#qi-hardware IRC log for Tuesday, 2011-08-02

wolfspraulwpwrak: yes, working on the news00:23
wolfsprauland yes you are right, there are many things from with the news :-)00:27
wolfspraulfirst, it's too much00:27
wolfspraulthen, too few readers00:27
wolfspraulfinally, the news are too much like a technical log book, like a university following various student projects00:28
wolfspraulsocial and marketing stuff is missing00:28
wpwrakyeah, you picked a rather dry style. i think bit more "juice" will make it more digestable00:32
wpwrakhow many readers are there ?00:32
wolfspraulmaybe 50000:33
wolfspraulI'm not exactly sure, of course we could put more efforts into pushing it out00:34
wolfspraulbut then we become a news organization00:34
wpwrakwell, that's still a decent number 00:34
wpwrak;-))00:34
wolfsprauloh sure, could be more even00:34
wolfspraulanother approach would be to come out with snippets that are reusable by blogs00:35
wolfspraulneed to discuss with rejon 00:35
wolfspraulI agree that the current style is dead-end, more like an archive00:35
wpwrakyeah, smaller pieces may work better than one big load every months or, worse, a cobweb-covered encyclopedia every two months00:36
wolfspraul:-)00:36
wolfspraulneed to continue today actually, I'm at most 50% done with cleanup00:36
wolfspraulI guess there will be no video of your talk, unfortunately00:36
wpwrakah well, at least we have the slides :)00:43
wolfspraulwpwrak: you will like this http://en.qi-hardware.com/wiki/File:M1_rc3_0x37_u20s_mark_is_faked.png00:46
wolfspraulAdam thinks he found the first root cause of our rc3 yield/testing problems, in the form of fake TI parts :-)00:47
wolfspraulabout 1/3rd of fake parts mixed into the reel00:47
wolfspraulbah00:47
rejonyeah00:47
rejongood to have a narrative intro and then the bullet points are good00:47
wolfspraulrc4 sourcing will use digikey much more, maybe I start boomifying the bom a little00:48
rejonwe can call it ChangeLog00:48
rejon:)00:48
rejonQiLog00:48
rejonyeah, i can help push it out today00:48
wolfspraulrejon: not sure about log. that implies that you can read it later.00:48
wolfspraulno no, wait00:48
wolfspraulit's not finished yet00:48
rejoni know00:48
rejonif you want to write something up top, good00:48
rejonwe can also just have as a long list and encourage others to spread it by writing something up top00:48
wolfspraulwe need to rethink the news approach anyway00:49
rejonabove the list00:49
wolfspraulwe have to make it 'news' that new software is released for the Ben or m100:49
wolfspraulwe cannot just wait until our news are factually so irresistible that someone just has to report about them. that's crazy.00:49
wolfspraulI wanted to wait with spending a seriously bigger amount of time on this again until after I have m1 in stock.00:50
wolfspraulwhich is real-soon-now00:50
wpwrakwow. fake parts. lovely.00:56
wpwrakmake sure you mention the distributor :)00:57
wolfspraulbought in Shenzhen. but it's not easy to blame, or rather to draw the right conclusions.00:58
wpwrakwell. someone's making a few cents, knowing that this costs you hundreds of dollars. it's not too hard to draw conclusions from that :-)00:58
wolfspraulfirst it's sad, I waste a couple more thousand USD and a few weeks delay. ok. got over that.00:59
wolfspraulthen...00:59
wolfspraulboom is the right idea00:59
wolfspraulI love boom.00:59
wolfspraulit's so right00:59
wolfspraulwe have sourced all/most parts of the jtag-serial run with boom (digikey), we should try the same with m1, asap00:59
wolfspraulno, drawing conclusions is not easy01:00
wpwrak(boom) ah yes, one of these days i'll have to teach it tape and reel ...01:00
rejonok yes, every news item should be important01:00
wolfspraulin hindsight vision is 20/20, that is easy to agree on01:00
rejonits hard to have so much though for people to grab onto01:00
rejonesp. news writers01:01
rejonnot like M1 released01:01
wolfspraulbut I can tell you endless individual items where you have to use different sourcing approaches to be successful01:01
rejonor nanonote has wireless01:01
wolfspraullike we had terrible problems with the ethernet connector01:01
wolfsprauland I ended up paying 8 USD for each one01:01
wolfspraulnot nice01:01
wolfspraulon rc3 we have a high quality Taiwan Ethernet connector that costs 1.10 USD01:02
wolfspraulthere are many things that are simply unavailable on digikey, and maybe always will be01:02
wolfspraulyou remember marcan the other day with his galvos...01:02
wolfspraultouch panels01:02
wolfspraullcms01:02
wolfspraulsome parts have high moq/lead-time on digikey01:03
wolfspraulbut anyway. boom is the right thing.01:04
wolfsprauldigikey (and competitors) are probably the best choice for any part that they have stocked in quantity01:05
wolfspraulif digikey stocks 1000 or more of something, sourcing it there is a safe bet, economically and quality-wise01:06
wpwrakand once you make enough money to hire some muscle, you may want to pay whoever sold you those fake parts a friendly visit :)01:08
wolfspraulno, all fine01:09
wolfspraulI know how this happens01:09
wpwrak;-)01:09
wpwrakonce :)01:09
wolfspraulthe problem is on my end, as a manufacturer I need to be good on sourcing01:09
wolfspraul'good' can mean many different things, depending on part and other circumstances01:09
wolfspraulwe should make a sourcing guide too, it could help many people01:10
wolfspraulyou cannot have it all ways, I see the responsibility for buying this reel 100% on my side01:10
wolfspraulsuch parts should have been sourced from digikey01:10
wolfspraulthey would even have been cheaper, as we paid too much again :-) 77 cents!01:11
wolfspraulat digikey it's 10-20 cents depending on quantity01:11
wolfspraulwpwrak: can boom be extended to schmitt-triggers, or they have too many parameters?01:12
wolfspraulI mean to automatically pick one :-)01:12
wolfspraulwhat I could do about this reel indeed is to try to follow back into the supply chain01:13
wolfspraulI want to visit the crappy little factory that makes fake TI parts and mixes them into reels of real TI parts01:13
wolfspraul:-)01:13
wolfspraulsince I'm one of their customers :-)01:13
wolfspraulactually this could even work, if I spend the time for such an investigation. there's a huge spot market and all sorts of things are happening there.01:14
wolfspraulbut I rather focus on the good side01:15
wpwrak(schmitt-trigger) you'd identify them by gate type, not really by parameter. you could add a parameter that describes the gate, but that may get confusing01:15
qi-botThe build has FAILED, see log here: http://fidelio.qi-hardware.com/~xiangfu/compile-log/openwrt-xburst.full_system-08012011-1629/01:24
rohwhats boom?03:25
wpwrakroh: my BOM procesor03:25
rohjust found parts of it... you should add a guide how to add more stuff to it03:28
rohand how to use it03:28
wpwrakonce it's a bit more "finished". right now it's pretty chaotic, for me too ;-)03:30
rohheh03:33
rohcan you make it scrape octoparts?03:33
rohthey also have some pretty nasty self produced items ;) http://octopart.com/ocpt-10003-octopart-283747203:35
wolfspraulno03:35
wolfspraulbad03:35
wolfsprauloctopart is very bad, werner's boom is 100 times better03:35
wpwrakroh: nice part ;)03:36
wolfspraulthe idea of boom is to start with simple passive stuff, resistors, capacitors03:36
wolfsprauland then connect it to distributors with real stock03:37
wolfspraullike digikey, mouser03:37
rohwolfspraul: hm? i thought octoparts is only a meta-searchmachine listing all sellers?03:37
wolfspraulyes but it's a mess03:37
wolfspraulI think we should start with quality distributors first, and factor in their stock levels03:37
wolfsprauloctopart throws together companies with totally different business models03:38
rohwrong data? it whats the issue.. i was thinking of only searching via it. deciding stuff in boom03:38
wolfspraulnothing in stock, high moq, long lead-times03:38
rohoctoparts doesnt sell parts themselves as far as i know03:38
wolfspraulsure I search with octopart too, it's better than google for sure03:38
wolfspraulbut it's also not very different to google :-)03:39
wolfspraulroh: at present, boom will not consider shipping costs or lead-times, I think03:47
wolfspraulwpwrak: does it consider MOQ and stock-level? I forgot ;-)03:47
wolfspraulroh: boom will also automatically look at volume discounts and pick the volume that is cheapest for the upcoming run03:48
wolfspraulbasically it helps remove a lot of tedious work from sourcing03:48
wolfspraulone of the next planned features was to add support for reels and digi-reels (@digikey) - maybe trays later03:49
rohwell.. cheap isnt always good as we learnt ;)03:49
wolfspraulwhere did we learn that?03:49
rohah. right.. you didnt even buy it cheap03:49
wolfspraul:-)03:50
wolfspraulsourcing is hard, I have no illusions03:50
rohi just checked the U20 on octoparts.. looks like theyre all there03:50
rohjust lots of variants of that thing...03:50
wolfspraulboom has the exact right approach to help reduce costs significantly, and increase quality03:50
rohi guess it doesnt matter if you buy at dk mouser or farnell.. but there are seldomly the cheapest03:51
wolfspraulimagine you have a free hardware project and with a few clicks you can have a complete shopping list for digikey - very cool I think03:51
rohi wonder whats the difference between the 80cent and the 11cent variant in the same case03:51
wolfspraulnah03:51
wolfspraulsee that's the thing. digikey may very well be the cheapest. or mouser, or farnell.03:51
wolfspraulthe problem is there are huge databases03:52
wolfspraulhow to efficiently scan them? software!03:52
rohexactly.. i was just thinking of using the advantage that somebody already did the gruntwork to unify access to all these distributors03:52
wolfsprauland you think octopart did that?03:52
wolfspraulme too, a while back :-) until I just tried to use them, and then you see it's all wrong.03:53
rohyes. they started with scaping and later made deals with the distros remaining03:53
wpwrakwolfspraul: yes, stock and MOQ (and quantity discounts). but not lead time.03:53
wolfspraulthey want to be like google, just a little more specialized03:53
wolfspraulroh: all wrong, really. does not match with what I've learnt in sourcing at all.03:53
rohalso removed some unwilling distributors who threatened them03:53
wolfspraulboom is the right approach, start with the bom out of kicad, try to parameterize. start with the cheapest components, connect to databases of known high-quality distributors.03:54
rohhttp://octopart.com/api/overview, then filter by a list of known good distributors.03:54
wolfspraultheir api will only obfuscate the underlying reality03:54
wolfspraulthey are not adding any value imo03:55
wolfspraulboom builds the shopping list out of the bom which is out of kicad03:55
rohit will add the value of them keeping it up to date with changes at all distributors for you.. abstraction.03:55
wolfspraulexactly what I don't want in sourcing :-)03:56
rohalso.. nasty legal trouble if you scrape too much03:56
wolfspraulI want to factor in reliable shipping and lead times as well, in addition to reliable MOQ and stock levels :-)03:56
wolfspraulI expect zero legal trouble for a tool like boom, in fact we should ask digikey to sponsor it.03:56
rohwolfspraul these you can cherry pick later when you have a list of selections done03:56
wolfspraulno it won't work, it's a crazy amount of work03:57
wolfspraulI'm a boom believer03:57
wolfspraul:-)03:57
wolfsprauloctopart was my sinner times03:57
wolfspraulI can check again, but I doubt their model will fly. we will see. I don't understand what value they add for sure.03:58
wolfspraulif I were digikey I would indeed not like something like octopart03:58
wolfspraulbut I wouldn't fight it either, just ignore. it's like google in the end.03:58
wolfspraulok, maybe block their IP if it gets too bad :-)03:59
wolfspraulthe reason I would not like it is not because they allow comparison with others, but because I would be worried that the quality of my service is ruined by them.04:00
wolfspraullet's say digikey's stock numbers are real, and the stock numbers of some other 'website' are fake04:00
wolfspraulok?04:00
wolfspraulwho will audit the other distributors?04:00
wolfsprauloctopart?04:00
wolfspraulno way04:00
wolfspraulI don't want to be in that pool, as a quality distributor with real stock.04:00
rohwolfspraul: they have some program running it seems to make sure they get real data04:01
wolfspraul'real data', ha ha04:01
wolfspraulI should setup a website just to proove my point. little 'incoming order redirection' business04:01
rohcheck their blog. seems the users put quite some pressure on the distributors to get listed after delisting themselves.04:02
wolfspraulyou mean octopart users want to see all distis? for sure04:02
wolfspraulI use it too04:02
wolfspraulquick and dirty overview04:02
wolfspraulbut useless for sourcing04:02
wolfspraulI tried to tell you why if I were at digikey, I would not like it.04:03
wolfspraulbecause for example take stock level04:03
wolfspraulwho audits the others?04:03
wolfspraulnobody!04:03
rohwolfspraul: why? it means more business coming your way.04:03
wolfspraulso that means my stock levels, let's say they are real, are compared with fake stock levels of my competitors04:03
rohwolfspraul: octopart will. because its in their interrest.04:03
wolfspraulnot fair04:03
wolfspraulI don't want to be in that market/pool.04:03
wolfspraulno they cannot04:03
wolfspraulI highly doubt that04:04
wolfspraulno distributor would let octopart people audit their warehouse04:04
rohfor me for example dk is pain. delivery times, buerocratic madness, shitty payment foo etc.04:04
wolfspraulwhich ones do you like?04:04
rohmeans for me boom is useless as long as it doesnt list real european sellers. not us sellers with a bad customercenter only in europe04:05
wolfspraulfully understood. that's valuable. so which ones do you like?04:05
rohi get most of stuff from segor, rs components, seldomly farnell, lots of conrad and voelkner (really fast shipping)04:05
wolfspraulwell great :-)04:05
wolfspraula lot of good work for boom04:05
rohi use mercateo a lot, but its not that well done for electronic parts, more for 'devices'04:06
wolfspraulthe boom approach is to start with cheap parts that are relatively easy to parameterize04:06
wolfspraulthat will clear out your bom a lot, and leave you with a much smaller set to source 'manually'04:06
rohespeciually because they do central billing. not having to have accounts in N vendor shops rocks.04:06
wolfspraulsure04:06
wolfspraulso you use octopart just for general searching as well, it sounds04:07
wolfspraullike google, just a bit more specialized04:07
wolfspraulit's not a sourcing tool04:07
rohi get one bill per order, even if ther are multiple distros in play that way04:07
wolfspraulone bill from octopart?04:07
rohi'd like to use octoparts more but its currently too us centric for me04:07
wolfspraulwho is giving you one bill for multiple distros?04:07
rohwolfspraul: mercateo. i dont order at conrad or so directly. i search and order via mercateo (b2b sap driven platform) and get one bill, and it optimizes for cost over the distros which are in direct competition04:08
wolfspraulah nice04:08
wolfspraulmaybe mercateo is doing this the right way :-)04:09
wolfspraulto be honest I don't understand octopart's model, so I should not qualify it either. they probably have a plan, and/or it may be changing over time as they learn. I don't know, I am not following closely.04:09
rohhelped me a lot. sadly its often not as fast to use as i'd like and there are sometimes 20 very similar or identical things which it doesnt get is 'one' and thus optimizing still needs some handywork to be perfect04:10
wolfspraulI can imagine that it's cracking quite a bit, with distis being kicked out, kicking out themselves, kicked back in, etc. etc.04:10
wolfspraulthat's not surprising :-)04:10
rohe.g. if ordernumbers for different packagesizes do'nt get put into the db right to match one product or so04:10
wolfspraulyep04:10
roha.. forgot some important sellers (non-mercateo) reichelt electronics and pollin04:10
rohmaybe i'd use digikey if they would do 24h shipping for 4E/package04:11
wolfspraulalso depends on where their warehouses are04:12
rohand dont bury me with import foobar paperwork and credit contracts04:12
wolfspraulfor example mouser has a warehouse in HK or Taiwan, so it's great in Taiwan04:12
wolfspraulwell, exactly. those details are important!04:12
wolfspraulif you talk about 'sourcing' as in 'getting the real stuff fast and cheap'04:12
wpwrakroh: digi-key are actually among the easiest to order internationally from :)04:13
rohwpwrak: maybe if you are already at the end of the world ;)04:13
rohwpwrak: as a european without a credit card its pretty annoying. they cant do timely bank transfer or even 'einzug' for some reason to its slow04:15
rohand if you order a beagleboard you get paperwork asking if youre a terrorist and why you import us stuff04:15
wolfspraul'without credit card' is unimaginable in the us04:16
wpwrakroh: duh. why don't you add other interesting constraints ? like payment only in cash, with reichsmark ? ;-)04:16
wolfspraulI remember seeing tourists stranded and crying, whole families, when finding out that they cannot get the rental car that was fully prepaid in New York, because the computer system insists on a credit card for insurance reasons04:17
wolfspraulwell, roh has the right to not like them and prefer others, for his reasons04:17
wpwrakdon't digi-key also accept bank transfers ?04:17
wolfspraulbut he is doing exactly what I think octopart will be weak in, he is zooming in on particular details of an actual purchase transaction04:17
wolfspraulas a US business I can imagine that they see little additional sales in the 'have no credit card' market04:18
wolfspraulmaybe only for large businesses, setup a contract etc.04:18
wpwrakwolfspraul: i think they do handle "no credit card", too. but of course, then you have to wire the money first.04:19
wolfspraulI'm just explaining the US perspective to roh.04:19
wolfspraulnot that he has to accept it, but that's their perspective, so yes, roh goes with German distis and it's great04:19
rohwolfspraul: well.. outside of the 3rd world there are well working debit systems. even in taiwan i got money with it. no reasons for prehistoric kludges working like electronic versions of a 'cheque'04:19
wpwrakwolfspraul: or the alternative would be to have a credit account with digi-key themselves, yes. of course that's even more difficult.04:19
wolfspraulI didn't ask you to accept it, or like it. just explaining :-)04:20
rohi mean.. i have _never_ filled in a cheque. nobody does that.04:20
rohmost people in germany only have visa or mastercards if they are for free and for ordering outside of europe via the internet or on the road. in germany there isnt any need for a creditcard. everybody takes debit and is happy. it gurantees you get your money as a merchant (when using card and pin, which basically every store does)04:21
wolfspraulyes but in the US it's not like that, and digikey is a US company04:22
wpwrakroh: now you just have to convince digi-key to move to germany and your problem is solved ;-)04:22
rohand its not ordering only from inland. its ordering preferrably inside europe. simply because money moving is easier and does cost not a cent (by law)04:22
rohwpwrak: well.. they got a german webshop. just no good deals (18E shipping if you order <65E)04:23
rohwpwrak: mostly ordering from the us is the same level or annoyment and work as ordering directly from china04:24
wpwrakroh: only EUR 18 ?? wow.04:24
rohwpwrak: well.. the market level is usually 4-6E shipping below 100E value. time is 2-3 working days, for some 24hours for stock parts04:24
wpwrakhere it's USD 40, and that's dirt cheap in comparison to what shipping costs at other places04:24
wpwrak"market level" = within germany, i guess04:25
wolfsprauldoes anybody know where digikey has warehouses?04:25
wolfspraulall in one place?04:25
rohwpwrak: lets make it central europe. france and similar isnt much more expensive (maybe 7euros or so04:25
rohwolfspraul: from what i can read the stuff gets sent from texas to cologne and then to the customer. no real stock here. only handling04:26
wolfspraulall I've ever seen at distis was one number for stock-level, as if they have 1 warehouse04:26
rohhm. i should talk to the us hackers on the camp if they know people at octoparts.. would like to meet them and pick their brains.04:27
wolfspraulI think any sourcing tool must be flexible and allow the user to override preferences. just the direction boom is taking.04:27
wolfspraulbecause the quality of those parameters may also change over time.04:27
rohafter all.. they are in the middle of something we all need somehow in some way04:27
wolfspraulsay digikey gets new management, and they decide to work a little differently with the stock levels in the database, to squeeze out more profits.04:28
wolfspraulwho will notice? how will customers react? it must be in the hands of those making the purchase decisions (the ones doing the sourcing)04:28
wolfspraulexactly like roh says he has 'his' distributors that he trusts and uses for this or that reason04:29
rohwolfspraul: my trust is not easy to earn. that only works by good service and not making complications. and if they happen there needs to be a will to fix stuff fast and make it not happen again.04:29
wolfspraulroh: yes, and they are in a very bad spot. because they deal with a lot of data without doing actual sourcing :-)04:29
wolfspraulthey can try to build feedback channels, but it will be all a quality mess04:30
rohmeans if some distro fucks up the same twice i will not use them again if possible, even if the alternative is more expensive to avoid the trouble04:30
wolfspraulquite normal04:30
wolfspraulmost people who work in sourcing have 'approved vendor lists' and what not04:30
wolfspraulroh: the alternative is not more expensive because it helps you save the cost of cleaning up the mess otherwise04:31
wolfspraulso you are actually trying to find the cheapest one, the actually cheapest one04:32
rohexactly. says somebody with experience. ;)04:32
wolfspraulwhich is an ongoing and never-ending process, in hardware04:32
wolfspraulmarcan: what is the largest display one could create with your laser projector? at what distance can it still project a recognizable image?04:50
wolfspraulcan you project a 20x20m big image? on something over a distance of 100m or more? :-) just curious how to calculate this and what factors affect it04:51
wolfspraulaw: hi :-)05:38
awwolfspraul, hi es05:38
wolfspraulI looked at the numbers a bit more http://en.qi-hardware.com/wiki/Milkymist_One_run_3_schedule#Classifications_of_Failure05:39
wolfspraul    Impedance err - 10 / 9005:39
wolfspraul    Current - 2 / 9005:39
wolfspraul    Flash - 11 / 9005:39
wolfspraul    VGA DDC - 6 / 9005:39
wolfspraul    No VGA Screen - 23 / 9005:39
wolfspraul    Audio Circuit - 2 / 9005:39
wolfspraul    Ethernet ARP - 1 / 9005:39
wolfspraul    Video I2C - 1 / 9005:39
wolfspraul    MIDI - 11/ 9005:39
wolfspraul    USB - 16 / 9005:39
wolfspraul    Rendering - 3 / 90 05:39
wolfspraulhow many of those do you think we can assume to be potentially fixable with a u7/u19/u20 replacement?05:39
awyup..the root cause of No Screen issue is by Schmitter-Trigger part is bad05:40
wolfspraulthe 'vga no screen' (23), and 'midi' (11) and 'usb' (16)?05:40
awsupposedly that we can fix most 'No VGA Screen' and 'MIDI' if they are related to that the same Schmitt-Trigger ship failure05:41
wolfspraulhow about usb, flash, impedance?05:41
awthe 'usb' I've not investigate it though05:41
wolfspraulok05:41
awflash...failures don't know yet05:42
wolfspraulimpedance?05:42
awthis afternoon I'll see if Schmitt-Trigger part influence with 'impedance' failure.05:42
wolfspraulok05:43
wolfspraulwell it looks like we don't need xray after all, right now05:43
wolfspraulfirst replace all u7/u19/u2005:43
wolfspraulinteresting about those 3 rendering cases05:43
wolfspraulso those were boards that already were rendering, but then there was a problem, right?05:43
awsince the Schmitt-trigger parts are used in MIDI & VGA circuits, so the most those failures must be caused by it.05:44
wolfspraulthat's clear05:44
awyes, temporarily no need to take X-ray now, we see later here what I can spot soon05:44
wolfspraulI'm looking at the list overall05:44
wolfspraulusb, impedance, flash - still some big blocks with unknown root causes05:45
wolfsprauland also those strange 3 'rendering' cases, let's see how this develops later05:45
awalso the Schmitt-trigger chips are connected relevantly to +5V as well as 3V3, so it's also possibly dependent to 'impedance' failures. I'll know soon.05:46
wolfspraulok05:46
wolfspraulwell it looks like you are on a good path05:46
wolfspraulyou won't run out of work soon :-)05:46
wolfspraulhow many boards do you have with all 3 u7/u19/u20 already replaced with good ones?05:47
wolfspraulor do you plan to replace all 270 again?05:47
awyes, ;-) I got see more boards then get solid conclusions.05:47
wolfspraulif you have boards where all three u7/u19/u20 are good, and you plan no further rework on them, and they pass 100%, then you can try to do long-term rendering testing with them05:48
awfrom those data I marked as OOO or OXX etc...the results are unreliable to the rc3 boards on both MIDI & VGA circuits05:48
wolfspraulbut only if you have 100% finished boards, otherwise we pickup too much noise05:49
awso I planed to rework them all (270pcs) with new parts later.05:49
wolfspraulsince you have enough power supplies now, you can just let 5 or 10 boards render for 24h or more05:49
wolfspraulah ok05:49
wolfspraulthen we wait with long rendering tests until after that05:49
wolfspraulotherwise too much noise05:49
awok05:50
wolfspraulin the meantime you can zoom in on all those other blocks, as you plan already05:50
wolfspraulusb, impedance, flash, also the more exotic ones like audio codec...05:50
awthe audio codec reason I knew already caused by L1's footprint is too big05:51
wolfspraulis your place still clean and everything under control organizationally?05:51
wolfspraulI mean in terms of cleanliness or chaos with all the boards...05:51
wolfspraulwatch out that your organization doesn't break down :-)05:51
awso the L1 is cold soldering while I tested...those ones we can fix it in rc405:52
wolfspraulcan you repair that rc3 board?05:52
awi could only step by step works now05:52
wolfspraul(with audio codec problems)05:52
wolfspraulsure05:52
wolfspraulbig blocks first05:52
wolfspraulkeep your place well organized05:52
awsure, for L1 is easy though so that I've not do it.05:52
wolfspraul90 is a little tough already without a real line :-)05:53
awI just focus on most high failures firstly05:53
wolfspraulwhen you go to minbo for the 270 u7/u19/u20, maybe they can do some other reworks on the side?05:53
wolfspraulsince you are there already, you can use them maybe :-)05:53
awthese two days I'll call if minbo can rework for it before the new parts are arrived maybe 3 days later.05:55
wolfspraulsure, sounds good05:56
awwell...the ods I need to make one column that shows up u7/u19/u20 I replaced all. and also is going to cancel the words of 'Available' to other words?05:57
wolfsprauldelete all 'available'05:57
awi think now the 'Available' must be include all pass though : rendering/parts/etc..05:58
awyup05:58
wolfspraulno, let's not overdo it05:58
wolfsprauljust delete all available for now05:58
wolfsprauland after the full u7/u19/u20 rework, you can delete the u7/u19/u20 column as well05:58
awalso good.05:58
wolfspraulit's a testing report, we don't need to record things like that which are basically 'extended' production steps05:58
wolfspraullike fix205:59
wolfspraulthe idea of the testing table is to give us a quick and efficient overview over quality issues, in preparation for the next run or future products05:59
wolfspraulwhich is exactly what it does now, also your grouping into categories is good06:00
marcanwolfspraul: the projector has a scan angle, after that you can project as large as you want (by taking the projector farther and farther away). Of course, brightness diminishes as you go farther, and also how collimated the laser is affects things12:41
marcan10mx10m sounds eminently doable12:41
marcan20m by 20m, if you have a decent laser or in near-darkness, why not12:41
wolfspraulmarcan: nice ;-) from what distance can you still project something?12:48
wolfspraulsay you want to project a 10x10m image, but from 100m distance12:48
wolfspraulwill that work?12:48
wolfspraul200m?12:48
wolfspraul500m?12:49
wolfspraul:-)12:49
wolfspraulwpwrak: brownout?13:49
wpwrakpc needs a new power supply. it dies every once in a while. let's see what i missed ...13:58
wpwrakah, just the sourcing fun14:02
wpwrakaw: btw, nice catch ! :)14:02
wolfspraulaw_: any news for the news hungry people here?14:04
larsc"pizza is ready!" ;)14:05
aw_oah~ i need pizza. ;-)14:08
wpwraklarsc:you shouldn't say such things. no everyone will be gone for a while now.14:09
kristianpaulnews news !14:09
aw_just rework few rc3 boards only but also realized that drying boards after cleaning is important...;-)14:10
wolfspraulkristianpaul: ah hi. I need your help with the news :-)14:10
wolfspraulthere's a lot of URLs missing14:10
wolfspraulcan you add some? Like you say you ported the namuru correlator to mm - source url?14:10
kristianpaulok14:11
wolfspraulthen the url to softgnss matlab code14:11
kristianpaulmom14:11
wolfspraulno rush14:11
wolfspraulI will edit the language, just help me with the URLs.14:11
wolfspraulaw_: what reworks did you do?14:12
aw_i reworked my 0x2f (the first one we reworked before) then seems not to let it fully dried after rework/clean u7u19u20 then I power up and can't even reconfigure...14:12
aw_just tried to use my rough rest schmitt-triggers(18pcs) on some boards...i need to very carefully soldering, since I broke new one. :(14:13
aw_phew~14:13
wolfspraulis 2F working now?14:13
aw_yes... ;-)14:14
wolfspraulwell then :-)14:14
wolfspraulany new discoveries on usb or impedance?14:15
aw_last 87pcs reworked, i cleaned then let them through over night...then not easy to cause it. so if not fully dried, seems that flash digital signals went wrongly though at the beginning. 14:16
aw_no news on usb and impedance14:16
GitHub6[milkymist] kristianpaul pushed 1 new commit to gps-sdr-testing: https://github.com/kristianpaul/milkymist/commit/8136a61d8e3e48770a786ee5c90c08cc45f3e45214:17
GitHub6[milkymist/gps-sdr-testing] New condition added to Counter, thanks Artyom G - Cristian Paul PeƱaranda Rojas14:17
aw_do reworks that i can only 1.) take them apart one time 2.) then soldering one time 3.) then clean...can't focus on checking others...brain will get messy. ;-)14:18
aw_seems that talking milkymist should have stayed in other IRC there? ;-)14:22
wolfsprauldoesn't matter much14:23
wolfspraulthis is something called 'off-topic'14:23
wolfspraulin some channels people are easily offended about someone talking about something 'off-topic'14:23
wolfspraul#qi-hardware is about copyleft hardware, manufacturing. #milkymist about the Milkymist SoC and all that connects to it14:24
wolfspraulso there is overlap :-)14:24
wolfspraulkristianpaul: thanks for the links! what is the osgps homepage?14:58
wolfspraulI googled but find a lot of inconclusive links...14:58
wolfspraulI think it's here http://sourceforge.net/projects/osgps/14:59
wolfspraulI guess they write it OpenSourceGPS14:59
kristianpaulah, yes i forgot http://home.earthlink.net/~cwkelley/15:00
wolfspraulwhich makes more sense, so if you don't mind I'll write that15:00
kristianpaulyes15:00
kristianpaulsure, go ahead15:00
wolfspraulearthlink or sf?15:00
kristianpaulboth? :_)15:00
wolfspraulI think sf.net is more active15:01
wolfspraulok, both15:01
kristianpaulyes, but is code, i mean sf15:01
kristianpaulbut  docs are in earthlink15:01
wolfspraulkristianpaul: perfect, I think the entry is perfect now. thanks for your help!15:07
wolfspraulwill move it around later15:07
wolfspraulstill more stuff from Werner, and a presentation from Sebastien15:07
wolfsprauloh and atben/atusb15:08
wpwrakwolfspraul: yup,  atben/atusb produced and available at tuxbrain :)15:20
wolfsprauloh sure, and sales numbers15:21
wpwrakwolfspraul: you also don't say explicitly that M1rc3 was smt'ed15:21
wpwrakwolfspraul: do you want to mention my nanonote test pint map ? http://lists.en.qi-hardware.com/pipermail/discussion/2011-June/008140.html15:21
wolfspraulwait wait15:22
wolfspraulthere's too many things15:22
wpwrakwolfspraul: maybe also the navigation board, for which volunteers are wanted ? http://lists.en.qi-hardware.com/pipermail/discussion/2011-June/008344.html15:22
wpwrakyeah, busy two months ;-))15:22
wpwraki think that's all i have15:23
wpwrakbtw, the large images are convenient but they also make it very hard to read the page. how about using relatively small thumbnails, so that the text doesn't get interrupted to badly15:23
wpwrak?15:23
wolfspraulwhich images?15:24
wpwraki.e., i think it would be nice if there was a bit of a logical flow in the news items. because some of them are connected. but it's hard to see this at the moment.15:24
wolfspraullet me first finish more links and images :-)15:24
wolfsprauloh sure15:24
wolfspraulall will be rearranged15:25
wolfspraulno worries15:25
wolfspraulthis is completely unreadable now15:25
kristianpauloh yeah..15:25
wpwrakmost of them ;-) anything taller than maybe 200 pixels interrupts the flow. also, if they all had about the same height, the structure may look tidier, too15:26
kristianpauli was to point same (thumbails)15:26
wpwrak(unreadable) okay, then we all agree ;-)15:26
wolfspraulthere will be 4 presentations, I like the idea of spreading them all out for quick overview, but it's big15:26
wolfspraulmaybe move elsewhere...15:26
wpwrakthe presentation overview collages look very nice. just not in the middle of text ;-)15:28
wolfspraulI need to pull in more images that you created on the downloads site15:31
kristianpaulhey, news collage not bad name :-915:31
kristianpauls/:-9/:-)15:31
wolfspraulwpwrak: just added another 36-page presentation :-)16:04
wolfspraulthis is just for fun, no worries16:04
wolfspraulI will probably remove it or move elsewhere, just playing with the concept16:04
wolfspraulnow that this works so nicely, I'll try the 4-page m1 brochure as well :-)16:04
wpwrak*grin*16:06
wolfspraulthere's a very nice picture on page 32 of that presentation that I hadn't seen before16:06
wpwrakyeha, the brochure/flyer/box design of M1 may also be interesting16:06
wolfspraulI think it's the infamouse m1 rc1 video-in rework16:06
wolfsprauloh, box design16:06
wolfspraultrue!16:06
kristianpaulcollage news for sure ;)16:07
wolfspraulseveral people attempted that rework, but only 1 got it to work :-) I guess that is documented in that picture16:07
wpwraki think if you can shrink/reduce each picture to a small thumbnail, the overall look should be quite nice16:07
wolfspraulhave to ask Sebastien for the original...16:07
wolfspraulI do the aesthetical stuff, together with more editing, later16:07
wolfspraulthe idea with the box design is really good16:07
wolfspraulwpwrak: check out that rework pic! :-)16:08
wolfspraulp32 of sebastien's presentation16:08
wpwrakyeah, nice :)16:08
wolfspraulthat was on rc1, for some reason some pins got wired up wrongly, and nobody had seen it in the reviews16:08
wolfspraulso we tried to verify that the chip worked with the correct pins, but that rework proved exceedingly difficult16:09
wpwrakoh, mwalle did it too a few days ago. so now the global count is 2 :-)16:09
wolfsprauloh an rc1 board?16:09
kristianpaulfor ac97 chip16:09
wpwrakdunno which board. but it sounded like that nasty rework, yes16:10
wolfspraulhmm. what did he do? replace the lm4550b with wm9707?16:10
kristianpaulyes16:11
wolfspraulok, nice16:12
wolfspraulhardware upgrade in the field16:12
wolfspraul:-)16:12
wolfspraulkristianpaul: I'm saying that the nyon cat is on a 'silicon wafer'. is that even accurate?16:15
wolfspraulwhat is the material it landed upon?16:15
kristianpaulyes is Si16:16
wolfspraulok good16:17
kristianpauli dont know what size is valid to say wafer..16:18
wolfspraulmaybe 2'' or 4'' wafer?16:19
wolfsprauldoesn't matter16:19
wolfspraulwe already spend 100 times more energy on verifying facts than any other news outlet :-)16:19
kristianpaulheh16:20
wolfspraulif people would realize the impossible task most journalists are facing every day, they wouldn't complain about news reports full of inaccuracies anymore :-)16:21
wolfspraulbut ok, we know these projects quite well, so it's relatively easy for us to make it quite accurate16:21
wolfspraulbut for example, I cannot even follow the Elphel cameras well16:21
wolfsprauleven though I see things happening16:21
wolfspraulbut it's too much16:21
wolfspraulcalling it a day, n8 everybody16:25
--- Wed Aug 3 201100:00

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