#qi-hardware IRC log for Saturday, 2011-01-15

qi-bot[commit] Joachim Steiger:  v1.3 http://qi-hw.com/p/m1/43ec4d201:26
wpwrakDocScrutinizer: do you still see anything I should try with the band-pass ? http://downloads.qi-hardware.com/people/werner/wpan/rework-20110113/sim.ps04:44
wpwrakDocScrutinizer: or is it crystal time now ?04:44
wpwrak(the very high parameter stability i have reached now does suggest to me that whatever the problem is, it won't respond to simple tweaks. but of course, maybe there *is* something magic that would work :)04:45
wolfspraulwhat software do you use for the simulation?04:55
wpwrakwolfspraul: me ? qucs04:57
wpwrakit's a bit like spice, but for people who've grown up with guis and not punch cards and fortran mainframes :)04:58
wpwrakin fact, i'd love kicad's eeschema to work like the qucs schematics editor - or to have a means to draw kicad schematics with it. it's extremely nice to use04:59
wolfspraulok thx. learning about a new tool :-)05:01
wpwrakit's great. been using it for years.05:09
qi-bot[commit] Wolfgang Spraul: removed absolute path http://qi-hw.com/p/m1/8afc8d505:32
wpwrakhmm, i just realize that one could make pcbs thinner by "embedding" tall components. now that would be something to win friends at the fabs :)05:32
wolfspraulroh: there is an absolute path to a png file in milkymist-assemble-howto.svg. I removed it before but your new commits brought it back, so I removed it again.05:43
wolfspraulsee http://projects.qi-hardware.com/index.php/p/m1/source/commit/8afc8d5/05:43
wolfspraulmaybe you can remove it in your original so that it won't come back in the next commit...05:43
wolfspraulthis is the link: xlink:href="file:///home/labor/Desktop/Kunden/Milkymist-Howto/Mm1_rc1_parts_on_pcb_farb.png"05:44
wolfspraulI just made it a local path href="Mm1_rc1_parts_on_pcb_farb.pnt", I think that should work05:46
wolfspraulthanks a lot for your updates btw, I'm still amazed by this assembly guide!05:46
wolfspraulI will definitely include a nice printout of the guide with every m1, even when they are assembled.05:46
wolfspraulit gives a lot of confidence on how to unassemble/repair/open the unit05:46
wpwrakvery good idea. maybe do the same for the ben ?05:51
wpwrakbtw, regarding the case scans. if you still remember them, it's been so long :) i put them on hold because i needed the mill from time to time for milling and thus didn't have entire half-weeks to assign it to scanning. i did try to finish the main PCB around christmas, but the pc controlling the mill died mid-way through the scan. it's now been replaced but the new machine can't run the proprietary scan software yet (still working on05:55
wpwrakthat). so i hope to get to doing the last scans once ben-wpan is finished. i didn't forget them :)05:55
wolfspraulwpwrak: oh totally fine. we have a number of 'long wave' stuff moving :-)05:57
wolfspraulI think I read about your misfortunes with that machine somewhere a while ago...05:58
wpwrakit actually kept going for about a day with the disk/controller dead. but then the virtualized windows finally crashed :-(05:59
wolfspraulI am thinking how we can put more momentum into your ben-wpan work.06:01
wolfspraulI want to avoid that they are just some obscure boards that don't really catch on.06:01
wolfspraulin my mind it's only a success if, let's say 12 months later, we have significant free software development happening to support them06:01
wolfspraulbut how to get there? how to get activity up in the Linux kernel?06:02
wolfspraulpretty much the only thing I can do is to try to lower the price, which only works if I simplify the whole thing.06:02
wolfspraulfor example I could only sell them in sets of 4 (2*usb, 2*8:10), but then make that dirt cheap.06:02
wolfspraulat least that would give everybody some incentive to make use of all those radiators :-)06:03
wolfsprauljust thinking...06:03
wolfspraulthis could be something for the diy/arduino/sparkfun crowd too06:04
wpwraki think an open "reference design" could already help the other diys06:07
wolfspraulthat's implicit, no? what would they do with the reference design?06:08
wolfspraulmy main concern is to kickstart practical (real world) applications06:10
wpwrakmost people are afraid of the wrong things with rf. e.g., if you just implement a reference design, you've basically absorbed all the benefit of a module already06:10
wolfspraulnot just theory, talk06:10
wolfspraulI think we are so deep in things, what most people are afraid of is 'hardware' in general.06:10
wpwrakwhat you still have to solve are antennas and the system integration. there, it helps to have reference data.06:10
wolfspraulI think we have to take your work directly into an application. Of course others can use it as a 'reference', it's all open.06:10
wolfspraulbut I doubt anybody will do that :-)06:10
wpwraki'm not much worried about the software side. that should be relatively easy. been there, have done it before ;-)06:10
wolfspraulwe have to show the application ourselves06:10
wpwrakwell, you can make a sensor board or such06:10
wpwrak(sw/app) yeah, i meant infrastructure. apps are a different story, agreed06:10
wolfspraulI realize more and more how much empty talk there is in hardware.06:10
wpwraki hope to use wpan for controlling electricity at my home, but that's of course something that's extremely hard to commercialize06:11
wolfspraulbecause so few know much about it, everybody projects their assumptions, but it sounds like gained from a lot of experience where it is actually not.06:11
wpwrakhehe :)06:11
wolfspraulso the only way out of that chicken-egg problem is to pull all the way through yourself.06:12
wolfspraulmake it work, add the software, get the certification, put a box around it, sell it06:12
wolfsprauland then 99% of talk will suddenly disappear06:12
wolfspraulthe day at reichl-emv lab gave me a lot new confidence in this area06:13
wolfspraul:-)06:13
wolfspraulI already told him I may show up with some strange 802.15.4 boards :-)06:13
wolfspraulwhy is it hard to commercialize controlling electricity?06:14
wpwrakdid he have any opinion on 802.15.4 ?06:14
wpwraklots of local regulations and certifications06:14
wolfspraulno [802.15.4]06:14
wolfspraulI'm not worried about that [certs]06:14
wolfspraulcompared to all the work we put into hardware, software, going through some tests/certification procedures is just some more work06:14
wolfspraulit's not overwhelming06:14
wpwrakand since its the kind of thing that is infrastructure, it may even be worse - only installable by certified electricians and that sort of stuff06:15
wolfspraulI'm in the real world.06:15
wolfspraulso one by one. first make it work, then bring it up in usability, including box, manual. and somewhere there are certs too :-)06:15
wolfspraulsome countries will have nasty regulations, ok fine.06:16
wolfspraulthere are a number of options to deal with them, country by country :-)06:16
wpwrakthe problem i would expect with this sort of certs is that you may need one for each country, not just CE for a while continent06:16
wolfspraulas always you pick the low hanging fruits first.06:16
wolfspraulmaybe, but I'm not worried.06:16
wpwrakperhaps the best would be to partner with companies that already make similar things, provide them with the core circuit and let them worry about the local issues06:17
wolfsprauldoesn't sound right to me.06:19
wolfspraulpull it all the way through yourself, set an example.06:19
wolfspraulonce you set an example, _maybe_ you can convince others to join.06:19
wpwrakwell, you can do that. there are examples for this approach, too06:20
wpwrakanother approach to ease things would be to approach a distributor of other systems with similar functionality. they ought to know about pitfalls.06:23
wpwrakthere's one in germany for plcbus ... lemme see if i can find them ...06:23
wpwrakhmm. seem to have vanished. here's one in the netherlands: http://www.elekhomica.nl/index.php?cPath=13706:29
Jay7wpwrak: all EU should be covered by CE06:32
Jay7and all USA by FCC afaik06:32
Jay7this zones will give you about 80% of income06:33
Jay7because of high purchase ability of that zones06:34
Jay7other interesting countries are Brazil, Argentina, China and Russia..06:35
Jay7well.. I should release kexecboot 0.606:37
wolfspraulJay7: from my NanoNote experience, I'd rate them slightly differently06:37
Jay7wolfspraul: It would be interesting to hear your rating :)06:38
wolfspraulEurope 40% US 30%, those two are easy06:38
wolfspraulafter that, Australia - very active scene. quite amazing considering that there's only 15 million people there?06:38
wolfspraulJapan, same06:38
wolfspraulvery active, actually if I would have a local Japanese distributor, I might even be able to match the US06:38
wolfspraulafter that, hmm.06:39
wolfspraulRussia?06:39
Jay7Japanese distributor should be easy to become06:39
wolfspraulnot easy06:39
Jay7hm.. then I'm wrong :)06:39
Jay7wolfspraul: China?06:39
wolfspraulthe problem is cultural06:39
Jay7and btw, India06:40
wolfspraulno way, China, India, Brazil - without hurting anybody I'd just say 'forget it'06:40
Jay7they growing06:40
Jay7hm..06:40
Jay7I'm totally wrong then06:40
wolfspraulyou can throw them in with all other destinations, of course there are always individuals here and there06:40
wolfspraullike Werner in Buenos Aires :-)06:40
wolfspraulI know _great_ individuals in all those places, but I cannot see something more consistent/structured06:41
Jay7yeah..06:41
Jay7we all need local groups like LUG's06:41
Jay7eLUG :)06:41
Jay7I have idea to create embedded linux [hardware] portal06:42
wolfspraulso in that list, there would be Belarus, China, South Korea, Brazil, Argentina (Werner :-)), Singapore, and so on. Lots of them.06:42
wpwraki'm a bit surprised brazil doesn't have more activity. they have a rather active free software scene06:42
wolfspraulI have never been there, so don't know.06:42
wolfspraulwhat I hear from the distance sounds very similar to my own extensive experience in China.06:42
Jay7Ukraine may be interested too06:43
wolfspraulyes06:43
wolfspraulhey we forgot Colombia :-)06:43
wolfspraulmaybe after the easy big ones like US/Europe/Japan/Russia/Australia, it's hard to pin down nowadays06:44
wpwrakzrafa: do you know where the jornadas regionales 2011 will be ?06:44
wolfspraulI am very glad that I have such a global approach, I will definitely continue with that.06:45
wolfspraulhave shipped nanos to Martinique, Philippines, Nigeria (yeah), and so on06:45
wolfspraulthat guy in Nigeria is really cool actually06:45
Jay7Nigeria have strong free software community06:46
wolfspraulI'm not surprised to hear that, but except for that one contact don't have any experience.06:46
Jay7according to KDE people blogs :)06:46
wpwrakfisl12 (the BIG brazilian free sw event) will be in june 2011 in porto alegre06:47
wpwrakthere are direct flights from buenos aires. now even from aeroparque :)06:50
wpwrakwolfspraul: so you can plan your 2011 south america trip accordingly :)06:52
wolfspraulmy samples are a bit too small for some places (just one or two contacts), but I feel good working with the data we've learnt in the first NanoNote year, and keep my mind and eyes open for new discoveries.06:52
wolfspraulthe main reason that Japan was so successful was one single journalist who did an extensive review in a large IT magazine06:52
wolfspraulfrom that you get all sorts of contacts, LUGs, etc.06:52
wpwrakso, milkymist to slashdot, the reg, and heise ;)06:52
Jay7hm.. Eiphel was founded by russian06:56
Jay7I didn't know that..06:56
wolfsprauloh sure06:58
wolfspraulhe's a great guy, I visited for 3 days last May06:58
wolfspraulthe richard stallman of hardware06:59
wolfspraulwpwrak: what do you think about writing some of the eeschema/pcbnew output files to stdout instead of a named file?07:13
wolfspraul(or accepting input files via stdin, although that's probably even less valuable...)07:18
qi-bot[commit] Wolfgang Spraul: convert dxf into svg instead of png http://qi-hw.com/p/m1/98f4ce107:45
kristianpaulwpwrak: control + j will save my life !07:56
kristianpaulso usefull :-)07:57
wpwrakwolfspraul: stdout could be nice, yes. or being able to pick a name in general. not very high priority, though. you can work around the fixed names easily enough.08:30
zrafawpwrak: (jornadas regionales) no yet, but I will check and then I will tell you (last year it was in San Luis)08:30
wpwrakkristianpaul: ;-))08:30
wpwrakzrafa: let's hope they didn't pick ushuaia ;-)08:32
zrafahaha08:33
zrafawpwrak: fisl in Brazil is huge right?08:34
zrafaI mean, a big conference08:34
wpwrakzedstar: oh yes. very large.08:34
wpwrakoops, zrafa08:34
kristianpauljornadas regionales de sofware libre?08:45
wpwrakyup08:50
kristianpaulargg i hate fedex deliver service here :/09:10
kristianpaulliers !09:10
wpwrakno saturday delivery then09:12
kristianpaul:-|09:12
wpwrakwell, i'm not surprised. it's not part of the normal service09:12
wpwrakif you're really in a hurry, what works best is to ask them to leave it at their distribution center and go to pick it up there09:12
kristianpauli'm not in  hurry i just like push for good service09:13
kristianpaulbut they advice me yday to waiy for the package for today, whey they dont say the truth then..09:13
wpwrakhow did you find out it's not coming ?09:13
Action: kristianpaul at phone call09:14
wpwrakah :)09:14
wpwrakyou can ask them if you can go and pick it up wherever it is now. then you'll at least sleep well tonight :)09:14
kristianpauli'm on that09:15
wpwrak(well, if you have a car. might be in a remote location in cali or such.)09:15
Action: kristianpaul calling thirdparty deliver enterprise09:15
kristianpauldont have car09:15
kristianpauli  was in cali yday i offer go ther and pick but they say, no sir dont worry tomorrow will be deliver09:16
kristianpaulargg09:16
kristianpaulany way..09:16
wpwrakyou can always make up a story. like you having to travel today and there's something in there you need. tell them that they already failed you yesterday, so now it's really an emergency.09:17
wpwrakwhatever works :)09:17
wpwrakif course, if they're closed today, then that won't help either09:17
wpwrakalso make sure you ask them what documentation you have to bring to identify yourself09:18
kristianpaulyeah np09:18
wpwrakaround here, they're surprisingly lax, but that may not be the same everywhere09:18
lekernelkristianpaul: getting your m1?09:22
kristianpaulhere are lax for everything.. :|09:22
kristianpaullekernel: fighting to get it sooner ;-)09:23
wpwrakevery minute counts :)09:23
kristianpaulok it is somwhere in buga...09:23
wpwrakwhee !09:24
DocScrutinizerwpwrak: so what are the results of RC and XTAL?09:24
wpwrakDocScrutinizer: no discernable change: it's at the bottom of http://downloads.qi-hardware.com/people/werner/wpan/rework-20110113/09:24
wpwrakDocScrutinizer: click on the simulation diagram for a better readable postscript version09:24
DocScrutinizerI got it09:25
DocScrutinizerXTAL?09:25
wpwrakDocScrutinizer: working on it, yes09:26
DocScrutinizerI'm not sure you'll see any improvement with that either, as to me it's more than unclear if CW really means unmodulated09:26
wpwrakDocScrutinizer: anything you'd suggest to try before the crystal ?09:26
DocScrutinizernope09:27
wpwrak(cw) well, consider the USB version. there, the "noise hill" is much lower09:27
DocScrutinizermay send data and see if spectrum noticeably changes09:27
wpwrakthat's the same chip, with the same settings09:27
wpwraksending data changes things a LOT ! here's data: http://downloads.qi-hardware.com/people/werner/wpan/tst-tx/09:28
DocScrutinizerooh, ok09:28
kristianpaullol the package from deal extreme i order month ago got today too :p09:28
wpwrakkristianpaul: what's in that one ?09:29
kristianpaulwpwrak: http://www.dealextreme.com/details.dx/sku.1466409:29
wpwrakkristianpaul: duh, serial cable :) real men build that while on the toilet09:31
kristianpauli built one09:31
kristianpaulwhen realized that may be the package was lost somwhere after taiwan ..09:32
wpwrakheh :)09:32
wpwrakstuff from taiwan can take a while. e.g., EMS tends to be on the way for more than a week09:33
DocScrutinizerheh, I got 2 nice critters here, gift of somebody who wrote a sw to decode: a white block ~20*20*10mm, with a 3.5mm plug for headset receptacle, and a slot to slide your credit card09:37
DocScrutinizerrather nifty09:37
DocScrutinizerclearly iPhart design, but hey09:38
wpwrakDocScrutinizer: the CC is for the built-in iTunes ? :)09:39
DocScrutinizerit seems CC transactions are cheaper when done "online"09:40
DocScrutinizerand you don't have to deal with customer signing papers, or you calling some AmEx clearance hotline to validate the card, or whatever might be involved09:41
DocScrutinizernot completely wrapped my head around the original business model that gave us these sugar cubes for free09:42
DocScrutinizera tape magnet head in a small white plastic block, connected to AV plug's mic ring09:43
wpwrakif you have to call to validate, you still need to do that for online transactions. of course, in practical terms, this means that you'll never get that transaction processed09:43
DocScrutinizerhonestly, I'm not too interested in the original usecase. It's just a nifty card reader for CC track 3(?)09:45
DocScrutinizerwondering if preparing my own cards with music on magnet tape would make some weirdo scratch machine09:46
DocScrutinizerthe digital info on my CC doesn't sound too interesting ;-)09:47
DocScrutinizeror building a real magnetic tape recorder ;-D09:49
wpwraka very short one :)09:49
DocScrutinizeroh no, it's only monophonic, that's not good enough for a tape recorder09:50
DocScrutinizernah, I thought about using real 1/2" tape on reels to run thru the sugarcube :-)09:50
DocScrutinizerjust for the fun of it - of course no real life usecase whatsoever, except baffling people09:53
DocScrutinizeryou could stick a 21cm tape snippet at the edge of every page of a book. Maybe nice for kids, or ornithologists09:55
wpwrakmore impressive if it can make sounds from a credit card09:55
wpwrakyou can take the CC as a source of random numbers, then feed this into a generator of good-sounding tones (following musical theory). that way, each CC would have its own characteristic melody09:57
DocScrutinizerevent cards for games (monopoly, whatever) come to mind as well09:57
wpwrakkristianpaul: so, how is the adventure going ?09:57
wpwrakDocScrutinizer: playing cards. with hidden information in a magnetic strip ;-)09:57
DocScrutinizerI wonder what a card deck for blind people might be like10:00
wpwrakbraille ?10:00
DocScrutinizeryou'd not want that to get read and maybe even seen by peers and the the card dealer10:01
DocScrutinizerso yeah, punch hole braille with a uniform backside second card glued to it might work10:02
DocScrutinizerhttp://www.marland.de/produkte/c/kartenspiele/p/kartenspiel-blinden-skat-franzoesisches-blatt/  hmm, they don't mention if it's punch-hole or usual braille - with "bubbles"10:10
wpwrakbetter single-sided bubbles. that way, they could also play with people who can see10:18
DocScrutinizerbubbles are bad when mixing the cards. You don't want the deck to pile up 3 times the usual hight10:18
DocScrutinizerso little dents, halfway depth into the card are better than protruding bubbles10:19
DocScrutinizerstill the card dealer easily could read ec10:20
DocScrutinizereach card he hands out10:20
DocScrutinizerwell, poker is a hard game to play, for vision impaired people, it seems10:25
qi-bot[commit] David Kühling: added swig: Simplified Wrapper and Interface Generator (for host toolchain) http://qi-hw.com/p/openwrt-packages/a6e6dff10:28
qi-bot[commit] David Kühling: plplot: make lua bindings work http://qi-hw.com/p/openwrt-packages/0f9e96c10:37
kristianpaulwpwrak: package is next to me now :-D11:26
Action: kristianpaul looks for a CRT monitor and a usb mouse11:26
kristianpaulthe package was deliver by an old man i a bycicle!!11:31
wpwrakkristianpaul: ;-))11:36
wpwrakkristianpaul: at least not by donkey rider ;-)11:36
wpwraksigh. board gets almost 6 mm longer :-(11:38
wpwrakand 0.4 mm wider11:44
kristianpaulargg crt and no image11:56
kristianpaullets try lcd display11:56
wolfspraulkristianpaul: you mean on the m1?11:56
wolfsprauldo you see the 3 leds next to the buttons?11:56
wolfspraulwhen you power it, one of them should go green11:56
wolfspraulthen you need to press one of the buttons, I think the middle one? forgot...11:57
wolfspraulthen it will actually turn on/boot11:57
kristianpaulah wait11:58
wolfspraulwhen it boots, a second led will go green11:59
wolfspraulthen after about 2-3 seconds, you should see something via vga11:59
wolfspraulit boots pretty fast11:59
kristianpaulyay !12:00
kristianpaullol flckrnoise is ugly indeed :p12:00
wpwrakkristianpaul: now you should say how much prettier X-based things are (-:C12:02
kristianpaulhmm i need a cable for the audio in..12:06
wolfspraulyes, I think in the future I will include a audio line cable12:06
wolfspraulsorry right now I don't have them yet :-)12:06
wolfspraulthe sufferings of the early adopters :-)12:07
kristianpaulnp12:07
wolfspraulit's one of those things that everybody needs with an m1, but many people will not have, or not find12:07
wolfspraulthen they have to buy one for 10-20 usd12:07
wolfspraulwhere I can source them most likely for < 1 usd12:07
kristianpaul:/12:07
kristianpauldont scare me !12:08
wolfsprauljust yesterday I was going through my piles of stuff here trying to find one - no avail12:08
wolfspraulkristianpaul: maybe you find one somewhere, just go through your household :-)12:08
wolfspraulI will probably include one at some point...12:09
lekernelwpwrak: this is how the only somehow working X based gui looks like: http://www.microwindows.org/SSWin32.html12:18
lekernelmay I reassure you, the flickernoise toolkit, even if not perfect, does a much better job than that12:18
lekernelthe fact that there is no alternative is just sad12:18
lekernelanyway the mtk appearance can certainly be improved a lot with some work12:19
lekerneland less work than porting X-windows, for a better result in the end12:19
lekernelbut since you like X-Windows so much, wpwrak, why not try to port it to get a taste of its crappiness?12:22
lekernelsame with linux...12:22
lekernelthough the linux kernel isn't crappy, but gcc is :)12:23
larscand the theobroma lm32 port12:25
lekernelyeah, that too, maybe even worse than gcc :)12:25
wpwraklekernel: i once had a quick look at kdrive and it didn't look too nasty. not sure if it has everything you'd want on mm1, though.12:26
lekernelget graphics accelerations and GTK/Qt to work on that then...12:26
larsckdrive is an xserver12:27
lekernelFYI the LM32 GCC toolchain is so broken that it mistakes some Qt C++ methods for the class's destructors, with the results you can imagine12:27
lekernelbesides, Qt is mega-bloated, a hello world app is 22M12:27
wpwraklekernel: rounding errors when calculating the index in the method table ? :)12:28
lekernelnot sure how you'd imagine fitting those 22MB + application payload + fpga bitstream + backup versions of those in a 32MB flash, that should also contain a user filesystem12:29
lekernelmtk/rtems, for all their faults, allow this very easily12:29
wpwraklekernel: i once had a bootable linux with the X11 server on a 1.44 MB floppy :) but o12:31
wpwraks/o/i'd rather not touch qt's c++/12:31
lekernelyeah, with twm and xterm?12:34
wolfspraullekernel: where are the rtems sources?12:37
wolfspraulI tried getting an overview over the sources but couldn't find them12:37
wolfspraulyou have 13 repositories here :-) https://github.com/lekernel/12:38
lekernelhttps://github.com/fallen/rtems-milkymist ... should be ready to be merged upstream soon12:38
wolfspraulahh, 'fallen'12:38
wolfspraulok thx12:38
kristianpaulok finally got/buy the cable for 1.2usd12:47
wpwraklekernel: not sure if there was space for that much ;-)12:47
lekernelwpwrak: I have a product for visual artists to get out of the door. not some thin client toy for nerds who play with cisco routers in their spare time. it's too expensive for them anyway12:48
wpwraklekernel: well, that was for compatibility testing12:49
kristianpauli guess is okay if i plug some usb powered speakers to the mm112:49
lekernelkristianpaul: yes it is12:49
kristianpaulah wait , dont have more usb ports avaliable >:-|12:49
lekernelthere's a ~1A combined current limit for both ports12:49
lekernelyup. maybe a nice accessory could be a passive USB port "duplicator"12:50
wolfspraulactually many people are 'overloading' that max usb current12:50
wolfspraulsometimes with non-standard chargers that can go to 1A12:51
lekernelyeah, that's why I put the limit on both ports, so they can draw 1A from one port if they don't load the second12:51
wolfspraulor sometimes because they just give more amps than 500mA, for example on a Mac Pro you can easily draw 750 mA12:51
wolfspraulthere you go, you do it too :-)12:51
lekernelthere's some margin too before the polyswitch trips, I don't remember how much...  at about 1.4A or something12:52
kristianpaulthats cool !12:52
lekernelso the USB current limit isn't really enforced, it's more a protection against short circuits12:52
kristianpauland toght i never used winamap efects12:52
wolfspraulthis is sometimes a problem one can overlook when designing a device. say you always test with your mac, but your device draws 700mA. then later with some other host it won't work...12:52
wolfspraulfrom my infinite collection of stupid hardware design mistakes...12:53
lekernelif you stick to the norm USB current limits are crazy, there is for instance the "suspend mode" you are supposed to go into when you don't get regular polls from the host12:53
lekernelthat one gave me some more headaches with the usb host design, because if you mess up the polls, your device goes into suspend mode and ignore whatever you send12:54
lekerneland it gives some headaches to device designers too, because there's a very low current limit for suspend mode12:54
lekernelhundreds of microamps iirc12:55
lekernelcounting the current draw from the detection pull-up resistor you must place on one of the the data lines12:55
lekerneland the USB norm mandates the suspend mode12:56
lekernelit's really a fucked up standard12:56
lekernelin practice this is never respected, many passive USB gadgets draw hundreds of milliamps while supposedly being a device in suspend mode12:59
lekerneland I have never seen one malfunction :)12:59
lekernel(i.e. coffee heaters, lamps, etc. that have no communication features whatsoever with the USB host)13:00
kristianpaulwow  interwomen for  patch13:06
kristianpaulwoven*13:07
rohwolfspraul: uh. i see.. how did you change the path? manually?13:07
kristianpaulhey, Milkymist One dint have implicit hardware RESET buton, is this the begin of a new era? ;-)13:15
lekernelit has, press the 3 buttons at the same time13:15
kristianpaulah13:16
lekerneland release the button close to the LEDs first, otherwise this will enable manual DRAM timing calibration mode13:16
lekernel(if you enable this mode, which you can notice from a strange pattern on the screen, just reset again)13:17
kristianpaulyup13:20
wpwrakwolfspraul: hmm, there's a small mechanical problem with the 8:10 card slot of the ben: when it "locks", it merely discourages the card from leaving on its own, but it doesn't really hold it.13:29
wpwrakwolfspraul: with wpan, this means that you can accidently loosen the card (or remove it entirely, if you apply enough force)13:30
wpwrakwolfspraul: you can also do this with a regular uSD card - it's just harder to grab it13:30
DocScrutinizerhmm, quite usual concept, it seems13:33
wpwrakprobably. could be considered a feature. i just noticed that it comes out quite easily, despite already being a bit thicker and wider than the original.13:38
DocScrutinizerwith my EeePC 701 I got in TPE, I complained about it's so hard to get out the SD card, and I sticked some tape to the card. Until somebody laughed at me and explained to my I need to push in the card to make it come out. Anyway no latch or whatever is fixing the card there13:49
DocScrutinizerand actually there's no notch or anything in a SD card, for a latch to hold it in slot. Only thing you could do is "close the slot door"13:54
DocScrutinizeron a slot like in EeePC where SD card protrudes ~0.5..1mm, that's not an option as well13:55
wpwraklook at the card carefully - there is a notch. it just doesn't go through the whole card13:57
wpwrakerr no, it actually does13:58
wpwrakon the side where one contact is retracted13:58
wolfspraulyes I know this already14:20
wolfspraulnot much we can do, I think14:20
wpwrakprobably not. just something to keep in mind.14:22
qi-bot[commit] Werner Almesberger: atusd: use crystal instead of trying to obtain an external clock from the Ben http://qi-hw.com/p/ben-wpan/c0941db15:37
qi-bot[commit] Werner Almesberger: atusd: added missing connection for SLP_TR http://qi-hw.com/p/ben-wpan/507bcaf15:37
wpwrakgrmbl. only 1.5 toner transfer sheets left. and it seems that HP paper no longer gets imported here :-(15:39
DocScrutinizerwpwrak: you're right, I missed that notchie16:15
qi-bot[commit] kyak: Tile: count moves http://qi-hw.com/p/openwrt-packages/cccd60c16:16
qi-bot[commit] kyak: qball: updated to 1.2, dropped merged patch http://qi-hw.com/p/openwrt-packages/c1fb2a316:46
qi-bot[commit] kyak: qball: really drop the patch :) http://qi-hw.com/p/openwrt-packages/e78146d16:48
kristianpaulhe tricky adjust buttons :-)17:39
erionhey guys i got a problem my pc keeps restarting constantly17:47
lekernelerion: wrong channel, sorry17:47
erionwat channel do i go to?17:47
Jay7ask on #help e.g.17:49
wpwrak_and this is what the board with crystal will look like: http://downloads.qi-hardware.com/people/werner/wpan/tmp/usdxtal.jpg18:28
LunohoDwpwrak_: wifi?18:31
wolfspraulLunohoD: much better. 802.15.4 with 6LoWPAN (later)18:38
zrafawolfspraul: dont forget the 64Mb of ram special edition for Ron18:41
zrafaand 1024x76818:41
kristianpaulwpwrak_: (board) is a monster !18:43
kristianpaulin the good sense :-)18:44
wolfspraulzrafa: hah yes!18:46
wolfspraul"isn't it time to discuss 64 mb"18:46
zrafaNO18:47
wolfspraulwhat inspiration!18:47
zrafamy nn has 32MB!18:47
zrafaif next bunch has 64mb our current nns will have still 32mB RON!18:47
zrafa:P18:47
wolfspraulvery good answer18:47
wpwrak_kristianpaul: (monster) yeah, it's quite big :-( maybe i can shrink it a little by rotating the crystal, but i'm not sure. the problem is that putting the crystal "upright" would mean that the traces would have a larger detour and this also means that they would need more room for the bend18:55
wpwrak_wolfspraul: sigh. how will you ever make money with that attitude ? :) make the old product obsolete, so people have to buy new ones. consume ! consume ! consume ! ;-)18:57
lekernelstrangely (and perhaps fortunately), Ron doesn't comment much on milkymist :)18:58
wpwrak_lekernel: enjoy the quiet while it lasts :)18:58
lekernelI cannot really imagine what he'd say and ask, but i'm a bit afraid18:59
wpwrak_lekernel: he's good at finding the weak spots. so just think of what doesn't work well :)19:00
wolfspraulyes, exactly19:06
wolfspraulbut for now he doesn't know what to compare it to19:06
wolfspraulwhen he has that it will start19:07
lekernelhow did you meet this guy btw...?19:09
wolfspraulhmm. forgot. he was always around :-)19:09
rohi think he somehow pop-ed up while moko19:13
wpwrak_he's our bad conscience. our collective moment of doubt. a bit like the monster in "forbidden planet" :)19:14
rohwpwrak_: heh... i sank another day into micromanagement yesterday19:15
rohcalculating details... .. sourcing sucks19:15
lekernelwell, i would rather describe him as a talkative slacker :)19:15
wpwrak_lekernel: i think the technical term is "heckler" :)19:16
wpwrak_roh: don't worry, soon it's tax time again. that will take your mind off the boring sourcing ;-)19:17
roh*grmbl*19:17
wpwrak_((-:C19:17
rohwpwrak_: been there just lately... not nice.19:17
rohneed to sell muuuuch more cases19:17
roh;)19:17
rohwpwrak_: i was thinking about cases for your 'stick' .. and ended up with millin molds and mixing some plastic in different forms for now. any ideas about materials?19:22
rohany removable form would be much bulkier.. i guess its a one-time-molded-in concept19:23
wpwrak_i was thinking of something elastic, rubbery. open on one side. you could then just put it over the board.19:30
wpwrak_with open hardware, it should be possible to open it when desired :)19:33
wpwrak_of course, a permanent casing would be easy to make. all you need is a bit of silicone and solvent. acetone, paint thinner, etc.19:34
wpwrak_well, the "baumarkt" type of silicone. there are others much nicer curing processes. with these, you'd mix two components, dunk the board, and it let it dry for a few hours. done.19:35
rohwould be interresting to certify that19:44
rohby means of toxic, electric, mechanic properties ;)19:44
wpwrak_;-))19:45
wpwrak_should isolate quite well. the toxicity should also be significant :)19:46
wpwrak_mechanical properties ... well ... that's probably the least of our worries :)19:46
rohi wonder how that stuff is called in germany: http://www.smooth-on.com/19:49
rohand if i can get that as easily at all19:49
wpwrak_(considering how fragile the whole thing is in general)19:49
rohyour pcb?19:49
rohwell.... thats a different issue. but solid plastic should fix that19:50
wpwrak_(fragile) the PCB-Ben assembly. that uSD holder is rather flimsy19:51
rohheh19:51
wpwrak_the board itself should survive some abuse. well, now that's getting bigger, a bit less, but still19:51
rohyeah. would be better integrated into the lcm half of the nanonote19:52
rohfor ya? (or how the next one is called)19:52
rohbbiab -> relocate ot19:53
rohto the lab19:53
wpwrak_(ya) yup, definitely19:54
rohre20:10
bartbesany news on ben-wpan?20:35
wpwrak_bartbes: the USB board looks pretty good now. the one for the ben has some noise issues: http://downloads.qi-hardware.com/people/werner/wpan/tst-cw/20:38
wpwrak_bartbes: i tried a lot of things to make the noise go away: http://downloads.qi-hardware.com/people/werner/wpan/rework-20110112/ and http://downloads.qi-hardware.com/people/werner/wpan/rework-20110113/20:38
wpwrak_bartbes: yet it's still there. the current hypothesis is that using the ben's clock isn't such a good idea after all. so i'm now adding a crystal.20:39
wpwrak_bartbes: this is what it'll look like: http://downloads.qi-hardware.com/people/werner/wpan/tmp/usdxtal.jpg20:39
wpwrak_(well, perhaps minus a few optimizations. i'll find those while soldering.)20:40
bartbesand what about a case?20:40
bartbesit looks awfully breakable20:41
wpwrak_if the board with crystal improves the noise, then all i need is a BER test and then it'll be good to go20:41
wpwrak_ah yes, the case issue is still unresolved20:41
bartbesso what could the issue with the clock be? is it too irregular?20:41
bartbesor using the wrong frequency?20:41
wpwrak_the board isn't too fragile despite being large. but it it'll act as a lever on the microSD holder in the ben20:42
wpwrak_the frequency is perfect. long-term clock stability if perfect. but maybe there's some noise on top of the clock.20:42
bartbestoo bad we sacrific uSD20:46
bartbesbut still20:46
bartbeshuge improvement20:46
wpwrak_yeah. it would be nice to have a 2nd slot20:46
bartbes(if I can walk around the house with teh internets)20:46
wpwrak_the technology is designed for a maximum range of about 10 m. so you can't walk very far. but still, "this and the next room" ought to be possible.20:47
bartbesthe room beneath me20:47
bartbesor above me20:47
bartbesI sure hope it's spherical ;)20:48
wpwrak_depends on the kind of floor you have :)20:48
bartbespure lead20:48
wpwrak_grin20:48
bartbesI happen to live in prypjat20:48
bartbesI don't even know what the range of my wlan router is20:49
bartbesbut I doubt it's very good20:49
bartbeshowever we have fast wireless around the house20:49
wpwrak_steel reinforced concrete already does a pretty good job at limiting this sort of radiation. also, terrace floors can be sealed with something that acts like a perfect shield. maybe it's actually lead.20:50
bartbesI have no clue20:50
bartbesbut since as I said our wlan does pretty good, I expect it not to be too bad20:51
wpwrak_my wlan has trouble crossing the kitchen. and it's about 50 times as strong (transmit power) as wpan.20:51
bartbesalso, in theory from my room to downstairs could use the windows ;)20:51
wpwrak_well, there's a tiny toilet in the path too. and the kitchen is quite long.20:52
wpwrak_windows are good, yes. RF always finds a way ;-)20:52
rohwpwrak_: http://www.kaupo.de/Produktkatalog/Glasklares/Crystal-Clear-Serie.html21:24
rohwhat about a transparent case?21:24
wpwrak_there isn't a lot to see inside. not even a led. so some sort of black (like the ben) would be fine too21:26
wpwrak_about the "crystal clear". that seems to be like acrylic. then you need to solve the mechanical issues of a rigid case.21:28
rohacryllic is 'PMMA'21:29
rohthats PUR .. not as brittle anyhow21:29
rohbut yes.. something simple and black would be nice21:29
wpwrak_black and rubbery. so you can pull it over the board and you're done21:30
wpwrak_no need to fit halves or have semi-rigid plastic that you snap or such21:30
rohhttp://www.kaupo.de/Produktkatalog/Polyurethangiessharz/C-1509-C-1515-C-1520.html21:34
rohdunno how rubbery it is. seems to be the 'gets hard' kind21:34
wpwrak_do you know the kind of PUR used for sealing ? similar in terms of use as silicone21:35
wpwrak_that one is more rigid than silicone but still stays flexible21:35
wpwrak_of course, silicone itself isn't too bad either. it does get fairly strong after a while.21:36
wpwrak_the main problems i have with the ones from the DIY shop are that they're too viscous and that i don't have a decent release agent21:37
wpwrak_hmm. maybe wax or a wax-like substance could work. warm it a little and it should liquify and come off easily.21:39
steve|mhttp://www.steve-m.de/pictures/pirelli_dpl10/ (still uploading)... gn821:50
kristianpaulnite21:56
zrafasteve|m: link: what is that?23:37
--- Sun Jan 16 201100:00

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