#qi-hardware IRC log for Tuesday, 2011-03-15

wpwraktuxbrain: the 2.4 GHz band. 2405-2480 MHz, to be exact00:07
tuxbrainwpwrak: mmm in the files there is no place where the componentes the name of the componentes apears in place? how the smt vendor will know where to put things?00:10
wpwraktuxbrain: (arduino.cc article) nice ! :) and the cool picture certainly helps to make it look more polished00:11
wpwraktuxbrain: these are still only the PCB files. SMT files are another can of worms00:12
wpwraktuxbrain: also, the component references are still missing. i'm still working on getting them on the silk screen as well. alas, kicad put them at another layer, together with other things. so i first have to separate them, then merge the layer into the silk screen00:13
tuxbrainwpwrak: then I have to ask the smt vendor what format do he wish to be provided for final producto to gain a little time a  no apear like noobs?00:14
tuxbrain(what in fact we are, at least me)00:15
tuxbrainproducto-> production00:15
wpwraktuxbrain: naw, don't ask about formats. that only causes more work.00:17
tuxbrainthen what is the excuse to not send them the files for ask for a quote?00:18
wpwraktuxbrain: ask them first about the finish and tell them we're still working on the file package00:18
tuxbrainfinish? gold  vs tin?00:19
wpwraktuxbrain: for now my goal is to finish the PCB files. there, i'm almost done. then the SMT files.00:23
wpwraktuxbrain: oh, and you should ask the SMT fab if they need a stencil for solder paste or if they apply solder paste by some other means.00:24
wpwraktuxbrain: the stencil would typically be made along with the PCB.00:24
tuxbrainstencil in spanish?00:25
tuxbrainI'm sending the mail to both PCB and SMT vendor, to make clear the have to interact with each other to to arrive to agreement on details00:26
tuxbrainwolfspraul:  http://arduino.cc/blog/2011/03/14/aduino-and-nanonote-put-together/00:27
tuxbrainwpwrak: anyway what is the reason of that stencil question?00:28
tuxbrainson another gerber with where to put the solder paste I guess00:29
tuxbrainson -> so00:30
wpwraktuxbrain: yes, that's for the solder paste00:30
wpwraktuxbrain: solder paste can also be deposited with a syringe (also by a machine). so they may or may not need a stencil.00:31
wpwrakmaybe better to wait until tomorrow morning, when i have the solder paste layer up to date as well00:35
tuxbrainok, I think as first contact is enough.... sended gerbers as just orienteative for quoting, asked about gold vs tin, give them the provisional bom for 100 units, asked if stencil or injection...00:36
wpwraktsunami coming (with more to follow)00:36
qi-bot[commit] Werner Almesberger: atben/atusb: added -pcb to PCB fab archives; added "upload" target http://qi-hw.com/p/ben-wpan/fffa6e900:36
qi-bot[commit] Werner Almesberger: atben.brd: specified which components to place (for SMT fab) http://qi-hw.com/p/ben-wpan/57e59c000:36
qi-bot[commit] Werner Almesberger: atben, atusb: added polarity markers http://qi-hw.com/p/ben-wpan/970695c00:36
qi-bot[commit] Werner Almesberger: atusb.brd: specified which components to place (for SMT fab) http://qi-hw.com/p/ben-wpan/c06ed9e00:36
qi-bot[commit] Werner Almesberger: atusb.brd (B1): moved component reference which overlapped with circle http://qi-hw.com/p/ben-wpan/34d236c00:36
qi-bot[commit] Werner Almesberger: atben.brd (B1): moved component reference which overlapped with circle http://qi-hw.com/p/ben-wpan/0e9312000:36
qi-bot[commit] Werner Almesberger: atben: use mlztx to copy the version number to the silk screen http://qi-hw.com/p/ben-wpan/bdfed8000:36
qi-bot[commit] Werner Almesberger: atben: set origin for fab files to lower left corner of board http://qi-hw.com/p/ben-wpan/073675a00:36
qi-bot[commit] Werner Almesberger: atusb: use cptx to copy the version number to the silk screen http://qi-hw.com/p/ben-wpan/695f6c200:36
qi-bot[commit] Werner Almesberger: atben: set origin for fab files to lower left corner of board http://qi-hw.com/p/ben-wpan/6b3a04900:36
qi-bot[commit] Werner Almesberger: atben/README-PCB, atusb/README-PCB: various minor clarifications http://qi-hw.com/p/ben-wpan/eac6e3c00:36
tuxbrainwpwrak:  well we don't have any real smt deliverable so that paste stencils can come later with the rest00:37
wpwraktuxbrain: yup, maybe they can already produce a quote with the information. for the pcb and - if needed - stencil, certainly00:38
wpwraktuxbrain: it just has to be clear that the package will be updated again. let me bundle up the latest version ...00:39
wolfspraultuxbrain: great article, congrats! will add it to the 04-01 news draft00:41
wpwraktuxbrain: new files at http://downloads.qi-hardware.com/people/werner/wpan/fab/00:43
tuxbrainok attaching00:44
wpwraktuxbrain: use the 110314 versions00:44
qi-bot[commit] Werner Almesberger: atben.brd: changed version from 110219 to 110314 http://qi-hw.com/p/ben-wpan/44bd93600:47
qi-bot[commit] Werner Almesberger: atusb.brd: changed version from 110314 to 110314 http://qi-hw.com/p/ben-wpan/1cb69d100:47
tuxbrainok mail sended... lets see what they reply, I bet one atben the reply will be a WTF :P00:48
wpwraktuxbrain:  did you also send them a link to the picture ?00:50
wpwraktuxbrain: great. then they can probably figure it out ;-)00:51
tuxbrainthat will be the first WTF00:51
tuxbrainsecond WTF will be sending mail to both at time00:51
tuxbrainthird the aks for quoting without smt gervers00:52
wpwrakoh, they probably get much worse stuff every day ;-)00:53
tuxbrainbut well, WTF is the most usual reaction on tuxbrain first contact so I'm fine with it00:53
tuxbrainon following contacts WTF->OMG :P00:54
wpwrakhehe :)00:57
wpwraknow .. let's see how i can tame that silk screen ...00:58
tuxbrainwell dudes , a lot of stuff is moving on tuxbrain , things visible avrnote, ubbs, wpan ...and  no other no so visible my mind blowing.. time for some rest.01:00
wolfspraultuxbrain_away: n801:02
wolfspraulis the non-visible stuff good?01:02
wolfspraulI hope it's good! :-)01:02
tuxbrain_awaysome is bring money, and other is cool stuff so yeah we can cosiderate them good :)01:03
kristianpaulokay, first pcb manufacture just gave up...  sorry but we cant mke 0.25 mm holes..01:36
kristianpauldevice, not bad for only 99¬01:39
kristianpaulha,! nice quote :-)01:39
kristianpaul"Making the little Ben the first AVR microcontroller programmer in the world able to edit the source code, building it, listen music or play Supertux at same time, in same device" sweet :-)01:40
kristianpaultuxbrain_away: you make it !! :-)01:40
wpwrakkristianpaul: (holes) did they read the spec ? hole sizes up to 15 mil / 0.4 mm are okay01:41
kristianpaulseems not01:41
wpwrakkristianpaul: of course, if they didn't read the spec, they may also not have realized that the board is 0.8 mm ...01:41
wpwrakkristianpaul: a 0.8 mm board may be "more exotic" than a 10 mil drill01:42
kristianpaulthey sell 0.8mm board01:42
kristianpaulactually i will buy one cut  from then when able01:42
wpwrakokay. then maybe ask them again, pointing out that they can go up to 15 mil / 0.4 mm for the hole01:44
kristianpaulokay,  so FYI, 1 PCB is 23800 COP01:44
kristianpaulyeah i already reply.. this people is lazy i pointed please, read the README !!01:44
kristianpaulor 13~ usd per PCB01:45
kristianpauland 10 days deliver time01:45
wpwrakwow. very cheap. what would be 10, 100 ?01:45
Action: panda|x201 waving from Shanghai01:46
panda|x201wolfspraul, morning :-)01:46
wpwrakalso, what surface finish ? ENIG ?01:46
kristianpaulwpwrak: 1001:46
wpwrakah, USD 130 for 10 PCBs then ?01:46
wolfspraulpanda|x201: hi, good morning01:47
kristianpaulwpwrak: http://www.kristianpaul.org/~paul/tmp/atben__iso-8859-1_Q_0110310_2_CotCristianPe_F1arandaPn.pdf__.pdf01:49
kristianpaulah, wait the second manufacurer askme more details... again... okay :-)01:53
wpwrakkristianpaul: USD 181 (with IVA) for 12 units is still a very reasonable price01:55
kristianpaulnot for me :/01:56
kristianpaulokay, i'll delay this a month01:56
kristianpaulbut is good to know i have good prices localy for small runs :-)01:56
wpwrakmay be cheaper to just get some from tuxbrain. he can spread out the setup cost01:56
kristianpaulbah.. the second pcb manufacurer is asking 394000 cop (tax included), ...01:58
wpwraktuxbrain's pcb fa has a high setup cost but a low per unit cost. so if you make larger quantities, the prices are good. yours has a much lower setup cost. there are places like batchpcb with even lower setup cost, but they can't do advanced boards like this one.01:58
wpwrakfor how many ?01:59
kristianpaulshould i ask 100? :D01:59
kristianpauljust for curiosity..01:59
wpwrakwell, it's not a lot more than for the other if the quantity is comparable. USD 217 vs. USD 18102:00
kristianpaullet see, as they said 50% initial, may be i can handle it..02:01
wpwrakyes, you should always ask for the next larger quantity, too :) e.g., in the case of tuxbrain's fab, 100 and 500 UBB cost almost the same. so it would be silly to just get 10002:01
wpwrakkristianpaul: (50%) initial. better make sure you have funding for the entire order before you make it.02:05
wpwrakkristianpaul: or wait for tuxbrain. may be easier.02:05
kristianpaulonce you you get one foot in the pool... :-)02:05
kristianpauli can wait, or ask for small quaitities, but not sure if is good for me..02:06
kristianpaulfor my position, may be i can asid, okay, lets try 4 boards with 0.4... and see how good is your work ;-)02:06
kristianpaulthe gave me a excuse02:06
wpwraktuxbrain_away: to answer your question about component placement: there is a separate placement file for that. they don't have to read the locations off the gerbers or such02:11
wpwraktuxbrain_away: they need to translate that file into what their machine(s) process(es), though02:12
kristianpaulxiangfu: wow, you mean with SDK there is no need to compile openwrt again in the next two years? :-)02:49
kristianpaulor at least for every upstream release..02:50
xiangfuas long as the gcc and uclibc don't update.02:51
xiangfukristianpaul: I have some question about MM1 boot.02:52
xiangfuwhen plug the power cable to MM1. the FPGA should load the VHDL code to itself  right?02:54
kristianpaulit loads a _standby_ bitstream generetated from a verilog code, yes02:54
kristianpaulso it waits you press the middle button to load the regular bitstream02:56
xiangfuthe LM32 cpu core is write by verilog code. when the FPGA load those LM32 cpu core?02:56
kristianpaulthat, then, will load from the flash memory the bios, and the bios will load flicernoise or other app02:57
xiangfuok. so it's standby-bitstream --> REGULAR BITSTREAM --> REGULAR BIOS --> REGULAR APP   (flickernoise)02:58
kristianpaulboard power  > fpga load standby bitstream > if right button pressed > regular bitstream is loaded (here comes LM32 cpu), then it loads bios from the nor flash02:59
kristianpaulyeap xiangfu02:59
xiangfuok. so REGULAR BITSTREAM is LM32 cpu core.03:00
kristianpaulwell, is more that than is the whole Milkymist SoC03:00
kristianpaulwich of course includes the LM32 core03:01
xiangfuok. understand the REGULARR BITSTREAM now.03:02
kristianpaulname is not self descriptive i should say03:02
xiangfuanother question. where is rtems system?03:02
kristianpaulis flickernoise03:03
xiangfufor my understanding. RTEMS is like Linux.  Flickernoise is Gmenu2x.03:03
kristianpaulremenber rtems, is not a linux like os03:03
kristianpaulwell yes, kind off...03:03
kristianpaulrtems is like a lib03:03
kristianpaulor a monolitic app, wich claims to be an OS03:03
xiangfuok. understand now. very thanks.03:04
kristianpaulfliernoise it self, is the mix of rtmes, plus MTK03:04
kristianpaulMTK is the graph toolkit for the fliernoise GUI03:04
kristianpaulalso you need some extra libs to driver the fpu03:04
kristianpauland dependecs for mtk and other graphical stuff as png and uudio too03:05
kristianpaulall those libs for rtems03:05
kristianpauland at the end, just on binary !03:05
xiangfuthat is why when I install the RTEMS. it's only copy libs and head files to /opt/rtems-4.11/...03:05
kristianpaulrtems is like the libc03:05
kristianpauli can say, i not sure id is the right aprouch03:05
kristianpaulplus the messahge queee and semaphores.. etc..03:06
kristianpauland some alrady integrated stacks like ethernet,..03:06
kristianpaulyou can ask Fallenou at #milkymist03:07
kristianpaulas he did gsoc lasy year for porting rtems to mm103:08
kristianpauli still learning rtems internals, so i dont have the last acurate info :-)03:08
kristianpaulfor example, just to you get an idea, a hello world app, is mainly a makefile plus a C file, with some defines and a main with a  printf/printk, :-)03:09
aw_kristianpaul, do you think that drawing/illustrating a functional block /or architecture about current all s/w repository is a good idea for like me can easy understood rather than describing?03:09
kristianpauland no. you cant load dinamically binaries on rtems yet (for the milkymist BSP)03:10
kristianpaulgcc issues mainlly03:10
kristianpaulaw_: oh, sure drawings are always good03:10
kristianpaulI see a timeline drawinf for boot processs and a shelllike drawing too for merelly software part03:11
aw_kristianpaul, when you are trying to introduce /or teach xiangfu about this which is almost what I would like to know. I think that you could hard to introduce like this everytime when  a beginner asking. :-) sorry that but I do not really know their structure/architecture.03:12
aw_kristianpaul, um..a timeline drawing about boot process is powerful/easy understood for me I think. sorry interrupt your talks.03:13
kristianpaulaw_: oh, no your comments are wellcome !03:15
kristianpaul"I think that you could hard to introduce like this everytime when  a beginner asking. :-)" You mean, I should make a introductory doc? sorry i dont get the "hard"03:18
kristianpaulaw_: yeah, i know there is a missing general structure/architecture for milkymist, but not be afraid to dig in, is not so hard as many people think03:19
aw_kristianpaul, sorry it's 'hardly'..03:19
kristianpaulah, i see, you wonder this, but not so many people ask what xiangfu did just before03:20
aw_kristianpaul, um..but for me...it's hard or even include yi i think.03:20
kristianpauland hey, this is qi-channel :-), good, we're more friendly here ;-)03:21
aw_yes, so if there's general structure/architect like http://wiki.rtems.org/wiki/index.php/Media:RTEMSArchitecture.png03:21
aw_it would be super helpful for us as a beginner.03:21
kristianpaulwell that already is helfull, but yes, not milkymist/flicernoise specific03:22
kristianpaulI'm taking your words, i agree, beginners need more intro doc, also will help wolfspraul do more sales ;-)03:22
kristianpaulSo people not will afraid buy mm1  and get alone in the wild...03:23
aw_yeah...well..that's just IMO. :( -- :)03:23
kristianpauli agree03:24
wolfspraulI suggest to use the milkymist.org wiki for that03:29
wolfspraulour resources are limited, so if we get carried away to draw difficult drawings for several days, I'm not sure that's good. The problem with drawings is that they often get outdated (maintenance effort is too high).03:30
kristianpaulkeep update is hard..03:30
kristianpaulre-use rtems graphic is not bad idea, and the boot process is not hard to describe as text on a wiki i think03:39
aw_i've not dig into such stuffs. once I finish sourcing then must run into them at some points. i have a lot of tasks waiting for me before rc3 run. :-)03:43
kristianpaulbtw, some not asked feedback, my mm1 power adapter get a bit warm when just pluged to main, no mather if is or not connected to the board. i guess is okay, just wondering at anycase, as other adapters i have dont get warm03:56
qi-bot[commit] Werner Almesberger: gerber/gmerge: new utility to merge multiple KiCAD Gerbers into one http://qi-hw.com/p/eda-tools/518781904:13
xiangfukristianpaul: same with my mm1 power adapter.04:15
xiangfukristianpaul: (rtems/flickernoise) very thanks.04:15
wpwrakaha ! found another bug in the scripted pcbnew. it ignores clearance settings05:12
wpwrakwolfspraul: btw, have you seen that i added another pair of patches ? --origin=abs|aux to set the origin of the coordinate system for fab files05:27
wolfsprauloh, no haven't seen that yet05:28
wolfspraullet's see05:28
wpwrakwolfspraul: meanwhile, i found another bug: seems that --plot doesn't pick up global clearance settings, so we end up with the hard-coded defaults05:28
wolfspraulhe, quite intrusive your patch05:30
wolfspraulI will try to merge it at the next opportunity, of course not sure I can find a cleaner way.05:30
wolfspraulwith 'global clearance settings' you mean what? where is the setting?05:30
wolfspraulyou mean a user/system-wide saved setting?05:30
wpwrakwolfspraul: yes, global. under Preference > Dimensions > Pads Mask Clearancs06:00
wpwrakwolfspraul: they're stored in .pro, but there's also an effect in the .brd file. not sure how they're related06:01
wpwrakwolfspraul: (intrusive patch) yeah, got a little messy. i also don't like that i had to second-guess what the algorithm does later on. maybe that's not necessary in more recent versions of kicad.06:04
deadmanHello, anyone able to help with some CPU questions?06:25
deadman8v) Really?06:26
deadmanI apologize this is way new to me here.06:27
tuxbrain_awaylarsc: cruel06:28
larscbut effective :)06:32
larschmpf, i should do more -rc testing nanonote support in 2.6.38 is broken06:39
Jay7larsc: so, 2.6.38 is unusable on NN?09:51
larscJay7: without patches yes10:10
Jay7have you patches in your tree already? :)10:10
qi-bot[commit] kyak: package libmpc http://qi-hw.com/p/openwrt-packages/bf897ad10:11
larscnot yet10:11
larsci'm still testing10:11
Jay7kyak: do you have that kernel patch around for kexec working?10:38
Jay7iirc, something about add_memory_region()10:38
tuxbrain_awayso lol ... http://www.youtube.com/watch?v=MI99t9k4aEE THE MAGICAL IPAD (Parody)10:44
larscJay7: i've just push jz-2.6.3812:04
viricwhat is better in 2.6.38, over 2.6.36, for example, for the nanonote?12:04
larscit boots faster12:05
viriclarsc: I wanted to understand what are the measures for kernel memory usage. /proc/meminfo should have all the numbers, but I even cannot guess what is the kernel image size from there. Do you understand that meminfo?12:05
viriclarsc: ah great12:05
larscwell jz-2.6.3812:05
larscnot upstream 2.6.3812:05
larscviric: sorry, no idea12:06
viricthank you :)12:06
Jay7larsc: so, jz tree is usable now?12:06
viricI was trying to get the ingenic new docs released, and I still could not access them12:08
viricdoes anyone have the link?12:08
viricand you can access that?12:10
virichere it stays forever12:10
viricah, after two minutes it worked12:12
larscit takes a moment though12:12
larscwolfspraul: do you know why the qi-hardware server is currently so slow?12:12
viriclarsc: the ftp does not have 4720 datasheets12:14
larscthe jz4720 is basically a jz474012:15
larscjust a different packaging12:18
viricthank y0ou12:18
Jay7larsc: http://en.qi-hardware.com/irclogs/qi-hardware_2011-03-03.log.html#t08:4412:22
Jay7please read some lines from this poing about add_memory_region()12:23
Jay7second problem is how to pass cmdline from kexec to chained kernel..12:25
Jay7we have some progress but not finished yet12:25
larscadd_memory_region is called from the 'mem' cmdline option callback12:27
Jay7so we should have mem option in kernel cmdline always?12:28
Jay7kyak: ^^^ :)12:28
Jay7well.. this is half of success then12:29
larschardcoding the memory size in prom.c is no option since there are multiple boards with different ram sizes12:29
Jay7here is conversation about passing cmdline from kexec to kernel12:34
Jay7kyak have success only when just passing old kernel cmdline w/o changes to new kernel12:35
Jay7larsc: may you suggest anything here?12:35
Jay7btw, previous day's log may be useful too12:35
viricJay7: what are you trying to achieve?12:36
Jay7viric: working kexec :)12:36
viricto achieve what? :)12:36
viricbetter debugging on hangs?12:36
Jay7working kexecboot :)12:36
virichm I don't know kexecboot12:36
Jay7it's GUI to choose kernel and boot it (simple explanation)12:37
wpwrak(ingenic docs) very interesting. they pulled all the 4720 material. good that we pulled full copies :)12:37
viricwpwrak: ah you have them? Where?12:37
viricJay7: ah ok12:37
wpwrakwolfspraul: something for your next meeting. ask them to restore at least 4720_ds12:38
viricJay7: why would someone want to switch the kernel there?12:38
viricwpwrak: if you can put a big zip somewhere, I'd gladly pick it12:38
Jay7viric: it's useful mostly to boot from any media12:38
wpwrakviric: i think xiangfu has a web-accessible mirror12:38
Jay7i.e. from SD e.g.12:38
viricJay7: ah ok12:38
viricwpwrak: ah ok. I'll ask him12:38
wpwrakviric: (kexec) the idea is to keep the dedicated boot loader (u-boot or such) as simple as possible and to move all the more complex things into a linux-based environment that then boots the real system with kexec12:40
viricwpwrak: hm ok12:41
Jay7next kexecboot release should have switch_root method support as well :)12:41
viricwpwrak: I hope that does not end in uboot mmc code unusable12:41
Jay7so we can use it as initramfs12:41
wpwrakviric: i hope it will end in the unceremonial burial of u-boot :)12:45
Jay7seems you are not happy with u-boot here ;)12:45
larscburn u-boot, burn! ;)12:47
larscJay7: this commit seems to add support for what you want http://dev.lemote.com/cgit/linux-loongson-community.git/commit/?h=tiny36&id=240c76841b26f1b09aaced33414ee1d08b6454cf12:50
wolfspraulwpwrak: I asked the CEO about permission to mirror, no response yet12:53
wolfspraulthat's unusual, he normally replied on the same day always. oh well :-)12:53
wolfspraulthere may be lots of reasons behind the documents on the server12:53
wolfspraulall the way to a clueless new server admin who just copied files around without knowing what was copied, and where it was copied to12:53
wpwrakwolfspraul: at least he didn't reply "oops, it was a mistake" and pulled them12:54
wolfspraulChinese will never admit a mistake.12:54
wpwrakbut most of the files are still there :)12:54
wolfspraulI'll find out one day.12:54
wolfspraulwell then, great12:54
wpwrakhehe ;-)12:54
wolfspraulplus it was public now, it will be mirrored, one way or the other12:54
wpwrakthe internet never forgets12:55
wpwrakwolfspraul: any idea about the default settings ? i see that you are loading the project file (or at least that's what the source suggests is happening)13:41
kristianpaulwolfspraul: hi13:56
wolfspraulnot sure, haven't looked yet14:08
wolfspraulthose parts are probably copied over from your original patch14:09
wpwrakheh, didn't even know you had copied things from my hack14:17
wolfspraulwpwrak: of course. on the shoulders of giants...14:43
wolfspraulyou made the first command line option, I just added some more14:44
kyakJay7: http://dpaste.com/503878/16:58
kyakthat add_memory_region patch :)16:58
kyakthe "MIPS: Get kernel parameters from kexec-tools" patch doesn't work.16:59
kyakwell, it works well  for fallbacking to default cmdline (i.e. the cmdline of running kernel)17:00
kyakbut you can't pass a new cmdline via kexec17:00
Action: kyak registered at identi.ca18:45
kyakit's only better than twitter because it's open source. Other than that, same ugly, slow, buggy piece of crap18:47
kristianpaulagplv3 is important too18:52
virickristianpaul: have you read "La herencia del coronel" (comic)19:25
kristianpaulviric: nope19:47
viric:) ok19:48
tuxbrainthere is any cronometer app on NN?23:30
--- Wed Mar 16 201100:00

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