#qi-hardware IRC log for Friday, 2010-11-12

kyakxiangfu: hi!02:18
xiangfukyak: hi02:19
kyakxiangfu: when things get updated in openwrt backfire, how they get into our repo?02:19
xiangfukyak: merge the upstream backfire branch to our 'master' branch.02:20
kyaki'm asking because there was a very important backport to backfire yesterday, but it's not in our git -\02:20
kyakxiangfu: i see.. so ot's not automatic02:20
xiangfukyak: no.02:20
xiangfukyak: I think it's like this "master" --(cherry-pick commit) --> "tracking_backfire" --(merge or send patch email) --> upstream "backfire" --(merge back to)--> our "master"02:22
xiangfukyak: I will try to merge the upstream backfire to our "master" today (in local)02:23
kyakxiangfu: thanks, i was going to ask you about this, cause my git powers are limited :)02:24
kyakand the patches flow as you explained seem reasonable to me02:25
kyakxiangfu: btw, there was a great finding yesterday.. ncursesw was not working unless locale support is enable in uClibc02:35
kyakafter i enabled locale support, rebuilt all things, ncurses-based programs were finally able to present wide chars!02:36
kyak(this also means that ncursesw package is kind of useless in openwrt trunk, because uClibc has not locale support there)02:37
kyakafter enabling locale support in uClibc i had to drop a patch to fbterm, because calls to iconv are not stubbed anymore02:39
kyakmaybe some other packages are related.. i will build with config.full_system and see02:39
kyakif it works, i think we should enable it.. it is as easy as adding to lines to toolchain/uClibc/config-
xiangfukyak: great.02:47
wolfspraulkyak: wow, very nice find!02:52
qi-bot[commit] jow: [backfire] merge r22279 and r22282 (#6011) http://qi-hw.com/p/openwrt-xburst/49bab2b02:58
qi-bot[commit] nbd: ath9k: merge compile error fix from r23912 http://qi-hw.com/p/openwrt-xburst/9fc27c602:58
qi-bot[commit] hauke: kernel: update from kernel to kernel and refresh patches http://qi-hw.com/p/openwrt-xburst/766ef7702:58
qi-bot[commit] jow: [backfire] merge r23952 http://qi-hw.com/p/openwrt-xburst/6771f8e02:58
qi-bot[commit] jow: [backfire] merge r23959-r23961 http://qi-hw.com/p/openwrt-xburst/cee2a4102:58
qi-bot[commit] mirko: [backfire] backport commit 22668 : http://qi-hw.com/p/openwrt-xburst/1b3cadd02:58
qi-bot[commit] Xiangfu Liu: this patch add uboot-xburst package http://qi-hw.com/p/openwrt-xburst/b51c58b02:58
qi-bot[commit] Xiangfu Liu: some cleanup about xburst target http://qi-hw.com/p/openwrt-xburst/8ab272102:58
qi-bot[commit] Xiangfu Liu: add ks7010 drivers http://qi-hw.com/p/openwrt-xburst/bc937d102:58
qi-bot[commit] Xiangfu Liu: copy data folder from history branch http://qi-hw.com/p/openwrt-xburst/704f20402:58
qi-bot[commit] kyak: gitignore vim temporary files (*~) http://qi-hw.com/p/openwrt-xburst/c84c7d202:58
qi-bot[commit] Xiangfu Liu: add Nanonoe special feeds.conf under data folder http://qi-hw.com/p/openwrt-xburst/c2a3c9b02:58
qi-bot[commit] Xiangfu Liu: [config.full_system] add package byobu http://qi-hw.com/p/openwrt-xburst/6d9d9c802:58
qi-bot[commit] Xiangfu Liu: cleanup the build file http://qi-hw.com/p/openwrt-xburst/0350d2702:58
qi-bot[commit] Mirko Vogt: add <http://downloads.qi-hardware.com/sources> as mirror for source http://qi-hw.com/p/openwrt-xburst/a50d3c502:58
qi-bot[commit] Xiangfu Liu: adjust the rootfs partiton size to 512MB http://qi-hw.com/p/openwrt-xburst/33490c502:58
qi-bot[commit] Mirko Vogt: deselect gnuplot since it does not compile (in our tree) http://qi-hw.com/p/openwrt-xburst/c23914502:58
qi-bot[commit] kyak: NAND DATA Partition offset and reflash_ben.sh fixed according to 512 http://qi-hw.com/p/openwrt-xburst/362da1c02:58
qi-bot[commit] Xiangfu Liu: [config.full_system] add gcc-mips, fixed in openwrt package: 8dbf61c, tested in build hosts http://qi-hw.com/p/openwrt-xburst/32f2e8b02:58
qi-bot[commit] Xiangfu Liu: [linux] support increase/decrease in screen brightness http://qi-hw.com/p/openwrt-xburst/eb8096a02:58
kyakguess it means we have synced with backfire --^ :)03:00
wolfspraulhe :-) script works - patch flood stopped after 20 items... (last time we had > 500 ...)03:00
kyakwolfspraul: yesterday wprak have noticed that there are some packages with mp3 support in my ~/people/kyak directory.. how do you think, is it bad to keep it there?03:02
kyaki removed them for now03:02
wolfspraulkyak: yes please, we have to remove them03:05
wolfspraulno mp3, no mpeg4 or h.26403:05
wolfspraulin openwrt we have this PATENTED flag that you should always turn on when building stuff you are uploading to a Qi server03:05
kyakok, i will be more carefull with that03:06
xiangfukyak: you can try to run "git pull", our commit still on the top.03:08
kyakxiangfu: just did it, going to rebuilt from scratch with full_system03:09
kyakxiangfu: just for my understanding, you saved our commits somewhere, than you checked out the backfire into our master, and then you applied our commits on top?03:12
kyakhow you managed to do this?03:12
xiangfukyak: create a local branch follow the upstream backfire.03:13
xiangfukyak: then 1. checkout to our "master",  2. git rebase backfire03:13
xiangfukyak: then git will take care the merge03:13
kyakwhat would you do if you had to resolve soem conflicts?03:14
kyakif some of our commits would not apply cleanly?03:14
xiangfukyak: when there is conflicts. it's will stop, then wait for your manual resole the merge.03:16
xiangfukyak: if people know there is a bit conflict. usually they will use the "git rebase -i"03:17
xiangfubefore the auto merge start, manually remove some commit. etc..03:18
xiangfukyak: since we just cleanup the 'master' branch. so I just use "git rebase" :)03:19
xiangfuhi kyak have you ever use the StarDict??03:22
kyaki stopped using it at some point when it got broken and then i discovered sdcv03:24
xiangfukyak: ok. I think I will add (English --> Simple Chinese) dictionary package :)03:28
xiangfukyak: now we only have Chinese and English font package. so the stardict display Chinese very well.03:30
xiangfukyak: maybe we can add Russian font and dictionary :)03:30
kyakxiangfu: is a good idea! btw. sdcv is already working correctly english-russian and russian-english both ways :) because cyrillic input is handled correctly in console03:32
xiangfukyak: for you understand the git rebase and branch with different remote URL. here is my git config -l :http://pastebin.ca/198888003:32
kyaki have no idea right now how to implement russian input in gtk app like stardict03:32
kyak(without X server)03:33
xiangfukyak: there is no russian dictionary in openwrt release . right?03:33
xiangfu(I forgot the cyrillic )03:33
kyakxiangfu: i'm still trying to figure out what you said about the git :) i started reading some git reference.. really i'm tired of git being stronger than me :)03:33
kyakxiangfu: there is no russian dictionary right now, i added some manually to my datafs03:34
xiangfukyak: me too.03:35
qi-bot[commit] Xiangfu Liu: [stardict] add Oxford Englihs -> Chinese dictionary http://qi-hw.com/p/openwrt-packages/a508a7504:21
kyakxiangfu: how does stardict know which fonts to use?05:09
kyakfor exmaple, one has WenQuanYi-chinese-font-ttc and unifont05:09
kyakthey both presumably support Chinese05:09
kyakwhich one will be used?05:09
kyakxiangfu: while i can imagine that cyrillic input (as well as accented characters) could be supported by stardict; how it could work for Chinese input?05:21
xiangfukyak:  I don't know how stardict deal with font. I just copy the Chinese font to system font folder. then stardict works fine.05:24
xiangfukyak: but there is one option stardict, we can modify the default font to whatever we want.05:25
xiangfubut I don't know how 'default' works. :(05:26
kyakhm ok, thanks for the hints05:26
xiangfukyak: no Chinese input method for now.05:26
kyakdo you have an idea about how this could be done?05:28
kyaki guess there is some kind of chinese "alphabet".. in this case such input method could work05:32
xiangfukyak: hmm... I know there are some Chinese input method like "SCIM" "IBUS" "FCITX"... we need port one to NanoNote :)05:32
kyakthere is often a problem for me when something other than ascii is needed :) i guess for Chinese this is a double problem05:34
wolfspraulxiangfu: I just saw the inclusion of the 'triggerhappy' hotkey daemon on the openwrt mailing list. can we bring it to the NanoNote? Can it be a replacement for the fn-handle power-off daemon?05:43
xiangfuwolfspraul: I think it's much better then fn-handle.05:49
wolfspraulgreat maybe we try and see whether we can use it as a replacement05:50
xiangfuwolfspraul: yes. I will do that.05:50
wolfspraulif it doesn't get added to any upstream openwrt package repo, we can add it to openwrt-packages.git as an intermediate (but let's way a week or so, I would think it gets included somewhere else if it works well)05:50
wolfsprauls/let's way a week/let's wait a week/05:51
xiangfuwolfspraul: I will wait a little bit. it's not yet commit to upstream yet.05:51
wolfspraulyes, right05:51
xiangfuwolfspraul: oh. yes. same thinking :)05:51
wolfspraulno rush05:51
wolfsprauljust keep an eye on it05:51
kristianpaulwolfspraul: soemthing wrong with qi-hardware.com server?07:43
kristianpaulwolf.. !07:44
kristianpaulah works now :)07:45
drizztwpwrak: I sent the transfer order to my bank07:54
wpwrakkristianpaul: (qihw.com) i've noticed that it occasionally has brief hickups where a http request times out07:54
drizztyou should get it in a few days07:54
kristianpaulwpwrak: seems07:54
wpwrakdrizzt: that was quick. thanks !07:54
drizztyou were as well :)07:55
drizztwolfspraul: wpwrak: I'll try to keep to the specs I chosed, though I may split the design in two parts, with the fix 'dock' part being much more simple than the tablet part07:58
wolfspraulwpwrak: can you describe those hickups? maybe keep an eye on it07:58
wolfspraulI know we can do a lot on server performance, it's just a time sink and hard to accurately track down the bottleneck07:59
drizztthen I may ask you for the design of the dock part07:59
drizztthis should be decided within the next three months07:59
wpwrakwolfspraul: it's just that occasionally an HTTP request times out. the next try works. i don't see it happen like this on other sites, so it doesn't seem to be a local problem.08:00
kristianpaulmay be install a nagios on fidelio wil help tto keep eye08:00
wpwrakkristianpaul: guess he just reproduced the cause ;-)08:01
wolfsprauldrizzt: where do you get the main tablet part from?08:06
drizztI don't know yet08:06
drizztthere's a lot of possibilities08:06
drizztworking with existing tablet maker, designing one, buying an existing one (worst case)08:07
drizztthe best would be the design, to integrate all the wirless interfaces08:08
wpwrakdrizzt: (integrate all the rf) that's when things get complicated again ...08:14
drizztyes, but this is the mandatory part08:14
wpwrakdrizzt: who not just make one ugly grey box with all the interfaces, then buy the sexiest tablet you can find and just communicate with the grey box over wifi ?08:15
drizztthat's the main purpose of the device : allow control of both wired and wireless08:15
wpwrakdrizzt: about half the domotics stuff is wire-bound, so you'll eventually need that box anyway :)08:16
wpwrakdrizzt: better have a sports car and a tractor than a ferrari with a plow ;-)08:17
drizztIf I can i'll choose a ferrary plus a tractor08:18
wpwrakdrizzt: also gives you more flexibility. the tablet can be anything. so you can just visit wolfgang and he can show you the gazillion of generic/clone/whatever tables you find in the chinese markets like potatoes in european markets ;-)08:20
wpwrakdrizzt: then you can patiently evaluate the beast. if you don't like it, there are many alternatives. no need to even solve their problems in case you hit any. just rinse and repeat.08:21
wpwrakdrizzt: for the grey box, pick something that uses simple, proven technology. maybe just some pc platform. or, if you don't like that, and old arm, an xburst, or whatever.08:22
wpwrakdrizzt: the less bleeding edge the technology, the more choices you have for getting it designed and produced.08:23
wpwrakwell, within reason. if you end up with a cpu consisting of vacuum tubes, you've overshot the target :)08:23
wolfsprauldrizzt: so you are interested in a tablet?08:48
wolfspraulany tablet?08:48
kristianpaulHanvon tables? :D08:50
kristianpaulmaybe with a Xburst chip? :O08:50
kristianpaulbbl i'm going to the lab now (finally !)08:51
wolfspraulI've seen a 4760-based tablet, if there is serious interest in it I guess I could try to find out more.08:51
wpwrakkristianpaul: the lab with a scope ? :)08:52
wpwrakwolfspra1l: just had one of those glitches when trying to access http://downloads.qi-hardware.com/people/werner/tmp/09:04
wpwraknow .. we all like out wlans strong and healthy. but this does suck when trying to measure something else in the vicinity ... http://downloads.qi-hardware.com/people/werner/tmp/ant.png09:05
wolfspra1lwpwrak: yes I am aware of glitches/problems, but it's hard to track them down.09:05
wpwrak(this is supposed to be a more or less straight line, with all of the ten runs yielding more or less identical results)09:06
wolfspra1lmaybe first I need to try to make them reproducible, without bringing the server down :-)09:06
wpwrakwolfspra1l: maybe dns ? the error came back very quickly, apparently from some "transparent" proxy09:07
wolfspra1lwpwrak: at least your graphic looks like something IMPORTANT is happening09:07
wpwrakwolfspra1l: yeah. now i have to filter those ~8 dB of junk out of my < 1 dB signal ...09:09
Guest81366i want a question09:33
Guest81366i have buy the hardisk samsung09:33
Guest81366but i see only 2 pins09:34
Guest81366how must do for configuration?09:34
zearGuest81366, how is this related to qi-hardware projects?09:34
wpwrakzear: it's the swap you need for running libreoffice :)09:42
zearand crysis ;)09:42
wpwrakhmm, that would be a nice project indeed. crysis, with a retro mode with ASCII graphics09:44
zearanyway, it's time for me, bbl09:45
wpwrakhmm, to estimate how much power an antenna puts in the air, i want to look at the amplitude in the spectrum close to the carrier (i can get a very narrow carrier)10:04
wpwraknow, what do i do on the FFT side ? my input are Q/I values. should i do a complex fft and then use hypot(Re(x_i), Im(x_i)) for the signal strength ?10:05
wpwrakor maybe do this conversion from complex to real before the FFT and do the FFT on real numbers ?10:06
drizztwpwrak: 12/11/2010      Virement vers l'étranger11:32
drizztc'est passé au niveau de mon interface de gestion de compte :)11:32
kristianpaulwpwrak:  yes !11:42
kristianpauland spectrum analizer11:43
Action: kristianpaul at home11:43
kristianpauli'll upload some pics11:43
kristianpaulnot so good probes... but the lab admin said he will find me some good ones11:46
kristianpauland mostly i having issues to visualize the data output http://en.qi-hardware.com/wiki/File:IMG_3143.JPG11:50
kristianpaulYEs the scope is almost 22pf load plust probes11:51
kristianpaulthe signal got triangled11:51
kristianpauli tested antenna and is detected and current and internal voltage regulator is under spected values11:52
kristianpauli'm so happy :)11:52
kristianpaulso i can witre it to SIE tomrrow11:52
kristianpaulcause i'm sort of delayed i think11:52
Action: kristianpaul listen to kevin jhojansen12:07
kristianpaulgood sync and clock seems look ok12:09
kristianpaulargg data signal still trciky to catch even stoping rendering on scope it looks messy/blurr, lets see how it looks witht the spectrun analizer :)12:17
kyakbartbes: here?12:18
kyakbartbes: in nlove Makefile, PKG_SOURCE_PROTO:=hg - what's this?: )12:19
kristianpaulwpwrak:( maybe do this conversion from complex to real before the FFT and do the FFT on real numbers) totally agreee12:19
kyakthe package can't be downloaded, build fails12:19
kristianpaulwpwrak: i'll do same with gps data before try get FFT and look PRN code matching12:20
kyakbartbes: k, seems i have to install hg-git.. first you made me install cmake, now that :) i guess you just nlove alternatives!12:24
bartbeskyak: ;)12:33
kyakbartbes: you could have a look at include/prereq-build.mk for an example of user-friendly build prerequisites12:38
kyaklike if your build requires svn, it would inform the user about it12:39
bartbeswhat is that?12:43
kyaka curse word12:43
kyakdon't get me wrong, i'm just hinting you :)12:43
bartbesoh, I'm not12:43
bartbesI just never even knew that existed12:43
bartbesso ehm, how does it work?12:43
kyaki don't really know :)12:43
kyakbut who needs to know, prereq-build.mk shows a good usage example12:44
kyak$(eval $(call RequireCommand,hg,Please install hg-git.))12:46
kyaki assume this should be enough12:46
kyakit will issue 'which hg' as per prereq.mk12:46
kyakyou could use a define Require/ block if you need more advanced checking12:47
wejpa while ago somebody said the flac package in openwrt has been updated, but either that's not the case or i am looking in the wrong place. how do i update my openwrt build environment correctly?13:15
kyakflac version in openwrt is 1.2.113:21
kyaki guess it's the new one13:22
kyakto update the feeds, make package/symlinks is good13:23
wejpyes, that's the new one, but how do i get that one13:23
wejpgit pull does not do the trick13:23
kristianpaulawesome link http://www.ctae.org/sdr/13:23
kyakyeah, git pull only updates the base system.. most packages are in so called feeds13:23
kyakso you have to 'make package/symlinks'13:24
wejpokay, thanks13:24
kyaknp! am i guessing correctly that you want to add flac support into gmu build for openwrt?13:25
wejpyeah, i'll enable flac by default in the next version of gmu for the nanonote, so the package that'll be build from that will include flac support13:25
wejpby the way, it will also come with speex support :)13:26
wejpfor speex libogg and libspeex are required, but both are available in openwrt, so no problem there13:27
kyakhow do you use speex? for voice control?13:27
wejpsome people have been asking about speex support. speex is useful for speech-only audio files, like podcasts13:27
kyakah, i see13:28
wejpone can create really small files with speex , whith a pretty good speech quality13:28
wejpabout 12 MB for an hour or so13:28
kyaksounds good; is there encoder available in openwrt, too?13:29
wejpprobably yes, the encoder is included with the speex libraray source13:30
kyakthis could be usefull, too.. because arecord is not compressing audio as it records13:30
wejpand as it is patent free it should be no problem to include it in the nanonote firmware13:31
kyakbtw, while we are discussing gmu - why is alt+enter (for exit) is only working from withing F1 help menu?13:32
wejpit is also working from the playlist view. unfortunately i accidentally have used alt+enter in the file browser for something else. but i'll change that in the next release to be more consistent13:33
wejpthen alt+enter should work in all screens13:33
wejpbut it is also easily possible to change the whole keyboard mapping13:34
kyak'q' would be awesome for exit :)13:34
wejpwithout recompiling anything13:34
kyakah, great13:34
wejpi can also change it to Q13:34
wejpq is cvurrently being used for the queue function though13:34
wejpbut it is no problem to change all that13:34
kyaki'm very satisfied with default gmu keymap, other than 'exit'13:35
kyakok, you made it on purpose :) you don't want people to ever quit gmu ;)13:35
wejpyeah, alt+enter isn't the most intuitive one. it basically is inherited from the dingoo version where the start button is the same as enter and you quit gmu on the dingoo with select+start ;)13:36
wejpbut you are right, people should never quit gmu :P13:36
rafawejp: now you need to build a gmu kernel.. people boot and just gmu appears, exit would be = shutdown/power off ;)13:38
wejpgood idea :D13:39
kyakdaemonizing gmu would be enough :)13:39
wejpi just run gmu instead of init when booting ;)13:39
kyaki have a gmush, gmu-compatible shell13:40
viricwpwrak: how do some small devices with microprocessors work with two 1,5V batteries?14:14
viricthat's 3V14:14
wpwrakdrizzt: kewl. then it will probably get credited to my account monday or tuesday. i'll let you know when it arrives.14:15
wpwrakkristianpaul: hmm, the GPS uses PM. not sure if you'll get much useful information if you just look at the amplitude.14:16
wpwrakkristianpaul: (scope) kewl. the signal in the picture is data, right ?14:19
qi-bot[commit] Xiangfu Liu: this patch add uboot-xburst package http://qi-hw.com/p/openwrt-xburst/293a2a014:20
qi-bot[commit] Xiangfu Liu: some cleanup about xburst target http://qi-hw.com/p/openwrt-xburst/f187a4814:20
qi-bot[commit] Xiangfu Liu: add ks7010 drivers http://qi-hw.com/p/openwrt-xburst/137de8614:20
qi-bot[commit] Xiangfu Liu: copy data folder from history branch http://qi-hw.com/p/openwrt-xburst/842fa6f14:20
qi-bot[commit] kyak: gitignore vim temporary files (*~) http://qi-hw.com/p/openwrt-xburst/7e7437d14:20
qi-bot[commit] Xiangfu Liu: add Nanonoe special feeds.conf under data folder http://qi-hw.com/p/openwrt-xburst/f1d5d3714:20
qi-bot[commit] Xiangfu Liu: [config.full_system] add package byobu http://qi-hw.com/p/openwrt-xburst/74016b114:20
qi-bot[commit] Xiangfu Liu: cleanup the build file http://qi-hw.com/p/openwrt-xburst/ef4105f14:20
qi-bot[commit] Mirko Vogt: add <http://downloads.qi-hardware.com/sources> as mirror for source http://qi-hw.com/p/openwrt-xburst/f88bb0114:20
qi-bot[commit] Xiangfu Liu: adjust the rootfs partiton size to 512MB http://qi-hw.com/p/openwrt-xburst/f1c496214:20
qi-bot[commit] Mirko Vogt: deselect gnuplot since it does not compile (in our tree) http://qi-hw.com/p/openwrt-xburst/589ead114:20
qi-bot[commit] kyak: NAND DATA Partition offset and reflash_ben.sh fixed according to 512 http://qi-hw.com/p/openwrt-xburst/14ac57714:20
qi-bot[commit] Xiangfu Liu: [config.full_system] add gcc-mips, fixed in openwrt package: 8dbf61c, tested in build hosts http://qi-hw.com/p/openwrt-xburst/1c3892a14:20
qi-bot[commit] Xiangfu Liu: [linux] support increase/decrease in screen brightness http://qi-hw.com/p/openwrt-xburst/edfd9f414:20
qi-bot[commit] kyak: Merge branch 'master' of projects.qi-hardware.com:openwrt-xburst http://qi-hw.com/p/openwrt-xburst/099ec9714:20
qi-bot[commit] kyak: added w3m to config.full_system http://qi-hw.com/p/openwrt-xburst/b404fd514:20
heberthwpwrak:  kristianpaul thi's sleeping14:21
wpwrakviric: hmm, just like devices that work with 5, 3.3, 2.7, 1.8. 1.5, or 1.2 V ? ;-)14:22
viricwpwrak: well, I don't know about that ;)14:22
viricIt's a matter of choosing ICs that work, let's say, at 2,7V?14:22
wpwrakviric: well, depends on what you need. low voltages are not a problem for modern technology. used for fast and slow systems.14:23
viricfor example, that 8051 mcu that had usb... can I find something close to that, that can work with two 1,5V batteries?14:24
wejpyou do not run thos ICs directly from the battery voltage14:26
wpwrakyes, probably.14:26
wejpyou rather use voltage regulators14:26
wpwrakwejp: why not ?14:26
viricI'd expect a very regular voltage from batteries14:27
wpwrakwejp: naw, battery generally is fine for mcus.14:27
wejpwpwrak, because the battery voltage varies, depending on its charging state14:27
viricwell, I don't think the mcus worry much on that14:27
virica regulator could take a lot of power from a battery14:27
wpwrakwejp: yes, but if your mcu has a voltage range from, say, 2-5 V, you don't really care ;-)14:27
wejpif you don't care about stability and don't care about using the battery efficiently, you cann use it without a regulator of course14:27
viricwpwrak: ah, really some have that range? That's great14:28
wpwrakviric: LDOs don't burn all that much power. but they cost money and space.14:28
viricwhat is a LDO?14:28
wejpand modern regulators are very efficient and do not use much energy themselves14:28
viricwpwrak: you said a cr2032 cannot power one of those f321?14:28
wejpswitching regulators that is, linear regulators suck for battery operation14:29
wpwrakviric: e.g., PIC18F13K50. 1.8 - 5.5 V14:30
viricsdcc can target that?14:30
viricI'm checking those f32114:30
wpwrakwejp: LDO vs. switching: depends again on what you're doing.14:31
wpwrakviric: the f321 unfortunately don't go very low. not ideal for battery14:31
wpwrakLDO = Low-DropOut regulator14:32
wejpah, well those are usually still linear converters14:33
wejptheir effiency is not the best14:33
wpwrakwejp: depends on what you're doing :)14:33
wejpyes, of course14:34
wejpbut if you are running on battery, most of the time you want to achieve a long runtime with those batteries ;)14:34
wejpif the circuit is using very little current, it probably doesn't matter much though14:34
wpwrakwejp: yup. when running from battery, the first thing you do is try to minimize your duty cycle.14:36
viricwpwrak: I was thinking on making some device to control whether to switch the flat heating on and off (it requires a relay as 'output'14:45
viricsome buttons, and two or three 7-segments should do the job...14:45
Action: kristianpaul back from snap14:47
kristianpaulwpwrak: yes is data14:47
kristianpaulbut i;m here at University lab and get better scope probes14:47
kristianpaulsignal is not triangled at least :)14:47
kristianpaulclock signal*14:48
kristianpauli'll try data again14:48
kristianpaulwpwrak: about GPS, yes of course i dont need look just to the amplitude14:48
wpwrakviric: you'll probably need a transistor for the relay. also, don't forget the diode :)14:50
viricwpwrak: I know. :)14:51
wpwrakkristianpaul: (clean clock) that's good :)14:51
kristianpauli see squares !14:51
Action: kristianpaul is usign brand new probes14:51
kristianpaulcan i suppose that the scope metter just miss the trigger bcause this is a "ramdon" pulse data siganl?..14:54
wpwrakgood. such a slow signal really ought to look tidy :)14:55
wpwraki don't think you miss the trigger. at least in the picture you posted, you got the trigger14:55
wpwrakjust the signal before and after (that is, one cycle later) is variable14:56
kristianpauli'm tunnign trigger right now14:56
kristianpauli jsut met this scope today14:56
wpwrak100 MHz ?14:57
kristianpaul150mhz 25GS/s14:57
wpwrak25 GS ? i doubt that :) 150 MHz is good14:57
kristianpaulsure i'm working ar 250MS/s14:58
wpwrakprobably a GDS-820. hmm. data sheet says 100 MSa/s for real-time. it switches to effective time for higher rates. not so good. ET only gives valid results for periodic signals.15:02
wpwrakperhaps that explains the pattern you're seeing in the [-oo, 0] and [1 cycle, +oo] from the trigger15:03
kristianpaulGDS-820S right15:03
qi-bot[commit] kyak: enable locale support in uClibc http://qi-hw.com/p/openwrt-xburst/1b577b715:08
wpwrakwell, it'll do. i think you can also disable ET15:10
wpwraki mean i'm afraid you _can't_ disable ET15:10
qi-bot[commit] kyak: fbterm iconv patch not needed since uClibc has locale support http://qi-hw.com/p/openwrt-packages/2d7e39815:10
qi-bot[commit] kyak: Merge branch 'master' of projects.qi-hardware.com:openwrt-packages http://qi-hw.com/p/openwrt-packages/334203a15:10
qi-bot[commit] kyak: add libncursesw package http://qi-hw.com/p/openwrt-packages/c9435fb15:12
qi-bot[commit] kyak: add libncursesw to config.full_system http://qi-hw.com/p/openwrt-xburst/9f6250e15:14
kristianpaul250MS/s not goof16:12
kristianpaulbut 100MS/s just fit my needs16:12
wpwrakkristianpaul: does the display look better when you go to 100 MSa/s ?16:43
Action: FrankBlues waves16:46
kristianpaulwpwrak: yes17:23
kristianpaulwait for min for a pic17:23
kristianpaulwpwrak: http://en.qi-hardware.com/wiki/File:Data_and_sync.JPG17:27
kristianpaulhey FrankBlues17:36
FrankBluesHey kristianpaul! How's it going?17:38
kristianpaulha i founded the doc for gps-sdr now i need to read 1024 pages, great !! (seriouslly)17:44
kristianpaulah suspiciously generated by doxygen17:48
drizztwpwrak: hi !18:07
drizzta question, about CE/FCC and CEM tests18:08
drizzthow do you manage them for Qi-hardware ?18:08
drizztand how does it cost ?18:08
wpwrakkristianpaul: very nice. these pictures look much better.18:09
wpwrakdrizzt: you'd have to ask wolfgang for the details18:09
wpwrakdrizzt: i think you just give the device to a certification lab, tell them what you want to have certified, then they run their tests ... and then comes the expensive part, fixing all the flaws they find :) in the end, they issue the certificate.18:10
drizztyes, I already did this part18:11
drizztbut as employee, so I never new (and carred) about the bill18:11
wpwraki don't know the cost structure either. i've heard something of a few kbucks, but don't have details -> wolfgang18:12
kristianpaulwpwrak: yeah just ready to wire SIE and SiGE EVB and begin the hard work with an already debuged envirioemnt :)18:13
wpwrakkristianpaul: you should tell the folks at the lab that they should get themselves a new scope and let you have their old junk :)18:13
kristianpaulnever use irssi with more than one window open !18:14
kristianpaulwpwrak:how the I/Q  dta provided by the URSP?18:14
drizztI have the details for medical devices (15Keuros for CEM tests only) but not for "simple" electronic devices, neither for both CE + CEM tests18:15
kristianpauli meant timinng and data per "cycle" or sync18:15
kristianpauli'm not aware of USRP signal adquisition process18:15
wpwrakkristianpaul: i/q ... oh, it's pairs of floats sent over Ethernet18:15
kristianpaulfloat ohh is BIG18:16
wpwrakactually ... floats is what the beastie outputs in the end. i think the usrp itself sends data in some 16 bit format.18:16
wpwrakyes, has to be 16 bits. the adcs are 14 bit.18:16
kristianpaulEthernet?? but latency is about miliseconds is that god enoguht?18:17
kristianpaulinteresting lets way URSP2 get cheaper (~700) to get one18:17
wpwrakdepends on what you're doing. most radio protocols don't mind a little bit of latency between stuff sent and received.18:17
kristianpauli see18:18
wpwrakit's probably also less than milliseconds18:18
kristianpaulhow come?18:18
wpwrakettus just released a new model, kinda like the usrp2, but USD 300 more expensive :(18:18
kristianpaultooo bad18:19
wpwrakconsider a 1000 bytes frame. that's something in the order of 10 kbits. at 1 Gbps, that would take only 10 us.18:19
kristianpauloh i forgot the 1Gps , is true18:20
wpwrakyou can still get the USRP. it's a bit slower and its ADCs/DACs have less resolution but it can use the same cards. it's only USD 700.18:20
kristianpaulyeap i'll may consider that for neaxt year18:21
wpwrakas an added bonus, it can even host two cards at the same time18:21
kristianpaulfull duplex usrp way?18:21
wpwrakdrizzt: if it says "medical" somewhere, it's probably more expensive than "regular" :)18:22
wpwrakfull duplex depends on the RF card. the USRP itself is always full-duplex-capable.18:22
wpwraknote for USB: if full-duplex-bw = 1/2 half-duplex bw.18:23
wpwrakso don't quite expect to run 2 TX + 2 RX 802.11g streams over an USRP :)18:23
kristianpaulyou got me !18:24
drizztwpwrak: yes, and this price is from a lab in Switzerland ....18:24
kristianpaulwpwrak: btw is MAX1003 the 6bit ADC 90Mspschip you pointed last time?18:25
wpwrakkristianpaul: don't remember. could be.18:27
wpwrakactually ... maybe not. i was looking for the usual DC-coupled type.18:29
wpwrakit does look like a nice chip to make SDR with, though18:30
wpwrak ah, it supports DC coupled operation too. that's nice.18:31
wpwrakjust a bit slow18:33
kristianpauli aks then for a EVB is 60USD, i think is expensive.. what do you think?18:36
kristianpaulis not that i going to buy it just i like/survery18:36
wpwrakthis is more like it: http://datasheets.maxim-ic.com/en/ds/MAX105.pdf18:37
kristianpaul800mps O_o18:37
wpwrakEVB prices can be pretty much anything18:38
kristianpaulhow is that? what rules it?18:38
wpwraklow volume ;-)18:39
wpwrakand for simple chips also captive or rich customers;-)18:41
wpwrakof course, if you consider an engineer's salary, it's hard to beat an EVB for only USD 60. you have to be very efficient to make your own EVB for less.18:45
drizztkristianpaul: if you find this expensive, have a look at this : http://focus.ti.com/docs/toolsw/folders/print/tmdxevm8168ddr2.html18:45
kristianpauldrizzt: expensive and too closed, not for me :^)18:47
wpwrakrafa: btw, was there a solution to the rootfs resizing conflict between openwrt and jlime ?18:47
kristianpauli miss that log18:48
qi-bot[commit] Andres Calderon: labels overlaped to pads has been fixed http://qi-hw.com/p/xue/5b6c73e18:48
wpwrakhmm .. the MAX105 has a DREADY signal. that could almost be a real english word ;-)18:48
kristianpaulData Ready Clock ?18:50
wpwrakkristianpaul: qi-hw's openwrt wants to change the rootfs size (or for all i know may have already done so). now, how are other distros to deal with this ? particularly if they allow you to mount the rootfs18:50
kristianpaulyes done18:50
rafawpwrak: no, I do not understand well the changes.. it seems that a time ago openwrt used 2gb, then used 512mb, then 256mb and now again 512mb.. No sure if that is the story.. but well18:50
wpwrakkristianpaul: yup. sounds like a mixture between dread and dreary :)18:51
kristianpaulrafa: it was i think18:51
rafawpwrak: I do not understand well the reasons and how that is decided18:51
kristianpaulrafa: reason was current image was running out space18:51
rafawpwrak: so.. jlime used the whole short life on nn just the 2gb.. and users will need to wait 5sec more to boot :)18:51
kristianpaulmostly because gcc inclusion so...18:51
wpwrakrafa: oh, i think it went something like this: at some point in time, the rootfs was full. then they discussed a bit about increasing the size (here, on #qi-hw)18:52
wpwrakrafa: then nothing happened for a while. and then xiangfu announced the immediate change :)18:52
kristianpaulrafa: i think that was the wrong way when resized the 2gb to 256mb, who cant wait 5s more18:52
kristianpaulwass  a too fast desicion at that time just for get 5 seconds of boot time18:53
wpwrakrafa: not sure if and how the 2 GB -> 512 -> 256 happened. just saw the 256 -> 512 change.18:53
kristianpauli think larsc remenber better that us18:53
wpwrakwell, 5 s is a long time if you're waiting ...18:53
kristianpaulyes sure18:53
rafakristianpaul: yes.. but well, if nn is some day a product more like a phone I would like it booting in just 2seconds.. but now yes, we can wait a bit more. Or it would be cool to know the reasons ideas about modifying that18:55
rafakristianpaul: so we can give our ideas/opinions as well18:55
rafakristianpaul: before the change happens :)18:55
kristianpaulrafa: about that i will remove gcc at first place18:55
kristianpauland just give it like an option18:55
kristianpaulbut want i see is that every new package is included as no option18:56
rafawpwrak: (announced) ah, I see then that is why I do not know well the whole story and their reasons18:56
kristianpaulwich is not bad , at least when it is 60MB big? i think is gcc, not sure at all18:56
kristianpaulany way..18:56
kristianpaulNight to take a beer18:56
kristianpaulsee you tomorrow18:57
rafajaja ;)18:57
kristianpaulabout ssh and my ilusinations i say, why not give disable automatica startup and give a key bidning to acitvate it18:58
kristianpaultha could save valious startup time too18:58
kristianpaulalso save memory18:58
kristianpaulcalled "transfer mode" just make posible able and disable it18:59
kristianpaulyou dont need it running all time18:59
kristianpaulbut you need it there later18:59
kristianpaulchao !18:59
wpwrakkristianpaul: it's nice to have ssh even if you don't have a display :)18:59
kristianpaulwpwrak: why? what for?19:00
kristianpauli SAID key binding19:00
kristianpaulbut is true19:00
wpwrakwhen you disconnect it :)19:00
wpwrakkey binding may still need some visual verification of the device state19:00
kristianpaulwhat about a beep ? ;)19:01
qi-bot[commit] Werner Almesberger: usrp/: move first "step" runs to directory usrp/step/ http://qi-hw.com/p/ben-wpan/d78dd4e19:01
qi-bot[commit] Werner Almesberger: usrp/d: cleanup and allow number of samples to be set on command line http://qi-hw.com/p/ben-wpan/38e5b8a19:01
qi-bot[commit] Werner Almesberger: usrp/p: cleanup and more command-line control http://qi-hw.com/p/ben-wpan/417ef7b19:01
qi-bot[commit] Werner Almesberger: usrp/plot: enable grid and accept center sensitivity from the command line http://qi-hw.com/p/ben-wpan/044c15219:01
qi-bot[commit] Werner Almesberger: usrp/: gave tools slightly more self-explaining names http://qi-hw.com/p/ben-wpan/e76af7d19:01
wpwraknaw, just launch it in parallel with the rest19:01
kristianpaulsecond try against ssh runing failed ;)19:02
kristianpaulnot parallelr19:02
kristianpaulmay be after 20 seconds19:02
kristianpauljust make sure you get a nice screen first with gmenu2x19:02
kristianpaullater lanuch ssh19:02
wpwrakit probably doesn't do much anyway. probably waits for /dev/random or such.19:02
wpwrakanyway, enjoy your beer ! ;-)19:03
rohhi wolfgang20:28
wolfspraulroh: pong20:28
rohi just found some milkymist-stickers here at the agency. also i heard that somebody of the devel team is in berlin atm?20:29
rohsomething about 'a mission to build a case for it' ?20:30
wolfspraulroh: ah, cool20:30
wolfspraulyes of course the founder of the Milkymist project moved to Berlin a few months ago20:30
wolfspraulSebastien Bourdeauducq20:30
wolfsprauland yes, a case would be really cool20:30
rohnice. he should visit us sometime, then i am sure we can work something out20:31
wolfspraulok let me make an email intro since he is not in irc right now (his irc nick is 'lekernel', and he is in #qi-hardware as well as #milkymist on freenode)20:32
rohah. ok.20:32
wolfspraulroh: your email is still @om.org ?20:32
rohjabber is nice ;)20:32
rohwolfspraul use @hyte.de20:32
rohom still works, but i rarely check it20:33
wolfspraulemail sent20:35
wolfspraulroh: what materials or process would you have for making the case?20:38
rohanything you can laser, mill or cut. just depends on how complex it will be20:40
rohi would go for finding a readymade plastic case and laser or mill the custom holes for the connectors20:40
rohalternatively one can always do sheet-matieral based cases with the laser.. e.g. acryllic20:41
rohlike this http://jjshortcut.wordpress.com/2010/08/16/my-logic-analyzer-and-case/20:41
wolfspraulhmm, the first case is not bad20:42
wolfspraulSebastien had a black metal case in mind, but I think everybody is flexible.20:43
rohit depends mostly on the 'how many' and 'how much can you invest'20:45
rohplasic molds are expensive. but cheap in use. laser is expensive and scales bad (doesnt get much cheaper in volume)20:45
wolfspraulroh: I'm not sure I like plastic20:49
wolfspraulI mean the injection process20:50
wolfspraulnot the material20:50
wolfspraulhow expensive (roughly) would it be to laser holes into a plastic case20:51
wolfspraulat the beginning we just need 30-50 or so20:51
rohdepends on the runtime and if we can place it easily20:51
rohwe usually take an euro a minute, but i guess we can work something out20:52
wolfspraulI count 16 connectors20:55
wolfsprauland then the problem of accessing the microSD, and the jtag-serial connectors20:55
wolfspraulplus the microphone which is inside, and the case probably needs some holes to not completely seal it20:56
wolfspraulthe board is 10x15 cm21:05
wolfspraulroh: have you seen it?21:06
rohah. that one21:07
wolfspraulthe microphone is not soldered in that picture, it's the round area right of the 2 vertical expansion headers21:07
wolfspraullower right side21:07
wolfspraulthe microSD connector is kindof in the middle of the board, to the left of it are the jtag/serial connectors21:08
wolfspraulwe have a small daughterboard with a ft2232h to plug in there, and then the usb conector will go out to the left side21:09
wolfspraulsee http://en.qi-hardware.com/wiki/File:Jtag-serial-board-on-m1.jpg21:09
wolfspraulthat whole jtag-serial thing is probably only necessary for the first units, or first 1-2 years21:09
wolfspraullater there will be quite robust 'reset to factory default' and jtag-serial will only be necessary for a few people21:10
wolfspraulso :-) do you have any rough idea/first guess on how long this would take, to laser cut into plastic...21:11
rohnope. its really extremely material dependant. and thickness21:12
rohone usually buys some sheets of some material to test and does runs to find  optimum parameters, then comes the real run to do something useful21:13
wpwrakthat board has connectors at all sides. that makes it difficult. probably impossible to find a ready-made case. (that is, unless one was considered when dimensioning the board)21:43
wpwrakfor a custom case, how about a base and top sheet of acrylic, mill grooves for the positions of lateral plates. then make 2D lateral plates. maybe laser first, mill metal later.21:44
wpwrakthen the only remaining problem would be the corners. maybe some flexible material could be found for them. seal the gaps with silicone ;-)21:45
rohwe'll see21:46
rohsometimes after some prototyping in plywood one gets the right idea21:47
wpwrakto keep it all together, use (in the four corners) plastic tubes of precise length at stand-offs, with a long screw through them21:47
wpwrakwood is nice. you can always fall back to nails ;-)21:48
roho. los23:03
--- Sat Nov 13 201000:00

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