#qi-hardware IRC log for Tuesday, 2010-11-02

qi-bot[commit] Xiangfu Liu: [gcal] update to 3.6, thanks to Otmar Struwe http://qi-hw.com/p/openwrt-packages/23a569402:59
qi-bot[commit] Xiangfu Liu: [byobu] new package,a multi-panel screens tools http://qi-hw.com/p/openwrt-packages/a85925104:19
qi-bot[commit] Xiangfu Liu: [byobu] new package,a multi-panel screens tools http://qi-hw.com/p/openwrt-packages/733b57d04:36
qi-bot[commit] Xiangfu Liu: [config.full_system] add package byobu http://qi-hw.com/p/openwrt-xburst/74016b104:41
qi-bot[commit] Xiangfu Liu: cleanup the build file http://qi-hw.com/p/openwrt-xburst/ef4105f05:23
nitin_guptais ther =e any web browser in NN ?06:09
rafanitin_gupta: you should find dillo, links and links-x11 surely08:08
qi-bot[commit] Werner Almesberger: Moved usd-card.lib over from f32xbase http://qi-hw.com/p/kicad-libs/cf2f4d610:06
qi-bot[commit] Werner Almesberger: Moved c2ben/components/usd-card.lib to kicad-libs project http://qi-hw.com/p/f32xbase/2951f9610:06
Action: xiangfu doing "apt-get install mercurial" for the new nlove package. the nlove is using mercurial10:41
Action: xiangfu "apt-get install mercurial" on build host: fidelio.qi-hardware.com10:42
xiangfukyak: Hi is your computer is 32bit?10:42
xiangfukyak: the gcc-mips compile fine in 32bit system. but have some errors in 64bit system.10:43
xiangfukyak: the error message is:10:43
xiangfukyak: http://pastebin.ca/197960410:45
qbjectbartbes: good morning. where can I find your recent work on nLove?11:08
qi-bot[commit] Werner Almesberger: Added top-level Makefile, for convenience. http://qi-hw.com/p/kicad-libs/7e3bcd011:19
qi-bot[commit] Werner Almesberger: ext/: added uSD extender board (draft) http://qi-hw.com/p/ben-blinkenlights/49c76e011:19
wpwrakxiangfu: 3s/512/256/ :-)11:25
bartbesqbject: http://bitbucket.org/bartbes/nlove11:26
xiangfuwpwrak: oh. sorry for typo.11:26
qbjectbartbes: Thanks!11:26
bartbesqbject: what good is a public repo if you don't share it? :P11:28
qbjectbartbes: true, true.11:29
qbjectwpwrak: last night you said MMC boot could be handled by Linux.11:29
qbjectwithout das u-boot.11:30
wpwrakqbject: yup, with kexec11:30
bartbesnot having to submerge my ben every time I want to boot from mmc is a huge bonus11:30
qbjectoh ho. look at that.11:31
qbjectWhat the heck do we need a bootloader for, then?11:31
wpwrakqbject: but even reading mmc doesn't seem to be too hard. e.g., the qi boot loader (on the openmoko freerunner) did it, and as far as i know, it had no problems.11:31
wpwrakqbject: all we need is a simple boot loader :)11:32
qbjectwpwrak: looks like. yay simple!11:32
wpwrakqbject: and the leaner and faster we can make the kernel and a basic userland, the simpler the boot loader can get11:32
qbjectwpwrak: "basic userland" - is that what busybox is giving us, or is it more than that?11:33
wpwrakqbject: e.g., if it's fast enough to just load a small linux from nand even for everyday use with booting the real system from MMC, then the boot loader doesn't have to do this.11:33
wpwrakqbject: yes, busybox gives something like this. or you could even have a custom userland with even less stuff.11:34
qbjectwpwrak: Mm. More of the things I want, less of the things I don't.11:35
wpwrakqbject: depend a bit on what you're after. but the good thing is that you have plenty of choices. if you need your boot environment to be feature-rich, that's easy to do.11:35
wpwrakheh ;)11:35
qbjectwpwrak: are you aware of a µSD breakout with a standard SD reader on it? (I know it wouldn't be hard to make, but I thought maybe someone had already.)11:37
wpwrakhmm, wolfgang had such a beast made once11:38
wpwrak(i have one. didn't use it yet, though)11:39
qbjectOkay, thanks.11:42
qbjectJust contemplating what hardware I want on my Ben MicroNote. ;)11:43
wpwrakthere it's probably easier to just make a little pcb with add-ons11:44
qbjectOr a tab that I can hook to this: http://www.sparkfun.com/commerce/product_info.php?products_id=20411:47
wpwrakqbject: well, if you're already making the uSD cardlet, then you can do the full SD/MMC side in the same process.11:52
qbjectwpwrak: good point.11:52
xiangfuqbject: the picture of uSD --> SD : http://en.qi-hardware.com/w/images/3/3a/MicroSD_to_full_size_SD_adapter.jpg11:56
qbjectwpwrak: See, that's gorgeous. I want one.11:57
qbjectWhy isn't wolfspraul selling them on sharism? =D11:57
xiangfuqbject: from you link that product only support 1bit data transfer.11:57
qbjectxiangfu: thank you for pointing that out. I am still learning about electronics.12:00
xiangfuqbject: me too. learning learning   :)12:00
xiangfuso many new things around me :)12:01
wpwrakxiangfu: (1 bit) naw, i think it supports 4 bit SD/SDIO. they only describe SPI mode, that's why it looks as if it did only one bit.12:03
wpwrakqbject: (not selling) ask him :) one issue is that each product has some overhead. so unless he can sell a lot or sell them at a high per unit price, then that doesn't make sense economically.12:04
wolfspraulno I'm not selling them12:08
wolfspraulI've made a few once and gave them away as gifts, I think12:09
qbjectwolfspraul: it makes sense that you wouldn't sell them. Nice looking piece, though, and exactly what I need.12:10
xiangfuqbject: what you want do with this adapter?12:14
wpwrakqbject: if you can make PCBs, it wouldn't be too hard to make your own.12:17
qbjectxiangfu: I want to take the electronic components from a Ben and repackage it in a very sturdy housing with a bigger keyboard. I will share my designs, of course. :)12:19
wpwrakqbject: design files for the uSD card shape are available and it shouldn't be too hard to translate this into something a CAD/CAM system can use. so you could machine the part with a CNC mill.12:19
qbjectAnd I would like to have the flexibility to use µSD or larger, cheaper SDHC12:19
xiangfuok. the DO, IRQ, NC(PIN 9), CS in SPI mode is DAT[0-3] in 4-bit mode. thanks wpwrak for point out. :)12:20
qbjectwpwrak: I don't have CNC capability but I live fairly close to Albuquerque. Sandia National Lab, Intel and Xilinx facilities, etc. I'm sure someone there can.12:21
wpwrakqbject: (no cnc) ah, pity. i was hoping you would. having a cnc mill at home is quite nice :)12:22
qbjectwpwrak: No joke. But Sherline makes CNC conversions (running on Fedora/EMC I think) for the manual mill I have, so maybe some day.12:23
qbjectwolfspraul wpwrak: is the µSD/SD adapter design open? is it on the wiki? can I help make it so?12:27
wpwrakqbject: dunno if it's open. but it should be quite simple to draw your own. do you know how to make PCBs ?12:33
wpwrakqbject: (draw your own) half of the work (the uSD side) is already done12:33
kyakxiangfu: hmm12:35
kyakmy system is 32bit, yes12:36
kyakunfortunately i don't have 64bit at hand12:36
kyakso can't really fix t -\12:36
xiangfukyak: do you have any idea about that error?   I will try to fix that tomorrow :)12:41
wpwrakqbject: all you would need is this part: http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=101-00708-64-ND12:49
wpwrakqbject: draw the footprint and a schematics symbol for it, then make a pcb that is uSD on the other end. done :)12:49
qbjectI'll give it a shot. My CAD experience is mechanical drawings so far, but I like learning new things.12:50
wpwrakqbject: do you have a laser printer ? ideally, just black and white.12:56
qbjectI have access to one. You talking about some kind of resist?12:57
wpwrakqbject: yup. with a laser printer, you can use the toner transfer method for making PCBs. it's a bit easier than the traditional photochemical process.13:00
qbjectDoes toner transfer produce a resist or actual conductors?13:00
wpwrakqbject: also, this allows you to cut/drill/mistreat the PCBs before etching, which can be convenient.13:00
qbjectwpwrak: Ah, okay, so it is still a resist/etch process?13:01
wpwrakqbject: it deposits toner on top of the PCB, which acts as an etch resist13:01
wpwrakyup. can't get rid of that one easily.13:01
qbjectwpwrak: still using ferric chloride, or is there something better?13:01
wpwrakheaven, no ! :)13:01
wpwraki now use a 2:1 mix of H2O2 and HCl.13:02
wpwrakH2O2 ... in the concentration you get at pharmacies. something like 10%, i think.13:03
qbject(ferric chloride is great for exposing the visible metallurgy in interesting handmade steel)13:03
qbjectAh, cool.13:03
wpwrakHCl around 35%, again, consumer grade concentration13:03
qbjectAnd I think they still sell "Muriatic" (HCl) at hardware stores around here.13:03
wpwraki hate ferric chloride ;-) stains everything, isn't transparent, relatively slow, and needs heating.13:04
kyakxiangfu: http://bugs.openembedded.org/show_bug.cgi?id=1951 - this should help you (hope you are reading logs) :)13:05
wpwrak(muriatic) yup. the ingredients are very very common.13:05
qbjecthehe. noted. I only know about ferric because knifemakers grump that you can't get it at Radio Shack any more.13:05
wpwrakthis acid is transparent and stays transparent. doesn't stain much (well, i didn't do extensive spill testing :), it fairly quick, and works at room temperature.13:06
wpwrakalso, when you mix it with NaOH, you get a water-soluble sediment. so it's easy to neutralize small quantities of the acid for disposal.13:07
qbjectSounds perfect. Do you need to add an offset around your conductors (are they called traces?) to account for the acid which works around the edges of the toner resist?13:07
wpwrak(with ferric chloride, you get a toxic swamp :)13:07
wpwraktraces, yes. well, there ought to be some under-etching but i never cared enough about it to measure.13:08
qbjectSounds good so far.13:10
wpwrak(ferric) good to be rid of it ;-) there are a few other acids that are common for DIY now, NH- and Na- based, I think. but why bother with something unusual when you can just get a nice acid with a trip to the drugstore and the hardware store ? :)13:10
wpwraktoner transfer needs some experimenting. what's important is the paper you use and the ironing technique. it works as follows: you print the layout on a sheet of paper. then you glue a bit of plastified paper on top of it, such that it covers the print.13:12
qbjectwpwrak: See my earlier statements re: bootloading. Simple is goood. :D13:12
wpwrakthen put all this into the printer again and print a second time, now on the plastified paper. remove it, put it on the copper-clad PCB, and use a clothes iron to heat the toner and make it stick on the PCB. after that, wash the board to remove anything13:13
wpwraksticky that came from the plastified paper, and it's off to the acid bath.13:13
wpwrakfor the plastified paper, you need to find some of the paper used for high-quality photo printing with ink printers. takes a bit of experimenting to find a good one. i like one made by HP,13:15
wpwrakthe "HP Premium Photo Paper, glossy", C6039A13:16
wpwrakthe "glossy" is important here. there's also a "HP Premium Plus Photo Paper", but that one's already different13:16
wpwrak(hmm, i just noticed that i have only 2.5 sheets left. let's hope i can still find it somewhere ...)13:17
qbjectWhat's the first printing on conventional paper for?13:17
wpwrakto get the position13:18
wpwrakif you can predict it accurately, you could skip this step. but then, why take chances13:18
qbjectOkay, thought so. Why not just reverse your layout and print it on the backside of the photo paper?13:18
wpwrakthe photo paper is expensive and you usually only have layouts that are much smaller than a sheet. so you want to use only small pieces of photo paper.13:19
qbjectI'm not sure I understand why that's an argument against my suggestion. Please explain? (If you can print the layout on the glossy side, why can't you print a reversed layout on the backside?)13:22
wpwrakoh, i thought you wanted to feed the whole sheet. the position things is to find out where the layout will be printed.13:23
wpwrakso that you put the patch of plastified paper on the right location on the "transport sheet"13:23
qbjectAhhhh. I get it now! :)13:25
wpwrakbefore the ironing, you also have to clean the PCB's surface from oxides and other dirt. fine steel wool works great for this. afterwards, wash off the steel dust with alcohol.13:26
qbjectDo you just watch the acid bath until all the traces are exposed?13:26
wpwrakwell, moving the board helps to accelerate the process :)13:26
qbject[wiggle wiggle]13:27
wpwrakor use a pencil-shaped object to push it around :)13:27
wpwrakfor the kind of small boards i'm making, a small beaker works perfectly. flat bottom, high side walls, transparent all around13:28
qbjectAnd you're just using copper-clad FR4, right?13:28
wpwrakfor the uSD project, you need a think PCB. 0.8 mm instead of the common 1.6 mm.13:29
wpwrakerr.. s/think/thin/ :)13:29
qbjectYup. McMaster-Carr has 1/32" (0.8mm)13:30
wpwraknot sure if radio shack has this. if not, digi-key does.13:30
wpwrakah, you already found a source. perfect :)13:30
qbjectOh yes. McM-C is like Digi-Key for machinists. ;)13:30
qbjectwpwrak: Thanks for taking the time to educate me!13:35
wpwrakqbject: you're welcome. good luck with your first pcb then ! :-)13:40
wpwrak(ah, we still have to cover the software tools. mainly kicad and fped)13:40
qbjectwpwrak: Well, I know I can do something as simple as this with qcad, taken as a mechanical part. But yes, I can see that it's probably not ideal for electronics in general.13:47
wpwraknaw, an EDA system does a lot of things to help you. doing this with qcad would be extremely painful in comparison13:52
qbjectAnd that's, what, when you can enter a schematic and it figures out the traces for you?13:55
wpwrakif could do that but you're better off drawing the traces manually. it will tell you what you can draw, though.13:56
jlmokoI know this isp sucks is telefonica :/. What can I do for you & myself?14:06
jlmokosorry wrong window14:06
qmasterrri try to get the latest git with 'git pull' but, it says i have unmerged files, how do i discard them?15:09
qbjectaisa: did you build the rootfs that you flashed onto your Ben, or did you download it pre-built?16:10
aisaI built it myself.16:10
zythrock hard, ride free (JP) and make v=9916:11
aisaqbject: I have the whole machine set up as a VM.16:11
aisaAlas, it is running VMWare, rather than something free,16:11
aisabut I believe you can convert back and forth.16:11
aisathis is you're not interested in building something from scratch.16:12
aisasomething==a build host.16:12
aisaah, but there is fidelio too.  the qi-hardware build host.16:12
aisaI have ready-to-go build images sitting there too.16:12
aisaso that is probably the better option than me giving you my VM.16:12
qbjectaisa: a pre-built image would be fantastic. I don't know that I could fully utilize a nifty setup like a VM.16:13
qbjectJust looking for something recent that includes the font-sizing tools.16:14
aisaThe recent images prepared by qi should work fine.16:14
qbjectDo you recall where those live? Because I'm having a devil of a time finding them on the wiki.16:15
aisayou only really need to build your own image if you want something weird at the kernel or configuration level.16:15
aisaah, ok.  let me see.16:15
aisaHere they are:16:15
qbjectWhat a logical place for them.16:15
aisaI was compiling some software today in preparing of getting another package ready,16:16
aisabut I was first compiling on my built host,16:16
aisaand I accidentally linked the code with my cross-linker.16:16
aisathat took me a little bit to figure out.16:16
qbjectMeaning it was targeted for your Ben instead of the intended platform?16:18
aisainstead of my build host, correct.16:18
aisaI had the correct compiler,16:19
aisaso it was creating all the intermediate files just fine.16:19
aisait was only when it went to link.16:19
aisaand it doesn't really create an error right away,16:19
aisayou have to create a little more space first by doing a bunch of other stuff :-p16:19
qbjectaisa: trixie gremlinses mucking about in there.16:22
aisaand like all trixie gremlins, it was a one line fix!  :-D16:22
qbjectHow many of my C++ compiles failed in college because of missing semicola? MANY.16:24
zythNanonote Harware 64 MB RAM Upgrade -->> http://www.flickr.com/photos/zyth/sets/72157624750968074/16:27
aisaI don't think it's an upgrade until all the pieces fit back together ;-)16:27
aisaactually, that is seriously cool.16:28
aisamy first thought was: 2 NanoNotes?16:28
aisazyth: What's going on there?  Are you sawing one chip off and reseating another?16:29
aisaor does the memory pop out?16:29
zythfirst one16:29
aisado you own two NanoNotes?  :-)16:30
aisais that because you broke one of them after doing this?  :-)16:31
zythjust updarde RAM16:31
aisait looks like you've got a parity error there.  :-)16:31
zythheh )16:31
zyth12. Worth Fighting For.mp3 (JP)16:34
wpwrakzyth: whee. how did you unsolder these beasties ?16:34
qmasterrri try to get the latest git with 'git pull' but, it says i have unmerged files, how do i discard them?16:38
aisagit rebase --skip?16:38
wpwrakzyth: ah, heavy weaponry ;-)16:38
aisaqmasterrr: that was for you.16:38
aisait depends on exactly what the problem is.16:39
wpwrakgit stash ?16:39
qmasterrraisa:   git rebase --skip / No rebase in progress?16:39
aisagit what wpwrak said, about git stash.16:39
aisaand I'll need the full error message if it doesn't work.16:39
qmasterrrok, mom16:40
zythqmasterrr: delete old git then get a new one16:40
aisaqmasterrr: I've got plenty else to do today.16:40
qbjectzyth: is that one o' them hot air rework stations everyone gets so excited about?16:41
viricwpwrak: I bought a ftdi cable that can do 3.3V too :)16:42
viric(say the specs)16:42
qmasterrrthis is what i get if i do a git pull16:42
aisagit status is going to show files that are unmerged or modified.16:43
aisayou need to run git add on them16:43
aisaI'm assuming you haven't actually changed anything?16:43
aisaif you have, got ahead and stash/commit your changes.16:43
qmasterrri have changed nothing usefull16:43
aisagreat.  so you'll need to add the files it is complaining about.16:44
aisadepending on where you are in your repo right now,16:44
aisathey might have merge conflicts in them.16:44
qmasterrrah, thanks :)16:44
qmasterrr git status | wc -l    outputs 118 lines, do i need to add everything by hand?16:45
aisagit status | wc -l | sed 'g/^#/d'16:45
aisaor something like that... I can't quite remember the output of git status.16:46
aisahold on...16:46
qmasterrri will nopaste the output, one moment16:47
aisarun "git status | grep -v '^#' > foo16:47
aisaclean up the few lines in foo that aren't files,16:47
aisathen run:16:47
aisaxargs git add < foo16:47
aisaah right.16:47
aisaok, one sec.16:47
viriczyth: great for the memory upgrade! congratulations16:48
viriczyth: why you chose 64 and not 128?16:48
aisaqmasterrr: your poor repo...16:48
zyth128 very expencive for nn )16:49
qmasterrraisa: i think i crashed it yesterday16:49
aisaqmasterr: without reading a bit more about git, I can't provide much help.  You are in the middle of a merge, and in theory you should be able to say "take one side of the merge and dump the rest."16:50
aisabut I'm not familiar enough with all the cases to know exactly how to get there from here.16:50
aisaI can describe what you could try, without being able to help with the commands.16:50
aisayou could try reverting your commit to the last common ancestor of your repo and the upstream.16:50
aisaand then a git pull should go without problem.16:50
qmasterrraisa: i will try that, thanks for your time :)16:51
aisabut the fact that git is talking about merge paths is a hint it is in a special mode.16:51
wpwraksometimes, a  git-whatsup would be nice to have ... :)16:52
zythsvn rulez16:52
wpwrakviric: ftdi is for weaklings ;-)16:53
wpwrakzyth: naw, git is a lot more powerful. but also harder to use.16:53
viricwpwrak: That's why I chose it! :)16:54
zythi prefer svn for my freerunner, dlink, tp-link routers & nn of coz16:54
aisazyth: I'm come to really depend on git add --patch16:55
aisaand having nothing like in in the svn side drives me crazy.16:55
aisaI imagine that you haven't really wanted something like that?16:55
aisaadd --patch means you can add a file, but not every change.16:55
aisabasically, I want to edit my patches before the commit.16:55
aisaI never thought I would want to do this, using svn.  But I really can't live without it anymore.16:56
wpwrakaisa: those things are actually bad. with svn, you can't do such stuff, so you work in a way that doesn't need it. with git, you can, and you end up with a much more complex workflow :)16:56
aisawpwrak: I remember reading a pro-git article that went something like "if your software said you shouldn't have done it that way, the software is bad."16:57
aisaand the use case cited was pretty much the one I use.16:57
aisaI hack on something, but then need to fix something else,16:57
aisaand after a crazy patch-fest, everything is stable, but the changelog is a mess.16:58
aisaso I clean that all up to commit bite-sized pieces.16:58
aisaeven stash unchanged stuff to run regression tests.16:58
aisaso I definitely see your point, it is more complex.16:58
aisabut I think about it less.16:58
wpwrakaisa: i notice the drawbacks of getting lured into complexity when i forget some of the stuff i've stashed/reset, etc.17:00
aisaah, indeed.  that is probably my single biggest mistake, forgetting that I stashed something.17:00
zythfuc&*&*% backfire not included s3c24xx, and nn git too17:04
zythmake[4]: Entering directory `/home/zth/openwrt/build_dir/target-arm_v4t_uClibc-0.9.31_eabi/sqlite-3.7.0'17:24
zythand nobody in channel17:24
zythwhere is rope and soap17:25
zyth*** Permission Denied - You're not an IRC operator17:27
zython /die17:27
zythgimme op17:27
wpwrakhmm, i have a puzzle for the chemically inclined: given a PCB that's covered with silicone, is there a way to strip the silicone such that the board can be reworked ?18:43
wpwrakexample of a board partially covered with silicone: http://downloads.qi-hardware.com/people/werner/f32x/c2ben-run1.jpg18:43
wpwrakthe "jelly" is silicone. now imagine the same with chips and all that.18:44
wpwrakthe underlying question is: could we use silicone as casing material for simple circuits ? for this to work, there would have to be a way to dissolve the silicone completely after curing. with reasonably accessible chemicals.18:48
wpwrak(and of course, they would have to leave the rest of the board intact)18:48
Destroyi need to know if i need a new power supply19:03
lisandropmIf the power supply is the one that powers your internet connection/pc... then I am afraid he does :-/19:04
kristianpaulqi webpage got out of size nor more PICS int he home page please !! ;-)23:21
--- Wed Nov 3 201000:00

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