#qi-hardware IRC log for Tuesday, 2012-10-16

qi-bot[commit] Xiangfu: mplayer disable libgsm (master) http://qi-hw.com/p/openwrt-packages/e0f74ab02:16
qi-bot[commit] Xiangfu: libggi: fix depends (master) http://qi-hw.com/p/openwrt-packages/5faa9e402:16
Action: xiangfu finial create my first ugly pcb design. 02:16
xiangfus/finial/finally :) 02:55
kristianpaulpicture picture !02:57
wpwrakmidnight ... marie antoinette remembrance day, the day when we honor our disconnected political elites, has just begun :)03:03
xiangfukristianpaul, I already print it on a photo paper.03:04
xiangfukristianpaul, this will not working on photo paper. since you can see the BGA ball short a lot.03:04
kristianpaulbga :o03:05
kristianpauli tought was tqfp03:06
kristianpaulyou are moving to slx45 now? ;-)03:07
kristianpaulit should not short on the paper...03:07
xiangfuno. it's slx9, ftg256.03:07
xiangfuthe short , you think it is my printer problem?03:08
kristianpauli bet03:08
qi-botThe build was successful: http://fidelio.qi-hardware.com/~xiangfu/build-nanonote/openwrt-xburst.full_system-20121016-0420 03:08
kristianpauli never trust my re-charged cardrige laser printed for those tasks..03:09
kristianpaulbut hey you need a more than two layer pcb anyway...03:10
xiangfugreat. this build will test for release.03:11
kristianpaulgood :-) lts release?03:11
kristianpaulyou should follow owrt release cycles, i havent update my router since years03:11
kristianpauland seems newest fimrmware will no run well anyway so.. ;-)03:12
wpwrakfor diy, i'd strongly suggest to stay away from bga03:12
qi-bot[commit] Xiangfu: mplayer: fix deps (master) http://qi-hw.com/p/openwrt-packages/fe641a303:12
kristianpauland ftg?03:13
wpwrakit can be painful enough to debug things where you don't have to worry so much about the contacts03:13
wpwraklet's see what ftg is ..03:13
kristianpaulsame :)03:13
kristianpaulto me eyes03:13
wpwraknaw, don't try bga for your first project. use something simple. tqfp or such.03:14
wpwrakthe next step is qfn / mlf. some are pretty friendly, i.e., the ones with inclined sides. others less so, i.e., the ones with vertical sides.03:15
kristianpauland let said you got soldering all the balls, you still need more layers to route all chip...03:16
wpwrakdepends on the pitch and on how many balls you actually need. but yes, in general, bga isn't suitable for diy pcbs. the best you can hope for is diy aoldering.03:18
xiangfuftg256 is bga.03:18
wpwrakxiangfu: how do you plan to solder it ?03:19
wpwrakxiangfu: and do you expect to make your own pcb, or have the pcb made at a fab ?03:20
xiangfudirect use the hot air or an oven. 03:20
kristianpauloven oven !03:21
kristianpaulhad you previous hot air esperience btw?03:21
kristianpaulbe careful03:21
xiangfuwpwrak, about pcb, I want do that parallel, etching and send the file out to some fab. 03:21
xiangfu100RMB give you 10cmX8cm. 10 pcs. it needs 4~5 days03:21
xiangfukristianpaul, hot air. not much. I have a useless pcba for test. :)03:22
wpwrakfab sounds reasonable for the pcb. that may allow you to have vias under the fpga. in general, it's almost unavoidable to have vias there (with a bga). and diy can't do that.03:22
wpwrakstill, i'd try to find a nice *qfp package if i was you :)03:23
wpwrak(or qfn, but i don't think they use that for fpgas. qfn is harder to solder than tqfp/lqfp but still manageable.)03:25
kristianpaulbut xilinx..03:25
kristianpaul7 series seems have all from bga03:25
kristianpaul:-/03:27
xiangfuwpwrak, my first plan even complicate. with all power ic and caps on pcb. 03:27
xiangfuthen I found I have problem on routing wires. 03:27
kristianpaultought csg is 0.8mm pitch.. 03:27
xiangfuso I remove again and again. then it only have the ftg256 chip and osc. :)03:27
kristianpaulah ftg is .2 bigger03:28
xiangfuwpwrak, yes. since I have those qfp pin-out board. I direct jump to bga. for new xilinx chips.03:28
xiangfukristianpaul, yes. that is 1mm pitch.03:29
wpwrakXC6SLX9-2TQG144C doesn't sound too evil. lqfp. 0.5 mm pitch is a bit tight, though.03:29
kristianpauli like that03:29
xiangfukristianpaul, most of xilinx chip is 1mm. only CSG and SBG is 0.8mm03:30
wpwrakyeah, only bga for series 703:30
wpwrakbut ... if experimenting with series 6 now, AND it's your first pcb, there's little point in making your own life miserable03:31
kristianpaultought you can follow azonenberg adventures on bga :)03:31
wpwraki'm sure the wright brothers dreamed of flying to the moon, but they had the good sense of not trying their first flight with a supersonic rocket engine ;-)03:32
kristianpaulwere diferent times ;-)03:34
xiangfuwpwrak, about XC6SLX9-2TQG144C, after follow your advices. I made 3 quiet stable board. now it is like: http://www.openmobilefree.net/wp-content/uploads/2012/10/IMG_1190.jpg and http://www.openmobilefree.net/wp-content/uploads/2012/10/IMG_1192.jpg03:34
xiangfuwpwrak, the cable that for osc. it have GND wrap the singal around. 03:35
wpwraknice :) it's getting better03:35
xiangfuwpwrak, (not trying their first flight with a supersonic rocket engine ) :-)03:35
xiangfukristianpaul, yes. I use azonenberg libs.03:36
wpwrakit'll still be a miracle if it works, but you're one step closer to something that might :)03:36
kristianpaulwrap?03:37
kristianpaulwhy?03:37
xiangfuwpwrak, I have a jtag-communication, can read the clock counter out. for 1000 time, it give 3 ~4 errors. still not stable enough.03:38
wpwrakwhy "why" ?03:38
kristianpaulwhy wrap things on a pcb with a fpga :)03:39
wpwrakxiangfu: yeah, that's roughly what i'd expect. circuits rarely fail 100%, but if you abuse them badly enough, they fail noticeably often.03:39
wpwrakkristianpaul: because he hasn't made his own pcb yet :)03:39
wpwrakalas, that adapter board is likely to do more harm than good03:40
wpwrake.g., i doubt is has correct bypassing03:40
xiangfuwpwrak, I have to jump to BGA now. :) or it will take me too much time . good thing is I don't want all 256 pins out. I only need ~10 of them. so I want keep it 2 layer.03:40
wpwrakand the long cables probably act are excellent antennas. so maybe the board works at daytime, when the neighbour is away, but fails when he's back in the evening, downloading porn over wifi03:41
xiangfuwpwrak, 4 layer will totally not fit for me. I can read the 4 layer pcb. I can not create it now by kicad. :(03:41
xiangfuwpwrak, :-D03:42
wpwrakwhy do you think bga will help you save time ?03:42
xiangfuwpwrak, (, i doubt is has correct bypassing) what do you mean ?03:43
xiangfuwpwrak, skip qfp-pcb, direct go to bga-pcb. I think it will same me 50% time of qfp-pcb. :)03:43
wpwrak(bypassing) capacitors, placed very near (not more than a few mm) to the fpga's pins03:43
xiangfuwpwrak, oh. yes.03:44
wpwrakwell ... in addition to the circuit problems you'll have to fight the soldering problems. and if you discover that you need to access some pin that's unconnected, you have to make another board03:45
xiangfuwpwrak, one of the 3 board have little improvement. the capacitors at the back of qfp144 pin-out board.03:45
wpwrakwith *qfp, you just solder a little wire to that "new" pin03:45
wpwrak(caps) ah, that's good. that should help a lot.03:45
wpwrakthe oscillator may need a cap, too03:47
xiangfuwpwrak, yes.03:47
xiangfuwpwrak, about this print [http://downloads.openmobilefree.net/tmp/ftg256_ugly_pcb.JPG]. do you think it is my printer problem?03:48
xiangfuwpwrak, I am using Brother HL-2140 laser printer.03:48
kristianpauluse foil paper at night :-)03:49
wpwrakthat looks rather odd indeed03:50
wpwrakdoes the postscript look fine ?03:51
xiangfu(with *qfp, you just solder a little wire to that "new" pin) I think I have experience on pin-out board. so I direct jump to pcb. and I this one not working. I have to learn more knowledge on pcb (via etc) 03:52
xiangfuwpwrak, too bad. I am still not read the book you advise me.03:53
xiangfuwpwrak, the postscript looks good.03:53
wpwraklooks like a printer driver problem then. maybe google around a bit. cups may also have more than one driver for your printer and selecting a different one may help03:54
wpwrakthe pin-out board was a detour. if you make your own board, you can improve a lot of things even if still using *qfp.03:55
wpwrake.g., you can have a nice big ground plane. and much shorter tracks for everything.03:55
xiangfuwpwrak, wait. I check the postscript agian. 03:56
xiangfuwpwrak, it looks same as the result.03:56
wpwrakhah ! :)03:56
wpwrakin pcbnew, did you "plot" or "print" ?03:56
xiangfu"print"03:57
wpwraki rest my case ;-)03:57
wpwrak"print" is broken. only use "plot"03:57
xiangfu(the pin-out board was a detour....  you can improve a lot...) yes. totally agree.03:57
xiangfuwpwrak, trying plot now..03:58
wpwrakyou should find the "make front" and "make back" targets in my various projects useful. they invoke pcbnew with all the right settings for toner transfer03:58
wpwrakin particular, they select the right mirroring03:59
xiangfuplot looks much much better.04:00
wpwrakkewl :)04:00
xiangfuhuge difference :)04:02
Action: kristianpaul take note about plot04:05
kyakxiangfu: qt4 build is still failing on buildhost?05:38
kyakxiangfu: also, why mpg123 gets built? It depends on BUILD_PATENTED...05:49
kyakthat's pretty damn strange, i see qt4 has built.. perhaps it's one of qt4's packages that fails to build05:50
kyakah, ok we are hitting this bug: https://dev.openwrt.org/ticket/11266 with BUILD_PATENTED05:52
xiangfukyak, part of qt4 failed.06:06
xiangfukyak, so we can just remove mpg123 instead of re-build. right?06:07
kyakxiangfu: just forget about mpg123, it won't get built06:13
kyakxiangfu: do you know which part of qt4 failed?06:14
kyakthere are two dependent packages, which have failed due to qt4 - these are qball and qstardict. Makes we wonder, because qball has exactly the same deps as Tile, which built just fine06:15
xiangfukyak, qt-xml. and I know qt-demos also failed.07:42
rohh10:16
rohre10:22
kyakxiangfu: strange, qt4-xml builds just fine here10:37
viricdamn serial port10:38
xiangfuyes. here[http://fidelio.qi-hardware.com/~xiangfu/build-nanonote/openwrt-xburst.full_system-20121016-0420/packages/] only have 3 qt4 packages.10:38
viricHas anyone ever connected two linuces by serial port, and then expect one side "echo hello > /dev/ttyS0" to emit hello at the other side, running "cat /dev/ttyS0"?10:38
viricwhat magic is there?10:38
xiangfuviric, I just do then on ben nanonoto and laptop. it works fine.10:41
viricmaybe I have some broken hw in the middle10:42
viricI just can't tell what10:42
viricxiangfu: does 'cat' and 'echo' work for you?10:42
viricxiangfu: can you paste 'stty -F serialdevice -a' ?10:43
xiangfuviric, laptop[/dev/ttyUSB0](running cat /dev/ttyUSB0) --> nanonote[/dev/ttyS0] (running echo hello > /dev/ttyS0)10:44
xiangfuwait..10:44
xiangfuhttp://dpaste.com/814225/ this is the laptop output.10:45
xiangfuhttp://dpaste.com/814229/ this is the nanonote output.10:45
xiangfu'cat' and 'echo' works fine here.10:48
virichm ok10:49
viricI might have something wrong10:49
rohviric: try using screen on both sides10:52
rohscreen /dev/ttyS0 960010:53
rohand make sure its a nullmodem cable ;)10:53
rohpin 5 straigh through (gnd) and 2 and 3 crossed10:53
roh+t10:53
viric2 and 3 are crossed10:54
rohsome weird serials loose state as soon as you close them (baudrate, handshake etc)10:54
rohso simply cat may npt work10:54
viricit's a full featured null modem cable.10:55
viricwith hw flow control even10:55
viricI think I have a bad mainboard-to-db9 cable11:01
viricor a broken mainboard port11:02
viricthe mainboard manual even does not include the pinout of the COM1 connector11:04
viriccrossing tx/rx doesn't give echo, in one end.11:15
viricon the other, yes.11:15
pcercueihi fellow neighbours21:24
pcercueiI want to introduce some change on GMenu2X, that might break things, I want your green light first21:24
pcercueieach app on gmenu2x can be given parameters21:25
pcercueiI want to drop that feature21:25
pcercueithat's unpractical, and redundant; if an app requires specific (and constant) parameters, then it's much more practical to wrap the binary using a script21:25
mthis "exec=application --arg" equivalent to the parameters feature?21:30
pcercueiI think in the current code, the value given to "exec" must be a file without arguments, as there is a check on the existence of the binary: if the binary is not present the link won't display21:32
pcercueibut that's fixable very easily21:32
mthsince in .desktop, "Exec" can include arguments, including things like %f for a file name21:33
mthgmenu2x doesn't use .desktop at the moment, but maybe could in the future (also outside of OPK support)21:34
whitequarkpcercuei: why drop that feature? it isn't hard to implement/maintain by any means21:34
pcercueiwhitequark: just like every other "feature" we dropped, because it's crappily implemented, we don't use it at all, and it's redundant21:35
pcercueimeh21:46
pcercueiGMenu2X will chmod() the binaries21:46
pcercueihow dirty is that?21:46
whitequarkow.21:47
kyakpcercuei: not a very good idea to drop params, since we've found a use to that feature: http://projects.qi-hardware.com/index.php/p/openwrt-packages/source/tree/master/nanonote-files/gmenu2x-icons/gmenu2x/sections/terminals/ash-default21:58
kyakyeah, we could just create a shell script wrapper for that :)21:58
pcercueikyak: all those params could be appended to the exec= line21:59
kyakthen we would have to re-write the code that check if a link is existing, just as you mentioned22:00
pcercueithat's not really a problem22:01
kyakwe've got 26 links using "params"22:02
kyakwell, that's fixable..22:03
kyakbut maybe better leave it as it is for good22:03
pcercueimmhhh22:03
pcercueiin that case, can I just remove the option to edit the parameters on the config menu?22:04
pcercueiI assume the user never need to modify the params22:05
kyaki don't even remember a shortcut for that option :)22:05
pcercueion the "edit" menu, you can modify the parameters22:06
pcercueiwhich shouldn't happen IMHO22:06
kyaki think you can remove it, it's easier to edit the link file itself (in my opinion)22:06
pcercueialso, are you aware that the "params" of gmenu2x are heavily parsed for keynames?22:07
kyaksorry, gtg..22:07
pcercueilike "[selFile]"22:07
pcercueiif you use "[selFile]" on the params, it will be replaced by the file selected22:08
pcercueibtw, could we add an "editable" option?22:24
pcercueiif editable=false is set on a link, then the "Edit" option on the contextual menu is hidden22:24
--- Wed Oct 17 201200:00

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