#qi-hardware IRC log for Thursday, 2011-10-13

qi-bot[commit] Xiangfu Liu: new package: gtypist a universal typing tutor (master) http://qi-hw.com/p/openwrt-packages/d7402ee02:47
qwebirc53148I am using openwrt in the console.  relative newbie to linux.  How do I tell which keymap is being used?  I am trying to get a backquote into the keymap.03:06
xiangfuqwebirc53148, by default it QWERT as the keyboard shows.03:11
xiangfuqwebirc53148, you can use 'dumpkeys' and 'setkeycodes' to change some keys.03:12
xiangfuqwebirc53148, or 'loadkeys' maybe.03:12
xiangfuqwebirc53148, you can install 'kbd-console-maps' for get more keymaps03:16
qi-bot[commit] Xiangfu Liu: config.full_system: include kbd-console-maps (master) http://qi-hw.com/p/openwrt-packages/ca0b3b503:26
qwebirc53148A week ago the login page gave me an option to pick a nickname.  Now I am just assigned one (qwebirc53148???).   Is it possible to get back to choosing one myself?  Sorry for off topic.  Thanks for any hints./03:50
xiangfuqwebirc53148, /nick NAME (http://freenode.net/faq.shtml#nicksetup)04:04
apinkxiangfu, thanks again04:20
wolfspraulafter over 2 years, NanoNote support has arrived in u-boot upstream!06:07
wolfspraulcongrats xiangfu!06:08
wolfspraulI think 2 drivers are still missing (lcm + mmc), but the first big step is made and the diff and maintenance overhead for upleveling u-boot will go down a lot06:08
wolfspraulthat's 'just' the bootloader, but any diff we get upstream makes me worry a little less about our ability to make cool new devices06:09
qi-bot[commit] Xiangfu Liu: frotz: remove license problem games (master) http://qi-hw.com/p/openwrt-packages/b01315409:10
qi-bot[commit] Xiangfu Liu: some BROKEN packages, maybe somebody can continue work on those package (master) http://qi-hw.com/p/openwrt-packages/3e5e71613:22
qi-bot[commit] Werner Almesberger: m1rc3/norruption/2/collda: corrected $n to be the number of standby failures (master) http://qi-hw.com/p/wernermisc/86e86dc14:09
qi-bot[commit] Werner Almesberger: m1rc3/norruption/2/plot-cdf: start empirical distribution at (0, 0) (master) http://qi-hw.com/p/wernermisc/9b2f2d114:09
qi-bot[commit] Werner Almesberger: m1rc3/norruption/2/aloop: lock the NOR again after restoring standby (master) http://qi-hw.com/p/wernermisc/88bf6b714:09
qi-bot[commit] Xiangfu Liu: new package: SDL-widgets: a GUI toolkit on top of the SDL libraries (master) http://qi-hw.com/p/openwrt-packages/1b3135c14:54
wpwrakwhee ! now let the widgetry begin ! ;-)14:57
kristianpaulncie !15:10
whitequarkyay! my components for lvds-board have arrived15:51
whitequarkI'm about to coin a new term, "yo dawg syndrome".15:51
whitequarkDigiKey have sent me a polyethyelene packet, inside there is a polyether antistatic/moisture package (which is absolutely unopenable), inside is a polyethylene package, inside is a polyethylene package, have I said that inside is a polyethylene package? And inside is a chunk of tape, and finally there is an IC.15:53
whitequarkand they've sent each IC in its own polyether package15:54
whitequarkfor some reason only known to them15:54
whitequarka package fetish maybe.15:54
whitequarkoh. trying to solder some TSSOPs and WSONs revealed a fun fact: I suck at soldering :/22:15
whitequark(also: there is never enough flux)22:15
whitequarkactually: no, I don't. also: I will pray for the inventor of Flux-Off for the rest of my life.22:39
whitequarkif there are still some remaining small sticky solder spheres, you've not used enough of it.22:39
wpwrakwhat type of flux do you use ?23:00
wpwrakwhy doesn't your ultrasound cleaner get rid of it nicely enough ?23:01
whitequarkwpwrak: RMA23:06
whitequarkI've understood several important things, namely:23:07
whitequarkflux, flux-off, pre-heating and ultrasonic cleaning are like violence. if they don't solve some of your problems, you are not using them enough23:07
whitequarkfinally, I've got some sense out of this f***king board23:08
whitequarkit was not SO bad23:08
whitequarkthat was just a whole lot of solder paste remains23:08
whitequarkthey looked awful23:09
wpwraki would go easy on the pre-heating, though. i've had some bad experiences with having too much heat around.23:09
wpwrakhmm, if you solder manually, you should need solder paste over in very few situations. e.g., for crystals.23:09
whitequarkdo I?23:09
whitequarkI've just poured it on every shining area and plucked the components on it23:10
whitequarkalso, flux23:10
wpwrakalso, for the flux, RMA is good but also very dirty. you may find "water soluble" more convenient.23:10
wpwraksolder paste should already include enough flux.23:10
whitequark(rma) flux-off gets rid of it COMPLETELY. the board is not sticky at all and it is clear like a mirror23:10
whitequark(solder paste) I use a Russian product.23:11
whitequarkit does not.23:11
wpwrak(rma) yes, but it's sticky while you're working on it :)23:11
whitequark(rma) I don't care much about that. I can always use more flux-off :D23:11
wpwrakare you sure your solder paste is for electronics ? ;-)23:11
whitequarkthat's written on it23:11
wpwrakmaybe they confused the label :)23:11
wpwraksolder paste without flux doesn't make sense :)23:12
whitequarknope. it's just done in a very Soviet way23:12
whitequarkit has _some_ flux23:12
wpwrakaha !23:12
whitequarkand I guess not the best kind of it23:12
whitequarkmaybe, worst23:12
whitequarkand it tends to split itself into layers23:12
wpwrakmaybe they thought the formula they used to fix ships will do for chips, too ...23:13
whitequarkyou're not very far from the truth23:13
whitequark(solder paste) what's your way of soldering things like tssop48 or 0603 caps?23:13
whitequarkor resistors23:14
wpwraki think you may be better off with just flux and regular solder (preferably 63/37)23:14
whitequarkthe paste is 63/37, btw23:14
whitequarkI guess it's done from the same materials that regular solder with rosin core is23:15
whitequarkit's not very good either23:15
Action: whitequark promises he will never be too greedy and save on solder. NEVER.23:15
wpwrak0402 and bigger: pre-tin the pads (options, but it helps. i do this one the whole board, before soldering any components). apply "water soluble" flux to the pads. place the component. hold it down with angled tweezers. touch a tiny drop of solder on one end. then on the other.23:15
whitequarkI've tried that method23:16
whitequarkthe results are not that different. some things are better, some are worse23:17
wpwraki haven't done tssop yet. ssop yes. there, i pre-tin (again, optional), apply a generous amount of flux on all the pads, place the component, hold it down, solder one or two pins (okay if they're bridged)23:17
whitequarkssop is easy as a pie23:17
wpwrakok :)23:17
whitequarktssop is not very hard too, as it turns out -- just get something to remove the remaining balls23:17
whitequarkbut I have a hirose connector on the board23:18
wpwrakyup. flux and heat. repeat. if it's still messy, flux and solder.23:18
whitequarkthat connector will melt immediately if heated from the component side23:18
wpwraklovely ;-)23:18
whitequarkso I've sit there for 15 minutes and heated the board from underneath it23:18
whitequarkboring as hell, but it works23:18
whitequark... I hope23:18
whitequarkat least it has kinda the same shape23:19
whitequarkmaybe not exactly :D23:19
whitequarkthat's what I've called "pre-heating"23:19
whitequarkI only have a SMD rework hot air gun, and no other smd-rework-related tools. so it's the only option to heat the whole board23:20
whitequarkit's thick. and I forgot the thermal relief pads, as I've said23:20
wpwrakmaybe pick a through-hole connector the next time, if possible23:21
whitequarkthat hirose stuff is a pain. it is rare, it cannot be identified when on board, and prior to this day, I've never managed to (de)solder it successfully23:21
whitequarkthat's not possible23:21
wpwrakmmh. then it's messy. did you pre-tin the pads/traces around the connector ?23:22
whitequarksecond time, yes23:22
qi-botThe build has FAILED, see log here: http://fidelio.qi-hardware.com/~xiangfu/compile-log/openwrt-xburst.full_system-10122011-2220/23:25
whitequarkI've got some images of the result23:27
whitequarkdarn. my hands are trembling :/23:29
wpwrakhmm, i think the #1 problem with the smt connector may have been the big ground plane and its numerous direct connections to it.23:31
wpwrakanyway, does it work ? :)23:31
whitequarkI don't have a host board23:31
wpwrakso you're doing this just for the art. nice ;-)23:32
whitequarkmy friend, who wants this board, has it23:32
whitequarkthis one: http://starterkit.ru/html/index.php?name=shop&op=view&id=4923:33
whitequarkthe board itself is an, ahem23:34
whitequarkI don't even know how to call that23:34
whitequarkit consumes something like 2 to 4A by 3.3V23:34
whitequarkand it is supposed to be powered by 5V23:34
wpwrakand what is the role of your thingy ?23:34
whitequarkguess what: they've put an LDO on it.23:34
whitequarkto make 3.3 from 5V23:34
wpwrakaah, level shifters23:35
whitequarkit is POWERED through an LDO.23:35
whitequarkit gets to full 125 deg. C while the board is idling23:35
whitequarkit heats even more when it calculates something23:35
whitequarkand pcb under it gets black soon23:35
whitequarkwe've mailed the vendor23:35
whitequarkhe said: it is normal condition (sic!)23:36
wpwraklovely ;-)23:36
whitequarkI, well23:36
whitequarkI don't have any words for it.23:36
wpwraknow i know why you need military-grade solder ;-)23:36
whitequarkmy board isn't something like that23:36
whitequarkI at least know that buck converters do really exist in this world23:37
wpwrakbut they're more expensive ;-)23:37
whitequarkyeah, extra $5 for a step-down on a $150 devboard is something unbearable23:38
whitequarklet's make some ARM toasts instead23:38
whitequarkalso: they frequently do not solder some passive components on the board23:39
whitequarkeach time some other ones23:39
whitequarksaying: "we don't have enough ones"23:39
wpwraknice :) russian design is famous for its rugged robustness. i think they make good use of it ;-)23:40
whitequarkthere's an Atmel ARM23:40
whitequarkon the other hand, the very same friend once (I don't know if that was an experiment or he has just missed the voltage. I fear that it's the latter) powered an ATmega8 with 12V23:41
whitequarkfor like a month, continously23:41
whitequarkit worked.23:41
wpwrakMCUs are tough23:41
whitequarkI guess Atmel has a whole lot of Russian engineers23:41
whitequarkto make their products Russian-proof23:42
whitequark(my board) it gets TFT video signal on 40 pin side and outputs an LVDS for one particular notebook LCD on the hirose side23:42
wpwrakaah, that's why you need that connector23:42
wpwrakmaybe try to find a 90 deg version if you need something similar in the future. should be easier to solder.23:43
whitequarkluckily, we found a leaked datasheet on the motherboard23:43
whitequarkit had a partnumber23:43
whitequarkin any other case, I doubt we would be ever able to find the right mating one23:43
whitequarkhirose are professionals of vendor lock-in23:44
whitequarkno, not DigiKey nor Hirose site contain any references to 90deg version23:44
whitequarkI think this particular connector was developed specially for this particular notebook series23:45
wpwraknice :)23:45
whitequark*specifically, or maybe not23:45
wpwrakconnectors are generally messy. very often, there's poor documentation23:45
whitequarkhirose has nice datasheets, but if you only have a part on your hands, it's impossible to figure out which one it is23:46
whitequarkeach notebook vendor has its own variant of that LVDS connector, almost entirely like any other, but not enough to be pluggable23:47
whitequarkand they like to play with the pinout, too23:47
whitequarkand with the order of bits inside of the LVDS stream23:48
whitequarkwithout any reason (except the lock-in) whatsoever23:48
whitequarkI just realized that I can test the buck23:49
whitequarklet's see if it will explode23:49
whitequark*the buck converter23:50
whitequarkit did not23:55
whitequarkdoes not work, either23:55
whitequarkeither I am hallucinating, or I have a voltage of 0.4V between different points of ground plane23:58
whitequarkah, of course it does not work23:59
whitequarkit is not soldered to the board.23:59
--- Fri Oct 14 201100:00

Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!