**** BEGIN LOGGING AT Thu Sep 15 02:59:58 2016 Sep 15 07:29:01 Hi guys Sep 15 07:29:14 I have a question regarding the power consumption of the beagle bone boards Sep 15 07:29:45 I want to use it in a battery driven unit, is it possible to reduce the current consumption of 7mW in the Deep Sleep Mode to the uW range? Sep 15 07:32:04 without looking into the specs I suspect not. 7mW is already very low. Sep 15 07:32:13 how and how often you want to wake up the board? Sep 15 07:41:02 It depends on the unit Sep 15 07:41:10 from once a week up to once a year Sep 15 07:42:51 ok. Sep 15 07:43:30 so why not add a little controler (maybe just a flip-flop or something) that can be used to switch the board on and off? Sep 15 07:44:24 Good idea, i need to think about that :) Sep 15 07:45:49 like a flip-flop: your external 'power interrupt' sets it and it is reset by a GPIO (with some kind of delay to allow the system to shut down). teh flip-flop drives a FET that actually switches the board on and off. Sep 15 07:47:07 some µcontrolers also reach the µA range on sleep modes with interrupts still enabled. could be handy as the controler could also do the delay. or some other stuff depending on the project (like driving a status LED or something). Sep 15 07:47:50 hello Sep 15 07:48:29 Thanks already for the answer Sep 15 07:48:53 Does anyone here know if you can use an eMMC image taken from a BBB black and copy that image to a BBB black industrial? Sep 15 08:00:16 USA_Swed: yes, it will work Sep 15 08:00:47 it is exactly the same configuration of parts except the BBB industrial can operate in a wider temperature margin iirc Sep 15 08:01:01 from the element14 page: "The element14 BeagleBone Black Industrial is also software and cape compatible with the original BeagleBone Black." Sep 15 08:05:29 hmmm, well I just tried it and the industrial failed to boot afterwards =\ Sep 15 08:06:12 I have a question to the boot sequence Sep 15 08:06:14 how did you install the image from eMMC Sep 15 08:06:32 preferably, you would take the eMMC flasher image and put that on a microSD, and use it to write the eMMC Sep 15 08:07:11 if we try to boot from emmc with an SD card with a bootable image on it, it always boots the kernel from the SD card Sep 15 08:08:00 the U-Boot is from the emmc, if we don't press the button Sep 15 08:08:08 used the guide from elinux.org and used a FAT formatted micro SD card. Created the dtbs folder and moved the correct file over. That did the trick to get it to boot on the black. Took about an hour for it to finish even though the guide said 10mins Sep 15 08:08:52 it seems, the Beaglebone always mounts the SD card as mmcblk0, if there is a bootable image on it Sep 15 08:08:54 why? Sep 15 08:10:11 Afterwards when doing the restore to the industrial, edited the autorun script but the board didn't boot with the card. Copied over the dtb file to the main folder and it then booted correctly and started the restore. After solid LED, unplugged board, took out card, restarted the board and nothing happened after multiple attempts Sep 15 08:10:46 hmm.. Sep 15 08:16:06 indeed Sep 15 08:17:32 reflashed the industrial just now with stock Ubuntu 14.04 img that works like a charm. No button pushes needed. Reflashing now again with same image from the black after a successful reflash to another black. If this doesn't work then there is something else going on Sep 15 08:31:36 laurittr: thanks, That's a good thing to know Sep 15 09:03:06 found out, that even the wrong uEnv.txt is used when booting from emmc with a bootable image on the SD card Sep 15 09:03:44 seems to be a bootloader problem Sep 15 09:05:38 kremlin: no go mate. Looks like for now I just have to use a separate img for industrial and only do an industrial > industrial copy and black -> black Sep 15 09:06:22 USA_Swed: that is super strange Sep 15 09:06:45 element14 purports (their) units to be 100% software compatible Sep 15 09:32:59 very possible I'm doing something wrong, but very odd that flashing a black with the black img works but not to an industrial. Will keep troubleshooting. Wondering if something is different with the uEnv.txt file Sep 15 10:00:50 industrial has a different u-boot version Sep 15 10:02:06 I noticed, that the prompt of the u-boot command line shows: U-Boot# at hte Beaglebone Black, at the industrial it is => Sep 15 10:03:00 interesting. Do you know of any sneaky tricks to get it to boot from a black img? Sep 15 10:03:29 didn't try it yet Sep 15 10:05:21 you could try to excange th u-boot with the newer version (the one originally on the industrial) Sep 15 10:07:55 you mean update the uboot on the black with the newer one from the industrial? Sep 15 10:08:16 yes Sep 15 10:15:28 could do, except the problem is that the master img I want to take is on the black. And I want to put that img onto both a black and an industrial. The problem is that I have a black img (which I know works when flashing to another black) that will flash to an industrial, but afterwards the industrial simply fails to boot Sep 15 10:17:06 Just now I have taken the same black img and flashed the sd card with it. Then popped it into the industrial. It boots normally and black img is now being used. However, it does not flash the eMMC, not even if I push the S2 button Sep 15 10:23:49 try the other bootloader on the black, if it works, you might have an image that works on both Sep 15 10:24:55 ich think, there's a possibility that the original black u-boot is to old to know the revision of the industrial Sep 15 10:25:39 having issues with the button, too at the moment Sep 15 10:26:40 our blacks don't boot anything from sd, if they have an image on the emmc Sep 15 10:27:13 but if there is a sd card in the slot, they boot the kernel from the sd card - with the u-boot from die emmc Sep 15 10:27:28 unchanged u-boot Sep 15 10:27:51 i found that my black only booted from sd if I created the dtbs folder and moved the dtb file into it (when copying the eMMC to SD card) Sep 15 10:28:25 ah ok. So what if we downgraded the uboot on the industrial? Sep 15 10:29:43 for kicks I'm starting over and recopying the black img to sd card. All fresh reformatting and copying of files. Sep 15 10:31:51 whats easiest way to pull across the industrial bootloader and put onto the black? Sep 15 10:36:00 copy the MLO ant the U-Boot.img from the industrial and put them on the BEAGLE_BONE partition on the black Sep 15 10:37:37 if it works, the black should boot Sep 15 11:25:46 Boot sequence with button pressed is SPI->MMC0->USB0->UART0 Sep 15 11:26:20 how long do I have to wait for it to boot from MMC0 (SD-Card)? Sep 15 11:29:08 funny you sent that, i JUST read that on a nice page explaining things about uboot :) Sep 15 11:31:10 How do you mean? What are you doing now? Sep 15 11:36:00 just found out, that it is normal for uboot to only change the u-boot environement choosen by pressing the button Sep 15 11:36:46 From that point onwards the current u-boot env checks to see if the SDCard is installed and if it is, u-boot boots from the SDCard FAT partition. If the SDCard isn’t installed, it boots from the eMMC FAT partition. Sep 15 11:38:14 the question , how long the boot sequence waits for SPI is because we didn't wait for several minutes, when we saw the cccccccccccccccccccccccccccccc from the uart Sep 15 11:39:25 ok just confirmed without a doubt, copying a modified Ubuntu 14.04 img from a black does NOT boot when flashed to an industrial. There is something more needed, and right now I'm thinking we should copy the bootloader from the industrial to the black Sep 15 11:42:17 that's what i suggested in the first place... Sep 15 11:47:25 yes, but we're trying to figure out to do it :D any guides? Sep 15 11:59:14 I only did it on an SD card Sep 15 12:01:39 works fine, we changed the MLO and the U-Boot.img to get the other version Sep 15 12:02:57 seems to be a hardware issue, got the black to boot from another power source Sep 15 12:04:01 the first one supplies the black over the power pins of the expansion headers, the second one over the power connector Sep 15 15:44:59 I'm trying to log to journald but I don't seem to be able to find the indicated systemd/sd-journal.h file anywhere, and naturally google is too retarded to let me search properly so I can't find anything about locating it. I would have expected since I have systemd that headers would be present. Perhaps install libsystemd-dev? Sep 15 15:50:08 Ragnorok: Probably in the -dev package, ges Sep 15 15:50:10 yes Sep 15 15:50:18 Thanks. Sep 15 15:50:30 Could have just intalled it but preferred not to guess. Sep 15 15:54:18 Personally, I'm attempting to comprehend how to get PIN_9_19 to stop being high when I don't want it high Sep 15 15:54:45 Ugh. This is reinstalling systemd. wtf? Sep 15 15:54:54 erm Sep 15 15:54:58 Because debian is moronic? Sep 15 15:55:19 That's about windows moronic there. Sep 15 15:55:47 I expect I'd completely raped at this point, it's changed so much. Sep 15 15:56:29 It's overwritten huge chunks of /etc/udev and /etc/systemd. Sep 15 15:57:03 Great. It used the wrong kernel version as well. What the literal fuck. Sep 15 15:58:03 Since it raped all this shit "for" me I see no going back. Thanks windows-like OS. Sep 15 16:04:34 It booted. Don't know what's broken, but it booted. Sep 15 16:04:51 Happy days Sep 15 16:05:43 Still no header. Sep 15 16:06:23 Or at least find / -name "*sd-journal*" -print gives nothing. Sep 15 16:07:19 libsystemd-journal-dev Sep 15 16:07:40 Count on debian to fuck up packaging. Sep 15 16:07:45 systemd-devel in the sane world Sep 15 16:07:49 How do I discover that? I looked all over for how to find the file and came up empty. Sep 15 16:07:57 https://www.google.se/search?q=systemd/sd-journal.h+debian&ie=utf-8&oe=utf-8&gws_rd=cr&ei=LcfaV_zmBYHw6ATj94nYAg Sep 15 16:08:07 packages.debian.org is my go-to otherwise Sep 15 16:08:37 Ah. I wasn't including the full path. Sep 15 16:08:51 Thanks! Sep 15 16:10:05 This is typical. I literaly just installed libsystemd-dev and it says libsystemd-journal-dev can't be install because libsystemd-dev is too old. Sep 15 16:23:36 I copied the systemd/ to linux-headers/include/linux on the cross-compile system. That looks like where all the standard include bits are, but it still can't find it. Sep 15 16:25:08 I'm a doofus. Check the include path, doofus. (smirk) Sep 15 17:18:52 Ragnorok: Yeah. Sounds like you're tired and need coffee :) Sep 15 17:19:06 Personally, I'm just doing a little dance cause I finally got shit to work(!) Sep 15 17:19:13 \o/ Sep 15 17:19:40 Now I need to make it go from "Works, so so" to "Solid, proper, actually works" Sep 15 17:20:23 I just force-installed it and put the systemd header dir in my project include path. Since I can't get sd_journal_printv() to work with my macro I don't know if it worked, but it's "progress". Sep 15 17:20:50 That solid part always seems harder than it should be. Sep 15 17:21:44 Well, the guy we have who's good at this kind of stuff has been ill for a month. Sep 15 17:22:04 Which seriously hurts us. ( we're three people. Being without one? Not good.) Sep 15 17:22:35 So I'm sort of fumbling at stuff here, all the while fighting to comprehend what the heck is going on Sep 15 17:22:39 lol Been there, done that. I know your pain. Sep 15 17:23:16 Seriously, I ended up having to design a new base dtb file for the BBB ... Sep 15 17:23:21 That was... a bit out of my league Sep 15 17:23:46 Fuck, I didn't even know how dts files worked before this, right now I'm having nightmares ;P Sep 15 17:24:29 Ewww. We just did that as well, but it wasn't me, it was other guy. He did hardware, dtb, and low-level libs for the hardware, I'm doing the higher level bits. Sep 15 17:25:35 Ah, right Sep 15 17:25:41 I'm normally doing server side stuff ;P Sep 15 17:25:59 And networking. Sep 15 17:26:17 I don't think I've ever committed anything to the kernel not in the networking parts.. Sep 15 17:26:32 I guess I'm ahead of many in that I do know how the build system works and such, but yeah. Sep 15 17:28:00 The two of us are doing a real-time acquisition system. I've learned about ten different things since January. (grin) Sep 15 17:28:20 Now I'd like to learn why sd_journal_printv() is unhappy. Sep 15 17:28:48 Ah Sep 15 17:28:50 Good luck Sep 15 17:28:57 I haven't gotten around to journald yet Sep 15 17:29:00 Thanks! I'll get there. Sep 15 17:29:06 First I guess I need to beat Python into doing the tagged syslog thing Sep 15 17:29:19 Once I have that, journal next. Sep 15 17:29:42 Next mission is getting ostree up and running on the BBB Sep 15 17:30:50 Looks spiff. Sep 15 17:31:32 I'll have to keep that in mind for later. Sep 15 17:37:00 Mmm Sep 15 17:37:06 Means migrating off Debian thouh Sep 15 17:37:12 Debian is ill suited for it, sadly Sep 15 17:37:35 Since debian doesn't have proper support for stateless Sep 15 17:38:00 Too much cruft that they never get around to fixing, and with a hostile attitude to contributions I'm not inclined to help Sep 15 17:40:40 Well poo. Sep 15 18:29:43 A question about Device Tree Overlays... I have a .dts file. Compile into .dtbo. Copy into /lib/firmware. And yet it doesn't get used on bootup unless I also go to /opt/source/bb.overlays and do a install.sh. Does that make sense? Wouldn't putting in /lib/firmware be enough? Sep 15 18:30:08 I have my cape plugged in. And the rev etc are ok. Sep 15 18:31:01 bb.org-overlays to be precise... Sep 15 18:32:47 GeofC_: It could be that it's not inside the initramfs Sep 15 18:34:09 Ah. And /opt/source/bb.overlays/install.sh puts it there (among other things). Is there a way for me to copy it there directly? Sep 15 18:35:08 Not quite sure. Why don't you just put your custom dts file into the bb.overlays repo? Sep 15 18:36:05 Do you mean do a pull request etc? Or just locally? Sep 15 18:36:21 Just locally Sep 15 18:38:08 I think the install.sh has that effect doesn't it? Once I do the install, everything works. Sep 15 18:38:22 Wow. This is tragically not working. I can't believe I need to include eight support libraries to write a message to journald from C. That's insane. Sep 15 18:40:00 really? Sep 15 18:40:32 O.o? Worked for me stuff that was already being built into my image. Sep 15 18:40:36 with* Sep 15 18:40:41 I have to be doing something wrong. Sep 15 18:41:20 Ragnorok: http://0pointer.de/blog/projects/journal-submit.html ? Sep 15 18:41:32 Found that hours ago. Sep 15 18:41:51 I don't have a build environment on arm here, so I can't help with that. Sep 15 18:41:52 That's when I couldn't find sd-journal.h. Sep 15 18:41:58 Mmm Sep 15 18:42:22 That page doesn't mention special libs, and it's the most informative one I've found on journal from code. Sep 15 18:42:34 pkg-config should cover that though? Sep 15 18:43:06 The current issue is the cross-compiler environment. I build on Debian and copy over to the BBB. Sep 15 18:43:15 Ah Sep 15 18:43:29 It's the cross-compiler that's demanding crap tons of libs just to build. Sep 15 18:43:34 Sorry, we build most stuff native due to that shit Sep 15 18:43:54 Got a 4-way arm system for builds, running builds inside docker containers. Sep 15 18:43:57 For the most part I haven't had issues. I drop stuff in the project libs, run ldconfig, drive on. Sep 15 18:44:26 Yea *nod* Sep 15 18:44:36 "wget not found" well stab you too Sep 15 19:13:37 Sweet. The magic is to drop the libs from the BBB into /usr/arm-linux-gnueabihf/lib and run ldconfig, then the cross-compiler will pick them up without having to list them all in the project. Sep 15 19:49:30 is there a gnd on the beagle that doesn't run through the pmic? Sep 15 23:23:56 Hi, I'm using a XM as a print server. It crashes (ssh: no route to host) from time to time. I've disabled DHCP, but the problem persists. Any hints? Sep 15 23:27:56 is that a new problem? Sep 15 23:28:07 have you examined it over the serial console? Sep 15 23:39:44 tbr: I don't have the serial D cable Sep 15 23:40:13 I've been observing this for a long time (several months, maybe more than a year) Sep 16 01:36:59 ayjay_t: there's no ground that *does* Sep 16 01:37:12 ayjay_t: why would gnd run through the pmic? Sep 16 01:46:19 I've only ever seen or heard of such an arrangement to measure total current flow of a battery-powered device, and even there I'm not sure why the ground was used instead of the positive lead Sep 16 02:02:51 oh really? Sep 16 02:04:38 its just the pmic was super hot when i had a backlight connected to ground Sep 16 02:04:44 and it says dgnd so i figured it was a gnd on the pmic Sep 16 02:06:23 which ground pin(s) did you use and how much current was flowing through into it? (and also from *where* exactly?) Sep 16 02:07:23 from a boost converter through a backlight to the DGND Sep 16 02:07:49 that answers exactly zero of the two questions I asked Sep 16 02:08:04 haha okay it was pin hold on Sep 16 02:08:07 although my second question was ambiguous I guess Sep 16 02:08:16 probably 50 ma Sep 16 02:08:36 I meant: where was it getting the power from (it being the boost converter in this case) Sep 16 02:08:47 external Sep 16 02:08:54 uhh Sep 16 02:08:59 5 v Sep 16 02:09:07 my own dc jack on a breadboard Sep 16 02:09:17 then why are you using the DGND of the beagleboard instead of whatever ground belongs to that power source? Sep 16 02:09:23 * veremit has visions of spaghetti wiring .. Sep 16 02:09:29 its a floating 5 v source Sep 16 02:09:36 its like ac-dc but theres no dc pin on the ac side Sep 16 02:09:42 i mean theres no gnd pin on the ac side Sep 16 02:09:45 power always comes in pairs Sep 16 02:10:08 yeah but it seems to be referential Sep 16 02:10:14 ? Sep 16 02:10:18 ill brb Sep 16 02:12:36 zmatt: sounds like an epic ground loop .. excited by a switching converter .. nice :D Sep 16 02:12:56 I'm wondering how it could have worked at all Sep 16 02:13:21 electronics is funny like that .. Sep 16 02:13:54 yeah especially a boost converter... maybe capacitive coupling and leakage was sufficent? Sep 16 02:14:31 okay so i mean, this is my theory, i know it sounds weird and i could be wrong.... it's a two prong ac-dc converted fed into a boost converter Sep 16 02:15:03 when i probe the ground or the vout on the boost converter it looks like it's still an AC signal, with a 5 V difference between the ground and vout Sep 16 02:15:17 probe relative to *what* ? Sep 16 02:15:28 the probe is relative to the actual ground Sep 16 02:15:33 whose actual ground? Sep 16 02:15:36 What ground? Where? Sep 16 02:15:37 measured at what point? Sep 16 02:15:37 the wall Sep 16 02:15:42 *shudder* Sep 16 02:15:53 there is no "actual ground" Sep 16 02:16:00 i mean as opposed to a virtual ground Sep 16 02:16:06 * veremit lol Sep 16 02:16:28 there is 'Earth' which is as it describes .. Earth potential .. the potential of the planet Earth. Sep 16 02:16:48 i mean, okay earth Sep 16 02:16:51 earth ground Sep 16 02:16:55 there's just Earth potential, which is quite variable and only matters for electrical safety, and otherwise you only have local islands of conductor that you declare to be ground Sep 16 02:17:05 earth is not a useful measurement reference Sep 16 02:17:07 yeah those are virtual grounds Sep 16 02:17:10 but zmatt is correct .. power or 'potential difference' is the difference between two points. Sep 16 02:17:20 voltage, not pwoer :P Sep 16 02:17:22 *power Sep 16 02:17:22 hence it comes in Pairs .. a 'send' and 'return' if you like Sep 16 02:17:24 zmatt: ahem. Sep 16 02:17:35 guys did i really say something so stupid that you feel this needs to be explained to me? Sep 16 02:17:39 ayjay_t: yes Sep 16 02:17:44 okay well it doesnt Sep 16 02:17:54 and i'm sory for misspeaking Sep 16 02:18:10 ayjay_t: if you're using earth as a measurement reference, you're pretty much guaranteed for seeing a big 50 Hz combined with tons of garbage Sep 16 02:18:51 isn't the BBB's ground and earth ground when it's usb'ed in? Sep 16 02:18:58 Nope Sep 16 02:19:14 oh thats the presumption i made Sep 16 02:19:31 "earth ground" ? you mean usb shield? Sep 16 02:20:11 i thought that if you had your BBB usb'ed to your computer, the DGND was connected to the USB gnd which was earth(wall) ground Sep 16 02:20:14 the bbb has no earth connection, nor any use for it since it has no dangerously high voltages Sep 16 02:20:42 usb has ground and shield, they are normally not connected together Sep 16 02:21:00 but i'm not probing the USB shield, i mean whatever ground is coming on the usb ground pin Sep 16 02:21:13 i never probed the USB Sep 16 02:21:20 that's signal ground and will connect to the BBB ground plane yes Sep 16 02:21:35 so signal ground = earth ground? Sep 16 02:21:40 i still want that to be true Sep 16 02:21:42 there's no earth ground Sep 16 02:21:56 what do you call the walls ground then? Sep 16 02:21:58 it might, eventually, be earthed somewhere, or not Sep 16 02:22:16 so nitpicky Sep 16 02:22:23 it's not a nit Sep 16 02:22:48 earth or "safety ground" is to protect people from electrocution, it's not a signal ground Sep 16 02:23:26 i figured that they were usually tied together Sep 16 02:24:10 ayjay_t: not recommended .. what would happen if there was a lot of 'noise' on the earth ground? Sep 16 02:24:19 isnt' that what caps are for Sep 16 02:24:23 or worse, a potential difference Sep 16 02:25:11 interconnecting two different earth potentials (say between buildings) can result in very very high currents Sep 16 02:26:01 i dunno guys i have a 5V AC-DC running into a BOOST then going into an LED chain and then back to the BOOST ground, but it behaves a lot more sensibly if i connect the boost ground directly to DGND, although the PMIC was getting supper hot Sep 16 02:26:09 I once read a nice post by someone who learned about this the hard way by being just in time to stop the smouldering cable from turning into a fire Sep 16 02:27:50 ayjay_t: if you need signals to cross between those two power domains, interconnect their grounds thoroughly Sep 16 02:28:27 thanks zmatt Sep 16 02:29:09 Yeah that could be messy otherwise. Sep 16 02:29:14 if this results in significant current flow through the ground interconnect (as measured by a current meter or if you notice anything getting unusually warm), disconnect them immediately -- your power supplies aren't properly isolated Sep 16 02:30:14 having said that, I've once seen 100 mA flowing between the signal grounds of two USB ports of the same computers, figure that Sep 16 02:30:26 anything getting hot is usually bad Sep 16 02:30:52 well, plenty of things can get legitimately hot Sep 16 02:30:53 yeah i mean, i might be a little out of my league here Sep 16 02:30:55 zmatt: guessing they weren't connected together .. Sep 16 02:31:03 i'm working but i don't think we properly studied grounding in school Sep 16 02:31:16 i need to read Sep 16 02:31:45 ayjay_t: you do know the notion of a closed circuit? as in, electricity always flows in loops? Sep 16 02:32:02 (ignoring radio signals for simplicity :P ) Sep 16 02:32:40 zmatt: yeah i do Sep 16 02:32:50 i mean, again i'm not that dumb Sep 16 02:33:05 and i could figure it out but i was really surprised that my ground... needed to be grounded Sep 16 02:33:08 ayjay_t: ok, hence a power supply always consists of a pair of pins, and that pair should be used to power something Sep 16 02:33:33 in general, ground should not be assumed to be at any particular potential Sep 16 02:33:42 right but again i've got a power supply and a pair of pins and the ground pin wants a ground pin coming from another device, for some reason. Sep 16 02:34:04 if i connect it to the ground latch on my scope, i get a clearer light. if i attach it to the ground in my multi, same thing, or my BBB Sep 16 02:34:17 'ground' enabled you to connect two different systems together .. and have a common reference point. Sep 16 02:34:19 then, provided at least one of the two is floating, you can interconnect their grounds Sep 16 02:35:12 voltage = potential _difference_ Sep 16 02:35:54 again like, i have a BA in EE Sep 16 02:36:14 like i know, V = |Edx Sep 16 02:36:33 and i know a BA doesn't mean much either Sep 16 02:37:01 yet grounding wasn't properly covered? ... Sep 16 02:37:09 it is so essential in so many ways Sep 16 02:38:50 yeah i mean not really, circuits 1 is kirchov's and whatnot but everything is ideal and we basically just solve systems Sep 16 02:39:37 note btw that even though you'd hope an ungrounded wall wart's output is properly isolated from the mains, this is actually not necessarily *quite* true. through some high-resistance path or capacitive coupling the output can be tugged around quite significantly compared to earth or compared to another supposedly floating device Sep 16 02:40:01 at least until you ground them together Sep 16 02:40:25 my other classes were mostly communications and some semiconductor physics Sep 16 02:41:04 (which btw if done "live" may result in a brief surge current of significant magnitude) Sep 16 02:41:13 this is interesting stuff Sep 16 02:42:47 if you can't ground stuff together without making a mess, there's of course always opto-couplers or for example these things: http://www.linear.com/product/LTC2862 Sep 16 02:43:15 it tolerates +/- 25V ground potential without affecting functionality Sep 16 02:43:42 i kinda love optos Sep 16 02:43:51 i should back in here more i love talking to you guys Sep 16 02:43:54 i gotta go for now though Sep 16 02:45:28 zmatt, do you recommend a PMIC? Sep 16 02:46:12 uhh, I recommend using a PMIC for applications for which a suitable PMIC is available and recommended :P Sep 16 02:46:25 yeah i guess there pretty varied Sep 16 02:46:33 your question is too vague Sep 16 02:46:36 just browsing TI's catalog now, i didn't expect so many options Sep 16 02:47:47 even the tps65217 for the am335x has at least four variants (differing in how they are programmed) Sep 16 02:49:36 e.g. since the BBB uses DDR3L memory it should actually have used the TPS65217D, but it uses the C version instead (intended for DDR3) so u-boot adjusts the DDR3 supply voltage in the pmic before releasing DDR3 reset Sep 16 02:50:11 the memory supports both DDR3 and DDR3L so this is tolerated by it Sep 16 02:52:57 why do you love optos? I thought they were typically rather slow and highly variable in actual parameters? Sep 16 02:53:03 (see e.g. second paragraph of https://en.wikipedia.org/wiki/Opto-isolator#Phototransistor_opto-isolators ) Sep 16 02:55:36 I'm actually quite curious though how Linear made a differential transceiver that works even if there's a 25V common mode potential Sep 16 02:55:55 (i.e. far outside the range of its supply pins) **** ENDING LOGGING AT Fri Sep 16 02:59:59 2016