**** BEGIN LOGGING AT Sun Oct 14 03:00:00 2018 Oct 14 06:13:13 Boot from SD is not a reliable mechanism - given a valid boot SD card? Oct 14 06:13:22 really? Oct 14 06:14:09 I thought it boots from SD first...then if no SD it continues on the flash. Oct 14 06:14:32 with boot from SD being reliable, can count on it every time. Oct 14 06:15:15 Cuz that allows for 2 different ways to boot. From SD or flash. Oct 14 06:16:00 Snert: ? no, the problem is that bootrom prefers to load u-boot from eMMC over sd card, even though u-boot subsequently prefers to boot linux from sd card over eMMC Oct 14 06:16:45 if the u-boot on eMMC is significantly older than the system on sd card, it may not properly understand all options in /boot/uEnv.txt Oct 14 06:16:57 so this results in a system that boots, but not quite right Oct 14 06:17:01 (e.g. no u-boot overlays) Oct 14 06:17:15 added insurance I guess. Oct 14 06:17:56 wiping eMMC, or at least wiping enough of it to make bootrom not recognize it as bootable anymore, makes bootrom load u-boot from sd card instead and fixes the problem Oct 14 08:07:58 I picture a fine working DNS and DHCP service using flash.....but if things go west I can always click in the SD card which has same everything and a copy of all dhcp and dns configs and zones. Oct 14 08:08:04 fallback insurance. Oct 14 08:08:27 so that really is not viable? Oct 14 08:09:51 my versions on SD card as well as flash would be zactly the same. Oct 14 08:27:19 perhaps I'd be better off with a 2nd bbb active as a dns slave. Oct 14 09:22:24 hmm? Oct 14 09:22:42 you can always boot from sd card Oct 14 09:23:34 if sd and eMMC have similar u-boot versions then the problem doesn't exist, or you can temporarily take eMMC out of the boot order (for bootrom) by holding down the S2 button during power-on (not merely reboot) Oct 14 12:37:38 Helloo Oct 14 12:37:59 I'm new to beaglebone black and I need help starting up, can anyone help me? thanks Oct 14 12:45:08 yeah i mean Oct 14 12:45:16 normally this channel is used to answer specific questions Oct 14 12:46:18 (bil) Oct 14 12:47:13 !topic Oct 14 20:08:53 ayjay_t: btw, you could also consider using libinput instead of manually dealing with the evdev devices. I did a little test with it an it's actually quite easy to use, and will get you events from all input devices (just like how input would work in X11 or Wayland) Oct 14 20:10:12 (or technically, all input devices that have appropriate udev properties set and belong to your "seat", which is seat0 by default for all devices) Oct 14 20:51:20 yeah zmatt, and hopefully theres an i2c driver for touch Oct 14 20:51:31 is libinput x or wayland Oct 14 20:52:08 wayland uses libinput for its input event handling Oct 14 20:52:59 x11 has a bunch of input drivers, but one of them is the libinput one Oct 14 21:03:01 in theory it should be able to replace both the x11-input-evdev and x11-input-synaptics drivers, but libinput is relatively new (1.0 in 2015) so I don't know if that's already the default in any distro Oct 14 21:05:26 lol Oct 14 21:05:29 "libinput uses udev tags to determine what a device is. This is a significant difference to the X.Org stack which determines how to deal with a device based on an elaborate set of rules, rules grown over time, matured, but with a slight layer of mould on top by now." Oct 14 21:13:11 im actually gonna start the uxmode code review relatively soon! Oct 14 21:13:22 not that i got done half the htings done this weekend i was supposed to Oct 14 21:13:41 i also have to politely send this guy from the other day who wanted help a denial Oct 14 21:13:58 they tried to outsource some SoC development to a far away place and now they want something done in 2 or 3 days :-/ Oct 14 21:14:38 I'm considering writing a post to the systemd mailing list titled "A lament on sd-event's refcounting" Oct 14 21:15:39 Do stylized comments like that help? I recently saw some stuff like that in the C/C++ RFC or w/e (not sure what it officially is called) for new standards Oct 14 21:15:51 ? Oct 14 21:16:00 but i was having trouble understanding the technical points because it was written in flowery language Oct 14 21:16:04 oh Oct 14 21:16:17 I wasn't planning on flowery language, I just thought the title was funny Oct 14 21:16:30 it is Oct 14 21:16:30 it is a hot mess though Oct 14 21:16:48 ("The hot mess of sd-event's refcounting" is another good candidate) Oct 14 21:17:05 haha yeah Oct 14 21:18:12 ugh actually i still have quite a bit to do today :-( Oct 14 21:29:20 ayjay_t: take a look at this https://github.com/HN-Vignolles/pruss-intermezzo/blob/master/Images/IMG_2117.JPG Oct 14 21:30:02 oh my god post that on the facebook group Oct 14 21:31:05 what even is this Oct 14 21:32:59 8 channel arb. waveform gen (a poor's man one) Oct 14 21:33:21 wow yeah Oct 14 21:33:27 whats the ad chip doin? Oct 14 21:33:51 is it a arb. waveform generator? a filter? Oct 14 21:36:41 ad5676, 8-ch 16-bit dac Oct 14 21:38:53 the *mess* above that chip was a filter once... cheap flux are hard to clean Oct 14 22:00:12 have you ever though about just trying to make your own switcher? Oct 14 22:00:29 i mean, maybe not getting 16-bits out of it, i don't know what type of bandwidth yuo're getting with that chip Oct 14 22:00:32 or how priecy it is Oct 14 22:00:54 but like, just a fast duty-ratio modulated pwm wave through a pi-filter or something? Oct 14 22:01:18 (btw #electronics is great) Oct 14 22:06:01 The performance of that dac is good, and, using one channel, I get 150Ksa/s (not so great considering high end arb.wavegen at 1Gsa/s, but I have 16-bits and high accuracy with a decent Vref) Oct 14 22:50:21 ayjay_t: you really don't want to duty-modulate PWM, the result sucks unless the modulation frequency is negligible compared to the pwm frequency Oct 14 22:51:19 datasheet says it's a "segmented resistor-string dac" Oct 14 22:53:10 the right way to do pwm with a rapidly changing duty cycle is called sigma-delta modulation Oct 14 22:57:24 when you say negligible do you mean like, 1/10 Oct 14 22:58:53 hmm i'm looking at sigma-delta its not too bad Oct 14 23:11:12 no idea what ratio is sufficient, it depends on the amount of distortion you tolerate. easy enough to do an experiment though: take a sine at 1 Hz, sample it at 10 Hz, turn those 10 samples into 10 pwm periods (at some high samplerate corresponding to the pwm resolution), and do an FFT on that Oct 14 23:13:52 high-frequency crap that appears in the spectrum can be filtered out, but any crap below 10 Hz would be a problem Oct 14 23:13:57 below 1 Hz I mean Oct 14 23:14:05 scale parameters to whatever you want of course Oct 14 23:16:46 (of course the conversion isn't linear so you can't conclude it's good enough based on one sine test input) Oct 14 23:31:38 i feel like we're going backwards Oct 14 23:34:05 i mean i can do a hard calculation of what filter quality i need Oct 14 23:34:42 i guess i can just do an fft of the pwm signal itself and see how clean it is Oct 14 23:35:16 my point was, I don't see any reason to expect that the spectrum is going to be clean in the passband, in fact I very much expect it won't be, and no amount of filtering can help with that Oct 14 23:35:36 yeah i hear you Oct 14 23:39:02 it'll be a pretty bad DAC to start, i'm okay with that Oct 14 23:39:11 we'll work on it over the course of a year Oct 14 23:39:19 i'd like to keep the form factors super super small Oct 15 00:13:14 oh I'm an idiot, there can't be any in-band crap, not for a sine whose period is an integral number of pwm periods Oct 15 00:14:53 since the result still a periodic signal (with period equal to the sine frequency), hence it has a fourier series (i.e. discrete spectrum, only energy at integer multiples of the sine) Oct 15 01:16:09 hello..anyone our there had any luck get the the click 5 gnns click board with the MikroCape working on BBB Wireless? Oct 15 01:16:35 as soon as I update the uEnv.txt file to include the overlay the device will not boot at all... Oct 15 01:17:58 I have gone through the steps at https://github.com/beagleboard/bb.org-overlays three times Oct 15 01:37:59 Benavides! Oct 15 01:38:37 Use a new image and try the config-pin utility. Oct 15 01:48:11 is 100 ft. for the BBBW WiFi a possibility? Oct 15 01:48:41 inline sight? Oct 15 01:49:04 I will test it but if you know, please do reply. Oct 15 02:50:50 ayjay_t: lol, it just occurred to me that since the conversion isn't quite linear, using a sine as test signal is a very bad idea. the spectrum doesn't look all that bad when you output a sine signal. on the other hand, if you feed the sum of two sines, helloooooo intermodulation distortion garbage: https://photos.app.goo.gl/9ZY4xxWS1pBLqruP9 (10 Hz sine, 7 Hz sine, combination of both, all with 100 ... Oct 15 02:50:56 ...Hz pwm) Oct 15 02:54:15 actually the output for a sine is pretty bad too... a low-pass filter with sharp cutoff just above 10 Hz would fix the output for 10 Hz, but it wouldn't fix it for lower-frequency signals **** ENDING LOGGING AT Mon Oct 15 03:00:01 2018