#qi-hardware IRC log for Friday, 2010-09-03

wpwrakheh, those wildfire threads on the list are funny. "pico c" :)06:00
kristianpaulah lets use llvm instead ;)06:40
GNUtoo|laptopkristianpaul, hi, tuxbrain is still not there, is it normal?09:12
wolfspraulGNUtoo|laptop: tuxbrain is going into super concentration mode :-)09:13
GNUtoo|laptopah ok09:13
wolfspraulI think all is fine, he just realized what a time sink chats can be.09:13
GNUtoo|laptopah but how can I contact him then?09:13
GNUtoo|laptopsomebody told me I should talk to them09:14
wolfspraulemail? what do you want, maybe we can help you here?09:14
GNUtoo|laptopor rather to the one in charge of the technical aspect(david)09:14
GNUtoo|laptopbasically it's about the phone they sell or will sell09:14
GNUtoo|laptopit was an android phone and they wanted SHR on it09:14
GNUtoo|laptopI'm very busy09:14
methril_workGNUtoo|laptop, it's related to GeeksPhone maybe?09:15
GNUtoo|laptopso I wondered if we could collaborate to evaluate the port09:15
GNUtoo|laptopthey point me to the sources etc...09:15
GNUtoo|laptopI evaluate09:15
methril_workthen contact them by e-mail09:15
GNUtoo|laptopbecause I was just pointed to a kernel source09:15
GNUtoo|laptopand even worse, I'm on 3g09:15
GNUtoo|laptopwith 100M limit09:15
GNUtoo|laptopbut on week-end the limit goes up to 2G09:16
GNUtoo|laptopI'll do then09:16
GNUtoo|laptopthe only problem is that I've tons of mails09:16
GNUtoo|laptopand reading them trough 3g can eat MBs09:16
methril_workGNUtoo|laptop, well, i try to contact them if you want09:16
GNUtoo|laptopok thanks09:17
GNUtoo|laptoptell him to contact me via irc or jabber09:17
GNUtoo|laptopalso tell him that I'm the guy he met at fosdem09:17
methril_worki would like to be at fosdem with them09:17
GNUtoo|laptoplol ok09:17
GNUtoo|laptopthe one with the bug device from buglabs09:18
GNUtoo|laptopit's pespin who contacted me on #openmoko-cdevel and told me about the geeksphone btw09:19
GNUtoo|laptopmethril_work, do you need my jabber account? in case they david doesn't want to use irc09:20
methril_workhe's going to connect09:20
methril_worki think i've your jabber from SHR ;)09:21
GNUtoo|laptopok thanks a lot!!!09:21
GNUtoo|laptoptuxbrain2, I'll PM you09:21
wolfspraulthere is tuxbrain2 ! :-)09:21
GNUtoo|laptopthanks a lot09:21
tuxbrain2hello all :)09:21
methril_workyou are wellcome09:21
tuxbrain2I can just stay for few mins09:21
wolfspraultuxbrain2: you are selling non-copyleft 'geeks' phones with tons of proprietary technology and no hope that that will ever change? :-)09:22
tuxbrain2well not selling it actually and I will not selling it until at least it has a GNU/Linux inside :P09:26
bartbesxiangfu: btw, have you tried the 'new' gmenu2x?09:29
qwebirc66005hello, the screen of my nanonote gone blank, the device is still reachable via usb. turning off/on doesn't solve the issue. any ideas?09:32
wolfspraulqwebirc66005: can you login with ssh?09:39
wolfspraulwhat do you mean with 'reachable via usb'?09:39
qwebirc66005I still can login via ssh on usb09:39
wolfspraulqwebirc66005: hmm. I have seen 1 or 2 cases where the FPC cable between LCM and main board got loose. It could be that.09:43
wolfspraulyou could open the case and make sure the FPC sits well in the connector, but it's not for the faint hearted.09:44
wolfspraulthere is a connector only on one side (on the side of the main board), on the other side the FPC is soldered in a hot-pressed process, so there is nothing to fix there, but also normally nothing will disconnect there.09:45
wolfspraulhave you tried a full cold start?09:46
wolfspraultry that too. remove the battery, unplug the USB cable. then plug the USB cable in, see whether the screen comes on (first should be some u-boot output)09:47
wolfspraulif it doesn't start automatically after connecting the USB cable, press the power-on button for a good solid 3-4 seconds09:47
wolfspraulsee whether the screen comes on09:47
qwebirc66005wolfspraul: trying...09:48
qwebirc66005wolfspraul: nope, but you gave me an idea, so I strongly pressed the display base and it worked. there is definitely something wrong with the cable. what do you suggest? the device is brand new (just unpacked...)09:52
wolfspraulhe. good that you pressed! :-)09:52
wolfspraulwhere exactly did you press?09:53
wolfspraulthe FPC is going through the hinge on the right side (above the microphone, left of the power button)09:54
qwebirc66005wolfspraul: I pressed just below the display, near the hinge on the center09:55
wolfspraulhmm, center. OK.09:55
wolfspraulis the LCM constantly on now, or only when you press there?09:56
qwebirc66005wolfspraul: it's intermittent, when I press above the mic it turns to full-white too09:57
wolfspraulI would suspect it's the FPC connector on the side of the mainboard.09:58
wolfspraulprobably if you would disconnect and reconnect that cable, it will be stable.09:58
wolfspraultoo bad. What do we do now?09:58
wolfspraulwe have a number of options. If it would be me, I would open the device and reconnect that cable. It's not that hard, but I don't know how confident you are with something like that. Or whether you want your brand new device to also work like that! (which I would understand)09:59
qwebirc66005wolfspraul:well, I'm an a techie so I think I could try to fix this. any suggestion?10:00
wolfspraulanother option is that you return it to where you bought it, we had one such case in the past and fixed it. But shipping a 99 USD device around always makes me cry because sometimes shipping cost can be higher than device costs :-)10:00
wolfspraulsure, great10:00
wolfspraullet's try!10:00
wolfspraulwhere did you buy it by the way?10:01
wolfspraulthis is still not how it should be, and in terms of customer support we should guarantee you a fully functioning device10:01
qwebirc66005wolfspraul: I bought it from hackable-devices.org10:01
wpwrakwolfspraul: what do you think causes this ? assembly forgetting to close the connector, so it passes inspection but eventually, the cable falls out ? or do they check for this ?10:02
wolfspraulwell I hope they don't mind if we move straight to the guts now...10:02
wolfspraulwpwrak: since I personally reflashed every nanonote (there is no other way Linux software could have gotten onto them), no, definitely. It did work.10:02
wolfspraulI have seen one other case exactly like qwebirc6600510:03
wolfspraulit must have worked when I reflashed, but then x months later the screen was dark10:03
wolfspraulI think it's the price you pay for having a fpc cable and connector10:03
wolfspraulqwebirc66005: let's start! :-)10:03
wolfsprauldo you have a small screwdriver?10:03
wpwrakwolfspraul: wow ! reflashed 1000 devices. where do they sell whatever motivation enhancer you took for that ? :)10:04
bartbesfpc being flexible printed circuit? (as I don't expect you to use federal prison camps)10:04
wolfspraulwpwrak: I will reflash another thousand in a week or so.10:04
wolfspraulfirst day was 280 I think, then 320, then 380. got better.10:04
wolfspraulit's a lot of fun to work with many devices10:05
wolfspraulhardware is really really interesting.10:05
qwebirc66005wolfspraul: please wait jus a moment, searching for right tool...10:05
wolfspraulwpwrak: unfortunately not all plastic parts are up yet http://www.almesberger.net/misc/ben/scan/10:05
wpwrakwolfspraul: sounds like connecot not fully close or cable not fully inseerted then. hmm, tricky to test10:05
wolfspraulso I could explain to qwebirc66005 how to lift off the plastic part above the keyboard10:05
wpwrakbartbes: how do you think they make them so cheap ? ;-)10:06
bartbeswpwrak: I always thought they used cosmic radiation to send messages :P10:06
wolfspraulman I hope the qwebirc66005 surgery will not end in a disaster. he is brave though I like that...10:07
wpwrakwolfspraul: (scans) yeah, mill is busy making boards these days :) not all the time, but often enough that i don't have a break long-enough for a big scan.10:07
wpwrakbah, opening the ben isn't too bad :) not having to take out the pcb actually helps. sometimes can get caught somewhere10:08
wolfspraulstill on the Ya I think we can improve some things10:08
wolfspraulnot those nasty little plastic hooks that break off etc.10:09
wpwrakwolfspraul: (ya) definitely. i'd go for a "no click" design. just simple parts and screws. may add a cent or two to assembly cost, though.10:09
wolfspraulI am even wondering whether we should do plastic on Ya or another material.10:09
wolfspraulyes I agree10:09
wpwrakwolfspraul: what materials are you considering ?10:09
wolfsprauldon't know10:09
wpwrak"keep refrigerated at all times"10:10
qwebirc66005wolfspraul: ok, I dissected the bug, so now what?10:10
wpwrakbartbes: i could offer lead ;-)10:10
wolfsprauldo you have a screwdriver?10:10
wolfspraulturn the nanonote around10:10
wolfspraulremove battery, unplug USB cable10:11
wpwraktuxbrain_away: speaking of which ... i still have those 20 counterweights waiting for you. how did the enquiry with TNT go ?10:11
wolfspraullooking from the bottom, you should see 4 screws.10:11
wolfspraulunscrew them, all 410:11
wolfsprauleven though we insisted on them not doing it, one of the 4 screws may be covered under a small white 'do not tamper' label10:11
qwebirc66005wolfspraul: I already opened (and by the way there were just 3 screws)10:12
wolfspraulthe one on the left side of the battery (above the battery connector)10:12
wolfspraulno there are 410:12
wpwrakwolfspraul: i'm afraid there may not be many alternatives to plastic. metal would be nice in some regards but also difficult to manufacture.10:12
wolfspraulqwebirc66005: look on the left side of the battery, above the battery connector10:12
wolfspraulit may be hidden under a small label10:13
qwebirc66005wolfspraul: well, there is a hole under the label, but no screw...10:13
qwebirc66005wolfspraul: really..10:13
wpwrakwolfspraul: wood would be fun for milling (you can do wood very rapidly so that should be cheap). but then you need thick walls or it'll break10:13
wolfspraulwas the label intact?10:13
wolfspraulyour device gets more mysterious...10:13
qwebirc66005wolfspraul: yes, as I can see10:14
wolfspraulok, so when all 4 (or 3) screws are out, turn the device around10:14
wolfspraulwait a sec I'm behind I am following here with mine...10:14
wpwrakfabrication date: monday, after the chinese equivalent of the oktoberfest :)10:14
wolfspraulqwebirc66005: ok now comes the hard part10:15
wolfspraulyou need to lift off 1 piece of plastic10:15
wpwrakwolfspraul: also, only things you can cast would scale to larger quantities10:15
wolfspraulit's the big piece that is all around the keyboard (and has another piece of metal sitting on top of it, over the speaker/microphone/power button)10:16
wolfspraulunfortunately Werner doesn't have a scan of that piece yet, but maybe you can imagine10:16
wpwrakwell, cast, stamp, or similar10:16
wolfspraulsince you unscrewed the 3 screws, the only thing that holds it down now are little plastic hooks on the sides10:16
wolfspraulwait let me try this here first10:17
wolfspraulyes I just took it off :-)10:18
qwebirc66005wolfspraul: I already disassebled it, I disconnected the cable too, opening the connector latch10:18
wpwrakwolfspraul: (scan) this one will be hard. i already looked into scanning it, but there's no way to mount it without the fixture showing. and even then there's the problem that there are few even surfaces. i'll probably have to construct a specific fixture for it. similar for scanning the outside of the bottom shell. but that one's much easier.10:18
wolfspraulqwebirc66005: wow!10:19
wolfspraulwell you don't need my assistance, you should become my teacher...10:19
wolfspraulwell then10:19
Ornotermessome new images and stuff -> http://slashhome.se/p/uSD_breakout/10:19
wolfspraulthere is actually no need to pull the cable out entirelly10:19
Ornotermeswpwrak: ^10:19
wolfsprauljust open the latch, pull it out a little, and push it all the way back in10:19
wolfspraulthen close the latch, turn on, see what happens10:20
wolfspraulcongratulations for the disassembly btw!10:20
qwebirc66005wolfspraul: waitasec...10:20
wpwrakOrnotermes: nice and tidy :-)10:20
wpwrakOrnotermes: next step: replace the header with a PCB plus header assembly. like the 0.1" connector i attached here:http://www.almesberger.net/misc/idbg/idbg-soldered.jpg10:21
wolfspraultuxbrain_away: should seriously consider offering this as a hacking kit for the NanoNote (usd_breakout)10:22
wpwrakOrnotermes: after that, you'd almost have industrial production quality. the only thing missing would then be the flat cable10:22
wolfspraulqwebirc66005: a word of caution. When you lift off that plastic piece, and also when you put it back in later, please watch 2 things10:22
wolfspraulfirst - there is a thin piece of metal glued onto it. Some people don't believe it but it is metal.10:23
wolfspraulwhat that means is if you bend it, you will not be able to bend it back, it will have some very ugly 'waves' or worse10:23
wolfspraulso never try to take that off the plastic10:23
wpwrakwolfspraul: (breakout board) i guess you'd have to produce the cables. then find a source for cheap and small prototype boards, like the one on the picture. then you'd have a nice package.10:24
qwebirc66005wolfspraul: sorry, same problem, I think the issue is on the other side10:24
wolfspraulsecond, if you look at the bottom side of the piece of plastic, you see a total of 5 plastic 'hooks' that produce this 'click in' effect10:24
wpwrakwolfspraul: add some components and you can even do experiments. "kosmos"-style, if you know them from germany10:24
wolfspraulwhen removing or putting the piece of plastic back in, try to be gentle to those hooks or they can break off (which is not that bad because the screws still hold everything quite well10:25
wolfspraulqwebirc66005: hmm. can't really be on other side.10:25
wolfspraulyou said when you pressed in the middle it worked for a while?10:25
wpwrakhairline fracture in the cable ?10:25
wolfspraulmaybe it's a problem there? anyway. sounds like you need a replacement.10:26
wolfspraulwell he said he pressed in the middle and the screen came on?10:26
wolfspraulthere's some beads there, maybe those have a problem? don't know10:26
qwebirc66005wolfspraul: yes it worked for a while, but not anymore10:26
wpwrakhairline fractures respond to the weirdest things ... you just need to push the cable in "the right way"10:26
wpwrakwolfspraul: you should remember the fun we had with the debug cable at openmoko. this one also suffered this sort of fractures.10:27
qwebirc66005wolfspraul: now if I press in the same point, the screen become full-white (what a progress...)10:28
wolfspraulwhere is 'the same point'?10:28
qwebirc66005wolfspraul: just above the hinge10:29
wolfspraulfrom the left, you first have the speaker, then NAND, then a few passive components and beads, then the FPC, then microphone, then power button10:29
wpwrakOrnotermes: (led) in R(on) we trust ;-)10:29
wolfspraulabove the hinge?10:29
wolfspraulyou mean on the side of the LCM?10:30
qwebirc66005wolfspraul: I press on the lid , below display, above the hinge10:30
wolfspraulyou need a replacement! that's what will help you :-)10:31
wolfsprauljust reassemble everything neatly, be as careful as possible we definitely would like to look into this a bit more once it's back one day...10:32
wolfspraulfor the replacement, since you bought from hackable-devices I'm not sure. maybe the easiest is if Tuxbrain just gives you a new one, and you return this one to Tuxbrain?10:33
wolfspraulis that acceptable to you?10:33
wolfspraulwhat do you suggest? (since you have the device in front of you and you know what you planned to do with it...)10:33
qwebirc66005wolfspraul: were is TuxBrain?10:34
qwebirc66005wolfspraul: okI think it's my best choice. How I should proceed?10:35
wolfspraulI suggest you email me, then we take it from there.10:36
wolfspraulsorry for the inconvenience, at least you got to do some beginner's hacking on your first day already :-)10:36
qwebirc66005wolfspraul: :-)10:36
wolfspraulfrom the distance it looks like the second or third device with this kind of LCM problem10:36
wolfspraulin about 95010:37
kristianpaulwolfspraul: whare are iphone case made off?10:37
kristianpauli like how it feels10:37
wolfsprauloh my god10:37
wolfspraulit's magic!10:37
wolfspraulit's probably a super special super secret super whatever thing10:38
wolfspraul200 material scientists, phd's, helped by engineering teams from another 50 partner companies, worked together on this magic substance10:38
wolfspraulwell but to give Apple credit, they do dig deep10:38
wpwrakwolfspraul: too bad they forgot the antenna designer :)10:39
wolfspraulso seriously, I totally don't know, and probably nobody but apple does10:39
kristianpaulok forgot it10:39
kristianpaulwpwrak: lol10:39
wolfsprauland if any of their partners who knows speaks even 1 word, no more apple business forever10:39
wolfspraulqwebirc66005: please email me at wolfgang@sharism.cc10:39
wolfsprauland also, please reassemble the nano nicely, I definitely want to find out more about what happened.10:40
kristianpaulwpwrak: there are ceramics cases?10:40
kristianpaulmay be something related with alumnion i dont know, any way seems plastic is cheaper isnt?10:41
wpwrakkristianpaul: haven't heard of any. at least traditional ceramics tend to be brittle.10:41
wpwrakkristianpaul: afaik, aluminium needs to be milled. so high per-unit cost.10:42
qwebirc66005wolfspraul: email sent10:42
kristianpaulyeah i tought that about ceramics is a shame :(10:42
kristianpaulwpwrak: plastic then10:42
wpwrakkristianpaul: there probably some fancy new ceramic that don't have such problems. but that's an entirely new world :)10:43
wolfspraulas usual, Apple is seriously investing lots of money and brain power into materials10:44
kristianpauli like that, i was told from an mechanic eng about that is been used in europe recently in some areas10:44
kristianpaulthat = ceramics10:44
wolfspraulwpwrak: for the breakout board, are you worried about the cable being too long in terms of the signal?10:44
kristianpaulalso re-usable mirenal i was told10:45
wolfspraulwhat if the cable is just an FPC? not stable/shielded enough? (the pictures show individually shielded round cables)10:45
wpwrakwolfspraul: eventually, cable length becomes a problem, yes. but i wouldn't worry about that for basic experiments. (led, button, this sort of stuff)10:47
Ornotermeswpwrak: you mean another board to make 50 mils ribbon to 100 mils pin head?10:47
wolfspraulqwebirc66005: didn't get the mail yet10:48
wpwrakwolfspraul: going to fpc is what i'm recommending for "serious" production. the ribbon cable on the pictures is unshielded. this is just the basic stuff you also have, say, in IDE connectors (before SATA)10:48
wpwrakOrnotermes: yup. should be easier to solder and more stable10:48
wolfspraulwpwrak: how long do you think the FPC should be?10:49
wpwrakOrnotermes: just solder everything flat on the PCB, SMT-style10:49
wpwrakwolfspraul: as short as possible but long enough for handling when you make experiments :)10:50
wpwrakOrnotermes: what do you think, 20 cm ?10:50
wolfspraulcan you translate that to centimeters?10:50
wpwrakmaybe 15. probably not shorter10:50
wpwrak.. than 10 :)10:51
Ornotermesi have 20cm now10:51
wpwrak20 looks decent10:51
Ornotermesi think thats pretty good10:51
wolfspraulwhat pitch do you have on the other side? (the breadboard side)10:51
Ornotermeson the pinhead10:52
wolfspraulthat's the most common?10:52
wpwrak100 mil, standard header10:52
wpwrakalso DIP chips have a 100 mil spacing10:52
wpwrakyou get a gazillion header types for 100 mil10:52
Ornotermesa importaint thing for cable lengt is how high frequencys it should handle10:53
wpwrakyup. of course, nobody will to microwave designs on a breadboard .. :)10:54
wolfspraulqwebirc66005: can you give me your email address here? still didn't get your mail. just in case, mine is wolfgang@sharism.cc10:54
qwebirc66005wolfspraul: dsplabs<at>dsplabs.net10:55
wolfspraulok thx got it10:55
Ornotermesi have som capacitace measures now :)11:01
wpwrakOrnotermes: and, how bad is it ? :)11:02
Ornotermeswpwrak: not too bad, working on table...11:07
Ornotermeswpwrak: its a table on my page now11:14
wpwraknot so bad11:19
wpwrakand now the inductance :)11:20
Ornotermesi think that's a bit hard for me to measure right now :P11:27
Ornotermesand i think it's a even lesser problem in the real wold since all ports have pull ups11:30
Ornotermesand to send clock signals D09 has the advantage of being between power leads11:31
qi-bot[commit] Juan64Bits: ERC report generated. http://qi-hw.com/p/xue/4402fb511:36
wpwrakOrnotermes: a good test would be so simply send out a signal and watch it on a scope.11:39
Ornotermesbut i'm a bit after on the software11:39
wpwrakthis program generates a 16 MHz MMC clock signal11:41
wpwrakit's a bit ugly, sorry about that. for "nice" software, see http://projects.qi-hardware.com/index.php/p/ben-blinkenlights/11:42
Ornotermeshow do i compile it?11:46
Ornotermesit was something about linking to the the compiler on the deskop system?11:47
wpwrakah, you need the toolchain. if you follow the openwrt build instructions, they will generate one for you. then you link the executables (mipsel-openwrt-linux-*) somewhere into your PATH. (or adjust the PATH)12:32
Ornotermesi think i have the binarys but i don't have any alias to the compiler12:35
wpwrakit's a pity there is no simple host toolchain package. in openmoko, we had a tarball with the toolchain and the most important libraries (you need these too), and i think it worked very well. someone should do the same for the ben.12:35
wpwraklemme see where it hides ...12:36
wpwrakyou can just ln -s wherever.../openwrt-xburst/staging_dir/toolchain-mipsel_gcc-4.3.3+cs_uClibc-0.9.32/usr/bin/*  /usr/local/bin/12:37
wpwrakwhen you run the things these, they will remember the include paths in the openwrt tree, so #includes pick up the right files12:38
Ornotermesoh, ok12:39
Ornotermesthis is one of the reasons to prefer languages like python :P12:39
wpwraknaw, there's nothing magic about cross-compilation. someone just has to set up the basics.12:52
wpwrakthe problem with not having an easily available toolchain is that people may start to develop irrational development models12:53
wpwrake.g., make their build processes depend on a distribution12:53
wpwraki've seen this at openmoko, where they used openembedded as their build system. so instead of just invoking "make" on their project, they would feed it into their local OE system and then run an OE build. naturally, with massive overhead.12:54
alcyfolks, has ruby been ported (or is present currently) to the nano-note ?12:55
wpwrakOE would take care of the "magic" of doing the cross-compilation, but just in a horribly inefficient way. and of course, many were unhappy about the problems they encountered in this model.12:56
wpwrak(this wasn't purely insane, since some of the work depended on libraries and such that were somewhat hard to obtain or even part of the development. but still.)12:56
kristianpaulwpwrak: so OE is blobbed , bitbake may be?13:03
kristianpaulalcy: ruby is openwrt repos13:06
alcykristianpaul, cool..thanks !13:07
kristianpaulalcy: you need supprot about how install it?13:08
kristianpaulwpwrak: i personlly see OE kidn of blobbed, comapred to openwrt build process, but damm they have thousands of packages13:09
Ornotermeswpwrak: i added it to the path instead13:09
alcykristianpaul...no, no...just that I am unable to give time to my nano-note...so, just to be involved with it ( for my sysadmin stuff), wanted to know about ruby...hopefully irb will run on it. :)13:09
emebcould someone please define how the term 'blob' applies to OE?13:18
emebI've always associated 'blob' with proprietary binary code, distributed but not open to review or change13:19
emebOE seems not to fit into that view13:19
kristianpaulohh sorry !!13:20
kristianpaulemeb: ^13:21
emebkristianpaul: thanks for the clarification13:21
kristianpaulemeb: you for point it13:21
emebno argument that OE is a huge buildsystem13:21
emebthat said - it seems to work pretty well (I use it on beagleboard)13:21
kristianpaulyeah me too on the nano13:22
kristianpauljust the build system is not so simple as in openwrt i think13:22
kristianpaulfor example bitbake..13:22
emebI've also been looking at buildroot for really simple stuff.13:23
kristianpaulwich files?13:23
kristianpaulkristoffer: hey13:23
kristianpaulkristoffer: did you documented how was added support for ben nanonote in OE?13:23
qi-bot[commit] Juan64Bits: Routing current sensors. http://qi-hw.com/p/xue/cb21a9a13:23
kristianpauli think i have clear about build the userland13:24
kristianpaulbut no else...13:24
kristianpaullike what to do if i want apply a patch to a package13:24
kristianpaulor build with gcc especific arcuitecture options13:24
kristianpaulyou know, ge tot the custom13:25
kristianpauls/ge tot/get to13:25
kristofferkristianpaul, one sec13:27
Ornotermeswpwrak: got it to work :)14:07
Ornotermesbut i didn't got 16 MHz14:07
Ornotermeswpwrak: http://gallery.slashhome.se/main.php?g2_itemId=3007 MSCCDR = 014:17
emebOrnotermes: how do you like that gw instek scope?15:05
Ornotermesemeb: i don't have that much experience in scopes, but it serves me well15:12
emebthx - trying to choose between that & Rigol15:13
Ornotermesthere is some things that feels odd but it doesn't have to be the scope it self15:13
Ornotermeslike if i measure something thats 0V it shows a slightly negative voltage15:14
Ornotermesand there is some over hearing between the channels15:15
emebdoes that happen at all vertical scales, or only on the sensitive settings?15:16
Ornotermesemeb: i try to find a dumt that shows what i mean15:16
wpwrakOrnotermes: nice signal. the clock frequency is odd, though.15:16
Ornotermeswpwrak: have you checked the signal output your self?15:17
wpwrakOrnotermes: yes, I get a perfect 16.00 MHz.15:17
Ornotermescan it be that my scope is too slow?15:18
Ornotermesit's a 40MHz scope15:18
wpwrakOrnotermes: doubt it. if it's too slow, this usually means that the analog bandwidth it soo low15:19
wpwrakwhat is the sample rate ?15:19
Ornotermeswpwrak: http://www.valuetronics.com/New_Instek_GDS-1042.aspx15:20
wpwrakemeb: some rigols have very deep memory. that's extremely useful for analyzing complex signals, particularly digital ones15:20
emebwpwrak: yes - even the low-end Rigols have 1Msamp15:20
wpwrakemeb: (i have a ds1102c) also the logic analyzer is a rather nice feature, even though it has a number of design flaws.15:21
emebwpwrak: I've seen that. Seems handy, but too much in one box15:21
emebI'd rather have a separate LA15:21
emeband spend more on analog BW15:22
emebLooks like the 150MHz Instek scope has a good price and nice features15:22
wpwrakOrnotermes: 250 MSa/s. so, sample rate should be no problem15:22
wpwrakemeb: 2 MSa, nice15:24
wpwrakOrnotermes: maybe my program is incomplete. you used clk.c, right ?15:25
wpwraklemme try is on a freshly booted system15:27
Ornotermesemeb: http://gallery.slashhome.se/main.php?g2_itemId=3010 and http://gallery.slashhome.se/main.php?g2_itemId=3013 is what i mean15:27
Ornotermesit might be from cheap probes or something like that so i can't really say it's the scopes fault :P15:28
wpwrakcheap probes generate distorted signals. yours looks quite clean15:31
emebOrnotermes: yes - that HF feedthru on the edges could be grounding issues with your probes. Not necessarily the 'scope.15:33
wpwrakemeb: "could" ;-)15:33
Ornotermesas i said, im not very experienced in scopes... but as always (exept apple) you get what you pay for15:34
wpwrakOrnotermes: ah yes, there's something wrong with clk.c . now i get 250 kHz :) let's see where i screwed up ...15:34
emebwpwrak: yes - deep trace memory, 1Gsps realtime, 150MHz BW for < $1k - seems good15:34
Ornotermeswpwrak: the importatint thig is that we get the same results :)15:35
wpwrakOrnotermes: (scope) 2 things to check: probe compensation and ground loop. if your probes some with little springs to attach to the probe head (after pulling off the clip), try once to measure with tese15:36
wpwrakOrnotermes: the wire with spring goes to ground, so you have to have one nearby. you should see quite a difference15:36
wpwrak(freq) i probably write to the wrong register somewhere15:37
Ornotermeswpwrak: i check probe compensation once in a while, but it might be the grounding15:37
wpwraki tried the settings first with "poke" and then updated clk.c, so if i missed a register that already had a good value from poke, that would explain it15:38
wpwrakOrnotermes: grounding problems are very common. if you're interested in the topic, you may want to read about resistive probes15:39
wpwraka while ago i had found a beautiful article about them that compared them with other probe types ... searching...15:40
Ornotermeswpwrak: thanks for the tip, it made _some_ difference15:41
Ornotermeshttp://gallery.slashhome.se/main.php?g2_itemId=3016 (both probes grounded near probe) :)15:42
wpwrakgood. i'm not sure about the dc offset. maybe the scope needs calibration. that should be somewhere in the menus.15:42
wpwrakyou may also want to check if there are any firmware updates. such small bugs are often fixed via firmware.15:43
Ornotermesprobably, when i use auto set it often changes but i haven't found any really manual settings15:43
wpwrakon any case, i wouldn't worry about the scope being off by a bit. they're not particularly good at measuring voltages anyway15:44
Ornotermesno i know15:44
Ornotermesthere is a reason why i don't just use a multimeter :P15:45
Ornotermesand why i some times does15:45
wpwrakcalibration is an item that should be in tje "utility" menu or so. it's a process that keeps the scope busy for some time. you also have to disconnect all probes, etc.15:46
Ornotermesyes, i tried it, but then abain, it might be my cables15:46
wpwraksome scopes also need recalibration during operation. some do this on their own (which can be a problem), others not.15:46
wpwrakmaybe check if your ground is really connected :) sometimes, it detached but you don't notice15:47
wpwrakfunny that you get 5 MHz. i found why i get 250 kHz.15:50
Ornotermesi got 5MHz when i changed the value15:52
Ornotermes"[...]hhome.se/main.php?g2_itemId=3007 MSCCDR = 0"15:53
wpwrakah !15:54
Ornotermesthanks, i try it a bit later15:55
wpwrakyes, that would be correct :)15:55
wpwraki wrote to the wrong register and thus didn't turn off the 1/64 divider15:55
wpwrakso i got 250 kHz. you changed the 1/21 divider to 1/1, so you got 250 kHz * 21 = 5.25 MHz :)15:56
Ornotermeseasy happened15:56
wpwraki picked the register from the wrong bank and didn't notice :-(15:57
Ornotermesbut shouldn't a scope have 10x higer frequency than the signal you measure?15:58
kakiharacan someone help me with cloning my hdd with northon ghost to a bigger one? i got some problems there...15:59
wpwrakOrnotermes: if you want to check signal properties, yes16:16
wpwrakOrnotermes: if you just want to check the frequency and see if attenuation isn't too bad, a slow scope will do16:17
wpwrakOrnotermes: a slow scope basically reduces your square wave to sinus16:17
Ornotermesbut then 16MHz might be a bit much for my scope16:18
wpwrakthe amplitude will still be indicative. and that's what we care about here.16:21
wpwrakit will not show a perfect square, no matter how good the signal is, though.16:21
lekernela slow scope basically reduces your square wave to sinus ...when you're lucky :p17:06
lekernelI had a digital oscilloscope that routinely displayed all kinds of crap when used over its designed bandwidth17:07
rohwell.. duh17:08
rohyou know the saying 'wer mist misst misst mist'?17:09
emebgerman is so full of wonderful farm-based aphorisms17:19
rohits quite yoda in the end... if you measure bullshit, your measurement bullshit will be... would be a loosely sensemaking translation ;)17:20
emebin english: garbage in, garbage out - GIGO17:21
emebanother good one: Wenn der Hahn kräht auf dem Mist, wird das Wetter ändern oder bleiben wie es ist17:22
Ornotermeshttp://slashhome.se/p/uSD_breakout/ <- all exept images har licences now17:42
Ornotermeswpwrak: still 250kHz :S18:00
wpwraklekernel: if you go over the sample rate, bad things happen :) but that shouldn't be a problem here18:31
wpwrakOrnotermes: (250 kHz) hmm, are you sure you got the new file all along the path ? (download, compilation, upload to Ben ?)18:32
wpwrakOrnotermes: to check the source, the important line is #defineMSC_CLKRTMSC(0x08)18:33
wpwrakOrnotermes: it used to say  ... CGU(0x08)18:33
Ornotermeswpwrak: i'm quite sure i done things right(line 49 ok, deleted binary by self, compiled, deleted files on ben, compared md5) but its still 250k :S20:06
Ornotermesoh well, time to sleep. bye for now20:26
wolfspraulhah, just got up20:35
wpwrakOrnotermes: hmm, then it would be time to try poke. http://svn.openmoko.org/developers/werner/poke/20:37
wpwrakcompile for MIPS, then run, on the Ben,  ./poke 0x1002100820:38
wpwrakif it responds with 6, then you're using the old version :)20:38
wolfspraulwpwrak: I had another question about the usd breakout board. What would be a good way to test the cable? data0-3, cmd, clk, vdd, gnd20:46
wpwrakyou mean at a factory ?21:02
wpwrakor to test a cable you just made ?21:02
wpwrakor to test a cable design ?21:03
wolfspraulit's hypothetical (I didn't make a cable), but I mean a cable I just made21:21
wolfspraulwhich of the wires can be controlled individually from userspace? and what would you connect them to on the breadboard to test. leds?21:21
wolfspraulto rephrase - what is a good test exercise for each of the wires?21:23
wpwrakyou can control all but GND. for VDD, you're limited to on/off. i.e., you couldn't send a 0, only 1 and Z. for the others, you have 0, 1, and Z.21:25
wpwrakto test, i would just use a led array and do the the blinkenlight. if you look at the schematics, it's very very simple.21:25
wpwrakif you want to optimize it for testing, do this:21:26
wpwrakconnect one LED with the anode to each signal line (data, cmd, clock), all the cathodes together, and through a 100 Ohm resistor to GND.21:27
wpwrakconnect one more LED with the anode to VDD and the cathode to the others21:28
wpwrakthat way, you have 7 LEDs, can control each of them, and can test the connectivity of all lines21:28
wpwrak(my blinkenlights don't use VDD and GND and use a multiplexing trick to control 10 LEDs with 6 lines.)21:29
wolfspraulhe, that perfectly answered my question. so that would be a setup where a simple program could go through all leds, and all wires are verified?21:30
wolfspraulthat program does not need a kernel module, just run as root in user mode, correct?21:30
wpwrakjust like my bbl.c, yes21:31
wpwrakat some point, we should upset the MMC kernel driver, though. not sure why this hasn't happened yet :)21:32
wpwraki.e., it should detect that *something* is going on, try to talk to the card, and get a bit frustrated by its lack of success. well, maybe it did that and i just didn't notice :)21:33
wpwraksince, in the DIY case, all the lines are assigned to GPIOs and not the MMC unit, there's little the kernel driver can do, though. except switching off power perhaps. not sure if it does that.21:34
wpwrakwhen using the MMC controller to output a clock signal, interference of the MMC driver could be a bit more of an issue. we'll see. if all else fails, you can always build it as a module and choose not to load it.21:35
wolfspraulunfortunately I don't think we can just always build it as a module, then we would need two different kernel images for NAND and SD boot21:36
wolfspraulso let's hope it behaves well for now... :-)21:37
wpwrakyou could use an initrd/initramfs :)21:37
wpwrak(heh, we solved that something like 14 times ago. how time flies ;-)21:38
larscyou can disable the mmc driver by unbinding it with `echo jz4740-mmc.0 > /sys/bus/platform/driver/jz4740-mmc/unbind`21:45
wolfspraulgreat, module problem solved...21:49
kristianpaulwhat cable are you talking about btw?21:57
kristianpaulcable from uSD connector?21:57
wolfspraulthe breakout stuff http://www.slashhome.se/p/uSD_breakout/21:59
wpwraklarsc: ah, right, that's even cleaner22:01
wpwrakthe mechanical details of the uSD holder are interesting. there's a little piece of spring-loaded plastic inside for the locking mechanism22:02
wpwraknow, i wonder if it's possible to return it to its original position when it's been pushed to the rear :)22:03
wpwrak(the uSD slot still works. just doens't click anymore)22:03
Action: kristianpaul likes the cable better than the messier gpio wires i have here :)22:04
wpwrakit's tons better22:04
kristianpauland cleaner22:04
kristianpaulmy ben dint look well after dissasembly22:05
kristianpaulwolfspraul: planning to manufacture that cable? :D22:05
wolfspraulI want to be careful to not overcommit. Yes I would like to but I don't know when I get to it, really.22:08
wolfspraulgood thing it's so simple I can probably give this job to any small street shop. but it's still a lot of work to pull it all off, oversee quality, ship stuff, etc. I added it to my todo list :-)22:09
wolfspraulplease don't hold your breadth22:09
kristianpaulok breading now :)22:10
kristianpauli think i will make one !22:10
kristianpaulnow i will be able to carry my ben on on the pocket with no wires getting out it !! :)22:11
wpwrakwe're getting there. first idbg, now uSD. not a lot missing :)22:13
kristianpaulahh damm the design was made in eagle?22:14
kristianpaulor my kicad got crazy..22:14
larscI'm wondering if we could find a clean solution for controlling the uSD break out board22:16
kristianpauli like wpwrak example22:16
larscit's pure evil22:18
kristianpaulbut have comments !22:19
larscbut it mmaps iomem regions which are already used by the kernel22:20
wpwrak(clean solution) a kernel driver than cleans up after closing a file descriptor would be nice.22:20
kristianpaulnot good22:20
wpwraklarsc: perfectly valid :)22:20
wolfspraulkristianpaul: about kicad vs. eagle - is there much (any) electrical design in this? what is the difference between blinkenlights and Ornotermes board? I thought it's just wires running out, no?22:20
kristianpaulwolfspraul: yeah not big deal22:21
kristianpaulwpwrak: fpade where you got the source?22:22
wpwrakbesides, you can just take the uSD card footprint from mine and add those wires22:22
wpwrakfped ? that's on svn.openmoko.org/trunk/eda/fped/22:22
kristianpaulahh last place i may consider :p22:23
wpwrakfedora has a nice page about it: https://fedorahosted.org/fedora-electronic-lab/wiki/PCB/fped22:24
wpwrakyeah, it's a little hidden :)22:24
kristianpaulyes i noticed that at first22:24
kristianpaulhey i also i think i can make a small case for it :)22:25
kristianpaulnothing fancy ofcourse22:25
kristianpaulwpwrak: yes i got our footprint22:26
kristianpauloh why there is a fped on qi projects too??22:27
kristianpaulwell is good :')22:27
wolfspraulwpwrak: I asked xiangfu to look into making a debian package around fped, and he created a project on the qi server as a repository for that. Once he is finished we can send you the patch, and you can apply it to the openmoko svn or we leave everything in the qi git. up to you. I didn't know he would create a public repository for the Debian packaging work :-)22:30
kristianpaulbot sound bad cosidering you have have an already xbusrt tools too for the repositoty22:32
wolfspraulthere are several things. first I want to have a debian package, makes it easier for me to spread the tool. Whether Werner wants to have that or not I don't know.22:34
wpwrakwolfspraul: (package) heh, cool :)22:35
wolfspraulthe official homepage for fped is svn.openmoko.org/trunk/eda/fped/ . some places already picked this up, like Chitlech's Fedora Electronics Lab, so why change it?22:35
wolfspraulI think if you googe for fped you also quickly find that url22:35
wpwraknaw, it's invisible on google. the only thing you get is fedora's page22:36
wolfspraulok, let's see when we have a debian package first22:37
wolfspraulthat's what I am focusing on22:37
kristianpauli was pointed to FEL at first22:37
rohwolfspraul do you think it makes sense to use the ubuntu ppa system for that?22:38
roha 'kicad, fped etc, qi-hardware edition' builds ppa ;)22:38
larscwpwrak: do you think something like http://pastebin.com/dWiFq7AY could work?22:47
wolfspraulroh: never heard of ppa. I'd rather help getting it into Debian, no?22:49
wpwraklarsc: looks good, yes. that way, one wouldn't have to worry about cleanup, particularly if the application doesn't terminate properly22:51
wpwraklarsc: you only need uint8_t, though. also, would that include power or would you switch it separately ?22:51
rohwolfspraul ppa is the ubuntu 3rd-party-binary-feed way of doing things. source-wise its the same for you. but i think it makes sense to make use of their buildbots22:52
rohwolfspraul also it makes installing unimaginable easy. one just pastes one line into his terminal or adds one into a click-tool. no manual editing of files22:53
rohhttps://launchpad.net/~paxer/+archive/ppa has some packages of kicad, but not with the patches you use22:54
larscwpwrak: ah, right, gnd can't be changed... not sure about power, what do you think?22:55
wpwraklarsc: another use is to emit a precise clock (from the MMC controller). but perhaps that's already too sophisticated for this interface.22:55
rohwolfspraul check the link 'read about installing'22:55
wpwraklarsc: i think power needs to be controllable. could be in the write or via some ioctl, though22:55
wpwrak(clock) i think that's the only MMC that can be meaningfully used for other things than MMC.22:57
wpwraklarsc: if you really want to get fancy, you could also add an interrupt mechanism :)22:58
wolfspraulroh: ok but I still don't understand why I should use this and not just a normal Debian package22:58
rohwolfspraul because i know of nearly nobody who gets a debian feed properly done?22:59
wpwraklarsc: e.g., an ioctl that makes the next read block unless a certain interrupt has happened. that way, one could also select/poll for it.22:59
wpwraklarsc: if you add all this, i can even write the wpan driver in user space ;-))22:59
rohwolfspraul ppa is including signing and buildbots with working environment, not some broken ones which are missing this and that. you get your own vm, with defined conditions, repeatable. building on some devels workstation will nearly always generate broken packages23:00
wpwraklarsc: ah, another thing: pull-ups23:00
wolfspraulroh: sure but we want to get fped included as a Debian package23:00
rohwolfspraul also installing is much easier. you need to paste one line, compared to adding a repo by hand, importing a rsa key by hand etc23:01
wolfspraulI don't want to host my own package repository23:01
wpwraklarsc: and of course input/output configuration23:01
wpwrakgetting complex :-(23:01
rohwolfspraul fped is only 1/4 of the needed work. what does fped help me if i cant even get a working kicad in the same config as you guys?23:01
rohwithout manual building it23:01
wpwrakroh: one step at a time :)23:02
rohwolfspraul thats the beauty of it. ppa allows you to not host it yourself. ubuntu takes care of that for you.23:02
rohand they dont care why you need this or that extra patch. its exactly made for that23:02
rohgetting kicad with qi-hw patches packaged will not happen easily, if the kicad guys dont want your patches i guess23:03
rohin debian that is23:03
wolfspraulone by one23:03
rohtrue.. the work on the package needs to be done first, anyhow (even if i guess most of it is already done)23:04
rohppa will not help you there. ppa is 'just' the component which make you not need to 'set up your own buildhost for N archs'23:05
wolfspraulI think I understand the Ubuntu ppa system now. looks like a way to create pools of packages with weird hacks and patches easier, remove the need of communicating with upstream about a unified solution. don't like it.23:06
rohwolfspraul wrong.23:06
wolfspraul:-) just thought since it's you I can afford to be a bit opinionated :-)23:06
rohits a pool of special interrest packages, which are not intended for 'the masses'23:06
rohits to enable developers e.g. distribute builds of test-releases in binary form to a bigger poool of testers23:07
rohwolfspraul do you have x86 and x86-64 buildhosts for multiple disto releases at home? i dont. and i dont know any devel who has.23:08
rohalso for nightly builds it nice. (same there.. test releases)23:09
rohof course its better to get your patches upstream. if thats possible.23:10
wolfspraulyes that's why packages need to go into Debian23:10
rohdebian is not upstream23:10
wolfspraulfor new software, get it there and you will have a buildhost23:10
rohand debian will not add your patches if upstream didnt like it for some reason (usually)23:10
wolfspraulfor patches of existing software, first communicate with upstream, if that totally doesn't work and you insist on your patches making a new piece of software, fork upstream23:11
rohalso getting it into debian will not fix any 'missing packages-' for existing distros. its not that you can get it into stable any way.23:11
wolfspraulanyway I think I got the ppa thing now, thanks for pointing me to it, but won't help with the next steps here wrt fped23:12
rohthats what i dont get. how would debian get us near 'there are packages in the repo which i can apt-get' anytime soon?23:13
rohi dont speak of 'you got a .deb file you can give me'23:13
rohmy debian versions of kicad are that old, that they crash and burn if i try loading any of your project files *sigh*23:14
rohdifferent question: whats the difference between fped and cvpcb ? (besides the origin)23:15
wpwrakcvpcb ? maybe you mean the module editor of pcbnew ?23:18
rohits a seperate app23:18
wpwrakyou mean, in addition to the cvpcb of kicad, there's another thing called cvpcb ?23:18
rohno.. i mean the cvpcb23:19
rohof kicad23:19
wpwrakcvpcb is for associating components with footprints23:19
rohor can cvpcb not edit, but really only select footprints?23:19
wpwrakfped it for drawing footprints23:19
wpwrakcorrect. in kicad, you draw footprints with the module editor in pcbnew23:20
rohwell.. cvpcb is crashing with segfault for me everytime i press the 3d-button23:21
wpwrakyou need wx with opengl support23:21
rohi am sure i got that.23:21
rohsince other apps can do 3d fine, also in wx23:21
wpwrakthen i don't know23:22
wolfspraulroh: Debian has the 2010 March 14 kicad release in testing and unstable23:22
rohwolfspraul unstable.. yeah right.23:22
rohwolfspraul you are talking to an admin. ;)23:22
rohto be fair, i dont have a single debian machine with screen anymore. only servers and sometimes its only some single vm left which is debian.23:23
wolfsprauland the rest is?23:24
rohservers are lts, most still hardy.23:24
rohcan't live with the bs unstable gives me. i need a working workstation, not something which breaks because of somebodys tests. testing on debian is still even worse that unstable (same as the last 5 years) since nobody uses testing if its not 'release in a few days'23:26
rohthe few servers i got left are all debian stable. some even worse (older) for some reason23:27
wolfspraulwho is maintaining the openmoko servers these days?23:27
wolfspraulwith Marco Knoebl leaving I am a bit worried about the wiki. he was deleting lots and lots of spam all the time.23:28
rohheh.. nobody? if something breaks gismo and i take a look and fix it. but it looks more and more like old oil-rigs in nigeria23:28
roh'we keep it flying' .. but can't invest any more than the absolutely neccessary time23:29
rohin the end nobody gives us even a beer for it.23:29
rohi am just doing it 'for the community' .. but bigger things like upgrading the wiki, the trac etc to recent versions, or do major cleanups are not really happening.23:30
wolfspraulsure understand. like I said, with Marco leaving I would be worried about the wiki.23:31
wolfspraulhe was the last strong spam blocker23:31
rohi did some debian updates every few month where easy... some stuff rots .. like the mail spamfilters (which are prone to break, so they are on hold)23:31
rohhm.. not much i can do (if nobody steps up)23:32
roh .oO(contrary to public believe, admins often dont care much about content, but about the stuff which holds it, and yes. they really need to eat food to survive and pay rent!1!!)23:33
larscroh: next time i'm in berlin i'll buy you a beer ;)23:33
rohlarsc ;)23:33
wolfspraulroh: Sean says (last in April this year) he pays gismo a monthly salary http://lists.openmoko.org/pipermail/community/2010-April/061601.html23:35
rohheh... not much of a salary23:35
rohatleast not enough to do more than 'take a look every 2 weeks'23:36
wolfspraulhe :-)23:36
rohserious work like back then, when we planned upgrades and so isnt possible in that time23:36
rohi don't know the exact number (and wouldnt name it here) .. but its not more than what i would name 'taschengeld'23:37
rohthe servers are more expensive, and they are only a few hundred euros23:38
wpwrakroh: the next time you make it to buenos aires, i'll throw a barbecue in your honor. that should at least satisfy your annual meat requirements :)23:38
rohwpwrak hrhr... looking forward to that (even if i dont know when that could happen)23:39
rohanyhow. dont get me wrong. i am happy that sean still pays the neccessary money to keep the stuff online.23:41
wolfspraulroh: we make it to Buenos Aires. Guaranteed! Have to...23:41
rohwolfspraul if you know a way to pay for the flight and some consulting hours i am with you ;) for sure.23:44
Action: roh didnt even make it to bavaria yet this year. only 3 days iceland.23:45
wolfspraulyeah I know. I see a rush of (legit) orders coming in one day...23:45
rohhrhr.. yes. if 'jobs' are there.. its always the same time and space. and one can barely handle it. other times its 'void' *waiting*23:46
wpwrakroh: this is how time for interesting things is created :)23:58
--- Sat Sep 4 201000:00

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