:27:01 ds2: it's about the only port I use May 02 04:27:09 I've actually disabled USB in my kernel since I have no use for it May 02 04:27:17 BT works for the few application I need networking May 02 04:27:32 zmatt: what are you doing that would want ethernet AND battery ? May 02 04:27:50 preventing eMMC corruption when the end user pulls the plug May 02 04:27:52 and they will May 02 04:27:59 zmatt .. it does? shit... May 02 04:28:12 zmatt .. yea they do that May 02 04:28:16 ahhhh as a UPS May 02 04:28:18 veremit: it actually supports rebooting while keeping the switch operational May 02 04:28:26 i much rather keep the eMMC RO May 02 04:28:28 well I da .. May 02 04:28:39 veremit: (that part is slightly buggy, but still) May 02 04:28:48 the imx6 I got only runs its eth at 0.5mb for some internal bottleneck :/ May 02 04:28:55 veremit: the switch also supports VLANs and shit May 02 04:28:57 but yeah .. 100mbit is enough :) May 02 04:29:13 it is not that hard to switch out the ethernet for gigE May 02 04:29:13 veremit: and packet timestamping for PTP / AVB May 02 04:29:52 veremit: and best of all, it's really easy to get the switch operational... piece of cake from baremetal code May 02 04:29:54 ds2 .. i should think its doable May 02 04:29:59 zmatt .. kewl May 02 04:30:03 (while I wouldn't touch the USB subsystem with a ten foot pole myself) May 02 04:30:03 I need to follow ya posts :P~ May 02 04:30:11 zmatt .. ah good old musb .. rofl May 02 04:30:14 now now... the USB stuff isn't THAT bad May 02 04:30:18 ds2: yes, it is May 02 04:30:21 ds2: really May 02 04:30:28 it is May 02 04:30:37 it really REALLY is .. its bodged ! May 02 04:30:38 ds2: half of the errata sheet consists of usb bugs May 02 04:30:42 zmatt: I have worked with that IP block on probally a dozen chips. it grows on you if yoi give it a chance :D May 02 04:31:03 ds2: lol .. oooook May 02 04:31:03 ds2: except TI frankensteined their own DMA controller onto musb May 02 04:31:18 zmatt: other folks have done stuff like that May 02 04:31:46 ds2: well, maybe those folks did it right, TI sure didn't May 02 04:32:13 zmatt: no better. May 02 04:32:14 hehehe May 02 04:32:23 you learn to deal with it and fix bugs May 02 04:32:42 half the problem is people fix it on one chip and break it on another May 02 04:32:50 well, have fun... I've looked at the USB subsystem chapter, and no thanks I'll pass May 02 04:34:01 the main thing annoying with the MUSB on the BBB is the integrated PHY May 02 04:34:39 huh, the tps65910 is actually cheaper than the tps65217 May 02 04:34:46 didn't expect that May 02 04:36:19 but no battery operation... or at least no charger May 02 04:39:37 eMMC 80 mA max for micron, kingston says 97 mA but that's at more than twice the xfer rate supported by the am335x May 02 04:45:43 the eth phy datasheet is confusing me a bit, but sounds like typ 90 mA for 100base-TX with traffic or 120 mA for 10base-T with traffic May 02 04:46:01 what else do we have? May 02 04:48:10 veremit: it does all add up to quite a bit... May 02 04:48:14 is that the max values? May 02 04:48:21 it does have to dry the magnetics May 02 04:48:38 drive May 02 04:48:46 yeah a large part is for the magnetics... more so for 10baseT for some reason May 02 04:49:05 probably 100baseTX is a bit more clever on the analog side May 02 04:49:21 interesting May 02 04:49:40 veremit: and the question is how much power people are drawing from the am335x IOs May 02 04:49:46 yep May 02 04:50:03 the max current rating for VDDHV1-6 combined is is 350 mA May 02 04:53:15 (50 mA each, except 100 mA for HV6) May 02 04:54:00 but e.g. HV5 will probably only draw something similar to the IO current of the ethernet phy, which is max 5 mA May 02 04:57:14 the phy does draw non-negligible current even in power down mode, I can imagine battery-users might prefer to kill it entirely May 02 04:57:29 (of course they can probably do exactly that with a sharp knife) May 02 04:57:43 :) May 02 04:57:54 ethernet == bad for battery anything May 02 04:58:21 ds2: well, there's no reason for any chip to not have a low-power shutdown mode, except for simply not caring May 02 04:58:48 it is a ethernet thing May 02 04:58:55 if you got a ethernet cable, you probally have wall power May 02 04:58:57 like hdmi probably uses a lot of power too, but the default of the hdmi chip is really negligble May 02 04:59:09 that's another thing I want to delete May 02 04:59:13 don't bother May 02 05:00:06 typical standby power: 18 μW May 02 05:00:27 this is in the context of doing a board May 02 05:00:40 no proper docs == no place for it on the board May 02 05:00:58 you can find their confidential driver sources leaked on the web though May 02 05:01:01 including all registers May 02 05:01:11 that is not "proper" May 02 05:01:21 * zmatt shrugs May 02 05:01:36 working drivers can sometimes be better docs than the actual docs :P May 02 05:02:43 working is used very loosely May 02 05:03:06 fair enough, I'm actually just assuming they're working May 02 05:03:11 zmatt .. so, what are our options for a gigabit phy !? I'm game for some bodgey :D May 02 05:03:54 veremit: you need to respin the board, IIRC. the lines aren't all brought out of the package May 02 05:04:12 other then that, it is pretty straightforward May 02 05:04:51 the pins are almost the same actually I think May 02 05:05:04 but rgmii will require more care with the traces obviously May 02 05:05:19 urgh damn ok May 02 05:05:23 zmatt: the last I checked there is something slightly different May 02 05:05:23 well .. the idea was nice.. lol May 02 05:05:47 actually May 02 05:05:55 rgmii 0 pins seem to be a subset of mii 0 pins May 02 05:06:43 see my spreadsheet ( https://docs.google.com/spreadsheets/d/1CK5c-Cs8G1RtzGo-J3VJsD9m5K-fp06AncgeYWsdjSU/view ), first tab (Pinmux), pins 66-80 May 02 05:07:38 actually 66-83 (everything on supply HV5) goes towards the PHY I think, though 81 ends in a DNI May 02 05:08:21 so you have rgmii 0 (or rgmii 1 if you like 1-based numbering) available in principle May 02 05:09:00 so in theory you could swap out the phy and double-check whether your traces are okay May 02 05:10:11 the good news is that rgmii uses DDR, so it's only 5 times the clockrate of mii May 02 05:10:31 I think it's worth a punt .. lol .. May 02 05:10:54 I must look when I'm not 2/3 exhausted May 02 05:11:04 I think there's a chance the am335x might need an external reference clock fed into pin 81 though, might need to look into that May 02 05:11:28 so its got dual-ethernet mac on the am335x? May 02 05:11:31 never woulda thought May 02 05:11:59 actually the am335x also supports etherCAT, though that uses different pins (the PRU MII pins) May 02 05:12:07 really? May 02 05:12:08 omfg May 02 05:12:15 I just had a killer idea .. May 02 05:12:44 Hello everyone. I'm fairly new and having an issue. I've searched the net for a few weeks and coming up empty. Question: I'm running Ubuntu 14.04 on my beaglebone and want to use the SPI bus. How do I enable it? May 02 05:12:45 actually .. hrm .. wonder May 02 05:12:54 w.r.t. clock, the issue was that they failed to hook up the ethernet subsystem to a low-jitter PLL May 02 05:13:04 zmatt .. bugger May 02 05:13:09 (in general the am335x PRCM looks like it was designed by an intern) May 02 05:13:22 or someone who was drunk May 02 05:13:22 rwien .. you'll need to investigate 'device-tree' files most likely May 02 05:13:27 or really really in a hurry May 02 05:14:09 rwien: if it's enabled in your device tree then it'll show up automagically in /dev May 02 05:14:11 veremit: I've used the device-tree with the original angstrom distribution but no cap-manager or dtc exists on the Ubuntu image. May 02 05:14:23 but are all those pins broken out? May 02 05:14:37 zmatt: Yeah, no spi buses show up there. May 02 05:15:03 wait... is the BBB on RMII or MII? May 02 05:15:04 rwien: many kernels don't support overlays, in which case you need to compile your own device tree blob May 02 05:15:16 which is better in the long run anyhow May 02 05:15:18 ds2: MII May 02 05:15:29 ds2: RMII sucks donkey balls May 02 05:15:46 zmatt: So that would mean installing the dtc for compiling and then where do I load / unload them? May 02 05:16:02 zmatt: Would that be in the uEnv.txt file? May 02 05:16:03 rwien ... have you seen http://hipstercircuits.com/enable-spi-1-0-and-1-1-with-device-tre-overlays-on-beaglebone/ ? May 02 05:16:15 veremit: that's assuming he has overlays May 02 05:16:24 d'oh May 02 05:16:34 which requires either the patchy 3.8-bone kernel or I think a 4.0 kernel May 02 05:16:36 I hate this kernel b**s** that keep changing May 02 05:16:45 we need rcn .. lol May 02 05:16:49 veremit: DT is very much in flux May 02 05:17:00 and overlays used to be a beaglebone-specific invention May 02 05:17:28 yes and the capemgr was a good hack :) May 02 05:17:47 rwien: the short story would be that you'd take the default device-tree file for the beaglebone black and customize it to your needs May 02 05:19:06 zmatt: Ok, I can do that. Just confused on where the default is and where it is loaded? May 02 05:19:30 rwien: the compiled blob lives at /boot/dtbs normally May 02 05:20:00 I think the default filename is picked by u-boot May 02 05:20:04 but you can override it May 02 05:20:57 it's all really easy, once you manage to do it the first time... maybe I'll one day get around to making a nice example/tutorial one day :P May 02 05:21:08 zmatt: I have 18 .dtb files listed for the boneblack. I'm assuming it would be the non-decorated file: am335x-boneblack.dtb? May 02 05:21:35 rwien: yeah, I think you can either motivate u-boot to use one of the decorated ones, or use a completely different one altogether May 02 05:22:20 I actually have spi configured in my dt... part of which looks like: http://gerbil.xs4all.nl/snippet.dts.html May 02 05:22:28 Those are all compiled, so I would need to track down the original source, modify it, compile? May 02 05:22:38 but those PIN_ macros are actually my own, lol May 02 05:22:57 yeah the sources of the defaults are in the linux kernel source tree May 02 05:23:14 arch/arm/boot/dts/ May 02 05:23:20 zmatt: You've been very helpful. Thanks. May 02 05:24:08 I tried to cross compile the 3.18 kernel today. That crashed and burned. Hopefully I'll get a solid win and after that it would become second nature. May 02 05:24:26 mine worked on the first try :D May 02 05:24:40 that was really weird and unexpected May 02 05:25:02 if you're gonna compile your own kernel anyway, why use an old one? May 02 05:25:52 zmatt: Just wanted to see how it worked. Looking through all those cool options. My thinking was maybe I could have a tailored kernel just for what I needed. May 02 05:26:05 yeah I dumped quite a bit of crap too May 02 05:26:10 zmatt: Noticed all the defaults and drivers that I'll never use. May 02 05:26:24 (like everything related to graphics/video) May 02 05:26:34 Absolutely. May 02 05:27:08 And virtualization, smp, pci bus, etc. May 02 05:27:21 hmm, smp is definitely off in the bone kernels May 02 05:27:50 ran make xconfig and it was one of the defaults. May 02 05:28:32 This is the repository I've using: https://github.com/beagleboard/linux May 02 05:28:32 I hope you mean make ARCH=arm xconfig and started with an existing config file for the beaglebone :P May 02 05:28:33 eewiki.net / beagle black page there .. :) May 02 05:28:45 very straightforward I find May 02 05:28:58 zmatt: Ah..... no. Just make xconfig. May 02 05:29:03 rwien: I don't think that's the bone kernel... note May 02 05:29:12 rwien: when cross-compiling, always specify ARCH May 02 05:29:54 *note that beagleboard != beaglebone May 02 05:30:24 Dang... Now I'm excited to try again. May 02 05:30:52 I used the rcn-ee build scripts May 02 05:31:21 see https://github.com/RobertCNelson/linux-dev May 02 05:32:03 I'm not sure how exactly you're *supposed* to use them though, so to make it build a bone kernel rather than its default kernel I did apply some violence May 02 05:32:18 follow the guide on his web page .. its simple ;D May 02 05:32:26 So am I using the wrong repository? May 02 05:32:33 https://eewiki.net/display/linuxonarm/BeagleBone+Black May 02 05:32:55 and the bonus .. if it don't work .. you just chase 'im down in here lol ;D May 02 05:33:04 indeed he's here often enough May 02 05:33:18 yes, he is. May 02 05:33:33 hey mrpackethead .. ltns :p May 02 05:33:45 usually we just get your flaky connection isues ;P May 02 05:34:21 veremit: as for a gigabit phy, you'd just need to browse around... I have no specific recommendations there May 02 05:34:31 had my head down and been way too busy May 02 05:36:19 sounds familiar May 02 05:36:22 veremit: for some reason vendors seem to treat their ethernet phy datasheets as if they're Top Secret May 02 05:36:27 zmatt .. what's on the board currently? May 02 05:36:36 do u really need 1GB? May 02 05:36:41 zmatt .. ah man its ridiculous. . pet hate of my mate May 02 05:36:43 if you only need 100M theres a ton too choose from May 02 05:36:55 mrpackethead .. its just the challenge :) May 02 05:37:04 veremit: you know you can just download the schematic and check yourself? ;P May 02 05:37:13 nice. i can provide you with challenges if you want May 02 05:37:21 zmatt : lazy and I been awake 1am-6:30am (currently) :P May 02 05:37:32 mrpackethead .. with some financial reward?! ;) May 02 05:37:57 its how commerical organisations work May 02 05:38:00 veremit: neither is an excuse May 02 05:38:05 veremit: :P May 02 05:38:12 we take money of people to solve problems, we use that money to pay people to solve problems May 02 05:38:24 veremit: I'll save you time by saying it's on page 9 of the schematics May 02 05:38:25 in the middle i make 0.50c per year May 02 05:38:30 sigh. May 02 05:39:12 zmatt .. cheers ;) May 02 05:39:59 veremit: also, whatever phy you choose in any board design, pick a different one May 02 05:40:22 true May 02 05:40:30 this one occasionally seems to get confused about its config strapping options May 02 05:41:08 some claim it has to do with reset timing, but I fail to see how exactly reset timing makes it sample an option high when that line has internal pull-down in the phy, internal pull-down in the am335x, and external pull-down May 02 05:41:10 -sigh- May 02 05:41:23 oh -that- cookie.. yes I hear about tht May 02 05:41:47 I also one day found my BBB with no ethernet May 02 05:42:03 both leds suddenly had inverted polarity May 02 05:42:12 link down on the BBB side May 02 05:42:44 link partner did report link up/down when plugging in, but said no auto-negotiation supported and ended up at 10base-T half-duplex May 02 05:43:16 (an inverted link led should be absolutely impossible, since that is supposed to mean the internal regulator is disabled) May 02 05:43:27 veremit: what can you do, build design? May 02 05:43:51 currently midway through a schematic design .. my boss does the board layouts normally May 02 05:44:01 then we get fab done out May 02 05:44:22 what kind of things May 02 05:44:31 not specifically, but generally May 02 05:44:43 integration .. interface May 02 05:45:04 doing some hd video stuff lately .. hd-sdi to rf to camera control with motor pid, etc May 02 05:45:40 if doing anything with video the dm814x is a really cool chip... especially if TI actually bothered to support it May 02 05:47:08 ti don't -do- support ..l ol May 02 05:47:23 I was doin bldc with luminary crap .. gawd that was bad May 02 05:47:30 dm36x is patchy May 02 05:47:31 Tony Lindgren did recently put in some effort in mainlining dm816x support, but no dm814x yet, and that's just the basic kernel - no video stuff May 02 05:47:51 on the bright side, most subsystems are documented if you cross-reference the omap4, omap5, and vayu TRMs May 02 05:47:53 you need to either spend a couple of years playin with it .. or just .. well .. I dunno lol hope! May 02 05:48:07 I spent about two years playing with it I think :/ May 02 05:48:13 not really any video though May 02 05:48:27 yea. Our Ip cam's are quiet at the moment. May 02 05:49:26 never figured serial control for that one yet lol May 02 05:49:27 I really would have wanted a C6A811x ... if I'd have known about its existence back then, and TI bothered to actually make it a catalog part May 02 05:49:44 but the whole C6-Integra line seems to have gone up in smoke May 02 05:50:10 it still exists as an automotive part apparently May 02 05:50:41 here, look at its massively informative product page -> http://www.ti.com/product/dra626 May 02 05:57:26 but it's basically a cross between an AM335x and DM814x ... somewhat less video power, but in return you get the real-time control stuff like PRUSS and PWMSS May 02 05:57:50 and still keep video in/out, C674x DSP, dual cortex-m3 subsystem, etc May 02 05:59:21 got the hardware encoder blocks? May 02 05:59:59 you mean HDVICP2 aka IVA-HD ? May 02 06:00:09 holy crap that page is useless rofl May 02 06:00:15 h264, etc? May 02 06:00:24 not sure, probably depends on device variant (i.e. eFuse) May 02 06:00:31 mhmm May 02 06:01:05 if you can find marketing material about "Jacinto 5 Eco", that might tell you May 02 06:01:13 there's also an evm -> http://support.spectrumdigital.com/boards/evm811x/revc/files/evm811x_Schematics_RevC.pdf May 02 06:01:42 but no sane person would touch something _this_ poorly supported May 02 06:02:01 hahaha May 02 06:02:02 (I'm personally not claiming to be sane and would love to have one) May 02 06:02:35 especially since I already know the DM814x better than my back yard May 02 06:03:10 even got its secretive Security Subsystem and its cortex-m3 running :) May 02 06:03:17 kewl May 02 06:03:18 :) May 02 06:03:29 pretty neat, basically can act like an embedded crypto coprocessor May 02 06:05:33 2 AES modules, 2 hash modules, DES module, RNG, public key accelerator, cortex-m3, 144 KB local SRAM, 5 timers, DMA controller, irq in/out routers, and a very fancy firewall covering both internal and external access May 02 06:11:23 but if you want H.264 stuff try the DM8168 ... apparently it can do real-time H.264 encoding (High Profile) of up to 7 streams at 1080p30 or 720p60, or 15 streams at 720p30 May 02 06:12:02 wow May 02 06:12:05 that's a chip! May 02 06:12:23 it has 3 instances of IVA-HD May 02 06:13:14 it's also I think the only TI SoC so far those evm came with a heatsink+fan mounted on the processor May 02 06:13:18 *whose May 02 06:13:23 that I've seen personally at least May 02 06:13:53 support for it is just as excellent as for the DM814x May 02 06:15:46 there's a state-of-the-art 2.6.37 kernel for it May 02 06:16:50 lol oh the good old 2.6.x series .. there's a classic May 02 06:17:05 everything arm used to be Stuck at that .. May 02 06:17:09 same as dm814x (I got the version number wrong before) May 02 06:19:12 basically there are a few third parties who have hardware+software solutions based on them, but if you're not using one of those you're totally on your own May 02 06:19:21 (e.g. http://www.udworks.com/shop/list.php?ca_id=10 ) May 02 06:20:07 that sounds familiar May 02 06:22:11 IVA-HD is also the only subsystem of which there's basically no public docs May 02 06:23:39 I could probably manage to get the two ARM9 cores in it running if I cared to, but for the hw accelerators you'd really need documentation (judging by the older IVA2.2 accelerators, which are documented in the omap3 TRM) May 02 06:25:59 yea May 02 06:26:05 otoh, codecs for it should be available... same subsystem (afaik) is present in omap4/5 and vayu May 02 06:30:09 I wonder when TI plans to actually release a datasheet for the am572x, or create a product page :P May 02 06:34:18 lol May 02 06:34:24 write one ! May 02 06:55:03 I am trying to boot a application from UEFI on beagleboard qemu, I am able to load the code and run it for a few instructions and then i am getting a data abort exception for instruction str r1, [r0]. I am trying to write r1(0xe59ff014) at r0(0x4020ffd0), which is the ram exception vector table. CPSR= (0x600001d3) May 02 06:55:41 But if i boot the same code using uboot, it runs without any problem May 02 07:21:44 <_av500_> nikhilk: you might take that to the uefi people May 02 07:22:00 <_av500_> most people here just use uboot as it comes with the board May 02 07:22:11 <_av500_> or jonmasters can help you :) May 02 07:22:38 _av500_, thank you May 02 07:23:03 u boot, i boot, we boot. everyone boots May 02 07:23:28 <_av500_> except uefi May 02 07:23:34 :) .. I just dont understand why it breaks.. May 02 07:24:29 I am not any uefi functionality when this happens, I load the elf file to memory and then do a goto. May 02 07:49:33 (gdb) x/10xh 0x4020ffc8; 0x4020ffc8: Cannot access memory at address 0x4020ffc8; So is it possible to lock some memory in ARM May 02 07:50:05 uefi probably enables the mmu ? May 02 07:50:28 let me check May 02 07:50:57 iommu? May 02 07:52:05 there's no iommu there, and no publicly accessible firewall (certainly not one that would be emulated by qemu) May 02 07:52:25 k May 02 07:54:06 unless things are different on the omap3, but even if that range is l3 memory rather than local, I doubt qemu would emulate firewalls (and I see no reason uefi might configure them) May 02 07:54:14 cortex-a8 mmu is much more plausible May 02 07:54:49 0x4020ffc8 is RAM vector table address in OMAP3 May 02 07:55:03 ram vector table address is whereever you set it May 02 07:55:11 it's configurable May 02 07:55:19 bootloaders typically change it May 02 07:55:50 yea true, you basically set the address you want to jump at that vector addr right May 02 07:56:04 when i use uboot i am able to print the same address location using gdb May 02 07:56:48 no, you configure a cp15 register to point at the vector table, which should be 4 bytes of code (typically a jump) per exception vector May 02 07:57:37 if you can't read or write the location then almost certainly the mmu has been enabled May 02 07:58:15 ldr r1, [pc, #176]; ldr r0, [pc, #168]; str r1, [r0]; this is how i am trying to write May 02 08:01:21 veremit: apparently it is l3 RAM, omap3 cortex-a8 subsystem doesn't have local ram like later SoCs do May 02 08:01:34 o,o May 02 08:01:37 (not that that changes anything) May 02 09:01:10 zmatt, As you mentioned i disabled the MMU, now it runs fine.. Thank you May 02 13:22:30 Hi is it still neccesary to remove Connman on Jessie to get WiFi to work? May 02 14:22:52 hi everyone, i have BBB and im trying to put rootfs as f2fs, anyone was able to achieve this? can anyone point me some directions please? May 02 14:33:39 rtanakaz: which kernel are you using? May 02 14:37:04 hi tbr, 3.14.39-ti-r61 May 02 14:37:34 and that has f2fs? May 02 14:39:01 not sure tbr, i installed f2fs and added f2fs on initramfs modules, would that be enogh? May 02 14:39:35 how can i check ? May 02 14:40:07 how did you "install f2fs"? May 02 14:41:09 sorry, not f2fs, f2fs-tools via apt May 02 14:45:49 are you trying this on a SD card or on eMMC? May 02 14:47:39 tbr, on emmc flashing debian 8 from sd May 02 14:49:55 rtanakaz: have you successfully used f2fs on the bbb yet? if not you should start with that instead. May 02 14:52:42 tbr, not sure if i understood correctly. but at this point i have a f2fs partition (mmcblk1p3) mounted at boot (on /media/f2fs) and it seems to be working nice May 02 14:53:11 ok May 02 14:53:32 tbr, is this a good start? May 02 14:53:38 yes May 02 14:54:58 by working nice i mean, i can copy to and from f2fs partition, should i do any other check? May 02 15:04:37 sounds good. the rest will be dealing with adapting the flasher May 02 15:05:19 rcn's wiki and git repositories should help May 02 15:09:20 tbr, thanks for the feedback, do you mean this "init-eMMC-flasher-v3.sh"? May 02 15:24:06 Hi is it still neccesary to remove Connman on Jessie to get WiFi to work? I just can't get my UWN100 to work (ra0 doesn't show up) May 02 16:53:09 Hi i am new to beagleboard i am missing the bonescript module in the cloud9 folder May 02 16:53:12 any help? May 02 16:57:02 How do i get the bonescript in my cloud9 folder May 02 20:18:17 So I feel completely lost on my BBB, basically I just want to set it up to play movies using a simple media player over hdmi on a TV, everything I find seems to be way more complex than the simple project I want to use it for. Any help would be greatly appreciated. May 02 20:29:46 Hi there, I want to bring the USR0,1,2 led signals to external leds that I will be connecting to GPIO pins. How can I achieve that since I don't find any pins on the expansion headers that correspond to the GPIO signals related to these leds May 02 20:50:31 why exactly those/ May 02 20:52:46 So I feel completely lost on my BBB, basically I just want to set it up to play movies using a simple media player over hdmi on a TV, everything I find seems to be way more complex than the simple project I want to use it for. Any help would be greatly appreciated. May 02 20:56:42 spaghettios: you asked that already. May 02 20:57:13 spaghettios: try telling us more about the specific problems you are having, what ypou have tried already, anything. May 02 21:01:01 I keep searching for vlc but it doesnt seem to be available May 02 21:01:15 do i need a new image to run off of? May 02 21:01:39 where are you searching? May 02 21:01:49 google mostly May 02 21:01:53 What distro are you running? May 02 21:02:12 out of the box BBB May 02 21:02:26 That means nothing. May 02 21:02:34 Sorry Dude May 02 21:02:45 I have it turned on 1 sec May 02 21:02:54 run "cat /etc/issue" and tell me what it said. May 02 21:04:22 I'm in the gui, I run that command line under run and nothing happens May 02 21:04:41 open a terminal and run the command in that. May 02 21:06:24 debian gnu/linux 7 \n \l May 02 21:07:45 OK, good, the distro (distribution) you are runnign is Debian 7. May 02 21:08:05 So. Run "apt-cache search vlc" May 02 21:09:56 alright did that May 02 21:10:03 And? May 02 21:10:24 came up with a lot of options May 02 21:10:45 so I think now i need to brush up on commands May 02 21:11:48 Well, then to install it, use "sudo apt-get install <>" May 02 21:12:03 You probably want the one labeled just vlc. May 02 21:12:13 thanks, you definitely helped me out a lot here May 02 21:12:14 Or vlc-nox May 02 21:12:26 spaghettios: No problem, everyone has to learn Linux sometime. ;-P May 02 21:13:05 Hopefully I'll pick up and start being helpful to other new people May 02 21:13:10 thanks again and take care May 02 21:40:00 any clues on my question above ? May 02 22:23:55 * GenTooMan isn't sure? May 02 23:48:54 nacef__: the led pins are not on the expansion headers, they are separate pins May 02 23:50:05 nacef__: you can connect external leds to any gpios, there's nothing special about the ones used for the USR0-USR3 leds May 02 23:50:56 (some expansion pins also support PWM so you'd be able to dim your leds -- the pins used for the on-board leds do not) May 03 00:27:30 zmatt, thank you for the reply but how about if I want to display the activity shown by the onboard leds through the external leds May 03 00:27:47 that's software configuration May 03 00:28:17 you can select the trigger for each led May 03 00:28:41 see /sys/class/leds/* May 03 00:29:11 if you declare some other pins to be leds in your device tree, they will appear there also May 03 00:33:03 for example, if you'd want to disable that annoying heartbeat led, just do echo none >/sys/class/leds/beaglebone:green:usr0/trigger May 03 00:33:25 (and no, I have no clue why they are named "green" when the leds are, in fact, blue) May 03 00:34:22 or why they have such verbose names anyway.... I think "usr0" .. "usr3" would have been fine May 03 00:37:45 zmatt: is the SPI port on the expansion toy a McBSP port? also I remember KoTh saying the clock wasn't partcularly "fine" on it (IE must be a multiple of 48mhz) May 03 00:38:16 McSPI May 03 00:38:29 divisor of 48 MHz you mean May 03 00:39:42 zmatt: indeed, hmmm bummer. Gah it's like the original beagle board IO 'just enough to be useful for something' :D May 03 00:40:55 well it's non-trivial to have it otherwise May 03 00:41:18 either it would need a very high functional clock speed, or a pll or something May 03 00:42:39 keep in mind 48 MHz is not exactly a low speed for a simple serial bus May 03 00:43:22 and it's rarely important to configure an SPI clock rate with precision May 03 00:44:16 ehh actually I was hoping to interface into an audio D/A with 192ksps with 64 bit frame. All audio stuff requires odd ball clock rates :D May 03 00:44:38 why aren't you using McASP for that? May 03 00:44:50 also, use an external clock May 03 00:45:51 McSPI can use one, McASP on the am335x requires one (that's why there's a 24.576 MHz osc for it on the BBB) May 03 00:47:05 also wtf @ that sample rate and frame size May 03 00:48:28 I'm hoping the reason for that insane sample rate is that you're abusing an audio adc for non-audio applications :P May 03 00:49:16 dac I mean May 03 00:50:15 since anything above 48k is a waste of time and usually degrades audio performance May 03 00:55:02 It's not a really insane rate for audio most sampling and digital audio systems operate at the sample rate. Definitely professional gear has changed as the difference between consumer and pro has narrowed in the last 10 years. Decimating the sample rate just seems like treason too me I guess? May 03 00:55:47 afaik >48 kHz occasionally shows up as a fad May 03 00:56:11 but unless your audience consists of bats, there's no benefit May 03 00:56:29 and the same filter requires 4x as much computational power for every doubling of the sample rate May 03 00:56:44 in practice that means higher sample rate leads to crappier filters May 03 00:57:41 in adc/dac datasheets you'll also often find worse acoustic performance numbers (such as THD+N) at 96k or 192k compared to 48k May 03 00:59:29 (also, if some adc or dac prominently advertises 32-bit resolution as if it's a feature, back away slowly) May 03 01:01:01 * GenTooMan thinks zmatt is the first bat he's seen use IRC. "Their is a reason they give the circuits for that kind of testing. sometimes the parameters they measure rely on the circuit they used." May 03 01:02:11 Actually MOST times they do. So often you just need to test it, which is actually what I was doing cause I was 'curious' if one maintain the SNR and THD using a better output buffer (or worse) May 03 01:02:31 although my hearing extends annoyingly high (.... ARGH switching power supplies) I'm still no bat :P May 03 01:03:48 * GenTooMan ahhs "just driven 'batty'", "oh yeah the low frequency ones can be a pain when the inductor begins mechanically reacting do the switching magnetic field. can you use "hight pitch audio transducer" May 03 01:04:23 * GenTooMan is not a fan of PFM as a consequency. May 03 01:07:37 the TPS65217 can actually be configured to always use PWM for its buck converters (by default it switches to pfm when power consumption is low) May 03 01:08:23 "For low-noise applications the devices can be forced into fixed frequency PWM" May 03 01:09:30 they do that because it's more efficient (FET) power wise to use a constant on time for the switch. By using PWM the filter can have better tolerances (my guess). May 03 01:10:36 yeah the default is for efficiency May 03 01:11:31 and "low power" of like 100mw or something? :D May 03 01:11:37 I have no idea May 03 01:12:23 * GenTooMan has been fighting for uwatts he just finds some peoples idea of low power ... alien ? "it uses a 1ma ... ok get that down to 3.2uA and maybe we are in business!" May 03 01:13:27 Low power is a real pain. But I digress. So basically use the McASP instead (looks at schematic to see if their is access). May 03 01:14:04 there's one decently usable McASP, it's also used for HDMI audio by default May 03 01:14:15 the other McASP is a bit lacking in pinout May 03 01:16:59 * GenTooMan ohs. "ok ... most of the HDMI signals are on the bus. The original beagle board didn't have the digital audio port accessible what so ever. May 03 01:17:37 the BBB HDMI is actually a CAPE.. or at least treated like one May 03 01:17:40 BBW didn't have it May 03 01:18:19 quiescent current of the buck converters is apparently 15 μA btw May 03 01:18:45 which is perhaps not as good as you'd like, but still, considering they can deliver 1.2 A max :P May 03 01:19:19 (max current of just the cortex-a8 running at 1 GHz is 1 A) May 03 01:20:18 that's presumably not typical though... the a8 aggressively performs internal clock gating May 03 01:21:05 The only way to keep power down in most CMOS circuits is not clock them. Low voltage circuits have issues with leakage however. http://www.silabs.com/products/analog/dc-dc-converter/Pages/default.aspx <-- really low power stuff. May 03 01:21:41 as stuff gets smaller and smaller, leakage becomes increasingly a problem May 03 01:23:43 before ya know it them darned electrons start tunneling all over the place ;) May 03 01:24:34 And all the diodes will become tunnel zeners! May 03 01:32:40 luckily people are also trying to make circuits using... *grabs something random from wikipedia* ... surface plasmon polaritons! maybe that helps May 03 01:34:03 * GenTooMan suspects the polariton union is totally against that, "always someone tries to stop progress I tell ya, tunneling diodes at least don't get under your skin!" May 03 01:36:15 I was actually surprised that "spintronics" wasn't merely a research-thing May 03 01:37:44 using spin-polarized currents for, well, anything really sounded still a bit fanciful, but apparently it's used in HD read heads May 03 01:38:52 I believe they used it for memory devices (everspin?) as well. **** ENDING LOGGING AT Sun May 03 03:00:00 2015