#qi-hardware IRC log for Friday, 2010-10-22

nitin_guptahi xiangfu00:55
nitin_guptahow much memory SD card can be supported by NN ?00:55
wpwrakkristianpaul, rafa: bbl from ben-blinkenlights ?01:57
wolfspraulwpwrak: I'm trying to standardize schematics values in the jtag-serial cable a little02:02
wolfspraulI'm not so sure though I get it all right :-)02:03
wpwraklets see ...02:03
wpwrakyou're using 47p but 4.7uF. gta02-core style would be 47p. i have to admit that i'm torn between the two myself. i tend to prefer the ...F variant. that's also i used in (all ?) my schematics on qi-hw.02:07
wpwrakand your multiplier for resistors is upper-cased, Kelvin instead of kilo :)02:09
nitin_guptahow much memory SD card can be supported by NN ?02:09
wpwrakwould be nice if the filters had some sort of value/specification02:10
qi-bot[commit] Wolfgang Spraul: tried to standardize components values in schematics http://qi-hw.com/p/mmone-jtag-serial-cable/39ebc8402:10
wpwraknitin_gupta: i think linux is happy with a big a card as you can get, but the boot loader may start to exhibit problems if you want to boot from anything larger than ~4 GB02:10
wolfspraulwpwrak: I'm already on some of those things, following what I see in other projects of yours.02:11
wolfspraulto it's 27pF now, and 10kR02:11
wpwrakthen, gta02-style would be 2k2 instead of 2.2k02:11
wolfspraulyes I know02:12
wpwrakah, i find the 10kR a little ugly :)02:12
wolfsprauldidn't you just write yourself that it should be like that?02:12
wpwrakfor C and L. for R, I stick with the gta02-core style :)02:13
wolfsprauldo the leds need a voltage?02:13
wolfspraullet's just settle on something for now, even if we change again later02:13
wolfspraulin CHARACTERISTICS, it says to always use R for Ohm02:13
wpwrak(leds) first, i would give them a color :) voltage can be nice to have, too, but it's not so critical02:14
wpwrak(CHAR...) i think that refers to the internal convention inside BOOM. there you also have dots, not 2k2 and such02:14
wolfspraulwhat should we do now in the schematics?02:15
wolfspraul10kR = ?02:15
wpwrak"internal convention" = what the .chr files have and what .sub eventually generates02:15
wpwrakjust "10k" ?02:15
wolfspraulno R?02:15
wpwrakno R02:15
nitin_guptathat means I can extend the SD card upto 4GB only02:15
wolfspraulalso for all other resistors, leave out the R?02:16
wpwrakexcept if you need it as a decimal point. 1R5 or such. (not the case here, i think)02:16
wolfspraulR2 is 2.2k02:17
wolfspraulso you would say 2k2?02:17
wpwraknitin_gupta: if it has to be bootable now. if you can boot from NAND for now, then your SD can be as big as you wish02:17
wpwrak2k2, correct02:17
nitin_guptawell SD card will be used for storage only02:18
wpwrakthen, gta02-core style would be to normalize things like 0.1uF -> 100nF02:18
nitin_guptaNAND will be used for booting NN02:18
wolfspraulwpwrak: I just did that (see my last commit)02:18
wpwraknitin_gupta: then i don't know of any limit02:19
wolfspraulwhat additional values would be good to specify for the filters and the crystal?02:19
wolfspraulwhat is the default precision/tolerance for capacitors, resistors, if nothing is said (like 1% or so)02:19
nitin_guptaok fine02:20
wpwrakresistors: 5%02:20
nitin_guptahow can I stop the GUI to pop up on booting NN02:20
wpwrakcaps: something like 10% i think.02:21
wpwrak(one of my screens is still trying to turn on ... backlight is dying :-( )02:21
wolfspraulmaybe we should specify a default in the boom documentation somewhere02:21
wpwrakthe default is a convention you set in the .sub :)02:22
wolfspraulwhich values should typically be given in schematics, and what typical defaults would be02:22
wolfspraulyes sure, the system is good/flexible02:22
wolfspraulbut a good 'default' can still help people02:22
wolfspraulI noticed a weird thing in eeschema - it just created the -cache.lib file at some point, and added it to the LIBS in the .sch file02:23
wpwrakin fact, with caps, a more common way of specifying their characteristics is to name the dielectric02:23
wolfsprauldon't know what triggered the creation of the -cache.lib file02:23
wpwrakfor example, X5R is a common choice02:23
wolfspraulit included copies of the components used in the schematics02:23
wolfspraulthat's very annoying of course02:23
wpwrakyeah, it does that ...02:23
wolfspraulbecause now that -cache.lib file is linked from the .sch, and the .sch with the -cache.lib pointer will be committed02:24
wpwrakschhist will weed it out :)02:24
wolfspraulso either you always commit the -cache.lib with it, creating redundancy and confusion, or you need to manually rip it out before commits02:24
wpwraknot sure myself when it does this. did you add a library through the GUI ?02:24
wolfspraulat the very least a cache should be transparent and not have a pointer inserted to it in the .sch file02:25
wolfsprauloh no02:25
wolfspraulonly fixing those values02:25
wpwrakhmm, odd. maybe some new "feature" then :)02:25
wolfspraulcan you help me identify under-specified components in the schematics?02:27
wpwrakit hasn't done that to me lately. but then, your version is a bit newer than mine02:27
wolfspraulwe already said: led color02:27
wolfspraulnewer? I thought I have exactly your version now, bzr2448 + your patches02:27
wolfspraulwhat's missing for the filters and crystal?02:28
wpwrak(version) ah, okay. i just thought you had something newer02:28
wpwrak(filters) are they beads ?02:28
wpwrakor some other kind of filter ?02:28
wpwrakfor the crystal, the capacitance could be nice. it's a bit of an annoyance parameter, though, because it's not something you design in but something that the component gives you. and you need to adjust the design for it.02:30
wpwrakso if you specify the capacitance, then you02:30
wpwrak're not saying "this should be XXX pF" but you're saying "the part I selected has XXX pF"02:31
qi-bot[commit] Wolfgang Spraul: cleaner resistor values http://qi-hw.com/p/mmone-jtag-serial-cable/3ce7c3002:32
wpwrakfor the caps, it should be fun to see what happens with them as they are. without specifying anything else, BOOM is allowed to pick the cheapest crap it can find. i'm kinda curious what will turn up. it now has a fairly wide selection of capacitors ;-)02:33
wpwrak(cheapest crap) Z5V and such. see also http://en.wikipedia.org/wiki/EIA_Class_2_dielectric02:33
wpwrakone problem with specifying dielectrics is that the technology changes over time. e.g., you now sometimes see X5R getting replaced with the better X7R.02:35
wpwrakso if you specify X5R, you may exclude some manufacturers or you may make BOOM pick a less common part. there's currently no mechanism for BOOM to automatically accept X7R as a replacement for X5R.02:38
wpwrak(with purely numeric values, you can do this. e.g., you could specify a tolerance of 5% it better, a voltage of 10V or more, etc.)02:39
wpwrak(X5R vs. X7R) it's not a big problem for now, but i think it may get worse with time02:42
wpwrakalso, some of the smaller values are more common as NP0 then X5R. very small ones (just a few pF) only exist as NP002:45
wpwrakso you have to be a bit careful not to over-specify02:46
wpwrakagain, NP0 could replace X5R or X7R, but BOOM doesn't know this02:47
wpwraknot quite sure how to teach Boom such tricks. hmm, should i call is BOOM or Boom ? :)02:48
wolfspraulsame to me. maybe I like Boom a bit better, in many cases I would lowercase it anyway02:51
wolfspraulI will follow whatever the official name is :-)02:52
wpwrakin the documentation, i carefully avoided calling it anything, but "BOM processor" or "BOM processing system" is a bit long :)02:52
qi-bot[commit] kyak: qijoe is now known as joe-full http://qi-hw.com/p/openwrt-xburst/f2065c402:54
wolfspraulwpwrak: what's wrong with boom?02:59
wpwrakhmm, it's a real word03:00
wpwrakalbeit not a very common one03:00
wpwrakwolfspraul: btw, the 47nF/250V syntax is something the current .subs don't recognize. they look for a the value to just say 47nF and some other field to say 250V. supporting 47nF/250V is left as an exercise to the reader ;-)03:06
qi-bot[commit] Xiangfu Liu: cleanup scripts/build a little. make it save un-commit stuff by git stash http://qi-hw.com/p/openwrt-xburst/1ceee3903:12
qi-bot[commit] Werner Almesberger: Include gpio-s3c24xx.h in gpio-s3c24xx.c http://qi-hw.com/p/f32xbase/df9fb5c03:42
qi-bot[commit] Werner Almesberger: c2usb/cam/pcb.pl: fix bogus tool parameters http://qi-hw.com/p/f32xbase/b1d1fcc03:42
qi-bot[commit] Werner Almesberger: Small board adjustments, mainly to improve room for isolation. http://qi-hw.com/p/f32xbase/8325a0603:42
wolfspraulwpwrak: in line with your other names maybe just plain bomproc?04:07
xiangfuwolfspraul: Hi05:15
wolfspraulxiangfu: hi05:17
xiangfuI added one line to Community news, something about "gcc-mips" package.05:17
xiangfuanother things is can you check your 8GB 16Gb sd card (which is not working) partitions tables05:18
wolfspraulit's a wiki, everybody can (and should) edit... :-) so that's cool05:18
xiangfuwhat is the "ID" of the partition type.05:18
wolfspraulok I will check, give me a bit of time then I'll let you know05:18
wolfspraulI probably formatted them with fdisk and set the 'type' (if that's what you mean) to 83 or so (it says "Linux")05:19
wolfspraulI'll check05:19
xiangfuwolfspraul: the last build with config.full_system. make rootfs.ubi goto 282M.05:21
kyak:) nice05:22
xiangfukyak: oh. you online :)05:22
kyaki am!05:23
xiangfukyak: the whole "bin/xburst" is 1G .05:23
kyakwell, there is SDK there and toolchain...05:24
kyakxiangfu: btw, have you tried "fbi"?05:24
kyakit seems more mature than imgv05:24
kyakthough latest imgv has added some new new functions05:25
xiangfunot yet.05:25
xiangfuxiangfu: will try it later.05:26
kyakno problem, whenever it's convenient for you ;)05:26
qi-bot[commit] kyak: removed myself from Makefile header http://qi-hw.com/p/openwrt-packages/05e38ea05:35
nitin_guptahi to all i am getting a problem while reflashing the kernel of NN06:09
nitin_guptageting msg "Cant get kernel image"06:09
nitin_guptaplz help06:09
kristianpaul" And in their paper, the fastest BC design"06:58
kristianpaul     that they present runs at 66 MHz06:58
wolfspraulxiangfu: hmm, that's great but I guess it means we need to increase the rootfs to 512 mb then.07:09
wolfspraulI mean the partition.07:10
B_Lizzardlarsc, sorry for the ball busting, did you commit that .config?09:13
B_LizzardAlso, back in 2.6.32, make uImage worked, now for 2.6.36 it doesn't09:14
B_Lizzardgrepping for uImage in arch/mips brings up nothing09:14
larscB_Lizzard: http://qi-hw.com/p/qi-kernel/cbf420a09:15
B_LizzardAre we supposed to make a vmlinux image and then run mkimage on that?09:15
B_LizzardOh, thanks for the defconfig09:16
qi-bot[commit] kyak: nupdf: initial port http://qi-hw.com/p/openwrt-packages/8bcbadf12:18
kristianpaullarsc: had you wrote documentation about about how to build kernel or how qi works with kernel?12:45
B_Lizzardlarsc, 2.6.36 boots, thanks12:45
kristianpaullarsc: i just wanted to copile kernel by hand no openwrt12:46
kristianpaulwiththe last linux kernel12:46
kristianpauland may try understand how make i2c/uSD in SIE may work with linux12:47
kristianpaulso if u wrote up some lines about waht i saidi can be very usefull12:47
kristianpaulnot just for me i think ;)12:47
larsckristianpaul: i usually build my kernel using openwrt12:50
kristianpaullarsc: even lasts hot version?12:50
larscbut it shouldn't be to hard to do it without it. clone the git repo12:50
kristianpaulsure tell me12:50
larscthen setup the CROSS_COMPILE environment variable to your toolchain bin dir12:51
larscrun make nanonote_defconfig12:51
larscand then run make12:51
larsckristianpaul: i think B_Lizzard just build a kernel without openwrt, so he might be able to help you12:52
B_LizzardYeah, I made a bb file for OpenEmbedded too12:53
B_LizzardSpeaking of which, I saw CONFIG_SOUND is off12:53
wpwraklarsc: no ARCH needed anymore ?12:53
B_LizzardI suspect it hasn't been merged?12:53
B_LizzardAlso, the keymap is borked, letters work but the rest doesn't (backspace, enter, shift etc)12:54
B_LizzardI'm not sure if these things are useful to you12:54
B_LizzardIf these things are my problem, tell me and I'll shut up.12:54
larscwpwrak: yes, sorry forgot about it12:54
larsckristianpaul: you'll also have to set ARCH=mips12:54
larscB_Lizzard: sound is off because we build the offical images with sound support as a modules12:55
B_LizzardShouldn't it be CONFIG_SOUND=m then?12:55
larschm, i guess yes12:56
larscregarding the keyboard: do you use the git from kernel.org?12:57
B_Lizzardkristoffer's clone, but yeah12:58
B_LizzardThe branches are synched12:58
larscyou'll need http://qi-hw.com/p/qi-kernel/5dd125312:58
B_LizzardAh, I see.12:58
B_LizzardDidn't make it in?12:59
B_LizzardIs there anything else that didn't make mainline?13:02
wpwrakah, and will the default target be right ? i.e., uImage ? (i shamefacedly have to admit that i have yet to build a kernel for the ben)13:02
larscB_Lizzard: http://projects.qi-hardware.com/index.php/p/qi-kernel/source/changes/jz-2.6.36/13:03
larscwpwrak: you'll have to create the uImage manually by using mkimage13:04
B_LizzardThanks larsc13:06
wpwraklarsc: ah, so still the same. thanks.13:06
kristianpaullarsc: thanks13:06
wpwrakif anyone is buindling a host toolchain, mkimage would make a nice addition13:07
wpwrakbundling even13:07
kristianpaullarsc: Talking about qi, what was made in linux to make Ben works?13:07
larsckristianpaul: well, we added drivers for all the different JZ4740 SoC components used on the ben13:12
larscand a driver for the backlight13:12
larscand few other peripherals13:12
kristianpaulwich ones?13:12
larscwhat do you mean?13:13
kristianpauldo you have the exact list of drivers created for the Ben hardware?13:17
kristianpaulno ather if each driver dont have documentation (tought)13:18
larsckristianpaul: http://kernelnewbies.org/Linux_2_6_36-DriversArch#head-f467e2a7fd169ab68f05a6cd5ac06c3165ea983d13:18
larscthe ben nanonote support was 1% of the changes in the 2.6.36 release13:22
wpwrakpretty big :)13:29
larscyup. was more then what canonical or nokia contributed.13:30
larscbut of course those companies contribute that much each kernel release and this was a onetime codedrop13:35
wpwrakseems that we have to hurry the ya along, so that we can keep on winning ;-)13:35
larscwell the milkymist kernel support should go upstream too13:38
wpwrakah yes, that ought to be a real biggie13:39
wpwrakeven a new architecture :)13:39
qi-bot[commit] Werner Almesberger: Cleaned up command-line parsing. Added option -n to disable target power. http://qi-hw.com/p/f32xbase/e614d1513:51
qi-bot[commit] Werner Almesberger: Changed default target to "ben" and added Ben upload. http://qi-hw.com/p/f32xbase/13aa36b13:51
qbjectzear: does your suggestion that I should play nethack-newt on jlime mean that you14:12
B_Lizzardlarsc, are you interested in feedback, then?14:12
qbject're not going to bother making it go on OWrt?14:13
B_LizzardI "ported" that, I can give you the source if you want14:13
B_LizzardSDL is dynamically linked, but it should run on OpenWRT14:13
B_Lizzardhahaha, it beeps!14:14
B_LizzardIn any case, the screen colors are completely borked14:14
B_LizzardI like the terminal beep, though.14:15
B_LizzardNice touch14:15
qbjectB_Lizzard: I tried running the mipsel.ipk you posted in here a few days ago and got :Packages ... found, but incompatible with the architectures configured.14:16
B_LizzardMaybe you can try extracting it without opkg14:16
B_LizzardYou know, the directory structrure14:17
rafaB_Lizzard: maybe CONFIG_SOUND=y is okey.. you can have the alsa drivers as modules after that.14:18
B_LizzardTell larsc14:19
B_LizzardIt doesn't matter to me.14:19
qbjectB_Lizzard: what type of compression am I trying to unpack?14:23
B_LizzardI think ipks are unpacked with ar14:23
B_LizzardSo, larsc, some things I noticed: No output on the screen other than the initial kernel messages. Not sure if that's normal. Some things under /sys do not exist, namely /sys/class/power_supply.14:29
B_LizzardChecking out the screen color issue.14:29
B_LizzardWell, that spi_3wires thing didn't fix the color issue.14:32
B_LizzardI suspect you could add the whole CONFIG_SOUND thing too, in case we're counting.14:33
larscB_Lizzard: for the color issue you need both the 3 wire patch and the display driver patch14:42
B_LizzardAh, sorry bout that14:42
larscit's strange thought that power_supply is missing from /sys14:42
B_LizzardMaybe I'm missing some patch or something14:44
B_LizzardWould all the patches in http://projects.qi-hardware.com/index.php/p/qi-kernel/source/changes/jz-2.6.36/ need to be applied?14:44
B_LizzardIs everything listed there an addition to mainline?14:44
B_LizzardCause some stuff says 5 months old and whatnot14:45
B_LizzardI didn't add the gpio-charger thing, maybe that's what's wrong?14:46
B_LizzardSorry for all the hand-holding, I just don't want to apply stuff I don't need14:48
larscwell, you don't need every patch. but i would recommend them for a fully functional system14:53
B_LizzardAh, so even though the jz4740 UDC thing is over 5 months old it's not in?14:56
larscthe udc driver is in a bad shape. not ready for mainline merging14:57
B_LizzardOK, so I'll add the cache quirks thing, the gpio-charger thing, the i2c thing and the dma thing15:02
larscskip the dma and the i2c patches15:03
kyaklarsc: do you have an idea why "red arrow + BackSpace" don't act like "Delete" on Ben (as it should)?15:08
B_LizzardYou can fix that with xmodmap15:11
B_LizzardUm, under X that is15:11
larsckyak: i have no idea15:11
kyaki mean in terminal15:11
kyakwell, ok...15:12
kyaki think, target/linux/xburst/patches-2.6.32/500-modifier-keys.patch, line 31, should be "+       altgr   keycode  14 = Remove"15:29
kyakthis way it works15:29
larsccan you make a patch out of it and send it to the mailinglist?15:30
kyaki can commit to openwrt-xburst if it's ok with you15:30
larscthats ok too15:32
kyakjust going to re-build and run the final check...15:32
B_Lizzardlarsc, last question, I swear :D15:44
B_LizzardWould copying the relevant CONFIG_SOUND stuff from the 2.6.32 config work for 2.6.36?15:45
B_LizzardOr were there driver changes/15:45
larsci think the codec driver config symbol got renamed from JZCODEC TO JZ474_CODEC15:46
B_LizzardLCD fixed, thanks15:47
B_LizzardNo battery under power_supply though15:47
B_LizzardI see a new entry, usb, though.15:47
larschm, for some reason i had "CONFIG_BATTERY_JZ4740 is not set" in the defconfig15:49
larsci'll fix it15:49
B_LizzardAdd the sound stuff too, if you can please15:49
B_LizzardElse I can do it and give you a diff15:49
larsci'll do it15:50
B_LizzardYou are a very helpful and kind person, thanks.15:50
larscbut it might take util tomorrow, i'm rather busy right now15:50
B_LizzardUsually people just tell me to fuck off15:51
B_LizzardNo problem, thanks again.15:51
kyaklarsc: what's the reason for build_dir/linux-xburst_qi_lb60/linux- i think it overwrites defkeymap.c instead of creating it from defkeymap.map15:53
kyakoh, i mean target/linux/xburst/files-2.6.32/drivers/char/defkeymap.c_shipped15:54
larsckyak: iirc there is some makefile rule which generates defkeymap.c from defkeymap.c_shipped15:57
kyakthese files are the same..15:58
kyakbut ok, i will just modify defkeymap.c_shipped, too15:58
zearqbject, about me porting to owrt. Never got owrt toolchain to work properly. The fact it's divided into two different dirs makes the whole porting process complicated15:58
zearbecause you have to modify Makefiles, set custom lib/include pathes, etc15:58
zeari just gave up15:58
zearand as for jlime, nanonote is not the first device i use it on15:59
zearso i'm more familiar with it15:59
larsckyak: yes. they are both the same because during the build defkeymap.c_shipped is copied to defkeymap.c. So if you modify defkeymap.c your change will be overwritten during the build process16:00
zearand just the fact they have a standalone toolchain i can just put on my hdd and use makes it more attractive16:00
larscyou should only modify defkeymap.c_shipped16:00
kyaklarsc: i'll modify defkeymap.c_shipped and the 500-modifier-keys.patch (though the latter seems pointless, but this way it at least reflects the changes in defkeymap.c_shipped)16:01
qbjectzear: thanks for filling me in. I can see why you'd choose that path.16:04
larsckyak: hm, i think if 500-modifier-keys.patch does nothing put patching defkeymap.c it can be deleted16:05
kyakseems so..16:06
zearqbject, before i was just using dingux toolchain and statically linking the binaries against it16:07
zearso it was working in owrt16:07
zearbut that wasn't too sane16:07
kyaki'd leave it though, it makes keymap more obvious, looking into C-file is not very helpfull16:07
zearqbject, well, current jlime toolchain is still somewhat buggy, as in, it does not work for me, but it works for rafa. I'd say it still has some pathes that point to his system if i didn't know him better. The thing is it works for him, it doesn't for me16:11
zearso i usually work on the port, and once it's done, i send him the code so he can compile it16:11
zearwe've ported commander genius like that today16:11
zeari still haven't tested it on my machine16:11
kyaklarsc: and it works! :)16:13
larsckyak: nice16:13
qi-bot[commit] kyak: Red arrow+BackSpace now indeed works as Delete http://qi-hw.com/p/openwrt-xburst/2f3b19816:14
qi-bot[commit] kyak: ben-cyrillic: keymap changed for altgr+bksp to act as Delete http://qi-hw.com/p/openwrt-packages/446437616:21
kyakmirko: hi, good to see you again!16:23
kyakmirko: is it possible to push upstream vim/Makefile that has "--enable-multibyte" instead of "--disable-multibyte"?16:25
kyakmirko: this is the only way to make utf-8 work in vim, and it doesn't affect non-utf-8 users :)16:26
kyakalso, i assume that those who want minimal vi, are satisfied with the one provided by busybox.. when they install vim-full they should expect multibye support16:27
qbjectzear: why do you say that your static-linking method wasn't sane?16:31
larschm, is qi-hw.com down?16:39
Ornotermeslarsc: works fine for me16:39
qbjectlarsc: fine here as well.16:42
larschm. something is wrong with my internet connection... this connection is still alive, but i can't seem to open new ones16:43
wpwraklarsc: NAT connection tracking table full ?16:44
larscwell, no idea. i've only got a e-mail client, webbrowser and an ssh connection open. but anyway, it seems to work again16:46
wpwrak/proc/sys/net/netfilter/nf_conntrack_count has the number of active entries, /proc/sys/net/netfilter/nf_conntrack_max the limit16:53
wpwrakif you don't have a donkey or torrent, and no neighbour borrowing your wifi, conntrack seems unlikely, though16:55
mirkokyak: should be, yeah17:03
mirkokyak: will take a look and ask the previous/initial committer17:03
kyakmirko: thanks!17:03
kristianpaulwow i was follwing this project http://www.funcubedongle.com/?p=6917:08
kristianpauland realize qi planet too17:08
kristianpaulany one around related to funcube?17:08
mirkokyak: http://pastebin.com/H8VJ36Mb - would that be okay for you (untested)?17:09
kyakmirko: looks very fine!17:11
mirkokyak: connection got interrupted...17:18
kyak< kyak> mirko: looks very fine!17:18
mirkohttp://www.funcubedongle.com/?p=69 - would that be of your intention (untested)?17:18
mirkokyak: tell me if it works as expected and i'll commit it at once17:19
kyakok, give me a minute17:19
mirkokyak: the patch may got scrambled in regard of tabs/spaces17:20
mirkokyak: http://downloads.qi-hardware.com/people/mirko/vim.patch17:21
rafamirko: hey man, long time.. how is the life?17:22
kyakmirko: ok!17:22
mirkorafa: fine - was in holiday and good ill afterwards, so i got a bit busy with non-technical stuff :)17:23
rafamirko: well... maybe you re charged batteries ;)17:27
mirkorafa: hope so :)17:28
rafamirko: haha.. well, no a lot of changes on software side I guess, just a lot of nice work on kernel from lars, but no from main rootfs I would say. Well, you know better than me surely. So it is a great that you are around ;)17:30
kyakmirko: tested the patch with vim-full, works as expected!17:31
mirkohehe.. spend some time on buildroot changes, the picture frame - indeed there's still lot's of work to be done regarding the actual nn rootfs17:31
mirkokyak: great17:32
kyakmirko: should i test vim or it's good to commit?17:37
mirkoyou just compiled yet?17:37
mirkoshould be enough, change is minor17:38
kyakmirko: i've built vim-full and run it17:39
mirkokyak: ok17:41
mirkokyak: that's fine by me17:41
kyakgreat, thanks!17:42
zeardamn it, something's wrong with my nanonote17:54
zearwhen i move the screen part, the colors on the screen change17:54
zearrevert like17:54
zearmust be something with the lcd cable17:54
qbjectzear: sorry to hear that./18:02
qbjectzear: why do you say that your static-linking method wasn't sane?18:03
zearqbject, because it means it uses the libs it's linked with, not the system ones18:03
zearwhen it uses system libs, you can update them18:03
zearwith static linking it's binded to the ones it's linked with, if they're old/buggy you won't be able to use different ones18:04
qbjectzear: http://sta.li/faq18:10
t_s_ohttp://en.akihabaranews.com/67208/laptops/king-jim-pimp-its-pomera-dm11g-with-three-gundam-inspired-design <- this with a full size usb host port would make for a interesting nanonote design ;)20:03
wolfspraulwpwrak: if the current design (mmone-jtag-serial-cable) uses a crystal with 12pF load capacitance, can I switch to one with 8pF ? or do I need to adjust some other values in the design too?21:34
wolfspraulI find crystals with 8 pF or 18 pF, but the only one with 12 pF is from EPSON and quite expensive21:35
wpwrakingeneral, you should be able to do this. let's see what the ftdi data sheet has to say about it ...21:38
wpwrakc1 and c3 look a little suspicious for 12 pF21:40
wpwrakin general, higher capacitance means more stable but also more power-hungry. in this case, you probably prefer stability. so all other things being similar, 18 pF is better than 8 pF21:44
wpwrakfor C1 and C3, i'd suggest to make them twice the crystal's value specified for the load capacitance21:44
wpwrak(round to the nearest E12 value, slightly favouring rounding down over rounding up)21:45
wpwrak2 * 18 pF = 36 pF may not be a common value. so 33 pF then21:46
wpwrakif you pick XTAL = 8 pF, that would be 15 pF for the caps21:47
wolfspraulYanjun Luo gave me a formula (was just looking it up), but I have no idea how to apply it :-) (cl=cg*cd/(cg+cd)+a)21:51
wolfspraulso if I adjust C1 and C3 in the way you describe, I can pick any any load capacitance for the crystal?21:52
wolfspraulwhat are the different variables in the formula? cl, cg, cd and a21:53
wpwrakgood question :) only C_L is more or less a "standard" name (using TeX syntax here)21:54
wpwrakhere we are with a formula with explanation: http://www.oscilent.com/spec_pages/PNDescrpt/Load_Cap.htm21:55
wpwrakyou should end up with my formula of C1, C3 = Cxtal*2-epsilon :)21:57
wolfspraulmaybe cl = load capacitance, cg and cd are c1 and c3 in our design22:00
wpwrakand a is the stray capacitance22:01
Action: kristianpaul always uses 22pf capacitors22:10
wpwrakgetting the caps wrong makes your crystal run fast or slow22:14
kristianpauljust some times is hard find the right cap :/22:16
wpwraksee also http://projects.qi-hardware.com/index.php/p/ben-wpan/source/tree/master/ecn/ecn0002.txt22:17
wpwrakyeah, you can only get within a few pF22:17
wolfspraulwpwrak: how do you feel about the BOOKSHELF file? (looking at the one in ben-wpan)22:22
wpwrakhmm, good in general :)22:24
wolfspraulwith that kind of confidence, already from the author, I doubt the success story will go very far ;-)22:27
kristianpaullol :)22:27
wolfspraulseems there are three fields, N: A: D:22:28
wolfspraulwhat are the aliases (A:) used for?22:28
wolfspraulit the idea that the N: field should match the bom reference in the kicad schematics?22:29
wpwrakN is the "main" name of the data sheet. A are convenience names. e.g., N could be the application note (AN1234) and A could be what it is, e.g., mcu (not necessarily a great name, but ...)22:30
wpwrakthe comonent link is in components/INFO :)22:31
wolfspraulcan you tell us a bit more about components/INFO :-)22:35
wolfspraulthere's another 3 fields there, S:, M:, N:22:35
wpwrakyup. S is the symbol name, as used in eeschema22:36
wolfspraulis that the same as the 'reference' in the bom .lst eeschema generates?22:36
wpwrakN links to the data sheet. if S already has the right name, you don't need N22:37
wpwrakhmm no, the only reference eeschema generates is Rxxx, Uxxx, etc. and the footprint. but that's modules/*22:38
wolfspraulhmm, yes. just realized :-)22:38
wolfspraulso how can you link the bom to the datasheets?22:38
wolfspraulnot at all I guess. good thing KiCad files are text format...22:39
wpwrakyou'd have to go search for the link in the .sch file(s)22:39
wpwrakor you link via the component that's selected at the end. eventually, we should have a link manufacturer-part/orderable-part -> data sheet as well22:40
wpwrakbut that doesn't exist yet22:40
wolfspraultrue actually, the datasheet is for an orderable part, not for something that is only spec'd in the schematics22:42
wolfspraulbut then components/INFO is even in the wrong folder, it should be in bom/22:43
wolfspraulyou need another extension in boom, .dsl data sheet links22:44
wolfspraulnamespace part-number url22:44
wpwraknaw, many parts have a unique schematic symbol. that's what components/INFO is about22:44
wpwrakyeah, something like this :)22:45
wolfspraulbut that symbol, no matter how unique it is, has to be mapped to an orderable part at some point, via .sub etc.22:45
wolfspraulso components/INFO looks a little parallel/redundant to boom22:46
wolfspraulit establishes a relationship between schematic symbol and manufacturer part-number / datasheet22:46
wpwrakther emay be some redundancy, yes22:47
wpwrakcomponents/INFO basically tells you what information to use for the design22:47
wpwrakat that time you don't have to know any orderable part22:47
wolfspraulyou are jumping ahead, making assumptions about a specific manufacturer's part-number that may not even be orderable22:48
wolfspraulso either your design is actually tied to that manufacturer's part number, or you shouldn't be making those assumptions22:48
wpwrakwell, for many parts this is the case. e.g., you pick the jz4720 even if you don't know yet where you'll buy it22:50
wolfspraulof course. so the design is tied to that manufacturer/part-number.22:50
wolfspraulif you cannot get that part anymore, you cannot manufacture the design.22:50
wpwrakyou may also skip the task of making the link at the BOM end if it's too messy. e.g., if you'd have to find and link the data sheet manually.22:51
wpwrakor some equivalent part. but for that, you also need to compare with the original spec.22:51
wolfspraulI think what components/INFO is doing right now should just be another extension in boom22:51
wpwrakit all depends on what sort of part you have.22:51
wolfsprauland dsv could parse from there22:51
wolfspraulBOOKSHELF could be merged in there as well22:52
wpwrakdunno. i don't want to overload boom. there are some integrity checks that could exist outside all this.22:52
wolfspraulthere's a bit too many names for my taste in BOOKSHELF and components/INFO (S: M: N: A:)22:52
wpwrakthe aliases for the things you can remember :)22:52
wolfspraulcan BOOKSHELF and components/INFO be merged?22:54
wpwraknot cleanly. there's stuff in BOOKSHELF that doesn't have a symbol. there are symbols that don't have a data sheet. also, footprints can point to data sheets.22:56
wolfspraulyou could use KiCad user fields for the data that is currently in components/INFO22:57
wolfspraulwho uses components/INFO?22:58
wpwrakthe INFO files are currently just accessed manually22:59
wolfspraulit seems to me that the comment lines in components/INFO actually carry a lot of information too22:59
wpwrakyes, i could hide a great many things in kicad. but why ? :)22:59
wpwrakindeed, the comments are useful :)23:00
wpwrakthere's a lot of stuff that doesn't have a "nice" place to go23:00
wolfspraulno reason, I just ping your brain to understand design decisions23:00
wolfspraulof course the text file is much easier to edit23:00
wpwrakand you can read it easily, too23:00
wolfspraulbut sometimes systems can be over-engineered, and cause more harm than good23:00
wpwrakthat's why i keep things nicely separated :)23:00
wolfspraulI keep this big 'free hardware design style guide' in my mind23:00
wolfspraulwe need to be able to describe how it all fits together, one day23:00
wolfspraulso nobody uses the fact that the S: field in components/INFO matches the KiCad schematics symbols?23:01
wolfsprauland if you edit the schematics symbol, you have to edit components/INFO manually to keep them in sync?23:02
wolfsprauldoes anybody use the 'package marking' M: field?23:02
wolfspraulor the connection to the data sheet name N: ?23:02
wpwrakas i said, these files are currently only for manual use. i look at them from time to time :)23:03
wpwrakif you rename the symbol, you need to edit INFO as well, correct23:04
wpwraki haven't bothered to maintain the package markings so far. i should, though. things can get confusing.23:04
wolfspraulwhat do you mean with package marking?23:06
wpwrakthe stuff that's printed on the package. e.g., SOTxxx components can have something cryptic like MJ523:08
wpwrakand then you need to decode this to the real part number, which can be totally different. XP1021, for example23:09
wolfspraulwhat is the plan there? you mean write down any text/characters you see on the component?23:29
wolfspraulor focus on some specific meaning, like datecode? (I don't think there is a datecode formatting standard, unfortunately)23:29
wolfsprauland why would you want to write this down? (into components/INFO) - when are you using it and for what reason?23:29
wpwraki would write down characters found in the data sheet23:30
wpwrakthe idea is to be able to identify items quickly. you may have many with identical or similar packages.23:31
wolfspraulyou mean you look at an image in the datasheet, and what is written on the chip there?23:31
wolfsprauldo you have a URL as an example?23:31
wpwrakthis would have M: 7M23:33
wolfsprauldo most datasheets have such a 'marking symbol' documented?23:36
wolfspraulit means something that is printed (visible) on the actual component, right?23:36
wolfspraulso it helps when you are dealing with the actual physical components in front of you, say you are fixing or manually mounting something, correct?23:37
wpwrakthey usually do, yes23:37
wpwrakcorrect. it's laser-printed on the component.23:38
wpwrakand yes, it's for finding/identifying a real comonent23:38
wolfspraulso you open components/INFO, and you have that one piece of information extracted there, so you don't have to go through datasheets...23:38
wolfspraulok I got it23:38
kristianpaulwpwrak: do you remener if TP24 is avaliable for use23:39
wolfspraulok I think I'm clear on BOOKSHELF and components/INFO now, thank you!23:39
wolfspraulhe, speaking about testpoints, could components/INFO also have entries for testpoints? just to document what they are hooked up to?23:40
kristianpauli have in my texpad like free TP35,36,25,7523:41
kristianpaulbut i soldered TP24 too lol23:41
wpwrak(tps) naw, put them elsewhere ;-) components/INFO describes components in general, unrelated to specific projects23:42
kristianpauljust because was next TP25, but i dont see in the schmatics..23:42
wpwrakkristianpaul: so your question is "what does TP24 do ?" ? :)23:43
kristianpaulwpwrak: right23:43
kristianpauli have my 4 gpio for the GPS EVB board but one pin more is wellcome just i case23:44
kristianpaulactually atena signal detection may be a candidate23:44
kristianpauli really complain kicad psd are not searchable..23:46
kristianpauls/psd/exported data as pdf23:47
wpwrakTP24 is an audio enable signal. POP. page 623:47
wpwrakyeah, non-searchable pdf sucks :-(23:47
kristianpaulhmm better no touch23:48
kristianpaulthanks wpwrak23:48
kristianpauljuan64bits: hey dude :)23:54
juan64bitstell me :)23:54
wolfspraulwpwrak: oh. components/INFO is not meant to be project specific?23:55
wolfspraulwell that's an important piece of information. I would have thought it is.23:55
kristianpauljuan64bits: just to save me some diff reading, wha u did in UNAL with SIE uboot?23:55
kristianpaulyou also get support for uSD and i2c in linux?23:56
wolfspraulin ben-wpan, it says for the antenna "Generic antenna with ground on pin 1, feed on pin 2". Isn't that project specific?23:56
wolfspraulfor the crystal, it says "Fairly generic crystal package with 4 pins (crystal connects pins 1 and 3, the rest is ground"23:57
juan64bitsin uboot, only uSD23:57
kristianpauland you verifed you can boot by uSD too?23:57
juan64bitsmmm no, we just did reding test23:59
--- Sat Oct 23 201000:00

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