#qi-hardware IRC log for Friday, 2011-03-04

wpwrakroh: in atben and atusb, there's an optional capacitor for antenna tuning that may or may not get equipped in the final product. for your boards, what would be your preference for having it equipped ? strong yes, strong no, or don't care ?00:15
larscgrml... this seems to be impossible00:17
wpwrakroh: the capacitor is 0402 and, if equipped, would be something around 0.5 pF00:17
larscwould probably be fast if i'd learn gcode and write the code for the board by hand00:19
wpwrakremember: each typo is equivalent to an a-few-percent-chance of a broken tool :)00:20
wpwrakaw: in atben and atusb, there's an optional capacitor for antenna tuning that may or may not get equipped in the final product (i still have to examine what works better)00:48
wpwrakaw: for your boards, what would be your preference for having it equipped ? strong yes, strong no, or don't care ?00:48
wpwrakaw: the capacitor is 0402 and, if equipped, would be something around 0.5 pF00:48
wpwrakaw: (my preference for your preference would be "don't care" ;-)00:52
awwpwrak, what's the goal of 'optional' capacitor on functionality?01:05
wpwrakaw: it's for antenna tuning01:05
awjust for ant tuning?01:06
wpwrakaw: yup. it's between the RF feed line and ground01:06
awwpwrak, hmm..eventually it may not be equipped? you hope? am i see clear difference on power performance measurement if it equipped but it's not?!01:09
wpwrakaw: i'm hoping that it may help getting rid of some little gremlins01:09
wpwrakaw: here's a full rf analysis of one board (with an improper test setup, so some things will change): http://downloads.qi-hardware.com/people/werner/wpan/20110303/01:10
wpwrakaw: as you can see, there are a lot of little spikes at around 0.3 MHz from the center. would be nice to get rid of them.01:11
wpwrakaw: (click on the small images to magnify)01:12
awokay, so those little spikes around 0.3Mhz is relevant to that 0402 optional capacitor? or not that meaning?01:14
awif that optional capacitor is not equipped, which is that we can easily know the frequency domain shown is wrong!?01:15
wpwrakaw: i think the cap may remove them, yes. but this still needs some experimenting01:15
wpwrakaw: my question is about the prototypes i'm about to send out. i don't want to delay them for this. i think i can examine the problem while the other boards are on their way01:16
awhm...sounds that you had have ideas on distinguishing what evidence will be shown when 'optional' capactot is missed. good.01:16
wpwrakaw: (if not equipped) i don't even know if you notice those spikes in the overall performance01:17
wpwrakaw: but we can figure that out later :)01:17
awwpwrak, it's ok..in parallel! yes..later. :-)01:17
wpwrakaw: for now, i'll collect all the data i can get with the existing boards, then decide who gets which board, send yours (via fedex) and roh's (by regular mail), wait until roh's clear customs, then send richard's (also by regular mail)01:18
wpwrakaw: ideally, yours and roh's would go out tomorrow, so that they can travel during the weekend. i hope customs don't delay this.01:19
awwpwrak, okay..please may just declare them as 'development samples' or 'kits' or so...no 'RF' words or so.01:21
wpwrakaw: yeah ;-)01:22
awwpwrak, thanks, :-)01:22
wpwrakaw: USB adaper "ATUSB", SDIO adapter "ATBEN", interface cable "ATBEN-PGM"01:22
wpwrakaw: and not "Microwave RF transmitter with AES encryption" ;-))01:23
wpwrakor they'd think it's military equipment ;-)01:24
awwpwrak, i think that you may not mention those words on description column, just like 'engineering samples' or 'development samples'. not need our internal model name now. ;-)01:25
awi did always do that way. no even formal 'model name' mentioned...;-)01:26
awwpwrak, yeah.01:26
wpwrakthey always ask for a "detailed description". i had been haggling with TNT endlessly when sending those counterweights to tuxbrain. they were not happy with just "counterweight (metal)"01:27
awwpwrak, umm...well-controlled in your country. well..do that you generally did before, just no 'RF'-related words...;-)01:28
wpwrakso i'd include the model name to make it look very "detailed". if they google for it, they'll only get confused, so it doesn't tell them much :)01:29
wpwrakaw: no RF, no microwave, no encryption, not "PCB" (Polychlorinated biphenyl) ;-)01:30
awwpwrak, okay. right exactly. good.01:30
wpwrakharmonized system codes 847330 and 85444201:32
awu need to prepare that whenever you send out from your country?01:34
wpwrakaw: i think someone has to - e.g., i or the courier01:35
awvia fedex, you need to prepare/determine them by yourself?01:35
wpwrakaw: and it's easier for me to look up the code than for be to explain them what exactly it is :)01:36
awprobably their computer system will always do such things internally. ;-)01:36
awwpwrak, okay.01:37
wpwrakaw: well, after my experience with TNT, i'm a bit cautious. they asked for a description, i sent it by mail, they asked me again when i scheduled the pickup, i put everything into the documents as well, then it was picked up but stuck at customs. then i called them, had to explain a third time, then they asked for me to send them an updated formal invoice with slightly changed wording, until they finally processed it.01:38
wpwrakaw: then, when it reached spain, it was held in customs too, apparently because there was more confusion about the content.01:39
wpwrakaw: i never had such troubles with fedex, but i think it can't hurt to help make things go smoothly01:39
awwpwrak, for the same goods in different countries import and export control practices vary. okay if you think that'll be smooth expectedly.01:42
xiangfuwpwrak: when I run aclocal;automake;./configure --prefix=/ --host=mipsel-openwrt-linux;make, the avrdude stop at "config.status: creating avrdude.conf.tmp". the make output: http://pastebin.com/BGK4FePN01:46
xiangfuwpwrak: full build message : http://pastebin.com/AG4Q1U4k01:48
wpwrakaw: i think fedex are more enlightnened than others, but still ...01:50
wpwrakxiangfu: yeah, something is wrong there. you need to run  make </dev/null  then run configure again01:50
wpwrakxiangfu: oh wait. maybe that's something noew. on openwrt, i didn't have that problem01:51
wpwrakxiangfu: maybe just try the make </dev/null and configure, just like on jlime01:52
wpwrakxiangfu: the symptoms look the same and i don't know why it worked for me without this hack on openwrt (or why it didn't work without it on jlime :)01:53
wpwrakroh: i'll also need your shipping address. hope regular mail is okay.01:55
awwpwrak, agreed. ;-) enlightnened means needing more processes. well..thanks that gift.01:56
xiangfuwpwrak: try to add </dev/null to make. testing01:56
wpwrakaw: (declarations) one problem is also that these things change from time to time. so if you could just write "gift" some years ago, they may now want a lot more details. perhaps also to check against the various lists of prohibited items (like terror-toner :)01:57
Action: kristianpaul checks sal02:01
wpwrakkristianpaul: i think the fft or some of the data acquisition is broken. i did some fft on samples obtained by another means (for the same kind of signal), and the results look much better.02:03
wpwrakkristianpaul: so this still need investigating ... at some point in time :)02:03
kristianpaulwpwrak: I dont seen a function related with coomand line arguments, how do you loaded the the "other means samples", ah you mean still using usrp2 for that i guess02:07
wpwrakkristianpaul: yes, it;s all usrp2. also, sal has a number of things hard-coded (such as the center frequency), because i haven't figured out yet how to change it after initializing the ustp202:10
kristianpaulstill nice if it if works for your requirements :-)02:12
wpwrakkristianpaul: the other one works fine: http://downloads.qi-hardware.com/people/werner/wpan/20110303/02:14
wpwrakwolfspraul_: good news (if you haven't spotted it in the logs yet): tuxbrain's UBB production run was successful.02:34
qi-bot[commit] Xiangfu Liu: avrdue: compile fine now http://qi-hw.com/p/openwrt-packages/61c7a8e02:35
qi-bot[commit] Xiangfu Liu: avrdude: apply qi_lb60-patches when build for nanonote http://qi-hw.com/p/openwrt-packages/bd3454902:35
wolfspraul_tuxbrain_away: wpwrak great, congratulations!02:35
wolfspraul_nobody realizes how  much risk is in those runs. in hardware you make a mistake, and the money is irretrievably gone.02:35
wolfspraul_in software you just reboot after a crash :-)02:35
methrilwolfspraul_, hi02:36
xiangfutuxbrain_away: http://downloads.qi-hardware.com/people/xiangfu/packages/avrdude_5.10-1_xburst.ipk. with werner's patch02:37
wolfspraul_methril: hi! good to hear from you02:37
wolfspraul_how is Brazil? :-)02:37
methrilwoakas, nice!!02:37
wpwrakxiangfu: kewl. thanks !02:37
xiangfutuxbrain_away: please test it. give some feedback .02:37
methrilwolfspraul_, i have less time that i would like... but is nice02:37
wpwrakmethril: hehe, caught by wo<Tab> ;-)02:37
xiangfuwpwrak: all the patches are from you. thank you :)02:37
methrilwpwrak, yes02:38
wolfspraul_methril: so, what's up? You want to jump full power into Milkymist One?02:39
wolfspraul_must be...02:39
methrilwolfspraul_, not full power, but i would like to hack something in the long term ;)02:39
kristianpaulright, lets flash this thing again03:25
qi-bot[commit] Xiangfu Liu: kbd: install more keymaps http://qi-hw.com/p/openwrt-packages/a1df6bb03:26
wolfspraul_xiangfu: I had this idea that if we reorganize the partition layout again one day, we may want to leave a separate 128 mb partition as well. the reason is that I'm thinking about encrypting partitions.03:29
wolfspraul_so there are 3 options:03:29
wolfspraul_1) the user doesn't care about encryption - everything stays like it is now03:29
wolfspraul_2) the user wants to encrypt everything, including rootfs, and doesn't mind enterning a password on every bootup03:30
wolfspraul_3) the user doesn't want to enter a password for every bootup, but wants to have an encrypted partition for certain files, and enter a password upon mounting that partition03:30
wolfspraul_for case #3, it would be a good idea to leave something like a 128 MB partition around...03:30
wolfspraul_just thinking, and see what people feel about it03:31
wpwrakwolfspraul_: you could also just loop-mount a container file03:31
wolfspraul_good point, maybe even better. don't know what is most easily implemented, I don't know much about encrypted file systems, partition, loop mounted blobs, etc.03:32
wolfspraul_I need to experiment with encryption a bit on my notebook, then I know better :-)03:32
wpwrakwolfspraul_: man losetup :) it's ancient03:33
wolfspraul_oh that I know03:33
wolfspraul_but how people actually, practically, encrypt partitions/filesystems I don't know03:33
wolfspraul_do you use encrypted filesystems?03:33
wpwrakno, i'm too lazy to worry much about such things03:34
wolfspraul_ok but I think for the Ben we should make it very easy to go secure03:34
wolfspraul_the details of which would probably need some thinking03:35
wpwrakeasy+secure is a good combination03:35
kristianpaul(encrypting partitions) wee !03:35
kristianpaul(loop-mount a container file) i thinks thats one openbsd handle ecryption..03:36
kristianpaulencrypt a whole system, requires work as wpwrak points, but a file, why not?03:37
wolfspraul_kristianpaul: seems like you also haven't used encrypted filesystems before. we need to find someone who has done it for a while and can give us some good guidance on how to start.03:39
wolfspraul_or if we don't find someone, we need to try ourselves, and see what works well03:39
kristianpaulwolfspraul_: I did03:39
wolfspraul_loop mounted fs sounds better than separate partition to me, so that's definitely good already03:39
wolfspraul_kristianpaul: great! how did you do it?03:40
kristianpaulwolfspraul_: but was too much work to maintain.03:40
kristianpaulwolfspraul_: using debian03:40
wolfspraul_you entered a password to access the partition? was it a partition or loop mounted something?03:40
wolfspraul_why was it too much work to maintain?03:40
kristianpaulllvm  + some crypto-something03:40
wolfspraul_how did you access the encrypted storage?03:40
kristianpaulyeah password when booting03:40
kristianpaulonce that all transparent03:40
wolfspraul_so every time you booted you had to enter the password03:40
wolfspraul_how about coming up from suspend?03:41
kristianpaulhmm i never use suspend..03:41
kristianpaulmuch work because something my computer restart, but i'm not at home..03:41
kristianpaulsecond i break my debian from time to time, so re-install take long03:42
wolfspraul_it was a computer you left somewhere?03:42
kristianpaulas the "formating" take some hours03:42
wolfspraul_with a mobile device you always carry with you, the need to enter a password should be less of an annoyance03:42
wolfspraul_of course it should be easy to turn this off if someone wants to remote operate a nn03:42
kristianpaulno, computer is at home, but i'm not there all time in the week03:42
kristianpaulI did with a file onces, i experimented with openbsd03:44
kristianpaullet me find doc03:44
kristianpaulis in spanish but you got the idea http://structio.sourceforge.net/guias/usuario_OpenBSD/configuracion-de-algunos-dispositivos.html#imagen-encriptada03:46
kristianpaulalso i think you need enable "crypto " suppor in linux03:46
kristianpaulrafa told me once, i asked this topic to him time ago for jlime03:47
xiangfuI have no idea on encryption partition. never use that before :(03:50
wolfspraul_we can find out slowly. it does make sense for a device like the Ben though, and it should be very easy to setup and use, or even come with a good default installation.03:54
wolfspraul_I will experiment on my notebook a little, see how that goes.03:55
kristianpaulwhole re-install?03:58
Action: kristianpaul reads a pdf on his milkymist04:01
kristianpaulhe, feels a bit slows when moving to next page tought..04:02
tuxbrain_awaywolfspraul wolfspraul_ have you seen the UBB pictures? :) from http://en.qi-hardware.com/wiki/File:UBB_production_01.JPG to http://en.qi-hardware.com/wiki/File:UBB_production_08.JPG08:24
wpwraktuxbrain_away: the pictures blew him away ;-)08:31
wolfspraulno just my 2nd identity08:31
wpwrakwolfspraul: ah, sneaky ;-)08:31
tuxbrain_awayso you achived my desired cloning machine! :p08:31
wolfspraulwow yes, nice!08:32
wolfspraul02 looks like world domination08:32
tuxbrain_awayhehehehe yeah I have chose it for the blog post :)08:32
wolfspraulwelcome to the world of hardware manufacturers!08:33
wolfspraulof course now you have to sell all that stuff...08:34
tuxbrain_away50% already sold :)08:34
xMffI've a feature request08:36
xMffI was surprised that the official images have the ks7010 driver already included, thats nice08:36
xMffbut can you disable its debugging?08:37
xMffit is spamming the terminal and actually has quite some performance impact08:37
wolfspraullars committed something along these lines just a few days ago08:38
xMffI see08:38
xMffsome kind of network manager would be nice :)08:39
xMffbut I suppose wpa_gui could be already enough, have to give that a try08:39
wolfspraulwhat gui toolkit is wpa_gui using?08:44
wolfsprauland 320x240 is enough screen space for it?08:45
xMffthat is another question :)08:45
xMffI think its not08:46
xMffat least the settings screen can get quite big08:48
xMffthe scanlist and statusview probably would fit08:48
kyakxMff: does WPA2 work fine for you with ks7010?09:43
xMffkyak: I didn't test recently09:43
wpwrakwolfspraul: ah, pricing. what would be a common factor between BOM cost and final price for boards with BOM costs in the USD 5-10 range ? (qty 100)10:03
wpwrakwolfspraul: i.e., atben/atusb10:03
wolfspraulnormally you do times 310:03
wpwrak(final = end customer)10:03
wpwrakgood. i had guessed 300-400% :)10:03
wolfspraultimes 3, where 1/3rd is the distributor/reseller, 1/3rd the manufacturer, 1/3rd the components/pcb/smt/assembly10:04
wolfspraulthat's just the ballpark, with our extremely low volumes everything is different, but of course we cannot offer boards for 200 USD...10:04
wpwrakokay, so more like 400% from the raw BOM cost. (well, PCB isn't in there either)10:04
wolfspraul(with a bom of 10 usd)10:04
wolfspraulsure, there are details, such as whether you have a difficult assembly/testing or not, and what the (known) yield is10:05
wolfspraulsometimes a manufacturer knowingly produces something with difficult testing, 60% yield, etc.10:05
wolfspraulbut then he better have a margin to survive through this chaos10:06
wolfspraulthe end customer won't care, they just want a functioning product10:06
wpwrakheh ;-) let's hope we don't run into such yields10:06
wpwrakatben looks pretty good for DIY yield. atusb is a bit of a slaughterhouse, but that's a) because it has large ground planes which make manual soldering and particular the vias hellish, and b) i have some issues with the tin chemistry, which make them even harder to solder.10:09
wpwraklemme update the results to illustrate this ...10:10
wpwrakat the end, atben-20110219-* are the five final atben boards10:12
xMffkyak: do you havei ssues with it?10:13
kyakxMff: yeah, had some issues.. In fact, i was only able to make WEP work. The some time later it didn't work at all :)10:14
xMffno, afaik it definitely worked with wpa10:14
xMffyou need to configure the supplicant though10:15
wpwrakat the top, atusb-4 through -7 are three of the five final atusb boards. the other two don't want to enumerate at the moment. maybe they need just reflashing.10:15
kyakanyway, i didn't have luck at that time -\10:15
kyak(and now i don't have the card, it's mechnically broken)10:16
xMffexpensive piece of hw to break10:16
kyakyes, pretty interesting how expensive and shitty can be combined together :)10:16
wolfspraulwow that's a chart (werner's antenna performance thing)10:34
wpwrakwolfspraul: kinda tricky to fit it all on a screen and still keep it readable :)10:41
wpwrakthe 8:10 card receptacle really want changing in a future nanonote ... just tried with a few more atusb-pgm. brrgh.11:09
qi-bot[commit] Lars-Peter Clausen: allow variable timer irq and timer id http://qi-hw.com/p/qi-kernel/d8312a112:21
qi-bot[commit] Peter Zotov: MIPS: JZ47xx: Fix a few naming issues. http://qi-hw.com/p/qi-kernel/fad288a12:21
qi-bot[commit] Peter Zotov: MIPS: JZ47xx: Moved jz4740 headers to a subdirectory of mach-jz47xx/. http://qi-hw.com/p/qi-kernel/ac7a6e412:21
qi-bot[commit] Lars-Peter Clausen: Split time.c into timer-cevt.c and timer-csrc.c http://qi-hw.com/p/qi-kernel/87c1d8512:21
qi-bot[commit] Lars-Peter Clausen: Generalize gpio support http://qi-hw.com/p/qi-kernel/1a111cf12:21
qi-bot[commit] Lars-Peter Clausen: Move 2nd level IRQ defintions to sub-arch specific files. http://qi-hw.com/p/qi-kernel/b57516e12:24
wpwrakroh: ayt ?13:34
wpwrakM_Rojas: i need a reality check: are there any holidays in the next days ? on the radio, they mentioned a "long weekend", but i can't seem to find any "official" holidays13:34
mthwpwrak: here in .nl we have carnaval, which is always 6 weeks before easter13:48
mthit's not an official holiday here though13:48
mthbut a lot of people take days off13:48
wpwrakmth: yeah, i see carnival listed as unofficial here as well. but what's coming is an official holiday. probably one of the populist holidays decreed late last year, so it's not in any printed calender yet.14:14
qi-bot[commit] Lars-Peter Clausen: Don't make any assumptions about which timers are used as sytem timers http://qi-hw.com/p/qi-kernel/d9ee7e917:34
qi-bot[commit] Lars-Peter Clausen: GPIO: Replace leftover JZ47XX_IRQ_GPIO(0) with jz_gpio_irq_base http://qi-hw.com/p/qi-kernel/1bbb2ee17:34
qi-bot[commit] Lars-Peter Clausen: Add support for more then one irq bank and move initialization into sub-archs. http://qi-hw.com/p/qi-kernel/7437e2217:34
qi-bot[commit] Lars-Peter Clausen: Add very basic jz4760 support http://qi-hw.com/p/qi-kernel/1768b0017:34
wpwrakroh_: ayt ?19:08
qi-bot[commit] Peter Zotov: MIPS: Add basic JZ4750 support. http://qi-hw.com/p/qi-kernel/b9dc8f819:35
qi-bot[commit] Peter Zotov: MIPS: JZ4750: Add workarounds for CP0 STATUS. http://qi-hw.com/p/qi-kernel/2308d1119:35
wpwrakmh ?21:14
kyaki wonder why suddenly the toolchain and everything have changed to "mips"21:14
kyakCONFIG_ARCH="mips" instead of "mipsel" - for some unknown reason...21:15
wpwraksound like fun :)21:16
kyaki copy the config.full_system and do yes "" | make oldconfig > /dev/null21:17
kyak$ env |grep ARCH21:18
kyaki was playing with qi-kernel21:18
kyakin this shell21:18
kyakheh, at least now i know that openwrt knows this env variable21:18
qi-bot[commit] kyak: gmenu2x: update to latest git http://qi-hw.com/p/openwrt-packages/5b28efc21:23
rohgoogle's self driving car... is mad.22:53
qi-bot[commit] Lars-Peter Clausen: Use common base address file for all sub-archs http://qi-hw.com/p/qi-kernel/6ab3c0623:21
qi-bot[commit] Lars-Peter Clausen: Add OST support http://qi-hw.com/p/qi-kernel/2c4cbdb23:21
Jay7larsc: ping23:22
larscJay7: pong23:22
Jay7larsc: have you seen here kyak's investigations about kexec?23:22
larscJay7: no23:22
Jay7wrt adding add_memory_region e.g.23:22
Jay7in short seems we need some kernel modification to make kexec working23:23
larscwhat kind of modifications?23:32
larsci've never worked with kexec before23:32
qi-bot[commit] Lars-Peter Clausen: Use common base address file for all sub-archs http://qi-hw.com/p/qi-kernel/6871d0723:37
qi-bot[commit] Lars-Peter Clausen: Add OST support http://qi-hw.com/p/qi-kernel/184d49123:37
Jay7larsc: sorry, can't say right now23:44
Jay7larsc: I'll remind you again later :)23:45
--- Sat Mar 5 201100:00

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