#qi-hardware IRC log for Monday, 2011-06-20

kristianpaulhum, how to reverser a downcovertion03:01
kristianpaulas the current state of namutu correlator seems to not be happy with SiGE serial out put, and in the case i re-reverse the serialization, does it mean is all what i needed to do?03:02
kristianpaulwell, i guess taking just the sign will be just be enought03:17
kristianpaul1 bit for now.. then03:25
kristianpaulthis guys is awesome http://www.s5tech.net/s53mv/navsats/04:03
kristianpaul1 bit ADC madewithh a FF04:04
kristianpaulis multiplexing the DSP (not for mobile, but still smart move)04:05
kristianpauluses  a LUT for correlator C/A .. 04:06
DocScrutinizer1 bit ADC is simple04:17
DocScrutinizerthat's why you use it ;-)04:17
tuxbrainI can stay connected briefly but two things:08:30
tuxbrainor maybe more :P08:31
tuxbrainwpwrak: http://www.tinyos.net/ sais it has a full stack of 6LoWPAN/RPL/CoAP stack may be worth the effort to take a look 08:32
kyaktuxbrain: have you checked if it is possible to order now with Fedex to Russia?08:36
tuxbrainAfter a brief meeting with an Tecnician of a RF (Wifi,BT...) integrator company, that I convice him to take a look at atben/atusb he advice me that whatever coverage(silicone, resine, pvc, even paper or wood) we put in the antenna will affect it badly his proposal is to use resine in mostly all the device but the antena part.08:37
kyak(before you disappear again :) )08:37
tuxbrainkyak, yes fedex is more reliable and at least as "cheap" UPS, but I have not time yet to upload the fedex info to the shop meanwhile we can arrage it via proforma and email.08:40
kyakshould i just order via tuxbrain.com and then we will continue by e-mail?08:41
tuxbrainkyak , how do you plan to pay? bank transfer? credit card? paypal?08:43
DocScrutinizertuxbrain: (antenna & resin) good advice08:43
kyaktuxbrain: it's paypal08:43
kyaktuxbrain: but i need to know the cost beforehand08:43
DocScrutinizertuxbrain: we thought you collapsed under the zillions of orders after /.   :-D08:43
tuxbrainsend me a mail with the items you want to buy and I will send you a proforma invoice with the prices08:44
wpwraktuxbrain: (partial resin cover) okay, that would also work. a bit fragile, but possible. did you have a chance to test the cured sugru ?08:44
kyaktuxbrain: ok, that i will do08:44
wpwraktuxbrain: oh, and please don't forget to send me the tracking number when you ship my devices08:45
tuxbrainDocScrutinizer: No not so , I have collapsed buy zillion of things to do apart of attend de some "atben/atusb" orders, + some questions rised + some proposals :)08:45
DocScrutinizerwpwrak: tuxbrain: my suggestion would be along the path of shoving PCB into some bag like thing, like shrink tube, or similar08:46
tuxbrainwpwrak: tracking sure08:46
tuxbrainwpwrak: (sugru test) not yet sorry, no any time window now to perform such test a pair of weeks.08:47
DocScrutinizeryou could build cases from two pieces of flat plastic sealed at the edge around with silicon or sth08:47
wpwraktuxbrain: btw, has the excellent press coverage rejon triggered led to any sales pickup so far ?08:48
wpwrakDocScrutinizer: very inexact :-(08:48
DocScrutinizergo for acrylic that you can get in any custom profile you order08:48
DocScrutinizerget a flat U profile, cut two pieces, glue together to get a nice rectangular tube08:49
wpwraktuxbrain: (tinyos) they seem to target devices that are still fairly big. e.g., an ATmega128. ours is a 32 (already the biggest model in that series)08:49
DocScrutinizerplace a chork on top 08:49
DocScrutinizeryou can get a meter of such profile for a few $$ at specialized shops08:50
wpwrakDocScrutinizer: sounds like a lot of work08:50
DocScrutinizersure not more work that messing with sticky resin08:50
wpwraktuxbrain: feel like cutting 2 * 100 u-profiles ? :)08:50
wpwrakDocScrutinizer: the idea is to outsource the work to the customer ;-)08:51
DocScrutinizerwait, 2*100 < 2^808:51
DocScrutinizerI'd cut that in like 15 min for you08:52
wpwrakDocScrutinizer: what kind of equipment do you have that can cut through 7 layers of profiles ? laser swords ? ;-)08:52
DocScrutinizerno problem08:52
tuxbrain(press coverage) I don't think it rise sales directly but @firefox and @JohnFMoore are following @tuxbrain on twitter :) , yes I believe a lot more people now know what qi-hardware is and what we are looking for, no is time to no disapoint the real target (fredoom lovers) the MS and Ithing lovers are lost from the begining :)08:53
DocScrutinizerbuzz saw08:53
DocScrutinizer(hail ding)08:53
DocScrutinizertuxbrain: indeed :-D08:54
wpwraktuxbrain: hehe, kewl. let's hope it also has an effect on sales. no money, no fun. well, unless your grocery store also becomes one of your follower and they let you have your food for free ;)08:55
DocScrutinizerdoesn't fly, tested that08:55
wpwraktuxbrain: (lost from the beginning) yes, no need to worry about those iDiots ;-)08:55
tuxbrainwpwrak: (tinyos) but the 6LoWPAN/RPL/CoAP stack cannot be a start point to port it to linux?08:56
DocScrutinizer(case) what's about revambing dirt cheap usb keyring dongles?08:56
kyaktuxbrain: i just sent the e-mail08:57
tuxbrainat least is the only place where I see it announced clear as a working implemented 6lowpan stack08:57
wpwrakDocScrutinizer: roh looked into that. didn't find anything that fit08:57
tuxbrainunder bsd licence08:57
wpwraktuxbrain: (tinyos) maybe08:58
DocScrutinizera task we need to outsource to our taipei branch of community ;-D08:58
DocScrutinizerwpwrak: you know how Pustefix soap bubbles came to market? They recycled WW-2 ammo, iirc the empty brass tubes left over from flak09:00
tuxbrainDocScrutinizer: dont worry about cover , if resin is the way to go we have some companies here that can do the job, and also Victor is quite good in such kind of manualities to do some tests, I this is more mater to have atusb working with standard 6lowpan than cover it or not that will make it a real success :)09:00
wpwrakDocScrutinizer: i think we just need to see how things go with simple substances that can act as a cover. while they certainly have some effect on the RF, it's not as if plastic-covered antennas  were unheard of.09:00
DocScrutinizerno, just they get detuned. Different dielectic constant09:01
tuxbrainwpwrak: As the RF guys tellme, but they where designed with the pastic cover in mind :)09:01
wpwrakDocScrutinizer: hmm, we could try brass tubes. they ought to produce very predictable results :)09:01
DocScrutinizerremember the advice about PCB material and thickness?09:01
DocScrutinizerwpwrak: indeed ;-P09:02
wpwraki guess we'll just have to try.it's not as if the antenna was perfectly tuned anyway. so it may actually improve it ;-)09:02
DocScrutinizeras somebody said yesterday he wants a N900 case of massive gold. I answered it will serve perfectly for the intended purpose09:03
wpwrakDocScrutinizer: like a rolex ;-)09:03
DocScrutinizermy words09:03
DocScrutinizer[2011-06-20 03:39:47] <DocScrutinizer> have you ever heard somebody complaining about weight of his Rolex, or about it not being accurate ;-D09:04
DocScrutinizerin a Rolls Royce they got the clock for sole purpose to listen to the ticking and muse about how silent the car is09:06
DocScrutinizerit doesn't really matter if it shows the correct time09:06
DocScrutinizer(the car clock, not Rolex)09:07
wpwrakanyway, the evil day star will raise soon. time for nap :)09:08
DocScrutinizerhey, I missed that point by far09:08
wpwrakggaaah. s/ride/rise/09:08
DocScrutinizerpondering siesta ;-)09:09
Action: DocScrutinizer just idly muses about his feeling facing hardware under a resin blob - depressing11:29
qi-bot[commit] Maarten ter Huurne: MIPS: JZ4740: USB: Fix packet read/write functions. http://qi-hw.com/p/qi-kernel/997934512:09
mthlarsc: thanks12:16
[g2]public logging sweet!14:21
[g2]Are the gerbers around for the atbin  ?14:21
[g2]doh... s/atbin/atben/14:21
jluisg2 look at http://downloads.qi-hardware.com/people/werner/wpan/fab/atben-pcb-110330.zip14:36
wpwrak[g2]: if you just want an overview, you may find this one more useful: http://downloads.qi-hardware.com/people/werner/wpan/tmp/atben-110324-overview.png14:43
[g2]jluis and wpwrak thx... Actually I was thinking about  have PCBs sent off today14:44
wpwrak[g2]: if you want to edit things, the kicad design files are here: http://projects.qi-hardware.com/index.php/p/ben-wpan/source/tree/master/atben/14:44
wpwrak[g2]: then the zip jluis mentioned is exactly what you want :)14:45
[g2]wpwrak, cool I'll have a look.  I'm also thinking of combining the Femtoduino and atben on one board. 14:46
wpwrak[g2]: sounds good. some people here have already been thinking of such a union. (although more in the sense "wouldn't it be nice if someone did this", not in the "let's do this" sense)14:47
[g2]wpwrak, I'm building Femtos so it's the same QFN packaging14:51
[g2]wpwrak, thx for all you work in this btw :)14:51
wpwrak[g2]: glad you like it and even consider extending it ! :)14:52
[g2]wpwrak, open hw and sw are the best IMHO14:53
wpwrakagreed :)14:55
[g2]wpwrak, have you thought about a U.FL connector at all for the boards ?15:00
wpwrak[g2]: of course :) if you remove the 0R resistor, scatch off a bit if the solder mask and a bit of the copper, you can solder an U.FL there. it's most likely to be a one-way modification, though15:03
wpwrak[g2]: like the atusb with the external antenna here: http://downloads.qi-hardware.com/people/werner/wpan/rflab-atusb.jpg15:05
wpwraklet me find a better picture ...15:05
wpwrakthis one has two boards with an U.FL: http://downloads.qi-hardware.com/people/werner/wpan/tmp/boards-so-far.jpg15:06
wpwraknaw, even three ;-)15:06
wpwrakthey're a bit earlier designs, but the modification should be similar for the production boards15:07
[g2]are you etching your own boards ?15:10
[g2]also, did you know that gerbv is not happy about the drl file ?15:12
wpwrak[g2]: (etching) yup15:12
[g2]cool... do you know about Laen's services ?15:12
wpwrakcan gerbv actually display excellon ? .drl isn't gerber15:12
wpwrakwho is laen ?15:12
wpwrakah yes, seen that one. such aggregators may be nice for final runs before going to the fab15:15
wpwrakfor prototyping, probably too slow and the shipping to expensive15:15
[g2]right, he does "medium" runs once a month15:15
wpwrakfor really simple stuff, i have turn-around times of a few hours :) RF boards are bit of a pain, though, with all those vias15:17
wpwrakalso, he doesn't say if he does 0.8 mm boards15:18
[g2]wpwrak, you mean thinner boards ?15:33
wpwrak[g2]: thinner than the usual 1.6 mm. e.g., atben/atusb both are 0.8 mm.15:34
wpwrak[g2]: a thicker board wouldn't fit into the 8:10 card holder. also, the RF traces (feed line and antenna) are designed for 0.8 mm, not 1.6 mm15:34
wpwraksee also README-PCB15:37
[g2]thx... sorry about all the silly questions, I just found out about the atben/atusb last night (via hackaday)15:40
wpwraknaw, they're good questions. you're the first one besides myself and then tuxbrain to make such boards15:42
wpwrak[g2]: if you plan on using DorkbotPDX (that is, if they can do 0.8 mm), the drill sizes will be okay. theirs is bigger than mine, but my vias are conservatively sized. no problem with trace width and trace-to-trace clearance. copper to edge clearance is several times a bit lower than their 15 mil, but you may accept the risk. the only area where it really matters is at the antenna, where you have about 11.5 mil before chopping off copp15:48
wpwrak[g2]: if you plan on making an atusb, you can also try an 1.6 mm board. RF performance is likely to suffer, but the USB connector will actually like the change :)15:49
wpwrak[g2]: for atben, you have no choice. anything much thicker than 0.8 mm won't go into the slot15:49
[g2]wpwrak, I won't be going into the slot.. I'll be connecting up to the ATMega32815:51
[g2]the atben is like the RFM12B module except has AES and other stuff15:51
[g2]and fully open  (except the chip itself)15:52
wpwrak(atmega328) ah, okay. then you only have to worry about RF performance if using a different board thickness. i don't know what sort of degradation to expect. most likely, you'll get some reflections on the then mismatched feed line, plus the antenna may show some frequency-dependent attenuation15:54
[g2]wpwrak, right I'll have to look closer at the RF.  I've got an ubertooth so I'll be able to do some spectral analysis16:01
[g2]see some levels and stuff16:01
methril_workwpwrak, (fisl) i think it's going to be impossible to got to FISL this year :(16:23
wpwrakmethril: aww ;-( for you it's almost next door16:31
qi-bot[commit] Werner Almesberger: atusb/fw/ep0.c (my_setup): fixed test for ATUSB_SPI_READ2 http://qi-hw.com/p/ben-wpan/3f7409816:44
rjeffriesg3 are you here?17:39
rjeffriessorry... g2   sheesh17:39
[g2]rjeffries, hi17:42
rjeffriesg2 I am the person wpwrak was referring to above about thinking about as opposed to doing a standalobe board that is arduino-ish plus the 802.15.4 radio17:43
rjeffriesthe reason is that Ben Nanonote could then be a human interface for remote sensor stations connected wirelessly17:44
[g2]rjeffries, it's all good.17:44
rjeffrieshave you been commenting on the femtoduino list?I read that list17:45
Action: [g2] thinks about a lot of stuff, but it's really hard to do _everything_17:45
[g2]rjeffries, yup that's me17:45
rjeffrieswell I am not an engineer but I think about doing stuff17:45
rjeffriesso have you built up a femtoduino yet? it is so clever. almost TOO small17:46
[g2]I've got several I've built17:46
[g2]both 3.3 and 5v17:46
rjeffrieshow difficult was the home brew surface mount stuff17:46
[g2]the QFN is a little tricky. I've got a jig for holding the board, laser cut solder stencils and a temperature controlled reflow oven (arduino based with thermocouple).17:48
[g2]I need to post some photos of the jig17:48
[g2]rjeffries, I've toyed with making a TQFN version17:49
[g2]rjeffries, also, I've only used leaded solder, haven't tried the silver RHOS stuff yet.17:50
wpwrakrjeffries: for an easier start, you may want to consider TQFP/SSOP/SIOC/etc. packages. they have little feet. they're quite easy to solder. QFN need a bit more experience, so they're not ideal for a smooth first-time experience.17:51
wpwrak[g2]: (RoHS) oh you're so missing out on the best fun ... ;-)17:52
[g2]wpwrak, are you using leadfree solder ?17:52
wpwrak[g2]: yeah, reluctantly. didn't want any troubles with shipping prototypes.17:53
[g2]got a good paste you like ?17:53
[g2]Everything is RHOS but the paste for me17:53
[g2]the QFN on the Femto's and 0402 are my first dip in the pretty small self-assembly stuff17:54
wpwrak[g2]: i normally solder with an iron. not too fond of the paste i got. but i didn't run comparisons, so maybe it's okay. my oven is a bit on the simple side, so it tends to overcook the boards a bit.17:54
[g2]wpwrak, homebrew oven ?17:55
[g2]rjeffries, around later ?  I've got run out for lunch.17:58
wpwrak[g2]: (oven) just the cheapest toaster oven i could find. wanted to see if it's easy to do oven-based smt. realized that no, some work would be needed. and probably a better oven as well :)17:59
[g2]wpwrak, I use a $19US oven from Target17:59
wpwrak[g2]: anyway, 0402 don't scale me anymore. qfn can still be a pain for the center pad and if i have lots of ground nearby, like in those RF designs18:00
wpwrak[g2]: you oven out-cheaped mine ;-)18:00
[g2]It does has a thermocouple sensor inside and that and the sensor probably cost 2x the oven18:00
[g2]well the sesnor is outside, but the thermocouple is in the oven :)18:01
wpwrak[g2]: nice :) just use the oven "as-is". didn't mod or instrumentalize it in any way.18:02
wpwrakhmm, that word probably doesn't exist :)18:02
[g2]wpwrak, you probably speak 4 or more languagues :)18:03
[g2]instrument is the work you were probably looking for18:03
wpwrakyeah :)18:06
wpwrak"instrumentalize" has more of a modding vibe to it, though. shame it doesn't exist :)18:06
dvdkkyak: huh?  the problem with ase crashing on safe is gone for me now.19:54
dvdkbut the only thing i did was to install uclibc compiled from the same openwrt-xburst.git revision as I compiled ASE with.19:55
dvdkso have we been chasing a libc binary compatability problem?19:55
dvdk(well, the uclibc was compiled with debug information, but that shouldn't change anything i guess, as the debug info is only kept on the host)19:56
xMffit does change things19:56
xMfffor example I recently fixed a segfault in gethostbyname_r() which did *not* occur when built with debug info19:57
dvdkgood to know.  too lazy recompiling libc yet again :)19:57
Action: dvdk thought he read 'it does NOT change things19:57
dvdkok, so not so good.19:57
xMffwhich obviously must not mean that this is the case here19:57
dvdkok, doing another run of toolchain/libc/compile19:58
dvdkbut if it does happpen again, i'm going to get crazy.  because the crash happens inside uclibc, and without debug-info (remote) gdb is completely lost19:59
dvdki notice it asks me about "ctype argument checking"20:01
dvdkon 'make toolchain/compile'20:01
dvdkand there is an option 'Issue a diagnostic and abort()'20:02
dvdkany chance that this was enabled in the nightly build trunk images?  20:02
dvdkbecause SIGABORT is how ASE dies20:02
dvdki set it to Detect and handle appropriately20:02
dvdk(i.e. not abort)20:02
dvdkshxx! without debug information it crashes again20:14
wpwrakdvdk: that's when the wavefunction collapses20:15
dvdka heisenbug20:15
dvdkand another round of libc compilation, again with debug info just to be sure20:16
dvdkhmm, now crashes with debug information, too.  maybe it's the gdb that kept it running20:23
dvdkhmm, looks like i only recompiled libc all the time, but the libc*.ipk wasn't rebuild.  what's the target for this?20:32
dvdktoolchain/uClibc/install doesn't help20:33
dvdkturns out it needs make package/base-files/{clean,install}20:38
dvdkuups, looks like upgrading uclibc was a bad idea after all. nanononet doesn't  boot any more.  caught in a loop printing 'segmentation fault'20:40
dvdkand reflashing20:40
wpwrakdvdk: this is your opportunity to switch to glibc ;-)20:43
dvdkbaah, then i could just switch to running debian20:44
dvdkflashing with the last nightly build.  hmm.  why does it have to be so slow...20:45
dvdkalso flashing bootloader, let's hope it doesn't break20:45
kristianpaulglibc? why not to try bionic :-D20:45
kristianpaulso wen can put android word in the makerking stuff and sell more nanonotes :p20:46
wpwrakdvdk: (debian) you should run gentoo, with local builds20:47
wpwrakkristianpaul: (android) yeah, "compatible with Android Bionic libc". of the few million people who google for this every day, a few ought to get sufficiently confused that they buy a ben ;)20:48
dvdkmaybe i should go for option #3 as suggested by Emeka: just run a bare bones forth interpreter wtihout any OS.  on more libc hassle  (and no more C-software that throws sigbort at me)20:48
kristianpaulwpwrak: (confused) lol20:49
wpwrakheh. great moments of busybox #419: http://pastebin.com/ADBGUkA120:57
wpwrakquestion: what will be in "out" after this ?20:57
dvdkno please don't tell me20:58
wpwrakwith jlime's busybox, "bar"20:59
wpwrakwith openwrt's busybox, "fobar"20:59
wpwrakkristianpaul: yes, "why" indeed ;-)21:01
kristianpaulthats a fair point to re-install jlime on my nanonote :)21:02
wpwrak(well, that's with openwrt from a while ago. 2011-02-23-ish.)21:02
dvdkdon't believe him, hi's going to tell you anything to lure you away from openwrt :)21:02
wpwrakdvdk: if you have the latest openwrt installed, maybe give it a try :)21:02
kristianpauli have my modem21:03
dvdki'm looking forward to it.  if flashing weren't that slow...21:03
kristianpauli;ll give a try right now21:03
wpwrakdvdk: that that was actually aimed more at busybox than at openwrt. it's just one of those typical little annoyances it creates.21:04
kristianpauloh, god !21:04
kristianpaulroot@nano:~# cat out 21:04
kristianpauland is last stable openwrt :/21:04
dvdkwpwrak: creates much more annoyances with complex software.  like emacs parsing 'ls' output etc.21:05
wpwrakat least now i understand why my MMC driver shutting-up script fails to miserably to silence the at86rc230 driver :)21:05
dvdkbut then it's leightweight, and we only have 32mb of ram21:05
wpwraki kinda wonder if it really matters that much. these tools are still quite small and you don't normally run a lot of them at the same time21:06
xMffseems a problem with ash21:06
wpwrak(my script where i originally encountered the problem) http://projects.qi-hardware.com/index.php/p/wernermisc/source/tree/master/ub21:07
xMffecho and printf result in weird output21:07
xMffcat and /usr/bin/printf (=not shell builtin) get ENOSPC21:07
wpwraka wpan kernel has no mmc driver, so the first echo fails. then, when it tries to turn off at86rf230, the echo writes nonsense, which the kernel ignores, leaving the driver still enabled21:07
xMffI know its annoying but I suggest to use /usr/bin/printf21:08
xMffit will behave as expected21:08
xMffor /bin/echo21:08
wpwraki'll just wait for the fix. i think i now got everyone's attention ;-)21:09
kristianpaullets see21:10
kristianpaulit works !21:11
kristianpaulroot@nano:~# cat out 21:11
kristianpaulthanks for the hint xMff 21:11
dvdkok, latest snapshot boots21:13
dvdkhuh?  flashed the wrong version?21:13
xMffbusybox and/or uclibc has other weird issues as well21:14
xMffat least on openwrt21:15
xMffor at least used to have21:15
dvdkah, just the gmenu2x doesn't have my latest commits (revision not updated in its makefile i guess)21:15
dvdkand still aborting21:16
xMffI guess we want that: http://lists.busybox.net/pipermail/busybox-cvs/2011-February/032261.html21:18
xMfflooks invasive21:18
dvdkwpwrak: 'fobar' with latest openwrt-trunk, too21:20
dvdkhmm, after flashing 'ubiattach' dropped from the image21:20
dvdkcannot mount my second ubi data partition21:21
wpwrakxMff: quite heavy indeed21:23
xMffI'll see if we can backport it21:24
kristianpaulis not a ash problem then, why patch is with echo?21:25
dvdknext try at installing debug-mode uclibc on top of running system21:26
dvdkhuuhhh.. segmentation fault.21:26
dvdkok, doesn't boot AGAIN.  need to reflash again.  21:27
dvdkah, it's just gmenu2x, console still running21:27
wpwrakhmm, coreutils, indeed21:29
xMffkristianpaul: ash calls into echo_main from echo.c21:34
xMffhowever the busybox version that openwrt uses, calls writev() to output its data21:34
xMffso I think its an uclibc bug :)21:34
wpwrakah, busybox repo. then it makes sense. still a busybox problem, as i suspected21:34
kristianpaulwpwrak: jlime uses coreutils from gnu insted? or..21:35
wpwrakxMff: why don't we have this fix yet ? is openwrt behind or is qi-hw lagging ?21:35
xMffwe use bb 1.18.421:36
xMffwhich was the official stable until six days ago21:36
xMffnow its 1.18.521:37
xMffmight be even fixed there21:37
wpwrakamazing that nobody else tripped over it before21:37
wpwrak"echo" is kinda crucial :)21:38
xMffI suspect its the uclibc21:38
xMffnot busybox21:38
xMffregarding the missing ubiattach.. mirko did split the mtd utils into separate packages, so the Qo config probably needs updates21:59
xMff*Qi config21:59
--- Tue Jun 21 201100:00

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