**** BEGIN LOGGING AT Fri Apr 22 02:59:58 2016 Apr 22 03:35:53 ok Apr 22 03:36:18 did it with python :) Apr 22 03:56:20 Can anyone point me to a simple c/c++ code snippet that opens a ftdi usb-to-serial device and writes a character or string? I'm using Linux beaglebone 4.1.15-ti-rt-r43 #1 SMP PREEMPT RT Thu Jan 21 20:13:58 UTC 2016 armv7l GNU/Linux, and for some reason cannot get anything to work. Help? Apr 22 03:56:43 I have the FT2232H mini-module. Apr 22 03:58:11 I'm trying to play audio to my hdmi device but "aplay -l" shows "no soundcards found". Am I missing a driver or something? Apr 22 04:01:56 Non_: so you are connecting the ftdi over usb to the bbb? Apr 22 04:05:02 tbr, yes, I have the BBB connected to the USB port of the FT2232H. the lsusb shows the device. The simplest example along with compile instructions would help me immensely Apr 22 04:05:28 I'm not an Embedded Linux wiz, unfortunately... used to Vistual Studio, not Cloud 9 Apr 22 04:14:09 Non_: when you plug in a device node /dev/ttyUSB0 should appear Apr 22 04:15:48 tbr, yes it does. But when I try and write a small C program the ftdi.h include file is not present. Don't know where to find it. Apr 22 04:20:27 Non_: you can then just open it as a file and write to it Apr 22 04:21:21 even works from the command prompt: echo "moo!" >/dev/ttyUSB0 Apr 22 04:21:52 I need to be able to set the speed of the device. FTDI provides those functions in the D2XX and using Visual Studio c#. But the VCP is builtin to the Debian kernel as I understand it, so I don't know which function call will let me set the speed/baudrate Apr 22 04:21:59 in c it would be fopen() and then write to that Apr 22 04:22:19 ah, ok Apr 22 04:22:54 Thanks for helping and being patient by the way Apr 22 04:25:46 somewhere I've seen the ftdi_setbaudrate or ft_setbaudrate functions being called in c/c++, but that assumes the ftdi calls are present. Apr 22 04:25:54 you'll have to import termios.h it seems Apr 22 04:26:10 it's a generic serial port Apr 22 04:26:19 you don't need ftdi specific functions Apr 22 04:26:30 http://man7.org/linux/man-pages/man3/tcsetattr.3.html Apr 22 04:29:37 The FT2232H will run at 12MBaud and I've never been able to get it to run that rate with anything less than the D2XX functions. But the FTDI tech said the standard VCP package that is part of the Debian Linux build should do it. So I set about trying to write a small app using the ftdi calls and no luck. Apr 22 04:30:39 you might need to use a multiplier Apr 22 04:32:58 you could just try using the setserial command to get a feeling for what works Apr 22 04:36:39 Will do tbr, thanks much! Apr 22 04:37:26 it looks like the ftdi kernel driver determines multipliers internally, so setting baudrate to 12000000 might work directly Apr 22 06:58:39 JOIN /qt-creator Apr 22 06:58:48 sorry Apr 22 08:53:14 hi Apr 22 08:59:24 hello Apr 22 11:17:58 Hi Guys Apr 22 11:18:38 I am triying to connect over USB to BBB, and cannot ping 192.168.7.2 Apr 22 11:18:45 I am triying to connect over USB to BBB, and cannot even ping 192.168.7.2 Apr 22 11:19:35 I use ubuntu... What is funny it sometimes work sometimes not so I need to reboot BBB few times and than works - I hate such situations - is thihs known bug ? Apr 22 11:20:19 ANd also from official BBB page this url link is dead: https://www.linux-usb.org/gadget/ Apr 22 11:21:16 IS anybody live here ? ;) Apr 22 11:34:18 So when I connect USB with BBB, sometimes ifconfig shows eth1 has inet addr:192.168.7.1 and sometimes only ipv6 is showed. Apr 22 11:34:45 in that later case no way to go to 192.168.7.2 Apr 22 11:35:09 so what could be the reason that this is not working 100% Apr 22 11:35:13 so what could be the reason that this is not working 100%? Apr 22 15:28:38 lots of reasons usb networking doesn't work .. but usually its (1) drivers. (2) config. Apr 22 15:29:03 (0) PEBKAC Apr 22 15:29:22 well .. that too Apr 22 15:29:44 there really isn't anything wrong with serial debug or ethernet .. :D Apr 22 15:30:06 i still don't understand why usb gadget networking is the "preferred" method .. Apr 22 15:30:14 so many pitfalls Apr 22 15:34:15 because in _theory_ it's easy Apr 22 15:34:44 also people are too cheap to buy FTDI usb2uart plugs Apr 22 15:37:03 theory .. haah,yea I did that once. Apr 22 15:58:34 hello Apr 22 15:59:57 I have a beaglebone black running debian jessie with linux 4.4.6 Apr 22 16:00:04 and I have an "odd" issue Apr 22 16:00:17 if I connect it to my laptop through usb Apr 22 16:00:30 I almost always have to hard reset it 3-5 times Apr 22 16:00:43 until the usb's eth interface on the laptop gets the ip address from the beaglebone Apr 22 16:01:32 I have tried with different boards, and it happens with all of them Apr 22 16:02:14 I have tried with both linux mint and ubuntu on several different computers Apr 22 17:07:36 ah, now i remember why i had grabbed the u-boot-ti patches ... beagle-x15 doesn't boot off of MMC with mainline u-boot. Apr 22 17:09:19 d'oh they're still lagging .. but then x15 does 'exist' yet Apr 22 18:24:38 hey there guys, out to build a high density current meter with a beaglebone black, luckily a lot of ppl have built these already so I have almost everything I need to build this. The idea would be to use the 8 analog inputs on the beagle to read data from non invasive current meters Apr 22 18:27:06 but these meters go up to 5V while the beagle has a 1.8V peak, so I was planning to use a voltage divider, a quick calculation here https://learn.sparkfun.com/tutorials/voltage-dividers gave me the values I needed Apr 22 18:28:40 I was trying to use low resistance resistors because I read somewhere that impedance was something to keep in mind, but then I found solutions elsewhere where they said that using high value resistors was better, so here comes my question Apr 22 18:29:31 you probably want either a current transformer or sense chip Apr 22 18:29:37 when building a voltage divider to feed data into the beagle ADC, should I use high or low value resistors ? and why ? Apr 22 18:30:01 Janos: I've used this .. http://www.linear.com/product/LT6106 .. works well Apr 22 18:30:19 esp. with micros Apr 22 18:33:38 veremit, thanks a lot for the recommendation, but but question is a bit more theoretical. What's the effect of impedance on the beagle's ADC and what value ranges should I target Apr 22 18:34:23 I don't know about the BBB's adc .. you'd have to consult the TRMs and associated docs Apr 22 18:34:40 veremit, thanks a lot, will do Apr 22 18:34:42 there's a chance zmatt might be able to shed some light on that .. i know he works in the analogue domain some Apr 22 18:35:11 awesome, i'll stay around and wait if he respond Apr 22 18:36:40 hmm? Apr 22 18:36:45 * zmatt reads scrollback Apr 22 18:37:45 adc's zmatt .. impedances Apr 22 18:38:01 I say sense chp or current transformer for bbb current amp Apr 22 18:38:49 Janos: "these meters" you're referring to are active devices? Apr 22 18:39:10 i.e. the output is provided by something with fairly low output impedance like an opamp? Apr 22 18:40:04 zmatt, they are current sensors, something like this https://www.sparkfun.com/products/11005 Apr 22 18:40:41 ok, passive Apr 22 18:40:45 I was planning to hook 8 of those to the 8 analog inputs of the beagle Apr 22 18:43:01 the datasheet shows a 10 ohm resistance being placed across the sensor Apr 22 18:44:08 and the sensor output voltage being 300 mV when the primary current is 60 A Apr 22 18:45:16 using voltage dividers with 680Ω in R1 and 377Ω in R2 there should be Vout=1.78V when Vin=5V Apr 22 18:45:31 maybe read what I just said Apr 22 18:46:27 given the output curve of the datasheet, 5V output would indicate 1000A current being measured Apr 22 18:47:11 ok Apr 22 18:47:29 so no need for the divider ? Apr 22 18:47:44 that specific sensor goes up to 30A anyway Apr 22 18:48:28 this sensor is just a transformer, it outputs a current, not a voltage Apr 22 18:48:37 yes Apr 22 18:48:43 http://www.electronics2000.co.uk/calc/potential-divider-calculator.php <= my fav. calc Apr 22 18:49:37 so you need to place a resistor across it to produce a potential... they use 10 ohm as example but you may want to use a larger value to increase the voltage swing Apr 22 18:49:52 if you're going to measure alternating current you also need to bias it Apr 22 18:50:27 in all cases, I strongly recommend using clamp diodes to protect the beaglebone Apr 22 18:50:43 zener warrior diodes no less .. >,< Apr 22 18:50:55 or just 1n4148 lol Apr 22 18:51:06 zener? I would normally expect schottky Apr 22 18:51:21 ah you don't -need- schottky's .. Apr 22 18:51:31 you just lose a teeny bit of range Apr 22 18:51:31 from vss to signal and from signal to vdd Apr 22 18:51:49 although I gues with 1.8v swing you ain't got much room :D Apr 22 18:52:14 you need low voltage drop diodes for the clamping to be of any use Apr 22 18:52:25 otherwise you're already violating the max ratings before it kicks in Apr 22 18:52:52 eh I gues .. used to 5v micros too much :) Apr 22 18:53:20 iirc it's -0.3 to vdd+0.5 Apr 22 18:53:44 at least for the digital IOs, would need to check ds for the adc inputs Apr 22 18:56:01 zmatt: sounds about right Apr 22 18:56:11 Janos: fortunately the sensor can't deliver much current so that makes damage unlikely anyway Apr 22 18:56:23 so maybe you can skip the clamp diodes altogether Apr 22 18:59:04 though not sure... especially if the sensor may also be connected while the BBB is off Apr 22 18:59:41 what kind of current are you especting to measure? Apr 22 18:59:41 zmatt: don't think you can be too careful with the bbb surely? Apr 22 19:01:31 veremit: dunno, thanks to the split voltage rails the am335x gets current injected on many pins every time you turn one off Apr 22 19:01:57 the sensor can deliver only 15 mA at its rated max primary current of 30A Apr 22 19:03:53 btw I just noticed the accuracy specs of the sensor only apply if the resistance across the sensor is max 10 ohm Apr 22 19:05:21 passively biasing this thing right also isn't trivial Apr 22 19:07:57 is there no ready-made amplifier board to turn this thing into a proper signal? Apr 22 19:08:46 it's probably designed for a multimeter shunt or something. Apr 22 19:10:27 hmm, here's a passive circuit f or it... https://openenergymonitor.org/emon/node/58 Apr 22 19:11:20 Janos: replace the 18 ohm resistor by 10 ohm and you're done Apr 22 19:14:04 they make a centered reference voltage using a voltage divider and a 10 uF capacitor... that could work I guess Apr 22 19:14:19 yeah its probably 'good enough' Apr 22 19:15:06 my problem with analog stuff: I know the theory, but I don't have the experience to have a gut feeling what suffices for an adequate result Apr 22 19:15:47 zmatt: you do msot of your work in the digital and power domains then? Apr 22 19:16:09 veremit: technically I do software :P Apr 22 19:16:21 ah. fine. Apr 22 19:16:23 lol Apr 22 19:16:37 and yet you're scared of the kernel lol :P Apr 22 19:16:44 the kernel *is* scary Apr 22 19:16:50 baremetal coding, no problem Apr 22 19:17:07 only if you think of it that way Apr 22 19:17:20 it *is* baremetal coding .. just a lot of it :P Apr 22 19:17:32 no, that's not the problem. and "scary" isn't the right word Apr 22 19:18:04 bare-metal coding is 'scarier' imho .. Apr 22 19:18:19 but I think I know what you mean. And the kernel is a behemoth Apr 22 19:18:21 it's big, has tons of rules of its own that you need to abide to, subsystems you need to interact with in the right way, locking you need to get right, and most of all this is often undocumented or wrongly documented Apr 22 19:18:41 *nod* Apr 22 19:19:08 it amuses me that uboot has a 'mini-kernel' of its own now Apr 22 19:19:19 if you look at the average driver, very few lines of code actually interact with the peripheral Apr 22 19:19:23 most of it interacts with the kernel Apr 22 19:19:34 yeah I can i,magine Apr 22 19:19:56 my "drivers" in baremetal code tend to be a header file that declares the peripheral as a struct :P Apr 22 19:20:22 its odd coming from an assembly microcontroller point of programming .. all that *bloat* lol Apr 22 19:20:43 and the need to be so universal Apr 22 19:20:53 (something not permitted in the linux kernel since it's not portable to architectures where I/O isn't directly memory-mapped... I DON'T GIVE A SHIT ABOUT THOSE ARCHITECURES !@#$ Apr 22 19:20:56 ) Apr 22 19:20:56 so *many* different hardware permutations Apr 22 19:20:58 that Apr 22 19:21:19 but yuo understand why thats the case, right? Apr 22 19:22:27 I mean .. you could right a 100% correct kernel for a specific processor/ram/etc combination .. but would you ever want to .. and what if you had to do it like 100 or 1000 times?! :) Apr 22 19:23:30 like much of life .. you learn the rules. . and pick the ones you wanna break .. and how badly XD lol Apr 22 19:24:02 yes, but *I* don't care about any of those architectures, and the peripherals I want to use aren't on them Apr 22 19:24:14 some of 'em will kill you .. some will get you locked up .. the rest .. well ... :] Apr 22 19:24:15 fortunately I don't need to care since I don't need to do kernel programming Apr 22 19:25:14 at least, so far I've mostly managed to avoid it Apr 22 19:25:53 I actually wrote one "driver" (that iirc consists entirely of some data declarations, maybe one copy/pasted function) Apr 22 19:26:40 two functions even, woo Apr 22 19:29:12 veremit: but you're coming from an assembly microcontroller point of view... yet "bare-metal coding is 'scarier' imho" ? Apr 22 19:30:58 zmatt: depends which 'sphere' you're looking at .. a micro has bugger all peripherals ;) Apr 22 19:32:59 the ton of peripherals is just a buffet from which you can pick stuff you like while ignoring the rest Apr 22 19:35:58 and the micros I know actually also tend to have plenty of peripherals :P Apr 22 19:39:39 I mean, I can see how a 5000-page TRM can be a bit... intimidating. but you can ignore most of it, possibly at the risk of finding yourself having just implemented something using GPIOs only to have me mention "by the way, did you know there's a peripheral that already does all that in hardware for you?" ;) Apr 22 19:42:47 though even if not doing baremetal coding I can actually recommend anyone who uses the BBB regularly to occasionally just pick some TRM chapter you're not familiar with and read its introduction and skim through the functional description Apr 22 19:48:44 zmatt: I've taken a look at a few things over the years .. but mining datasheets for fun vs profit .. doesn't cut it with me lately lol Apr 22 19:49:40 heh Apr 22 19:50:04 if you're paying me to pour through .. different matter .. I'll read the whole TRM XD Apr 22 19:50:29 just don't quiz me afterwards .. I'm likely to be frazzled Apr 22 19:50:32 XD Apr 22 19:51:17 I've already done that "btw, there's a periperal for that" thing once here XD twice if I count the oddball who was hooking up an external 555 timer Apr 22 19:51:56 and lol, the TRM isn't exactly something I'd recommend reading cover to cover Apr 22 19:52:45 precisely lol Apr 22 19:54:14 but e.g. it's worth knowing the display controller can also directly interface with a variety of character display controllers (e.g. HD44780) _before_ writing a driver that implements the bus protocol using GPIO Apr 22 19:54:49 eh .. depend whether you can get to the driver control registers or not . Otherwise I'd bit-bang for speed. Apr 22 19:55:18 otherwise you're sweating around dodgy inexplicable kernel code for days when you could have a 'driver' in a couple of hours Apr 22 19:55:42 commercial realities .. I don't even have *spare* time to find out how to do things The Right Way any more. Apr 22 19:55:57 memory-map it (see http://pastebin.com/GrHwgYiR but replace adc by lcdc) Apr 22 19:56:00 I've read probably 3/4 of the TRM, for the same reason zmatt said; I wanted to know what's in there. For instance when I started with BBB dev I had no clue what a PRU was. I found it in the TRM. Apr 22 19:56:12 1) google, dont' reinvent the wheel 2) hack .. last resort (3) read manual .. (4) write a paper on how it should be done. Apr 22 19:56:44 cynical? me .. noooo.... XD Apr 22 19:57:52 veremit: few lines of initialization, and then you can just do a read/write to two special registers to create bus read/write transactions (with the 'Register Select' line high or low depending on which of the two registers you accessed) Apr 22 19:58:28 I agree much of this stuff is in need of examples though Apr 22 19:59:06 ah your crazy uio driver ;p Apr 22 19:59:16 which, granted, *is* quite funky Apr 22 20:00:08 not much crazy about it Apr 22 20:00:52 it's like /dev/mem, but safe, usable by non-root users, kernel handles power/clock management for you, and you can even receive irqs via the file descriptor Apr 22 20:01:19 most peripherals are then just as easily usable from userspace as they'd be on a microcontroller Apr 22 20:01:30 Uio works. Can't say the same about remoteproc. Apr 22 20:01:45 except you could probably also use python :P Apr 22 20:02:17 there ya go, https://docs.python.org/3.5/library/mmap.html Apr 22 20:02:18 :D Apr 22 20:02:20 I have barely tried python beyond 'hello world' .. which I've also managed in lua and ruby recently. Apr 22 20:03:07 mmap in python .. loosk good :D Apr 22 20:03:27 only question is whether it'll do a word-access if you read/write a word-aligned word Apr 22 20:03:32 rather than 4 byte-accesses Apr 22 20:03:46 but with a bit of luck it uses memcpy, which will be optimized Apr 22 20:04:00 probably .. if its configured right .. although the skim i did was bytearray ,,, I'd imagine it would be optimized, yeah Apr 22 20:04:27 ok I gotta shoot before the store closes .. milk stock is 'critical' .. Apr 22 20:04:42 go go go Apr 22 20:04:45 :D Apr 22 20:06:13 ok now I gotto try this Apr 22 20:06:30 I'll try the L3 interconnect registers, those are fussy as hell about access size Apr 22 20:08:06 otherwise there's still always ctypes of course Apr 22 20:11:26 hmmm Apr 22 20:12:05 Ragnorok: note btw that I'm not referring to uio_pruss Apr 22 20:12:46 Ah. I thought perhaps the mechanism may be similar. Not like I've looked. Apr 22 20:13:06 yeah it is Apr 22 20:14:19 in fact you could also use PRU using uio_pdrv_genirq just fine Apr 22 20:14:34 * Ragnorok recalls seeing examples to that effect Apr 22 20:14:36 has the benefit that this driver has been in mainline linux for ages Apr 22 20:15:34 and enabled in all of rcn-ee's kernels (all the way down to 3.8 series afaik) since quite a while Apr 22 21:10:17 I'm trying to play audio to my hdmi device but "aplay -l" shows "no soundcards found". I shouldn't need to run xwindows to get hdmi audio, should I? Apr 22 21:12:31 zmatt, veremit thanks a lot guys for the guidance, took me a while but turns out there are some current sensors that output volts and others current (i had no idea). Here are a couple of examples, volts: http://garden.seeedstudio.com/images/b/bc/SCT013-030V.pdf, current: http://garden.seeedstudio.com/images/1/10/SCT013-000_datasheet.pdf Apr 22 21:13:12 the later goes up to 100A and seem to have some diodes for protection when it's disconnected, are those the diodes zmatt was talking about ? Apr 22 21:13:12 Janos: the only difference is the built-in resistor in the former Apr 22 21:13:18 correct Apr 22 21:13:38 no, those diodes are to protect the sensor if no external resistor is connected Apr 22 21:13:50 ahh ok Apr 22 21:13:52 the former has a built-in resistor hence no need for the diodes Apr 22 21:14:02 awesome Apr 22 21:15:02 like I said, take the arduino example I linked to, replace the 18 ohm resistor by 10 ohm and you should be good to go Apr 22 21:15:15 according to your link which is great, there is a 141.4mA peak-peak current if combined with a burden resistor of 10Ω should be good Apr 22 21:15:25 jejeje yeah sorry :P Apr 22 21:15:37 JoeJulian: http://www.raspberry-pi-geek.com/Archive/2013/02/HDMI-and-the-BeagleBone-Black-Multimedia-Environment/(offset)/2 Apr 22 21:15:57 Janos: mV you mean Apr 22 21:15:58 you need the right display mode .. oddly enough Apr 22 21:16:18 and I think it's zero-peak ? Apr 22 21:16:35 mm mA i think, this sensor gives current, burden resistor needs to be added to convert to V Apr 22 21:16:52 then it was 15 mA, at least in the datasheet you linked to Apr 22 21:17:07 combined with 10 ohm resistor yields 150 mV Apr 22 21:17:55 this is the diagram with the current sensor https://openenergymonitor.org/emon/sites/default/files/Arduino%20AC%20current%20input%20A.png Apr 22 21:18:19 that's a different CT then Apr 22 21:18:59 Well I /thought/ I had the right display mode. I guess I'll try them all. Apr 22 21:19:01 though indeed it will have been rms, my bad Apr 22 21:19:36 this is the datesheet for that CT http://garden.seeedstudio.com/images/1/10/SCT013-000_datasheet.pdf Apr 22 21:19:42 Janos: and lol, looks like someone used google translate or something Apr 22 21:19:46 "Burden" instead of "Load" Apr 22 21:20:25 hey zmatt... how much do you know about the jtag stuff on the AM335x? Apr 22 21:21:04 ds2: enough that I could almost manage to do a bus transaction with switches connected to TDI/TMS/TCK and a led on TDO :P Apr 22 21:21:28 maybe a quickref for some IR codes, I don't know all of those by heart Apr 22 21:21:46 zmatt: okay - how important are the EMUn lines? Apr 22 21:22:07 so in the end if I replace the 33Ω burden/load resistor with a 10Ω it should give me around ~1.4V max which should be within range for the beagle's 1.8V, so far so good ? Apr 22 21:23:38 ds2: well that's a subjective question... other than that EMU0-EMU1 need to be sampled high by the cpu at POR-release for normal operation Apr 22 21:24:09 zmatt: yes, that is the official line - trying to figure out what happens if EMU0/EMU1 are floating :( Apr 22 21:24:17 they have internal pull-up Apr 22 21:24:40 someone told me those are very weak and may not charge up the stray cap before reset is done :( Apr 22 21:26:57 I've never heard of that happen, then again I've never seen a board where they don't have external pull-up Apr 22 21:27:12 hehe... indeed Apr 22 21:27:21 the internal pull *is* very weak, especially if VDDHV is 3.3V Apr 22 21:27:23 missed item...wasn't on the Ti check list Apr 22 21:27:36 Janos: I'll get back to you in a moment btw Apr 22 21:27:40 hoping all I would loose is the JTAG and not the entire board not working Apr 22 21:28:00 or maybe it comes up with all security disabled ;) Apr 22 21:28:14 no, holding EMU0 low brings the cpu up in wait-in-reset mode Apr 22 21:28:27 it will not boot Apr 22 21:28:29 but EMU0 = EMU1 = 0 is reserved Apr 22 21:28:29 zmatt, no rush, take your time Apr 22 21:30:08 EMU1 low is reserved on this one yeah. on some SoCs emu1 low is used to bypass the clock voting logic and pulling both low was used e.g. on the omap4 for a special boundary scan mode of the chip-to-chip interface Apr 22 21:30:26 am335x probably doesn't have any clock voting logic since none of the TAPs have rtck Apr 22 21:30:54 keep in mind btw that the caps do have a lot of time to charge up Apr 22 21:31:21 zmatt: so I should not scrap the boards cuz of this? Apr 22 21:31:23 there's iirc more than 20 ms between vddhv going up and POR being released Apr 22 21:34:49 back-of-envelope calculation says that if capacitance < 190 nF then even at the first-case spec of 19 uA pullup current it'll reach logic high in that time Apr 22 21:35:50 eh Apr 22 21:35:54 s/first-case/worst-case/ Apr 22 21:35:59 weird Apr 22 21:37:49 so dunno, it depends on how credible the source is of the claim this might be a problem, and his or your definition of "stray capacitance" Apr 22 21:38:13 I don't have much intuition on how much capacitance a typical PCB trace is, but I'm pretty sure it's nowhere near that amount Apr 22 21:39:27 of course I also don't know how much credibility the pull-up current specification has when it's given as "min 19 typ 100 max 243" uA Apr 22 21:40:10 for comparison, the leakage of a high-Z input is given as max +/- 18 uA Apr 22 21:41:32 hmmm 190nF is pretty hugh for strays Apr 22 21:41:52 if I had 0.190uF of strays i can save on a lot of caps :D Apr 22 21:42:21 yeah that was my reasoning also: if a pcb trace had that much capacitance, they wouldn't litter PCBs with 10 nF caps Apr 22 21:42:22 but then this is the source that was arguing the leakage onto VDD_3V3B is not an issue Apr 22 21:42:37 it isn't Apr 22 21:42:43 *nod* Apr 22 21:42:43 leakage *from* 3V3B however... Apr 22 21:42:54 sorry, from Apr 22 21:43:05 I pretty much dumped the 3V3B rail Apr 22 21:43:25 well you know what my solution was Apr 22 21:43:46 thanks... just wanted someone to bounce it against... it is annoying to discover missing pull ups a day AFTER the board has been released to fab :( Apr 22 21:44:32 yeah... the weak-as-shit pull-ups are more a concern w.r.t picking up glitches Apr 22 21:44:38 but that's not a problem for the EMU pins Apr 22 21:44:49 (unless timing is really really really bad) Apr 22 21:45:15 I guess the only reason they are on the jtag connector is for the wait in reset mode for the debugger Apr 22 21:45:23 not the only reason Apr 22 21:45:32 the debugger can setup alternative functions Apr 22 21:45:53 thought all 4 pins is needed to read the ETB? Apr 22 21:45:54 <_av500_> emu pins were always on TI jtag Apr 22 21:46:01 it can link them to the cross-triggering matrix (especially useful for multi-cpu debugging) Apr 22 21:46:04 <_av500_> we had them on DSC21 Apr 22 21:46:07 <_av500_> I recall Apr 22 21:46:11 DSC21? what's that? Apr 22 21:46:16 <_av500_> :) Apr 22 21:46:16 act like dbgreq/dbgack on the arm Apr 22 21:46:25 <_av500_> ds2: arm7 + c54 Apr 22 21:46:27 or use them for hw trace output Apr 22 21:46:29 <_av500_> for digital cams Apr 22 21:46:37 hw trace is also the only use of emu2-5 Apr 22 21:46:37 <_av500_> c54 + hw accell Apr 22 21:46:38 av500: wow... early Davinci? Apr 22 21:46:47 <_av500_> yes Apr 22 21:46:55 hwtrace != etb btw Apr 22 21:47:11 for basic break point, that isn't needed, right? Apr 22 21:47:15 <_av500_> DSC21, DSC25, DM270, DM320 and then davinci dm6441 Apr 22 21:47:15 neither is Apr 22 21:47:37 hwtrace is for non-intrusive debugging and performance profiling Apr 22 21:47:40 <_av500_> and some other DSC and DM inbetween Apr 22 21:49:02 ds2: https://e2e.ti.com/support/arm/sitara_arm/f/791/p/321919/1620784#1620784 Apr 22 21:49:07 ds2: https://e2e.ti.com/support/dsp/davinci_digital_media_processors/f/716/p/377700/1330420#1330420 Apr 22 21:49:39 ds2: the am335x debug subsystem is btw identical to that of the dm814x and dm816x Apr 22 21:50:08 which is useful to know since the dm816x has an unusually informative TRM chapter on the debug subsystem Apr 22 21:51:23 of course since the am335x has no DSP or IVA you can ignore bits about that (the corresponding jtag tap IDs are unused) Apr 22 21:51:43 zmatt: I saw your e2e posting Apr 22 21:53:17 it would still be interesting to see if the trace port can be made to work with a PRU as receiver.... if I ever find the time Apr 22 21:53:48 bbl...got to fix a tire :( Apr 22 21:58:55 Janos: do you require the ability to measure such high currents? since the sensor you showed me first, with max 15 mA output, is definitely safer to use Apr 22 22:00:31 (max 15 mA rms, max 21 mA peak that is) Apr 22 22:01:18 I'd still be concerned about leaving that connected to the BBB when it's powered off if there are no protection diodes in place Apr 22 22:02:28 but sourcing/sinking 50 mA would definitely be bad Apr 22 22:03:37 at least use a series resistor to the BBB if you go for that one (even though it may reduce the quality of your measurement, that's balanced against the improved quality due to larger voltage swing) Apr 22 22:17:14 zmatt, well the highest breaker i have at home is 60A so yeah Apr 22 22:19:07 besides since the 100A CT is different than the rest of them in the sense that it needs a load resistor, I was hoping to build everything around that idea Apr 22 22:20:32 and be able to monitor up to 100A, but in reality 60A would be the point where the breaker goes off, and most breakers are 30A only like 3 are 600 Apr 22 22:20:59 Janos: http://i.stack.imgur.com/OxORp.png <-- that's what I mean with protection diodes btw Apr 22 22:22:01 nothing arduino-specific about it of course :P and the upper diode always goes to the supply of whatever power domain the input belong to (in this case vdd_adc) Apr 22 22:22:49 ahh ok, will add them the to circuit Apr 22 22:23:12 the series resistor may also be a good idea, though with protection diodes in place it may not be needed Apr 22 22:24:15 will do Apr 22 22:25:31 now to the last part of the original circuit, the extra resistors and the capacitor are supposed to be for bias, no idea what purpose that serves, can you explain a bit on that or suggest some links about it ? Apr 22 22:26:36 overall you have: voltage divider and capacitor to make vdd_adc/2 (this may be shared between multiple sensors), sensor with low-value resistor across it (if not already integrated), series resistor, input with protection diodes Apr 22 22:26:56 the sensor output is AC Apr 22 22:27:26 the analog inputs do not tolerate negative voltages Apr 22 22:29:17 btw I would suggest connecting the vdd_adc/2 rail you create to one of the analog inputs if you have one spare Apr 22 22:29:19 that's the thing, i thought the voltage divider was no longer needed since the CT + the 10Ω load resistor will give me about 1.4V which is less than the 1.8V max of the ADC, or am I just lost here ? Apr 22 22:29:44 CT + load resistor will give an AC voltage output Apr 22 22:29:54 across the two terminals Apr 22 22:30:18 by connecting one terminal to vdd/2 you nicely center it within the range of the ADC Apr 22 22:31:02 the image you linked to also shows this quite clearly Apr 22 22:32:12 60 A RMS means a sine from -85 A to +85 A Apr 22 22:34:53 so with a 2000:1 ratio transformer and 10 ohm load that gives 300 mV RMS, i.e. a sine from -424 mV to +424 mV Apr 22 22:36:40 by connecting one terminal to a vdd_adc/2 = 0.9 V supply, the resulting range is 0.476 - 1.324 V Apr 22 22:40:32 the supply is quite feeble since it's made using a voltage divider, but that's okay since we're not really drawing any power from it... Apr 22 22:42:24 is this pinout correct for the black http://dev.iachieved.it/iachievedit/wp-content/uploads/2015/11/cape-headers-spi.png? Apr 22 22:42:36 btw, I would also suggest a low-value capacitor on the analog input to filter out high-frequency noise... mains supply line may have quite a bit of that from all the equipment that's on it Apr 22 22:42:55 (it'll form an RC filter together with the resistor between the sensor and the analog input) Apr 22 22:43:49 I guess that would C1 in this pic https://openenergymonitor.org/emon/sites/default/files/Arduino%20AC%20current%20input%20A.png ? Apr 22 22:43:55 correct ? Apr 22 22:43:56 it is correct but incomplete of course since most pins have multiple functions Apr 22 22:44:13 see the P9 and P8 tabs of my spreadsheet for a complete overview -> https://goo.gl/Jkcg0w Apr 22 22:44:27 no Apr 22 22:45:10 that's stabilizing the midpoint-voltage-supply you created with the voltage divider Apr 22 22:47:24 what I'm suggesting is taking the line labeled "arduino input" in that schematic, through a series resistor, to the adc input with a small cap to ground and the two protection diodes Apr 22 22:47:42 ahh ok Apr 22 22:48:42 if you have a spare adc input, I suggest connecting it to your midpoint-supply in a similar way Apr 22 22:48:57 the ADC on the BBB can do differential measurements, and that may be more accurate Apr 22 22:49:17 that may be total overkill though Apr 22 22:50:40 the adc has 7 inputs right ? Apr 22 22:50:43 as I said earlier, I don't have much intuition on what suffices for an adequate result... you'd need some engineer with a big grey beard for that :) Apr 22 22:51:13 8 technically, but only 7 available on the expansion header Apr 22 22:52:33 no worries, this is just me with some free time wanting to play around, the main purpose is to just monitor current on those cables and see some graphs about it, nothing serious Apr 22 22:53:37 you could probably get to the last one if you have a fine-tipped soldering iron and aren't afraid to wield it ;) for some reason they hooked it up to the 3.3V supply via a voltage divider, so you could remove those resistors and solder a wire onto one of the pads Apr 22 22:54:02 (I see very little use in the ability to monitor the 3.3V supply via the ADC) Apr 22 22:55:49 yeah you'll need to play with it a bit Apr 22 22:56:16 the adc also has parameters you can play with that influence your measurement Apr 22 22:58:20 e.g. the sample-time should be enough to allow the ADC's internal caps to achieve equilibrium with the signal you're measuring Apr 22 23:01:53 about the extra input, 6 should be enough and I could use the 7th for this differential measurement you mentioned which I have no idea what it does, sounds awesome :) Apr 22 23:02:36 literally what it says: it measures the (signed) difference Apr 22 23:02:56 which is the voltage across the sensor Apr 22 23:03:56 so then you're not very much dependent on the stability of your midpoint voltage, other than to keep the voltages within acceptable limits Apr 22 23:06:34 every time the adc performs a measurement a (small) amount of current briefly flows... this charge comes primarily from the Vmidpoint capacitor Apr 22 23:06:50 which therefore may fluctuate slightly Apr 22 23:08:06 I don't know how much effect this is and whether it's relevant... you could dig up numbers from the am335x datasheet I guess, or just see whether it makes a difference in practice Apr 22 23:08:35 the faster you sample, the stronger this effect will be obviously Apr 22 23:10:49 but a differential measurement across the sensor terminals would not be affected by this Apr 22 23:12:09 ahh ok got it, and yeah, it does sounds useful depending on the sampling rate then Apr 22 23:14:27 yeah I wouldn't be able to say for sure... I'd need to dig for numbers and I'm too lazy for that right now :) Apr 22 23:16:25 same for the right values for the series resistor and noise-filtering cap Apr 22 23:21:09 though it can't be hard to find a site which can calculate the response of an RC filter Apr 22 23:21:42 or try first without, maybe I'm just overthinking Apr 22 23:28:34 keep the resistor though, it's also for extra protection of the adc Apr 22 23:30:31 especially when the BBB is turned off, and to cope with the adc startup bug (see am335x errata advisory 1.0.32) Apr 22 23:37:39 lol, the advisory also has a formula to calculate the max series resistance permitted for a given sample time... but even for a sample time as short as 1 microsecond it's still okay to have 8420 ohm series resistance apparently Apr 22 23:42:04 Janos: oh, a final complication to add for you: if you want to know your *power* consumption you actually need to simultaneously measure the voltage on the line, since reactive loads can cause a phase shift between current and voltage Apr 22 23:42:18 https://openenergymonitor.org/emon/node/58 actually includes a voltage measurement circuit for that reason Apr 23 00:24:06 zmatt, ouch that does complicate things, even at the theoretical level, had no idea that current and voltage could be out of proportion Apr 23 00:27:09 i think i'm going to have fun with current measure for now, and improve from there Apr 23 00:27:32 sorry :) you can of course hope that your "power factor" is close to 1 Apr 23 00:27:57 it should be, power companies don't appreciate heavily reactive loads Apr 23 00:28:41 zmatt, Janos that was an interesting point discussed in the arduino article that zmatt linked :) Apr 23 00:29:48 power factor will quite likely be different from 1 - especially from switching power supplies and such loads. This won't really affec the balance of power at your house (hence the power companies don't care) unless you're running high power heavy machinery which feeds BACK power into the grid Apr 23 00:30:08 they don't lkike that kinda thing .. makes metering difficult, and surges all-too likely :P Apr 23 00:30:42 actually, at least electromechanical meters will automatically do the right thing and run backwards Apr 23 00:30:58 also, if you're running from an inverter.. they don't like that either .. we actually blew a couple until we figured the compact fluorescents were drawing power completely out of phase from the voltage waveform .. a few suppression caps soon fixed that :) Apr 23 00:31:09 zmatt: yes, we need more of that ;) Apr 23 00:31:35 the reason power companies care about power factor is because those meters measure energy transferred Apr 23 00:31:48 but current, in either direction, induces losses in the power cabling Apr 23 00:32:14 that cabling also has max current ratings Apr 23 00:32:41 so if you're flossing current through them without consuming energy that you're billed for, they're not happy Apr 23 00:33:42 but yes, loads that are at least somewhat reactive are all over the place Apr 23 00:34:44 veremit: cheap CFs Apr 23 00:35:14 ds2: hey, some of us have to work to a budget man. Apr 23 00:36:59 cost difference between a cheap CF and a PF corrected one isn't that big Apr 23 00:37:22 correction is either an extra inductor or cap Apr 23 00:37:32 well .. we didn;'t spot it as an issue at the outset, anyway. Long time ago. But yeah... lesson learnt. Apr 23 00:39:45 veremit: do you follow eevblog? Apr 23 00:39:55 ds2: not actively Apr 23 00:40:09 barely passively lol Apr 23 00:40:11 veremit: he has a related video on this very subject Apr 23 00:40:23 which one? pf correction or ... Apr 23 00:40:25 you do have very fancy CF fixtures though... dimmable in 5%-100% range, "Lamps turn on to any dimmed level without flashing to full brightness", etc :) Apr 23 00:40:39 veremit: he was "evaluating" one of those energy saver thingies Apr 23 00:40:50 sorry, F, (well I guess they might exist for CF too) Apr 23 00:40:59 we had some funkty ordinary fluorescent control systems at the uni lecture theatre .. would fire up then dim to 50% Apr 23 00:41:08 but you can't really 'dim' fluorescents lol Apr 23 00:41:12 now its all LED Apr 23 00:41:15 you can Apr 23 00:41:30 what happens when the gas deionises? Apr 23 00:41:33 erm ..yanno Apr 23 00:41:41 'goes out' Apr 23 00:41:42 lol Apr 23 00:41:46 I can't dscribe it Apr 23 00:41:52 I don't know all the fine details, but http://www.lutron.com/en-US/Products/Pages/FluorescentBallasts/EcoSystemBallasts/Overview.aspx Apr 23 00:42:01 EEVblog #848 - Home Energy Savers BUSTED! Apr 23 00:42:01 I think HV pulses may be involved Apr 23 00:42:19 yeah I could believe you had some HV to 'hold' the state Apr 23 00:42:40 ds2: will take a look later .. gotta hit the .. erm .. you know I was going to the store about 5hrs ago zmatt?! ;) Apr 23 00:42:53 it is a 24hour store, right? :D Apr 23 00:42:58 it is now :D Apr 23 00:43:09 so, plenty of time Apr 23 00:43:15 started a whole bunch of tidying shit up under the desk .. lol Apr 23 00:43:45 the dimming to 5% requires fixtures with separate power and control signals though Apr 23 00:44:23 so initial diagram, surely wrong is here if you want to give it a look, and correct me :P Apr 23 00:44:24 https://drive.google.com/open?id=0Bxje4d_cB_sTVExhSmxYSEtiUTA Apr 23 00:46:14 it's missing the filter capacitor and no idea if I should connect the mid-point to AIN6 Apr 23 00:46:51 and just about everything is wrong with the diodes :P Apr 23 00:47:10 LOL no wonder there, sorry Apr 23 00:47:36 including the symbol, which is that of a zener rather than a schottky :P Apr 23 00:47:55 ahh ok i can fix that really quick Apr 23 00:49:07 the diodes are from gnd to input, and from input to vdd_adc Apr 23 00:49:15 ok here we go Apr 23 00:49:21 R1 and R2 should have the same value Apr 23 00:49:53 you're missing the series resistor between the sensor and the input Apr 23 00:53:11 zmatt: go easy bro :p Apr 23 00:53:17 ok .. going, going .. gone .. Apr 23 00:53:40 if you're connecting Vmidpoint to ain6, do so in the same way as the other inputs Apr 23 00:54:19 I'll check back in once you've fixed it up .. oh, and do a write-up somewhere .. we have a cronic lack of bbb application docs on the web :) Apr 23 00:54:20 (series resistor, protection diodes, cap to ground if you decide to use them) Apr 23 00:54:24 * veremit scoots. Apr 23 00:56:05 sounds like basic EE stuff Apr 23 00:57:00 it may seem odd to protect that input, but if the BBB gets powered off then C1 would initially still hold quite some charge that needs to go somewhere Apr 23 00:59:02 couldn't you get away with using a resistor to limit current and rely on the ESD diodes to drain it? Apr 23 01:00:42 possibly, but from what I've seen of the IBIS data the esd diodes kick in rather late... which makes perfect sense if they're meant to block esd rather than sustained mild overvoltage Apr 23 01:01:05 also they fry easily into shorts (= dreaded power led blip of doom) Apr 23 01:01:57 a large enough series resistor would fix that, but it may also cause measurement error Apr 23 01:02:39 so there's kind of an uncomfortable trade-off with some unknown variables there, one that's easily avoided by using external diodes Apr 23 01:06:48 hi Apr 23 01:08:44 zmatt: is there any docs on the esd diodes (like the Vf for them) Apr 23 01:09:20 the ibis model Apr 23 01:09:42 ahh Apr 23 01:09:45 (if you find a good free visualization tool for ibis data, let me know) Apr 23 01:10:06 haven't looked at ibis stuff Apr 23 01:10:30 do I need a couple of extra diodes for AIN6 too ? Apr 23 01:10:46 ds2: some of the ibis models seem to contain nonsense though, unless I don't know how to correctly interpret it Apr 23 01:10:52 Janos: I already answered that Apr 23 01:11:33 * Janos reads back Apr 23 01:11:45 ds2: here's my attempt at visualizing some models... https://docs.google.com/spreadsheets/d/1L2Meyzb9H9B8_OMjMM_TemsEmZY2sWzeCeNchJRLKQ0/pubhtml Apr 23 01:12:37 I have no idea why the I/V curve for driving high intersects the 0 mA at a lower voltage than 3.3V Apr 23 01:13:41 model 107 seems to have its data offset entirely there... the weird drop is due to missing datapoints Apr 23 01:17:07 what's the colors? Apr 23 01:17:46 min/typ/max Apr 23 01:22:23 if I understand correctly, there's usually not much process variation within a single die, so all IO may be on the blue line ("weak silicon") or all on the orange line ("strong silicon"), but not a mix thereof Apr 23 01:22:36 that's just something I picked up by osmosis though, I'm no chip designer :P Apr 23 01:22:39 ok guys, thanks a lot for everything, got a run, I'll check up tomorrow if you still have the patience for to tutor a noob like me :P, is the the documentation I could definitively come with a blog about the process when it works Apr 23 01:22:54 wonder how much variation is there with stuff in 1.8V vs 3.3V Apr 23 01:23:37 yeah I don't have any 1.8V curve yet... Apr 23 01:24:37 nor any curve with pull-up or pull-down enabled :P Apr 23 01:24:49 as the numbers show, there are quite a few models in the IBIS file Apr 23 01:26:26 I can see if I still know how to do this and fish some particular curve from the file Apr 23 01:27:18 I/O model names can be found in the "Model" column in my pins spreadsheet Apr 23 01:29:14 { model, supply voltage, supply tolerance (5%/10%), slew (fast/slow), pull (up/down/disabled) } yields a set of IV curves Apr 23 01:29:50 (and also step response curves, but I don't have anything to extract and visualize those yet) Apr 23 01:31:28 but I got a bit put off by the initial results since either the dataset or my visualization thereof is evidently of dubious quality :P Apr 23 01:38:58 i wanna design a breakout pin size (like 5 mm or w/e the dimensions of the headers are) 5-3.3v bidirectional converter who's gonna fund me Apr 23 01:40:39 we're solvin the bone's biggest problem here lets go Apr 23 01:44:52 I've seen what bidirectional converters can do to signal integrity though :/ Apr 23 01:45:42 yeah it's actually a pretty tough problem now that i'm trying to draw it out Apr 23 01:45:59 and your probably looking at high speed stuff right? Apr 23 01:46:10 if you call i2c "high speed" Apr 23 01:46:31 well yeah i wasn't gonna say RF Apr 23 01:46:53 cause i literally ment like >10khz Apr 23 01:47:04 (my math is better than my spelling, i promise) Apr 23 01:47:11 well iirc they were running i2c at a pretty crappy speed Apr 23 01:47:27 i think 100k is nom Apr 23 01:47:33 bit-banged from some 8-bit uC iirc Apr 23 01:48:03 it is, but this was way slower Apr 23 01:48:37 i think the hardest (impossible) would be preserving line impedances Apr 23 01:48:47 yep, the problem is that a bidir level-shifter becomes a high resistance as the voltage rises Apr 23 01:50:11 so 1. on a rising transition you inject current into a line and then raise a shield for any incoming reflections, allowing copious ringing if it's not a terminated line (say, for example, i2c) Apr 23 01:50:34 2. the pretty much undriven line easily picks up cross-talk and other noise Apr 23 01:50:36 i dont really understand the architecture for the bidir shifter Apr 23 01:51:48 afaik it works like a bus switch, with integrated pull-ups that automatically activate Apr 23 01:52:46 i figured the first part but the second part i'm curious Apr 23 01:53:05 like either way your adding some logic to the line and slowing things down Apr 23 01:53:19 that's the least of the problems though Apr 23 01:53:56 but it's relatively straightforward... it just detects a high level and on both sides activates pull-ups to the respective supply Apr 23 01:54:31 there's inherent feedback in this, but since they're merely pull-ups they can be overriden by a low-driven signal Apr 23 01:54:48 and then u can detect the low-driven signal and take off the pullup Apr 23 01:55:51 yeah i'm not seeing any situations where that wouldn't work all though i thought it would be more complex when i first thought of it Apr 23 01:56:41 i was thinking about impedance matching as far as the pullup impedance Apr 23 01:57:28 in theory you can do perfectly matched bidirectional level shifting Apr 23 01:57:32 and then what happens if a pin becomes high impedance w/ a pull down for a read when it was previous high Apr 23 01:57:42 matched to the line at least Apr 23 01:58:34 I'm thinking here of gigabit ethernet transceivers which simultaneously transmit and receive on all four pairs Apr 23 01:59:09 never studied ethernet Apr 23 02:00:00 the ethernet part isn't really important anyhow, it just shows you can receive on a transmission line on which you're simultaneously transmitting, not just in theory but also in practice Apr 23 02:00:21 but that involves pretty amazing analog hardware Apr 23 02:01:50 it sounds frequency modulated Apr 23 02:02:38 they're not separated in frequency domain, time domain, or by code modulation or anything Apr 23 02:04:05 it really receives and transmits at the same time, on the same chunk of spectrum Apr 23 02:04:11 that sounds really interesting aka impossible Apr 23 02:04:30 it isn't Apr 23 02:05:29 no i believe you Apr 23 02:05:34 it just *sounds* impossible Apr 23 02:06:44 it's easy to model theoretically: a driver with output impedance Z can be viewed as a 0ohm driver with a series resistor of value Z Apr 23 02:06:56 now measure the voltage across Z :) Apr 23 02:07:08 across the resistor I mean Apr 23 02:07:28 oh yeah thats obvious now Apr 23 02:08:02 so basically u know the impedance on both ends Apr 23 02:08:06 so its just a voltage divider Apr 23 02:08:10 and u know the voltage ur driving Apr 23 02:08:27 and the voltage on the middle node Apr 23 02:09:14 if it's of any help, I scribbled this a while ago... http://www.texpaste.com/n/o0vss5bn Apr 23 02:13:21 haha yeah Apr 23 02:13:22 thanks Apr 23 02:13:42 a transformer also performs passive level-shifting, albeit not down to DC. gigE transceivers show active circuits can also transmit and receive simultaneously. these things considered I'm inclined to conjecture that true bidirectional level shifting is in fact possible Apr 23 02:14:49 (of course a transformer also scales impedance, but that's because it's a passive element and can't make energy appear out of nowhere... an active component wouldn't have that constraint) Apr 23 02:16:06 given that I don't see any application of such a component where the complexity would be justified however, I doubt it exists as a product Apr 23 02:16:45 but who knows :) Apr 23 02:17:05 oh man yeah you could really use that transformer to your advantage Apr 23 02:17:27 you could basically filter the dc part and shit that the less-nice way and then let the transformer take care of the ac components Apr 23 02:17:28 maybe Apr 23 02:17:40 you've got tons of ideas tho i like this Apr 23 02:17:52 i was trying to build an avr programmer with my beaglebone Apr 23 02:18:39 there was a time when i thought atmega needed to be 5v Apr 23 02:19:13 so that inspired the bidirectional level shifter idea Apr 23 02:19:38 well, I can report they do exist, but yuck :P Apr 23 02:20:40 (and apparently even work most of the time) Apr 23 02:21:40 yeah i found them on adafruit Apr 23 02:21:53 i was thinking you could build them so that they just looked like header pin extensions Apr 23 02:21:59 thats for the sexy factor Apr 23 02:24:11 i feel like thats a big obstacle for the beaglebone to break into the arduino market Apr 23 02:24:54 since so many of the existing designs and peripherals are designed for the 5v logic ( i guess this reflects the same industry design issues) Apr 23 02:25:58 funny thing is that the am335x actually has to jump through complicated hoops to make 3.3V... it isn't really comfortable with anything above 1.8V Apr 23 02:26:53 if you check the datasheet you find a 2V abs max limit appearing in many context (including max difference between 3.3V and 1.8V supplies during power-up/down) Apr 23 02:27:13 the I/O supplies are called VDDHV .. High Voltage :) Apr 23 02:27:51 and apparently use tricks with cascaded transistors to avoid more than 2V across any individual transistor Apr 23 02:28:59 cool, it works! accessing peripheral registers directly from python Apr 23 02:30:36 this opens up possibilities Apr 23 02:40:03 did u write a lib? Apr 23 02:44:18 Hi.. Is someone here? Apr 23 02:46:59 =( Apr 23 02:48:21 theres usually people here bud Apr 23 02:48:31 most channels just want you to ask Apr 23 02:49:23 Ok. Sorry.. I'm new here Apr 23 02:50:25 yeah it seems like it would be rude to just come in and start asking questions i guess Apr 23 02:50:31 if uve never done it before Apr 23 02:51:42 I think so Apr 23 02:51:44 :) Apr 23 02:53:10 then sorry and sorry if my english is not perfect.. I'm learning Apr 23 02:55:15 Well I'm Camilo and I'm learning how to use Beaglebone Black because I need to build a recognition robot Apr 23 02:55:44 hi camilo Apr 23 02:55:47 my robot grows plants Apr 23 02:57:22 but I have a problem.. My Beaglebone Black can't connect to 192.168.7.2.. When I connect this by USB, just only light on D1 and D2, unblinking Apr 23 02:58:13 I'm sure all of you may have impressive projects Apr 23 02:58:14 what OS are you connecting from? Apr 23 02:58:17 :) Apr 23 02:58:27 hi dsockwell Apr 23 02:59:18 on IRC it's OK to skip all those polite things, I want to hear about your OS Apr 23 02:59:40 i'm connecting by windows 10, but this is so strange because I've been conected well before **** ENDING LOGGING AT Sat Apr 23 02:59:58 2016