#qi-hardware IRC log for Sunday, 2011-01-16

lekernelzrafa: looks like a mobile phone being reverse engineered05:42
wolfspraulsteve|m: oh nice05:42
wolfspraulyes, here's the story...05:42
lekernelsteve|m: how do you scrape those pcb layers?05:42
wolfspraulsteve|m noticed that I did a number of delayerings, and he wanted to give me some more boards05:42
wolfspraulI explained the whole background to him, street jobs for 100 USD, or big reverse corp (ma-tek.com) for 2000 USD, etc.05:43
wolfsprauland in the end he decided he can just try it himself! :-)05:43
wolfspraulso he took a sandpaper and went through the pcb05:43
wolfsprauljust manually, that's how the 'street jobs' folks will do it too05:43
lekernelah, just sandpaper?05:43
wolfspraulthen I asked him to document that a little, also to write a short mail somewhere saying which sandpaper he used, what he learnt...05:44
wolfspraulI guess those pictures are the first part of it.05:44
wolfspraullekernel: do you know www.ma-tek.com ?05:45
wolfspraulit's a nice analysis company, maybe 50 people or so, and they have all equipment right there05:45
wolfspraulyou would like the place :-)05:45
wolfspraulthey do analysis (mostly ic, of course they can do pcb too) for many customers (see their website)05:45
lekernel"The Best R&D Partner"... this says a lot about what kind of R&D is carried out in Asia :)05:46
wpwrak_.cn = copy nation :)05:47
wolfspraulnot true05:47
wolfspraulthis is a good company05:47
wolfsprauland it's Taiwan, not China05:47
wolfspraulbut they have offices in China of course, the Chinese government gives huge subsidies to this type of analysis/reversing company05:47
wpwrak_wolfspraul: now they'll arrest your for sedition05:48
wolfspraulanyway they do ANALYSIS05:48
wolfspraulyou think way too narrow if you think that's all just copying05:48
lekerneloh, I never said they're not a good company... they must be good at their analysis job :)05:48
wolfspraulthey are used to analyze the failures of particular ics05:48
wolfspraulof course they are also used for competitive analysis05:48
wolfspraullike "what are the others doing"05:48
wolfspraulthey are used for quality control05:49
wolfspraulthey are used for ip litigation, i.e. to proove that another chip has my technology inside05:49
wolfspraulanyway they are just analyzing, they are agnostic as to what their customers do with the data they create05:49
wolfspraulit's a good company, really05:49
wolfspraulif we ever move into ic land, I'm sure we will have some jobs for them.05:49
wolfspraulthey are small, focused, run by an amazing founder, reasonably priced, etc.05:50
wolfspraulreally cool people05:50
wpwrak_(when going into chips) i wouldn't be surprised if that service came as part of the package from the chip manufacturer05:56
wpwrak_agreed on the relevance of the other cases, though05:57
kristianpaulwow just sandpaper?07:13
wolfspraulkristianpaul: yes. it's amazing how often simple technology works :-)07:27
kristianpaulit is07:35
zrafawolfspraul: nice ! :) (about steve job)08:17
wolfspraulkyak: you there? building config.full_system I get an error about not finding libgfortran.so.*08:44
wolfspraulhave you seen something like this before?08:44
lekernelbtw, can the nanonote run Windows CE?08:46
wolfspraulno idea08:46
lekernelhe, apparently yes08:46
kristianpaullekernel: yes08:51
kristianpaulbut i never tried of course :p08:52
lekernelwell, if you have GNU patches to merge upstream, putting windows CE forward could be a nice way to motivate the FSF maintainers imo :)08:52
lekernelmicrosoft hatred is the only thing that moves them08:53
wolfspraulfor now it seems when I remove octave, plplot and libgfortran, things work again09:36
kyakwolfspraul: this might have something to do with fortran support, recently added by David - i think he should know better, because i didn't build gfortran09:50
wolfspraulyes I am preparing an email to the list, I think that's the easiest way to reach him09:55
wolfspraulprobably easy to fix...09:55
wolfspraulkyak: did the iconv/gettext situation improve? is there anything that bugs you?09:56
wolfspraullet me look at our troublemaker tickets :-)09:57
kyakthe situation hasn't improved, i.e. there is no solution to switch between full and stub iconv/gettext; but it doesn't really bug me, because we have these overriden09:59
wolfspraulok but I think a more general solution will be introduced at some point10:00
wolfspraulI keep an eye on it. the tickets are also still there...10:01
kyakyes, like mirko said10:01
kyakso let's hope it will be introduced in next weeks10:01
wolfspraulok I see mplayer has not yet made it into config.full_system?10:03
kyakyes, it's not there.. i'm still wondering if it's realy clean :)10:04
wolfspraulwhy did you disable mencoder?10:06
kyakgood question.. at first i thought "let's see how mplayer would build, leave mencoder apart" and than i thought "who needs mencoder on Ben?"10:09
wolfspraulah ok10:15
wolfspraulthat's natural :-)10:15
wolfspraulfor some reason over the years I ran into many systems/distros that for some unexplicable (to me) reason left out mencoder10:16
kyakit's just usually packaged separately :)10:16
kyakthe case in mandriva10:17
B_LizzardHey, how do I produce a NAND-ready kernel image from a uImage used for MicroSD installations?10:37
B_LizzardI wonder if that made any sense.10:38
larscnot to me10:41
B_LizzardI think rafa's helping me, in any case10:44
lekernelat least some interesting stuff about the kinect: http://www.futurepicture.org/?p=11610:54
lekernelbetter than all the buzz about this stupid open source DRIVER... lol!10:55
lekernelthose micro patterned lenses would also be something fun to play with...10:57
zrafakyak: you could need mencoder to scale some video to fit nn screen?11:03
zrafakyak: just thinking.. no idea if it would be really useful for people11:03
zrafakyak: and maybe it is really hard to build without any codecs (and just ogg and theora)11:04
kristianpaulmilkynect? :-)11:04
wolfspraullekernel: that's a very nice blog! I'll add it to the Qi planet...11:04
kristianpaulwow yes11:07
kristianpaulwolfspraul: please add this if you like usrp/gnu radio blog for qi http://www.oz9aec.net/index.php/gnu-radio/gnu-radio-blog11:09
wpwrak_DocScrutinizer: first result with the crystal:  http://downloads.qi-hardware.com/people/werner/wpan/redesign-xtal/11:43
DocScrutinizerHEH \o/11:43
wpwrak_DocScrutinizer: doesn't get much cleaner, does it ? :) the signal seems a bit weak, though. could be a soldering problem.11:43
DocScrutinizerlooks *much* better to me, no?11:44
wpwrak_yay !11:44
DocScrutinizeraah, doesn't get... yes, as good as it gets :-)11:45
wpwrak_ah, caught the antenna at a bad angle. taking a new spectrum ...11:45
DocScrutinizerwell, mission accomplished :-D11:47
wpwrak_DocScrutinizer: reload. peak is a nice 40 dB above noise. maybe i can get rid of the thingy at +0.5 MHz, too.11:47
wpwrak_(may be leakage from the PLL, though)11:48
DocScrutinizerit's single sideband, that's most likely an artefact11:48
wpwrak_it's at the center frequency11:48
lekernelwpwrak_: what spectrum analyzer do you have?11:49
wpwrak_the large peak is at center -0.5 MHz. the constant wave is center +/- 0.5 MHz11:49
wpwrak_lekernel: none :) it's a USRP211:49
DocScrutinizercheck your USXRY11:49
wpwrak_my what ? :)11:50
lekernelwpwrak_: what's the highest frequency there's a usrp daughterboard for?11:50
wpwrak_lekernel: i think ~5.8 GHz (the board says it goes up to 6.1 GHz)11:52
lekernelmh, ok. a bit too low for my purposes :(11:52
wpwrak_lekernel: if you're looking for spectrum analyzers up to 9.4 GHz, here are some that are quite low-cost in comparison: http://www.spectran.com/12:01
lekernelhe, not too bad. thanks for the link!12:01
lekernelusrp would be nicer, as the purpose is to measure the frequency variation induced by a vibrating piezo in a microwave cavity12:02
lekernelwith potential corrections so get the frequency modulation waveform that I want at the end12:02
kyakzrafa: heh, you are right, it's doubtful that someone would upload a video to Ben, encode it there and then watch it :) btw, are you aware about other open codecs other than ogg and theora?12:03
lekernelbut perhaps I could get away with just a spectrum analyzer or frequency counter... need to check12:03
lekernelbasically depends on the linearity of the piezo and absence of mechanical resonances...12:05
kyaki understood "basically" -]12:06
lekernelalso I'm working around 9.35GHz, so it's near the limit of those devices :(12:07
kristianpaulask ettus12:08
wpwrak_lekernel: you could design your own RF frontend. 9.4 GHz ought to be fun :)12:11
lekernelnope. I just want my radar to work. not to involve in those things12:11
lekernelso it's going to be with OTS equipment12:12
lekernellater i'll try to get my hands on a car cruise control system to scavenge the 77GHz radio. _this_ is fun (and would increase the radar resolution a lot) and what I want to spend my time on. not design 9.4GHz frontends12:14
wpwrak_77 GHz sounds nice :)12:20
lekernelthere I think I'll have to design some equipment from scratch :p12:21
lekernelthe DRO running at 19 GHz provides a highly accurate signal to the PLL circuit12:22
wpwrak_you could chain a few plls. start with a 32.768 kHz watch crystal ;-))12:24
lekernelthose chips seem to require an interview to order, so i'll probably need to use some reverse engineering as well...12:24
lekerneloh, i'll use pre-mounted boards from actual car cruise control systems12:25
lekernelmuch easier to find than the chips12:25
lekerneland easier to work with (at this frequency you work with bare dies and micromanipulators)12:25
lekernelit's going to be a nice challenge :)12:25
lekernelbut I'm exercising with a cheaper and easier 9.35GHz design first12:26
wpwrak_i'll just wait for your 2012 version that uses neutrinos :)12:27
kristianpaulif you get evil please dont make an ADS :-)12:31
lekernel(crystal) DROs are pretty stable already. the (super cheap) 9.35GHz one I have on my desk atm has a specified < 5MHz frequency drift12:31
zrafakyak: (codecs) no much.. .. BTW, on beta4 jlime on qi servers I added a theora video player. It is the sample examle with libtheora. I added a bit of crappy code to control it a bit.. if you are interested to put it on openwrt I could give you the details. But if you have a clean mplayer (without patented stuff) then you do not need the theora example I have.. mplayer is 1000 times better.12:41
wolfspraullekernel: do you know a good blog about diy radars?12:45
lekernelblog, no12:45
wolfspraulI think futureproject.org is basically just a syndication of danreetz.com and mattikariluoma.com, and dan seems to be the one blogging the most interesting stuff12:58
wolfspraulso I'll include danreetz.com in the planet, see how it goes. Also I will email him to notify him of the syndication, in case he doesn't like it...12:59
wolfspraulwpwrak: congrats for the signal quality!13:11
wolfspraulI understand your idea about reusing the clock was kia, but oh well13:11
wolfspraulI think it will also make the board more robust, even if the Ben clock would have been good, other devices may have different clock characteristics.13:12
wolfspraulit's just too much to reuse a digital SD clock in an RF application, I guess...13:12
kyakzrafa: thanks, will keep it in mind about theora player13:19
wpwrakthanks ! yes, apparently the clock is a bit too sensitive13:19
wolfspraulkyak: which feeds.conf are you using now?14:08
wolfsprauloh, I just see there is a feeds.conf committed in data/qi_lb60/conf now14:09
wolfspraulwith qipackages first, but no revision on the openwrt packages feed. alright the, I'll try that one...14:10
kyakwolfspraul: yep, that one...14:54
wolfspraulkyak: I just got a compile error in fetchmail 6.3.19, oh well15:01
wpwrakwolfspraul: for version 2.0, we can do the board with 4 layers and microvias. then i could route all the traces underneath the crystal :)17:39
kristianpaulany one can recommend a commad line tool for generating html index of a directory?18:06
Jay7for i in *; do echo "<li>$i</li>" >> index.html; done :)18:11
kristianpaullol :-) thanks !18:13
kristianpauli need read man bash more often18:13
Jay7yeah, it's useful sometimes :)18:15
wpwrakwolfspraul: btw, what kind of text do you want on the board ?18:26
wolfspraulhard to say.18:26
wolfspraulI think in general the cc-by or cc-by-sa logo is nice18:27
wpwrakpheew ... do you have that for kicad ? :)18:27
wolfspraulthen I like to credit creators, so don't hesitate to put (c) Werner Almesberger on it :-) (although your name is long :-))18:27
wolfspraulfor kicad, not sure18:28
wolfspraulwe can do it later18:28
wolfspraulwithout logo, you can just put cc-by-sa on it as text18:28
wpwraki can put "cc-by-sa". in principle, there's a lot of room for text on the silk screen18:28
wolfspraulyeah just put the text there then, and it's fine18:29
wpwrakjust because i don't have a silk screen at the moment, it all is on the copper layers18:29
wpwrakpick and place likes component outlines on the silk screen, don't they ?18:30
wolfsprauldon't know18:31
wpwrakwhat other information to put ? the project name, QI-HW ? [IEEE] 802.15.4 ? or WPAN ? both ? version/date code for sure. i think i'll scrap the "8:10 CARD". it's kinda obvious when you look at it anyway.18:33
wolfsprauldate code is good, yes18:44
wolfspraulcc-by-sa, we had that18:44
wolfspraul802.15.4 - sure why not?18:44
wolfspraulyou can put a url there, for example the short qi-hw.com18:45
wpwrakah, cool. that's handy, yes.18:47
wpwrakgrrr. bloody zone construction algorithm crashes pcbnew every time i try to redo the digital zone. @#*%$!18:49
wpwrakthat would be in fact one item that's probably fixed in a more recent version. at least i'd very much hope so :)18:50
wpwraklet's see if my electricity is good again ...18:52
wolfspraulwpwrak: btw it turns out the new jtag-serial boards we made only work in full-speed, not high-speed18:55
wolfspraulthe ones yanjun luo made worked in high-speed too (after a fix)18:55
wolfspraulso we took on a few risks too many, something got wrong either in my schematics cleanup, or in boom, or manual sourcing, or production, or ...18:55
wolfspraulwe'll track it down18:56
wolfspraulthe good news is that full-speed is good enough right now for everybody to just use them, so it's only an annoying bug, but no drama18:56
wpwrakgreat that it's not a big issue. production tolerances and other such small things can be evil indeed.18:58
wpwrakwe'll also have to figure out a way to production-test the wpan boards. hopefully some BER/PER measurements will enough. (thus not require a spectrum analyzer or usrp)18:59
wpwrakanyway, gotta run for now19:00
wpwrakadamw_: when making a low-cost board, do you normally put a silkscreen on both sides of a pcb ? or just the front ?20:32
adamw_wpwrak, no regarding the cost on silkscreen on both sides. They are quite the same on two layers manufacturing.20:33
wpwrakany preference ?20:34
adamw_just put them on both side you think they should be at there for more clearly and visualable.20:34
adamw_no,, but if you want, I can ask them.20:35
wpwraklet's see if i can fit everything on one side ...20:36
adamw_but from my points. you don't need to take into such considerations since it's a consideration. :-)20:36
kristianpauljust shirnk names :-)20:36
adamw_yeah..if you could.20:36
wpwrakkristianpaul: they're already near the low end of my process :)20:37
wpwrak7 mil line width, character cell nominally 40 x 40 mil (whatever kicad does with this .. the letter aren't square)20:39
kristianpauladamw_: what for is the small spring that came in the probe bag?20:45
kristianpaulHi there btw :-)20:45
wpwrakkristianpaul: when you remove the probe's clip-on cap, you see the needle-shaped tip and behind it a wide metal ring20:48
wpwrakkristianpaul: you can slide the spring on the metal ring such that the straight part of the spring is in parallel to the probe tip20:48
adamw_kristianpaul, hmm...just wpwrak wrote to you. :)20:49
kristianpaulah so is spare part?20:49
wpwrakkristianpaul: then you can touch ground with the spring and the signal with the tip. the advantage over using the crocodile clip for ground is that you have a MUCH smaller loop between signal and ground20:49
kristianpauli need a picture..20:49
wpwrakkristianpaul: it's not a spare. it's a different mode of operation20:50
wpwrakkristianpaul: mode 1: clip probe to the signal, crocodile to ground20:50
wpwrakkristianpaul: mode 2: crocodile to ground, but you remove the cap and touch the signal with the "needle"20:50
wpwrakkristianpaul: mode 3: touch signal with "needle" and touch ground with the spring20:50
wpwrakmode 3 is the most difficult to use but gives you the best signal fidelity20:51
wpwrakkristianpaul: by the way, have you noticed the ethernet port of the scope ?20:52
kristianpaulyes yes20:52
kristianpauli got a sofware for that, i just dint tried yet..20:52
wpwraki think there's a little web server running in it. so it's quite easy to use.20:52
wpwrakthere's more sophisticated software as well, yes20:53
kristianpaulif that allow me driver scope from computer will nice, the SCALE pot is tricky20:54
kristianpaulit like plays between min and max just touching it :-)20:54
kristianpauls/pot/ajust pot20:54
wpwrakah yes, i heard that20:54
kristianpauli'll check webserver. let see20:55
wpwraki'm not sure if it has dhcp. you may have to set the ip address via the scope's setup. (don't remember what it's called)20:55
kristianpaulwpwrak: scope httpd, nice :-)21:06
kristianpauli had to use static ip dunno, why dhcp dint worked..21:08
qi-bot[commit] Xiangfu Liu: enable the toolchain option for Octave and libgfortran http://qi-hw.com/p/openwrt-xburst/68914e021:10
wpwrakroh: tuxbrain once floated the idea of putting a "nose" under large ben extension boards. the nose would go into the lanyard hole and add a bit of mechanical stability.21:16
wpwrakroh: with atusd having grown so much, i'm beginning to consider this idea again ...21:17
wpwrakroh: the lanyard hole has a depth of about 3.3 mm. we need ~1 mm clearance to push-to-insert/release the board. this would leave about 2-2.3 mm for the nose. not a lot, but perhaps better than nothing.21:19
wpwrakroh: the lanyard hole is about 3.4 mm wide. so one could make such a nose by cutting a 1.6 mm PCB and gluing two pieces together. i like easy prototyping :)21:20
wpwrakroh: the atusd pcb would probably need a dent to help with positioning the nose. a hole would be better but may eat too much into the ground plane21:21
wpwraksilk on one side is enough. and i managed to shave 1.4 mm off the board's length :)21:25
rohwpwrak: i see21:33
qi-bot[commit] Werner Almesberger: dirtly made the atrf tools run with the new atusd board http://qi-hw.com/p/ben-wpan/ebe667122:14
qi-bot[commit] Werner Almesberger: atusd.brd: moved SLP_TR via a bit further from the chip http://qi-hw.com/p/ben-wpan/bf3189f22:14
qi-bot[commit] Werner Almesberger: atusd.brd: re-layout to slightly reduce size, and added more text http://qi-hw.com/p/ben-wpan/62994da22:14
wpwrakhmpf. orthography ...22:15
wpwrakso, if all testing goes well, this will be more or less the final circuit and layout. i may still shave off some of the ground plane left of the antenna and perhaps make that dent for the nose.22:19
Action: kristianpaul git pull22:19
wpwraki realize that i should create a setup that lets me determine the directional characteristics of the antenna. at least close to the receiver, it changes quite a bit even for small movements.22:20
kristianpaul40ppm do you measured that?22:22
rohwpwrak: heh. if you send me boards i can make you some acryllic lasercut holder to rotate it without having metal parts or so near it22:22
kristianpaulis a small (in components) design even with Xtal, i like :-)22:24
kristianpaulhe, wonder if i can attach it to my sky atenna .. and see how long it goes :-)22:24
qi-bot[commit] Xiangfu Liu: config.full_system-add-more-php5-modules.patch http://qi-hw.com/p/openwrt-xburst/15de82922:25
qi-bot[commit] Xiangfu Liu: config.full_system: add package sqlite2-cli http://qi-hw.com/p/openwrt-xburst/caf02b822:25
kristianpaulwpwrak: is it working the diff tool for schmatics?22:25
kristianpauli want see diffs :-)22:26
qi-bot[commit] Xiangfu Liu: feeds.conf add @revision to feeds.conf http://qi-hw.com/p/openwrt-xburst/b2cc65922:29
wpwrakkristianpaul: shold work .. lemme check ...22:35
wpwrakkristianpaul: yup. http://projects.qi-hardware.com/schhist/atusd/22:35
wpwrakkristianpaul: (40 ppm) of course i did. http://projects.qi-hardware.com/schhist/cntr/22:35
kristianpaullooks nice :-)22:36
wpwrakroh: thanks :) but i think it's a lot faster if i cut a piece of wood ;-)22:36
kristianpaulbtw how do you cut wood? i mean code..22:38
wpwrakl8r .. gotta run22:40
qi-bot[commit] Xiangfu Liu: add the MAINTAINER to package http://qi-hw.com/p/openwrt-packages/40912c622:41
--- Mon Jan 17 201100:00

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