#qi-hardware IRC log for Friday, 2011-08-05

wpwrakwolfspraul: hmm, the toped screen shot needs more contrast. i can only see at night that it's not just black ;-)05:19
wolfspraulhow can a screenshot need more contrast? but yes, you are right05:20
wolfspraullemme see :-)05:20
wpwrakbtw, do you want to mention the navigation board, for which volunteers are wanted ? http://lists.en.qi-hardware.com/pipermail/discussion/2011-June/008344.html05:21
wpwrakfor the free CAD tools comparison, i'd suggest to add "scripted". they're not your usual CAD tool ;-)05:22
wpwrakalso, the captions are somewhat misleading. maybe "OpenCSG" could be "OpenCSG, showing artefacts" (i don't know if it's supposed to show them of if this is a problem with my setup)05:24
wpwrakthe inside/reversed images also show artefacts that are caused by errors in the model / weaknesses in the modeling erm model, not really bugs in the CAD system05:25
wolfspraulwell sure, thanks. I know the captions are bad, it's hard for me to hit the right ones05:26
wolfspraulI need precise instructions though :-)05:26
wolfspraulwait, opencsg...05:26
wpwrakmaybe for "inside", "Modeling artefact (internal structure)"05:27
wpwrakand for "reversed", "Modeling artefact (reversed face)"05:27
wolfsprauldone05:29
wolfspraulScripted?05:29
wpwrakno GUI. you write a script. (e.g., in the OpenSCAD language or, in the case of Cadmium, in Python)05:30
wpwraksomething like this: http://projects.qi-hardware.com/index.php/p/wernermisc/source/tree/master/cad/test2/button.py05:31
wolfspraulwhat do you want me to edit? I don't get it :-)05:31
wolfspraulfor the toped screenshot, the software draws the lines 1-pixel wide05:31
wolfspraulnot sure how I can easily fix that in gimp, trying...05:31
wpwrak(scripted) maybe simply s/free CAD tools/free scripted CAD tools/05:32
wpwrakmaybe contrast-enhance the scaled image ?05:32
wolfspraulnah05:32
wolfsprauldoesn't work05:33
wolfspraulthe problem is toped itself05:33
wolfspraultheir own visuals are not good05:33
wolfspraulif I increase contrast and brightness the cat becomes clearer, but then all menus disappear :-)05:33
wolfspraulthe problem is the 1-pixel wide line05:33
wolfspraulah, maybe I just increase it for that region...05:34
wpwrak(off-topic) nice article on "heroic scientific development": http://web.archive.org/web/20080512050604/http://apperceptual.wordpress.com/2007/01/15/the-heroic-theory-of-scientific-development/05:34
wpwrak(region) yeah ;-)05:34
wpwrakcontrast is often a bit of a problem. i also need to add some knobs to fped - it displays quite a number of things in very dark colors that look great at night but are almost invisible at daytime05:35
wpwrak(or maybe just my screens are crap :)05:36
wolfspraul[free scripted] done05:52
wolfspraul[cat contrast] done05:52
wolfspraulanything else?05:52
wolfspraulneed to rearrange, then push it out. speed speed speed :-)05:52
wpwrak(rearrange) the thumbnails ;-)05:53
wpwrak(cat contrast) excellent !05:54
wpwrakwolfspraul: btw, once the news are done, you ought to post about the fake components. supply chain security is an interesting issue, and open hardware would encounter it at several points06:06
wolfspraulyes I agree, interesting story06:29
wolfspraulbtw, I am not 100% sure they are fake. it's very difficult to distill really hard facts.06:30
wolfspraulbut still feel comfortable saying 'fake', if I mix in my experience and the evidence we have06:30
wpwrakwell, a trace of uncertainty remains with everything :)06:31
wolfspraulwpwrak: with you the well known SD lover, have you read Bunnie's article about microSD cards? http://www.bunniestudios.com/blog/?p=91806:32
wolfspraul"Armed with this evidence, we confronted Kingston  both the distributor in China as well as the US sales rep. First, we wanted to know if these were real cards, and second, if they were real cards, why were the serialization codes irregular? After some time, the Kingston guys came back to us and swore these cards were authentic, not fakes, but at least they reversed their position on not offering an exchange on the cards  they took back the p06:34
wolfsprauljust read the entire article if you haven't in the past... I feel our TI schmitt-trigger story is very similar.06:34
wpwrakyeah, i've read it a while ago. nasty.06:34
wolfspraulalso interesting how Kingston (!) first swears they are authentic (not fake), then takes them back which they would not have done before (of course those cards will be discarded)06:35
wolfspraulthere's a lot of dishonesty on all sides06:35
wolfspraulsales people doing business on the sides06:35
wolfspraul'extra shifts'06:35
wolfspraulre-marking06:36
wolfsprauland so on, it's endless06:36
wolfspraulso actually I correct that. the cards will probably not be discarded but end up yet somewhere else on the 'spot market' :-)06:36
wpwrakwhat's bad there is that they seem to have come straight from the official chinese distributor. so even if you follow the supply chain quite meticulously, you get screwed06:37
wolfspraulbecause the dishonesty goes all the way up06:37
wolfspraulI have no illusions06:37
wolfspraulall the way to the semiconductor fabs and foundries06:37
wpwrakwell, but kingston have a strong motivation to prevent junk being sold under their brand06:37
wolfspraulI don't know. that's the thing, there is so much dishonesty everywhere that at some point you don't know anymore who is fooling whom :-)06:38
wolfspraulI can say for sure that the sourcing shop we bought from is 'honest' (VC40 in the wiki)06:38
wolfspraulthat's just a feeling though, and it's a very deep supply chain06:39
wolfspraulthere could be TI people involved in this as well, who knows06:39
wolfspraulrogue cops everywhere :-)06:39
wolfspraulbut from the pattern, my guess is it's a cheap chinese backyard job, done with full knowledge of whom? don't kno06:40
wolfspraulknow06:40
wolfspraulbut yes, we should write up the facts in a sentence or two and make a little story out of it. I agree it is important for others to consider.06:40
wpwrakfor TI, the case would be even stronger. they're considered an original source, even more so than kingston (who don't produce chips themselves)06:41
wolfspraulsimilar to the Bunnie article which I greatly enjoyed. Also his insistance on finding the truth :-)06:41
wpwrak(bunny) yeah, quite some investigation :)06:41
wpwrakerr, s/bunny/bunnie/06:42
wolfspraulI'm not sure about ti 'original', I don't know06:42
wolfspraulmaybe they license? subcontract?06:42
wolfspraulwho knows, really06:42
wolfspraulthe entire industry is not known for transparency06:43
wolfspraulone time I bought xilinx parts 75% below list price in China06:43
wpwrakyes, this is unfortunately so06:43
wolfspraulok?06:43
wpwrakheh ;-)06:43
wolfspraulI was in Silicon Valley and wanted to talk to Xilinx about it06:43
wolfspraulbecause I want to come clean, I want to understand whether they are really 'real', and if not where I should buy 'clean' parts06:43
wolfspraulthey are absolutely not interested in discussing this :-)06:44
wpwrakyeah, differential pricing confuses customer expectations06:44
wolfspraulthey may very well be behind some activity that the native observer may consider 'illegal' or 'fake'06:44
wolfsprauls/native/naive/06:44
wpwrakoh, can be just different prices for different customers06:44
wolfspraulthey do _not_ want to talk about real vs. fake06:45
wolfsprauland I am sure they have very good reasons for that06:45
wolfspraulso I've learnt my lesson on this. what is fake? who knows.06:45
wolfspraulI know those parts failed, and caused me rework trouble in my run. I don't like that.06:46
wpwrakluckily, they were relatively easy to replace. imagine the same with a BGA06:46
wpwrakor COB :)06:46
wolfspraulI see suspicious things on markings, which even though it may look helpless, those little things may give more clue about what's going on in reality than investigating into the supply chain.06:46
wolfspraula lot of people are very aware of the slightest variation in silkscreens, little labels, little special dots and lines and rounded edges and what not06:47
wolfspraullike identifying fake currency :-)06:47
wpwrakin the future, part numbers will be replaced with high-resolution pictures ;-)06:47
wolfsprauloh absolutely06:48
wolfspraulI totally agree with Bunnie's approach06:48
wolfspraul1) microscope06:48
wolfspraul2) use the Linux kernel to look at every last byte of ID that can be read06:48
wolfspraul3) look at dies, just delayer to see them06:48
wolfspraulit may sound silly, but that's how it works :-)06:48
wolfspraulnot talk to your supply chain, you will just get more lies06:49
wolfspraulonce we are sourcing nothing but silicon wafers, I think we are safe06:52
wolfspraul(just kidding)06:52
wpwrakyeah. they've cheated you already once, so why trust them another time ?06:53
wolfspraulwe need to make some products before that...06:53
wolfspraulalso nobody really knows06:53
wolfsprauland is surrounded by more strange things06:53
wolfsprauland all those lots are flowing, it's hardware06:53
wolfspraulit leaves little traces06:53
wolfspraulsome sales guy may do something once, to test whether he can make extra money06:54
wolfspraulthen it stops06:54
wolfspraulhe waits to see whether there is any backlash06:54
wolfsprauland so on06:54
wpwrak(wafers) yeah. the day you get to open your own fab will be a liberating moment. 06:54
wpwraki wonder if public shaming works. post all the details, including supply chain contacts. then switch suppliers.06:55
wpwrakif enough people do this, there may be a perception that those tricks don't pay off in the end06:55
wpwraknot for everyone, of course, but for people who are in a better position to enforce compliance than you as the customer06:56
wolfspraulnah, won't work06:57
wolfspraulillegal acts are done in secrecy06:57
wolfspraulobviously06:57
wolfspraulyou are always surrounded by people who claim they like transparency06:57
wolfspraulsome of them are lying06:57
wolfspraulwe have nothnig but friends when we go public, this is totally fine06:57
wolfspraul:-)06:57
wolfspraulthe bad guys won't care, they depend on their actions remaining anonymous/untraceable before and after publicity06:58
wolfspraulhunting them down is hard, see how the Kingston story ended06:58
wolfspraulnobody is really interested in cleaning it all up06:58
wpwrakwell, you can expose at least the last step in the chain. make it their problem.06:58
wpwrakthey can in turn work upstream. they may be big enough to be heard.06:59
wolfspraulagain, the TI part problem is _not_ related to our supplier (VC40, a 2-person sourcing company in Shenzhen)06:59
wolfspraulyes sure, of course06:59
wolfspraulEva (the owner) would love to find out about it06:59
wolfspraulbut her tracking is weak06:59
wolfspraulit may have been a one-off thing (see above)06:59
wolfspraulany investigation will just waste time06:59
wolfspraulif Eva's tracking is good, maybe she can trace back where she bought that reel07:00
wolfsprauland then?07:00
wpwrakwell, if you trust your supplier 100%, you can also include that, plus all the information they are willing to give you in the public shaming07:00
wolfspraulonce you hit the bad guy they will respond with a lie, that simple :-)07:00
wolfspraulwhat do you do to cover up a lie? invent another lie! :-)07:00
wolfspraulit won't work, trust me07:01
wolfspraulwhat was the result of Bunnie's work?07:01
wolfspraulhis blog post07:01
wolfspraulenlightenment for the world07:01
wolfspraulanything else?07:01
wpwrakprobably too obscure for a major impact07:01
wolfspraulI have already made my conclusions practially for rc407:01
wolfspraulwe will source everything that is easily sourcable (or even cheaper) on digikey, instead of the Chinese spot market07:02
wolfsprauland then we see07:02
wpwrakof course, it means that the Ya should ship with cards from sandisk, not kingston :)07:02
wolfsprauland in parallel sure we could contact Eva and see whether she can trace back the reel, for fun07:02
wolfspraulyou bet07:02
wolfspraulexcept for 'fake' Sandisk cards on the market?07:02
wolfspraulah no, wait. nobody fakes those.07:02
wpwrakbuy from sandisk directly ? or from a distributor they recommend ?07:03
wolfspraulyou will always be a believer, I can tell :-)07:03
wolfspraulthe bad guys are in the system07:03
wpwrakif you allow the supply chain to be broken, you're lost07:03
wolfspraulit's not broken, it's full of humans07:03
wolfspraulit's hardware, not software07:03
wolfspraulno md5sum07:03
wolfspraula sales guy urgently needs extra money, mom is sick. he decides to make an exception, only once, when shipping stuff to one customer07:04
wolfsprauldoesn't happen?07:04
wolfspraulbah07:04
wolfspraul:-)07:04
wolfspraulthen he likes the money he made, does it again07:04
wpwraki think these are the exception :)07:04
wolfspraulsometimes it goes big, sometimes stays small07:04
wolfsprauloh no07:04
wpwrakmost of them are a little more systematic07:05
wolfspraulnot my experience07:05
wolfspraulbest is strict IQC, everywhere07:05
wolfsprauland yes, only work with people you trust, lots of personal visits07:05
wolfspraulit helps07:05
wolfspraulanything suspicious - stop working with them07:05
wpwrakyes, IQC is great. border controls07:05
wolfspraulyour sales guy suddenly drives a Ferrari? stop working with them07:05
wolfspraulthat kind of thing07:05
wpwrakheh ;-)07:05
wolfspraulit's hardware, really07:06
wolfspraulI wish there were a magic way out, but there is not07:06
wpwrakapplying social control, why not07:06
wolfspraulI can say that for sure after all these years and endless meetings from the smallest sourcing shops on the streets in SHenzhen to the uber-sales guys at Samsung HQ in Seoul07:06
wolfspraulit's a mess, really. hardware is dirty. real world.07:06
wolfspraulno md5sum, really07:07
wolfspraulevery shipment is new07:07
wolfsprauleverybody has some secrets to cover up, at least sometimes, temporarily :-)07:07
wolfspraulyou constantly have to work against that, it won't stop07:07
wpwraki think a manufacturing company can protect itself against shipping fake goods07:08
wpwrakso the question is what pressure it takes to make them implement enough internal checks07:09
wolfspraulbtw, I can also say that overall the problem is small. a few percent. it pops up, you fight it.07:09
wpwrakremember all the issues with fake x86 CPUs ? those seem to have largely disappeared. not sure how exactly they solved it. perhaps just tightened the supply chain.07:10
wolfspraulanybody from small to large manufacturers has the chance to survive, it's not a deadly threat if you are aware of it and take some precautions.07:10
wpwraka few percent is certain doom on a complex device07:10
wolfsprauland I would say 95% or more of people around you will honestly try to weed this out, all the time07:10
wolfspraulof course, otherwise there would be no functioning consumer electronics :-)07:11
wpwrakhopefully a lot more than 95% ;-)07:11
wolfspraula few percent of sourcing jobs07:11
wolfspraulno I don't think so07:11
wolfspraulthere are bad guys, they try to cover something up07:11
wolfspraulit starts small07:11
wolfspraulthe chip has some bug, but the sales guys are told to cover it up07:11
wolfspraul"take it back silently if the customer runs into the bug"07:11
wpwraki wonder who the typical victims are. may be small operations like sharism.07:12
wolfsprauleverybody, it's just in the whole system. I am sure Foxconn is affected too, talk to Tony :-)07:12
wpwrakthe big ones certainly have their IQC in place, so there's little point in probing them. will just piss them off.07:12
wolfspraulwell but the fake parts may work07:12
wolfspraulor the bug they are trying to cover up may not manifest itself for that particular customer they are selling the chip to07:12
wolfspraulthere's a lot of grey07:13
wolfspraulif Foxconn gets fake parts and they work, do you think they care?07:13
wolfspraulshould they?07:13
wpwrakah, so they're almost real. i see. i thought they were just completely broken07:13
wolfsprauldo they proactively investigate this?07:13
wolfspraulhow many resources for taht?07:13
wpwrak(broken/wrong chip)07:13
wolfspraulI don't know07:13
wolfspraulI will not investigate it forever, throwing more good cash after bad cash.07:13
wolfspraulno point07:13
wolfspraulthis reel was bad07:14
wolfspraulI change rc4 sourcing07:14
wolfsprauldone07:14
wpwraksend the reel to bunnie ;-)07:14
rohhrrr07:14
rohbbl.. will send out the 2nd shipment now07:14
wolfspraulnice07:14
wolfspraulI am sure Tony could tell you lots of stories from Foxconn07:14
wolfspraulthere is a lot of money to be made in this07:14
wolfsprauland if you think nobody tries, oh well07:14
wolfspraul:-)07:14
wolfspraulpeople do crazy things for a few USD07:14
wpwrak(foxconn caring) yeah, if it passes IQC, then it's beyond their radar too. at least they're keeping the barrier high. that in turn helps everyone.07:15
wolfspraulIQC is very important07:15
wolfspraulcatch it early, fight back hard07:16
wolfspraulsounds like a Chinese 'enforce the law' campaign :-)07:16
wolfspraulstrike the offenders07:16
wpwrakdidn't they "solve" their drug problem that way ? :)07:16
wolfspraulone time I bought a microSD card, just one for myself07:20
wpwraknow wolfgang admits his dirty secrets :)07:21
wolfspraulwhen I was at the vendor, the price was too low, ridiculously low. like 30 US cents, instead of 4-5 USD.07:21
wolfspraulno wait07:21
wolfspraulmy Chinese is not very good, so I try to communicate with that guy07:21
wolfspraulI say "it's too cheap", "do you have a real one"07:21
wolfspraulbla bla, I don't understand him07:21
wpwrak;-)07:21
wolfspraulI say, OK, whatever, 30 cents, gotta go07:21
wolfspraulof course it didn't work07:22
wolfspraulthen I realized - he is selling broken cards07:22
wpwrakpity07:22
wolfspraulwhy not?07:22
wolfspraulhe knows, he will be open about it07:22
wolfspraulit's a business for broken microsd cards07:22
wpwrakhehe07:22
wolfspraulthe price is 80-90% under what it should be, just scrap value07:22
wolfspraulwho is he selling to?07:22
wolfspraul(except for exotic cases like myself)07:23
wpwrakcriminals07:23
wolfspraulprobably :-)07:23
wolfspraulmaybe artists who are planning an installation with microsd collage?07:23
wolfsprauland they need 10,000 cards, just for their artwork?07:23
wpwrakbig market, that one :)07:23
wolfspraulcould be, right?07:23
wolfspraulwell07:23
wolfspraulis this an illegitimate business? (the guy selling)07:23
wolfspraulhe did not betray me, that's for sure07:23
wolfspraulif I would have understood him better, he would have said "may not work"07:24
wolfspraulof course most likely 90% of the cards he is selling will end up in some 'shady' use case07:24
wolfspraulbut we don't know07:24
wpwraki'd call it legitimate. the problem is what happens next07:24
wolfspraulit still could be an artist who needs cards for an installation...07:24
wolfspraulyes correct, I think so too. legitimate. but barely :-)07:25
wolfsprauland then somewhere is the guy who cannot resist, and will take the money07:25
wolfsprauland the price goes from 20 cents to 3.50 USD :-)07:25
wpwrakof course, that vendor may make an easier target to hit, so making his biz illegal may help to fight this sort of fraud07:25
wpwrakthen he has to operate underground, increasing his risk and maybe driving his prices up, which in turn reduces the margin07:26
wpwrakso as long as uSD cards themselves stay legal, that approach may work :)07:27
wpwrakah, and what would work even better is getting rid of patents ! ;-)07:27
wolfspraulyou have some crazy ideas there ;-)07:28
wpwrakthen almost all the money can go into the production, greatly reducing the margin for fraudulent products07:28
wolfspraulI'm just telling you some examples of grey, a lot of grey. and then someone somewhere will make the last step and take the money.07:28
wolfspraulbut the probelm is contained, really07:28
wolfspraulthe industry is healthy, 95% of people are honestly trying to weed this out07:28
wolfspraulbut if you are doing sourcing on a regular basis, yes, you will run into this kind of thing07:29
wolfspraulstrong IQC, lots of microscope pictures, always work towards trustworthy sources, react swift and hard if you are fed crap07:30
wolfspraulask Tony for more advice :-)07:30
rohwolfspraul: the package is at the shop now. will go out 16:30ish (usually the pickup time)07:30
wolfspraulperfect07:31
wolfspraulI think the first package is on the way to Taiwan already, crossing my fingers07:31
wolfspraulroh is a source I trust07:31
wolfspraulhow about that?07:31
wolfspraulbut maybe there's a bad customs agent in between somewhere? nah... :-)07:31
wpwraki think when you get the first fake M1, that's a day to celebrate ;-)07:32
wolfspraulseriously, shipping people are swapping goods too!07:32
wolfspraulwhen you have a tray full of bad microsd cards (above example), it's not that hard to convert it to cash07:32
wolfsprauland some trays of this and that can be quite expensive, as we all know07:33
wolfspraulopen the box, slip it in, close the box07:33
akiwiguy8:10*07:33
wolfspraulsell the now good tray on the spot market07:33
wolfsprauldoesn't happen?07:33
wolfspraulha ha07:33
rohwolfspraul: the tracking shows the first has left germany.. atleast i think it has07:33
wolfspraulby the time anybody has foudn out about that bad tray the connection and evidence leading to the shipping company is long gone07:34
wolfspraulso there's a lot of quite sophisticated sealing going on too07:34
wolfspraulI really should sit down with Tony for some stories... :-)07:35
wolfspraulwe always missed, tried to meet a few times but he is running around too fast07:35
wolfspraulakiwiguy: the bad stuff only happens with microSD, not with 8:1007:36
rohwolfspraul: first package is 19.5 kg, second is 6.9kg07:36
wpwrakyou're making a good case for smuggling ;-) that way, you could ship completely sealed packages, and just discard any that have been opened.07:36
rohsecond one has a few spare shielding sheets, a dozend buttons and some 'cap missing' buttons07:36
akiwiguywolfspraul: xD07:36
wolfspraula lot of nasty things are happening, seriously there is a lot of easy money in this industry07:36
wolfspraulthose little thingies are just too valuable, too easily exchanged etc.07:37
wolfspraulroh: I will send some material back to raumfahrtagentur, a few boxes, also brochures, stickers, etc.07:37
roh:)07:37
wolfsprauljust promotional material, use it as you wish07:37
rohyeah.07:37
rohhow much are these cameras? if they are good i could use some of them here at the agency (mount them to machines to remote keep watch)07:38
wolfspraul25 USD07:38
wolfspraulmy most expensive accessory :-)07:39
rohi see.. well.. then they HAVE to be good ;)07:39
wpwraka big problem is that customs are allowed to open things. otherwise, you could easily create the right incentives.07:39
wolfspraulno no. this wouldn't work. maybe they are 'allowed' but for any serious import/export company they will work directly with customs for things like sealed boxes.07:40
wolfspraulotherwise customs would destroy everything07:40
wolfspraulmaybe this happens in North Kore, but not China the 2nd largest economy in the world07:40
wpwrakyou could still do a few other things, like putting individual random laser markings on the devices. then scan the tray. that would be expensive to duplicate.07:40
wolfspraullet's say shipments of bare dies, shipments of all sorts of things that are sealed, even simple vacuum wrapping07:41
wolfspraulif customs has an issue, they will contact the agent, xray, etc.07:41
wolfspraulis not like crazy "rip open everything"07:41
wolfspraulyes correct! [laser markings]07:41
wolfspraulthose little things work best07:42
wolfspraulvery effective07:42
wpwrakyeah, but they can open the box. so you can't just ensure integrity of the shipment.07:42
wolfspraulbecause in the end you are hunting down rare cases, individuals (mostly) who try their luck07:42
wolfspraulso you need low-tech approaches, laser markings, frequent little changes to logo or how/where date codes are written07:43
wolfspraulthose are very effective, and what most people use to keep the quality of the supply chain up07:43
wolfspraulroh: what do you mean with 'good'? I don't understand07:44
wolfspraulyou mean the mechanical housing?07:44
wolfspraullens? CCD sensor?07:44
wpwrakyeah, drive the cost of making an authentic-looking device up and drive the whole device cost down. at some point, the fraudster's biz just collapses07:44
wolfspraulit's an analog camera07:45
wolfspraulthe one we are including has a waterproof metal housing, bullet type07:45
wolfspraul1/3 '' Sony CCD sensor07:46
wolfspraul20 tpi (standard) tripod threading07:46
wolfspraulquite a nice little thingie07:46
wolfspraulone problem I found is that if you turn it a lot, say in your hand, the cable that comes out from the end may slip out, exposing the inner wires (still shielded)07:49
wolfspraulbut even in that state it's still very nice and fully functional (not waterproof anymore though)07:49
wolfspraulI will wait for feedback07:49
wolfspraulalso the software on m1 can improve a lot with regards to the camera, I think before making more improvements on the cam we would want to make more improvements to the software using the cam data.07:49
wolfspraulwpwrak: I like Bunnie's terminology. I will call the TI parts 'irregular', not 'fake'. And I may even down-play the TI association because from looking at the data, I think it's unlikely and unfair. So I will say 'irregular schmitt-triggers'.08:22
qwebirc49919hello ?08:40
qwebirc49919anybody here?08:40
qwebirc49919¡º(08:42
wolfspraulqwebirc49919: hi08:46
xiangfuqwebirc49919, Hello08:55
qwebirc49919I have no name?09:00
kodeinit's "qwebirc49919"09:01
wolfspraulit's an automatically assigned name, you can try to type: /nick new_name09:02
wolfspraulstart with the / in a new line09:02
xiangfuqwebirc49919, from your IP address, you are in Beijing. cool.09:07
rohheh. YAY09:37
rohi just realizied i now have shipped over 100 case kits :)09:37
kyakkristianpaul: i did really have a chance to test at* stuff for the last month, so the current status for is "ssh somehow works with dirtpan" :)10:03
kyak*didn't10:03
kyaki have some ideas about pluggin atusb thing into my 'wdtv live hub' media player, which recently got rooted and so i'm able to build kernel modules for it. The problem is that wdtv is running very old kernel (2.6.22) and atusb driver can't be yet compiled as module10:05
kyakbut that would be cool.. I won't have to keep my laptop running to provide wireless to Ben10:05
xiangfukyak, hi, I tried last dailybuild image. by report bug the 'at' package compile fine now. 10:11
xiangfubut 'stardict' 'mediatomb' 'python' not working. compile fine. but when I start them just some 'seg fault'10:12
xiangfukyak, do you know if the early daily build works fine?10:12
xiangfuI am download the early version now: 07052011.10:14
kyakxiangfu: the latest build i flashed was probably from one month ago10:25
kyakxiangfu: the stardict is not working here, too :(10:27
kyakprobably it's been broken at some point, but i'm using qstardict10:28
kyakwhich does work and doesn't hit OOM killer with the same dictionary as stardict10:28
wolfspraulwe need only one stardict10:31
wolfspraulif qstardict works better, why not just remove the other one?10:31
wpwrakwolfspraul: (to association) naw, full disclosure, please :)12:11
wpwraks/to/ti/12:12
wpwrakif using "irregular", you may want to clarify what you think has happened. e.g., that a reel of bad parts has been sourced somewhere along the way, but that you don't believe they're original TI parts that are just messed up12:16
wolfspraulwpwrak: well, it's difficult. I am thinking from the headline backwards. I think we are doing a better service by first saying 'irregular schmitt-trigger parts', because if we say 'fake TI parts' that's sensationalist and will mislead the casual reader, imho12:25
wolfspraulof course we ran into the 'irregular schmitt-trigger parts' trying to source TI schmitt-triggers12:25
wolfspraulI don't want to hide that at all12:25
wolfspraulbut it's the second step of the explanation, imho12:26
wolfspraulsmall detail, maybe I'm over-cautious12:26
wolfspraulthis is like police work, you don't want to spread accusations all around you, in the end you may just burn good people12:26
wolfspraulI very much like the approach Bunnie took in his post, you can tell he's relaxed and all and has the right focus. that's why I feel better with 'irregular schmitt-trigger', as the first line.12:27
wpwrakbut "fake TI parts" will draw more attention :-)12:27
wolfsprauldefinitely12:27
wolfspraul:-)12:27
wpwrakif this gets big headline news, it's much more likely that someone will make the effort of hunting down that fraudster12:28
wolfspraulI won't make any big effort on that news item, it's a small detail in the end12:29
wpwrakwell, unless it gets too much attention, in which case they'd need to "show quick results", and hang the next best scapegoat they can find12:29
wolfspraulthat's a very small and crappy backyard shop that mixed some bad parts in, it's really not worth that much attention12:30
wpwrakof course, even that could serve to "encourager les autres" :)12:30
wolfspraulbut I will write it up nicely (actually Adam already did most of it) because it's a nice little experience that should be remembered12:30
wpwrakwell, it is. you lost some money and were at risk of losing a lot more. consider this happening with a 1000 pieces order and distributors already out of stock12:31
wolfspraulbut it's totally normal in sourcing12:31
wolfsprauljust under-documented a little maybe12:32
wpwrakif that shop hadn't proven to be untrustworthy now, they may have bitten you much worse in the future12:32
wolfspraulbecause most people that run into this on a daily basis, and fix it on a daily basis, won't blog12:32
wpwrakperhaps that's why it keeps on happening :)12:32
wolfspraulpossible, although experience sourcing people share the knowledge with each other12:32
wolfspraulonly we are new :-)12:33
wpwrakit's like kidnapping. as long as you just pay and go on, that biz model remains attractive12:33
wolfspraulnah don't worry. it's not increasing, spreading, anything.12:33
wolfspraulbut I agree we should document it well, which we do12:33
wolfspraulbtw, a lot of Adam's new pictures come from a 100 USD USB microscope he bought before the rc3 run12:34
wolfspraula good investment :-)12:34
wpwrakhehe ;-)12:34
wolfsprauldon't know the magnification, but it's nice12:34
wolfsprauland quick too I would think12:34
wpwrakplays nice with linux ?12:34
wolfspraulaw: you there?12:34
wolfspraulis that right that you bought a USB microscope before the rc3 run?12:34
wolfspraulwhich model is it? how do you copy pictures to your Linux notebook?12:35
wpwraki once looked around for USB microscopes but got mixed results. it seems that the ones that give you good focus also tend to need proprietary drivers. but that was some years ago. maybe things have changed since.12:36
awwolfspraul, yes, i bought it before rc312:37
awi shared my windows's folder so that my ubuntu laptop can access it. ;-)12:37
wpwrakaw: you had a moment of cleirvoyance ;-)12:38
wpwrak(windows) awww :-((12:38
awwpwrak, hehe12:39
awwolfspraul, http://www.dino-lite.eu/index.php/en/digital-microscopes/dino-lite/45-dino-lite/256-am201112:40
wolfspraulwpwrak: yeah! we are open about everything, even our sins12:42
wolfspraulhow about that?12:42
wolfspraulmaybe not all of our sins?12:42
wolfspraullet's add the copyleft hardware microscope to the todo list...12:42
wpwrakaw: here's a nice accessory: http://www.microscope.com/dinolite-stand-ms21w-flexible-versatile-stand-p-522.html12:46
wpwrakaw: or, fancier: http://www.microscope.com/dinolite-stand-ms52b-articulating-boom-with-clamp-p-662.html12:47
wolfspraulaw: have you made any new discoveries about boards with flash or impedance problems?12:49
awwpwrak, yes, i was thought that to buy with stand and clamp...well...the results of rc3 was out of my plan though. ;-) exactly if having yes stand/clamp is good.12:49
awwpwrak, are you using one of them?12:50
awwolfspraul, not yet though..i am fixing that VIN: I2C fail. ;-)12:50
awi was thought this one should be easiler to fix then still hehe...;-)12:51
awVIDEOIN_SDA came from fpga is bad. ;-) http://en.qi-hardware.com/wiki/File:M1_rc3_0x4d_ng_VIDEOIN_SDA.JPG12:52
awthe good one must be like this: http://en.qi-hardware.com/wiki/File:M1_rc3_0x4e_good_VIDEOIN_SDA.JPG12:52
wolfspraulhow can you fix it?12:53
awthere's no any parts between fpga and U21's 53 pin(ie. TP15 = VIDEOIN_SDA .12:53
wolfspraulso how can it be fixed then? that pin under the fpga is badly soldered?12:55
awi took out video decoder then the same bad SDA siganl. So i just threw out my final step (using hot air to blow fpga body), not yet tested. ;-)12:55
rohcan it be a bad pullup?12:56
rohbadly soldered resistor12:56
awwpwrak, i do believe that sda is needed slave to be activated(for ack) from sender 12:56
awroh, yes, i thought there's no pull up successfully.12:57
awroh, no any resistors placed between U21's 53 pin(VIDEOIN_SDA) and fpga(ball AB17 )12:58
rohhm.12:59
wpwrakaw: (scope) no, i just use my digital camera. it's has a pretty good macro. but lighting is a big issue if i try to get close.12:59
awso there's internal pullup in fpga though, also i just thought all functionalities from fpga all work well, why this SDA signal is bad? I checked VIDEOIN_SCL, HS, VS, FIELD are all okay.13:00
wolfspraulaw: can we replace the entire fpga?13:00
awwpwrak, i see.13:00
wolfspraulI mean of course we can, but what would be the cost, and would it result in a sellable board?13:00
rohi see.. internal pullup. what happens if you add an external pullup?13:01
awi can't but factory can do this. No cost needed, i can request minbo they do this without charge show that I have no idea why SDA came from fpga no good. 13:02
wolfspraulnice13:02
awroh, i've not used external pullup13:02
wolfspraulif this results in another sellable board, without excessive work on our side, let's do it13:02
wolfspraulbut if the board is too ruined from too many reworks to be considered sellable, then let's just forget it13:02
wolfspraulyour call13:02
rohaw: can you try and see what happens if you do? maybe the internal pullup is broken and the pin is ok otherwise13:02
wolfspraulroh: but in that case we could not sell the board still13:03
wolfspraulI would not want to sell it13:03
awroh, but should not be like that, the SCL's level stays very well @ 3.3V .;-)13:03
wolfspraulso trying this rework only brings the board one step closer to unsellable state13:03
wolfspraulwe are not doing design work...13:03
rohheh. ok.13:03
wolfspraulinstead, replacing the entire fpga _may_ get us to a sellable board13:03
wolfspraulwell that's just my thinking13:03
wolfsprauladam is in charge of the production13:03
wolfspraulif he wants to experiment with external pullups, fine :-)13:04
wolfspraulI don't see the value from a production perspective, though13:04
wolfspraulI'm learning...13:04
awha~ i won't actually  spend time to add it.13:04
rohtrue. was just thinking that its less work to add a resistor instead of swapping a bga chip13:04
awsince if I still get no work I'll gather boards later then send back to factory.13:05
Action: roh will sleep now, then find an atm and pack for camping13:06
wpwrakindeed, an external pull-up seems to be a fairly easy thing to try. the voltages are a little strange, so one more data point could be intresting. if only to satisfy curiosity :)13:36
wpwrak(strange voltages) one theory would be that the pin is shorted to another pin driven low. but the pull-up vs. Ron ratio is off by about one order of magnitude for this.13:38
awwpwrak, i checked ball AB17's surroundings balls under fpga. there's two possible conditions could make this route as a lower magnitude. one is the nearest adjacent ball  to AB17 is exactly ground (AA17). the other is the pcb's ground may close to this.13:57
awwpwrak, i'll reserve this to take xray when I need later. ;-)13:57
awwpwrak, surely roh's idea is good for experiment of course. ;-)13:58
awwell...i stop on checking this board. next to check the most bigger problems: impedance.14:00
wpwrakhmm, pcb ground seems unlikely. the internal pull-up is something like 10kOhm or maybe a bit more, if i remember correctly14:00
wpwrakinto PCB GND, you'd have something like a 10k:1 voltage divider. that would then be 0.33 mV instead of 200 mV :)14:02
awwpwrak, yeah..me too think it's unlikely though. since liked i found that all other functionalities are okay, so I would prefer to think it's a pcb somewhere problem14:02
wpwrakinto another pin that's driven low you'd have ~10k:100. still only 33 mV instead of 200 mV, but closer :)14:02
awwpwrak, exactly though.14:02
wolfspraulbad fpga?14:03
wolfspraulwhat's more likely? bad fpga or bad pcb?14:03
wolfspraul(assuming it's not a soldering problem)14:03
wpwrakthere's only one way to find out :)14:03
wpwrakah, #1 suspect would be soldering/contamination14:03
awi quite think it's not said a fpga ball problem until i take xray.14:04
wpwrakyou can test the internal pull-up with an external pull-up. if the internal pull-up is broken, but all the rest is okay, the external pull-up should fix the problem.14:04
wpwrakif you remove the fpga, you destroy all other evidence with it, so that may fix the board but it won't help to gain any new insights14:05
wolfspraulthere's not always a valuable insight to be found14:05
wpwrakthat's of course true14:05
wpwrakbut the external pull-up should be a 5 minutes low-risk experiment. compare this to swapping the entire fpga :)14:06
awhi...i temporarily reserve this board late to fix. It's good surely that all rest are okay and seems adding external pullup then pass, isn't a good idea from roh? ;-) not bad.14:07
awphew~i take rest firstly, tomorrow tell you the good news though. ;-)14:08
wpwraki like the idea. as long as we don't know what to look for, tests that allow for incremental examination are good. any chances are that anything you hit once with 90 boards is something that will eventually come back when you make 1000s of them ;-)14:09
wpwrakbtw, what's the impedance thing ?14:09
wolfspraulwpwrak: not always, with 90 you are already slowly in 'mass production' territory. you can do the math and be surprised how easy it is to spend too much money on a board.14:10
wolfspraulof course, I agree, with 90 you may still take every one kind of serious14:10
wolfspraulbut with 250 you definitely don't do that anymore14:10
wolfspraulit's more important to move to the next 250 :-)14:11
awwell...time to take rest. I'll keep irc posted tomorrow. :) night.14:11
wpwrakwolfspraul: hmm, the news are still hard to read. how about just putting one slide of each presentation, shrunk to no more than one Ben (320 x 200 pixels), and then link to the full collage ?14:47
wolfspraulwait still moving around14:48
wpwrakmany of the pictures in the nanonote section have a good size. it think the set with tuxbrain making the wpan boards is nearly perfect14:50
wolfspraulyes some stuff is too big, no worries14:51
wolfspraullet me finish it first14:51
wpwrakso .. maximum height, 166 pixels :) no more than one line of images. then it looks good.14:51
wolfspraulwell it depends14:51
wpwrakthe tumbnails don't really need to be readable14:51
wolfspraulAdam's testing setup is crowded, hard to see the m1 board if it's too small14:52
wpwrakthey just need to show you what you'll find there14:52
wolfsprauland you don't want too many different image sizes either, that looks unstable and unnerving14:52
wolfspraulbut let me just go through first, I am slow14:52
wpwraki think a relatively constant small image size would be ideal. "click here to see a bearded guy making wpan boards", "click here for adam's crowded lab"  :)14:53
wolfspraulthe box is also difficult to make smaller14:53
wpwrakwhy ? make it small. people can still recognize the outline and see that there are more details. again, it's just one click away :)14:54
wpwrakso you write the news directly in the wiki or do you have some source that you process into wiki syntax ?14:54
wpwrakin the latter case, you could generate an alternative version with all images "expanded". in case someone wants this.14:55
wpwrakcheck after ~3 issues to see if people still click on the "expanded" thing :)14:55
wolfspraulI'll go through tomorrow, mostly I agree14:57
wolfspraulbut some things cannot be too small14:58
wolfspraulfor example the videos are probably either watched inline or not at all14:58
wolfspraulso I need to pick a size that is pleasant to watch14:58
wolfspraulI'll see what the youtube default is :-)14:58
wpwrakoh, and i'm not sure having two levels of hierarchy really works that well. i was actually about to suggest trying that, but now that i see it, i've changed my mind :)14:58
wpwrakyoutube has really small thumbnails ;-)))14:59
wpwrakalso, the inline viewer doesn't really work for me. if i watch a video, i have to click my way through to the file, wget it, then view it with mplayer15:00
wpwrakso at least for me, a teaser image plus a direct link to the file (without intermediate html page) would be ideal. not sure what works in other people's setups, though. i would expect many browsers to start a more or less usable viewer15:02
wpwrakthose intermediate html pages are a bit of a disease. most of the time, they don't really add value, but only make it harder to get at the real content. i can understand their legal purpose for recording authorship, mandatory credits, and such, but they're still a pain.15:05
wpwraki think what looks "right" is a certain ratio of text block height to picture height. if the text blocks get too thin (and you usually have only one line), they get lost among the images. i think a clear visual separation between image and text is also important.15:13
wpwrake.g., the tuxbrain images are clearly separated form the text, because they have a "frame". as a counter-example, my apartment drawing is not clearly separated from text, also because it has text of its own paralleling the text in the news. thus it gets harder to mentally segment the page into the text that provides the context and structure, and the images that go one level deeper.15:15
qi-botThe build was successfull, see images here: http://fidelio.qi-hardware.com/~xiangfu/compile-log/openwrt-xburst.full_system-08042011-1755/21:53
kristianpaulhow come... accum interupt is asserted low if i read a register first..22:12
kristianpaulno make sense because documentation said, it is overwritten !!22:13
kristianpaulif you dont reed it before int..22:13
larscmaybe the documentation is wrong22:15
larsci was working on a sound chip yesterday where the documentation had left and right swapped in a few places, was quite annoying22:17
kristianpauli think documentation is right and HDL implementation is outdated, i cant be  that the interrupt that tell cpu sample the accumalator dependent of a soft read procedure..22:22
GitHub93[milkymist] kristianpaul pushed 1 new commit to gps-sdr-testing: https://github.com/kristianpaul/milkymist/commit/14055f8514fb416553c69679b9bd36565e9685ff22:34
GitHub93[milkymist/gps-sdr-testing] missing status_read initialization, also added some debug signals - Cristian Paul Peñaranda Rojas22:34
ignatius_Is it possible to assign/use the 1.7GiB as the rootfs... is it a kernel setting, or something? Several years ago, I installed Debian on my Nanonote, and it worked perfect fline. It was able to use the entire NAND (~1.7GiB) as the rootfs, so I know it's possible.23:06
ignatius_Sorry to be a pest.23:07
kristianpaulno no, no pest no way :)23:16
ignatius_Cool.23:16
ignatius_:)23:17
kristianpaulyou can implement ubifs but i never tried it with debian23:17
kristianpaulhttp://en.qi-hardware.com/wiki/Ubifs23:17
kristianpaulroh: you awake?23:18
mthignatius_: NAND partitioning can be specified in arch/mips/jz4740/board-qi_lb60.c in the kernel sources23:27
mthbut reading the code there, I think it already uses as much as possible for the rootfs23:28
ignatius_Ah... Hmm.. I wonder why when I installed the Debian (Lenny) instructions the first time, I was able to boot the system, and its capacity was the entire "free" MTD (??) partition.23:31
ignatius_I messed up and deleted that kernel source... :(23:32
mthI'm looking at the 2.6.39 sources, not sure what version released kernels are built from23:33
ignatius_Since then, i've tried many, many different kernel sources.... no luck. Then I found one that did work, and it will boot the kernel from the first time, but not others that people have offered. And setting up the sources properly again, I am not able to product a kernel that boots without the "VFS" error.23:33
ignatius_The only thing about that kernel tree I remember, is that it was version "2.6.32.10" and it was part of a tool chain, and had the directory of "openwrt-xburst-xburst"23:37
ignatius_GIT/CVS... not sure..23:37
mthyes, you can build the kernel using the openwrt build system23:38
mthI build kernels for the Dingoo though, not for the NanoNote, but they're similar in hardware23:38
ignatius_Yeah. I know. I'm just saying that I don't remember WHICH toolchain I used.23:39
ignatius_Nod. I just got a Dingoo the last week.23:39
mthah, I told you this before?23:39
mthI remember I told someone, but not the nick23:39
ignatius_Nah. You didn't tell me that you got a Dingoo.23:40
mthyou probably used this last time: http://projects.qi-hardware.com/index.php/p/openwrt-xburst/23:42
mthby the way, you can find more Dingoo users on #dingoonity23:44
ignatius_Thanks.23:45
mthI'm going to get some sleep now23:48
ignatius_'Night.23:49
--- Sat Aug 6 201100:00

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