#qi-hardware IRC log for Sunday, 2012-08-19

wolfspra1lprogramming question: any downside to using designated c99-style initializers to initialize union members?02:35
wolfspra1lI haven't done that before but c99 feels kinda stable and well supported now, no?02:35
wolfspra1lunless I find a reason not to, I'll go ahead with .xxx = ...02:35
whitequarkthere was some really weird incompatibility with some software03:37
whitequarkbut if you stick to gcc/clang it should be fine03:37
whitequarkI think that was armcc... not sure actually03:38
wpwrakwolfspra1l: hmm, when you take something to FCC approval checking, how does that work ? let's say a simple circuit without radio. i.e., how "finished" does the device have to be ? and do you send the device to the test lab when it has reached that state of maturity or do you send them development prototypes to accelerate the process ?07:56
wolfspra1ltesting, hmm10:39
wolfspra1lwpwrak: that's a long story :-)10:40
wolfspra1lthe FCC works with approved labs10:40
wolfspra1lso for authentic fcc testing, you need to work with an approved lab10:40
wolfspra1lthere may well be one in Buenos Aires, I don't know. or maybe not. in China they are everywhere (the real ones, in addition to a flood of fake ones)10:40
wolfspra1lmostly those labs (=companies) are offering all sorts of engineering and consulting services, so what you use them for is really your own decision10:41
wolfspra1lthey love to sell you as much as possible, naturally10:41
wolfspra1lonce you passed the testing, you can still make modifications to your design or production process, but at which point exactly this falls out of the first testing then and a re-testing is necessary is a grey area and comes down to your own understanding of quality, at some point10:42
wolfspra1lsome people are just letting whatever 'brick' pass the testing, and sell something entirely differently10:42
wolfspra1land nobody will ever check or care10:42
wolfspra1lI've seen cases where people starting selling a product that was totally different under the name of an old one because "renaming is expensive"10:43
wolfspra1lother take the testing more serious, and once they make a change (in design, production process or sw) that they think might affect the test result, they will voluntarily resubmit and want to work toward a better product anyway10:44
wolfspra1lthe external engineering is not too different from any in-house engineering anyway, since a normal company will pay their employees, they may as well pay an outside service provider10:44
wolfspra1lI think the category you are looking at is 'unintentional radiator'10:45
wolfspra1lyou should be able to get a certification for about 1000 USD or so, and yes it really only makes sense with case imho10:45
wolfspra1lthat's about it :-)10:45
wpwrak(long story) yeah :) at OM, i got to see the tips of that iceberg, but never had the entire picture.13:09
wpwrak(unintentional radiator) yup :)13:09
wpwrakso for a completely new design, perhaps it would make sense to have an early prototype "pre-certified" to make sure it doesn't have major issues. USD 1k isn't all that bad anyway.13:11
wpwrakhow much time does such a certification (or attempt) usually take, from sending the device until the result is known ? i think there's an extra delay until the actual certificate is issued, isn't there ?13:16
wolfspra1lcalling it a day, but quick... it mostly comes down to getting an 'appointment'/free time for their lab13:36
wolfspra1lwhich can be anywhere from a few days later if they are not very busy to a month later in high season13:37
wolfspra1lI think if it's a lab you never worked with before, it will be difficult to arrange it all remotely, too many communication issues13:37
wpwrakah yes, better to avoid the pre-xmas rush :)13:37
wolfspra1lthe key is to find a friendly small lab that is officially certified as a FCC lab and then start working with them13:37
wolfspra1lyes but the xmas rush may be around now actually :-)13:38
wpwraki suppose you know such (a) lab(s) ?13:38
wolfspra1llast time I worked with one in nanjing, the docs are uploaded somewhere, which I can locate tomorrow13:38
wpwrak(xmas) hmm. people have too much money. they should spend it only once per year :)13:38
wolfspra1lbut I do think you should first look locally, ideally even in buenos aires13:38
wolfspra1lwell the factories are in xmas season now, more or less13:39
wolfspra1land the cert is always last minute...13:39
wolfspra1lbut I haven't worked with labs in a while, not really up to date now13:39
wolfspra1lan unintentional radiator will not get an FCC ID13:39
wolfspra1lso I think you can keep this all very low key13:40
wolfspra1lin fact most people would not bother at all...13:40
wolfspra1ldefinitely not on the hobbyist or even other 'open hardware' products without naming them now (or pointing fingers...)13:40
wolfspra1lit's more 'in style' to not be certified :-)13:40
wolfspra1lespecially for unintentional radiators (the others most people in that scene wouldn't touch anyway)13:41
wolfspra1lwhy do you think it might radiate at all?13:41
wolfspra1lyou worry too much :-)13:42
wolfspra1lunfortunately I heard horror stories about FCC labs in Europe asking for 25k usd upfront etc.13:42
wolfspra1lthat is quite possible, given that they want to keep the whole flood of hobbyist/garage projects away that would never turn into good customers anyway13:42
wolfspra1lin China the volume is so big in all this, and there are so many (real and fake) labs, they take the small fish as well13:43
wpwrakoh, anything can radiate :) all it takes is switching a few signals quickly enough ... (i think the limits go down to pretty low frequencies)13:43
wolfspra1lsure I know, but still. I just try to put it into perspective.13:43
wolfspra1lthat's why it's called 'unintentional radiator'13:44
wpwraki suspect that, if i find anything here, it'll be prohibitively priced13:44
wolfspra1ltheoretically I would assume mostly everything radiates, at whatever frequency etc. but does it matter? who cares? etc.13:44
wolfspra1lpossible, yes13:45
wolfspra1lbut it will be hard to remote-work with a chinese lab especially first-time13:45
wpwrakyeha, one would need an EE on the ground to act as interface13:45
wpwrakit's basically a form of poverty tax: with proper equipment and a bit of time to learn the rules, one could do all these tests by oneself. that way, there wouldn't be surprises from the certification lab.13:47
wolfspra1lmaybe you do check, I think on the fcc website they have a list of their certified labs somewhere13:48
wolfspra1lhundreds I think, or more13:48
wolfspra1lbut again, maybe you don't need all this, really13:48
wolfspra1lyou would not get an FCC ID anyway13:48
wolfspra1lso you only get some paperwork that shows that your device meets the requirements for unintentional radiators13:48
wolfspra1leven when importing into the US, there are a number of exceptiosn (aside from just incorrect paperwork which many do as well) to get devices without FCC whatever into the country13:48
wolfspra1ljust read what sparkfun and friends write about those little nasty paper-whatever :-)13:48
wolfspra1lfor unintentional radiators, I think you are trying to be too perfect13:48
wolfspra1lpeople build multi-million USD business with intentional radiators and won't spend a penny (or second) on any of this13:48
wolfspra1lk getting crazy tired here, backlog tomorrow13:48
wolfspra1lI will find my old cert paperwork for the ben (which is uploaded somewhere), and also a list of labs maybe13:48
wolfspra1lbut tomorrow...13:49
wpwrakthanks ! sweet dreams ! :)13:49
wpwrakfunny. fcc.gov has a bad certificate ;-)13:50
wpwrakthis seems to be the place with the "known" labs: https://apps.fcc.gov/oetcf/eas/reports/TestFirmSearch.cfm14:00
wpwrakbut there's not a single one in the whole region. not even in brazil or mexico. so this is of limited use.14:01
wpwrakmay well be that everything that gets fcc-certified around here just gets sent to some lab in the US, maybe with a friendly interface person who speaks fluent spanish and/or portugese :)14:02
larscust as a side note, the ben cert paperwork is here http://downloads.qi-hardware.com/hardware/certification/2009_09_30_final/14:19
wpwrakthanks !14:22
qi-botThe build was successful: http://fidelio.qi-hardware.com/~xiangfu/build-nanonote/openwrt-xburst.minimal-20120818-1408 16:26
paul_boddieOpenWrt really doesn't seem to like Qt and Gtk+ these days.16:28
paul_boddieAnyone here have any eglibc experience? I'm trying to figure out the real cause of "libgcc_s.so.1 must be installed for pthread_cancel to work". I think Ubuntu switched to eglibc and now there are thousands of upset people on the Internet.16:32
--- Mon Aug 20 201200:00

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