**** BEGIN LOGGING AT Fri Jul 31 02:59:59 2015 Jul 31 03:00:04 zmatt: whatever is in the published wiki. Jul 31 03:00:18 yeah, the ground plane is broken up Jul 31 03:00:32 ddrown: what are you calling "earth"? Jul 31 03:00:40 ddrown: and multiply-connected rather than a tree Jul 31 03:01:11 zmatt: well, it's a tree structure on the bottom Jul 31 03:01:26 topside is more broken up Jul 31 03:01:56 but no ground plane underneath the NS-T module? Jul 31 03:02:01 ds2: grounding connection on my scope, which should be connected to the house's ground Jul 31 03:02:37 * stt_michael quietly snickers to himself Jul 31 03:02:38 zmatt: would this be easier to see if I exported the top and bottom as images? Jul 31 03:02:38 ddrown: keep in mind that any loop of wire will pick up crap... the bigger the loop, the worse Jul 31 03:03:24 ddrown: to see what guidelines related to high-frequency signals look like, see for example this appnote on ethernet -> http://www.micrel.com/_PDF/Ethernet/app-notes/an-111.pdf Jul 31 03:05:11 ddrown: ah, yellow is the bottom layer? Jul 31 03:05:26 http://imgur.com/a/vtY6Q Jul 31 03:05:28 ddrown: are you in the US? Jul 31 03:05:32 ds2: yes Jul 31 03:05:47 how is the Bone being powered? Jul 31 03:05:55 2 prong wall wart? Jul 31 03:06:01 ds2: yeah, 2 prong to barrel Jul 31 03:06:40 hmmm.... it could be it is floating or it could be you have a grounding issue in your facility Jul 31 03:06:42 hrm, those images didn't come out very well Jul 31 03:06:54 ddrown: hmmm, both planes are mostly-ground but partly sliced up or carved in? Jul 31 03:07:02 that intuitively feels like a bad idea Jul 31 03:07:27 zmatt: yeah, would it be better to redesign and have the bottom be just ground? Jul 31 03:07:45 might check to see if earth ground is really earthed as prescribed Jul 31 03:08:11 ds2: doesn't really matter anyhow, he's measuring a floating BBB... Jul 31 03:08:25 yeah, it's 38V if I measure from ground or from 5V Jul 31 03:08:33 and no current from either to earth ground Jul 31 03:08:42 random data Jul 31 03:08:44 so I think it's floating properly Jul 31 03:09:23 ds2: that question sounds ominous though Jul 31 03:09:36 zmatt: it does if he plugs in things like USB Jul 31 03:09:58 you mean, like a USB with external power? Jul 31 03:10:10 ddrown: usb will ground the BBB to your PC Jul 31 03:10:12 say an external hard drive Jul 31 03:10:18 oh, that usb Jul 31 03:10:35 external hard drive -- depends on whether it has a grounded power supply Jul 31 03:10:55 I can measure BBB ground to PC ground if that's something interesting Jul 31 03:11:10 none of those values are interesting Jul 31 03:12:04 ddrown: RF_GND being floating also can't possibly be right I would imagine Jul 31 03:12:23 I wonder what exactly their intentions are Jul 31 03:12:31 ohhh lookie Jul 31 03:12:33 it'd be floating if it was plugged into usb Jul 31 03:12:34 it's a split ground plane Jul 31 03:12:37 on the module Jul 31 03:13:58 I'm afraid I'm gonna have to admit I have no RF design experience, so I'm also not really qualified to help you redesign your pcb Jul 31 03:14:05 heh Jul 31 03:14:22 I can mess with RF_GND as an experiment Jul 31 03:14:25 a module like this will need very carefully adhere to design rules Jul 31 03:14:31 except I don't know what they are Jul 31 03:16:47 general guidelines with high speed signals are: avoid making big loops (like page 2 of that an-111.pdf I linked to showed, to prevent making a loop antenna), having a ground plane as pcb layer is a good idea Jul 31 03:17:22 if you have multiple ground planes, do not let a trace an another layer (especially high-speed signal traces) cross over the split between the ground planes Jul 31 03:17:33 yeah, I tried to make the TIMEPULSE/PLL lines as short and as straight as possible Jul 31 03:17:55 there's much more to it though Jul 31 03:19:06 hm Jul 31 03:19:41 keeping them short (so both sides are "lumped together") at least helps having to pay close attention to line impedance Jul 31 03:19:49 *helps avoiding having to Jul 31 03:20:55 put a series resistor on any high-speed output (unless that's already taken care of internally) Jul 31 03:21:35 but all this doesn't help with the question about what to do with the module's RF ground plane Jul 31 03:21:41 yeah, there's one on the PLL but not on the TIMEPULSE Jul 31 03:21:58 TIMEPULSE should have one on the module if it's needed Jul 31 03:23:08 but, based on location, it would seem TIMEPULSE2 and TXD1 are referenced to RF_GND... which is awkward Jul 31 03:23:55 you may need to avoid running your (digital) ground plane undernearth the module's RF ground plane Jul 31 03:24:11 hm Jul 31 03:24:14 if you didn't need TXD1 and TIMEPULSE2 I'd say let that part of the module stick out :P Jul 31 03:24:16 just on the layer directly below it? Jul 31 03:24:58 I don't know, the amount of capacitive coupling will depend on surface area and distance (and the properties of material inbetween) Jul 31 03:25:09 ok Jul 31 03:25:31 but you don't want to capacitively transfer all the shit on your digital ground to the RF ground of the module Jul 31 03:25:59 that makes sense Jul 31 03:26:00 but really, I have no idea how you need to design this cape right Jul 31 03:26:09 sorry Jul 31 03:26:26 no reason to be sorry Jul 31 03:26:43 trial and error is all I got Jul 31 03:27:04 no it's not Jul 31 03:27:08 you also have google Jul 31 03:27:12 true Jul 31 03:27:16 research is step 1 here Jul 31 03:28:06 yeah, I'm working on that too Jul 31 03:30:49 btw, eQEP modules can be used for counting and frequency measurement if you connect a signal to its "A" input Jul 31 03:31:44 yeah, I'm using the dmtimer modules hardware capture Jul 31 03:31:52 and TCLKIN pin Jul 31 03:32:04 which can run the dmtimer's clock base (sampled by 100MHz) Jul 31 03:32:51 so that github gist is the BBB measuring TCLKIN with the PPS Jul 31 03:32:57 dmtimer is quite probably the least capable module for the purpose :P Jul 31 03:33:18 eh, it's limit is 24MHz which is fine for me Jul 31 03:33:25 I mean in terms of functionality Jul 31 03:33:49 oh, all I need is a counter and hardware event capture Jul 31 03:33:56 which it has Jul 31 03:34:13 either-or though Jul 31 03:34:23 dmtimer does both at once Jul 31 03:34:46 don't need PWM output Jul 31 03:34:51 that's the either-or part of it Jul 31 03:35:07 only one pin assigned to each timer Jul 31 03:35:19 which makes comparing gps modules harder Jul 31 03:35:54 yeah but I meant it can't do time-measurement (using a stable clock source as reference) and pulse-counting on the same signal Jul 31 03:36:00 since for counting you need TCLKIN Jul 31 03:36:14 TCLKIN is a different pin Jul 31 03:36:30 while for capture you need the timer to use a stable clock source (not TCLKIN) Jul 31 03:36:33 it drives a timebase for all dmtimers to optionally use Jul 31 03:36:36 I know Jul 31 03:37:00 oh, you mean TCLKIN might be unreliable, throwing your measurements off? Jul 31 03:37:24 never mind, I think we're just failing at communication Jul 31 03:37:30 heh yeah Jul 31 03:37:33 take a look at eCAP and eQEP though Jul 31 03:37:42 eCAP can capture events and has a 4-entry fifo to avoid overruns Jul 31 03:38:14 my typical latency from event to interrupt handler is ~16us Jul 31 03:38:21 so I'm not too worried about 1 PPS :) Jul 31 03:38:26 eQEP's quadrature decoder can be disabled, and then it can count on rising/falling edge or both Jul 31 03:38:48 and it can measure the interval between events Jul 31 03:38:57 (the interrupt handler delay is column 4 in the gist) Jul 31 03:39:20 and it can generate a warning irq if no event has been seen for some time Jul 31 03:39:38 that could be useful Jul 31 03:39:43 (if TCLKIN disappears, counters that use it become unusable and depending on how they're configured may stall CPU access indefinitely) Jul 31 03:40:00 yeah, I've seen that Jul 31 03:40:09 reads are ok, writes stall the bus Jul 31 03:41:22 you can configure write-posted mode (depending on frequency range, see docs)... then at least you can notice the write-in-progress bit fails to clear Jul 31 03:41:31 yeah, I get a kernel panic Jul 31 03:41:47 ah then it's in write-posted mode Jul 31 03:41:57 there's an appnote about the dmtimers only working in one mode Jul 31 03:41:59 (in write-non-posted mode the cpu would freeze) Jul 31 03:42:03 must be posted Jul 31 03:42:26 hmmmm, iirc there are frequency restrictions on one or both modes Jul 31 03:43:17 bingo, posted mode requires timer freq < ocp freq / 4 Jul 31 03:43:37 ah, hence max 25 MHz Jul 31 03:43:44 I'm pretty sure another part says timer freq < ocp freq / 4 Jul 31 03:44:55 or actually, I think it was in period, ocp peroid * 4 + 1ns Jul 31 03:45:30 another benefit of the PWMSS modules: none of this bullshit :P Jul 31 03:45:37 heh Jul 31 03:46:03 the PWMSS modules were actually lifted from their C2000 line of real-time controllers Jul 31 03:48:01 it's clear that someone put actual thought in how to design modules that can be used without race conditions Jul 31 03:50:58 anyhow, I've got stuff to focus on Jul 31 03:51:11 yup, thanks for the talk Jul 31 03:51:21 I have new stuff to consider now Jul 31 03:52:34 when looking for pins and associated functions, I made a spreadsheet to get better overview, it may be of interest -> https://docs.google.com/spreadsheets/d/1CK5c-Cs8G1RtzGo-J3VJsD9m5K-fp06AncgeYWsdjSU/edit Jul 31 03:52:40 (particularly the orange tabs) Jul 31 03:53:13 handy Jul 31 03:53:43 note however that I generally drop embellishments from peripheral names that merely state how fancy they are, i.e. eHRPWM is just pwm :P Jul 31 03:54:48 and the TRM and linux sometimes use 1-based indexing for peripheral (inconsistently, both internally and with each other), I always use 0-based Jul 31 03:55:59 the BBB tab also has filter views (menu Data -> Filter views...) to sort the list in various ways Jul 31 05:18:57 o.O I think rather than caps to ground, maybe the lcd pins should have used some series resistance Jul 31 05:19:37 alternating white/black pixels results in > 5V peak-to-peak Jul 31 05:20:15 found a bit of resonance?! Jul 31 05:20:55 just overshoot/undershoot I presume Jul 31 05:21:27 the stubs I added to the expansion connector probably aren't helping either, but I need something to attach probes :P Jul 31 05:22:48 it's not really a surprise to see ringing in this situation, I hope all parties involved are okay with it Jul 31 05:22:57 hdmi framer seems quite tolerant though Jul 31 05:25:00 also if I set the reference-voltage of the digital probes below 1.2V, my alternating pixel sequence started showing up on the vsync pin... I'm really hoping that's just a measurement artifact Jul 31 05:28:38 it was Jul 31 05:42:39 stt_michael: http://gerbil.xs4all.nl/lcdc/2015-07-31-073710.png Jul 31 05:44:22 (pixel data ...0101010101010010 ) Jul 31 05:45:02 lol niiiiice Jul 31 05:45:34 grid lines are the sample points Jul 31 05:46:22 or at least, where the (pretty sinusoidal) pixel clock crosses the V_IH(min) of the framer... Jul 31 05:48:10 unfortunately I can only ground a subset of the digital probes... normally that's never been a big problem, but clearly it does affects things since grounding the VSYNC probe severely reduced the cross-talk from the alternating pixel lines Jul 31 05:48:58 of course, all 16 pixel lines are switching at the same time here... that's probably not nice to the signals sitting right next to them Jul 31 05:52:30 mhmm Jul 31 05:53:54 I really don't understand why there are no series resistors near the am335x on these lines for impedance matching, like you'd normally do on any high-speed output Jul 31 05:55:05 I doubt they'd interfere with alternate uses of the pins Jul 31 05:55:48 I'd love to see a photo of your measurement setup. I'm just getting into dabbling with fancy scopes and seeing how more experience folks do it is always cool. Jul 31 05:56:19 would you also see power savings from limiting the current dumped into the trace capacitance? Jul 31 05:57:04 heh, this one isn't too bad actually... when I was doing measurements on the power up/down sequences at some point the poor BBB looked like it was on an intensive care unit or something Jul 31 05:57:41 oh yeah :) Even the tidy braided probe cable still terminates in a little Medusa. Jul 31 05:58:29 I do like this scope though (Agilent MSO7054B) Jul 31 05:58:38 the UI is sometimes a bit... quirky Jul 31 05:59:15 then again, you can also fully configure and control it via the network Jul 31 06:01:58 (I don't have anything to take a pic of the setup btw, sorry :) Jul 31 06:02:31 I think someone did take a pic of the power up/down measurement setup though, iirc Jul 31 06:08:53 yup, that's a nice scope alright. I'm still learning my way around the msox4154a in the lab. Looks like nearly the same UI between the infiniivision/infiniium lines. Jul 31 06:09:39 oh wait, they're both infiniivision, yeah, brainqueef. (There's an infiniium in the lab where I used to work, it's a bit different.) Jul 31 06:10:02 yeah, in fact the help screens not uncommonly contain conditional parts ("On some models...") ... which annoy the hell out of me, lazy ass programmers Jul 31 06:10:18 yeah, that's a bit... cheap. Not like the scopes are bargain units! Jul 31 06:10:46 You could always edit the help text; the filesystem is pretty easy to get at. ;) Jul 31 06:11:07 there are also some finer points that are non-obvious Jul 31 06:11:08 (telnet to the scope, login infiniivision password skywalker1972) Jul 31 06:11:21 lol seriously? Jul 31 06:11:26 that's.... pretty bad Jul 31 06:11:38 telnet: Unable to connect to remote host: Connection refused Jul 31 06:11:50 http://imgur.com/8rVbcq0 Jul 31 06:12:38 oh that's odd. the 4000 series anyway shares firmware with the 3000 series, and there's quite an extensive scope-hacking thread on the eevblog forums. I don't know how different the 7000 is firmware-wise. Jul 31 06:13:01 maybe they closed it in a firmware revision Jul 31 06:13:18 nmap it ;) Jul 31 06:14:41 too lazy Jul 31 06:14:52 (to look up the right options to pass to nmap) Jul 31 06:15:26 still I wonder, do other people *not* have the 1-pixel shift in their HDMI image? or did they just not notice? Jul 31 06:15:38 or will I find a long forum thread if I bother to check the forums? ;) Jul 31 06:16:10 I bet they just shifted the image one pixel and went on with their day! Jul 31 06:16:35 not that simple, the rightmost pixel of the screen shifts to the leftmost pixel of the next line Jul 31 06:16:44 last pixel becomes first pixel of the next frame Jul 31 06:16:54 oooooh Jul 31 10:11:53 Where do I get the flasher image for the BBB? (elinux.org is down) Jul 31 10:14:42 beagleboard.org Jul 31 10:14:53 /latest or somesuch Jul 31 10:15:13 http://beagleboard.org/latest-images Jul 31 10:15:17 @tbr They point to elinux.org. Jul 31 10:15:35 "For testing, flasher and other Debian images, see http://elinux.org/Beagleboard:BeagleBoneBlack_Debian" Jul 31 10:15:52 well wait a bit, elinux will come back. Jul 31 10:16:12 Any mirror. I'm a bit under pressure with setting that thing up. Jul 31 10:16:15 or peruse archive.org Jul 31 10:16:25 *sigh* Jul 31 10:16:27 as the images are not hosted on elinux IIRC Jul 31 10:17:54 I already tried Google cache for , but exactly for that page it shows an error. :-( Jul 31 10:19:40 about 30s of perusing the searchengine of my least distrust yields: https://rcn-ee.com/rootfs/2015-07-08/flasher/ Jul 31 10:19:52 where do I send the bill? Jul 31 10:19:53 tbr: or more broadly, https://rcn-ee.com/rootfs/bb.org/release/ Jul 31 10:20:25 feklee: for relieving pressure, you should maybe ramp up your google-fu skillz a bit :-) Jul 31 10:22:01 LetoThe2nd: your link is better btw, has not only console images but also lxde and 4gb lxde. not that I personally would want to run a desktop on a bbb... Jul 31 10:22:21 * LetoThe2nd bows Jul 31 10:22:23 Are there MD5 sums or sth. so that I can trust this source? Jul 31 10:22:38 you serious? Jul 31 10:23:15 if you want trust, rebuild everything from source (after reviewing it.) Jul 31 10:23:30 (including the toolchain that built the toolchain, of course.) Jul 31 10:24:15 I normally just don't download from random sources. elinux I would trust because it's linked from the BB web site. Jul 31 10:24:19 md5 (or any other checksum) only tells you about the integrity of the download. no trust involved at all. Jul 31 10:24:49 feklee: well it was you complaining about pressure, not us, right? Jul 31 10:28:06 feklee: and for integrity, the bmap files seem to be useful too. no idea about how to use them, though (and won't read up on it now.) Jul 31 10:29:42 anyways - friday now. Jul 31 10:33:54 feklee: if you want to establish the provenance of the images, I'd recommend to look at rcn's eewiki. He prepares the official images. He also provides instructions for building from scratch. Jul 31 10:34:38 * tbr also notes that elinux is a wiki and a rather sizeable amount of people has accounts and all accounts have write access AFAIU Jul 31 10:46:50 @tbr Interesting - thanks! Jul 31 13:48:22 Anyone have some advice for a noob looking for a proper sd card? Size, class? Jul 31 13:51:13 as big as you need, as fast as the device can support. Jul 31 13:51:39 and if you want to run an OS from it, forget the speed claims. it will be the random IO numbers that limit it Jul 31 13:52:37 it's largely hit and miss. I think Transcend used to be ok-ish. Some very specific samsung cards were good (later larger type was horrible tho). etc Jul 31 13:52:40 meh Jul 31 14:36:07 Hey guys, What is the most stable kernel I can run on my BBB? I've been having troubles with my custom one, and want a good baseline. Jul 31 14:39:09 * tbr has been happy with a 4.0 kernel from rcn (near vanilla, just a few patches) Jul 31 14:49:25 tbr, which one are you running? I see 4.0.(0-8) Jul 31 14:50:22 I think in 4.1 rcn has working capes and dtbo Jul 31 14:52:58 my custom kernel was based off of 4.1.3, and I have been experiencing random restarts(without clean shutdown). capes and dtbo aren't necesary for my application, but staying online is. Jul 31 15:47:06 fyi: OSUOSL is still working on restoring the database that elinux.org uses. Jul 31 15:47:39 tbr was it a full hardware failure? Jul 31 15:48:14 rcn-ee: no, they had some migration mishap and are splicing things together from an old backup and binlogs Jul 31 15:48:32 at least that was the plan that they emailed Jul 31 15:48:44 ahh... poor guys.. ;) Jul 31 15:49:02 yeah, I don't envy them. supposedly they were pulling an allnighter Jul 31 15:58:09 hello people, My beaglebone is not able to resolve DNS though it can ping to IP addresses, what to do? Jul 31 15:58:36 I have nameserver 8.8.8.8 Jul 31 16:01:33 Can you talk to google? Jul 31 16:02:47 no, but I can ping 8.8.8.8 Jul 31 16:03:12 tired of resolving this issue Jul 31 16:05:08 What does `dig @8.8.8.8 agmlego.com` say? Jul 31 16:06:03 cant resolve the DNS Jul 31 16:07:22 That is definitely not the actual text. Please past ethe exact text. Jul 31 16:08:03 it doesnt say anything... Jul 31 16:08:09 it stays blank Jul 31 16:08:20 but I think its a strange address Jul 31 16:08:37 it says nothing when I ping www.google.com Jul 31 16:09:14 ...it "stays blank". So, it exits immediately without printing anything, or it has not stopped running yet? Jul 31 16:09:51 no, it waits for Ctrl+c to be terminated Jul 31 16:10:04 doesnt print anything Jul 31 16:10:56 OK, so let it run until it times out or completes. Jul 31 16:10:59 Be patient. Jul 31 16:12:10 ohh.. ok.. Jul 31 16:13:25 says unknown host Jul 31 16:14:12 Interesting. COuld you please start pasting the exact text? Jul 31 16:15:27 root@beaglebone:/etc# ping www.google.com ping: unknown host www.google.com Jul 31 16:15:45 OK, please try the command I told you to run. Jul 31 16:16:11 dig @8.8.8.8 agmlego.com Jul 31 16:16:45 root@beaglebone:/etc# ping @8.8.8.8 agmlego.com ping: unknown host @8.8.8.8 Jul 31 16:17:39 ...please read and follow the *exact* command I gave. Jul 31 16:17:54 I did not say to use "ping" I said to use "dig" Jul 31 16:18:00 ok Jul 31 16:18:02 These are different tools. Jul 31 16:18:19 root@beaglebone:/etc# dig @8.8.8.8 agmlego.com -bash: dig: command not found Jul 31 16:18:28 ...what. Jul 31 16:18:37 alok_: What are you running on your bone? Jul 31 16:18:46 debian image Jul 31 16:19:38 OK. How is your bone connected to the Internet? Jul 31 16:19:55 through laptop... Jul 31 16:20:01 shared my internet Jul 31 16:20:11 That seems like a poor choice. Jul 31 16:20:12 saw that video on the youtube Jul 31 16:20:23 I know, it is poor Jul 31 16:20:55 Try plugging it into your network switch. Jul 31 16:23:00 well, my router doesnt share internet... but I can do that in the college tomorrow, I have dont that before Jul 31 16:23:16 done Jul 31 16:23:30 thank you so much agmlego Jul 31 16:25:13 Your router likely has a switch built into it... Jul 31 16:27:41 yes, I have tried it before, but didn't work. It works in my lab Jul 31 16:27:46 will have to go there Jul 31 16:28:01 I dont know whats the problem with my router Jul 31 16:28:06 Oh I hate this Jul 31 16:28:53 Might be time to fix your router. Jul 31 16:29:26 yup, trying that now Jul 31 17:36:30 what is the default password for debian ? Jul 31 17:37:05 http://elinux.org/Beagleboard web is down ! Jul 31 17:38:18 Yup, sure is. Jul 31 17:39:07 the beagle tells you the default password... Jul 31 17:39:17 mine does Jul 31 18:00:34 nerienna: Without knowing the password how can I ssh into the system Jul 31 18:01:30 get the error message - with the password Jul 31 18:03:01 temppwd Jul 31 18:03:51 just googled beaglerbone default password debian and noticed that I've read that before Jul 31 18:24:57 aswin temppwd Jul 31 18:26:21 aswin?! Jul 31 18:26:27 debian** :P Jul 31 18:27:00 beagler :P Jul 31 18:27:01 lol Jul 31 20:17:03 Hi I was posting a new topic about our open-source platform that beaglebone users could use as a resource but it was denied, is there a better place to announce this? this is a really cool project, very advanced software and not selling anything. thanks Jul 31 20:21:46 have you tried contacting elinux, jwgorman Jul 31 20:22:07 ping jkridner also Jul 31 20:23:20 Hi sst_michael, no I have not, let me do that but am not all that familiar with irc Jul 31 21:06:01 hii Jul 31 21:06:18 The Ethernet adapter on my Beaglebone Black board is disabled when I have a USB WiFi adapter connected to the board. How can I active both Ethernet and WiFi mode. Jul 31 21:07:01 nitin, wicd only handles one device, enable eth via /etc/network/interfaces ... Jul 31 21:13:09 wow people actually use wicd .. lol ;p Jul 31 21:24:03 stt_michael, yeah, it's the default "wireles" utilty in the wheezy image.. Jul 31 21:24:28 ew .. it didn't make it to connman? although if you want a gui .. I really think networkmanager works .. Jul 31 21:25:14 connman was really old... too old... network manager required too many extras... Jul 31 21:25:27 hrm oh well. Jul 31 21:25:35 didn't you 'fix' connman? Jul 31 21:25:41 or build from source!? Jul 31 21:25:56 well, we backported a newer connman for jessie... and using "cmst" for the gui.. Jul 31 21:26:06 aha Jul 31 21:26:30 I've yet to do a full update to jessie I think Jul 31 21:26:41 oh wait .. perhaps my wand build is there **** ENDING LOGGING AT Sat Aug 01 02:59:58 2015