**** BEGIN LOGGING AT Fri Jul 26 02:59:59 2019 Jul 26 04:44:08 nowadays it appears as two network interfaces actually one CDC-ECM and one RNDIS Jul 26 04:44:48 snice Mac only supports the usb standard (CDC-ECM), and Windows doesn't give a fuck about standards and only support's microsoft's proprietary alternative (RNDIS) Jul 26 04:44:54 Linux supports both Jul 26 04:46:28 ksft: personally I prefer connecting to beaglebones via ethernet... I use usb networking pretty much never, and on most of our beaglebone's it's not even enabled Jul 26 04:47:02 the script which sets them up is a significant contributor to startup time, and I prefer beaglebones that boot quickly Jul 26 04:47:21 the script which sets the USB connection up? Jul 26 04:47:35 yeah, among other stuff it does (none of which is important) Jul 26 04:48:00 does that mean it'll boot faster if I connect the miniusb port only to power? Jul 26 04:48:06 nope Jul 26 04:48:59 the script doesn't care whether usb is connected to a pc or not, it just does setup stuff (and it's the script itself that makes things slow, not the usb networking per se) Jul 26 04:49:44 how do you get it to boot quickly, then? Jul 26 04:49:46 this is how fast a beaglebone *can* boot with a bit of effort (but not an enormous amount of effort): https://liktaanjeneus.nl/boot.svg Jul 26 04:49:51 (I have noticed that it takes a while) Jul 26 04:51:23 systemd sure does take a while Jul 26 04:51:44 on a beaglebone with a cleanly installed console or iot image, I can achieve <10s boot with very little effort, just by replacing the network manager by systemd-networkd and disabling unnecessary (imho) services Jul 26 04:52:12 less than a second here, mostly on loading config files from eMMC (i.e. it's mostly waiting for eMMC probably) Jul 26 04:54:41 probably the biggest impact however will come from updating the bb-customizations package to its latest version, and then removing initramfs (I install a dummy initramfs package to work around the annoying unnecessary dependency of the kernel package on the initramfs package) Jul 26 04:55:41 or, if you really want to use initramfs for whatever reason, at least update it after updating the bb-customizations package Jul 26 04:57:49 how can I ssh to it over an ethernet cable, then? Jul 26 04:58:05 I thought that's what you're doing right now? Jul 26 04:58:22 I was using a router in the middle Jul 26 04:58:42 I mean, yeah I plug it into the existing network, which has a router Jul 26 04:58:46 I assume it would be faster (and more convenient) to connect it directly to my computer? Jul 26 04:58:58 for me it would be neither Jul 26 04:59:10 the connection would be faster, I mean Jul 26 04:59:15 I mean, it would be neither faster nor, for me, more convenient Jul 26 04:59:20 oh Jul 26 05:00:00 I might not always be connected to a network I control Jul 26 05:01:15 connecting directly to the bbb via ethernet should work too. if you have avahi-daemon installed or systemd-resoolved enabled on the bbb then it can be located by hostname via mDNS. if you have systemd-resolved enabled it can also be located using LLMNR (used by Windows). Jul 26 05:01:33 so then ssh'ing to it by hostname should still work, even if it only have a link-local ipv6 address or whatever Jul 26 05:02:00 maybe it would be simpler to not care about the boot time and use the USB connection I'll have anyway for power? Jul 26 05:02:15 alternatively, I'd just add g_ether to /etc/modules and add a config file for systemd-networkd to setup the usb networking Jul 26 05:03:39 I'm not actually sure which protocol is setup by that (RNDIS or CDC-ECM), but since I use Linux on my laptop I don't care, it supports all options Jul 26 05:10:22 how do I configure systemd-networkd? Jul 26 05:11:20 and if I do it right, will it just cause the bbb to appear as a network adapter to my laptop? Jul 26 05:14:21 basically it's just a matter of adding some appropriate config files, enabling (using systemctl) systemctl-networkd and systemd-resolved, disabling every other service that messes with networking, rebooting, and hoping you can still connect to it after rebooting Jul 26 05:15:12 if you add g_ether to /etc/modules and a suitable config file to /etc/systemd/network/ (something like https://pastebin.com/3EQUxN5y -- untested) then it'll appear like a usb network interface yes Jul 26 05:15:39 so I don't even need g_ether on my laptop? Jul 26 05:16:00 g_ is on the device side (the beaglebone), not on the host side (your laptop) Jul 26 05:16:10 also, your laptop will automatically load appropriate drivers for usb devices Jul 26 05:16:19 I take it you're using linux? Jul 26 05:18:51 yeah Jul 26 05:19:48 yeah so then you don't need to worry about whether g_ether will work with your laptop... I *think* the beaglebone kernels are configured to make g_ether use RNDIS, which means it'll work with Linux and Windows hosts Jul 26 05:20:52 I think that requires a kernel option I'd have to enable Jul 26 05:27:11 where? Jul 26 05:27:33 g_ether is compiled (as module) in every standard beaglebone kernel Jul 26 05:28:22 the driver on the host side (dunno its name) is probably compiled as module in every remotely standard linux kernel for every major linux distribution Jul 26 05:29:16 no, I mean on my laptop Jul 26 05:29:26 I use Gentoo Jul 26 05:29:28 why do you think that? you already have it working *right now* Jul 26 05:29:34 right? Jul 26 05:29:55 no, I've been sshing to the bbb through the router, not over USB Jul 26 05:30:01 ohhh right Jul 26 05:31:06 well dunno, even for gentoo it would seem silly to me if it's not in your kernel config... there's no real downside to compiling drivers for usb devices as kernel modules, it just adds a little bit of compile time and a tiny bit of disk space Jul 26 05:31:33 I didn't even have support for ethernet at all until I got a USB ethernet adapter and noticed it didn't work Jul 26 05:31:54 it only takes a minute or two to find the right option and recompile the kernel Jul 26 05:31:57 CONFIG_USB_NET_RNDIS_HOST Jul 26 05:31:57 usually Jul 26 05:31:59 I think Jul 26 05:32:33 which should be =m Jul 26 05:52:01 ah, wouldn't have guessed that one Jul 26 05:53:07 I'd also recommend CONFIG_USB_NET_CDCETHER=m just ni case Jul 26 05:53:28 I think RNDIS_HOST is the one you want in this case, but I'm not 100% sure Jul 26 05:54:08 besides, from a moral point of view, if you're enabling a driver for microsoft's proprietary alternative to an usb-standard protocol, you should also enable the usb-standard one :P Jul 26 05:54:39 seems to have automatically been done by USB_USBNET Jul 26 05:54:44 and I definitely agree Jul 26 05:54:44 ah ok Jul 26 05:55:01 but in what sense is the protocol proprietary? Jul 26 05:55:16 it must be publicly documented sufficiently for Linux to have it Jul 26 05:55:18 is it patented? Jul 26 05:55:38 it was probably implemented based on usb sniffing / reverse engineering Jul 26 05:55:41 I mean, it being developed by Microsoft is enough reason for me not to like it Jul 26 05:55:48 would be my guess Jul 26 05:56:00 that's impressive Jul 26 05:56:11 I mean, I don't think it's a complicated protocol Jul 26 05:56:51 I didn't know Microsoft ever called things standards unless they were sufficiently overcomplicated Jul 26 05:57:10 they don't call it standard Jul 26 05:57:17 it's just RNDIS :P Jul 26 05:57:28 ah Jul 26 05:57:46 "A partial RNDIS specification is available from Microsoft, but Windows implementations have been observed to issue requests not included in that specification, and to have undocumented constraints." -- wikipedia Jul 26 05:58:15 of course Jul 26 05:59:07 note that CDC-ECM provides equivalent functionality (afaik), is part of the usb standard, is older than RNDIS, and is supported by every OS *except windows*. Fuck microsoft. Jul 26 05:59:07 rebooting with new kernel... Jul 26 05:59:17 of course it is Jul 26 06:00:57 hey, that's cool, the bbb didn't lose power when I rebooted Jul 26 06:02:01 what's the RNDIS module called? Jul 26 06:02:52 *shrug* Jul 26 06:03:19 since it gets loaded automatically I've never needed to care, so dunno Jul 26 06:03:38 does it automatically get loaded automatically? Jul 26 06:03:59 yeah, unless your system is thoroughly broken Jul 26 06:04:07 there's a decent chance of that Jul 26 06:04:11 but I'll assume it's working Jul 26 06:04:17 lol Jul 26 06:04:39 I don't see a new network interface Jul 26 06:04:48 basically if it doesn't auto-load, I'd just suggest a quick "depmod -a" and otherwise wouldn't have any idea how to fix your broken system Jul 26 06:05:40 running depmod should normally already be part of installing a kernel though Jul 26 06:05:56 (otherwise not even modprobe would work) Jul 26 06:08:23 hmm, dmesg says "Product: RNDIS/Ethernet Gadget" Jul 26 06:21:34 hmm, I should do some shopping now, before it gets (even more) disgustingly warm outside Jul 26 06:22:12 bye Jul 26 06:22:50 just because I've come to the conclusion I should doesn't mean I've already succeeded in getting up and actually going :P Jul 26 06:23:15 oh Jul 26 06:23:48 yeah, I guess I've come to a lot of conclusions that I should do things Jul 26 06:24:23 like learn how to solder or at least find a glue gun so I'm not holding the bare wires up to the back of the XLR connector with my hands Jul 26 06:24:42 or learn how to use my operating system properly Jul 26 06:31:26 soldering isn't hard Jul 26 06:31:56 I'm really not a hardware guy myself, but soldering wires to the back of an XLR connector shouldn't be too hard Jul 26 06:32:44 everyone tells me that Jul 26 06:32:59 I'll get around to it Jul 26 06:33:05 :) Jul 26 07:00:36 the trick is to heat the terminals and wire, thne let the solder melt onto that. don't melt the solder and let it dribble like glue onto cold terminals. it won't stick. Jul 26 07:05:53 huh, I thought you were supposed to melt the solder Jul 26 07:07:42 yes, but you melt it onto the hot wire Jul 26 07:07:58 I did not know that Jul 26 07:08:08 otherwise it 'freezes' before it can stick Jul 26 07:08:40 having a bit of molten solder on the tip can be helpful, but it will still not stick on the terminals or wire as long as these are cold Jul 26 07:09:32 also no "carrying around" solder on the tip, you touch the solder to the object directly so that the flux inside the hollow core can work its maaaagic Jul 26 07:09:45 yeah that too Jul 26 07:09:51 I think I have a soldering iron somewhere Jul 26 07:10:36 basically you just need to get all three things (soldering iron tip, the wire/terminal, your solder) in the same place at the same time :) Jul 26 07:11:04 the iron heats the target, the solder melts against these and introduces flux Jul 26 07:12:12 when soldering X against Y you may first do this on X and Y individually before trying to do it with X and Y together, it can make things easier Jul 26 07:13:15 yeah, with solder pre-applied to both surfaces you only need to reflow (melt it on both sides, hold still, let solidify) Jul 26 07:13:16 also, probably just disregard anything I say about soldering since I've done it only very rarely :D Jul 26 07:39:09 no, the presoldering (it;s called tinning) is correct and very helpful Jul 26 07:40:08 especially to avoid melting the wire insulation, since you need to apply heat to the big XLR terminal when tinning that for much longer than to the wire Jul 26 07:40:53 why? Jul 26 07:41:27 a relatively beefy piece of metal conducts heat away more quickly than a thin wire Jul 26 07:41:31 it's a big bit of copper. takes a while to get it hot enough to melt the solder Jul 26 07:42:13 I don't think it's copper... Jul 26 07:42:23 but that makes sense Jul 26 07:42:39 it's electrically conductive, therefore (in general) thermally conductive Jul 26 07:43:53 the little bit of solder on the tip is also useful. also called tinning. but it's not carying solder around as tbr rightly forbids, it's so you've got something liquid in place to fill the gap and improve thermal contact between iron and terminal Jul 26 07:45:02 a lot of XLR terminals are plated copper. but some are pressed or even turned steel. If it's thick, there's a good chance it's copper underneath the coating Jul 26 07:46:01 the pins themselves are often stainless steel though, or something nickel-plated Jul 26 07:46:27 I like XLR. nice solid parts :) Jul 26 08:05:25 how does everyone know so much about exactly what I'm doing? Jul 26 08:05:57 and if everyone knows this much about how to build DMX controllers, why do they cost like $10000? Jul 26 08:06:20 ksft: because when you sell them, you need someone to pay your time Jul 26 08:06:37 (also, ours didn't) Jul 26 08:06:44 (*doesn't) Jul 26 08:07:20 what company was it, if you don't mind sharing? Jul 26 08:07:39 ksft: so if you spend a year developing, but only have market perspective of 50 units a year or so, each one will cost 10k€. roughly. same applies to any kind of specialized gear. Jul 26 08:08:05 lanbox.com. I'm amazed they're actually still being sold... I've heard that they're now running into the issue that the ancient microprocessor in them just isn't being manufactured anymore Jul 26 08:08:17 nor anything compatible Jul 26 08:08:28 I guess I'm just disappointed that the free software community doesn't have much overlap with lighting designers Jul 26 08:08:33 or many other communities Jul 26 08:08:46 DMX is very widely used. lots of 'embedded' people have come across it in some way Jul 26 08:08:49 I mean, there's low-cost stuff for dmx I think Jul 26 08:08:54 most of it crap probably Jul 26 08:09:15 but the hardware shouldn't be expensive Jul 26 08:09:24 there's overlap aropund OLA and openpixel and artnet i think Jul 26 08:09:49 many have mused DMX for sure. but either you're a hobbyist then you have rather limited means of doing things - or you're a pro, then you'll probably not exactly share everything Jul 26 08:10:09 but also a lot of commercial stuff that's possibly based around a desire to sell big systems Jul 26 08:10:22 and so isn't very open Jul 26 08:10:27 also, if you're a pro then the controller probably isn't a particularly expensive part, especially since you probably only need one Jul 26 08:10:37 the top-of-the-line ETC consoles are a fancy custom mechanical keyboard, a low-power computer running Windows for no reason, and monitor or two Jul 26 08:10:52 lol Jul 26 08:10:53 that should be maybe a couple hundred dollars Jul 26 08:11:08 "fancy custom mechanical keyboard" that alone sounds expensive Jul 26 08:11:14 zmatt: ++ Jul 26 08:11:18 yeah. they used to be full custom but have gone pc-ised Jul 26 08:11:22 switches are, what, a dollar each? Jul 26 08:11:40 ksft: let me guess, you've never done a series product in hardware? Jul 26 08:11:44 not decent switches that will stand up to years of roadies Jul 26 08:11:44 a single large custom PCB to attach them to is a few dozen more dollars? Jul 26 08:11:51 LetoThe2nd: no, I know nothing about hardware Jul 26 08:12:19 I don't know what a "series product" is Jul 26 08:12:32 serious ? Jul 26 08:12:44 once everyone who lifted a finger in bringing a product into this world wants to get paid, by the hour, shit gets expensive Jul 26 08:13:10 the sort of switches that pros like to fondle on a big expensive console can be more like 20-30 $ Jul 26 08:13:22 ksft: then here's the rundown. rough estimate, take whatever you think the parts cost is. multiply with at least 20. depending on series size, can easily also be 50. the result is the price you have to ask from the end user in order to make the püroduct prifitable Jul 26 08:13:45 LetoThe2nd: that's still well below how much an ETC console costs Jul 26 08:13:53 so, 50 switchs @ 1$, plus 10$ pcb, plus 20$ enclosure. go figure. Jul 26 08:13:56 and let's not even start about software development cost /o\ Jul 26 08:14:04 free software exists Jul 26 08:14:08 even as in freedom Jul 26 08:14:14 whenever someone happens to care enough Jul 26 08:14:20 maybe I care enough Jul 26 08:14:21 ksft: well, welcome to the real world. i for one want to be paid. Jul 26 08:14:22 someone who also has the skill and time Jul 26 08:14:26 or maybe I just have nothing else to do Jul 26 08:14:29 then maybe you're going to be the one to bring change Jul 26 08:14:39 to this particular field/product Jul 26 08:14:45 ksft: do it. :) Jul 26 08:14:49 ksft: go go go! Jul 26 08:14:51 and the ability to learn about fascinating niche concepts like "datasheets" Jul 26 08:15:06 and "voltage" Jul 26 08:15:15 calling datasheet a "niche concept" is funny and highly arrogant at the same time. Jul 26 08:15:26 (I was joking, in case that wasn't clear) Jul 26 08:15:27 hmm. Maybe learn more and then do it ? :) Jul 26 08:15:53 but no better way to learn than on the job ! Jul 26 08:15:56 learn by doing Jul 26 08:16:08 ksft: to be honest. most of what you type is both funny and awkward to somebody who has spent several years in professional embedded. Jul 26 08:16:09 nobody knows how to read datasheets... until you're forced to read datasheets Jul 26 08:16:12 thanks to this incredibly helpful channel (zmatt), I've got most of the hardware stuff working enough to start more serious software stuff Jul 26 08:16:33 the software is the fun part anyway Jul 26 08:16:53 needing to understand something so you can solve a problem is a much better motivator for me than just learning to get through a course Jul 26 08:16:53 I don't think I seriously plan to sell what I do, let alone try to make moneye off it Jul 26 08:17:04 money* Jul 26 08:17:29 even better, yay for open source hardware! Jul 26 08:17:50 I have a friend who is a very interesting person in many ways, including that he says he writes large programs for fun just to delete them Jul 26 08:18:09 tbh i thought you were talking about an open source control console, with all the details handled instead of just a pile of unconnected software and random docs. That's entirely a worthwhile concept even if nobody ever builds one exactly the same Jul 26 08:18:28 yah, that's what I want to do Jul 26 08:18:30 in fact, perhaps especialy if noone ever builds it the same Jul 26 08:18:30 yeah* Jul 26 08:18:57 I barely understand the concept of open-source hardware, because I barely understand the concept of hardware Jul 26 08:19:20 the reference implementation of RDM is just such a thing, unfortunately implemented on slightly weird hardware Jul 26 08:19:31 but I've managed to connect wires to a breadboard mostly at random until I managed to turn on a fixture Jul 26 08:19:41 it's basically a pile of metal and semiconductor, and then you put in electricity and either smoke comes out, or if you're lucky it does something funny Jul 26 08:19:45 (a PIC dev kit .. would be far better on a beaglebone or a teensy) Jul 26 08:20:00 zmatt: i love funny smoky! Jul 26 08:20:10 like, what exactly do I publish to make my hardware open source? Jul 26 08:20:20 everything, obv. ! Jul 26 08:20:43 schematics and any other design files Jul 26 08:20:53 I don't have anything like that Jul 26 08:21:07 have a look at https://www.oshwa.org/definition/ Jul 26 08:21:21 you know people who understand the syntax of Python and that's it? Jul 26 08:21:38 you will, though, when you finish. Document as you go. Draw schematics in kicad Jul 26 08:21:41 like, they can write a program to add up all the even numbers from 1 to 100 Jul 26 08:21:59 but they couldn't possibly contribute to free software Jul 26 08:22:12 (nor will they *know* the syntax of python at that point) Jul 26 08:22:20 they can contribute to manual-writing, probably. very valuable. Jul 26 08:22:33 I was trying to help someone like this develop a pretty simple web app recently Jul 26 08:22:55 he made the comment that seemed to him insightful that he could use an if statement to check whether a condition is true Jul 26 08:23:09 I was like that until maybe a year or two ago Jul 26 08:23:16 I'm sure all programmers are when they're first learning Jul 26 08:23:24 anyway, I feel like that but with hardware now Jul 26 08:23:34 yeah, teaching people to program when they don't have the concepts is hard Jul 26 08:24:03 ksft: give it time, and just do stuff... the only downside of hardware is that it has this annoying tendency to break if you're not careful enough Jul 26 08:24:21 also, it costs money Jul 26 08:24:27 yep Jul 26 08:24:31 but yeah, I'm going to keep trying to do stuff Jul 26 08:24:42 so those obstacles make it easier to get into software than to get into hardware Jul 26 08:25:03 you can can get an awful long way these days juist by sticking modules together Jul 26 08:25:27 that doesn't necessarily prevent stuff from breaking though Jul 26 08:25:48 the extent of my "design" so far is that I connected an RS-485 controller to some pins on a bbb and (sorta) to an XLR connector Jul 26 08:25:59 beagebones, rs485 drivers, voltage regulators .. you just need a few concepts about power and signalling levels to avoid letting too much smoke out Jul 26 08:26:11 zmatt: true, but easily replaced Jul 26 08:26:27 and then lesson learned :) Jul 26 08:26:29 I figure if the whole thing draws its power from a USB port on my laptop, it can't do too much damage Jul 26 08:26:59 not much power is required to fry hardware Jul 26 08:28:17 like, connect a wire to the 5v supply, swipe the other end across the beaglebone's PCB... voila, hardware fried Jul 26 08:28:21 and a surprising amount is used if you have a panel with loads of leds or something Jul 26 08:28:36 ...huh Jul 26 08:28:42 well, I'll try not to do that Jul 26 08:28:52 (almost nothing on the beaglebone is 5v tolerant) Jul 26 08:29:31 https://xkcd.com/908/ Jul 26 08:29:33 I should probably be working on software on my laptop and copying it over instead of having my only copy on the thing I might accidentally blow up Jul 26 08:30:06 absolutely Jul 26 08:30:26 in fact, the tiny tiny transistors in the am335x do not tolerate more than 2V across the atoms-thin insulator of the base, so it actually uses Magic(tm) to have 3.3v I/O in the first place (but all the logic inside it is at 1.8V or lower, most of it at 1.1V actually) Jul 26 08:30:42 my friend said that what he does is nice because he controls when he loses the code instead of doing that accidentally Jul 26 08:30:52 nice thing about a beaglebone vs. something like an arduino is that you can do that copying with git, if you have the disciplne for it Jul 26 08:31:47 I should make friends with electrical engineers Jul 26 08:33:29 find a hackerspace or makerlab or something like that? Jul 26 08:33:38 good idea Jul 26 08:33:48 hackerspaces are cool Jul 26 08:33:56 as stupid as every term for that concept is, the concept is cool Jul 26 08:34:17 eh, you get used to terms Jul 26 08:35:34 calling a pcb after a dog is stupid too but it works :) Jul 26 08:37:12 w00t, pretty things in the mail :) Jul 26 08:37:33 https://www.crowdsupply.com/unexpected-maker/tinypico Jul 26 08:37:51 I'm at NURDspace every now and then.... https://www.facebook.com/NURDspace/photos/a.386682241383028/1771925519525353/ (I don't think I've ever seen it in such a clean and organized state with my own eyes) Jul 26 08:38:43 you were just making fun of Microsoft, and you use Facebook? Jul 26 08:39:17 you can view the photo in an incognito window Jul 26 08:39:18 haha Jul 26 08:39:31 it was just the first decent photo I found on the interwebs Jul 26 08:39:49 I refuse on principle to intentionally have my computer make any requests to Facebook domains Jul 26 08:40:00 that looks so like our hackspaces' first space Jul 26 08:40:22 the photos on the nurdspace wiki are all way obsolete Jul 26 08:40:25 I'm not completely sure what that principle is exactly, but I stand by it firmly Jul 26 08:40:50 it's a principled principle Jul 26 08:41:05 something righteous about whatever their most recent scandal was (they're all the same anyway) Jul 26 08:42:35 I mean, you're not wrong, but again... incognito window :P Jul 26 08:43:38 Hi all Jul 26 08:44:23 I want to install and use a RTOS and be able to use the ethernet Jul 26 08:44:56 ksft: photo of the previous nurdpsace location: https://www.flickr.com/photos/dvanzuijlekom/24531053521 that chain of lights wired in a bus on the ceiling are actually dmx-controlled RGB led moving heads Jul 26 08:45:10 they were so awful XD Jul 26 08:45:27 ROS and Bela are the RTOS environments I've particularly seen targeted at beaglebone Jul 26 08:46:03 that actually looks bigger zmatt Jul 26 08:46:06 ksft: made from the cheapest parts you can find on ali-express along with 3d-printed parts. they broke like *all the time* Jul 26 08:46:07 was it ? Jul 26 08:46:45 thanks artag Jul 26 08:47:02 artag: it probably was, at least ths main room was. they have a few number of separate rooms in the new location though. regardless, they didn't move to expand, they moved because the building they were in was going to get demolished Jul 26 08:47:20 I rather like the one-room hackspaces. I think once we separated dirty stuff from clean we started to get a morlock vs eloi culture war Jul 26 08:47:41 there are pros and cons Jul 26 08:48:01 the size where everyone gathers around a table and shares food is way more sociable Jul 26 08:48:10 the new location has one big main room, and some small ones Jul 26 08:48:17 but certainly has limitiations Jul 26 08:48:59 the main social divider is the smoking room... at the same time, I enjoy being able to breathe in the main room Jul 26 08:49:25 is that for flux fumes ? Jul 26 08:49:30 no as in, smoking Jul 26 08:49:52 tobacco and/or weed Jul 26 08:49:53 heh .. how quaint ) Jul 26 08:51:15 it has a fair number of smokers, so sometimes you might be left wondering "where is everyone?" and it turns out everyone is in the designated smoking room (which has its own ventilation I'm pretty sure) Jul 26 08:51:55 please just let me know if somebody can online can help me with ti-RTOS and ethernet comunication using CCS and BBB i will be around hitting my head with the keyboard :) Jul 26 08:52:25 we currently have two floors here https://wiki.london.hackspace.org.uk/view/Ujima_House Jul 26 08:52:36 MarineroBBB: TI-RTOS is supported on TI's own support forum, not here Jul 26 08:52:44 thought there were more photos but most seem to be of the previous space Jul 26 08:52:47 I don't think I've ever seen or heard of anyone here using TI-RTOS Jul 26 08:53:23 (using CCS is also very rare, it's been a looong time since I've touched that IDE myself) Jul 26 08:53:56 what's the application, MarineroBBB ? Jul 26 08:54:05 ROS and Bela and hopefully somebody else :) Jul 26 08:54:34 ROS and Bela are linux-based, not TI-RTOS Jul 26 08:55:23 i mean, what do you want to use the thing for ? ROS is probably the way to go for robots, Bela form analog i/o, I think both are based on rtlinux Jul 26 08:57:43 artag I´m just starting, the idea is to run the RTOS with network. i just want to be able to read some values digital input also analog input and save it in memory and later to be able to transfer it by some kind of download utility in a browser Jul 26 08:58:22 MarineroBBB: easiest would be to just use linux Jul 26 08:58:49 yeah, don't bother with an RTOS until you want to do precise scheduling Jul 26 08:59:19 you haven't yet mentioned anything that sounds like it requires real-time constraints, and even if such requirements show up there are still plenty of options when using linux (e.g. using an RT kernel and/or using PRU for the hard real-time bits) Jul 26 08:59:37 if you specifically want to get experience with TI-RTOS, then again... this is the wrong support forum for that Jul 26 09:00:03 go to e2e.ti.com in that case Jul 26 09:00:44 the network has no time constrains but the response to the analog input has to be quick Jul 26 09:00:57 define "quick" Jul 26 09:01:36 also what "response" .. so far you've just mentioned data recording Jul 26 09:01:45 payroll is a real-time application. If it doesn't run on time huge problems happen :) Jul 26 09:02:05 i will be measuring analog input and changing the output according to it as soon as possible Jul 26 09:02:14 but in us ? ms ? Jul 26 09:02:20 ns ? Jul 26 09:02:38 "quick" means very different things to different people. give actual numbers please Jul 26 09:02:43 xD Jul 26 09:03:45 ms can be enough but I´m not sure yet Jul 26 09:03:55 ms isn't quick, it's slow as fuck Jul 26 09:04:06 XD with BBB Jul 26 09:04:18 ms can mostly be handled by python, shellscripts etc Jul 26 09:04:33 mmm Jul 26 09:04:57 i will check how fast i need Jul 26 09:05:09 thanks for the advice guys Jul 26 09:05:18 adobe xd ? Jul 26 09:06:10 if it's for human i/o then tens of ms is usually ok Jul 26 09:06:28 unless something like musical instrument interfaces Jul 26 09:06:36 no no is a automatic machine Jul 26 09:07:13 if you want 1ms response time then using a thread/process with real-time scheduling to handle the "real-time" (*cough) part on a plain linux kernel should be able to give a high degree of reliability with little effort Jul 26 09:07:19 i want to change the power of the machine according to the measuring I´m getting Jul 26 09:08:04 if you want hard real-time then there's always the option of using PRU, although it would mostly be waiting for the ADC conversion :P Jul 26 09:08:26 (for digital signals, PRU can give a response time in the order of a few dozen nanoseconds) Jul 26 09:08:36 mmm Jul 26 09:08:52 beaglebone is very flexible like that Jul 26 09:09:00 but I wouldn't go for that option unless easier options have been evaluated and found to be inadequate Jul 26 09:09:13 from shell script at the slow end to pru at the fast end Jul 26 09:09:16 but the option is there if you need it in the future Jul 26 09:09:42 and I guess an fpga cape if you're pushing it Jul 26 09:10:53 well at that point it's not really the bbb anymore... you can fix the response time of pretty much anything by plumbing an fpga into it ;) Jul 26 09:11:11 xD Jul 26 09:11:41 thanks for the support guys Jul 26 09:13:05 i think I'd start with a simple C program at user level and migrate to ROS if you need multithreaded realtime Jul 26 09:13:19 with lots of subsystems Jul 26 09:13:48 armv8 board from ti, WHEN! Jul 26 09:14:09 then lets say i will start with a BBB with linux and test it. Jul 26 09:15:05 kremlin: http://www.ti.com/tool/tmdx654gpevm Jul 26 09:15:14 where are the examples. where i can start. examples for eth a/d i/o etc Jul 26 09:15:45 MarineroBBB: networking on linux is just networking on linux... nothing BBB-specific about it, use whatever Jul 26 09:15:58 zmatt: aha! Jul 26 09:16:04 gorsh. Jul 26 09:16:15 kremlin: or http://www.ti.com/tool/tmdx654idkevm Jul 26 09:16:16 i'm going to need to mail TI and ask for a sample Jul 26 09:16:18 or at least a discount Jul 26 09:17:16 MarineroBBB: as for using the analog inputs... depends on the programming language I guess Jul 26 09:17:28 i like MQTT for simple channel-based i/o to rest of world. not especially fast but very flexible and easy to do adhoc Jul 26 09:18:35 i will go for c to have better performance Jul 26 09:18:51 in the simplest form you can use the sysfs interface to read the ADCs, which is really simple in any programming language (you just read the value from a file basically), but not great performancewise probably. Or you can use the iio device, but I don't know much about how that works exactly. Presumably there are examples and/or higher-level wrappers for it... somewhere... Jul 26 09:19:08 i prefer that, I'm not really into python Jul 26 09:19:23 yeah I don't like python either Jul 26 09:20:58 but tbh it's used for gnuradio plumbing so probably reasonable performance Jul 26 09:22:05 so far I've actually used the ADC by reserving a piece of physical memory and manually (from userspace, using uio and/or /dev/mem) setting up the ADC to sample the inputs at a fixed rate and an EDMA channel to transfer the readings to the buffer, which can then be mmap()ed by applications that care about the values Jul 26 09:22:27 nah, the plumbing probably just didn't require any performance Jul 26 09:22:35 python is pretty slow Jul 26 09:22:40 well, true Jul 26 09:22:48 its semantics make it impossible to optimize in any shape or form basically Jul 26 09:23:47 that (ADC stuff) sounds like something Bela would make very easy Jul 26 09:24:00 basically anything you do... reading a variable, writing a variable, calling a function, requires doing a lookup by name at runtime Jul 26 09:24:27 that's also why it can almost never report typos as errors until execution hits the typo Jul 26 09:24:42 which is something I really hate about python Jul 26 09:24:56 i thought it did a fair bit in some sort of jit compiler ? Jul 26 09:25:10 the semantics *require* runtime lookups Jul 26 09:25:16 so there's nothing to optimize there Jul 26 09:26:06 also I don't think the main python implementation does jit Jul 26 09:26:43 it just parses to bytecode, but the problem is that name resolution isn't done at that time, it's done when the bytecode is executed (each time) Jul 26 09:27:00 i guess it's really only tokenization. so it doesn't have to look up the next execution unit but still has plenty to do in terms of variable access Jul 26 09:27:17 it's more than tokenization Jul 26 09:27:25 but not enough :P Jul 26 09:30:14 i also dislike python for its multiple implementation details. I think people tend to grow their installation as the application develops and it can be hard to replicate their environment when you want to run it yourself. and they've 'fixed' this by some sort of containerisation so you can have lots of independent environments. Jul 26 09:30:43 but often, I want to create something by merging other things, and that doesn't fit the model well Jul 26 09:31:10 artag: https://pastebin.com/254m2EiN Jul 26 09:34:08 I haven't really run into that issue myself yet, but then my main experience with python is writing py-uio, I haven't done much with other people's python code Jul 26 09:34:45 they can be equivalent at some level, I'm sure. but when I pick up some reasonably sized preexisting python project I always seem to spend way too much time trying to resolve dependencies Jul 26 09:34:54 that sucks Jul 26 09:35:08 plus so often people still use obsolete versions of python -.- Jul 26 09:35:20 *cough* python3 *cough* Jul 26 09:35:22 *python2 Jul 26 09:35:29 damnit, bad typo Jul 26 09:35:56 .. and don't necessarily make it clear (to non-pythonistas) which one Jul 26 09:37:09 anyway. I must go and do some REAL hardware involving a hacksaw Jul 26 09:37:18 later Jul 26 09:37:21 .. Jul 26 09:37:45 kremlin: but, instead of getting an am6 board, maybe just wait for am7/j721e Jul 26 09:38:57 kremlin: https://liktaanjeneus.nl/j721e-block-diagram.png Jul 26 09:42:21 AND it's got a PRU! Jul 26 09:42:24 awesome possum Jul 26 09:42:43 kremlin: look at those cores... 2x cortex-a72, 6x cortex-r5f, 1x c71x, 2x c66x, 8x pru, an m3 (though I wonder if it can be used for custom code on GP parts), probably more I'm overlooking Jul 26 09:43:05 it's like four computers Jul 26 09:43:20 actually I think the am65x has more pru cores Jul 26 09:43:52 it sure does Jul 26 09:44:28 am65xx has three quad-core pru subsystems, j721e has "only" two of 'em Jul 26 09:45:02 are they still 200MHz? Jul 26 09:46:30 on am654x they're 250 MHz Jul 26 09:46:39 nice Jul 26 09:47:14 ok, I was checking the Analog input sample rate I need. Jul 26 09:47:18 almost certainly ditto for j721e since that frequency is related to their use for gigabit ethercat Jul 26 09:47:27 MarineroBBB: sample rate was never the question Jul 26 09:48:13 response time was Jul 26 09:48:21 i need to store the value of the analog imput every 50us Jul 26 09:49:23 that's of no importance, the hardware can sample the input every 50us and store the values in memory Jul 26 09:49:46 in itself that doesn't impose any significant constraints on software Jul 26 09:50:19 nice. i will check a little bit more to see some critical time restrictions Jul 26 09:50:31 thanks again zmatt Jul 26 09:50:46 and artag Jul 26 09:51:59 the only constraint it would impose on software is that the data being recorded to memory is consumed in some way (processed and/or written to storage and/or network) Jul 26 09:52:12 fast eonugh to prevent buffer overrun Jul 26 09:52:28 but that's a bulk processing requirement, not a real-time latency requirement Jul 26 09:53:57 but, if 2 bytes are used pre sample, and one analog input is used, then that's only 40 KB/s of data... it's really not much Jul 26 10:46:53 hi Jul 26 10:49:21 how can i use bbb with linux code Jul 26 10:49:45 you put linux code into bbb, and then use it! Jul 26 10:52:01 Just do the ssh and get into your beaglebone Jul 26 12:10:57 zmatt are you suggesting some do something? to get results? Jul 26 13:47:12 m Jul 26 14:30:51 GenTooMan: ? Jul 26 15:26:22 zmatt "mehrdad: how can i use bbb with linux code?" "zmatt: you put linux code into bbb, and then use it!" "GenTooMan: are you suggesting someone do something? to get results?" Jul 26 15:27:40 context helps I guess. Jul 26 15:45:35 GenTooMan: I just saw your most recent two messages before I saw the conversation they were about Jul 26 15:45:47 I assumed you *must* be jokingly summarizing a conversation Jul 26 15:46:06 not literally quoting the whole thing Jul 26 17:45:18 Hey, what's up? Jul 26 17:47:13 * zmatt looks up Jul 26 17:47:15 ceiling. Jul 26 17:48:47 makes sense Jul 26 17:51:17 I'm trying to boot off a micro sd card. Is there any OS other than Debian that's supported by Beaglebone Black? I've tried Ubuntu and Fedora but Debian is the only that booted successfully. Jul 26 17:52:07 debian is the only one officially supported anyway... I'm sure it's possble to get any distro running if someone puts in enough effort Jul 26 17:52:33 rcn also makes unofficial ubuntu images which should work, but I'd recommend sticknig to debian Jul 26 17:53:24 unless you enjoy debugging obscure problems noone can help you with because they don't happen on debian. if you use an uncommon distro (uncommon for the platform that is), then you put yourself on the frontline for debugging issues Jul 26 17:55:23 also, if you want to boot from sd card instead of eMMC, I strongly recommend wiping eMMC (using sudo blkdiscard /dev/mmcblk1 after booting from sd card) ... this ensures that the bootloader itself will also be loaded from sd-card and not from eMMC (which is the default). this can avoid issues that may result when the bootloader on eMMC doesn't really match the system on sd card Jul 26 17:56:35 alternatively you can temporarily bypass the bootloader on eMMC by powering on while holding down the S2 button (the button closest to the sd card slot). you can (and should) let go of the S2 button once the power led turns on Jul 26 17:59:16 This is helping me out a lot. I really appreciate it. When I do the temporary bypass method with an OS other than debian it doesn't even try to boot. The power light comes on but that's it. Jul 26 17:59:40 sounds like whatever images you tried were not remotely compatible with the beaglebone, i.e. missing a suitable bootloader Jul 26 18:00:15 unless the image is *for* beaglebone (or at the very least for "TI AM335x" boards), it's not going to work Jul 26 18:01:17 the kernel can be broader in scope (e.g. one kernel for a variety of ARM-based boards is possible, although it's probably suboptimal for all of them), but the bootloader has to be pretty specific Jul 26 18:01:55 userspace can be mostly hardware-independent (other than having to be compiled for armhf obviously) Jul 26 18:09:48 You pretty much answered all the questions I had. Thanks so much for your help :):] :] . Jul 26 18:16:18 I now see a network interface from the bbb Jul 26 18:16:30 I can't figure out how to use it, though Jul 26 18:17:10 I'd like to keep my wifi connection and just ssh over it Jul 26 18:20:38 the getting started page seems to say the IP address it uses depends on the host OS? Jul 26 18:58:21 ksft: one is RNDIS, one is CDC-ECM. both options should work for you Jul 26 18:59:39 there shouldn't really be any "how to use it" steps... you just ssh to the IP (or to "beaglebone.local" if you have libnss-mdns installed, or libnss-resolve if you use systemd-resolved and MulticastDNS is enabled in it) Jul 26 18:59:57 Tan7: do not send unsolicited private messages, ask your questions here in the channel Jul 26 19:00:16 ksft: of course I'm assuming here your OS is well-behaved and not broken ;) Jul 26 19:00:27 right, possibly bad assumption in my case Jul 26 19:00:50 what network manager do you use? Jul 26 19:00:56 NetworkManager? Jul 26 19:01:22 and it doesn't automatically bring up the interfaces? Jul 26 19:01:28 I think it does Jul 26 19:01:36 but I don't know what address to connect to Jul 26 19:01:39 is it 192.168.7.2? Jul 26 19:01:54 ksft: http://beagleboard.org/getting-started#step2 Jul 26 19:01:56 I can't connect to that one -- I think it's trying to use my wifi adapter Jul 26 19:02:06 it shouldn't Jul 26 19:02:24 can you share (via e.g. pastebin) the output of the "ip addr" and "ip route" commands? Jul 26 19:04:11 this is the part of ip addr's output that I think I relevant: https://pastebin.com/raw/bNAX8J8C Jul 26 19:04:35 o.O Jul 26 19:04:40 ip route just lists entries on my wifi adapter Jul 26 19:05:05 is it supposed to assign an IPv4 address? Jul 26 19:05:27 yes, and there should be two interfaces Jul 26 19:05:34 huh Jul 26 19:05:48 (one RNDIS, one CDC-ECM) Jul 26 19:06:19 assuming you enabled both options in your kernel config, and we're talking about a BBB in default configuration (i.e. not one where you tried to switch to using g_ether) Jul 26 19:07:32 state UNKNOWN also feels weird to me Jul 26 19:07:48 I think I loaded g_ether Jul 26 19:09:12 you can't use g_ether without also doing more configuration on the beaglebone side, including disabling the startup script that normally sets up the usb networking Jul 26 19:09:23 otherwise neither g_ether nor the stuff that would normally be setup wlil be functional Jul 26 19:09:52 ah Jul 26 19:10:09 which would explain why 1. you only get one interface (as expected with g_ether) rather than two (as expected with the default) 2. why no address is configured on that interface Jul 26 19:11:54 Sorry for interrupting but zmatt could you provide me that eMMC wiping command again? Jul 26 19:12:09 sudo blkdiscard /dev/mmcblk1 Jul 26 19:13:52 Thank you Jul 26 21:57:04 To recap...the other day I was describing something. This something was the idea of backfeed into the BBB via a simple LED circuit. Jul 26 21:57:05 ... Jul 26 21:57:13 Okay. Jul 26 21:57:24 Now, what is the backfeed of current into the BBB called? Jul 26 21:58:46 I mean...there has to be a name or some type of description that describes the wrong direction of current in the wire/line that feeds back into the BBB. Right? Jul 26 21:59:20 I came up w/ backfeed but it is mostly for generators and grids. Jul 26 22:02:45 backwards flow of charge? Jul 26 22:04:49 negative flow? Jul 26 22:05:07 Negative flow sounds like a poor mannered lady. Jul 26 22:05:08 Ha! Jul 26 22:11:34 Come on. Jul 26 22:12:58 alternating? Jul 26 22:26:30 So, in Direct Current, the current only moves in one direction. Jul 26 22:30:08 So, if I was using the BBB w/ a <= 2.0Ah power bank (DC), would I need to worry about "backfeed" from a LED circuit to "hit" my GPIO pin that is protected via pull-down resistor? Jul 26 22:31:48 Indian people, that I will not name, are stating things differently and their idea is that the 6mA to 8mA per GPIO will not damage a LED that can handle 20mA. Jul 26 22:34:30 Oh well. Win some, lose some. BBL! Jul 27 01:14:45 you can have "backfeed" from all kind of devices set_ Jul 27 01:15:12 for instance if you have a big beefy mosfet you need to pump out the charges from it when the voltage goes low Jul 27 01:15:26 but that's what drivers are for, to not have to directly interface mosfet gates with your mcu outputs Jul 27 01:15:33 so you should be safe if you use a proper led driver Jul 27 01:26:10 Okay. Jul 27 01:26:19 mawk: Okay. Jul 27 01:27:05 what am I saying, this is not what backfeed is Jul 27 01:27:20 the backfeed will happen when the charge of that mosfet is dynamic Jul 27 01:27:30 but anyway yes even with DC you will often have reverse voltages Jul 27 01:29:29 what I was saying with reverse charge damaging the IC won't happen, the IC will be damaged first in the charging phase Jul 27 01:29:36 it will never live up to being destroyed by the discharge Jul 27 01:42:47 Oh. Jul 27 01:43:06 Okay. Jul 27 01:43:17 For application purposes... Jul 27 01:44:06 This tutorial stated to me that the seven seg. display would not need any resistors when dealing w/ the BBB. Jul 27 01:44:41 zmatt had told me over and over that the resistor is to protect the LED and the GPIO on the BBB. Jul 27 01:45:02 This PMIC might be damaged in the process. Jul 27 01:45:16 B/c of what we are describing. Jul 27 01:45:54 Sorry for the sentence fragments. Jul 27 01:47:13 Which IC are you referring to? Jul 27 01:50:27 mawk: Which IC are you discussing? The PMIC or the entire circuit (IC)? Jul 27 01:52:19 I don't know what zmatt exactly said but if you're talking about just powering LEDs then yes a resistor is required Jul 27 01:52:34 Okay. Jul 27 01:52:37 a led is a device that above a certain threshold will look just like a plain wire, eating all possible current Jul 27 01:52:48 so you need to be limiting that current, that's all Jul 27 01:52:53 Okay. I got it. Jul 27 01:52:57 and the easiest thing for this is to use a resistor Jul 27 01:53:03 but for bigger leds you use a driver Jul 27 01:53:44 it's not exactly all possible current, but the current will rise exponentially with voltage applied through the LED Jul 27 01:54:03 I've written a program to run on the PRU that takes some input in the form of several bytes written to its memory before it's started Jul 27 01:54:05 when I said IC it was GPIO pin I guess Jul 27 01:54:17 mawk: I have a seven seg. display and this fellow online is going around stating that the BBB is okay w/out the resistors for this seven seg. display. Jul 27 01:54:18 Okay. Jul 27 01:54:24 mawk: Okay. Jul 27 01:54:35 I want to run it periodically, probably a few dozen times a second Jul 27 01:55:05 It is not smart to go w/out resistors for LED displays w/out a driver. Got it. Jul 27 01:55:05 ! Jul 27 01:55:22 I'd like to periodically change the input and write over the memory at the next available opportunity Jul 27 01:55:34 what's the best way to set this up? Jul 27 01:56:01 I am lucky I am not disabling my GPIO pins. mawk: thank you for clarification on this odd tutorial. Jul 27 01:56:07 it only takes a few milliseconds for the program to run, so I guess I don't need to worry about a buffer filling up while it's waiting Jul 27 01:56:41 Do not look at me, ksft. I am still new to PRU instances but Godspeed! Jul 27 01:56:48 when it's a led *display* I would hope it has a driver, but I don't know which kind of display we're talking about set_ , like the little 16×2 LCD ? Jul 27 01:56:52 Someone should be around soon. Jul 27 01:56:54 these ones have no drivers afaik Jul 27 01:56:59 you just have led cathode and led anode pins Jul 27 01:57:08 Cathode common. Jul 27 01:57:15 led forward voltage and expected current should be in the datasheet Jul 27 01:57:23 maybe I'll have one process running it repeatedly and another that communicates with the first one by a pipe or something? Jul 27 01:57:24 Generic! Jul 27 01:57:36 the current you feed into should not exceed the capacity of BBB gpio outpus Jul 27 01:57:41 which is lower than with microcontrollers Jul 27 01:57:43 I should invest in better gear. Jul 27 01:57:52 you have a per pin maximum, and a grand total maximum Jul 27 01:57:53 Okay. Jul 27 01:58:01 Oh! Jul 27 01:58:09 So, at 250mA? Jul 27 01:58:17 lol Jul 27 01:58:21 250mA is quite enormous Jul 27 01:58:28 if I had to guess I'd say 5 mA Jul 27 01:58:34 I think it is 6 to 8mA per GPIO, right? Jul 27 01:58:35 my stm32 is 20mA per pin, 45mA total, something like this Jul 27 01:58:43 Oh. Okay. Jul 27 01:58:56 you can make a quite lame driver using a transistor Jul 27 01:59:05 that way the gpio won't be stressed out, only the power line of the bbb Jul 27 01:59:16 which can handle higher currents Jul 27 01:59:22 Okay. Jul 27 01:59:33 or just use a led driver Jul 27 01:59:52 which is normally a constant current source but you shouldn't have to know that Jul 27 01:59:58 anyway your display doesn't require 250mA does it ? Jul 27 02:00:03 Got it. This is good info. I will just a LED driver. I am going to invest one day in learning more w/ better equipment. Jul 27 02:00:05 I doubt it. Jul 27 02:00:31 The fellow gave me a link to purchase a cheap LED display. Jul 27 02:00:38 do you have the link ? Jul 27 02:00:42 Sure. Jul 27 02:00:43 I'd look at the datasheet Jul 27 02:00:44 Please hold. Jul 27 02:00:56 mawk: Do not worry about it. Jul 27 02:01:04 There is no datasheet. Jul 27 02:01:15 yeah but if it's generic enough I can work things out Jul 27 02:01:25 I looked and contacted these people. It is a very generic thing. Jul 27 02:01:45 They do not have a datasheet plus they did not reply at all. Jul 27 02:02:32 Okay mawk: I have an idea. There is a commmon cathode segment display. It is basically just some leds w/ a cover on it. Jul 27 02:02:43 so your thing is that ? Jul 27 02:02:48 7 segment led display? Jul 27 02:02:52 Yes. Jul 27 02:02:58 I see it used with a driver typically yes Jul 27 02:03:04 No idea. Jul 27 02:03:06 all of that is cheap Jul 27 02:03:10 Oh. Jul 27 02:03:53 mawk: I can look into a driver one day. I do not have one right now and the tutorial online did not have any recommendations for a driver. Jul 27 02:05:07 So, for now, I have no driver for this display nor do I have any resistors hooked up to it, e.g. say on a breadboard or the like. Jul 27 02:05:22 I can use resistors. I have some. Jul 27 02:06:24 I have 1k ohm resistors. I can use them for this project. The idea of this project is that the "common cathode" uses one ground for all the LEDs on this display. Jul 27 02:06:54 yeah Jul 27 02:07:04 well if you do the calculation and you don't exceed the grand total use gpios for now Jul 27 02:07:13 and don't exceed individual driving limits tho Jul 27 02:07:20 you said 8mA per gpio, that sounds too low for leds Jul 27 02:07:33 Oh. Jul 27 02:07:41 That is what is happening I guess. Jul 27 02:07:45 you can use a transistor per led Jul 27 02:07:56 Okay. I can try that... Jul 27 02:07:59 if you just want to play Jul 27 02:08:05 I only have about three or four right now. Jul 27 02:08:07 Okay. Jul 27 02:08:22 They are lost in action for now, though. I will have to dig for them. Jul 27 02:08:22 with common cathode you need PNP transistors Jul 27 02:08:28 I don't think you have these Jul 27 02:08:28 Okay. Jul 27 02:08:37 I think you are right. I have NPN. Jul 27 02:08:58 npn are good for connecting to ground, but you have a common ground Jul 27 02:09:03 Right. Jul 27 02:09:15 I have a common GND. Jul 27 02:09:32 can still use npn in some kind of inefficient way I guess, but anyway you don't have enough of them Jul 27 02:09:43 Right-o. I guess I am ill-prepared for now. Jul 27 02:09:49 Dang it. Jul 27 02:10:18 I need to make a list of items to purchase when I get big time in life. Jul 27 02:10:31 Trans. caps. and so on... Jul 27 02:10:57 I wonder why these people were acting like they were so familiar w/ the BBB and LEDs or displays. Jul 27 02:11:05 ... Jul 27 02:11:08 Odd days. Jul 27 02:11:14 Anyway mawk: Thank you. Jul 27 02:11:24 I will need to make a purchase one day. Jul 27 02:11:33 I cannot do anything right now. Jul 27 02:11:36 Aw! Jul 27 02:12:13 And the dang store for electronics closed in our town, i.e. Radio Shack... Jul 27 02:12:20 The shack is down. Jul 27 02:19:59 here's a crude driver set_ http://tinyurl.com/y37rkpy6 Jul 27 02:20:15 of course bipolar transistors aren't the best Jul 27 02:20:32 the mosfet is what comes close to a true electrically commanded switch Jul 27 02:21:02 the important thing is it allows gpio to sink/source less current than what the leds receive Jul 27 02:21:06 it's like a current amplifier Jul 27 02:21:16 note that the logic is inverted here Jul 27 02:21:23 if the gpio is low, segment is on, if the gpio is high, segment is off Jul 27 02:22:49 here if you know a bit about bipolar transistor theory you see that a pnp is infinitely better since we're able to drive it to full saturation, so it has a low voltage drop and doesn't heat up Jul 27 02:23:10 while if you do the same kind of configuration with a npn you will have a ~ 0.6V voltage drop Jul 27 02:23:20 instead of the 0.05V here Jul 27 02:25:00 or get a different polarity 7seg Jul 27 02:37:42 Okay. Thank you for the heads up. Jul 27 02:38:10 Dang ideas and my lack of edu. on these facts. Jul 27 02:38:58 see how I made the nice little alignment Jul 27 02:39:01 it's a work of art Jul 27 02:41:47 Nice. Jul 27 02:42:24 mawk: Thank you. **** ENDING LOGGING AT Sat Jul 27 02:59:57 2019