**** BEGIN LOGGING AT Thu Sep 10 07:36:15 2020 Sep 10 08:36:08 Greetings, zmatt. Your dts file (https://pastebin.com/ERDuqARR) mentions 4 pins for spi - how does the driver / device tree know, which pin he has to use as which function (miso, mosi, etc.)? Sep 10 08:44:51 To be more exact, there is data sent from the bb, and data sent TO the bb (oscilloscope says so), but the encx24j600 driver does not receive them **** BEGIN LOGGING AT Thu Sep 10 09:23:04 2020 **** BEGIN LOGGING AT Thu Sep 10 09:49:35 2020 Sep 10 09:54:38 mindbyte: which pin is which function is mentioned in comments of the pinmux block (lines 43-46 of the paste) Sep 10 09:56:43 Okay but how does it know which pin is used for which function? I just enter these 4 pins and he magically knows it? Sep 10 09:57:34 there's no magic involved, each of these pins only has exactly one of these signals of spi0, the one listed in the comments Sep 10 09:58:12 Okay Sep 10 09:58:30 I mean... it works: https://wille.io/img/IMG_20200910_115625.jpg but the beaglebone doesn't seem to receive what is sent by the slave device Sep 10 09:58:40 if you want to know what signals are available on what pins (using what mux modes), see e.g. my pins spreadsheet ( https://goo.gl/Jkcg0w ) Sep 10 09:59:00 what does it receive instead? all zeros? all ones? Sep 10 10:00:17 Sadly I can't see where the module stops. It seems to run into this timeout after sending EUDAST_TEST_VAL (0x1234): https://github.com/beagleboard/linux/blob/7a920684860a790099061b67961d0b5ffa033fdf/drivers/net/ethernet/microchip/encx24j600.c#L465 Sep 10 10:02:49 any chance it's just a bad connection to the BBB? poor contact / broken wire? Sep 10 10:05:28 accidently used wrong pin? Sep 10 10:06:37 you could try testing these same transfers from userspace (using the spidev driver) to see what response you're getting Sep 10 10:16:23 Good idea, I'll test it with spidev, thanks Sep 10 11:14:11 trying to update the debian image Sep 10 11:14:24 and downloaded this Sep 10 11:14:26 AM3358 Debian 10.3 2020-04-06 4GB SD IoT Sep 10 11:15:18 but, couldn't find the "cd/boot" directory Sep 10 11:15:41 HELP! HELP! HELP! Sep 10 11:15:52 @zmatt data looks fine with spidev... the slave answers with predicted answers: https://pastebin.com/pz3tpe9Z Sep 10 11:16:29 a good, obedient slave Sep 10 11:41:15 newGuy: what do you mean? Sep 10 11:41:42 newGuy: also, screaming "HELP! HELP! HELP!" is appropriate when your house is on fire or something Sep 10 11:42:06 not having trouble updating an embedded linux system Sep 10 11:44:51 newGuy: if you want to reflash your beaglebone to the latest image (note: this procedure completely wipes eMMC and installs a fresh system) the easy way is to download the latest "Flasher" image (AM3358 Debian 10.3 2020-04-06 4GB eMMC IoT Flasher), write that to SD card (preferably using Etcher, see etcher.io), and boot the beaglebone from that card Sep 10 11:44:59 it will automatically proceed to reflash Sep 10 12:01:20 mindbyte: when testing with the overlay for the encx24j600, did you also have cape-universal enabled? Sep 10 12:07:05 mindbyte: since your userspace test uses mode 0 while cape-universal inexplicably puts spi-cpha; on &spi0/channel@0 Sep 10 12:09:36 and your overlay would not (in fact cannot) fix/override that, and I don't think the driver overrides the spi mode selected in DT either Sep 10 12:10:18 so be sure to test with cape-universal disabled Sep 10 12:17:20 (if that fixes the issue and you do want cape-universal enabled then you'll need to do a 1-line fix in am335x-bone-common-univ.dtsi and recompile the base dtb) Sep 10 12:32:20 Anyone familiar with a 20V(or at least 8.4v)input pmic that can supply the power domains needed by am3358? Looking to make a typeC based device with a dedicated battery charger **** BEGIN LOGGING AT Thu Sep 10 12:44:10 2020 Sep 10 12:45:25 i found TPS650860 in the regulators section of drivers in the kernel, but it is way too intergrated, i really dont need a kitchen sink in this portable device. Sep 10 12:46:51 "AM335x PMIC Selection Guide" maxes out at 5.5V Sep 10 12:48:07 Konsgnx: why do you need to run 2 cells in serial? Sep 10 12:48:46 design requirement. it's a battery pack that has no 1-cell options Sep 10 12:48:59 ic Sep 10 12:49:06 yea i saw that, bah! Sep 10 12:50:34 annoyingly they make sure to mention that the tps62570... (whatever is in the beaglebone's) pmic is 20v tolerant, but then don't recommend you exceed 5.8V Sep 10 12:53:13 tps65217 Sep 10 12:53:25 ya Sep 10 12:54:22 it can *withstand* 20V, as in absolute maximum ratings Sep 10 12:55:03 but it goes into overvoltage protection mode around 5.8-6.4 V Sep 10 12:55:48 there you go: https://e2e.ti.com/support/power-management/f/196/t/768180 Sep 10 12:56:29 so just add a dropper resistor like in ye olde days :) or 1-2 tube heaters Sep 10 12:56:41 or more sanely a buck converter :P Sep 10 12:57:24 in a pinch Sep 10 13:01:59 so why "20V(or at least 8.4v)input" when a 2S li-ion battery is not going to get any higher than about 8.4 V ? Sep 10 13:09:28 20v would be better as the design could then accomodate a larger battery. I think 4 cell is max for the bq25710 Sep 10 13:09:41 just for future-proofing Sep 10 13:11:10 what do you guys think of the LM61460? it seems to have a high conversion efficiency even when the consumed current is negligible. Sep 10 13:11:36 https://www.ti.com/product/LM61460 Sep 10 13:12:58 do you need 6A? Sep 10 13:13:49 probably not, but the quiescent current is really good on this one. Sep 10 13:14:10 how much current do you need? Sep 10 13:16:20 if 1A is enough, https://www.ti.com/product/LMZM23601 is very convenient Sep 10 13:17:52 hahah, I used that one on a previous design. did not work well... was my setup though, it doesn't drive analog power domains particularily cleanly. Sep 10 13:18:19 It is a really nice chip. Sep 10 13:18:33 you might want a linear post-regulator for sensitive analogue stuff Sep 10 13:18:50 it should be fine for powering a beagle Sep 10 13:19:24 I like it because it's so simple Sep 10 13:19:24 if it did 2A it would be great. yea,,,lessons learned, granted the entire circuit fit into a tiny 10mm tube, and as is everything was crammed in, so no space left at all. Sep 10 13:20:19 I would want to have higher current available in case a usb hub that i plan to add gets something plugged in. Sep 10 13:22:08 is 17V input enough? https://www.ti.com/product/TPS82130 Sep 10 13:23:43 those all-inclusive modules are convenient if you're lazy Sep 10 13:24:17 yea, that looks good, gotta compare the light load efficiency with this one:https://www.ti.com/product/TPSM53603. since this will always be on, it needs minimum draw. Sep 10 13:27:59 gah, I accidentally limited the search earlier Sep 10 13:29:09 they have loads of modules in that general area Sep 10 13:29:38 might be worth checking what ADI has to offer too Sep 10 13:30:33 Nope, the tps82130 is better..ahhhh, nvm, gunna stick to the lm61460, i just copied its ref design. Sep 10 13:30:46 Do they have something like that for boost? I expect so, I just haven't seen one. Sep 10 13:31:27 When you have multiple converter ic's, to keep noise down, should you try to syncronize their switching frequencies? Sep 10 13:31:50 on the contrary Sep 10 13:31:53 you should spread them Sep 10 13:31:56 Ragnarok, https://www.ti.com/power-management/non-isolated-dc-dc-switching-regulators/step-up-boost/products.html#p238min=2;4.5&p238max=1.8;40 Sep 10 13:32:14 spread, by use of the sync pins? Sep 10 13:33:01 by not syncing them **** BEGIN LOGGING AT Thu Sep 10 13:34:55 2020 Sep 10 13:35:08 for emc compliance you want to avoid concentrating noise at a single frequency Sep 10 13:35:08 so for the ones that have variable switching, would you use the sync to force the 2(or more ) frequencies to not overlap? or just leave them be? Sep 10 13:35:16 that's why spread-spectrum clocking was invented Sep 10 13:35:47 Yea, Ragnarok, those filters on the side are actually quite user freindly. Sep 10 13:36:57 mru, that makes sense, I'm curious about how much effort typically goes into avoiding overlapping frequencies. Sep 10 13:37:33 I'd say that depends on how close to the limit you are Sep 10 13:37:50 and how much you care about regulations in the first place Sep 10 13:38:01 Konsgnx: It's the top quick search doesn't work. I found what I needed on the side, and it turns out only two "module" style are found. Good to know. Thanks again! Sep 10 13:38:36 mru, gotcha, and thanks! Sep 10 13:39:13 NP , I find TI's search one of the better ones for finding what you are looking for. Sep 10 13:39:30 do the regulators you're planning to use have configurable frequency? Sep 10 13:41:07 Nagh, I was just asking in general/ just realizing that the pmic and the lm61460 need to be compared for switching frequencies. Sep 10 13:47:23 do you have emc testing equipment? Sep 10 13:49:45 oscopes yes, and i have made a couple simple probes for emc testing, but nothing calibrated. Sep 10 13:50:13 you need a spectrum analyser Sep 10 13:50:49 I think we have one of those here. Sep 10 13:51:09 if you're going to sell the device, it'll need to be tested in a proper lab Sep 10 13:51:29 true, it will be sent out once we are happy with it. Sep 10 13:51:52 some quick bench testing can still avoid wasting expensive lab time Sep 10 13:54:17 just to double check, a short exposed center conductor from a coax hooked up to a spectrum analyzer is enough to be a point test probe. right? Sep 10 13:54:53 if the spectrum analyser has the usual 50-ohm input, you'll probably need an RF amp as well Sep 10 13:55:46 haha, would a hackrf be considered a spectrum analyzer in a pinch? Sep 10 13:56:07 no idea what that's capable of Sep 10 13:56:24 but don't worry too much about that now Sep 10 13:57:27 nearfield probes are useful for finding out exactly where emissions are coming from Sep 10 13:57:42 you need something else to measure the total emissions **** BEGIN LOGGING AT Thu Sep 10 14:28:50 2020 Sep 10 14:38:51 dodged a bullet yesterday Sep 10 14:39:11 I had my logic shifter plugged in backwards for my encoder Sep 10 14:39:44 of my shaft spun it would sent 5V back at the black gpio pins Sep 10 14:40:29 fortunately I forgotten to tighten the set screws on the hollow shaft encoder so when the shaft spun it didn't sparing me 70 bucks Sep 10 17:03:45 when trying to compile dts in /opt/source/, there is a syntax error regarding __timestamp__ Sep 10 17:04:30 e.g. /opt/source/bb.org-overlays/src/arm/AM335X-PRU-RPROC-4-14-TI-PRUCAPE-00A0.dts Sep 10 17:05:14 do i cp the dts to a new file and enter it in manually? Sep 10 17:05:55 use the makefile Sep 10 17:06:19 "make src/arm/AM335X-PRU-RPROC-4-14-TI-PRUCAPE-00A0.dtbo" in the bb.org-overlays directory Sep 10 17:06:30 (or just "make" to compile all dtbos) Sep 10 17:07:32 ok, and my goal is to load am335x-pru1-fw Sep 10 17:08:00 am I on the right track, by loading a new dtb? i noticed in lib firmware that am335x-pru1-fw is 0 bytes Sep 10 17:08:19 uh no, why would those things have anything to do with each other? Sep 10 17:09:32 if remoteproc-pru is enabled (it is by default) then the kernel driver will automatically load firmware (if present) onto the pru cores at boot Sep 10 17:10:07 is firmware only from ti, or is there a package? Sep 10 17:10:17 and yeah /lib/firmware/am335x-pru0-fw and am335x-pru1-fw are by default 0-byte dummy files, which you'd replace by your own Sep 10 17:10:21 what firmware? Sep 10 17:10:32 either am335x-pru1-fw Sep 10 17:11:03 ?? that's just the filepath where you need to put whatever software you want to load onto the pru cores Sep 10 17:11:27 i was trying to demo out the pru example code Sep 10 17:11:40 "the pru example code" ? Sep 10 17:11:47 hold on a sec, the path is... Sep 10 17:12:20 /usr/lib/ti/pru-software-support-package/examples/am335x Sep 10 17:13:05 (I should note I have close to zero experience with remoteproc-pru, I use uio-pruss instead... I wrote a python library that uses uio to let you load firmware onto the pru cores as well as interact with them) Sep 10 17:13:34 i understand there is uio, and rproc. are people still using uio? Sep 10 17:14:11 is one deprecated, or are they both being used Sep 10 17:14:42 yes, why not? it's more stable and remoteproc-pru still doesn't have the functionality it needs to actually be an adequate replacement Sep 10 17:16:39 (notably shared memory) Sep 10 17:18:37 i think i was mistaken. I was trying to Sep 10 17:18:38 remoteproc's api also kept breaking incompatibly with every major kernel release... though I think it may have stabilized now (not sure, I don't actually follow its development) Sep 10 17:19:20 poke around in /usr/lib/ti/pru-software-support-package/examples/am335x/PRU_ADC Sep 10 17:19:55 i found that the userspace code would load the fw. I was trying to run through the steps Sep 10 17:20:09 but the fw, it looks like is compiled in the PRU_ADC folder Sep 10 17:20:13 maybe Sep 10 17:20:28 yep, there's a makefile there Sep 10 17:21:03 software for the pru, as if it's a micro. then userspace calls to the pru from linux. Sep 10 17:21:06 i get it now. Sep 10 17:21:29 you'd need to copy it e.g. to your homedir to avoid permission errors Sep 10 17:22:13 is it possible to program the pru all on its own? Sep 10 17:22:33 as if it were just a micro without an os on the arm cpu Sep 10 17:23:05 no, the arm cpu needs to enable pruss, load firmware into a pru core's instruction ram, and then enable the core Sep 10 17:25:03 (when using remoteproc-pru this is done by the kernel driver, when using uio this is done from userspace) Sep 10 17:25:13 or get an arduino Sep 10 17:25:15 lol Sep 10 17:25:34 an adruino is no replacement for pru though... nor vice versa Sep 10 17:25:53 or make the pru core into an arduino Sep 10 17:25:59 it would make using the pru easier if just for learning purposes if it could be separated was my thought Sep 10 17:26:11 I don't see how Sep 10 17:26:28 being able to load code onto it from linux actually makes it easier to study Sep 10 17:26:45 well you may be right. i'm just guessing. Sep 10 17:27:01 even more so with my py-uio library, which lets you interactively poke around in the pru subsystem, e.g. you can view and modify a pru core's registers whenever the core is halted Sep 10 17:27:13 unless you don't know linux :} Sep 10 17:28:02 ( https://github.com/mvduin/py-uio ) Sep 10 17:30:16 remoteproc-pru does add more complication, with its resource tables and the excessively slow and complicated rpmsg protocol it tries to convince you to use to communicate between linux and pru Sep 10 17:30:39 how would you make a core into an arduino Sep 10 17:30:50 also there has to be some benefit to remote proc right Sep 10 17:30:55 cannot be all garbage Sep 10 17:30:59 ? Sep 10 17:31:26 I understand the idea behind remoteproc.. it's just not right for pru imho Sep 10 17:32:40 add a core definition to the boards manager mastermind, handle pin-map , digitalwrite, digitalread, and you can set/clear pins using arduino enviroment Sep 10 17:33:23 things like tone, printf, pwm, will be more complicated to implement but should be doable. Sep 10 17:33:43 printf to *what* :D Sep 10 17:34:18 also, reminder that each pru core only has 8KB of instruction ram (i.e. 2048 instructions) Sep 10 17:34:28 haha, isnt there a uart internal to the pru's, otherwise to shared mem perhaps Sep 10 17:35:14 ooh wait that would be serial.print Sep 10 17:35:29 I don't even want to think about this abomination being suggested ;) Sep 10 17:36:23 you could compile the program from arduino running on the beagle, compile and run all in one. Sep 10 17:36:56 in seriousness, maybe this would be something more useful for the arm cores that beagle-ai has... Sep 10 17:39:21 don't know that it's really that useful even if it did work completely, but there is a start for that in cloud9-examples (compile a .ino file). Sep 10 17:40:18 as in compile an ino for an arbirtary board definition? Sep 10 17:42:46 it compiles .ino sketch files to the board it is running on using Linux userspace interfaces. Sep 10 17:43:13 the project hasn't been updated and doesn't include BeagleBone AI support, but could be updated to use the GSoC generated cape compatibility layer. Sep 10 17:43:58 haha, interesting, so the main arm processor is the "core"? Gotcha. Sep 10 17:44:01 so, getting it all to "work" wouldn't be that hard, but there is question if that is a "good" way to use Beagles. Sep 10 17:44:12 I'll grab some repo pointers. Sep 10 17:45:01 https://github.com/beagleboard/cloud9-examples/blob/v2020.01/.c9/runners/sketch.run Sep 10 17:45:23 https://github.com/beagleboard/cloud9-examples/blob/v2020.01/extras/Userspace-Arduino/Makefile Sep 10 17:45:46 https://elinux.org/Userspace_Arduino Sep 10 17:46:29 I think being able to compile for the sub-proccessors in the sitara family makes some sense. But it does break the intended use of do one thing fast and efficiently. Sep 10 17:46:46 also, Arduino has added support for BeagleBone Black in their cloud IDE. Sep 10 17:47:01 targeting PRU would make more sense. Sep 10 17:47:16 there have been several failed attempts by students who don't understand the complexities. Sep 10 17:47:49 another project: https://github.com/AbhraneelBera/wiringBone/issues/2 Sep 10 17:49:04 probably the best way to fix one of these projects to work on both BeagleBone Black and BeagleBone AI: https://deepaklorkhatri.me/GSoC2020_BeagleBoard.org/ Sep 10 17:49:28 the new test images should include the dt updates natively. Sep 10 17:50:21 * jkridner[m] goes afk **** BEGIN LOGGING AT Thu Sep 10 23:29:35 2020 Sep 11 00:56:37 I need wiring advice Sep 11 00:56:46 this is a mess Sep 11 00:56:55 I always struggled with wires when building my computers Sep 11 00:56:59 and this is 100x worse Sep 11 00:57:01 please help Sep 11 00:57:38 https://imagebucket.io/uploads/view/5kzzp/ Sep 11 00:57:54 I need a way to cut down on the wires or lengths but I am not sure how do do so Sep 11 00:58:10 I needed a level shifter a ttl-usb chip Sep 11 00:58:20 load cell digitizer Sep 11 00:58:23 encoder Sep 11 00:58:32 all require like 4 or 5 wires Sep 11 00:58:40 how can I make this less messy Sep 11 01:15:14 ribbon cable Sep 11 01:16:02 oh thats much large wire than i thought Sep 11 01:19:29 are they drawing much current do the wires need to be that large Sep 11 01:19:51 MattB0000ne: ^ **** BEGIN LOGGING AT Fri Sep 11 02:54:30 2020 Sep 11 02:54:55 zip ties would help alot Sep 11 02:55:09 if nothing else Sep 11 02:55:10 true Sep 11 02:55:27 i may try nailing components to a board or something Sep 11 02:55:40 not pretty Sep 11 02:56:21 https://www.amazon.com/Alex-Tech-25ft-Protector-Sleeving/dp/B07FW3MKGH Sep 11 02:56:26 or something like that Sep 11 02:57:03 that would help Sep 11 02:59:24 you can get spiraly stuff at like walmart **** ENDING LOGGING AT Fri Sep 11 02:59:57 2020